Theory and Analys~s
S E C O N D
LAMINATED COMPOSITE PLATES and SHELLS
E D I T I O N
J . N . REDDY
CRC P R E S S
Boca Raton London New York Washington, D.C.
Library of Congress CataloginginPublication Data
Reddy, I. N. (Junuthula Narasimha), 1945Mechanics of laminated composite plates and shells : theory and analysis I J.N. Reddy.2nd ed. p. cm. Rev. ed. of: Mechanics of laminated composite plates. c1997. Includes bibliographical references and index. ISBN 0849315921 (alk. paper) 1. Plates (~n~ineerin~)Mathematical models. 2. Shells (Engineering)Mathematical models. 3. Laminated materialsMechanical propertiesMathematical models. 4. Composite materialsMechanical propertiesMathematical models. I. Reddy, J. N. (Junuthula Narasimha), 1945. Mechanics of laminated composite plates. 11. Title.
This book contains information obtained from authentic and highly regarded sources. Reprinted material is quoted with permission, and sources are indicated. A wide variety of references are listed. Reasonable efforts have been made to publish reliable data and information, but the author and the publisher cannot assume responsibility for the validity of all materials or for the consequences of their use. Neither this book nor any part may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, microfilming, and recording, or by any information storage or retrieval system, without prior permission in writing from the publisher. The consent of CRC Press LLC does not extend to copying for general distribution, for promotion. for creating new works, or for resale. Specific permission must be obtained in writing from CRC Press LLC for such copying. Direct all inquiries to CRC Press LLC, 2000 N.W. Corporate Blvd., Boca Raton, Florida 33431.
'kademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation, without intent to infringe.
Visit the CRC Press Web site at www.crcpress.com

O 2004 by CRC Press LLC
No claim to original U.S. Government works International Standard Book Number 08493 1592 1 Library of Congress Card Number 2003061067 Printed in the United States of America 1 2 3 4 5 6 7 8 9 0 Printed on acidfree paper
To the Memory of
My parents, My brother, My brother inlaw, My father inlaw, Hans Eggers, Kalpana Chawla, . . .
About the Author
J. N. Reddy is a Distinguished Professor and the inaugural holder of the Oscar
S. Wyatt Endowed Chair in the Department of Mechanical Engineering at Texas A&M University, College Station, Texas. Prior to his current position, he worked as a postdoctoral fellow at the University of Texas at Austin (197374), as a research scientist for Lockheed Missiles and Space Company (1974), and taught at the University of Oklahoma (197551980) and Virginia Polytechnic Institute and State University (19801992), where he was the inaugural holder of the Clifton C. Garvin Endowed Professorship. Professor Reddy is the author of over 300 journal papers and 13 text books on theoretical formulations and finiteelement analysis of problems in solid and structural mechanics (plates and shells), composite materials, computational fluid dynamics and heat transfer, and applied mathematics. His contributions to mechanics of composite materials and structures are well known through his research on refined plate and shell theories and their finite element models. Professor Reddy is the first recipient of the University of Oklahoma College of Engineering's Award for Outstanding Faculty Achievement in Research, the 1984 Walter L. Huber Civil Engineering Research Prize of the American Society of Civil Engineers (ASCE), the 1985 Alumni Research Award at Virginia Polytechnic Institute, and 1992 Worcester Reed Warner Medal and 1995 Charles Russ Richards Memorial Award of the American Society of Mechanical Engineers (ASME). He received German Academic Exchange (DAAD) and von Humboldt Foundation (Germany) research awards. Recently, he received the 1997 Melvin R . Lohnmnn Medal from Oklahoma State University's College of Engineering, Architecture and Technology, the 1997 Archie Higdon Distinguished Educator Award from the Mechanics Division of the American Society of Engineering Education, the 1998 Nathan M. Newmark Medal from the American Society of Civil Engineers, the 2000 Excellence i n the Field of Composites Award from the American Society of Composite Materials, the 2000 Faculty Distinguished Achievement Award for Research, the 2003 Bush Excellence Award for Faculty i n International Research award from Texas A&M University, and 2003 Computational Structural Mechanics Award from the U.S. Association for Computational Mechanics. Professor Reddy is a fellow of the American Academy of Mechanics (AAM), the American Society of Civil Engineers (ASCE), the American Society of Mechanical Engineers (ASME), the American Society of Composites (ASC), International Association of Computational Mechanics (IACM), U.S. Association of Computational Mechanics (USACM), the Aeronautical Society of India (ASI), and the American Society of Composite Materials. Dr. Reddy is the EditorinChief of the journals Mechanics of Advanced Materials and Structures (Taylor and Francis), International Journal of Computational Engineering Science and International Journal Structural Stability and Dynamics (both from World Scientific), and he serves on the editorial boards of over two dozen other journals.
Contents
Preface to the Second Edition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .xix
. Preface to the First Edition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxi
1 Equations of Anisotropic Elasticity. Virtual Work Principles.
and Variational Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1
1.1 FiberReinforced Composite Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .I 1.2 Mathematical Preliminaries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 . 1.2.1 General Comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3 1.2.2 Vectors and Tensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .12 . 1.3 Equations of Anisotropic Entropy
1.3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 1.3.2 StrainDisplacement Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .13 1.3.3 Strain Compatibility Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 1.3.4 Stress Measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 1.3.5 Equations of Motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .19 1.3.6 Generalized Hooke's Law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 . 1.3.7 Thermodynamic Principles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .38 1.4 Virtual Work Principles . 1.4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 1.4.2 Virtual Displacements and Virtual Work . . . . . . . . . . . . . . . . . . . . . . . .38 1.4.3 Variational Operator and Euler Equations . . . . . . . . . . . . . . . . . . . . . . .40 . 1.4.4 Principle of Virtual Displacements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 . 1.5 Variational Methods 1.5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 1.5.2 The Ritz Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 1.5.3 WeightedResidual Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 . 1.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 . Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
. References for Additional Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .78
2 Introduction to Composite Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
2.1 Basic Concepts and Terminology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .81 . 2.1.1 Fibers and Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 . 2.1.2 Laminae and Laminates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .83 2.2 Constitutive Equations of a Lamina . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .85 . 2.2.1 Generalized Hooke's Law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85 86 2.2.2 Characteristics of a Unidirectional Lamina . . . . . . . . . . . . . . . . . . . . . . .
X
CONTENTS
2.3 Transformation of Stresses and Strains . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .89 2.3.1 Coordinate Transformations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .89 2.3.2 Transformation of Stress Components . . . . . . . . . . . . . . . . . . . . . . . . . . .90 2.3.3 Transformation of Strain Components . . . . . . . . . . . . . . . . . . . . . . . . . . .93 2.3.4 Transformation of Material Coefficients . . . . . . . . . . . . . . . . . . . . . . . . . .96 2.4 Plan Stress Constitutive Relations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .99 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103 References for Additional Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 3 Classical and FirstOrder Theories of Laminated Composite Plates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .109 3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .109 3.1.1 Preliminary Comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109 3.1.2 Classification of Structural Theories . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109 3.2 An Overview of Laminated Plate Theories . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110 3.3 The Classical Laminated Plate Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .112 3.3.1 Assumptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 . 3.3.2 Displacements and Strains . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113 3.3.3 Lamina Constitutive Relations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117 3.3.4 Equations of Motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .119 127 3.3.5 Laminate Constitutive Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.6 Equations of Motion in Terms of Displacements . . . . . . . . . . . . . . . . 129 3.4 The FirstOrder Laminated Plate Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132 3.4.1 Displacements and Strains . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .132 3.4.2 Equations of Motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .134 3.4.3 Laminate Constitutive Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .137 3.4.4 Equations of Motion in Terms of Displacements . . . . . . . . . . . . . . . . 139 3.5 Laminate Stiffnesses for Selected Laminates . . . . . . . . . . . . . . . . . . . . . . . . . . . 142 3.5.1 General Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142 3.5.2 SingleLayer Plates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .144 3.5.3 Symmetric Laminates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148 3.5.4 Antisymmetric Laminates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .152 3.5.5 Balanced and QuasiIsotropic Laminates . . . . . . . . . . . . . . . . . . . . . . . 156 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157 References for Additional Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .161
4 OneDimensional Analysis of Laminated Composite Plates . . . . . . . . . 165
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .165 4.2 Analysis of Laminated Beams Using CLPT . . . . . . . . . . . . . . . . . . . . . . . . . . . 167 ... 4.2.1 Governing Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . .,. . . . . . . . . 167 4.2.2 Bending . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .169 4.2.3 Buckling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .176 4.2.4 Vibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
4.3 Analysis of Laminated Beams Using FSDT . . . . . . . . . . . . . . . . . . . . . . . . . . .187 . 4.3.1 Governing Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187 . 4.3.2 Bending . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .188 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192 . 4.3.3 Buckling . 4.3.4 Vibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .197 . 4.4 Cylindrical Bending Using CLPT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .200 . 4.4.1 Governing Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200 . 4.4.2 Bending . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .208 . 4.4.3 Buckling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .209 . 4.4.4 Vibration . 4.5 Cylindrical Bending Using FSDT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214 . 4.5.1 Governing Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .214 4.5.2 Bending . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215 . 4.5.3 Buckling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .216 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .219 . 4.5.4 Vibration 4.6 Vibration Suppression in Beams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222 . 4.6.1 Introduction . 4.6.2 Theoretical Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222 4.6.3 Analytical Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227 4.6.4 Numerical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230 4.7 Closing Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232 . References for Additional Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
5 Analysis of Specially Orthotropic Laminates Using CLPT . . . . . . . . . .245
. 5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .245 5.2 Bending of Simply Supported Rectangular Plates . . . . . . . . . . . . . . . . . . . . .246 . 5.2.1 Governing Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .246 5.2.2 The Navier Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247 5.3 Bending of Plates with Two Opposite Edges Simply Supported . . . . . . . 255 5.3.1 The Lkvy Solution Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .255 5.3.2 Analytical Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .257 5.3.3 Ritz Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .262 5.4 Bending of Rectangular Plates with Various Boundary Conditions . . . . 265 5.4.1 Virtual Work Statements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265 . 5.4.2 Clamped Plates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .266 5.4.3 Approximation Functions for Other Boundary Conditions . . . . . . .269 5.5 Buckling of Simply Supported Plates Under Compressive Loads . . . . . . .271 . 5.5.1 Governing Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .271 5.5.2 The Navier Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .272 5.5.3 Biaxial Compression of a Square Laminate ( k = 1) . . . . . . . . . . . . . 273 5.5.4 Biaxial Loading of a Square Laminate . . . . . . . . . . . . . . . . . . . . . . . . . .274 5.5.5 Uniaxial Compression of a Rectangular Laminate ( k = 0) . . . . . . . 274
xii
CONTENTS
5.6 Buckling of Rectangular Plates Under InPlane Shear Load . . . . . . . . . . . 278 5.6.1 Governing Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .278 5.6.2 Simply Supported Plates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .278 5.6.3 Clamped Plates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .280 5.7 Vibration of Simply Supported Plates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282 5.7.1 Governing Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .282 5.7.2 Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282 5.8 Buckling and Vibration of Plates with Two Parallel Edges Simply Supported . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285 5.8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .285 5.8.2 Buckling by Direct Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287 5.8.3 Vibration by Direct Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .288 5.8.4 Buckling and Vibration by the StateSpace Approach . . . . . . . . . . .288 5.9 Transient Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .290 290 5.9.1 Preliminary Comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.9.2 Spatial Variation of the Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290 5.9.3 Time Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .292 5.10 Closure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293 . References for Additional Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .296
6 Analytical Solutions of Rectangular Laminated Plates Using CLPT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 297 .
6.1 Governing Equations in Terms of Displacements . . . . . . . . . . . . . . . . . . . . . .297 6.2 Admissible Boundary Conditions for the Navier Solutions . . . . . . . . . . . . . 299 6.3 Navier Solutions of Antisymmetric CrossPly Laminates . . . . . . . . . . . . . . 301 6.3.1 Boundary Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301 6.3.2 Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .304 6.3.3 Bending . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .308 6.3.4 Determination of Stresses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .309 6.3.5 Buckling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .317 6.3.6 Vibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .323 6.4 Navier Solutions of Antisymmetric AnglePly Laminates . . . . . . . . . . . . . . 326 6.4.1 Boundary Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 326 6.4.2 Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 328 6.4.3 Bending . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .329 6.4.4 Determination of Stresses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .330 . 6.4.5 Buckling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 335 . 6.4.6 Vibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 337 6.5 The L&y Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 339 6.5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .339 6.5.2 Solution Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .342 6.5.3 Antisymmetric CrossPly Laminates . . . . . . . . . . . . . . . . . . . . . . . . . . . .348 6.5.4 Antisymmetric AnglePly Laminates . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
6.6 Analysis of Midplane Symmetric Laminates . . . . . . . . . . . . . . . . . . . . . . . . . . .356 6.6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .356 . 6.6.2 Governing Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .356 6.6.3 Weak Forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 357 . 6.6.4 The Ritz Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .358 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 358 6.6.5 Simply Supported Plates . 6.6.6 Other Boundary Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .360 ................................................... 6.7 Transient Analysis 361 . 6.7.1 Preliminary Comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .361 6.7.2 Equations of Motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361 . 6.7.3 Numerical Time Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 362 ............................................. 6.7.4 Numerical Results 364 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .371 . 6.8 Summary . Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 371 . References for Additional Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .375
7 Analytical Solutions of Rectangular Laminated Plates . Using FSDT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 377
7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 377 7.2 Simply Supported Antisymmetric CrossPly Laminated Plates . . . . . . . . 379 7.2.1 Solution for the General Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .379 . 7.2.2 Bending . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .381 . 7.2.3 Buckling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 388 7.2.4 Vibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 394 7.3 Simply Supported Antisymmetric AnglePly Laminated Plates . . . . . . . . 400 7.3.1 Boundary Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 400 . 7.3.2 The Navier Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .402 7.3.3 Bending . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 404 . 7.3.4 Buckling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .405 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .406 . 7.3.5 Vibration 7.4 Antisymmetric CrossPly Laminates with Two Opposite . Edges Simply Supported . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .412 7.4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 412 7.4.2 The L6vy Type Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .413 7.4.3 Numerical Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 415 7.5 Antisymmetric AnglePly Laminates with Two Opposite . Edges Simply Supported . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .421 . 7.5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 421 . 7.5.2 Governing Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 421 7.5.3 The Lkvy Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 423 . 7.5.4 Numerical Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 425 . 7.6 Transient Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .430 7.7 Vibration Control of Laminated Plates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .437 7.7.1 Preliminary Comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 437 . 7.7.2 Theoretical Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 438
xiv
CONTENTS
7.7.3 Velocity Feedback Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 438 . 7.7.4 Analytical Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 439 7.7.5 Numerical Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .441 7.8 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 442 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 444 445 References for Additional Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8 Theory and Analysis of Laminated Shells . . . . . . . . . . . . . . . . . . . . . . . . . . . .449
8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .449 8.2 Governing Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .450 8.2.1 Geometric Properties of the Shell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 450 8.2.2 Kinetics of the Shell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .454 8.2.3 Kinematics of the Shell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 455 . 8.2.4 Equations of Motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 457 8.2.5 Laminate Constitutive Relations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .461 8.3 Theory of DoublyCurved Shells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 462 8.3.1 Equations of Motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .462 8.3.2 Ana.lytical Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .463 8.4 Vibration and Buckling of CrossPly Laminated Circular Cylindrical Shells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 473 . 8.4.1 Equations of Motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .473 475 8.4.2 Analytical Solution Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.4.3 Boundary Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .479 8.4.4 Numerical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 480 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 483 References for Additional Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 483
9 Linear Finite Element Analysis of Composite Plates and Shells . . . .487
9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 487 9.2 Finite Element Models of the Classical Plate Theory (CLPT) . . . . . . . . . 488 9.2.1 Weak Forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .488 . 9.2.2 Spatial Approximations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 490 9.2.3 Semidiscrete Finite Element Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . .499 9.2.4 Fully Discretized Finite Element Models . . . . . . . . . . . . . . . . . . . . . . . .500 9.2.5 Quadrilateral Elements and Numerical Integration . . . . . . . . . . . . . .503 9.2.6 PostComputation of Stresses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 510 9.2.7 Numerical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .510 9.3 Finite Element Models of Shear Deformation Plate Theory (FSDT) . . . 515 9.3.1 Weak Forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .515 9.3.2 Finite Element Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .516 9.3.3 Penalty Function Formulation and Shear Locking . . . . . . . . . . . . . . .520 9.3.4 PostComputation of Stresses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 524 . 9.3.5 Bending Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 525 9.3.6 Vibration Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .540 9.3.7 Transient Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 542
CONTENTS
XV
9.4 Finite Element Analysis of Shells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .543 9.4.1 Weak Forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 543 . 9.4.2 Finite Element Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 546 . 9.4.3 Numerical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 549 . 9.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .558 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 560 References for Additional Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .560 .
10 Nonlinear Analysis of Composite Plates and Shells . . . . . . . . . . . . . . . .567
10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .567 . 10.2 Classical Plate Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 568 10.2.1 Governing Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .568 10.2.2 Virtual Work Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 569 . 10.2.3 Finite Element Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .572 10.3 FirstOrder Shear Deformation Plate Theory . . . . . . . . . . . . . . . . . . . . . . . .575 . 10.3.1 Governing Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 575 10.3.2 Virtual Work Statements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .576 10.3.3 Finite Element Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 578 10.4 Time Approximation and the NewtonRaphson Method . . . . . . . . . . . . . .583 . 10.4.1 Time Approximations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .583 10.4.2 The NewtonRaphson Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .584 10.4.3 Tangent Stiffness Coefficients for CLPT . . . . . . . . . . . . . . . . . . . . . . .586 10.4.4 Tangent Stiffness Coefficients for FSDT . . . . . . . . . . . . . . . . . . . . . . .590 10.4.5 Membrane Locking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .594 10.5 Numerical Examples of Plates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 596 10.5.1 Preliminary Comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 596 10.5.2 Isotropic and Orthotropic Plates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 596 10.5.3 Laminated Composite Plates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .601 10.5.4 Effect of Symmetry Boundary Conditions on Nonlinear Response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 604 . 10.5.5 Nonlinear Response Under InPlane Compressive Loads . . . . . . . 608 10.5.6 Nonlinear Response of Antisymmetric CrossPly Laminated Plate Strips . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 608 . 10.5.7 Transient Analysis of Composite Plates . . . . . . . . . . . . . . . . . . . . . . .612 10.6 Functionally Graded Plates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .613 . 10.6.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 613 10.6.2 Theoretical Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 615 10.6.3 Thermomechanical Coupling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 616 10.6.4 Numerical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 617 10.7 Finite Element Models of Laminated Shell Theory . . . . . . . . . . . . . . . . . . .621 10.7.1 Governing Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 621 . 10.7.2 Finite Element Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .622 . 10.7.3 Numerical Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .625
xvi
CONTENTS
10.8 Continuum Shell Finite Element . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 627 10.8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .627 10.8.2 Incremental Equations of Motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .628 10.8.3 Continuum Finite Element Mode: . . . . . . . . . . . . . . . . . . . . . . . . . . . . .631 10.8.4 Shell Finite Element . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 633 10.8.5 Numerical Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .638 10.8.6 Closure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 644 10.9 Postbuckling Response and Progressive Failure of Composite Panels in Compression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .645 10.9.1 Preliminary Comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 645 10.9.2 Experimental Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 645 10.9.3 Finite Element Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .647 10.9.4 Failure Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 648 ......................................... 10.9.5 Results for Panel C4 650 10.9.6 Results for Panel H4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 655 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .658 10.10 Closure . Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 658 References for Additional Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 664
11 ThirdOrder Theory of Laminated Composite Plates and Shells . . 671
11.1Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .671
11.2 A ThirdOrder Plate Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .671 11.2.1 Displacement Field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .671 . 11.2.2 Strains and Stresses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 674 11.2.3 Equations of Motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .674 11.3 HigherOrder Laminate Stiffness Charac:teristics . . . . . . . . . . . . . . . . . . . . . 677 11.3.1 SingleLayer Plates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .678 11.3.2 Symmetric Laminates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .680 11.3.3 Antisymmetric Laminates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .681 11.4 The Navier Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .682 11.4.1 Preliminary Comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .682 11.4.2 Antisymmetric CrossPly Laminates . . . . . . . . . . . . . . . . . . . . . . . . . . .684 11.4.3 Antisymmetric AnglePly Laminates . . . . . . . . . . . . . . . . . . . . . . . . . .687 11.4.4 Numerical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .689 11.5 Lkvy Solutions of CrossPly Laminates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .699 11.5.1 Preliminary Comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 699 11.5.2 Solution Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .701 11.5.3 Numerical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .704 11.6 Finite Element Model of Plates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .706 11.6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 706 11.6.2 Finite Element Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 707 11.6.3 Numerical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 712 11.6.4 Closure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 714
CONTENTS
xvii
11.7 Equations of Motion of the ThirdOrder Theory of DoublyCurved . Shells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .718
. Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .720 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .721 . References for Additional Reading
12 Layerwise Theory and Variable Kinematic Models . . . . . . . . . . . . . . . . .725
12.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 725 . 12.1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .725 12.1.2 An Overview of Layerwise Theories . . . . . . . . . . . . . . . . . . . . . . . . . . .726 12.2 Development o f t h e Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 730 12.2.1 Displacement Field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 730 . 12.2.2 Strains and Stresses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .733 12.2.3 Equations of Motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .734 12.2.4 Laminate Constitutive Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 736 . 12.3 Finite Element Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .738 12.3.1 Layerwise Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 738 12.3.2 Full Layerwise Model Versus 3D Finite Element Model . . . . . . . 739 12.3.3 Considerations for Modeling Relatively Thin Laminates . . . . . . . 742 12.3.4 Bending of a Simply Supported (0/90/0) Laminate . . . . . . . . . . . .746 12.3.5 Free Edge Stresses in a (45145), Laminate . . . . . . . . . . . . . . . . . . . .753 12.4 Variable Kinematic Formulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 759 . 12.4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 759 762 12.4.2 Multiple Assumed Displacement Fields . . . . . . . . . . . . . . . . . . . . . . . . 12.4.3 Incorporation of Delamination Kinematics . . . . . . . . . . . . . . . . . . . . .764 . 12.4.4 Finite Element Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .766 12.4.5 Illustrative Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 769 780 12.5 Application to Adaptive Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 780 12.5.2 Governing Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .783 . . 12.5.3 Finite Element Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 785 12.5.4 An Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 787 ............................... 794 12.6 Layerwise Theory of Cylindrical Shells . 12.6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .794 . 12.6.2 Unstiffened Shells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .794 12.6.3 Stiffened Shells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 798 806 12.6.4 Postbuckling of Laminated Cylinders . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.7 Closure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 812 . References for Additional Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 816
Subject Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 821
which underwent significant revisions to include laminated shells. Chapter 8. respectively. The subject of nanoscience and technology. the material from the first three chapters is condensed into a single chapter (Chapter 1) in this second edition to make room for the new material. In particular. functionally graded materials (FGMs). It should be an integral part of any study on laminated composite structures. Thanks are also due to Mr. Chapters 9 and 10 (corresponding to Chapters 10 and 13 in the first edition) are devoted to linear and nonlinear finite element analysis. on theory and analysis of laminated shells is added to overcome the glaring omission in the first edition of this book. and nanoscience and technology each topic deserves to be treated in a separate monograph. Very little material has been deleted. The focus for the present edition of this book remains the same the education of the individual who is interested in gaining a good understanding of the mechanics theories and associated finite element models of laminated composite structures. These chapters are extensively revised to include more details on the derivation of tangent stiffness matrices and finite element models of shells with numerical examples. Chapters 11 and 12 in the present edition correspond t o Chapters 11 and 12 of the first edition. Chapters 2 through 7 correspond to Chapters 4 through 9. the first edition of this book did not contain any material on the theory and analysis of laminated shells. of laminated plates and shells. Foremost among these developments have been the smart materials and structures. While the author's expertise and contributions in these areas are limited. Thus Chapter 1 contains certain mathematical preliminaries. New material has been added in most chapters along with some rearrangement of topics to improve the clarity of the overall presentation. It is a pleasure to acknowledge the help of my colleagues. and they have been revised to include smart structures and functionally graded materials. from the first edition. it is felt that the reader should be made aware of the developments in the analysis of smart and FGM structures. a study of the equations of anisotropic elasticity. A completely new chapter. The problem sets essentially remained the same with the addition of a few problems here and there. The acknowledgments and sincere thanks and feelings expressed in the preface to the first edition still hold but they are not repeated here. and an introduction to the principle of virtual displacements and classical variational methods (the Ritz and Galerkin methods). for their help with the proofreading of the manuscript. ZhenQiang Cheng. of course. is outside the scope of the present study.Preface to the Second Edition In the seven years since the first edition of this book appeared some significant developments have taken place in the area of materials modeling in general and in composite materials and structures in particular. R o m h  . especially Dr. respectively. Also.
Reddy College Station. J.XX PREFACE T O T H E SECOND EDITION Arciniega for providing the numerical results of some examples on shells included in Chapter 9. N. Texas .
Some mathematical preliminaries. The motivation for the present book has come from many years of the author's research and teaching in laminated composite structures and from the fact there does not exist a book that contains a detailed coverage of various laminate theories. buckling. buckling loads. where a complete derivation of the equations of motion of the classical and firstorder shear deformation laminated plate theories is presented. Here analytical solutions are developed for bending. lamination scheme. and plate aspect ratio on bending deflections and stresses. Readers who have had a basic course in composites may skip Chapter 4 also. electronic circuit boards. material scientists. and analytical and finite element solutions he and his collaborators have developed over the last two decades. Here. Chapter 7 deals with the analysis of specially orthotropic rectangular laminates using the classical laminated plate theory (CLPT). Chapter 6 includes applications of the classical and firstorder shear deformation theories to laminated beams and plate strips in cylindrical bending. the parametric effects of material anisotropy. equations of anisotropic elasticity. analytical solutions. followed by a discussion of the constitutive equations of a lamina and transformation of stresses and strains.g. The number of students taking courses in composite materials and structures has steadily increased in recent years. The courses offered at universities and the books published on composite materials are of three types: material science. and virtual work principles and variational methods are reviewed in Chapters 1 through 3. The present book belongs to the mechanics category. automotive. The book is largely based on the author's original work on refined theories of laminated composite plates and shells. mechanical engineers. and design. and underwater structures. natural vibration. and to analyze and design structural components made from composite materials. The subject of composite materials is truly an interdisciplinary area where chemists. and finite element models. and laminate stiffness characteristics of selected laminates are discussed. where certain terminology common to composite materials is introduced. . aerospace.. The major journey of the book begins with Chapter 5. and sports equipment) and the number of journals and research papers published in the last two decades attest to the fact that there has been a major effort to develop composite material systems.PREFACE T O THE FIRST EDITION xxi Preface the First Edit ion The dramatic increase in the use of composite materials in all types of engineering structures (e. A reader who has had a course in elasticity or energy and variational principles of mechanics may skip these chapters and go directly to Chapter 4. and transient response of simple beam and plate structures. and the students are drawn to these courses from a variety of disciplines. chemical engineers. vibration frequencies. as well as in medical prosthetic devices. and structural engineers contribute to the overall product. and transient response are discussed. mechanics.
The first course may cover Chapters 1 through 8 or 9. Nicholas J . The author wishes to thank all his former doctoral students for their research collaboration on the subject. Govind Rengarajan for their help with the proofreading of the manuscript. and technical as it does. Chapters 11 and 1 2 are devoted to higherorder (thirdorder) laminate theories and layerwise theories. and finite element analysis of equivalent singlelayer and layerwise theories of composite laminates. The book is suitable as a reference for engineers and scientists working in industry and academia. Krishna Murty of the Indian Institute of Science. V. Onedimensional (for beams and plate strips) as well as twodimensional (plates) finite element models based on CLPT and FSDT are discussed and numerical examples are presented. National Science Foundation (NSF). Without this support. and the Oscar S. and it can be used as a textbook in a graduate course on theory and/or analysis of composite laminates. and numerical results are presented for some typical problems. Indeed. Army Research Office (ARO). analytical solutions. Chapters 7 through 13 contain results of the research conducted by Drs. Filis Kokkinos for his dedication and innovative and creative production of the final artwork in this book. It is also the author's pleasure to acknowledge the help of Mr. The research of the author in composite materials was influenced by many researchers. The author is indebted to Dr. and Mr. it would not have been possible to contribute to the subject of this book.xxii PREFACE TO T H E FIRST EDITION Analytical solutions for bending. In particular. buckling. An introductory course on mechanics of composite materials may prove to be helpful but not necessary because a review of the basics is included in the first four chapters of this book. the National Aeronautics and Space Administration (NASA Lewis and NASA Langley). Jones of the Virginia Polytechnic Institute and State University. the U S . Chapter 10 deals with finite element analysis of composite laminates. natural vibration. without his imagination and hundreds of hours of effort the artwork would not have looked as beautiful. the U. The RayleighRitz solutions are also discussed for laminates that do not admit the Navier solutions. P. Analytical as well as finite element models are discussed. Professor Robert M. The material included in these chapters is up to date at the time of this writing. Ahmed Khdeir. Jr. Praveen Grama. on the development of theories. Bert of the University of Oklahoma. The author gratefully acknowledges the support of his research in composite materials in the last two decades by the Office of Naval Research (ONR). and a second course may cover Chapters 8 through 13. respectively. respectively. Displacement finite element models of laminated plates with the von KBrmAn nonlinearity are derived. Finally. professional. The author wishes to thank Professor Charles W. and transient response of rectangular laminates based on the Navier and Lkvy solution approaches are presented in Chapters 8 and 9 for the classical and firstorder shear deformation plate theories (FSDT). and Dr. a colleague .S. Mr. Chapter 13 is concerned about the geometrically nonlinear analysis of composite laminates. Professor A. Stephen Engelstad. It can also be used for a course on stress analysis of laminated composite plates. and Donald Robbins. The author is also grateful to Professor G. Pagano of WrightPatterson Air Force Base. Wyatt Chair in the Department of Mechanical Engineering at Texas A&M University. the Air Force Office of Scientific Research (AFOSR). Asghar Nosier. Dakshina Moorthy. Peterson.
the suddenness makes it more difficult to accept. Most of these hours came from evenings and holidays that could have been devoted to family matters. While death is imminent. J. it should be recorded that the author is grateful to his wife Aruna for her care. brother. Texas A l that i not given i lost l s s . the author has lost his father. The writing of this book took thousands of hours over the last ten years.and friend. for his encouragement and support of the author's professional activities at Texas A&M University. Reddy College Station. This book is dedicated to the memory of these individuals. brother inlaw. devotion. While no words of gratitude can replace the time lost with family. During the long period of writing this book. and love.all suddenly. and a friend (Hans Eggers) . father inlaw. and to his daughter Anita and son Anil for their understanding and support. N.
.
fibers have near crystalsized diameter and a very high lengthtodiameter ratio. electrical resistance. acts as a loadtransfer medium between fibers. agents that are used to enhance bonding between fibers and matrix. Reinforced steel bars embedded in concrete provide an example of fiberreinforced composites. and so on. no attempt is made here to present basic material science aspects. Virtual Work Principles. and Variational Met hods 1.g. we consider a basic unit of material as one that has properties. and protects fibers from being exposed to the environment (e. Fibers provide an example of such materials. called whiskers. singlecrystal level. However.. such as the molecular structure or interatomic forces those hold the matter together. Materials are "processed" such that the basic units are aligned so that the desired property is maximized in a given direction. etc. For the purpose of gaining some insight into the material behavior. etc. it is necessary to know certain aspects of material science. for example. contain high strength and high modulus fibers in a matrix material. whereas matrix materials have their usual bulkform properties. and the matrix material keeps the fibers together. humidity. matrix materials.). When a property is maximized in one direction. paradoxically exhibit better structural properties than long fibers.1 Equations of Anisotropic Elasticity. a group of crystals.1 FiberReinforced Composite Materials Composite materials consist of two or more materials which together produce desirable properties that cannot be achieved with any of the constituents alone. it may be achieved at the expense of the same property in other directions and other properties in the same direction. Geometrically. strength. nanolevel. Short fibers. and other properties of fiberreinforced materials. whose magnitudes depend on the direction. an abstract understanding of the material behavior is useful. Fiberreinforced composite materials. fibers are the principal loadcarrying members. which are "stronger" in one direction than in other directions. It is known that fibers are stiffer and stronger than the same material in bulk form. The directional dependence of properties is a result of the interatomic bonds. such as the modulus. To gain a full understanding of the behavior of fibers. Materials are studied a t various levels: atomic level. thermal coefficient of expansion.. When materials are processed such that the basic . moisture. Since the present study is entirely devoted t o mechanics aspects and analysis methods of fiberreinforced composite materials. In these composites.
the analysis requires a knowledge of anisotropic elasticity. steel. such as bars.e. kinematics of deformation) of laminates. The fibers and matrix materials used in composites are either metallic or nonmetallic. Anisotropic Elasticity Structural Theories Analysis of Laminated Composite Structures Methods Damage 1Failure Theories Figure 1. boron. the resulting material tends to have the same value of the property. in all directions.1. and organic materials like glass. structural theories (i. Structural elements. in an average statistical sense. beams or plates are then formed by stacking the layers to achieve desired strength and stiffness. Material scientists are continuously striving to develop better materials for specific applications. copper. Analysis of structural elements made of laminated composite materials involves several steps. and titanium. called lamina. The fiber materials in use are common metals like aluminum.. A matrix material.1. and represent the material response of laminated structural elements. and graphite materials. As shown in Figure 1. Fiberreinforced composite materials for structural applications are often made in the form of a thin layer.2 MECHANICS OF LAMINATED COMPOSITE PLATES AND SHELLS units are randomly oriented. Fiber orientation in each lamina and stacking sequence of the layers can be chosen to achieve desired strength and stiffness for a specific application.1: Basic blocks in the analysis of composite materials. analytical or computational methods t o determine solutions of the governing equations. iron. which is made in bulk form. Such materials are called isotropic materials. provides an example of isotropic materials. nickel. It is the purpose of the present study to develop equations that describe appropriate kinematics of deformation. and failure theories to predict modes of failures and to determine failure loads. .1. govern force equilibrium. A detailed study of the theoretical formulations and solutions of governing equations of laminated composite plate structures constitutes the objective of the present book. A lamina is a macro unit of material whose material properties are determined through appropriate laboratory tests.
Nonscalar quantities require not only a magnitude specified. 1. When vector notation is used. Some nonscalar quantities require the specification of magnitude and two directions. such as direction. A brief discussion of vectors and tensors is presented next. Sometimes a vector is referred to as a tensor of order one.2 Mat hematical Preliminaries 1. and it is "an element from a linear vector space. and various physical quantities involved in the description are expressed in terms of measurements made in that system. and mass density provide examples of scalars. temperature. however. as they are needed in the sequel. Vector in modern mathematical analysis is an abstraction of the elementary notion of a physical vector. We also encounter thirdorder and fourthorder tensors in the discussion of constitutive equations. according to the information needed to specify them completely: scalars and nonscalars. and we may seek to represent the law in a nianner independent of a particular coordinate system. and acceleration are examples of nonscalars . and virtual work principles and variational methods is presented. equations governing a deformable anisotropic medium. . The description thus depends upon the chosen coordinate system and may appear different in another type of coordinate system. Time.EQUATIONS OF ANISOTROPIC ELASTICITY 3 Following this general introduction. In this book. For example. integral relations. temperature gradient. a particular coordinate system need not be introduced. The laws of nature. in nearly all cases of practical interest the vector is endowed with a magnitude. 1. use of vector notation in formulating natural laws leaves then1 invariant to coordinate transformations." While the definition of a vector in abstract analysis does not require the vector to have a magnitude. and a tensor of order 2 is also called a dyad. vectors and dyads) will be of primary interest in the present study (see [l81 for additional details). but also additional information. The term vector is used to imply a nonscalar that has magnitude and "direction" and obeys the parallelogram law of vector addition and rules of scalar multiplication. force. but also an area upon which the force acts. Consequently.2. the specification of stress requires not only a force. Readers familiar with these topics can skip the remaining portion of this chapter and go directly to Chapter 2.2 Vectors and Tensors In the analytical description of physical phenomena. a review of vectors and tensors. moment. The scalars are given by a single number. a coordinate system in the chosen frame of reference is introduced.1 General Comments The quantities used to express physical laws can be classified into two classes. A stress is a secondorder tensor. we need only vectors with magnitude.. A way of doing this is provided by vector and tensor notation.e. Displacement. volume. Firstand secondorder tensors (i. should be independent of the choice of a coordinate system.2.
e. x3) = (x.e2. 3 ) is the orthonormal basis. i.en. We shall always use a righthand coordinate system.e 2 . ( X I . . The Cartesian coordinates are denoted by The familiar rectangular Cartesian coordinate system is shown in Figure 1.. When the basis vectors are of unit lengths and mutually orthogonal. Then the vector and tensor quantities are expressed in terms of their components in that coordinate system.1. as) and basis vectors (el. the components have the same physical dimensions as the vector). The general Cartesian system is oblique. es) (ei are not necessarily unit vectors) as When the basis vectors of a coordinate system are constants. they are called orthonormal. a vector A in a threedimensional space may be expressed in terms of its components (a1.2. When the Cartesian system is orthogonal. with fixed lengths and directions.ey. and Ai and Bi are the corresponding physical components (i.e3) = (e2. it is called rectangular Cartesian. 2 .1: A rectangular Cartesian coordinate system. F i g u r e 1. In many situations an orthonormal basis simplifies calculations. &) (1. For example.e. a2. ey. z ) . the coordinate system is called a Cartesian coordinate system.4 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS Vectors Often a specific coordinate system is chosen to express governing equations of a problem to facilitate their solution. (el. e 3 ) or ( e x . y. We denote an orthonormal Cartesian basis by (el. z2. Gz) are the unit basis vectors.2.2.3) For an orthonormal basis the vectors A and B can be written as where Gi (i = 1 ..
the component form of vector A where (el. in the present context the range of j. k are not in cyclic order and not repeated (i # j # k) 0.e 2 . Differentiation of vector functions with respect to the coordinates is a common occurrence in mechanics. the Kronecker delta and the permutation symbol are related by the identity. In a rectangular Cartesian system it has the form . es) are basis vectors (not necessarily unit). j . known as the €4identity.2. In an orthonormal basis the scalar product (also called the "dot product") and vector product (also called the itcross product") can be expressed in the index form using the Kronecker delta symbol Sij and the alternating symbol (or permutation symbol) ~ i j k : where 1. For example. For example. Most of the operations involve the "del operator.7) Further. can be expressed in the form 3 The repeated index is a dummy index in tne sense that any other symbol that is not already used in that expression can be employed: The range of summation is always known in the context of the discussion. j ." denoted by V. j. k are in cyclic order and not repeated (i # j # k ) 1. if i .EQUATIONS OF ANISOTROPIC ELASTICITY 5 S u m m a t i o n Convention It is convenient to abbreviate a summation of terms by understanding that a repeated index means summation over all values of that index. k are repeated (1. if i . k and m is 1 to 3 because we are discussing vectors in a threedimensional space. if any of i .
in the summation convention. because it is an operator.2): z and all other derivatives of the base vectors are zero.or. y. For instance V . we have It is important to note that the del operator has some of the properties of a vector but it does not have them all. V is a scalar differential operator. z ) and cylindrical coordinates (r. whereas A . see Reddy and Rasmussen [5] and Reddy [6].2. We have the following relations between the rectangular Cartesian coordinates (x. A is a scalar. For more on vector calculus. ) (see Figure 1. . Thus the del operator does not commute in this sense.2. among other references. called the divergence of A . The operation Vq5(x) is called the gradient of a scalar function 4 whereas V x A(x) is called the curl of a vector function A. x Figure 1. 8.2: Cylindrical coordinate system.
2. Because of Newton's third law for action and reaction. The surface force acting on a small element of area in a continuous medium depends not only on the magnitude of the area but also upon the orientation of the area. If tl. we have by Newton's second law for the mass inside the tetrahedron.2.t ( n ) .3: (a) Force on an area element. The component of t that is in the direction of n is called the normal stress. To fix the direction of the normal. and AS. AS3. Then the area A can be denoted by A = Aii.EQUATIONS O F ANISOTROPIC ELASTICITY 7 Tensors To introduce the concept of a secondorder tensor. the stress vector can be defined as follows: t ( n ) = lim ns+o  A F(n) AS We see that the stress vector is a point function of the unit normal n which denotes the orientation of the surface AS. not in the direction of n.3a). t2. It is customary to denote the direction of a plane area by means of a unit vector drawn normal to that plane. where AV is the volume of the tetrahedron. (b) Tetrahedral element in Cartesian coordinates. The direction of the normal is taken by convention as that in which a righthanded screw advances as it is rotated according to the sense of travel along the boundary curve or contour. It is useful to establish a relationship between t and n. also called a dyad. Since the total vector area of a closed surface is zero F i g u r e 1. .n ) = . The component of t that is normal t o n is called a shear stress. we assign a sense of travel along the contour of the boundary of the plane area in question. To do this we now set up an infinitesimal tetrahedron in Cartesian coordinates as shown in Figure 1.2. t3. in general. AS2. Note that t ( n ) is.3b. we consider the equilibrium of an element of a continuum acted upon by forces. and a the acceleration. Let the unit normal vector be given by ii. respectively. If we denote by A F ( n ) the force on a small area n A S located at the position r (see Figure 1. and t denote the stress vectors in the outward directions on the faces of the infinitesimal tetrahedron whose areas are AS1. p the density. f the body force per unit mass. we see that t ( .
2.&)AS. we have t(n) =n .20) in (1. Hence.25) is known as the Cauchy stress formula.ASlel  AS2e2 .18) The volume of the element AV can be expressed as where Ah is the perpendicular distance from the origin to the slant face. ASn it follows that AS.2. AS2 = ( n . AS3 = ( n . and ts into their orthogonal components. Ah + 0.2.2.2. 2 . and is termed the Cauchy stress tensor.17) and dividing throughout by A S reduces it t o In the limit when the tetrahedron shrinks to a point.3). we are left with It is now convenient to display the above equation as The terms in the parenthesis are to be treated as a dyadic.24) a r eltl + eztz G3t3 + t* Thus. called stress dyadic or stress tensor (we will not use the "double arrow" notation for tensors after this discussion) : (1. The stress vector t represents the vectorial stress on an area perpendicular to the direction ii.AS3e3 = 0 (1. = (1.2.4).2. tz. the stress dyadic can be expressed in summation notation as The component aij represents the stress (force per unit area) on an area perpendicular to the ith coordinate and in the j t h coordinate direction (see Figure 1.19) . It is useful to resolve the stress vectors t l . Equation (1.2..(see Problem l. e2)AS. Substitution of Eqs. We have for i = 1 . (1.19) and (1. & ) A S (1.a t* and the dependence of t on n has been explicitly displayed. 3 . .
4: Notation used for the stress components in Cartesian rectangular coordinates.31) .2. by Thus the dot operation with a vector produces another vector. When the ordering is understood. One of the properties of a dyadic is defined by the dot product with a vector. respectively.2. or what we shall call a secondorder tensor. The components are thus said to be ordered. the explicit writing of the dyads can be suppressed and the dyad is written as an array: (1. The two operations in general produce different vectors. dot products of a secondorder tensor @ with a vector A from the right and left are given. has nine independent components in general.EQUATIONS OF ANISOTROPIC ELASTICITY 9 Figure 1. each component associated with a certain dyad pair. For example. The transpose of a secondorder tensor is defined as the result obtained by the interchange of the two basis vectors: It is clear that we have We can display all of the components CPij of a dyad to the right and the i index run downward: by letting the j index run This form is called the nonion form.30) illustrates that a dyad in threedimensional space.2. Equation (1.
Let R denote a region in space surrounded by the surfa. vectors are called firstorder tensors and dyads are called secondorder tensors. + Here we have selected a rectangular Cartesian basis to represent the tensor.e2. . its components referred to another rectangular Cartesian basis (6.. The two sets of components are related according to where tij are called the direction cosines.e3). The doubledot product between tensors of second order and higher order is encountered in mechanics. An nthorder tensor can be expressed in a short form using the summation convention: (1.2. Let ) be a scalar function and A be a vector function defined over the region R.ce I'. We record them here for future reference and use. (1.2. Scalars are called zerothorder tensors.: 6.. Tensors are sometimes defined by the transformation law for its components. triadics.5): .) are Aij. . The doubledot product between two secondorder tensors @ and 9 is defined as Integral Relations Relations between volume integrals and surface integrals of the gradient (V) of a scalar or a vector and divergence (V.2. The generalization to thirdorder tensors thus leads. For example. 6. but it is taken to mean the same.33) = dijke. A unit second order tensor I is defined by In the general scheme that is developed. or is derived from. a vector A has components Ai with respect to the rectangular Cartesian basis (el. then it is not a tensor. or three vectors standing side by side. Similarly.. Let t dv be a differential volume element. It follows that higher order tensors are developed from polyads. the components of a secondorder tensor @ transform according to the rule =! a m or [a']= [L] [ L ] ~ [@I If the components do not satisfy the above transformation law.30). and let ds be a differential element of the surface whose unit outward normal is denoted by n.10 MECHANICS OF LAMINATED COMPOSITE PLATES AND SHELLS This representation is simpler than Eq.) of a vector are needed in the later chapters. Then the following integral identities hold (see Figure 1. eiejekee .
37) and (1..38).A ds niAi d s (vector form) (component form) (1. divergence or curl operation.2.2.2.38a). A du = d .38) can be expressed concisely in the single statement 6 where * denotes an appropriate operation.2.$ d s Divergence Tlleorem (component form) /' v .37) and (1. and the comporient forms refer to the usual rectangular Cartesian 2 coordinate system. gradient.e.2.2. denotes the integral on the closed boundary r of the domain ! .2. (1.2. (1.2.38a) In the above integral relations. i. Equations (1. (1. Gradient Theorem n.38) are valid in two as well as three dimensions.37) and (1.5: A solid body with a surface normal vector n. Some additional integral relations can he derived from Eqs. and obtain . where p is a scalar function. and F is a scalar or vector function. Let A = V p in Eq.EQUATIONS OF ANISOTROPIC ELASTICITY 11 Figure 1. The integral relations in Eqs.
1 Introduction The objective of this section is to review the governing equations of a linear anisotropic elastic body. V p is called the normal derivative of p on the surface I?. without the consideration of forces causing the deformation. The straindisplacement relations. The thermodynamic principles are concerned with the conservation of energy and relations among heat. equations of motion. In the following sections.3. 1.12 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS or. The equations governing the motion of a solid body can be classified into four basic categories: (1) Kinematics (straindisplacement equations) (2) Kinetics (conservation of momenta) (3) Thermodynamics (first and second laws of thermodynamics) (4) Constitutive equations (stressstrain relations) Kinematics is a study of the geometric changes or deformation in a body. A solid body B is a set of material particles which can be identified as having onetoone correspondence with the points of a region Q of Euclidean point space R3. and they relate the dependent variables introduced in the kinetic description to those in the kinematic and thermodynamic descriptions. Kinetics is the study of the static or dynamic equilibrium of forces and moments acting on a body.. no change in the temperature of the body) are presented first. V p (invariant form) an = ni axi 8 9 (rectangular Cartesian component form) The integral relations presented in this section are useful in developing the socalled weak forms of differential equations in connection with the Ritz method and finite element formulations of boundary value problems. mechanical work. Subsequently. and the constitutive equations for an isothermal state (i.3 Equations of Anisotropic Elasticity 1. These equations are supplemented by appropriate boundary and initial conditions of the problem. in component form The quantity n .e. . and thermodynamic properties of the body. and is denoted by 3 = ii . an overview of the governing equations of an anisotropic elastic body is presented. This leads t o equations of motion as well as the symmetry of stress tensor in the absence of body moments. The constitutive equations describe thermomechanical behavior of the material of the body. the thermodynamic principles are considered only to determine the temperature distribution in a solid body and to account for the effect of nonuniform temperature distribution on the strains.
Whether the deformation is time dependent or not.The particles of B are identified by their timedependent positions relative to the selected frame of reference. i. and its position x at any time is referred to the material coordinates X. The analytical description of configurations a t various times of a material body acted on by various loads results in a set of governing equations. Xy) are called the material coordinates. In the spatial or Eulerian description of a body B.3.2 StrainDisplacement Equations The phrase deformation of a body refers to relative displacements and changes in the geometry experienced by the body. This curve is called the path of X and is given parametrically by Eq. Thus. during a motion of a body B.X2. Consider a deformable body 23 of known geometry.1).e. whereas in the Lagrangian description the coordinate system X is fixed on a given body of matter in its undefornied configuration. the motion of a body B is referred to a reference configuration C R . the particle X moves to a new position x = ( x l .e.3. xy) are expressed in terms of the reference coordinates ( X I .. x3). The spatial description focuses attention on a given region of space instead of on a given body of matter. X3).. When the body is deformed under the action of external forces. the geometry of the body will change continuously with time. If the loads are applied slowly so that the deformation is only dependent on the loads. x2. In the referential or Lagrangian description. At time t > 0. in the Lagrangian description the current coordinates ( x l . 1. Thus.. the deformation of the body will be a function of time. the reference configuration C R is chosen to be the unstressed state of the body. The coordinates ( X I . the body assumes a new configuration C and the particle X occupies the new position x. the motion is referred to the current configuration C occupied by the body B. i. Under given geometric restrictions and loading. There are two commonly used descriptions of motion and deformation in continuum mechanics.e. every particle X in the body corresponds to a set of coordinates X = ( X I . Xy ) and time t as Often. C R = CO. If the applied loads are time dependent. in which a particle X of the body B occupies a position X. (1. Xy). X2. the body will undergo motion and/or deformation (i. the body will take a definitive shape at the end of each load application. a representative particle X occupies a succession of points which together form a curve in Euclidean space. Referred to a rectangular Cartesian frame of reference ( X I . the forces acting on the body will be in equilibrium at all times. Note that X is the name of the particle that occupies the location X in the reference configuration. Suppose that the body B under consideration at time t = 0 occupies a configuration CO. 2 2 . X2. The simultaneous position of all material points of 23 a t a fixed time is called a configuration of the structure. The displacement of the particle X is given by . and is the description most used in fluid mechanics.. and loading. constitution. geometric changes within the body).X2.
Here we use the standard strain measure of solid mechanics. which lies along a curve in the deformed body. The vector dx can be interpreted as the position occupied by the deformed material vector dX. Consider two neighboring material particles P and Q which occupy the positions P : (Xl. and dx is the vector connecting P t o Q. The particles are separated by the infinitesimal distance dS = (sum on i ) in CO.3. z3) and (xl dxl.If the displacement of every particle in the body is known. X2 dX2. In the Lagrangian description. z2 dx2. X2. z3 dx3). in the undeformed configuration C0 of the body B. The deformation (i.. The two particles are now separated by the distance ds = in the deformed configuration C.1). respectively. relative motion of material particles) of a deformable body can be determined only by considering the change of distance between any two arbitrary but infinitesimally close points of the body.1: Kinematics of deformation of a continuous medium. and d X is the vector connecting the position of P to the position of Q. The deformation (or strains) in a body can be measured in a number of ways. the line vector dx in general does not coincide exactly with the deformed position of dX. and we have A rigidbody motion is one in which all material particles of the body undergo the same linear and angular displacements. Suppose that the positions of P and Q are ( z l . A deformable body is one in which the material particles can move relative to each other. which is defined such that it gives the change + + + + + + x3. nanlely the GreenLagrange strain E.3. . X3 dX3). x 3 Particle X A3 Co(time t = 0) Xl 4 Figure 1. X3) and Q : (XI dX1. respectively. When the material vector d X is small but finite. These two particles move to new places P and Q. in the deformed body (see Figure 1.e. we can construct the current (deformed) configuration C from the reference (or undeformed) configuration CO. the displacements are expressed in terms of the material coordinates Xi. respectively. 2 2 .
.3. its total differential is given by [using the chain rule of differentiation and Eq. (I VU)]. [dX . dX dX .dx&. The rectangular Cartesian component form is given by . E = E~ (Eij= Ep). (I + O u ) .3. (1.3. X.X~. In order to express the strains in terms of the displacements.7) Thus the Green (or GreenLagrange) strain tensor E is given in terms of the displacement gradients as Note that the GreenLagrange strain tensor is symmetric.ILK = [dX .dX=dx. the summation conwntion on repeated indices is used.4b) and in the equations that follow. (1.in the square of the length of the material vector dX and in rectangular Cartesian component form we have In Eq. (1.3.3. (1. Now the strain tensor or its components from Eqs.8) are called finite strain components because no assumption concerning the smallness (compared to unity) of the strains is made. dX = d X . (I + VU)*.X~.b) can be expressed in terms of the displacement vector or its components with the help of Eq. (I + + VU)] dX . (I + V U ) ~ I] d X  (1.2) and write (1. (1. strain components defined in Eq.6): 2dX.3. and the range of summation is 1 to 3.5)] where V denotes the gradient operator with respect to the material coordinates. we use Eq.4a. (1. The . dX = dX .E.3.3.5) x =X u(XI. [(I+V u ) .~) + Since x is a function of X.
3. The only nonzero strain component is ( e = 0. Then the GreenLagrange strain tensor reduces to the infinitesimal strain tensor. ++au. ax. au3 ++atLl au2 au2 aul ax.1: (a) A square block is deformed as shown by dotted lines in Figure 1. ax.11) is given by (yij denote the engineering shear strains) Example 1.2a. a x . lVul << 1. If the displacement gradients are so small. (1. aua au2 The explicit form of the infinitesimal strain components (1.3. x3) in the deformed body. Assuming that the body is very thin and the strains (due to the Poisson effect) associated with the thickness direction are negligible. X3) in the undeformed body takes the position (xl. ax.3.3. A material particle which occupied position ( X I . 2 2 .2cm and a = 10cm) . E = E : au.10). ax. we wish to determine the twodimensional strains. X2. that their squares and products are negligible compared to IVul. The current coordinates of the material particle can be expressed in terms of its original position as The displacements are Then the GreenLagrangian strains can be computed using Eq. ax.Explicit form of the six Cartesian components of strain are given by a x .
3.2 2 = X 2 $ z 3 = X 3 (1. we use only the linear strains and the von KBrmAn nonlinear strains derived from Eq.2: Undeformed and deformed configurations of a solid square block.10). This completes the kinematic description.X2. deforrried as shown by dotted lines in Figure 1.X3) in the undeformed body can be expressed as z l = X l + e ~ l . (1.3.02 percent.2b. . In the coming chapters.16) a The displacerrierits are The only nonzero Lagrangian strain is The strain is nonlinear.3. (b) Consider a square block. (b) Pure extensional deformation. (a) Pure shear deformation.Figure 1. The nonlinear part of the strain is 0.3. The current coordinates of the material particle occupying position (XI.
the components of the strain tensor can be computed from a differentiable displacement field using Eq.2.3. a. 1.3.3 Strain Compatibility Equations By definition.1. The stress tensor P is called the first PiolaKirchhofl stress tensor. Thus the Ca. (1. For the twodimensional case. (1. Expressing df in terms of a stress times the initial undeformed area d A requires a new stress tensor P.3. The surface area in turn depends on the orientation of the plane used to slice the body.25).11).3. is used.19) reduces to the following single compatibility equation It should be noted that the strain compatibility equations are satisfied automatically when the strains are computed from a displacement field. and it gives the current force per unit undeformed area. t = a . stress vector t at a point in a deformed body is measured as the force per unit area in the deformed body. if the six components of strain tensor are given and if we are required to find the three displacement components.3. one needs to verify the compatibility conditions only when the strains are computed from stresses that are in equilibrium.22) where N is the unit normal to the undeformed area d A . In the above discussion. where da = da n (1. The existence of a unique solution is guaranteed if the infinitesimal strain components satisfy the following six compatibility conditions: for any i .2. df =d A . j .3. .4 Stress Measures Stress at a point was introduced in Section 1. (1. It was shown that the state of stress at a point inside a body can be expressed in terms of stress vectors on three mutually perpendicular planes. the strains given should be such that a unique solution to the six differential equations relating the strains and displacements exists.2. The area element As in the deformed body corresponds to an area element A S in the reference configuration. The first PiolaKirchhoff stress tensor is not symmetric. P. in much the same way x is the position of a material particle X in the deformed body whose position in the reference configuration was X. Eq. say planes perpendicular to the rectangular coordinate axes by Cauchy's formula in Eq.21) where Cauchy's formula. Thus.uchy stress tensor a is defined to be the current force per unit deformed area: df = t da = d a . n .16) indicates that the stress vector at a point depends on the force vector (its direction and magnitude) and the surface area. m. Equation (1.2 as a measure of force per unit area. n = 1. (1.3. where d A =d A N (1.3. However.8) or Eq.
a2u (vector form) at2 aaji + fi= pat2 (Cartesian component form) a2ui ax:.. between o and S and between E and E.e. the second PiolaKirchhoff stress tensor gives the transformed current force per unit undeformed area. Analogous to the transformation between X and x .3. (1.3. FPT.3. and we use the first symbol of each pair.27b) = 0 (Cartesian cornporient form) For kinenlatically infinitesimal deformations.The second PiolaKirchhoff stress tensor S is introduced as follows. d x = dx .3. The principle leads to the following equations of motion:  + f = p.. (1. (1. we introduce the deformation gradient tensor F and Vo is the gradient operator with respect to X. In much of this book we deal with kinematically infinitesimal deformations (i.11).3.o where p is the density in the deformed configuration and f is the body force vector (measured per unit volurne).3.3. D .27a) (1.5 Equations of Motion The principle of conservation of linear moir~entum states that the rate of change of the total linear momentum of a given continuous medium equals the vector sum of all the external forces acting on the body B.. i. we can transform the force df on the deformed elemental area d a to the force dF on the undeformed elemental area d A (not to be confused between the force d F and deformation gradient tensor F) Thus. linearized elasticity). where F . The second PiolaKirchhoff stress tensor is symmetric whenever the Cauchy stress tensor is symmetric. lVul << 1.8). we do not distinguish between x and X.e.24) and V is the gradient operator with respect to x. We also have d X = F' . which initially occupied a configuration CO.= ~  dX dx = VX (1.26a) and . The straindisplacement relations and the equations of motion in any coordinate system can be obtained from the vector forms in Eqs. provided Newton's third law of action and reaction governs the internal forces. 1. The equations of equilibrium are obtained by setting the time derivative term to zero: V . =0 (vector form) (1. First.a +f dojTji + fi alc.
having two base vectors) and it is symmetric: 0 = e .and f3 = 0.3). 0 .3. f.2: Consider the following stress field in a body that is in equilibrium: and all other components of stress are zero.e. The principle of conservation of angular momentum.3.20 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS (1. and V in the chosen coordinate system.. 2 2.2 and constitutive relations or stressstrain relations to be discussed in the next section.28a) U This notation is meaningful and descriptive of the nature of the tensor. we may write it with a "double arrow" notation as 0 (1.. We have Thus. We assume that the body experienced only a small deformation. the first two equations of equilibrium are identically satisfied for any choice of constants.3.3. .3.27a) by expressing 0. These include the straindisplacement relations discussed in Section 1. independent of the choice of the coordinate system.3.2813) t* A Note that the equations of motion or equilibrium contain three equations relating six stress components and therefore cannot be solved for all six components uniquely. i. The third equation of equilibrium is trivially satisfied. and cq. Example 1. fl = 0. The bending moment about the xaaxis at any distance X I is given by = P ( L .1 3 (1. u.e. Then the stress component 011 can be calculated using the flexure stress formula from elementary strength of materials: . in the absence of any distributed body couples.3. Example 1.3: Consider the cantilevered beam under an end load (see Figure 1. We wish to determine if the stress field satisfies the equations of equilibrium in the presence of body forces. Since the Cauchy stress tensor is a secondorder tensor and symmetric. c2. Additional equations are required. the notation indicates that the quantity is a dyad (i. cg. The vector forms of equations are invariant. leads to the symmetry of the stress tensor: Thus there are only six independent components of the Cauchy stress tensor. cl. f2 = q.XI).e .
. Assuming a twodirnerisional state of stress (with respect t o the X I and 23 coordinates) in the beam.3: A cantilevered beam (i..3. arid components a l g and as3 in the absence of body forces.c. Vanishing of a33 a t 23 = fh / 2 gives which imply t h a t (If = 0. The second equation of equilibrium is trivially satisfied. we wish t o determine the stress a12. or f = c2 and g = 0 dx1 Vanishing of u13 a t x3 = *h/2 gives cl h2 c2 = 8 Thus the twodimensional state of stress is given by Figure 1. The third equation of equilibriuni gives 8~33 8 ~ 1 3 df ax3 ax1 dxl Integration with respect to x3 yields The functions f and g can be determined using the boundary conditions of the bearn.e. Since the stress compor~erits 023 are assumed t o be zero.. the first equation of equilibrium yields Integration with respect t o 23 gives where f is a function of X I only. Note that 013 and 033 must be zero on the top and bottom surfaces of the bearn (i. u22. a t 23 = * h / 2 ) .EQUATIONS OF ANISOTROPIC ELASTICITY 21 where 1 2 2 is area moment of inertia about the x2axis. g = 0. fixed a t one end and no support at the other end) under an end load.
20)]. independent of position).e. we obtain Thus the strains are compatible only if S15 = 0. the material properties are a function of position. ~ 3 are related t o the stress components 0 1 1 . These equations are called the constitutive equations. Suppose that the strains ~ 1 1~ . 0 1 3 . a structure composed of several uniform thickness layers of different materials stacked on top of each other and bonded t o each other is heterogeneous through the thickness. In the special case in which the work done by the stresses during a deformation is dependent only on the initial state and the current configuration. 1 3 and . .3. which is the case when the material is isotropic or orthotropic with respect to the problem coordinates. An isotropic or anisotropic material can be nonhomogeneous or homogeneous. 1.6 Generalized Hooke's Law The kinematic relations and the mechanical and thermodynamic principles are applicable t o any continuum irrespective of its physical constitution. An anisotropic body is one that has different values of a material property in different directions at a point. Materials for which the constitutive behavior is only a function of the current state of deformation are known as elastic. i.20) is satisfied. it is necessary to see if the strain compatibility condition in Eq. An isotropic body is one for which every material property in all directions at a point is the same. A material body is said to be homogeneous if the material properties are the same throughout the body (i. For example. Here we consider equations characterizing the individual material and its reaction to applied loads. and 033 by the relations (see the next section for 3 details) Then Substituting these strain components into the compatibility equation [see Eq.e.22 M E C H A N I C S OF LAMINATED COMPOSITE PLATES A N D SHELLS Since the stress field is derived from stress equilibrium equations.. In a heterogeneous body.3. (1.3. the material is called hyperelastic.. material properties are directiondependent. (1.
~ . syrrirnetry of E. Then if the stress corriponents are assumed to be linear furictions of the corr~poiients strain. We account for the thermal expansion of the rnaterial. Here.wo subscripts. This reduces the uumber of independent material stiffness components to 21. in general. To show this we express Eq. the principle of conservation of angular momentum requires the stress tensor to be symmetric. then the most general of form of the linear constitutive equations for infinitesimal deforrriations is where C is the fourthorder tensor of material parameters and is termed s t i f n e s s tensor. The linear constitutive model for irifinitesirnal deformation is referred to as the generalized Hooke's law.35) in an alternate form using single subscript notation for stresses and strains and two subscript notation for the ~ E ~ . Hence.3.e. and use S = 0 for the stress tensor and E = E for strain tensor in the inaterial description used in solid mechanics. The constitutive equations described here do not include creep at constant stress and stress relaxation at constant strain. we discuss the constitutive equations of linear elasticity (i. Then it follows from Eq. I11 the absence of body couples. the hotly recovers its original form corripletely upon removal of the forces causing deforrriation. There are.3. to If we also assume that the material is hyperelastic.ant during the deformation.e.35) that C. as discussed next 161. Suppose that the reference configuration has a (residual) stress state of a'. we will not distinguish between various measures of stress and strain.oone relationship between the state of stress and the state of strain in the current configuration. under isothermal conditions. relations between stress and strain are linear) for the case of infinitesimal deforrnation (i. (1. s4 = 81 scalar components of a fourthorder tensor. Since the strain tensor is synirrietric (by its definition). ( E ~ ~ ) that such we have Since the order of differentiation is arbitrary. IOU(<< 1). the material coefficierits that specify the coristitutive relationship between the stress and strain corrlporierits are assumed to be const.A material body is said to be ideally elastic when. This does riot automatically imply that we neglect temperature effects on deforrnation.jke must be syrrirnetric in the first t. arid there is a onet. The number of indeper~deritcomponents of C are considerably less because of the of symmetry of cr. (1. further reducing the mmiber of independent material stiffness corriponent~s 6 x 6 = 36.e. ~ i = ~ j i then Cijka must j . Thus.. crij = aji. and syn~nietry C . .. d2~o/dEijdEke d 2u ~ / ~ E ~ it = ~ follows that CiJke= Ckgij. i. Herice the number of independent rnaterial stiffness componerits reduces to ~ ( 3 = ~ ) 54.. be syrrimetric in the last two subscripts as well. there exists a strain energy density function U . which can produce strains or stresses as large as those produced by the applied mechanical forces.
In matrix form Eq. We assume that the stressstrain relations (l. (1. (1. In matrix notation.38a) can be written as Now the coefficients Cij must be symmetric (Cij = Cji) by virtue of the assumption that the material is hyperelastic. Equation (1. the components of strain are related to the components of stress by where Sij are the material compliance parameters with [S]= [C]' (the compliance tensor is the inverse of the stiffness tensor: S = C . Hence.3.35) now takes the form where summation on repeated subscripts is implied (now from 1 to 6).39a) becomes In the following discussion we assume that the reference configuration is stress free.38a. a: = 0 and strain free E! = 0.e.b) are invertible. we have 6+5+4+3+2+ 1 = 21 independent stiffness coefficients for the most general elastic material. ai ..3.material stiffness coefficients: It should be cautioned that the single subscript notation used for stresses and strains and the twosubscript components Cij render them nontensor components (i.l ) . .3.3. subscript notation for stresses and strains is called the engineering notation or the VoigtKelvin notation. The single . Eq. Thus. ~ i and Cij do not transform like the components of a vector or tensor).
symmetry of internal structure due to crystallographic form. say (x. Note that the phrase "material coordinates" used in connection with the material description should not be confused with the present term. 2 2 .38a.35) are valid only for the tensor components. In the remaining discussion. Suppose that (xl.g.40) Cijkl = t i pe j q e k r els Cpqrs where lijare the direction cosines associated with the coordinate systems ( x l . x2. but we may use one rnaterial coordinate system. Thus.). respectively. we will use the material description for everything. transforms according to the formula (1. xb. We note that the symmetry under discussion is a directional property and not a positional property. the material is called a monoclinic material. Both are fixed in the body.2. y.39a.b) and (1. in singlesubscript notation . with the X I . x i ) be two coordinate systems. etc. z) used to write the equations of motion and straindisplacement equations will be called the problem coordinates to distinguish them from the rnaterial coordinate system. In the following we discuss various planes of symmetry and forms of associated stressstrain relations. x2. 23) denote the coordinate system with respect to which Eqs. regular arrangement of fibers or molecules. y. Choose xiaxis such that x i = 23 (never mind about the lefthanded coordinate system as it does not affect the discussion) so that one system is the mirror image of the other. 2 2 . For example. Note that use of the tensor components of stress and strain is necessary because the transformation laws of the form (1.Material Symmetry Further reduction in the number of independent stiffness (or compliance) parameters comes from the socalled material symmetry. xs) to describe the stressstrain behavior.3. When elastic material parameters at a point have the same values for every pair of coordinate systems that are mirror images of each other in a certain plane.b) are defined.3. xk. for example.. The coordinate system (x. The definitions and sign conventions of the stress and strain components show that or. The fourthorder tensor. and Cljkland Cpqrs the components of the fourthorder tensor C in the primed and unprimed coordinate systems. let ( x l . and the two systems are oriented with respect to each other. (1. a material may have certain elastic symmetry at every point of a material body the properties may vary from point t o point. Monoclinic Materials When the elastic coefficients at a point have the same value for every pair of coordinate systems which are the mirror images of each other with respect to a plane. z ) . x2plane parallel to the plane of symmetry. that plane is called a material plane of symmetry (e. to describe the kinematics as well as stress state in the body and another material coordinate system ( x l . 2 3 ) and (xi. 23) are and ( x i . Positional dependence of material properties is what we called the inhomogeneity of the material. x i ) . We shall call them material coordinate system.3.
Thus out of 21 material 2 = parameters. The stressstrain relations for an orthotropic material take the form . for example.while all their independent stress and strain components remain unchanged in value by the change from one coordinate system to the other. e3.3. i. Similar discussion with the two alternative expressions of the remaining stress components yield C24= 0 and C 5 = 0.38b).43) Orthotropic Materials When three mutually orthogonal planes of material symmetry exist.3. we can write But we also have Note that the elastic parameters Cij are the same for the two coordinate systems because they are the mirror images in the plane of symmetry.3.8 = 13 independent parameters. (1. the number of elastic coefficients is reduced to 9 using arguments similar to those given for single material symmetry plane. principal axes of stress do not coincide with those of strain.42) can also be obtained using the following transformation matrix (which converts the unprimed coordinate system to the primed one) in Eq.3. as indicated below Note that monoclinic materials exhibit shearextensional coupling. = O for i fj ) (1. From the above two equations (subtract one from the other) we arrive at CI4E4 CI5&5= 0 for all values of + €4 and ES The above equation holds only if CI4 = 0 and C15= 0. = 1.. (1. (1.e. the can produce a normal stress. The result in Eq. and such materials are called orthotropic. we only have 21 . C34 = 0 and C35 = 0. a shear strain Therefore.40): [L] = 0 I: :I : 1 0 (orell =e22 = 1. l. and C4fj 0 and C56 = 0. all = Clses = 2C16~12. Using the stressstrain relations of the form in Eq.
Hence. These constants are measured using simple tests like uniaxial tension test or pure shear test.3. the extensional strain Ej:) in the material coordinate direction X I due to the stress all in the same direction is all/E1. Next we discuss how to relate the compliance coefficients SZI. For example. the material properties are determined in a laboratory in terms of the engineering constants such as Young's modulus. and so on. the strains of the same type produced by the application of individual stress components can be superposed. the engineering to constants. engineering constants are used in place of the more abstract stiffness coefficients Cy and compliance coefficients StJ. That is. One of the consequences of linearity (both kinematic and material linearizations) is that the principle of superposition applies. In particular. the sum of the displacements (and hence strains) produced by two sets of loads is equal to the displacements (and strains) produced by the sum of the two sets of loads. ratio V2l = El1  E22 .V ~ ~ ~ ~ ~ a l/ is the Poisson where V E ~ .The transformation matrices associated with the planes of symmetry are hlost simple mechanicalproperty characterization tests are performed with a known load or stress.44). shear modulus. The extensional strain &(I:) due to the stress a 2 2 applied in the 22 direction is . it is convenient to write the inverse of relations in (1. if the applied loads and geometric constraints are independent of deformation. Because of their direct and obvious physical meaning. The strainstress relations of an orthotropic material are given by where STj the cornpliur~cecoeficients ([C]= are [SIP') Most often.where El denotes Young's modulus of the material in the 21 direction.
E 1 produces a strain E\. and 3 material directions. (a)(d) in matrix form..2. j) Ej for i.47)]: v21 . G23.E3 are Young's moduli in 1. the compliance matrix [S]is the inverse of the stiffness matrix [C]and the inverse of a symmetric matrix is symmetric.. j = 1.v ~ ~ ~ ~ ~ / the ~total strain ~ 1 due to . we can write &33 =  011v13 El  E 2 022v23 +033 E 3 (4 The simple shear tests with an orthotropic material give the results Recall that 2cij (i # j) is the change in the right angle between two lines parallel to the xl and x2 directions at a point. and G Z 3G13. Writing Eqs. The 9 independent material coefficients for an orthotropic material are El. 2. E3.G12 are shear moduli in the 23.v13. aij (i # j) denotes the corresponding shear stress in the xixi plane. vij is Poisson's ratio. defined as the ratio of transverse strain in the j t h direction to the axial strain in the ith direction when stressed in the ith direction. 13. This in turn implies that the following reciprocal relations hold [see Eq. v32 .) equal to . G13.3. E2. Ea. in short  vij Ei vji (no sum on i.v12.49) . Similarly. (1.. .3.v23 E 2 El ' E3 El ' E3 E2 or.and E2 is Young's modulus of the material in the x2 direction. we obtain where El. Since . v31 . (1. Similarly. and 12 planes. and Gij (i # j) are the shear moduli in the xixj plane. respectively.3. ~ 1 2 ~ 1 3 v23 . the simultaneous application of all three normal stress components is where the direction of loading is denoted by the superscript. 033 Hence. G12.. respectively. it follows that the compliance matrix [S]is also a symmetric matrix.
3. First a stress a is applied in the xldirection as shown in Figure 1.48) gives u p ) = u y ) . However. The resulting strains are where the direction of loading is denoted by the superscript and negative sign indicates compression. which is the statement of Betti's reczproczty theorem (see Reddy [6]).4b. For example the difference between y 2 and for an orthotropic material is illustrated in Figure 1.4: Distinction between ~2 and v21 . between vij and vji for i # j for an orthotropic material [lo].4a. the displacements associated with the two magnitudes of E$:) and E~~ loads are ~g) and the reciprocal relation (1. for example. Figure 1.It is important to note the difference.3.3. Next. The strains are < if El > E2.4 with two cases of uniaxial stress for a square element of length a.we have no clue about the relative While it is obvious that (2) .3.3. the same value of stress is applied in the x2direction as shown in Figure 1.
5). 2113 = 0.25.036. (1.46) the stiffness coefficients can be expressed in terms of the engineering constants Example 1. Application of a normal stress t o a rectangular block of isotropic or orthotropic material leads to only extension in the direction of the applied stress and contraction perpendicular to it.3. whereas an anisotropic material experiences extension in the direction of the applied normal stress. (1. are: El = 25.Comparing Eqs.36 x lo6 psi .47 x lo6 psi = 0. (1.3. 2123 = 0. E g = 0. GZ3= 0.45) and (1.4: The material properties of graphite fabriccarbon matrix layers. the application of a shearing stress t o an anisotropic material causes .8 x lo6 psi .3.3. as well as shearing strain (see Figure 1.2 x lo6 psi . we note that and using Eq.54) as A qualitative understanding of the anisotropic behavior of a material can be obtained by simple tension and shear tests [lo]. contraction perpendicular to it.3.47). 2112 E2 = 4. Conversely. which are characterized as orthotropic.171 The matrix of elastic coefficients for the material can be calculated using Eq.75 x lo6 psi GI2 = 1.1 x lo6 psi .3. G13 = 1.
53).e.3. analysis.. Isotropic Materials When there exist no preferred directions in the material (i.47).55) 2 Consequently. Such materials are called isotropic.Normal Stress Shear Stress I I I Isotropic and Orthotropic j J fl Anisotropic Figure 1. For isotropic materials we have El = EP = E3 = E.3. the material has infinite number of planes of material symmetry).3.3. take the form   . (1. Eqs.44) and (1.54) and (1. in view of the relations (1.5: Deformation of orthotropic and anisotropic rectangular block under uniaxial tension. Normal stress applied to an orthotropic material at an angle to its principal material directions causes it to behave like an anisotropic material. arid design of composite rnaterials.55). (1. the number of independent elastic coefficients reduces to 2.3. The coupling between the two loading modes and the two deformation modes plays a significant role in the testing. ~ 1 = V ~ : J= ~ 1 3 u (1.3. Glz = G13 = G23 G. shearing strain as well as normal strains.3.
a =adilatation. a' = a  = E. Eqs.. the stressstrain relation for the isotropic case takes the form The strainstress relations are We note the following relations between the Lam6 constants X and p and engineering constants E.3.3. deviatoric strain.3. 22 (1.3. In view of the relations between the Lam6 constants and engineering constants.where A= (1 + v ) ( l .2v) E Alternatively. E' = E .1 mean stress. the stressstrain relations can be written in more compact form using the fact that a fourthorder isotropic tensor can be expressed as where X and p are called Lame' constants. (1. Therefore.63)  61.61) can be written in terms of engineering constants: .t r ( ~ ) (1. e 3 22' deviatoric stress. v and G for a n isotropic material [8]: The following definitions and constitutive relations are of interest in the sequel: .60) and (1.64) 3 1 where K is the bulk modulus and p = G is the shear modulus.
70a) are inverted to obtain the stressstrain relations where the Qij. ~ 3 is not zero. E l .~ 2 ) ) ua3 = ~ a 3 ( 5 1x2). Although a g g = 0. v12.3. and Gl2.47)] and the transverse normal strain is given by The strainstress relations (1.3. The transverse shear stresses are related to the transverse shear strains in an ~rt~hotropic material by the relations . called the plane stressreduced stiflnesses. E2.EQUATIONS OF ANISOTROPIC ELASTICITY 33 The strain energy density for a linear isotropic material is given by Plane StressReduced Constitutive Relations A state of generalized plane stress with respect to the x1x2plane is defined to be one in which cap = acrp(xl. are given by Note that the reduced stiffnesses involve four independent material constants. .3. a33 = 0 (1. (1.69) 3 where cr and 0 take the values of 1 and 2. The strainstress relations of an orthotropic body in plane stress state can be written as [see Eq.
4 is the specified boundary flux. = 8. Equation (1. The thermal problem for the solid requires the temperature or the heat flux to be specified on all parts of the boundary enclosing the heat transfer region as where is the total boundary enclosing the heat transfer region. The constitutive equation of the thermal problem is the well known Fourier's heat conduction law. which states that heat flux is proportional to the gradient of temperature: where k denotes the thermal conductivity tensor of order two. Q is the internal heat generation (measured per unit volume). The thermodynamic principles can be expressed. equations of motion of Section 1. is a reference (or sink) . thermodynamic equations of this section. Thermoelasticity The thermoelastic problem is governed by the straindisplacement equations of Section 1.. T. The first law of thermodynamics. I'T n r.3. The viscous dissipation couples the thermal problem to the stress problem.4. = rT U r. (1.7 Thermodynamic Principles Of the four principles of thermodynamics. as where T is the temperature. states that the time rate of change of the total energy is equal to the sum of the rate of work done by applied forces and the change of heat content per unit time.5. temperature for convective transfer. the thermal problem is coupled to the stress problem through constitutive relations. The second law of thermodynamics places restrictions on the interconvertibility of heat and work done. q is the heat flux vector. the second law states that the entropy production is positive.3. . h. is the convective heat transfer coefficient. and the constitutive equations to be given in this section. Even when the viscous dissipation is neglected. The negative sign in Eq.34 MECHANICS OF LAMINATED COMPOSITE PLATES AND SHELLS 1. is used to determine the temperature distribution in the body. in the Lagrangian description of deformation of solid bodies.3. c.74). termed the generalized heat conduction equation. as explained in the next section. is the specific heat at constant volume or constant strain. also known as the principle of conservation of energy. For irreversible processes. and i is the strain rate tensor (or time rate of the strain tensor).51) indicates that heat flows from higher temperatures t o lower temperatures. the first law of thermodynamics and the second law of thermodynamics are important in the study of deformable solids. a is the stress tensor.3.3. and s denotes the position of a point on the boundary. p is the density.
7 is the entropy density. and a i j are the thermal coefficients of a expansion and related to Pij by Pij = CiJke k e . The moisture concentration problem is mathematically similar to the heat transfer problem. it is irnportant to determine the temperature and moisture concentration in composite laminates under given initial and boundary conditions.To is the reference temperature. ( s . The negative sign in Eq.77b). the heat conduction problem described by equations (1.79a) where D denotes the mass dzffusitivity tensor of order two.7713) is known as the DuhamelNeumann law for an anisotropic body. and Pij are material coefficients. and rl n r 2= 0 and quantities with a hat are specified functions on the respective boundaries.3. It is assumed that and aij are initially zero.3. Therefore.3.79b) indicates that moisture seeps from higher concentration to lower concentration. and 4f is the moisture source in the domain. t ) on (1. Inverting relations (1. The moisture concentration c in a solid is described by Fick's second law: ac at  =Wqf +of (1.80a) where = rl U r 2 . The moistureinduced strains {E)" are given by .To. Equation (1. we obtain are where SZjke the elastic compliances.3. The boundary conditions involve specifying the moisture concentration or the flux normal to the boundary: c = i .3.EQUATIONS OF ANISOTROPIC ELASTICITY 35 The constitutive equations of thermoelasticity are derived by assuming the existence of the Helmholtz freeenergy function Qo = Q0(&ij. (see [ll141) T) such that where 8 = T .3. q f is the flux vector. Hygrothermal Elasticity Temperature and moisture concentration in fiberreinforced composites cause reductions of both strength and stiffness [15181.74)(1.3. As described in the previous section. (1.76) can be used to determine the temperature field.
we will use only thermal strains to show their contribution to governing equations in the sequel. The piezoelectric effect is described by the polarization vector P. the change in polarization vector A P is given by where p is the vector of pyroelectric coefficients. Electroelasticity Electroelasticity deals with the phenomena caused by interactions between electric and mechanical fields.3). the hygrothermal strains have the same form as the thermal strains [see Eq.where { a M )is the vector of coeficients of hygroscopic expansion.2. The piezoelectric effect is one such phenomenon. k = 1. It is related to the stress tensor by the relation (see [14171) where d is the thirdorder tensor of piezoelectric moduli. (1. For a small temperature change AT. Analogous to the strain energy function Uo for elasticity and the Helmholtz freeenergy function @o for thermoelasticity. the electric charge that is applied to actuate a structure provides an additional body force to the stress analysis problem. which represents the electric moment per unit volume or polarization charge per unit area. we assume the existence of a function . and the piezoelectric laminae send electric signals that are used to measure the motion or deformation of the laminate. and it is concerned with the effect of the electric charge on the deformation [14161. thermal. A laminated structure with piezoelectric laminae receives actuation through an applied electric field. The total strains are given by where To and co are reference values from which the strains and stresses are measured.3. Thus. In these problems. The coupling between the mechanical. In view of the similarity between the thermal and moisture strains. The pyroelectic effect is another phenomenon that relates temperature changes to polarization of a material. The inverse effect relates the electric field vector £ to the linear strain tensor E by Note that dkijis symmetric with respect to indices i and j because of the symmetry j of ~ i (note that i .76)]. much the same way a temperature field induces a body force through thermal strains.j. and electrical fields can be established using thermodynamical principles and Maxwell's relations.
3. .3. such that where aij are the components of the stress tensor. D iare the components of the electric displacement vector. eijk are the piezoelectric moduli.88).3. Maxwell's equation governing the electric displacement vector is given by It is often assumed that the electric field & is derivable from an electric scalar potential function $: & = V$ (1. (1. 2 . . . and To is the reference temperature.86ac) can be expressed as Note that the range of summation in (1.3.87b). is the specific heat per unit mass.3. (1.3.k . Eqs. eij are the dielectric constants. there are 21 independent elastic constants.85a) in Eq. (1. .3. Pij are the stresstemperature expansion coefficients. In the coming chapters reference is made to many of the equations presented here. and 17 is the entropy. c.3. In singlesubscript notation. 3 pyroelectric constants.8513) gives the constitutive equations of a deformable piezoelectric medium: where Cijke are the elastic moduli.2.89) This assumption allows us to write Eq. (1. j = 1 . l = 1. 18 piezoelectric constants. 6 . pk are the pyroelectric constants.EQUATIONS OF ANISOTROPIC ELASTICITY 37 which is called the electric Gibbs freeenergy function or enthalpy function. (1. .87ac) is different for different terms: i. in view of Eq.3.90b) This completes a review of the basic equations of solid mechanics. For the general anisotropic material. 6 dielectric constants. Use of Eq. as where f e =  a dxk (eke&! + pko) (1. and 6 thermal expansion coefficients.
These configurations are restricted to a neighborhood of the true configuration so that they are obtained from infinitesimal variations of the true configuration. These displacements need not have any relationship to the actual displacements that might occur due to a change in the applied loads. The displacements are called virtual because they are imagined t o take place (i. equations of equilibrium or motion of the system). The variational approach. it is useful to study variational principles and methods (see Reddy [6] for additional details)..e. and it is this configuration that satisfies Newton's second law (i. the governing equations are obtained by the principle of virtual displacements or by seeking the minimum of the total potential energy of the system.4. In a variational approach.e. Of all the possible configurations. the principle of virtual displacements will be used to derive the equations of motion of laminated plates. is known in mechanics as Newton's second law and it is also derivable from a variational principle. The virtual displacements at the boundary points at which the geometric conditions (or displacements) are specified.4 Virtual Work Principles 1. The use of Newton's laws to determine the governing equations of a structural problem requires isolation of a typical volume element of the structure with all its applied and reactive forces (i. which requires that the vector sum of all applied forces acting on a body be equal to the total time rate of momentum of the body. The work done by the actual forces moving through virtual displacements is called virtual work. 1.4.2 Virtual Displacements and Virtual Work From purely geometrical considerations.. For complicated systems the procedure becomes more cumbersome and intractable. During such variations. The set of configurations that satisfy the geometric constraints but not necessarily Newton's second law is called the set of admissible configurations. In the context of the present study. the type of boundary conditions to be used in conjunction with the derived equations is not always clear. When a mechanical system experiences such variations in its configuration.1.. We begin with the concepts of virtual displacements and forces. and obtaining approximate solutions by variational methods. Hence. the geometric constraints of the system are not violated and all the forces are fixed at their actual values. the freebody diagram of the element). it is said t o undergo virtual displacements from its true or actual configuration. is useful both in deriving governing equations and boundary conditions.e. a given mechanical system can take many possible configurations consistent with the geometric constraints of the system. hypothetical) while the actual loads acting at their fixed values. For example. applicable to linear or nonlinear theories. In addition. The virtual work done by actual forces F in a body noin moving through the virtual displacements 6u is given by . only one corresponds t o the actual configuration. are necessarily zero.1 Introduction In solid mechanics some of the laws of physics take several alternative forms. the principle of conservation of linear momentum.
? I For example. for example.3. the work done by the force due to stress a12 in the body is Thus. (1.r. in moving through the virtual displacement 6ul = 6 ~ ~ is~ d x ~ j Here ~ i denote the strain components and o. is given by ? ? . where c is an arbitrary constant. The work done by these internal forces in moving through displacements of the material particles is called znternal work. Therefore.4. It is understood that the displacements are specified on the remaining portion I?. ~ due to the virtual displacements Su.. The internal virtual work due t o the virtual displacement Su can be computed as follows. of the boundary I?. provided bu(0) = 0. The forces associated with the stress field move the material particles through displacements corresponding to the strain field in the body. Similarly. The negative sign in Eq. where [see Eq. Note that the work done on the body is responsible for the internal work stored in the body.j the stress components. one may select 6u(x) = cx.EQUATIONS OF ANISOTROPIC ELASTICITY 39 where du denotes the volume element dv = d ~ ~ d x ~in the ~ d z : material body 0" The external virtual work done due to virtual displacements 6u in a solid body Ro subjected t o body forces f per unit volume and surface tractions t per unit area of the boundary I.12)] The work done by the force due to actual stress all. and hence work is done. Thus.2) indicates that the work is performed on the body. Recall that the deformation of solid body acted upon by forces can be measured in terms of strains and that the body experiences internal stresses. (1. = r . a bar fixed at one end (x = 0) and subjected to an axial load at the other end ( z = L) can be irnagined to have a virtual displacement 6u(z). Suppose that an infinitesimal material element of volume dv = dx1dx2dx3 of the body experiences virtual strains S E .. because the actual displacement is specified at x = 0. the total virtual work done by forces due to all the stresses in a volume element (that originally occupied the material element dv) in moving through their respective displacements is . the virtual displacements are zero on . where ds denotes a surface element and I denotes the portion of the boundary on which stresses are specified. irrespective of whether u is specified to be zero or not.
5) is called the virtual strain energy of a deformable body. We have .6) is also known as the virtual complementary strain energy. and Su is called the first variation of u.e. for fixed x. (1. (1.4. Thus S is an operator that produces virtual change or variation Su in a dependent variable u. It is used to denote a variation (or change) in a given quantity. Su denotes a variation in u. is The first variation of F is Since Su is small.The total internal virtual work done is obtained by integrating the above expression over the entire volume of the body Equation (1.4.6) The expression in Eq. The internal virtual work done by virtual stresses Soij in moving through the j actual strains ~ i is SU* = Lo EQ Soij du (1. terms involving squares and higher powers of Su can be neglected.27b)l and stress boundary conditions [see Eq. The expression in Eq. i. The operator proves to be very useful in constructing virtual work statements and deriving governing equations from virtual work principles.. There is an analogy between the variational operator 6 and the total differential operator d.25)] are satisfied: In the present study we will not consider complementary energy principles. 1. To see this consider a function F of the dependent variable u and its derivative u' = duldx in one dimension. as will be shown shortly. (1.3.3 Variational Operator and Euler Equations The delta symbol S used in conjunction with virtual displacements and forces can be interpreted as an operator. The virtual forces (Sfi. (1. in much the same way as dx denotes a change in x.4.5) is valid for any material body irrespective of its constitutive behavior.2. called the variational operator.4. The total differential of F.4. Sti) and virtual stresses (Soij) should be such that the stress equilibrium equations [see Eq.
. A functional is said to be linear if for all constants a and /3 and dependent variables u and v. Note that I ( u ) is a number whose value depends on the choice of u. we have dx = 0 in Eq. the integral expression qualifies as a functional for all integrable and squareintegrable functions u(x).17) where.8) and the analogy between S F in Eq. the laws of variation of sums.G (1.4. is a differential operator with respect t o the dependent variable. products. (1.8) becomes apparent: the variational operator.4. v . a functional I(. are completely analogous to the corresponding laws of differentiation. u. Thus.Since x is fixed during the variation of u to u+6u.G+6. for example. (1.10) and d F in Eq. the total variation is the sum of partial variations: 6G = 6. powers. w) is a function of several If dependent variables (and possibly their derivatives). ratios. S denotes the partial variation of G with respect to u. Functionals Integral expressions whose integrands are functions of dependent variables and their derivatives are called functionals. and so forth. R: For example. a functional is a real number (or scalar) obtained by operating on functions (dependent variables) from a given set is (or vector space).G+ 6. Indeed. G = G(u. (1. The following properties of the variational operator should be noted: where Fl = F l ( u ) and F2 = F2(u). A quadratic functiorlal is one which satisfies the relation .) an operator which maps functions u of a vector space H into a real number I ( u ) in the set of real numbers. Mathematically.4. 6.4.
b). b).2413) In most of our study in this book. A mathematical proof of the lemma can be found in most books on variational calculus. then Ja then G=Oina<x<b and B ( a ) = O (1.e.10). A more general statement of the fundamental lemma is as follows: If q is arbitrary in a < x < b and q(a) is arbitrary. Since q is arbitrary. minimum or maximum) of the functional The necessary condition for the functional to have a minimum or maximum is (analogous t o minima or maxima of functions) that its first variation be zero: . The first variation of a functional I ( u ) of u (and its derivatives) can be calculated using the definition in Eq. For instance consider the functional I ( u ) defined in the interval (a.b) because they provide the means to the determination of the governing equations and boundary conditions and their solution by the variational methods. Thus. (1. that G = 0 in (a. then it follows .24a. it can be replaced by G. We have Since an integral of a positive function is positive.. of x . the above statement implies that G = 0. the variation of a functional can be readily calculated Fundamental Lemma of Variational Calculus The fundamental lemma of calculus of variations can be stated as follows: for any integrable function G.4.for all constants a and dependent variable u. u and du/dx functional I is = u'. b) I(u) = lb F ( x .4. in general. we shall be interested in the use of Eqs. u. ZL') dx (1.4. Consider the question of finding the extremum (i. A simple proof of the lemma follows. (1. if the statement holds for any arbitrary continuous function ~ ( x )for all x in (a.21) The first variation of the where F is a function.4.
The variable I L that is subjected to variation is called the primary variable. is called a secondary variable. it follows that Thus the necessary condition for I ( u ) to be an extremum at u = u ( z ) is that u ( z ) he the solution of Eq. u < z < b. (1.28).e. In solid mechanics. the expression next to 6u in the boundary term. we integrate the second term by parts and obtain Let us first examine the boundary expression: There are two parts to this expression: a varied quantity and its coefficient.4.4. All admissible variations must satisfy the homogeneous form of the essential (or geometric) boundary conditions: &L(U) 0 and 6u(b) = 0. We have which must hold for any Su in (a.4.4. We cannot use the fundamental lemma in the above equation because it is not in the form of Eq. (1.27).10) we obtain Note that Su' = S(du/dz) = d(Su)/dz.24). Returning to Eq. (1. these are known as the geometric and force boundary conditions. . i. Elsewhere. we note that the boundary terms drop out because of the conditions on 6u. (1.24).EQUATIONS O F ANISOTROPIC ELASTICITY 43 Using Eq. respectively. b).4. The coefficient of the varied quantity. The specification of the primary variable at a boundary point is terrned the essential boundary condition. (1. To recast the above equation in the forrn of Eq. and the specification of the secondary variable (aF/3u') is called the natural boundury condition.. The product of the primary variable (or its variation) with the secondary variable often represents the work done (or virtual work done). = Su is arbitrary. In view of the fundamental lemma of calculus of variations (7 = nu).
the virtual displacements Sul. . For a discussion of the principle of virtual forces and its special cases. respectively. consult Reddy 161. .28).F. and Su(b) is arbitrary (i.4 Principle of Virtual Displacements Recall that the virtual work due t o virtual displacements is the work done by actual forces in displacing the body through virtual displacements that are consistent with the geometric constraints. Su2. In addition. we have n 6~ = i= 1 F.4. and suppose that the points of application of these forces are subjected to the virtual displacements 6ul.. . (1.= 0 (1. This implies that the total virtual work. Note that the boundary conditions that are a part of the EulerLagrange equations always belong to the class of natural boundary conditions. Thus. Su. 6U SV.4. Now we have all the necessary concepts and tools in place to study the principles of virtual work. The virtual displacement 6ui has no relation t o Suj for i # j . (1. the principle of minimum total potential energy. F2..e. hui =  ( Fi) f k l Su and hi.. . is equal to zero.4.. . 1. for a rigid body.. we discuss the principle of virtual displacements and its special case. All applied forces are kept constant during the virtual displacements. the vector sum of the forces acting on a body in static equilibrium is zero. then &(a) = O and we have from Eq.32) But by Newton's second law. Sun should all be the same. Consider a rigid body acted upon by a set of applied forces F1.30b) are called the EulerLagrange equations.27) the result Since Su is arbitrary in (a. that both the integral expression and the boundary term be zero separately: Both Eq. in view of Eq.4.30a) and Eq. . b) and 6u(b) is arbitrary. say Su.4. . for a body in equilibrium the total virtual work done due to + .44 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS x If u(a) = u. (1.4. . In the next section. (1. u(a) is specified but u is not specified a t = b). . . Su2. Thus. The external virtual work done by the virtual displacements is The internal virtual work done 6U is zero because a rigid body does not undergo any strains (hence virtual strains are zero). the above equation implies.
they must belong to the set of admissible variations. Eq. The set of admissible configurations are defined by sufficiently differentiable functions that satisfy the . The virtual displacements are arbitrary. tractions are specified to be t. deformable bodies. ? .4... the principle of virtual displacements and its special case are described since they play an important role in the formulation of theories (e.. plate theories) and their analysis by variational methods of approximation. the virtual work of all actual forces in moving through a virtual displacement is zero: SU+SVSW=O (1. For a solid body. In order to determine the equations governing the equilibrium configuration C. The principle of virtual displacements can be stated as: if a continuous body is in equilibrium. the actual one corresponding to the equilibrium configuration makes the total virtual work done zero.35) of the principle of virtual displacements are nothing but the equilibrium equations of the . Let u be the displacement vector corresponding to the equilibrium configuration of the body. whose volume is denoted as n o . do not overlap). the external and internal virtual work expressions are given in Eqs. and their sum is the total boundary I'. and on portion r the . and let a and E be the associated stress and strain tensors.g. The boundary portions I and I are disjoint (i. (1.5). then of all admissible configurations.34) takes the form where the summation on repeated subscripts is implied. we can derive them for the statement in Eq. continuous functions except that they satisfy the homogeneous form of geometric boundary conditions. Consider a continuous body B in equilibrium under the action of body forces f and surface tractions t.e. for which SU is not zero. (1. geometric boundary conditions: u = u on r. (1. i. Suppose that over portion I'.4. respectively. The principle of virtual work is independent of any constitutive law and applies to both elastic (linear and nonlinear) and inelastic continua.e. ? .EQUATIONS OF ANISOTROPIC ELASTICITY 45 virtual displacements is zero. respectively." no is the volume of the undeformed body.4. If the body is in equilibrium. This statement is known as the principle of virtual displacements.33) Just as we derived the EulerLagrange equations associated with the statement 6 1 = 0. However. The EulerLagrange equations associated with the statement (1.33).4. In this section. we let the body experience a virtual displacement 6 u from the true configuration C.4.2) and (1.4. Writing in terms of the Cartesian rectangular components. first we must identify SU and SV for a given problem. Let the reference configuration be the initial configuration c'. The principle can be expressed as where a : SE denotes the "double dot product.of the total boundary r of the region no the displacements are specified t o be u. and d v and d s denote the volume and surface elements of no. The principle also holds for continuous.
one obtains (aij= aji) Since I? = I?. z) along the x.11). (1.40) are the natural boundary conditions.38). Example 1.e. (1.2.4. w=w0(x) (1. (1. (1. Under the assumption of small strains and displacements.40) are the EulerLagrange equations associated with the principle of virtual displacements for a body undergoing small deformation.4. which assumes that straight lines perpendicular to the beam axis before deformation remain (1) straight. The stress boundary conditions in Eq. and (3) inextensible after deformation. 0. y and z coordinates..3.1). and (uo.4. respectively. Eq.41) dx where (u. v.4.wo) are the displacements of the point (x.1: (EulerBernoulli beam theory) Consider the bending of a beam of length L.O). Eq. Young's modulus E and moment of inertia I. dw0 v=O. U r. we derive the governing differential equation of the beam using the EulerBernoulli hypotheses. (2) perpendicular to the tangent line to the beam axis.4.27b)l r.35).4. (1. and using the divergence theorem. 4 . and subjected to distributed axial force f ( x ) and transverse load q (see Figure 1. elastic or inelastic).. we have Because the virtual displacements are arbitrary in Ro and on the following equations [cf. Eq. (1...z .4.4. The principle of virtual displacements is applicable to any continuous body with arbitrary constitutive behavior (i. The virtual strains 6&ijare related t o the virtual displacements 6ui by Substituting from the above equation into Eq.38) yields Equations (1. h ) u = u o ( x ) . w) are the displacements of a point (x.3.. and 6ui = 0 on r. Recall the straindisplacement equations from Eq. Under the assumption of smallness of strains . These assumptions lead to the displacement field (see Figure 1 .39) and (1.3D elasticity theory. to transfer differentiation from 6ui to its coefficient. y.
. on a cross section as (see Figure 1 . which are known as the stress resultants.EQUATIONS OF ANISOTROPIC ELASTICITY 47 Figure 1. ~ d A . and a. t h e only nonzero strain is First we derive the equilibrium equations using Newton's second law of motion. arid they are defined in terms of the stresses u.dA (1.1: Bending of beams. (b) Equilibrium of a beam element.4..dA.4. and V(x) the shear force. V(X)= S. .4.4. (1.. and rotations.4.43ac) reduce t o .4.43b) and (1.. (c) Definitions (or internal equilibrium) of stress resultants. Summing t h e forces and moments on a n element of the beam (see Figure 1. u. Equations (1.lb) gives the followirig equilibrium equations: where N ( z ) is the net axial force. M ( x ) the bending moment. 4 .43~)can be combined into the single equation so that Eqs..44) Here A denotes the area of cross section. (a) Kinematics of deformation of an EulerBernoulli beam. 1 ~ ) N(r)=Ln:. M(x)= I* ' u .
4.42)] First.. we derive the governing equations (1.dx.4.z(d2Suio/dx2). Using the relations in Eq.2) . Note that the xaxis is taken through the geometric centroid of the cross section so that JA zdA = 0.b) using the principle of virtual displacements. The total internal virtual work done is where all other stresses are assumed to be zero. is The virtual work done by any applied point loads (and moments) must be added to 6V in Eq.4. (1.4.e. we obtain Next.dA..4. Substituting this expression into (1.The stress resultants (N. dA in moving through the virtual displacements SE. Hence.. dx is given by u. we obtain The virtual work done by the external distributed forces f (x) and y(x) in moving through the displacements 6uo and Swo. i. (1.. The virtual ~ strain SE. the virtual work done by the counterclockwise moment ML at x = L in rotating through the virtual rotation (L) is and the virtual work done by an axial point load PL in moving through 6uO(L)and a transverse point load FL in moving through the virtual displacement 6wo(L) is (see Figure 1. the internal virtual work done per unit length of the beam by the actual internal force u... For example.6wo) by S E =~(d6uo/dx) . (1. The actual strain in the EulerBernoulli beam theory is given by Eq..45a.42). (1.4. 6~. Note that for the problem at hand the only nonzero stress is a.4. (1. M ) can be related back to the stress a.48) in Eq. using the linear elastic constitutive relation for an isotropic material as [see Eq. is related to the virtual displacements (6uo..50).51b). respectively.4. note that where I is the moment of inertia about the axis of bending (yaxis) and z is the transverse coordinate..46). the EulerBernoulli assumptions are invoked.
wo and d w o / d x are primary variables arid N. the total external virtual work done is The principle of virtual displacements states that if the beam is in equilibrium we must have bU+bV=Oor To obtain the EulerLagrange equations associated with the virtual work statement (1.54).2: A cantilever beam with distributed loads f and q.4. Figure 1.b). d h f l d x = V and M are the secondary variables of the problem.4.45a. Since 6uo and bwo are independent and arbitrary in 0 < x < L. consider the integral expressions in (1. and concentrated loads PL.EQUATIONS OF ANISOTROPIC ELASTICITY 49 Thus.4. integrate the first term by parts once and the second term by parts twice and obtain Note from the boundary terms that u o . .47). (1. we obtain the Euler equations which are the same as those in Eqs. We have First. FL and M L at the right end.4.
t .34). The statement of the principle of virtual displacements.57) together define axial deformation.4. The Principle of Minimum Total Potential Energy A special case of the principle of virtual displacements that deals with linear as well as nonlinear elastic bodies is known as the principle of m i n i m u m total potential energy.48)].6u d v + l. (1. the virtual displacements 6uo and 6wo must satisfy the conditions and they are arbitrary a t x = L. The strain energy density Uo is a singlevalued function of strains at a point and is assumed to be positive definite.6u d s ] =0 (1. (1.55~~) (1.4. For elastic bodies (in the absence of temperature variations) there exists a strain energy density function Uo such that Equation (1. fourth and sixth boundary expressions vanish.4.4.4.4. (1.Next. These sets of equations can be solved independently as N is only a function of uo and M is a function of only UIO [see Eq. (1.59) describe bending deformation of the beam.54).60) represents the constitutive equation of an hyperelastic material.58) and (1. If the beam is fixed a t x = 0 and subjected t o forces PL. consider the boundary expressions in (1.4. and we have the (natural) boundary conditions resulting from the virtual work principle: We note that Eqs.61a) or. in component form. can be expressed in terms of the strain energy density 1Jn : " : 6E d v  [Lo f .4.4. Consequently. The first integral is equal to E where U is the internal strain energy functional U= 0 Uo ddv .4. the second. and F L .M L . Eq.55b). while Eqs. and (1.
2). (1. PL and FL is The total potential energy of the beam is given by The total potential energy principle requires that 6(U + V) = 0: . (1. sum of the strain energy arid potential energy due to applied loads) of the beam and set its first variation to zero to obtain the EulerLagrange equations of the functional.48) is used to write the last expression for U.4. The nliriirrlurri total potential energy pririciplc requires us to construct the total potential energy (i.4. It means that of all admissible displacements. The total strain energy stored in the beam is where Eq. The equality holds only if u = u. g . ATL. those which satisfy the equilibrium equations make the total potential energy a m i n i m u m : + where u is the true solution and u is any admissible displacement field. The work done by external applicd loads f .4.63) is known as the principle of m i n i m u m total potential energy.Suppose that there exists a potential V whose first variation is Then the principle of virtual work takes the form The sum U V = II is called the total potential energy of the elastic body. Example 1.e.4.4.2: We consider the cantilever beam problenl of Example 1..1 (see Figure 1. The statement in Eq.
4.. when the beam constitutive equations are used.Integration by parts of the first two terms.55a.57)(1. w) u=uo+avl.56) and the property that Suo and Swo are arbitrary both in (0.e. vl(0) = 0 + pvz. (2) (1.59) are the same as above when N and M are replaced in terms of uo and wo using Eq.4. L) and at x = L. (1. /? small. and use of Eq.4. v2(0) = 0.4. consider the second integral and the boundary terms .69a) =O (1. ( a . i.47a.4.b). yields the Euler equations Equations (1.4.4.69b) z=o For the example problem we have Now. (1. = wo a small. and (1. The minimum property of the total potential energy can be established by considering an arbitrary admissible displacement field.b).
69a. z ) = 0 for all x t (1. Thus. Hamilton's principle may be considered as dynamics version of the principle of virtual displacements [6]. (1.4. we considered axial deformation of a bar (set 'wo = 0) as ~ well as pure bending of a beam (set I L = 0).4. The actual path u = u ( x . where Su is the admissible variation (or virtual displacement) of the path.72) where rn is the mass.4. One may note that in this example. Newton's second law of motion applied to deformable bodies expresses the global statement of the principle of conservation of linear momentum. the energies can be expressed in terms of the dependent variables (which are functions of position) of the problem. Thus n ( u > is greater than n ( u o . Thus.li # uo.t i ) = SU(X. These equations are uncoupled for the case of small strains. For deformable bodies. Hamilton's Principle Hamilton's principle is a generalization of the principle of virtual displacements to dynamics of systems. (1. Since (uo. We suppose that the varied path differs from the actual path except at initial and final times.ulo) is the true solution of the problem. an admissible variation Su satisfies the conditions. Newton's second law of motion for a continuous body can be written in general terms as Fma=O (1. Su = 0 on S1 for all t Su(x. The total potential energy is the minimum with respect to both uo and 7lio. (1. consistent with kinematic (essential) boundary conditions.4. t ) . respectively. to u + Su. a kinetic energy K and a potential energy II. Eq. all terms in Eq. and F is the resultant of all forces acting on the body.70a) becomes 'a # rno and the equality holds only when u = uo and w = wo. the kinematic conditions and constitutive equations discussed in the previous sections are needed to completely determine the motion. it should be noted that Newton's second law of motion for continuous media is not sufficient to determine its motion u = u ( x .4.EQUATIONS OF ANISOTROPIC ELASTICITY 53 The boundary terms a t x = 0 are zero because of the conditions in Eq.73b) .70b) are zero.4. However.b). establishing the minimum character of the total potential energy of the beani. t l and t z . t) followed by a material particle in position x in the body is varied.U J ~when w) ) and . The principle assumes that the system under consideration is characterized by two energy functions. a the acceleration vector.73a) (1.
4.3. because F.76).74) represents the virtual work of internal forces stored i n the body. The last term in Eq. The work done by the inertia force ma in moving through the virtual displacement S is given by u where p is the mass density (can be a function of position) of the medium.4.4. Equation (1.54 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS where S1 denotes the portion of the boundary of the body where the displacement vector u is specified. We have the result In arriving a t the expression in Eq. and and tt E are the stress and strain tensors.76) is known as the general form of Hamilton's principle for a continuous medium (conservative or not. The strains S are assumed to be F compatible in the sense that the straindisplacement relations (1.76).72) with 6u gives work done a t point x.4. For an ideal elastic body. we obtain .b) into Eq. Integration of the product over the volume (and surface) of the body gives the total work done by all points. (1. a.7313).4. (1. integrationbyparts is used on the first term. and u are vector functions of position (whereas the work is a scalar). the integrated terms vanish because of the initial and final conditions in Eq. we recall from the previous discussions that the forces f and t are conservative. (1.4. Note that the scalar product of Eq.11) are satisfied. and that there exists a strain energy density function Uo = & ( E ~ ~ such ) that Substituting Eqs. (1. and elastic or not).4.77a. (1. (1. The work done o n the body at time t by the resultant force in moving through the virtual displacement S is given by u where f is the body force vector. t the specified surface traction vector.
t ) where c : ~= 4/(3h2).b) were used in arriving at Eq. through the beam height.hr . (II = U V) can be obtained from Eq.78): + + 0=6 1:' L ( u .4.cd ~ next. Now suppose that the beam is subjected to tlistributed axial force f ( z ) and transverse load of q(z. (1.4.4. The displacemcrlt field (1. of the body. and 4 the rotation of a point on the cerltroidal axis x of the beam.3 ( Thirdorder beam.81) are the EulerLagrange equations for an elastic body.t ) (i. V u . Thus.4. Recall that the sum of the strain energy arid potential energy of external V. L = K .e. t ) = 'uo(z.80) from Eq.73a. only restriction tiom the EulerBernoulli beam theory that is kept is ~ ( zz .!t ) = wo(x. Hamilton's principle (1.78) reduces to the principle of virtual displacements.82). The procedure t.II...4. (1. 'uo is the axial displacement. t )  qz" (4 + 2) ) = wo(z. we will riot consider specific geometric or force boundary corltiitioils here.iorl to lxcome (cubic) curves with arbitrary slope at z = 0. forces are applied sufficiently slowly such that the niotiori is independent of time. . (1.. is called the total potential energy.raight lines normal to the beam axis before tleforrrlat.4. transverse deflection is independent of the thickness coordinate 2). Since we are primarily interested iri deriving the equations of motiorl and the nature of the boundary conditions of the beam that experiences a displacerncnt ficld of the form in Eq. U involving no motion (i. t ) along the length of the beam.4. For bodies forces.t ) u?(:I. tl and for x in V and also on S2. arid Eqs.EQUATIONS OF ANISOTROPIC ELASTICITY 55 where K and U are the kinetic and strain energies: Equation(1. Example 1.4. theory) Consider the tlisplacenmlt field u(x: z . Because 6u is arbitrary for t .78). and (b) reql~iring transverse shear stress to thc vanish a t the top and bottom of the beam. 7 4 the trarisvcrse displacement. The EulerLagrange equations associated with the Lagrangian.78) represents Hamilton's principle for an elastic body (linear or nonlinear). t z + z$q:c. (1. (1.o obtain the equations of motion and boundary cor~ditiorisirivolvrs t.c. II.. and the inertia forces are negligible). gradient theorems.82) acconirnodates quadratic variation of transverse shear strain E . u) dt where integrationbyparts.4.it follows that Equations (1.4. The disp1accmr:nt field is arrived by (a) relaxing thc EulcrBcrrioulli hypotheses to let the st. and~ shear stress u. as can bc seen from the strains cornput.
:1 vanish on account of the assumption that all variations and their derivatives are zero a t t = 0 and t = T. Although one can use the general nonlinear straindisplacement relations. a. Note that y.following steps: (i) compute the strains... From the dynamic version of the principle of virtual displacements (i. (ii) compute the virtual energies required in Hamilton's principle. = Gy. and (iii) use Hamilton's principle. derive the EulerLagrange equations of motion and identify the primary and secondary variables of the theory (which in turn help identify the nature of the boundary conditions).. Hence. Hamilton's principle) we have where all the terms involving [ . The linear strains associated with the displacement field are where and cz = 4 / h 2 . & is a quadratic function of z . is also quadratic in z . here we restrict the development to small strains and displacements. = 2 .e. and the new variables introduced in arriving at the last expression are defined as follows: .
4. .87) and (1.b) by setting cl = 0 (but not c2): These equations are lowerorder than those in Eqs. The corresponding secondary variables are the coefficients of Sue..92) where K is the shear correction factor.87) and (1.4.82)..88a. by Q..4. .b)..85). the EulerLagrange equations are The last line of Eq. A simplified thirdorder beam theory can be obtained from Eqs.e.4. (1. (1.@)and Note that the Timoshenko beam theory accounts for transverse shear strain (N.b) by setting cl = cz = 0: The primary and secondary variables of the Tirnoshenko beam theory are: (uo. : y. which indicate that the primary variables of the theory are (those with the variational symbol) uo..4. and 36wo/dx: When cl = 0 in Eq. In the Tinloshenko beam theory Q. (1. Thus. II = I3 = I5 = 0) Thus.Q. (1.. 6wo.4. (1.88a. = y and hence Q. = K /' ozzd~ (1.4. uio.4.EQUATIONS OF ANISOTROPIC ELASTICITY 57 Note that I.4.and dwo/dx. the equations of motion of the Tirnoshenko beam theory can be obtained directly from Eqs.wo.aory.. @. 64..4.n'I. it corresponds to the displacement field of the Timoshenko beam th.84) includes boundary terms. are zero for odd values of i (i.88a. is defined. in place of the definition (1.).87) and (1.
5 Variational Met hods 1. The weightedresidual methods will be visited only briefly. The methods to be described here are known as the classical variational methods. and material behavior. Interested readers may consult the references at the end of the chapter for additional details 161.5. including the geometry.4 the principle of virtual displacements gives the equilibrium equations as the EulerLagrange equations.. EulerLagrange) equations. Therefore. Since the solution of a continuum problem in general cannot be represented by a finite set of functions. such as those provided by the principles of virtual displacements or the minimum total potential energy.2 The Ritz Method As noted in Section 1. The assumed solutions in the variational methods are in the form of a finite linear combination of undetermined parameters with appropriately chosen functions. we saw how virtual work and variational principles can be used to obtain governing differential equations and associated boundary conditions. One such direct method was proposed by Ritz [26]. Here we study the direct use of the variational principles in the solution of the underlying equations.e.1. In these methods. Such solution methods are called direct methods because the approximate solutions are obtained directly by applying the same variational principle that was used to derive the governing (i. and so on). and the weightedresidual methods (e. There exists a number of approximate methods that can be used to solve differential equations (e. The variational methods of approximation to be described here are limited to the Ritz method. As the number of linearly independent terms in the assumed solution is increased. the leastsquares method. it is understood to be relative to exact solutions of the governing equations that inherently contain approximations. etc.4. These governing equations are in the form of differential equations that are not always solvable by exact methods of solution. which are . the error in the approximation will be reduced. This amounts to representing a continuous function by a finite set of functions. The approximations are introduced by several sources. The most direct methods are those which bypass the derivation of the EulerLagrange equations. the solution obtained is an approximation to the true solution of the equations describing a physical problem. when one thinks of permissible error in an approximate solution.g.. 1. finitedifference methods. The Ritz method is based on variational statements. and the assumed solution converges to the exact solution. error is introduced into the solution. It should be understood that the equations governing a physical problem are themselves approximate. and go directly from a variational staternent of the problem to the solution of the equations. collocation method.5. the finite element method.g.). Therefore.1 Introduction In Section 1. representation of specified loads and boundary conditions.. we seek an approximate solution to the problem in terms of adjustable parameters that are determined by substituting the assumed solution into a variational statement equivalent to the governing equations of the problem.
cj are termed the Ritz coeficients. i. . and the approximate solution converges to the true solution of the problem as the number of parameters N is increased. . In the Ritz method we approximate a dependent unknown (e. .1) into II(u) for u and the minimization of J2(cj) results in a set of algebraic equations among the parameters cj. cpo has the principal purpose of satisfying the specified esser~tial geometric) (or boundary conditions associated with the variational formulation. the sequence {pj) is said to be complete if there exist.2. . A sequence of algebraic polynomials. . N ) should satisfy the following three conditions: (a) be continuous as required in the variational statement (i..1) can be viewed as a representation of u in a finite component form. j = 1. Equation (1.5. pj should be such that it has a nonzero contribution to the virtual work statement).EQUATIONS OF ANISOTROPIC ELASTICITY 59 equivalent to the governing differential equations as well as the natural boundary conditions.5. which are appropriately selected functions of position x. .2) (c) the set {pj) is linearly independent and complete. we must choose cpj ( j = 1 . I/ denotes a norm in the vector space of functions u. and they are also known as the weak forms. N ) arid cpo such that they meet the following requirements: 1. 2. N ..e. 2 . .e. In Eq.. The basic idea of the Ritz method is described here using the principle of virtual displacements or the minimum total potential energy principle. (1. cpo plays the role of particular solution. Properties of Approximation Functions Substitution of Eq... Given a function u and a real number E > 0.5. the displacement) u of a given problem by a finite linear combination of the form and then determine the parameters cj by requiring that the principle of virtual displacements holds for the approximate solution. The set {cpj) is called the spanning set. (1. The completeness property is defined niatheniatically as follows. The selection of cpj is discussed next.?. In order to ensure that the algebraic equations resulting from the Ritz procedure have a solution. . .g. . cpj ( j = 1 . . 3 .s an integer N (which depends on E ) and scalars el. (1.5. 2 . It should be the lowest order possible for completeness. . minimize l I ( U N ) with respect to cj. for example. is complete if it contains terms of all degrees up to the highest degree ( N ) .. . c. . arid cpo and cpj are the approximation functions. (b) satisfy the horrlogeneous form of the specified essential boundary conditions.1) cj denote undetermined parameters. c~ such that where 11 .
cps should be a cubic polynomial. Since the natural boundary conditions of the problem are included in the variational statements. i. if cpl is a linear polynomial.1). ii = 0. i. This is done by selecting cpi to satisfy the homogeneous form and cpo to satisfy the actual form of the essential boundary conditions. there is no need to include cpo (or equivalently. N Note that when the specified values are zero. therefore. coordinate functions should be selected from the admissible set. As a general rule.f(L)= 0 for j = 1 . from the lowest order to a desirable order without missing any intermediate admissible terms in the xy=. The conditions in Eq.5. . we require The requirement on cpi to satisfy the homogeneous form of the specified essential boundary conditions follows from the approximation adopted in Eq. For example. Thus no function is expressible as a linear combination of others in the set. it follows that cp. and so on (but each cpj need not be complete by itself): The completeness property is essential for the convergence of the Ritz approximation (see Reddy [29]. For instance. if u is specified to be u on the boundary x = L. For polynomial approximations functions. cpo = 0). p. however. we have and.2) provide guidelines for selecting the coordinate functions. (1. the relation holds only for all a j = 0. (1.Linear independence of a set of functions { c p j ) refers to the property that there exists no trivial relation among them.e. .. it follows that ~ ~ (L) = 0. we require the Ritz approximation UN to satisfy only the specified essential boundary conditions of the problem. 2 .e. they do not give any formula for generating the functions. . 262). . cp2 should be a quadratic polynomial. ~ 9 9 Since this condition must hold for any set of parameters cj.. Since UN = ii and cpo = ii at x = L..5. the linear independence and completeness properties require cpj to be increasingly higherorder polynomials. cpj are still required to satisfy the specified (homogeneous) essential boundary conditions.
cj will be different for different values of N . Of course. c2. The function cpo has no other role to play than to satisfy specified (nonhomogeneous) essential boundary p conditions.5) remain unchanged. there are no continuity conditions on c o Therefore. satisfy the completeness property).g. c2. the assumed approximation for the displacements converges to the true solution with an increase in the number of parameters (i.. If the resulting algebraic equations are symmetric. Some general features of the Ritz method based on the principle of virtual displacements are listed below: 1. the previously computed coefficients Aij and bi of the algebraic equations (1.5. Ariq  bi or [A]{c) = {b} where Aij and bi are known coefficients that depend on the problem parameters (e. then the resulting algebraic equations will also be nonlinear in the parameters ci.5.representation of UN(i. 3. The symmetry of the coefficient matrix depends on the variational statement of the problem. One must add only the newly computed coefficients to the system of equations. 291. Equations (1.5. A mathematical proof of such an assertion can be found in [2022. a variety of numerical methods are available (e.. geometry. which will be discussed later in this book (see Chapter 13). (1.. the parameters cj in Eq. .e.5) are then solved for {c) and substituted back into Eq.2). el. Note that H(UN) . . 2.. If the variational (or virtual work) statement is nonlinear in u. (1.. [A]. (1.1) to obtain the Nparameter Ritz solution. and loads) and the approximation functions. . material coefficients. .. If the approximate functions pi are selected to satisfy the conditions in Eq. is now a realvalued function of variables.. the Picard method). These coefficients will be defined for each problem discussed in the sequel. one should select the lowest order p that satisfies the essential boundary conditions.1) are determined by requiring UN to minimize the total potential energy functional n (or satisfy the principle of virtual work) of the problem: SII(UN) = 0. . For increasing values of N . To solve such nonlinear equations. the NewtonRaphson method. provided the previously selected coordinate functions are not changed.g. . as N + co). 4. cN) 0 =  an = aci J=. .e. Newton's method. c ~ Hence minimization of the functional n ( U N ) is reduced to the minimization of a function of several variables: This gives N algebraic equations in the N coefficients (el. Algebraic Equations for the Ritz Parameters Once the functions cpo and pi are selected.5.5. one needs to compute only upper or lower diagonal elements in the coefficient matrix.
For example. < exact).4. cpo = 0.4.67) of Example 1. (1. and p moment Mo.2..5. The applied loads will have no bearing on the selection of c o and 9. Example 1.2.. Next. The equilibrium equations of the problem are satisfied only in the energy sense.(x) is chosen in the form Since the specified essential boundary conditions are homogeneous. For this case the geometric (or essential) boundary conditions are The force (or natural) boundary conditions can be arbitrary. for M > N where UN denotes the Nparameter Ritz approximation of u obtained from the principle of virtual displacements or the principle of minimum total potential energy.. The Ritz method can be applied. concentrated point load Fo.5. we must select cp.1: Consider the cantilever beam shown in Figure 1. must be differentiable as required by the total potential energy functional in Eq.2. Since there are two conditions to satisfy.a form that is equivalent to the governing equations and natural boundary conditions of the problem. which will alter the expression for the coefficients F.5. t o satisfy the homogeneous form of the specified essential boundary conditions dcp cpz(0)= 0 and 2 (0) = 0 dx and p. we begin with cpl = a bx cx2 and + + . the Ritz method can be applied to all structural problems since a virtual work principle exists. the strains and stresses are generally less accurate than the displacement.4. 8... It should be noted that the displacements obtained from the Ritz method based on the total complementary energy (maximum) principle provide the upper bound. the beam can be subjected to uniformly distributed transverse load q(x) = go. Therefore the displacements obtained from the Ritz approximation. Consequently.4. In particular. in general do not satisfy the equations of equilibrium pointwise. The applied loads will enter the analysis through the expression for the external work done [see Eq. not in the differential equation sense. < UN < UM. Since a continuous system is approximated by a finite number of coordinates (or degrees of freedom). in principle. the displacements obtained using the principle of minimum total potential energy by the Ritz method converge to the exact displacements from below: UI < Uz < . as in Figure 1. uo = 0). We consider the pure bending case (i.5).52)].. Since the strains are computed from an approximate displacement field. (1.4. 6. An Nparameter Ritz approximation of the transverse deflection w. We set up the coordinate system such that the origin is a t the fixed end. (1.e. of Eq. to any physical problem that can be cast in a weak form . unless the solution converged to the exact solution. 7. the approximate system is less flexible than the actual system..
we can determine 9 2 . b. 6Il = 0. CN + : FL[(.67) we obtain Il as a function of the coefficients c l . + cNxN+' Substituting Eq.~T~  (1. . The third constant will remain arbitrary.8) into Eq. = ~ ' ' ' ~ The ith eqnation in (1. ... cps. + ( N + I ) c N z ~ ] . (1.5.and twoparameter approximations we have the following equations: I" q(~)s"~dz+ FLL"' + M L ( i + l ) L Z (1.: $ D ~ = L c ~l . We can arbit.7)..5..4.11a) 3=1 where j ( j + l ) z ~ .8) W N = x 2 c 2 z 3 + . . + +..11b) ... since they are already accounted in the preceding p.7) give a = b = 0. and cpl(n:) = c z 2 .. c2. . .10) has the form 0 = = dc. N ) (1. .. f 2 = Z I 3 ( P 3 = Z4. + N ( N + l ) c ~ z ~i(i ' 1 ) ~ ' . (1. Conditions (1.5.5. One may set the coefficients of lower order terms to zero..' .9) Using the total p~terit~ial energy principle. (1.. . C N . = Ebr one. c2. .q xZt1 ~ ] + = c l A t l + c 2 A z 2 + . (PN=z~+' The Ritz approximation becomes (1. etc.5. which requires that rI be a minimum with respect to each of c l .l i ( i + l ) z F 1 d z .5. 2.rarily take c = 1.5. .. + c ~ A I N F= z x~~~~~ N  h " ((i= 1 . an LL { E I [2c1 + Bqr + . we arrive a t the coriditions A I L [ ~ c ~ 3c2z2 z + c2Z3 + + c ~ z ~ + ~ ] ~ .5.deterniirie two of the three constants using Eq. . Using the same procedure.
If we were to choose trigonometric functions for cp.l). For go # 0.rrx/2L]. The threeparameter solution.14) is given by r2 rl is the point x = 0. a proper choice of the coordinate functions is important in realizing the exact solution. u is the dependent variable. (1. would be exact for this problem. a weightedresidual method.3 WeightedResidual Methods Consider an operator equation in the form B l ( u ) = u on rl.5. we may select the functions cp. An specified values on the portions rl and example of Eq. with 4 3 = x4.5. B 2 ( u ) = g on r2 (1.5. because the applied load go. This particular choice would not give the exact solution for a finite value of N.14) where A is a linear or nonlinear differential operator. 1. and Q and g are of the boundary of the domain.. .The exact solution is The twoparameter solution is exact for the case in which go = 0.(x) = 1 . for obvious reason. would involve infinite number of terms. Of course. r2is the point x = L We seek a solution in the form where the parameters cj are determined by requiring the residual of the approximation be orthogonal to N linearly independent set of weight functions I + + : The method based on this procedure is called. B1 and B2 are boundary operators associated with essential and natural boundary conditions of the operator A. the solution is not exact for every x but the maximum deflection W 2 ( L )coincides with the exact value wo(L).cos[(22 . Thus. when expanded in terms of pi. f is a given force term in the domain R. both algebraic and trigonometric functions would yield acceptable results with finite number of terms.
gi can be any linearly independent set.5. Equation (1.16b). such as (1.5.16b) becomes Note that Gij is not symmetric in general. (1.2). the resulting algebraic equations will be nonlinear. (15.5. both essential and natural.17) The variational statement referred to in Property 2a of (1. pi should satisfy homogeneous form of all specified boundary conditions.16b) provides N linearly independent equations for the determination of the parameters ci. It is symmetric when A is a linear operator and Qi = A(cpi) (the leastsquares method). . (1. (1. are not included in Eq.). the operator A is of the form that permits the use of integration by parts to transfer .5. If A is a nonlinear operator. . Properties in (1.5. we have and Eq. (1.17) are required because the boundary conditions.) denotes the Dirac delta function.5.xi) Here S(. and cpi in a weightedresidual method should satisfy the properties in Eq.16b) (with gi # p i ) is known as the PetrovGalerkin method. The weightedresidual method in the general form (1.The most commonly used weight functions are Galerkin's method: Leastsquares method: Collocation method: $i = pi $i = A(cpi) gi= S(x . .5. Whenever A is linear. Various special cases of the weightedresidual method differ from each other due to the choice of the weight function qi. and no continuity requirements are placed on &.5. even when qi = cpi (Galerkin's method).16b).2) is now given in Eq. It should be noted that in most problems of interest in solid mechanics. Both properties now require to be of higher order than those used in the Ritz method. except that they should satisfy all specified boundary conditions: 90 should satisfy all specified boundary conditions. On the other hand.The coordinate function cp. x.
5. Next.20) we have . The Galerkin Method The Galerkin method is a special case of the PetrovGalerkin method in which the coordinate functions and the weighted functions are the same (pi = $i). $i = 2(aRN/aci) [compare Eqs.20) provides N algebraic equations for the constants ci. the Ritz method is most suitable.5. The specified boundary conditions of the problem are all essential type. (1. If the Galerkin method is used in such cases. The leastsquares method is applicable to all types operators A but requires higherorder differentiability of pi. and therefore the requirements on pi in both methods are the same. Therefore. it is possible to construct a weak form of the equation. the weight function $. Equation (1. First we note that the leastsquares method is a special case of the weightedresidual method for the weight function. 2. When the governing equation has even order of highest derivative.5. For problems for which there exists a quadratic functional or a virtual work statement.16b) and (1.16a).5. The problem has both essential and natural boundary conditions.5. but the coordinate functions used in the Galerkin method are also used in the Ritz method.half of the differentiation to the weight functions gi and include natural boundary conditions in the integral statement (see Reddy [6]). becomes Then from Eq. it would involve the use of higherorder coordinate functions and the solution of unsymmetric equations. The Ritz and Galerkin methods yield the same set of algebraic equations for the following two cases: 1. the coordinate functions pi should satisfy the same conditions as in the case of the weightedresidual rncthod. and use the Ritz method. (1. It constitutes a generalization of the Ritz method. (1. if the operator A in the governing equation is linear. LeastSquares Method The leastsquares method is a variational method in which the integral of the square of the residual in the approximation of a given differential equation is minimized with respect to the parameters in the approximation: where R N is the residual defined in Eq.20b)l.
XC(u) = O Timedependent problem At(u) A(u) = f ( ~ .). we require the residual to vanish at a selected number of points xZin the domain: which can be written.or where Note that the coefficient matrix is symmetric.(x) = S ( x .16b) with $.uulue problem A(u. t ) + . one must choose as many collocation points as there are undeterrrlined parameters.14). Note that in the leastsquares rnethod the boundary conditions can also be included in the functional. consider Eq. Eigenvalue and TimeDependent Problems It should be noted that if the problcrn at hand is an eigenvalue problem or a timedependent problem. the collocation method is a special case of the weightedresidual rnethod (1. The leastsquares functional is given by Collocation Method In the collocation method. these points should be distributed uniformly in thc domain.14) takes the following alternative forms: Eigea.x L ) . the operator equatior~in Eq. with the help of the Dirac delta function.5. whereas the Ritz and Galerkin methods yield unsymmetric coefficient matrices. In general.5. The leastsquares rnethod requires higherorder coordinate functions than the Ritz method because the coefficient matrix LtJ involves the same operator as in the original differential equation and no trading of differentiation can be achieved. For firstorder differential equations the leastsquares method yields a symmetric coefficient matrix. For example. (1. as Thus. (1. Otherwise.In the collocation method.5. illconditioned equations among cJ may result.
we have Application of the weightedresidual method to Eqs. i. For additional details and examples.27) with Eq. (1.. the reader may consult [6].In Eq.5. area of cross section.5. (1. The minimum value of P is called the critical buckling load.167. (1.5. The problem involves determining the value of P and mode shape u(x) such that the governing equation and certain end conditions of the beam are satisfied. If the same function is used for cpl in the oneparameter Ritz solution.26) follows the same idea. and A and C are spatial differential operators.5.25) and (1. (1.5.5.26) are provided by the equations governing the axial where u denotes the axial displacement. The lowestorder function that satisfies the two conditions is + The oneparameter Galerkin's solution for the natural frequency can be computed using which gives (for nonzero cl) X = 50112 = 4. In this case. (1. A. An example of the equation is provided by the buckling of a beamcolumn where u denotes the lateral deflection and P is the axial compressive load. Comparing Eq. p the density.5.(l) = 0.25). and f body force per unit length.. must satisfy not only the condition cpl(0) = 0 but also the condition i p : ( l ) cp. E Young's modulus. we note that In Eq. parameter X is called the eigenvalue. Example 1 5 2 . Examples of Eq. which is to be determined along with the eigenvector u(x). . (1.16b) holds.5.e. (1. cp. we obtain the same result as in the oneparameter Galerkin solution.: Consider the eigenvalue problem described by the equations In a weightedresidual method.26) A is a spatial differential operator and At is a temporal differential operator.25). Eq.
This indicates that the leastsquares method with & = A(cp. To determine p2(x).6825.30). Note that .0 arid (d2cp1/d:x2)= 4. p2}. The oneparameter leastsquares approximation with $1 = A ( p l ) gives and X = 4. = A(cp. Let us consider a twoparameter weightedresidual solution to the problem X1. (1.XC(p. If we use y'il = A(pl) .) .116. we obtain whose roots are 25 1 + XI = 7. X2 = 0. and cp2 beconies p2(x) = a bx cx2 dx3 = 4x2 .37a) + + + 011 the other hand. we begin with a polynomial that is one degree higher than that used for p l : and obtain We can arbitrarily pick the values of b and c.01 which gives X = 4. and cpz becomes The set {pl .). we have d = 2.2 =  + a where p l ( z ) is given by Eq. if we choose b = 1 and c = 2.) is perhaps more suitable than $.5. Thus we have infinite number of possibilities. If we pick b = 0 and c = 4.5)= 1. p 2 ) is equivalent to the set {pl . we obtain [cp1(0.6508 (1.5. we have d = 3.5.8.3x3 (1. except that not both are equal to zero (for obvious reasons).Xpl.35) 6 6 Neither root is closer to the exact value of 4.For oneparameter collocation method with the collocation point a t z = 0.5.
we set the integral of the weightedresidual to zero and obtain In matrix form.5. for the choice of functions in Eqs.37a). we have [Kl{c) . we have Evaluating the integrals.37a). (1. (1.4Ml{cl = (01 where First.5. we obtain and . Using pl from (1.5.5.Comparing the two relations we can show that Hence. either set will yield the same final solution for U2(x) or A .30) and (1.30) and p2 from Eq. we compute the residual of the approximation as For the Galerkin method.
the choice of the approximation functions in the finite element method is limited to algebraic polynomials.e. The single most difficult step in all classical variational methods is the selectiori of the coordinate functions. cl # 0 and ca # 0. Recent trend in computational mechanics is to return to traditional variational methods that are meshless and find ways to construct approximation functions for arbitrary domains [31361. and linear constitutive equations of elasticity. we may select z = 113 and x = 213 as the collocation points. The last feature proves to be very helpful in the derivation of higherorder plate theories.6 Summary In this chapter a review of the linear and nonlinear straindisplacement relations. If we were t o use the collocation method. The ideas introduced in connection with classical variational methods are also useful in the study of the finite element method (see Chapter 9). The principle of virtual displacements will be used in this book to derive governing equations of plates according to various theories.479 (1. the principle of miriimuni total potential energy. We leave this as an exercise t o the reader.121.5. 1.. we set the determinant of the coefficient matrix t o zero t o obtain the characteristic polynomial which gives X1 = 4. They also yield the natural boundary conditions and give the form of the essential and natural boundary conditions. The traditional finite element method is discussed in Chapter 9. equations of motion in terms of stresses and displacements. called finite elements. Traditionally. The exact value of the second cigenvalue is 24. among other choices. The selection of coordinate functions becomes more difficult for problems with irregular domains or discontinuous data (i.40) Clearly. loading or geometry).139. The virtual work principles provide a means for the derivation of the governing equations of structural systems. These limitations of the classical variational methods are overcome by the finite element method. as will be shown in the sequel. leastsquares. In the finite element method. an introduction to the principle of virtual displacements and its special case. the domain is represented as an assemblage (called mesh) of subdomains. provided one can write the intcrnal and external virtual work expressions for the system. and the Ritz arid Galerkin methods will be used to determine solutions of simple beam and plate problems. the generation of coefficient matrices for the resulting algebraic equations cannot be automated for a class of problems that differ from each other only in the geometry of the domain. thermoelasticity and electroelasticity is presented. boundary conditions. and collocation methods) is also included in this chapter. Further. or loading. . X2 = 25. A brief but complete introduction to the Ritz method and weightedresidual methods (Galerkin. compatibility conditions on strains. is also presented.For nontrivial solution. that permit the corlstruction of the approximation functions required in Ritz and Galerkin methods. the value of X1 has improved over that computed using the oneparameter approxirriatiori. Also.
. ( C x D ) ] B .4 Derive the following integral identities: i where w.e3) can be expressed using the index notation as where (a) (c) tz3k is the permutatzon symbol.[ B . 9 = A x 9. t ~ .6 k j i . xg) in the deformed body (solid lines) in terms of its coordinates in the undeformed body (broken lines) and compute the nonlinear Lagrangian strains for the body shown in Figure P1. The summation convention on repeated subscripts is used.3 Use the integral theorems to establish the following results: t . ~ .. ( C x D ) ] A (b) ( A x B ) .3b). x2.2 Prove the following vector identities using the summation convention and the (1. = 6. 23 1.E . (b) Show that A V = + A S (see Figure 1. e 2 .6 . C ) (c) ( A x B ) .6 Write the position of an arbitrary point (xl . Prove the following properties of 6.k k over a range of 1 to 3) (dl E .( A . and u are functions of position in R.1 The nine crossproduct (or vector product) relations among the basis ( e l . D ) . C ) ( B . show that (a) (I x A ) .Problems 1. . [(B x C ) x ( C x A)] = [ A . .. (B x C)I2 (d) (AB)T = (B)T(A)T.6 identity (a) The total vector area of a closed surface is zero. 2 k 3 .where A and B are dyads 1.2.t j k z= tjtk = . Figure P1. 1. D ) ( B .2. In the first three identities A . B . and ttJk: =Fzk F&k (b) St. ( C x D ) = ( A .5 If A is an arbitrary vector and 9 is an arbitrary secondorder tensor. . ~ A ~ =~ A0 (e) etjk = c k . and I? is the boundary of 0. C and D denote vectors: (a) ( A x B ) x ( C x D ) = [ A . ~ ~ ~ = k (for z 6.8). 1. I = unit tensor (b) ( 9 x A)T = A x aT 1.&.6.
associated with the displacerrient field where e .10 Consider the uniform deformation of a square of side 2 units initially centered a t X = (0. 2 2 % x3) in the deformed body (solid lines) in terms of its coordinates in the undeformed body (broken lines) and compute the nonlinear Lagrangian strains for the body shown in Figure P1.O). The deformation is given by the mapping (a) Sketch the deformed configuration of the body (b) Compute the components of the deformation gradient tensor F and its inverse (display them in matrix form).7 1.EQUATIONS OF ANISOTROPIC ELASTICITY 73 1. Figure P1.. 1. Figure P1.8 Compute the axial strain in the line element A B and the shear strain at point 0 of the rectangular block shown in Figure P1. a . 1.11 Find the linear strains associated with the 2D displacement field .8 using the engineering definitions.7 Write the position of an arbitrary point ( x l .8 1. .7.9 Compute the nonlinear strain components E. (c) Compute the Green's strain tensor components (display them in matrix form). and b are constants.
and EI are constants.0 . 2 2 ~ 1 = 2 Bx1z2 where A and B are constants.11) and the vector form of the displacement field and the del operator (V) in the cylindrical coordinate system * l a u = ~ ~ ~ ~ + ~ e e Q = +& u + e e e.kj 1.26~~) the cylindrical coordinate system: in 1. 1 ~ 2 = AX:.3k = %. h. 1. .12 Find the linear strains associated with the 2D displacement field (u3 = 0) where co. and ~ i .2. singlevalued displacement field that corresponds to this strain field. cg are constants 1..14 Show that in order to have a valid displacement field corresporidirlg to a given infinitesimal strain tensor E . are the Cartesian components of the strain tensor.15 Consider the Cartesian cornponents of an infinitesimal strain field for an elastic body [a]: ~ 1 = AX.5b) and (1.17 The components of a stress dyadic a at a point. v . . (1. referred to the rectangular Cartesian system .7)] and E. (b) Determine the most general form of the corresponding displacement field with the A and B from Part (a). z) to express the equations of motion (1.13 Use the definition (1.2.3. (c) Determine the specific corresponding displacement field that is fixed at the origin so that u = 0 and V x u = 0 when x = 0. Hints: Begin with V x E and use the requirement '%.. it must satisfy the compatibility relation where t i j k is the permutation symbol [see Eqs. .~ ~ dr r 80 a d to compute the linear straindisplacement relations in the cylindrical coordinate system: 1. . 1.16 Use the del operator (V) and the dyadic form of a in the cylindrical coordinate system ( r . (a) Determine the relation between A and B required for there t o exist a continuous.+~ .74 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS where P.3. e l .
Here the hollow cylinder represents thc matrix arourltl the fiber while the. (c) Determine the work done by the load P. is applied to the rigid core along its centroitial axis.2e2 +B:< passing through the point. Thc outer surface of the hollow cylindcr is assim~rti to be fixed and its inner surface ideally bonded to a rigid circular cylintlrical core of radius b arid length L. (a) Find the axial displacenient 5 of the rigid corc by assuming the following displaccrnt:nt. The problem of pulling a fiher irnbedded in a matrix makrial can be idealized (in the int. Here Qi denote the hasis vectors in (xl.18 1.20 Write expressions for the total virtual work done.191.18. x 3 ) systerri. P1. and length L.or arid thc normal to the plane.EQUATIONS OF ANISOTROPIC ELASTICITY 75 Find the following: (a) The stress uector acting on a plane perpendicular to the vector 2e1 . Suppose that an axial force F = Pe.crcst of gaining qualitative understarlding of the stress distributions at the fibermatrix interface) as one of studying the following problem [8]: consider a hollow circular cylinder with outcr radius a: inner radius b. 6W = 6U beam structures shown in Figs. fiber is idealized as the rigid corc. field in the hollow cylinder: (b) Firid the relationship between the applied load P arid tlisplacernent 5 of t h r rigid core. as shown in Fig.20. + 6V. 2 2 . Figure P1. (c) The magnitudes of the normal and tangential cornporlents of the stress vector.for each of the. Figure P1. (d) Principal stresses. (b) The magnitude of the stress vector and the angle bctween the strcss vect.19 anti P1.19 . P1.
variables. The dependent variables are listed as the arguments of the functional. All other variables are not functions of the dependent.Figure P1.25.21 through 1. wo = 0 .20 Find the EulerLagrange equations and the natural boundary conditions associated with each of the functionals in Problems 1. = an 0 on the boundary I' .
1. Use the operator definition to be A = (d2/dx2) to avoid increasing the degree of the characteristic polynomial for A. Consider a uniform beam fixed at one end and supported by an elastic spring (spring constant k) in the vertical direction.1. Take k = L = 1 and go = 3. k) at the right end.67) to determine a twoparameter Ritz solution of a simply supported beam subjected a transverse point load Po at the center. dx" 0 < x < 1. You may use the symmetry about the center (2 = L / 2 ) of the beam to set up the solution. v . dl. and $ B. respectively.. and use algebraic polynomials. Assume that the beam is loaded by uniformly distributed load qo. Use collocation points x = L/2 and r = L. . z) in a laminated beam can be expressed as where (uo. Find the first two eigenvalues associated with the differential equation  d2u = Xu.iy x . Determine the oneparameter Galerkin solution of the equation that governs a cantilever beam on elastic foundation and subjected t o linearly varying load (from zero at the free end to qo a t the fixed end). (1.5. 4. v .. u(1) + u'(1) = 0 Use the leastsquares method.u = 0 on the boundary using the following Nparameter Galerkin approximation UN = cZi sin i ~ sin j. Use the total potential energy functional in Eq.5. Determine a oneparameter Ritz solution using algebraic functions. 0) along the x and z directions.. Solve the Poisson equation v2u = fo in a unit square. Assume that the beam is subjected to a distributed load q(x) a t the top surface of the beam. u(0) = 0. @. w)along the three coordinate axes (x. wo) denote the displacerrients of a point (x. and 8. Determine a twoparameter collocation solution of the cantilever beam problern in Example 1. y. for the theory.EQUATIONS O F ANISOTROPIC ELASTICITY 77 Suppose that the total displacements (u. denotes the rotation of a transverse normal about the yaxis. are functions of x.4. (b) A beam clamped a t the left end and simply supported a t the right end. Determine a twoparameter Galerkin solution of the cantilever beam problem ill Example 1. Give the approximation functions y q and cpo required in the (i) Ritz and (ii) weightedresidual methods to solve the following problems: (a) A bar fixed a t the left end and conriected to an axial elastic spring (spring constant. Construct the total potential energy functional .
C. The Linearized Theory of Elasticity. G. Z.. M. MA (2002).tin.uou. Introduction to the Mechanics of a Con.78 MECHANICS OF LAMINATED COMPOSITE PLATES AND SHELLS References for Additional Reading Aris. J .. NJ (1962). 40. M. Vectors. London. R. Boca Raton. T .. Research Monograph No. and Hermann..897903 (1915). and Thompson. 7. Oxford University Press. The Netherlands (1978). Kreyszig. A. H. New York. Mechan. Second Edition. R. Carslaw. 1.. I. Sijthoff & Noordhoff. G. and Jaeger. Second Edition. L. L. W. Jones. Acadernic Press. "Berechung der frei galagerten elliptschen Platte auf Biegung. J . Birkhauser.yherAnalysis (translated by C. Energy Principles and Varzational Methods i n Applied Mechanics. New York (1988). M. 6th Edition. Pagano. The Netherlands (1994).ics of Composite Materials. Nowinski. American Elsevier. Hildebrand. B.tional Methods i n Mathematical Physics. Lekhnitskii. S.. G. N.. The M. Lanczos. G. Chapman & Hall.. P. Parton. V. John Wiley. Engineering Mechanics of Composite Structures.. London. The Problem of the Minimum of a Quadratic Functional (translated from the 1952 Russian edition by A. CRC Press. Melbourne. "SeriesSolutions of Some Cases of Equilibrium of Elastic Beams and Plates" (in Russian). Philadelphia.. (1923). Klnwer. Methods of Applied Mathematics. New York (2002). Englewood Cliffs.: Linear Piezoelectric Plate Vibrations. Conduction of Heat i n Solids. Kantorovitch... New York (1952). Noordhoff. Z Agnew..s Medium. Englewood Cliffs. Penfield. Advanced Enyzneering Analys~s. (translated from the 1957 Russian edition by T.. D. B. Second Edition. The Netherlands (1958). S. N. Theory of Thermoelasticity with Applicatiom. Plenum. E. Englewood Cliffs. and Kudryavtsev. I. S. Liquids.. PrenticeHall. London. Cambridge." Math. and Rasrnussen. MA (1967). Reddy. J . UK (1965). and Gases. Theory of Elasticity of an Anisotropic Elastic Body. 1990. reprinted by Krieger.. Electrodynamics of Moving Media. F.. Benster). and the Basic Equations of Fluid Mechanics.. J . Mikhlin. J r . UK (1992). A n Advanced Course of Mathematical Physics. UK (1959). Inshenernov. W. Press. S. Alphen aan den Rijn. Jost. CA (1965). V. F. Tensors. L. New York (1970). Galerkin. S. S. and Krylov. San Francisco. FL.. H.. H. HoldenDay. The Variational Principles of Mechanics. Approximate Methods of H~. . Gandhi. Cambridge University Press. Tiersten. Mech. V. C. B. HoldenDay. G. E. Mechanics of Composite Materials. L. (Ed. Feinstein). Diffusion i n Solids. B. V.. Second Edition.John Wiley.. Taylor & Francis. Smart Materials and Structures. Advanced Engineering Mathematics. Selected Works of Nicholas J. G. Vestn. NJ (1969). The University of Toronto Press. B. FL (1993). John Wiley. S. Reddy. Boston.. CA (1963).). Galerkin. Malvern. New York (1964). A. Reddy. New York (1969). Cartesian Tensors. Jeffreys. 1982. N.. San Francisco. H. Boddington) The MacMillan Company. Toronto (1964). Mikhlin. Slaughter. Mikhlin. Varia. Prentice Hall.. PA (1999). NJ (1965). PrenticeHall.
1 3 5 .$ Engineering. 1 6 1 (1908).: "The Partition of Unity Finite Element Pvletliotl: Basic Theory arid Applications.. 237 262 (1996). Math. Liew. 289314 (1996). "Harmonic Reprod~icing Kernel Particle Method for Free Vibration Analysis of Rotating Cylindrical Shells. Belytschko.EQUATIONS OF ANISOTROPIC ELASTICITY 79 26. Reine Angew. Washizu." Cornputer Methods i n Applied Mechanics an. K. J.. Applzed Functional An(~lysis Hill. L. 34. Y..p Adaptive Method llsirlg Clouds. (to appear). E. "A Hybrid Moving Least Squares and Differential Quadrature (MLSDQ) Meshfree Method.... and Gu. Y.Y. N. New York (1986). Lu." International Journal . T. Variational Methods zn Theoreizcal Mechanics. 1 3 9 . Pergarriori Press. Huang. Q. M.. and Reddy... T. 27. X. and Redcly. and Variational Methods i n Engineering. Mechanics of Elastic Structures. Zhoa. J. K . Q. Ritz. "Ubcr eirle neuc Pvlethotlr zur Losung gewisser Variationsprol~ler~~e der rnathernatischen Physik. M. Licw." J . K. M. "Moving Least Square Differential Quadrature Method and Its Application t o the Analysis of Shear Deforniablc Plates. T. 3(1). 33. Second Edition.. P.. N. G. N. 139. Melbourne. A. Reddy. N. . and Babuska. J .urnal of Corr~p~ut~~tion..al Engzneering Science. arid Oden. Springer Verlag.. Second Edition. Duartc.. New York (1981). 37. A... Huang. M. J . J. Methods i n Applied Mechanics and Engineering. J . J . and Rcddy. 112 (2002). I. N. hZcGraw~ 29. New York (1982).. Y. Liew. reprinted by Krieger. 28. Variational Meth.. Oderi. J.. T." Comprter.for N u m e ~ l c a lMethods i n Engineering. and Reddy." International Jo." Compwter Methods i n Applied Mechanzcs and Engineerrng.. T. 31. Y. J . 36. 229 256 (1994). "An h. 30.for Numerical Methods rn Engineerzng... FL (1992). (to appear). Ng. Berlin (1982).ods i n Elasticity and Plasticity. Melenk. Third Edition. T." International . C. 32. "Ele~rientFree Galerkin Methods. 35. J o ~ ~ r n a l . Zou.. W. arid Ripperger. Hemisphere. Oden. K.
.
weight reduction. on the matrix. for example. and its magnitude decreases from a high value at the end of the fiber to zero at a distance from the end. which are composed of macro size particles of one material in a matrix of another.1. and metallic in metallic.e. nonmetallic in nonmetallic. (2) particulate composites. Whiskers are about 1 to 10 microns (i. and they are handled as a bundle of several thousand fibers. starting from zero value at the end of the fiber to its maximum at a distance from the end. which consist of fibers of one material in a matrix material of another. which are made of layers of different materials. P. there exist four possible combinations: metallic in nonmetallic.la). and protects fibers from being exposed to the environment. When the applied load P on the matrix is tensile. corrosion resistance.Introduction to Composite Materials 2. The tensile stress a in the fiber cross section has the opposite trend.wo or more materials on a macroscopic scale such that they have better engineering properties than the conventional materials.1 Basic Concepts and Terminology 2.005 inches. The stiffness and strength of fibrous composites come from fibers which are stiffer and stronger than the same material in bulk form. acts as a loadtransfer medium between fibers. The basic mechanism of load transfer between the matrix and a fiber can be explained by considering a cylindrical bar of single fiber in a matrix material (see Figure 2. Thus. called matrix material. The two stresses together balance the applied load. The matrix material keeps the fibers together. including composites of the first two types. Shorter fibers.. nonmetallic in metallic. strength. thermal properties. exhibit better strength and stiffness properties than long fibers. metals. Most manmade composite materials are made from two materials: a reinforcement material called fiber and a base material. micro inches or p in.) in diameter and 10 to 100 times as long. The particles and matrix in particulate composites can be either metallic or nonmetallic. and (3) laminated composites.1. The distance from the free end to the . The load transfer between the matrix material and fiber takes place through shear stress. and wear resistance.1 Fibers and Matrix Composite materials are those formed by combining t. fatigue life. Composite materials are commonly formed in three different types: (1) fibrous composites. shear stress r develops on the outer surface of the fiber. Some of the properties that can be improved by forming a composite material are stiffness. Some forms of graphite fibers are 5 to 10 microns in diameter. Matrix materials have their usual bulkform properties whereas fibers have directionally dependent properties. Fibers may be 5 microns to 0. called whiskers.
1: Load transfer and stress distributions in a single fiber embedded in a matrix material and subjected to an axial load.lc).1. The pure tensile state continues along the rest of the fiber. Characteristic distance Springs representing the lateral restraint provided by the matrix material A Broken fiber Figure 2.50.. When a fiber is broken. When a compressive load is applied on the matrix. At this stage. rest of the fiber length. the load carried by the fiber is transferred through shear stress to the neighboring two fibers (see Figure 2.point at which the normal stress attains its maximum and shear stress becomes zero is known as the characteristic distance.lb). the matrix provides a lateral support to reduce the tendency of the fiber to buckle (Figure 2. much like a wire subjected to compressive load. elevating the fiber axial stress level to a value of 1. .1. the fiber tends to buckle.1. in the compressive region. i. the stresses in the region of characteristic length are reversed in sign.e.
2 Laminae and Laminates A lamina or ply is a typical sheet of composite material. and fiber buckling. A poor bonding between a fiber and matrix results in poor transverse properties and failures in the form of fiber pull out. unidirectional fiberreinforced laminae can be stacked so that the fibers in each lamina are oriented in the same or different directions (see Figure 2.inate is a collection of laminae stacked to achieve the desired stiffness and thickness. The layers are usually bonded together with the same matrix material as that in a lamina. A lam. unidirectional. or randomly distributed (see Figure 2.1. It represents a fundamental building block. which can be a metal like aluminum. or a nonmetal like thermoset or thermoplastic polymer. The larnination scheme and material properties of individual lamina provide an added flexibility to designers to tailor the stiffness and strength of the laminate to match the structural stiffness and strength requirements.2). If a laminate has layers with fibers oriented at 30' or 45O. Discontirluous fiberreinforced composites have lower strength and modulus than continuous fiberreinforced composites. For example. The fibers can be continuous or discontinuous. but they have very low strength and modulus in the direction transverse to the fibers. (a) Unidirectional (b) Bidirectional (c) Discontinuous fiber (d) Woven Figure 2. coupling (chemical) agents and fillers are added to improve the bonding between fibers and matrix material and increase toughness. . A fiberreinforced lamina consists of many fibers embedded in a matrix material.1. The sequence of various orientations of a fiberreinforced composite layer in a laminate is termed the lamination scheme or stacking sequence.3). fiber breakage. Often.1.2. it can take shear loads. Unidirectional fiberreinforced laminae exhibit the highest strength and modulus in the direction of the fibers. bidirectional. woven.1.2: Various types of fiberreinforced composite laminae.
damaged fibers.3: A laminate made up of laminae with different fiber orientations. because of the mismatch of material properties between matrix and fiber. material defects such as interlaminar voids. and variation in thickness may be introduced. civil. during manufacturing of laminates. analysis and design methodologies must account for various mechanisms of failure. This book is devoted to the theoretical study of laminated structures. The theoretical concepts and analysis methods presented herein can help structural engineers in aerospace.Figure 2. may cause delamination. delamination.1. and boundary conditions constitutes the major objective of the study. incorrect orientation. therefore. Also. and buckling characteristics of fiberreinforced composite laminates with different lamination schemes. It is impossible to eliminate manufacturing defects altogether. Determination of static. Because of the mismatch of material properties between layers. Laminates made of fiberreinforced composite materials also have disadvantages. and mechanical engineering industries to select suitable materials and the number and orientations of fiberreinforced laminae for the best performance in a particular application. thicknesses. fiber debonding may take place. loads. the shear stresses produced between the layers. . vibration. especially at the edges of a laminate. Similarly. transient.
The generalized Hooke's law for an anisotropic material under isothermal conditions is given in contracted notation [see Eq. The following discussion of constitutive equations is independent of whether the material is homogeneous or not. If fibermatrix debonding and fiber breakage.3.6. and the fibermatrix interactions in a unidirectional lamina are discussed.6) for an orthotropic material. When materials possess one or more planes of material symmetry. From the macroscopic point of view. the number of material parameters is reduced to 9 in threedimensional cases. and Cij are the material coefficients. composite materials are assumed to be homogeneous. viscoelastic. Transformation of stresses. etc.2. all referred to an orthogonal Cartesian coordinate system (x1.e. the number of independent elastic coefficients can be reduced. which is the basic building block of a composite laminate..x3). and for materials with three mutually orthogonal planes of symmetry. treating it as an orthotropic. plastic. In formulating the constitutive equations of a lamina we assume that: (1) a lamina is a continuum. strains. eij ( E ~ )are the strain components. no gaps or empty spaces exist. there are 21 independent elastic constants for the most general hyperelastic material as discussed in detail in Section 1. called monoclinic materials. because the stressstrain relations hold for a typical point in the body. the elastic coefficients of an orthotropic material are expressed in terms of engineering constants of a lamina. (1.In general.g.b)]by where aij (ai) are the stress components.x2.2 Constitutive Equations of a Lamina 2.1 Generalized Hooke's Law In this section we study the mechanical behavior of a typical fiberreinforced composite lamina. 2. i.In the remaining portion of this chapter. for example.. The generalized Hooke's law is revisited (see Section 1. and elasticity coefficients from the lamina material coordinates to the problem coordinates are also presented. we study the mechanical behavior of a single lamina. called orthotropic materials.) behavior of a lamina. then we must consider the micromechanics approach. The first assumption amounts to considering the macromechanical behavior of a lamina. which treats the constituent materials as continua and accounts for the mechanical behavior of the constituents and possibly their interactions. (2) a lamina behaves as a linear elastic material.3. fiber and matrix.3. It should be noted that both assumptions can be removed if we were to develop micromechanical constitutive models for inelastic (e. linear elastic continuum. wherein the material properties of a composite are derived from a weighted average of the constituent materials.37a. For materials with one plane of material symmetry. are to be included in the formulation of the constitutive equations of a lamina. there are only 13 independent parameters. The second assumption implies that the generalized Hooke's law is valid. . Composite materials are inherently heterogeneous from the microscopic point of view.
Fibers are parallel. The applied loads are either parallel or perpendicular to the fiber direction. urn = Poisson's ratio of the matrix v = matrix volume fraction . the x2axis transverse t o the fiber direction in the plane of the lamina. 5. The matrix is free of voids or microcracks and initially in a stressfree state. 2. called a micromechanics approach.2 Characterization of a Unidirectional Lamina A unidirectional fiberreinforced lamina is treated as an orthotropic material whose material symmetry planes are parallel and transverse to the fiber direction. fiber direction). Em = modulus of the matrix uf = Poisson's ratio of the fiber.1 and 2.2. The material coordinate axis xl is taken to be parallel to the fiber. and uniformly distributed throughout. Then it can be shown (see Problems 2.2. 3. The moduli and Poisson's ratio of a fiberreinforced material can be expressed in terms of the moduli. Perfect bonding exists between fibers and matrix.1: A unidirectional fiberreinforced composite layer with the material coordinate system (xl. To this end.2. used to determine the engineering constants of a continuous fiberreinforced composite material is based on the following assumptions: 1. x3) (with the xlaxis oriented along the x2. Poisson's ratios. The theoretical approach. The orthotropic material properties of a lamina are obtained either by the theoretical approach or through suitable laboratory tests. uf = fiber volume fraction. let E f = modulus of the fiber.2.2) that the lamina engineering constants are given by Figure 2. and the xsaxis is perpendicular t o the plane of the lamina (see Figure 2.1). Both fibers and matrix are isotropic and obey Hooke's law. 4. and volume fractions of the constituents. .
The specimen is then loaded along the longitudinal direction and strains along and perpendicular to the fiber directions are measured using strain gauges (see Figure 2. G23.2ad. and U23 of an orthotropic material can be determined experimentally using an appropriate test specimen made up of the material. E2. ul2 is the major Poisson's ratio. Gla.2a. The engineering parameters El. These are shown schematically in Figure 2.2. Interested readers may consult Chapter 3 of Jones [3] and the references given there (also see [IS201). For example.E3. and GIP is the shear modulus. as opposed to mechanics of materials approaches. The specimen consists of several layers of the material with fibers in each layer being aligned with the longitudinal direction.2: Tests required for the mechanical characterization of a laminate. E2 is transverse modulus. U13.2. . the .E3 and GI2 and the longitudinal strength X. At least four tests are required to determine the four constants El. longitudinal strain Ee = ~1 and transverse strain ~t = ~ 2 we can calculate q  p x l (a) Figure 2. and Other nlicrorriechanics approaches use elasticity. G13. E2. ul2 and X of a fiberreinforced material are measured using a uniaxial test shown in Figure 2. By measuring the applied load P. u12. the crosssectional area A.2. E l . transverse strength Y and shear strength S (and additional tests to determine Gl3 and G2:3).where El is the longitudinal modulus.2.2e).
Ep (AS) = graphiteepoxy (AS13501).Ep *Moduli are in msi = million psi.Ep (T) = graphiteepoxy (T3001934)./OF.Ep (1) GI. and using the transformation equation (4a) of Problem Et. GPa = lo9 Pa.Ep = boronepoxy.1 and 2. kPa = lo3 Pa. Material Aluminum Copper Steel Gr. The shear strength S is determined from the test shown in Figure 2. 2 ~ measuring by El = P / A E ~ . GI. MPa = lo6 Pa.2d: m 27rr2h where T is the applied torque.894./in. u2l and Y can be determined from the test shown in Figure The shear modulus is determined from the test shown in Figure 2 .88 MECHANICS O F LAMINATED COMPOSITE PLATES A N D SHELLS where Pult the ultimate load (say. ~aterialt Aluminum Copper Steel Gr.Ep = glassepoxy. S 1 ult = Tult =  Table 2. t The following abbreviations are used for various material systems: Gr.2.2: Values of additional engineering constants for the materials listed in Table 2. Similarly. and r and h are the mean radius and thickness of the tube.2. and the units of cul and a2 are 10P6 in. Table 2.Ep (T) GI. 3 2.1".2.Ep (2) Br.2. E2. P a = N/m2.1: Values of the engineering constants for several materials*. load a t which the material reaches its elastic is limit).Ep (1) G1. wherein Get is the only unknown.76 N/m2. 1 psi = 6. Gr.Ep (AS) Gr.2.2.Ep (2) Br. 2 . respectively.Ep (AS) Gr. Et and vet. The values of the engineering constants for several materials are presented in Tables 2. .Ep (T) GI.2. Br.Ep E3 v13 v23 Ql a2 * Units of E3 are msi.
the ~ 1 x 2 plane and the xyplane are parallel) and the XIaxis is oriented at an angle of +O counterclockwise (when looking down on the lamina) from the xaxis (see Figure 2.I.3 Transformation of Stresses and Strains 2.3. In forming flat laminates. there is a need to establish transformation relations among stresses and strains in one coordinate system to the corresponding quantities in another coordinate system. each with different orientation of their material coordinates with respect to the laminate coordinates.1) is cos0 sin 0  sin0 0 cos 0 O] {ii} {ii} = (2.1 Coordinate Transformations The constitutive relations (1. Thus. The coordinate system used in the problem formulation.3. Thus we have a special type of coordinate transformation between the material coordinates and the coordinates used in the problem description..45) for an orthotropic material were written in terms of the stress and strain components that are referred to a coordinate system that coincides with the principal material coordinate system. z ) denote the coordinate system used to write the governing equations of a laminate.I e3 .3. y. The transformation relations (2.3.1) and (2. fiberreinforced laminae are stacked with their ~ 1 x 2 planes parallel but each having its own fiber direction. (2. the xscoordinate of each lamina we will always coincide with the zcoordinate of the problem. x3) be the principal material coordinates of a typical layer in the laminate such that xaaxis is parallel to the zaxis (i.2) are also valid for the unit vectors associated with the two coordinate systems: {:I}{!.44) and (1.1). These relations can be used to transform constitutive equations from the material coordinates of each layer to the coordinates used in the problem description.2.3. {.I = [L] = [LIT e3 ez ez {:.3. does not coincide with the principal material coordinate system. and let (XI. composite laminates have several layers. Let (x.3.3. The coordinates of a material point in the two coordinate systems are related as follows (z = xs): The inverse of Eq. Further.e.2) [L]' Note that the inverse of [L] is equal to its transpose: [L]' = [LIT. in general. If the zcoordinate of the problem is taken along the laminate thickness.x2.
3.1: A lamina with material and problem coordinate systems.y. are the orthonormal basis vectors in the material and problem coordinate systems.4) can be expressed in matrix forms. a.4) hold among tensor components only. x3). in the problem (p) coordinates (x. a33 in the material (m) coordinates (xl.x2. First. .x3) coordinate systems. and (ei). Since stress tensor is a secondorder tensor. we introduce the 3 x 3 arrays of the stress components in the two coordinate systems: Then Eqs. (2. x3) and components a. which has components all. . Note that the tensor transformation equations (2.. y. z ) and (x1.3.3. Let a denote the stress tensor. 012.3. 2 ) . 2. . whereas (aij). are the components of the same stress tensor a in the problem coordinates (x.4) can be expressed in matrix form as where [L] is the 3 x 3 matrix of direction cosines ti? . . a. z ) . . x2. . Equations (2. . respectively. and eij are the direction cosines defined by and (ei).. y..2 Transformation of Stress Components Next we consider the relationship between the components of stress in (x.Figure 2. x2. y.. it transforms according to the formula are the components of the stress tensor a in the material coordinates where (xl.3.
3.sin2 8 sin 0 cos 0 0 0 0 or { ~ ) = p [Tl{a)m The inverse relationship between {a).e.he stresses act..ing on the face with the area of the surface. a.Equation (2.iplying t. y.. Suppose that the thickness of t.2 ~ . (2.: a. Suppose that we wish to determine 0 2 2 in terms of (a. (2.9) can also be obtained from Eq.6a.3.3. Equations (2. and hence it holds for the special transfornlation in Eqs.).6b).3. and {a).sir18 cos 8  sin2 8 cos2 0 0 0 0  0 0 1 0 0 0 0 0 cos8 . is given by 0 0 0 0 1 0 0 0 cos8 0 0 0 sin0 0 . z ) and (xl. with [L] defined by Eq. Example 2. Then by surr~rrlirlg forces all acting on the wedge along coordinate 2 2 (i. we obtain cos2 8 sin2 0 0 0 0 .. respectively.. Consider a wedge elernrmt whosc slant face is parallel to the fibers.1). (2.sin2 8 The result in Eq. Carrying out the matrix multiplications in Eq.I ) .3. (2.6a). Eq.he lamina is h.b) hold for any general coordinate transformation.3. (2.613) allows computation of stress components referred to the problem coordinates in terms of stress cornponents referred to the material coordinates. and rearranging the equations in terms of the singlesubscript stress components in (x.. equilibrium of forces along x2) we obtain a 2 2 = 'T .3..sin 28 0 0 .2). 2 3 ) coordinate systems..cos. The forces acting on any face of the wedge are obtained by rnult. sin2 H + a. Then the horizontal and vertical sides of the wedges are of lengths AScosB and ASsiriB.1: The stress transfornlation equations (2. cos2 B .3. (2..sin 8 cos 6' sin Qcos0 0 0 sin2 0 cos2 0 0 cos2 8 sin2 8 0 0 sin 28 0 .3.B sin B ~ .3.. 2 2 .3.3. and the length of the slant face is AS.9) can he derived directly by considering the equilibrium of ari element of the larnina (see Figure 2.6a) provides a means to convert stress components referred to the problem (laminate) coordinate system to those referred to the material (lamina) coordinate system. while Eq.7) by replacing 8 with 0. (2.sin8 0 cos0 0 0 cos2 8 ..sin8 0 0 0 sin0 cos0  sin 28 sin 20 0 0 0 cos2 8 .
and ulz = 0.. the thickness is about onetenth of the radius). .3.. E2 = 10 MPa (1. a. respectively.(a.. summing the forces along x l coordinate. in Eq.) internal diameter and pressurized to 1.3. We wish to determine the shear and normal forces per unit length of filament winding. (cos2 0 . (2. in terms of ("113 0 2 2 .and ycoordinates we can obtain stresses a.3.3. The vessel is of 63.sin2 0) Clearly. ~ 1 2 ) . and use the following material properties. Assume a filament winding angle of 0 = 53.AS cos 0 h) sin 0 (a.AS  sin 0 h) cos 0 + (a.45 Msi). we obtain u12AS h + (a.02 Msi).AS sin 0 h) sin 0 .. a.2: Consider a thin (i..3 Msi).379 MPa (200 psi).Figure 2.2)..894.9).125" from the longitudinal axis of the pressure vessel.. A S cos 0 h) cos 0 = 0 or 012 = a. typical of graphiteepoxy material: El = 140 MPa (20. Note that MPa means mega (lo6) Pascal (Pa) and Pa = N/m2 (1 psi = 6.2: A freebody diagram of a wedge element with stress components.) by considering a wedge element whose slant face is perpendicular to the fibers (see Figure 2. By summing forces along the x..3).5 cm (25 in.... the expressions for 0 2 2 and 012 derived here are the same as those for a1 and as. and a. filamentwound. Similarly. sin 0 cos 0 + a.e.3. GI2 = 7 MPa (1. closed cylindrical pressure vessel (see Figure 2. cos 0 sin 0 + a.76 Pa).. The stress component a11 can be determined in terms of (a. Example 2.3....
transformation equations derived for stresses. The equations of equilibrium of forces in a structure do not depend on the material properties.8 = h Thus the normal and shear forces per unit length along the fibermatrix interface are F22 = 0. (2..9) or from the equations derived in Example 2.0'2189 MPa .INTRODUCTION T O COMPOSITE MATERIALS 93 Figure 2.8)2 = 0. 6 ) 2 = 0.3.4378(0.3.379 x 0.4378 h 0.3: A filamentwound cylindrical pressure vessel. 2. Hence.0. Next we determine the shear stress along the fiber and the normal stress in the fiber using the transformation equations (2. We obtain 1.6a.2189 h MPa h 0.3590 h h 0'2189 (0.1.3.3. is zero.43'i8( 0 .1051 MPa x 0. whereas the force per unit length in the fiber direction is Fll = 0. We obtain 0. and h is thickness of the pressure vessel.) stresses in a thinwalled cylindrical pressure vessel are valid here: where p is internal pressure.3 Transformation of Strain Components Since strains are also secondorder tensor quantities.1051 MN.635 . = 1.2189 (0... D iis internal diameter.6)2 + .2977 022 = 0..379 x 0.) and circumferential (cr.3.2977 MN and F I 2 = 0.3.b).359 MN.711 = h  MPa h 0.635 . Eqs. are also valid for tensor components of strains: [~lm = [A][EIP [LIT (2.11a) .4378 gzz = 4h h 2h h The shear stress rr.6 x 0. equations derived for the longitudinal (u.8)2+ 0. cr.
longitudinal strain in the fiber is 1 1 = E ~ = 0 0 2cry .9) are valid for strains when the stress components are replaced with tensor components of strains from the two coordinate systems. (2. and vlz = 0.3. (2. The fibers are oriented a t 45O to the horizontal.sin 20 0 sin28 0 sin0 o o 0 cos2 o .3: A square lamina of thickness h and planar dimension a is made of glassepoxy material (El = 40 x lo3 MPa.3. = 0.7) and (2.3.14).01. We have cos2 0 sin2 0 0 0 sin 20 0 0 1 0 0 0 . (2.sin2 0  [i} (2. (2.7) and (2.= 0.sin0 0 0 0 0 sin0 cos 0 0 sinQcosQsin 19cos 0 0 0 0 cos2 0 . (2.l4). However.3. From Eq.4.13) EXX E~~ The inverse relation is given by cos2 0 sin2 o sin2 0 cos2 o 0 o o 0 o 0 o cos0  sin 0 cos I9 sin0 cos 0 11 1 J IE:: I 0 . Hence.10): Example 2. (2. the transformation matrix in Eq. (2. we wish to determine the longitudinal strain in the fiber and shear strain a t the center of the lamina.3.sin2 0 We note that the transformation matrix [TI in Eq. Similarly.01 cm/cm I + + JZJZ and the shear strain is given by .94 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS [~l= p [LIT [&I m [L] (2.3.3.3.sin20 0 sin2 0 cos2 0 0 0 0 0 cos8 . the only nonzero strain is E. E2 = 10 x lo3 MPa.3. the singlecolumn formats in Eqs.9) will yield the proper relations for the engineering components of strains. GI2 = 3.11b) Therefore.9) for stresses are not valid for singlecolumn formats of strains because of the definition: Slight modification of the results in Eqs.3. (2..3.7) and (2. When the lamina is deformed as shown in Figure 2.5 x lo3 MPa.3.13) is the transpose of the matrix [R] in Eq. Eqs.8) is the transpose of the square matrix in Eq.25).3.3.3.
Then the stress field in the material coordinates becomes u l l = 17. ulz = 5. (2. uz2 = 14. We have ( U ~ ~ / lE /~ 1 ) =v 2 E The strains in the (z.2 is h = 2 cm.4: Suppose that the thickness of the cylindrical pressure vessel of Example 2.3.13): .47).95 MPa.4: Deformation of a fiberreinforced lamina.3.255 MPa The strains in the material coordinates can be calculated using the strainstress relations (1. y) coordinates can he computed using Eq.INTRODUCTION T O COMPOSITE MATERIALS 95 Figure 2.3. Example 2.3.885 MPa.3.
Carrying out the matrix multiplications in (2. (2.3. with all their variables and coefficients. for orthotropic as well as anisotropic).17) for the general anisotropic case. [TI is the matrix based on the particular transformation (2. (1.17) is valid for general constitutive matrix [C] (i. the same result can be obtained by using the stressstrain and strainstress relations (l.b). .4 Transformation of Material Coefficients In formulating the problem of a laminated structure. in the problem coordinates. in the material coordinates by the tensor transformation law Cuke = aimajnakpaeqCmnpq However.38a. and the stress and strain transformation equations in (2.2 ~ 3 cos sin o c23 o cos2 6 sin2 C16= C16cos4 o (cll c12 2 ~ 6 6cos3 0 sine 3(cz6.e.3.Cz6)C O S ~ sine + ( ~ 1 1 + CO Q ~ .C16)cos2 sin2 o ) (2cs6 c12 ~ 2 2 cos o sin3 o .3.3.3. stresses. we obtain Cll = cI1O S ~ . Thus the transformed material stiffness matrix is given by ([C] = [CIp and [C] = [C]. and strains.35)] are the components of a fourthorder tensor.3.4C26cos o sin3 o + cz2 0 sin4 sin2 o 2 2 C16)cos B sin3 c12 o sin4 C13= c13 o .2.8) and (2. In the previous section we discussed transformation of coordinates (which are also valid for displacements and forces).3. The material stiffnesses Cij in their original form [see Eq.) Equation (2.3. the above equation involves five matrix multiplications with foursubscript material coefficients.3. Hence. the tensor transformation law holds. (1. Alternatively. is the 6 x 6 material stiffness matrix [see Eq. The only remaining quantities that need to be transformed from the material coordinate system to the problem coordinates are the material stiffnesses Cij and thermal coefficients of expansion a i j . Of course.4C16 c0s3 o sin e + 2(c12 + 2 ~ 6 6cos2 C e ) C12 = c12 S ~ + 2(c16 .Cz6sin4 o ) Cz2= cz2O S ~ 4C26 cos3 B sin 2(c12 2cs6) cos2o sin2 o C e  + 2(cz6 + + + + + 4C66) cos2o sin2 o + + 4Cls cos 8 sin3 8 + Cll sin4 8 = C23cos2 6 + 2C36 cos 0 sin 6 + C13 sin2 8 + + + .1) (rotation about a transverse normal to the lamina).38a)l in the material coordinates and [TI is the transformation matrix defined in Eq.3.. we must write the governing equations. The fourthorder elasticity tensor components Cijke in the problem coordinates can be related to the components C.15): where [C].8).
C46)em2 0 sin B B Cq6sin3 0 + + (C24  When [C] is the matrix corresponding to an orthotropic material. C35.2 ~ 1 2 2CG6) Q sin2 Q 2 cos2 ( .C26) cos3 Q sin Q + ( c ~ I +~ 2 .16) has the explicit form [cf.in the material coordinates by ([SIP [s] and [S].C24) cos2 4 o sin o + ( ~ 1 5  C2.INTRODUCTION TO COMPOSITE MATERIALS 97 626 = + (cI1 c12 2666) cos Q sin3 Q .3. (1. (2. r [S])  [RIT{&)m= [RIT ([SIm{a)m) = [S]pb)p [ ~ I ~ l (IRI{c)p) ~ l m (2. and C56 are zero for an orthotropic material.C56)cos sin2 o (2.2cd5cos H sin 0 5 sin2 +2 + + C14 = C14em3 o + (c152 ~ 4 6cos2 o sin 0 + (C24 2 ) ~cos 0 sin2 8 + ~ 2 sin3 8 ~ ~ ) 5 C15= CI5cos3 Q (C14 + 2 ~cos2 Q sin~ + (C25+ 2C46)cos Q sin2 8 ~ 2 sin" 8 ~ 8 ) 4 C 4 = C2* cos3 Q + (C25 + 2C46)cos2 Q sin Q + (C14 + 2CS6)cos 8 sin2 8 + cis sin3 0 2 c 2 5 = C25c0s3 e + (2c56 C24)cos2 8 sin o + (C15 2C46) cos o sin2 o c14 8 sin3 C34 = C34 cos 0 + C35 sin 8  + + 635 = C35 cos 0 . then Eq.CI6)cos Q sin3 Q + C@(COS~ sin4 Q) ~ ~ ~ Q+ C44 = ~ 4 cos2 B 4 ~ 5 sin2 8 5 2 ~ 4 cos o sin 8 5 C45 = c45(cos20 . C45.3. C15.3. we use the strain transformation equation in Eq.3. z ) coordinate system. Eq.44).20a)  [Sl = [RlT[s1 [R] (2.3.C25 . c36. Note that C14.15): {&Ip = Thus the compliance coefficients Sijreferred to the (x.3.C56 sin3 8 + + +~ 1 . (724. C16.(726)cos2o sin2 o 2 )  C33 = C33 C36 = (C13 .sin2 8) (CS5. which are related to the elastic coefficients in the material coordinates Cij by Eq.C34 sin Q C6 = 4 C46 cos3 0 (C56 . it has the form shown in Eq.3. z ) system are related to the compliance coefficients Sil. C46. y. (2. y.C4*)cos Q sin 0 C55 = ~ 5 cos2 0 c~~ 0 .sin2 8) C(js = 2(c16 . C26. (2. (1.5 .42) for monoclinic materials] where the Cij are the transformed elastic coefficients referred to the (x. C34. (225.3.18) C56 = C56 C O S ~ (CIS . In order to relate compliance coefficients in the two coordinate systems.C23)cos Q sin Q + C36(cos28 .20b) .cis sin4 0  C26cos4 o + (c12 ~ 2 + 2 ~ 6 6cos3 Q sin 0 + 3(c16.C46)cos 0 sin2 o C14 .18).
(2..912 2 ~ 2 2 S(j6) + cos3 o sin 8 + 3(s16 cos o sin3 o S16sin4 o + ( 2 s l 2sI2  + ( 2 ~ 1 2 S(j6) o sin2o + cos2  ~ 2 6 cos2 8 sin2 Q ) s 3 3 = 5'33 s3(j = S66 = + 4(Sll + S 2 2 .2S12) cos20 sin20 cos 0 sin 8 + S55sin2 8 Sq4 S 4 4 cos2 8 + = = S ~ ~ ( C O S ~ sin2 0) + (S55 .. and the strainstress relations in the problem coordinates are given by .3.912 + $33) 2S26cos 8 sin3 0 Sz2 0 sin4 Slz = .914 sin3 6 5'34 = S34 cos 6 + S3.3.2s25 .21) For an orthotropic material.912 cos4 8 (S16 S2(j) o sin 8 cos3  cos2 8 sin2 o  + ~ 1 sin4 0 2 SI3= S13 e .SS6 o c0s3 sin3 + + S5(j = (2s15 .sin2 0) cos ~ ~ ~ (sin2 Q )0 ~ 4(S16 .S34 sin 0 Sq6 ( 2 ~ 1 4 2 ~ 2 4 SS6) 0 sin o + (2S15 .Ss6) cos3 8 sin 8 3 ( ~ 2 6 S16) cos2 Q sin2 o cos o sin3 8 .S 4 4 ) cos 8 sin 8 0 S55= s 5 5 cos2 o + sqq o sin2 2s45 cos o sin 6 SI4= SI4 cos3 0 + (S15 S46) cos2 6 sin 8 + (S24 $56) cos 8 sin2 8 + S2. (1.5 cos3 B + (S24 + S56) cos2 o sin o + (S15 = cos 8 sin2 o .S4(j) o sin 8 cos2 SS6 cos3 o s4(j 8 sin3 + ( 2 ~ 2 4 2s14   SS6) 8 sin cos 2 o (2. the compliance matrix [S]has the form shown in Eq.20b) is Sll = sllC O S ~o . sin3 8 SI5= SI5c0s3 e (sI4 S56) cos2 o sin o + (S25 S46) cos o sin2 o .&(j)  2(sI3 .936 cos o sin 8 + S23sin2 8 cos2  + (S26 + + + + (sll+ .~ 2 6 (cos2 o .922 + S66) cos2 8 sin2 o SI6) o sin3 o cos  S16= S16 o + (2sll c0s4 (S(j(j + 2s12 o+ 2s12 .sin2 8) cos 8sin o 0 ~ s ~ ) + + S35 = 5'35 cos 0 .S d 6 ) cos o sin2 8 = + cos2 + Sd6 B .3.S 2 4 sin3 o + + S 2 4 = S24 cos3 e + (S25 + S4(j) cos2 8 sin o + (S14 S5(j) o sin2 o + s15 o + cos sin3 s~~S2.45). sin 0 .S23) 8 sin o s36(cos2o .2s16 cos3 o sin 8 + (2.2 ~ 2 .S2(j 8 sin4 sin o S 2= ~ 2 cos4 2 2 cos" s 2 3 = S23 cos2  + 2Sls cos 8 sin3 0 + SI1sin4 8 o + S3(j B sin o + sI3 o cos sin2 = S26 cos4 o + (2.Expanded form of the relations in Eq.
and a2:3 (see Figure 2. 2 3 ) which is oriented at an angle 0 (in the xyplane) from the (2.x2.3. a 2 2 = a 2 .44) represents the stressstrain relations in the principal rnaterial coordinates (xl. (2. In the context of the present study. . the transverse shear stress components are not neglected in shear deformation theories. Then the constitutive equations must be modified to account for this fact. = all sin 0 a 2 2 cos2 0 2aTY 2 (all. and { a T )and {awr) are vectors of thermal and hygroscopic coefficients of expansion. and 012. azz= a 3 3 The same transformations hold for the coefficients of hygroscopic expansion. z ) coordinate system (see Figure 2.3. (2. However. only nonzero components of thermal expansion tensor are all ail.1).22) relates stresses to strains in the problem coordinates while Eq.x3).3..l8). The transformation relations (2.y. (2.18).3.aZP) 0 cos 0 = sin 2asz = 0. In summary. (1.3.3.21). the hygrotherrrioelastic stressstrain relations in the laminate coordinate system can be written as + + (1~22sin2 0 where all quantities are referred to the (2. (1.19) represents the stressstrain relations in the (2.1). 2ayz = 0. = a 1 cos2 0 1 2 a. and therefore they transform like the strain corrlporlerlts (because a s = 2alz. respectively. All other components are zero. z) coordinate system. xa. a d = a i 2 j = 0)  ax. The orientation angle 0 is measured counterclockwise from the xaxis to the xlaxis.y.4 Plane Stress Constitutive Relations Most laminates are typically thin and experience a plane state of stress (see Section 1. a l s . 2. and so on). following Eq.INTRODUCTION T O C O M P O S I T E MATERIALS 99 Note that Eq. for the kth layer of a laminate. (2. y. they can induce failures because fiberreinforced composite laminates are weak in the transverse direction (because the strength providing fibers are in the xlzaplane). are the components of a secondorder tensor.3. and Eq. y. z ) coordinate system are related to rnaterial coefficients in the material coortliriates by Eq. a s = a13= 0.23) are valid for a rectangular coordinate system (21.3. Eq.45) relates the stresses to strains in the material coordinates. and (2. Although these stress components are small in comparison to all. In general..6). we can write the transformation relations (as = a 1 2 = 0. (2. in most equivalentsingle layer theories the transverse normal stress a33 is neglected. Hence.3. The thermal coefficients cw. a i d C Q ~ = ag. For this reason. z) coordinate system.3.3. a 2 2 . The material coefficients of the lamina in the ( T . For a lamina in the zlzzplane.7).4. the transverse stress components are a s s .
eij ) are the piezoelectric moduli. a1 and a 2 are the coefficients of thermal expansion along the X I and 2 2 directions.4. (gi. x3). Di) are the stress. referred to the material coordinate system (xl.1: A lamina in a plane state of stress.ei. xz. Ei. . electric field. and electric displacement components.100 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS Figure 2. The condition 033 = 0 results in the following therrnoelastic constitutive equations for the kth layer that is characterized as an orthotropic lamina with piezoelectric effect: (k) (k where Qij are the plane stressreduced stzffnesses. respectively. and cij are the dielectric constants of the kth lamina in its material coordinate system. strain.
.4. (1.2Qss) sin3 0 cos 0 + (Ql2 Q22 + 2Q6s) sin0 cos3 0 Q I I = Q11 cos4 Q22 = QII (Qii Q12 Q12   (Q11  Q44 Qss = (Q1i Q22 .3.To. are a.413) Note that the reduced stiffnesses involve six independent engineering constants: El. G12.9) .72) that Qij ) are related to the engineering constants as follows: Qss .8) and axY the transformed thermal coefficients of expansion [see Eq. aZy (al . sin2 0 +  + + Qs6(sin48 + cos4 0) Q45 = (Q55 Q44) cos 0 sin 0 Q55 = Qs5 cos2 8 + Q44 sin20 2 (2.3. The transformed stressstrain relations of an orthotropic lamina in a plane state of stress are (the superscript k is omitted in the interest of brevity) where $ denotes the scalar electric potential [see Eq.4..INTRODUCTION TO COMPOSITE MATERIALS 101 respectively. (2. ayy.2Q12 .4Qm) sin2 0 cos2 0 + sin^ 8 + cos4 8) sin4 0 + 2(Q12 + 2Qss) sin2 0 cos2 0 + Q22 cos4 0 Qls = 2Qss) sin 0 cos3 0 + (Q12 Q22 + 2Qs6) sin3 0 cos 0 &as = .a2) 0 cos 0 = sin (2.3. ayY a1 sin 0 = + a 2 cos2 0.2Qss) sin2 0 cos2 0 = Q44 cos2 0 Q5.. and G23.89 + 2(Q12 + 2Qt3j) sin20 cos2 0 + QZ2sin4 0 Qia = (QH + Qzz .4. Recall from Eq.G12 (k)  (k) ? (k) Q44  G23 (k) ? (k) Q55  (713 (k) (2.23)] a. G13. and A T is the temperature increment from a reference state. (1. = n l cos 0 2 + a z sin20. AT = (k T . E 2 r V12.
4. tXy.365 0.esz) sin 0 cos 8. G23 = 0. E3 = 0.355 0 0 1. E2 = 4.696 0 0 7.4.8) for various values of 0 as The transformed coefficients for various angles of orientation are given below: rnsi 8.365 0 0 0 0.4.0175 0 O 7.l cos B el5 sin 2 8.1: The material properties of graphite fabriccarbon matrix layers are (see Example 1.835 0.4.390 15. = €11 cos2 8 €22 sin2 0 eyy = €11 sin2 0 €22 cos2 0.3.203 8.4. and transformed dielectric coefficients t.007 4. tXy (ell .. Example 2.E ~ sin 0) cos 0 = ~ + + + e33 = e33 + + + (2.1 sin2 0 G25 = (e15 .4.75 x lo6 psi lo6 psi.007 1.785 0 0 0.10) This completes the development of constitutive relations for an orthotropic lamina in a plane state of stress.4) and (2.076 [QIs=45 = 0 0 0.1 x lo6 psi.8 x G12= 1. e 3 = (esl . e15= el5 cos26' e2.203 0 0 5.2 x lo6 psi.36 x lo6 psi.and try are Esl = egl cos2 0 e32 sin 2 8.51 4.14) I (2.923 6. E32 = e y l sin2 # e32 cos20.076 5. E. el4 = (el5 . G13 = 1.3161 [Q]B=~o = 0 0 0.102 MECHANICS OF LAMINATED COMPOSITE PLATES AND SHELLS and Cij are the transformed piezoelectric moduli.076 0 0 7.835 0 5. (2.15) .4): El = 25.3161 1.785 0 0 5.6525 0.47 x lo6 psi The matrix of plane stressreduced elastic coefficients for the material can be calculated using Eqs.883 I msi (2.ea4)sin 8 cos 0 ~ 2 E24 = e2.923 0 0 5.076 6..e24) sin 0 cos 0.696 5.
2 Consider the composite lamina of Problem 2. (2. P2.1 Consider the composite lamina subjected to axial stress u1. v Let Ef. can be evaluated using the engineering constants in the rnaterial coordiiiat.3.2 2. as shown in Fig. are the transformed compliances defined in Eq.47) in the material coordinates. we can write strainstress relations in the problem coordinates as Comparing Eq.3 (Apparent moduli of an orthotropic material) Note that the transformed material corripliance matrix [S]is relatively full and is in the same form as that for a nlonoclinic material. Derive the result 2. we note that and so on. Thus.Problems 2. .1 but subjected to axial stress a2 alone.. ( I ) . volume fraction and area of cross section of the fiber. A. and (E.1 Figure P2. Figure P2.. T I . f and Af denote Young's modulus.2.) be the same qnantities for the matrix. (2) with Eq. derive the law of mzxtures. .e system: .. we have where S. in the problem coordinates...3. Guided by tht.1 below.. Assuming that plane sections remain plane during the deformation process and both matrix and fiber undergo the same longitudinal deformation A x l . for example. the equivalent modulus of elasticity E. form of the strainstress relations (1. P2.21). For an orthotropic material. as shown in Fig.
cos 48) + 8 2. by S11. S12. by definition.3) Show that the coefficient of mutual influence is zero at Q = O0 and 0 = 90°.6 (Continuation of Problem 2. Make use of the following trigonometric identities: 1 sin4 = . S l 2 . E2. S16represents the normal strain in the xdirection caused by the shear stress in the xyplane. when all other stresses are zero. G12. "12. and Ss6 and the lamination angle 8: S which was zero in the material coordinates.Thus. '9229 and S66: Physically. y. &ti for o.4 (Continuation of Problem 2. to the coefficients r/xy. 2 2 .yby 2.x and qzy. Guided by these observations. and 0. .(3 . Since 4 6 = 361.(1 .i =characterizes shearing in the xixjplane caused by a normal stress in the 2. Lekhnitskii [4] introduced the following engineering constants. # O and all other stresses being zero (7) The compliance S16and S2s are related. z ) coordinate system is contributed by the compliances Sll.3) Derive an expression for Gxy in terms of E l . when all other stresses are zero. it also represents the shear strain in the xyplane caused by the normal stress along the 2direction. the apparent compliance Sll in the (x. is contributed We note that the compliance S16.3) Show that Gxy is a maximum for Q = 45". 2. called coeficients of mutual influence: vij.direction (i # j )  %.4 cos 28 cos 48) 8 1 cos2 0 sin2 Q = ..5 (Continuation of Problem 2.
49. 2.10 Show that the following cornbinations of stiffness coefficierit. and v23 = 0. E 2 . a2. when 6. ...3) Show that the moduli E. width 2 in. is along the vector 61 . vl2 = vl:3 = 0. Determine the transverse normal strain €3. E 2 .4. .xk. are given as follows: 6.. e3) of the system ( x l ..13 Verify the transformation relations for the piezoelectric moduli given in Eq.10)..3) Derive the expression for G.15 Compute the numerical values of the reduced stiffriesses Q. and use El = 20 ~ n s i Ez = 1. Ans: .3. + + +  2.9 nisi. v12. direction cosines) relating the orthonorrnal basis vectors (61. but it can either exceed or get smaller than both El and E2. r u l . and 0 for the nonisothermal case..11 Rewrite the transformation equations (2.. G l z = G I 3 = 1. in terms of E l .14 Consider a square. 2. and subjected to an axial load of 1000 lbs. 2.3 msi.8) as where 2. for the graphiteepoxy material of Probleni 2.6. (2..62 63 and 8. Assume that the load is applied parallel to the fibers.005 in.14.3) Derive the expression for E. 4 a1..8 (Continuation of Problem 2. GZ3= 0.4. 2.2 2 . GI2. C Y ~ and 0 for the nonisothermal case. G I 2 .. and thickness 0.7 (Continuation of Problem 2. is perpendicular to the plane 2x1 322 2 3 5 = 0..12 Determine the transformation matrix (i.s are invariant: 2 . in terms of E l . 2.9 (Continuation of Problem 2.2. graphiteepoxy lamina of length 8 in.) varies between El and E z . e2.e. x 3 ) to the orthonormal basis (6.03 rnsi. Z 3 ) of the systeni ( x i . x i ) . (and E. 2.
CRC Press. M. Vinson. John Wiley. FiberReinforced Composites. R. R. S. Nauka. Second Edition. and Kudryavtsev. H. NASA Report T T F118 (1964). Principles of Composite Material Mechanics. J. 7. 3. New York (1988). CT (1969). McGrawHill. M. The Netherlands (1986). K.0 x m/m/OK.. Parton. Ambartsumyan. New York (1980). R. Agarwal. S. R. Gibson.. New York (1979). 2. Theory of Anisotropic Plates. arid Hahn. Analysis and Performance of Fiber Composites. Taylor & Francis.. B. V. Marcel Dekker. P. FL (1993).3 a2 = 30 1. G. T. Moscow (1982). Lancaster.. New York (1994). A. L. PA (1980). 8..2. John Wiley. and Broutman. Reddy. Mechanics of Composite Materials.. J. Christensen. The Behavior of Structures Composed of Composite Materials. Technomic. Boca Raton. A. Introduction to Composite Materials. compute the transformed thermal coefficients of expansion for 0 = 45" References for Additional Reading 1. Mechanics of Composite Materials. Moskva (1967). vlz = 0... Engineering Mechanics of Composite Stmctures. A. PA (1999). Theory of Elasticity of a n Anisotropic Body. English translation by Technomic. B. Mir Publishers. 11. 6. 9. R. x l o p 6 m/m/"K Show that (1 GPa = lo3 MPa = lo9 Pa) The transformed coefficients for various angles of orientation are given below: (:Pa GPa Also. J. S. G23 = 7 x lo3 MPa. D. Stamford. Lekhnitskii. Second Edition. and Sierakowski. 12.. ... Izdat. Theory of Anisotropic Shells. New York (2002). GI3 = 7 x a1 = E2 = 10 x lo3 MPa. 4. Kluwer. Ambartsumyan. N. 10. L. S. Z. Energy Principles and Variational Methods i n Applied Mechanics. Tsai. 5. F.. Jones.16 The material properties of AS13501 graphiteepoxy material layers are El = 140 x 10"~a. W. G I S= 7 x lo3 MPa lo3 MPa. Mallick. John Wiley.
Koyarna. . Halpin. P.. 18. Dayton. K. R. S.. and Doncr. "An Analytical and Numcrical Study of Fiber Microbl~ckling.. Pugarm." Journal of Composite Materials. C. (Ed. Selected Works of Nicholas J . F. "Transverse Normal Loading of a Uriidirectional Composite. 16. and Tsai. D. "Critique on Theories Predicting Therrrioelastic Properties of Fibrous Composites.. "Longitudinal Shear Loading of a Uriidirectional Composite. Tsai. N. S. 17. 15. R. Kluwcr. R. 332358 (1968). F. and Doner.. "Effects of Environmental Factors on Composite Materials. C. "Thcrmal Expansion Coefficients of Unidirectional Composites. 1 . (1964) 19.). W. Zhang. S. The Netherlands (1994). Charnis. C. J. 51. W." Journal of Composite Materials. R. OH (1969). Air Force Flight Mechanics Laboratory." Journal gf Composite Materials. 152164 (1967). Adarns. 95 109 (1994)." AFMLTR67423.13. NASA CR71. G. Mechanzcs of Composite Materials. G. 1 ." Composite Scie~rceand Technology. Structurul Behavior of Composrte Materials. D. and June. 12. T." Journal of Co7nposite Materials. D. and Sendeckyj. Ishikawa. J. 14. and Kobayaslii. Rcddy.. 153168 (1978)... 20. D.. Adarris. 417 (1967).
.
each layer is modeled as a 3D solid. Therefore. .2 Classification of Structural Theories Analyses of composite plates in the past have been based on one of the following approaches: (1) Equivalent singlelayer theories (2D) (a) Classical laminated plate theory (b) Shear deformation laminated plate theories (2) Threedimensional elasticity theory (3D) (a) Traditional 3D elasticity formulations (b) Layerwise theories (3) Multiple model methods (2D and 3D) The equivalent single layer (ESL) plate theories are derived from the 3D elasticity theory by making suitable assumptions concerning the kinematics of deformation or the stress state through the thickness of the laminate.1. In the threedimensional elasticity theory or in a layerwise theory. In this chapter. These assumptions allow the reduction of a 3D problem to a 2D problem. Often laminates are used in applications that require membrane and bending strengths.1 Introduction 3. an overview of pertinent literature on laminate plate theories is included here. The objective of this chapter is to develop two commonly used laminate plate theories. 3.Classical and FirstOrder Theories of Laminated Composite Plates 3. Literature reviews and development of the governing equations of the thirdorder shear deformation plate theory and the layerwise theory will be presented in later chapters (see Chapters 11 and 12). composite laminates have their planar dimensions one t o two orders of magnitude larger than their thickness. composite laminates are treated as plate elements. To provide a background for the theories discussed in this chapter. we present the classical plate theory and the firstorder shear deformation plate theory as applied to laminated plates.1.1 Preliminary Comments Composite laminates are formed by stacking layers of different composite materials and/or fiber orientation. namely the classical plate theory and the firstorder shear deformation plate theory. By construction.
which depend on the assumed displacement functions. Since all functions are explicit in the thickness coordinate. and Ro denotes the undeformed midplane of the plate. laminated or not.SV. virtual work done by external applied forces. pi and their variations. For plate structures. ! When pi are displacements. reducing the 3D continuum problem to a 2D problem. R:. z the thickness coordinate. y) the inplane coordinates. These quantities are determined in terms of actual stresses and virtual strains. the integration over plate thickness is carried out explicitly. In . and pi are functions to be determined. which is chosen as the reference plane.2 An Overview of Laminated Plate Theories The equivalent single layer laminated plate theories are those in which a heterogeneous laminated plate is treated as a statically equivalent single layer having a complex constitutive behavior. Consequently. C ( Z ) ~ P : ( ~ .2. except that the basis of the derivation of the governing equations is the principle of virtual forces. the integration over the domain of the plate is represented as the (tensor) product of integration over the plane of the plate and integration over the thickness of the plate. respectively. The ESL theories are developed by assuming the form of the displacement field or stress field as a linear combination of unknown functions and the thickness coordinate [l131: ~ i ( x . (x.2) consist of differential equations involving the dependent variables p:(x. More complete development of this procedure is forthcoming in this chapter. see Section 1.110 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS 3. and SK denote the virtual strain energy. y.1) N j=O where pi is the ith component of displacement or stress. The same approach is used when pi denote stress components."): The resultants can be written in terms of pi with the help of the assumed constitutive equations (stressstrain relations) and straindisplacement relations. then the equations governing p are determined by the principle of virtual displacements (or its dynamic version when time dependency is to be included. t the time. (3. l =t ) .2. ~ ) (3. because of the explicit nature of the assumed displacement field in the thickness coordinate: where h denotes the total thickness of the plate. and the virtual kinetic energy. the EulerLagrange equations of Eq. reducing the problem to a two dimensional one. Y .4): where 6U. t ) and thicknessaveraged stress resultants. ~ .
Such extensions lead to second.2. The higherorder theories introduce additional unknowns that are often difficult to interpret in physical terms. The firstorder shear deformation theory requires shear correction factors (see [28321)..5) implies that straight lines normal to the xyplane before deformation remain straight and normal to the midsurface after deformation.e.CLASSICAL A N D FIRSTORDER THEORIES 111 the present book.. Second. the stressbased theories will not be developed.vo. but also on the loading and boundary conditions. y. deforrriatiori is due entirely to bending and inplane stretching. The Kirchhoff assumption amounts to neglecting both transverse shear and transverse normal effects.e.e. Readers interested in stressbased theories may consult the book by Panc [14].which is based on the displacernent field where 4.1)] in the expansion of the displacernent components through the thickness of the laminate (see 133383.and higherorder ESL laminated plate theories use higherorder polynomials [i. The simplest ESL laminated plate theory is the classical laminated plate theory (or CLPT) [15201. N > 1 in Ey. Inclusion of this rudimentary form of shear deformation allows the norniality restriction of the classical laminate theory to be relaxed. of a point 011 the rnidplane (i.. The next theory in the hierarchy of ESL laminated plate theories is the firstorder shear deformation theorly (or FSDT) [2127].e. the transverse shear strain is assumed to be constant with respect to the thickness coordinate. i.2. among many others). wo) are the displacement components along the (:c.arid higherorder theories of plates.. It is based on the displacement field where (uo. The FSDT extends the kinematics of the CLPT by including a gross transverse shear deformation in its kinematic assumptions: i. respectively. In both CLPT and FSDT. which are difficult t'o determine for arbitrarily laminated composite plate structures. (3. z = 0). The shear correction factors depend not only on the lamination and geometric parameters. the planestress state assumption is used and planestress reduced form of the constitutive law of Section 2.4 is used. and & denote rotations about the y and x axes. respectively. The displacernerit~ field (3. which is an extension of the Kirchhoff (classical) plate theory to laminated composite plates. z ) coordinate directions. In both theories the iriexterlsibility and/or straightness of trarisverse normals can be removed. The secondorder theory with transverse inextensibility is based on the displacement field .
gross deflections. the FSDT with transverse extensibility appears to provide the best compromise of solution accuracy. e. The thirdorder theories provide a slight increase in accuracy relative t o the FSDT solution. the accuracy of the global response predicted by the ESL models deteriorates as the laminate becomes thicker. For additional discussion and references. 3D theories or multiple model approaches are required (see Chapter 12 for the layerwise theory and multiple model approaches). Complete derivations of the govkrning equations of the thirdorder laminated plate theory and their solutions are presented in Chapter 11.39] with transverse The displacement field accommodates quadratic variation of transverse shear strains (and hence stresses) and vanishing of transverse shear stresses on the top and bottom of a general laminate composed of monoclinic layers. the ESL models are often incapable of accurately describing the state of stress and strain at the ply level near geometric and material discontinuities or near regions of intense loading the areas where accurate stresses are needed most. the ESL models often provide a sufficiently accurate description of global response for thin to moderately thick laminates. First. This completes an overview of various ESL theories. critical buckling loads.. . whereas a restriction is not a necessary condition for the development of the theory. In such cases. finite element models of thirdorder theories that satisfy the vanishing of transverse shear stresses on the bounding planes require continuity of the transverse deflection and its derivatives between elements. In the classical laminated plate theory (CLPT) it is assumedt that the Kzrchhofl hypothesis holds: t An assumption is that which is necessary for the development of the mathematical model. Second.3.  3. In the remaining sections of this chapter. the ESL models have limitations that prevent them from being used to solve the whole spectrum of composite laminate problems. and simplicity. one may consult the review articles [40431.112 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS The thirdorder laminated plate theory of Reddy inextensibility is based on the displacement field [38. and fundamental vibration frequencies and associated mode shapes. In addition to their inherent simplicity and low computational cost. economy. Further.1 Assumptions The classical laminated plate theory is an extension of the classical plate theory to composite laminates.3 The Classical Laminated Plate Theory 3. at the expense of an increase in computational effort. However.g. we study the classical and firstorder shear deformation plate theories for laminated plates [44521. Of the ESL theories. Thus there is no need to use shear correction factors in a thirdorder theory.
x.(1) Straight lines perpendicular to the midsurface (i. x i ) of the kth lamina oriented a t an angle Qk to the laminate coordinate. = 0. . The lcth layer is located between the points z = zr. it is convenient to take the xyplane of the problem in the undeformed midplane f10 of the laminate (see Figure 3. (3) The transverse normals rotate such that they remain perpendicular to the midsurface after deformation.2 Displacements and Strains Consider a plate of total thickness h composed of N orthotropic layers with the principal material coordinates ( x f . E ~ = 0. The zaxis is taken positive downward from the midplane.3.1: Coordinate system and layer numbering used for a laminated plate.. (2) The transverse normals do not experience elongation (i. E.1).. transverse normals) before deformation remain straight after deformation.3. The third assumption results in zero transverse shear strains.3. is zero. The first two assumptions imply that the transverse displacement is independent of the transverse (or thickness) coordinate and the transverse normal strain E.. and z = zk+l in the thickness direction. they are inextensible). Figure 3..e.e. Although not necessary. z!j. 3. .
z w) in the deformed laminate. a combination of generalized forces and boundary generalized displacements. we make certain assumptions or place restrictions.vo. + veY+ we. y v.e.e. z) in the undeformed laminate moves to the position (x u. . no distinction is made between the material coordinates and spatial coordinates.1) allows reduction of the 3D problem to one of studying the defornlation of the reference plane z = 0 (or midplane). Different parts of the are subjected to.e.z) in the 3D continuum can be determined using Eq. as stated here:  r + 0 The layers are perfectly bonded together (assumption). The transverse shear stresses on the top and bottom surfaces of the laminate are zero (restriction). orthotropic) (restriction). y. y. h/2) of the laminate. ey.3.2).3. nyey. The strains and displacements are small (restriction). and between the second PiolaKirchhoff stress tensor and the Cauchy stress tensor (see Chapter 1). h/2). Once the midplane displacements (uo. 0 0 0 0 By the Kirchhoff assumptions. The Kirchhoff hypothesis requires the displacements (u. the displacements of any arbitrary point (x.3. in general. wo) are the displacements along the coordinate lines of a material point on the xyplane. wo) are known.1) where (e. In formulating the theory.vo. a material point occupying the position (x. where (u. The boundary of fiO consists of top surface St(z = h/2) and bottom surfaces I? x (h/2. v. z) coordinates. (3. between the finite Green strain tensor and infinitesimal strain tensor. w) are the components of the total displacement vector u along the (x. (3.2) where (uo. v. Due to small strain and small displacement assumption. We have + + + u = ue. w) to be such that (see Figure 3.3. Note that the form of the displacement field (3. In general. The material of each layer is linearly elastic and has three planes of material symmetry (i.The total domain fro of the laminate is the tensor product of Go x (h/2.) are unit vectors along the (x. y.. z) coordinates. and the edge a curved surface. Each layer is of uniform thickness (restriction). A discussion of the boundary conditions is presented in the sequel. with outward normal n = n. I? is Sb(z = h/2).y..
12).3.2: Undefornled and deformed geometries of an edge of a plate under the Kirchhoff assunlptions. The nonlinear strains are given by .2) can be computed using either the nonlinear straindisplacement relations (1.10) or the linear straindisplacement relations (1.Figure 3.3.3. The strains associated with the displacement field (3.3.
then the following terms are small but not negligible compared to 6 : and they should be included in the straindisplacement relations. Terms of order c2 are (El2.e. (3. 72) : ( (E) (g) (E)):( (g)(E) . the notation ~ i is used in place of Eij. the strains in Eq. then the small strain assumption implies that terms of the order e2 are negligible in the strains. (3.(E)2.3.e..4)(3.3. aw/i3z = 0.3) take the form where.If the components of the displacement gradients are of the order E.7) reduce to .3. i.3. for this special case of geometric nonlinearity (i. (3. If the rotations awo/ax and awo/dy transverse normals are moderate (say 1O0of 15"). For the assumed displacement field in Eq.6). small strains but moderate rotations).3.2). The corresponding second Piolaj Kirchhoff stresses will be denoted aij. In view of the assumptions in Eqs. Thus for small strains and moderate rotations cases the straindisplacement relations (3..
(1) el.3: Variations of strains and stresses through layer and laminate thicknesses. (3.10) the flexural 3.8) are called the von Khrma'n strains.10).. strains at any point (x.9) that all strain components vary linearly through the laminate thickness. in general. y. and they depend on time t for dynamic problems. z) in the plate can be computed using Eqs..3. . E. (b) Variation of corresponding stress. .3.) are zero by definition. nonlinear functions of x and y.3.) are identically zero in the classical plate theory... The first three strains in Eq. with their xlx2plane oriented arbitrarily with respect to the xyplane (x3 = z ) .3.3.:E ! ( (bending) strains.. and . For a laminate composed of orthotropic layers. . all three transverse strain components (E. Note that the transverse strains ( E ~ . E.. E. Once the displacements (uo.3. Figure 3. where . For a fixed value of z. and the associated plate theory is termed the won Ka'rma'n plate theory. (a) Variation of a typical inplane strain. the strains are. (3.3.9) and (3.. (3.3.. and they are independent of the material variations through the laminate thickness (see Figure 3.3 Lamina Constitutive Relations In the classical laminated plate theory..3.8) have the form are the membrane strains. y. (E:.3a). (3. Note from Eq.The strains in Eqs. known as the curvatures. wo) of the midplane are known. E ~ .(1) E.(1)) are (3.vo..
..4. AT = TTref.. The linear constitutive relations for the kth orthotropic (piezoelectric) lamina in the principal material coordinates of a lamina are where Q (. z ) relate the stresses (a. the planestress reduced constitutive relations of Section 2. a thin or moderately thick plate is in a state of plane stress because of thickness being small compared t o the inplane dimensions. z ) . x3).4. The coefficients Qij ) are known in terms of the engineering constants of the kth layer: Since the laminate is made of several orthotropic layers. and AT is the temperature increment from a reference state.3. as explained in Section 2. The stressstrain relations (3. However..a1 and a 2 are the coefficients of thermal expansion along the xl and x2 directions. strain. electric field components.. the part containing ( (k the piezoelectric moduli eijk ) should be omitted. ny.(ai.4a1b)]. (2.3.. Eq. = 0. and .1l a ) when transformed . the constitutive equations of each layer must be transformed to the laminate coordinates (x. although not zero identically... the transverse normal stress a. (2. Consequently. E ~T ~ ~ and components of the electric field vector (Ex. When piezoelectric effects are not present. to the laminate coordinates (x. y. in theory.118 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS the transverse shear stresses (n. Thus we have.EZ)in the laminate .) are also zero. respectively. from practical considerations.5)] . ~ ) coordinates [see Eq.Ey. Since E. y.~ iEi) are the stress. Hence... it amounts to neglecting the transverse normal stress. k ) are the plane stressreduced stiffnesses and e: are the piezoelectric l) moduli of the kth lamina [cf. 2 2 . ayyaxy) t o the strains (E.4 may be used. referred to the material coordinate system (xl. respectively. a case of both plane strain and plane stress. does not appear in the virtual work statement and hence in the equations of motion. with their material axes oriented arbitrarily with respect to the laminate coordinates.
they will have different linear variation in different material layers when Q!:) change from layer to layer (see Fig.3b)..23)] a..1213) Q11 = Qii Q12 cos4 6' (Q11 Q22  Q22  Qii (Q11   (Qlz  Q12 = Q22  and a. The transverse normal stress a. 3. yy. consistent with the mechanical strains..3. is not zero by the constitutive relation because of the Poisson effect. (3.3. Consequently. a..a 2 ) sin Q cos 0 and Eij are the transformed piezoelectric moduli + + + e32 sin2 8 2 2 e32 = egl sin 0 + eg2 cos 0 egl = e ~cos2 8 l e36 = (egl .3. the transverse strains (y. all three stress components do not enter the formulation because the virtual strain energy of these stresses is zero due to the fact that kinematically consistent virtual strains must be zero [see Eq.3. = 2 ( a l . the transverse shear stresses (a. we can write and the total strains are of the form in Eq.) are identically zero in the classical plate theory.9) with 3...e32) sin 0 cos Q Here 8 is the angle measured counterclockwise from the xcoordinate to the XIcoordinate. = a1 cos2 0 a 2 sin2 19 ay:y a1 sin2 8 a 2 cos2 0 = 2a.and aZv the transformed thermal coefficients of expansion [see Eq. (3..4 Equations of Motion As noted earlier.. however.3. If we assume that the temperature increment varies linearly. Note that stresses are also linear through the thickness of each layer. E. a v v .3. are (2.where + 2(Q12 + 2Q66) sin20 cos2 8 + Q22sin4 8 = + 4Qss) sin2 8 cos2 0 + Q12(sin48 + cos4 0) = sin4 Q + 2(Q12 + 2Q66) sin2 Q cos2 0 + Q22cos4 0 Qls = Q12 ~ Q M ) 8 cos3 6' + sir1 Q22 + 2Qs6) sin3 8 cos 0 Q26 = (QII ~ Q M ) 8 cos 0 + (Qlz Qa2 + 2QCiG) 0 cos3 6 sin3 sin Qss (QII + 2Q12 2Qm) sin2 8 cos2 Q + sin^ Q + cos4 0) (3...8)]: .) are zero for a laminate made of orthotropic layers if they are computed from the coristitutive relations. However...
17) where qb is the distributed force at the bottom (2 = h / 2 ) of the laminate. (1. In addition. ST = 0 and S& = 0). the governing equations are derived using the principle of virtual displacements. but they must be accounted for in the boundary conditions and equilibrium of forces. and the virtual kinetic energy SK are given by  kv1.3. these stress components may be specified on the boundary.Swo I dzds (3.78)) is where the virtual strain energy SU (volume integral of duo). the transverse stresses do not enter the virtual strain energy expression. Thus temperature and electric fields enter the formulation only through constitutive equations [see Eq.4. en.3.12a)l. (en.. virtual work done by applied forces SV. moisture) and piezoelectric effects only with the understanding that the material properties are independent of temperature and electric fields.. qt is the distributed force at the top (z = h/2) of the laminate. we account for thermal (and hence. The dynamic version of the principle of virtual work [see Eq.ens.120 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS Whether the transverse stresses are accounted for or not in a theory. In the derivations. Here. they are present in reality to keep the plate in equilibrium. and that the temperature T and electric field vector & are known functions of position (hence. h [ennbun + e n s b ~ ~ senzhw] drds + + 6. (3.) are the . Thus.
If a stress component is specified only on a part of the boundary. uo = auo/at.respectively. (1.3.15) and integrating through the thickness of the laminate. and SK from Eqs. respectively. (3.) are related to (uo.h curved boundary.. contributing nothing to the virtual work done. (3. gzy)..16)(3.3. For timedependent problems.4.. po is the density of the plate material..vo) and (a. SV.3. (6uo.3. The virtual displacements are zero on the portion of the boundary where the corresponding actual displacements are specified.) and (a. Since we are interested in the governing differential equations and the form of the boundary conditions of the theory. 6 ~ ~ . will be presented shortly. on the remaining part of the boundary the corresponding displacement must be known and hence the virtual displacement must be zero there. Details of how ( u o . the admissible virtual displacements must also vanish at time t = 0 and t = T [see Eq.4: Geometry of a laminated plate wit. a.4). of the boundary I?. we can assume that the stresses are specified on either a part or whole of the boundary. ~ ) are the virtual displacements along the normal and tangential directions. on the boundary r (see Figure 3.CLASSICAL A N D FIRSTORDER THEORIES 121 specified stress components 011 the portion r.73b)l. and a superposed dot on a variable indicates its time derivative. ayy.18) into the virtual work statement in Eq.uo. Substituting for SU.3... we obtain Figure 3. . .
3 . and Mi in lbinlin.3. denotes 11.). the transverse force resultant. 12)are the mass moments of inertia. and ( I o .5: Force and moment resultants on a plate element..g. Figure 3..) are called the inplane force resultants. .122 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS where q = qb + qt is the total transverse load and The quantities (N. Mz. and (M. Ni and Qiin Ib/in. 5 ) ...) are called the m o m e n t resultants (see Figure 3 . All stress resultants are measured per unit length (e.N.. N y y .. Mw. Q.
10)]: Substituting for the virtual strains from Eq. (3. so that we can use the fundamental lemma of variational calculus. (3./ax. = aN. 6wo) in no of any differentiation.21) into Eq. we obtain where a comma followed by subscripts denotes differentiation with respect to the subscripts: Nx.Sv".The virtual strains are known in terms of the virtual displacements in the same way as the true strains in terms of the true displacerrlents [see Eq.19) and integrating by parts to relieve the virtual displacements (duo. (3.. Tlie terms obtained . Note that both spatial and time integrationbyparts were used in arriving at the last expression.3.3. and so on.3.%.
23) to zero separately: . Svo. Svo. and Swo over Ro of Eq.Swo) together and noting that the virtual displacements are zero on I. where The EulerLagrange equations of the theory are obtained by setting the coefficients of Suo. we obtain ' . Collecting the coefficients of each of the virtual displacements (Sue.in Ro but evaluated at t = 0 and t = T were set to zero because the virtual displacements are zero there.3. (3.
.The ternis involving I2 are called rotary inertia terms. in the righthand side of the above equation are equal to N. In order to collect the coefficients of the virtual displacements and their derivatives on the boundary. wo.) are related to the derivatives (wo.) and ( w o .. = cos 0 and n..) are related to (a.. + cos 0 e. = sin 0 en  sin 0 e. Hence. y.. respectively.. the transformation between the coordinate system ( n .23). the normal and tangential derivatives (wo. r ) and (x.. wo. then its direction cosines are n. duo. Next we obtain the boundary conditions of the theory from Eq...). (3. 0 ..) by Now we can rewrite the boundary expressions in terms of (uo. s We have ~ ) We recognize that the coefficients of 6uo.9): Hence we have .. z ) is given by e.. = e... (2.. = sin 0. the displacements (uon. The term can contribute to higherorder vibration or frequency modes. a. If the unit outward normal vector n is oriented at an angle 0 from the xaxis.uo. and N. a. arid 6uo. a. s. e. Therefore.. SvO)in terms of (Sue..3.vo) by Similarly.$ e. This follows from the fact that the stresses (a..uOs) are related t o (uo.. we should express (Sue. = cos 0 e.3. ~ . and are often neglected in most books.) by the transformation in Eq...
M. wo).3..vo..23) can be written as The natural boundary conditions are then given by Mnn . N.e.25) have the total spatial differential order of eight.. Q.. We note that the equations in Eq. giving a total of ten boundary conditions.In view of the above relations. = 0 on . generalized forces) of the theory are primary variables: secondary variables: u. ? which constitutes the essential I (or geometric) boundary conditions.an ' a s N. Mns dwo awe (3... if the equations are expressed in terms of the displacements (uo. they would contain secondorder spatial derivatives of uo and vo and fourthorder spatial derivatives of wo. .. where Thus the primary variables (i.3. Hence.. =0 r.32) shows five essential and five natural boundary conditions. the boundary integrals in Eq.e. the classical laminated plate theory is said t o be an eighthorder theory.Mn. (3. us.3. wo. To eliminate this discrepancy.. one integrates the tangential derivative term by parts to obtain the boundary term .. Mr. This implies that there should be only eight boundary conditions.32) The generalized displacements are specified on . generalized displacements) and secondary variables (i. whereas Eq.. (3. Mns .3. In other words... (3.
Eq.e. To this end..e.3.. stresses are not. = Q. This boundary condition.25) are applicable to linear and nonlinear elastic bodies. (because it is a coefficient of Swo): which should be balanced by the applied force Q.12a) holds for the kth lamina in the problem coordinates. is known as the Kirchhoff freeedge condition.3. in addition to setting the nonlinear terms in the straindisplacement equations to zero. each lamina). (3. since the constitutive equations were not utilized in deriving the governing equations of motion. Equations (3.20a) to the strains of a laminate.25) by setting all terms involving time derivatives to zero. Although the strains are continuous through the thickness.. we assume that each layer is orthotropic with respect to its material symmetry lines and obeys Hooke's law.3. As a special case.. This term now must be added to Q. V. $ . 3. Hence.5 Laminate Constitutive Equations Here we derive the constitutive equations that relate the force and moment resultants in Eq. For the moment we consider the case in which the temperature and piezoelectric effects are not included. the integration of stresses through the laminate thickness requires laminawise integration. we set N ( w o ) and P(wo) to zero. For linear analysis. (3.. due to the change in material coefficients through the thickness (i. We note that both the displacement and velocities must be specified. The boundary conditions of the classical laminated plate theory are The initial conditions of the theory involve specifying the values of the displacements and their first derivatives with respect to time at t = 0: where variables with superscript '0' denotes values at time t = 0.3.3. E$$ d~ r' + vx. The force resultants are given by Qii Q12 Q22 Q I ~ ]) (' Q26 k=l Q12 QIG Q26 QG { EL? + } + ZE!) ZE. i. This completes the basic development of the classical laminated plate theory for nonlinear and dynamic analyses.The term in the square bracket is zero since the end points of a closed curve coincide.. one can obtain the equations of equilibrium from (3.
3.37') Adyy = + M~~ B 6 a 2 6 B66 1 yxy Dl6 0 2 6 D66 yxy where Aij are called extensional stiffnesses. and therefore A's.36) and (3. Equations (3. functions of position (x.3. (3.3. [B].37) can be written in a compact form as where { E ' ) and {E') are vectors of the membrane and bending strains defined in Eq.3. and [A]. y).10). and D's.and [Dl are the 3 x 3 symmetric matrices of laminate coefficients defined in Eqs.b).3. (3. are.14) and the laminate constitutive equations (39) become ) where { N T ) and { M ~are thermal force resultants . (3. and Bij the bendingextensional coupling stzffnesses. which are defined in terms of the lamina (k stiffnesses Qij ) as { } [i::it: {zg} [gt: :: (0) {ZR } (1) Note that Q's.3.38a. in general.5. the strains are given by Eq. For the nonisothermal case. Dij the bending stzffnesses.Mxx (3. Values of the laminate stiffnesses for various stacking sequences will be presented in Section 3. B's.
41a.6 Equations of Motion in Terms of Displacements The stress resultants (N's and M's) are related to the displacement gradients. nonhomogeneous plates).43) and (3. y) (i. temperature increment. the equations of motion (3. For homogeneous laminates (i. B's..3. 3.3.3. (uO..and {N'} and {' M} are the piezoelectric resultants Relations similar to Eqs. ~ &Q ax + 1&2 2 ( ax ) ay + &Q + &!&b ax axay The equations of motion (3. wo) by the relations ~ 1 1 ~ 1 2 ~ 1 6 ax 2( ax ) 24 + l ( & 4 ) 2 ay 2 a y awoawn ay + av.b) can be written for hygroscopic effects.44).e. and D's). for laminates with constant A's.25) can be expressed in terms of displacements VO. In general. the force and moment resultants can be expressed in terms of the displacements (uo. + a x a y ax duo + dWg 2 duo ~ 1 2 l c .e.3.3.vo. (3.25) take the form . (3. the laminate stiffnesses can be functions of position (x. and electric field.3. tug) by substituting for the force and moment resultants from Eqs. In the absence of the temperature and electric effects.
.
The nonlinear partial differential equations (3.hat the orily ) = ED in. These cases will be considered in the sequel.3./iri. the governing equations (3.47).a..3... = ~0 2 6 0 .3. a  "" a2M& + 2. the plate becorries a plate strip.3. (3.45)(3.3. the stresses iri kth lamina are given by !.tic analyses. In such cases.~. and = .24a).. analytically or numerically for a given problem. st. (3. Assume that layers are nonzero strains at a point ( 2 .47) can be sinlplificd for linear analyses. Consider a plat. the deflection UI" and displacernents ( u o : of the plate are functions of only x. all derivatives with respect to y are zero.r~(t~zcul bending. Eq = 2.8 psi.12). We wish to determine the state of stress (~.3.erial properties El = 7..3. arid it is referred to as thc cy2a. Hence.(! strip that has a finite dimension along the zaxis and subjected to a transverse load q(z) that is uriiforr~l at any section parallel to the zaxis. Therefore. G l z = GI3 = 1. ~ . Bij. with mat.cr. + .47) reduce to tlo) Example 3.5 psi. ~ are of thickness 0.3 psiir GZ3= 0. Once the displacements are determined by solving Eqs."/in.Dij) are zero.cr.1: (Cylindrical Bending) If a plate is infinitely long in one direction.6 psi. (3. For this case.3. respectively.) arid forw resultants in the laminate. The only nonzero strain is E . a. arid vl2 = 0. In such a case.2: Suppose that a sixlayer (d3i0/0).(3..45)(3.3.25. Example 3. the deflected surface of the plate strip is cyliridrical.E :! !! :E .45). symmetric laminate is subjected to loads such t.nd lamination schemes for which some of the stiffnesses (Aij.3.10) and (3.005 in.3. the strains and stresses in each lamina can be computed using Eqs.9' + F ) ayax where N ( w o ) was defined in Eq.
where
The stress resultants are given by
If
EO =
1000 x
lop6
in./in. and
KO
= 0, we have
If
EO
= 0 in./in. and no = 1.0 /in., we have
3.4 The FirstOrder Laminated Plate Theory
3.4.1 Displacements and Strains
In the firstorder shear deformation laminated plate theory (FSDT), the Kirchhoff hypothesis is relaxed by removing the third part; i.e., the transverse normals do not remain perpendicular to the midsurface after deformation (see Figure 3.4.1). This amounts to including transverse shear strains in the theory. The inextensibility of transverse normals requires that w not be a function of the thickness coordinate, z. Under the same assumptions and restrictions as in the classical laminate theory, the displacement field of the firstorder theory is of the form
where (uo,vo, wo, 4x,dy) are unknown functions to be determined. As before, (uo,vo, wo) denote the displacements of a point on the plane z = 0. Note that
which indicate that 4, and q4y are the rotations of a transverse normal about the y and xaxes, respectively (see Figure 3.4.1). The notation that 4 denotes the , rotation of a transverse normal about the yaxis and q4y denotes the rotation about the xaxis may be confusing to some, and they do not follow the righthand rule. However, the notation has been used extensively in the literature, and we will not
Figure 3.4.1: Undeformed and deformed geometries of an edge of a plate under the assumptions of the firstorder plate theory. depart from it. If (P,, &) denote the rotations about the x and y axes, respectively, that follow the righthand rule, then
The quantities (uo,vo, wo, &, &) will be called the generalized displacements. For thin plates, i.e., when the plate inplane characteristic dimension to thickness ratio is on the order 50 or greater, the rotation functions 4, and q59 should approach the respective slopes of the transverse deflection:
The nonlinear strains associated with the displacement field (3.4.1) are obtained by using Eq. (3.4.1) in Eq. (3.3.7):
Note that the strains (E,,, E ~ , , yzy) are linear through the laminate thickness, while the transverse shear strains (y,, ,yy,) are constant through the thickness of the laminate in the firstorder laminated plate theory. Of course, the constant state of transverse shear strains through the laminate thickness is a gross approximation of the true stress field, which is at least quadratic through the thickness. The strains in Eq. (3.4.3) have the form
3.4.2 Equations of Motion
The governing equations of the firstorder theory will be derived using the dynamic version of the principle of virtual displacements:
virtual work done by applied forces SV, and the where the virtual strain energy SU, virtual kinetic energy SK are given by
+ w O S w O dz d x d y
I
(3.4.8)
where all variables were previously introduced [see Eqs. (3.3.16)(3.3.18) and the paragraph following the equations].
Substituting for SU, SV, and SK from Eqs. (3.4.6)(3.4.8) into the virtual work statement in Eq. (3.4.5) and integrating through the thickness of the laminate, we obtain
where q = qt, qt, the stress resultants (N,,, Nyy,Nzy,M,,, Myy,Mzy) and the inertias (Io, 12) as defined in Eq. (3.3.20), (N,,, Nns, M,,, Mn,?)are as defined 11, are in Eq. (3.3.29a,b), and
+
The quantities (Q,, Q y ) are called the transverse force resultants.
Shear Correction Factors
Since the transverse shear strains are represented as constant through the laminate thickness, it follows that the transverse shear stresses will also be constant. It is well known from elementary theory of homogeneous beams that the transverse shear stress varies parabolically through the beam thickness. In composite laminated beams and plates, the transverse shear stresses vary at least quadratically through layer thickness. This discrepancy between the actual stress state and the constant stress state predicted by the firstorder theory is often corrected in computing the transverse shear force resultants (Q,, Qy) by multiplying the integrals in Eq. (3.4.10a) with a parameter K, called shear correction coeficient:
This amounts to modifying the plate transverse shear stiffnesses. The factor K is computed such that the strain energy due to transverse shear stresses in Eq. (3.4.10b) equals the strain energy due to the true transverse stresses predicted by the threedimensional elasticity theory. For example, consider a homogeneous beam with rectangular cross section, with width b and height h. The actual shear stress distribution through the thickness of the beam, from a course on mechanics of materials, is given by
where Q is the transverse shear force. The transverse shear stress in the firstorder theory is a constant, oiz = Qlbh. The strain energies due to transverse shear stresses in the two theories are
f The shear correction factor is the ratio of ~ , to U,C, which gives K = 516. The shear correction factor for a general laminate depends on lamina properties and lamination scheme. Returning to the virtual work statement in Eq. (3.4.9), we substitute for the virtual strains into Eq. (3.4.9) and integrate by parts to relieve the virtual generalized displacements (6uo,6vo,6wo,S$,, 64,) in Ro of any differentiation, so that we can use the fundamental lemma of variational calculus; we obtain

(
y
+M y , Q
2
6

1 1 ~ 0 )64y
 (Qz,z
+ Qy,, + N(wo) q  IOGO) dxdy + 6w0
I
where N(wo) and P(wo) were defined in Eq. (3.3.24), and the boundary expressions were arrived by expressing 4, and 4, in terms of the normal and tangential rotations, (4n, 4s): (3.4.12) 4s = n d n  nY4, , 4, = n Y 6 h+ n264s The EulerLagrange equations are obtained by setting the coefficients of 6uo, 6vo, 6wo, 64,, and 64y in Ro to zero separately:
The natural boundary conditions are obtained by setting the coefficients of 6u,, SU,~, 6wo, S4,, and 64, on r' to zero separately:
Nnn  Nnn = 0 , NnS  NnS 0
M,, where
Qn
Qn  Q, = 0
 Mnn = 0
Thus the primary and secondary variables of the theory are primary variables: secondary variables: N,, u,,, us, wo, &, , Nns, Q, , M,,

,
Mns  Mns = 0
QZ%
+ Q,ny + Wwo)
4s
, Mns
(3.4.15)
Note that Q,, defined in Eq. (3.4.1413) is the same as t,hat defined in Eq. (3.3.31b). This follows from the last two equations of (3.4.13). The initial conditions of the theory involve specifying the values of the displacements and their first derivatives with respect to time at t = 0:
for all points in
no.
3.4.3 Laminate Constitutive Equations
The laminate constitutive equations for the firstorder theory are obtained using the lamina constitutive equations (3.3.12a) and the following relations:
where [see Eq. (2.4.10)] Q44 = Q44 cos2 Q
Q45
+ QS5sin2 19 + QS5cos2 Q
0
= (Q55  Q44)cos Q sin Q
Q55 = Q44 sin2 0 (el5  e24) sin 0 cos 0, 215 = el5 cos2 0 e24 sin2 8 ,
E14 =
2 E24 = e24 cos
+
+ el5 sin2 0
En5 =
(e15 e24)sin Q cos 13
The laminate constitutive equations in Eqs. (3.3.36) and (3.3.37) are valid also for the firstorder laminate theory. In addition, we have the following laminate constitutive equations:
where the extensional stiffnesses A44,A45, and As5 are defined by
=XI
N
k=l
zk+l
'"
(Q44.Q15,Q55)d~
(k) (k)
(k)
and the piezoelectric forces Q: and Q: are defined by
When thermal and piezoelectric effects are not present, the stress resultants (N's and M's) are related to the generalized displacements (uo,vo, wo, &, 4y) by the relations
When thermal and piezoelectric effects are present, Eqs. (3.4.20) and (3.4.21) take the same form as Eq. (3.3.40), and Eq. (3.4.22) will contain the col~mrl of piezoelectric forces given in Eq. (3.4.18).
3.4.4 Equations of Motion in Terms of Displacements
The equations of motion (3.4.13) can be expressed in terms of displacements (uo,vo, wo, &, 4,) by substituting for the force and moment resultants from Eqs. (3.4.20) (3.4.22). For homogeneous larninates, the equations of motion (3.4.13) take the form (including thermal and piezoelectric effects)
Equations (3.4.23)(3.4.27) describe five secondorder, nonlinear, partial differential equations in terms of the five generalized displacements. Hence, the firstorder laminated plate theory is a tenthorder theory and there are ten boundary conditions, as stated earlier in Eqs. (3.4.14) and (3.4.15). Note that the displacement field of the classical plate theory can be obtained from that of the firstorder theory by setting
Conversely, the relations in Eq. (3.4.28) can be used to derive the firstorder theory from the classical plate theory via the penalty function method (see Chapter 10).
Example 3.4.1:
T h e linearized equations of motion for cylindrical bending according t o t h e firstorder shear deformation theory are given by setting all derivatives with respect to y in Eqs. (3.4.23)(3.4.27):
3.5 Laminate Stiffnesses for Selected Laminates
3.5.1 General Discussion
A close examination of the laminate stiffnesses defined in Eqs. (3.3.38) and (3.4.lga) show that their values depend on the material stiffnesses, layer thicknesses, and the lamination scheme. Symmetry or antisymmetry of the lamination scheme and material properties about the midplane of the laminate reduce some of the laminate stiffnesses to zero. The book by Jones 1441 has an excellent discussion of the laminate stiffnesses for various types of laminated plates. In this section, we review selective lamination schemes for their laminate stiffness characteristics. Before we embark on the discussion of laminate stiffnesses, it is useful to introduce the terminology and notation associated with special lamination schemes. The . lamination scheme of a laminate will be denoted by ( a / / 3 / y / G / ~./.), where a is the orientation of the first ply, /3 is the orientation of the second ply, and so on (see Figure 3.5.1). The plies are counted in the positive x direction (see Figure 3.3.1). Unless stated otherwise, this notation also implies that all layers are of the same thickness and made of the same material. A general laminate has layers of different orientations Q where 90" Q 90". For example, (0/15/35/45/90/45) is a sixply laminate. General angleply laminates (see Figure 3.5.2) have ply orientations of Q and 8 where 0" 5 8 90°, and with at least one layer having an orientation other than 0" or 90". An example
<
<
<
Figure 3.5.1: A laminate with general stacking sequence.
of angleply laminates is provided by (15130/0/90/45/45). Crossply laniirlates are those which have ply orientations of 0" or 90" (see Figure 3.5.3). An example of a crossply laminate is (0/90/90/0/0/90). For layers with 0" or 90" orientations, the layer stiffnesses Q I 6 , Q26, Qd5 are zero. Hence, AI6 = A26 = Ad5 = DI6 = 0 2 6 = 0. When ply stacking sequence, material, and geometry (i.e., ply thicknesses) are symmetric about the midplane of the laminate, the laminate is called a symmetrzc lamznate (see Figure 3.5.4). For a symmetric laminate, the upper half through the laminate thickness is a mirror image of the lower half. The laminates ( 451451451 45)=(45/45), and (45145145145) = (45145),, with all layers having the same thickness and material, are examples of a symmetric angleply laminate, (0/90/90/0) = (0/90), is a symmetric crossply laminate, and (30145/0/90/90/0/45/30)=(30/45/0/90), is a general symmetric laminate.
Figure 3.5.2: A general angleply laminate.
T2
Figure 3.5.3: A crossply laminated plate with the 0" and 90" layers.
Figure 3.5.4: A symmetric laminate.
Note that symmetric laminates are also denoted by displaying only the lamination is scheme of the upper half. The symmetric laminate (25/35/0/90/90/0/35/25) denoted as (25/35/0/90),. An unsymmetric or a s y m m e t r i c laminate is a laminate that is not symmetric. An antisymmetric laminate is one whose lamination scheme is antisymmetric and material and thicknesses are symmetric about the midplane. Examples of antisymmetric angleply and crossply laminates are provided, respectively, by (30/30/30/30/30/30)r (30/30)3 and (0/90/0/90/0/90)= (0/90)3. Laminate stiffnesses Aij depend on only on the thicknesses and stiffnesses of the layers but not on their placement in the laminate. On the other hand, laminate stiffnesses Dij depend not only on the layer thickness and stiffnesses but also on their location relative to the midplane. For example, both (0/90), and (90/0), laminates will have the same inplane stiffnesses Aij. However, (0/90), laminate will have larger bending stiffnesses Dij about an axis perpendicular to the fiber direction than the (90/0), laminate, because the 0" layers are located farther from the midplane in the (0/90), laminate. Both Aij and Dij are always positive. Laminate stiffnesses Bij also depend on the layer thickness, stiffnesses and location relative t o the midplane, and they can be negative, depending on the lamination scheme and the number of layers.
3.5.2 SingleLayer Plates
Here we discuss some special cases of singlelayered configurations and their stiffnesses. The special single layer plates discussed here include: isotropic, specially orthotropic (i.e., the principal material coordinates coincide with those of the plate), generally orthotropic (i.e., the principal material coordinates do n o t coincide with those of the plate), and anisotropic. The bendingstretching coupling coefficients
Bij and the shear stiffnesses AIF, AZ6, DIG, and Dzs can be shown to be zero for all singlelayer plates except for generally orthotropic and anisotropic singlelayer plates. The units of Ni and Mi, the U.S. in Customary System (USCS), are lbin. and lbinlin., respectively. Single Isotropic Layer
E For a single isotropic layer with material constants E and u [G = and thickness h, the nonzero laminate stiffnesses of Eqs. (3.3.38) and (3.4.19a) become
rn]
1v 1v Eh All, A44 = A55 = All All =  A12 = uAll, A22 = A l l , A66 = 2 1u2' 2
The plate constitutive equations for the classical and firstorder theories become
) The nonzero thermal stress resultants { N ~and { M T ) are given by
NT =NT
ZZ
=

AT dz,
MT~= M& =

Single Specially Orthotropic Layer For a single specially orthotropic layer, the stiffnesses can be expressed in terms of the Qij and thickness h. The nonzero stiffnesses of Eqs. (3.3.38) and (3.4.19a) become Ail = Q i i h , A12 = Q12h, A22 = Q22h
where Qij are the planestressreduced stiffnesses, and they are given in terms of the engineering constants [see Eq. (3.3.1l b ) ] as
The plate constitutive equations for the classical and firstorder theories become
The nonzero thermal stress resultants are given by
Single Generally Orthotropic Layer For a single generally orthotropic layer (i.e., the principal material coordinates do not coincide with those of the plate), the stiffnesses can be expressed in terms of the transformed coefficients Qii and thickness h. The nonzero stiffnesses are (Bij = 0)
The plate constitutive equations are
The thermal stress resultants for this case are given by
A similar expression holds for { M T ) .
If the temperature increment is linear through the layer thickness, A T = To+zTl, the thermal stress resultants have the form
Single Anisotropic Layer For a single anisotropic layer, the stiffnesses are expressed in terms of the coefficients Cij and thickness h. The nonzero stiffnesses are (Bij = 0)
for i, j = 1 , 2 , 3 , 4 , 5 and 6 [see Eq. (2.4.3a)I. The plate constitutive equations are the same as in Eqs. (3.5.13)(3.5.16) with the plate stiffnesses given by Eq. (3.5.18).
Example 3.5.1:
The material properties of boronepoxy material layers are
El = 30 x lo6 psi, E2 = E3 = 3 x lo6 psi, G12= G I 3 = 1.5 x 106 psi
G2, = 0.6 x 10"si,
ul2
= 0.25, ulu = 0.25, u2j = 0.25
The rnatrix of elastic coefficients for the material is [see Eq. (1.3.44)]
The plane stressreduced elastic coefficient rnatrix in the material coordinates is
30.189 0.755 0 0,755 3.019 0 0 0.6 0 0 0 0
0 0 0 0 1.5 0 0 1.5
The transformed stiffness matrix [Q] for 0 = 60" is given by
1
msi
The laminate stiffnesses Ai, and D,, for i , j = 1 , 2 , 6 may be computed using Eq. (3.5.12). The A45,and A55 are given by Aij = Q,,h for i,j = 4 , 5 . transverse shear stiffnesses A44, Suppose that the thermal coefficients of expansion of the material are
The transformed coefficients are
3.5.3 Symmetric Laminates
When the material properties, locations, and lamination scheme are symmetric about the midplane, the laminate is called a symmetric laminate. If a laminate is not symmetric, it is said to be an unsymmetric laminate. Due to the symmetry of the layer material coefficients Q::), distances z k , and thicknesses ha about the midplane of the laminate for every layer, the coupling stiffnesses Bij are zero for symmetric laminates (see Figure 3.5.5). The elimination of the coupling between bending and extension simplifies the governing equations. When the straindisplacement equations are linear, the equations governing the inplane deformation can be uncoupled from those governing bending of symmetric laminates. Further, if there are no applied inplane forces or displacements, the inplane deformation (i.e., strains) will be zero, and only the bending equations must be analyzed. F'rom production point of view, symmetric laminates do not have the tendency to twist from the thermally induced contractions that occur during cooling following the curing process.
tz
Figure 3.5.5: A symmetric crossply laminate.
about the midplane of the laminate does not exhibit coupling between bending and extension . (3. If A T = To zTl.4.The force and moment resultants for a symmetric laminate. (3. (3.15)l.13)(3. then Eq. the relations between strains and resultants can be further simplified.5.3. have the same form as the generally orthotropic singlelayer plates [see Eqs.5. as explained next. in general.5. the resulting laminate will have the following laminate constitutive equations for the classical or firstorder theories: All A12 0 (3. For certain special cases of symmetric laminates.21a) where the laminate stiffnesses Aij and Dij are defined by Eqs. Symmetric Laminates with Multiple Isotropic Layers When isotropic layers of possibly different material properties and thicknesses are arranged symmetrically from both a geometric and a material property standpoint.38) and (3. both from a material and geometric properties standpoint.lga) with The thermal stress resultants for this case are given by and similar expression holds for { M ~ ) .5.23) can be written as + Symmetric Laminates with Multiple Specially Orthotropic Layers A laminate composed of multiple specially orthotropic layers that are symmetrically disposed.
).5. DI6. p. Symmetric Laminates with Multiple Generally Orthotropic Layers Laminates can be composed of generally orthotropic layers whose principal material directions are aligned with the laminate axes at an angle 6' degrees. a general symmetric crossply laminate can have either an even or odd number of layers: (0/90/0/90/90/0/90/0) or (0/90/90/0/0/90/90/0) (see Figure 3. the values of A16..21ac). . Stiffnesses A16. where 0. A common example of specially orthotropic laminates is provided by the regular symmetric crossply laminates. Az6. and the laminate constitutive equations are given by Eqs. Of course. .6).15). 0" < a < 90" (see Figure 3. If the thicknesses. . otherwise. and Dzs are not zero).and D26 of a regular symmetric angleply laminate are the largest when the number of layers N is equal to 3. Bij = 0.e. A general symmetric angleply laminate has the form (6'/P/y/.19a) with Such laminates are also called specially orthotropic laminates.5.e. (3. where the laminate stiffnesses Aij and Dij are defined by Eqs. The regular symmetric crossply laminates necessarily contain an odd number of layers.e. The thermal stress resultants have the same form as those given in Eq.). but they should be symmetrically placed about the midplane.. Dl6. and twisting moment and normal curvatures is not zero for these laminates (i.5). Note that the coupling between normal forces and shearing strain. Regular symmetric angleply laminates are those that have an odd number of orthotropic laminae of equal thicknesses and alternating orientations: (a/a/a/a / a / .. Azs. . and material properties of the layers are symmetric about the midplane of the laminate.23). for symmetric angleply laminates with many layers. . Dl6.5. xl and x2) alternating at O0 and 90" to the laminate axes x and y: (0/90/0/90/. and they decrease in proportion to 1/N as N increases. It can be shown that the stiffnesses A16.. D16.4. where the subscript denotes the number of layers of the same orientation and thickness. and Dz6 are not zero. which consist of laminae of the same thickness and material properties but have their major principal material coordinates (i.3.13)(3. normal moments and twist. . locations. Thus.5. which holds for all symmetric laminates. and Dzs can be quite small compared to other Aij and Dij. (3. and they do not necessarily go to zero as the number of layers is increased. (3. A laminate composed of multiple anisotropic layers that are symmetrically disposed about the midplane of the laminate does not have any stiffness simplification other than Bij = 0. and y can take any values between 9 0 " and 90°. Als.i. and each layer can have any thickness.).38) and (3.5. . the coupling between bending and extension is zero. they are not symmetric. shearing force and normal strains. Azs. Bij = 0.5. An example of a general symmetric laminate with generally orthotropic laminae is provided by (30/603/155/603/30). The laminate constitutive equations are again given by Eqs. (3. A26.
of boronepoxy layers has the stiffncsses [A] = [ 16.5. TI = 0) A symmetric crossply laminate (0/90/0/90).959 0 0 ] lo6 lbin.5.5 I [I808 0..755 16. and rnade of boronepoxy layers [see Eqs. [Dl = 0.2: A general symmetric laminate (30/0/90/45).5. (3.19) and (3. Symmetric laminates are much easier to analyze than general or unsymmetric laminates. symmetric laminates are preferred wherever they meet the application requirements. The transverse shear stiffnesses are (in lo6 lb/in.604 0.063 0.755 0 0. of total thickness 1 in. symmetric laminates do not have a tendency to twist due to thermally induced contractions that occur during cooling following the curing process.20) for material properties] has the following laminate stiffnesses: The trar~sverse shear stiffnesses are (in lo6 lb/in.6: A symmetric angleply laminate In general.O63 0.Figure 3.604 0 0 1. Further.) The thermal stress resultants are (To # 0. Example 3.125 lo6 lb/in.5.) .
0151 0.481 0. The only nonzero strain is E.1. [Dl = 0 7. = E ~ : ) Z .126 0. 0" 5 8 5 90". The general class of antisymmetric laminates must have an even number of orthotropic laminae if adjacent laminae have equal thicknesses and alternating orientations: ( 1 3 .3321 0 0 0 0. the shear stiffness of laminates can be increased by orienting the layers at angle to the laminate coordinates..256 0.0151 0. All moments will be zero on account of the fact that there are no bending strains and the coupling stiffnesses Bij are zero. We wish to determine the forces and moment resultants.461 0. the only nonzero strain is E.03 1 { :I } 1. A symmetric angleply laminate (30130145145).O ) . Due to the antisymmetry of the lamination . The transverse shear stiffnesses are Example 3.4 Antisymmetric Laminates Although symmetric laminates are more desirable from an analysis standpoint./in.122 1.379 6. ~ ~ = EL:).000 = { b/in.481 0.256 0. Then the force resultants are zero. Another example that requires coupling is provided by turbine blades with pretwist.376 :] lo6 lb/in.5.470 0. For example.3:21 0. of boronepoxy layers has the stiffnesses 14..543 1 lo6 lbin. Now suppose that the laminate is subjected to loads such that it experiences only nonzero strain of EL:) = 0. Suppose that the laminate is subjected to loads such that it experiences only nonzero strain of E:. here h is the total thickness of the laminate. they may not meet the design requirements in some applications.126 0..5.005 in. and the moment resultants are given by 3.Note that the crossply laminate considered here is equivalent to ( 0 / 9 0 / 0 / 9 0 / 0 / 9 0 / 0 ) where all layers except the middle layer having a thickness of h / 8 and the middle layer (90) has a thickness of h / 4 . Moreover. made of boronepoxy layers of thickness 0. A12 A22 Hence the force resultants in the laminate are given by } = All [Al2 0 Nv x = [0. a heat shield receives heat from one side and thus requires nonsymmetric laminates to effectively shield the heat. Hence.3: Consider a symmetric laminate ( 0 / 9 0 ) . = 1 0 in.
.5. Figure 3. Foa general antisymmetric laminate. they go to zero as the number of layers is increased. The coupling stiffnesses Bij are not all zero.7: An antisymmetric laminate.scheme (see Figure 3. this class of antisymmetric laminates has the feature that AI6 = A26 = D16 = 0 2 6 = 0.7) but symmetry of the thicknesses of each pair of layers.5. the relations between the stress resultants and the strains are given by The thermal force resultants are given by Similar expression holds for { M ~ ) .
laminates that have an even number of orthotropic laminae. For these laminates. the coupling stiffnesses Bij have the properties Bz2= Bll.5. there is a 90" layer of the same thickness and location on the other side of the midplane (see Figure 3. An tisymme tric Crossply Laminates A special case of antisymmetric laminates are those which have an even number of orthotropic layers with principal material directions alternating at 0" to 90" to the laminate axes. Such laminates are called antisymmetric crossply laminates.e.8: An antisymmetric crossply laminate.In the following pages. . .28) The relations between the stress resultants and the strains are * z Figure 3. Examples of antisymmetric crossply laminates are (0/90/0/90/ . we discuss some special cases of the class of antisymmetric laminates described above (i. . and all other Bij = 0 (3.5.) with all layers of the same thickness.5.8).. each pair having equal thicknesses and alternating orientations). and (0/90/90/0/0/90) with layers of the thicknesses (hl/h2/h3/h3/h2/hl). Note that for every 0' layer of a given thickness and location.
For these laminates. then the coefficients B.755 16. An arltisyrrlrnetric angleply laminate (45/45/30/0/0/ 301 45/45) of boronepoxy layers has t h e laminate stiffnesses 17. Example 3. the stiffnesses B16and B26 go t o zero as the nurnber of layers in the laminate increases.281 5.194 0. [B] = 0. Antisymmetric Angleply Laminates An antisymmetric angleply laminate has an even number of orthotropic layers with principal material directions alternating at 8 degrees to the laminate axes on one side of the midplane and corresponding equal thickness laminae oriented at 8 degrees on the other side.050 0 1. An example is given by (45/40/15/15/40145). 5 ] lo6 l i n { = A5.. would vanish (why?).194 0 106 lb/in.063 0 6 3 1. 8 should be interpreted as 90" or vice versa.067 .755 0.A regular antisymmetric crossply lam. 1.172 0 5.384 0 0 0 0 .050 0 0 b/n.5. Note t h a t if t h e same 0" and 90" layers are positioned differently.604 0 0 10"b/in.384 0. the stiffnesses can be simplified as The relations between the stress resultants and the strains are For a fixed laminate thickness. the coupling coefficient Bll approaches zero as the number of layers is increased..5 { } 1. For antisymmetric angleply laminates without 90" layers.. say (0/90/90/0/90/0/0/90).inate is one that has an even nurnber of layers of equal thickness and the same material properties and which have alternating 0" and 90" orientations. A regular antisymmetric angleply laminate is one that has an even number of layers of equal thickness and material properties.917 1 0.4: A regular antisyrr~metriccrossply laminate (0/90/0/90/0/90/0/90) laminate stiffnesses of boronepoxy layers has the ['0 "I4 [A] = 0. When 8 = 0.5 I [B] = [ I= [ 1.
. {ti:} { = A55 } 1 lo6 1b 1o61b/in. Laminates consisting of three or more orthotropic laminae of identical material and thickness which are oriented at the same angle relative to adjacent laminae exhibit inplane isotropy in the sense that All = A22.491 3. 0 0.. The two layers are not necessarily symmetrically located with respect to the midplane. 3.425 lo6 lb/in.88 1 0 1.A12)/2. lbin.311  = 0. somewhere in the laminate. the unsymmetric laminate (f35/O)T =(35/3510) as well as the symmetric laminate ( f 35/O).565 0 4. However.366 ] ] 0. the laminate constitutive relations are not that much simpler than for a general laminate.2).13)(3.0 0. are balanced laminates. Thus.5. When the bendingstretching coupling coefficients are zero..637 ] 10. The stress resultants are given by .233 0 lo6 lbin.5.5. (2:' { } A55 A general antisymmetric laminate (30/0/90/45)as (30/0/90/451451 0/90/30) of total thickness 1 in.5 Balanced and QuasiIsotropic Laminates A laminate is said to be balanced if for every layer in the laminate there exists. and AI6 = A26 = 0.9938 0 1. The reason is that QI6 and Q26 from opposite orientations of the pair of layers are of opposite sign and therefore the net contribution from the pair to AI6 and Azs is zero: For a general balanced laminate.842 0 0.425 0. for a symmetric balanced laminate they are given by Eqs. (3. [ B ] = 0.233 0.1063 0. another layer with identical material and thickness but opposite fiber orientation. Such laminates are called quasiisotropic laminates. The characteristic feature of any balanced laminate is that the inplane shear stiffnesses A16 and AZ6are zero.3. Examples of quasiisotropic laminates are provided by (90/45/0/45) and (60/0/60) (see Example 3. and composed of boronepoxy layers has the following laminate stiffnesses and thermal resultants: 15.15) with AI6 = = 0. the relations between force resultants and membrane strains are the same as those for isotropic plates. = (All .219 1 0 lb/in.842 0.
1) Use the principle of virtual displacements to derive the equations of equilibrium and the natural and essential boundary conditions associated with the displacement field of Problem 3. when the beam is subjected to axial distributed load p(x) and transverse distributed load q(x). t ~ . respectively.Problems 3. cl = 0) and the firstorder (Timoshenko) beam theory (co = 0. hence the equations of equilibrium (3).1 Suppose that the displacements ( u . y.3 (Continuation of Problem 3.1) Assurrie linear elastic constitutive behavior and show that the laminated beam's constitutive equations are given by where . 0) along the x and z directions.wo) denote the displacements of a point (x. cl = 1). y . w ) along the three coordinate axes (x. and 4 denotes the rotation of a transverse normal about the yaxis. contain those of the classical (EulerBernoulli) beam theory (co = 1 . Show that the nonzero linear strains are given by where 3. z ) in a laminated bean1 can be expressed as where (uo.1.2 (Continuation of Problem 3. 3. In particular. show that and the boundary conditions are of the form Note that the displacement field ( I ) .
) using . x3 = z ) . F3) determine the functions (Fl. and then use the equilibrium equations of the threedimensional elasticity theory to determine the transverse stresses (a. in the absence of body forces.z k f l ) with respect to z = 2 3 to obtain: for lc = 1 . orthotropic layer plate (Q45 = 0).3.. Fl .7 Consider a single. 2 ) is implied.) as a function of the thickness coordinate. Compute the stresses (u. ./3 = 1 . and 3.z2 = y... can be expressed in index notation as % +3k3 a a dz. where summation on repeated subscripts ( a .. . the constitutive equations of the firstorder plate theory. u. . F3) such that the Kirchhoff hypothesis holds: 3... where N is the total number of layers. 2 (zl = x.4 The 3D equilibrium equations of a lcth layer. Integrate the equations over the thickness (zk. a..4) Multiply the equilibrium equations with z and integrate over the lamina thickness to obtain the third equation 3. o .5 (Continuation of Problem 3.. F2.6 Starting with a linear distribution of the displacements through the laminate thickness in terms of unknown functions (uo. a. ax3  . and assume that the material coordinates coincide with the plate coordinates. 2 . N and a . wo... vo. . P = 1 . . F2.
using the transverse shear stresses obtained in Problem 3.3. According to the firstorder theory.3.. (1.7 from the threedinlensional elasticity. 3.26)] in the absence of body forces: Integrate the above equations with respect to z over the interval (h/2. K. 3. orthotropic layer plate (Q45= 0).3.9 Consider the equations of motion of 3D elasticity [see Eq. the equation of motion governing the transverse deflection wo in the classical laminate theory is .. the strain energy due to transverse shear stresses is given by Compute U . the applied inplane forces induce bending moments for laminates with nonzero coupling coefficients [B]. 3.g.20~~). Use the following boundary conditions: Next.3.8 Consider a single. multiply the eqnations of motion with z and integrate with respect to z over the interval (h/2. h/2) and express the results in terms of the moment resultants defined in Eq.10 Show that the membrane strains { E O } and the moment resultants {M} in the classical or firstorder laminated plate theory can be expressed in terms of force resultants { N } and bending strains {&I} as These equations bring out the bendingextensional coupling for larriinates with nonzero [B]. h/2) and express the results in terms of the force resultants defined in Eq. to determine the shear correction coefficient. and equate it with U . for symmetric laminates). and assume that the material coordinates coincide with the plate coordinates. = 0 (e.20a). For exa~nple.11 Show that if B. (3. (3. when the bending strains are zero.
Assume that each layer is of thickness 0.18 The material properties of AS13501 graphiteepoxy material layers are: El = 140 x lo3 MPa.25.) and force resultants.25 mm each.. Interpret the results you obtain in light of the assumed strains.22.18 for the material properties). 3.28 Determine the displacement associated with the assumed strain field in Problem 3. symmetric laminate is subjected to loads such that the only = 103. show that the stiffnesses [A]and [Dl remain unchanged.19 is made of four layers (45145145145) of thickness 0.. 3. .BllrB22rB12. . Determine the state of stress (a. symmetric laminate is subjected to loads such that the only f = 1o3. EL:) 3.18 is heated from 20" to 90°. 3. 3. EL:) . Compare the stiffnesses B.) with respect to the laminate coordinates in each layer.24 Compute the stress resultants N's and M ' s for the problem in Problem 3.3 G13 = 7 x lo3 MPa.22 for the case in which the laminate is subjected to loads such that the only nonzero strain at a point (x. The material properties of a lamina are nonzero strain at a point (x.27 Compute the stress resultants N's and M's for the problem in Problem 3. Determine the state of stress (a.22 3. Z 6 Dl6. 3.14 Show that for antisymmetric laminates the stiffnesses. A Z 6B16.15 Show that for antisymmetric crossply laminates.Dl6..20 If the laminate of Problem 3. G12= 7 x lo3 MPa vl2 = 0. if it were restrained from free expansion./in. BZ2= B l l . 3.5 mm) layers. the coupling stiffnesses Bij have the properties: Bz2 = . y) is E!& = ((1112) /in.B I 1 and all other B. B . the AZ6. 3. BZ6.5 mm) layers of AS13501 graphiteepoxy layers (see Problem 3. the following stiffnesses are zero: A161A26.25 Repeat Problem 3. B16..005 in. y) is (typical of a graphiteepoxy material) El = 20 msi. and DZ6are zero.25..A26. and D22 = D l l . and 3. and DZ6 are zero. 3.22. and that Az2 = A l l . B66. for the two laminates (do they increase or decrease in values?).29 Suppose that a sixlayer ( 45/0).21 If the laminate in Problem 3. coefficients Bij are zero. determine the thermal forces and moments generated in the laminate.3. A I 6 . and DZ6 are zero. 3. E2 = 10 x lo3 MPa. a. 3.3. vl2 = 0./in. 3.u in.19 Determine the stiffnesses [ A ][ B ] and [Dl for an antisymmetric laminate (45145) composed .. GZ3= 7 x lo3 MPa. = 0.13 Show that for a general laminate composed of multiple specially orthotropic layers.23 Compute the stains and stresses in the principal material coordinate system of each layer for the problem in Problem 3. and [Dl for the antisymmetric laminate (0190) composed of equal thickness (0.u in. E2 = 1...22 Suppose that a fourlayer (0/90). and the coupling stiffnesses Bij are not zero. 3.30 msi. a.03 msi.ayya.. . Dl6. of equal thickness (0. G I 2= 1. Determine the stiffnesses [ A ][ B ] .16 Show that for antisymmetric angleply laminated plates..26 Compute the stains and stresses in the principal material coordinate system of each layer for the problem in Problem 3. the laminate stiffness A l 6 .y) is of a lamina are the same as those listed in Problem 3. laminate stiffness A16. . The thickness and material properties nonzero strain at a point ( 2 .25 3.12 Show that for a general laminate composed of multiple isotropic layers. 3.D16rD26.17 Show that for laminates ( a / P / P / a / P / a / a / P ) where 90" 5 a: 5 90° and 9 0 " 5 P 5 90°.
. References for Additional Reading 1. 357570 (1829)." Transactzons of the Royal Society of Canada.. "Elastic Wave Propagation in Heterogeneous Plates. Poisson." NACA TN1833. Suisse. B.) and stress resultants." Bull. G. 23.E. Y. 969975 (1962). S. Noordhoff. Reissner. E../in." Ing. y ) is ESP. "On the Theory of Bending of Elastic Plates. 433480 (1890). and Taylor. Reissner. C T (1969). Reissner. R... 32. 12. Vlasov. 28. 4 . 3138 (1951)." Izv. E. Mindlin.. R... (1949). "Uber das Gleichgwich und die Bewegung einer Elastischen Scheibe. Cheron.." J. "The Effect of Transverse Shear Deformation on the Bending of Elastic Plates. V. 2. Determine the state of stress (a. "Memoire sur l'equilibre et le mouvenlent des corps elastique. 40. 6588 (1938). Akd. Norris. "On the Problem of the Beam and the Plate in the Theory of Elasticity. 8. 7276 (1947). The thickness and material properties of a lamina are the same as those listed in Problem 3. K. 328355 (1828).. C. E. 16. 3.. J. Reissner. 73. Panc. Nauk SSR.3. y) is EL:) = (1/12) /in." Journal of Mathematical Physics.." Engineering Mech. and Stavsky. Stavsky. Pister. 5. A. Theories of Elastic Plates. Romande... 8(2). "On the Theory of Laminated Anisotropic Shells and Plates. 184191 (1944). 38(11). F. "Bending and Stretching of Certain Types of Aeolotropic Elastic Plates.. 4. H. B.. 3. Tech. 87 (EM6). C. 16. Boll&. Reissner. O T N . Dong." Journal of Applied Mechanics. 19. "Notes on the Foundations of the Theory of Small Displacements of Orthotropic Shells. L. B.. Arch. "Influence of Rotatory Inertia and Shear on Flexural Motions of Isotropic.. Kirchhoff. 13. "Bending and Stretching of Laminated Aeolotropic Plates. 10.. N. S. (London) Ser. Acad. A. 18. Ambartsumyan. Stamford. D.. translated from Russian by T . 9.. "Reflections on the Theory of Elastic Plates. 181 (6). Exercises de 2. Theory of Anisotropic Plates.." Journal of Applied Mechanics.." Journal of Aeronautical Science. 11. Journal of 17." Applied Mechanics Reviews. Washington. 665684 (1966). 3156 (1961). P. 29(8). Cauchy. Hencky. 5188 (1850)." Mem." Journal of Applied Mechan. The Netherlands (1975). "Uber die Berucksichtigung der Schubverzerrung in ebenen Platten. S. "Sur l'equilibre et le mouvement d'une plaque solide. Sci. E. Math. 6. Y. A. L..30 Repeat Problem 3. 402408 (1961). 281285 and 293298 (1947).. B. 7. Basset... 14. 102109 (1958). "Contribution au Probleme Lineare de Flexion d'une Plaque Elastique. Elastic Plates. and Stavsky. 6977 (1945). F. Yang. ASCE.. "Oh uravneniyakh teovii isgiba plastinok (On the Equations of the Theory of Bending of Plates)..31 Suppose that a threelayer (64510) unsymmetric laminate is subjected to loads such that the only nonzero strain a t a point ( z . u. 15. H.29 for the case in which the laminate is subjected to loads such that the only nonzero strain at a point (z.u..C. Leyden..zcs.22. ?3ansactions of ASME. 12. B. ." Mathematique.anics.. Hildebrand. G. E. S. Goodier. and Thomas. Technomic. 14531464 (1985).) = in. 18." Philosophical Transactions of the Royal Society. "On the Extension and Flexure of Cylindrical and Spherical Thin Elastic Shells. D. 3.. A. D. Y. Angew.." Internatzonal Journal of Solzds and Structures.
10311036 (1970).. A." Journal of Composite Materials. 10(5). 3.. "A General NonLinear ThirdOrder Theory of Plates with Moderate Thickness. Murthy. J. M. E. 18231824 (1977). V.. Reddy. Reissner. A.. Sept.. 20." Computer Methods i n Applied Mechanics and Engineering. J . Srinivas. 534547 (1969). 24.). Reddy.. 306319 (1971). Elishakoff and H." A I A A Journal. J. and Sun. K. Berlin. Noor. 302304 (1973). V. Krishna Murty. 8597 (1973). 261 . and Wu. and Whitney. R. and Reddy. 3213 (1987). Elastostatics and Kinetics of Anisotropic and Heterogeneous ShellType Structures.. J. "An Exact Analysis for Vibration of SimplySupported Homogeneous and Laminated Thick Rectangular Plat. Whitney.. 40. 36. 137 (1981)." Journal of Sound and Vibration. 22. Chow. M. 40(1). T. Joga Rao. Bert. Reissner. 31. A. John Wiley. Refined Dynamical Theories of Beams. N." Applied Mechanics Reviews. E. N. Librescu. "A Simple HigherOrder Theory for Laminated Composite Plates.. The Netherlands (1975). 26. "Higher Order Theory for Vibration of Thick Plates. Noordhoff. Whitney." NASA Technical Paper 1903. Plates and Shells and Their Applications. H..." A I A A Journal. 66:)676 (1977). and Burton. 7.. Christensen. Sun. J. New York (2002). 677686 (1990). "A Higher Order Theory for Extensional Motion of Laminated Composites. 37(4).. and Rao." International Journal of Solids and Structures. "An Assessment of Mixed and Classical Theories on Global and Local Response of Multilayered Orthotropic Plates. "A Consistent Treatment of Transverse Shear Deformations in Laminated Anisotropic Plates. Whitney.. 38. Kassel. Carrera. M." Journal of Sound and Vibration. C." Journal of Composite Mater. Librescu. 15(12). S.266 (1969)." Journal of Applied Mechanics. "Shear Correction Factors for Orthotropic Laminates Under Static Load. J. "Note on the Effect of Transverse Shear Deformation in Laminated Anisotropic Plates. 34. E. 41. S. E.." A I A A Journal. 183198 (2000). 525529 (1973).... 178183 (1973). S. 50. Wittrick." Euromech Colloquium 219. A." Composite Structures. 37. Reddy. Energy Principles and Variational Methods i n Applied Mechanics. J. pp. J.. 203209 (1979). SpringerVerlag. Part 2. 32. Irretier (Eds. V. 33. Leyden." Journal of Applied Mechanics. Germany. C. "On the Propagation of Flexural Waves in an Orthotropic Laminated Plate and Its Response to an Impulsive Load. 39. 1986. 5. J. "A Higher Order Theory of Plate Deformation. N. 113 (1989)." Journal of Applied Mechanics. 30. .es. M. M. 25(6). Whitney. "Shear Deformation in Heterogeneous Anisotropic Plates. Whitney. "Assessment of shear deformation theories for multilayered composite plates." Journal of Composite Materials. W. 44. 21. J. M. N. K. 745752 (1984). 1 2 . K. M. H. 29. 187199 (1970).. J." Journal of Applied Mechanics. and Leissa.. "Theories for the Dynamic Response of Laminated Plates. M." International Journal of NonLinear Mechanics.. 28. "A Critical Review and Generalization of Transverse Shear Deformable Anisotropic Plate Theories. 23. 36(2).20. "An Improved Transverse Shear Deformation Theory for Laminated Anisotropic Plates. "Analysis of Heterogeneous Anisotropic Plates. 30. W.. L. "Analytical ThreeDimensional Elasticity Solutions t o Some Plate Problems and Some Observations on Mindlin's Plate Theory. 716718 (1972). W. "The Effect of Transverse Shear Deformation in the Bending of Laminated Plates. 23. and Pagano. R. N. Laminated Plates. M. 51. 27.. C. "Simplified Analysis of Static Shear Correction Factors for Beams of NonHomogeneous Cross Section. V. 25. T. Lo." Journal of Applied Mechanics. I .. L. T.als... 42(1). W. 35. Second Edition. C. 11(2). 441464 (1987).
Finite Elenrent Anu. V. 0.. Washington. Theory of Laminated Plates.lysis of Composite Laminates. Structural Analysis of Laminated Aarsotropic Plates. Anisotropic. Technornic.es and Shells. 0. Reddy. 45. Marl. Kluwer. 51. 46. Jones. E. Newark. C T (1970). 8 7 140 (2002). R. D C (1988). PA (1987).res. and Reddy. V. . "Theories and Finite Elements for hlultilayered. Translated from Russian by S. Cheron. Selected Wor. and Sierakowski. G. Mechanics of Composite Structu. R. E. T h e Netherlands (1994). W . T h e Netherlands (1992).s J. "Developments. J . Mechanics of Composite Matersuls." Applied Mecha. 9 ( 2 ) . Vasiliev. Technonlic. M. 54(4). R.. 43... Gordon and Breach. 49. Pagano. Philadelphia. Ideas. Translated from Russian by L. Second Edition. N. M. Anisotropic Plates. a n d Evaluations Based upon Reissner's Mixed Variatiorml Theorem in t h e Modelirlg of Mu1t)ilayered Plates arld Shells.. 50. 44. Klilwcr. Laricaster. J. S. Composite Plat. 301329 (2001). Ochoa.ri. J. (Ed.ics Rewie~~ls. Taylor and Francis. 48. L.). M.."Archives of Computationml Methods in Engineering. I. N J (1968). 47. J.42. Whitney.. Lekhnitskii. and Whitney.. J . Mechamcs of Cornposzte Materials. N. T h e Netherlands (1986). Carrera. E. Kluwer. Stamford. Carrera. Ashton. Tsai and T . Taylor and Francis.. The Behavzor of Structures Conposed of Composite Materials. Virlson. J.ks of Nicho1a. PA (1999).
.
The solutions obtained with any of the variational methods (see Chapter 1) and numerical methods. .1 Introduction There are two cases of laminated plates that can be treated as onedimensional problems. A numerical solution is one that is obtained by satisfying the governing equations and boundary conditions of the problem in an approximate sense. the deflection wo and displacements (uo. the laminate is treated as a beam (see Figure 4. and (2) cylindrical bending of laminated plate strips. In such a case.. whereas the beam problem is a plane stress problem. An exact solution of a problem is one that satisfies the governing equations at every point of the domain and the boundary and initial conditions of the problem. and all derivatives with respect to y are zero.1. and boundary element methods. and loading is such that the displacements are functions of x only.1. such as the finite difference. finite element. the displacements are functions of just one coordinate: (1) laminated beams. An exact solution can be either Figure 4. The transverse load q is assumed to be a function of x only. the laminated plate is assumed to be a plate strip that is very long along the yaxis and has a finite dimension a along the xaxis (see Figure 4.1.2). i. In cylindrical bending.vo) of the plate are functions of only x . are termed numerical solutions.4 OneDimensional Analysis of Laminated Composite Plates 4.1).1: Geometry of a laminated beam. The cylindrical bending problem is a plane strain problem.e. When the width b (length along the yaxis) of a laminated plate is very small compared to the length along the xaxis and the lamination scheme. In this chapter we develop exact analytical solutions for the two classes of problems.
In addition. analytical .x 3x2 4sinn7rx is a closedform solution. The finitesum series solution uN(2) = n=l x N a.1. Since the series solution.1. in a sense. u(x) = 2 .2: Geometry of a plate strip in cylindrical bending.1) are included in (4. Closedform solutions are those that can be expressed in terms of a finite number of terms. sin n r x (4..exact as well as numerical solutions can be developed for a number of laminated beams and plate strips. sin nnx (4. is evaluated for a finite number of terms. The analytical solutions presented here for simple problems serve as a basis for understanding the response. it is "exact . is not a closedform solution because the number of terms in the series is not finite. For example.1. it is. although it is approximate because not all terms of the series (4.1. approximate.1. in reality. For all practical purposes. the results can serve as a reference for verification of computational methods designed t o analyze more complicated problems.2). whereas a solution in the form of a convergent series + + 0 0 u(x) = n=l a." Due to their onedimensional nature.1) where an are real numbers. closedform or an infinite series.  .Figure 4.2) will be termed an analytical solution.
vo) are zero.1 Governing Equations Here we consider the bending of symmetrically laminated beams according to CLPT.1). For angleply laminates this ratio must be rather large to make the twisting curvature negligible. we have where D. the equations for bending deflection are uncoupled from those of the stretching displacements. The classical laminated plate theory constitutive equations for symmetric laminates.2. when the lengthtowidth ratio is large). in inverse form.7j denote the elements of the inverse matrix of D i j . If the inplane forces are zero.These effects can be neglected only for long beams (i. or. assumption (4.2. The lengthtowidth ratio for which the transverse deflection can be assumed to be independent of y is a function of the lamination scheme.3a) indicate that the transverse deflection wo cannot be independent of the coordinate y due to the Poisson effect (DT2) and anisotropic shear coupling (DT6).. .4. For symmetric laminates.e.2. the inplane displacements (uo.2 Analysis of Laminated Beams Using CLPT 4. in the absence of inplane forces. In deriving the laminated beam theory we assume that everywhere in the beam. we have In view of the where Equations (4. are given by [see Eqs. and the problem is reduced to one of solving for bending deflection and stresses.
2. (4. we have where I ? . (4. f2 = b12. (3.3.In the following derivations we assume that the laminated beam under consideration is long enough to make the effects of the Poisson ratio and shear coupling on the deflection negligible.5) In order t o cast Eq. . M . 2 p(z)' dz ( i = O.2. Then the transverse deflection can be treated only as a function of coordinate x (along the length of the beam) and time t: Then we can write =  a2wo ax2 DTl M. I. for symmetrically laminated long beams.  aw0 ax  dM dx . = b l . The equation of motion of laminated beams can be obtained directly from Eq. and h  4 = bq..25) by setting all terms involving differentiation with respect t o y to zero: or.2. is the applied axial load. fo = bIo.5) as and the shear force and bending moments are related by where b is the width and h is the total thickness of the laminate. (4.1. we introduce the quantities and write Eq.2) The boundary conditions are of the form Geometric : Force : specify specify wo Q .5) in the familiar form used in the classical EulerBernoulli beam theory.
(4. the form [cf. use of Eq.2.b) are obtained by direct integration. 4.4.4513).2. and cl through c4.1 for the sign convention] where q = bq.813) take . to those of the EulerBernoulli beam theory of homogeneous. = 0 o dx (4. the solutions available for deflections of isotropic beams under various boundary conditions can be readily used for laminated beams by replacing the modulus E with E!& and multiplying loads and mass inertias with b.2.8ac).. (1. b2.3.2.9) are identical. dx M=O w =0 . Hence. in form. General Solutions The general solutions of Eqs.2. Eqs.10a.2 Bending For static bending without the axial force.Equations (4. bl.2. For indeterminate beams.4713) and (1. (4. Equation (4.4.2b)l .10a) and from Eq. isotropic beams. (4.2.7a) and (4.10a) is the most convenient when it is possible to express the bending moment M in terms of the applied loads.10b) The constants of integration. We obtain from Eq.12a) and (4.2. Eqs.2. N ~ = 0.7)(4. The boundary conditions for various types of supports are defined below: Free: Simply Supported : Clamped : dill QE=O.2.see Figure 1. can be determined using the boundary conditions of the problem. (3.2. Note that the rotary (or rotatory) inertia I2is not neglected in Eqs.2.2. (4.4. M =0 o dwo w = 0 .11~) Calculation of Stresses The inplane stresses in the kth layer can be computed from the equations [see Eqs. (4. (4.10b) is more convenient.
In the classical beam theory..27)]: For each layer. zk+1): where constants. They are not valid especially in the freeedge zone. o&). the maximum stress does not occur at the top or bottom of a laminated beam.12). The stresses given by Eq. the interlaminar stresses (a.1213) are approximate for the purpose of analyzing laminated beams. (1. (4. (4. and H(') are . the 0" layers take the most axial stress. and ~ ( ~ F(". As will be seen later in this section. However. 1 . where the stress state is three dimensional. (ai:). these stresses do exist in reality.2.3. these equations may be integrated with respect to z to obtain the interlaminar stresses within each layer (zk 5 z L..2.. The maximum stress location through the beam thickness depends on the lamination scheme. a. and they can be responsible for failures in composite laminates because of the relatively low shear and transverse normal strengths of materials used.og))are known from Eq.) are identically zero when computed using the constitutive equations.In general. Interlaminar stresses may be computed using the equilibrium equations of 3D elasticity [see Eq. The width of the edge zone is about the order of the thickness of the beam.
For layer 1.. the constants should be such that a.. and a.12b). Hence. which are evaluated using the boundary and are interface continuity conditions. For example. The ( constants G(" and H(" for k = 2 . equal the shear and normal stresses at the bottom face of the laminate.14a.2. from Eqs. we have G(') = 0 and H(') = 0. if the laminate bottom is stress free.7b) are used to replace d M / d x with Q = bQ. .. all variables are independent of y and v = 0. .1): ai'C.2.2.1: (a) Sign convention.2.c) and (4. (4.. are determined by requiring that a. . derivatives with respect to y are zero.2. =0 Sign convention Figure 4. and G ( ~ ) and H ( ~ ) the integration constants. (b) Equilibrium of interlaminar stresses in a laminated beam. we obtain where Eqs.For beams. x .2. 3 .. (4. For example.k ) and be continuous at the layer interfaces (see Figure 4.) L a  +q . .6) and (4.
This gives.. for k = 1..2.2. The deflection is symmetric about the point x = a/2. The expression for the bending moment is a M(X) = . . The distributions are described by different functions in different layers but they are continuous across layers.2. (4.1 (Simply supported beam): Consider a simply supported beam with a center point load (see Figure 4.2. (4.2). the result Note from Eqs. This case is known as the threepoint bending. for 0 5 % (F$)x 5 2 Substituting this expression into Eq. cz = 0) The deflection is the maximum at x = a/2. which is given by This expression can be used to determine the modulus of the material in terms of the measured center deflection w. ..15a. is cubic through the thickness of each lamina.b) that the transverse shear stress a.. occurs at x = a/2 (M(a/2) = Foba/4) . we obtain The constants cl and c2 are evaluated using the boundary conditions of the problem We obtain (el = Foba2/16.11a) and evaluating the integrals. applied load Fo.. . Example 4.and the geometric parameters of the laminated beam in a threepoint bend test: The maximum inplane stress a.2. is quadratic and normal stress a.
2.2. For the full beam case we have and for the half beam model we have Either set of boundary conditions will yield the same solution.2. q = qo (see Figure 4.3).la for the sign convention).2. Example 4. clamped at both ends.2: Threepoint bending of a laminated beam (see Figure 4. cs = c4 = 0) .11b) the result The constants cl through c4 are evaluated using the boundary conditioris of the half (because of the symmetry) or full beam.Figure 4. c 2 = qoba2/12.2. (4. We have from Eq. and subjected to uniformly distributed load acting downward. We obtain (el = qoba/2.2 (Clamped beam): Consider a laminated beam. The deflection is symmetric about the point x = a / 2 .
The maximum deflections and bending moments are also listed (note that the loads are assumed to be applied in the downward direction). The deflection is the maximum a t x = a/2.6) and (4.2.. When both point load and uniformly distributed load are applied simultaneously.e. clamped edges.1...ributed load. a: .3: Clamped beam under uniformly dist.2. (4.2. Expressions for other boundary conditions can be found in textbooks on a first course in reformable solids.Figure 4. and clampedfree (cantilever) supports and subjected to a transverse point load or uniformly distributed load are presented in Table 4. I. Recall that wo(x) is taken positive upward and M ( x ) is positive clockwise on the right end. occurs a t x = 0. . Expressions for the transverse deflection of laminated beams with simple supports. which is given by The maximum bending moment. adding) the expressions corresponding to each load. and hence the maximum inplane stress a . as can be seen from Eqs.313).2. = b/DT1. The effects of material properties and stacking sequence are accounted for through the bending stiffness Ek. the solution can be obtained by superposing (i.
..Table 4. and "0" refers t o x = 0. wo(.c) w. The constants in the expressions for the deflection are defined as .2... 0 HingedHinged Central point load Uniform load 0 FixedFixed Central point load Uniform load 0 FixedFree Point load at free end Uniform load Superscript "c" refers t o the center (at x = a / 2 ) .1: Transverse deflections of laminated composite bcarrls with various boundary conditions and subjected to point load or uniformly distributed load (acting downward) according to the classical beam theory. and w.. Laminated Beam Deflection... "a" to the end x = a .
26).2.2.8b).0. it results in a large deflection and the beam seeks another equilibrium configuration. subjected to threepoint bending (Fo = 1.25 (4. shortens as the load increases from zero to a certain magnitude.2. in proportion to their axial stiffness. . h = 0.3 Buckling A beam subjected to axial compressive load N ~ = N& remains straight but .12b). then the beam is said to be stable. h = 0.5E2.7 show the effect of stacking sequence on maximum transverse shear stress. as predicted by Eq. The magnitude of the compressive axial load at which the beam becomes unstable is termed the critical buckling load. respectively. (90" corresponds to outer layers) laminated beams. The equation governing buckling of laminated beams is also given by Eq. the beam is said to be unstable.] . The parabolic distribution of transverse shear stress through an orthotropic beam is shown in dashed lines for comparison. 2 . Here we determine critical buckling loads for laminated straight beams.0. and it is termed buckling deflection. through the thickness of (0/45/45/90).b = 0.. If the load is increased beyond this critical buckling load.1). It is clear the 0" layer carries the most axial stress while the 90° layer carries the least axial stress.2.8). (4. where w is the original equilibrium (prebuckling) 6 by substituting w = w satisfies the equation deflection and W is the buckling deflection. If a small additional axial or lateral disturbance applied to the beam keeps it in equilibrium.Figures 4. Equation (4.5 show the maximum normal stress distribution. The maximum stress value is dependent on the stacking sequence and considerably different from that in a homogeneous beam.2.0. Note that wfj + [The reader is asked to verify the result in Eq. ~ 1. the deflection is measured from onset of buckling. Figures 4. (4. and axial force is assumed to be unknown. q = 0. (0' corresponds to outer layers) and (90/45/45/0). (4. for laminates (0/45/45/90). the load at which a beam becomes unstable is of practical importance in the design of structural elements. a = 1.2E2. E2 G I 2 = GI3 = 0.2. respectively (Fo = 1.1). Thus.2.8b): we obtain the equation where W denotes the buckling deflection. wherein the applied transverse load and inertia terms are set to zero.6 and 4.25) The maximum normal stress distribution in an orthotropic beam (with eight 0" layers) is shown in the figures by dashed lines. (4. The following layer material properties are used = (E2 = 1 msi): 2 = 25.4 and 4.2.2. The onset of instability is called buckling (see Figure 4. and (90145145/0).2. as predicted by Eq. v = 0.2. and all inertia terms to zero in Eqs.2.. 4. (4.2.26) is obtained from the nonlinear equilibrium equation o 6 W. b = 0 .15a). In addition. GZ3= 0. If the small additional disturbance results in a large response and the beam does not return to its original equilibrium configuration.0. Setting N ~ = N:.2.
(a/2.1 8 ill 0..5 I I I I ~ I I I I I I I ' I I ~ 400 300 200 100 0 100 200 300 400 Stress. distribution through the thickness of a symmetrically laminated ( 0 1 545/90).2. a..4:  1 : I I I I  0. . distribution through the thickness of a symmetrically laminated (go/* 45/O). ( a 12.(a/2. beam.3 Figure 4.Orthotropic 4 2 0 0.z ) 0 Figure 4. z ) ..2. 0. z ) . beam.0 0.4: Maximum normal stress. a.5: Maximum normal stress. .
0. beam subjected to threepoint bending (Fo = 1. b = 0 . 2 .0 1 2 3 4 5 6 7 Transverse shear stress.1).) through the thickness of a symmetrically laminated ( 0 / f 45/90). ~ ) ( 8 Figure 4. = 1. ~ Figure . h = 0. oXZ0 ..2.6: Variation of transverse shear stress (a.0.
26) twice with respect to x. c ~cy. The buckling shape (or mode) is given by W(x).2. is the critical buckling load.27) is W(x) = c~sin Xbx where + c2 cos Xbx + C ~ X c4 + (4. In the following.. . of the beam. we obtain The general solution of Eq.2. We are interested in determining the values of Xb for which there exists a nonzero solution W ( x ) .(a) Simply supported beam (b) Clamped clamped beam (c) Clamped free beam Figure 4.2. Once such a Xb is known (often there will be many). i. the buckling load is determined from Eq. (4. which is given by the smallest value of Xb.8: Buckling of laminated beams under various edge conditions. (4. when beam experiences deflection. (4.29): The smallest value of N t z .2. and c4 can be determined using the boundary conditions . we consider beams with different boundary conditions to determine Xb and then the critical buckling load for each beam.28) and the constants cl. Integrating Eq.e.2.
= 0 which implies cz = 0. .4 (Clamped beam): When the beam is fixed at both ends. 2 .= 0 (0) dx2 dx2 We have W(O)=O: ~ " ( 0=0 : ) c2+c4=0  c2X. W ( a ) = O ..3 (Simply supported beam): For a simply supported beam. n = 1 .180 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS Example 4.2.= 0. the boundary conditions are which can be expressed as We have c2 + Cq =0 clXb cg = 0 cl sin Aha c2 cos Xba cga + c4 = 0 clXb cos Aha . the condition cl sin Aha = 0 implies that Aha = nx. q = 0 W(a) = 0 : cl sin Aha + csa = 0 ~ " ( a = 0 : cl sinXba = 0 which implies c3 = 0 ) For a nontrivial solution. . cl # 0 a The critical buckling load becomes (n = 1) and the buckling mode (eigenfunction) associated with it is W(x) = cl sin a 'Tx Example 4. and the buckling load is given by b ~ = E:xIyy & (?) 2 The buckling mode is nrx W(x) = cl sin .cZXb Aha+ cg = 0 sin + + + .2. the boundary conditions are These boundary conditions imply d2W d2W (a) W(0) = 0 .
. i.) = e. 0..9868.2. (4.. for the critical buckling load of a cantilever beam (i. fixed at one end and free at the other end).2. 6 ~ .2. sine. . 4 5 0 5 .2832(= 27r). A plot of 2cose.2. nonlinear equation involving trigonometric functions. 2 against e .cz sin Xba =O  + c2 (cos Xba For a nontrivial solution. and cq for several combinations of simply supported (hinged). known as the characteristic equation. clamped (fixed).30)] +  Table 4. M. the determinant of the coefficient matrix of the above two equations must be zero (eigenvalue problem): sin Xba .1) . Equations (4.. ca.38b) is a transcendental equation.e. gives the eigenvalues en = & a .6.3813) T h e solution of equation (4..30). which are equivalent to dx (0) = 0.la = 2 n ~ ) Hence. 8. smallest) buckling load is [see Eq.Expressing these equations in terms of constants cl and ca. the boundary conditions are wo(0)=0.5664(= 47r). and freeedge conditions. For example. dwo Q. and the buckling load is calculated from Eq.e. with some typical values. = the function f (e. . cg. 1 5 .&a) 1) =O cl (cos Xba .2 contains governing equations for Xb.12. .2.2. and values of the constants el. (4.38b).e.1 sin Xba  =Aha sin Xba + 2 cos Xba  2 (4.. (a) = 0 The critical buckling load is given by .Xba cos Xba 1 cos Xba . shows that f ( e n ) is zero a t e. (Xz. (a) = 0. we obtain cl (sin h a . the critical (i.
(Ie~ /? .7. 4 .2. the governing equation (4.2.2: Values of the constants and eigenvalues for buckling of laminated composite beams with various boundary conditions ( x ~ = ~ N ~ ~ / E ~ . *For critical buckling load. = . sin en = 2(1 . . = nrr HingedFixed c1 = l / e n cos e.cos en) e. ~ . cg = 1 cz = cq = 0 tan en = en e.43) has the general form where . the solution is assumed to be periodic In the absence of applied transverse load q.8b) reduces Equation (4.493. Characteristic equation and values* of en = X. 9 8 7 .2. = nrr End conditions a t x=Oandx=a HingedHinged FixedFixed e.28): W ( s ) = cl sin Xbz c2 cos Xbz c32 cq. = 4.2.2. = 0 e.) ~ )The classical laminate theory is used.a sine.~ . only the first (minimum) value of e = Xa is needed. t See Eq.725. FixedFree 1 FreeFree sin en = 0 e. .Table 4.. = 2 ~ .4 Vibration For natural vibration. + + + 4. (4. 8 .
a. the reader is referred to Table 4. . and r from Eq. The smallest frequency w is known as the fundamental frequency.2.2. c2.2. c3.2..2. (4.46)(4.46~~) and cl. From Eqs.45) into Eq.b) are the same and hence either one can be used to calculate the frequency once X is known.2.48) can be used to find w.47a. we have X = p and In the following discussion beams with both ends simply supported or clamped are considered to illustrate the procedure to evaluate the constants el through c4. For other boundary conditions. (4. (4. For boundary conditions other than simply supported. which are to be determined using the boundary conditions.The general solution of Eq.44) is W(x) = cl sin A x + c2 cos Ax + CQ sinh px + c4 cosh px (4.3. the frequency of vibration can be calculated from It is clear from the first expression that rotary inertia decreases the frequency of natural vibration. If the rotary inertia is neglected.. one must solve a transcendental equation for e. we obtain The two expressions for w in Eqs.2. (4. (4. (4.46b).2. and c4 are constants. q.2. and more importantly. to determine X so that Eqs.48a. When the applied axial load is zero. = X. we have Substituting for p.b) and solving for w2.
069. the second term under the radical in Eq.) c2 = ~4 = l/(cosen . rotary inertia is used. e.coshe. which implies X = a Substituting for X from Eqs. . .2.) cl = c3 = l/(sin e.069. we have 7 . say a cable under large tension. (4.2..1 = 0 e.) cl = c3 = l/(sin en .a sin e.46a): W(x) = cl sin Xx + c2 cos Xx + cs sinhpx + c4 coshpz.1 = 0 e. (4.730. c2 = c4 = l/(cose.2. = 4.~/ ~ ) ~ )The classical laminate theory wzthout ~ e ~ ..2. cos en cosh e.sinh e. . cosencoshe.853.2.. we obtain If the rotary inertia is neglected. if n is not large. . FixedFixed FixedFree FreeFree HingedFixed HingedFree nlhr t  + sinh en) + coshe.694..5 (Simply supported beam): For a simply supported beam.48a).. See Eq.is to increase the natural frequencies.53b) becomes very large in comparison with unity.7.) ~2 = ~4 = 1/(cos en . tan en = tanh en e.3: Values of the constants and eigenvalues for natural vibration of laminated composite beams with various boundary conditions (A: G W ~ I ~ / E = . .cosh en) tan en = tanh e. (4. = 3. (4. = 3.730. .45) and (4. we obtain Thus the effect of the axial tensile force NZ.2. End conditions at z=Oandx=a HingedHinged Characteristic equation and values of en = X. If we have a very flexible beam.Table 4.46a) into Eq.( I .2. (4. Example 4.927.7..927.. = 4.7.2. the boundary conditions in Eq. .sinhe.875.31b) give C2 = C 3 = C4 = 0 nr cl sin Xa = 0. = 0 eTL nr = cl = cg = l/(sine.853. + 1 = 0 en = 1.4.7. cos en cosh en ..
In Table 4. Thus. We also note from Eq. the first row corresponds to deflections due to point load Fo. If the rotatory inertia is neglected.4.56) are used to eliminate c3 and c4. where qo is the intensity of the distributed load (forcelunit square area).2.e. When N ~ = 0.2. and cantilever (clampedfree) boundary where Fo is the line load across the width conditions. (4.48~~) (4.59) Equation (4. frequencies corresponding to a l h = 100 . (4.6 (Clamped beam): For a beam clamped a t both ends. Also. we obtain Example 4. we require the determinant of the coefficient matrix of the above equations t o vanish. which yields the characteristic polynomial 2 + 2 cos Xa cosh p a + (P  $1 sin hasinh p a = 0 (4.5313) that frequencies of natural vibration decrease when a compressive force instead of a tensile force is acting on the beam. Eq.2. For natural vibration without rotatory inertia and applied inplane force (i.2. the boundary conditions in Eq. and the second row corresponds to deflections due to uniformly distributed load go.48b).( : sinh p a ) where relations (4.36) lead t o and the eigenvalue problem sinh p a cos Xa  cosh pa cos Xa . rotatory inertia decreases frequencies of natural vibration.sin Xa .2.2.2. (4. For nonzero cl and c2. are presented in Table 4.2.2.2. 9 = 0 in Eq. clamped (fixedfixed). critical buckling loads.53~~) .58) takes the simpler form cos Xa cosh Xa  1=0 (4. (4.2. we obtain from Eq.4 for simply supported (hingedhinged). and the distributed line load along the length is gob.2.59) is satisfied for the following values of A : Maximum transverse deflections. if the applied axial force is zero.4613) and X = p ) . of the beam (forcelunit length). according to the classical beam theory. Eq. on the second and third rows. (4.2..49) can be or used to calculate the frequency of vibration. In the case of bending. (4. and fundarnental natural frequencies of various laminated beams. Then the natural frequency of vibration can be calculated from Eq.2.2.cosh p a . the point load is Fob. (4.which are natural frequencies of a stretched laminated cable.58) The solution of this nonlinear equation gives X and p.
. and laminate B is stiffer than laminate C. All other frequencies were computed by neglecting the rotary inertia. the bending stiffness increases with (cube of) the distance of the 0' layers from the midplane. it has smaller deflection and larger buckling load and natural frequencies when compared to the (90/0). (9010)s (451 45)s  Laminate A Laminate B Laminate C Laminate A = (0/+45/90). Also.W N w  w  N w  90 (0/90). G23 = 0. G12 = G13 = 0. ~ 1= 2 0. Similarly. beams.4: Maximum transverse deflections. Since the 0' laminae are placed farther from the midplane in (0/90).5E2. Laminate C = (90/*45/0)s.2E2.and a l h = 10 are listed when rotary inertia is included. the 0'laminated beam is stiffer in bending than the 90'laminated beam. Table 4. Symmetric angleply laminated beams ( Q I Q ) .2. laminate A is stiffer than laminate B.. . due to the placement of the 0' layers. critical buckling loads.. Laminate B = (45/0/45/90). and therefore. Thus. ClampedClamped ClampedFree HingedHinged Laminate 0 w N w  . laminate. and fundamental frequencies of laminated beams according to the classical beam theory (E1/E2= 25.have the same stiffness characteristics as (O/Q). The following nondimensionalizations are used: The stiffness in a laminate is largest in the fiber direction because El > E2. and they are less stiff compared t o the symmetric crossply laminated beams.25). 0" beam has smaller deflection and larger buckling load and natural frequencies when compared to the 90' beam.
Except for second and third rows..4. and ATj.2.22)] or.2. (3.4.2.3 Analysis of Laminated Beams Using FSDT 4.We note that for clampedclamped and clampedfree beams.3) 4. From the frequencies listed in rows 2 and 3 of Table 4.3.ary inertia. 6 ) denote the elements of the inverse of [Dl. = A44 A45 g . As in Section 4.. 4.2.s) denote elements of the inverse of [A]: A.3. all other frequencies listed in the table were calculated by neglecting the rotary inertia. the calculation of natural frequencies require the solutions of transcendental equations for A. (4.2.4. FSDT is known as the Tzmoshenko beam theory. The governing equations can be readily obtained from the results of Section 3. are given by [see Eqs. = A55 A . A. For the case where rotary inertia is negligible. we assume that M y y = Mxy = Q y = 4y = 0 and both wo arid are functions of only x and t: .3 arc applicable. 2 . in which case the values of A1 given in Tablc 4. To see the effect of rot.21) and (3.58) were solved for X and the frequencies were calculated.. D. ( i . the roots of these equations are given in Table 4. ( i . j = 4. Eq. it is clear that the effect of rotary inertia on fundamental frequencies is negligible for small lengthtoheight ratios.3. A& = A .4. in the absence of inplane forces.1 Governing Equations Here we consider the bending of symmetrically laminated beams using the firstorder shear deformation theory. A = A44455  Add45 (4.j = 1 . When applied to beams. we have where K is the shear correction coefficient. The laminate constitutive equations for symmetric laminates. in inverse form.
I2= b12 Note that when the laminated beam problem is such that the bending moment M ( x ) and Q(x) can be written readily in terms of known applied loads (like in statically determinate beam ~roblems).. and then wo can be determined using Eq. 4.7b).3. (4. the following relations prove t o be useful. .3.7) in Eq. I. (3. (4. (4. h3 b a4x b 12 The equations of motion from Eq.2a.8).In the latter case. Eqs.2 Bending = bIo.9alb) are used to determine wo(x) and dX(x).b) we have E X X I Y Y ( x). (4.3.1) the displacement field takes the form (when the inplane displacements uo and vo are zero) and the linear straindisplacement relations give From Eqs.4.3. When M ( x ) and Q(x) cannot be expressed in terms of known loads.7a) can be utilized to determine 4. (4.188 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS From Eq.3.13) are Using Eq. the equations of motion can be recast in terms of the displacement functions: where 6 = bq. M ( x ) = bMxxl Exx= =M~ D.3.4. (4.. (3. Eq.3.
10b) and integrating with respect to x. we obtain Substituting for 4(x) from Eq.3.10a) with respect to x.3.13) that the transverse deflection of the Timoshenko beam theory consists of two parts. (4. (4. we obtain Substituting the result into Eq.9a. (4. (4.12b) into Eq. It is informative to note from Eq.3.3. Eqs.3. (4. (4. one due to pure bending and the other due to transverse shear: where .11).b) reduce to Integrating Eq.For bending analysis. we arrive at where the constants of integration cl through c4 can be determined using the boundary conditions of the beam.3.
(4.3. is independent of transverse shear stiffness).7b). However.7a)]. and therefore the axial stress.3. Consequently.(a/2) = 0.7a) and integrating with respect to x. For this case.1 (Simply supported beam): Here we consider the threepoint bending problem of Section 4. into Eq. and the Timoshenko beam theory solutions reduce t o those of the classical beam theory. Substituting for 4. (4.b) for transverse shear stresses derived from 3D equilibrium are also valid for the present case.3. In fact. (4. This implies that $.17) that the rotation fiinction &(x) is the same as the slope dwo/dx from the EulerBernoulli beam theory (i. we obtain By symmetry.16) for M in Eq.2.2 (see Figure 4. 4.12b)l. (4. the bending moment [see Eq. the shear deflection wi(x) goes to zero. (4.. (4.17)] and shear forces are Fo . 4. (4. (4.36).2).e. These relationships enable one to obtain the Timoshenko beam solutions from known classical beam solutions for any set of boundary conditions (see Problems 4. We have Example 4. When the transverse shear stiffness is infinite.33 and 4.O < z < M (x) = 2 dz 2 2 Using Eq. ( 4 2 l b ) ] .3. for general statically indeterminate beams. one can establish exact relationships between the solutions of the EulerBernoulli beam solutions and Timoshenko beam solutions (see [27291).The pure bending deflection wi(x) is the same as that derived in the classical beam theory [cf.2. the bending moment [see Eq. The expressions for inplane stresses of the Timoshenko beam theory remain the same as those in the classical beam theory [see Eq. is independent of shear deformation. is independent of shear deformation for all statically determinate beams and indeterminate beams with symmetric boundary conditions and loading (see Wang [27]). Hence and the solution becomes It is interesting to note from Eq. u = u l o + z4.2. The expressions given in Eqs. The transverse shear stress can also be computed via constitutive equation in the Timoshenko beam theory. Eq. the rotation 4.bh (see Problem 4.= ~ bx a dM F b . is zero a t x = a/2. Q ( Z ) = . we obtain .11). In fact.3.3. will depend on the shear stiffness KG:.15a..2.
= bh3/12) (4./Gi. in contrast to the classical beani theory. . The effect of shear deformation is negligible for thin and long beams.3.18)].5b)l Note from Eq.Iv.3.3.2 (Clamped beam): Consider a laminated beam fixed at both ends and subjected to uniformly distributed transverse load qob as well as a point load Fob a t the center.) is the same as that obtained in the classical beam theory [cf.2. as well as the ratio of thickness to length h. Note that the first part (211.3. (4..4 (z) 2 ] = ( b z(z) In light of Eq. For this case.ds 16E$. a t the expression is zero at x = a/2. the first part of Eq.3. the boundary conditions are (using half beam) which in turn imply that . we arrive where the constant of integration is found to be zero on account of the boundary condition wo(0) = 0. dwildx = 4./a.ONEDIMENSIONAL ANALYSIS OF LAMINATED COMPOSITE PLATES 191 Let us denote the first expression in (4..22) shows that the effect of shear deformation is to increase the deflection. The maximum deflection occurs at x = a/2 and it is given by Equation (4.3.18a) can be viewed as the slope (or rotation) due to bending arid the second one due to transverse shear strain: Indeed.3. both acting downward.14a). dwildz can be interpreted as the transverse shear strain [cf. (4. (4. Eq. the slope dwo/dx a t the center of the beam in the Timoshenko beam theory is nonzero.3. (4.18a) by dw:. [l .20) However. (4. (4. We have (Iu.18a) with respect to x. The contribution due to shear deformation to the deflection depends on the modulus ratio E$. Integrating Eq. Eq.18a) that.3. Example 4.
2.3. (4. By comparison to the classical theory (see Table 4.3. it is clear that the shear deformation increases the deflection.2 contains maximum transverse deflections t of various laminated beams according to the Timoshenko shear deformation beam theory.9a.3.23)) where S is the positive parameter that characterizes the contribution due to the transverse shear strain to the dis~lacement field Table 4. (4. The material properties of a lamina are taken to be those in Eq.. 2Ti = ~ r n r n z E ~ h ~ / ~ = a ~ (Po ~ qoa).2.e.. the inertia terms and the applied transverse load q in Eqs.1 and 4. Thin or long beams do not experience transverse shear strains. Eq.2. For buckling analysis.25).1 contains expressions for transverse deflections and maximum transverse deflections of laminated beams according to the firstorder shear deformation theory. The effect of shear deformation on maximum deflection can be seen from Figures 4. i Table 4. accounting for the transverse shear strain).3.1)..3. Clamped beams show the most difference in deflections due to transverse shear deformation (i. The effect of shear deformation is more significant for beams with lengthtothickness ratios smaller than 10. The effect of lengthtoheight (or thickness) ratios of the beam on the deflections can be seen from the results. where the nondimensionalized maximum deflection. respectively. (4. 4 3 3 Buckling .: = .2.The solution is The maximum deflection is at x = a / 2 and is given by [cf.b) are set to zero to obtain the governing equations of buckling under 0 compressive edge load Nxx Nx. of a simply supported beam is plotted as a function of lengthtoheight ratio a / h for various laminated beams under a point load and uniformly distributed load.
Deflection HingedHinged Central point load F" Uniform load FixedFixed Central point load Uniform load FixedFree Point load at free end Uniform load . wo ( x ) Max. Laminated Beam Deflection.3.Table 4.1: Transverse deflections of laminated composite beams with various boundary conditions and subjected to point load or uniformly distributed load (acting downward) according to the shear deformation theory.
~~x(x) = (~~.25).29) to obtain the result The general solution of Eq. The deflection is nondimensionalized as w = .3.2: Maximum transverse deflections of laminated beams according to the Timoshenko beam theoryt (E1/E2 = 25. w .Table 4. ul2 = 0. which must be evaluated using the boundary conditions.2810) with respect to x and substitute for dX/dx from Eq. (4. HingedHinged Laminate t ClampedClamped 10 100 20 10 ClampedFree 100 100 20 20 10 t ~ h first row of each laminate refers to nondimensionalized maximum deflections under point load e (Fob) and the second one refers to rnaximum deflections under uniformly distributed load (gob).3. G23 = 0.5&. (4.3.3.32) and cl through c4 are constants of integration.3.bh b~:)) dW + K~ dx  (4.. a~ Solving Eq.31) is W(x) = cl sin Ax where + c2 cos Xx + c3x + cq (4.30) Next differentiate Eq.. (4. . (4.28a) for dX/dx one obtains Integration with respect to x yields KG.2E2.3. ( ~ ~ h ~ / ~x ~l o 2 ()F o = qoa). G12 = GI3 = 0.3. . .
.1: Transverse deflection versus lengthtothickness ratio ( u l h ) of simply supported beams under center point load. d h ( 3 w ) Figure 4. 0 10 20 30 40 50 60 70 Sidetothickness ratio. 0.3. d h 80 90 100 Figure 4.00 0 10 20 30 40 50 60 70 80 90 100 Sidetothicknessratio.3.(0145/45/90).2: Transverse deflection (a) versus lengthtotliickrless ratio ( u l h ) of simply supported bearm under uriiforrrily distributed load.
(4.30) can be shown (see Problem 4.31a)l In view of Eq. and for cl # 0 the requirement (4.3.3.29).3. the boundary conditions are W(0) = 0 . c1sinha+c2cosha+csa+cq=O  (1  a) KG!&bh ( h q cos ha  hc2 sin ha) . Example 4.).3. W ( a ) = O .33) for N L .bh (Y) 2 + EkzIvy (?) I The critical buckling load is given by the minimum (n = 1) It is clear from the result in Eq.3. dx2 (a) d2W =0 (4.35) sin ha = 0 implies Xa = n. (4. (4.4 (Clamped beam): For a beam fixed at both ends.3.30). noting that and then setting the determinant of the resulting algebraic equations among cl and c2 to zero.3.3.38) In order to impose the boundary conditions on X.34b) The boundary conditions in Eq. we use Eq. (4.3.2. the boundary conditions are [see Eq.3.. x ( a ) = O (4.35)].2. Eq. The constant K l appearing in Eq. (4.3.3. we obtain =EizIYy (:12 [I ELIYY KG$. the above conditions are equivalent to W(0) = 0 .3 (Simply supported beam): For a simply supported beam.Example 4. we obtain .3413) lead to the result c2 = cy = cs = 0.35) into Eq. The boundary conditions yield cz+cq=O. (4.c j = 0 Expressing cl and cz in terms of cy and c4.rr Substituting for A from Eq.37) that shear deformation has the effect of decreasing the buckling load [cf. (4. (0) d2W dx2 = 0. W ( a ) = O . (4. (4.10) to be equal to K1 = cs(bN&. x(O)=O.
c2.4 Vibration For natural vibration. 4.3.3. Equations (4.3. Note that we have . and cq are constants.3.3.3.39).b) take the form We use the same procedure as before to eliminate X from Eqs.40a.3. (4. the buckling load can be readily determined from Eq. which are to be determined using the boundary conditions. we have Substitute the above result into the derivative of Eq. we assume that the applied axial force and transverse load are zero and that the motion is periodic. (4. (4. From Eq.3.33). (4. (4.Once the value of Xu is determined by solving the nonlinear equation (4.9a.b).40a).4013) for d X / d x and obtain the result where The general solution of Eq. c3.42b) is W ( x ) = el sin Ax where + c2 cos Ax + c3 sinh px + cq cosh px and cl.
(4.3. When the rotary inertia is neglected. (4. (4. we have P = O and the frequency is given by Example 4.3. Eq.3. S l l c l .Alternatively. which then can be integrated with respect to x to obtain an expression for X .3.! = c3 = c4 = 0 and cl sin Xa = 0.3. (4. ~ ~ ) . Example 4. (4.S l l c 2 .5 (Simply supported beam): For a simply supported beam.3. = a nT (4.3. (4. we obtain . with W given by Eq. and setting the determinant of the resulting equations among cl and cz to zero (for a nontrivial solution).48) Substitution of X from Eq.= x (w sz2 &" +2~~q.S 2 2 ~ q 3 =O where sll = (fOw2  X ~ K G : . c1 sin Xa + c2 cos Xa + c3 sinh pa + cq cash pa = 0 (4.3. When the rotary inertia is neglected. (4. (4.46a).47) and the result into Eq. (4. The fundamental frequency will come from Eq. (4.40a) and expression (4. and therefore the frequency given by the first equation is the smaller of the two values.43a) for W ( z ) .47) the result Thus.bh) Eliminating c2 and c4 from the above equations.3.3.3.4 P R > 0 (and PQ > O).6 (Clamped beam): d Using Eq. the boundary conditions in Eq. (4. in terms of w as PW~QW~+R=O (4.S 2 2 ~ : 3 .55)].48) into Eq. X / d x can be determined in terms of the constants cl through c4.38).S 2 2 ~ =O.43).3.5013) Sllcl .3. there are two (sets of) roots of this equation (when f2 # 0) It can be shown that Q~ .46a.3413) yield c.2. shear deformation decreases the frequencies of natural vibration [see Eq.50a) (4.45a) where Hence.3. we obtain from Eq. we obtain cz + cq = 0. which implies A.b) gives two frequencies for each value of A.3.3. Using the boundary conditions in Eq.3.42a) can be written.
This is primarily due to the assumed infinite rigidity of the transverse normals in the classical laminate theory.3.25). The first row of each laminate refers to the nondimensionalized critical buckling load. Transverse shear deformation has the effect of decreasing both buckling loads and natural frequencies.2.3 contains critical buckling loads and fundamental frequencies of various laminated beams according to the Timoshenko beam theory. The following nondimensionalizations are used: The frequency equations (4.3.59)] are independent of the beam geometry or material properties. i. buckling loads and fundamental frequencies according to the Timoshenko beam theory from those of the EulerBernoulli beam theory [29]. Once again we note that the relationships between the classical beam theory and the Tirnoshenko beam theory may be used determine the deflections. there are two different things that influence the frequencies in the Timoshenko theory: (i) the effect of transverse shear deformation [see Eqs. for clampedclamped and clampedfree boundary conditions. if one designs a beam for buckling load based on the classical laminate theory and if no safety factor is used.3.51) of the Timoshenko theory depend.49)].58) and (4. Thus. Thus. Note that the assumption does not yield a conservative result.e.. (4. it appears that rotary inertia may actually increase the frequencies slightly. and the reader may find it challenging to develop the relationships for bending. and (ii) the values of A. on the lamination scheme and geometric parameters (through StJ). Also. it will fail for a working load smaller than the critical buckling load.whereas those of the classical laminate theory [see Eqs.Table 4.2. The second effect is not significant. the effect of rotary inertia on the frequencies is not as obvious as it was in the case of simply supported beams.3.3. the classical laminate theory overpredicts buckling loads and natural frequencies. From the results presented in Table 4. The effect of lengthtoheight (or thickness) ratios of the beam on critical buckling loads N and fundamental frequencies w is shown in Figures 4. for clampedclamped and clampedfree boundary conditions.3. Such relationships exist only for isotropic beams. as can be seen from rows 3 and 5 of Table 4. .3 and 4. respectively. (4. for various lamination schemes.4.3. and the fourth row refers to fundamental frequencies without rotary inertia. buckling and vibration of laminated beams (see Section 5. The material properties used are those listed in Eq.2.47) and (4. which are governed by different equations than those of the classical theory (for clampedclamped and clampedfree beams). the frequency equations are the same in both theories). The numbers in rows 3 and 5 refer to the fundamental frequencies calculated using the frequency equations of the classical laminate theory (for the simply supported boundary conditions. the second row refers to nondimensionalized fundamental frequencies with rotary inertia.5 of [29]). (4.3.3.3. where the rotary inertia would decrease the frequencies.
3. and subjected to a transverse load q(x) that is uniform at any section parallel to the xaxis.1. G23 = 0.4 Cylindrical Bending Using CLPT 4.200 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS Table 4. all derivatives with respect to y are zero.1.3. Therefore.3: Critical buckling loads (N) and fundamental frequencies (6) of laminated beams according to the Timoshenko beam theory (E1/E2= 25. Gl2 = G13 = 0. (3. the governing equations of motion according to the linear classical laminate plate theory (CLPT) are given by [see Example 3. ~ 1 = 0.2E2.48)] .25). Suppose that the plate is long in the ydirection and has a finite dimension along the xdirection.2). the deflection wo and displacements (uo. 2 HingedHinged Laminate + ClampedClamped ClampedFree 100 20 10 100 20 10 100 20 10 4. In such a case. and the plate bends into a cylindrical surface.3. and let the x and y coordinates be parallel to the edges of the strip.1 Governing Equations Consider a laminated rectangular plate strip.vo) of the plate are functions of only x.4.5E2. For this cylindrical bending problem (see Figure 4. Eqs.
0 .t o .ONEDIMENSIONAL ANALYSIS O F LAMINATED COMPOSITE PLATES 201  (451451.3.t h i c k n e s s r a t i o . 1 l l 1 . a/h Figure 4.4: Nondimensionalized fundamental frequency (w) versus lengthtothickness ratio ( a l h ) of simply supported beams.3. 1 l l l 10 20 30 40 50 60 70 80 90 100 S i d e . l l l l ~ l l l l ~ l l l l . 10 20 30 40 50 60 70 80 90 100 Sidetothickness r a t i o . Figure 4.3: Nondimensionalized critical buckling load (N) versus lengthtothickness ratio ( a l h ) of simply supported beams. l ' 1 l ~ l l l l ~ l l l l ~ l l l l .  /   0 0 I l I I . / / (90145/45lO).
2.lac) can be expressed in an alternative form by solving the first two equations for u f f and vf' and substituting the results into the third equation where Note that C = O for a crossply laminate (Al6= B16= D16= 0). For a general lamination scheme. Eq. Eq. Eqs.2. In the case of crossply laminates. Therefore. the solutions developed in Sections 4. the second equation becomes uncoupled from the rest.4. (4.2 through 4. (4.4.202 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS where N ~ is an applied axial load. (4. (4.813).2~) have will the same form as Eq. and v is identically zero unless N$ is at least a linear function of x. In the general case. In the absence of thermal forces and axial loads.2. If the inplane inertias are neglected.4.2~) wo is uncoupled from those for of uo and vo. the three equations are fully coupled. and .4 are also valid for cylindrical bending with appropriate change of the coefficients. .
3b) to determine uo(x) and vo(x): where Further integrations lead to . (4. 4 .4.3a) and (4. (4.4.b).3a. satisfaction of the boundary conditions places a restriction on the lamination scheme.4.2.1013) mainly in the bending stiffness term. it can be integrated.4. When both edges are simply supported or clamped. it differs from Eq. ( 4 . Hence. (4.2ac) reduce to Equation ( 4 . (4. 3 ~is uncoupled from Eqs. for ) given thermal and mechanical loads.4.2.3~) closely resembles that for symnietrically laminated beams [see Eq.10b)l.2 on exact solutions applies to Eq. For clampedfree laminated plate strips.4. as will be seen shortly.2 Bending For static bending analysis.4. 3 ~ ) governing wo is uncoupled from those governing (uo.3~). and the result can be used in Eqs. Eqs. 4 . exact solutioris can be developed without any restrictions on the lamination scheme. (4. (4. The limitation on the lamination scheme in cylindrical bending comes from the boundary conditions on all three displacements of the problem. much of the discussion presented in Section 4. to obtain wo(x). (4. Since Eq.4. While Eq.4.3~) valid for more general laminates (symmetric as well is as nonsymmetric). Equation (4.vo).
then we have . [[i' i hd i ) ( d?] d i + ~2 ~ 2 ( i ) d+ ~2 i ix N&(W If the temperature distribution in the laminate is of the form where To and Tlare constants.and AUOx ) = B ( Ax[l'(1' g(i) 4) d ~4 ] + Gi lx ix N& ( i ) d i + p i Lx N& o x = i.
(4.12) where .1 (Simply supported plate strip): For a plate strip with simply supported edges a t x = 0 and x = a .14)]. UQ = 0. The inplane stresses in each layer can be cornputed using the constitutive equations..4.4. = 0.2.7) become The constants of integration ail bi. A& = 0 .4.13) and (4. and the transverse stresses can be determined using equilibrium equations of 3D elasticity [see Eqs. expressions in Eqs.2.where In addition. (4.. the boundary conditions are (see Table 4.1) N . For a crossply laminate the only nonzero strain is E. and ci can be determined using the boundary conditions.4. (4. Example 4. if q = qo.
(4. the solution for uo and vo can be determined only with an arbitrary constant (i.4. and (4.4.16~~) (4.4. for an arbitrary lamination scheme and it dvo/dx = 0. which give ag = b3 = 0.13~) duo M x y = BIGdx f &j6O  dv dx d2w dx2  M:y (4. (4. (4. (4. The constants can be determined by setting uo (0) = 0 and vo (0) = 0.4.14~~)follows that.Nxy = Al6.4.4.4. we obtain 4oa3 IQ(X)= .14~) From Eqs. (4.4. a Since only the derivatives of ? L o and vo are specified a t the boundary points.12).13a).e.16b) (2 ma3 [2 q. x + a 3 (4.15) in Eq. rigid body motion is not eliminated).4.. Using boundary conditions (4.13) and (4.2 ( ~ ) " 3 ( : ) ~ ] AD 12 [ +hi'.4.(x) = AD 12 wo(x) = 24 D ( E ) ~3 ( E ) ~+]hg  + MT A 2 [(z)) [(z)' 2 a2  (E)] (I)] (4.4.16) into Eqs.11ac).16~) where the constants a3 and by can be interpreted as rigid body displacements. we must have at x = 0. The stress resultants for any x are then given by substituting Eqs.14): .dx duo + A66duo dx  BI6dx2 d2uio  Nzu (4.4.4.
B = 0 . we have . for antisymmetric crossply laminates. the difference is in the expression containing Poisson's ratios.4. can be very large. = E.11ac).$. and it is given by In order to see the effect of the bendingstretching coupling on the transverse deflection.$. thc rnaxirnum deflection beco~nes ~~  + In the case of a~itisyrrmetric angleply laminates. the rriaxirnum deflection can be expressed in the form For syrr~rnetric laminates the coupling terms are zero.B f 6 / A s s . The bending stiffriess Dll used in cylindrical bending is given by whereas the bending stiffness used in the beam theory is E. The rnltxinlurn deflection becomes r Note that when the bendingstretching coupling terms are zero (e. where v.2d)l is expressed as Hence.Dh3/l2. which is due to the plane strain assurnptiori used in cylindrical bending compared to the plane stress assumption w e d in the beam theory.4..7ad).g. For example. B = B l l / A l l .rI. one must use Eqs. (4. for syrrirnetric larrii~lates).. and D = D l l B F ~ / A Thus.. C = 0.c = 012. it follows that the effect of the coupling is to increase the maximurri transverse deflection of the plate strip.4.416 = = B I G= B2fj = Dl6 = D26 = 0. the reciprocal of the bending stiffness D [see Eq. .. we have A16 = = = B l l = B22 = B12 = BGG D I 6 = DZfi= 0 . The difference between the two solutioris will be the most for laminates contair~irigangleply layers. The difference is only in the beriding stiffness term. Analytical solutions for beams under uniform transverse load with other boundary conditions may be obtained from Eqs. (4. C = B 1 6 / A 6 6and D = Dll . Therefore. For loads other than uniformly distributed transverse load.The maximum transverse deflection occurs a t .he same form. and the rnaxirriuni deflection is given by 6 It car1 be shown that the expression B ~ ~B IB C is always positive. Thus. the cylintlrical bending and laminated beam solutions have t. (4.
o .3 Buckling The equilibrium of the plate strip under the applied inplane compressive load N ~ = . V.26).4.24) and (4.=O . (4.25).=O simple support uo=O wo=O uo=O wo=O 4.N : ~ can be obtained from Eqs.=O w0=o N.2ac) by omitting the inertia terms .4. Edge Condition CLPT FSDT t 't roller W O = O .o 0 du dx N. Equation (4. can be integrated twice with respect to x to obtain .4.4.1: Boundary conditions in the classical (CLPT) and firstorder shear deformation (FSDT) theories of beams and plate strips. and thermal resultants where (U.4. W ) denote the displacements measured from the prebuckling equilibrium state. which is uncoupled from (4.4.0 du dx M.208 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS Table 4.=O M. The o boundary conditions on u and vo are only for laminated strips in cylindrical bending.
Eq.4.15a) we have Use of the boundary conditions on W gives c2 = cy = c4 = 0 and the result The critical buckling load N.where K1 and K2 are constants.B I ~ For .4. and X are determined using (four) boundary conditions of the problem.4. q = K ~ / X ~ and c . Recall from Section 4.33) The smallest root of this equation is X = 27r.2~) reduced to is where T2 = I2 . (4. a periodic motion. (4. and transverse load.2: When the plate strip is simply supported a t x = 0. Here we consider only the case of simply supported boundary conditions for illustrative purposes. c4.4. Example 4.2. the buckling load can be determined using Eq..3 that when both edges are clamped. and the critical buckling load becomes 4. a .29). Once X is known.2 = 0 (4. from Eq.4.2. = D. c3. + c2 cos Xx + c3x + cq (4. is given by (n = 1) Thus the effect of the bendingextensional coupling is to decrease the critical buckling load.4. The results of Section 4.3 are applicable here with b = 1 and E&.4. X is determined by solving the equation Xa s i n X a + 2 c o s X a .4. thermal forces.4 Vibration For vibration in the absence of inplane inertias. we assume . (4..28) The three of the four constants cl.27) is W ( x )= cl sin Xx where cg = K1/X2. c2. The general solution of Eq. (4.
the natural frequency of vibration. (4. The general solution of Eq..N. For natural vibration without rotary inertia and applied axial load. which are determined using the boundary conditions. Hence.4. we have Example 4. with rotary inertia included. is given by When rotary inertia is neglected.4. I ~ .4. We = Section 4.42) it follows that Note that the rotary inertia has the effect of decreasing the natural frequency.4. q = 12w . D. and c4 are integration constants.35) has the same form as Eq. (4.4 are applicable here with b = 1 (fo = Io. the equation for X = p reduces t o If the applied axial force is zero. When the rotary inertia is zero..4. A r = Iow2 (4. and C I .2. C ~ c3.37) is 4) W (x) = el sin Ax + cz cos Ax where + cg sinh px + c4 cosh px . (4.43).40) . A. = 7 and from Eq. = summarize the results here for completeness. is given by A. we have . all of the results of f2 and E ~ .35) becomes Equation (4.3: For a simply supported plate strip.210 MECHANICS O F LAMINATED COMPOSITE PLATES AND SHELLS where w is the natural frequency of vibration.2.2 p = D . (4. Then Eq.4.
X p ) .4. (4.47) The roots of Eq. (4.ONEDIMENSIONAL ANALYSIS OF LAMINATED COMPOSITE PLATES 211 For a plate strip clamped a t both ends. a l h . (4.2. .56)(4.4. The frequencies obtained from Eq.47) are In general.4.4.48) can be used t o determine the natural frequencies of vibration with rotary inertia from Eq. When rotary inertia is neglected.42) with the values of X from Eq.48) are only a n approximation of the frequencies with rotary inertia.  Figure 4. rotary inertia is more significant in reducing the frequency than for thin and long plate strips. (4.4.4. a/h Figure 4. the frequencies arc givcrl by Eq.60)] For natural vibration without rotary inertia. (4.4. X must be determined from [see Eqs. the roots of the transcendental equation in (4.43) with X as given in Eq.46) takes the simpler form cos An cosh Xrr  1= 0 (4. If one approximates Eq. o. (4. the roots in Eq.4.2.42).4.1: Effect of rotary inertia on nondimensionalized fundamental frequency of a simply supported (45145) laminated plate strip.47) (i. Eq. (4.46) are not the same as those of E g (4.4.46) as (4.4.48)..4. For small values of a l h . 1 3 Fundamental mode.4. \   Plate strip  4. (4. (4.47).4.70 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 ~ 1 1 1 1 ~ 1 1 1 1 1 1 1 1 ~ 1 1 1 1 10 20 30 40 50 60 70 80 90 100 Sidetothickness ratio.1 contains a plot of the nondimensionalized fundamental frequency = w a 2 J w of a simply supported plate strip with rotary inertia versus lengthtothickness ratio. (4.4.e.
laminates have larger bending stiffness than the (90/0). both have the same axial stiffness.4.4.205 32.3 show the effect of lamination angle on maximum deflections w = . GI2 = GI3 = 0. ~ 1 = 0.2 are for the case where rotary inertia is included and a / h = 10.. deflect less and carry more buckling load.. laminated plates in cylindrical bending undergo smaller displacements and have larger buckling loads and frequencies.4).740 57.807 0. Laminate B: ( 0 / f 45/90). This is because there are two 0' layers and they are placed farther from the midplane in the first laminate than in the second laminate. It should be noted that antisymmetric angleply laminates with more than two plies are stiffer.2 and 4.4.. (0/90/0) laminates undergo smaller deflections and have larger buckling loads and natural frequencies.wm a z ( ~ 2 h 3 / q o a x )lo2. and fundamental frequencies (G)of laminated plate strips according to the classical laminate theory (E1/E2 = 25. i.316 5. w = .645 26.2. Table 4. Hence.). The antisymmetric laminates have some of the Bij # 0 and thus are relatively flexible when compared to symmetric laminates.2: Maximum deflections (w) under uniform load.. ( ~ ~ h ~ / ~ ~ critical buckling load N.179 12. Compared to laminated beams (see Table 4. All of the frequencies listed in Table 4.w .4.e.25).584 11. frequency G of twolayer antisymmetric angleply (010) plates.613 w 0.623 N 20.809 = symmetric.I. laminates. and fundamental natural frequencies of simply supported and clamped (at both ends) laminated plate strips with various lamination schemes.5E2.035 0.169 Laminate A Laminate B 4.212 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS Table 4.125 N 82.2E2. 4 w = wa2 JIo/Ezh3 . critical buckling loads (N). . 2 Laminate HingedHinged ClampedClamped w  w 0 0. and fundamental a~). This is due to the Poisson effect discussed earlier.2 contains nondimensionalized maximum deflections. critical buckling loads. G23 = 0.897 3. The (0/90/0) laminates have larger bending stiffness as well as axial stiffness compared t o the (90/0/90) laminates.264 12. N = N:=(a2/E2h3).453 w  14. Laminate A: (go/* 45/0). Figures 4. (.185 14. . The (0/90).4. antisymmetric = (four layers).838 0.
\ simply supported . /  Uniform load Lamination angle.2: Nondimensionalized maximum transverse deflection (G) versus lamination angle (8) of a simply supported (BIB) laminated plate strip in cylindrical bending (CLPT). . 0 Figure 4. 0 80 90 Figure 4.4.3: Nondimensionalized critical buckling load (N) and fundamental frequency ( 3 )versus lamination angle (0) of a simply supported (018) laminated plate strip in cylindrical bending (CLPT).4.. 0 10 20 30 40 50 60 70 Lamination angle.Center point load ' .
lb) A16 a2uo+ A66d2vo+ B16 a24. + 8 1ax 6 ax2  ahT& ~  = Io a2uo+I1a24.5. (4.5.la) (4.4.5. (3.lae) are simplified to Next. we neglect the inplane inertia terms and assume that there are no thermal effects.a24.ld) from further consideration.2~): . For the purpose of developing analytical solutions. and omit Eq.23)(3.27)l: d2~o All ax2 a2(Pv + A16a2vo+ Bll. we consider the equations of motion for cylindrical bending according to the firstorder shear deformation theory (FSDT) [see Eqs. ax2 ax2  .4. we eliminate uo and vo from Eqs. Then Eqs.5. (4. at2 at2 d2vo at2 (4.2a) and (4.2ac) by solving (4.213) for uo and vo in terms of 4 and substituting the result into Eq. + B66d24.5..= 10 aN& ax + 11a24y at2 For cylindrical bending we further assume that q5y = 0 everywhere. (4.5. ax.1 Governing Equations In order t o see the effect of shear deformation on bending deflections and buckling loads. .4.5 Cylindrical Bending Using FSDT 4.5.5. (4.5.
Example 4.5.5. 4.5.3.3. Next we illustrate the procedure to determine the constants for beams with both edges simply supported or clamped.9a.Equations (4.4) are similar to Eqs. Eqs.3) and (4.3.3) and (4.4) reduce to Following the procedure of Section 4. we obtain [see Eqs.2.12)(4.2 Bending For static analysis. (4. (4.3. and therefore all developments of Section 4.3 would apply here.b) for laminated beams. The solutions developed are general in the sense that they are applicable to any symmetrically laminated beams.5. we obtain The rriaximum deflection occurs at x = a12 and it is given by .5.5.1 (Simply supported beam) : For a plate strip simply supported a t both ends and subjected to uniforruly distributed load q = qo as well as a downward point load Fo a t the center. (4.14)] the general solution for the rotation and transverse deflection where the constants of integration cl through c4 can be determined using the boundary conditions.
3. ( E ~ h ~ / ~of simply supported. The effect of orthotropy on deflections is shown in Figure 4. on the lamination scheme. Figure 4.2 (Clamped beam): Consider a laminated plate strip fixed at both ends and subjected to uniformly distributed transverse load go and a point load Fo at the center. and K = 516). For this case. shear deformation is relatively more significant for a l h 5 10. = I2 = 0 in Eqs.2 contains plots of nondimensionalized maximum deflection versus lengthtoheight ratio for twolayer antisymmetric crossply (0190) and angleply (451 45) laminates ( K = 516) under uniformly distributed load and with simply supported edges as well as for clamped edges. and lamination scheme on nondimensionalized deflections w = ~ . we obtain .3. The effect of shear deformation is to increase the deflections. 4. The shear correction factor has little influence on the global response for the antisymmetric: laminates analyzed. The most commonly used value of K = 516 is based on homogeneous. (4. Values of K for various special cases are available in the literature (see [481).3 ( G 1 2 = G I 3 = 0. Antisymmetric angleply laminates are relatively more flexible than antisymmetric crossply laminates.5. crossply ~ a ~ ) (0190) and angleply (45145) laminates under uniformly distributed load.4).5. geometry. .5.5. For clamped boundary conditions. and material properties.4): .5.1 shows the effect of shear deformation. Finure 4. we set q = 0 and (4. .Example 4.5. and I. the solution is given by The maximum deflection is given by The determination of the shear correction coefficient K for laminated structures is still an unresolved issue. shear correction coefficient. N~~= N:~. especially for a l h 5 10. in general.5E2. G Z 3 = 0.3) Following the procedure of Section 4.3 Buckling For stability analysis. vl2 = 0.5.2E2.25. both acting downward. although K depends. isotropic plates (see Section 3.
2: Transverse deflection (w) versus lengthtothickness ratio (alh)of simply supported (SS)and clamped (CC) plate strips. d h Figure 4. 0 2 0 1 10 20 30 40 50 60 70 80 90 100 Lengthtothickness ratio.0 0 .0.5.5. SS = Simply supported a t both ends CC = Clamped at both ends 0.07 Figure 4.1: Transverse deflection (w versus lengthtothickness ratio (a/ of ) h) simply supported plate strips (K = 1. .2/3).5/6.
0 1 1 1 1 1 1 1 1 1 1 .3: The effect of material orthotropy and shear deformation on transverse deflections of simply supported crossply (0190) laminated plate strips under uniformly distributed load. which are evaluated using the boundary conditions. (4. 1 1 1 1 1 1 1 1 . The general solution of Eq. the critical buckling load is given by .5. Example 4.17) is where and el through c4 are constants of integration.3: For a simply supported plate strip. a 1h Figure 4.5.0. 1 1 1 1 0 10 20 30 40 50 60 70 80 90 100 Lengthtothickness ratio.5.
25a) .2E2. r KA55 = l o w2 The general solution of Eq.2.40a. q = .5. we assume solution in the form where w is the natural frequency of vibration. Substitution of the above solution forms into Eqs. For a plate strip fixed at both ends.b)] Following the results of Section 4. (4. and u/h < 20 in the case of clamped boundary conditions. (4.Thus. = G I 3 = cm). and W(x) and X(x) are the mode shapes. v = 0. (4.4) yields [cf.hence. 4.25.InD w2. (4. in the classical laminate theory the structure is represented stiffer than it is.5E2.5 show the effect of shear defornlation and modulus ratio on riondirriensio~lalized critical buckling loads N = N&(a2/E2h3) of twolayer antisymmetric angleply (45145) and crossply (0190) plate strips (E1/Ea = 25.38b)I.5. Orriissiori of the transverse shcar deforrnatiori in the classical theory amounts to assuming infinite rigidity in the transverse direction (i.3) and (4.4 and 4. K = 516). the critical buckling Figures 4. In Figure 4. The effect of shear deforrr~atior~ significarlt for a l h 5 10 in the case of is simply supported boundary conditioris.e. Ey. G23 = 0. X is governed by the equatiori The roots of the equatiou are approximately the same as for the case in which shear deformation is neglected [see Eq.3.5). The first root of the equation is X1 = 27r. The effect of shear deformation is more for materials with larger modulus ratios (see Figure 4.4 Vibration For a periodic motion.5.24a) is W (x) = cl sin Ax + c2 cos Ax + c:3 sinh px + c4 cosh p z (4.5.4.3. GI2 = GI3 = 0.5. Hence.5.4 results are preseuted for simply supported as well as clanipcd bouridary coriditio~~s.5..5. the effect of the transverse shear deformation is to decrease the t)ucklirig load. we obtain where p = D .5.
0 10 20 30 40 5 0 6 0 70 80 9 0 100 Lengthtothicknessratio.CC SS = Simply supported at both ends C = Clamped at both ends . a l h Figure 4.5. a l h Figure 4. (01901. l l l l ~ l l l l 0 10 20 30 40 50 60 70 80 90 100 Lengthtothicknessratio.5: The effect of material orthotropy and shear deformation on critical buckling loads of simply supported crossply (0190) laminated plate strips. .5.S S 1  4  2 : 0   f \(45/45).220 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS \ (451451.4: The effect of shear deformation on the critical buckling loads of simply supported (SS) and clamped (CC) crossply and angleply plate strips. SS   l 1 1 1 ~ " l l ~ l l " ~ l l l l ~ I I I I ~ I I I I ~ I I I I ~ I I I I .
e.5.C Z . and cq are integration constants.~ SZ2= X ( I ~ W ' ) +p 2 K ~ s 5 ) Once the value of X is known. C Q . The fnndamental frequency will come from Eq.4 for details).ONEDIMENSIONAL ANALYSIS OF LAMINATED COMPOSITE PLATES 221 where and q . (4.27) the result By neglecting the shear deformation (i.5. (4. the effect of shear deformation is to reduce the frequency of natural vibration. For a laminated strip with clamped edges. As5 = GI3 = m) we obtain the result which is the same as in Eq. and an equation governing X and p (see Section 4.45).4: For a simply supported plate strip. the following equation governs A: 2 + 2 cos Xu cosh pa + sin Xu sinh )m  (8 . (4.4. Thus. frequencies of vibration can be determined from Eqs. (4.b) gives two frequencies for each value of A..31a) (4.5.5.28) Substitution of X from Eq. . the boundary conditions give c2 = cs = cs = 0. we have P = 0 and the frequency is given by Example 4. (4. (4.5. The frequencies w can be determined from where When the rotary inertia is neglected.3. (4.26a). and sin X = 0.5.26a. Use of the boundary conditions leads to the determination of three of the four constants.5. = u a nr 7 (4.b). When the rotary inertia is neglected.5. or A.28) into Eq. we obtain from Eq.31b) SI1= p ( I ~ W ' X ' K A ~ .26a.5. the fourth one being arbitrary.
6). E 1 / E 2 = 25. 4. 4. This is known as the e1t:ctrostriction. whereas for clamped boundary conditions the effect is felt for a l h 15. and Reddy thirdorder beam theories.6. The use of TerfenolD for vibration suppression has some advantages over other smart materials. For example. = wa2 of twolayer antisymmetric angleply (45145) and crossply (0190) plate strips ( K = 516. lamination scheme. + . We wish to study the problem of vibration suppression in these beams using the EulerBernoulli. namely.5.6 Vibration Suppression in Beams 4. Although there have been important research efforts devoted to characterizing the properties of TerfenolD material. u12 = 0. To facilitate the development of all three theories in a unified manner.2 layers can be made of any fiberreinforced materials with varying fiber orientation 8 but symmetrically disposed about the midplane of the beam. and dysprosium.5. it has easy embedability into host materials.2 Theoretical Formulation Displacement and strain fields Consider a symmetrically laminated beam of n layers. (magnetostriction). Here we present a generalized beam theory that contains the classical EulerBernnoulli beam theory as well as the firstorder and the thirdorder beam theories. we introduce tracers whose values will yield the results for a particular theory [29]. iron.5. a commercially available magnetostrictive material TerfenolD is an alloy of terbium. such as TerfenolD particles embedded in a resin (see Figure 4.1 Introduction The grains of certain materials consist of numerous small.. the effect of shear deformation is more for < materials with larger modulus ratio.25). From Figure 4. such as the modern carbon fiberreinforced polymeric (CFRP) composites. The electric (magnetic) orientation brings about internal strains in the material. Few studies [33351 report experimental evidence of significant variation in material properties such as Young's modulus and magnetomechanical coupling coefficient. the mth and (n .222 MECHANICS O F LAMINATED COMPOSITE PLATES AND SHELLS Figures 4. Also.7. The remaining n . in particular.7 show the effect of shear deformation and rnodulus ratio ( E 1 / E 2 ) on nondimensionalized fundamental frequencies i.m 1)th layers. fundamental information about variation in elastomagnetic material properties is not available. Jw G23 = 0.6 and 4. and bring out the effects of material properties of a lamina. Timoshenko. as can be seen from the results of Figure 4. are made of magnetostrictive material. G 1 2 = GI3 = 0. and placement of the actuating layers on vibration suppression time. randomly oriented magnetic domains that can rotate and align under the influence of an external electric or magnetic field.5E2.6 it is clear that shear deformation effect in decreasing frequencies is felt for a l h 5 10 for simply supported boundary conditions.2E2. without significantly affecting the structural integrity.6. Considerable effort is spent to understand the interaction between magnetostrictive layers and composite laminates and the feasibility of using magnetostrictive materials for active vibration suppression (see [30321).5. Suppose that two of the layers.
alh Figure 4.6: The effect of shear deformation on the fundamental frequencies of simply supported and clamped crossply and angleply plate strips. 3 0 1 10 20 30 40 50 60 70 80 90 100 Lengthtothickness ratio.SS Lengthtothickness ratio.SS = Simply supported at both ends = Clamped at both ends (0/90). .7: The effect of material orthotropy and shear deformation on fundamental frequencies of simply supported crossply (0190) laminated plate strips.5.5. a l h Figure 4.
CO=O (4.e. respectively. q=1. y .1: Layered composite beam with embedded actuating layers. The functions f I (2) and f2 (z) are given by The displacement field (4.w) are the displacement components along the (x.1a) can be specialized to various beam theories as follows: EulerBernoulli beam theory (EBT): Timoshenko beam theory (TBT): Reddy beam theory (RBT): co = 1. wo is the transverse deflection of a point on the midplane (i. and $(x.layer hctuating layers Z Figure 4.2) . t) is the rotation of a transverse normal line. 4 CQ=. z = O).v.. 3h2 cl = c3 = 0 co=c~=O C I = ~ . Consider the displacement field where (u.6. z ) coordinate directions.6.6.
The nonzero linear strains are given by where Constitutive relations The constitutive relations of the lcth fiberreinforced (structural) layer are Q'..) 1.V l 2 (1) ( k ) V21 The constitutive relation for an actuating (say. a magnetostrictive) layer is where H is the magnetic field intensity. (k) Q22  E. E ( ~being the modulus of the magnetostrictive layer ( e ( m )= ~ ( ~ ) d ( ~ ~ ) ) . (k) Q12  (k) (k) V12 E 2 2 (k) (k) 1 .V l 2 V21 . rnagnetostrictive layer s ( ~ )the is compliance of the mth ) and d(") is the magnetomechanical coupling coefficient.:) = E!:) 1 (k) (k) u12 u21 . .
Various symbols introduced in Eq.Velocity feedback control Considering velocity proportional closedloop feedback control. :1 vanish on account of the assumption that all variations and their derivatives are zero a t t = 0 and t = T.6.11) are defined as . which can be expressed in terms of the coil width b... and number of turns n. coil radius r. Equations of motion Using Hamilton's principle (or the dynamic version of the principle of virtual displacements). (4. t ) as and I ( t ) is related to the velocity wo by where kc is the coil constant. we obtain  LT lL q6wo dxdt  lT 2 + lL [(K~ ~ 3 4 ) aswo + ( ax awo + ~ ~ 2 3 644 +~10zb06zb0 dxdt I where all the terms involving [ . the magnetic field intensity H is expressed in terms of coil current I ( x . in the coil by and c(t) is the control gain.
. awe 32. Pr:r:..6. 4 PC. .Qz .6.3 Analytical Solution First we write the equations of rriotiori (4.6. [see Eqs.6. P.17) in terms of the displacement variables (wo. We have .. Q.13a.6.6.16) and (4.) . (4. 4) by expressing AIL.18b) V.6...K3) are the mass inertias The equations of rnotion are The prirriary and secondary variables of the formulations are PrimaryVariables : SecondaryVariables : where Wo.12)]. and R ..18a) (4.where (hfzz. 4. arid ( K 1 K 2 . denote the corlverltiorlal and higherorder stress resultants R. f?f.b) and (4. (4.
4). a rmx 4(x.6. Assuming solution of the form nrx wo(x.20) and (4.This completes the development of the governing equations in terms of the displacements (wo.22) and substituting into Eqs. we obtain where the coefficients sij = Sji and Mij = Mji are defined by . Of course. (4.. Here we discuss the Navier's solution of these equations for the case of simply supported boundary conditions.21). t ) = X ( t ) cos  a (4. t ) = W (t) sin . the equations can be specialized to any of the three theories.6.6.
6. cl = c3 = 0) Timoshenko beam theory (TBT) (co = 0.c3 = 0) .Equation (4. el = 1.24) can be specialized to various theories as follows (only nonzero coefficients are listed) : EulerBernoulli beam theory (EBT) (co = 1.
GlassEpoxy and BoronEpoxy] are listed in Tables 4. Table 4.6. and for the Timoshenko and thirdorder beam theories. where Equation (4. (4. for nontrivial solution.6. The numerical values of various coefficients (namely.4 Numerical Results Numerical studies were carried out to analyze damped natural frequencies.6. GraphiteEpoxy (AS). A typical eigenvalue can be expressed iwd The lowest imaginary part (wd) corresponds to the transverse as X = a motion. and they are compared with the results obtained by Krishna Murty et al. Different lamination schemes were used t o show the influence of the position of magnetostrictive layer from the neutral axis on the vibration suppression time. and we can write + In arriving at the solution (4. The damping and frequency parameters for transverse modes n = 1 to n = 5 are shown in Table 4.dH 4. All values of the material and structural constants are indicated in the tables. A time ratio relation between the thickness of the layers and the distance to the neutral axis of the laminated composite beam is also found. we assume q = 0 and solution of the ordinary differential equations in Eq.32). and the vibration suppression time. the inertial and magnetostrictive coefficients) based on different layups and material properties [CFRP.31) gives two sets of eigenvalues. the following initial conditions were used: The actuation stress is ad = E. damping coefficients.6.1 and 4.2. using the three theories [29].2 also shows the damping coefficients and natural frequencies for different materials and layups.23) in the form and obtain. [32] using the EulerBernoulli beam theory (EBT). the result for the EulerBernoulli beam theory. The material properties used are the same as those used in [32].6.6.For vibration control. There is some difference between the numerical .6.6.3.
38 5.94k143. p. Gla= G134.6 GPa.62k102.52 52. 4 5 GPa .=9250 kg.80F2597.1=7. for Various Lamination Schemes  Layup [45/m/45/0/9O]s [m/k45/0/90]s [m/90rls [m/Oals Murty et al 4.85 13.4: Damping and Frequency Parameters a and o.6.mode 1 EBT 4.Layup [k45/m/0/90ls TBT 3.22k2601.30.1: Coefficients for Different Lamination Schemes and Materials (from Reddy and Barbosa [301) CFRP : E11=138. v12=0. Eaz4.25.26.621t102.68f943.12 GPa . 1 GPa.93f98. 2 7 GPa.78 GPa.30k104. Gln=Gm=0.9 GPa.m 3 .68 80.48f940.72 82.21 GPa.14 GPa. G 2 ~ 3 .94k98.93 GPa.11 5.=104.6E22.57 + TBT 4.82 13.64 5. a = l m Magnetostrictive layer : E.6E22.90k143.Table 4. and Parameters a and cod for Various Laminates Table 4.ms 2 c(t). Gz=6.27*1669. 2 p=1824 kg. G 1 ~ 4 .v12=0.67~10~m/A.m" kg.6 Ez2.65 5.02 CFRP : E11=138. G2:3=4. d~1.5 Gpa. v12=0. dk=1.2: Mass Inertias and Magnetostrictive Coefficients.80 29.=9250 kg. GPa.73k1676. Gu=4.=O.6 GPa Eza4.p=l824 kg.12 5.90k64.v. as Predicted by Various Theories (see Reddy and Barbosa [301) Mode 1 2 3 4 5 a k Murty et a1 3.m.m3 2 c(t).6.93k143. E 2 ~ 4 . .R=lO.27 GPa G 1 ~ 4 .87 U .6Gpa.49 RBT 4.83 82.90 29.62k102.29k104.64 5.05 52.44 CFRP :E11=138. ~ 1 ~ 0 .19k419.40 52.m'j GraphiteEpoxy (AS) : El1=137. 9 6 GPa.5 GPa.39 5.86k1678.94f98.69 GPa.p=1950 Table 4.=O.90f98.30f104.50 29.44 5. 1 GPa G13=G23=0.6.32 81.Ea2=20. a = l m Magnetostrictive layer : E.27 GPa. M (radls) . E22=17.15 5.10k1667.m BoronEpoxy : E11=206.59k2621.96 GPa.=26.09 EBT 3.70k943. p. E 2 ~ 8 . G13=G23=0. p=1900 kg.65 5.94k64.04 RBT 3.67x108mlA.26. p=1824 kg.94k64.93f 143.3: Comparison of the Damping and Frequency Parameters a and w.37 29.82 13.16&418.R.6.17 5.601t102.88 52.m 3.j 6.30k 104.30.53k941.34k2619.58 a wdn (radts) .G~=Gln=8.=26.94k64.17i418.9 GPa. ~1~=0.p=1450 kg. Table 4.10 GlassEpoxy : Ell=53.v. vr2=0.42 5.88 13.9GPa.20f419. Gra=G1.
Figure 4. and it remains nearly the same in the laminates with different stiffness.6 shows that the vibration suppression time decreases very rapidly for higher modes. The uncontrolled and controlled motions at the midpoint of the beam. Analytical solution for the simply supported beam is presented to bring out the effects of the material properties of a lamina. Timoshenko. the difference between the predictions of the two theories is not significant. indicating faster vibration suppression. as predicted by EBT and RBT.5 for the first mode when the actuating layer (m) is placed at different distances from the midplane of the laminate. natural vibration. Clearly. when exact closedform solutions cannot be developed. ( A n s : The maximum deflection is the deflection wo(x) w. they are preferred over the series solutions. Figures 4.results predicted by the three theories only in the higher modes. The value of a: increases when the magnetostrictive layer is located further away from the xaxis.) 4.. lamination scheme.6.3 Show that the critical buckling load of a clampedfree laminated beam using the classical beam theory is given by . the series solutions are the best alternative.7 shows the controlled motion of the beam.4 shows the influence of the position of the magnetostrictive layer in the zdirection and the influence of the lamination scheme in the damping and frequency parameters. The layup [m/904]. A unified formulation for laminated beams with embedded actuating layers is presented.6. are shown in Figures 4.. The formulation includes the EulerBernoulli. Use the symmetry about x = a/2 to determine using the classical beam theory. Table 4. numerical solutions based on the finite element method (see Chapters 9 and 10) can be used to determine the solutions..6. Analytical solutions are presented for static bending.I. However. When closedform solutions can be derived. obtained using the three theories show that there is no significant difference between the results. for mode n = 5. 4. When analytical solutions cannot be derived at all.24.6. and Reddy thirdorder beam theories as special cases. and buckling problems under a number of boundary conditions.1 Consider a simply supported laminated beam under point loads Fo at x = a/4 and x = 3a/4 (the socalled fourpoint bending). as predicted by RBT. 4. represents the softest beam and the layup [m/04].6.7 Closing Remarks In this chapter analytical solutions are developed for laminated beams and plate strips in cylindrical bending using the classical and firstorder shear deformation theories. = llFoa3/384E$. the A comparison of the fundamental transverse and axial modes. stiffest beam. Problems 4. These figures show that the vibration suppression time decreases when the distance to the neutral axis is increased. and placement of the actuating layers on vibration suppression.2 Determine the static deflection of a clamped laminated beam under uniformly distributed load go and a point load Fo at the midspan using the classical beam theory.
Uncontrolled .3: Comparison of uncontrolled and controlled maximum deflection (at midpoint of the beam) for (45/m/45/0/90).2: Comparison of uncontrolled and controlled maximum deflection (at midpoint of the beam) for (f 45/m/O/gO).6.Controlled Time (s) Figure 4. Uncontrolled Controlled Time (s) Figure 4. laminate. .6. laminate.
Controlled Time (s) Figure 4..5: Comparison of uncontrolled and controlled rnaximum deflection (at midpoint of the beam) for (m/904). .Uncontrolled .4: Comparison of uncontrolled and controlled rnaximum deflection (at midpoint of the beam) for ( m l f 45/O/gO)..6.. 2 V 0... laminate.Controlled E Time (s) Figure 4.Uncontrolled . laminate.h E ...02 .6..
0.20 0. as predicted by EBT and RBT for mode n = 5. for modes n = 1 and n = 2.40 0.02 Time (s) 0.. .I I .00 0.010 """"'""""""'""'"""""""""""..6.04 Figure 4...03 0.01 0.. as predicted by RBT.00 0.00 Figure 4. I 'I :I _ I I mode n = l 0...0...7: Controlled motion of the laminated beam (f 45/m/O/9O). ..6: Controlled motion of the laminated beam (f45/m/0/90).6..60 Time ( s ) 0. .80 1.
4 Show that the characteristic equation governing buckling of a clampedhinged laminated beam using the classical beam theory is given by 4 5 Show that the characteristic equation governing natural vibration of a clampedfree laminated .10b)l with a equivalent load given by the righthand side of the above equation. I. is given by where a = E $ .10 Show that the equations governing the stability of a laminated beam according to the Timoshenko theory can be expressed as K G : & ~ X+ (KG:#  dW b ~ & = K .6 Show that the characteristic equation governing natural vibration of a clampedhinged laminated beam using the classical beam theory.3. 4.~N.10a. ~ ~ ~ ~ . (3) is 4. (4.O. = K l x W dx dx +K2 Combine the above two equations to arrive at Show that the general solution of Eq. I ~ ~ / K G $ . when rotary inertia is neglected. 4. (4. beam using the classical beam theory is given by 4.11 Show that the solution to the equations governing the bending of a hingedfixed beam according to the Timoshenko beam theory. under uniformly distributed transverse load.7 Show that the characteristic equation governing natural vibration of a hingedfree laminated beam using the classical beam theory. is the same as that for a clampedhinged beam.4. this effect will come through the boundary conditions..b) can be reduced to the single equation This equation shows that the deflection of the Timoshenko beam theory can be obtained from that of the classical beam theory by replacing the load q [see Eq. when rotary inertia is neglected. 4. Although the effect of shear deformation is zero when the load variation is linear or less.9 Show that Eqs. is sin Xa cosh Xa  cos Xu sinh Xa = 0 4. .2.. ) dx (1) E:. . when rotary inertia is not neglected.8 Derive the characteristic equation governing natural vibration of a clampedhinged laminated beam using the classical beam theory.
bh ) 0 Ans: The boundary conditions give wO(0)= 0 gives c2 4.13 Determine the critical buckling load of a clampedfree laminated beam using the Timoshenko beam theory. These equations are a static version of those in Eqs. REbh)+ hC1 . l3a).4. C3 = 0 wo(a) = 0 gives cl sin Xa (a) a..4. is given by 4.4. dx =0 gives X2 ( 1 + c2 cos Xu + c3a + cq = 0 :$bh) (cI sin ha + c2 cos Xa) = 0 In addition.. (0) = 0 gives ( + c4 = 0 1 . when rotary inertia is neglected.1).16 Derive the equations of equilibrium for cylindrical bending using the principle of virtual displacements.4. 6 + Nxy (2) Use the laminate constitutive equations (4.14a) to express the resulting EulerLagrange equations in terms of the displacements and the thermal stress resultants. KG6. SW = 0. l k ) . (4. note that 4. 4. . and (4.4.14 Show that the characteristic equation governing natural vibrations of a clampedfree beam according t o the Timoshenko beam theory. where 6W = la [21(%)'I + {N. (4.12 Show that the characteristic equation governing the buckling load of a hingedfixed beam according t o the Timoshenko beam theory is given by Xa cos Xa  sin Xa (' + X2ELIY.15 Show that the characteristic equation governing natural vibrations of a clampedhinged beam according to the Timoshenko beam theory is given by Sll cos Xa sinh pa + SZ2 Xa cosh p a = 0 sin (1) 4.
Use the total potential energy functional to construct a oneparameter Ritz solution to determine the natural frequency of vibration. Use the total potential energy functional n(uO.4NgI. Use algebraic polynomials for the approximate functions. cantilever beam).Consider the equations of equilibrium of crossply laminates in cylindrical bending in the absence of thermal effects: Show that the Navier solution of these equations for the simply supported boundary conditions is given by where D = AllDll coefficient Q./(5/310) [ ( 1 2 ~ $1. (Ans: w = ( l l a ) . (Ans: w = (l/a)./a2b) ./(l0/1~)[(12~~. ) N  Repeat Problem 4.19 for a laminated beam with clamped boundary condition a t x = 0 and free a t x = a (i.. (Ans: a1 = Bqoa2/12AD. of a simply supported laminated beam with compressive load N&.wo) for a simply supported plate strip..) ..wO) =l[T (2) a A11 2 duo duo Ass +. load y(x) is also expanded in sine series with The For the cylindrical bending problem of crossply plates (see Problem 4. ./a2b) L ] . .  B ? ~ am = and 7 . cl = qoa2/24D..e. . bl = Cqoa2/12AD. w. Use algebraic polynomials for the approximate functions.17). and a..vo. a t z = &h/2 and the stress continuity conditions a t the interfaces are satisfied.%.1.) .= A" + (2) to construct a oneparameter (for each variable) Ritz solution of (uo. show that (a) the stresses in the kth layer are given by and (b) the transverse stresses from the 3D equations of equilibrium are given by where G%nd H%re constants to be determined such that the strcss boundary conditions on a.
= 0 arid free at Repeat Exercise 4. 4 .. w .o t.. Use algebraic polynoniials for the approximate functions.= Repeat Exercise 4. cl = qOa2/12D.  (El%) .he trarisversc deflection. Determine the critical buckling load of a beam clamped at one end arid simply supported a t the other end.].) . of a plate strip with clamped boundary conditions at . Use the total potential energy functional to construct a oneparameter (for each variable) Ritz solution to determine the critical buckling load N.r. is the rotation. Consider a laminated beam of length L. Use algebraic polynoiriials for the approximate functions..24 for a plate strip with clamped boundary condition at a.he following geometric boundary conditions arid force boundary conditiorls (z d I ) z=o . S is the shear stiffness. (Ans: N. The total potential energy functional for the problern can be written as where UI"(T)is t.. / ( I ~ ) N&] )  1. Use oneparameter RaylcighRitz approxitnation for each variable.) J. (Anst w = ( 1 / a )J ( l o / 1 ~ ) [ ( 1 2 ~ / a ~ )NJ!. D is the flexural stiffness. Suppose that the beam is subjected t. = 3D/a2.h clamped boundary conditiorls at z = 0 and free 1)onndary conditions at z = (L. fiexural stiffness EI=constant..:= L Q ( I ) s=o =Q2 " d"" dz ( E l z ) =Qa.Repeat Prohlenl 4. and subjected to uniforrrily distributed transverse load q(x) = qo. Consider the buckling of a uniform hear11 according to the Tirnoshenko heam theory. (Anst nl = B q o a 2 / 6 A ~bl = Cqoa2/6AD. = I> =Q4 .) Use the total potential energy fimctiorlal to constnict a oneparameter (for each variable) Ritz solution to determine the natural frequency of vibration. (Am: w = ( ~ / ( L ) J ( ~ o / ~ I ~ ) [ ( ~ D ..25 for cylindrical bending of a plate strip using the firstordcr shear deformation theory but neglecting rotary inertia.r = 0 and free boundary conditions a t z = u. o f a simply supported plate strip with edge cotnprcssive load N&. a d N k is the axial cornpressive load.21 for a plate strip wit.
in terms of the values of 4.31 Equations (4.29 (Continuation of Problem 4.c2.3. and x = L and obtain . and wo of the Timoshenko beam theory suggest that they can be approximated with quadratic and cubic polynomials Rewrite the constants a. and c4 in terms of u l . 4. Note that ui and Qi are introduced into the formulation to have the convenience of specifying a geometric or force boundary condition. at x = 0.2.29 is valid for any boundary conditions. it can be used to determine solutions (which turn out to be exact) even for indeterminate beams.12b) and (4. (d) of Problem 4.30 Since Eq. boundary conditions (a) and rewrite (c) in the form Define the functions p. x = 0. cg.Q4) are the associated shear forces and bending moments at the same points.) 4. Assume Ritz approximation of the form (the exact solution of the homogeneous equation.2 (Ans: The stiffness matrix [ K ] for the EulerBernoulli beam element. In particular.5L. u z ) and (ug.3. determine the displacement in the spring that supports the right end of a beam when the left end is fixed and the beam is subjected to uniformly distributed transverse load go. are the same as the Hermite cubic interpolation functions given in Section 10. show that and force vector {q) are the same as those given in Section 10. a constant. In particular. of the stiffness matrix and qi of the force vector when E l = constant and q(x) = go. ~ 3and u4 using the geometric . u 2 .(x) (i = 1.3. respectively.1310) for 4. u4) denote the transverse deflections and rotations (clockwise) at the left and right ends.Here ( u l . ~ l d ~ w ~= 0dsuggests this polynomial) l x ~ and express the constants cl.28) Substitute the approximation into the total potential energy functional associated with the EulerBernoulli beam theory and express it in the form (a) Define and evaluate the coefficients K. and (Q1. These functions can serve as the approximation functions for the RayleighRitz method (see the next exercise).4) that you derived. (Ans: p..2. Q3) and (Q2.) 4. and (b) use the total potential energy principle to determine the fourparameter Ritz solution for the problem.
( j = 1 . and subjected to a uniformly distributed transverse load go. In order t o establish these relationships.33 are given by where R = D.. Then use the total potential energy principle to derive the Ritz equations for the problem. Show that for cantilevered beams all C.3. The boundary conditions of the EulerBernoulli and Timoshenko beam theories for the problem are as follows: EBT : dwE wF(0) = w. 3 ) are the moments corresponding to the rotations Q 3 . Show that for simply supported beams all C.L~). Use Eq. and superscripts E and T on variables refer t o the EulerBernoulli and Timoshenko beam theories.(L) = L ( 0 ) = M & ( L ) = O dx (1) TBT : wf(0) = w:(L) = q 5 T ( ~ ) = MTx(L) = 0 (2) Show that the constants of integration in Problem 4.29 t o express the total potential energy functional in terms of u.: where P. and shear force of the Timoshenko beam theory can be expressed in terms of the corresponding quantities of the EulerBernoulli beam theory (see [27.2./(A.K. clamped (or fixed) a t the left end and simply supported at the right. (a) of Exercise 4. Show that where GI.28]). z / ( A .31 and Eq. we use the following equations of the two theories: where K . of Problem 4. The deflection.where cP1 = 4. 2 . and a. except C4 = h f ~ ( 0 ) D .33 are zero. bending moment. is the shear correction coefficient.C2:C3. and C4 are constants of integration. z K s ) of Problem 4. ..(0) etc.33 are zero. Show that &(x) (i = 1. which are t o be determined using the boundary conditions of the particular beam. (a) of Problem 4.3) are the quadratic Lagrange interpolation functions derived in Section 10. Consider bending of a beam of length L.
. 10. 15. N. "Shear Correction Factors for Orthotropic Laminates Under Static Load. "Exact Solutions for Composite Laminates in Cylindrical Bending. Heyliger. 41. 19711980 (1994). 6. New York (1959). 12. New York. Franklin. Timoshenko. Y. "Elastic Wave Propagation in Heterogeneous Plates. 2. 126(2). and Wang.. Transactions of A S M E . Englewood Cliffs. P.. 7. 8. Harper and Row. W. 13.References for Additional Reading I.).. Pagano. 11." Journal of Applied Mechanics. W. N. S." Journal of Sound and Vibration. S.. "Influence of Rotatory Inertia and Shear on Flexural Motions of Isotropic Elastic Plates. W. New York (1970). E.. Second Edition. Fifth Edition. S. Vibration Problems i n Engineering. 51. A. . 19. Third Edition. L. 8187 (1981). PA (1987). S. S. 16." Journal of Composite Materials. R. 287300 (1948) (in Russian).. Clough. Systems of Ordinary Differential Equations. J. D. P. and Schwartz. 23. 74.. 309326 (1988). "The Propagation of Waves in the Transverse Vibrations of Bars and Plates. and Penxien.." Journal of Applied Mechanics. Pagano. A. R.." Philosophical Magazine. D.. Theory of Elastic Stability. 18... Mindlin. J. Jr. Second Edition. P. and Gere. N. B. 17. M. 5. (Ed. Technomic. C. J. A. New York (1975). New York (1990)." Journal of Applied Mechanics. Timoshenko. and Stavsky.. H. 6977 (1945). Selected Works of Nicholas J." International Journal of Engineering Science. Timoshenko. ." Journal of Composite Materials. 5 . P. 3. Yang. P. J... J. J. J. NJ (1985). A. 14. Whitney. Kluwer. Reissner. Norris. N. A n Introduction. "Free Vibration of CrossPly Laminated Beams with Arbitrary Boundary Conditions. R. Energy Principles and Variational Methods i n Applied Mechanics.. Theory of Plates and Shells. "Influence of Shear Coupling in Cylindrical Bending of Anisotropic Laminates. PrenticeHall. N. J. Khdeir. W. "On the Correction for Shear of the Differentid Equation for Transverse Vibrations of Prismatic Bars. Singapore (1970). M." Journal of Sound and Vibration. Reddy. 40(1). J. "A New Rectangular Beam Theory... J.... and Young. "The Effect of Transverse Shear Deformation on the Bending of Elastic Plates. Whitney. N. Matrix Theory. Mekh. Timoshenko. 25. McGrawHill. 744746 (1921).. J. P. A. S. Modern Control Theory." Journal of Applied Mechanics. Prikl Mat. "A Consistent Higher Order Beam Theory. and WoinowskyKrieger. 125131 (1922).. 4. Mechanics of Composite Materials. D.. PrenticeHall. 12. N. 302304 (1973). 665684 (1966). 21. The Netherlands (1994). R. Reddy. Ya." International Journal of Solids and Structures. 11. Lancaster. 521528 (1971). J. L. P. and Reddy. Nauk SSSR.. 2 .eoretica1 and Applied Mechanics. J. New York (2002). Timoshenko.. 24. 18: 3138 (1951). Pipes. Reddy. Goldberg. 3. L. and Harvill. "Further Study of Composite Laminates Under Cylindrical Bending. Applied Mathematics for Engineers and Physicists. Pagano. S. "A Simple HigherOrder Theory for Laminated Composite Plates.. 22. John Wiley. M. 26. John Wiley. Pagano. "A HigherOrder Beam Finite Element for Bending and Vibration Problems.745752 (1984)." Journal of Composite Materials. McGrawHill. J. Levinson. McGrawHill.. Bickford. Uflyand. P. C. H. 12. 3 2 (12). 20. Brogan. and Reddy. 137150 (1982).. NJ (1968). 1972. 398411 (1969). 4." Philosophical Magazine. Structural Analysis of Laminated Anisotropic Plates. Englewood Cliffs. Dynamics of Structures. S. 330343 (1970). L. "On the Transverse Vibrations of Bars of Uniform Cross Section. N." Developments i n Th." Akad. 9. McGrawHill. Weaver. 43. J. N.
B.. Arijanappa. C.l. N. and Calkins. V." Smart Materials and Stmctures. "Displacement Control of Timoslienko Bcarns via Induced Strain Actuators. K. Y. p. J ..  . 36.s. Lagoudas. 28." Journal of Sound and Vibration. F. 14 (2000). C. paper 42. .. J. K. 37. and Reddy.. 206(2).Journal of Enginee~. K. 3373 3384 (1997). Bhattacharya." . "Statistical Analysis of TerfenolD Materials Properties" Proceedings of SPIE Smart Structures and Materials. and Bi.. J.rt Materials and Struct. and Calkins.. . "Vibration Suppression of Laminated Corriposite Beams Using Enlbedded Magnetostrictive Layers. 83 (1994)...27. H." . Oxford.. Qidwai.JournalAS. S. N. "The Use of Magnetostrictive Particle Actuators for Vibration Atteriuatiorl of Flexible Beams. J. U K (2000). Reddy. Arljanappa. Shear Deformable Beams and Plates. "Timoshenko BeanlBending Solutions in Terrns of EulerBerr~oulli Solutions. and Lee. 32. N. I. J. M. M.. T. "Relationship Between Bending Solutions of Classical and Shear Deformation Beam Theories. M.F. 29.. M. N. "Modeling of t.M." Internationml Journd of Solids W Structures. A. A. M. Wu. 121(6). Dakshirla Moorthy. 763 765 (1995).:Wang. Flatau. 31. M.. 34. C. 9. B. 3327 (1998).. Arijanappa. and Barbosa... 3844 (1998). Dapino. 133149 (1997). J.. Dapino.." Smart Materials and Structures. 476 488 (1997). "On Vibration Slippression of Magnetostrictive Beams. 35.. 30. Wang. A. Elsevier. A. N. M. Wang. 3.ar~g Me(:har~zc. 8 . C."Sma.: and Bhat. Flatau. Ang. 49 58 (2000). and Wu." IE (I) . Y. 9.. ASCE. "A Theoretical and Experimental Study of Magnetostrictive Mini Actl~ators. arid Wang.he Thermomechanical Response of Active Laminates with SMA Strips Using the Layerwise Finite Element Method. M. Reddy. H. D.... T. "High Bandwidth Tunability in Srriart Absorber" Proceedings of SPIE Smart Structures and Integrated Systems. K. A.Journal of 17~telligentMaterm1 Systems and Structures. Reddy. M. 33. Krishna Murty. and Lee.F. V.. J. M.. B. Reddy. F. paper 304120 (1997). C. M.. C.ures.. 78. Krishrla Murty. 34(26).J.
.
simple support. In this chapter we develop analytical solutions of specially orthotropic plates.Analysis of Specially Orthotropic Laminates Using CLPT 5. Analytical solutions can also be developed for rectangular laminates with certain lamination schemes and boundary conditions. This class of laminates will be used to gain a basic understanding of the response. Analytical solutions were developed in Chapter 4 for certain onedimensional problems. The Ritz method can be used to determine approximate solutions for more general boundary conditions. Stresses can be determined using either the constitutive equations or the 3D equilibrium equations expressed in terms of stresses. The L6vy solutions can be developed for plates with two opposite edges simply supported and the remaining two edges having any possible combination of boundary conditions: free. plates for which the bendingstretching coupling coefficients Bij and bendingtwisting coefficients D I 6 and D26 are zero. using the classical laminate theory. and the Ritz method. i. they may represent reasonable approximations to more complex laminates. Although most laminates of practical interest do not qualify as specially orthotropic plates because of the presence of bendingtwisting coupling terms DI6 and Dzs. In the subsequent chapters. as long as we can find suitable approximation functions for the problem. the solutions obtained for more complicated laminates will be compared with those of the specially orthotropic plates to assess their behavior. These equations can be solved either analytically or numerically for the generalized displacements and strains. . The analysis of specially orthotropic laminates is greatly simplified because the bending deformation is uncoupled from the extensional deformation and the fact that DI6 = = 0. namely laminated beams and cylindrical bending of laminates..e. The solution methods used here are the Navier method. The Navier solutions can be developed for a rectangular laminate when all four edges of the laminate are simply supported. the Lkvy method with the statespace approach.1 Introduction The governing equations of composite laminates according to various laminate theories were developed in Chapter 3. or fixed support.
2 ~ ~ ~ axay a2wo . and Mnn are defined in Eqs.2 Bending of Simply Supported Rectangular Plates 5. We have  [ 2+ Dl17 2 (Dl2 + 2D66) ar"ayi +D2z7 a4wo a4w01 ay +q Equation (5. (3. and approximate solutions using the Ritz method when exact solutions cannot be developed. in conjunction with appropriate boundary conditions [see Eq. for the desired response.47) by omitting the nonlinear terms.1.1.3. and bendingtwisting terms.1) must be solved. (3. we wish to determine static deflections and stresses. The boundary conditions at any point on the boundary are of the form and where Q. and buckling loads under inplane compressive or shear loads of specially orthotropic plates. frequencies of natural vibration.The equation of motion governing bending deflection w of a specially orthotropic o plate can be deduced from Eq.3.3. We seek exact solutions whenever possible. In this chapter.29b).2. respectively.34)] and initial conditions of the problem.1) for this case reduces to The simply supported boundary conditions on all four edges of the rectangular plate (see Figure 5.1) can be expressed as where the bending moments are related to the transverse deflection by the equations M~~ = .3113) and (3. Equation (5. 5.2. bendingstretching terms. (3.1 Governing Equations Here we consider the static bending in the absence of thermal effects and inplane forces.3.
1: Geometry. Otherwise. and Wmn are coefficients to be determined such 1 that the governing equation (5.2) are satisfied by the following form of the transverse deflection w"(z. the Navier solution cannot he developed for the problem.y ) is also expanded in double trigonometric series. 3 = nxlb.2.2. sin a x sin P y where a = m n l a and .2. Substitution of the displacement and load expansions into the governing equation should result in an invertible set of algebraic equations among the parameters of the displacement expansion. y ) = W. The load q(x. and simply supported boundary conditions for a rectangular plate. The origin of the coordinate system is taken at the lower left corner of the midplane (see Figure 5.2.1) is satisfied everywhere in the domain of the plate. and ycoordinate directions and a and b denote the inplane dimensions along the zof the rectangular laminate.1). The simply supported boundary conditions in Eq.2. The boundary conditions in Eq. 5. The choice of the trigonometric functions in the series is restricted to those which satisfy the boundary conditions of the problem.2) admit the Navier solution for specially orthotropic rectangular laminates.2. (5. We assume that the load can also be expanded in the series form as . coordinate system..Figure 5.2 The Navier Solution In the Navier method the displacement wo is expanded in a double trigonometric (Fourier) series in terms of unknown parameters. (5.
y) = yo sin . (5. y) of the domain 0 < x < a and 0 < y < b. Then the solution in Eq.4) and (5.2. we have Qmn = 16qo for m.2.2.xo. the Navier solution is a series solution. For other types of loads.COS  m=l m7rx a n7ry b . n . s = bla. a constant.sin ~ ) a b  (5.sin a b is a oneterm solution (Qmn = qo and m = n = I ) .12~~) (5.2. the load coefficients are given by [q(x.2.2.where Qmn y (z.1) yields   4/ b (5.YO)^ 4Qo mxxo nryo (5.2. For example.yo). y) = yo. the Navier solution for a sinusoidally distributed transverse load 77z 7ry q(x.2.6) Since the equation must hold for every point (x.4) becomes 03 00 Qmn sin a x sinpy n=l m=l The load coefficients Qmn for various types of loading [see Eq.1.sin ab a b The bending moments can be calculated from Mxx = n=l m=l 03 00 C C (Dlla2 + 00 03 00 ~ 1 2 m7rx n7ry Wmn~ sin . (5.5b)l are listed in Table 5. This yields where s denotes the plate aspect ratio.11) Qmn = sin . odd 7r2mn For a point load Qo located at (xo. Y .2.12b) Mvli = n=l m=l C C (Dlza2 + D22P2) Wmn sin m7rx a C C c~PD66 Wmn n=l 03 sin n"Y b Mxy = 2 COS . In particular.y) = QoS(z . The effect of thermal moments can be easily incorporated into the calculation. which can be evaluated for a sufficient number of terms in the series.y) sin az sin py dniy ab o o Substitution of the expansions (5.2. for uniformly distributed load q(x. (5. and therefore it is a closedform solution.5) into Eq. the expression inside the curl brackets (or braces) should be zero for every m and n.
3 . n = l . 5 . 4 =4 0 Qmn = 16qo (m. n = 1 . ran sin (m = 1 . . Y) = qo: Qmn = 8qo cos m r nlmn (m. 3 . Line load. 5 . 3 .) Point load. .Table 5.) ... . . .. . Uniform load. 4(x.2. 2 . . . n = l . 3 .) Y Hydrostatic load.. .1: Coefficients in the double trigonometric series expansion of loads in the Navier method. 5 . Q ~ = .
y. a. h/2) = 0 and continuity of stresses at layer interfaces. b/2. (5.(x. h/2) = oyz(x.2. h/2) = o. y. z) = (a/2.(x.The inplane stresses can be computed from Eqs.2. they can be computed using the 3D stress equilibrium equations [see Eqs. and C. and the shear stress is maximum a t (x.. h/2) and other three corners.13)] for any zk 2 zk+l: < < (k) ( k ).12a) The maximum normal stresses occur at (x.lib) . ( z ) [$+ ( 2  3z()] (5. h/2). (4.l. We obtain og) where ( z ) =( 2 ) ( ). (k) where the stresses ox.y. b. (4.. However. y.13). y. The interlaminar stresses are identically zero when computed from the constitutive equations in the classical laminate theory. z) = (a. .2. gzY and are known from Eq. are functions to be determined using the boundary conditions.
the deflection and stresses are nondimensionalized as follows: i ="(0.O : ) a 90 x1 0 8 .oyz...15a) can be simplified to c cTI:) w~:.2 contains the nondimensionalized maximum transverse deflections and stresses of square laminates under various types of loads. a. y). = . uniform.. In integrating the stressequilibrium equations it is assumed that the stresses (ox. or point . Y 5.) are zero a t z = h/2. y ) . (5. ox.cos .. = o.2. = a b h 5. CO CO n=l m=l 0 0 03 rnrx my cos . the expressions in Eq. A T = zTl(x. B= a 1 a b h ~ Y = 0v:q(2. l) E2 ..q at z = h/2. For the case of mechanical loading.. Because of the assumptions of the laminate plate theory.(a. a b h . and the thermal load consists of linear temperature distribution through the laminate thickness.For singlelayer plates.o. Both q and TI are assumed to be sinusoidal. .) 2 2 2 (&) a b h For the thermal load case.sin n b y a 00 00 .. l) (g) 5)() h 132 pa The mechanical load consists of only the transverse load q(x.0)8 x lo2. n=l m=l n=l m=l rnrx r C C $:)wmnsin .sin a b rnrx a rmy b T $ ) W ~ sin . Table 5. the nondimensionalized quantities are defined as = wn(0.2.6 .(. = oZ. () pa .
25). and 1 4 2 = 0. the maximum deflection and (negative) stress occur for an aspect ratio around 1. is the maximum at ( x . The material properties of the lamina are taken to be: E1/E2= 25..0.y. through the thickness for laminates (0/90/0) and (0/90/90/0) under sinusoidally distributed transverse load.2. O).5E2. G23 = 0. Plots of nondimensionalized maximum transverse deflection w and normal stress a.functions.2E2.25. t From equilibrium equations (mechanical load). and Figure 5.5. 242 = 0. Figures 5. whereas for sinusoidally distributed load the maxima are reached around a/b = 2.. For uniformly distributed load. as a function of the plate aspect ratio a/b are shown in Figures 5. respectively..4 and 5.5Ez.2 and 5.5E2.3..2.2. a y . Table 5.6 shows the distribution of the maximum transverse shear stresses through the thickness for the two laminates ( a l b = 1. and the transverse normal stress a. a = 3a2. Z ) = ( 4 2 . W ) . O). is the maximum at ( 2 1 Y.2: Transverse deflections and stresses in specially orthotropic square laminates subjected t o various types of mechanical and thermal loads (E1/E2= 25.y. To = 0).5 show the distributions of the maximum inplane normal stresses a. z) = (a/2. UDL=Uniformly distributed load.5. ul2 = 0. . CPL=Central point load. El = 25E2. In the case of uniform and point source distribution. all laminates are of the same total thickness. G12 = GI3 = 0. b/2. z) = (0.2.2. b/2. 1 Laminate Mechanical Thermal SSL* * SSL=Sinusoidal load. is the maximum at ( x . the first ten terms of the double trigonometric series are evaluated. G12 = GI3 = 0. G12 = GI3 = 0. for symmetric crossply (0/90/90/0) laminates under uniformly distributed (UDL) and sinusoidally distributed (SSL) loads.25. The transverse shear stress a. the number in parentheses denotes the number of terms used in the double Fourier series to evaluate the series.2. respectively. and ayy.
rectangular (0/90/90/0)laminates UDL = Uniformly distributed load SSL = Sinusoidally distributed load   0.0 Figure 5. .2.000 0.2.. a 1b Figure 5.0 1.3: Nondimensionalized maximum normal stress (a.0 2. 0.20 Simply supported.0 4.0 P l a t e aspect ratio.0 3.2: Nondimensionalized maximum transverse displacement w = ~ ~ ( ~ ~ h plate aspect ratio (~ l b ) of symmetric crossversus ~ / a ~ ~ a ply ( 0 / 9 0 / 9 0 / 0 ) laminates. rectangular (0/90/90/0)laminates UDL = Uniformly distributed load SSL = Sinusoidally distributed load P l a t e aspect ratio.4 SSL UDL  Simply supported. a 1b 5.) versus plate aspect ratio ( a l b ) of symmetric crossply ( 0 / 9 0 / 9 0 / 0 ) laminates.
bl2..Simply supported square laminates under sinusoidally distributed load All laminates are of the same total thickness 4. (al2.2. (al2. Simply supported square laminates under sinusoidally distributed load... All laminates are of the same total Stress.. . z ) Figure 5.4: Variation of nondimensionalized maximum normal stress (azx) through the thickness ( z l h ) of square crossply laminates under sinusoidally distributed load.5: Variation of nondimensionalized maximum normal stress (ayv) through the thickness ( z l h ) of square crossply laminates under sinusoidally distributed load. z ) Figure 5.2. bl2... Stress.
0. b to reduce the partial differential equation (5.1) to an ordinary differential equation with respect t o the coordinate x. and (a/2.6: Variation of nondimensionalized maxirnuni transverse shear stresses. 5.2. 0 . ply laminates.2. The other two edges a t x = 0.(0. independent of the other. the Navier solution cannot be developed. the load is represented as .z) Figure 5. b and subjected to a transverse load q. can each be free.Stresses. the idea of the Navier method can be applied with respect to the simply supported boundary conditions at y = 0.@ through the thickness ( z l h ) of square cross. @y..3 Bending of Plates with Two Opposite Edges Simply Supported 5. The solution to the problem of a rectangular plate with two opposite edges simply supported arid the other two edges having arbitrary boundary conditions can be represented in terms of single Fourier series as n= I Similarly.~) . which may then be solved exactly or approximately.3. simply supported. a . For such problems. The stresses are the same in both laminates. and . This procedure is known as the Lkvy method.1 The Lkvy Solution Procedure Consider a rectangular plate with simply supported edges along y = 0.612. or clamped. However.
. 4(x) = ( 4 o ~ l b ) Y Qn = a (.1: Coefficients in the single trigonometric series expansion of loads in the L6vy method. 2 . Q(X) = Qo at Line load. Load q(x) Uniform load.) Point load. 4 = 40 Coefficients Q.where Q.(x) are given by (see Table 5. x (n = 1 . .1) Table 5.l ) n + l .3. Hydrostatic load. 3 .. .3.
5) can be solved either analytically or by an approximate method. b.h( z ) = C exp (Xz) where X denotes a root of the algebraic equation (5.3.3. it follows that the expression in the square brackets must be zero: The ordinary fourthorder differential equation (5. As for approximate methods. we obtain Since the result must hold for any y. the roots are real and unequal: ) 0 . We consider three cases.3. (5. The true form of the solution depends on the nature of the roots. i e .2.1). the solution (5.1) into Eq. the coefficients Q .1) satisfies the simply supported boundary conditions on edges y = 0.5) consists of two parts: homogeneous and nonhomogeneous (or particular) solutions. . . Eq. the Ritz.2 Analytical Solutions The general form of the analytical (exact) solution to the fourthorder differential equation (5.3.2a) and (5. finite difference. analytical solution by the statespace approach. Here we discuss direct analytical solution.The assumed solution in Eq.3. The homogeneous solution is of the form W. Case 1: Roots are real and distinct When (Dl2 + 2 D ~ s > ~ 1 1 0 2 2 .12]). Analytically.6) Since there are four roots. are given by Substituting Eqs. In the case of uniformly distributed load of intensity qo. (5.3.3. real or complex and equal or distinct. (5.3. and approximate solution by the Ritz method. and finite element methods are good candidates. (5.5) can be written as a linear combination of functions of these four roots. 5.5) can be solved directly or by the socalled statespace approach used in control theory (see [11.3.
3. or Case 3 ) in a problem is dictated by the plate stiffnesses. Dij.5).3. cos X2x + D. cos X2x x( + B. + B. X I > 0.. X 2 > 0): + The homogeneous part of the solution is of the form w . sinh X3x +W z )sin p y ( A .15) . P particular solution becomes ~ ~ the ~ The four constants A. Hence.e.3.13) can be determined using the four boundary conditions associated with the edges x = 0. Case 2. The solution in this case is given by 0 0 w~(x) = y.z) cosh Xx x( Case 3: Roots are complex + ( C . sin X2x)cosh X1x + ( C .l l ) .3. We obtain ~ D = Q.. B. cosh XSx + Drlsinh Xsx (5. Case 1.3. C.5) in the general case in which Q. sinh X1x + C . a (in addition to the simply supported boundary conditions on the edges y = 0.. the roots are complex and they appear in complex )~ conjugate pairs X I f iXz and Al f iXa ( i = &f. Here we illustrate the procedure for simply supported and clamped boundary conditions in the case of real and distinct roots.3. Note that the particular case (i. sinh X12: + C .13) The particular solution of the fourthorder differential equation (5. and D. cosh X3x +D. ~ (= ) A .3. b). sin X2x)sinh Xlx (5. (5.When Q .3. (5. is a constant the particular solution is a constant k . and the homogeneous part of the solution is of the form w .9) Case 2: Roots are real and equal When ( D l a +2 ~ ~ =~D ) I '~ Dthe roots are real but equal ~ ~ .11) When (Dl2 2 0 ~ 3 < 0 1 1 0 2 2 .. h ( x= A.z) sinh Ax (5.. ~ (= ) A . cosh Xlx ) + B .9). and (5. is a function of x can be determined using the method of undetermined coefficients (see Pipes and Harvill [ l o ] ) .The homogeneous part of the solution is of the form ~ . and it is determined by substituting it into Eq.3. (5. in Eqs. n=l + B. + D. cosh Xlx (5..
Qn A. / ..X: cosh Xsa + D. 0 4 ~ 2 zBy virtue of the first two equations. we obtain A.3.A:)  Simply supported at y=O. cosh Xsa + D.17) (5. cosh X3a + D. ( A: AA ) A: (1 .. sinh Xla + C.3. s i n Xla 0 cosh Xla A sinh Xla : Xi 1 cosh A3a A. a are Using (5.  A?) B n = Qn A: (1 . in the last two equations are identically zero. x : ) ~ X I a sinh Xsa. D l l (A. The solution of the matrix equation yields sinh I{ ] D.cosh Ala) .coshXsa) Dn = Qn sinh Asa (A: .:$a sinh X3a 0 X3 sinh Xsa X g cosh X3a 1 1$1 D.  {} (5. slnh Xla ( X i .16).In the following discussion we assume that the applied transverse load is uniformly distributed.17) is (A: The determinant of the 4 x 4 coefficient matrix in Eq. we require which yield I 1 cosh Alu 0 X I sinh Xla O sin11 Xla XI X1 cosh Xla 1 0 o h . =  {f ] An = . = Q .X: sinh hla + C. Xla + C.b and clamped at x=O. a. Simply supported plate The simply supported boundary conditions on edges x = 0.3. sinh X3a + Q.a For clamped boundary conditions on edges x = 0.X: cosh Xla Dl2 + B..20) .3. The four equations can be expressed in matrix form as 0 1 cosh h. hi sinh h3 ~ ) (An Ala + B.=Q. 1 (A.. the coefficients of D I 2 .3. cosh X3a 0 sill 0 sinh X3a (5.A:) c. sinh sinh Xsa +9 =0 .15) in (5.) p2 = 0 cash where Q . cosh Xla + B.
firstorder matrix differential equation The general solution of Eq.5) with constant coefficients can be expressed in the form of a single.l)] n Bn = ~ En h [AS sinh Xsa (cosh Xla  1) + X1 sinh Xla (1  cosh Xsa)] where En is the determinant of the coefficient matrix En = .3. and its solution is obtained using matrix methods in terms of the eigenvalues of the matrix operator.(AY sinh Xla .The solution of the matrix equation (5. [E]' denotes its inverse. .3.3.3.3. (5.2.cosh Xsa) (cosh Xya . Xj ( j = 1. (5.cosh la)^ a  Xy sinh X3a) (5.20) is An =~ n k j + [(A1sinh Xsa .X1 sinh Xya) (A1 sinh Xla +A1& (cosh ~ g .5) is provided by the statespace approach [12].22) An alternative method of solving Eq. The approach involves writing a higherorder ordinary differential equation as a firstorder matrix equation.Xg sinh Xla) sinh Xya En X1 (cosh Xla .4) are the eigenvalues associated with matrix [TI.23) is given by e G(x)K + H(x) 0 Here eTx denotes the matrix product eTx = [El ex4x Here [El is the matrix of distinct eigenvectors of matrix [TI.and {K) is a vector of constants to be determined using the boundary conditions of the problem.3. the linear ordinary differential equation in (5. In the present case.
As an example.24. square laminates under uniformly distributed transverse load. of course.3.2 msi. edges y = 0. 4 ) : j=1 4 x 4 GI. for example. The lamina material properties used are El = 19.2.26). the procedure is algebraically complicated.2813) These four conditions in turn yield. 2 . crossply (0°/900/00).. (a)K j + Hl (a) = 0 + H2 ( a ) = 0 G2j(a)K j j=1 These equations can be solved for the four constants. is used to denote a plate with edge x = 0 is simply supported (S) and edge x = a is free (F).2 contains numerical results for threelayer.56 msi.l)] The clamped boundary conditions (5. The transverse deflection and stresses are nondirnensionalized as follows: The notation SF.e.3.19) a t x = a imply Wn(a) = 0. E2 = 1. consider the case of simply supported boundary condition a t x = 0 and clamped boundary condition a t x = a. In general. the following four nonhomogeneous algebraic equations among Ki(i = 1 . and therefore all calculations. G12 = 0.2a.3. and 242 = 0.b are simply supported. matrix multiplication. are made using a computer. .b) at J: = 0 imply [see Eq. (5. Table 5.3. (5.3. and evaluation of the solution. The simply supported boundary conditions (5. in view of Eq. 3 . WA(a) = 0 (5. i.82 msi. determination of eigenvalues and constants Ki.
5) can also be solved using the Ritz method.5) be satisfied: where SW.. and ayy) of symmetric crossply (0°/900/00) square plates subjected to uniform distribution of transverse load and for various boundary conditions. we obtain Since the above expression must hold for all arbitrary values of Sci.3.3.33) into (5.3. we seek solution of (5.3. In the Ritz method. (1.31) and (5.2)]. We have N 0=x A u c j j=1  Fi or [A]{e} { F ) = .3.3. Variable SS SC CC FF FS FC 5. (5.2: Nondimensional center deflections (w) and inplane normal stresses (a. Substituting (5.32). it follows that the expression in the curly bracket must be zero.5.3 Ritz Solution Equation (5.3. The parameters cj are then determined by requiring that the weak form of Eq. denotes the virtual variation in W.5) in the form where cpj(x) are approximation functions that must meet the continuity and completeness conditions and satisfy the homogeneous form of the geometric boundary conditions [see Eq.3.Table 5.
the approximation functions cpi must be selected such that cp. we have . The constant is arbitrary and may be set to unity. a are clamped. = 0 and (dp. a .3. The geometric boundary conditions are given by Eq. (5. As an example.35a) represents a set of N algebraic equations among ci. We obtain The ith function can be written as For the choice of pi(x) in (5. If an algebraic polynomial is to be selected./dx) are zero atx = 0. one may begin with the fiveterm complete polynomial and determine four of the five constants Kiin terms of the remaining constant using the four boundary conditions.wherc Equation (5.19): Hence. we consider the case in which the edges x = 0.3.37).3.
3.2 msi. we have cpi (x) = sin Xix .001988qoa4.1) becomes with p = nnlb and The center deflection is given by a b wo(.sinh Xix + ai (cosh Xix  cos X i 2 ) (5. (5.001774qoa4. for h = 0. are Dl1 = 1.35a) gives and the solution (5. For example.3.031347. we use the eigenfunctions of a beam with clamped ends. . and Dss = 0. (5.82 msi. From Eq.46a) and Table 4.2.42a) . G12 = 0.For N = 1. the maximum deflection becomes The series converges slowly unless we also increase the number of parameters in the xcoordinate [see Eq.46a)l.31)].25. (4. (4. .2. 3 .5528.18531. Dl2 = 0.3. . and ul2 = 0. (5. 1. For a square plate.068333 lbin. The "exact" solution for a square laminate under uniformly distributed load is whereas the oneterm (n = 1 and N = 1) solution predicted by Eq.40) is 0.2. 3 and N = 1) is 0.56 msi.01. 5 .3. we have (s = alb) E2 = for n = 1 .. the bending stiffnesses. for clamped boundary conditions. The twoterm solution ( n = 1 . For uniformly distributed load qo. Dza = 0. ) 2 2  l C O 16 n=l nn cl(n) sin 2 For a symmetric crossply laminate (0/90/0) with ply properties El = 19.3. Other choices of cpi(x) are provided by the eigenfunctions W(x) of beams developed in Chapter 4 [see Eq.3. Eq.
4 Bending of Rectangular Plates with Various Boundary Conditions 5. the center deflection ( X l ( a / 2 ) = 1.2.8532 cp2 ( x ) = sin .3.4. In this section.61637) predicted for n = 1 is 0 .* dx and = X i [cos Xis  cosh Xix + ai (sinh X ~ X sin X i x ) ] + (5. Recall from Table 4.99922.3.3. (4. a. Eq.2.3.43) are given by The corresponding values of cul = ai are ai = 1. . (5. a:! = 0. pi and ( d p i / d x ) are zero at x = 0.44) Clearly.732 4.0178. 0 0 1 7 9 5 ~ ~ a ~ .sinh a a 7.8532 7.47) For N = 1.3.60)] that the roots Xi of the characteristic equation (5.sinh a a . Therefore.59) ct!i = sinh Xia cosh Xia sin Xia .cos  a  cos a (5. 5. 1 for i >2 Hence.2.4213) where Xi are the roots of the characteristic equation (4.1 Virtual Work Statements The Navier and Lkvy type solutions do not exist for rectangular plates with all four edges clamped or when two parallel edges are not simply supported.cos Xia   cosh Xia sinh Xia  + cos Xia sin Xia (5.3 [also see Eq. the first two eigenfunctions are 4. an approximate method must be utilized to determine solutions of these plates. 0 0 2 0 0 9 ~ ~compared to the exact a~ solution of 0 .732 cpl ( x ) = sin . we discuss applications of the Ritz method to determine the bending deflections of specially orthotropic rectangular plates with various boundary conditions.35a) yields For the symmetric crossply laminate considered above.
42a): . (5. the approximation functions pij(x.3a. (5.4. a (5. For this problem.37) or (5.4.3a) We assume the Ritz approximation in the form whe:e the approximation functions pij satisfy all the (homogeneous) geometric boundary conditions in Eqs. y ) . The boundary conditions associated with the clamped plate are wo = 0 and  aw 0 ax =O at x = O .2 Clamped Plates Consider a rectangular plate with all edges clamped and subjected to distributed transverse load q(x. (3.3. y) can be expressed as a tensor product of the onedimensional functions given in Eq.19)] and The above expressions should be appended with appropriate terms due to any additional applied edge forces and moments. both the Galerkin and Ritz methods give the same solution for the same choice of approximation functions. therefore.3. In view of the rectangular geometry and clamped boundary conditions.b). 5 4.3.The virtual work statement (or weak form) and the total potential energy expressions for a specially orthotropic rectangular plate are [see Eq.
4. . .4). (5. and they involve evaluating five different integrals La Xi dx. The parameters Xi and ai are defined in Eq.. m. .4. .8b) represents m x n algebraic equations among the coefficients cij.where Xi(x) = sin Xix T ( y ) =sinXjy  sinh Xix . the expression inside the curly bracket should be zero for all p. Note that all integrals in (5.: Equation (5. (5. 2 . .3. q = 1 . . and into Eq.44). (5. la XiX. respectively.5). with cpij given by Eq.4. . dx. (5. we obtain d2y+ Da2Xi+Xp$] dy dxdy Since the statement should hold for any arbitrary variations hepq. n. j = 1 .810) are line integrals. .4.sinhXjy + ai (cash Xix  cos Xix) + n j (coshXjy  cosXjy) (5.7) for i = 1 . . 2 .4.43) and (5. 2 .3. . Substituting Eq.4. laz2 dr .1).
73~ 4.9a) for this case are given by Substituting the integral values into (5. the functions in (5.7).sinh a a 4. The maximum deflection occurs at x = a/2 and y = b/2: The algebra involved in evaluating the integrals in Eq.913) gives  COS  a  CoS 4.4. An algebraic manipulator (e.g.4.73~ 4.6) with m = n = 1 and q = qo (uniformly distributed load).b b and substitution into (5.4.7) are given by 4.73~ Yl(y) =sin . For m = n = 1. (5. dx2 " d2xid 2 x dx As an example we consider the algebraic functions in (5.8b)..73~ X I (x) = sin .4.9a) is quite tedious for the choice of approximation functions in (5. Maple or Mathematica) may be used to evaluate them.sinh .73y b j (5.4.4. (5.4. The integrals in Eq.13a) .4.Lax i 2 d2x dx. we obtain and the oneparameter solution becomes where s = a/b denotes the plate aspect ratio.
3) of beams with corresponding boundary conditions.3 Approximat ion Functions for Other Boundary Conditions Here we discuss the approximation functions pij = X i ( x ) q ( y ) required in the Ritz approximation (5. For an isotropic square plate ( a l b = 1 .cos Xia .4.531348 a3 (5. the maximum deflection (5. (5.sin Xia cosh Xia .2.8b) becomes The maximum deflection is given by ( X l ( a / 2 )= Y l ( b / 2 ) = 1. Clamped at x = 0 . a and simply supported at y = 0.4.6164) where s = a / b denotes the plate aspect ratio.4.12) gives The "exact" solution (see Timoshenko and WoinowskyKrieger [ 6 ] ) is 5.15) becomes + whereas Eq.dx= 12.13b) Then Eq.4) of specially orthotropic rectangular plates with a variety of boundary conditions (see Hearman [ 8 ] ) . Dl1 = D22 = Dl2 2Dss = D ) .4. b X i ( x ) = sin Xix  sinh Xix + aii (cosh Xix  cos Xix) 5( 9 ) = sin Pw b where Xi are the roots of the characteristic equation and Qi = sinh Xia .4.4.7442 a a d 2 x 1d 2 x 1 d x = 518. (5.The choice is restricted to the products of eigenfunctions (see Table 4.
b. a and simply supported at y X ~ ( X= sin X i % ) + sinh Xix  + cos Xix) q ( y ) = sin j.iry b where X i are the roots of the characteristic equation cos Xia cosh Xia . and clamped at x = a X ~ ( X= sinh Xis sin Xix ) + sin Xia sinh Xiz where X i are the roots of the characteristic equation tan Xia  tanh Xia = 0 Simply supported at x = 0 and y = 0. b sinh Xix + ai (cosh Xix  cos Xix) U.cos Xis Simply supported at x = 0 and y = 0.(y) = sin j w where X i are the roots of the characteristic equation cos Xia cosh Xia and ai = + 1= 0 sinh Xia cosh Xia + sin Xia + cos Xia = 0.sin Xis sinh XZx U.sin Xia cosh Xia . b. and free at x = a X i ( x ) = sinh Xia sin Xix .270 MECHANICS OF LAMINATED COMPOSITE PLATES AND SHELLS Clamped at x = 0. free at x = a.( y ) = sin ~ T Y b where X i are the roots of the characteristic equation tan Xia  tanh Xia =0 .1 = 0 and Qi = sinh Xia . and simply supported at y X i ( x ) = sin Xix  = 0. b CX~ (cosh Xix Free at x = 0.
Of course.21a) % (y) = sin pI.22a) Y.~(cosh Xiz i cos Xix) (5. and free at x = a and y = 0. the Ritz method for general rectangular laminates with arbitrary boundary conditions is algebraically more complicated than a riurnerical method.4.(y) = sinh pjb sin pjy  sin pjb sinh p j y where Xi and p j are the roots of the characteristic equations cos Xia cosh Xia + 1 = 0.4. The phenomenon of changing the equilibrium configuration at the same load and without drastic changes in deformation is termed bzfurcation.4. the stable state of the plate is disturbed and the plate seeks an alternative equilibrium configuration accompanied by a change in the loaddeflection behavior. such as the finite element method. = 0. .1 Governing Equations When a plate is subjected to inplane compressive forces.4. One may use a syrribolic manipulator. called the buckling load. sin pi b cos b (5.4. tan pjb and Qi  tanh p.5.Clamped at x = 0. The plate remains flat until a certain load is reached. the equilibrium of the plate is stable (see Figure 5. 5. to evaluate the integrals.4. and free boundary conditions on the four edges of a rectangular plate. cos pjbcoshpjb sinh pa b  (5.22b) = sinh Xia cosh Xia + sin Xia + cos Xis Similarly. b X i ( x ) = sin X ~ X sinh Xix  + ai (cosh Xix   cos Xix) (5. one can construct the approximation functions for any combination of fixed.y + sinh / L j ? j pi (cosh p j y + cos pjy) 1= 0 where X i and pll are the roots of the characteristic equatior~s cos Xia cosh Xia and ai = + 1 = 0. and if the forces are sufficiently small. N ~ < 0.z + ( . products of the beam eigenfunctions can still he used for the approximation of the transverse deflection with appropriate functions for the inplane displacements. N~~ < 0.813).jb = 0 (5.21b) sinh Xis cosh Xia + sin Xia + cos Xis  "= cosh p j b = 0.1). In general.5 Buckling of Simp1 Supported Plates Under Compressive Loa s B 5. (5. hinged. fir. At that load. and . the most difficult part is to evaluate the integrals of these functions as required in Eq. such as Mathernatica or Maple. simply supported at y X i ( z ) = sin Xiz  and free at z = a and y = b  sinh X. When general laminated plates are considered.21~) Clamped at z = 0.5.
(5.2. (5.2) 5. sin a x sin py (5.3) Substituting Eq.5. (5. The magnitude of the buckling load depends.1) when the inplane forces are N~~ = No.5. (5. sin a x sin Py (5. Here we determine the critical buckling loads of simply supported specially orthotropic plates using the Navier method. NYY k: = &x (5. is given by (see Section 4.1). as well as on the buckling mode shape. we assume that the only applied loads are the inplane forces and all other mechanical and thermal loads are zero.4) Figure 5. as will be shown shortly. For the buckling analysis.. we obtain (for any m and n) x W.2.5. . and the edges are simply supported. We wish to determine a nonzero deflection wo that satisfies Eq.The loaddeflection curve for buckled plates is often bilinear.3) into Eq.5.3) For simplicity. Since the prebuckling deformation wo is that of an equilibrium configuration.~).2 The Navier Solution As in the case of bending. NyY = kNo.5.5. we will omit the superscript "b" on buckling deflection w.5.. material properties.y) = W..2) wo(x. and the equation governing buckling deflection w.1: Buckling of a plate under inplane compressive edge forces = N. it satisfies the equilibrium equations. (fizz= .5. we select an expansion for wo that satisfies the boundary conditions in Eq.~ g4. on geometry. .
5a) yields Dz2.n).. For a given laminate this value is dictated by a particular combination of the values of m and n. for each choice of m and n there corresponds a unique value of No. The critical buckling load is the smallest of No(m. we obtain R = n = 1. D12 = vD. and the critical buckling load becomes For an isotropic (Dll = D22 = D .. # O). Then D l l m 2 increases more rapidly than the decrease Now suppose that D l l in ~ ~ with an increase of m. (5. the minimum of No occurs when m = 1: ~ / m > The buckling load is a minimum when n is the nearest integer to the real number R For example.e. biaxial compression with k = I ) . 5. W. We investigate critical buckling loads of various laminates next. (5. and 2Dss = (1 .5.~Thus. Eq. we obtain R = 1.u)D) square plate under biaxial compression.52 or n = 2.3 Biaxial Compression of a Square Laminate (k = I) For a square laminate subjected to the same magnitude of compressive load on both edges (i.5.5.6): .e. the critical buckling load becomes fi or For modulus ratios of MI = 12 and M2 = 1. the buckling can be calculated from Eq. Hence. for modulus ratios of MI = 10 and M2 = 1.. This yields where Thus.y) of the domain for nontrivial wo (i.Since the equation must hold for every point (x. the expression inside the curl brackets should be zero for every m and n.
(5. For example. Since the value of m from Eq.= No and the edges y = 0. the minimum buckling load occurs at m = 1 and n = 1.3.and the critical buckling load occurs at m = n = 1.5. we have = 0) An examination of the expression in Eq.5a) becomes for n2 < m2/k.16) when m is the nearest integer value given by Eq.5. and it is equal to 5. (5. Eq.16) with respect to m. (5. (5.17) into Eq. the minimum buckling load cannot be predicted by substituting the value of m from Eq.5.16).5. The minimum value of No is given by Eq.5. a are subjected to compressive load N ~ . we have 5.5. occurs for n = 1: The critical buckling load is then determined by finding the minimum of No = No(m) in Eq.17) is not always an integer.5.5. (5.5.5. (5. (5.5. b are subjected to tensile load Nyy = kNo.17). For the isotropic material properties used in Section 5.15) shows that the smallest value of No. for any m. Since the value of m depends on the ratio of the .5.5 Uniaxial Compression of a Rectangular Laminate (k When k = 0 (Nyy= 0). (5. when k = 0. We have = dNo dm 0 gives m = Dl 1 The second derivative of No with respect to m can be shown to be positive.4 Biaxial Loading of a Square Laminate When the edges z = 0.
5. For aspect ratios less than 2. The critical buckling load of a laminate with is given by N 1 1 = 7r2 1 ( ) + 2 + (i)2] 9 (5.66.66 and 4. we must investigate the variation of No with aspect ratio a l b for different values of m for a given laminate. we have from Eq.r2D22 b b2 ' b b2 It can be shown that if the laminate aspect ratio a l b is greater than 2. (5.1 x (1.5.66 but less than 4. for aspect ratios (alb) less than 2. = 9.5. Thus larger aspect ratios lead to higher modes of buckling.778)~ 0. we have a 7r2D22 a 2 : N. for a l b = 3. we have Thus the closest integer is m = 1.5. Then we have = r2D22[ L( 71 0 m 2 ( f ) 2 + 2 + m2 ~ ) 2 ] (5.3). the plate buckles into two halfwaves in the xdirection (and one halfwave in the ydirection). For example.5.5 : N.21) .17)]..18a) with m4 = Dl 1 D22 (x) 4 = 0.5. the buckling load is the minimum for n = 1 and m = 2 [using Eq. = 8.principal bending stiffnesses as well as plate aspect ratio.44.18b) In fact.44. consider a laminate with Dll/Dz2 = 10 and a l b = 1.5.2 contains a plot of the nondirnensionalized buckling load No = ~ ~ b ~ /versus plate ~ ~ ~ ) a l b ( r ~ aspect ratio for laminates whose material properties are Dll/Dzz = 10. Figure 5. As an example.= Thus.21) For various aspect ratios.. . (Dl2+2Dfj6) = 0 2 2 .= 2. the plate buckles into more and more halfwaves in the xdirection.5. (5. for aspect ratios between 2.778.85. As the aspect ratio increases.9994 = =1 (5. the plate buckles into a single halfwave in the xdirection (see Figure 5.5. Note that intersections of two consecutive modes .
NO.0 2. N = NO (. alb b2/ Figure 5.5 4.0 Plate aspect ratio.0 1.5 3.3: Nondimensionalized buckling load.I Simply supported rectangular laminates m=l (m = mode number) 16.ir2 ~ plate aspect ratio a/b. wavelengths m in the xdirection.5 1.0 Number of half wavelengths in the xdirection.2: Nondimensionalized buckling load. versus number of half . 2 2 ) .0 3. m Figure 5.5.0 0.versus I I I I Simply supported rectangular laminates (ah = plate aspect ratio) 0.5 2.5.
. 20. 2 4 2 = 0. Thus.5. = 40E2.correspond to certain aspect ratios (see Figure 5. for two different materials: Material 1: Material 2: El El = 25&.5.4 shows plots of nondimensionalized critical buckling load N = as a function of the plate aspect ratio. all layers of equal thickness). For the data in Eq. 242 = 0. for each of these aspect ratios. this limiting value of the critical buckling load is For a square isotropic plate (Dll = 0 2 2 = D .17)]. and 2Ds6 = (1.1 shows the effect of plate aspect ratio and modulus ratio (anisotropy) on the critical buckling loads N = N O ~ ~ / ( . under uniform compression (k = 0) and biaxial compression (k = 1) (E1/E2varied. except for a/b = 0.2). we have m = 1 [from Eq.5E2. and 40. The No versus a l b curve gets flatter with the increasing aspect ratio.5.3) for modulus ratios 5. Figure 5. Table 5.5 and k = 1. for which case the modes are (1.5. G23 = 0. ~~D~~) under uniform compression (k = 0) and biaxial compression (k = I ) . and it approaches the value which is obtained from Eq.25 There is a mode change around a l b > 2. The nondimensionalized buckling load increases as the modulus ratio increases. and the critical buckling load from Eq. GI2 = GI3 = 0. respectively.l)to (1.of rectangular laminates (0/90). (1.v)D).2).2 from (1. . 10.5. (5. vl2 = 0.16) after substituting for m2 from Eq.2E2. In all cases the critical buckling mode is ( m . (5.16) is Table 5.25.6E2. (5. n ) = (1. D12 = vD. (1. (1.20). GIZ = G13 = 0. and (1.5.5. (5.1).25 G12 = GI3 = 0.17). there are two possible buckled mode shapes.5.5E2. alb.5.3).l ) .2).2).1: Effect of plate aspect ratio and modulus ratio on the nondimensionalized buckling loads N of rectangular laminates (0/90). (5. 25.
the Navier solution does not exist because the cross derivative term involving N& will have a different coefficient (cos a x cos py) than the rest of the expression in Eq. we use a variational method to solve the problem. = 0 and &. = 2N& Dl1 . When N ~ = fiy.2 Simply Supported Plates When the plate is simply supported on all its edges and subjected to inplane shear. N.6. a l b Figure 5.0 2.1) 5.1). Hence. (5. = N& (see Figure 5.5 Plate aspect ratio. we will seek the solution by a variational method.6.6 Buckling of Rectangular Plates Under InPlane Shear Load 5.1).6.1 Governing Equation In this section we consider buckling of specially orthotropic rectangular plates under inplane shear load.3) for wo satisfies the geometric boundary conditions of the problem.1) takes the form W~ a4w. the governing equation (5.0.The problem does not permit the Navier solution.. for two different modular ratios.5 2.0 1.5 3. the same functions are admissible in the Ritz method: . therefore.5. (5.0 3.5 1.0 0. .&.5. (5.5.5. 5.6.4: Nondimensionalized uniaxial critical buckling load (N) versus plate aspect ratio ( a l b ) of symmetric crossply laminate ( 0 / 9 0 ) .. Since the expression given in Eq.2 (D12 2DGs)axay ax4 ax2aY2 + DZZ ay4  a4 + + a4~o a2~.
dx = a { 2. Using the Galerkin method. m f n m=n sin Ax cos px dx = we arrive at + (96.8b) .2) in the total potential energy functional for the Ritz method or tlie weightedintegral statement for the Galerkin method would lead to the same equations for the coefficients em. Since the approximation functions (5.6.6. 0.. substitution of Eq. (5.m2)(q2.sin Y) a b also satisfy the natural boundary conditions of the problem.n2) q2 # n2 (7. Thus.6..where ol =mr/a and /3 = n7i/b. the Ritz and Galerkin solutions are the same.7) where Pq for p2 # rn2 and (p2 . nrx sin . we obtain Using the identities a sin mrx .3) mrx nry (~rnn(x. = sin .
6.3 Clamped Plates The total potential energy expression for the clamped rectangular plate under inplane shear load NZv is o = 2 0 // o [ I ( )+ a2wo 2 2" " a2wo a2wo ax2 ay2 + 46 6 .7]).6.and the coefficients are zero when p = m .sinh Xjy  + ai ( C O S ~ . We have a6wo awe + ) dwo a6wo ay dx dy ] dxdy We assume a Ritz approximation of the form where with or X ~ ( X ) sin Xix = sinh Xix 5( 9 ) = sin Xjy .9) converges very slowly with increasing values of M and N (see [3.e.9) requires an eigenvalue routine that is suitable for nonpositivedefinite matrices.. () a2wo axay 2 The minimum total potential energy principle requires that SII = 0.6. or when q = n.6. p f m even. q f n even. the solution of (5. Note that [A] is a diagonal matrix while [S]is a nonpositivedefinite matrix. hence.8a) define an eigenvalue problem which has a nontrivial solution (i. It is found that the solution of (5.cos X ~ X ) Xix + (cash Xjy cos X j y ) ~j  . cmn # 0) when the determinant of the coefficient matrix is zero. 5. The set of m n homogeneous equations (5.
45) and (5.8532 7. The parameters X i and a of Eq.sinh b b 7. .853~ Y 2 ( y ) =sin . Using the approximation with [see Eq.3. .532b Dll t a3 or in matrix form where . n = 2 and m = 2.732 XI ( x ) = sin . 2 .3. 2 .46).6.73~ Yl( y ) = sin . . n. Substituting Eq. (5.47)] 4.cos  b we obtain [ 3791."~ zero for m = n = 1. .sinh b b  cos a  cos .10b) we obtain When functions in Eq. (5.6.8532 X2( x ) = sin .6.73~ 4. (5.853~ 7. a t least two terms should be used because the coefficient of N. (5. (5.sinh a a .sinh a a 7.13) are used. j = 1 .6.73~ 4. . . n = 1. respectively.13) are i defined in Eq.. .for i = 1 . (5. m.a  cos  b .11) into Eq.. 4.3. . other coefficients are zero for is m = 1 .
we obtain Y) The f sign indicates that the shear buckling load may be either positive or negative. the determinant of the coefficient matrix should be zero.1 Governing Equations For natural vibration.6. (5.6. we have a = b and Dll = D22 = ( 0 1 2 2Ds6) = D .2 Solution We assume a periodic solution of the form .7.a 1 2 a 1 2 ( ~ ~= 0.18) is over 21% in error. The variational solutions presented here for buckling under inplane shear are only for illustrative purposes. a symbolic manipulator proves to be effective in evaluating the integrals in the variational methods. 5.7 Vibration of Simply Supported Plates 5. Once again. alla22 . More than twoterm variational approximations are required to obtain accurate buckling loads.For a nontrivial solution.17) is + whereas the "exact" critical buckling load is The twoterm Ritz solution (5. all applied loads and the inplane forces are set to zero in Eq.1) where where L denotes the total number of layers in the laminate. This concludes the discussion of shear buckling of rectangular plates. For an isotropic square plate. and the shear buckling load predicted by Eq.2 Solving for the buckling load N&. (5.7.1. 5.
4) reduces to w:.3) Since the equation must hold for every point (x.7.. (5. = fland w is the frequency of natural vibration. y) of the domain 0 < x < a and 0 < y < b..2) in i (5. the frequency of a rectangular specially orthotropic laminate reduces to and for a square plate we have = The fundamental frequency occurs at r r ~ n = 1: .7. Substituting (5.~ m7rx sin a  n7ry sin b where is the amplitude of the vibration mode (m. For square laminates. When the rotatory inertia I2 is not zero. n)..7.7. = w:. Eq. we obtain (for any m and n ) x W. the expression inside the braces should be zero for every m and n. The rotary inertia has the effect of reducing the frequency for any m and n. When the rotary inertia I2 is neglected. This yields where For different values of m and n there corresponds a unique frequency wmn and a corresponding mode shape w0 y) (x.la).where .. sin a x sin py = 0 (5. it is not simple to find the lowest natural frequency (fundamental frequency).
1 shows a plot of nondimensionalized fundamental frequency G l l as a function of plate aspect ratio for symmetric (0/90).3)..8) becomes and the fundamental frequency is given by Nondimensionalized frequencies.7. (2. w . n ) = ( l . (1. laminates for various aspect ratios and modulus ratios. 4 2 = 0.1 are for m . Table 5.2E2.w. 2 . .7. ( m .7. and the effect is negligible for this case. l ) . wmn (b2/. The fundamental frequency increases with modular ratio.7.2 contains nondimensionalized fundamental frequencies of symmetric (0/90). .1). and (1.25).2). whereas for symmetric crossply plates the first four frequencies are provided by the modes: (m.(b2/.3). Gag = 0.25).1).n ) = ( l .ir2) JphlDzz). The results presented in Table 5.5E2. The effect of including rotary inertia is to decrease the frequency of vibration.7. Table 5. G12 = GIS = 0. n = 1 . the frequency equation (5.5E2. and (2. 242 = 0. when the rotary inertia is neglected. l ) . Figure 5.rr2) q'phlD22. (1. of specially orthotropic square laminates are presented in Table 5.1: Nondimensionalized fundamental frequencies of symmetric cross& = ply laminates according to the classical plate theory ( . (1.7.20 (G12 = G13 = 0. graphiteepoxy laminate (E1/E2= 40.284 MECHANICS OF LAMINATED COMPOSITE PLATES AND SHELLS For a rectangular isotropic plate. 3 . and for the case in which the rotary inertia is neglected.1 for modulus ratios E1/E2 = 10. The first four frequencies for an orthotropic (0") plate correspond to the modes.2).
8. and free boundary conditions on the edges of rectangular plates.7.7.3. as described in Section 5. one may use the Ritz method with the approximation functions suggested in Section 5.1: Nondimensionalized fundamental frequency as a function of plate aspect ratio a l b for symmetric (0/90).2: Nondimensionalized fundamental frequencies G l l of symmetric crossply laminates (0/90). (0/90) symmetric laminates Plate aspect ratio. according to the classical plate theory. rectangular.Simply supported. hinged. a l b Figure 5.4.1 Introduction The L&y method can be used to deter~nine natural frequencies and critical buckling loads of rectangular laminates for which two (parallel) opposite edges are simply supported and the other two edges have any boundary conditions.3 for bending analysis.8 Buckling and Vibration of Plates with Two Parallel Edges Simply Supported 5. . Table 5. laminate. Without Rotary Inertia a/b With Rotary Inertia 40 10 20 30 40 2 = 10 20 30 5. For other combinations of fixed.
1): < < Recall that in the L6vy method the partial differential equation (5.1. Here we assume that the edges y = 0. b are simply supported. The laminate coordinate system (x. or free boundary conditions. The equation governing buckling under inplane normal forces and natural vibration of a specially orthotropic laminated plate is given by Eq.8. Figure 5.1) is reduced to an ordinary differential equation in x by assuming solution in the form of a single Fourier series which satisfies the simply supported boundary conditions on edges y = 0. b. z) is taken such that a12 5 x 5 a/2.8. and the other two edges each have simply supported.Consider a rectangular laminate with inplane dimensions a and b and total thickness h. h/2 5 z h/2 (see Figure 5. . The ordinary differential equations obtained in the L6vy method can be solved either by direct integration or by means of the statespace approach.8. We discuss both procedures in the following sections. y. clamped.1: Geometry and coordinate system of a rectangular plate used in the Lkvy method. (5.1).0 y 5 b.
.8. Nyy = N ~ Y (5.. (5. in theory. to determine the critical buckling load of the plate.02. the complexity of (5. and D . we obtain Dl1 d. (5. must be determined using where N : ~= N OYY /.11) Since A1 and A2 contain the buckling load N:~. # 0.10) to zero.. Eq.10) D.8.8. a .8.. sinh Xlx + C. For clamped boundary conditions on edges x = 0 . We have [cf.8. we require which yield the eigenvalue problem 1 0 cosh Xla X 1 sinh Xla 0 A1 1 0 sinh Xla X I cosh Xla 0 cos X2a A2 sin X2a sin X2a X2 X z cos X2a 1 {g} {i} = (5.2 ( 0 1 . . and they are given by 0 2 NYypX =0 The constants A. B. # 0. the boundary conditions at x = 0 .8.) sin Xla sinh Xza = 0 (5.2 Buckling by Direct Integration Here we consider buckling under uniaxial compressive forces Nzz = 0. and D . a.8.4) Substituting (5. cos X2x + D . ( x )= A. A. # 0.2) and (5.2. C. (4. Eq.cosh Xla cos Xza) + (A: . cosh X1x + B.. for any y..11) makes it less useful in readily computing the buckling loads.A.8. However. we set the determinant of the coefficient matrix in (5.1) with the inertia terms zero.= o )  We assume the general solution of Eq. B. 2 + 2 D ~ t 302dx2 + oz2p4wnN 0~ . sin X2x + 2DG6)p2X2 + D2204  where Xi are the roots of the characteristic equation DllX4 . " dx  2 (Dl2 d2w. .5) in the form W 7 . # 0.8. For a nontrivial solution.58)] 2X1X2 ( 1 .8. C .ANALYSIS OF SPECIALLY ORTHOTROPIC PLATES 287 5.4) into the governing equation (5. for example.8.11) can be used. Qwn.
8. and ( 0 ~ ~ 2. a.2). Here w. the governing differential equation in (5. For periodic motion. to a system of a firstorder matrix differential equation where N = N = N i Y . Substituting (5.8.2) and (5. and the procedure described in Section 4.y)eiwt where i = fl and w is the frequency of natural vibration. .17) for buckling analysis and mode.5. Then the amplitude of vibration wo is approximated as in Eq.8.B212.8.2. (5.8.8.14a) is of the same form as Eq.2). we assume that (5. with Eq.8. for any y and t we obtain where Equation (5.2. t ) = w 0 ( x .N ~ ~ ] Dl 1 (5.1) can be reduced.D2(Di2 2066) . 5. denotes the frequency of vibration of the m t h & = I.8.1).44).8.y. (4.12) w 0 ( x .p 2 f i Y y ) 2 C = 1  Dl 1 .3 Vibration by Direct Integration Here we consider natural vibration of a specially orthotropic plate. + . for free vibration analysis.and i = fl.8.3. with the inplane forces zero.4 Buckling and Vibration by the StateSpace Approach As explained in Section 5.12) into the governing equation (5.4 can be used to determine the natural frequencies for various boundary conditions on edges x = 0 . (5.8. (32 = [2.
.8. As before.8.8.8.2 contains numerical results for various boundary conditions (see [16]). G = w ($) Uniaxial Critical Buckling Loads. Table 5. Substitution of Eq. Note that the nondimensionalized frequencies and buckling loads are the same for any odd number of layers n = 3.8. (5. crossply laminates are presented in Table 5.5. The L6vy type solution procedure is used to evaluate the natural frequencies and critical buckling loads under uniaxial compression of specially orthotropic rectangular laminates. Laminate = 3 g: 10 20 30 40 Rlndamwtul Frequencies. the determinant of the coefficient matrix in (5. .21) The roots of the above equation are the squares of the frequencies of natural vibration.7. they denote the buckling loads. (5. the notation SF. The lamina material properties used are Numerical results for the nondimensionalized fundamental frequencies and critical buckling loads under uniaxial compression of square.The solution of Eq. symmetric. is used to indicate that edge x = a/2 is simply supported (S) and edge x = a12 is free (F).8.15) is given by and the vector K of constants is to be determined from the boundary conditions. (with the total thickness of all laminates being the same). or. N = N!rb2/Ezh3 .8.1: Nondimensionalized fundamental frequencies and critical buckling loads under uniaxial compression of simply supported symmetric crossply square plates as a function of the modulus ratio.1 for various ratios of principal moduli of the material. .19) into the set of boundary conditions (expressed in terms of Zi) results in a homogeneous system of equations For a nontrivial solution. Table 5. for example. in the case of buckling.20) should be zero: lMijl = 0 (5.
t) = 0 for t = 20 . Fundamental Requencies. assuming no applied inplane and thermal forces. The resulting ordinary differential equations in time can be solved exactly when possible or numerically using a timeintegration method.9 Transient Analysis 5.Table 5.O.t) 0 .8.b. t Mode m in which the lowest buckling load occurs (otherwise.9. wO(x.2 Spatial Variation of the Solution The equation of motion governing bending deflection wo of a specially orthotropic plate.1 Preliminary Comments In this section we will develop transient solutions to specially orthotropic plates.1)] Suppose that the plate is simply supported with the boundary conditions wO(x. is [see Eq. The same methods can also be used to approximate the spatial variations of the transient solutions of plates. m = 1) 5. 5. (5. the Lkvy method. * Denotes the mode number m.y. Here we consider simply supported plates to illustrate these ideas (see Reddy [21]).t)= 0 .y. 3 = w ($) Biaxial Critical Buckling Loads.9. wO(O.t)= 0 . and the Ritz method.2: Nondimensionalized fundamental frequencies and critical buckling loads under uniaxial compression of symmetric crossply (0°/900/00) square plates for various boundary conditions and modulus ratios. wo(a.1. Recall that in the static bending analysis of plates we developed the analytical solutions using the Navier method.
and initial velocity can be expanded as 0 0 q(x.2) for any time t 0 > where a = ( m ~ l a and /3 = ( n ~ l b ) Similarly. we assume that the transverse load..9.5) into Eq.y.b Qmn(t)= ab o o x x D. t ) sin a x s i n y dxdy (5. y. sin az sin Dy Vmn sin a x sin By 1 la q(x. (5. y. for example.3) where do and vo are the initial displacement and velocity. (x.8) Substituting the expansions (5. respectively.9.4) and (5.O) at = V O ( X .9.1).y.9. ) .9..O)= dO(x..9.and assume that the initial conditions are awo wo(x. Y ) for all x and y (5..y) = uo(x. We assume the following expansion of the transverse deflection to satisfy the boundary conditions (5. Qmn are given by 4 . it follows that + m d a 2 p 2+ &d4] + [lo + 4(a2+ p2)] wmn Q~~~= 0 +  or where .y).y) = n=l m=l where. t ) = x 0 0 Qmn(t) sin a x sin fly do(x. we obtain + [I"+ I2(a2+ p2)] wmn Qmn} Wmn [ h a 4 2(&2 sin a x sin By = o Since the above expression must hold for all x and y. initial displacement.
9. we first write Eq.9. cos pt n=l m=l + vmn sin pt + Qkn   (1 .9.10a) in the form d2Wmn dt2 + () Mmn Kmn wmn = 1 pQmn(t) Mmn Qmn (t) (5.9. For a step loading. The numerical time integration methods will be discussed in the subsequent chapters. is known. To solve it exactly.(t) = A cos pt + B sin pt + Wg.(t) is the particular solution with r l ( t ) = eXlt and r2(t) = exzt.9.. we obtain Thus the final solution (5.9.3 Time Integration The ordinary differential equation (5.(t) (5.16) .14a). W$..and XI and X2 are the roots of the equation x2 + Kmn = Mmn 0.1313) The solution becomes W.(:OS pt) p Kmn I sin Q:C sin DY (5.9.9.. (5. (5. both spatially and with time. Eq.10a) can be solved either exactly or numerically.11) The solution of Eq. X2 = ip.3).5...9. where H ( t ) denotes the Heaviside step function. the solution can be determined from Eq.9. (5. i = a.14a) Once the load distribution. Kmn 1 0 Using the initial conditions (5..11) is given by where C1 and C2 are constants to be determined using the initial conditions. (5.9. Qmn(t) = Q.9. A1 = ip. p= (5.4) is given by Dm.14a) takes the form wmn(t) = A cos pt + B sin pt + Q.H(~).
even the "exact" solutions become approximate because of the truncation of an infinite series or roundoff errors in the solution of nonlinear equations. In most cases. e.3 Derive the expressions for transverse shear stresses from 3D cquilibriurn equations when tlie plat. It is assumed that the plate rnotion ensues from rest... (5.2.1 for various types of distributions. h2 = h/2. Eq.3. Thus.15) of a specially orthotropic rectangular plate with simply supported edges at ?j = 0: h and z = a . do = 0 and uo = 0. The exact solution of the differential equation (5..1 Determine the displacenlent field of a sirriply supported platc strip under a concentrated (line) load Fo a t the center using tlie Navier solution method. The . h = 5 cm) under a step loading that is sinusoidally distributed (SSL) or uniformly distributed (UDL) over the plate surface.9. a t least qualitatively.. 5. (5. a = b = 25 cm.5E2.9. (5.. 5.1 contains plots of the nondimensionalized center deflection w = w o ( ~ 2 h " a ~ as ~ )function of time for a simply supported (SS1) symmetric cross~ a ply (0/90/0) laminate (hl = hs = h/4. EI/E2= 25. B. Once the solution .The coefficients QL. Figure 5. y ) . The dashed curve corresponds to the solution when the rotary inertia is neglected.2. Figure 5. ?I.10 Closure In this chapter analytical and Ritz solutions for bending. stresses can be computed using Eqs..2 Derive the Navier solution of a simply supported rectar~gular plate under the following temperature distribution T(J. natural vibration. solution of a transcendental equation.10b) could have been obtained using the Ritz method or other methods. and D..9. .e is subjected to the temperature distribution of the form giver1 in Problem 5. The rotary inertia has the effect of increasing the wavelength slightly. The solution is plotted to show one complete wavelength.e.2.. and V. It should be noted that the procedure outlined above is valid irrespective of how one arrives at Eq.25. and clamped a t x = 0.. Assume uniformly distributed transverse load... Note that the stress variation for the uniformly distributed load case is not as smooth as for the sinusoidally distributed load case. i. = a. arc given in Table 5.. Problems 5. which can be expanded in double Fourier series in the same way as the mechanical loading q(z. 2 ) = To(x. 5. and transient response of specially orthotropic plates are presented. the numerical determination of actual solutions require evaluation of a series solution. in the Lilvy solution (5. buckling.2 contains plots of nondimensionalized center normal stress a.. 5. 242 = 0. the behavior of laminated plates. The analytical solutions developed herein serve to help one understand.9.g. ?I) + where To and TI are known functions of z and y only.(h2/a2qo) as a function of time for the same laminates.10b).L. same holds for D. ?I) z T I ( ~ ..4 Determine the constants A. C. Assurric that To and TI can be expanded in double sine series.9.10b) can also be obtained using the Laplace transform method.13). or determination of eigenvalues (in the statespace approach). .wo is known. GI2 = GI3 = 0.
. laminate.2: Nondimensionalized maximum normal stress (a.I = rotary inertia 0 20 40 60 80 100 120 Time. t (p) Figure 5.9.9. Uniformly 1.1: Nondimensionalized maximum transverse deflection (3) versus time for a simply supported symmetric crossply (0/90).. laminate.) versus time for a simply supported symmetric crossply (0/90).4 load (UDL) under step load I 0 20 40 60 80 Time. t (ps) 100 120 Figure 5.
in the L6vy solution (5. t ) = q o S ( t .10 Verify the result in Eq.15) of a specially orthotropic rectangular plate with sinlply supported edges at y = 0 . .y. clamped at x = 0. b and z = 0 .y .12 Determine the critical buckling load of a rectangular orthotropic plate simply supported on edges y = 0 .. and clamped on edge z = n using the oneparameter Ritz approximation of 5.11 Verify the result in Eq.6 Determine the constants A.15 Solve Eq.C. and D. 5..B.t o ) and ( b )q ( x . 5. b and clamped on edges x = 0 . 5.11).t o ) . a using the oneparameter Ritz approximation of the form 5.. B.3. Assume uniformly distributed transverse load. (5..6) is also given by Eq.14).3. and D.C.15) of a specially orthotropic rectangular plate with simply supported edges at y = 0 . (5. t ) = q o H ( t . (5.6. The boundary conditions for the free edge are These boundary conditions can be expressed in terms of the transverse deflection as 5. 5. (5.4. and free a t z = a. and free a t x = 0. 5.14 Determine the transient response of simply supported specially orthotropic plate under transverse loading (a) q ( x .16)..4.. in the Lkvy solution (5.7 Use the following oneparameter Ritz approximation to determine the deflection of a simply supported rectangular plate: Ans: The parameter cl is given by cl = @ with R11 5. Assume uniformly distributed transverse load.9 Use oneparameter Ritz approximation of the form ( xy) % 1 (1 0 s 27rx )a (1  cos  b to determine the deflection of a rectangular plate with clamped edges and subjected to unifor~rily distributed transverse load. (5. 5.9.4.. where H ( t ) denotes the Heaviside step function and 6 ( t ) is the Dirac delta function.8 Show that the oneparameter Galerkin solution with the algebraic functions in Eq.5 Determine the constants A.10) using the Laplace transform method. b and x = a.. b .5.13 Determine the critical buckling load of a rectangular orthotropic plate simply supported on edges y = 0 ..
the on^ and Analysis of Elastic Plates." Journal of Sound and Vibration. and Librescu. P. A. N.. A.. A. R." Journal of Composite Materials. and Librescu. Khdeir. "Analysis of Symmetric CrossPly Laminated Elastic Plates Using a HigherOrder Theory: Part I: Stress and Displacement. N." Composite Structures. NJ (1968).).. Khdeir. "Analysis of Anisotropic Plates 11. F. A. "Free Vibration and Buckling of Symmetric CrossF'ly Laminated Plates by an Exact Method. Pagano. J . New York (2002). Taylor & Francis. 448453 (1950). P. PA (1999). Cheron. 9. J . "The Frequency of Flexural Vibration of Rectangular Orthotropic Plates with Clamped or Supported Edges.296 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS References for Additional Reading Reddy. L. Theory and Analysis of Plates.. 189213 (1988). 740742 (1987). 49. Reddy. New York (1959). T. Khdeir. "Lkvy Type Solutions for Symmetrically Laminated Rectangular Plates Using FirstOrder Shear Deformation Theory. Second Edition. Tsai and T .. J... and Gere.S.. NJ (1974). PrenticeHall. 1 7 . The Netherlands (1994). 54. Theory of Plates and Shells. R. M. Lancaster. A n Introduction. PrenticeHall. J. Harper and Row. "On the Solutions to Forced Motions of Rectangular Composite Plates. Theory of Elastic Stability. Timoshenko. Franklin. McGrawHill.. and Schwartz. PA (1987). E. Matrix Theory. 3. 148165 (1969)." Composite Structures. E. L. 3." Journal of Applied Mechanics. M." Journal of Applied Mechanics. and Harvill.. G. E. and Librescu. John Wiley. J . A." Journal of Composite Materials. 447461 (1988). Ashton. and WoinowskyKrieger. A. J. Hearman. "Vibrations of Rectangular Plates by the Ritz Method.. . Translated from Russian by S. Reddy. Selected Works of Nicholas J. J. l 2 6 ( 3 ) . Ashton.. J... Szilard. New York (1972)." Journal of Applied Mechanics. Wang. A. N.. McGrawHill." Journal of Composite Materials.1 (1968). J. L. New York (1959). S. Mechanics of Composite Materials. Third Edition. 470479 (1969). "Analytical Solution of a Refined Shear Deformation Theory for Rectangular Composite Plates. 23.. R. 403408 (1982). "Analysis of Symmetric CrossPly Laminated Elastic Plates Using a HigherOrder Theory: Part 11: Buckling and Free Vibration. N. N. "On the Solution of Plates of Composite Materials. Gordon and Breach. L. N. (Ed. "Analysis of Anisotropic Plates. 26(4). Second Edition. N. Classical and Numerical Methods. Reddy.. Timoshenko. Khdeir. Anisotropic Plates. S. A. New York (1970). A. 9. S. Structural Analysis of Laminated Anisotropic Plates. 537540 (1959). 259277 (1988). 590592 (1969). Newark. Goldberg.. and Waddoups. A. Kluwer. J . Lekhnitskii.. McGrawHill.. 14471463 (1987). L. Reddy. L. J. J. Reddy. Whitney. Englewood Cliffs.. 3. A. N. Systems of Ordinary Differential Equations. Khdeir." Journal of Applied Mechanics. Philadelphia. D. Technomic. P. S. J. Young. Pipes. Applied Mathematics for Engineers and Physicists.. L. W. J. A. S. Energy Principles and Variational Methods i n Applied Mechanics.." International Journal of Solids and Structures.. Englewood Cliffs. and Librescu..
3. it is useful to express the governing equations in terms of the generalized displacements of the theory. and the Ritz method are used.3.p axay ay2 d2vo + B163x3 + (B12 + 2 & j 6 )+ 3B26&ayz dx2du a3~. depending on the boundary conditions.1 Governing Equations in Terms of Displacements In this chapter analytical solutions of antisyrnmetric crossply and angleply laminated plates based on the classical laminated plate theory (CLPT) are developed. d3vo + "2 a3vO 893 . the von KBrmBn nonlinear terms in the straindisplacement relations are omitted. In all cases considered in this chapter. Before we begin with the derivation of the exact solutions. The Navier method.47) b y setting the noriliriear terms to zero: A1 1 d2u0 d2uo .+ M16a x & ax a2uo + A16d 2vo + (A12 +&6)  + A 2 d2vo 0. (3. The linear equations of motion of the classical laminated plate theory (CLPT) can be obtained from Eqs.(3.Analytical Solutions of Rectangular Laminated Plates Using CLPT 6.. a"..45). the L6vy method with the statespace approach.
1. .1). = In order to make the coefficient matrices [C] and [MI symmetric.~. (3.3) can be cast in differential operator form as' fizz. N.where NT and MT denote thermal resultants defined in Eq.N:.1.1. 4. where coefficients cij are defined by Figure 6.3.1. Equations (6.).1: A plate with applied edge forces ( N ~= .41). the third equation is multiplied with a negative sign. and and fiyy denote the applied edge forces (see Figure 6.1)(6.
not all laminates permit the Navier solution. The Navier solutions can be developed for rectangular laminates with two sets of simply supported boundary conditions. (NZ.. (3. Simply Supported (SS1): The displacement boundary conditions are +' ) M (.  and da..coefficients rnij and f are defined by T f3 = . The geometry.2). . laminate coordinate system. ill.b).3. Note that the thermal forces and moments.2. the Navier solution cannot be developed for the problem. set of algebraic equations among the parameters of the expansion.41a. 6. invertible. The choice of the functions in the series is restricted to those which satisfy the boundary conditions of the problem. The two types of boundary conditions are given below. ~ax2T .'. a 2 ~ T ya2iM. and de denote the differential operators N&. Even for these boundary conditions.1.2 Admissible Boundary Conditions for the Navier Solutions In the Navier method the generalized displacements are expanded in a double trigonometric series in terms of unknown parameters (see Section 5. Substitution of the displacement expansions into the governing equations should result in a unique. Otherwise. are known in terms of the temperature distribution and material coefficients as defined in Eqs. M & ) . and the two types of simply supported boundary conditions are shown in Figure 6. We will determine which lamination schemes permit such solutions. N$) and (A!!&. ayax a y 2 .2. +. db.
1: Types of simply supported boundary conditions.2. used in the analytical solutions of rectangular laminated plates. SS1 and SS2.The boundary conditions associated with stress components (for a plate theory) are J'S J'S at y=0 and y=b Figure 6. .
(6. uo(x. symmetrically laminated plates with multiple specially orthotropic layers. t ) = 0 = 0. WO(X. As will be shown in the following sections. BI6.2. t ) dz = 0 (6. b.1)(6.B66.e.t ) = 0. t ) = 0. Similarly. .4) In Eqs.A26. B22.2. a and b denote the inplane dimensions along the x and y directions of a rectangular laminate. BI2. i.t) = 0. the Navier solutions for the SS1 boundary conditions can be developed for laminates with a single generally orthotropic layer.y.2. 0. and A45 are zero. zcyy(x. D26. ~ ( 5 t.1 Boundary Conditions The stress boundary conditions in Eqs.2) imply. the following SS1 boundary conditions on the displacements and stress resultants of the classical laminate theory: uo(x. the Navier solutions using SS2 boundary conditions can be obtained only for laminates whose stiffnesses A16. and Aq5 are zero. Z .3 Navier Solutions of Antisymmetric CrossPly Laminates 6. A26. t ) dz = 0. vo(a. and antisymmetric angleply laminated plates. in view of Eq.2.3.3.) b. for laminates with a single generally orthotropic layer. 6.1. The origin of the coordinate system is taken at the lower left corner of the midplane. (6.2). and antisymmetric crossply laminated plates. Y.b.y.B26. BI1. (3. Dl6. 9.t) = 0.4). Thus. Z .t) = 0. wo(0. as shown in Figure 6.t) = 0 wo(a. vo(O.O.DZ6. 0..DIG. symmetrically laminated plates with multiple specially orthotropic layers.Simply Supported2 (SS2): The displacernent boundary conditions are The boundary conditions associated with stress components are h t~ zoyV(x.2. the Navier solutions using SS1 boundary conditions can be obtained only for laminates whose stiffnesses A16.
43) and (3. To see if the boundary conditions (6.3.3~) n=l m=l where a = m ~ / a and p = n ~ / b and (Urn.The displacement boundary conditions of SS1 in (6. (3.3.3.3.3.and Mxy given in Eqs. Nyy NxY.y.1) are satisfied by assuming the following form of the displacements n=l m=l (6.Wmn) are coefficients to be determined. we substitute expansions (6.Mxx. Myy. Vmn..2) on the stress resultants are also satisfied.44): wo(x.B11 a2wo ax2  BlzT 8~ a2wo  2B16 a2wo axay  Nxx T .3) into the expressions for Nxx.t) = 1 Wmn(t)sinax sinpy x 03 03 .3.
the thermal force and moment resultants must satisfy the boundary conditions in Eq. D2fj are zero (because g ( x . AIG.6) Note that the boundary conditions in Eq.3. y ) = cosan: cosPy (6. (6. a or y = 0. . in addition. Mzz.Dl6. y) # 0 for x = 0. and Myy can be satisfied only if the laminate stiffnesses A26.where f ( x . y) = s i n a x sinpy . Thus. (6.2).3. NYY. g ( x .2) on the stress and moment resultants Nrz. b).3. the Navier solutions for rectangular laminated plates with SS1 boundary conditions may exist only when the laminate stacking sequences are such that .B I 6 B26.
approximate or numerical solutions may be constructed. admit the Navier solutions for the SS1 boundary conditions. An examination of Eqs.2 Solution Substitution of Eqs. In addition. with no restrictions on laminate stiffnesses. it follows that plates with a single generally orthotropic layer. as shown later in this chapter or in subsequent chapters.3. + IlaWmn cos a x sin B y a ~ = (%+) I "Ty 8~ + (B12a2p+ ~ 2 2 Wmn. which include the former cases as special cases.lovmn ~ ~ ) + IIPWmn I sin OLDcos P y where ~ 1 = B12 2 +2Ba. (6.e.1.3.3) yields + ( ~ l l + B12ap2) Wmn. (6. Although the Navier solutions cannot be developed for general laminates. i.8ac) should also be expanded in the same form as .2. 6. the Navier solution does not exist).1. (6.7) into Eqs.3.lo&..3.8) shows that the mechanical force q and thermal forces and moments of Eqs. inertia Il must be zero.3) and (6. 012 =012 + 2Ds6 (6.3.9) Note that the edge shear force NZyis necessarily zero (otherwise. (6. for the class of lamination schemes admissible here. and antisymmetric crossply laminated plates.1)(6.From Section 5. symmetrically laminated plates with multiple specially orthotropic layers.3.2.
b) Thus we have = 1 [aNAn(t) cos as sin By + /?Nzn(t) sin a x cos /?y] C (6.3. is the expression in the square bracket of Eq.8~). t) but they enter Eq. side of Eq. y .3. 8 ~ ) with different derivatives. (3. q(x.14a) .their counterparts on the left side of the equality in Eqs. (6. z . the left side of Eq. and M& are defined in terms of the same temperature increment AT(x. should also be expanded in double sine series.3.8~) the form has Hence. (6. For example. 3 . t ) must be expanded as Since the thermal moments M&.t ) =ab o la ib AT(^. ( 6 .8c). t ) sin a x sin ay dzdy then we have from Eq. (6. If the temperature increment is expanded as Trnn(2. it is expected that not all of them will contribute to the solution. which consists of the transverse load and thermal moments. M&.3. y .3. . Thus. the right where C . (6.3.8ac). y.41a.
the conditions in Eq. (6. (6. In the temperature distribution should be expanded order to include N$.8ac)] fT = f : = 03 x k. ... . and M must be zero.(t) sin a x cos &] (6. and M$. Then N.(t) cosnx sinpy sin a x cos Py sin a x sin Py f : = xx 03 This requirement places a restriction on the lamination scheme in order for the Navier solution to exist in the presence of temperature changes.(t)) sin a x sin py + 2apMkn(t) cos a x cos py J (6.14b) = x xI . .3. and antisymmetric crossply laminated plates. .. The lamination scheme must be such that For singlelayer plates with a generally orthotropic layer. = MAn = 0 into Eq. Substituting the expansions (6. symmetrically laminated plates with multiple specially orthotropic layers.8)..(t) f..3. MA.. N&...3..l6a.3.= n=l m=l xx 03 03 [ a ~ : ~ ( t )cos a x sin Py + pN.3. in a double cosine series.(t) xx f. be zero because they must be of the form [see Eqs.3. (6.14~) This particular expansion of temperature distribution necessarily requires that N .[a2 .b) are automatically satisfied.3. . we obtain expressions of the form x xx ~ a .14) with N&. ( t ) ~ + p2M:.10a) and (6. and M. ( t )cos a x sin Py =0  bmn (t) sin a x cos By = o cmn( t )sin a x sin By = 0 .
and W.19) provide three secondorder differential equations in time among the three variables U..... are coefficients whose explicit form will be given shortly.. for any fixed values of m and n.where a... For transient (i. the differential equations in time can be solved either exactly or approximately. b.. . (6. b... n. = 0. and y. 3= n ~ l b .. it follows that a. and c.3..e.. time dependent) response. .c . The explicit forms of the coefficients a.17) must hold for any m. and c.... Since Eqs. = 0. are given by or in matrix form where tijand mij are and a = m ~ / and . and .. x.. V. a 1 Equations (6. b .3. = 0 for every m and n.
Nmn. {A') and {A2).21) can be written as MA. First.. The latter allows the elimination of a selected set of variables and retains a desired set of variables. where [K"] = [el1 el2 '12] 222 . [ K ~ ~tJ3 5 3 ] + !3 l = {F'} = { .24a) for {A1}.21) can be solved using Cramer's rule or by the method of static condensation. }. MA. (6. and Wmn in terms of the 1 coefficients Qmn.3.' [ {a2) )  [K'~]~[K~~]'{F'} . [ K ' ~= ] = { c23 } . . Vmn.iNin{F2) Qmn + a'&$. Then Eq. and Then the final solution is given by Eqs. (6.3. Equations (6. according to what is to be eliminated and what is to be retained.3.3. (6.3.~ k .22) represents a pair of two matrix equations: [K1']{A') + [ K ' ~ ] { A ~= {F'). (6. we obtain = (~"1' ({F'}  [KL2]{A2)) Then substituting the result into (6. and therefore it is described here. we obtain K ~ ~= {F') ] ([KZ2] [ K ~ ~ ] ~ [ K " ] .24b). The method is useful in later discussions of this book. } {A') [ K ' ~ ] ~ { A+)[ K ~ ~ ] { = ~ ) ' A {F2) Solving Eq.3.. the column of unknowns is subdivided into two parts.3ac). Nmn + p2Mkn Equation (6.6.19) by setting the time derivative terms to zero: which can be solved for the coefficients Urn. Suppose that we wish to eliminate the coefficients associated with the inplane displacements and retain those associated with the transverse deflection.3 Bending The static solution can be obtained by solving the algebraic equations resulting from Eqs. which is to be eliminated.3.3..
27a) for each m. solving for { A 1 )next using Eq.3). V. (6. . = Note that for antisymmetric crossply laminates. and then. BG6 0 and the coefficients in Eq. urn. Using the definitions (6.26) we obtain (when 5"33 = 0) where a.3.2. v . = Qmn .  wm. The calculation involves solving for { A 2 )first. (6. if desired. = t33 a1 a2 + t 1 3 a0 + t 2 3 a0  Solution of Eq..3.3.3. W. If there are no thermal loads.~ { F(6. Accounting for only mechanical and thermal effects.25). we obtain ..3.3. am71 aoarnn = 6. n = 1. (6.3. .).12a).~ [ K ~ ~ ] [ ] { F ~ = { F ~) [ K ~ ~ ] ~ [ K ~ ~ ] .27a) can be simplified. the solution becomes a2 Qmn aoam7. which can then be used to compute the solution (uo... gives (U. = a1Qmn .where [K"] = [ K ~ ~ K ~ ~ ] ~ [ K ~ ~ ] .3. (3.23) in (6.26~) ) ~ ) This procedure of eliminating (or condensing out) a subset of unknowns is known in structural mechanics as the method of static condensation.4 Determination of Stresses The inplane stresses in each layer of a laminate are calculated from constitutive relations in Eq. wo) from Eq. (6. .3.vo..
z. b/2. we obtain (6. b. (6. where o. (5. z) = (a/2. integrating with respect to z.3.. The transverse stresses in a laminate can be determined using the 3D equilibrium equations [see Eqs.3.. h/2).31) and ..) sin a x sin j3y (6.29b) The inplane stresses of a simply supported (SS1) crossply laminate (i.3.14)] for any zk 5 z 5 zk+l (k) (k) .3.oZy and are known from Eq. h/2).. when Q16 = Q26 = 0 and aZy 0) are then given by = The maximum normal stresses occur at (x. and determined using the boundary conditions og) c. y.e.30) into Eq.29). + ZT.2. y. (6.where temperature increment A T is assumed to be of the form AT(x.(~)functions to be are and continuity of stresses at layer interfaces: Substituting for the inplane stresses from Eq. z) = (a. . and the shear stress is maximum a t (x. y. t ) = m=l n=l xx 03 03 (TO.
.y. .3.33) result in  for k = 1 . 1 + (z 2 2  z i ) ~ $ ~cos ar sin/?y ] . .and a comma followed by x or y denotes differentiation. The boundary conditions (1) (1) (6. 2 .3.32) yield C1 .3ac) and (6. (6.3. .3413) yields the following expressions for interlaminar transverse shear stresses: o$k)(x. Substitution of the displacement and temperature expansions from Eqs. where n denotes the number of layers.C2 = 0. (6.3. z) = C C [(z m=l n=l Cx2 rn  zk)dZ. The interface continuity conditions (6. n.14ac) into Eq.3.
.) are determined using the boundary and interface continuity conditions.. n The bending moments can be calculated from Eqs. = (a/2.37) (k where the functions Em. and the maximum of a. for k = 1 .0). We obtain . z) = (0. b / 2 ..5ac) MXX m = l n=1 Rgn sin a x sin py REn sin a x sin @y R z ?C O S a X COS@y S&\ sin a x sin py Sgnsin a x sin py S&\ cos a x cos py . occurs at (2. y. The transverse normal stress is given by 4 x sin a x sin py (6. . O). 2 .3. . (6.3. occurs at (x.0.Y.The maximum of a. .
h. h / 2 ) of (0/90)k and (0/90)k. eZY gzY b. = ( b/2.Y) = (WO. The maximum values of Mzz and MYyoccur at (x. antisymmetric crossply laminates with four or more layers are more desirable than twolayer laminates due to the reduction in deflections as well as stresses.. 3 . From these results it is clear that.5E2. The following notation is used: (0/90)2 = (0/90/0/90) and (0/90)2. For A16 = comparison.3013). For square .3. Table 6.1: Transverse deflections and stresses in composite laminates subjected to sinusoidally distributed transverse load (E1/E2 25. which are dominant in the case of two layers. '30. laminates under uniformly distributed transverse load can be seen from the results presented in Table 6. The material properties used are GIZ = GI3 = 0. see Eq. for the same laminate thickness. 2 ) = (a. b/2. = a. and the rnaxirnum of hfzy occurs a t (. (a. (2) Tables 6. y ) = ( a / 2 . h/2).3.3. b/2. b/2). (a. and the series is evaluated using m. h/2) b.or eightlayer laminates is due to the bendingstretching coupling coefficients Bij.b/2.ANALYTICAL SOLUTIONS OF RECTANGULAR LAMINATES USING CLPT 313 For the definitions of R's and S's. The effect of stacking sequence on nondimensionalized maximum deflection w x lo2 and inplane normal stress ex.25). The following nondimensionalizations are used in presenting the numerical results: 6~~ g~~ ~ 1 2 .. we have a. and 2 4 2 = 0. (6.. n = 1 .z) (2) . .. The difference between twolayer and four. and = Dl6 = = 0) under various types of mechanical loads. 2 Laminate (bla = 1)* (bla = 3)t * (a/2. (a/2. are zero. the Bij decrease and the laminate essentially behaves like a specially orthotropic plate. G23 = 0.3. results of symmetric laminates are also included. .2 contain nondimensionalized deflections and stresses for antisymmetric crossply laminates (all Bij. = (0/90/0/90).h/2). except for BI1 = B22.3.5E2. and aritisyninietric crossply laminates.2E2. = GI2 = G13 = 0.3. (a/2. ~ 1 = 0. As the number of layers increase.. G23 = 0.1 and 6. h/2).25. a..2E2.
Wx102 u. For (0°/900)k. Gl2 = GIs = 0.1 for stress locations. a. GI2 = GI3 = 0..3: Effect of lamination scheme on the deflections x lo2 (first row) and stresses a.3.20 are used to evaluate the series). For the uniformly distributed load there corresponds an aspect ratio.3.. . See the foot note (*) of Table 6. Figures 6.3. UDL Laminate .25. and 1 4 2 = 0. Table 6.25.2: Transverse deflections and stresses in square laminates subjected to uniformly distributed transverse load (UDL) or central point load 2 (CPL) (E1/E2 = 25. both heterogeneity and anisotropy ratio influence the deflections.5E2..25 for (0/90)2 and a/b = 3. m.. Also. stresses a = [h2/(qob2)]a(a/2.1 through 6.5 for (0/90). .314 MECHAIifCS OF LAMINATED COMPOSITE PLATES A N D SHELLS Table 6. zo) for a fixed z = zo. for which the deflection is the maximum of all aspect ratios.. . which decrease as the number of layers is increased.5E2. ~ 1 = 0. laminates. a. The magnitude of deflections and stresses of symmetric laminates (0/90/90/0) are about two to three times that of antisymmetric (0/90/0/90) laminates for a/b > 1. for antisymmetric laminates the deflections decrease and stresses increase with the number of layers for a fixed anisotropy ratio. Y *XY CPL w x lo2 a. G2s = 0. around u/b = 2.4 show the effect of bendingstretching coupling and and ) plate aspect ratio on the transverse deflection ul = w o ~ ~ h ~ / ( q o b ~ normal b/2. (second row) in square laminates subjected to uniformly distributed load (UDL).2E2. deflections decrease and stresses increase with increasing value of E1/E2.. The material properties used are El/E2 = 25.. . heterogeneity has little effect on deflections and stresses. The anisotropy ratio affects deflections and stresses. [(0/90)kl* [("/go)kslt 20 25 %=5 10 15 20 25 Ic G =5 10 15 For the (0/90)k (antisymmetric crossply) laminates.3. 3 . n = 1 .3.
(0/90/0/90) = (0/90).3. b / 2 .(a/2. .. UDL L ' /" +0/90/90/0) = (0/90).h / 2 ) ) versus plate aspect ratio ( a l b ) for simply supported (SS1) laminates. L<0/90) J All laminates have the same total thickness '  1 2 3 4 Plate aspect ratio. .2: Nondimensionalized normal stress (a.1: Nondimensionalized center transverse deflection ( G ) versus plate aspect ratio (alb) of simply supported ( S S .3.. a 1b 5 Figure 6.1 ) laminates.0 1 2 3 4 Plate aspect ratio. a 1b 5 Figure 6.
4 Figure 6.b l 2 .50 1Q b rn / 1 .+ 5 r n m 2. l l l l l l l l . a 1b 5 e < 0.3.4: Nondimensionalized normal stress ( a y v ( a / 2 . (0190)  0.20 UDL f h 0 \ UDL .../ . Plate aspect ratio..50  I I I I I I I I ~ I I I I / I I I I I I I I I I I I I I I I I q 2._ 0.b / 2 .. .3.05  ._ _ FDL .. / . All laminates have the same total .00.3: Nondimensionalized normal stress ( F y y ( a / 2 ..1 ) laminates. l l l l . SSL    .h / 2 ) ) versus plate aspect ratio ( a l b ) for simply supported ( S S .1 ) laminates.00i k  .:  S 1. l l l l l l l l .  f '  g 1. l l l l .. _ _ UDL  All laminates have the SSL 3 SSL .:  1 . Il/' "I/ /"/ \ 0.A thickness /  I I ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ /   \  . a 1b Figure 6.' UDL (0/90/90/0)= (0/90). ! ! /ye..00 0 l l .50. . l l l l l l H .UUDDLL  e / 1 / e  __    5%. . l l l l ' SSL 1 2 3 4 Plate aspect ratio.#    S_s_L. h / 2 ) ) versus plate aspect ratio ( a l b ) of simply supported ( S S ..
25).5 Buckling For buckling analysis.2E2.3. 3 .8 show the distribution of the nondimensionalized maximum normal stress and transverse shear stress computed using the 3D equilibrium equations. the determinant of the coefficient matrix in (6.42) should be zero: . through the thickness of twolayer and eightlayer antisymmetric crossply laminates under sinusoidal loading (alb = 1.. For a nontrivial solution.5E2.3.. 242 = 0. G12 = GI3 = 0.25) for the sinusoidal load. 4 ) laminates under sinusoidal transverse loading (El = 25E2.5 for (0/90)k ( k = 1 .19) we have the eigenvalue problem where tijare the coefficients defined in Eq.3.3. El = 25E2. Figures 6. # 0. and W. The coupling coefficients BZJdecrease in magnitude (hence the effect of coupling decreases) with the increase in the number of layers (for the same total thickness of the plate) in antisymmetric crossply laminates.. (6.. G23 = 0.6. ul2 = 0.. Thus. the effect of the bendingstretching coupling present in twolayer plates on stresses is to increase the magnitude of stresses.The effect of coupling is to increase the deflections and stresses. and the stress concentration is reduced in the latter.3. G12 = GIS = 0. a l h = 100.5E2. # 0. From Eq. and vl2 = 0. 6. The nondirnensionalized center deflection w = woE2h3/(qob4) versus the aspect ratio a l b is shown in Figure 6. U.3. The twolayer plates experience larger stresses than eightlayer plates.20). (6...3. V. we assume that the only applied loads are the inplane forces and all other mechanical and thermal loads are zero. # 0.5E2. The dependence of the coupling effect on the modulus ratio is illustrated in Figure 6. where the maximum nondimensionalized deflection is plotted against the modulus ratio E1/E2 (GI2 = GI3 = 0. 2 .3. The nondimensionalized deflections of the sixlayer and eightlayer plates approach the limiting case of an orthotropic plate.25). The solution rapidly approaches that of an orthotropic plate for increasing number of layers..7 and 6.
0 1 2 3 4 Plate aspect ratio. .3.5: Nondimensionalized center transverse deflection versus aspect ratio for simply supported crossply laminates. a l b 5 Figure 6.6: Nondimensionalized center transverse deflection versus modulus ratio for simply supported crossply laminates. 0 5 10 15 20 25 30 35 Modulus ratio.3. E1/E2 40 45 Figure 6.
2 0.All laminates have the same total thickness / 4 0. obtained from equilibrium) versus plate thickness ( z l h ) for simply supported crossply laminates (SSL).4 0.8 Stress.2 0.3 0.3. & Figure 6.) versus plate thickness ( z l h ) for simply supported crossply laminates (SSL). & Figure 6. .8: Nondimensionalized maximum shear stress @ (.0 0.8 0.0 0.3. ' :  All laminates have the same total thickness a   1 0    0.6 0..1 0.4 0.4 Stress.2 0.50 I I I I I I I I I I I I I I I I I I I ~ I I I I I I I I I I I I I I I I I I I 0.7: Nondimensionalized maximum normal stress (a.6 0.
except for the antisymmetric crossply laminate. and material 2: El/E2 = 40. with aspect ratio a/b = 1.22)(6. Therefore. Hence the buckling load for antisymmetric crossply laminates is given by Eq. The nondimensionalized buckling load increases for symmetric laminates whereas it decreases for antisymmetric laminates as the modulus ratio increases.5E2.5E2.5. Antisymmetric crossply laminates have special stiffness characteristics given in Eq. neither sheartwist coupling nor bendingextension coupling exists (t13 223 = O). No(m. and therefore U.3.t12t12 (6. G12 = G13 = 0. (6.4413) Alternatively.44a) or (6. u12 = 0.n<cc {No(m. . using the static condensation procedure described in Eqs.3. l ) .3. we obtain Since Wmn # 0. (5. the mode is ( 2 .4 shows the effect of stacking sequence. in uniform compression. we obtain Clearly.3. (6.where .5. GI:! = G13 = 0.3.7a).25. . 242 = 0.3.45) with coefficients Zij from Eq. (6. In all cases (also see Figures 6. n)) Since tij depend on m and n. and V. for each pair of m and n . l ) . ( ~ ~ / T ~ D ~ ~ ) of rectangular laminates under uniform compression (k = 0) as well as biaxial compression (k = 1). The critical buckling load is the smallest of all No = No(m. 613 623 233 ao = 211222 . The following material properties were used: material 1: E1/E2 = 25. we have ( c f .5a)) Table 6.9 through 6.11) the critical buckling mode is (m. For specially orthotropic plates.3. (6. plate aspect ratio.. n ) = ( l . are zero = . there is a unique value of No.n): Ncr = min l<m.25. and modulus ratio on nondimensionalized critical buckling loads N = N .3. prior to onset of buckling.26). Eq.n) is a complicated function of both m and n n.3.20). In the latter case.3. and no simple conclusions can be drawn about the mode ( n ~ ) at buckling.
1). (1.2).0 3. and 40.3.5 1. 1 < 4 4.0 2.4: Effect of lamination scheme. 10.5 Figure 6. arid modulus ratio on the nondimensionalized buckling loads N of rectangular laminates under uniform compression and biaxial compression (material 1).Table 6. 25.5 and k = 0: mode is ( 2 . material 2 0.n) = ( 1 .2). (1.0 0. respectively.. (b) (0"/90°).material 2 (0/90)2. Uniaxial conipression (k = 0) Biaxial compression (k = 1) The mode number is (m.0 2. material 1 (0/90)4. aspect ratio.0 1.9: Nondimensionalized buckling (N) load versus plate aspect ratio (alb) for simply supported (SS1) antisymrrletric crossply laminates (0/90). 20. except for the following: (a) (0°/900)2.3. and (1. l ) . u / b = 1.3) for modulus ratios 5. l ) for all cases.5 and k = 1: modes are (1.material 2 (0190) .5 2. % 3 r( + 3 (0/90)2.0 3.0 0.5 Plate aspect ratio. a 1b 3.0 All laminates have the same total thickness 0 . trlb = 0. . (1:2). under uniaxial compression.
2 .0 0 6.11: Nondimensionalized biaxial buckling load (N) versus plate aspect ratio ( a l b ) for antisymmetric crossply laminates ( 0 / 9 0 ) .0 3.0 0 5 10 15 20 25 30 Modulus ratio. 0 0. 5.0 1. E1/E2 35 40 Figure 6.3.0 4.3. (n = 1 . 3 ) under biaxial compression ( k = 1). . All laminates have the same total thickness kN.0 0 2.alb = 1 (m = n = 1) 1. a l b .10: Nondimensionalized buckling load (N) versus modulus ratio ( E 1 / E 2 ) for antisymmetric crossply laminates ( 0 / 9 0 ) 2 under uniaxial compression.0 ~ Figure 6.0 Plate aspect ratio.
Hence.2.3. . Nondinlensionalized frequencies.43) we have with + Note that if the inplane inertias are not neglected.e. All layers are of equal thickness. The smallest of the frequencies is called the fundamental frequency. the determinant of For a nontrivial solution..2E2.19) where i = reduces to the eigenvalue problem a : ~ # 0. (6.n). the smallest frequency might occur for values other than rn = n = 1 .6.6 Vibration For free vibration.. Results are presented for rn.25). # 0. and irrespective of whether the rotary inertia is zero.. and we assume a periodic solution of the form and w is the frequency of natural vibration. Eq.3. u . u12 = 0.. If the inplane inertias are neglected (i. and . V.3. from Eq. (b2/'ir2) JphlDzz.:. w .3. wll is not the fundamental frequency. . n = 1.50) takes the same form as Eq. which yields the characteristic polynomial I/x3 q ~ TX $S =0 2 w&..3.3. lib11 = m22 = 0).49) should be zero. (6.3. all applied loads and the inplane forces are set to zero. (6. # 0. the coefficient matrix in (6.3. the eigenvalue problem cannot be simplified to a single equation even if the rotary inertia is zero. + where X = w2 is the eigenvalue and The real positive roots of this cubic equation give the square of the natural frequency w. Then Eq.G23 = 0.5E2.43) kp2) replaced by w2fi33. associated with mode ( m .wTrLn . of specially orthotropic and antisymmetric crossply square laminates are presented in Table 6. (6. In general.5 for modulus ratios E1/E2=10 and 20 (GI2 = G13 = 0.
(1. Also. Figure 6. The bendingstretching coupling has the effect of lowering the vibration frequencies. and vl:! = 0. . (2. (2. l ) . (1. and 7 4 2 = 0.5: Nondimensionalized frequencies 6 of crossply laminates according to the classical plate theory.for the case in which the rotary inertia is neglected.2). the effect of including rotary inertia is to decrease the frequency of vibration. we note that w antisymmetric laminates. (1.2).3). (1. With an increase in the number of layers. G12 = GIY = 0.12 shows a plot of fundamental frequency w versus aspect ratio a / b for symmetric (0/90). The material properties used are E1/E2 = 40. For symmetric crossply plates the first four frequencies are provided by . Note that the first four frequencies for an orthotropic (0') plate correspond to the modes. As noted earlier. l ) .3). whereas for antisymmetric crossply plates they are (m.1). and (2. ( m .3. and (2.1). Table 6. G12 = G13 = 0. the frequencies approach those of the orthotropic plate.25. The fundamental frequency increases with modular ratio. the twolayer plate has vibration frequencies about 40 percent lower than those of eightlayer antisymmetric laminate or orthotropic plate with the same total thickness.6E2.5E2. crossply and antisymmetric (0/90)2 crossply laminates. n ) = ( l . =w . n ) = ( l . For example.25.2).1).3.2). l ) . for the modes: ( m . n ) = ( l . The material properties used are E1/E2 = 25.13 shows the effect of coupling between bending and extension on the fundamental frequencies of antisymmetric crossply laminates.3. and (1. Figure 6.
0 Plate aspect ratio.0 0. 0.3.0 1. l l l .0 6. .0 Figure 6. l l l l ~ l l l l ~ ' l l l .3.13: Nondimensiorlalized fundamental frequency ( w ) versus plate aspect ratio ( a l b ) for antisymmetric crossply laminates.5 1.5 3. a 1b 2.12: Nondimensiorlalized fundamental frequency ( w ) versus plate aspect ratio ( a l b ) for crossply laminates. a 1b Figure 6.0 0.0 5.la u s 7.0 3 s & All laminates have the same total thickness ' Plate aspect ratio.5 2. All laminates have the same total thickness  l I I I ~ l " l .
4.. Myy.7) imply the following conditions on the generalized displacements and stress resultants of the classical laminate theory: The displacement boundary conditions in (6.. (6. Nzy.ir/a) and l = ( n r l b ) . y. we obtain + n=l m=l xx 0 0 0 0 [A16 (PUmn + a%. expressions for N. M.4 Navier Solutions of Antisymmetric AnglePly Laminates 6. and M z Y .4. / 3 ~ sin BY (6. t ) = (x.6. t ) = wo y. Nyy. n=l m=l xC 0 0 0 0 Vmn(t) cos an: sin .4.la) are satisfied by assuming the following form of the displacements vo(x..4.2.1 Boundary Conditions The SS2 boundary conditions in Eqs. Substituting the expansions (6.2b) C C Wmn (t) sin where a = (m.2) into the ? ..) + (&la2+ ~ 1 2 wmn] ) a x sin By ~ ~ sin .
Thus.AIxr. D and D26 are zero. B22.Note that the boundary conditions in (6. B12.4.lb) on the stress resultants Nxy. can be satisfied only if the laminate stiffnesses A16. and My. M .A2G. BGG. B11. the Navier solutiorls for rectangular laminated plates with .
2) into Eqs. . must be zero.3..2. then it follows from Eq.4.1)(6. MA. In addition.. 6.5) can be expressed in matrix form as . N.SS2 boundary conditions exist only when the stacking sequences are such that 1 In addition.10)(6. N.2. and M&. (6. symmetrically laminated plates with multiple specially orthotropic layers.3.2 Solution Substitution of Eq.y .4.3) yields  IOUmn sin a:.13)]. for dynamic problems. Thus plates with a single generally orthotropic layer. cos .14) that NA. Equations (6. (6.d x 7YI sin az sinpy ++2 ~ M F . inertia Il is zero.4. If the transverse load and thermal resultants are expanded as before [see Eqs. aaxay + d ay2~ 2~& 2 ) Note that the edge shear force N~~ is necessarily zero..' I =  + ) aNz ay  o n c oa sin py = (%+? (0 1 ay  (a2fixx p2fiYY) + Wmn  + 1 2 ( a 2 + p2)) & . and antisymmetric angleply laminated plates admit the Navier solutions for the SS2 boundary conditions.3. (6. we must have I = 0. (6. . and M& must be zero. If the temperature field is expanded in double cosine series.
. which can .6) are valid for the case in which the temperature field is expanded in double cosine series.. n = 1 ..4.3.22)(6.4. we can determine the solution to Eq.4.9a) for each m.3 Bending The static solution can be obtained by setting the time derivative terms in Eq.wo) from Eq.8) (when S33 = 0) as where Solution of Eq. there are If no thermal loads.4. The second column of thermal forces in Eq. . 6. .4. (6.6) to zero: Using the static condensation procedure presented in Eqs. . W.3.where and u: = m r l a and /3 = n r l b . (6. (6.4.2~~c).. (6. gives (U. then be used to compute the solution (uo. the solution becomes . 2 . (6.).vo. V.26).. (6.
4.Z) + ~ g ) ) sin a x cos PY 6 + [ 2 a p ~ j : ) ~ m n (a2 + ii2 ~(k) ~ vmn] ) s i n i i ~ + 2 ~ n n ]  C O S ~ Z ~ 3 %= .1 and E.4. The transverse stresses are determined as described in Section 6.29)] fmn = sin a x sin py .. Also.3.O) contributions from E.4 Determination of Stresses The stresses in each layer of an antisymmetric angleply laminate can be calculated from [see Eq.11b) Note that the inplane stresses in angleply laminates will have nonzero contributions from Q&. and they have 0 The values of shear stress ax. a. and relative maximum depends on the specific laminate construction. (6.(Qi.. b/2).E. g . b/2) may be comparable. at (0.12~) ..4.) occur at (a/2.4..)a3  [a2. and (a/2.6. For the isothermal case.~..!? (Q[:) + 3Q$)ap2) wTnn a x sin Py cos (k) + 2 ~ k ) +?!..3. = cos a x cos fly (6. 3 Q22] Wmnsin a x cos fly ) (6. the maxima of (o. we obtain where A ~ = [(a A (k) + P2 QG6) Umn + ap (Qji..
occur at (x.151 0.251 0.214 0. when the coupling coefficients are zero. b/2. respectively. cos a x cos py Note that the locations of the maximum values of M. and MXg cannot be determined in the general case. My.157 0..4. sin a x sin By w.211 20 30 40  0.. 2 1 terms. and 4. y) = (0. vl2 = 0. O). (6. G23 = 0. are not zero. y) = (0..  Table 6.a]{ 0 aUmn cos a x cos py pvmncos~xcospy . and the layer properties are: E1/E2 varied.732 0.4.283 0.25.1 for antisymmetric angleply laminates (45/45)k for k = 1.. y) = (a/2.3). G12 = GIS = 0.415 0. . n = 1 .. However. although their values at (x.. y ) = (0. y) = (a/2.(a/2.278 0.5E2.462 0.308 0.181 0. y) = (a/2. The effect of bendingextension coupling and the dependence of the coupling on 4) the modulus ratio can be seen from the deflections ti^ = w o ( ~ 2 h 3 / q o b and stresses osx = ax. 2~mnsinaxsiny 2a/3Wm.283 0. . in square angleply laminates (45/45)k. maximum values of M. b/2). b/2) and (x. . and a. The location of the maximum value through the thickness depends on the lamination scheme.190 0.4.(/?Urn. Load 30 40 0.) sin a x sin py + m=l n = l a2 . occur at (x. and My. respectively.0) and ( x . All laminates are of the same total thickness.467 0.278 t Denotes k in the laminate (45/45)k .2E2.693 0. The bending moments in an antisymmetric angleply laminate can be calculated from Eq.153 0. The series for uniform load is evaluated using m. 3 .1: Effect of lamination scheme on the transverse deflections stresses Laminate E &=1 a and a.245 SSL UDL SSL UDL SSL UDL SSL UDL E 1 a x lo2 10 20 cm 10 0. b/2). and subjected to sinusoidal load (first line) and uniformly distributed load (second line). and it is given by the expression { M y m=ln=l B16 BZ6 [ I :.278 0.2.O).217 0.190 0.The maxima of a.. aV.174 0.. Note that with increasing number of layers the laminate solution does not tend towards the orthotropic plate solution. and the maximum of MZyoccurs a t (x. h/2)(h2/qob2) presented in Table 6. .190 0.
Lastly. All laminates have the same total thickness 0. G12 = GI3 = 0.(0.4. y) = (a/2.030 0 1 2 3 4 Plate aspect ratio.4.25).2 contains w as a function of the lamination angle 8 for square laminates (8/8)k under sinusoidal load. although small in magnitude compared to that at (x. (a/2. (n = 1.. a l b 5 Figure 6.5E2. .4.2. z). nondimensionalized transverse deflections as a function of the modulus ratio for square laminates under sinusoidal transverse load are presented in Figure 6.0.4) laminates under sinusoidal load. b/2).5 show the plots of nondimensionalized transverse shear stresses a.3.. z) and a. Figures 6.1: Nondimensionalized maximum transverse deflection ( a ) versus plate aspect ratio (alb) for antisymmetric angleply (45/45). The parabolic type variation shown in Figure 6. The effect of coupling is significant for all modulus ratios except for those close to unity. the stress a. z) = ayz (a/2.4 is consistent with that of an orthotropic plate. y) = (0. respectively. b/2. Figure 6. The material properties used are E1/E2 = 25. z) = ay.. Unlike in antisymmetric crossply laminates. is not zero at (x. bl2.4.1 contains a plot of the nondimensionalized deflection w versus plate aspect ratio for simply supported (SS2) antisymmetric angleply laminates (45/45)k under sinusoidal load. Note that throughthickness variations are significantly altered when the number of layers are increased (for the same total laminate thickness). .) under sinusoidally distributed transverse load (alb = 1. for twolayer and eightlayer antisymmetric angleply laminates (451451451.. Clearly.4.4 and 6. the bendingextension coupling is quite significant for twolayered plates.Figure 6.3. and ul:! = 0.4.0. u12 = 0. G12 = Gl3 = 0. (0.25.4.0). but the coupling decreases very rapidly as the number of layers is increased.5E2. El = 25E2.
E1/E2 40 45 Figure 6.2. ..3: Nondimensionalized maximum transverse deflection (w) versus modulus ratio ( E 1 / E 2 for antisymmetric angleply (45/45).4.0074 All laminates have the same total 0.2: Nondimensionalized maximum transverse deflection (w) versus lamination angle (0) for antisymmetric angleply (O/O). 0. (n = 1 . 8 40 45 Figure 6. 2 .4) laminates under sinusoidal load.000 / l i l l l l l l l l l l l l . l l l l I l l l l l l l ' l l l l l l l l l l l I l l l l / 0 5 10 15 20 25 30 35 Modulus ratio.002 ~ 1 1 1 1 1 1 1 1 1 1 1 ~ 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 .ANALYTICAL SOLUTIONS OF R E C T A N G U L A R LAMINATES USING C L P T 333 0.4. (n = ) 1. 4 ) laminates under sinusoidal load. 1 1 1 1 1 1 1 1 1 0 5 10 15 20 25 30 35 Lamination angle.
20 0.4.0.5: Variation of the nondirnensionalized maximum transverse shear stresses through the thickness of antisymmetric angleply (45/45).z) Stress.10 0. laminates.z) = 0. laminates.z) = ol. & (O. & (a/2.00 0.b/2.15 0.25 (a/2.05 0.20 Figure 6.(O.0.4: Variation of the nondimensionalized maximum transverse shear stresses through the thickness of antisymmetric angleply (45/45).10 0.b/2y) 0. 0.4.10 Stress.30 Figure 6. .334 MECHANICS OF LAMINATED COMPOSITE PLATES AND SHELLS 0.00 0.20 0.
As6.DZ2. The critical buckling load is the smallest of all No = No(m.4.3.5 a/b 5 4. ( m .4. n ) = ( 4 .15) to zero.6) we have where tijare the coefficients defined in Eq.5 5 a/b 5. Since tij depend on m and n . n) = ( 5 .4.5 Buckling For buckling analysis.45). 7 ~ ~ ) . 4 . The buckling mode associated with the critical buckling load is ( m .4. BI6. ( 6 .5E2.4. For biaxial compressive inplane loading. and Dss Thus. l ) for 4.Dl6. and the orthotropic solution is rapidly approached as the number of plies is increased. The specially orthotropic solution for antisymmetric angleply laminates is the one that corresponds to the case in which Als. ( m . No(m. a plate made up of a single specially orthotropic layer or a laminate consisting of specially orthotropic layers that are symmetrically arranged about the laminate middle surface). b2 / E~ h:3)of antisymmetric angleply laminates under uniaxial and biaxial inplane compressive loads. D12. l ) for 2.6 for E1/E2 = 40.ANALYTICAL SOLUTIONS OF RECTANGULAR LAMINATES USING CLPT 335 6.4. A2s. n ) = ( 3 .4.4.l)for a/b 5 1.n) is a complicated function of both m and n arid no conclusions can be drawn about the mode ( m . for each pair of m and n. neither shear or twist coupling nor bendingextension coupling exists.e. n )= (1. and ~ 1 = 0.25. and ul2 = 0.5E2. G12 = GI:< = 0.. The buckling mode is (1.5 a/b 2. n/b 3. except for uniaxial compression with aspect 2 ratio equal t o 1. The effect of bendingstretching coupling is the most for twolayer laminates.n).there is a unique value of No.. For specially orthotropic laminates (i. and ( m .5 (m. l ) for 1. we obtain Clearly.25. (6.B26. the only nonzero stiffncsses are All.2 contains nondimensionalized buckling loads ( N = N. Alz. The material properties used for a typical lanlina are G12 = 0.5. we assume that the only applied loads are the inplane forces arid all other mechanical and thermal loads are zero: From Eq. the buckling load is given by Eq. (6. Plots of nondimerlsiorialized critical buckling loads versus plate aspect ratio (alb) for simply supported (SS2) angleply laminates (45/45)k under uniaxial compressive inplane loads are presented in Figure 6. Setting the determinant of the coefficient matrix in (6. Aa2.n ) at buckling. Dl1.4. Table 6.1). n )= ( 2 . < < < < < < . l ) for 3. and D2s are zero.
Table 6.888 41.285 22.320 18.518 6.5* 12.001 11.. 22h3) .963 16.5 1.660 23.0 3.2: Effect of coupling.353 18.841 17.856 5. a l b 5.685 67. y 2 = 0.0 4.6: Nondimensionalized buckling load (N = N. t h e mode is (1. N = Ncrm. G12 = GI3 = 0. versus plate aspect ratio (alb) of antisymmetric angleply laminates under uniaxial compressive edge load.222 8.746 43.020 64. Uniaxial cornpressiori ( I c = 0) 0.475 21.270 7. 250 All laminates have the same total thickness orthotropic plate   \   (45145)  0 1 1 1 1 ~ " l l ~ l l l l ~ l l l l 1 ' l i l l l ~ i l / l ~ l l l l ~ l l l l ~ l l l l ~ l l l l 0.713 9. of rectangular laminates under uniform compression and biaxial compression (E1/E2 varied.25). * Mode is (2.l) for this row (alb = 1.578 32.129 14.5 1.343 6.0 Plate aspect ratio.5 11.779 23.373 15. for all other cases.076 43.060 13. and modulus ratio on b2 the nondimensionalized critical buckling load.809 9.140 20.l).0 2.738 10.565 34.045 7.4.0 1.877 21.633 18.893 4.0 Figure 6.021 16.813 16.999 35.637 33.607 Biaxial compression (k = 1) 0.909 53. plate aspect ratio.530 3.5E2.110 67.5).743 t Modulus ratio.166 43.091 84.4.0 1.603 23.692 9.251 13.825 28.O 1.660 20.
4. Note that.4. we obtain the cubic characteristic polynomial in the eigenvalue X = w2.17) yields Note that w is a function of the mode numbers (m.8.ANALYTICAL SOLUTIONS OF RECTANGULAR LAMINATES USING CLPT 337 Nondimensionalized critical buckling loads versus the lamination angle for uniaxial compression ( I c = 0) and biaxial compression (k = 1) antisymmetric angleply square laminates are shown in Figures 6. (6. where If the inplane inertias are neglected (i.4.7). Eq. The effect is negligible for eight or more layers. and q 2 = 0.6 Vibration For free vibration Eq.25. The buckling load is the maximum for Q = 45".17) to zero.7 and 6. for El = 40E2. Setting the determinant of the coefficient matrix in (6.8 are symmetric about Q = 45". (6. n) because the coefficients t i j depend on m and n.5E2.e. respectively. (6.4.6) reduces to the eigenvalue problem where Eij and mij are defined in Eq.4..4.7a). 6.4. m l l = m22 = O). once again.4. the bendingstretching coupling severely reduces the buckling load for the twolayer plate. Gla = 0. The plots shown in Figure 6. .4. (6. as shown in Eq.
.4.8: Nondimensionalized buckling load (N) versus lamination angle ( 8 ) of antisymmetric angleply square laminates under biaxial compressive edge loads ( k = 1). l i l l 10 20 30 40 50 60 70 Lamination a n g l e .7: Nondimensionalized buckling load ( N ) versus lamination angle (8) of antisymmetric angleply square laminates under uniaxial compressive edge load ( k = 0).orthotropic plate 60 / i  1 All laminates have the same total thickness \ (010) 80 0 0 l l l l ~ l i l l ~ 1 1 1 1 ~ l l l l ~ l l l l ~ l l l l ~ l l l l ~ l l l l . 0 90 Figure 6. f orthotropic plate (010) All laminates have the same total thickness 0 0 10 20 30 40 50 60 70 Lamination a n g l e .4. 0 80 90 Figure 6.
1 Introduction The Lkvy method can be used to solve the governing equations of various plate theories for rectangular laminates for which two (parallel) opposite edges are simply supported and the other two edges can have any boundary conditions.5. The coupling is the maximum for twolayer plates.4. and the total thickness h.25 and a / b = 1 are shown as a function of lamination angle in Figure 6.5. The bendingstretching coupling due to the presence of B16 and B2fj has the effect of lowering the frequencies. .04  All laminates have the same total thickness  0. details are presented for only crossply laminates. The same conclusions hold for results presented in Figures 6.4.. G12/E2 = 0.ANALYTICAL SOLUTIONS OF RECTANGULAR LAMINATES USING CLPT 339 Nondirnensionalized fundamental frequencies ~2= w(b2/. antisymmetric crossply laminate). z ) is taken F 5. At 0 = 45". ~ 4 = 0. The laminate coordinate system (x.5 The L6vy Solutions 6.11.e.9: Nondimensionalized fundamental frequency versus lamination angle (0) of antisymmetric angleply square laminates. and it rapidly decreases with increasing number of layers. 6.9.0 0 l l l l ~ l l l l ~ l l l l ~ l l l l ~ l l l l ~ l l l l ~ l l l l ~ l l l l ~ l l l l 5 10 15 20 25 30 35 Lamination angle.10 and 6. y. However.4.ir2)JphlD22 of graphite2 epoxy composites with E1/E2= 40.4. The effect of coupling is significant for all modulus ratios. the fundamental frequency of the twolayer plate is about 40 percent lower than that of the eightlayer laminate. 0 40 45 Figure 6. and the difference between the twolayer solution and orthotropic solution increases with modulus ratio. Here we describe the Lkvy solution procedure for crossply and antisymmetric angleply laminates using the classical laminate plate theory (CLPT). The planar dimensions are taken to be a and b. Consider a rectangular laminate which has an even number of orthotropic layers with principal material directions alternating at 0" and 90" to the laminate axes (i.
0 4. l l l l ~ l l l l ~ l . All laminates have the same total thickness 0.0 2. a 1b 5.0 Plate aspect ratio. l l l l 0. E11E2 50 Figure 6.I8 5  All laminates have the same total thickness   s  orthotropic plate 200.0 3. 250.0 1.4.4. (45/45)2  0 l l l l ~ " " ~ " l ~ " " ~ " I l l ~ l l l l .11: Nondimensionalized fundamental frequency (6) versus modulus ratio E1/E2of antisymmetric angleply square laminates.10: Nondimensionalized fundamental frequency (a) versus plate aspect ratio (alb) of antisymmetric angleply laminates. 3 & g  8 150 a c 1 !i   a 100  50.0 Figure 6. l l .0 0 5 10 15 20 25 30 35 40 45 Modulus ratio. .
etc.5.3 [see Eq.l thickness h.34)]. Note that only one quantity in each of the following pairs should be specified on the boundary: < < < < where n refers to the normal and s to the tangential directions at the boundary point. h/2 z h/2. The type of the boundary conditions for the classical laminate plate theory were derived in Section 5. . s = f x . x = ( n = f x and s = f y ) : ~5 Simply supported (S): Clamped (C): aw0 ug = 0.such that to be a and b. simply supported.... U . The simply supported boundary conditions on edges y = 0. Here we assume that the edges y = 0. y.) are expressed as follows: One of the following three types of boundary conditions may be used on the remaining two edges. b (n = f y ..5. as shown (x. (3. z) is taken such that a12 < x in Figure 6. and the tota.  2 x )    I simply supported edge (same at y=O) Figure 6..= 0 ax Free (F): (6.1. b are simply supported. vg = 0 .1: The coordinate system used in the Lkvy solution. wo = 0. = ug .3.5. The laminate coordinate system a/2. . and the other two edges can each have arbitrary boundary conditions (e. clamped.g.4) f b  I . = v ~U . or free).0 < y b.
or higherorder derivatives of the unknown coefficients of the displacement expansion. we have the following equations of motion of the classical laminate theory for the isothermal case: 6. from Eqs. we assume the following representation of the displacements: . For the case of antisymmetric crossply laminates.2.1)(6. which involve usually second.The basic idea of the L6vy method is to seek a solution that satisfies the boundary conditions along the simply supported edges exactly.5. we have Consequently. These ordinary differential equations are then solved using the socalled statespace approach (see Brogan [2] or Franklin [3]). This results in ordinary differential equations in x.2. and thereby reduce the twodimensional problem to a onedimensional problem with respect to the coordinate x.3).2 Solution Procedure In the Lkvy type procedure. (6.
) and their derivatives with respect to x satisfy the following boundary conditions: Simply s u p p o r t e d ( A l l D l l .r. The stress resultants derived from the displacement field (6. V... These expansions satisfy the simply supported (SS1) boundary conditions (6. The transverse load is expanded as vo. W.ANALYTICAL SOLUTIONS OF RECTANGULAR LAMINATES USING CLPT 343 where p = ( r n r l b ) .). b. w.and Qm denote amplitudes of (ua..10a) are given by require that The boundary conditions in (6. and q .. w.=o. v.=o.2) on edges y = 0 .=o . Wv.5.5. w o ) .5..=o.. V.5.5) on edges x = ~ a / 2 (U.B f l # 0 ) : Clamped: u.. where (U.. respectively.3)(6.
. . + C~W: + DIU. + ~ 1 2 ~ : CoQm (6.344 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS Free: Substituting Eq. (6. + D6Vm + D7Wm + D The coefficients Ciappearing in Eqs. + + D5u.5. terms of the highest derivatives u v and w into the expression for w .: we obtain (when fiZy = 0) U = CiU.5.15~) where The linear system of ordinary differential equations in (6. expressing the results in .15b) D ~ ~ .5. firstorder.5.: .c ~ u ~ + c ~ v ~ .5..5.15) with constant coefficients can be expressed in the form of a single.5. + C11W.15a) v ~ = .. .D~W.D w~+ WK = C~U.15) are ~ W ~ (6. (6.9).10) into Eqs.c ~ w ~ + c ~(6.5. (6.C~W.: .: and substituting for U and V : .5.7)(6. matrix differential equation (6. : + C~V..17) (2') = [ T W ) + { F ) . + CloV.
14). (6. Cz. (6.5. we discuss the solution of Eq.18) hold with In addition. (6.17) separately for bending.5. Bending In the case of static bending. . .5.[ E ] denotes its ~ ~ inverse. is given by (see Franklin [3]. A. Chapter 3) + where eTZ represents the matrix product [El denotes the matrix of distinct eigenvectors of the matrix [TI.17) or its special cases. when T is independent of x.5.5. and buckling problems.17) and (6. .and K is a vector of constants to be determined from the boundary conditions (6.3. Z' = T Z F.5. we solve Eq.5.2. In each case. and N~~ appearing in the definition of the coefficients el2 and e l l are assumed to be zero.17).ANALYTICAL SOLUTIONS OF RECTANGULAR LAMINATES USING CLPT 345 where [TI = and Next.12) (6. The solution of Eq. . all variables are independent of time. Equations (6. ( i = 1.8) are the eigenvalues associated with matrix [TI. vibration.
For uniformly distributed load (static bending case).13)] that Gsj(a/2).5.20a) into any combination of boundary conditions (6. M2j = Similarly.23a) The free boundary condition at x = a12 requires [see Eqs. assumed to be zero. (6.20) can be written as Now the components of the matrix [MI and vector {R) in Eq.. t ) =Vm(x) cosPy eiWmt 9 W(X. The clamped boundary condition at x = a12 requires [see Eq.5.5. t ) =Um(z)sin pg eiWmt 9. (6. (6.21) can be defined in terms of the coefficients Gij and Hi.5.Substitution of Eq.5.5. the coefficients Riare defined by Natural Vibration In the case of natural vibration.12)(6. and the solution is of the form ~ ( x .) are 9. For example. the solution (6. t ) =Wm(x)sin Py eiwmt . NY. consider the case in which the edge x = a/2 is clamped and the edge x = a12 is free.5. ~ ( x .14) on edges x = ?a12 yields a nonhomogeneous system of equations which can be solved for the vector { K ) . evaluated at 3: = a12 and x = a/2. as described below.14ac)] that Mlj = Glj(a/2). (6. NXx. M3j = Gsj(a/2). M4j = Gsj(a/2) (6. the applied mechanical loads (Q.5.
and i = n.26a) with  0 G [A1 = 0 0 0 0 0 0 (6.5. the applied mechanical load Q. Equation (6.5. Substitution of Eq. 1 = Cl2 2 D8w (6. and are determined.28) should be zero: lMijI = 0 (6.. (26a) is given by Z(z) = e A x ~ (6.  ~ f = j c6  Dgw 2 D6w 2 ~ 1 = ClO 0 2 D ~ w ~~ . and NYu The operator equation for this case is .2613) where c1= c1 c7= C7 ~ 1 = Cl1 1  6'3 D ~ u ~ .27) and the vector K of constants is determined from the boundary conditions. (6. ~9 = C9 .5.5. the determinant of the coefficient matrix in (6. = c3 0 2 w 2 .15) becomes (6.o5w2. is zero.0 4 w 2 . .27) into the set of boundary conditions results in a homogeneous system of equations For a nontrivial solution.5.5.26~) The solution of Eq.5.5. denotes the frequency of vibration of the mth mode. The solution is assumed to be of the form NZ.ANALYTICAL SOLUTIONS OF RECTANGULAR LAMINATES USING CLPT 347 where w. Buckling In the case of buckling.29) The roots of the above equation are (the squares of) the frequencies of natural vibration.
Another source of difficulty in the numerical evaluation of the eigenvalues of the matrix [TI or [A] is due to the fact their diagonals have zero entries.5.22). The eigenvectors in both cases are the same.18b).16b)I. The roots of this equation are the buckling loads.5..3 Antisymmetric CrossPly Laminates Here we present numerical results obtained with the L6vy method and the statespace solution approach. and N ~ ~ .5. This can be circumvented by adding a nonzero constant to all diagonal elements (i. for example. Note that the buckling loads enter the matrix through the coefficients Cll and C12.21).5. {K) = [ E ] '{ K ) (6.28). add c[I]). respectively [see Eqs.16a) and (6. which contain ell and el2. and (6. (6.5.33) should be zero.5. Computational Issues Some comments are in order on the numerical solution of Eq. the determinant of the coefficient matrix in (6. the matrix [MI appearing in Eqs. It should be noted that.33) For a nontrivial solution. (6. This can be overcome (see.e.. (6. while { K ) and [MI are realvalued. 6.5.5.31) is given by (6. {K) and [MI are complexvalued arrays.32) Z(x) = e T x ~ and the vector K of constants is determined from the boundary conditions.5.where [TI is the matrix defined in Eq. Khdeir and his colleagues developed solutions for static and dynamic (natural vibration as well as transient response) analyses and buckling . Nosier and Reddy [4]) by rewriting Eq. Due to the sparse nature of matrix [TI or [A].5.5.22) as and [MI K } { + { F )= (0) . (6.33) is often illconditioned and results in computer overflow or underflow. (32) into the set of boundary conditions results in a homogeneous system of equations [MI{ K ) = ( 0 ) (6. (6. (6. The eigenvalues of the original matrix [TI or [A] are obtained from the eigenvalues of the modified matrix by subtracting the same nonzero constant.5. N. Substitution of Eq.5.5. The solution of Eq. (6.3413) The matrix [MI is not illconditioned and therefore can be easily inverted to solve for {K) and {K) = [ E ] { K ) .
z ) = zTl (z. 242 = 0. are used in all numerical examples presented here: The loading. In the tables and figures.5.35).2. in all cases. is assumed to be sinusoidal q(x.5. y.2 and 6. with a 2 = 3 a l .25) under various boundary conditions on edges x = fa/2. .5.38) are presented in Table 6.1. The notation used for rectangular laminates with different boundary conditions on edges x = fa/2 is as follows (see Figure 6. SC in place of SSSC. The notation SF.x 1 0 a.5.) x 10 2 ' 2 b2qo 2 b2qo where h is the total thickness of the laminate and qo is the intensity of the distributed transverse load. is used to denote a plate for which edge x = a12 is sirnply supported (S) and edge J: = a12 is free (F). Bending The following lamina properties. The degree of orthotropy has less influence on the deflections for large ratios of El to E2.5.and tenlayer laminates. showing the effect of material orthotropy on the deflections. The following nondirnensionalizations are used: ..5E2. Figures 6. (6. y) = ZTI cos a x sin /3y (6.6 = ( n r l b ) . Thus SS is used in place of SSSS.ANALYTICAL SOLUTIONS OF RECTANGULAR LAMINATES USING CLPT 349 of rectangular composite laminates with various lamination schemes and boundary conditions. =  ( b 0  T ( x . = o. one should refer to Figure 6.3713) .5. b are always sirnply supported. and so on.5.(O.r/a) and . CC in place of SSCC. Numerical results for deflections and stresses of crossply laminates subjected to sinusoidal distribution of temperature ox.1 contains numerical results of deflections and stresses for two. The reader niay consult the papers cited in the bibliography for detailed derivations and additional numerical results. For the coordinate system used in the nondimensionalization. we also use the notation SSSF to denote SF.5.3 contain plots of versus E1/E2 for twolayer antisyrnnletric rectangular (bla = 2) laminates (GI? = GIS = 0.5.36) where a = (mi.. the results for deflections and stresses are presented in the following nondimensional form: 1 h2 1 b h h2 (6. Since edges y = 0. for example. y) = qo cos a x sin oy (6.1). Table 6. typical of graphiteepoxy material. The material properties used are the same as those in Eq.5.
10 0.20 0. . E11E2 Figure 6.3: Nondimensionalized maximum transverse deflection (w) versus modulus ratio ( E 1 / E 2 ) for antisymmetric crossply (0190) laminates ( b l a = 2 ) subjected t o sinusoidal load.5.05 0.15 0.5.25 1 3 + Q % Y 0 g' 0.2: Nondimensionalized maximum transverse deflection ( a ) versus modulus ratio ( E 1 / E 2 ) for antisymmetric crossply (0190) laminates ( b l a = 2 ) subjected to sinusoidal load. E 1IE2 Figure 6.0. 0 10 20 30 40 50 Modulus ratio.30 0.00 0 10 20 30 40 50 n Modulus ratio.
) are presented in Table 6.2675 FF 2.2: Nondirriensionalized center deflections (w) and inplane normal stresses (a.2639 2.1504 0.0546 1.6645 1. and @yy) of antisymmetric crossply square plates stresses @ subjected to sinusoidal distribution of transverse load and for various boundary conditions. .4635 11.2935 1.1: Nondimensionalized center deflections (w) and inplane normal (.8393 0.6148 1.6067 1. number of layers.0681 1.4.2512 15.. a.0312 1.1916 0. .7183 5..Table 6. Table 6.0526 SC 0.2152 1.8217 w  *YY (0190)~ (019010) u) w   gxx Vibration and Stability The L6vy type solution procedure is used to evaluate the natural frequencies and critical buckling loads of antisymmetric crossply rectangular laminates. Results for fundamental frequencies and buckling loads are presented for various boundary conditions and aspect ratios in Table 6.5. antisymmetric. The fundamental frequencies increase with increasing orthotropy E 1 / E 2 as well as number of layers.0312 0.0331 1.4543 0. and ayy) of crossply square plates subjected t o sinusoidal distribution of temperature distribution and for various boundary conditions. The natural frequencies increase with an increase in the aspect ratio as well as the number of layers. and ratio of principal moduli of the material.3800 0.. Similar results for critical buckling loads are also presented in the same table.4681 8.064 7. crossply laminates ( 0 / 9 0 / 0 / .4489 FS 1.157 7. . of Layers 2 UI  Variable SS 1.5.4684 1.1091 1.2443 0.157 ax.5.3 for various boundary conditions.6222 0. Laminate 0 (0190) Variable w  SS 1.3914 0.1264 CC 0.5. No. The following material properties are used in the analysis (material 2): Numerical results for the nondimensionalized fundamental frequencies of square.
N = N.25). vl2 = 0. t Table 6. . of Layers bla FF FS FC SS SC CC Natural Frequencies (wl 1)I Critical Buckling Loads (k = 0) t Frequencies with rotary inertia included ( a l h = 10). and dimensionless fundamental frequency. G23 = 0. No. of antisymmetric crossply plates with various boundary conditions (El = 40E2. N = ~ .. (b2/ ~2 h3). of simply supported antisymmetric square laminates (E1/E2 = varied.5.4: Dimensionless fundamental frequencies. No. w = w(b2/h) and uniaxial critical buckling loads.25). the nondimensionalized frequencies depend on the ratio a l h .Table 6.5E2. the frequencies are reported for a l h = 10. J * P I E 2 .5E2. G12 = GI3 = 0. ( b ~ / ~ (k h ~0). of Layers 3 Natural Frequencies (wll)i 10 El IE2 20 30 40 Critical Buckling Loads (k = 0) Fundamental frequencies obtained with (first row) and without (second row) rotary inertia. 1 4 2 = 0. When rotary inertia is included.3: Effect of degree of orthotropy of the individual layers on the = w(b2/h)J=.5. G23 = 0. ~ = ) critical buckling loads. .6E2.6E2. G12 = G13 = 0.
6 contains results for two. 102. (6.82 x lo6 psi (5. As one might expect. . vl2 = 0.ANALYTICAL SOLUTIONS OF RECTANGULAR LAMINATES USING C L P T 353 6.56 x lo6 psi (10. G23 = 0. free vibration.5.2 x 10"si (132.5.24 G12 = G13 = 0.523 x 10"si (3.5.5. w = wo b/2) ~ ~ h "x ~ ~ ofb square.5. Table 6.41)]. . for which the coupling is negligible.76 GPa).5.6: Effect of ply angle (8) and number of layers (n) on dimensionless deflection w of a square plate [(8/8/8/. ~ antisymmetric angleply laminates (451451451 45) for various boundary conditions and uniformly distributed load of intensity qo are presented in Table 6. (6. and its effect is to make the laminate more flexible and hence deflects more than the tenlayer plates.and tenlayer antisymmetric angleply laminates as a function of the lamination angle arid for different boundary conditions.5.5.61 GPa)(6. . and inplane compressive buckling of rectangular laminates are presented (see Khdeir [18]).40). Bending Nondimensionalized deflections.5: Effect of orthotropy on dirrlensionless deflections w of a (451 45145145) square laminated plate. Table 6.41) It is clear that the bendingstretching coupling is the most significant for twolayer laminates. material properties are as given in Eq. E2 = 1.5. In this section numerical results of bending.4 Ant isymmetric AnglePly Laminates The Lkvy solutions in conjunction with the statespace approach can also he obtained for antisymmetric angleply laminated plates. Table 6.65 GPa). (0. The material properties used are the same as those presented in Eq. plates with a combination of free and simply supported boundary conditions deflect the rriost and those with simply supported arid clamped boundary conditions deflect the least. The material properties used in this case are El = 19.38 GPa). 119).
Table 6. plate aspect ratio.5. Table 6. = . . The material used in all these cases is assumed to be a high modulus graphite epoxy with the properties listed in Eq.9: Effect of aspect ratio dimensionless fundamental frequency w w $ d m of a (45145145145) square laminated plate. 3 = w d*.5.354 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS Vibration and Buckling Numerical results for nondimensionalized frequencies. The parametric effects of the lamination angle.5.7 through 6.13.8: Effect of ply angle (0) and number of layers (n) on dimensionless fundamental frequency 0 of a square laminate (01O/O/ . are presented for various laminates in Tables 6.5. (6. N = Nxx&. 10).40): g The fundamental frequencies presented are for the case in which rotary inertia is neglected. Table 6.5. . and boundary conditions on frequencies and buckling loads can be seen from the results presented in these tables. and dimensionless uniaxial buckling loads.7: Effect of inplane orthotropy ratio on dimensionless fundamental frequency 0 of a (45145145145) square laminated plate.5.
12: Effect of inplane orthotropy ratio on dimensionless uniaxial ~ h (45145145145) square buckling loads N = ~ .5. . b ~ / Eof a ~ laminated plate. alb mode n=2 n=4 n=6 n=8 Table 6. ~ ~ h3 of a square plate [ ( O / /E~ 0 / 0 / .5.5.13: Effect of ply angle ( 8 ) and number of layers (n) on dimensionless uniaxial buckling loads N = N .. . r ~ ~ ~ ~ ) . Table 6. . ( b ~ / i .5. . N = ~ . Table 6.10: Dimensionless frequency w for various mode numbers (m) of (451. . 45145145) square laminated plate. .ANALYTICAL SOLUTIONS OF RECTANGULAR LAMINATES USING CLPT 355 Table 6. . .) laminates. 1O)].11: Effect of plate aspect ratio ( a l b ) and number of layers (n) on uniaxial buckling load of simply supported angleply (451451451.
The specially orthotropic plates considered in Chapter 5 are a special case of symmetric laminates. h2 = h4. offer both analysis simplifications and practical advantages over more general laminates.A26. symmetric angleply laminates offer more shear stiffness than crossply laminates. orthotropic layers whose principal material axes are not parallel to the plate axes) that are symmetrically placed about the midplane fall into the class of symmetric laminates. vibration.6.2 Governing Equations The governing equations of motion of symmetric laminates according to the classical laminate theory can be obtained from (3. < 6. and D26 are small. (018/19).e. they influence the laminate behavior significantly. we have Bij = 0 and Dl6 and D26 are not zero. and they decrease with increasing N .1 Introduction In the previous sections of this chapter we considered analytical solutions of bending.3. For example. and buckling of antisymmetric crossply and angleply rectangular laminates.Dl6. The symmetric angleply laminates. Thus the coupling stiffnesses are the largest when N = 3 for symmetric angleply laminates. Dl6. and Dz6 are proportional t o 1/N. A more general example of symmetric angleply laminate is provided by (30160/15/60130) with thicknesses hl = h5. and 0 2 6 . In such symmetric laminates. and the midplane of the plate coincides with the midplane of the 15" ply. in general. An example of symmetric laminates is provided by the class of regular symmetric angleply laminates. For symmetric angleply laminates the coupling terms A16. A26.6. For linear analysis. D16. The regular symmetric angleply laminates should contain an odd number of plies.6. In these laminates. we obtain d2uo d2vo + A12 axay All + a2vo + a2uo a2vo a ~ & ay & ay26 axay ( ++.45)(3. Even when A16.6 Analysis of Midplane Symmetric Laminates 6.47) by setting Bij = 0 and I I = 0. the bendingstretching coupling stiffnesses Bij were not zero. Laminates containing multiple generally orthotropic layers (i. 0 5 0 90 with equal thickness layers. but the bendingtwisting coupling stiffnesses D16 and Da6 were zero.. In this section we consider laminates that are symmetric in both geometry and material properties about the middle plane.3. ( a2uo =hTa t . where N is the total number of layers in the laminate. A26. with Bij = 0 and small A16.
6. the Navier solutions of Eq.3) cannot be developed. For bending. the first tvlro equations yield zero inplane displacements everywhere. For bending we set all terms involving the inplane edge forces and frequency of vibration to zero. forcing us to use the Ritz.+ 4D66axay a2Sw0 ay2 ay2 axay  w 2 [lowo6wo + 12 dx + dy ay dxdy (6. Because of the presence of the bendingtwisting coupling stiffnesses.3 Weak Forms We can use the Ritz method to determine an approximate solution to the bending.4) where w denotes the frequency of natural vibration. and natural vibrations of symmetric laminates.. . Galerkin. (6. .ANALYTICAL SOLUTIONS OF RECTANGULAR LAMINATES USING CLPT 357 where N. = Nyy = NZy = 0 for natural vibration. ~are the applied edge forces. the virtual work done t o applied edge forces and moments should be added t o the expression a2woa26w0 a2w0 + 0 2 2 .6.6. We set q = 0 and w = 0 for buckling analysis. or the finite element method. and natural vibration problems is given below. fiyv.v o ) are uncoupled from the third equation governing wo In the absence of any inplane loads. In the following sections we discuss the Ritz solutions for symmetrically laminated plates. and N .6. buckling. Clearly the first two equations governing (uo. and q = 0 and N. The weak form or the statement of the principle of minimum total potential energy for bending. buckling..
dY.5) into Eq. dy + 2fi& ( ~d y~d l i + 5. dXi dY. (6.7) 6.d ~ .6. d X . irx .6.5 Simply Supported Plates Recall from Eq.d x d y 6 y d2xiYx J d2yq P FY ) + 2D16 M N ( dXidY. Substitution of Eq.4) results in the following equations: b O i=l j=1T ~ { L = 1 a d2x d X i dY.4) that the choice of the double sine series p i j ( 2 .& $ y q + 4 ~ 6 d x. [ ~ l l ~ % . (6. xy5.j ~ d~ )dxdg}cii x i d X5 ] P y i=l j=1 + x ~y % x d IYu)] d x d y } c q d~  lb La qXPY. dxdy (06. ddXxp ddYy q ) d x d y cij dy2 dy dy ] } +fi ' dY.6. (5. dx dy d2yq d2 + xiY2.4 The Ritz Solution We begin with the Ritz approximation of the form and X i and Yj denote any admissible approximation functions for the problem. j ~ y sin sin a b .6.2. The choice is dictated by the essential (or geometric) boundary conditions of the problem.6.dx2 + dXpdYq dx dy AX.y ) = X i (x)? (y ) .d2xP d2xi d x d y d x 2 Yq Y.
8) into (6.b Suppose that the load q(x.6. g.ANALYTICAL SOLUTIONS OF RECTANGULAR LAMINATES USING CLPT 359 satisfies the simply supported (SS1) boundary conditions. y) sin a p z sin Pqydxdy cos a i z sin ojg cos a. 0.p.x sin Pqy + sin six cos Pjy sin a p z cos &y)] dxdy where a" = i ci.7) we obtain o= M N i=l j=1 [illa (hjPq sin ail sin b y sin li..x sin &y + I2 (cos a i x sin pj y cos a.y) is also expanded in double sine series M N 4(x. y) = i=l Qij sin a i x sin 6y j=1 In view of the integral identities Sb" sin a i x sin a j x dx = {.x sin  i=l j=1 ll' F{Lla z p:  q(x. Substituting (6. i i=j . and I i na sin h y sin n.x sin &y &  i=l j=1 { l a ~.6. 3 .x sin pqy + NyY a j x cos Pj y sin a p x cos Pqy sin +2 M N ~ (sin air cos /3jy cos a. i # .
. D12 2D(j6 = 1. it is found that the orthotropic plate solutions for buckling loads and natural frequencies of vibration are overpredicted in comparison to the solutions obtained with the bendingtwisting coupling in place. and it can only give an approximate solution to the symmetrically laminated plates when many terms in the series are used.12~) i # j and i + j odd Eq. because the solution does not contain the stiffness terms Dls and D2s due to the vanishing of the integrals. Similarly. the oneterm Ritz solution does not exist for a general symmetric laminate. 6.46a) and Table 4. The parameters Xi are the roots of the characteristic equation cos Xia cosh Xia . Thus the double sine series solution is incomplete. the maximum deflection under uniformly distributed load. for a square plate with Dz2 = O.6.6 Other Boundary Conditions Equation (6. In particular. .6.cos Xix) + aij (cash Xjy  5(Y) = sin Xjy cos Xjy) (6.6. j = 1.16) . M . for a symmetric laminate with all edges clamped. the deflection is underpredicted by 23. when D16 = D26 = 0. N .5D11. .2. Only the choice of the approximation functions Xi and is different for different boundary conditions.4.6. . obtained with M = N = 7 in the series. When D16 and Dz6 are nonzero.lDll. and D16 = D26 = 0. As reported by Ashton and Whitney [5]. the eigenfunctions of the EulerBernoulli beams can be used for these functions (see Eq. In general. is + For the same case.2.0.9) gives the Navier solutions presented in Chapter 5. we can use the eigenfunctions of a beam with both ends clamped: 5 Xi(x) = sin Xix  sinh Xix sinh Xjy + ai (cosh Xix . the task of computing the Ritz solutions is algebraically complicated. (4. also see [22. .1 = 0 (6.6. For example.3. Eq.5D11. i = j and i + j even cos a i x sin a j x dx = (6. As discussed in Section 5. (6.23]).2.6.7) is also valid for other boundary conditions. .76% when the bendingtwisting coupling is neglected.6. when D16 and Dz6 are neglected the maximum deflection is Thus.15) for i = 1 . . and many terms have to be included to obtain accurate results. (6.9) can be simplified.3. . 2 .
9.7 Transient Analysis 6. For example.. such as the statespace approach (see Khdeir and R.17) We will not consider the topic of solving symmetrically laminated plates for bending deflections. (6. The choice of a separable solution form as above implies that the general spatial variation is independent of time. coefficients to be determined such that 4(x. (6.. Interested readers may consult [19231.6) for antisynlmetric angleply laminates. are y. These are given by Eq... the Navier solution method can be used to reduce the governing equations of motion to differential equations in time. In the absence of thermal effects and applied inplane forces..3.19) for antisymmetric crossply laminates and by Eq. and (2) solve the ordinary differential equations exactly if possible or numerically. 6. The first step is amply illustrated in the preceding sections of this chapter. Here we discuss a method which takes advantage of the static solution form for spatial variation and which uses a numerical method to solve the resulting differential equations in time (see Reddy [27]). these equations are of the form .eddy 124261).7. The equations of motion can be solved using analytical solution methods.2 Equations of Motion For simply supported crossply and antisymmetric angleply laminates. 6.a (6. (6.3. buckling loads.3)] where F.4.and a.1 Preliminary Comments Here we discuss the procedures to determine the transient response of composite larninates.6. = sinh Xia cosh Xia  sin Xia cos Xia  cosh Xia sinh Xia  + sin &a cos X. are assumed to be functions of time. Thus a typical dependent variable @ zy. there are two major steps in the solution process: (1) assume a spatial variation of the displacements and reduce the governing partial differential equations to a set of ordinary differential equations in time. and its amplitude may vary with time. As described in Section 5. and vibration frequencies by the Ritz method further in this book. t) satisfies its governing equation.7. t) is expanded as [see Eq. The only difference is that the coefficients of the double Fourier series (. the Navier solution method can be used to determine the spatial variation of the transient solution. are suitable functions that satisfy the boundary conditions and T..
20) and (6. Then we have where and VA.+l . 2 . St.7.t. 3 ) can also be expanded in the double Fourier series in the same way as the corresponding displacements.4. respectively. Therefore we will not attempt them here. (6. DL. respectively. the function (of time) and its first derivative are approximated using Taylor's series and only terms up to the second derivative are included: where St is the time increment. (6.7). In this numerical integration method.where the superposed dot denotes differentiation with respect to time. for any fixed m and n. We assume that the solution a t time . Alternatively. are the coefficients in the Fourier expansion of the ith initial displacement and velocity. In the Newmark method. as in the statespace method. = t. and The coefficients t i j and m i j of Eq. can be solved exactly using either the Laplace transform method or the modal analysis methods.1) are defined in Eqs.2) using the wellknown family of Newmark's integration schemes for secondorder differential equations (see Reddy [27]). and ts is the current time and t.7. for the two classes of laminates..3.+l is the next time at which we seek the solution.7. (6.3 Numerical Time Integration The set of three equations in (6. Equation (6.2). the time derivatives are approximated using difference approximations (or truncated Taylor's series). 6.7. Both methods are algebraically complicated and require the determination of eigenvalues and eigenfunctions.1) is subjected to the initial conditions We assume that the functions di and vi (i = 1 .7. and therefore solution is obtained only for discrete times and not as a continuous function of time. we seek numerical solutions to Eq.
In such cases.y)f awmm where w .7. torsional. T = 2 ~ 1 It should be noted that the frequencies of vibration for different modes.. the scheme is said to be conditionally stable. denotes the maximum frequency of the discrete eigenvalue problem associate with Eq. The critical time step for the element is the smallest of the critical time steps calculated using the maximum frequency of each mode of vibration. The following choices of a and y define some of the widely used schemes: a=. for example. ~ . the FoxGoodwin scheme (conditionally stable) 2 ' 1 a=y = 0 . is known. Substituting the third equation into the first two in Eq.. (6.6) and solving for {A). the central difference method (conditionally stable) 2 ' 8 3 a=y = . When the error introduced is bounded (hence the solution is bounded). (6. the constantaverage acceleration method (stable) 1 y = . All schemes for which y > u: > 112 are unconditionally stable. and shear modes. denotes the value of the enclosed vector a t time t. such schemes are said to be numerically stable schemes. we obtain where and {. there is a restriction on the size of the time step that would make the error remain bounded..5 are conditionally stable. = ( a . bending..2): > The critical time step can also be expressed in terms of the period of vibration. and the stability condition is 1 ~t 5 at.t. The Newmark family contains several wellknown schemes as special cases.' the linear acceleration method (conditionally stable) 2 ' 3 1 1 a=y = 6 .7. The parameters a and y are selected such that the error introduced in the approximation (6. the backward difference method (stable) 2 ' . Sometimes.7.the Galerkin method (stable) 2 ' 5 ' 3 a=y = 2 . are different.6) does not grow unboundedly as the scheme is applied at each time step to determine the solution at the next time.I. axial. Schemes for which y < a and a 0. y 1 a= 1 = 1 5.
7.. {A)o = ( ~ ( 0 ) is not known at time t = 0. and as are defined in Eq. However..14a).2) An alternative form of Eq.7.5) of the problem.7.(t.12) is given by where where ag. Equation (6. At the first time step (i.e.7.7. Thus the values A l ( t ) = Umn(t).4 Numerical Results Several examples of applications of the methodology descrilled in this section are presented here.) = A&. (6. .7. aq. . (6.7. . (6. the values {A)o = {A(O)) and {A)o = (A(0)) are known from the initial conditions (6. zero initial conditions were .7. Y. ts) = m71x m71y Wmn(ts) sin .obtain we and using Eq. Note that for the central difference scheme (y = 0).12).7) with at t = t.14a) represents a system of algebraic equations among the (discrete) values of {A(t)) at time t = t. . the Newmark method ) is not a selfstarting scheme.8) in terms of the time step 6t and the parameters a and y. In all of the numerical examples. it is used to determine {A)o at t = 0: The transient solution. ... s = O). 6. for the transverse deflection at time t.)) wo (x.+l to replace [ M ] . + ~ .+l in terms of known values at time t = t.sin b a m=l n=l 0 0 03 > 0. + ~ { ~ ) . Although Eq. (6. t. s is given by (Wmn(t. .2) is not valid for t = 0. and As(t) = Wmn(t) are determined at time t = t l . (6.Premultiplying the second equation in (6. . by a repeated solution (or marching in time) of Eq. (6.7. Thus.7. t a . A2(t) = Vmn(t). for example.7. it is necessary to use Eq.12) or (6.
7. except that the nondimensionalized deflection plotted in the figures is w = ~ ~ ( ~ ~ h " l o~ ~ a * ) x/ 2 (note the multiplicative factor).1 shows the nondimensionalized center transverse 4 lo2. the difference is not noticeable on the graphs. For all time steps below lops. 6 = w o ( ~ z h 3 / q o ax ) S t = 5.7. h = (6. at selective times for three different time steps: deflection.7. El = 25E2. E2 = 2.7.x(h2/qob2) presented in .5E2. Gla = GI3 = 0. The effect of larger time step is to reduce the amplitude and increase the period. ~ 1 = 0. . and 50ps (ps = 10@s).7. which is larger than that at z = h/2 (see Figure 6.1 x 10" N/crn2. The effect of coupling on the transient response can be seen from the twolayer and eightlayer results. a / h = 25) 2 p = 8 x lop6 ~ .5E2. Figure 6. S t = 5ps is used. Table 6. Figures 6.20. @ = g. E1/E2= 25.25).1 x lo6 N/cm E2 .3).7.) square plates under suddenly applied transverse load.7.s ~ / c m ~= 2. . In all the following examples. The nondimensionalizations used are the same as listed in Eq. .assumed. The maximum deflections and stresses for the static case are summarized next.2 through 6.1.25 2 = b = 25 cm. Plots of the nondimensionalized center deflection versus time for the same problem are shown in Figure 6.1 Nondimensionalized center transverse deflections ( a ) in simply supported (SS1) crossply (0190) laminates subjected to uniformly distributed transverse load (h = lcm. The following data (in dimensional form) were used in all of the computations: a 1 cm ( a l b = 1.5.39).17) The values of ai and y in the Newmark integration scheme are taken to be 0. Table 6. It has the effect of increasing the amplitude as well as the period. ul2 = 0. t Denotes time in microseconds (ps).4 is computed at z = h/2.7. (6.2E2. Glz = G13 = 0. G2y = 0. which correspond t o constantaverage acceleration method.3. The normal stress .5 contain nondimensionalized transverse deflections and normal and shear stresses in twolayer and eightlayer antisymmetric crossply (0/90/0/. The effect of the time step on the accuracy of the solution was investigated using a simply supported antisymmetric crossply (0190) laminate under uniformly distributed step loading.
.   I 0 200 400 600 Time.57  3. . I .. . C. (6. 2 6 U  1. .5 I3 All laminates have the same total thickness .1: Nondimensionalized center transverse deflection (w) versus time ( t ) for simply supported (SS1) antisymmetric crossply (0190) laminates subjected t o uniformly distributed step loading.5 1.0 0. n .0. t ( ~ L s ) .7.3.U r( g 2.0: 0.0.5 3.5: 0.7.0: All laminates have the ame total thickness 1 2. ..2: Nondimensionalized center transverse deflection (w) versus time ( t ) for simply supported (SS1) twolayer and eightlayer antisymmetric crossply laminates.7.0 . I .. t (ys) 800 1000 Figure 6.0 1. I' 800 "l""1 1000 Figure 6.. .$2.5: 1.5x 3 C.14) for the data.  3. 0 200 400 600 Time. see Eq.0 2..5 0. . .
3: Nondimensionalized normal stress (a.7.) versus time ( t ) for simply supported (SS1) twolayer and eightlayer antisymmetric crossply (0190) laminates.4: Nondimensionalized normal stress (a.7. . Figure 6...0. at the bottom of the laminate) versus time ( t ) for simply supported (SS1) twolayer and eightlayer antisymmetric crossply (0190) laminates.28r 1 T ~ ~ ~ T T T ~ T 1 1 / 1 1 1 ~ / 1 r l i ~ ~ i r ~ ~ ~ ~ i ~ r 1 ~ 1 1 1 1 I 1 1 All laminates have the same total thickness 0. UDL Figure 6.24 h ( 0 / 9 0 ) .
The same observations made for crossply laminates also apply for angleply plates. have smaller maximum deflections. Laminate (45/45). for the same material and geometric dimensions. and periods of oscillation. The angleply plates. UDL: The maximum transient deflection for the twolayer plate is 2.084 Laminate (0/90).114 and it occurs at t = 305 ps. Similarly.8 contain nondimensionalized transverse deflections and shear and normal stresses in twolayer and eightlayer antisymmetric angleply (0/90/0/.035 times that of the static stresses.056 times that of the static deflection. . UDL: Note that the maximum transient transverse deflection of (0190) laminate under UDL. h/2) = 0.6 through 6.7 times that of the static deflection.035 times that of the static deflection. . .7. (a/2. b/2.).064. the maximum transient deflection is 0. . UDL: Laminate (0/90/0/ . . which occurs at t = 400 ps.7. The maximum static deflections and stresses are given below. SSL: w = 1.) .7988 and it occurs at t = 190 ps. stresses. Figures 6.) square plates under suddenly applied transverse load.Laminate (0/90). it is about 2. a . the stresses are also about 2. it is 2. In the case of eightlayer laminate. . is 2. . UDL: Laminate (451451451.
1 All laminates have the same total thickness 1 (0/90).7.5: Nondimensionalized shear stress (ifzy) versus time ( t ) for simply supported (SS1) twolayer and eightlayer antisymmetric crossply (0190) laminates. UDL / ! 0 200 400 600 800 1000 Time. 1.7. t (ps) Figure 6. laminates. .604 \ (45145).15 Figure 6.6: Nondimensionalized center transverse deflection (w) versus time ( t ) for simply supported (SS2) twolayer and eightlayer antisymmetric angleply (45/45). UDL lb' 0.
7: Nondimensionalized shear stress (axy) versus time ( t ) for simply supported (SS2) twolayer and eightlayer antisymmetric angleply (45/45).7.(45/45). laminates. UDL 4 Figure 6.) versus time ( t )for simply supported (SS2) twolayer and eightlayer antisymmetric angleply (45/45).7.8: Nondimensionalized normal stress (a. 0 0 . t (ps) 2 800 7 1000 Figure 6. laminates. .. 200 2 0 400 600 Time.
.e. (6. . Numerical results were presented for static bending. for example. respectively. and it decreases gradually as the number of layers is increased for fixed total thickness. is also presented. The coupling dies out as the number of layers is increased for fixed total thickness. y) only and functions of time t only (i.3) and assuming that conditions in Eqs. In this procedure. Analysis of such laminates by the Ritz method is characterized by slow convergence.6. (6.2. 6. SS1 and SS2. and natural frequencies of general laminates. The Ritz solutions for symmetric laminates are discussed in some detail.2. In general. and transient response of antisymmetric crossply and angleply laminates. The presence of bendingextensional coupling in a laminate generally reduces the effective stiffnesses and hence increases deflections and reduces buckling loads and natural frequencies. unsymmetric laminates. one must use approximate methods.3.8 Summary In this chapter analytical solutions for bending. and the time variation is determined using the Newmark time integration scheme. The effects of bendingstretching coupling and twistcurvature coupling on deflections. buckling. buckling under inplane compressive loads.2. Problems 6. buckling loads.4) by casting Eqs. Lastly. For such laminates.27). (6. A discussion of symmetrically laminated plates. 6. (6. such as the Ritz method or the finite element method because the Navier solutions do not exist for symmetric laminates. (6. decreasing buckling loads. The Navier solutions were developed for two classes of laminates: antisymmetric crossply laminates and antisymmetric angleply laminates. each for a specific type of simply supported boundary conditions.2 Verify Eq. separation of variables).3) into Eqs.3 Verify the solution in Eq.2. The Lkvy solutions with the statespace approach were developed for these classes of laminates when two opposite edges are simply supported with the other two edges having a variety of boundary conditions of choice.1 Verify Eq.3) in operator form. The spatial functions are the same as those used in the static case. which are characterized by nonzero bendingtwisting coupling terms. The presence of twistcurvature coupling in a laminate also has the effect of increasing deflections. and natural vibration of rectangular laminates with various boundary conditions were presented based on the classical laminate theory. and decreasing natural frequencies. the bendingtwisting coupling in symmetrically laminated plates has the effect of increasing deflections and decreasing buckling loads and natural frequencies of vibration. (6. natural vibration.7) hold.1)(6. The coupling also increases the period of oscillation in the transient problems. The coupling is the most significant in twolayer laminates.1)(6.3.19) by substituting expansions (6. a transient solution procedure for antisymmetric crossply and angleply laminates is presented.3.3. Such laminates can be analyzed only with approximate methods of analysis.2. can only be assessed by specific studies. the solutions are assumed t o be products of functions of spatial coordinates (x.
MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS Derive the expressions for transverse shear stresses from 3D equations of equilibrium for the case of isothermal. having arbitrary boundary conditions. Verify Eqs.4.3) and assuming that conditions in Eqs.Y.5. Assume solution of the form 0 3 m=l and load expansion in the form m m=l where a = rnxla.2. and the coefficients Ci are defined as .15).4. Show that the equations of equilibrium of the classical laminated plate theory for such laminates (without any applied inplane loading) can be reduced to the following ordinary differential equations where the primes indicate differentiation with respect to y.4.1)(6. Verify Eq.10) into the definitions of the resultants in Eqs.9).5. (6. (6.2) into Eqs. antisymmetric crossply laminates. Consider antisymmetric angleply rectangular laminates with edges x = 0 and x = a simply supported and the other two edges. Verify the solution in Eq. (6.3. y = fb/2. Y) z) = Assume that both To and Tl can be expanded in double sine series (similar to the mechanical load). Verify the expressions in Eq.3. (6. (3.4) hold. Derive the expressions for transverse shear stresses from 3D equations of equilibrium for the nonisothermal case of antisymmetric angleply laminates when the temperature distribution is of the form WX. (6.6) by substituting expansions (6. (6.4.2.44).11) by substituting expansions (6.43) and (3.5.To(x1y) + zT1(x.
12 Repeat Exercise 6.4).. (6.11 Repeat Exercise 6.10 hold with exception of e l l . are the inplane compressive forces. ( 3 ) of Problem 6. All definitions in Problem 6. 6. 6.z6= w:. is the frequency of vibration associated with mode m.10 for the case of free vibration.6. governiug equation for static bending analysis is given by The weak form (or the virtual work statement) of the same equation is given by Eq. which are modified as where N& and N.10 hold with exception of e l l and e l a . All definitions in Exercise 6... Z8 = wll (I) express Eqs. z7 = wk . are defined as 6.ANALYTICAL SOLUTIONS OF RECTANGULAR LAMINATES USING CLPT 373 and the coefficients e.10 as a firstorder matrix equation of the form where the matrix T and the column vector F are given by 6.14 Consider a symmetrically laminated rectangular plate under the transverse load q(s. Show that the Ritz solution of the form .10 for the case of biaxial buckling. which is modified as (when I2 = 0) where w. without the inplane force and inertial terms.13 Defining the state vector Z(y) as Z5 = KTm. The y).
w o ( x .x ) ( b .15 Consider a symmetrically laminated rectangular plate with simply supported edges. The boundary conditions are given by where the bending moments are related to the transverse deflection by the equations d2wo d2wo + 2Dz6dxdy Find a twoparameter Ritz approximation using algebraic polynomials. Ans: For the approximation of the form the Ritz coefficients are given by .374 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS requires the solution of the algebraic equation [Rl{c) = (91 where dxdy (3) 6.y) = c l x y ( a .y) does not give a solution for the case in which D16 and DZ6 are not zero. Note that the oneparameter approximation.
Cheron. Matrix Theory. 8. 4(1). New York (2002). W. J. "Exact Solutions for Rectangular Bidirectional Composites and Sandwich Plates. Reddy. Pagano. Anisotropic Plates. Khdeir. 14471463 (1987). 10(7). 11. Stamford. 1 5 7 ( I ) . A. . Khdeir. 18081817 (1989). 189 213 (1988). Khdeir. A. A. S. Asliton... 377388 (1988). NJ (1968). 740742 (1987). A. 9. Pagano. Kluwer. Mechanics of Composite Materials." Internatzonal Journal of Solids and Structures." Journal of Composite Matemals. 245258 (1989). J. "Analysis of Anisotropic Plates. J. NJ (1985) 3. 15." Composzte Structures. 931933 (1972). A. Selected Works of Nicholas J.. 122(2). 2034 (1970). 7. J. Khdeir.. Second Edition. Technornic. 21. A. E. and Waddoups. 159172 (1989). 6. M. "Analytical Solution of a Refined Shear Deformation Theory for Rectangular Composite Plates.. W. 64. "Free Vibration of Antisymrnetric AnglePly Laniinated Plates Including Various Boundary Conditions. 13. "Analysis of Anisotropic Plates 11. Reddy.. 18. and Librescu. and Chao. C T (1970).." A I A A Journal.. J.. "Analysis of Synirrietric CrossPly Laminated Elastic Plates Using a HigherOrder Theory: Part 11: Buckling and Free Vibration. and Reddy. N." Composzte Structures. N. M. 470479 (1969). Pagano. Reddy. L. A. The Netherlands (1994). A." A I A A Journal. N. Buckling. 148. " L h y Type Solutions for Symmetrically Laminated Rectangular Plates Using FirstOrder Shear Deformation Theory. A. Khdeir. J. A. Reddy. Modern Control Theory.urnal of Composite Materials. N. "Free Vibration and Buckling of Unsymrnetric CrossPly Laminated Plates. 377395 (1989). J. and Librescu. "Elastic Behavior of Multilayered Bidirectional Composites. and Khdeir. Khdeir. 259277 (1988). 3." Composzte Structures." Jo. Translated from Russian by S. "A Comparison of Closed Form and Finite Element Solutions of Thick Laminated Anisotropic Rectangular Plates.. "An Exact Approach to the Elastic State of Stress of Shear Deformable Antisymmetric AnglePly Laminated Plates." Journal of Sound and Vibration. J. 13. John Wiley. A. and Librescu. Ashton. 12. N.. Franklin. PrenticeHall.. Newark.. 54. E. 10. A. J.... E. Nosier. Englewood Cliffs. J. Englewood Cliffs. Reddy. 9. Lekhnitskii. (ed." Journal of Applied Mechanics. NJ (1968) 4. 128 (3). J. 27(12). 5. 11... L. Tsai and T. N. Reddy. 17. "Analysis of Symmetric CrossPly Laminated Elastic Plates Using a HigherOrder Theory: Part I: Stress and Displacement. "Comparison Between Shear Deformable and Kirchhoff Theories for Bending. N.Journal of Sound and Vibration. and Hatfield. L. Khdeir." . Khdeir.165 (1969). 23. 20.... A. A. J. E." Composite Structures. and Whitney. A. and Vibration of Antisymmetric AnglePly Laminated Plates. N. "Buckling and Vibration of Laminated Composite Plates Using Various Plate Theories. 2. C. 16. L.. L.). W. A." Journal of Composite Materials. A. 139159 (1992). arid Librescu. 3 . 9. N.. J. G." Nuclear Engineering and Design.ANALYTICAL SOLUTIONS O F RECTANGULAR LAMINATES USING C L P T 375 References for Additional Reading 1. A. 19. Gordon and Breach. N. Energy Principles and Variational Methods i n Applied Mechanics. J. 14." Journal of Sound and Vibration. J. PrenticeHall. 153167 (1981).. Theory of Laminated Plates. A. Brogan. Ashton.. "Vibration and Stability Analyses of CrossPly Laminated Circular Cylindrical Shells. S.
26. J. 437445 (1988). A. R. University of Texas. S." Journal of Applied Mechanics. 537540 (1959). and Reddy. J. Young." Journal o.22.mgular Orthotropic Plates with Clamped or Supported Edges.. 4913 (1949). 28. Khdeir A. J.f Pressure Vessel Technology. F. N. 570578 (1991).. and Reddy. Khdeir. D.. Reddy. A... . 499510 (1989). Publication No. 23. N. A.. and Felgar. 205224 (1989). F. "Dynamic Response of Antisymmctric AnglePly Laminated Plates Subjected to Arbitrary Loading. N. "The Frequency of Flexural Vibration of Rect.~408(1982). and Reddy. 113(4)." Journal of Applied Mechanics. A. 3 4 . 3 1 . Tables of Characteristic Functions Representing the Normal Modes of Vibration of a Beam. N. 403. N." Journal of Sound and Vibmtion. "Exact Solutions for the Transient Response of Symmetric CrossPly Laminates Using a HigherOrder Plate Theory. 1 2 6 . J. 24. Khdeir A. 27. Khdeir A. "On the Solutions t o Forced Motions of Rectangular Composite Plates.. 25. Hearman. "Analytical Solutions of Refined Plate Theories of CrossPly Composite Laminates." Composites Science and Technology. 26(4). 49. P. "On the Forced Motions of Antisynlrnetric CrossPly Laminates. A. J.. and Reddy." International Journal of Mechanical Sciences.
Eqs. a / h < 20). they are not accurate when plates are relatively thick (i. they are not always accurate.23) through (3. are expressed in terms of the 4z generalized displacements (uo. As noted in Chapter 6. Although such stresses can be postcomputed through 3D elasticity equilibrium equations. For this reason alone it is necessary to use the firstorder theory in the analysis of relatively thick laminated plates. the transverse shear stresses derived from the equilibrium equations are quadratic through lamina thickness. the classical laminate theory underpredicts deflections and overpredicts frequencies as well as buckling loads with plate sidetothickness ratios of the order of 20 or less. whereas those computed from constitutive equations are constant. in general.1 Introduction The classical laminate plate theory is based on the Kirchhoff assumptions. The more significant difference between the classical and firstorder theories is the effect of including transverse shear deformation on the predicted deflections. In this chapter..wo. as was shown in Chapter 6 for CLPT. the equations of motion of the firstorder plate theory. frequencies. frequencies.7 Analytical Solutions of Rectangular Laminated Plates Using FSDT 7. a constant state of transverse shear stresses is accounted for. The primary objective is to bring out the effect of shear deformation on deflections.4.e. and buckling loads. in which transverse normal and shear stresses are neglected. In fact. those derived from equilibrium equations. which is quite simpler than deriving them through equilibrium equations. In the firstorder shear deformation theory (FSDT). and &) as . we develop analytical solutions of rectangular laminates using the firstorder shear deformation theory.27). stresses. It should be noted that the interlaminar stresses derived from constitutive equations do not match.4. The equilibriumderived transverse stress field is sufficiently accurate for homogeneous and thin plates. The FSDT allows the computation of interlaminar shear stresses through constitutive equations. and often the transverse normal stress is neglected.vo. and buckling loads. To discuss the Navier and other solutions. (3.
are defined in M&. M&). .where the thermal resultants. N&) and (MZ. Eqs.b).N&. (NZ. (3.3.41a.
1 Solution for the General Case The SS1 boundary conditions for the firstorder shear deformation plate theory (FSDT) are (Figure 7.3) Wm.2.1): The boundary conditions in (7. .1: The simply supported boundary conditions for antisymmetric crossply laminates using the firstorder shear deformation theory (SS1).y.213) (7. x xC Vmn (t) sin QX COS py (7.2. (t) sin an: sin By n = l m=l where a = m r / a and p =nrlb.2.2. t ) = n=l m=l 0 0 0 0 wo (x7 t ) = y.2.ANALYTICAL SOLUTIONS OF RECTANGULAR LAMINATES USING FSDT 379 7. Figure 7.lb) are satisfied by the following expansions v0(2.2.2 Simply Su ~arninated'blates orted Antisymmetric CrossPly 7.
e.. For such laminates the coefficients (Urn.b).1. (7. V.3. Wmn.2...2.The mechanical and thermal loads are also expanded in double Fourier sine series where Qmn(t)= 5 J" Jo ab o b q ( x .5) into Eqs...I)(?.613) T m n ( z . NA.5) will show that the Navier solution exists only if i.. I. t ) sin a x sin py dxdy A T ( x . m n .. N$. MA. y . 1. (7..13a. for the same laminates as those for the classical laminate theory.t ) sin a x sin /3y d x d y (7. are defined in Eqs. and M$.2.t ) = ab o 1"lb Substitution of Eqs.2)(7. .y . Ymn) the Navier solution can be X of calculated from 511 212 $12 g22 0 $14 1 5 0 $24 225 $33 + 234 235 0 0 $14 224 533 234 $44 $45 $15$25 $35 $45 $55  mil 0 0 0  0 m22 0 0 m33 0 0 0 0 0 mq4 0 0 0 0 0 0 0 xmn 0 where B i j and mij where the thermal coefficients (6.
(7.2.2. V which can then be used t o compute the solution (uo.29a.2.10~) The transverse shear stresses from the constitutive equations are given by Note that the stresses are layerwise constant through the thickness. Antisymmetric crossply laminates have the following additional stiffness characteristics [see Eqs.e.5. gives (U.. .. &) from Eqs.3. W. The stresses in each layer can be computed using the constitutive equations (see Section 6.4). when Qls = Qzs = Q45 = 0 and a. .10~~) where ternperature increment AT is assumed to be of the form AT(x. T : + ZT. Xmn. wo.2. (7. = 0) are then given by (Rzn (RF& (RZn where + zSEn) sin a x sin /3y + z S e n ) sin a x sin p y + zSk\) cos a x cos Py (7. 2 . &..713) can be simplified. The bending moments are calculated from .2 Bending The static solution can be obtained from Eqs..2. Solution of Eq.7) by setting the terms and edge forces to zero: me derivative (7. (7. .2)(7.. t ) = m=1 n=l xx 0 0 0 0 (. the matrix coefficients in Eq..7.Y..2.b)]: Hence.2. n = 1 . z. y.vo.2...2. (3.)... The inplane stresses of a simply supported (SS1) crossply laminated plate (i.4). sin a x sin p y (7.8) .8) for each m.). (7.
and Dm. defined in Eq..31)(6. Consequently. Specially orthotropic plates (k) (k) (k) (6.Z Y ('1 (x. (5.3. iI5 0. (7.): Decoupling the inplane displacements from the bending displacements.8) that Umn and Vmn are uncoupled from (wmn.z:)Bgi] cos a x sin Py 11 + . z) = m=l n=l C C [(z .)(X.3. coefficients Amn.15b) The solution of Eq. Cmn. and y. y. It is clear from Eq. i14 0.2. and gas = 0. zl) = 0.15a) is given by . we have Qmn (7. i24 0.3.2. (7. Y.37) with the (k) Specially orthotropic plates differ from antisymmetric crossply laminates in that = = = all Bij are zero. the transverse stresses can also be determined using the equilibrium equations of 3D elasticity. Bmn.rk)Cmn+ 2 (r (k)  1 2 .2.(x.r2 k ) D S sin a x cos by I (0) where P~.As discussed in Chapter 6. we obtain (x. Following the procedure outlined in Eqs. The transverse normal stress can be computed using Eq. z) = C C [(z 0 0 0 0 0 0 03  1 z ~ ) A % ~ (z2 .. y. y.. (6. zl) = gY.1313).37). X.2.
 w0 y) (x.1513) can be solved either directly (by inverting the 3 x 3 coefficient matrix) or by using the static condensation procedure outlined in Chapter 6 [see Eqs.26)]. the bending deflections are given by .i12i12.3.3. y) = n = l m=l C C Ymn sin ax cos /?. Equation (7. The inplane deflections are identically zero when the thermal (and inplane edge) forces are zero.18~) with a = m O ~ / a p = n ~ / b . .2.. = x . (x. = . (6. Wmn sin a x sin y 4 . we arrive at where When the thermal forces are zero.where a.. and The bending moments are given by The inplane stresses are given by . 00 00 (7.2.22)(6. Using the latter.
are given by .( 8  ] X m .23a) reduce to ? [I .+ T(k) Y ) cos tll sin /3g + l~2l) IJ.) + 2 ~ 8 )+ . computed using the 3D stress equilibrium equations. 0 = [ Z3 + ( Z  3 ) ( T X +Y ) sin a x sin ~y ~4:' a 3 ~ i +)o r p 2 ( 2 ~ k+) Q!.2. Eqs.)).2. + TI2Ymn) cos a x out = . zk) (x. ) Z ( x .By For singlelayer plates.and the transverse shear stresses are given by The interlaminar stresses.2313) sin . The following nondimensionalizations are used t o present results in graphical and tabular forms: Table 7.! [I f 8 (T) '1 ( T 2 X m n+ T22Ymn)sin a z cos fly Numerical results for the maximum transverse deflection and stresses of symmetric laminates are discussed next.2.038$) ) (7.1 contains the maximum nondimensionalized deflections and stresses of simply supported square symmetric laminates (0/90/90/0) and (0/90/0) under . =( g . t = 0x2 = T$) = a2/3(~i. (7.
( 0 / 9 0 / 9 0 / 0 ) 10 SSL UDL 0.4165 0.2227 0.sinusoidally distributed load ( S S L ) as well as uniformly distributed load (UDL) and for different sidetothickness ratios ( E l = 2 5 E 2 .6627 1.4989 0. is evaluated a t (x.4205 0.5387 0.0338 0.7684 0. and a.7758 20 100 CLPT SSL UDL SSL UDL SSL UDL Symmetric Laminate.3393 0.0425 0.4921 0.0464 0.0352 0.3558 0.0222 0.0213 0.3968 0.3806 0.0221 0. a..1 ) symmetric crossply square plates.0267 0.7262 0.5384 0. G 2 y = 0. alh Load w x lo2 a. $).4836 0. and (TZY(al $).7719 0.0539 0.2704 0.4912 0.1804 0.6796 0.0223 0.b / 2 ) in layers 1 and 3.4312 0.5273 0.8236 0.0426 0.7986 0.7577 0.7828 0.0250 0.0395 0.5387 0.2694 0.0487 0.0470 0. and a y .7548 0.0241 0.9519 0. G12= GI3 = 0.4312 0. For the ( 0 / 9 0 / 0 ) laminate.4315 0.5385 0. h / 2 ) ] 10 SSL UDL 0.6383 0. b . is evaluated at (x.7014 0.4337 0.0246 0.0213 0. y ) = ( 0 .0463 0.7572 0.3540 0.0420 0.7706 0. calculated from equilibrium equations (at z = 0 ) .8072 0. 0. ( 0 / 9 0 / 0 ) 10 SSL UDL 0.. y ) = ( a / 2 .6147 0.5318 0. y.3452 0.1796 0.5350 0.25.4089 0.8420 0.6497 0.4312 0.7697 0.0286 0.5387 0. K = 516). z ) = ( a / 2 .1997 0.0213 0.4333 0.6081 0.1912 0.7865 0. The transverse shear b.8305 0.0245 0.1925 0.5382 0.3518 0. 2)) Table 7.2956 0.8420 0. 0 ) in layer 2.0514 0.7983 0.4247 0.0272 0. / 2 .7744 0.3614 0.0213 0.2536 0.6206 0.4370 0.2E2.5248 0.1: Effect of transverse shear deformation on nondimensionalized maximum transverse deflections and stresses of simply supported ( S S . is computed at (x.7866 OMY  O51/ arz Orthotropic Plate [a. The membrane stresses were evaluated at the following = locations: (TZZ(a/2.0219 0.4448 0. stresses are calculated using the constitutive equations.3501 0..6833 0.3181 0.0267 0.8075 0. .5E2.6697 0.3072 0.2.4337 0. ~ q 2 0. b/2.0453 0. a y Y ( a / 2 b / 2 .6404 20 100 SSL UDL CLPT SSL UDL SSL UDL Symmetric Lammate.0396 0.0244 0.6194 0.0213 0.5134 0.3951 0.4398 0.7191 20 100 CLPT SSL UDL SSL UDL SSL UDL t o .7694 0.6693 1.0213 0.0252 0.6660 0..8045 0.6528 0.5006 0.
4107 0.9643 0.5E2.5387 0. Table 7.4630 0.6874 0.0436 0.2.7575 0. Symmetric Laminate.4188 0.4332 0. The material properties used are El = 25E2.3683 0.0215 0. G12 = GIY = 0. Shear deformation has different effects on different stresses. for a / h = 10 and sinusoidal loading.4844 0.7605 0.5230 0.4)bqo bqo (7.26) Table 7.5387 0.3416 0. yz = 0. (0/90/0/90/0) 10 20 100 CLPT SSL UDL SSL UDL SSL UDL SSL UDL 0.2.b/2.0226 0. The difference between the deflections predicted by the firstorder shear deformation theory and classical plate theory increases with the ratio hla.7322  0.8260 0.The nondimensionalized quantities in the classical laminate theory are independent of the sidetothickness ratio.4312 0.4559 0.2.1591 0.5044 0. The influence of transverse shear deformation is to increase the transverse deflection.0213 0.7649 0.3535 0.4814 0..3232 0.8272 0. For example. The same nondimensionalization as before [see Eq. =a.1770 0.0422 0.0217 0. whereas it is only 12% for a / h = 20.7267  0.0381 0.4410 0.3746 0.1774 0.2: Effect of transverse shear deformation on nondimensionalized maximum transverse deflections and stresses of simply supported (SS1) symmetric crossply square plates. both laminates of the same total thickness.8264 0.1998 0.5).3685 0.4796 0. k = 2.5276 0.6277 0.3852 0.0396 0.4438 0.2E2.0213 0. G23 = 0.5475 0.5383 0.3598 0.1840 0.0385 0.0386 0.6927 0.3459 0.5021 0.3.9727 0.5241 0.7212 0.1519 0. 0 .0403 0.0213 0.6844 0.6867 0.3748 0.(O.3617 0. and K = 516.0380 0.0213 0.4312 0.5285 0. oyz = o y z ( a / 2 .6016 0.3240 0.7581 0.4108  .8080 0.2 contains results for crossply laminates (0/90/90/0/90/90/0) and (0/90/0/90/0).7166 0.4546 0.4365  SSL UDL Symmetric Laminate.6896 0.5525 0.6213 0.. the classical plate theory underpredicts deflections by as much as about 35'36.3591 0.8059 0.k= h h 1.4333 0.0221 0.2.25.5382 0.6901 0.8270 0. (7.25)] is used except for the following quantities: a. (0/90/90/0/90/90/0) 10 20 100 CLPT SSL UDL SSL UDL SSL UDL 0.
9660 0. h/2) and a. The firstorder theory results are slightly different from those of the classical plate theory.9468 1. Table 7.7582 1.25.4913 0.7150 t Maxinium stresses derived from equilibrium. as the number of layers is increased. The stresses are nondimensionalized as in Eq.2 clearly shows the diminishing effect of transverse shear deformation on deflections.4496 0. The influence of transverse shear deformation is less in the case of the laminates presented in Table 7.where k denotes the layer number.2.1070 1.2E2..6216 0. The locations of the maximum stresses.0653 1. Figure 7. computed using the constitutive equations.] of antisymmetric crossb/2. h/2) = ayy(a/2.2373 1. Thus. ~ 4 = 0.3: Effect of transverse shear deformation on nondimensionalized maximum transverse deflections and stresses of simply supported (SS1) antisymmetric crossply square plates (hk = hln.7175 0. the effect of transverse shear strains on deflections and stresses decreases. K = 516). .5E2.2.. = a.2.2.6955 Antisymmetric Lammate.2. and a t z = 0 for (0/90)4 laminate.0636 1. (0190) 10 20 100 CLPT SSL UDL SSL UDL SSL UDL SSL UDL 1.2.25). the effect being negligible for sidetothickness ratios larger than 20. GI2 = GIs = 0. 2 Antisyw~w~etric Laminate.7776 0. T h e reported values are a t z = &h/4 for (0190) laminate. are as follows: Table 7.6980 1. (7. El = 25E2.4479 0. G23 = 0. (0/90)4 10 20 100 CLPT SSL UDL SSL UDL SSL UDL SSL UDL 0. ply laminates subjected to sinusoidally and uniformly distributed transverse loads.3 contains nondimensionalized transverse deflections w and stresses [@xx(a/2. b/2.
. G23 = 0. From Eq. b/2. The results in the table correspond t o To = 0 and TI # 0. Gag = 0. through the thickness of simply supported square laminates (0/90/90/0) under sinusoidally distributed transverse load. and K = 516.1 and 7. of ~ ~ ) supported plates subjected t o the temperature field of the form given in Eq. and the results for the eightlayer laminate are much the same as those of symmetric laminates in Tables 7.2.. Figure 7. vl2 = 0. the equilibrium equations predict a stress variation that is inconsistent with that predicted by constitutive relations.2.2.2.7) we have Following the condensation of variables procedure t o eliminate the inplane displacements Umn and Vmn.2. Figures 7. and maximum transverse shear stresses. due to bendingextensional coupling. we obtain .5E2.7 show plots of maximum normal stresses. qualitatively.2.2.(0. In Figures 7.10~).2. The material properties used are El = 25E2. (7.. z ) .4 contains nondimensionalized deflections.2.. b/2. G12 = G13 = 0.4 through 7. a. G12 = GI3 = 0.We note that the twolayer laminate exhibits quite different behavior.3 Buckling For buckling analysis. UI = w ~ / ( Q I ~ T ~simply.25. stresses computed using the constitutive relations are also included.0.2. We note that the effect of shear deformation on thermal deflections is negligible. from the eightlayer laminate. The dashed lines correspond to classical plate theory solutions.2. (7.. the correct stress variation. z) and @y.2E2.(a/2. In the case of a. z) and @. we assume that the only applied loads are the inplane forces and all other mechanical and thermal loads are zero.5E2. 7. It turns out that (see Pagano [6]) the constitutive equations yield. @.2.25. The material properties of orthotropic layers are assumed to be El = 25E2. Table 7.3 shows the effect of transverse shear deformation and bendingextensional coupling on deflections. while the constitutive equations predict maximum stress in the outer layers. and a 2 = 3a3.2E2.6 and 7. The eightlayer antisymmetric crossply plate behaves much like an orthotropic plate (results are not shown in the figure).(a/2.7. equilibrium equations predict the maximum stress to be at the midplane of the plate. K = 516. (a/2. z) . b/2. vlz = 0.
16 O 13 0.2.  @ All laminates are of the same total thickness  0.2: Center transverse deflection (w) versus sidetothickness ratio for simply supported (SS1) symmetric crossply (0/90/90/0) square laminates subjected to uniformly or sinusoidally distributed transverse load.025 Classical plate theory SSL = Sinusoidal load UDL= Uniform load \ (0190).2.014 .UDL 4 0.012 B 0.J SSL = Sinusoidal load 0 . dashed lines correspond to the classical plate theory (CLPT) solutions. 0.3: Center transverse deflection (w) versus sidetothickness ratio for simply supported (SS1) orthotropic and antisymmetric crossply (0190) laminates under sinusoidally distributed transverse load.010 n (O/9O/9O/O)=(O/9O). a l h Figure 7.000 0 same total thickness 10 20 30 40 50 60 70 80 90 100 Sidetothickness ratio. UDL  0 10 20 30 40 50 60 70 80 90 100 Sidetothickness ratio. a 1h Figure 7. .
2.. alh=lO Stress. & (a/2.. Figure 7. a/h=4 A .FSDT.b/2.5: Nondimensionalized normal stress (ayy) versus thickness ( z l h ) for simply supported (SS1) symmetric crossply (0/90/90/0) laminates.2.4: Nondimensionalized normal stress (a. . FSDT.) versus thickness ( z l h ) for simply supported (SS1) symmetric crossply (0/90/90/0) laminates.z) Figure 7.
0..Stress.612. & (0. Stress. .2.7: Nondimensionalized shear stress (8v.z) Figure 7.) versus thickness ( z l h ) for simply supported (SS1) symmetric crossply (0/90/90/0) laminates.2.~) . Figure 7.6: Nondimensionalized shear stress (a. (al2.) versus thickness ( z l h ) for simply supported (SS1) symmetric crossply (0/90/90/0) laminates.
5 a l b = 3 SSL UDL Orthotropic 10 10 SSL 10 20 100 CLPT UDL 10 20 100 CLPT Laminate. Load alh alb = 1 a l b = 1.5 a l b = 2 a l b = 2. TI = constant). . To = 0.4: Effect of the aspect ratio and sidetothickness ratio on the deflection of simply supported ( S S . both CLPT and FSDT solutions are the same and independent of a l h .3.1 ) plates subjected to temperature field that is uniform in the xyplane and linearly varying through the thickness (qo = 0. SSL 10 20 100 CLPT 10 20 100 CLPT UDL Laminate. ( 0 / 9 0 / 9 0 / 0 ) SSL 10 20 100 CLPT 10 20 100 CLPT UDL tv = 0. (0190) SSL 10 20 100 CLPT 10 20 100 CLPT UDL Laminate.2.Table 7.
3313) is of the form i.ANALYTICAL SOLUTIONS OF RECTANGULAR LAMINATES USING FSDT 393 Repeating the procedure to eliminate X .. Equation (7.2. Eq.55. (7. (7..3. and Y..2. and Y . see next section for details.. we can eliminate X. to obtain an expression equivalent to the one given in Eq.2.7).2.. we can write Clearly.47a) obtained using the classical plate theory. (7. The expression in (7. we have from Eq.2.31). when the effect of transverse shear deformation is neglected.30b) i14 iI5= 0 arid = ia4 i 2 5 = 0.r. we obtain Alternatively.33b) yields the result (6. (7.33 1 + + where kl < k2 k2 from which it follows that E33 + kl > t 313+ k2 indicating that transverse shear deformation has the effect of reducing the buckling load (as long as i. first and then eliminate U.31) takes the form Using the definitions of iijfrom Eq. and V. bl = bz = bs = b4 = 0 and 344 = = s 4 5 = ti?45... consequently.33 > 1). and 355 = 2.2.. Specially orthotropic plates For specially orthotropic plates. ..
Hence. the critical buckling load becomes Table 7. except for orthotropic plates in biaxial compression. For an isotropic plate. and substitute Umn(t)= ~ : ~ eVmn(t) =~ . 0. For free vibration.125. For example. 25. and the error is less for thin plates.25).2. for which the mode is (2. except that the mode at critical buckling is (m.2. For the sidetothickness ratio of 10. ~ ~ ~ ~ ~ ~ in Eq. Figure 7. a parametric study is carried out to determine the minimum buckling load.1.125. 0.394 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS No conclusions can be drawn from the complicated expression of the buckling load concerning its minimum.2. 0. In these laminates the 0" layers and 90" layers have the same total thickness. 7 2 4 Vibration .5E2.8 shows the effect of transverse shear deformation on critical buckling loads of symmetric (0/90/90/0) laminates under uniaxial and biaxial compression (alb = 1. G23 = 0.~. E1/E2 = 25.2. 0.125. and 0. 0. G12 = GI3 = 0. 0. The critical buckling loads in all cases occurred in mode (1. vl2 = 0.1. 0.2.2E2. (0/90/0/90/0/90/0).9 shows the effect of transverse shear deformation and bendingextensional coupling on critical buckling loads ( a l b = 1. The effect of shear deformation on buckling loads is not as significant as for deflections. .. l ) .1. G23 = 0.1). GI:! = G13 = 0.l). (7.5E2. n ) = ( 3 . where ill 1 $22 [S]= 1 $12 0 $14 0 0 $33 $14 $24 $44 $15 $25 0 $24 i34 $35 $45 g34 $35 $25 $45 $55 .: ~ eWmn(t) = . respectively. we set the thermal and mechanical loads to zero. Figure 7. E1/E2 = 25.5 contains nondimensionalized critical buckling loads of a square orthotropic plate and symmetric square laminates (0/90/0).. which occurs at m = n = 1.6 for modulus ratios E1/E2=10. The eightlayer antisymmetric crossply plate behaves much like an orthotropic plate. vl2 = 0. The effect of shear deformation is clear from the figure. the classical laminate theory overpredicts the critical buckling loads by as much as 48% for orthotropic plates.125.25). Critical buckling loads of twolayer and eightlayer antisymmetric crossply laminated plates under uniaxial and biaxial loading are presented in Table 7. and (0/90/0/90/0/90/0/90/0) under uniaxial and biaxial loadings.1. Note that the same critical buckling loads are valid for a rectangular laminate with aspect ratio a l b = 3. in the case of the ninelayer laminate the individual layer thicknesses are 0.7) and obtain w L ~ ~ " ". and 40. (0/90/0/90/0).1.2E2. for example.
= Table 7.c).l) 10 20 100 CLPT 2.184 3.363 23.839 8.5E2. K = 516).747 CLPT t Mode for orthotropic plates in biaxial compression is (nr.367 6.526 Biaxial Compression (k = 1).576 4.046 23. ) .187 4. Uniaxial Compression ( k = 0) 10 20 25 50 100 CLPT 15.2E2.746 6.555 7.6: Effect of shear. ~ 2 . N = N.2E2.189 9.5: Effect of shear deformation on nondimensionalized critical buckling of loads.800 23. x:.381 23.489 11.374 9.The coefficients iijand mij are defined Table 7.289 20.755 4. simply supported (SS1) symmetric crossply square plates (El = 2532.495 Biaxial Compression ( k = 1) 10 20 25 50 100 5. G12 = G I 3 = 0.568 22.n) ( 2 ..257 8.2.873 3. mode: (1.369 8. ( a 2 / ~ 2 h 3 )of simply supported (SS1) antisymmetric crossply square plates (G12 = GI3 = 0.102 3. buckling loads.495 15.837t 7.094 4.7b.ANALYTICAL SOLUTIONS OF RECTANGULAR LAMINATES USING FSDT 395 and {A)T = in Eqs.511 9.864 8.2. K = 516). {u:.407 7. v:.158 10.644 10. N = Ncr(a2/E2h3).25.deformation on nondimensionalized critical .5E2.682 11. w$.422 5.25. ~2 = 0. G23 = 0.314 10.628 21.205 6.2. (7. G23 = 0.874 20. ~ 1 = 0. 2 Unzaxzal Compression (k = 0).784 11.953 21.432 4. l ) . mode: (1.579 5.978 23.153 9.1) 10 20 100 CLPT 5.843 10.190 5.763 .380 10.
2.0 uniaxial compression alh Figure 7.20.8: Nondimensionalized critical buckling load (N) versus sidetothickness ratio ( a l h ) for simply supported (SS1) symmetric crossply (0/90/90/0) square laminates.2. square laminates. orthotropic alh Figure 7.9: Nondimensionalized critical buckling load (N) versus sidetothickness ratio ( a l h ) for simply supported (SS1) antisymmetric crossply (0/90). .
37).2.2..e. Similar results are presented in Tables 7. Eq. (7. Specially orthotropic plates For specially orthotropic plates. The effect of rotary inertia is negligible in FSDT and therefore not shown in the figure. Eq. the smaller are the frequencies.2. m l l = m 2 2 = md4= m 5 5 = O). (6..10 shows the effect of transverse shear deformation and rotary inertia on fundamental natural frequencies of orthotropic and symmetric crossply (0/90/90/0) square plates with the following lamina properties: The symmetric crossply plate behaves much like an orthotropic plate.8 and 7. (say.ANALYTICAL SOLUTIONS OF RECTANGULAR LAMINATES USING FSDT 397 When rotary inertia is omitted.36) can be simplified by eliminating X. (7.2.7 contains frequencies of isotropic plates.2.52)] If frequencies of inplane vibration of specially orthotropic laminates or natural frequencies of flexural or inplane vibration of antisymmetric laminates are required. Eq. Eq. (6.9 for symmetric crossply laminates.e. .2. The rotary inertia (RI) also has the effect of decreasing frequencies.38) gives the same frequencies of flexural vibration as Eq.37a). The effect of the shear correction factor is to decrease the frequencies. one must use Eq.. We obtain the following 3 x 3 system of eigenvalue problem [cf.2. (7. the smaller the K .49)]: where (sij = qi) If the inplane and rotary inertias are omitted (i. the inplane displacements are uncoupled from the transverse deflection. i.37a) for this case.3. and Y. and therefore the natural frequencies of vibration are given by Eq.2.. Table 7. Figure 7. (7. we have [cf. (7. using the static condensation method).2.3.
E l = 25E2.Table 7. v = 0.7: Effect of shear deformation.2. a / h = 10). CLPT~ w/o RI CLPT with RI FSDT w/o RI FSDT with RI m n K t w/o RI means without rotary inertia. Table 7.3.8: Effect of shear deformation on dimensionless natural frequencies of simply supported (SS1) symmetric crossply plates (G = w ( a 2 / h ) m .5E2. Gl2 = GIY = 0. q 2 = 0. alh 5 10 20 25 50 100 Theory FSDT CLPT FSDT CLPT FSDT CLPT FSDT CLPT FSDT CLPT FSDT CLPT 0' Threeply Fiveply Sevenply Nineply . and shear correction coefficient on nondimensionalized natural frequencies of simply supported (SS1) isotropic square plates (6 = w ( a 2 / h ) m . G23 = 0. K = 516. the total thickness ' of all 0 layers and all 90" layers is the same. h / 2 ) .25.2. rotary inertia is included.2E2. rotary inertia.
G23 = 0.299 56.522 t T h e first line corresponds t o shear correction coefficient of K = 1.5E2. The eightlayer antisymmetric crossply plate behaves much like an orthotropic plate.2E2. R.299 56. . and shear correction coefficient on nondimensionalized natural frequencies (w = w ( a 2 / h ) m ) of simply supported (SS1) symmetric crossply (0/90/0) square plates (hk = h/3.2E2. The effect of rotary inertia is negligible in FSDT and therefore not shown in the figure. rotary inertia.9: Effect of shear deformation.25).885 60. El = 25E2. vl:! = 0.911 66.5E2. and rotary inertia on fundamental natural frequencies of twolayer and eightlayer antisymmetric crossply laminates (E1/E2 = 25.10 contains numerical values of fundamental frequencies of antisymmetric crossply laminated plates for various modular ratios. vl2 = 0.885 60. GI:! = GI:<= 0.11 shows the effect of transverse shear deformation.228 22.877 40.911 66.2. CLPT w/o RI CLPT with RI FSDT w/o RI FSDT with RI 15.522 15.ANALYTICAL SOLUTIONS OF RECTANGULAR LAMINATES USING FSDT 399 Table 7.esults for both twolayer and eightlayer laminated plates for square and rectangular ( a l b = 3) geometries are presented. Table 7.754 71. G12 = GI3 = 0.228 22.2. G23 = 0.0 and t h e second line corresponds t o shear correction coefficient of K = 516.2. Figure 7.754 71.877 40. bendingextensional coupling.25).
832 7.7) imply the following SS2 boundary conditions on the generalized displacements and resultants of the firstorder laminate theory (see Figure 7.1 Boundary Conditions The boundary conditions in (6.454 7.3. K = 516).3 Simply Supported Antisymmetric AnglePly Laminated Plates 7.962 4.908 4.Table 7.926 7.2.2.751 4.2E2.956 4.5E2. blh Theory (0190) (0/90)4 (0190) (0/90)4 (0190) (0/90)4 Square Plate ( a l b = 1 ) 10 20 100 FSDT CLPT FSDT CLPT FSDT CLPT 7.906 7.3.964 7.930 4.10: Effect of shear deformation on nondimensionalized fundamental frequencies of simply supported (SS1) antisymmetric crossply square plates (G12 = GI3 = 0.802 7.931 Rectangular Plate ( a l b = 3 ) 10 20 100 FSDT CLPT FSDT CLPT FSDT CLPT 4. G23 = 0.1): .25. vlz = 0.
I I I I ~ I I I I / I I I I . CLPT.2.11: Nondimensionalized fundamental frequency (G)versus sidetothickness ratio ( a l h ) for simply supported (SSI). IROT + 0 0/90/9010.10: Nondimensionalized fundamental frequency (w) versus sidetothickness ratio (alb) for simply supported (SSI). [rrOo. FSDT. FSDT. . WRI _______ 0°. CLPT. CLPT. I l I I . orthotropic and symmetric crossply (0/90/90/0) laminates. WRI l1 1 (01901. I I I I / / I I I / I I I Oo. FSDT. WRI 3 0 10 20 30 40 50 60 70 80 90 100 alh Figure 7. IROT = 0 13 Oo. CLPT. FSDT. I I I I ~ I I l l . antisymmetric crossply (0190) laminates. WRI (0190)4.2.CLPT. IROT # 0 O0. WORI  Oo. IROT # 0 alh Figure 7.ANALYTICAL SOLUTIONS OF RECTANGULAR LAMINATES USING FSDT 401 I I I I . 17 15 13 .
7..3.2a.5) shows that the Navier solution exists only if i.3.3.b). t ) = n=1 m=1 C C Wmn(t)sin a z sin py (7.Vmn. (7.5.2 The Navier Solution The boundary conditions in (7.y. for antisymmetric angleply laminates [see Eq.Figure 7. (7.3.3).1) are satisfied by the expansions u ~ ( x .1: The simply supported boundary conditions for antisymmetric angleply laminates (SS2). The coefficients Umn. can be determined from the equations . (7..3.e. Xmn.3.1.4a7b)into Eqs.3l)I.2a) wo(x. t) = 2/.3) Substitution of Eqs.W.Ym.3.I)(?. C C Umn(t)sin CYXcos jjy 0 0 03 (7. and (7. (3. 1.3.
as was discussed for antisymmetric crossply laminates. and natural vibration.. Q45 Q45 Q55 (ymn PWmn)sin a x cos p y (X. The inplane stresses in each layer can be computed from the equations  m=l n=l  + azzT&.c). as discussed before.11)(6.) fmn (PXmn + aYmn) gmn + 2~2 T Imn f mn (aXmn XY (7.3.sin .10) m=l n=l Note that the stresses are layerwise constant through the thickness.3.7 gmn = COSa b a cos b (7. and Sij and miJ are defined in Eqs (7. They are ' _ '" c c 0 0 0 0 [Q4.2.3.3. Equation (7.16).3. The transverse stresses can also be determined from the equilibrium equations of 3D elasticity.5) can be specialized for static analysis.3.7b. buckling under inplane compressive loads. aWmn)cos ax sin ~y + + } .9a) mrx nry mrx nrY fmn = sin .9b) The transverse shear stresses from the constitutive equations are given by (7.where and the thermal coefficients are defined in Eqs. (6.
the effect is quite significant. Figure 7. .3. antisymmetric. . . .and eightlayer antisymmetric angleply laminates ( a l h = lo).25.). .3 Bending Table 7. E1/E2= 25. . .2 contains plots of the nondimensionalized transverse deflection versus sidetothickness ratio ( a l h ) of various angleply laminates subjected to uniformly or sinusoidally distributed transverse load (alb = 1.3.3 contains plots of the nondimensionalized deflection as a function of the lamination angle for two. G12 = GIS = 0.3. Table 7. subjected to sinusoidally distributed transverse load. The nondimensionalizations and locations of maximum quantities are as follows: where k = 1. . For values of a l h less than 10.2. . Both constitutive and equilibrium based transverse shear stresses are included in the table. n denotes the ply number. ul2 = 0.2 contains the maximum transverse deflection and inplane normal stress as a function of the modulus ratio of simply supported (SS2) square.7.) angleply laminates ( a l h = 10) subjected to uniformly distributed load. subjected to sinusoidally and uniformly distributed loads.). stresses are not.1 contains numerical results of nondimensionalized maximum deflections and stresses of simply supported (SS2).3. K = 516).5E2. (451451451. twolayer (45145) and eightlayer (451451451. While the deflections are sensitive to the transverse shear deformation. The effect of bendingstretching is significant in twolayer laminates. . (818181. The effect of transverse shear deformation is negligible for all values of a l h greater than 10.3. twolayer and eightlayer antisymmetric angleply square laminates. Figure 7.
K = 516.2E2. Figures 7.4 Buckling Table 7.3. The material properties used are: El = 25E2. Note that for certain modulus ratios.4 and 7. antisymmetric angleply laminates for various modulus ratios and two lamination schemes (45145) and (451451451.3 contains critical buckling loads of uniaxially and biaxially compressed simply supported (SS1) square.5E2.).5E2.3. K = 516). The sidetothickness ratio is taken to be a l h = 10. and the larger of the two is reported.25. G23 = 0. GI2 = GI3 = 0. the shear deformation theory predicts buckling modes different from the classical laminate theory.3. The effect of transverse shear deformation is negligible on the stresses.3.7 show the influence of shear deformation.3.6 and 7. (45145) 10 20 100 CLPT SSL UDL SSL UDL SSL UDL SSL UDL Antisymmetric Laminate. vl2 = 0.1: Effect of transverse shear deformation on nondimensionalized maximum transverse deflections and stresses of simply supported (SS2) antisymmetric angleply square plates (hi = h l n .25. the shear stress derived from constitutive relations will have two values at each interface. Figures 7.2E2.G12 = Gl3 = 0. . 7. 2 Antisymmetric Laminate. The values reported are at z = kh/4 for (45145) laminate.3.3. sidetothickness ratios and lamination schemes. G23 = 0.Table 7. arid a t z = 0 for the (45/45)4 laminate. (45/45)4 10 20 100 CLPT SSL UDL SSL UDL SSL UDL SSL UDL t Maximum stress derived from equilibrium. . El = 25E2. ~ 1 = 0.5 contain nondimensionalized maximum transverse shear stress distributions through laminate thickness for an eightlayer antisymmetric subjected to uniformly or angleply square laminate (45130145/0/0/45/30145) sinusoidally distributed transverse load..
G23 = 0.8 and as a function of lamination angle in Figure 7. The sidetothickness ratio for the laminates in Figure 7.e.2E2. 7. transverse shear deformation (i.829 CLPT 4. a l h = 10. As the number of layers increases. (45145) FSDT 4.7 is taken to be a l h = 10.3.5E2.251 0.) are presented in Table 7.412 Laminate.172 1.759 Laminate.3.902 number of composite layers (bendingextensional coupling).. and rotary inertia is to lower the fundamental frequencies.3. and the lamination angle on critical buckling loads of antisymmetric angleply square laminates under uniaxial compressive loads (alb = 1. The effect of bendingstretching coupling (i. 2 1 in the series are used to calculate the solutions for uniform load). E1/E2varied.25).480 CLPT 4.577 1. G23 = 0. = h l n .3.. G12 = G13 = 0. .e. (45/45)4 FSDT 4.577 2.2: Effect of lamination scheme and shear deformation on the transverse deflections and stresses in square antisymmetric angleply laminates subjected to uniformly distributed transverse load (h.153 0. vl2 = 0. G12 = Gl3 = 0. Theory %=1 E2 10 w* 20 30 40 a=l E2 ffxx  t 10 20 30 40 Orthotropic Plate FSDT 4..25. n = 1 .999 Laminate.829 CLPT 4. The effect of shear deformation decreases with increasing .5E2.Table 7. E.4 for two different materials.2E2.. vl2 = 0. (45/45)2 FSDT 4.. . # 0 and E~~ # O). Numerical results for twolayer (45145) and eightlayer (45/45)4 plates with E1/E2 = 25. vlz = 0.678 1. G12 = GI3 = 0.9.829 CLPT 4.3. .5E2.010 1. the coupling decreases. B16 and B 2 6 ) . and K = 516 are given as a function of sidetothickness ratio in Figure 7. 3 . E1/E2= 25. a l h = 10.25.5 vibration Numerical results of nondimensionalized fundamental frequencies of antisymmetric angleply laminates (451451451. m. .577 1.3.
3. 0 Figure 7. . 0 0 .3: Center transverse deflection w versus lamination angle 0 for simply supported (SS2) antisymmetric angleply (19/8).2 ) antisymmetric angleply (45/45).3.0 10 20 30 40 50 60 alh 70 80 90 100 Figure 7. 4 ) square laminates ( a l h = 10). 4 ) square laminates. 5 0 10 0 15 20 2 25 30 0 35 40 1 45 Angle. ( n = 1 . ( n = 1 .2: Center transverse deflection w versus sidetothickness ratio a / h for simply supported ( S S .
(O.1 0. z) through 0. 0.3 0.3.4: Distribution of transverse shear stress @.5 Figure 7.0.4 0. 0..4 (a/2.5: Distribution of transverse shear stress cy. b / 2 .3.(a/2.b/2.z) 0. thickness z / h for simply supported (SS2) antisymmetric angleply (45130145/0/0/45/30145) square laminates ( a /h = 10). .0 0.2 Stress.5 Stress.z) 0. G (O.1 0.2 0.0 0. z) through thickness z/h for simply supported (SS2) antisymmetric angleply (45/30/45/0/0/45/30145) square laminates.6 Figure 7.SSL i 0.3 0.
   1  0   I I l I ~ l l l 1 ~ l 1 1 l ~ 1 1 l l ~ 1 l l l ~ l l 1 ' ~ l l ' l ~ l 1 l l ~ l l 1 1 ~ l l l l 0 10 20 30 40 50 60 alh 70 80 90 100 Figure 7.6: Critical buckling load ( N ) versus sidetothickness ratio ( a l h ) for simply supported (SS2) antisymmetric angleply (45/45). laminates.3. square laminates.7: Critical buckling load (N) versus lamination angle (0) for simply supported (SS2) antisymmetric angleply (8/19). . 8 Figure 7. 0 5 10 15 20 25 30 35 40 45 Angle.3.
5141 50.6E2.495 20.25.397 17.771 8.067 17. 2 Uniaxial Compression (k = 0).) square plates [G = w ( a 2 / h ) d m .717 7.5Ea. This decrease is slower for eightlayer plates than for twolayer plates.727 9. G12 = G13 = 0. whereas it is significant in CLPT only for very thick plates.854 15.437 20.644 63.364 4.115 7. of simply supported (SS2) antisymmetric angleply.628 21. vl2 = 0.709 24.257 7.910 9. Material 1 Material 2 a/h Theory n =2 n=8 n =2 n=8 5 10 20 100 FSDT CLPT FSDT CLPT FSDT CLPT FSDT CLPT Material 1: El = 25E2. Table 7.322 31. The effect of rotary inertia is negligible in FSDT. square plates (Gl2 = G13 = 0. G 2 = 0.336 25. (45/45).974 64. K = 516).1)].3: Effect of s h e a r deformation on nondimensionalized critical 2 buckling loads.l).163 15.: natural frequencies of simply supported (SS2) symmetric angleply (451451451. N = N.581 7.199 8.l) 10 20 100 CLPT 3.683 Biaxial Compression (k = 1). G23 = 0.values of a / h . ~ 1 = 0.. . G12 = Gl3 = 0.2E2.542 16..637 12.5E2. ~ Material 2: E l = 40E2.774t 19.738 12.25.847 8.476 21.990 40.513 15.818 6. Table 7 3 4 Effect of shear deformation on nondimensionalized .923 4. Gag = 0.25..052 9. mode: (1. mode: (1. K = 516. mode: (1.082t 34.987 32. vlz = 0.435 15. .875 41.066 13.526 4.584 17.930 10.231 14.533 6. .3.l) 10 20 100 CLPT 7.2E2.810 10.5E2.341 t Mode is (2. ( a 2 / ~h3).861 21.792 8. rotary inertia is included.
  2layer. square laminates.9: Nondimensionalized fundamental frequency (w) versus lamination angle (8) for simply supported (SS2) antisymmetric angleply (19/0). (45145) (Rotary inertia included) 1  0 10 20 30 40 50 60 alh 70 80 90 100 Figure 7.3.8: Nondimensionalized fundamental frequency w versus sidetothickness ratio ( a l h ) for simply supported (SS2) antisymmetric angleply (45/45). . square laminates. Figure 7.3.
1: The coordinate system and boundary conditions used on the simply supported (SS1) edges for the L6vy solutions of rectangular crossply laminates using the firstorder shear deformation theory. and then the partial differential equations of equilibrium are reduced to ordinary differential equations in the coordinate parallel to the simply supported edges.4. For additional details and for a discussion of the free vibration and buckling analyses. the discussion is limited to the bending case. natural vibration. the reader may consult References 8 and 23. . As described earlier. The equations of equilibrium appropriate for the antisymmetric crossply laminated plates. clamped.1 Introduction In this section we present the L6vy type solutions for bending. We now proceed to describe the procedure for bending of crossply laminates. the reader may consult the references a t the end of the chapter. For additional details.4. and simply supported boundary conditions (see Figure 7.7. while the remaining edges x = a / 2 and x = a12 have any combination of free. The ordinary differential equations are then solved using the statespace approach. In the interest of brevity. the L6vy solution technique involves choosing a solution form that satisfies the simply supported (SS1) boundary conditions on two parallel edges of a rectangular laminate. Suppose that the edges y = 0 and y = b are simply supported (SSl). can be expressed in matrix form as Figure 7.4.4 Antisymmetric CrossPly Laminates with Two Opposite Edges Simply Supported 7. according to the firstorder shear deformation plate theory. and buckling of antisymmetric crossply laminates.1).
in conjunction with the statespace concept. The transverse load is also expanded as .4. b (see Figure 7.4. (7. can be used to develop analytical solutions of Eq. ~ . = mrlb. It can be easily verified that the boundary conditions (7.5) are satisfied by the displacement field in (7. The generalized displacements are expressed as products of undetermined functions and known trigonometric functions so as to satisfy the simply supported boundary conditions at y = 0. 4 ~ . b and the remaining edges x = fa12 have any boundary conditions. o ) and .1) when the plate is simply supported on the edges y = 0.1): The displacement field is represented as ! where ?. F ~ } . the coefficients and Note that the classical plate theory can be obtained as a special case of the firstorder shear deformation theory by setting 7.4. w o . v ~ .6). o .2 The Lbvy Type Solution The Lkvy method.where { A } ~= { ~ ~ .4.4. o . { 4} ~ = Lij Lji defined as = are { o .
4.8) to a system of firstorder equations (i. the components of the state vector Z ( x ) are defined as .8) are given by In order to reduce the system of equations (7.Substitution of the displacement field (7. The coefficients in Eq. (7..6) into governing equations (7.4.e.4.4.1) results in five ordinary differential equations where the primes denote the derivative with respect to x. use the statespace approach).
and for additional information see [8. clamped (C).4. 7.9. and buckling of rectangular.12) where the matrix T is the 10 x 10 matrix and the load vector r is defined as The solution to Eq.12) is Here K denotes constant column vector. and free (F) boundary conditions a t the edges x = fa/2 are Boundary conditions in (7.4.15271. are used: .8) may be converted t o the form Z'=TZ+~ (7.4.4.11).12) to obtain ten equations for the ten constants Ki.4. The simply supported (S). the following two sets of lamina properties. which is to be determined from the boundary conditions on edges x = fa/2. crossply laminated plates. (7.15) can be used in (7.Using the definitions (7. the systems of equations (7.4. typical graphiteepoxy material.4. For the purpose of comparison. vibration. The same procedure can be used to study natural vibration and buckling under inplane compressive forces.3 Numerical Examples Here we present numerical results for a number of example problems of bending. The procedure was discussed earlier for the eigenvalue problems in Chapter 6.
480 0. Figures 7. Bending The loading in all cases considered here is assumed to be sinusoidal 7rx 7ry q(x.e.064 FSDT FSDT 1. Layers 2 5 10 5 10 Theory SS SC 1.687 1.4.758 FSDT 1. b) are simply supported. (7.897 1..4 and 7. The notation SC.25 (7.045 0. indicating that the shear deformation effect is more significant in anisotropic plates. refers t o the boundary conditions used on the edges x = fa/2.5E2.4.471 1. The classical laminate theory always underpredicts deflections because the plate is modeled as infinitely stiff through the thickness.4.237 C L P T ~ 1.5 contain plots of w vs.615 0.665 FS 2.223 0.5E2.1: Nondimensionalized center deflection (w) of antisymmetric crossply square plates with various boundary conditions (Material 1).579 FC 1.Material 1 : El = 25E2.6Ea.258 0. Gla = G13 = 0. material I ) . the results for deflections and stresses are presented using the following nondimensional form (see Khdeir and Reddy [23]): where h is the total thickness of the laminate. Gas = 0.777 1.385 0. Ga3 = 0. As the degree of orthotropy increases.137 0.1 through 7. Tables 7. y) = qo cos sin a b In the tables and figures.915 0. for example.e.777 2.883 0.429 0. (a = b/2) with various boundary conditions (see Khdeir and Reddy [23]). while the other two edges (i.2 and 7.4.028 1.664 1. the difference between the deflections predicted by the classical and the firstorder shear deformation theories increases. .266 CC 1.16) = 0.800 0. 242 Material 2 : El = 40E2.477 0.4.335 1.980 1.663 0. Table 7.4. .4..17) The shear correction coefficient for the firstorder theory is taken to be K = 516. and a = b/2. E1/E2for the same load.2E2.25 (7.612 0. .656 0.4.4.442 10 CLPT~ t Results are independent of blh.167 FF 2..460 0. Figures 7. b/h = 10.945 0.257 0.4 contain numerical values of deflections and stresses in square plates obtained using the L&y method. y = 0. Glz = GIs = 0.4. ula = 0.3 contain plots of deflections versus sidetothickness ratio ) crossply laminates b/h of twolayer and tenlayer antisymmetric (0/90/ . The material properties used are those listed in Eq.16) (i.380 FSDT 1.
2: Nondimensionalized center transverse deflection (w) versus sidetothickness ratio (blh) for antisymmetric crossply (0/90). laminates (Material 1.60 0 2 4 6 8 10 12 14 16 18 20 blh Figure 7.4.3: Nondimensionalized center transverse deflection (w) versus sidetothickness ratio (blh) for antisymmetric crossply (0/90). 5 ) laminates (Material 1.. layers: (0/90)5 10 0.4. . 0 2 4 6 8 10 12 14 16 18 20 blh Figure 7. bla = 2). bla = 2). (n = 1 .
bla = 2). .4. blh = 10. blh = 10. Figure 7. bla = 2).4.5: Nondimensionalized center transverse deflection (w) versus modulus ratio (E1/E2) for antisymmetric crossply (0190) laminates (Material 1.4: Nondimensionalized center transverse deflection (w) versus modulus ratio (E1/E2) for antisymmetric crossply (0190) laminates (Material 1.Figure 7.
790 3.135 1.986 FC 2.4.157 7.628 3.494 5. Layers h/h SS SC CC FF FS FC Table 7.660 3.338 5.723 1..566 6.009 5.009 5.430 4.7. Similar results for critical buckling loads under uniaxial compression are included in Table 7.4. square laminates can be seen from the results presented in Table 7.4.957 2.450 4.Table 7.157 7. The fundamental frequencies increase with an increase in degree of orthotropy arid number of layers (or decrease of coupling).800 2.157 7.884 11.642 3.4.483 4.009 1 Results are independent of blh.848 9.829 CC 3.560 5.. The effect of transverse shear deformation and boundary conditions on the fundamental frequencies of twolayer and tenlayer antisymmetric crossply laminates ( n l h = 10) are examined in Table 7. Vibration and Buckling The effect of orthotropy and number of layers (i.403 1.442 2.442 2. Layers 2 5 10 5 10 Theory FSDT FSDT CLPT~ SS 7.009 FSDT 5.042 1.707 3.911 4. In all cases.153 3.594 1.029 4.6. .157 SC 6.e.025 CC 5.837 6. bendingstretching coupling) on the fundamental frequencies of simply supported.531 FC 8.212 3.4.434 2.) of antisymmetric crossply square plates with various boundary conditions (Material 1).533 7. the classical plate theory overpredicts frequencies and buckling loads.4.157 5.4: Nondimensionalized axial stress (ayy) antisymmetric crossply of square plates with various boundary conditions (Material 1).3: Nondimensionalized shear stress (ay.914 4.847 9.435 4.009 C L P T ~ 5.904 3.480 FS 9.8.) of antisymmetric crossply square plates with various boundary conditions (Material 1).712 1.911 FF 11.583 7.157 7.865 10 CLPT~ Table 7.284 10 FSDT 5.009 SC 5.047 7.275 2.725 FS 4.109 4.692 3.150 6. crossply.590 6.343 1. Theory FSDT FSDT CLPT~ FSDT FSDT SS 7.907 11.706 5.034 5.5 (see Reddy and Khdeir [8]).469 2. The critical buckling loads for the same laminates under uniaxial compression are presented in Table 7.4.849 7.799 2.968 2.167 FF 2.2: Nondimensionalized axial stress Layers 2 5 10 5 10 (a.
4.610 14.044 23.779 18. Theory FSDT CLPT FSDT CLPT FSDT CLPT FSDT CLPT Layers = 3 gk 10 20 30 40 Table 7.4.152 18.267 10.881 7. of simply supported (SS1) antisymmetric crossply square laminates ( a l h = 10) under uniaxial compression (Material 2).906 FC 7. a = w .6: Effect of degree of orthotropy of the individual layers on the 2 dimensionless critical buckling loads.862 13.5: Effect of degree of orthotropy of the individual layers on the dimensionless fundamental frequency of simply supported antisymmetric crossply square laminates: a / h = 5..471 31.779 SS 10.900 12. N = N.543 20.971 CC 15.7: Effect of number of layers and transverse shear deformation on the = ( ~ b ~ / h ) ( ~ /of antisymmetric ~ ~ ) ' / ~ dimensionless frequencies crossply square plates ( a l h = 10) with various boundary conditions (Material 2). b2 / ( ~ h3).215 7.4.741 8.709 10 . Theory Layers 10 20 30 40 FSDT CLPT FSDT CLPT FSDT CLPT FSDT CLPT Table 7.223 18. .492 SC 12. Layers 2 Theory FSDT CLPT FSDT CLPT FF 6.079 12.Table 7. I " m (Material 2).228 11.154 15.636 11.680 FS 7.473 11.
we present the Lkvy solution procedure for natural vibration and buckling analyses..232 SC 16.389 SS 11.Table 7.2 Governing Equations Consider a rectangular laminated plate composed of an even number of identical layers having the principal material directions of orthotropy oriented at angles +Q and Q with respect to the xaxis of the laminate (i.450 35.968 14.5 Ant isymmetric AnglePly Laminates with Two Opposite Edges Simply Supported 7.957 25.280 34. antisymmetric angleply laminates).524 17.851 5. b2/ E~h of " antisymmetric crossply square plates ( a l h = 10) with various boundary conditions (Material 2).8: Effect of number of layers and transverse shear deformation on dimensionless critical buckling loads. buckling.1.166 6.116 32.5. the Lkvytype solutions of the firstorder theory can be developed for bending.003 12.5.5) associated with the firstorder theory take the form .770 FSDT CLPT FSDT CLPT 7.1 Introduction As in the case of classical laminate theory.437 21. and the differential equations (7.288 CC 20.4.837 89. 7. and free boundary conditions.1.614 59. simply supported. N = N.023 FC 6.425 12.1)(7. The laminates exhibit twistingextensional coupling..067 31.353 12. In this section.092 16. Layers 2 10 Theory FF 4.e.351 6. and natural vibrations of antisymmetric angleply laminated rectangular plates with two opposite edges simply supported and the remaining ones subjected to a combination of clamped.358 19.426 FS 5.
a : Simply supported (SS1) at edges y = fb / 2 : I at x=0 and x=a I Figure 7.The following boundary conditions are considered (see Figure 7.5.5. .1: Boundary conditions used on simply supported edges for the L6vy solutions of rectangular angleply laminates (FSDT).1): Simply supported (SS1) at edges x = 0 .
The edges x = 0. a are assumed to be simply supported while the remaining edges.5. = $hy = 0 7. y = f b / 2 .3 The L6vy Solution Here we present the Lkvy type solution procedure in conjunction with the statespace concept to determine the compressive buckling loads of rectangular plates (ax b). The following representation of the displacement field is used: .Clamped (C) a edges y = fb/2: t U o = vo = wg = 4 . having an arbitrary set of boundary conditions.
can be expressed as where the primes indicate differentiation with respect to y. and fizyt o zero. are defined as Introducing the components of the state vector Z = Z(y) as . (7. and the constants C.5)' after setting the inertia terms. yields a system of ordinary differential equations in the ycoordinate.where a = mrla. q . (7.5. after some elementary algebraic manipulations.1)(7.5. These equations.5. Substitution of Eq.11) into Eqs.
5. The loading in the case of bending is assumed to be sinusoidal. (7. and Eq. The above solution procedure is also valid for the free vibration case.15) in conjunction with the boundary conditions yields a homogeneous system of equations for the buckling problem and setting the determinant of [MI to zero allows determination of the buckling loads associated with the rnth mode for the boundary conditions a t y = fb / 2 . In the present case all eigenvalues of matrix [TI are distinct.4. (7.Eq.16) and (7. Equation (7. . The shear correction coefficient for the firstorder theory is taken to be K = 516.5. (7.5. the elements of [TI are modified by setting the inplane force terms to zero.4 Numerical Examples In the examples presented here the two sets of lamina properties given in Eqs.5.14a) and equations for the determination of the constant vector K will be modified accordingly.17) are used. except that the elements of the operator [TI should be modified to account for the inertia terms. the Jordan canonical form must be used.12a) may be reduced to the matrix form (Z' = T Z )  0 1 0 0 0 0 0 0 0 0 0 0 IT]= C1 0 0 0 0 0 0 0 c19 c6 c7 0 0 0 0 0 0 0 0 0 0 c 4 1 1 c 5 0 0 C2 1 0 0 0 0 0 0 c20 c8 0 C1l 0 0 C3 0 0 1 0 0 0 0 cl 2 0 C4 0 0 0 0 1 0 0 c2 2 C5 0 0 0 0 0 0 0 c23 0 0 C10 cg 0 C12 0 0 C13 0 c8 1 0 c6 1 0 c 1 7 0 0 0 0 1 0 A formal solution to Eq. (7. In the case of repeated eigenvalues. as discussed in Chapter 6. 7. The deflections and stresses are nondimensionalized as given in Eqs.5.4.19).14a) is given by where K is a constant column vector to be determined using the boundary conditions.4. In the case of static analysis. (7.
and "12 = 0.5.531 0.0)~ 2 x h of simply supported.737 FS 6.701 0.735 10.160 1.and sixteenlayer angleply (818/8/81.747 1.76 GPa). .075 0.010 1.284 3.623 1.470 6.5. a l h = 10).285 CC 1.471 0. angle. G23 = 0. The material properties used are those of Material 2 listed in Eq.2 shows the effect of sidetothickness ratio ( a l h ) on the nondimensionalized center deflection of a square antisymmetric angleply laminate (45145145145) under uniformly distributed transverse loading and for various boundary conditions.944 0.558 0. under various boundary conditions (see Khdeir [21]). .234 2. Theory FSDT CLPT FSDT CLPT FSDT CLPT FSDT CLPT 1 2 2 10 20 30 SS 3. two.446 2. Table 7.412 0.38 GPa). fourlayer antisymmetric angleply square plates (45145145145 under uniformly distributed transverse load (Material 2.2 x lo6 psi (132.and with different values of E1/E2 (Material 2).602 0.379 1. and sidetothickness ratio for simply supported (SS2) angleply square plates under sinusoidally distributed transverse load.743 4.61 GPa).531 0.423 2.049 5.82 x lo6 psi (5.17).771 0.343 1.345 1.1 contains maximum nondimensionalized deflections w = wo (a/2.747 0.25.313 0.) square plates under sinusoidal loading (see Reddy and Chao [lo]).5.523 x lo6 psi (3.518 0. E2 = 1.000 0. The effect of the ratio of principal moduli (Material 2) on the nondimensionalized center deflection is shown in Figure 7. The material properties used are El = 19.225 0. (7.753 1.Bending Table 7.611 2.2: Nondimensionalized deflections of simply supported (SS2 .109 1.5.65 GPa). Table 7.657 3.56 x lo6 psi (10.447 6.5.542 0.214 0.571 4.422 2.375 3. G12 = GI3 = 0.1: Nondimensionalized deflection w as a function of number of layers.417 0.372 SC 2. Figure 7.218 FF 10.712 .2 contains nondimensionalized deflections 6 of angleply laminates subjected to uniformly distributed transverse load.943 FC 4.214 1.5.4. ~ Table 7.3 for the same laminate.
2: Nondimensionalized center transverse deflection (w) versus sidetothickness ratio ( a l h ) for square. Figure 7. a l h = 10).5. .5.3: Nondimensionalized center transverse deflection (t3) versus modulus ratio ( E 1 / E 2 ) for square. antisymmetric angleply ( 4 5 / 45145145) laminates (Material 2 . antisymmetric angleply (45145145145) laminates.Figure 7.
48 10. 0. The material properties listed in Eq.37 4.).45 9. N = ~ .  Table 7.75 20.64 19.82 29. .81 27.69 12.57 22.5.33 15.00 19. included in Table 7.44 19. .95 13.5. a ~ / for ~ h same laminates under uniaxial compressive load are ~ the ~ .4.60 13. The critical buckling loads.24 18. under various boundary conditions and with different values of 6 are presented in Table 7. .88 14.96 Table 7.68 14.51 23.26 8.87 6.68 14.48 15.86 CC 14. .84 7.38 7.62 FF 6./6) square plates (Material 2.5. (7.95 SC 13.23 16./6) square plates (Material 2.57 7.33 3. N.52 8. 0" n Theory FSDT CLPT FSDT CLPT FSDT CLPT FSDT CLPT FSDT CLPT FSDT CLPT SS SC CC FF FS FC . 8" 30 10 2 45 10 2 60 10 n 2 Theory FSDT CLPT FSDT CLPT FSDT CLPT FSDT CLPT FSDT CLPT FSDT CLPT SS 12.3.58 15.12 6.11 12.51 23.11 25.92 FC 8. .5.24 18.4 shows the effect of sidetothickness ratio ( a l h ) and Figure 7.5. of antisymmetric angleply (196 / 8 / .89 3. a/h = 10).04 14.38 25.63 19.76 5.65 9. a l h = 10).52 17.25 33.95 7.54 9.59 14. Figure 7. 0 w(a2/h) of antisymmetric angleply laminates (81 6/O/ .31 21.91 14.Vibration and Buckling The dimensionless frequencies.82 4. of antisymmetric angleply (818/6/ .74 19.27 28.46 15.21 37.03 5.32 FS 8.70 7.32 16.47 12.3: Effect of ply angle (8) and number of layers (n) on dimensionless fundamental frequency.47 3.17 6.22 11.58 10.13 7.17) were used.79 10.5 shows the effect of the ratio of principal moduli on the nondimensionalized fundamental frequencies of a square antisymmetric angleply laminate (45145145145) for various boundary conditions (see Khdeir [17. 211).53 9.4.4: Effect of ply angle (8) and number of layers (n) on dimensionless critical buckling load.48 21.5.35 12. . .41 17.
antisymmetric angleply (45145145145) laminates ( a l h = 10). antisymmetric angleply ( 4 5 / 45145145) laminates (Material 2 ) .5. .Figure 7. Figure 7.5: Nondimensionalized fundamental frequency (w) versus modulus ratio ( E 1 / E 2 )for square.4: Nondimensionalized fundamental frequency (w) versus sidetothickness ratio ( a l h ) for square.5.
3. A plot of the deflection under sinusoidal load (obtained with FSDT) is also included for comparison.6 contain plots of nondimensionalized center deflection for twolayer angleply (45145) laminates under sinusoidally and uniformly distributed step loadings. h/2) and shear stress axy(a.2) are replaced with those in Eq. Note that the effect of shear deformation on the amplitude of stresses is negligible. (7. The effect of shear deformation is to increase the amplitude and period of the waves. for a l h = 10. For crossply laminates under uniformly distributed load. it still increases the period.7a) for crossply laminates and in Eq. however.6. Angleply laminates show larger increase in the period due to shear deformation. 4.. the ratio of maximum transient deflection to static deflection is found to be 2.12 show plots of nondimensionalized transverse shear stress a.062 for CLPT. (7.5E2.6. G12 = GI3 = 0.6.6 Transient Solutions Transient solutions of antisymmetric crossply and angleply laminates using the first order theory can be developed either by the statespace approach or the combination of the Navier solution procedure and the Newmark time integration scheme as discussed in Section 6.5) for angleply laminates.3 and 7.11 and 7.6. 7u = ~ ~ [ ~ ~ h lo2. in twolayer plates (0190) and (45145) under uniformly and sinusoidally distributed transverse loads ( a l h = 10).6.6.7 show a plot of the maximum inplane displacement (uo or vo) versus time.049 for FSDT. whereas it is 2.7 is valid here when the coefficient matrices [MI and [K] in Eq. E2 = 2. G23 = 0. versus timeafor antisymmetric crossply laminates (0190) under ~ / ( ~ ~ ~ ) ] sinusoidally distributed step loading (see Figure 7. while Figures 7.6. h/2) for the uniformly distributed load case ( a l h = 10).6.1 contains the results of transient and static analysis.6. Figures 7.1 shows plots of nondimensionalized center deflection. (7. From this. Figures 7. Figure 7. q5y). Similar results are presented in Figure 7.7. Table 7..3.s ' / c m ~a l b = 1. ul2 = 0. (6.2.1 x lo6 N/cm 2 .7. the ratio of maximum transient deflection to static deflection is found to be wd/wS = 2. Figures 7.6. Here we present numerical results based on this procedure.6.6.b/2.b = 25 cm .7.vo. The same material properties as in Eq. Figure 7. The solution vector {A) consists of the amplitudes of the five generalized displacements.wo.1 for the coordinate system used here). respectively. For angleply laminates under uniformly distributed load.b.1) are used. whereas it is 2.4 contain the nondimensionalized center normal stress axx(a/2. Lastly.2E2.6.1) Results obtained with CLPT and FSDT are presented for two values of sidetothickness ratios.8 through 7. one may conclude that the effect of transverse shear is greater on dynamic response than static response in the twolayer crossply plates.25 p = 8 x 10@ ~ . a l h = 10 and 25.5 and 7.041 for FSDT.10 show plots of normal stress at top and bottom of the laminate and shear stress at the bottom of the laminate for the uniform load case. The application of the statespace approach to the transient analysis of shear deformable theories can be found in the papers of Khdeir and Reddy [25271. The plots in dashed lines correspond to CLPT. The procedure of Section 6. (7.2 for uniformly distributed step loading.6.017 for CLPT. This indicates that the effect of transverse shear is less on dynamic response than on static response in the twolayer angleply plates. . (uo. The material properties used are El = 25E2.
1: Nondimensionalized center transverse deflection ( a )versus time (t) for simply supported (SS1) antisymmetric crossply (0190) laminates. 0 200 400 600 800 1000 Time. .2: Nondimensionalized center transverse deflection (w) versus time (t) for simply supported (SS1) antisymmetric crossply (0190) laminates.6.0 200 400 600 Time.6. t (ps) Figure 7. t (ps) 800 1000 Figure 7.
. .4: Nondimensionalized normal stress (ayy) versus time ( t ) for simply supported (SS1) antisymmetric crossply (0190) laminates.Figure 7.3: Nondimensionalized normal stress (a.) versus time ( t )for simply supported (SS1) antisymmetric crossply (0190) laminates. Figure 7.6.6.
.6.5: Nondimensionalized center transverse deflection (w) versus time ( t ) for simply supported (SS2) antisymmetric angleply (45145) laminates.6.6: Nondimensionalized center transverse deflection (w) versus time ( t ) for simply supported (SS2) antisymmetric angleply (45145) laminates. t (ps) 800 1000 Figure 7. Figure 7.ANALYTICAL SOLUTIONS OF RECTANGULAR LAMINATES USING FSDT 433 FSDT 0 200 400 600 Time.
7: Nondimensionalized inplane displacement ( u ) versus time (t) for simply supported (SS2) antisymmetric angleply (45145) laminates.8: Nondimensionalized normal stress (a.6. 0 200 400 600 800 1000 Time. at the top of the laminate) versus time (t) for simply supported (SS2) antisymmetric angleply (45145) laminates. t (p) Figure 7. t (ps) Figure 7. ..6.434 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS Time.
10: Nondimensionalized shear stress (azy) versus time ( t ) for simply supported (SS2) antisymmetric angleply (45145) laminates. 0 200 400 600 Time. . t ( p s ) 800 1000 Figure 7.6.6. t (p) 800 1000 Figure 7.20 j l l l l ~ l l l l ~ l " l ~ " l l ~ l l l l ~ l l l l ~ l l l l ~ l l l l [ l l l l ~ l l l ~ 0 200 400 600 Time.9: Nondimensionalized normal stress ( 8 at the bottom of the laminate) versus time ( t ) for simply supported (SS2) antisymmetric angleply (45145) laminates.ANALYTICAL S O L U T I O N S O F R E C T A N G U L A R L A M I N A T E S U S I N G F S D T 435 0.
.6.) versus time ( t ) for simply supported antisymmetric crossply (0190) and angleply (45145) laminates.11: Nondimensionalized transverse shear stress (a.) versus time ( t ) for simply supported antisymmetric crossply (0190) and angleply (45145) laminates.0 200 400 600 800 Time. 0 200 400 Time.t (ps) Figure 7. ..6.12: Nondimensionalized transverse shear stress (a.t (PSI 600 800 Figure 7.
Ti) 0 3 ) .442 t T h e first line is t h e transient solution and t h e second line is t h e static solution 7.ANALYTICAL SOLUTIONS OF RECTANGULAR LAMINATES USING FSDT 437 Table 7.25. The effects of material properties. lamination scheme.6. Glz = GI3 = 0. and electrorheological fluids provide examples of such materials. Piezoelectric materials exhibit a linear relationship between the electric field and strains for low field values (up to 100 V/mm). TerfenolD.427 0. a drift from zero state of strain is observed under cyclic electric field conditions.127 0. E2 = 2.028 0.947 3. The advantage of incorporating these special types of materials into the structure is that the sensing and actuating mechanism becomes part of the structure so that one can monitor the structural integritylhealth of the structure.611 1.120 1. piezoelectric materials show dielectric aging and hence lack reproducibility of strains. and they exhibit nonlinear behavior and hysteresis for large electric fields [37].275 0.181 0. and lead zirconate titanate or PZT (Pb (Zr.093 2.096 0. Furthermore. Examples of piezoelectric materials are Rochelle salt.1 x lo6 psi. a magnetostrictive material [38]. a = 25 cm. Theory (0/90) (45/45) FSDT~ CLPT 3.832 0. There are a number of materials that have the capability to be used as a sensor or an actuator or both.e. has the characteristics of being able to produce strains up to 2500 pm and energy density m ~ as high as 25000 ~ / in response to a magnetic field. Piezoelectricity [36] is a phenomenon in which some materials develop polarization upon application of strains. shape memory alloys.695 0. In this section.6 for references).181 0. Among these. Beneddou [39] surveyed more than 100 papers and discussed the research trends in piezoelectric finite element modeling.421 1. a / b = 1.813 0.126 0. vl2 = 0.5E2.274 0. G23 = 0.279 2. a / h = 10).7 Vibration Control of Laminated Plates 7.41].7. . A simple negative velocity feedback control is used to actively control the dynamic response of the structure through a closedloop control.795 0.351 0. Piezoelectric materials. piezoelectric and magnetostrictive materials have the capability to serve as both sensors and actuators.. quartz.1: Nondimensionalized deflections and stresses of simply supported crossply (0190) and angleply (45/45) square plates under uniformly distributed transverse load (E1/E2= 25. control of the transient response of laminated composite plates with integrated smart material layers is presented [40.2E2. electrostrictive materials.990 1. magnetostrictive materials.799 0.1 Preliminary Comments The study of smart materials and structures has received considerable attentions in recent years.348 0. There have been a number of studies on vibration control of flexible structures using smart materials (see Section 4. and placement of the smart material layer on deflection suppression are studied [4l]. i.
6.7. the magnetic field intensity H is expressed in terms of coil current I (x. and number of turns n. (4. (3.7. The constitutive relations of the kth lamina take the form [see Eq. the force and moment resultants are related to the strains by where K is the shear correction factor.1) and (7.4. 7.2). coil radius r. In view of the constitutive equations (7.. (3.3.17a)l (k (k) where Qij ) are the transformed plane stressreduced stiffnesses. t ) as [see Eqs. in the coil by and c(t) is the control gain. y.7.27)]. and the actuation stress resultants { N M ) and { M M )are defined by = ck.7.3 Velocity Feedback Control Considering velocity proportional closedloop feedback control.nmfl . electrostrictive..8)(4.6. or magnetostrictive coupling moduli of kth lamina.4.23)(3.10)] where kc is the coil constant.4. which can be expressed in terms of the coil width b. and eii are the transformed piezoelectric.2 Theoretical Formulation The governing equations of motion for FSDT remain the same as before [see Eqs.12a) and (3.7. k=m.
displacements (uo.q5y) by substituting for the force and moment resultants in terms of the generalized displacements. + ayax ) The simply supported boundary conditions for the firstorder shear deformation plate theory (FSDT) are The mechanical load and magnetostrictive moments are also expanded in double Fourier sine series as . The equations of motion of the firstorder theory can be expressed in terms of the generalized vo. ( a24x a2&.4 Analytical Solution For simply supported plates. neglect the inplane contributions). 4x. For homogeneous laminates. w".e.ANALYTICAL SOLUTIONS OF RECTANGULAR LAMINATES USING FSDT 439 7. we can develop the Navier solution.7. Here we consider the pure bending case (i. the equations of motion take the form Dll a24x ax2 + ~ 1 2ayax + D66  a24..
7.MZ( x . t ) sin a x sinpy dxdy ( x . For vibration control.for a nontrivial solution we obtain the result where .y.8)(7.17) into Eq. (7.15).12) into Eqs. . (7.7. (7.7).7.7.7. y .15) in the form Substituting Eq.y. (7.Y . we assume q = 0 and seek solution of the ordinary differential equations in Eq. ( t )sin a x sin y 00 00 where.7. for example Qmn ( t ) = MA. t ) = x x MA.By dxdy 0 ~ g ( x . sinoxsinpy dxdy t) 0 Substituting Eq. ( t ) = ab ~ : ~ (= ) t ab la 1 4 Sa MZ ASa1 4 b b 4(x. (7. Cij and A .7. = Mji are defined by where the magnetostrictive coefficients ES1 and &32 are defined in Eq.14)we obtain where sij= S j i . t ) sin a x sin . (7.
Numerical studies are carried out to obtain the natural frequencies.27 8.30 0. p. magnetostrictive damping coefficients and the vibration suppression time. The lowest one corresponds t o the transverse motion. A typical eigenvalue can be expressed as A= a iwd.9 4. = lo4 (7.78 206. moment of inertia. = 9250 kgm3.96 6. Magnetostrictive damping coefficients and natural frequencies for various materials and lamination schemes are also listed in Table 7. j = 3 .9 53. Table 7.5 GPa.22) The numerical values of various material and structural constants (e.93 20. All values of the composite material and structural constants are tabulated.7.26 0.96 6.96 7. CFRP GrEp(AS) GIEp BrEp 138.7. = E. magnetostrictive material constants) based on different lamination schemes and material properties (CFRP. so that the damped transverse deflection is given by + In arriving a t the last solution. 5 .96 6.0. Also. the following initial conditions are used: 7. and damped and undamped frequencies are presented in the form of figures.1.g.5 Numerical Results and Discussion Numerical results are obtained using the formulation presented above.14 4. dl.96 17.7.3. c r .6 137.30 1824 1450 1900 1950 .67T8 m ~ . glassepoxy (GIEp).21 3.for i .3. a time ratio relation between the thickness of the layers and the distance to the neutral axis of the laminated composite plate is obtained.20 8. Various lamination schemes are used to show the influence of the position of the pair of magnetostrictive layers from the neutral axis on the vibration suppression time. graphiteepoxy (GrEp)(AS).' .20 8.45 4.7. v = 0.56 0. = 26. The plate is taken t o be a unit square of l m x lm.2 and 7.69 4. 4 . Magnetostrictive material properties are taken to be 1. boronepoxy (BrEp)) are listed in Tables 7.7.1: Material constants of various composite materials.9 0. This equation gives three sets of eigenvalues. The composite lamina material properties are listed in Table 7.12 7.7. .9 8.7.
889 BrEp [*45/m/0/90]s 5.729 4.09 33. and transient response of rectangular laminates with various boundary conditions are presented based on the firstorder shear deformation laminate theory.53 1.015 3.09 30.62 6. natural vibration.954 GrEp(AS) [f45/m/0/90Is G1Ep [ ~ t 4 5 / m / O / 9 0 ] ~ 2.614 7.0921 2. It can be observed that attenuation favors the higher modes. Figure 7. 7.13 a fwd.7.62 7. laminate ( m denotes the magnetostrictive layer).196 2.157 3.83 22.13 30.8 Summary Analytical solutions for bending. The value of a! [see Eq. I.432 3.052 1. The Navier solutions were developed for two classes of laminates: antisymmetric crossply laminates and antisymmetric angleply laminates.7.015 1..13 22.09 33.257 1. (rad sl) CFRP [f45/m/O/90]s [45/m/ .1 shows a comparison of uncontrolled and controlled amplitude of the center deflection of ( m / f 45/0/90).303 1.62 6.3: Mass inertia coefficients and parameters Material Laminate a! and wd.897 0.62 6.461 3.73 CFRP [f45/m/0/90]s [45/m/ .09 33.78 7.62 6.45/0/90Is [ m / 31 45/0/90Is [m/g04ls Table 7.514 2.029 2. respectively.552 3.54 2.816 1. .352 4. each for a specific type of simply supported boundary conditions.538 3.979 2. Material Laminate &(lo3) IN ml D12(1O3) D22(1O3) D66(103) A44(107)A55(1O7) [N ml [N ml [N ml P/mI [N/ml 2.7 34.432 [ d o 4 ]s 7.54 4.274 0.1 Figure 7.739 3.7146 0.7.974 7.442 MECHANICS OF LAMINATED COMPOSITE PLATES AND SHELLS Table 7.62 6.62 7.54 4.09 33. buckling under inplane conlpressive loads.435 2.974 7. I2(1OP4)ES1 [kg/ml [kg ml 2. indicating faster vibration suppression.2: Coefficients for different laminates and materials.751 6.614 7.83 39.0921 0. SS1 and SS2. The Lkvy solutions with the statespace approach were developed for these classes of laminates when two opposite edges are simply supported and other two edges having a variety of boundary conditions of choice.20)] increases when the magnetostrictive layer is located farther away from the neutral axis.528 2.215 3.45/0/90]s [m/ f 45/0/90]s [m//904]s [m/041s GrEp(AS) [f45/m/O/90]s G1Ep [f45/m/O/90]s [&45/m/O/90]s BrEp 33.1 33.13 22.066 3.149 3.066 6.83 39.2 contains controlled amplitudes of the center deflection for modes 1 and 2 for [f 45/m/0/90Is laminate.55 22.7.62 6.7.62 6. This is due to the larger bending moment created by actuating force in the magnetostrictive layers.7146 2.221 1.62 6. (7.98 39.
ANALYTICAL SOLUTIONS OF RECTANGULAR LAMINATES USING FSDT 443 0.7. laminate.4 Figure 7.1: Comparison of the uncontrolled and controlled center deflection of (m/ f 45/0/90).1 0.3 0. laminate. . 1 mode n= 1 mode n=2 111.~ Figure 7.2 Time ( s ) 0.2: Comparison of the uncontrolled and controlled center deflection of (f45/rn/O/gO).7.006 0 0.
eg = P 2 ~ s 6 I ~ W & e l l = P2A22 el5 = P'K +low:. . Additional results can be found in [41]. relationships between deflections. show that the operator [TI in Eq.23) 7. The effect of transverse shear deformation on transient response is to increase both amplitude and period of oscillation. numerical results are also presented for vibration suppression of simply supported laminated plates with magnetostrictive layers./ 3 2 ~ 2 2 e2l = ~ ~ + + +  . and transient response of antisymmetric crossply and angleply laminates. e29 = . In particular. = .KA55 +I~w?.P 2 ~ A q. The bendingextensional coupling and transverse shear deformation in a laminate generally reduce the effective stiffnesses and hence increase deflections and reduce buckling loads and reduce natural frequencies. In a series of papers.4. (7. (7.e22 = 6 w 6 e28 = p2B22 + ~ l w : . where 7.2 Verify the expressions for transverse stresses presented in Eq. A ~ ~ IOwL + I ~ ~ : ~ . buckling loads and vibration frequencies predicted by the firstorder shear deformation plate theory and the classical plate theory of isotropic plates were presented (see [42451 and references therein). KA44 + 1 2 ~ : ~ 7.4.. show that the operator [TI in Eq./ ! ? 2 ~ y y el5 4 .P 2 ~ 2 2 IIW?. In particular. The coupling is the most significant in twolayer laminates.2.48] must be carried out for various types of lamination schemes. Analytical solutions for bending. Extension of these ideas to composite plates has not been done. buckling and vibration of stepped laminated plates [46]or laminated plates with internal hinge [47.13a) holds with e l 3 = KA55 + N~. (7. and it decreases gradually as the number of layers is increased for fixed total thickness.3 Formulate the Lkvy type solution procedure for the natural vibration o f antisymmetric crossply laminates.4 Formulate the L6vy type solution procedure for the buckling o f antisymmetric crossply laminates under inplane compressive loads. Lastly. el2 = .444 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS Numerical results were presented for static bending.13a) holds with eg = /!?2~66 + IOU&. natural vibration. buckling.
The Netherlands (1994).5. 931 933 (1972). John Wiley. . A. "Exact Solutions for Rectangular Bidirectional Composites arid Sandwich Plates. 7. 27(12). Brogan. L. 6. A . Englewood Cliffs. PrenticeHall. and Hatfield: S. Matrix Theory. J." Acta Mechanics. N. Englewood Cliffs.14b) holds with and all other 6..ANALYTICAL SOLUTIONS OF RECTANGULAR LAMINATES USING FSDT 445 7. References for Additional Reading 1. N. show that the operator [TI in Eq. 94(3. 10. Mechanics of Composite Materials.eering and Design. L. 18081817 (1989). A. "Buckling and Vibration of Laminated Conlposite Plates Using Various Plate Theories.. Reddy.5 Verify the expressions for transverse stresses presented in Eq. 20 34 (1970). 7. (7. Reddy. N. 3. J." Journal of Applied Mechanics. . 2. New York (1984). 10(7). J." A I A A Journal.4).. N. 64.Journal. J.11). N . Franklin. Reddy. Pagano. J . = N.." Nuclear Engin. J . = 0. 4(1).12b) with N. Energy and Variational Methods zn Applied Mechanics. and Khdeir.6 Formulate the L6vy type solution procedure for the natural vibration of antisyrrirnetric angleply laminates. N. N J (1968). Selected Works of Nlcholas J. N. Pagano.. N J (1985)..). J. A. PrenticeHall. 54. In particular. 153167 (1981)." Journal of Conposite Materials.. J . 123 170 (1992). J.5. 640642 (1987). and Librescu. J .. . Modern Control Theory. "A Comparison of ClosedForm and Finite Elerrlent Solutioris of Thick Laminated Anisotropic Rcctangdar Plates. "On Vibration and Buckling of Symmetric Laminated Plates According to Shear Deforrriatiori Theories. 8. Nosier. Khdeir.. N. W.3. 9. Pagano. 5." A I A A . 4. "Elastic Behavior of Multilayered Bidirectional Corriposites.. A . and Chao. and Reddy. W.. (ed. Reddy. C. Reddy.. (7. are as defined in Eq. N. "Litvy Type Solutions for Symmetrically Laminated Rectangular Plates Using FirstOrder Shear Deformation Theory. (7. Kluwer.
11.. N. C. and Librescu. J. and Reddy. S. J. "Exact Solutions for the Transient Response of Symmetric CrossPly Laminates Using a HigherOrder Plate Theory. and Reddy. and Hsu.245258 (1989). 113(4)." Journal of Sound and Vibration... 14. A." Journal of Sound and Vibration. N. "Free Vibration of Antisymmetric AnglePly Laminated Plates Including Various Boundary Conditions." Journal of Sound and Vibration. "Analysis of Symmetric CrossPly Laminated Elastic Plates Using a HigherOrder Theory: Part IStress and Displacement. and Reddy. N. Khdeir. Reddy. 22. S. 9. 28." International Journal of Solids and Structures. N. 30." Composites Science and Technology. Y. 178183 (1973). Srinivas. Khdeir. 20. Khdeir A. A.. 17. L. 570578 (1991). N. J. and Rao. V. A. "Dynamic Response of Antisymmetric AnglePly Laminated Plates Subjected t o Arbitrary Loading.. "Buckling of Thick Rectangular Plates. J." A S C E Journal of Engineering Mechanics. "An Exact Approach to the Elastic State of Stress of Shear Deformable Antisymmetric AnglePly Laminated Plates.. J. 15. 9 . 31. "Dynamic Response of Antisymmetric AnglePly Laminated Plates Subjected to Arbitrary Loading. N. 377395 (1989). A. and Librescu. 23. Khdeir. A. Khdeir. "Free Vibration and Buckling of Symmetric CrossPly Laminated Plates by an Exact Method. "Effects of Shear Deformation and Anisotropy on the Thermal Bending of Layered Composite Plates. J. 189213 (1988). 403408 (1982). 447461 (1988).. 12. L. 122(2). A. J. Khdeir. A. 13. 14471463 (1987). . A. 499510 (1989). A. 437445 (1988). 126(3). Khdeir A. 25. 29. 16. Khdeir. 49.. Joga Rao." A I A A Journal. T. and Rao. 475493 (1980). Reddy." Journal of Sound and Vibration.." Composite Structures. 128(3). A. A.." Composite Structures. "An Exact Analysis for Vibration of Simply Supported Homogeneous and Laminated Thick Rectangular Plates. "Stability of Antisymmetric AnglePly Laminated Plates. 419438 (1991).. 27.. and Reddy. K. N. 205224 (1989). "Comparison Between Shear Deformable and Kirchhoff Theories for Bending. and Reddy. 23(10). 7. C." Journal of Pressure Vessel Technology. A. 34. Khdeir A. Buckling and Vibration of Antisymmetric AnglePly Laminated Plates." Composite Structures." Journal of Applied Mechanics. A. "Free Vibration and Buckling of Unsymmetric CrossPly Laminated Plates Using a Refined Theory. J. 14(4). 187199 (1970). Khdeir. "On Theories for the Dynamic Response of Laminated Plates. 126. and Reddy. 11(2)." Journal of Thermal Stresses. N. "On the Forced Motions of Antisymmetric CrossPly Laminates. A. 259277 (1988). N. A. M. L. J. 11. Khdeir. 26. A. 18... A. "Thermal Stresses and Deflections of CrossPly Laminated Plates Using Refined Plate Theories.. Srinivas. S.. A. Khdeir." International Journal of Mechanical Sciences. A." Journal of Thermal Stresses. A. "Analysis of Symmetric CrossPly Laminated Elastic Plates Using a HigherOrder Theory: Part 11Buckling and Free Vibration. and Whitney. 24. 3. Khdeir. A. 12. "On the Solutions t o Forced Motions of Rectangular Composite Plates. A. Librescu. 159172 (1989). A. "Analytical Solution of a Refined Shear Deformation Theory for Rectangular Composite Plates. 126(3). 13. A. Sun.." Journal of Sound & Vibration. 19. J. 377388 (1988).. 437445 (1988). Khdeir.. A. Khdeir. "Analytical Solutions of Refined Plate Theories of CrossPly Composite Laminates." A I A A Journal. 1645 (1969)." Journal of Sound and Vibration. 115. K. and Reddy. A.. A." Composite Structures. A. 952962 (1989). A. 21..
10. and Reddy.A. "Bending." International Journal of Solids and 1481 (1970). A. "Flexure of Thick Rectangular Plates.. 245254 (1992). 1463~ Srinivas. J. 647652 (1986). S. for Active Vibration Control. J. and Rcddy. ." Journal of Intelligent Material Systems and Structures... M. Elsevier. Symmetric CrossPly Laminated Rectangular Plates. "Deflection Relationships Between Classical and ThirdOrder Plate Theories. Pradhan.ANALYTICAL SOLUTIONS OF RECTANGULAR LAMINATES USING FSDT 447 Srinivas. 385 408 (2001). "On Canonical Bending Relationships for Plates. 3039 3067 (2003). hlaugin. "Buckling and Vibration of Orthotropic Plates with an Internal Hinge. Vibration. 21. P. N. C. 3. J. Y. G. C. C. and Reddy. N. 45. 130(34). K. NorthHolland. "Electrostrictive Actuators: Materials and Applicatio~ls..: Lim. 495507 (1973). S. "Natural Vibration of Rectangular Plates with Internal Line Hinge Using the FirstOrder Shear Deformation Plate Theory. and Rao." Smart Materials and Structures. Relatiomhzps with Classical Solution." Journal of Franklin Institute. N. Reddy. C.. TerfenolD. 838849 (2001). 568593 (1999). N. and Lee..: Reddy. and Shoop. K. J. "On Laminated Composite Plates with Integrated Sensors and Actuators... and Reddy. 291.. Structures. T. 6 . A. Y. "Some Results from an Exact Analysis of Thick Laminates in Vibration and Buckling.. Continuum Mechanics of Electromagnetic Solids. S. J. Wang.. 30. 298299 (1973). Srinivas. Srinivas. 457486 (2002)." Journal of Applied Mechanics. "Relationships Between Bending Solutions of Reissner and Mindlin Plate Theories.. 111 (2001). N. 1(3). J. H. K.." In. "Advances in Piezoelectric Finite Element Modeling of Adaptive Structural Elements: A Survey. Goodfriend. Joga Rao. J . M. Uchino. G. Gupta. C." Engineering Structul'es. Wang. Reddy. Y.. "A Refined Analysis of Composite Laminates. A. "A ThreeDimensional Solution for Plates and Laminates. and Rao.al Journa.. 285297 (2003). Xiang. "Adaptive Characteristics of the Magnetostrictive Alloy. 868870 (1970)." Engin. J.... S. 40. 2(4). Ng. K. M. 40. 48." Journal of Sound and Vzbratzon.." Journal of Applied Mechanics... A..d Dynamics.. Y." Acta Mechanics. 469481 (1971). K. Srinivas. J. Xiang. A. U K (2000). Lam. S. "Control of Laminated Composite Plates Using Magnetostrictive Layers. 199208 (1998)." Computers and Structures." Journal of Sound and Vibration. G. and Rao. and Reddy..eering Structures. 347 363 (2000).. and Buckling of Simply Supported Thick Orthotropic Rectangular Plates and Laminates. 44. N. N. Amesterdam.ternationa1 Journal of Structural Stability and Dynamics.." Ceramic Bulletin." Internation. "Buckling and Vibration of Stepped. Benjeddou. 76.. 43. S. R. J.. and Reddy. Lim. 23. 263. K. T. V.." Internatzonal Journal of Solzds and Structures.l of Structural Stability an. M. and Rao.. 46. Wang. 65. N. 37. Shear Deformation Theorzes of Beams and Plates. 47. H. K. The Netherlands (1988).. K. N. T. and Lee. M. K.
.
the individual layers are. pipes. and Taylor [15] formulated a theory of thin shells laminated of anisotropic material that is an extension of the theory developed by Stavsky [28] for laminated anisotropic plates to Donnell's shallow shell theory (see Donne11 [17]). concrete roofs. A shell theory for the unsymmetric deformation of nonhomogeneous.251 and a detailed study of thin isotropic shells can be found in the monographs by Ambartsumyan [I31. anisotropic. Dong. . Shells are common structural elements in many engineering structures. Cheng and Ho 1291 presented an analysis of laminated anisotropic cylindrical shells using Fliigge's shell theory [6]. elastic cylindrical shells was derived by Widera and Churig [30] by means of the asymptotic integration of the elasticity equations. in general. Therefore. in the present study. and curved beams as special cases. exteriors of rockets. we consider the theory of laminated composite shells. ship hulls. missiles. The first analysis that incorporated the bendingstretching coupling (owing to unsymmetric lamination in composites) is due to Ambartsumyan [I31. Pister. anisotropic. Ambartsumyan assumed that the individual orthotropic layers were oriented such that the principal axes of material symmetry coincided with the principal coordinates of the shell reference surface. Ambartsurnyan's work dealt with what is now known as laminated orthotropic shells.271. including pressure vessels. also known as the Sanders shell theory 126. better known as shells. and many other structures. rather than laminated anisotropic shells. can be found in Kraus 181. and the principal axes of material symmetry of the individual layers coincide with only one of the principal coordinates of the shell (the thickness normal coordinate). A number of theories exist for layered anisotropic shells [l251. the theory reduces to Donnell's equations. automobile tires. Other shell theories can be found in the works of Naghdi 124. Fliigge [6] and Kraus [8]. Thus. isotropic material. submarine hulls. wings and fuselages of airplanes. in laminated anisotropic shells. containers of liquids. In his analyses. We now extend the theory to curved plates and surfaces.1 Introduction In the preceding chapters. For a homogenous. we studied the theory and the analysis of flat plates.Theory and Analysis of Laminated Shells 8. Many of these theories were developed originally for thin shells and are based on the KirchhoffLove kinematic hypothesis that straight lines normal to the undeformed midsurface remain straight and normal to the middle surface after deformation. The firstorder shear deformation theory of shells. The theory of laminated shells includes the theories of ordinary shells. flat plates.
The theory accounts for transverse shear strains and the von KBrmAn (or Sanders) nonlinear strains. are derived in the next section. the application of such theories to layered anisotropic composite shells could lead t o 30% or more errors in deflections. 8.3. In Section 8.2) is called the surface metric tensor and a. known as the Love's firstapproximation theories (see Love [22]). and frequencies.2. on a) (no sum (8. & . The square of the distance ds between points (El. stresses. laminated curved shell. g! = 0 when the lines of principal curvature coincide with the : coordinate lines. Whitney and Sun [33] developed a shear deformation theory for laminated cylindrical shells that includes both transverse shear deformation and transverse normal strain as well as expansional strains. were considered by Gulati and Essenberg [31] and Zukas and Vinson [32]. The finite element models of shells will be considered in Chapters 9 and 10.2. < 2 . kinematic (straindisplacement) and constitutive (Hooke's law). (2) the dynamic excitations are within the lowfrequency range. 0) on the middle surface is denoted by r . ( a = 1. 0) is determined by . where <) denote the orthogonal curvilinear coordinates such that El and curves are the lines of curvature on the middle surface (< = 0).2) Note that g l .2 Governing Equations 8.1 Geometric Properties of the Shell Figure 8.All of the theories listed above are based on KirchhofiLove's hypotheses. For additional references and applications to composite shells. = 6 . in which the transverse shear deformation is neglected. doublycurved anisotropic shells. as well as thermal expansion through the thickness of cylindrical shells.p = 1.lb).la shows a uniform thickness.2) are a. These theories. the kinetic (equilibrium). and the position of an arbitrary point ([I. and (3) the material anisotropy is not severe.36]. The position vector of a point ( < I . c2 + + : where the vectors g l and g! are tangent to the El and <! coordinate lines. analytical solutions of the static equations for some cases will be discussed. see Bert [35.2. t2 d&. 5 ) is denoted by R (see Figure 8. Reddy [34] presented a generalization of the Sanders shell theory [26] to laminated. namely. &J0) and (J1 d& . However. The unit vector normal to the middle surface can be determined from . three basic set of equations. gap : (a. Following this introduction.2. are expected to yield sufficiently accurate results when (1) the radiustothickness ratio is large. The effects of transverse shear deformation and transverse isotropy.
1: Geometry of a doublycurved laminated shell. .2. (a) Shell geometry. Figure 8.Further. (no sum on a ) (theorem of Rodrigues) R. we have the WeingartenGauss relations  ga . (c) A differential element of the shell (dS1 and dSz denote the arc lengths). (b) Position vectors of points on the midsurface and above the midsurface.
. A. 2 . J2. general.2... hold: From Figure 8 . and A1. n .2.d 1 dal 6 ' 1 aa2 (Codazzi conditions) (8. G p Hence.2.2. and =" %a = (I+ &)&.10).1~). In view of the Codazzi conditions (8. dc) is given by <+ in which dR = G i d<l + G2 d& + n d c .5) The values of the principal radii of curvature of the middle surface are denoted by R1 and R2 (see Figure 8.2. 1 ~ elements of area of the cross sections are the . the square of the distance dS between points (J1.1~) Note that vector G. (no sum on a ) (El + dCI.C) and dt2. and A3 are the Lam6 coefficients (see Fig. is parallel to the vector g. (8. (8.2.4) we obtain G.lb) By differentiation we have and using Eq. In The position vector R of a point at a distance 5 from the middle surface can be expressed in terms of r and n by (see Figure 8. R1 and Rz are functions of t1 and C2.2.5) and Eq. and A. it can be shown that the following relations between the derivatives of a.E2 + Gap r G. 8.
n d < = d A C d <= A1A2 d<l d<2 d< (8.Middle surface F i g u r e 8. n = and an elemental area of the surface at dE1 d l 2 = a l a 2 d t l dC2 (8.2a) $Ao = d r l x d r 2 . (b) Area element on a surface at +<.13) C is given by (see Figure 8.2.THEORY AND ANALYSIS OF LAMINATED SHELLS 453 An elemental area of the middle surface (< = 0) is determined by (see Figure 8.2.2: Surface area elements of a doublycurved shell.2. .15) t 5 .2. (a) Area element on the rnidsurface.213) The volume of a differential element above the midsurface is given by d V = d R 1 x d R 2 .2.
= M I Figure 8.2 Kinetics of the Shell Next. The tensile force coordinate line on a cross section perpendicular measured per unit length along a coordinate line (see Figure 8. and Nll is the membrane force per unit length in El direction.2.2. acting on a surface perpendicular to the <lcoordinate (see Figure 8.1~) all dSz. .2.3): < < Similarly the moment of the force all dS2 about the laaxis is All resultants on these sides 6 are equal but with opposite signs to those on the parallel y e d g e s M. = h/2 and = h/2 denote the bottom and top surfaces of the shell.3: Stress resultants on a shell element. we introduce the stress resultants acting on a shell element.2.454 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS 8. The total tensile force on is to a the differential element in the direction can be computed by integrating over the entire thickness of the shell: c2 el where h is the total thickness of the shell..
5' ( a .2.THEORY A N D ANALYSIS OF LAMINATED SHELLS 455 Similarly.2.3 Kinematics of the Shell The linear normal and engineering shear strain components in an orthogonal curvilinear coordinate system are given by (no sum on repeated indices.b) and making use of conditions (8. a 2 = 022. NaO # Np. see [37]) where Substituting equation (8. one obtains . for a # . as = a13): Note that for shells. one can neglect (/R1 and </R2 and obtain Nap = No. in general. However./3 = 1.10) and (8. the remaining stress resultants per unit length (see Figure 8.2.2).20a. as in a plate theory..2. for shallow shells. and Map # Mp.2. The shear forces Qiare defined by where is the shear correction factor 8. a s = a l 2 . a 4 = 0 2 3 .3) can he defined as follows (al = a l l . arid Map = Mp.11).21) into (8.2.
. They are outlined below [6. The shell deflections are small and strains are infinitesimal.14]: 1. ES 0). and (41.9.In developing a moderately thick shell theory we make certain assumptions (as we did in the case of plates). 2. 3. The transverse normal is inextensible (i. Consistent with the assumptions of a moderately thick shell theory. 4. ) Substituting the displacement field into the straindisplacement relations (8. 2 . The transverse normal stress is neglible so that the plane stress assumtion can be invoked.8. 0) on the midsurface of vo.2. Normals to the reference surface of the shell before deformation remain straight but not necessarily normal after deformation (a relaxed KirchhoffLove hypothesis). we obtain where .e. the shell. we assume the following form of the displacement field: 1 in which (uo7 wo) are the displacements of a point (tl.22).4 ~ are the rotations of a normal t o the reference surface.
Then we have .23) can be used to derive the governing equations of the firstorder shear deformation theory of shells laminated of orthotropic layers by means of Hamilton's principle (or the dynamic version of the principle of virtual displacements). coordinate (and circle on the integral implies that it includes the total boundary of the shell.2.2. with rl.4 Equations of Motion The displacement field (8.THEORY A N D ANALYSIS OF LAMINATED SHELLS 457 8. To write the expressions for these virtual energies. We have where 6K denotes the virtual kinetic energy. and 6V the virtual potential energy due to the applied loads. let R denote the midsurface and r its boundary. being the boundary normal to the t. 6U the virtual strain energy.
642) of any derivatives with respect to (1. (Nap. p is the mass density. we substitute the expressions for SK.26).2. Q. 1 2 and t: . To derive the EulerLagrange equations (or equations of motions of the shell) from Eq. 6U and 6V.) are the stress resultants defined in Eqs.19). (8. (8. and then integrate the expressions by parts (or use the GreenGauss theorem) to relieve the varied generalized displacements (6uo.641.6vo.18) and (8.where q is the transverse load on the upper surface (( = h/2) of the shell.2.2. and Ii are the mass inertias A caret ('1 on the stress resultants indicate that they are specified quantities.6wo. Map.
2 .. 1 ~yields an additional relation among the ) twisting moments and surface shear forces: which must be accounted for in the formulation. a rigid body rotation gives a nonvanishing torsion except for flat plates and spherical shells). for a # 0. the equations of nlotiori and the natural (or force) boundary conditions are obtained by setting the coefficients of the varied 2 generalized displacements to zero in ! and on rl and r2: The natural boundary conditions are obvious from the boundary terms in Eq. Vanishing of the moments about the normal to the differential element (see Figure 8 . we have not assumed that NOB = NP.31).THEORY A N D ANALYSIS OF LAMINATED SHELLS 459 Noting. This amounts to modifying WE . otherwise. it will lead to inconsistency associated with rigid body rotations (i.e. To account for this discrepancy. and w.2. by the hypothesis of Hamilton's principle. transverse normal to the shell surface. In deriving the equations of motion. (8. that the virtual generalized displacenlerits are zero at t = 0 and t = T. we must add the term (see Sanders [28] and Budiansky and Sanders 1291) to the virtual strain energy functional SU. Here $ denotes the rotation about the . in Eq. and Map = M0.
2] ~ 2a1a2  1 (8.25) as follows: The rotation vector u 4.(8.2.2.39) in Hamilton's principle yields the following equations of motion: where .40)' one can show that Use of the modified straindisplacements relations (8.2.(uoa1).40) In view of Eq. is equal to the normal component of the curl of the displacement 4n (v x U ) . f ~ = [ ( ~ o G J ) . (8.2..
Q44) Q55 = Qij5cos2 Q + cos Q sin 6 ' Q44 sin2 0 The superscript k on Qij. in the shell coordinate system are given as (k. 0. Aij # Aji. vl2 and so on.55 sin2 0 + + + (Q12 .2. stacked on each other with the principal material 1 axis of the kth layer is oriented at an angle Or.2. the stress resultants defined in Eqs.e. (8. In equation (8. Hc denotes the intensity of the electric or magnetic field and eij are the electro.2. and Qii are the lamina stiffnesses referred to the principal material coordinates of the kth lamina + Q22sin4 0 (212 + + sin^ 0 + cos40) Q22 = Q11 sin4 Q + 2(Q12 + 2Qs6) sin2 0 cos2 I9 + Qz2c0s4 0 Q1i = Qii cos4 Q 2(Q12 2QG6) sin2 0 cos2 0 = (Qii Q22 . for a # 0. as well as on the engineering constants E l .8.Jk) = C.es2)sin Q cos 8 Using the lamina constitutive equations. However. whether structural layer or actuating/sensing layer.1 where Qi3 are the transformed stiffnesses. and Map # Mp.49a) Q45 = (Q55 . from the shell xl coordinate in the counterclockwise sense and . the laminate constitutive equations do not exhibit the symmetry among the stiffnesses (i.Q22 + 2Q66) sin 0 cos" 0 2Q12 .2Q66) sin2 8 cos2 0 + sin^ I9 + cos4 8) (8.48).18) and (8.. Bij # Bji and Dij # Dji) primarily due to Nap # Np.5 Laminate Constitutive Relations Suppose that the shell is composed of N orthotropic layers of uniform thickness.19) can be expressed in terms of the membrane strains E: and bending strains E:.4Q66) sin2 0 cos2 6'  + + 2Q66) sin 0 cosd Q Q2s = ( Q I ~ Qiz . .Q22 + 2Q66) sin" cos Q + (Q12 . E2.2. is omitted for brevity.or magnetostrictive material coefficients + e32 sin2 0 2 e32 = e32 cos 0 + e31 sin2 0 E3l = esl cos2 Q e3t3 = (e31 .2Qss) sin3 0 cos 0 Q16 = (Qi1 Q12  Qss = (QII Q22 Q44 = Q44 cos2 0 Q.2. The stressstrain relations of the kth lamina.
For thin shallow shells.25) and (8. we have and we have N I 2 = NZ1and M12 = M2.39)] and and the magnetostrictive stress resultants { N ~and { M M }are defined as ) . The laminate constitutive relations become where the laminate stiffness coefficients (Aij.3.2) (i.3 Theory of DoublyCurved Shells 8.2.D i j ) are defined by and the strains are given by [see Eqs. constant radii of curvatures)../3 = 1..8.p = 0 ( a .2. (8.1 Equations of Motion If we omit the term z / R in the definition of the stress resultants and assume that a. the equations can be simplified substantially [14].e. Bij..
4 2 ) by substituting for the force and moment resultants from equations (8. axl +A12 (  a2vo 8x18x2 + R2 ax1  .10): a2uo 1 dw AH (+L) ax: R.6)(8.la.b) into Eqs.vo.3..3.5a) and ml. denote the layer numbers of the magnetostrictive (or any actuatinglsensing) layers. k=nrl.m2. wo. 8. .2 Analytical Solution Analytical solutions of the equations (8.3.. m a . j = 1.10) can be obtained for simply supported crossply laminated shells [34]. (8.3.3. i = 3.10) in terms of displacements (uo. ~3 = <) (note that N12 = N21 = NG and &Il2 = A121 = M 6 ) are and p(" being the density of the kth layer and n is the number of layers in the laminate. x e$'  ck).3. Towards using the Navier type solution. Then equations of motion of the simplified theory in the Cartesian coordinate system ( x l . first we write the equations of motion (8.6)(8.3. x2.where Aij = ck.6)(8.3. 4*.3.2 (8.
+ B ~ ~ 8x1 I8x2 .ax.13) .12) + B12a a241 + B22 x 1 8x2 ax. + (x "UO.3. ) + .'"~) Rl 8x1 + ( A26 a . R~ % + R2ax1 ) + ( a2vo + ) a2uo .Io a2uo at2  11.3.L 1{ ~ l l a x* + ~ ) + ~ 1 2 ( ~ + ~ ) + ~ 1 6 ( ~8x2 * ) R ( 1 8x2 + 842 + B I 841+ B ~ ~ .RP 1 2 ( 2 + 2 ) {A 1 +A22 (z+z) dxldxz A~~$} + ~ 2 6 ( 2 . a2qh1 at =o (8.14) at2 ( 3ax.3. R2 ax1 z v 2"") ( a2uo + B26.2 ) + B12841 +&2ax1 d2wo +q11=o 842 a x1 +B26 avo auo (8.a x 1 ax2 a2vo a242 1 0 ~ at2 = O 11 d2vvo  (8.
.3. a242 + D66 . xa.+ D16.17) are satisfied by the following expansions of the generalized displacement field: $1 ( x l . t ) = x2 7 n=lm=l Ymn ( t ) sin ax1 cos Pxa .a2ax2 + D26* ax 41 ax.a 2 h axlax2 + [ The simply supported boundary conditions (SS1) for the firstorder shear deformation shell theory are The simply supported boundary conditions in (8. t ) = CC W O O X m n ( t ) cos ax1 sin Px2 $2 (xl.
12)(8.3.16) yields the 0 0 1 4 0 0 2 2 0 0 M25 0 3 3 0 0 0 M41 0 0 M44 0 .3. A32.3.0 5 2 0 0 M55~ where Sij = S j i . are defined in Eqs. (8. 5 ) are defined by (only nonzero and all other Cij = 0. B31 and B32 .J' elements are given) 1 + .3. Here the magnetostrictive coefficients A 3 1 .466 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS Substituting the expansions (8.18) into equations (8.5a7b). Cij and Mij = Mji (i.
. As an example consider simply supported crossply laminated spherical shell panels (R1 = R2 = R) under uniforrnly distributed load of intensity qo.3.3.1 contains nondinlensionalized center deflections... we set the time derivatives terms in Eq.25 (8.2E2..23) Table 8.3.. W...36] El = 25E2.19) to zero and solve the resulting equations for the amplitudes U.. The total solutiori is obtained by substituting the amplitudes into Eq...3..3... Figure 8. (8.Static Analysis For static analysis. and Y. Table 8.. G13 = G12 = 0. = lo3 w o ~ z h 3 / ( q o a 4 ) .18). X.3.3.. (8. The lamina material properties are assumed to be [15. V. G23 = 0.5E2.1: Geometry of a doublycurved shell panel.1: Nondimensionala center deflection versus radiustothickness ratio of spherical shells under uniformly distributed load (19term Navier solution). for various values of radiustoside ( R l a ) ratios and two values of sidetothickness ( a l h ) ratios (see Figure 8.1). ~ 1= 2 0.
w = w(a2/h) for crossply laminated spherical shell panels [14. we set all Cij = 0 and assume solution of the form urn.. (t) = V e .(t) in (8. versus radiustoside length ratio of spherical shells (a/b = 1. For nontrivial solution. the determinant of the matrix in the parenthesis is set to zero.2 contains nondimensionalized fundamental natural frequencies. . Table 8.3. Substitution of (8. The case R / a = lop3' corresponds t o a square plate.3..24) in (8.Natural Vibration For natural vibration without structural damping. 0 iwt . Table 8.e 0 (t) = iwi w0. and Ks = 516) .34].19).3. = Y. This gives values of w2. Y.2: Nondimensionalized fundamental frequenciesa (w =wa2Jm/h).eiwt.3. Vm. w.(t) = U~. e i ~ t Xmn(t) = x.eiwt..3. Results for three different laminates and two different thickness ( a l h = 10 and a l h = 100) are presented using a shear correction factor of Ks = 516. R1 = R2.19) yields which is an eigenvalue problem. Jn. .
v = 0.. GlassEpoxy (GlEp).(t). .j = 1 .39]. GraphiteEpoxy (GrEp).20) in the form = 0 and seek solution of the ordinary differential Substituting Eq. BoronEpoxy (BrEp)] are listed in Tables 8.1 23 (8.4.. the following boundary conditions are used: Numerical results are obtained for various lamination schemes to show the influence of the position of the pair of magnetostrictive layers from the neutral axis on the amplitude suppression time [38.26) into Eq. s. Magnetostrictive damping coefficients and natural frequencies for various materials and lamination schemes are also listed in Table 8. This equation gives five sets of eigenvalues.+ X ~ M .3. The eigenvalue with the lowest imaginary part corresponds to the transverse motion.28) for i .3.5 GPa.20).7. we assume q equations in (8.22)] Em = 26.3.. 5 .. we obtain...1. The eigenvalues can be written as X = a iwd. 3 . (8.0.3. C(t)r.l . Magnetostrictive material properties (for TerfenolD material) are taken to be [see Eq. 6 T 8 m ~ . = 9250kgmP3.4. dl. moment of inertia.Vibration Control For vibration control. p.3. 2 . (7. so that the damped motion is given by + In arriving at the last solution. magnetostrictive material constants) based on different lamination schemes and material properties [CFRP. the result where S.1 2.3. = l .3 and 8.g. 4 .+ XC.3. All values of the composite material and structural constants are tabulated and both damped and undamped frequencies are presented in the figures.7. = ZJ 2. for a nontrivial solution. . (8. = l o 4 The numerical values of various ~naterial and structural constants (e. W. The composite lamina material properties were given in Table 7.
3.5.3. .3. (8.3: Stiffnesses for different laminates arid materials. It can be observed that attenuation favors the higher modes.4 shows a plot of vibration amplitude for mode 5. where modes 1 and 2 are superposed and it is obvious that mode 2 attenuates at a significantly faster rate.3.29)] increases when the magnetostrictive layer is located farther away from the neutral axis.. This is clearly seen in Figure 8.2 shows the uncontrolled and controlled deflection amplitude at the center of the laminate.3. Present results also show that the vibration suppression time decreases very rapidly as mode number increases. Figure 8.3.3.4: Mass inertias and parameters a: and wd. indicating faster vibration suppression. Figures 8. [kg/ml I2(10~) [kg ml &31 a & wd.3 that [m/0/90/0/90]s 5 [m/(0/90)2]s has the maximum vibration suppression. GrEp [0/90/m/0/90]s BrEp [O/90/m/O/90]s Table 8.470 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS Table 8. for a tenlayered [0/90/m/0/90]s laminate. Material Laminate I. It is observed from Figure 8. (rad sl) GrEp [0/90/m/0/90]s BrEp [0/90/m/0/90]s The value of a: [see Eq.
0012 Time. 5 0..3. ..3: Controlled motion at the midpoint of the shell for different lamination schemes.0012 1 Time..) motion at the midpoint of the shell for the layup [m/0/90/0/90].0012 7 Controlled  0. t Figure 8.3.0. t Figure 8.2: Comparison of controlled () and uncontrolled (..
. .3.) motion at the midpoint of the shell for the layup [m/0/90/0/90]s. t Figure 8. mode n = l .4: Comparison of controlled () and uncontrolled (..mode n=2 Figure 8. 2 (layup [0/90/m/0/90]s).5: Comparison of original and controlled motion at the midpoint of the shell for modes n = 1.3..Time.
1 .4.10) with Go = 0.1 Equations of Motion The equations of motion of the firstorder shear deformation shell theory (FST) of a laminated circular cylindrical shell are (see [14. vo and wo are the displacement components along the X I . (i = 0 .41]. and 4 g are the rotation functions.3.40. p(k) is the material mass density of the Icth Figure 8.4.THEORY A N D ANALYSIS OF LAMINATED SHELLS 473 8. and I.1). uo. l/R1 = 0 and Rg = R): where R is the radius of the cylinder. a superposed dot indicates differentiation with respect to time t . 2 ) are the mass inertia terms defined as < N is the total number of layers and layer. (8. .6)(8.4. cf.1: Laminated shell geometry and coordinate system.3. N is the axial compressive load (positive in compression).4 Vibration and Buckling of CrossPly Laminated Circular Cylindrical Shells 8. x2 and = axes (see Figure 8.
1): (0) (8.4.3) and the subsequent results into equations (8.a3 A22 d L23 = (B12 + 2B66) . (8.4. R ax. wo.5) The equations of motion can be expressed in terms of generalized displacements $ 2 ) by substituting Eqs.a2 a2 8x2 a2 (B12 + 2B66) a3 ++I1. 4 2 = 8. woIT a2 + A66. = [Ll{A> Classical shell theory (CST): { A }= { u o . and the strains are defined as a w0 0 E~=+I+(8.4) 8x2 8x1 The equations of motion of the classical shell theory ( C S T ) are obtained from equations (8. a2 a2 at2 d4 +Dzza 3x2 + (N2%) $ at2 (t$ +7 :i2) .4.w0  ~ (8. (8.4.4) [and Eq.B2z7 L11 = All8x1 ti    axpx2  L33 = d4 Dl17 8x1 R 2 + 2(D12 + 2066) x d4 R 8x2 dx2at2 A22 B22 +2+IoI2 a2 n ax..1) by setting E:=~z+.+ 11.a2 10a2 L22 = A667 8x1 ax2 at2 a3 a3 + .5) in the case of C S T ] ( u o .vo.vo.. the stress resultants are given by Az6 = = B16= Bz6 = D16 = where K& and K& are the shear correction factors.8 A~~ a3 L13 = B11 8x1 axlax.474 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS For a general crossply laminate (i.4.e.6) where the coefficients of the linear operator L ( L i j = L j i ) and displacement vector { A }for FST and CST are given below. axlat2 a2 + A22.a2 Io@ L12 = (A12 f A66).4.4.4. a laminated shell with stiffnesses A16 = = O). $1 awe d =G w0 . into equations (8. aspx.
. (8.~ ) I8x2 (8. i = fl and &. = (R K ~ B22 ~ ~ A..4.).wo)T vo. = . L23 = L14  1  at . ax$ at.I at ~ . . w.. a at 2 a2 a2 L33 = KS5A55 + Dll6 ax.4. a2 a2 A22 3 L25 = R ax2 L34 = (Dl2 a + D66)  a2 L44 = ~ i 4 ~ 4 . = eiwmt.. when performing an eigenfrequency analysis (we keep in mind that there are denumerable infinite frequencies for each value of m ) . (8. Firstorder shell theory (FST): {A) = { u o . = I +BIXI a2 + A22a2 . we assume the following representation for the generalized displacement components: being the natural frequency corresponding to the mth where T.+ D 6 a 7 L I I = Ail7 8x1  L12 = (A12 a2 + A66)axlax2 a2 L14 = (B12 + B66)L15 =  A12 R a ax. a2 = ax.10. Since the solution technique presented for these equations is general.rn/R(rn = 0 . +1 0 ax2 I2 .+ ( N R  d2 5 ) 5ax:  a2 a2 ~&A44.THEORY A N D ANALYSIS OF LAMINATED SHELLS 475 Note that the longitudinal. For the circular cylindrical shell with arbitrary boundary conditions at X I = i L / 2 . we present only the equations of FST and include the numerical results of CST for the sake of comparison. = .4. at2 Lq:.8) a 8. .9) into Eq. mode.4.6) results . a2 + I 3 2 2 7 II?. a2 a2 ax. circumferential and rotary inertia terms are included. Substitution of Eq..2 Analytical Solution Procedure Herc we discuss the Lkvy type solution procedure. 41. 2 . 4 + a2 + D22 ax2 8x1 ~ 2 5 ~ a2  A22 L5. a2 a2 L24=&ax. .42. at a2 a2 a2 h3 B ax I ~ ax. 1 . a2 + A 6 6 7 Io?.
However.10). it is made sure that only one unknown variable with its highest derivative appears in equations (8. . .32). . it is more practical to introduce new unknown variables and replace the original system of equations by an equivalent system of firstorder equations (to be able t o use the statespace approach). . There exists a number of ways to solve a system of ordinary differential equations. .where a prime indicates a derivative with respect to X I . .10). we let w. when there are more than three governing equations. .10).2. + 0 in ej ( j = 1 . This will save the computational time required in the method that we will introduce for solving equations (8. (8.25) are given for free vibration analysis by where In the stability analysis. The coefficients Cj ( j = 1.4. We introduce the following variables: .4.4. 2 . . With some simple algebraic operations (addition and subtraction). as in Eq.
4. with ( I ) = ( ) ( 0 ) and {D} = + .4. (8. the columns of which consist of ten linearly independent solutions of equations (8. (8.4. a homogeneous system of algebraic equations can be found: .13).4. (8. 2 .4. Since equations (8.10) can be defined from Eq.17) such that are also the solutions of equations (8.4.17).19) Since [A] is a cor~staritmatrix. However. as will be seen later..14) (see [43.10) and q ( 0 ) is a nonsingular constant (known as the state transition matrix) matrix. a special fundamental matrix @(zl) for Eq.13) can be expressed in the form where and the coefficient matrix [A] is A formal solution of Eq. are in general complex. .4. The nonsingular fundamental matrix 8 ( x 1 ) is not unique. Some or all conlponents of this vector.4.' ( 0 ) { ~ ( 0 ) } (8.20) By imposing the ten boundary conditions at x l = fL/2 on the solution given by Eq.44])is given as where Q(. all fundamental matrices differ from each other by a multiplicative constant matrix. (8.4.4.cl) is a fundamental matrix.4.l0).THEORY A N D ANALYSIS OF LAMINATED SHELLS 477 for m = 1 .4. the state transition matrix is given by a matrix exponential fur~ction as = e[AIrl (8. .10) along with relations in (8.17) are the solutions of equations (8. In view of the definitions in (8.4.10) and {D) is an unknown constant vector. . Eqs. (8.
Eq. Hence. .4.4. (8. When the eigenvalues of the coefficient matrix [A] are distinct. Instead of imposing the boundary conditions on equations .4.25) were used in [40].4. (8. appearing in equation (8. A nonzero compressive (or tensile) edge load can also be included in the free vibration analysis.43].4.19) yields Equations (8. are identical. When the eigenvalues are repeated. dAIx1. which in general can be real and complex.20). Xj ( j = 1 . the fundamental matrix Q(xl) is given by @(xi) = lUI[Q(xi)l. (8. in a trial and error procedure. .23) where [Q(xl)] is another fundamental matrix.. We note that the eigenvalues of [A] are the same as the roots of the auxiliary equation of (8. (8. The axisymmetric buckling problem and axisymmetric eigenfrequency problem (when all inertia forces.g. are neglected) of a cylindrical shell are two examples where two of the eigenvalues. Numerous methods are available (e.19). due t o the occurrence of an illconditioned determinant I M I in Eq. except the radial inertia force. . In Eq.For a nontrivial solution of natural frequency or critical buckling load. defined as and [U] is a modal matrix that transforms [A] into a diagonal form (i. Nosier and Reddy [41] proposed the following approach. it is found that I M ( becomes ill conditioned when the ratio of the characteristic length of the structure to its thickness is near or larger than 20. However. (8. However. the j t h column of [U] constitutes the eigenvectors of [A] corresponding to the j t h eigenvalue of [A]).4. 2 . l o ) are the distinct eigenvalues of [A].18) and (8.4. as will be seen.e. see [43. regardless of any method used.4. (8.23) into Eqs. one can easily find the correct value of natural frequency (in a free vibration problem) or of critical buckling load (in a stability problem) which would make IMl = 0.20).22).. This is also the case in the Lkvytype eigenfrequency and stability problems of laminated plates and shell panels when shear deformation theories are used. the determinant of the coefficient matrix [MI must be set to zero Since the constant vector {Z(O)) is real.44]) for determining the matrix exponential. the determinant of [MI is also real. Substituting Eq.24) is no longer valid and a Jordan canonical form of [A] must be used [42. . These eigenvalues in most eigenfrequency and stability problems of plates and shells are distinct.4.4.
27) to obtain [K] U ] . after assuming a trial value for natural frequency (in free vibration analysis) or for buckling load (in stability analysis) for a particular m. . (8.4. = u = vo = 0 (8. rather than evaluating the determinant of ( [ K ][ u ] .27) to exist. For very thin shells (or long shells). to evaluate the determinants of [K] and [U]separately. 8. (8.29) is satisfied. we can subtract a nonzero constant number from the diagonal elements of [A] and compute the eigenvalues and eigenvectors of the new matrix. S4: W" = hf1 = 4.28a) to zero: In this way the determinant in equation (8. in general.l ) .17).4. impose the boundary conditions on equations (8.4. in this way. (8. (8.27) For a nontrivial solution of Eq.29) is never zero.17). MI = 4 .27) and check whether Eq.4. However.2813) as that is.30) 5 o . (8. it will have the same computational problem as IMI in Eq.28b) will always be a real number. which have a simpler form.4. A remark must be made concerning the computation of eigenvalues and eigenvectors of the coefficient matrix [A]. Since the diagonal elements of [A] are all zero. (8. S2: W" = MI = 42 = u g = N6 = 0 = N1 = ~o = 0. It should be remembered that IUj appearing in Eq. To resolve this problem. The key point in overcoming this difficulty is to rewrite Eq. it should be kept in mind that the determinant of [K] never becomes ill conditioned. [K] can be complex. The eigenvalues of [A] can then be obtained by adding the same number to each eigenvalue of the new matrix.4.18). it may be computationally more convenient to substitute Eq. since the eigenvectors in [U] are independent of each other.4. (8.~ { Z ( ~ = (0) [ )) (8.22).4. The eigenvectors of [A] will be identical to those of the new matrix (see page 52 of [43]).4. Since.4. (8.THEORY A N D ANALYSIS OF LAMINATED SHELLS 479 (l8.4.4.3 Boundary Conditions A combination of boundary conditions may be assumed to exist at the edges of the shell. we will impose the ten boundary conditions a t X I = fL/2 on Eq.4. the inverse of [U] is never needed. Here we classify these boundary conditions for the FST according to [41]: Simply supported S1: S3: W" = W" = = 4 = N1 = N6 = 0.4. In summary. computer overflow and underflow may occur when we evaluate the elements of the coefficient matrix [K] This problem is addressed in detail in 1411. derive Eq. This way we come up with a set of homogeneous algebraic equations of the form [Kim = (0). It should also be noted that. (8.4. However.4.4. the determinant of the generally complex coefficient matrix [K] must vanish. during the computation of eigenvalues and eigenvectors computer overflow or underflow may occur.26) into Eq. (8.28a) and set the coefficient matrix in equation (8.
Similar boundary conditions may also be classified for the CST (see [41]). the reader may consult 1411 8.33) ( u o . In the above discussion it was assumed that m # 0 (nonaxisymmetric case).awe ax 1 (8. . .4. 2 . 2 . . . (8. = (Uo. . and Eqs.4.4..34) are where o In the stability problem. . .32) The boundary type S3 is referred to as a shear diaphragm by Leissa [44].e. when m = 0) we have vo = 4 2 = 0.Wo)To(t) WO) and where To(t) = eiWot for free vibration and To(t) = 1 for stability problems.Clamped Free edge F : Nl = N4= MI =M4=Q1 N=0 . we will have For additional details and discussion.4.10) become (8.9) and (8.4. XO. . 9 ) appearing in Eq.4. (8. ~ . when only the radial inertia is included. The coefficients Cj ( j = 1 .13). In the vibration problem. we let w + 0 in ej ( j = 1 . For axisymmetric mode (i.4 Numerical Results Numerical results are presented here for an orthotropic mat'erial with the following : properties [42] .
. ~ 4 " ~ Table 8. N = 0 and G. Furthermore. unless the shell is extremely short.0155 (0) 6.2.4.8535 (6) 2. L I R = 1.7200 (5) 1. It should be noted that the effect of imposing various in Table 8. ( L 2 / 1 0 h ) ~ .and for an isotropic material with Poisson ratio v = 0.9745 (6) 2.. L I R = 1.8497 (6) 2. in all the numerical examples.2090 (6) 6. indicates that this exception for boundary type S4F does not occur.2659 (ti) 3.0368 (0) 2.1882 6.4076 (3) S3F 0. = 516 and the total thickness h of the shell is equal to 1 in.25.1: The effects of lamination and various boundary conditions on the dimensionless fundamental frequency w of a shell. The effect of altering the lamination scheme on the fundamental frequency of a crossply shell with various boundary conditions is shown in Table 8. Isotropic 1. except for the S4F case. = w . R l h = 60.7788 (6) S3C4 3. w.4071 (3) 0. It is assumed that = K& = K . an analysis based on a more accurate theory.0358 (6) 2.2: Comparison of the dimensionless fundamental frequency with dimensionless minimum axisymmetric frequency of a shell according = to FST: R l h = 60.4096 (3) 0. However... Laminate (0190) Theory FST CST FST CST FF 0. f i = 0 and a.0196 6. It is observed that.1 (a number in parentheses denotes the circumferential mode number m). known as the generalized layerwise shell theory [45] (also see Chapter 12).0291 (6) 3..4.4585 (3) 0.1657 6.2424 (6) 3.4542 (3) 0.7193 (5) 1.2762 (fj) 3.1657 6. the fundamental frequency for a (9010) laminated shell is slightly smaller than that of a (0190) laminated shell.4579 (3) 0. Note that in a (9010) laminated shell.7233 (5) S3S3 2.0155 (0) 6.4.0155 6.9805 (6) C4C4 3.1685 (0) . the minimum axisymrnetric frequency is always quite larger than the fundamental frequency of crossply and isotropic shells.1657 (0) 6.0368 (6) (0) (0) (0) 1.4098 (3) 0.4.1685 (5) (0) (0) (0) 2.7158 (5) 1. This is particularly true for crossply shells as can be seen from Table 8.4545 (3) C4F 1. the fibers of the outside layer are in the circumferential direction and those of the inside layer are along the longitudinal axis of the shell.0155 6.2508 (6) (90/0) The numerical results indicate that.(L2/l0h)J=.7747 (6) 2. where the results are tabulated for cases C l C 1 through C4C4.9928 (6) 6. all layers are assumed to be of equal thickness.1657 (0) 6.
plane boundary conditions is more severe for isotropic shells than crossply shells. two minimum axisymmetric frequencies are presented. Table 8. corresponds to the case when only the radial force is included.4.3: The effects of various simply supported conditions on the dimensionless critical buckling load N of crossply shells [N = ~ ~ ~ / ( 1 0 0 h % ~ an isotropic shell [N = f i ~ ~ / ( 1 0 h R / h)= . In Table 8.6512 (0) 1.5451 (4) 3. Also. as can be seen from the results of Table 8. Indeed. A22 = All and D22 = D l l . B22 = BII. and ) ] ~~ ] 40 and L / R = 2.4.5923 (0) CST (019010) FST CST Isotropic FST CST .7535 (1) 9.3. the effect of the coupling dies out rapidly as the number of layers is increased.4. for antisymmetric crossply shells we have BI2= Be6:= 0.8234 (4) 5. For additional discussion. Furthermore. Laminate Theory S1S1 S2S2 S3S3 S4S4 (9010) FST 1.5233 (0) 1. see [41]. the actual computations indicate that only for extremely short crossply shells the axisymmetric buckling load will be the actual critical load.5705 (4) 3. As in the frequency problem. the minimum axisymmetric buckling load in isotropic shells is relatively larger than the critical buckling load in crossply shells. it is seen that various inplane boundary conditions have more severe influence on the critical buckling load of isotropic shells. The influence of various simply supported boundary conditions on the critical buckling load of laminated and isotropic shells can be studied with the help of Table 8.4.2.5074 (0) 4. Note that we have not generated any numerical results for unsymmetric crossply shells.8062 (1) 9.8396 (5) 5.4. for antisymmetric crossply shells. The second number. It should be noted that the bendingextension coupling induced by the lamination asymmetry substantially decreases the buckling loads.7693 (0) 1. which is slightly larger than the first one. However.7739 (0) 4.
NASA Report T T F118 (1964). Theory of Anisotropzc Sh. Ser.... John Wiley. V. . . / I 0 0 layers) FF S3F C4F S 3 S3 C 4 C4 Problems 8. Lamination (9010) (0190) (90/0/90/00 (0/90/0/90) (90/0/90/0/90/0) (0/90/0/90/0/90) (90/0/. P.22).. S.15 (1953). Vlasov. New York (1967) 9. . V. "Calculation of Laminated Anisotropic Shells. Arnbartsumyan. . Theory of An.. Washington. H.2.2 Verify t h e straindisplacement relations in (8. A . Arnbartslimyan. Tekh. and Rz = R. Novozllilov. A.3 Show t h a t t h e equations of motion associated with a cylindrical shell of radius R are whcre 2 1 =2. Noordhoff.4. 2 2 = RH. (Translatiori of Obshchaya teoriya obolocheck i yeye przlozheniya v telchnike). C . New York (1956). SpringerVerlag. R1 = oc. Grijningeri (1959). 8.Table 8. L I R = 1 and N = N L ~ / ( ~ o ~ ~ E ~ ) . Kuhn. NASA T T F99. 8. McGrawHill. p. D. (1964). Est.iin Nauk Armenskoi S S R . S. References for Additional Reading 1. New York (1974) .'' Izvestiia Akurtern. 2. 3./I00 layers) (0/90/. . Than Elnstzc Shells. 4. Fliigge. Pergamon. W.4: The influences of lamination and boundary conditions on the dimensionless critical buckling load N of a shell according to FST: R l h = 80.lsotropic Shells. 7. Fiz. General Theory of Shells and Its Applications i n Engineering. English translation.. Natiorial Aeronautical and Space Adniinistratiorl. 6(3). Stresses in Aircmft and Shell Structules. NASATTF118 (1964).1 Verify t h e straindisplacement relations in (8. The Theory of Thin Shells. Z.. I. Kraus.24).C. V. Introd7~ctzonto the Theory of Shells..S. 6. 8.2. Nauk. Mat. Berlin (1960).ells.. A. A~rhartsurnyan. 5. Dym. Moscow (1961). Stresses i n Shells..
Heyman. 491546 (1888). "An Improvement of Donnell's Approximation of ThinWalled Circular Cylinders. 34. Morley. Niordson (Ed. L. The Netherlands (1975). Green. 19." Journal of Aerospace Sciences. L. 659 666 (1967)." Quarterly Journal of Mechanics and Applied Mathernatics. W. 169176 (1959). J . EM2. T. T. "A Theory of Laminated Cylindrical Shells Consisting of Layers of Orthotropic Laminae. . Reddy. L... C. 25. 18. and Wang. The Prager Anniversary Volume. F . 27. Hill (Eds. Series A." Progress i n Solid Mechanics. R. Copenhagen. McGrawHill. K. J. T. 1(4)." NASA TRR24 (1959). 0 ." Journal of Applied Mechanics. 30. E. 2141 (1970). New York (1984)." 2. 29. 9(9)." A I A A Jo. Widera. 10751090 (1980). 11. 13.. B. L.." Journal of Applied Mechanics. "Stability of Heterogeneous Aeolotropic Cylindrical Shells Under Combined Loading. "A Theory for NonHomogent:ous Anisotropic Cylindrical Shells. Logan. S." Applied Mechanics Reviews. G.. Naghdi. 365368 (1956). L. 17... "On the 'Best' First Order Linear Shell Theory.. 29. Elastostatics and Kinetics of Anisotropic and Heterogeneous ShellType Structures. S. B. "Stability of Thin Walled Tubes in Torsion. NorthHolland. Macmillan. K. 4 . 400407 (1971).. Amsterdam. E.. 24. Dong. O. N . L. Sneddon and R. and Tso.10.. 23." A I A A Journal. P. J. Boston. and Ho. Ser. Zukas. "Laminated Trarisversely Isotropic Cylindrical Shells. T. and Taylor. Leyden. 32. D.. arid Esscnberg. and Chung. A." Progress i n Applied Mechanics." Journal of Applied Mechanics. Physik. Timoshenko. Equilibrium of Shell Structures. 28. and Sanders.. 969975 (1962). Dong.898 (1963). MA (1982) 14. A. 16. Oxford University Press. Stavsky. 129140 (1963).. 1 (1963). The Netherlands. 26. Naghdi. Love. "A Survey of Recent Progress in the Theory of Elastic Shells. I . 21. F." Journal of Engineering Mechanics Division.. and WoinowskyKrieger. M. S.. D. 892. S.Theory of Thin Elastic Shells. Cheng. Pitman. H." Theory of Shells.. "On the Linear Theory of Thin Elastic Shells..uriml. S.. 93105 (1967). J. 89105 (1963). "Thermoelasticity of Heterogeneous Aeolotropic Plates. L. W.. W." Philosophical Transactions of the Royal Society (London). Donnell. A. N. S.. "On the Small Free Vibrations and Deformations of the Elastic Shells. pp.Angew Math. 8(12). 39.. Y. Librescu. "An Improved First Approximation Theory for Thin Shells. "Foundations of Elastic Shell Theory. M. J .. 31. A Survey of Recent Progress. 21. P. 266 (1962). "Effects of Anisotropy in Axisymmetric Cylindrical Shells. Gulati. E. S. (First Edition) John Wiley. "Refined Theories for Nonhomogeneous Anisotropic Cylindrical Shells: Part 11Application. Theory of Plates and Shells. 10911096 (1972). P. and Vinson." NASA Report (1933). P.3787399 (1970). New York (1959). B. UK (1977). R." Proceedings of the Royal Society. S.. I. S.). New York. A. J. Sanders Jr.. Budiansky. Energy and Variational Methods i n Applied Mechanics. S. 17. B. 20. N. 8. "On a Laminated Orthotropic Shell Theory Including Transverse Shear Deformation. Koiter. 12.). 106(EM6). IUTAM Symposium. P. "Foundations and Basic Equations of Shell Theory. Dikeman. E. 15." Journal of the Engzneering Mechanics Division. M. Pister. and Widera. "On the Theory of Laminated Anisotropic Shells arid Plates. Noordhoff. Hsu. J. 22.. M.
J . "Exact Solutions of Moderately Thick Laminated Shells. S. 1524 (1976). "Analysis of Shells. Part I." A I A A Journal. W. 553 558 (2002).. 28. NASA. Vibration and Buckling of CrossPly Circular Cylindrical Shells with Various Shell Theories. 207258 (1974)." . Washington. Gopal. K.Jor~rnal of Engirreeri7~gScience. "Asymptotic Theory for Larriiriated Piezoelectric Circular Cylindrical Shells. and Reddy. "A stndy of Bending. A." (corrected title). Cheng.. A. 8(10). "Vibration Control of Composite Shells Using Ernbedded Actuating Layers.). DC (1973) 45. 38.eddy. 157(1). Leissa. "A Refined Theory for Laminated Anisotropic. H. R. C. St& (1967). PrenticeHall. Krieger. 37. 35. 3748. NASA SP288. J .. 39." A I A A Journal. N. Khdeir. New York (1984). 7.: Reddy. 36. W. Ogata. A. Saada. Vol. Nosier. 40.33. C. J. 42. J. 7 of Composite Materials. .. Modern Control S y s t ~ m Theory. Vibration of Shells.. N. J . C. 1976. l l O ( 5 ) . 44. 41. A. 471476 (1974). N. and Sun.. Z. C. 8(11). Cliarnis (Ed. Krock (Eds. Pradhan.. and Frederick. C. &I. 27. 41.. T. Bert. and Reddy." Structural Deszgn and A ~ ~ a l y s i s . M. L. Barbero.Q. E. Li. Broutnian and R. W. J.) Academic Press. N. N. 794809 (1984). A. 13371351 (1989). Reddy. D." Journal of Sound and Vzbratzon. N.." (to appear). H.gineering Mechunics.. Cylindrical Shells.. Space Analyszs of Control Systems. 34." Journal of Applied Mechunzcs. New York. Second Edition. and Reddy.Parts I and 11. C. 139 159 (1992). 544 553 (1990). "Vibration and Stability Analyses of CrossPly Larriinatcd Circular Cyliridrical Shells.. S...Jurnal of En. Wiley Eastern.." Internatzonal . Eriglewood Cliffs. Boca Ratori. Bert. "General TwoDimensional Theory of Laminated Cylindrical Shells. FL (1993). 40(9).. "Dynamics of Composite and Sandwich Panels . Whitney. J. Elasticity: Theory and Applications. Shock & Vibration Digest. N J 43. J.
.
g. Lkvy. called finite elements. and hence they are also called interpolation functions. For a given differential equation. and eigenvalue problems arising in various fields of engineering. Therefore.. The finite element method is a powerful computational technique for the solution of differential and integral equations that arise in various fields of engineering and applied science. initial. the Navier. or nonlinearities are involved. Ritz) and weightedresidual (e.) methods [l51.. collocation. . The basic idea of the finite elernent method is to view a given domain as an assemblage of simple geometric shapes.e. Thus the finite elernent method is a piecewise (or element wise) application of the variational and weightedresidual methods. depending on the choice of a particular variational or weightedresidual method. etc.Linear Finite Element Analysis of Composite Plates and Shells 9. the reader is advised to consult R. one must resort to approximate methods of analysis that are capable of solving such problems. However. Since most realworld problems are defined on domains that are geometrically complex and may have different types of boundary coriditions on different portions of the boundary of the domain. For a detailed introduction to the finite element method.1 Introduction In Chapters 4 through 8. Galerkin.eferences 15. and variational (Ritz) solutions t o the equations of composite beams. The approximation functions are often constructed using ideas from interpolation theory. arbitrary boundary conditions. plates and shells were presented for simple geometries. it is possible to develop different finite element approxiinatioiis (or finite element rnodels). for which it is possible to systematically generate the approxirnation fuilctioris needed in the solution of differential equations by any of the variational arid weightedresidual methods. exact analytical or variational solutions to these problems cannot be developed when complex geometries. The method is a generalization of the classical variational (i. leastsquares. The ability to represent domains with irregular geometries hy a collection of finite elements makes the method a valuable practical tool for the solution of boundary. it is difficult to generate approximation functiorls required in the traditional variational methods.
In this chapter. Even t o develop proper input data to a computer program requires a good understanding of the underlying theory of the problem as well as the method on which the program is based.3.1. Assembly of finite elements to obtain the global system (i.2 Finite Element Models of the Classical Plate Theory (CLPT) 9. finite element models of Eqs. Solution of equations. The value of the theory and analytical solutions presented in the preceding chapters to gain insight into the behavior of simple laminated beam and plate structures is immense in the numerical modeling of complicated problems by the finite element method or any numerical method. 9. the method allows coupling of various physical problems because finite elements based on different physical problems can be easily generated in the same computer program.1. The above steps of the finite element method make it a modular technique that can be implemented on a computer. Postcomputation of solution and quantities of interest. we develop finite element models of the linear equations governing laminated composite plates and shells.The major steps in the finite element analysis of a typical problem are (see Reddy [IA) Discretization of the domain into a set of finite elements (mesh generation).1 Weak Forms In this section. Eqs. independent of the shape of the domain and boundary conditions.3) governing the motion of laminated plates according to the classical laminate theory are developed. Weightedintegral or weak formulation of the differential equation over a typical finite element (subdomain). Imposition of boundary conditions.3). For the sake of brevity. (6. Development of the finite element model of the problem using its weightedintegral or weak form.1. Those who are quick to use a computer rather than think about the problem to be analyzed may find it difficult to interpret or explain the computergenerated results. . The objective is to introduce the reader to the finite element formulations of laminated composite structures.e. it helps the reader in gaining an understanding of the plate and shell finite elements used in the analysis of practical problems. are used t o develop the weak forms. for this is primarily a textbook. The finite element model consists of a set of algebraic equations among the unknown parameters of the element. In addition.2. (6.25). (3. While the coverage is not exhaustive in terms of solving complicated problems.1)(6.1.. which are expressed in terms of the stress resultants but equivalent to Eqs. It is important to note that any numerical or computational method is a means to analyze a practical engineering problem and that analysis is not an end in itself but rather an aid to design. for the total problem) of algebraic equations.1)(6.
25) with Sue. Integration by parts to weaken the differentiability of uo..  2 d2~...FINITE ELEMENT ANALYSIS OF COMPOSITE PLATES A N D SHELLS 489 Multiplying three equations in (3. vo.4  a2~yy ($)I  d~dg av2 ax ax where N ~ . and Nyyare inplane edge forces. The stress and moment resultants N.vo.y ayax . Svo..N. are known in terms of the displacements ( u o ..6wo) take the role of weight functions in the development of weak forms. etc.40). and integrating over the element domain.3.. 6vo.. we obtain 3Nx. all I d2vo at2  I at O W [ a2Mz. and Swo. (3. Note that the virtual displacements (Sue. aN. + I 0 6 uat2p o 8~  dxdy . Mz..3. and w results in the expressions o Nz. wo) through Eq. respectively. . asfuo a2uo + N.
Second.vo) must be approximated using the Lagrange interpolation functions. Let . the primary variables uo.where (n. and awo/dy are the primary variables (or generalized displacements). whereas wo should be approximated using Hermite interpolation functions over an element Re. vo. the displacements (uo. 9. and dwolay must be carried as the nodal variables in order to enforce their interelement continuity.2 Spatial Approximat ions First.2) that uo. Integration by parts of the inertia terms in the last equation is necessitated by the symmetry considerations of the resulting weak form.2. and are the secondary degrees of freedom (or generalized forces). wo. wo.2. the polynomial expansion for wo should be a complete quadratic. We note from the boundary terms in Eq. (9. we note that the stress and moment resultants contain firstorder derivatives of (uo. Also. Thus.. vo. which leads to symmetric mass matrix in the finite element model. to satisfy the constant displacement (rigid body mode) and constant strain requirements. finite elements based on the classical plate theory require continuity of the transverse deflection and its normal derivative across element boundaries. dwo/ax. n y ) denote the direction cosines of the unit normal on the element boundary re. awo/dx. Thus.vo) and secondorder derivatives of wo with respect to the coordinates x and y.
y) used for the inplane displacements (uo.where (u:. The simplest Lagrange element in two dimensions is the triangular element with nodes at its vertices (see Figure 9.~.vo) at the j t h node of the Lagrange elements. element with linear variation of the dependent variables) can represent only a constant state of strains: (b) Figure 9.) Lagrange Interpolation Functions The Lagrange interpolation functions $$(x. complete. can be derived as described for the onedimensional functions (see Reddy vo) [l].2. kth node.1). and have nonzero first derivatives with respect to x and y. . and respectively. ($7.) denote the values of (uo. v. The linear triangular element (i.. Chapter 9). and its interpolation functions have the form The functions are linear in x and y. A i denote the values of wo and its derivatives with respect to x and y at the are the Lagrange and Hermite interpolation functions.e.2.1: Linear Lagrange triangular element and its interpolation functions.
For this reason the linear triangular element is known as the constant strain triangle ( C S T ) . The quadratic triangular element represents a state of linear strains: 6 2 ~ = ~ [sf(4 : i=l C 6 + dix + 2fig) + U. A triangular element with quadratic variation of the dependent variables requires six nodes.2: Quadratic Lagrange triangular element .2.2): where Liare the area coordinates defined within an element Figure 9.2. and the other three nodes are placed at the midpoints of the sides (see Figure 9.2. because a complete quadratic polynomial in two dimensions has six coefficients: The three vertex nodes uniquely describe the geometry of the element (as in the linear element).2.2). Li (see Figures 9.1 and 9. (bi + 2 % + dip)] ~ The interpolation functions for linear and quadratic triangular elements are presented below in terms of the area coordinates.
2.2.2.e. which define the geometry.3: Linear Lagrange rectangular element and its interpolation functions.. at least linear in one coordinate) Note that the shear strain is represented as a bilinear function of the coordinates. The interpolation functions for this element have the form $f (2.FINITE ELEMENT ANALYSIS OF COMPOSITE PLATES A N D SHELLS 493 The simplest rectangular element has four nodes at vertices (see Figure 9.y) = ai + biz + ciy + d i z y (9.8a) The strains in the linear rectangular element are partially linear (i.3). Figure 9. .
called the natural Figure 9. .4: Ninenode quadratic Lagrange rectangular element. The linear and quadratic Lagrange interpolation functions of rectangular elements are given below in terms of the element coordinates ( E .2.2. q ) .494 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS A rectangular element with a complete quadratic polynomial representation contains nine parameters and hence nine nodes (see Figure 9.4). In these elements the strains are represented at least as bilinear: and the shear strain is represented as a biquadratic function of the coordinates.
. Although the iriterpolatiori functions are not complete because the last term in Eq. dwo/8z. see References 627). dwo/8n. Cheung.2. (9. is not satisfied. There are two kinds of C1 plate bending elements. and a nonconforming element is one in which the continuity of the normal slope. as discussed in Reference 1. an alternative procedure must be employed.2. The interpolation functions for the linear triangular element can be expressed in terms Figure 9.g. and it consists of three degrees of freedom (wo.5). and Zienkiewicz [7]. dwo/8y. practical applications (serendzp~t~ Hermite Interpolation Functions There exists a vast literature on triangular and rectangular plate bending finite elements of isotropic or orthotropic plates based on the classical plate theory (e.9a) is omitted. The interpolation functions of the serendipity elements are not complete. The interpolation functions for the quadratic serendipity element are given in Eq. Irons. Here we discuss triangular and rectangular C1 plate bending elements.6). A conforming element is one in which the interelement continuity of uio. .FINITE ELEMENT ANALYSIS OF COMPOSITE PLATES A N D SHELLS 495 The sewndzpaty family of Lagrarige elements are those elements which have no interior nodes. Instead. An effective nonconforming triangular element (the BCIZ triangle) was developed by Bazeley. Serendipity elements have fewer nodes compared to the higherorder Lagrange elements. and they cannot be obtained using tensor products of onedinlerisional Lagrange interpolation functions.2.2.12) below (also see Figure 9. the serendipity elements have proven to be very effective in most !) . and 8wo/ay (or awo/dn) is satisfied. (9.5: Eightnode quadratic serendipity rectangular element. dwo/dx) at the three vertex nodes (see Figure 9.2.
3 ) where (E. as shown in the Figure 9. The element was developed by Melosh [18] and Zienkiewicz and Cheung [19]. by equating the variables from the vertices of each subtriangle at the common points and normal slope between the midside points of subtriangles. V) are the local coordinates. (xi. In each subtriangle.yj.2..8).7. yi) being the global coordinates of the i t h node. The normal slope varies cubically along an edge whereas there are only two values of awo/an available on the edge.5L1Lz L3. awo/ay) per node.2. A nonconforming rectangular element has wo.2. A conforming triangular element due to Clough and Tocher [22] is an assemblage of three triangles as shown in Figure 9. The thirty coefficients are reduced to nine.2. the transverse deflection is represented by the polynomial (i = 1 . awo/ax. xij = xi . ~ w o / a x . The interpolation functions for this element can be expressed compactly as . and yij = yi .Figure 9. and By as the nodal variables (see Figure 9.6: A nonconforming triangular element with three degrees of freedom (wo. The normal slope continuity is enforced a t the midside nodes between the subtriangles. the cubic polynomial for the normal derivative of wo is not the same on the edge common to two elements. 8.7. 2 . Therefore.x j . ~ w o /at y ) ~ each vertex of the triangle. three ( w o . of the area coordinates as where f = 0.
2. and (x. Figure 9.2.. of the center of the rectangle. a w o / a y ) per node.7: A conforming triangular element with three degrees of freedom.8: A nonconforming rectangular element with three degrees of freedom (wo. where 2a and 2b are the sides of the rectangle. a w o / a x . Y) are the global coordinates .FINITE ELEMENT ANALYSIS OF COMPOSITE PLATES AND SHELLS 497 Figure 9. .
and Schmidt [20]. awo/ay. and the nonconforming element the total number of degrees of freedom per element is 12.2.dwo/dx.9: A conforming rectangular element with four degrees of freedom (wo. whereas the nonconforming element has a total of five degrees of freedom per node.A conforming rectangular element with wo . For the conforming rectangular element (m = 4 and n = 12) the total number of nodal degrees of freedom per element is 24. Figure 9.9) are where In this book we will use the Lagrange linear rectangular element for inplane displacements and the conforming and nonconforming rectangular elements for bending deflections to present numerical results. The interpolation functions for this element (see Figure 9. awo/ax. and a2wo/axdy as the nodal variables was developed by Bogner.2. a2wo/axay) per node. . The combined conforming element has a total of six degrees of freedom per node. tIwo/ay. Fox.
. . =Lp (ZNT xY +N. FT~ a$. Swo p i ) into the weak forms.2. .T 3 = Le ax a$.FINITE ELEMENT ANALYSIS OF COMPOSITE PLATES AND SHELLS 499 9.2. m .4) for the displacements and the i t h interpolation function for the virtual displacement (Suo $i. . n. +N xx ay T . dzdy.3 Semidiscrete Finite Element Model Substituting approximations (9. 2 .) .2. mass matrix M : = FTa are defined as follows: MP The coefficients of the stiffness matrix (symmetric). .) ay dzdy az2 axay . aIIj pa ~' Kzj = Kji . Svo Gi. k = 1. . and force vectors F r and M zk~= ' 11$' ay dxdy LNT F. we obtain the i t h equation associated with each weak form    where i = 1 . . .
y)) to a node will add up from elements connected at the node and remain as a part of the force vector. Of course. ( .18.) ~ P and [ .2. In matrix notation..17) can be expressed as This completes the finite element model development of the classical laminate theory. q . Tke ~ = ~ neaEdq K a2yB ~ < dxdy (9.X [ K ' ~ ][~ K ~ [K33]~ ~ ] [Ol 1 1 [GI 0 ["l2IT "I' I' '" { Pe) } = {F2} {{F1}} P3) . Buckling In the case of buckling under applied inplane compressive and shear edge loads.2.19) reduces to 'Kl31] ['OI '01 '011) [ K ~ ~ ] ~ ~ ] [0] [O] [O] [ K .2. It should be noted that the contributions of the internal forces defined in Eq. (9. Eq.2.4 Fully Discret ized Finite Element Models Static Bending In the case of static bending under applied mechanical and thermal loads.2. They will remain in the force vector only when the element boundary coincides with the boundary of the domain being modeled (see Reddy [I].41)].17) or (9. 313318). (9.2. (9. (3. etc. (9. (9. and the generalized displacements are the primary nodal degrees of freedom. are the thermal force and moment resultants [see Eq. 9.where N&.3.3) to the force vector will cancel when element equations are assembled. M&. and a2vZ dxdy. q(x.2.y) and AT(x. The finite element model in Eq.19) reduces to where it is understood that all timederivative terms are zero. and p can be equal to x or y. Eq.2. Eq.e.19) is called a displacement jinite element model because it is based on equations of motion expressed in terms of the displacements. pp. the contributions of the applied loads (i.
(9..23) into Eq.23) where {A0) is the vector of amplitudes (independent of time) and w is the frequency of natural vibration of the system. { v ) = {vO)eiwt.28]for a more general equation than that in (9. (9. Consider matrix equation of the form . the response of the plate is assumed to be periodic {u) = {uo)eiwt. To fully discretize them (i.2.19) can be written symbolically as where [Ke] (which may contain [Gel) and [ M e ]are the stiffness and mass matrices appearing in Eq.19).2. (9. reduce them to algebraic equations). i = a (9. Eq. and Equation (9.2. Natural Vibration In the case of natural vibration.25) represents a set of ordinary differential equations in time. we must approximate the time derivatives.2. Here we discuss the Newmark time integration scheme [1.2.25). Substitution of Eq.2.2.19) yields Transient Analysis For transient analysis.2.where and all timederivative terms are zero. {A) = {nO)eiwt. (9.e.
Q{A}.2. the the the the the constantaverage acceleration method (stable) linear acceleration method (conditionally stable) central difference method (conditionally stable) Galerkin method (stable) backward difference method (stable) (9.2.~t {A). E. 0.s+I (9. The global displacement vector {A) is subject to the initial conditions In the Newmark method [28]. 7 a as=pit ' a = P  a8=6t(~1) (9. and [ K e ] stiffness matrix.33) Note that in Newmark's scheme the calculation of [K] and {F) requires knowledge of the initial conditions {A)o.31) where and ai . to a set of algebraic equations relating {A). and St is the time step. it can be calculated from the assembled system .2. For a = 0. In practice.5.+I = {A). the following values of y define various wellknown schemes: . and {A)o.2.As an approximation. 8 .a ) { ~ } . {A}S+Q= (1 .+. {A)o.where [Ce] denotes the damping matrix (due to structural damping and/or velocity the the proportional feedback control). .2.2. 5. We have [ ~ s + I { ~ ) s= I + {F)s. .+l to {A).2913) (9. i = 1 ' 2 . are defined as (y = 20) 1 as=1.+~ + + (9.2..27) can be reduced. 2. .29ac).30) The set of ordinary differential equations in (9. (9.29~) and a and y are parameters that determine the stability and accuracy of the scheme. one does not know { A ) ~ . .the function and its time derivatives are approximated according to {d). with the help of Eqs. [ M e ] mass matrix.
results y) in algebraically complex expressions. The transformation of the geometry and the variable coefficients of the differential equation from the problem coordinates (x. however. it can be represented by triangular and quadrilateral elements. numerical integration is used to evaluate such complicated expressions. and {F) (often {F) is assumed to be zero a t t = 0): At the end of each time step. defined in Eq. Numerical integration schemes. However. the fully discretized system is given by 9. and it is easier to evaluate integrals over rectangular geometries than over irregular geometries.5 Quadrilateral Elements and Numerical Integration Introduction An accurate representation of irregular domains (i.31) using initial conditions on {A). A coordinate transformation between the coordinates (x. jj) used to derive the interpolation functions of rectangular elements is introduced for this purpose. such as the GaussLegendre numerical .+l and acceleration vector are where al and a...2.y) to the local coordinates (z. (9.33). {A).of equations associated with (9. exact) evaluation of the integrals. it is useful to select one that is also convenient in the numerical evaluation of the integrals. a nonrectangular region cannot be represented using all rectangular elements..25). (9. While the element coordinate system. called global coordinates.2.?+I are computed using the equations {A}. Therefore.y) used in the formulation of the problem. and the element coordinates (33.2.e. the new velocity vector {&).2. Returning to Eq. For example. and they preclude analytical (i. also called a local coordinate system. Therefore. domains with curved boundaries) can be accomplished by the use of refined meshes and/or irregularly shaped elements. it is easy to derive the interpolation functions for a rectangular element. it is practical to use quadrilateral elements with straight or curved sides but have a means to generate interpolation functions and evaluate their integrals over the quadrilateral elements.e. can be any convenient system that permits easy construction of the interpolation functions.
2. Superparametric ( m > n): The polynomial degree of approximation used for the geometry is of higher order than that used for the dependent variable.2. requires l the integral to be expressed over a square region f of dimension 2 x 2 and the (J.27) and by (9. require the integral to be evaluated on a specific domain or with respect t o a specific coordinate system. The coordinates (J. y) in a typical element Re of the mesh to a point (E. different from used in the approximation of the geometry. q) of a given integral expression defined over a quadrilateral element Re t o one on the domain fi facilitates the use of GaussLegendre quadrature to evaluate integrals. < < Coordinate Transformations The transformation between Re and transformation of the form fl is accomplished by a coordinate y) while a typical dependent variable u(x.28). Isoparametric ( m = n): Equal degree of approximation is used for both geometry and dependent variables. the transformation between (x. Although the Lagrange interpolation of the geometry is i~nplied Eqs. (9. in general.27) maps a point (x. The element fl is called a master element (see Reddy [I]. one can also use Hermite interpolation of the geometry and/or the solution as required. we may use linear Lagrange interpolation of the geometry . y) and ([. For example. the finite element formulations are classified into three categories. Gauss quadrature. q) be such that 1 called normalixed or natural coordinates.2. Thus. q) in the master element fl. 421448).2. Depending on the relative degree of approximations used for the geometry and the dependent variable(s). Subparametric ( m < n): Higherorder approximation of the dependent variable is used. is approximated by where denote the interpolation functions of the master element fl and $$ are interpolation functions of a typical element Re over which u is approximated. The positivedefinite requirement of the Jacobian dictates admissible geometries of elements in a mesh (see Reddy [I]. for example. 3. q) 1.Chapter 9). The transformation (9.integration scheme.q) are coordinate system (J. in the finite element analysis of the EulerBernoulli beams. used for the approximation of the dependent The interpolation functions variable are. 4 7 $7 4 7 1.pp. and vice versa if the Jacobian of the transformation is positivedefinite.
and XI denotes the global coordinate of the I t h global node of the mesh.+l.2. = x.whereas the Hermite cubic interpolation is used to approximate the transverse deflection 4 WII(X) = j=1 A~P~(X(E)) (9. Master elements of different order interpolation define different transformations and hence different collections of finite element meshes. can be written in terms of the natural coordinate with the help of the linear coordinate transformation (9.2. N o transforrnation of the physical d o m a i n or.+l . Thus. and x. the coordinate transforrnations have the sole purpose of numerically evaluating the integrals inside the computer program.x. (9. The transforrnation (9. is the element length.40) Then we say that subparametric formulation is used for the transverse deflection wo. Note that the Lagrange and Hermite interpolation functions defined in Eqs.2.40). the solution i s involved in the finite element analysis.41a) can be expressed directly in terms of x and [: where h. respectively. In the Timoshenko beam element we can use the same degree of interpolation for both geometry and dependent variables.2. However. An example of the coordinate transformation in one dimension is provided by the linear transformation. . the transforrnations of a master element should be such that there exist no spurious gaps between elements and no elerrlent ovcrlaps occur. any given element of a mesh can be generated. once the approximations of geometry and solution are selected.2. which maps straight lines into straight lines where x: = x. The resulting algebraic equations of the finite element formulation are always arnorig the nodal values of the physical domain arid the nodal values are referred to the global coordinate system. = x.41): It should be noted that. Then we say that isoparametric formulation is used for the transverse deflection wo and rotation 4. Different elements of the finite element mesh can be generated from the same master element by assigning the global coordinates of the elements. with the help of an appropriate master element.39) and (9.. x: being the global coordinate of the ith node of the eth element.
37).g. (9. Note that the integrand contains not only $. (9. Negative nonzero values of 3 imply that a righthand coordinate system is transformed to a lefthand coordinate system. Here we discuss the Gauss quadrature.45).Numerical Integration: the Gauss Quadrature Recall that a finite element model is a system of algebraic equations among the nodal values of the primary variables (generalized displacements) and secondary variables (generalized forces).45) is called the Jacobian matrix of the transformation (9.46) and its determinant 3 is called the Jacobian.2.2. we obtain . y). we have < < which gives the relation between the derivatives of $ with respect to the global : and local coordinates. in general. but also their derivatives with respect to the global coordinates (x. known as numerical quadratures. which is the most widely used method for master elements of rectangular or prismatic geometries.(x.2. The integral expressions are.37) (9.2. y) can be easily expressed in terms of the local coordinates and q by means of the transformation in Eq. By the chain rule of partial differentiation..2. 1 q 5 1) so that the Gauss quadrature can be used. as was shown for onedimensional Lagrange and Hermite functions in Eqs. (9.43)."/aq using the transformation (9. which should be avoided.2. /l ! and d$. which must be greater than zero in order to invert Eq. material properties) and functions used for the approximation of the primary variables. The coefficient matrix in Eq.37). We illustrate the essential elements of the Gauss quadrature by considering the following representative integral expression We wish to transform the integral from Re to the master element fi = {(E. The coefficients of these algebraic equations contain integrals of the physical parameters (e. Inverting Eq.45). are used to evaluate them. complicated algebraically due to the spatial variation of the parameters or coordinate transformations. numerical integration methods. (9. (9. Therefore. y).2. The functions $f(x. q) : 1 5 { 1.2.2. We must first develop relations a$f/ax and a$z/ay to &$d .42) and (9.
functions (9. (9. The Jacobian can be be determined using the transformation (9.This requires the Jacobian matrix [J] nonsingular.2.2.47) where JT1 J22 =  J . J Jil= . from used in the approximation of the dependent variables. Re is transformed to d A = dxdy = .2. JT2=  Jl2 J . given the global coordinates (xj.37) in Eq.5013) Returning the integral in Eq.48). we can write it now in terms of the natural coordinates as where the element area d A = dxdy in element Jdrdq in the master element 6. Note that @ are different.46). (9.2. JG2 Jl 1 = . We have Thus.44). The Jacobian is given by 4 7 $5 We have from Eq. yj) of element nodes and the interpolation used for geometry. (9.J 2 l J (9. the Jacobian matrix can be evaluated using Eq. in general.2.2.
. N or M Points (1 or T ~ J Weights W I or W J The selection of the number of Gauss points required to evaluate the integrals accurately is based on the following rule: a polynomial of degree p is integrated exactly employing N = + I)]. which are the same as those for the onedimensional quadrature. the interpolation functions are of the same degree in both ( and 7. Table 9. ( t I . . see [29]. and therefore one has M = N. i(p + < .2.2.1 contains Gauss point locations and associated weights for N = 1 . For Gauss point locations and weights for N > 5. that is. the smallest integer greater than 1). When the integrand is of different degree in and 7. Table 9. we obtain where M and N denote the number of Gauss quadrature points in the I and 7 directions. . 2 . The minimum allowable quadrature rule is one that computes the mass of the element exactly when the density is constant. In most cases.1: Weights and points for the GaussLegendre quadrature in one coordinate direction.Using the Gauss quadrature formulas for integrals defined over a rectangular master element fi.5. q J ) denote the Gauss points. . and W I and W J denote the corresponding Gauss weights. the number of Gauss points is selected on the basis of the largestdegree polynomial in one of the coordinates.
their functional variations must be approximated by a suitable polynomial in order to determine the polynomial degree of the integrand. and c and in general may not be polynomials. = 3 ) 4 Cubic (7. quadratic. The N x N Gauss point locations are given by the tensor product of onedimensional Gauss points tI. Ji5 Table 9. the coefficients a.42). b. quadratic.2: Selection of the integration order and location of the Gauss points for linear.2.ion points and weights for each coordinate direction.2 contains information on the selection of the integration order and the location of the Gauss points for linear. Element Type Maximurn Polynomial Degree Order of Integration (7. Of course.2.2. The maximum degree of the polynomial refers to the degree of the highest polynomial in [ or 7 that is present in the integrand of the element matrices of the type in Eq. e . and cubic elements.1 for thc integrat. and cubic quadrilateral elements (nodes not shown). x r) Order of the Residual Location of Integration points? in Master Element Linear (7. Note that the polynomial degree of coefficients as well as J$ and J should be accounted for in determining the total polynomial degree of the integrand.= 2) 2 Quadratic (7.2. (9.Table 9. = 4) 6 t ~ e Table 9. In those cases.
if the solution exhibits biaxial symmetry. the membrane strains at any point (x. Eq. y. displacements and rotations. geometry.6 PostComputation of Stresses Once the generalized displacements at the nodes are determined.29a). and the results obtained with the two meshes would be identical.2. Note should be made of the fact that the finite element model developed herein is not restricted to any particular lamination scheme. (b) the material properties. For the case of small strains. Solution symmetries available in a problem should be taken advantage of to identify the computational domain because they reduce computational effort. Additional results are presented in Section 9. including boundary conditions. the values of any strain component computed from different elements connected at a node are different. the first derivatives of the inplane displacements and the second derivatives of the transverse deflection are.3. within the roundoff errors of the computation.3. z ) in a typical element Re can be computed from the equations Recall that only (uo. including nodes. not continuous across element interfaces.3. . dwo/ay) are continuous across element interfaces.vo) and (wo.(9. in general.angular finite elements to analyze laminated plates for bending and natural vibration.9. The boundary conditions along a line of symmetry should be correctly identified and imposed in the finite element model. When one is not sure of the solution symmetry.7 Numerical Results Here we use the conforming (C) and nonconforming (NC) rect. It was shown by Barlow [30.31]that stresses computed at the Gauss points associated with the Gauss rule used to evaluate the stiffness matrix of an element are the most accurate. A solution is symmetric about a line only if (a) the geometry. as given in Eq.2. 9.awo/tlx.3. and (c) the loading are symmetric about the line.5. (6. The stresses at any point in the plate can be computed from the constitutive equations of a lamina. it is advised that the whole plate be modeled. boundary conditions. or loading. the stresses are also discontinuous across element interfaces.10). The notation m x n mesh denotes m subdivisions along the xaxis and n subdivisions along the yaxis with the same type of elements. In particular. a 2 x 2 mesh in a quadrant of the plate is the same as 4 x 4 mesh in the total plate. Additional numerical results will be presented in Section 9.4) can be used to determine the strains using the straindisplacement relations (3.2. Since the strains are discontinuous. For example.
17b/32) for uniform meshes 2 x 2. and (15a/32.10).=O O . the results will be less accurate. I x=O I y=O uo=O uo=O uo=o ($ly=o ($ly=o O .o aw SS1 FSDT SS2 SS1 Uo=O uo=O uo=o $. and uyy are computed at (5a/8. and 8 x 8.2. as shown in Figure 9.e. the origin of the coordinate q system is taken at the center of the laminate (see Figure 9.o . 4 x 4.=O @.C. . a quadrant of the plate was used in the finite element analysis.5b/8).2.o aw CLPT SS2 ax ay u o = ~dx o u o = ~. The synmctry boundary conditions can be identified from the Navier solutions for each case.2. a quadrant of the plate may be used as the conlputational domain of the finite element analysis..76/16). The stresses in the finite elernent analysis were computed at the Gauss points nearest to the locations at which the stresses were evaluated analytically. For the norlcorlformirig and conforming rectangular plate elements used here. at the center of the element.3 shows a comparison of finite element solutions with the analytical solutions of simply supported orthotropic and twolayer crossply and angleply (45145) square laminates under a uniformly distributed transverse load.9b/16). The boundary conditions along the symmetry lines for the crossply and angleply laminates are different. Theory I B. a is computed for the same meshes at (3a/8. Table 9. i.15b/32).3b/8).2. When one is doubtful of the boundary conditions along the lines of symmetry.Bending Analyses For antisyrrmetric crossply and angleply rectangular laminates with their respective simply supported boundary conditions. Stresses a. and (17a/32. Otherwise. the strains and stresses are corriputed using the onepoint Gauss rule.10.O aw 3v Figure 9. (9a/16.10: Symmetry boundary conditions for antisymmetric crossply and angleply laminates. it is safe to use the full plate model. respectively. (7a/16.. In the case of conforming elernent. In all cases. it is necessary that the crossderivative i3wo/&& be also set to zero at the center of the plate when a quarterplate model is used.
7203 0.7709 0. b/2) and (0. It is clear from the results presented in Table 9.0421 CrossPly (0190) Plate (SS1) PU gxx   ~ Y Y  0 .0440 0.3935 1. and the rate of convergence of stresses is two orders less than that of displacements for the CLPTbased element. the parts add up at the Gauss . however.4264 AnglePly (45/45)4 Plate (SS2) t The stresses are computed a t the center of each finite element. vl2 = 0. CLPT solutions).0751 1.2.9436 0.Y 1. The slower convergence of stresses in twolayer angleply plates compared to the eightlayer laminate is due to the presence of bending stretchingcoupling. convergence of the displacements is always faster than stresses for the displacementbased finite elements.3279 0.3 that the convergence of the finite element results to the analytical solutions is very good. They are mutually exclusive at points (a/2.0337 0. E1/E2 = 25.  = .Table 9. they are expected to be different.3: A comparison of the maximum transverse deflections and stressed of simply supported square plates under uniformly distributed transverse load (h.0296 0. The conforming element (d2wo/dxdy = 0 at the center of the plate) yields slightly better solutions than the nonconforming element.0253 0.7148 0.25.0341 0. Variable 2x 2 Nonconforming 4x4 8x8 2x 2 Conforming 4x4 8x8 Analytical solution Orthotropic Plate (SS1) W c'xx  ~ Y Y c'x Y 0. G23 = 0.0524 0.@ . and both elements show good convergence. Since the stresses in the finite element analysis are computed at locations different from the analytical solutions.2E2.2600 0.2.0872 AnglePly (45145) Plate (SS2) w  a .6635 0. Recall that in angleply laminates the analytical solution for stresses is the sum of two parts: one is a double sine series and the other is double cosine series. However.O).7082 0.5E2. G12 = GIS = 0.1230 1.7937 0.1109 0. Qzy 1.. resulting in better agreement with the true solution. = h l n . but the Gauss point locations also get closer to the node point locations (but never become the nodal locations). Mesh refinement not only improves the accuracy of the solution.
2. = wo = 6.5E2. = 0) boundary conditions on twolayer crossply and angleply laminates is investigated using full plate models and 8 x 8 uniform mesh of conforming elements (no boundary condition on the crossderivative was imposed) and the results are presented in Table 9. = 0) and SS2 (u.8125a..(xo.3486 and a.2Ez. CrossPly SS1 and SS2 [a. = 8. aX.zo.4: Maximum transverse deflections and stressest of square laminates under uniformly distributed transverse load and for different boundary conditions (h.46875) and (O.0.4312 at the Gauss points (0.h/2)]: CrossPly Clamped [a.. = 0) and clarnped (u. Glz = GI3 = 0.2. G23= 0. h/2)] : aYY(0. 0. and any laminate for clarnped (CC) boundary conditions.(0.yo.(xo. E1/E2 = 25. (mo. Recall that crossply laminates admit the Navier solutions for the SS1 boundary conditions whereas antisymmetric angleply laminates admit for the SS2 boundary conditions. ~ 1 = 0.0. Navier or Lkvy type solutions) are not available for crossply laminates with SS2.O3125.. h/2) = o. Analytical solutions (i.0.4375a. h/2) Table 9.25). which shows better agreement between the analytical and finite element stress values. The locations of maxirrium stresses are indicated below for various cases. yo.. = 0.0. = us = wo = 8. the analytical results of stresses for the (45/45) laminate are a. h/2) .46875. 2 Variable (0/90) (45145) t The stresses are computed at the ceritcr of each fiuitc dement: 8 x 8 unsforrn rrlcsh of conforrriing elcrnents is used In the full plate.c.0.. The effect of the simply supported SS1 (us = wo = 0. (yo.. yo.03125).46875) and (0. h/2) = a.0625a. antisymmetric angleply laminates with SS1.8125a.points (0.03125). .03125. = h l n .4. For example. = 0.. respectively..46875.
25. Similarly. 12 TL Variable SS SC CC FF FS CrossPly Laminates (0/90/0. = 0). .2.2E2. Table 9. n = number of layers). . = h l n . . The clamped boundary conditions make both plates quite stiff compared to the simply supported boundary conditions. . 21 = 0. The stresses are computed at the Gauss points and the locations of the stresses are indicated in the footnote of the table.2.4) it is clear that SS2 boundary conditions make the crossply laminate stiffer because they restrain the bidirectional composite from having normal inplane displacements (u.5: A comparison of finite element (second row) and analytical (first row) solutions of antisymmetric crossply and angleply square plates subjected to sinusoidal distribution of transverse load and for various boundary conditions (h. The finite element solutions are obtained using 8 x 8 uniform mesh of conforming elements in the full plate.5. E1/E2= 25. G23 = 0. Glz = GI3 = 0. A comparison of finite element solutions with the analytical solutions of antisymmetric crossply and angleply laminates under sinusoidal loading is presented in Table 9.45.5E2.) (FEM only) .From the results (see Table 9. SS1 boundary conditions make the antisymmetric angleply stiffer by restraining the inplane tangential displacements (us = 0).2.) AnglePly Laminates (451451 .
2.2.i 5 = w(b2/h) of simply supported square plates (h. and 6g5y. We consider the linear equations of motion of FSDT from Eqs.6E2.6: A comparison of the natural frequencies.6.1. and integrating over the clement domain.1 Weak Forms Following the procedure described in Section 9.504 23.1) through (9.4. vl2 = 0. G12 = G I : ~ 0.13). we can develop the weak forms of the equations governing the firstorder shear deformation plate theory.vo. (9.Natural Vibration The finite element solutions are compared with the analytical solutions of antisymmetric crossply and angleply laminates in Table 9.25.1. G23 = 0.052 14. a@. a quarterplate model with appropriate symmetry boundary conditions was used in the finite element analysis.294 25. E1/E2 = 40.360 22.024 14. .888 14. = n1 Nonconforming 2x2 4x4 Conforming 2x 2 4x 4 Analytical solution CrossPly (0/90/0/. 6wo.439 23.304 25.699 t The rotary inertia is included.5E2.3 Finite Element Models of Shear Deformation Plate Theory (FSDT) 9. generalized displacements of FSDT are ( u o . wo. (5. = hln.13) are obtained by multiplying them with 6uo.4. 9. Table 9. respectively.. The weak forms of the five equations in (5.168 24.168 24.. We obtain .3.659 23.413 23.5).2. The dC.821 23.883 14. which are in terrns of the stress resultants but equivalent to Eqs. The finite element solutions show convergence to the analytical solutions with mesh refinements.1. . a / h = 10). In = Numher of layers in the laminate.6vo.) Plates (SS1) AnglePly (451451 2 4 8  451. d y ) .) Plates (SS2) 14. In all cases. .
wo. 4.vg. 4. only the first derivatives of the dependent variables (uo.3. the sets (uO. at the most.) are independent of wo. &). they can all be approximated using the Lagrange interpolation functions. Therefore. Let . vO). and (#. we use the same interpolation for all variables..) are the primary variables (or generalized displacements).. 4. the rotations (4. d y ) can be approximated with differing degrees of functions..vo. Unlike in the classical plate theory.We note from the boundary terms in Eq.2 Finite Element Model The weak forms of the firstorder theory contain. (9.. Note also that no derivatives of wo are in the list of the primary variables. In principle.3. The secondary variables are 9.lae) that (uo. 4.wg. wo. For simplicity.
11 [Ol 0 11 [Ol [Ol 0 11 11 [Ol 0 0 [GI [0] [0] PI PI PI [Ol [Ol [Ol  [K"] A " }+ [ M e { A " } { ] = {Fp) (9. quadratic. One can use linear. 2 . (uo. wo.3.5) for (uo. we obtain the semidiscrete finite element model of the firstorder theory:  [K"] [K12] [K13] [KI4] [K15] . (9. In general.vo). 5 ) by the expressions .[Ol [Ol [ K ' ~ ][ ~K ~ ~[ ]K ~ ~[ ]K ~ ~ ] ~ ~ ] [()I [ol [ K [ K " ~ ] ~ K ~ [ K] ~ ~ [ K S 4 ] [ K ~ + ] [0] [0] [ ~ ~ ] ~ ~ [ K [KI4lT [ K ~ [ ~ K] ~~ [ K] ~ ~~ ] ~ ~ ] [Ol PI [ ~ 1 5 ] T[ ~ 2 s ] T[K%]T [ ~ 4 5 ] T [K%] . or higherorder interpolations of these sets. 4. .3. . &) into the weak fornls in Substituting Eqs. (9.where $ are Lagrange interpolation functions. .3)(9.c. . wo. and (&.vo. &) : may be interpolated with different degree of interpolation.3./3 = 1 .I ) .3. Eq.6b) where the coefficients of the submatrices [ ~ " p ] and [ b f a p ] and vectors { F " ) are defined for ( a .
. and QyJ for a = 1 . . 2 . M. . 5 and I = 1 . 2 . .518 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS The coefficients NP.. 6 are given by .
2.l o ) . Figure 9.where A[&.3. q52. 4y). When the bilinear vo. the element stiffness matrices are of the order 20 x 20. (9. and for the ninenode quadratic element they are 45 x 45 (see Figure 9.11). (9.1).6) is often referred to in the finite element literature as the Mzndlzn plate element due to the fact that it is based on the socalled Mindlin plate theory.12) may be used to approximate the displacenlent field (wo. wo. Any of the Lagrange interpolation functions presented in Eqs. are the thermal force and moment resultants.3.2. The displacementbased C0 plate bending element of Eq. ~ 2 . which is labeled in this book as the firstorder shear deformation plate theory.2. . etc. rectangular element is used for all generalized displacements. and (9.1: Linear and quadratic Lagrange rectangular elements for the firstorder shear deformation theory. (9.3.
lg)]: 0= / (Mxx6~:! + MYy&$ + ~ ~ ~ 6 ~ 96~10)dxdy 1 .3. axay Po] dxdy Equation (9. However. numerical results for bending.6) can also be derived using the penalty function approach (see Reddy [32. consider the weak form of the classical plate theory without membrane strains.3. that the plate under consideration is orthotropic.9) Qo Next assume.44). i )  (9.10) is a statement of the principle of the minimum potential energy.Equation (9. as described in Eqs. buckling. The simplifications are obvious and therefore are not repeated for the FSDT element.24). (9.9) in terms of the generalized displacements wo 4 D G 6 dxay (a2 )  qwo] dxdy 6IIo(wo) where no denotes the total potential energy functional 46 D . For the sake of simplicity in discussion. 9.3. buckling.2. again for the sake of simplicity.2. nonlinearity. Using the plate constitutive equations (3. (9. natural vibration. and transient analyses.33])applied to the classical plate theory. and Oy such that .3. and inertia terms [see Eq.3. (3.3. natural vibration.3 Penalty Function Formulation and Shear Locking A finite element model equivalent to that in Eq.20)(9.6) can be simplified for static bending. Introduce the variables 0.3. which is a special case of the principle of virtual displacements when the material of plate is assumed to obey Hooke's law. we rewrite the weak form (9.3. and transient response will be discussed.
Q.) np  rI(w0. (9. multipliers can be shown to have the meaning of shear forces X1 The finite element model based on the functional IIL is called a m. 0. 8.) + ~ . 0. In the Lagrange multiplier method we assume that there exist Lagrange multipliers X I and X2 such that the constrained minimization problem is equivalent to GIIL(wo. [ h 1 ( ~ .. the principle of the minimum potential energy must be stated as one of minimizing the functional II(wo. (9. Thus. However. 0. SIIL = 0.3. N dxdxy (9. where HL . In the penalty function method.. O. the square of the error in each constraint is minimized along with the origirial functional.) +2  Lo (2 [y1 Oz)2+y2 (2 0. Therefore. because displacements are mixed with forces as element degrees of freedom.0 . in the penalty function method the constraints are .15) The weak form. e.3.8.Q. 0.3. The Lagrange Q. ) + h 2 ( $ . 8. The difference between IIo(wo) and II(wO..16) where yl and 7 2 are called the penalty parameters.~ ( w oo x . Oy) is that the latter contains at the most only the first derivatives of the dependent variables.13) does not include the fact that the dependent variables 0.14) can be solved either using the Lagrange multiplier method or the penalty function method..12): minimize subjected to the constraints The constrained minimization problem (9.. and Xz.ixed finite element model. ) ] . X2) = ex.) in Eq.)2] dxdy (93. and 0.O. Q y . and the penalty parameters represent weights with which the constraints are minimized relative to the original functional. the potential energy functional II(wo.0 .3..Then the potential energy functional takes the form O. X I . and Xg . (9.. X I . which are preselected positive functions of (x. In the penalty function method the constrained problem is posed as one of O. and therefore will require Cointerpolation in the finite element model. are related to wo by Eqs.12).y).3. minimizing the functional IIp(wO.) subjected to the (constraint) conditions in Eq.. can be used to construct a finite element model with Cointerpolation of all dependent unknowns: wo.
we rewrite the weak form (9. and assuming orthotropic material behavior.4. we recover the firstorder shear deformation theory from the penalty formulation of the classical plate theory. there are computational problems. (3.4.9): Using the plate constitutive equations (3.3. Before we embark on the discussion of shear locking.satisfied only approximately and the minimum character of the problem is penalized. The total potential energy functional for the firstorder theory.3.3. namely shear locking.21) and (3. A finite element model based on the penalty functional (9.3. arising from the finite element implementation of the model based on (9.4. the smaller the error in satisfying the constraint conditions.22). However. for a particular choice of the penalty parameters. At this juncture we should remind ourselves that the problem we are trying to formulate is the classical plate bending. can be derived from Eq. The larger the values of the penalty parameters.16) is expected to give an approximate solution to the classical plate theory for sufficiently large values of the penalty parameters.19).16). it is useful to note the similarity between the functional IIp and that of the firstorder shear deformation plate theory.3. omitting membrane effects and the von K&rman nonlinearity.16) and They are the same with the following correspondence n in (9. Thus.17) in terms of the generalized displacements which is the first variation of the functional \iVe note the similarity between the functionals IIp of (9. . A desirable aspect of the penalty function method is that no new variables are introduced in addition to those in the original functional and the constraint equations.
Thus. This type of behavior is known as shear locking. in order to obtain a nontrivial solution. and force components. Thus the penalty parameters are of the order h2. of the textbook by Reddy [l]for a discussion of modeling considerations). highly distorted elements tend to have slower rates of convergence but they give sufficiently accurate results. pp. This is due to the fact that Dij are proportional to h h h e r e a s Aij are proportional to h. use of the functional in (9. reduced integration of transverse shear stiffnesses (i. The Coplate bending elements based on the firstorder shear deformation plate theory are among the simplest available in the literature. mass coefficients. When a l h = 100. When thin plates are analyzed by the shear deformable elements.12) are satisfied accurately. the shear stiffness matrix must be singular. and Ass) is necessary. the penalty parameters Aij are lo4 times larger than Dij.3.e. terms containing A44. and As5) are evaluated using reduced integration. In this chapter only rectangular or quadrilateral elements based on the firstorder shear deformation theory are used. the elements do not accurately represent the bending behavior as the sidetothickness ratio of the element becomes large (i.3.for large values of the penalty parameters. Therefore. most commercial codes issue warning messages when the element is highly distorted (e. Ads. the classical plate theory is realized.. 439448. i. the classical plate theory is recovered. that contain All. Numerically this is equivalent to requiring the product of the shear stiffness matrix and the displacement vector be zero.. and E~~ are required to vanish. Of course. but sometimes a t the expense of rate of convergence. Shear locking is due to the inability of shear deformable elements to accurately model the bending within an element under a state of zero transverse shearing strain. Indeed.. There are a number of papers on the subject of shear locking and elements developed to alleviate the problem (see [32571).e. One way to achieve the singularity of the transverse shear stiffness matrix is to use an order of numerical integration lower than is necessary to evaluate the integrals exactly. > .. Unfortunately. in theory. They are expected. and hence the constraints (9. Equal interpolation of all generalized displacements is employed.g. Stiffness coefficients associated with the transverse shear deformation (i.. the shearing strains E. Higherorder elements or refined meshes of lowerorder elements experience relatively less locking.e. With the suggested Gauss rule.e.. see Chapter 9. one should avoid using highly distorted elements. yielding displacements that are too small compared to the true solution. For thin plates. thin plate limit). the energy due t o transverse shear strains must vanish.19) is more appropriate because it naturally gives rise to the classical plate theory as the plate thickness is reduced in relation t o the plate inplane dimensions. to give the thin plate theory solution when the sidetothickness ratio a l h is very large ( a / h 100). and the plate elements based on the firstorder theory become excessively stiff. all coefficients ~' in K . Als. and full integration is used for all other stiffness coefficients. when lowerorder (quadratic or less) equal interpolation of the transverse deflection and rotations is used.
along a boundary common t o two elements. z ) . where N x N is the exact Gauss quadrature rule used to evaluate the bending stiffness coefficients. 4x. We assume that there are no temperature effects. (9. to check for failures. the strains and stresses are the most accurate if they are computed at the ( N . reduced integration points)in the global coordinates using the constitutive relations If stresses and strains are required in the lamina principal material coordinates. Since the displacements in the finite element models are referred to the global coordinates (x. the strains and stresses of Eqs.3. Once the nodal values of generalized displacements (uo. Similarly.3.3.3. the linear rectangular plate bending element of the firstorder theory requires 2 x 2 integration to evaluate the bending stiffnesses exactly. (2.5)].3. for a quadratic rectangular element the reduced integration rule is the 2 x 2 Gauss rule. the strains can be transformed using Eq. strains. for example. Then the onepoint integration should be used t o evaluate the transverse shear stiffness coefficients.21) and (9. the stresses are computed a t the Barlow points (i. However.10). the strains are evaluated in each element by differentiating the displacement expansions [see Eqs.vo.3.4 PostComputation of Stresses Here we discuss the evaluation of stresses from the known displacement expansions. Here we give the equations for the computation of stresses in an element.3)(9.1) Gauss points. Since only the displacements and not their derivatives are continuous across the element boundaries in the C0 finite element formulations.14) and (2. That is.&) have been obtained by solving the assembled equations of a problem. strain continuity across the boundaries is not ensured.3. wo.e. For example. the strains and hence stresses take different values on the two sides of the interface.3. and stresses.9.14) .. As noted earlier. y.22) should be transformed to material coordinates associated with each layer using the transformation relations (2.1) x ( N . (9. Alternatively. strains and hence stresses are continuous within an element.
)2 ' 2 ' 2 b2qo 2 2 2 b2qo . and Pagano [59. y) = qo cos . and h/2 5 z 5 h/2. The following nondimensionalizations of the quantities are used: < < @zy = cJzy( a b h h2 a b h h 2 . it is equivalent to (0/90/90/0) laminate with equal thickness plies.sin or. cos ok 0 .3.sin Ok cos 6' k .60] developed the 3D elasticity solution for the problem. The laminate consists of three plies (0/90/0) of thicknesses h/4.26) where the origin of the coordinate system (x. vibration.cos a b (9. and h/4.3. y) is taken at the center of the plate. h/2. a12 5 x a/2.sin2 Ok (9. .)= oxy(. a12 5 y a/2.5 Bending Analysis First the effect of integration rule and the convergence characteristics of the Co finite element model based on equal interpolation is investigated using a simply supported (SS1) crossply square laminate under sinusoidally distributed transverse load [58]. For this problem. we have developed closedform solutions in Chapters 5 and 7. Also see [61771 for analytical solutions for bending. where h denotes the total laminate thickness.sinek 0 cosOk 0 0 cos2 Ok . and stability of shear deformation plate theories.3.2310) and then the lamina constitutive equations are used to compute the stresses: 0 cos2 Ok sin2 O~ 0 sin2 Ok cos2 Ok cosdk [R](~) = 0 0 0 sin& 0 0 _2~inO~cos8~ 2sinOkcosOk  o 9. The material properties used are those typical of graphiteepoxy material (Material 1) The transverse load in all cases is assumed to be (sinusoidal on the whole plate) nx Try q(z.
28) Table 9. the stresses in the finite element analysis are computed a t the reduced Gauss points. The Gauss point locations differ for each mesh used.A) in layers 1 and 3.2: Effect of reduced integration on the nondimensionalized maximum deflections G and stresses u of simply supported (SS1) crossply (0/90/90/0) square plates under sinusoidal load (see [58]).ed below: ax. The notation nL stands for n x n uniform mesh of linear rectangular elements.As noted earlier. Table 9. B) in layer 2 (9. Finite Element solutionst Analytical Solutions 10 CFS ELS (table is continued on the next page) .3. nQ8 for n x n uniform mesh of eightnode quadratic elements.1: The Gauss point locations at which the stresses are computed.3. The stresses in FEM are evaluated at the Gauss points as indicat. The finite element solutions (FES) are compared with the 3D elasticity solution (ELS) and the closedform solutions (CFS) in Table 9.2 for three sidetothickness ratios a l h = 10. irrespective of the Gauss rule used for the evaluation of the element stiffness coefficients.3.1. and 100.3.(B.(A.20. and nQ9 for n x n uniform mesh of ninenode quadratic elements in a quarter plate. The Gauss point coordinates A and B are shown in Table 9.3. ay.
$ The values of transverse shear stresses irl parentheses are obtained using the 3D equilibrium equations. S = selective integration. R = reduced integration. * The CLPT solution is independent of sidetothickness ratio.(table is continued from the previous page) Finite Element Solutions Analytical Solutions 20 CFS ELS Finite Elenlent Solutions Ar~alyticalSolutions 100 CFS ELS CLPT* t F = full integration. a / h .
3. ~ 1 = 0.K = 516). the error is relatively more for the fivelayer case compared to the threelayer case shown in Table 9. The element behaves uniformly well for thin and thick plates when the reduced (R) or selectively reduced integration (S) rule is used. Shear locking is evident when the element is used to model thin plates ( a l h 2 100) with full integration rule (F).3. G12 = G13 = 0. . 2 4 ELS CFS FEM 10 ELS CFS FEM 20 ELS CFS FEM 100 ELS CFS FEM CLPT t Stresses computed from 3D equilibrium equations. h2 = h4 = h/4) crossply (0/90/0/90/0) square laminates under sinusoidally distributed transverse load are compared in Table 9.3.3: Comparison of nondimensionalized maximum deflections and stresses of simply supported (SS1) fivelayer (0/90/0/90/0) square plates under sinusoidal loading (El = 25E2.25. The finite element results were obtained with 4 x 4 mesh of eightnode quadratic elements in a quarter plate are in excellent agreement with the closedform solutions. the stresses are in good agreement with those predicted by the 3D elasticity theory. Also. Nondimensionalized maximum deflections and stresses in fivelayer (hl = ha = h5 = h/6. G23 = 0. which is expected because the rate of convergence of gradients of the solution is one order less than the rate of convergence of the solution.3.An examination of the numerical results presented in Table 9.2 shows that the FSDT finite element with equal interpolation of all generalized displacements does not experience shear locking for thick plates even when full integration rule is used.2E2.5E2. Table 9. higherorder elements show less locking but with slower convergence. The displacements converge faster than stresses.2. Although the firstorder shear deformation theory underpredicts deflections for small values of a l h .3. The finite element results are in excellent agreement with the closedform solutions of the firstorder shear deformation theory.
27) are used. The same locations and nondimensionalizations as given in Eqs.3. The face sheets (i.3.5E2. the stresses predicted are in general agreement with the firstorder shear deformation theory and elasticity theory.e. Also.Table 9.2E2.26) and (9. with the material properties listed in Eq. the transverse shear stresses predicted through equilibrium equations.25. we consider a sandwich plate subjected to sinusoidally distributed transverse loading.2E2. Next.3. The layers are of equal thickness. K = 516).4 shows a comparison of the elasticity solution of Pagano [60]. E2 = lo6 psi.016 x lo6 psi ) . are very close to those predicted by the elasticity theory. Table 9. G12 = GIS = 0. ~ 1 = 0. 10 ELS CFS FEM 20 ELS CFS FEM 100 ELS CFS FEM CLPT t Values computed from equilibrium equations. El = 2532.25 (9. with the closedform and finite element solutions of a threelayer crossply (0/90/0) square plate under sinusoidally distributed transverse load. While the classical laminate plate theory underpredicts deflections for small values of a l h . (9.. The finite element results obtained with 4 x 4 mesh of eightnode quadratic elements in a quarter plate are in excellent agreement with the closedform solutions.4: Comparison of nondimensionalized maximum deflections and stresses of simply supported (SS1) threeply (0/90/0) square plates subjected to sinusoidal loading (hi = h/3. (9.5E2. GI3 = G23 = 0.25).06 x lo6 psi. ul2 G12 = = 0.3.29) and the core material is transversely isotropic and is characterized by the following material properties: El = E2 = lo6 psi. 2 GI2 = GI3 = 0.3. 242 = 0. layers 1 and 3) are assumed to be orthotropic with the following material properties: El = 25E2.3. G23 = 0. G23 = 0.25 El 2(1 + ~ 1 2 = 0. for the laminates studied so far.
hz = 0. is significantly overestimated by CLPT.5.3.8h.2 and 9. .5: Comparison of nondimensionalized maximum deflections and stresses in a simply supported (SS1) sandwich plate subjected to sinusoidally varying transverse load (hl = hs = O.3. K = 516).Each face sheet is assumed to be onetenth of the total thickness of the sandwich plate (a = b).3.10.3.3 show the variation of the transverse shear stresses through the thickness of the sandwich plates for sidetothickness ratios a l h = 2. Figures 9. The transverse shear stress component ay. The stresses are nondimensionalized as before.lh. and 100. The finite element results obtained with 4 x 4 mesh of eightnode quadratic elements with reduced integration (4Q8R) are compared with the closed form solution and elasticity solution of Pagano [60] in Table 9. Table 9. The equilibriumderived transverse shear stresses are surprisingly close to those predicted by the elasticity theory for a l h 2 10. while those computed from constitutive equations are considerably underestimated for small sidetothickness ratios. and their locations with respect to a coordinate system whose origin is a t the center of the plate are as follows: The results indicate that the effect of shear deformation on deflections is significant in sandwich plates even at large values of a l h . 4 ELS CFS FEM ELS CFS FEM ELS CFS FEM ELS CFS FEM CPT t Values computed from equilibrium equations.
3.0.3. % (a/2.4 Figure 9.025 0.3: Distribution of transverse shear stress nv.100 Stress.b/2.3 Stress.0 0.125 Figure 9. 0.z) .000 0. 0. & (O.050 0.z) 0.0.075 0.2 0.2: Distribution of transverse shear stress u. through the thickness of a simply supported (SS1) sandwich plate under sinusoidally distributed transverse load. through the thickness of a simply supported (SS1) sandwich plate under sinusoidally distributed transverse load. .1 0..
2654 0.5401 0. The figure also shows the effect of using incorrect boundary conditions along the lines of symmetry.0108 0.The same sandwich plate as discussed above is analyzed for simply supported and clamped boundary conditions when uniformly distributed load is used.2400  0.5347 0.0514 0.2785 0. the laminate is close to being a crossply laminate. The material properties of an individual layer are assumed to be (Material 2) The finite element results obtained with a 4 x 4 mesh of ninenode elements in a quadrant are identical to the closedform solutions (CFS) for all angles and sidetothickness ratios.0030 0. .2406 0.5018 0.3. number of layers.0916 0.5356 0.0550 0.5978 1. Once again a quarter plate model is used with 4 x 4 mesh of quadratic FSDT elements and 8 x 8 mesh of CLPT conforming cubic elements.0906 0.5830 0.0120 0.3359 1.0605 0.3671 1. when the correct symmetry boundary conditions (SS1) are used.0526 0.7 shows maximum nondimensionalized deflections for angleply (018/8/8. Table 9. the symmetry conditions of SS1 (see Figure 9.~ 4Q9R 4Q9R ~CCFS t 1.3.3.2394  0.1160 0. Figure 9.1148 .2433 0.0906 0. .10) were used to obtain solutions of the two.8h.1445 0. Simply supported plate under uniformly distributed load 10 50 100 CLPT 4Q8R 4Q8R 4Q8R ~CCFS 2. and the lamination angle on the nondimensionalized maximum deflection of simply supported (SS1) antisymmetric angleply plates. h2 = 0..and sixteenlayer laminates.0881 0. The effect of shear deformation on the deflections is even more significant in clamped plates than in simply supported plates.0094 0.1136 0.3.4 shows the effect of sidetothickness ratio.2396 0.6: Nondimensionalized maximum deflections and stresses in a square sandwich plate with simply supported (SS1) and clamped boundary conditions (hl = ha = O. K = 516). The results are presented in Table 9.0039 0. and 45". t The 4Q9S element gives the same results as 4Q9R.3111 0.0145 0.6.2.0880  Clamped plate under uniformly distributed load 10 50 100 CLPT 4 ~ 9 .0509 0. Table 9. $ 8 x 8 mesh of conforming cubic elements with full integration for stiffness coefficient evaluation and onepoint Gauss rule for stresses.5964 1. the laminate behaves like an orthotropic plate.0991 0.lh.5430 1. for which the SS1 boundary conditions are valid.2951 0. . the error due to the incorrect symmetry boundary conditions is small.2318 0. It is clear from the results that when the lamination angle is small or the number of layers is large.0883 0.) square plates under sinusoidal load for I9 = 5".3296 1. and when the number of layers is large (n 2 8). This is expected because for very small lamination angle.3370 1. 30°.
12 0.ri ClosedForm Solution (CFS) E ! 0~~ 0.3.06 0.04 0.4: Effect of transverse shear deformation. and sidetothickness ratio for simply supported (SS2) angleply square plates under sinusoidally distributed transverse load (h. ' % Q g 5 10152083035404550 Sidetothickness ratio.5E2.25. = h l n .FINITE ELEMENT ANALYSIS OF COMPOSITE PLATES AND SHELLS 533 Table 9.14 0. . w x lo2.02 0. 12 a/h Source 8 = 50 8 = 30" Q = 45" 2 4 CFS FEhl CFS FEM CFS FEM CFS FEM CFS FEM CFS FEM 10 20 50 100 CLPT 0. number of layers. lamination angle.0=45") 1 3 .3.10 FEM* (n=4.7: Nondimensionalized deflections. alh Figure 9. K = 516. El = 40E2. angle. and symmetry boundary conditions on the deflections of simply supported (SS1) antisymmetric angleply laminates under sinusoidal load. G12 = G13 = 0. 21 = 0. n = Number of layers in the laminate). as a function of number of layers.00 0 ' : Finite Element Model with the symmetry boundary cond~tions ~mplied CFS by c .GE2. G2y = 0.
ys. ~ 1 = 0. Next.0233 0. 2 CLPT 0. Figure 9. To = O. the effect of shear deformation on thermal deflections and stresses is not as significant as in mechanically loaded plates.2411 0.1440 0. uxy = ~ x y ( x . 4Q9R.534 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS The stresses obtained with the same mesh as above for tenlayer antisymmetric angleply laminates are presented in Table 9. 2 b2qo 2 b2qo The finite element stresses (second row) are in excellent agreement with the closedform solution (first row). h h2 h h2 g y y ) ( x c .5 shows the effect of sidetothickness ratio a l h on the nondimensionalized deflections and stresses of crossply and angleply plates subjected to temperature distribution that is linear through the thickness and varies sinusoidally in the plane of the plate (q = 0.1206 0. ~)c. GZy= 0.3.0873 0. Clearly. = h l n . )s. For t9 = 45" it is found that the stresses are independent of sidetothickness ratio. .8: Nondimensionalized stresses for simply supported (SS2) angleply (O/Q/O/.1402 t The stresses are independent of sidetothickness ratio.@yy) = (gxx.) square plates under sinusoidal transverse load (h.0207 0.3. The deflection and stresses are amplified to show the effect of thicknessshear strain.62]).25.8. results for thermal bending are presented (see [61. n = number of layers). n = 10 El = 40E2. GI:! = GI3 = 0. . . .3. K = 516.6E2.5E2. The stresses in the finite element analysis as well as in the closedform solution are evaluated at the Gauss points: (@xx.Tl # 0). Table 9.5290 0.
12 '3 0. O O ~ ~ l l l l ~ l l . bla=l 2 0 Ei' Q. .10 0.FINITE ELEMENT ANALYSIS OF COMPOSITE PLATES A N D SHELLS 535 (0190190/0). 3 n 0 5 (451451451451.06 0.3. alh (b) Stresses at the Gauss points F i g u r e 9. Material 2 q. bla=2   101520253035404550 Sidetothickness ratio.04 (0190190/0).= 0 .5: Effect of shear deformation on the nondimensionalized (a) deflections and (b) stresses of simply supported antisymmetric crossply and angleply laminates subjected to thermal loading. Material 2 0.08 0. alh (a) Center deflections o . ~ l l l l l l l l l l l 0 5 10 15 20 25 30 35 40 45 50 Sidetothickness ratio. Material 2 qo=O.
b.3.3. refers to simply supported boundary condition on the edge x = .3. CC. and the material properties of each lamina are assumed to be those of graphiteepoxy material with the properties listed in Eq. Theory FSDT CLPT SS SC CC FF FS FC h 5 1 10 FSDT CLPT 5 FSDT CLPT FSDT 10 CLPT t The second row corresponds to finite element results. a 4 x 4 mesh is used for the classical plate theory. twolayer and tenlayer crossply square plates (0/90)k.a / 2 and clamped boundary condition on the edge x = a/2. while the remaining two edges.5.9: Nondimensionalized center deflections k a  a x 10' of antisymmetric. In all cases a sinusoidally varying transverse load is used.11 contain numerical values of deflections and stresses of antisymmetric crossply square plates (0/90/0/ . while 2 x 2 mesh of ninenode quadratic elements is used for the firstorder theory when a quadrant of a plate is analyzed (for SS.) with various boundary conditions and sinusoidally distributed transverse load. For the finite element analysis. y = 0. numerical results for transverse deflections and stresses of rectangular plates for a variety of boundary conditions are presented. The analytical results reported were obtained using the Lkvy method with statespace approach (see Sections 7.9 through 9.4 and 7. and equivalent 8 x 4 and 4 x 2 meshes were used for the halfplate models (in the case of SC. for example. . and F F boundary conditions). The notation SC.25). In all cases the finite element results are in good agreement with the analytical solutions. The analytical (first row) and finite element solutions (second row) are presented. Table 9. The deflections and stresses are nondimensionalized as follows: Tables 9. . FS. and FC boundary conditions).Lastly. are simply supported. and References 6376). . (9.3.
6E2. ..10: Nondimensionalized stress (a.) of antisymmetric crossply square plates (0/90/0/.3.12 contains nondimensionalized deflections of angleply laminates subjected to uniformly distributed transverse load under various boundary conditions and with different values of E1/E2 (G12 = GI3 = 0.5.3.25). .). The results were obtained using the Lkvy method with statespace approach discussed in Section 7.3.) square plates. n denotes the total number of layers. .5E2. .FINITE ELEMENT ANALYSIS OF COMPOSITE PLATES A N D SHELLS 537 Table 9. . n denotes the total number of layers n a FL Theory SS SC CC FF FS FC 5 FSDT CLPT 2 10 FSDT CLPT 5 FSDT CLPT 10 10 FSDT CLPT Table 9. Gpg = 0. ul2 = 0.11: Nondimensionalized stress (ayy) of antisymmetric crossply (0/90/0/. 7 L  t~ Theory FSDT CLPT SS SC CC FF FS FC 5 2 10 FSDT CLPT 5 FSDT CLPT 10 10 FSDT CLPT Table 9.
12: Nondimensionalized deflections w lo2/ (a4qo) of simply supported (SS2).3. = wo(0. K = 516. and type of element on the nondimensionalized fundamental frequency = w ( a 2 / h ) J n of simply supported (SS1) crossply (0/90/90/0)square plates.7 shows the nondimensionalized center deflections obtained by the CLPT and FSDT ( a l h = 10) as functions of the lamination angle for simply supported (SS2) symmetric threelayer (0/0/0) plates under uniformly distributed load.13: Effects of sidetothickness ratio. symmetric laminates under uniformly distributed transverse load (El = 25E2. fourlayer antisymmetric angleply square plates [(45/45145145). a l h = 10). ul2 = 0. the midplane symmetric plates are characterized by nonzero bendingtwisting coupling coefficients D16 and Dz6.5E2.3. a/h Serendipity Element Lagrange Element CFS . The plots are symmetric about 45". square. 4Q9R).3. Also.6 contains plots of the nondimensionalized center deflection versus sidetothickness ratio for threelayer (45145145) simply supported (SS1) and clamped. G12 = G13 = 0.2E2. The bendingtwisting coupling has the effect of increasing deflections.Table 9. Figure 9. The finite element results are obtained using 4 x 4 mesh of ninenode elements (i. a / h = 101. The solution obtained with omitting DI6 and D26 is also shown in the figure..25.6. Figure 9. Table 9. GZ3= 0.e. integration.3. the effect of shear deformation is more in the clamped plate than in the simply supported plate. b/2) E~ x h3 FF FS Theory FSDT CLPT FSDT CLPT FSDT CLPT FSDT CLPT 2 2 SS SC CC FC 10 20 30 As noted in Section 6.
_ .7: Nondimensionalized deflection versus lamination angle for simply supported (SS1) symmetric threelayer (O/O/0) plates under uniformly distributed load.. 8 Figure 9.= . FSDT (alh=lO) Lamination angle..3.3... . simply supported (SS1) and clamped symmetric plates under uniform loading..6: Nondimensionalized deflection versus sidetothickness ratio for threelayer (45145145)._ 0) A 0 10 20 30 40 50 60 70 80 90 100 alh Figure 9.CFS (with D16_ D26 = .004 ...F I N I T E E L E M E N T ANALYSIS O F C O M P O S I T E PLATES AND SHELLS 539 0.
From the results obtained.2.2E2. of crossply plates ( a l h = 10. E1/E2 = 25. However. While the first two fundamental frequencies are very close.13./p/Ez.25.3. 3 = w(a2/h)JphlEz. Table 9. Similar results are presented for angleply plates in Figure 9. Gas = 0. the reduced integration and selective integration rules both give good results for a wide range of sidetothickness ratios. Material 2.3.14 contains the lowest six natural frequencies of two. 2Q8R in full plate). whereas reduced integration (R) gives the best results for thin plates ( a l h 2 100). rotary inertia included) are presented in Table 9.3.and fourlayer crossply and angleply laminates with clamped edges.813 corresponds to the case in which the SS1 (incorrect) symmetry boundary conditions were used.3.14: The lowest six nondimensionalized frequencies of crossply and angleply square plates with clamped boundary conditions (a = w(a2/h). . Laminate alh "J1 "Jz "J3 "J4 " J 5 W6 Occur in pairs Figure 9. and type of element on the nondimensionalized fundamental frequency L2 of simply supported (SS1) crossply (0/90/90/0) square plates (Material 2.3.6 Vibration Analysis The effect of reduced integration and the use of eightnode and ninenode elements on the accuracy of the natural frequencies are studied using threelayer crossply (0/90/0) laminate used to obtain the results in Table 9.5E2. "12 = 0. it is clear that both full (F) and selective (S) integrations give good results for thick plates ( a l h 5 l o ) .3. The dashed line in Figure 9. the higher frequencies are quite different for these laminates.77]. hi = h l n . a / h = 10.3. A 2 x 2 mesh in a quarter plate is used to obtain the results [58. integration.8a shows the effect of sidetothickness ratio on the nondimensionalized fundamental frequencies. K = 516). Table 9. which also contains a plot of the fundamental natural frequency versus the lamination angle for fourlayer antisymmetric angleply laminates.813. Gla = G13 = 0.3.9. Effects of sidetothickness ratio.
(451451... (0/90/0). bla=3 0 W 0 10 20 alh 30 40 50 (b) Angleply plates Figure 9.3. . (451451. bla=3 Material 2 . CFS FEM.Material 1 0 CFS FEM. ( 4 5 / 4 5 1 4 ...8: Nondimensionalized fundamental frequency versus sidetothickness ratio of simply supported antisymmetric (a) crossply (SS1) and (b) angleply (SS2) plates. . alb=l FEM.......... (0190/9010)...alb=l FEM....alb=l FEM.. (0190/90/0). (0/9010). CFS FEM.. . (4514514. alb=l 0 F E M .bla=3 W 0 10 20 alh 30 40 50 ( a ) Crossply plates Material 1 0 CFS FEM..bla=3 Material 2 ....
( a b h h2 . The finite element results are compared with the closedform solutions (CFS) developed in Section 6. The boundary conditions for a quadrant are (see Figure 9.7 ).9 through 9. The following geometric and material properties were used: a = b = 25 cm. ayy.12 show plots of center deflection 6 and maximum and of stresses a.3.3.3..15 contains nondimensionalized center deflection and stresses for (0190) laminate under sinusoidal load.3. The finite element results are in excellent agreement with the analytical solutions of the firstorder shear deformation plate theory.. . We note that a a ~~h~ x lo2.2. For additional results. At = 5 ps (9.36) and (9. The results are in good agreement with the analytical solutions. under suddenly applied sinusoidally (SSL) or uniformly (UDL) distributed transverse step load are analyzed.. The constantaverageacceleration scheme ( a = y = 0. In all cases initial conditions were assumed to be zero.12 ' 2 ' 2 b2qo h h2 7 ' = "y(O7 ) 2 b2qo a b h = ayy(57 2 7 2 h b ) = ax. (9. the reader may consult References 7880. Figures 9.7 Transient Analysis Here we present results of transient analysis obtained using the shear deformable finite element..37). The geometry and material properties are the same as listed in Eqs. aXy twolayer and eightlayer antisymmetric crossply square plates under suddenly applied sinusoidally or uniformly distributed transverse step load.3.3.a 2 2'2' ) 12 h I 2 h Table 9.(.9. b4qo a b h h2 = ayy2 7 .5) of Newmark is used for time integration. The ninenode quadratic element with selective integration rule is used in the examples discussed here.b2q0 7 (..7 (also see Reddy [78]).. Simply supported (SS1) antisymmetric crossply square plates (0/90). "XY = a.10): The finite element results were obtained using a 2 x 2 mesh of ninenode quadratic FSDT finite elements in a quadrant.36) The following nondimensionalizations are used: w =w0(2 ' )2 a.3. h = 5 cm. qo = 10 N/cm 2 .( a.yy 2 ' 2 ' a.2 2 a b 2'2' a b 0 (.
10) (xl = x and x2 = y): ( N. (8. can be derived in a manner similar to that of plates. 9. In fact.3.15: Comparison of transverse deflection and stresses obtained by the finite element method with closedform solution of twolayer crossply square plate under suddenly applied sinusoidal load.10).R2+Io6voi32vo+ h6voat22 ] 8~ at2 Q2 dxdy . Eqs.FINITE ELEMENT ANALYSIS OF COMPOSITE PLATES AND SHELLS 543 Table 9.4.3. Time  U J OST ffry  (PSI CFS~ FES CFS FES CFS FES t Closedform solution with Newrnark's scheme for time integration. (8. We begin with the weak forms of Eqs.3.6)(8.CO&) asvo a24 + N2 6110.3.3.6)(8.1 Weak Forms The displacement finite element model of the equations governing doublycurved shells. the main equations are presented here. For the sake of completeness. the finite element model of doublycurved shells is identical to that of FSDT with additional terms in the stiffness coefficients (see pages 465468 of Reddy [89]).4 Finite Element Analysis of Shells 9.
. ..U D L 0 100 200 Time. S S L (0/90)4. E z = 2. t (ps) 300 400 Figure 9. t (ps) 300 400 Figure 9. (0/90). S S L (0/90)4. h = 5 cm p = 8x kg/cm3 I R O T = l .. ~ .1 x lo6 N/cm2 a = b = 2 5 cm.10: Plots of center normal stress a. SS1. SS1.3.544 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS (0/90).9: Plots of center deflection versus time for simply supported (SS1) twolayer and eightlayer crossply square plates under sinusoidal or uniform step loading.. ~ O ~ I I I I I I I I I I I I I ~ ~ I I I I I I I I I I I I I I I I I I I I I I I I J (0/90). versus time for simply supported (SS1) twolayer and eightlayer crossply square plates under sinusoidal or uniform step loading. U D L (0/90). h = 5 cm p = 8x kg/cm3 I R O T = l . At=5 p s .UDL 0 100 200 Time. UDL (0/90).. (0190). q 0 = 10 N/cm2 Material 1. go = 10 N/cm2 Material 1. At=5 ps .3.1 x lo6 N/cm2 a = b = 2 5 cm. Ez = 2.
FINITE ELEMENT ANALYSIS OF COMPOSITE PLATES A N D SHELLS 545 Figure 9.3. UDL .12: Plots of center normal stress aTV versus time for simply supported (SS1) twolayer and eightlayer crossply square plates under sinusoidal or uniform step loading. SS1.. UDL 1 0 100 200 Time. SSL (0/90)4.11: Plots of center normal stress ayy versus time for simply supported (SS1) twolayer and eightlayer crossply square plates under sinusoidal or uniform step loading.1 x lo6 Nlcm a = b = 2 5 cm. (0/90). A t = 5 p s (01901..3. ..(01901. h = 5 cm p = 8 x 10"g/cm3 IROT=l. E p = 2. t ( p s ) 300 400 Figure 9. q o = 10 N/cm2 Material 1.
41. M i and Qi are defined by Eqs. $7 . 2 . Note that in the case of shells. We note from the boundary terms in Eq. wo.3. (8. .4.2 Finite Element Model Using interpolation of the form where are Lagrange interpolation functions.vo.lae) that (uo. .42) are the primary variables. we can use the CO interpolation of the displacements. Therefore.b).5(1/R1 . surface displacements are coupled to the transverse displacement even for linear analysis of isotropic shells. l / R 1 = O and 1/R2 = 0). The secondary variables are ~ ) where ( N ~ . with p = 1 .1/R2) and the stress resultants Ni.where Co = 0. (9.4. 2~ . In the present study equal interpolation (m = n = p) of five displacements.la. Note that the finite element model developed here for doublycurved shells contains the FSDT plate element as a special case (set Co = 0. 9. 6 are the applied surface loads that are introduced to study the buckling problem. . is used.
vo. wo.FINITE ELEMENT ANALYSIS OF COMPOSITE PLATES AND SHELLS 547 Substituting Eqs. 2 . 5 ) by the expressions in (9. 42) into the weak forrris in Eq.4. .7). The stiffness coefficients [KaB] are defined as follows: and the nonzero coefficients N E .5) for (uo.3. . 2 .lae). . .4. (9. M E . 6 are . . .p = l . and Q$ for a = 1 . 5 and I = 1 . we obtain the semidiscrete finite element model of the firstorder shear deformation shell theory: ] where the coefficients of the submatrices [ M N f iand Gij are the same as those defined for ( a . . . (9. 2 .4. $1.3)(9.
. are the thermal force and moment resultants. etc. MT.where NF.
[91] 0. ~ 4 = 0. The results are compared. with those reported in the literature. Quadrilateral elements with selective integration rule to evaluate the stiffness coefficients (full integration for bending terms and reduced integration for bendingmembrane coupling terms and transverse shea.4.3 Numerical Results Here we present numerical results for a number of problems.25 2 (9.r tJerms) are used.) of a clamped cylindrical shell with internal pressure. The reference solutions by Rao [90] and Timoshenko and WoinowskyKrieger [91] did not account for the transverse shear strains. r R / 2 . The shell is clamped at its ends (see Figure 9.1870 0. E2 = 2 x 10"si. . The geometric and material parameters used are R1 = 10" (& = 0) .9a) El = 7.3754 0.25 x 10"si GI3 = G23 = 0.1: Clamped cylindrical shell with internal pressure. [go] 0. In all examples presented here we set Co = 0. ~ ~ = 4 ~ = O 2a=t0 . See Chapter 10 for a discussion of the socalled membrane locking.1.5 x 106psi.4. Clamped cglindrical shell Consider the deformation of a cylindrical shell with internal pressure [89].625 x 10' psi.. R2 = R = 20in.1803 Figure 9.4. isotropic as well as composite shells (mostly cylindrical shells).3727 0.4.367  0.1).4. a n d u o = v o =wo =41 = 4 2 = O a t x1 = a / 2 ) x of the shell are presented in Table 9.. Present Solutions Laminate 0 0190 4 x 4Q4 2 x 2Q9 Ref.4. The numerical results obtained using The pressure is taken to be po = ( 6 . (9. a = 20in.FINITE ELEMENT ANALYSIS OF COMPOSITE PLATES AND SHELLS 549 9.1: Maximum radial deflection (wo in. Table 9. 4 1 / ~ ) 4 x 4 mesh of fournode (linear) quadrilateral elements (4 x 4Q4) and 2 x 2 mesh of ninenode (quadratic) quadrilateral elements (2 x 2Q9) in an octant (uo= 41 = 0 a t = 0 .4.3666  Rcf.9b) ksi. GI2 = 1. when available. h = l i n .
2: Simply supported spherical shell panel under central point load.2: Maximum radial deflection ( .2.4. The finite element solution converges with refinement of the mesh to the series solution of Vlasov [92].9373 0190 45145  0. [92] 0.2).0349 1.who did not consider transverse shear strains in his analysis.) of a simply supported spherical shell panel under central point load.4.4. The numerical results obtained using various meshes of linear and quadratic elements in a quadrant of the shell are presented in Table 9.3904  4 x 4Q9 Ref. [go] Nonuniform 0.2376  Ref.3956  Isotropic 0.3726 1. Table 9.3506 Orthotropic 0. The geometric and material parameters used are The point load is taken to be Fo = 100 lbs.2644 1.5504 4 x 4Q9 Uniform 0.Doublycurved shell panel Next. The shell panel is simply supported at edges (see Figure 9.4.3866    Figure 9. .w o x 10 in. we consider a spherical shell panel (R1= R2 = R) under central point load [89].0217 0.3935 1. Present Solutions Laminate 4 x 4Q4 Uniform 2 x 2Q9 Uniform 0.
are used in a quadrant of the shell with different p levels. A45or A44).. those containing A44. for p = 1 the mesh is 8 x 8Q4.3): cr =0.. one with 81 nodes (405 DoF) and the other with 289 nodes (1. "Reduced integration" rule is one in which one point less than that in the full integration rule is used.4. For example. where K denotes the number of Gauss points (i.45 x 10"si. q = 0.04psi Two sets of uniform meshes.4. One may use full integration for all terms.3)(9. (9... Doubling the above  . With 5 degrees of freedom at each node. The values of I.e.e. J and K used in the present study for different p levels and integration rules are listed below. "Full integration'' means using a Gauss rule that evaluates an integral exactly. a = 20 in. K x K Gauss rule) used to evaluate the transverse shear terms (i. (9. for p = 2 the mesh is 4 x 4Q9.5)].. h = 0.125 in.445 DoF). p level Full integration Selective integration Reduced integration Clamped cylindrical shell panel First we consider an isotropic cylindrical shell panel with the following geometric and material parameters and subjected to uniformly distributed transverse (normal to the surface) load q (see Figure 9. and for p = 8 the mesh is 1 x 1Q81 all meshes have a total of 81 nodes.The remaining example problems of this chapter are analyzed using various p levels [see Eq.3. R = 100 in.1 sad.4. the nunher of degrees of freedom per element for different p values is as follows: Element type p level DoF per element The numerical integration rule (Gauss quadrature) used is I x J x K.4. or selective integration where reduced integration for transverse shear and coupling terms and full integration for all other terms in the stiffness matrix.11a) (9. reduced integration for all terms. J denotes the number of Gauss points to evaluate the bendingmembrane coupling terms (which are zero for the linear analysis of plates).4.11b) E = 0. and I denotes the number of Gauss points used to evaluate all remaining terms in the stiffness matrix. v = 0.
1577 1. q. q acts vertically down. Selective integ. The results obtained with selective and reduced integrations are in close agreement with those of Palazotto and Dennis [93] and Brebbia and Connor [94].1349 1.1349 1.3).625psi ..4.1352 1.3: Vertical deflection (wAx 10' in. = q cos  9 q R' = 0.1562 1.e.1349 1.4.1349 1.144 x lop2 in. The geometric and material data of the problem is (see Figure 9. Selective intcg.4.4) ..7456 1.1349 1.1349 1.1349 t Palazotto and Dennis [93] reported 1.1349 1. Reduced integ. The problem consists of a cylindrical roof with rigid supports at edges x = fa / 2 while edges at y = fb/2 are free. meshes will have 289 nodes.4. known as the ScordelisLo roof [95]. Y 0.3378 1.3: Clamped cylindrical shell panel under uniform transverse load.1427 1. Table 9. while Brebbia and Connor [!I41 reported a value of 1.1404 1.3. A solution to this problem was first discussed by Cantin and Clough [96] (who used u = 0.1351 1.1348 0.1721 1.1401 1. The vertical displacement at the center of the shell obtained with various meshes and integration rules are presented in Table 9. Full integ.)t at the center of the clamped cylindrical panel under uniform transverse load. Mesh of 81 nodes P levcl Mesh of 289 nodes Full integ.Figure 9.1347 1. Reduced integ.1349 1.. Barrel vault This is a wellknown benchmark problem. The shell is assumed to deform undcr its own weight (i. u = R' q.1349 1.0.1 x 10p2 in. not perpendicular to the surface of the shell). 1 2 4 8 0. = q sm  E = 3 x 10' psi.1348 1.1348 1.
6429 Reduced integ. At y = b/2 : Two sets of uniform meshes. The displacement at y = fb/2 (middle of the free edge) of the shell.6434 3.6431 3. Atx=u/2: vo=~o=$~=O Free (9.6393 3.4: A cylindrical shell roof under its own weight.6428 3. Figure 9. Figure 9.6430 3.)+ at the center of the free edge of a cylindrical roof panel under its own weight.445 DoF) and the other with 1. Table 9. 3.6419 Full integ. while Figure 9. are used in a quadrant of the shell with different p levels.4: Vertical deflection (wB in. 4 .6428 3. Fox and Rifai [97].445 DoF).6430 3. The results obtained with selective and reduced integrations are in close agreement with those reported by Simo.6399 3. one with 289 nodes (1.F I N I T E E L E M E N T ANALYSIS O F C O M P O S I T E PLATES A N D SHELLS 553 Figure 9..8387 3.6429 Reduced integ.2681 3.4.y) and ~ ~ ( 3 0 0 . The boundary conditions on the computational domain are Atx=O: uo=q!q=O.6430 3.4.6425 3. 3. are presented in Table 9. To avoid shear and membrane locking one must use at least a mesh of 4 x 4Q25 (p = 4). 1.4. 3. .6428 3.6367 3.5 shows the variation of the vertical deflections wo(O. Fox and Rifai [97] reported w.4.6374 3. obtained with various meshes and integration rules.6429 1 2 4 8 t Simo.9002 3. 3. Mesh of 289 nodes P level Mesh of 1.4.089 nodes Full integ. 2 .6429 Selective integ.f = 3.5 also contains the results of Zienkiewicz [98]. 8 .4.5415 3.6392 Selective integ. 0.089 nodes (5.6170 3.6288 in. for deep shells.13) At y = 0 : vo = 4a = 0. y) as a function of xz = y.4.4.6 shows the convergence of the vertical displacement WB for p = 1 .
4 Figure 9.4. . (b) Displacement uo(300. 4 40 4x4Q81Present Angular distance. wB/wref.6: Convergence of the relative vertical deflection.y).554 MECHANICS OF LAMINATED COMPOSITE PLATES AND SHELLS .1 . y).5: (a) Vertical deflection wo(O. o I 1 I 0 10 20 30 Angular distance. 0 5 10 15 20 25 30 Number of nodes per edge 35 Figure 9.4.
4. as shown in Figure 9. Crossply laminates Layers Angleply laminates S = Rlh 13 QLL u~~  U) om   OYY Pinched cylinder This is another wellknown benchmark problem [991011. .16) Three different meshes with 81 nodes.. .4. u is set to zero at x = y = 0 to eliminate the rigid o body mode.4. .4.5: $2 = Table 9. It is clear that the problem requires a high p level to overcome shear and membrane locking. S = R l h .4. 289 nodes and 1.4. . R = 300 in.6 contains radial displacement at the point of load application.) laminated shells for different radiustothickness ratio.8 shows the convergence characteristics of the problem. The circular cylinder with rigid end diaphragms is subjected to a point load at the center on opposite sides of the cylinder. Figure 9. The following dimensionless quantities are presented in Table 9.5 contains the nondimensionalized deflection and normal stresses for twolayer and tenlayer antisymmetric crossply (0/90/0/90/ .4. The geometric and material data of the problem is The boundary conditions used are: At y = 0. a = 600 in.FINITE ELEMENT ANALYSIS OF COMPOSITE PLATES A N D SHELLS 555 The barrel vault problem is also analyzed when the shell is laminated of a composite material. b/2 : vo = $2 = 0 (9. Table 9. The solution of Fliigge [99]is based on classical shell theory. and The full panel is modeled with 4 x 4Q81 mesh and boundary conditions vo = wo = 0 a t x = fa / 2 .) and angleply (451451 451 . In addition.089 nodes (with different p values) are used in the octant of the cylinder. . The data of the problem is a = 40°. Table 9.7.5: Maximum transverse deflections and stresses of crossply and angleply laminated cylindrical shell roof under its own weight.
[loo] reported the value .4. Full Selec. Full Mesh of 1. Mesh of 81 nodes p level Mesh of 289 nodes Full Selec...4. wA/wref.05 1  Number of nodes per edge F i g u r e 9. Reduc.F i g u r e 9.4. t The analytical solution of Fliigge [99] is 1.f = 1.7: Geometry of the pinched circular cylinder problem.6: Radial displacement (wA x lo5)] at node 1 of the pinched circular cylinder problem. 1. Reduc.089 nodes Selec. Table 9. Cho and Roh w.8248 x lop5 in.8541 x in. Reduc.8: Convergence of the relative radial deflection.
(9. R = 300 in. The panel is simply supported a t all its edges.16)]. a = 4.) and angleply (451451451. we would obtain erroneous results.14b). a : vo = w = 4 2 o CrossPly : J: = = 0. (9. angle 2a and radius R. as shown in Figure 9. The geometric and material parameters of the problem are a = n / 8 rad. The data of the problem is o = 45". a = 600 in. CrossPly Layers AnglePly  S = R/h W U UZJ.7: Displacements and normal stresses a t point A of the laminated pinched circular cylinder problern. a : vo = w = 4 = 0 .9. o 2 u = 4a = 0 o Table 9.7 contains the nondimensionalized deflections and normal stresses for twolayer and tenlayer antisymmetric crossply (0/90/0/90/ . The results were obtained using 4 x 4Q81 mesh in half cylinder and full integration [and boundary conditions given in Eq.4. .4..F I N I T E E L E M E N T ANALYSIS O F C O M P O S I T E PLATES A N D SHELLS 557 The pinched circular cylinder problern is also analyzed when the shell is laminated of a composite material. R = 1. and the lamina material properties used are the same as those in Eq.4. .. q(x.16~~) 0. S = R l h . The following nondimensionlization is used: The boundary conditions used are: AnglePly : x = 0.4. guy u  nm   . . b : vo = 42 = 0 y = 0. y) = qo sin  nx ny sin a b . . and subjected to distributed transverse load q. b : (9.) laminated shells for different radiustothickness ratio.~ Y Y S i m p l y supported cylindrical panel The last example of the section deals with the bending of a crossply laminated circular cylindrical panel of length a. y = 0. Table 9. If the same mesh and boundary conditions as those used for the crossply laminated cylinder to analyze the angleply laminated cylinder.4.4.
and transient response of rectangular plates and bending of doublycurved (mostly cylindrical) shells are presented in tabular and graphical forms. and various ~ s Rlh] stresses [@. For additional details. The boundary conditions used for the panel are: A mesh of 4 x 4Q25 is used in the full panel and the stiffness coefficients were evaluated using full integration.p = for different radiustothickness ratios. buckling. 9. The present results are compared with the 3D analytical solutions of Varadan and Bhaskar [102] and closedform solution developed by Cheng. . The finite element models developed herein are general in that they can be used for any lamination scheme. Numerical examples of bending. b/2.Figure 9.8 contains the maximum displacement [. the reader may consult the references listed at the end of the chapter. Table 9.9: Geometry of the simply supported circular cylindrical panel. [I031 using the thirdorder shell theory (see Chapter 11) for (90/0/90) and (0190) laminates. and boundary conditions. natural vibration.4. et al. at the center of the panel. o ) ( ~ o E ~ / ~S = " .5 Summary In this chapter.4.w = wo(a/2. They are in good agreement with each other. geometry. The Sanders shell theory accounts for transverse shear strains in much the same way as the firstorder shear deformation plate theory. linear finite element models of the classical and firstorder shear deformation plate theories and the Sanders shell theory for doublycurved shells are developed.
[l02] Ref.are computed at the bottom . crossply laminated circular cylindrical panel under sinusoidally distributed load. variable? 5 = R / h Ref.4. [I031 ' Present Ref. 11031 Prcscrit o:.8: Displacements and stresses in simply supported. [I021 Ref.are computed at the top and €3 of.and oz.F I N I T E E L E M E N T ANALYSIS O F C O M P O S I T E PLATES A N D SHELLS 559 Table 9. o:.
t) is the distributed transverse load. (1). Note that you must select a complete polynomial containing six parameters and derive the Hermite interpolation functions. and d2wo/dx2 as the nodal variables of an element with two (end) nodes.1 The equation of motion governing the bending of symmetrically laminated beams according to the classical laminate theory is given by (see Chapter 4) where N ~ is the axial load and . with a total of six degrees of freedom per element.3 Derive finite element Hermite interpolation functions using wo and 0 = dwo/dx as the nodal variables of an element with three nodes (two end nodes and the middle node).5 Consider the following set of equations governing the classical plate theory: d2wo M x x = . 9. assume interpolation of the primary variables.1. q(x. 9. and I z are mass inertias Develop the weak form and finite elernent model of Eq.2 (a) Derive finite element interpolation functions using wo.e.Problems 9. I2 = bIz Construct weak forms of Eqs. you must select a complete polynomial containing six parameters and derive the Hermite interpolation functions. I . and I. lo= bIo. the Timoshenko) bean1 theory are where q = bq. q = bq. 9.. (1) and (2) over the typical finite elernent.4 The equations of motion governing symmetrically laminated beams according t o the firstorder shear deformation (i. 9. and develop the finite element model.2. As in Problem 9. lo= bIo.. = 2066 d2wo dxdy . (b) Use the finite element approximation to compute the stiffness matrix [ K e ]derived in Problem 9. with a total of six degrees of freedom per element. = b12 (2) and b is the width of the beam.( D H + ~~2 d2wo dy2 M. 0 = dwo/dx.
. R. R... 10. 7 2 . M. Y. Argyris. A n Introduction to the Finite Element Method.d Structures. M. The Finite Element Method. K. 3. N J (1987).5. "Triangular Elements in Plate Bending Conforming and NonConforming Solutions. 6. MA (1987). .. 547576 (1965). from equations (1)(4) of Problem 9. ... J. 0 . Reading. 101122 (1969). C. 2. T. Fraeijis de Veubeke. G. B." International Journal for Numerical Methods i n Engineering.M z c . "A Refined Triangular Plate Bending Finite Element. 9." Internmtional Journal of Solids an. 5.  8.. H.) are the independent variables. Englewood Cliffs. Reddy. 4(1). Linear Static and Dynamic Finite Element Analysis. Air Force Institute of Technology. M." Proceedings of the Conference o n Matrix Methods i n Structural Mechanics. 95108 (1968). and Zienkiewicz." Computers and Structures 19(3). The Finite Element Method.. K. Irons. PrenticeHall. Irons. Ohio. Vol. MV. 1...6 A simplified mixed model can be derived by eliminating the twisting n~ornent M. B. D.. N. "A Conforming Finite Element for Plate Bending. 7.where DL. New York (1993). McGraw Hill.MX. Energy Prznczples and Variational Methods i n Applied Mechanics. New York (2002). S." International Journal for Numerical Methods i n Engineering. AFFDLTR6680. W. and Hrudey. J. J. We can write the resulting equations as where h denotes the plate thickness and Develop the weak forms of the equations and associated finite element model. New York (1989). John Wiley. Reddy. Second Edition..(4) in an alternative form (curvatures in terms of the moments) such that (mo. WrightPatterson Air Force Base. "A Conforming Quartic Triangular Element for Plate Bending. References for Additional Reading 1. McGrawHill.. Hughes.2945 (1969). 479495 (1984). develop the weak form of the equations and develop a mixed finite elenlent model of the equations. D." Journal of the Royal Aeronautical Soczety.. N. Cheung. C.J. and Scharpf. Hrabok. Bell. Fried. M. (1). "The TUBA Family of Plate Bending Elements for the Matrix Displacement Method. 1: Linear Problems.. 701709 (1969). are the bending stiffnesses of a specially orthotropic plate (see Chapter 5) Rewrite Eqs. and Taylor. Zienkiewicz. L. 4. Burnett. 11. I.. T. P. B.. 9. 1. "A Review and Catalog of Plate Bending Finite Elements. Finite Element Analysis: AddisonWesley. Bazeley. 0 . M.. Second Edition.
WrightPatterson Air Force Base. Newmark. K. and Wilkes. Batoz. 21(7).. Tisdale. C. L. R. 243251 (1976).." A I A A Journal. D. 7(1). J . 21. Air Force Institute of Technology. D. 18(7)." Technical Note. "Finite Element Stiffness Matrices for Analysis of Plate Bending. 471488 (1964).. 1 0 . 0 . R. New York. L. 18.. 0 . L. Carnahan.. 28. 1 8 . Air Force Institute of Technology. 30. "A Refined Quadrilateral Element for Analysis of Plate Bending. Ahmad. 0 . N. 17. 10311055 (1968).. J. J. Ohio." Journal of Engineering Mechanics. and Tocher. 6(1). S. 13(4). 16551677 (1982). C. A. "A Rapidly Converging Triangular Plate Element. S. Zienkiewicz. Haisler. K." A I A A Journal. WrightPatterson Air Force Base. Applied Numerical Methods. C.. AFFDLTR6680. G. Barlow. 16311637 (1963). 575586 (1971). C. K. "An Explicit Formulation for an Efficient Triangular Plate Bending Element. L.. l 5 ( l 2 ) . 27. .. C. 14. Morley. 1969. 16." Proceedings of the Conference on Matrix Methods i n Structural Mechanics. A. J. "Basis of Derivation of Matrices for the Direct Stiffness Method.. "An Alternative Formulation for DKT Plate Bending Element. "The ConstantMoment Plate Bending Element. J. Air Force Institute of Technology. 1. J ." Proceedings of the Conference on Matrix Methods i n Structural Mechanics. Trondheim (1967). "The Triangular Equilibrium Element in the Solution of Plate Bending Problems. O.. WrightPatterson Air Force Base. for Plate Bending Problems. "Vibration and Stability of Plates Using Finite Elements. 19. and Cheung. and Tahar. "Evaluation of a New Quadrilateral Thin Plate Bending Element.. 6794 (1959).. Y. "The Finite Elernent Method for Analysis of Elastic Isotropic and Orthotropic Slabs. 10771089 (1985).. F . "A Study of ThreeNode Triangular Plate Bending Elements. Irons. and Gunderson. Batoz. 13. B. 515545 (1965). U K . Clough. Ohio." Journal of Strain Analysis. M. 8 5 . and Ho. B." International Journal for Numerical Methods i n Engineering. Anderson. "The Generation of InterElementCompatible Stiffness and Mass Matrices by the Use of Interpolation Formulas. 25. 399440 (1968). M. Jr. Journal of the Aero~~autical 26.. A.. John Wiley & Sons. Morley. Clough. L. Bathe. J. L. Bogner." Proceedings of the Second Conference on Matrix Methods i n Structural Mechanics. L." International Journal for Numerical Methods in Engineering. 7 2 (1968). S. 17711812 (1980).. Ohio.. 2024 (1971).. L. Civil Engineering Series 46. 24. and Felippa. A.. Batoz." International Journal for Numerical Methods i n Engineering.12. "Analysis of Plates in Flexure by a Simplified Finite Element Method. Melosh. Morley. R. Irons. 29. Hellan.. 397443 (1965).. M. R. 4(10). S. W. J." Proceeding of the Institute of Ci7iil Engineers.. and Zienkiewicz. and Zienkiewicz.. 15. "A Method for Computation of Structural Dynamics. London..." Journal of the Aeronautical Society. B. R. W. Stricklin. 28. AFFDLTR6680. J." International Journal for Numerical Methods i n Engineering. R. 20. 12891293 (1985). 22. and Schmidt.. 180181 (1969). 7 1 (1967). P. L. M." International Journal for Numerical Methods i n Engineering. "Optimal Stress Location in Finite Element Models. K. Jeyachandrabose. Fox. W. W.. and Kirkhope. L. L.." Acta Polytechnica Scandinavia." International Journal of Solids and Structures. AFFDLTR6680.. D... 23. Luther." International Journal for Numerical Methods i n Engineering.. B. J. "A Triangular Equilibrium Element with Linearly Varying Bending Moments Society. "Analysis of Thick and Thin Shell Structures by Curved Finite Elements.
36. 11(10). 47. 0. "More on Optimal Stress Points Reduced Integration Element Distortions and Error Estimation." Engzneerzng Computatrons." International Journal for Numerical Methods i n Engineering. 269 295 (1984). Park. R. 42. and Tsay. "On a Hierarchy of Conforming Tinloslienko Beam Elements. and Rcddy. T. 445450 (1978)... and Liu. Barlow. 48.. Reddy. 275290 (1971)." International ." Computer Methods z 7 ~Applied Mechanics an." Engzneenng Corrcputations..J. and Cohen. 49." Computer Methods i n Applzed Mechanics and Engineering. 9(5). 335344 (1981). C. "Stabilization Matrix for the Bilinear Mindlin Plate Element. .. 41. 833852 (1984).. 5774 (1990). 45.. M. University of New South Wales. S.. "A Stabilization Procedure for the Quadrilateral Plate Bending Element with Onepoint Quadrature. Belytschko. 409431 (1986). Hughes.. 1 9 405420 (1983). T." Nuclear Engineering and Design. and Averill. M." Computers and Structures 23(3). 44. 0 ." International Journal for Numerical Methods i n Engineering.d Engineering. 44. "A Synibolic Fourier Synthesis of a Onepoint Integrated Quadrilateral Plate Element. H. MI. 369379 (1984). Malkus. 6381 (1978). arid Huang.31. "A Penalty PlateBending Elenlent for the Analysis of Laminated Anisotropic Composite Plates... E. Taylor. T. S. Zienkiewicz. "A Family of Quadrilateral Mindlin Plate Elements with Substitute Shear Strain Fields." Computers and Structures 1 4 . Tsay.. 7. B. 46. "Simple Finite Elements with Relaxed Continuity for NonLinear Analysis of Plates. D." Computers and Structures. 15(8). J.. Pugh. Ong. W. Hinton. "A NineNode Lagrangian Plate Element with Enhanced Shear Iriterpolation. J. Averill. 2(5/6). K." Proceedzngs of the Third Internatio~cal Conference i n Australia o n Finite Element Methods. L. R. Crisfield. N. R. D.  32. J. E. 39. 12(7). D. T. 33. C. K. 1. W. 14861504 (1989). 43." Computers and Str. L. 203222 (1981).. . R. 3. 38. "Mixed Finite Element MethodsReduced and Selective Integration Techniques: A Unification of Concepts. Reddy. 35.. 313 327 (1981). and Liu. 15291543 (1977). S. 18. J . J. N.J. "Reduced Integration Technique in General Analysis of Plates and Shells." International Journal for Numerical Methods i n Engineering. C. Hughes. arid Zicnkiewicz. R. C.. 34. Belytschko. J ." International Journal for Numerzcal Methods i n Engineering. Hinton.. "A Simple and Efficient Element for Plate Bending. Belytschko. R. anti Dong... N. Huang. J.ics and Engineering. l 5 ( l ) . S. W. 46. and Kanoknukulchai.. 28.. and Hughes." Computer Methods zn Applied Mechan.. "A Quadratic Mindlin Element Using Shear Constraints. 10591079 (1978).." Computer Methods i n Applzed Mechamcs and Engineering. Sydney (1979). "Advances in the Modeling of Laminated Plates. M. L... H.. 29." Concputzng Systems i n Engineerrng. K. "The 'Heterosis' Finite Element for Plate Bending. 11871206 (1980).: "Reduced and Selective Integration Techniques in the Finite Element Analysis of Plates. A. T . hl. S. "A Study of Quadrilateral Plate Bending Elements with 'Reduced' Integration. Cohen. C. R. J . "A Consistent Control of Spurious Singular Modes in the 9node Lagrange Element for the Laplace and Mindlin Plate Equation.Journal for Numerical Methods i n Engineering. C. 203236 (1985). 37. and Taylor. R. J.. J . E." International Jourl~alfor Numerical Methods i n Engineering. A. C. 48(2). and Hinton. and Flaggs. and Haroun. J . "Behavior of Plate Elements Based on the FirstOrder Shear Deformation Theory.. Reddy. 40. Hughes. T.uctures. 541555 (1991). Tessler.. N. C. Too. T . E.. C. L.
Nosier. 93120 (1983). 587596 (1981). J. A. and Reddy. 54. T . L.. 437445 (1988). Khdeir." Journal of Composite Materials. I. 2034 (1970). R." Computer Methods i n Applied Mechanics and Engineering. J. and Munir. 18081817 (1989). A. N. L. J. "Buckling and Vibration of Laminated Composite Plates Using Various Plate Theories. 475493 (1980). J. N. "A HybridStress Quadratic Serendipity Displacement Mindlin Plate Bending Element. and Reddy. A. . 64. "Exact Solutions for Rectangular Bidirectional Composites and Sandwich Plates." Composite Structures.. 63. N. M. L. 419438 (1991). and Reddy. 62. A. A. 1121 (1980). "Exact Solutions for Composite Laminates in Cylindrical Bending. N." Acta Mechanica. "Dynamic Response of Antisymmetric AnglePly Laminated Plates Subjected to Arbitrary Loading. 123170 (1992). 3. Khdeir. 311335 (1983). 153167 (1981). N. 64. 14471463 (1987). 9 . Tessler. 14(4). N. L. J. and Chao. 71101 (1985). A. 66. 23(10).. 189213 (1988). A." Computers and Structures. 60. R. A. "A FourNode Plate Bending Element Based on MindlinIReissner Plate Theory and Mixed Interpolation. L. A. 367383 (1985). 15(8). "An Improved Treatment of Transverse Shear in the MindlinType FourNode Quadrilateral Element." AIAA Journal. Spilker. '(ThreeNode Mindlin Plate Element with Improved Transverse Shear. and Khdeir. A Modified Version of Lyons' Element. Bathe. S. Khdeir. and Reddy. 9. Pagano. "Effects of Shear Deformation and Anisotropy on the Thermal Bending of Layered Composite Plates. N. J. 50. "A FourNoded Plate Bending Element Using Shear Constraints. N. 38. and Librescu. 48(3). T.. 58.50. 51.. Spilker. 15(8). 4." Nuclear Engineering and Design. 259277 (1988). "Analysis of Symmetric CrossPly Laminated Elastic Plates Using a HigherOrder Theory: Part IStress and Displacement.." Internat~onal Journal for Numerical Methods i n Engzneering." Journal of Thermal Stresses.. 126(3). W. J. Pagano. J. 55. Reddy. 57. "A Comparison of ClosedForm and Finite Element Solutions of Thick Laminated Anisotropic Rectangular Plates. L. "Analysis of Symmetric CrossPly Laminated Elastic Plates Using a HigherOrder Theory: Part 11Buckling and Free Vibration. Reddy. and Munir. 65. R.. T... A. N. Khdeir. "A Serendipity CubicDisplacement HybridStress Element for Thin and Moderately Thick Plates. T. R... 59. "Finite Elements Based Upon Mindlin Plate Theory with Particular Reference to the FourNode Bilinear Isoparametric Element. and Librescu. 68.. N.. Reddy. J. Crisfield. J. and Tezduyar. 67." Journal of Composite Materials." Journal of Applied Mechanics.." Journal of Thermal Stresses. A. and Hughes. Spilker." International Journal of Solids and Structures.. 2 7 (12). 94 (34). Tessler. 56. Hughes. 61. 21. E. R. K. N. 398411 (1967). R. A. 52. "On Vibration and Buckling of Symmetric Laminated Plates According to Shear Deformation Theories. Y. 1239I260 (1980). and Munir. J. C.. J. Khdeir. N. E. J. "Thermal Stresses and Deflections of CrossPly Laminated Plates Using Refined Plate Theories. A. 12." Computer Methods i n Applied Mechanics and Engineering. "The HybridStress Model for Thin Plates. 53.." International Journal for Numerical Methods i n Engineering." Journal of Sound and Vibration. 39. A." International Journal for Numerical Methods i n Engineering. 12611278 (1980). I.. 3 .. I.. A." Composite Structures. A. and Hsu. and Dvorkin. and Hughes. "Analytical Solution of a Refined Shear Deformation Theory for Rectangular Composite Plates. N." Computer Methods i n Applied Mechnnics and Engineering. Librescu.
A. 347363 (2000). 111 (2001). R. A. 583591 (1997). J. P. 13. 65. A. A. J. A. T h e Net. Benjeddou. J. A. "Electrostrictivc Actuators: Materials and Applications. Uchino. "A Rectangular Laniiriated Anisotropic Shallow Thin Shell Finite Element. 1333 (1978). 6(5).. and Cheng. Y. 3. M. TerfenolD. 570578 (1991). K. "On the Solutions t o Forced Motions of Rectangular Composite Plates.." Journal of Intelligent Material Systems and Structures. "Comparison Between Shear Deformable and Kirchhoff Theories for Bending.. Reddy..d Vibration. Reddy. A. N." Computer Methods i n Applzed Mechanics and Engineering. 3 . "Deformations of Piezothernloelastic Laminates with Internal Electrodes. N. X. Ng. Khdeir." International ." Journal of Sound a7. N.." Journal of Applied Mechan. 122(2). N. Y. N. "Analytical Solutions of Refined Plate Theories of CrossPly Composite Laminates. J . "Free Vibration of Antisymmetric.." Composite Structures. 21(4). 19." Smart Materials and Structures.. AnglePly Laminated Plates Including Transverse Shear Deformation by the Finite Element Method. 647. Lam.Journal for Numerical Methods i n Engineering. G. Reddy... Buckling and Vibration of Antisymmetric AnglePly Laminated Plates." Composite Structures. and Reddy. "Free Vibration arid Buckling of Symmetric CrossPly Laminated Plates by a n Exact Method. 245254 (1992). A. 15. A. 159 172 (1989). . K. Khdcir. and Reddy.652 (1986). A. "Free Vibration and Buckling of Unsyrnrnctric CrossPly Laminated Plates Using a Refined Theory." Journal of Sound and Vibration. 11.. Reddy.ce and Technology. Peng. for Active Vibration Control. 403408 (1982). 205 224 (1989). Continuum Mechanics of Electromagnetic Solids. N. 66(4). G.." Composite Scien.. J. K. A. New York (1984)..herlarids (1988). "Exact Solutions for the Transient Resporise of Syrrirrietric CrossPly Laminates Using a HigherOrder Plate Theory. N. 128(3). 34. Lam. A. Khdcir." Journal of Sound and Vibration." Journal of Sound an." A S C E Journal of E n p e e r i n g Mechanics. Khdeir. "Advances in Piezoelectric Finite Element Modeling of Adaptive Structural Elements: A Survey.. Z." A I A A Journal. M. N. 347359 (2001). 115. 377395 (1989). "Geornetrically Nonlinear Transient Analysis of Laminated Conlposite Plates. Pradhan..d Vibration." Smart Materzals and Structures." Ceramic Bulletin. N. T. "Control of Laminated Composite Plates Usirig Magnetostrictive Layers. John Wiley & Sons.. Y.: "A FiniteElement Model for Piezoclectric Composite Laminates. J. "An Exact Approach t o the Elastic State of Stress of Shear Deformable Antisymmetric AnglePly Laminated Plates. 113(4).. J. Liu. "Stability of Antisynimetric AnglePly Laminated Plates.. 377388 (1988). N." Computers & Structures.. 76. Maugin. K. A. A.. Q. Khdeir. A. "Free Vibration of Antisymmetric AnglePly Laminated Plates Including Various Boundary Conditions. 21." Journal of Pressure Vessel Technology. Khdeir. Reddy." Z A M M . 126(3).ics. 4 9 . Reddy.. "Adaptive Characteristics of the Magrietostrictive Alloy. arid Shoop. 237255 (1983). 568593 (1999).Q. 447461 (1988). "On Laminated Composite Plates with Integrated Sensors and Actuators.Khdeir. NorthHolland. 565576 (1979). A. N..245258 (1989). 952962 (1989). Rao.. S. Khdeir. Reddy. K. C. 81(5). J. A. J . and Reddy. Amesterdarn." Engineering Structures. A. 10. "Dynamic (Transient) Analysis of Layered Anisotropic CompositeMaterial Plates.: Energy and Variational Methods i n Applzed Mechanrcs. J. J... 621629 (1983). Goodfriend.. and Reddy.
94. M. Part 11: The Linear Theory. 93. "Development of Geometrically Exact New Elements Based on General Curvilinear Coordinates. McGrawHill. General Theory of Shells and Its Applications i n Engineering. 100. 538560 (1964).. H.. (1992).. NASA T T F99. Second Edition. Washington. K. Theory of Plates and Shells.. New York (1977). J." International Journal of Solids and Structures. Germany (1973). Stresses i n Shells. T. N. I. Y. New York (1959). "A Curved Cylindrical Shell Finite Element.. 11231142 (1997).. J.91. S. 141156 (1991). Schmidt. Simo.. Fliigge.. G. AIAA Education Series. "Bending of Laminated Orthotropic Cylindrical Shells . Timoshenko. . 463483 (1969). (1964). C. S. S.. 17. J. 32(6). W.. S. V. "Geometrically Nonlinear Finite Element Analysis.. and WoinowskyKrieger. S. 0 . Kreja." Computer Methods i n Applied Mechanics and Engineering 73. "Computer Analysis of Cylindrical Shells. D. K. Washington." Journal of American Concrete Institute.C. 95. T. 101. R.. and Connor. Brebbia. A.. 21272150 (2000). Cheng. D. 92. A. "Finite Elements Based on a Firstorder Shear Deformation Moderate Rotation Theory with Applications to the Analysis of Composite Structures. H. 102. W. The Finite Element Method. 1057 (1968). and Lo. Q. K.. Z. SpringerVerlag. 99. D. and Kitipornchai." A I A A Journal. and Clough. 98." Composite Structures. McGrawHill. 37. 96. Varadan. and Rifai. Zienkiewicz. Cho. and Dennis. 81115 (2003). 6. National Aeronautics and Space Administration.C. and Reddy. 56. Fox. He. Cantin. "Influence of Imperfect Interfaces on Bending and Vibration of Laminated Composite Shells. L. C. D. (Translation of Obshchaya teoriya obolocheck i yeye prilozheniya v tekhnilce).." Journal of Engineering Mechanics. Scordelis. 97. R. M.. S.An Elasticity Approach. C. and Roh. Berlin." International Journal for Numerical Methods i n Engineering. 103." International Journal of NonLinear Mechanics. Vlasov... and Bhaskar. C. N.. "On a Stress Resultant Geometrically Exact Shell Model. 5392 (1989). Nonlinear Analysis of Shell Structures. Palazotto. Z.
The first approach is based on a laminate theory. the geometry is updated between load increments.Nonlinear Analysis of Composite Plates and Shells 10. The use of numerical methods facilitates the solution of such problems.e. Among the numerical methods available for the solution of nonlinear differential equations defined over arbitrary domains. as described in Chapter 3. perturbation method. The second approach is based on the 3D continuum formulation. In the nonlinear formulation based on small strains and moderate rotations. . the classical variational methods (e. We shall term the elements based on such assumptions the laminated elements (see [13201).23. the Ritz and Galerkin methods) are limited to simple geometries because of the difficulty in constructing the approximation functions for complicated geometries.1 Introduction The nonlinear partial differential equations governing composite laminates of arbitrary geometries and boundary conditions cannot be solved exactly. for example. where any kinematic assumptions are directly introduced through the spatial finite element approximations. and the equations are derived in an incremental form directly. Galerkin method. Full nonlinear strains or only the von KBrman nonlinear strains are included as desired. and only rectangular or cylindrical geometries were considered. Thus. two distinct approaches have been followed in developing nonlinear finite element models of laminated structures. the finite element method is the most practical and robust computational technique. For example.. The latter restriction is a direct result of the methods of analysis used. Approximate analytical solutions to the largedeflection theory (in von KBrmh's sense) of laminated composite plates were obtained by many (see. In most of these studies the effects of shear deformation and rotary inertia were neglected. Historically.26321). and the double series method cannot be applied to plates of complicated geometries. The formulation accounts for geometric changes that occurred during the previous increment of loading. and the geometric nonlinearity in the form of the von KBrmAn strains is included.. Ritz method. the geometry of the structure is assumed to remain unchanged during the loading.~ 121).g. in which the 3D elasticity equations are reduced to 2D equations through certain kinematic assumptions and homogenization through the thickness. i. [ I . Finite elements based on this formulation are called continuum elements (see [21.
The shear deformable nonlinear finite element models presented herein are used to study nonlinear bending. configuration.1 Governing Equations The equations of motion of the classical theory of laminated plates are given by [see Eq. Then a formulation of the continuum shell element is presented.. develop nonlinear laminated plate elements.2 Classical Plate Theory 10.e. mixed and hybrid finite element models. i.3. The major objective of this chapter is to study the geometrically nonlinear behavior of laminated plates and shells. geometry) of the structure for the current load increment is determined from a previously known configuration. Alternative finite element models to the displacement model.26321. The stress and strain measures used in this approach are the Cauchy stresses and the infinitesimal (Almansi) strains. In these formulations. Towards this objective we develop the displacement finite element models of the classical laminated plate theory (CLPT) and the firstorder shear deformation plate theory (FSDT) when the von K&rm&n nonlinear strains are accounted for. the configuration (i..2. one may consult [21. and postbuckling of laminated structures. often the undeformed. and changes in the displacement and stress fields are determined with respect to the reference configuration.25)] where the nonlinear expression N and the mass moments of inertia Ii are defined .. Additional details and numerical examples may be found in [17. 10.23. et al. [26] and Reddy [32]): (1) the total Lagrangian formulation and (2) the updated Lagrangian formulation. i.e. For additional discussion of the continuum finite elements.1925.There are two incremental continuum formulations that are used to determine the deformation and stress states in continuum problems (see Bathe. transient behavior. In the total Lagrangian formulation. and the updated configuration is used as the reference configuration for the next increment. The strain and stress measures used in this approach are the GreenLagrange strain tensor and 2nd PiolaKirchhoff stress tensor. the geometry of the structure from the previous increment is updated using the deformation computed in the current increment. all of the quantities are referred to a fixed. A direct consequence of this is that differentiations and integrations are performed with respect to this reference configuration.321.e. In the updated Lagrangian formulation. can be found in [33421. (3.
Nxy) and (Mx.2.MZy)are defined by where { N T } and { M T } are thermal force resultants and { N P ) and { M P } are the piezoelectric (or other actuation field) resultants 10.2.3). in general.2 Virtual Work Statement The stress resultants (N's and M ' s ) in (10.2. and electric/rnagnetic (or any . temperature increment. are related t o the displacement gradients.and the stress resultants (Nxx..Nyy.1)(10. n/fyy.
3) can be expressed in terms of displacements (uo.6). (10.2.6).2.2. the equations of motion (10. wo) by substituting for the stress resultants from Eqs.3) over a typical laminated plate finite element Re are given by (here only the thermal stress resultants are included) + Le( I ~a3w0 6uo d x d y at2ax  ) .actuation) field by Eqs.2.1)(10.2.5) and (10. Then the weak forms of (10.1)(10.2.vo.2. Therefore. (10.
[$++ax ax ] ay .)ax + *26 2] 1  [z+ 1 aw0 5 (.+ 1ax [ awO + 2 ( ..( + B26 [$ 5 + a2w0 + DlliP wo a2 1 1 (%12] + 2D66 axay awO ay2 ) 1 d 2wo ah. awo + a h 0 awO at2ax ax at2ay ay ) where and (nz. denote the direction cosines of the unit normal on the element boundary ny) I".)2] a2wo a2aw0  2 8 ~x1 8 { 6 ~ [2+ 5 1 awO (K)2] " +B. ..
11) where are the Lagrange interpolation functions.(x.11) into Eq.2. A. Y) = j=1 C AJelpjehY) n (10. (10. Y) = j=1 C u?d$(x.$. we obtain $7 [K"] [K12] [K13] [ [ K 2 ' ] [K"] [ K 2 3 1 ] [K31] [ K ~ [ K ~ ~ {]A } ~ ] { }+[ [OI [ M ? ~ ][ M 2 3 ] ] [O] [ M ' ~ ] ~ M ~ [~M ] ~ ~~ {]A } [ [ ~ l l ] wi3] { :ii} or. Y) = j=1 C v.2. the specific form of which will depend on the geometry of the element and the nodal degrees of freedom interpolated. are the values of wo and its derivatives a t the nodes.2. m Y). m vo(x.2.3 Finite Element Model Assume finite element approximation of the form un(x.9). w d x .10. in compact form [Ke]{ae} + [ ~ ~ ]= { F6 }~ } { e The nonzero elements of the stiffness matrix [ K e ] and mass matrix [ M e ]= [MeIT and force vectors { F ) and { F T ) are defined as follows: . Y). Substituting approximations (10. and cpje are the interpolation functions.
NONLINEAR ANALYSIS OF PLATES AND SHELLS 573 .
For example.vo). dwo dp: +B26+B66 a~ a~ dxdy M~=L I~$!~c.13) can be reduced to a set of nonlinear algebraic equations by means of the Newmark time integration scheme. d2vj" + a~ axay d2ve + 2.ax ax dwo dp. linear interpolation of (uo.7a.2.7. The plate bending elements discussed in Chapter 9 for the classical plate theory can be used here with a choice of Lagrange interpolation of the inplane displacements (uo.2.574 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS +%( ax a.MZ.2 + B267 d2v. (10. dxdy. . A discussion of iterative methods for the solution of the nonlinear algebraic equations resulting from (10. NZ. This completes the development of the nonlinear displacement finite element model of the classical plate theory in the rectangular Cartesian coordinate system.vo) and Hermite cubic interpolation of wo will have 20 element degrees of freedom for nonconforming rectangular element and 25 degrees of freedom for conforming rectangular element. see Reddy [32]).13) is presented in Section 10. as shown in Section 6.a x a y kl6. M: =  acp" kerideds dxdy where etc.4 (also. are the thermal (or hygrothermal and/or actuation) forces and moments defined in Eqs.b). Equation (10..2.
3 FirstOrder Shear Deformation Plate Theory 10. are defined by Eq.b).ASS.4a.NONLINEAR ANALYSIS OF PLATES AND SHELLS 575 10.1 Governing Equations The equations of motion of the firstorder shear deformation plate theory are given by where by N and I.2. and the stress resultants are given and the shear stiffnesses (A44.3. (10.A 4 ~ are defined by ) .
10.3.2 Virtual Work Statements
The weak forms of the equations of motion associated with the firstorder shear deformation plate theory are
+%{~16
a~
[=
auo a2uo
+
awo
(z)2]
+A26
[$+5
1
awo
(F)2]
+
Le(%
+ Il
g) ,,
dxd,
NONLINEAR ANALYSIS OF PLATES A N D SHELLS
577
0
. ne (,{Bl1
/
as$,
+ Bl6
duo [u+ dx + d x

(2+awo()awo 'ax I avo
1 2
awe
dy
+2
+ B6 .
[
+
1 2
()'I
awe
a~
dx
 
+ { a~
[
[ASS
+
($1
[$+  j 2 ]

1 awo 2 ay
+ K&
(h+
2)+
A45
(&+

)I$
+ 0 2 6 864 + D f i ~ ay
dTdY
J
(I'
ax
+ M ) d x d y jkeh f r L h & ds $
(10.3.9d)

J
3r
+ *M:,) a!)
dxdy

jrcA L 6 4 , d s k
where the secondary variables of the formulation as
10.3.3 Finite Element Model
The virtual work statements in Eqs. (10.3.9ae) contain at the most only the first derivatives of the dependent variables (uo,vo, wo, &, 4,). Therefore, they can all be approximated using the Lagrange interpolation functions. In principle, (uo, vo), wo , and (&, 4,) can be approximated with differing degrees of functions. Let
where ( ( a = 1 , 2 , 3 ) are Lagrange interpolation functions. One can use linear, quadratic, or higherorder interpolations of these variables. Substituting Eqs. (10.3.11)(10.3.13) for (uo,vo, wo, q5,, &) into Eqs. (10.3.9ae), we obtain the following finite element model:
q)
(10.3.14)
or, in compact matrix form
[ K e ] { A e+ [ M e ] { b e= { F ' ) ) )
(10.3.15)
NONLINEAR ANALYSIS OF PLATES A N D SHELLS
579
where the coefficients of the subrnatrices [ ~ " p ] [MaP]and vectors { F a ) and { F " ~ ) , are defined for ( a ,/3 = 1 , 2 , 3 , 4 , 5 ) by the expressions
ax ay +ay ax
dxdy
where N: N& and N& are thermal forces and 1 , ,, 1 and ~ f are the , thermal moments. When the bilinear rectangular element is used for all generalized displacements (uo, 710, wo. $ , T , &), the element stiffness matrices are of the order 20 x 20, and for the ninenode quadratic element they are 45 x 45 (see Figure 9.3.1).
10.4 Time Approximation and the NewtonRaphson Method
10.4.1 Time Approximations
Here we discuss the solution of equations of the form in (10.2.13) and (10.3.15). Equation (10.3.15), when generalized to include damping (structural or otherwise), has the form 1321 (10.4.1) [ ~ ] { h )] { A ) [K]{n) { F ) [c =
+
+
The fully discretized equations using Newmark's scheme are
where
and a are defined as (y = 2P) i
In Eqs. (2.4.2) and (2.4.3~~)' notation (.), indicates that the enclosed quantity the is evaluated at time t,. The new velocity vector {A),+l and acceleration vector { A ) , + ~at the end of each time step are computed using the equations
10.4.2 The NewtonRaphson Method
Equation (10.4.2) represents a system of nonlinear algebraic equations at time t,+l. These equations must be solved using an iterative method. Here we discuss the NewtonRaphson iteration method, which is based on Taylor's series (see [32, 434711. The NewtonRaphson iterative method is based on Taylor's series expansion of the nonlinear algebraic equation (10.4.2) about the known solution. Suppose that Eq. (10.4.2) is to be solved for the generalized displacement vector at time t,+l. Due to the fact that the coefficient matrix [K({A),+~)]depends on the unknown solution, the equations are solved iteratively. To formulate the equations to be solved at the r 1st iteration by the NewtonRaphson method, we assume that the solution at the r t h iteration, {A):+,, is known. Then define
+
where {R) is called the residual, which is a nonlinear function of the unknown solution {A),+l. Expanding {R) in Taylor's series about {A):+l, we obtain
where O(.) denotes the higherorder terms in {SA), and stzffness matrix (or geometric stiffness matrix)
[KT]is known as the tangent
Equations (10.4.5)(10.4.7) are also applicable to a typical finite element. In other words, the coefficient matrix in Eq. (10.4.7) can be assembled after the element tangent stiffness matrices and force residual vectors are computed. The assembled equations are then solved for the incremental displacement vector after imposing the boundary and initial conditions of the problem [see Eq. (10.4.6b)l
The total displacement vector is obtained from
Note that the element tangent stiffness matrix is evaluated using the latest known solution, while the residual vector contains contributions from the latest known and previous time step solution solution in computing element [K({A):+~)] in computing element {F),,,+l. After assembly and imposition of the boundary conditions, the linearized system of equations are solved for {SA}. At the beginning of the iteration i.e., r = 0), we assume that {A)' = (0) so that the solution at the first iteration is the linear solution, because the nonlinear stiffness matrix reduces to the linear one. The iteration process is continued [i.e., Eq. (10.4.8) is solved in each iteration] until the difference between {A);+l and { A } reduces t o a preselected error tolerance. The error criterion is of the form ~ ~ ~ (for the sake of brevity the subscript ' ( s 1)' on the quantities is omitted)
+
where N is the total number of nodal generalized displacements in the finite element mesh, and E is the error tolerance. The velocity and acceleration vectors are updated using Eqs. (10.4.4a,b) only after convergence is reached for a given time step. In the NewtonRaphson method the global tangent stiffness matrix and residual vector must be updated using the latest available solution {A);,, before Eq. (10.4.8) is solved. If the tangent stiffness matrix is kept constant for a preselected number of iterations but the residual vector is updat,ed during each iteration, the method is known as the modified NewtonRaphson method. The approach often takes more iterations to obtain convergence. The NewtonRaphson method fails to trace the nonlinear equilibrium path through the limit points where the tangent matrix [KT] becomes singular and the iteration procedure diverges. Riks [44] and Wempner [47] suggested a procedure to predict the nonlinear equilibrium path through limit points. The method, known as the RiksWempner method provides the NewtonRaphson method and its modifications with a technique to control progress along the equilibrium path. The theoretical development of this method and its modification can be found in [32,43461.
10.4.3 Tangent Stiffness Coefficients for CLPT
The NewtonRaphson iterative method involves solving equations of the form (10.4.11a) where
3 A: = u 2.) A? = v i , Ai z
=ai

(10.4.11b)
are the coefficients of the submatrices [Tap] defined by
the components of the residual vector {R") are
and na denotes n or m, depending on the nodal degree of freedom. Thus, we have
The only coefficients that depend on the solution are K:?, K?, K g , K g , and K g , and they are functions of only A: = &. Hence, derivatives of all stiffness coefficients with respect to A' = uj and A2 = vj are zero. Thus, we have
n* 8% T ; ~ = ~ ~  A ' + K ? ~ = K T ? ~ =x ? ~ x
y=l k=l
n*
auj
'3
2.7
'
23
_ a y + K ? ? = ~ ? ? z3 y=l k=l
~KZ?
avj
~ T 1 3 =aa; x  ~ ; +aaag;;; / = x _ A 2 + ~ $ x ~
n*
dKZ2
n

21 .
y=l k=l
n*
k=l
T23 ? ~ = ~ ~  A : + K $ = C ~ ~ ; + K $ y=l k=l k=l
a~:? aag aag
n
aK;;

aag
T27. ? = ' ~ ? ~  A ;   + K $ ?
~ = k=l l
n*
8~32
aag
Thus, we must compute the following derivatives of the element stiffness coefficients (only the nonzero parts are shown in the calculation):
dxdy
8wo89;
+'"6
+
(tx (u
awe %)I
dpj"
dxdy
Note that qy is given by combining the expressions in Eqs. (10.4.I8)(lO.4.21) and K?. One may find that the tangent stiffness matrix is sym~netric.This completes the finite element model development of the classical plate theory.
10.4.4 Tangent Stiffness Coefficients for FSDT
Since the source of nonlinearity in the classical and firstorder shear deformation plate theories is the same, the nonlinear parts of the tangent stiffness coefficients derived for the classical plate theory are also applicable to the firstorder theory. For the sake of completeness, they are presented here again. The coefficients of the submatrices [T"P] ( a ,p = 1 , 2 , . . . , 5 ) are defined by
where the components of the residual vector { R a )are given by
and n* denotes n , m, or p, depending on the nodal degree of freedom. Thus, we have
It should be noted that only coefficients that depend on the solution are K?, K?, K g , and Kjy. Since they are functions of only w o (or functions of wJ), derivatives of all stiffness coefficients with respect to u j , u j , and S are zero. ; Thus, we have
s:,
5
n*
~ 1.7 f
i
=
~
auj~
ka
~ 2.7 ~
y = l k=l
+ J ' ~ ~ 2.ll l~ =l x x=a l T : : 1~ y l+ ~ ~ ? = K , !v? Z ~ l.7 y=l k l
5
n*
at),
dx
dx
dx
+
A 62
awo a y 8~
a$,( 2 )
d x d y + K$
ax ay +ay ax
awo
+
A 21
awo W (j 2 ) 8 8 Y Y
NONLINEAR ANALYSIS OF PLATES A N D SHELLS 593 .
A45.594 MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS ax n* y=l k=l dxdy + ~f as.e. The membrane locking can be explained by considering. Clearly. it should experience no axial (or membrane) strain: = 0 (for pure bending) . the Timoshenko beam finite element (see Problem 10. As discussed earlier. When the element is used to analyze pure bending deformation. shear locking is avoided by using selective integration: full integration t o evaluate all linear stiffness coefficients and reduced integration to evaluate the transverse shear stiffnesses (i. yielding displacements that are too small compared to the true solution. that contain ~' A44. the FSDT elements become excessively stiff in the thin plate limit. for simplicity.4. occurs in plates and shells due t o the inconsistency of approximation of the inplane displacements (uo. ~~A~+K?P=K?S 8s.5 Membrane Locking Recall that when lower order (quadratic or less) equal interpolation of the generalized displacements is used.vo) and the transverse displacement wo. all coefficients in K . k‘ zJ V y=l k=l a~:. known as the membrane locking. Another type of locking.12). This type of behavior is known as shear locking (see Reddy [32] and references therein). 10. the tangent stiffness matrix is symmetric.. and A55).
4. the twopoint Gauss rule would yield an exact value for the first term + + arid at the same time the second term . (10.4. the correspondence (lO. when quadratic interpolation of both ILO and wo is used. If uo is interpolated with cubic polynomials and wo is interpolated with quadratic polynomials. when both uo and wo are approximated using linear polynomials.26). threepoint Gauss quadrature yields exact values of both integrals.NONLINEAR ANALYSIS OF PLATES AND SHELLS 595 In order that the above constraint be satisfied for independent approximations of uo and wo.4. then is linear and is quadratic and there is no possibility of canceling the coefficient in quadratic term.26) is again satisfied.26). For quadratic interpolation of uo and wo.4. it is necessary to find a way to avoid membrane locking of the element. however. the term should cancel the second term in Eq.27). is automatically satisfied. Thus. This in turn requires that degree of polynomial of  dx  degree of polynomial of (2) 2 (10.26). for constant E:. the coefficients in the polyriornials get adjusted to satisfy the constraint (10.")""" (see Problem 10. for this case. Also. Thus the exact evaluation of the first term requires N G P = (p 1)/2 = 2 and the second term requires NGP = [(p 1)/2] = 3.A. hence constraint (lO.14): the form in E. It is found that. consider the coefficient (KZ?.. Since quadratic approximation of uo and wo is common in practice. the membrane locking can be avoided by using selective integrations of the terms of For example.. we have % (2)'  duo dz (quadratic)  (quadratic) Thus the constraint (10. where NGP denotes the number of Gauss points. the element does not experience membrane locking for the following two cases: (1) uo is linear and w is linear (2) uo is cubic and wo is quadratic and it experiences locking when both uo and wo are interpolated using quadratic polynomials.4. In summary.4.27) If both variables are approximated with sufficiently higherorder polynomials. the first term is a cubic (p = 3) polynomial and the second term is a fourthorder (p = 4) polynomial. However.
(3) postbuckling response of laminates under inplane compression. Stresses are calculated at the center of the element. For this choice of mesh. This amounts to using an interpolation for w that satisfies the constraint E::. (4) nonlinear transient response of composite laminates. if the mesh of quadratic elements is sufficiently refined.is approximated as the same degree polynomial as the first term.2. Unless stated otherwise. 10. a uniform mesh of 4 x 4 ninenode quadratic elements is used in a quarter plate for the FSDT.5. square plate with Two types of simply supported boundary conditions are studied. 10.1 Preliminary Comments Here we present some numerical examples of laminated plates and shells using the nonlinear shear deformable laminated plate finite element presented in Section 10. A tolerance of t = convergence in the NewtonRaphson iteration scheme to check for convergence of the nodal displacements. The problems presented here illustrate certain features characteristic to composite laminates. o The discussion presented above for the Timoshenko beam element also applies to membrane locking in CLPT and FSDT plate elements. and (5) postbuckling and progressive failure analysis of composite panels subjected to inplane compression. All of the problems are selected from the author's publications. full integration (F) is to use 3 x 3 Gauss rule. Example 10.2 Isotropic and Orthotropic Plates In this section several examples of isotropic and orthotropic plates with various edge conditions are presented to illustrate the use of the CLPT and FSDT elements in the geometrically nonlinear (in the von KBrmAn sense).5 Numerical Examples of Plates 10.1: Consider an isotropic. see Reddy [32]).5. conditions used for SS1 and SS3 are The displacement boundary . The shear is used for correction coefficient is taken to be K. = 516. (2) the use of biaxial symmetry boundary conditions in quarter plate models of rectangular laminates. the membrane locking disappears. = 0. Of course. and additional examples can be found in the references cited at the end of the chapter (in particular. and reduced integration (R) is to use 2 x 2 Gauss rule. These include: (1) the effect of geometric nonlinearity on static deflections. A shear correction coefficient K = 516 is used here.5. The effect of the integration rule to evaluate the nonlinear and transverse shear stiffness coefficients is investigated in the first example.
317 8.1330 (3) 1.6902 (3) 0. The linear FSDT plate solution for load go = 4875psi (or P = 6..9570 (3) 1. The number of iterations taken for convergence are listed in parenthesis.3125.085 14..5.4774 (2) 1.4619 (3) 0. . .25.458 13.8673 (4) 1.1). Using the load parameter.0.085 14. = 0 (symni.5.5186 (3) 0. = u .25.O) and normal stresses if.882 9.25. ( a 2 / ~ h 2 ) t the center a of the first element for various integration rules (also see Figure 10.2813 (3) 0.743 1.002 9.634 13.5681 (2) FR FF Deflections.25.3808 (2) 1.5.162 11.2688 (3) 1.6399 (2) 0.610 16.307 14.5629 (2) 1.1333 (3) 1.2686 (3) 1.984 11.3) SS3: uo = vo = wo = 0 on simply supported edges Uriiforrrlly distributed load of intensity go is used..755 * The first letter refers to the integration rule used for the nonlinear terms while the second letter refers to the integration rule used for the shear terms.1 contains the deflections wo(O.NONLINEAR ANALYSIS OF PLATES A N D SHELLS 597 At y = b/2 : uo = .6911 (3) 0. the 4 x 4Q9 meshes are not sensitive to shear or rnenlbrarie locking.5629 (2) 1.856 3.8687 (3) 2.0) Table 10.0.3151in.779 3.320 8. for SS3.0758 (2) 2.398 15.3809 (2) 1.3125.1: Center deflection w and stresses a.25) is wo = 0.25. P = qOa4/Eh4.4630 (3) 0.2) (10..733 1.299 14. Table 10.6399 (2) Normal stresses.398 15.553 18.O) 0. u.2917in.878 16.305 5.0.6241 (3) 1. .159 11. SS3  SS1 P RR* FR FF RR 0.293 17.2790 (4) 0.u~o & = 0 = (10.462 13.5.779 3. The boundary conditions along the symmetry lines for both cases are given by At z = 0 : uo = 4 = 0.5.634 13. . the incremental load vector is chosen to be { A P ) = {6.6.396 5.396 5. lines) (10.2567 (2) 2. As discussed earlier.610 16. wo (0.5.4) It is clear that SS3 provides more edge restraint than SS1 and therefore should produce lower transverse deflections.572 18.3149 (4) 1. h/2) 1.743 1.4194 (2) 2. . of simply supported (SS1 and SS3) plates under uniformly distributed load.890 16.983 11.300 5. At y = 0 : vo = 4.278 17.861 3.4774 (2) 1. for SS1 and wo = 0. and therefore the results obtained with various integration rules are essentially the same.2780 (3) 0.(0.882 9.9575 (3) 1.001 9.5..12.
simply supported square plates under uniformly distributed load (4 x 4Q9 for FSDT and 8 x 8C for CPT). versus load P for isotropic (v = 0.. SS1 (FSDT) 16 12   stresses 1 250 0 50 100 150 200 Load parameter.1: Plots of center deflection wo versus load P and center normal stress a. F Figure 10.5. .3).
vl2 (10. the C P T deflections were obtained using 8 x 8 mesh of the nonconforming elerrients and the FSDT deflections were obtained with 4 x 4Q9 mesh (mesh of ninenode Lagrange elements).05 is w o = 0.. 1 3 8 i n . The boundary conditions of a clamped edge are taken to be The geometric and material parameters used are the same as those listed in Eq. q = qo=constant) are analyzed.5 is w = 0.5. Plots of load qo (psi) vs..3). i. all generalized displacemerlts are set to zero on the boundary. and a tolerance of t = 0. center deflection w o (in. 2 8 ~ 1 0 ~ psi = 0. .32 GI2 = GI3 = GZ3= 0.3 contains center deflections and stresses for the problem (see [36381).5. The figures also show the results obtained using 8 x 8 mesh of conforming C P T elements.2: Center deflection w and normal stress a.5.3: Here. Table 10.. = a..01140 for SS3.5. E = 1.5) arid subjected t o uniformly distributed transverse load (i.5.) and qo versus normal stress (total as well as membrane) a. Table 10.5). g The linear solution for load qo = 0.3 contains plots of load versus center deflection of an isotropic plate ( h = 0.2 contains the center deflection and total normal stress as a function of the load for SS1 and SS3 boundary conditions.2 for SS1 and SS3 plates. . El = 3 x lo6 psi: E 2 = 1 .e.(a2/E2h2) are shown in Figure 10.01 is used for convergence.. and v = 0.NONLINEAR ANALYSIS OF PLATES A N D SHELLS 599 Example 10. The linear FSDT solution for load qo = 0. for simply supported o orthotropic square plates under uniformly distributed load (4 x 4Q9). h = 0 .h clamped edges. A unifornl mesh of 4 x 4Q9 elements with reduced integration is used in a quadrant. Table 10. A uniformly distributed load of intensity qo is used.5.5.2: Orthotropic plates with a=b=12in.01132 for SS1 and w o = 0.e.28 x lo6 psi. (10.37 x 106psi.5.138 in.0301. SS1 SS3 FSDT  40 CPT w0 FSDT UJ0 CPT w 0 FSDT W0 FSDT ST czr Example 10. Figure 10. The incremental load vector is chosen to be Twelve load steps are used.. we analyze an orthotropic plate wit.5..
as functions of the load go for simply supported. .0 0. orthotropic.0 2. square plates under uniformly distributed load..4 Load.8 1.SS1 (FSDT) Load.2 1.2: Center deflection wo(O.5. P Figure 10.4 0. 40 0.0) and stresses a.6 2.
25 (10.4 for thc gcorrictry and boundary conditions) made of high modulus glassepoxy fiberreinforced material El = 25E2.3: Nonlinear center deflection w versus load parameter qo for o clamped. v12 = 0.5.5.5.3: Center deflection wo and normal stress a.5. Example 10.. Unless stated otherwise. square plates under uniform load.5.7) .0 4 8 12 16 20 24 Load.5E2.2E2. Table 10.5. orthotropic plate (see Figure 10. E2 = x106.3 Laminated Composite Plates In this section examples of laminated plates with various laminations schemes and edge conditions are presented.4: This example is concerned with the nonlinear bending of a square. G2:3 = 0. for clamped orthotropic square plates under uniformly distributed load (4 x 4Q9). G I 2 = G I 3 = 0. orthotropic. all example problems arc analyzed using the FSDT element. 10. 40 Figure 10. simplysupported (SSl).
5.  h2 h/2)E2a2 ' a.5). The linear deflections predicted by the CPT and FSDT elements for P = ( q o a 4 / E 2 h 4 )= 10 and plate sidetothickness ratio of a l h = 10 are zi..0653 a analytical solutions (see Reddy [40]). These values coincide with the h 0. 4 3 7 5 ~ .y(B.4 contains results of the nonlinear analysis for a l h = 10 (also see Figure 10. = ~ ~ ( 0 .4: Nondimensionalized maximum transverse deflections and stresses of simply supported (SS1) square plates.5. h E2a C. 90a h (10. Table 10.4: Geometric boundary conditions for SS1 type simply supported rectangular plates. ( B .Figure 10.5. and subjected to sinusoidal or uniform load. .A). Table 10. = o Y z ( A B)..5. Uniform meshes of 8 x 8 C P T nonconforming elements and 4 x 4 ninenode quadratic FSDT elements in a quarter plate. B . 0 6 2 5 ~ and B = 0 . = u X .0952. 0 ) ( E ~ = ~ / ~ ~and ~w) = 0.8) where A = 0 . The following nondimensionalizations are used: uxy = u.5.
024 in. orthotropic.8315x lo6. Results for both clamped and SS3 boundary coriditioris under uniform load are obtained. Table 10.3125 x 10" ul2 = 0. and FSDT for a/h=100) 0 20 40 60 80 100 120 140 Load parameter.5: This example is concerned with the nonlinear bending of a square. E l = 1 . Table 10. The geometric parameters used are: a = b = 12 in and h = 0. thick). square plates under uniform load.0) = 0.Nonlinear (FSDT.07611 in.5.8). A load i~lcrerrlent of Aye = 0.Cy io2a. E2 = 1. (10.096 in (each layer of 0. Example 10. The rriaxirriuni linear deflection for the clamped case is wo(O.5. 8 2 8 2 ~ lo6. G12 = G13 = G23 = 0. synmetric crossply (0/90/90/0) laminated plate made of layer properties.c.5.6 contain nondimensionalizcd deflections and stresses for the plate. P Figure 10.5.5.2395.5.04102 in and for SS3 it is ulo(O. Clamped Plate  Simply Supported Plate YO lowo ~ Y Y UX Y ioa. a / h = l O Linear (FSDT) (CLPT.0) = 0. A mesh of 4 x 4Q9 FSDT elements in a quarter plate is used. The stresses are nondimensionalized as in Eq.  10~1~ UV 9 if..5 and Figure 10.5: Nonlinear center deflection w versus load parameter qo for simply o supported (SSl). .5: Maximum transverse deflections of clamped and simply supported (SS3) crossply (0/90/90/0) square plates under uniform load.2 psi is used.
6 2.0 ) = 0. h = 0.5. If the symmetry conditions implied by the Navier solutions are used in the linear finite element analysis of a quarter plate. E x a m p l e 10.6: Here we consider nonlinear bending of square.22683 in. E2 = l o 6 . The maximum linear deflection for two layers (0190) is wo(O.. ~ 1 = 0. uniform mesh of 4 x 4Q9 F S D T elements in a quarter plate is used. G23= 0. Again. Table 10.25 2 (10. the Navier solutions of the linear theories can be developed for antisymmetric crossply plates with SS1 boundary conditions and antisymmetric angleply laminates with the SS2 boundary conditions.wolayer laminate (for the same total thickness of the laminates). antisymmetric. as was .5.7). The Navier solutions can be used to determine the conditions on deflections and forces along the lines of biaxial symmetry.3 in. and for six layers ( 0 / 9 0 / 0 / 9 0 / 0 / 9 0 ) it is wo(O. 6. along the lines x = a / 2 and y = b/2 of a rectangular plate of dimension a x b and with the origin of the (x.0 2.5E2.5.0 0. . 10. 0. i.5. A load increment of Ago = 200 psi is used.8 1.O) = 0. Note that the sixlayer laminate is relatively stiffer than the t..) laminated plates made of material E l = 40 x l o 6 .4 Effect of Symmetry Boundary Conditions on Nonlinear Response As discussed in Chapters 5.5. y) coordinate system being at the lower left corner of the plate.096 in.2 1. G12 = G13 = 0. and 7.6: Loaddeflection curves for symmetric crossply ( 0 / 9 0 / 9 0 / 0 ) laminates. 40 Figure 10.08669 in.4 0.9) The geometric parameters used are: a = b = 12 in and total thickness h = 0.5.4 Intensity of t h e distributed load. crossply (019010190. a = b = 12 in.6E2.6 contains results of the nonlinear analysis (also see Figure 10. Results for clamped boundary conditions under uniform load are obtained.e.
with those of the corresponding fillplate models (see Reddy 15111. in general. aritisyrnmetric angleply (45145) square larriiriate (a = 1000 riirn.5. antisymmetric crossply (0/90/0/90/ .Linear ( 6 layers' Nonlinear. results do not agree. CC [FSDT. To illustrate this point.) square plates under uniform load. .3 in. Whcn quarterplate models with the geometric boundary conditions implied by the Navier solutions on the lines of symmetry are used in the nonlinear finite element analysis. h = 2 mrn). one obtains correct full plate solutions. . a twolayer. .5. a = b = 12 in ( a l h = 40) Intensity of the distributed load. Table 10.6: Maximum deflections of twolayer and sixlayer (0/90/0/90. . crossply done in Chapter 9.) laminated square plates under uniform load. q. Figure 10. (0190j1 h = 0. .7: Loaddeflection curves for clamped. under uniform transverse load is considered.
5 4.25 (10. W O (mm) Figure 10. The following boundary conditions along the lines of symmetry were used: where the coordinate system is fixed at the lower left corner of the laminate. The discrepancy increases with the intensity of the transverse load.5. ul:! = vl3 = 0.5. Note that the force boundary conditions are included in the finite element model in an integral sense.0 1.0 Maximum deflection. the symmetry I l l I I I I I I 1 1 1 1 I +Full plate model .0 0.Quarter plate model 0 I 1 0.0. G12 = Gl3 = 10 GPa G23 = 4 GPa. Meshes of 2 x 2 and 4 x 4 ninenode quadratic elements based on the firstorder shear deformation plate theory are used to model the quarter and full plates. respectively. As noted earlier. It is clear from the results that the use of a quarterplate model with the symmetry conditions (10.10) The loaddeflection curves obtained from the quarterplate and fullplate analyses are shown in Figure 10.5 2.11) yields larger deflections than those obtained from the fullplate model.8.5 1.5.0 3.The following layer properties are used: El = 250 GPa.11) used to model the quarter plate.0 2.5. This discrepancy can be explained in the light of the symmetry conditions (10. E2 = 20 GPa.5 3.8: Loaddeflection curves (A vs. .5. wo) fullplate and quarterplate for models of simply supported (SS2) antisymmetric angleply (45145) laminates.
(x..n=1 mnx sin cos niry a b ' wo = m. indicate that the force have boundary conditions in (10.11) are not satisfied.. = m.. The quarterplate model with zero inplane forces N. b/2) =A12 rn27r2 2a' wL. nny Wmn sin sin a b 0 0 $1:= m. (10.n=l mnx nny cos sin a b mnx . in which the inplane forces are not taken to be zero 011 the lines of symmetry. for the nonlinear case... in the finite element analysis it is implied that the natural boundary conditions Nzz = 0 on x = a/2 and NYv= 0 on y = b/2 are specified.. 0 0 m. b/2) simulates the plate as more flexible than the fullplate model. Bll = BIZ= B22= B6g = 0.. and Dl6 = D26 = 0) A16 = A2g = The expressions in Eqs.5. a 4. a quarterplate model with the symmetry conditions implied by thc Navier solution gives the same solution as the fullplate . 0 0 mnx nny cos sin . For the angleply case.3.n=l s. For antisyrrinletric crossply laniiriates.conditions are derived from the Navier solution for the linear theory..5. N. 0 3 m. and NYV the form N.(a/2. cos mnx # 0 a When a quarter plate model is used without specifying u on line x = a12 and v" on o line y = 612.n=l v" = x v~. (7. and moment MZy are given by (note that = 0. the assumed solution is of the form [see Eqs. For example. rnnx nny sin cos a b The resultant forces NT:. and N.n=l c s&. y ) and Nyy(x.2)] s = o x u.13ac).
and the same mesh is used to determine the nonlinear response under applied inplane compressive ~.52E2 For SS2 type simply supported boundary conditions. h = 2 mm El = 40E2. the uniaxial buckling load can be determined analytically (see Chapter 7).6 Nonlinear Response of Antisymmetric CrossPly Laminated Plate Strips Unlike isotropic metallic plates. Here we use 4 x 4 mesh of ninenode elements in the full plate to determine the critical buckling load. depending on the lamination scheme and boundary conditions (see [24."~ 6. and eightlayer laminates.5. For example.5.9 contains a plot of the maximum outofplane deflection wo (mm) versus load parameter X (N& = 10. G23 = 0.82E2. The critical buckling loads can also be determined from geometric nonlinear analysis.5 Nonlinear Response Under InPlane Compressive Loads Another interesting characteristic of composite laminates is their behavior under compressive loads.5. the geometric nonlinear effects could be very significant even at small loads and deflections. composite plates exhibit quite different nonlinear behavior. The SS1 type simply supported boundary conditions and 2 x 2 mesh of ninenode elements in a quarter plate are used to determine the critical buckling loads No by eigenvalue analysis and loaddeflection curves in the nonlinear analysis under inplane load = N. This is due to the fact that the zero force boundary conditions are satisfied in an integral sense for crossply laminates. the crossply laminates do not exhibit clear limitload behavior. load Nyv = x ~ N . The critical buckling loads are indicated on the loaddeflection curves for comparison. 10. Figure 10. The geometry and materials properties used are the same as those used for the angleply laminate..25) N/m). " where Xo is the critical buckling load determined from the eigenvalue analysis.25. Next we consider antisymmetric crossply laminates. E2 = 6. Most often the critical buckling loads are determined through an eigenvalue analysis. It is clear the loaddeflection curve exhibits a limit point. and subjected to uniformly distributed transverse load. To illustrate the point we analyze an antisymmetric crossply square laminate (9010) with two opposite edges pinned (uo = 0) or hinged (uo # 0) and the other two edges free. = XONo (N. Figure 10.model for both linear and nonlinear theories.52]).5. six.85) N/m). GI2 = = 0.10 contains loaddeflection curves for YV two. First we consider an angleply (45145) laminate with the following geometric parameters and material properties: a = b = 1. Unlike the angleply laminates. where the critical buckling load is taken to be the socalled limit load.000 mm. . Sour.25 GPa. 10. which is the same as the critical buckling load determined from the eigenvalue analysis.
5. . 0.0 Maximum deflection.10: Loaddeflection curves (A vs. .[ I l l (45/45).6 Maximum deflection. 0.0 3. .0 4.) laminate under uniforrrily di~tribut~ed inplane compressive edge load.2 0.5 0.0 W 0 Eigenvalue 1.0 0. Nonlinear analysis Eigenvalue analysis . .3 0. ~ u plate ll I.4 0. . wo) a simply supported (SSof 1) antisymmetric crossply (0/90/.0 Figure 10.0 2.9: Loaddeflection curves (A vs.5.1 I 0 . w o Figure 10. ss2. / 0. w o(mm) 5. wo) of a simply supported (SS2) twolayer antisymmetric angleply (45145) laminate under uniformly distributed inplane compressive edge load.
850 4. the solution is independent of the sign of the applied load. is positive.847 0. is not zero.70 11.35 4.665 3.56 59.82 30.50 35.705 2.550 :1.33 *For negative load values.3.50 0. the nonlinear solution is larger than the linear solution for small values of the load. The material properties and geometric parameters used are El = 20 msi.88 2. For a pinned plate strip. G12 = GIS = G23 = 0.00 94. it is . Therefore. the axial force N .078 2.005 0. the axial force N .5. this yields For a negative load.55 3.02 0. Load PO 0. .00 141. finite element mesh.17 24. becomes positive.56 59.0 3.710 2.82 30.  For small values of the positive load.37 4. and the two terms in N..03 0.075 2.4 msi. add up.255 0.609 1. the expression containing the All coefficient is small compared to the expression containing the Bll coefficient.190 7.5.710 2.402 3.0 5. and N.159 0.19 7.0 4. The load deflection curves for the first few load steps are shown in Figure 10.00 75.7 msi The results of the nonlinear analysis are presented in Table 10.470 0.12.10 0.920 16.00 Hinged Nonlinear* 0. the plate strip is essentially in pure bending and hence .480 0. E2 = 1.675 onl line art 0. is compressive and increases the transverse deflection analogous to the transverse deflection of a plate strip under an axial compressive load and a transverse load.075 3.327 1.16 24. As the load is increased. t For positive load values.25 47.117 3.87 35.5. Hence. N.100 1.845 1.618 0.33 Case 2t 0.69 49.525 3.92 16.05 0.870 2.7: Transverse deflections.04 0. = 0. the All expression becomes larger than the BI1 expression.7.11. For hingedhinged boundary conditions.532 3.65 68.940 1..429 0. of cylindrical bending of a (9010) laminate under uniformly distributed transverse load.370 4.332 2.954 1.5.87 35.235 0.00 188.41 1. Thus. and boundary conditions for the pinnedpinned and hingedhinged cases are shown in Figure 10..034 1.00 235. which is negative for 0 < x < 4.555 0.00 75.65 68. % Table 10.5.0 Pinned Linear* 0.858 1.386 0.75 23. wo/h.125 Case l * 0.475 0.The geometry. Therefore the deflection is lower than that for the case of positive load.75 1.673 0. y 2 = 0.25 0. Bll a larger axial force and therefore a stiffer structure than for the positive load case.0 2.858 1. This stiffens the structure and the nonlinear solution becomes smaller than the linear solution.429 0.01 0.69 '19.233 1.
0.02 0. . loading.12: Loaddeflection curves (qo vs.h Pin supported (u = 0) Roller supported (uo+ 0 ) +a+ +a+ Figure 10.05 Load.5. q o (lbslin) Figure 10.01 0.03 0.11: Geometry. wo) cylirldrical bending of a for crossply (9010) plate strip with pinnedpinned edges. and boundary conditions used for cylindrical bending of a crossply plate strip.5.04 0.00 0.
. The iterative loop for convergence of the solution is the innermost. Here we present a few examples from this paper. The following geometric.. The first example involves a simply supported isotropic square plate subjected to suddenly applied uniformly distributed transverse load.80 1.30 2 0. In the nonlinear transient analysis..547 x Ns2/cm4 Figure 10.5 m) . The amplitude increases while the period of response decreases with an increase in the intensity of load.10. . a k 0. U G  I 0. the reader may consult 1201. and load parameters were used: a = b = 243.5 ms) . Load versus the maximum deflection is also shown in the figure. suddenly applied transverse step loads are considered. t ( s ) 120 150 40 (At=5.60 0. I I 1.20 0..7 Transient Analysis of Composite Plates Nonlinear transient response of laminated composite plates was reported by Reddy 1201.2 1. .8 cm..O ms) 5q0 (At = 2.90 0 .13: Center deflection versus time for nonlinear transient analysis of an isotropic.O ms) 2q0 (At=5.8 1.6 Deflection. .4 0. followed by loops on time increments and the load increments. p = 2.30 0.13 shows the center deflection w as a function of time for four different o values of the load and two different time steps.Ei 0.90 0 30 60 90 Time.. simply supported.00 a 0. This problem may serve as a reference for verification of the geometric nonlinear option of a finite element program.5. For additional examples.. square plate. there are three loops. h = 0..50 r p \ I ' I l l 1 1 I ' l l 10 3 U 1.5....  Figure 10.635 cm.0 0.. wo (cm) a . In all examples discussed here.....60 C.5. 10qO (At=2. and the initial conditions are taken to be zero. material.
Functionally gradient materials (FGM) are a class of corrlposites that have a cont.inuous variation of material properties from one surface to another and thus alleviate the stress concentrations found in laminated composites. and stress concentration factors. matrix cracking.2.1. we consider nonlinear transient analysis of simply supported crossply (0190) and angleply (451 45) plates. They are typically manufactured from isotropic corriporients such as metals and ceramics since they are mainly used as thermal barrier structures in erivirorirnerlts with severe thermal gradients (e. R.5E2. p = 2.5. y) 0. i. turbines and other machine parts are susceptible to failure from buckling.1 was used. The effect of m geometric nonlinearity is obvious from the figure.25 2 (10. Background While laminated composite materials provide the design flexibility to achieve desirable stiffness and strength through the choice of lamination scheme.19) were used.5.eferences 5469 provide a background and insights into thermorneclianical and transient analysis of FGM structures. or excessive stresses induced by thermal or combined thermomechanical loading. including thermal shock a situation that arises when a body is subjected to a high transient heating or cooling in a short time period. Thinwalled members.8 cm. Figure 10. (10. A nonuniform 4 x 4 mesh of ninenode quadratic elements in quarter plate was used.e. in English units of inches and pounds) under uniforndy distributed transverse patch loading of intensity qo = over the area 0 (x. The material properties were assumed t o be El = 25E2. the anisotropic constitution of laminated composite structures often results in stress concentrations near material and geometric discontinuities that car1 lead to damage in the forrn of delamination. The three degrees of freedom element models the plate stiffer than the five degrees of freedom element. meanwhile the metal provides the strength arid toughness.5. Figure 10.  .2E2.5.2. The gradual variation results in a very efficient material tailored to suit the needs of the structure and therefore is called a functionally graded material. G23 = 0. angleply \45/45) plates under uniformly distributed pressure loading (qo = 50 x 1 0 . The material properties in Eq.~ ~ / c ).14 coritairis plots of center o deflections ( G = w o x lo3 and 6 = w x lo2) of simply supported (SS1) crossply laminated rectangular plate (a = b = 1.635 cm.h = 0.g.69] for plates is presented here.0. A brief review of the work carried out in [54. ~ 1 = 0. residual stresses. < < A time step of At = 0.5. thermoelectric devices for energy conversion.Next. 10.15 contains plots of center deflection versus time for simply supported (SS2).547 x 10V6~s2/crn4). serriicorlductor industry). Figure 10. p = 1. h = 0. and adhesive bond separation. plates and shells. In such applications the ceramic provides heat and corrosion resistance. large amplitude cicflcctions.6 Functionally Graded Plates 10. used in reactor vessels.19) 4 .. The main applications of functionally gradient materials have been in high temperature environments.6. The gradation in properties of the material reduces thermal stresses. square (a = b = 243.14 contains results of both five and three degrees of freedom models.. Glz = Gl3 = 0.
Time.0 3. h = 0.DoF = 3) ..LO...UI.0 6.psi. nonlinear (qo..0 Time.5.0 F i g u r e 10.. t (s) 5.14: Center deflection versus time for nonlinear transient analysis of a simply supported (SS1) crossply (0190) plate. linear (qo.DoF = 3: Isotropic plate . DoF = 3) . &=wox102  Crossply (0190) plate U). nonlinear (go.15: Center deflection versus time for nonlinear transient analysis of a simply supported (SS2) angleply (45145) square plate under suddenly applied uniformly distributed transverse load.0 1.DoF = 5 ) .L. .DoF = 31 0. t (ms) F i g u r e 10.2 in qo= u.0 4.5.=woxlO3 . nonlinear (4q0. nonlinear (4q0.&.0 2.
Here we assume that moduli E and G. the governing equations of motion as well as the finite element models derived for the CLPT and FSDT are valid for the FGM plates. Suppose that a typical material property P is varied through the plate thickness according to the expressions (a power law) where Pt and Pb denote the property of the top and bottom faces of the plate. { M ~= } /f 2 {IJI}T(z) dz z (10. as the properties of the ceramic and metal.) dz. The above power law assumption reflects a simple rule of mixtures used to obtain the effective properties of the ceramicmetal plate. However. not all Bij = 0). (10. and to the case in which the temperature field is known. Hence.6. Further. thermal coefficient of expansion a . and n is a parameter that dictates the material variation profile through the thickness.e. while v is assumed to be a constant.3a) The plate stiffriesses are given by . The metal content in the plate increases as the value of n increases. exhibiting bending stretching coupling (i. density p.2 Theoretical Formulation Consider a plate of total thickness h and made of an isotropic but inhomogeneous material through the thickness of the plate. respectively. We take Pi = PC and Pb = P.1). The value of n = 0 represents a fully ceramic plate.10.. small strains and displacements.6. the temperature distribution through the thickness must be calculated by solving the equation The temperature field T ( z ) is then used in computing the thermal forces and moments { N ~= } /f 2 {BIT(. respectively. and thermal conductivity k vary according to Eq. The throughthickness functionally graded plate is an inhomogeneous (through the thickness) isotropic plate. we restrict the formulation to linear elastic material behavior.6.
Therefore. for the boundary conditions given in (lO.b). this term is included in the stiffness matrix.For the purpose of computational efficiency. Next. (10.6.2a) is of the form (see Reddy P I Due to the dependence of the conductivity k on z.3 Thermomechanical Coupling The finite element model associated with Eq. The modulus E and the thermal coefficient of expansion a . vary through the plate thickness according to Eqs. respectively. is a nonlinear function of z. we wish to examine the contribution of the temperature field to the nonlinear finite element equations. (10. 'm' and 'c' correspond to the metal and ceramic. 10.la. the temperature distribution through the thickness of a FGM plate.6.6. The thermal contributions to the finite element equations associated with the five generalized displacements are: The thermal contribution associated with 6wo is nonlinear in wo.where quantities with subscripts.6. and the elastic coefficients Qij.2b). .
bending under applied temperature field is studied. reduced integration is used in the numerical evaluation of the nonlinear and the shear terms of the stiffness matrix (see Reddy [32]).01 n m.4. In order to avoid membrane and shear locking. v = 0. Because of the biaxial symmetry of the problem.3 and 10. showing the stiffening effect due to the development of inplane forces that make the plate stiffer with increasing load. A regular mesh of 4 x 4 fournode elements is used.6. a = 23 x 10K"c The plate considered is a square plate with side a = 0.tern1 of the stiffness matrix can he expressed as (superscript 2 on $. One may note that the nonlinear deflections are smaller than the linear ones. As expected. Next. 7 = wo/h and load parameter P = q o a 4 / ( ~ m h 4 )First.2 r and thickness h = 0.6.6. The metal surface is exposed to 20°C and the ceramic surface is exposed to fixed but different temperatures. Figures 10.707Kg/m3. Thus. bending of FGM plates under transverse mechanical load is investigated. but the purpose is to establish the bounds for the FGM analysis. p = 2. k = 204 W/mK.6.3. Note that the value of powerlaw index n = 0 corresponds to the ceramic plate and n i oo corresponds to the metal plate. the deflection response of FGM plates is intermediate. The metal is taken to be aluminum and the ceramic used is zirconia.1 and 10. The nondimensionalized quantities used in reporting the results are: center 3 . The melting point of pure aluminum is 600°C and that of zirconia is 2600°C. using 0 to 600°C for aluminum plate is not realistic (the modulus and other properties of aluminum will change long before its temperature reaches 600°C). The properties for the two materials are listed below. Also. whose melting point is about 1900°C. is omitted for simplicity) The K" 10. both for linear and nonlinear response. Typical property variations as well as the temperature variations through the thickness for various values of n are shown in Figure 10. to that of the ceramic (stiffer) and metal (softer) plates. aluminum reacts with oxygen and forms aluminum oxide. E = 70 GPa.4 Numerical Results Numerical results are presented for ceramicmetal FGM plates. deflection. respectively. the computational domain is taken to be the positive quadrant.2 contain plots of nondirnensionalized deflection w versus the load parameter P for sirnply supported plates for various values of the powerlaw index n under distributed transverse load. The boundary conditions considered are all sides simply supported (SSI). .6.
P Figure 10. .6.Load Parameter.6. P Figure 10.2: Nondimensionalized center deflection for a simply supported aluminumzirconia FGM plate for various values of'volume fraction exponent (mechanical load and nonlinear analysis). I I I I I I I I I I I I 5 10 15 20 I I 25 30 Load Parameter.1: Nondimensionalized center deflection for a simply supported aluminumzirconia FGM plate for various values of volume fraction exponent (mechanical load and linear analysis).
6.Volume fraction function.6.. .7)]. This is again due to the fact that the inplane forces developed due to the geometric nonlinearity are negated by the thermal forces and moments [see Eq.e.6. thermal forces as well as bending moments) that develop in FGM plates are smaller than those of the monolithic plates. n=O2 trn=IO n=20 ++ metal Temperature (in " C) Figure 10. Another interesting observation is that the nonlinear deflections are larger than the linear deflections under thermal loads (for FGM as well as for monolithic plates).5 and 10. The intermediate behavior observed for mechanical loads is not present in the thermal load case. The FGM plates experience less transverse deflections due to the thermal forces than their monolithic counterparts. (lO. Plots of the nondimensionalized deflection as a function of the temperature of the ceramic surface are presented in Figures 10. respectively. This is due to the fact that the thermal resultants (i. linear or nonlinear analysis. V ( z ) Figure 10.6.3: Variation of the material property.6. making the overall plate stiffness reduced.6 for linear and nonlinear analysis.4: Variation of the temperature through the plate thickness.
Tc[OC] 500 600 Figure 10.6.6.6: Nondimensionalized center deflection for a simple supported aluminumzirconia FGM plate for various values of volume fraction exponent (thermal load and nonlinear analysis). 100 200 300 400 Ceramic Temperature.5: Nondimensionalized center deflection for a simple supported aluminumzirconia FGM plate for various values of volume fraction exponent (thermal load and linear analysis). .1 1 I I I 0 100 3 00 400 200 Ceramic Temperature.+ Ceram~c 0 n=02 n=l t n=05 t n=2 x Metal " 1 0. Tc[OC] 500 600 Figure 10.
7. . and is the nonlinear contribution to the equilibrium equations due to the von KBrmAn nonlinear strains.1 for the coordinate system used): where p being the mass density.7. Here we present a brief development of the finite element model [71]. Figure 10. The equations of motion of the firstorder shear deformation shell theory (see Chapter 8) are summarized here for the case Co = 0 (see Figure 10.7.1: Geometry and coordinate system of a doubly curved shell.10.1 Governing Equations The finite element model of the laminated shallow shell theory with the von K&rman nonlinear strains can be developed in the same way as for the plate (or Sanders [70]) element.7 Finite Element Models of Laminated Shell Theory 10.
(10.1)(10. (10. The stress resultants are related to the strains (in the absence of thermal and other influences) 10.It should be noted that the nonlinear straindisplacement equations of the Sanders nonlinear shell theory [70] are modified here for shallow shells by omitting the nonlinear terms of the form Otherwise.2 Finite Element Model The weak forms of Eqs..7.4.Nzy). and N~ replaced by Nxx.7.7.1)(10. (9. The finite element model is of the form . with the understanding that XI = x.4) will contain additional nonlinear terms (see Reddy [32]).Nyy.lae). 2 2 = y and Nl = Nxx.7.5) were presented in Eqs. the governing equations in Eqs.7. etc.
where the linear stiffness coefficients.7) and (9. (9. are the only ones that have additional nonlinear terms when compared to the plate element: .7). mass coefficients and force coefficients are as defined in Eqs.8ac).4.4. only the stiffness coefficients that contain nonlinear terms are given here. Since all of the linear stiffness coefficients are already defined in Eq. Note that K. (9.4.
.
The following boundary conditions are used: SSl:Atx=a: uo=wo=&=O.. R = lo3 in. qo. twolayer (0/90). The second example consists of a clamped (CCI).. = 4. .2 contains the loaddeflection response of the shells for the two sets of material properties. as functions of the load qo. Figure 10. and h = 1 in.The tangent stiffness coefficients can be computed as in the case of plates.7. cylindrical shell panel under uniform load qo.=0 CC1: vo = uo = wo = 4. respectively) : The first example is concerned with the bending of a simply supported (SS1).7. The geometric parameters used are: a = b = 50 in. The following two sets of material properties of a lamina are used (labeled as Material 1 and Material 2. 10. At y = b : vo=wo=q5. .). spherical panel under uniform transverse load. = 0 along the clamped edges Along the symmetry lines the normal surface displacement and normal rotation are set to zero (for the crossply laminates discussed here). . ninelayer (0/90/90/ . Material properties used are those of Material 1.3 Numerical Examples Here we present two numerical examples.. normal stress ayy and transverse shear stress a.3 contains the center deflection wo.7. Figure 10.
h = 2.54in.. normal stress and transverse shear stress versus load for clamped crossply (0190) laminated cylindrical shell panel under uniform load.) laminated spherical shell panel under uniform load. ..wdh Figure 10.. . l rad. R = lo3 in.7. Figure 10.) spherical shell panel A Material 2 h = 1 in.. a = b = 50 in.7. .. 8 = 0 .. Deflection. . $ R = 1 0 .9layer (0190101901. o. x (lo2) at x = 227. a l h = 1 0 0 .3: Center deflection.2: Center deflection versus load parameter for simply supported crossply (0/90/90/ .3 in.
it is called the updated Lagrangian description.). The magnitude of load increments should be such that the computational method used is capable of predicting the deformed configuration at each load step. displacements. and C2. A left superscript on a quantity denotes the configuration in which the quantity occurs. . strains. is that used by Bathe [31]. It is assumed that all variables.8. . . and the geometry and the displacement fields of the structure are directly discretized by imposing certain geometric and static constraints t o satisfy the assumptions of a shell theory (see [21. 2 .10. C1.1 Introduction The plate and shell finite elements developed in Chapter 9 and previous sections of this chapter were based on laminated plate and shell theories. Such theories are limited to geometrically linear analyses and nonlinear analysis with small strains and moderate rotations. namely.1). C1. such as the displacements.. the total .. A practical way of determining the final configuration C from a known initial configuration Co is to assume that the total load is applied in increments so that the body occupies several intermediate configurations. The finite element model to be developed in this section is based on 3D elasticity equations. the left subscript may not be used. and the accumulated deformation of the body from Co to C1 can be arbitrarily large but continuous (i. For example. prior to occupying the final configuration. strains. and assume that the body may experience large displacements and rotations. neighborhoods move into neighborhoods). The left superscript will be omitted on incremental quantities that occur between configurations C1 and C2. We wish to develop a formulation for determining the displacement field of the body in the current deformed configuration C2. We consider three equilibrium configurations of the body. If the initial configuration is used as the reference configuration with respect to which all quantities are measured. Consider the motion of a body in a fixed Cartesian coordinate system. It is difficult to determine the final configuration of a deformed body subjected to loads with large magnitude. Co. When the quantity under consideration is measured in the same configuration in which it occurs. If the latest known configuration CiPl is used as the reference configuration. .8. Thus i Q indicates that the quantity Q occurs in configuration C. stresses. (i = 1 . The three configurations of the body can be thought of as the initial undeformed configuration Co. and CiPl as the reference configuration. and a left subscript denotes the configuration with respect to which the quantity is measured. the Lagrangian description of motion can use any of the previously known configurations Co.e.8 Continuum Shell Finite Element 10. which correspond to three different loads (see Figure 10. In the determination of an intermediate configuration Ci. and the current deformed configuration Ca to be determined. it is called the total Lagrangian description. Here we use the total Lagrangian description to formulate the governing equations of a continuum. but measured in configuration CJ. The development presented here is based on the material in Chapter 9 of the author's nonlinear finite element book [32]. C. the last known deformed configuration C1. The notation used for positions. . .72401). It is assumed that the deformation of the body from C1 to Cg due to an increment in the load is small.2332. etc. and stresses are known up to the C1 configuration.
Figure 10.2 Incremental Equations of Motion The principle of virtual displacements requires that the sum of the external virtual work done on a body and the internal virtual work stored in the body should be equal to zero: where 6 2~ denotes the virtual work done by applied forces .8. displacements of the particle X in the two configurations C1 and C2 can be written as and the displacement increment of the point from C1 to C2 is 10.8.1: Three different equilibrium configurations of a body.
we simplify the virtual work statement (10. In a large deformation analysis special attention must be given to the fact that the configuration of the body is changing continuously.and d 2~ denotes the surface element and 2f is the body force vector (measured per unit volume).ui and ui). (9.8. 2eiJ is the infinitesimal strain tensor and .6). (10.321: where fi and i t i are the body force and boundary traction components referred to the configuration C o Using Eqs. we note that (see Eqs. We use the following identities 131. The stress and strain measures that we shall use are the 2nd PiolaKirchhoff stress tensor S and the GreenLagrange strain tensor E.3. First. Hence.lb) we arrive a t where Next.3.8.8. which are "energetically conjugate" to each other (see [32]).la) cannot be solved directly since the configuration C2 is unknown. This change in configuration can be dealt with by defining appropriate stress and strain measures.8. In the total Lagrangian formulation. This is an important difference compared with the linear analysis in which we assume that the displacements are infinitesimally small so that the configuration of the body does not change.) = 0 because it is not a function of the unknown displacements.3)(10. (10. all quantities are measured with respect t o the initial configuration Co. Equation (10.t is the boundary stress vector (measured per unit surface area) in configuration Ca The variational symbol '6'is understood to operate on unknown displacement variables (. 2a is the Cauchy stress tensor.5) in Eq.15) and (9.l a ) must be expressed in terms of quantities referred to the reference configuration. (10.16) of 1321) where ~(AE.8.8.. the virtual work statement in Eq. .
The virtual strains are given by
Substituting Eqs. (10.8.8) for G ( ; E ~ ~ ) and using the decomposition
2 s . .  1 . ()Sij s + . 0 211  0 211
for ;Sij into Eq. (10.8.6), we arrive at the expression
where is the virtual internal energy (in moving the actual internal forces through virtual displacements) stored in the body at configuration C1
~(AR)
Since the body is in equilibrium a t configuration C1, by the principle of virtual work applied to configuration C1 we have
and therefore 6(bR) =
1
0
v
QfiSui d
OV
+
(10.8.14)
We need only to replace oSijin terms of the strains and ultimately the displacement increments using an appropriate constitutive relation. The first term of Eq. (10.8.11) represents the change in the virtual strain energy due t o the virtual incremental displacements ui between configurations C1 and Cz. The second term represents the virtual work done by forces due to initial stresses ASij. The last two terms together denote the change in the virtual work done by applied body forces and surface tractions in moving from C1 to C2. This is primarily due t o the geometric changes that take place between the two configurations. Equation (10.8.11) represents the statement of virtual work for the incremental
deformation between the configurations C1 to C2, and no approximations are made
in arriving at it.
For dynamic analysis, the principle of virtual displacements (10.8.11) can be written as [32]
where 6 ( 2 ~ i= 624. )
10.8.3 Continuum Finite Element Model
Equation (10.8.15) can be used to develop the nonlinear displacement finite element model for any continuum. The basic step in deriving the finite element equations for a shell element is the selection of proper interpolation functions for the displacement field and geometry. In the case of beam and shell elements, the approximation for the geometry is chosen such that the beam or shell kinematic hypotheses are realized. First we derive the finite element model of a continuum and then specialize it to shells [24261. It is important that the coordinates and displacements are interpolated using the same interpolation functions (isoparametric formulation) so that the displacement compatibility across element boundaries can be preserved in all configurations. Let
where the right superscript k indicates the quantity at nodal point k, gk is the interpolation function corresponding to nodal point k , and n is the number of element nodal points. Substitution of Eqs. (10.8.16) and (10.8.17) in Eq. (10.8.15) yields the finite element model of a 3D continuum
where {Ae) is the vector of nodal incremental displacements from time t to time t+& in an element, and {be),[KL]{A~), [KNL] {Ae), and A{F) are obtained by evaluating the integrals, respectively:
A [MI
A
Various matrices are defined by ~[KL] =
1
LA
~ [ B L ] ~ o[C]@L]
~ [ B N L o[S] ] ~
dOv
dOv
(10.8.19a) (10.8.19b)
~ [ K N=] L
ly
A[BNL]
and L In the above equations, ~ [ B L ] ~ [ B ~ are] the linear and nonlinear straindisplacement transformation matrices, o[C]is the incremental stressstrain material property matrix, h[S]is a matrix of 2nd PiolaKirchhoff stress components, :{s} is a vector of these stresses, and '[HI is the incremental displacement interpolation matrix. All matrix elements correspond to the colifiguration at time t and are defined with respect to the configuration at time t = 0. It is important to note that Eq. (10.8.18) is only an approximation to the actual solution to be determined in each time step. Therefore, it may be necessary to iterate in each time step until Eq. (10.8.15), with inertia terms, is satisfied to a required tolerance. The finite element equations (10.8.18) are secondorder differential equations in time. In order to obtain numerical solutions at each time step, Eq. (10.8.18) needs to be converted to algebraic equations using a time approximation scheme, as explained in previous sections. We have
where {A} is the vector of nodal incremental displacements at time t , { A ) = t+At{A} ' { A ) ,and 
2 { ~ = 2 { ~ )}
;{F}
+ ;[MI
(a3 ' { A ) aa ' { A } a5 "(6)
+
+
(10.8.21b)
Once Eq. (10.8.20) is solved for { A ) at time t vectors are obtained using
t+At
+ At, the acceleration and velocity
(10.8.23)
{ A )= as{A) a4 t { ~ ) a5 ' { A ) t+At{h){ A )+ a1 '+ At{A} a2 ' { A } =' +
where a1 = aAt and a2 = (1  a ) A t . The finite element equations (10.8.20) are solved, after assembly and imposition of boundary conditions, iteratively at each time step until Eq. (10.8.15) is satisfied within a required tolerance. The NewtonRaphson method with RiksWempner algorithm (see Reddy [32]) is used in the present study.
10.8.4 Shell Finite Element
The FSDT shell finite element can be deduced from the 3D continuum element by imposing two kinematic constraints: (1) straight line normal to the midsurface of the shell before deformation remains straight but not normal after deformation; ( 2 ) the transverse normal components of stress are ignored in the development. However, the shell element admits arbitrarily large displacements and rotations but small strains since the shell thickness is assumed not to change and the normal is not allowed t o distort [31,32.78,79]. Consider the solid 3D element shown in Figure 10.8.2. Let (t, ) be the q curvilinear coordinates in the middle surface of the shell and be the coordinate in the thickness direction. The coordinates (C, rl, C ) are normalized such that they vary between 1 and +I. The coordinates of a typical point in the element can be written as
<
where n is the number of nodes in the element, and $ k ( C , q ) is the finite element interpolation function associated with node k. If +n(<, q ) are derived as interpolation functions of a parent element, square or triangular in plane, then compatibility is achieved at the interfaces of curved space shell elements. Define v = (x;)top  (xt)bottorn. $ = v.~/Iv~l ; i (10.8.25) where v!j is the vector connecting the upper and lower points of the normal at node k . Equation (10.8.24) can be rewritten as
E,
;ode k
Figure 10.8.2: Geometry and coordinate system of a shell element.
where hk = I v ~ I is the thickness of the shell element at node k. Hence, the coordinates of any point in the element at time t are interpolated by the expression
The displacements and the displacement increments are interpolated by
Here 'u: and U: denote, respectively, the displacement and incremental displacement components in the xidirection at the kth node and time t. For small rotation di2 at each node, we have 1 1 k di2 = 8, el Q1 e2 O3 l  k e3 (10.8.30)
+
+
the increment of vector
le!
can be written as
Then Eq. (10.8.29) becomes
The unit vectors
lef
and
le:
at node k can be obtained from the relations
where E~ are the unit vectors of the stationary global coordinate system (Ox1,Ox2,Ox3). Equation (10.8.35) can be written in matrix form as
where {Ae) = {u: O f (i = 1 , 2 , 3 , k = 1 , 2 , .. . , n, and n is the number of nodes) is the vector of nodal incremental displacements (five per node), and [HI is the incremental displacement interpolation matrix
Qg)T,
For each time step or iteration step one can find 3 unit vectors at each node from Eqs. (10.8.31) and (10.8.33).
The linear strain increments { o e ) = {oell
expressed as
0e22 oe33
2oe12 2oeI3 2oe23IT can be
(10.8.36a)
{oe) = '[A]{ou)
where
{OU)
is the vector of derivatives of increment displacements,
and ouij = a u i / a o x j . The vectors { ~ uand {Oe) are related to the displacement ) increments at nodes by
{oe) = [A] {ou) = [A] N ] [HI [
{ne) =
[B~] {ae)
(10.8.37a)
6[BLI
=
[A1[Nl [HI
where [ N ]is~the operator of differentials
The components of ' [ A ] include ;ui,,?. From Eq. (10.8.28) the global displacements are related to the natural curvilinear coordinates (<, 7 ) and the linear coordinate C. Hence the derivatives of these displacements hui with respect to the global coordinates Oxl, Ox2 and ' 2 3 are obtained through the relation
The Jacobian matrix '[J] is defined as
and is computed from the coordinate definition of Eq. (10.8.27). The derivatives of displacements lui with respect to the coordinates J,q and ( can be computed from Eq. (10.8.28). In the evaluations of element matrices in Eqs (10.8.6ad), the integrands of ;[BL], 0[C], '[HI and ;{s) should be expressed in the same coordinate system, namely the global coordinate system (Ox1,Oxa,Ox3) or the local curvilinear system (xi, 2'3). xk, The number of stress and strain components are reduced to five since we neglect the transverse normal components of stress and strain. Hence, the global derivatives of displacements, [;uif] which are obtained in Eq. (10.8.26), are transformed to the local derivatives of the local displacements along the orthogonal coordinates by the following relation
A B ~ ~ ] A[s], ,
where [@IT is the transformation matrix between the local coordinate system ( x ~ , x ~ , xand the global coordinate system (0x1,0x2,0x3).The transformation $) matrix [O] is obtained by interpolating the three orthogonal unit vectors ('el, l e 2 ,'e3) at each node:
Since the element matrices are evaluated using numerical integration, the transformation must be performed at each integration point during the numerical integration. In order to obtain ;[BL], the vector of derivatives of incremental displacements {uo) needs to be evaluated. Equations (10.8.38) and (10.8.40) can be used again except that lui are replaced by ui and the interpolation equation for ui, Eq. (10.8.41), is applied. Next we discuss the matrix of material stiffness. For a shell element composed of orthotropic material layers, with the principal material coordinates (xl, xz,2 3 ) oriented arbitrarily with respect to the shell coordinate system (xi, xk, x; = x3). For a lcth lamina of a laminated composite shell, the matrix of material stiffnesses is given by c c;, c;, 0 0 ci2 c c 0 0 0[Ct](lc) = Cifj Ck, C&7 O O 0 0 0 c;, Ci,  0 0 0 c c;,
where
where Qij are the surface stressreduced stiffnesses of the Icth orthotropic larnina in the material coordinate system. The Qij can be expressed in terms of engineering constants of a lamina
where E, is the modulus in the x, direction, G,:, (i # j ) are the shear moduli in the 2,x:, surface, and v,:, are the associated Poisson's ratios. To evaluate element matrices in Eqs. (10.8.19ad), we enlploy the Gauss quadrature. Since we are dealing with laminated composite structures, integration through the thickness involves individual lamina. One way is to use Gauss quadrature through the thickness direction. Since the constitutive relation o[C] is different from layer to layer arid is riot a continuous function in the thickness direction, the integration should be performed separately for each layer. This increases the computational time as the number of layers is increased. An alternative way is to perform explicit integration through the thickness and reduce the problem to a 2D one. The Jacobian matrix, in general, is a function of ( E , rl, <). The terrris in may be neglected provided the thickness to curvature ratios are small. Thus the Jacobian matrix ' [ J ] becomes independent of arid explicit integration can be employed. If terms are retained in ' [ J ] Gauss points through the thickness should , be added. In the present study we assume that the Jacobian matrix is independent of in the evaluation of element matrices arid the internal nodal force vector. Since the explicit integration is performed through the thickness, the expression for
<
<
<
<
is now expressed in an explicit form in terrris of C. Hence, we can use exact integration through the thickness and use the Gauss quadrature to perform rlunierical integration on the rriidsurface of the shell element. For thin shell structures, in order to avoid "locking" we use the reduced integration scheme to evaluate the stiffness coefficierits associated with the transverse
shear deformation. Hence we split the constitutive matrix [C'] into two parts, one without transverse shear moduli O[C1IB,and the other with only transverse shear moduli o[C']s. Full integration is used to evaluate the stiffness coefficients containing OIC']B,and reduced integration is used for those containing o[C']s If a shell element is subjected to a distributed load (such as the weight or pressure), the corresponding load vector 2 { ~ ) from Eq. (10.8.19ad) is given by
i where 2 ~ is the component of distributed load in the Oxi direction at time t At, O is the area of upper, middle or bottom surface of the shell element depending on A the position on the position of the loading and the loading is assumed deformationindependent. Substituting [HI into Eq. (10.8.45) yields
+
NP where h = C k = lE $k([, q)hk is the shell thickness at each Gauss point, and W is the weight at each Gauss point, and 1°Jl is the determinant of the Jacobian matrix in Eq. (10.8.39) at each Gauss point. Here the C terms are retained in Jacobian matrix and let 5 equal to 1, 1 or 0, respectively, when the distributed loading is at the top, bottom or middle surface.
10.8.5 Numerical Examples
A number of numerical examples of laminated plates and shells are presented. Only
static bending problems of plates and shells are included. The RiksWempner method is employed for tracing the nonlinear loaddeflection path (see Appendix 1 of [32]). For most of the problems the reduced/selective integration scheme is used to evaluate the element stiffness coefficients. The following three sets of boundary conditions are used in the numerical examples presented here (see Figure 10.8.3).
Figure 10.8.3: Geometry and coordinate system for a plate or shell panel.
Orthotropic plate u n d e r u n i f o r m load
Here we consider a simply supported, orthotropic, square plate under uniform transverse load qo. The geometry and material parameters used are
a = h = 12in., h=O.l38in.,
El = 3 x 106 psi, E2 = 1.28 x 1 0 ~ ~ s i
(10.8.50)
Glz = GI3 = Gz3 = 0.37 x lo6 psi, yz = 0.25
A quarter of the plate with BC1 boundary and symmetry conditions is modeled with the 2 x 2Q9 mesh of continuum shell elements. The present results shown in Figure 10.8.4 are in good agreement with the experimental results of Zaghloul and Kennedy [8].
S i m p l y supported spherical shell panel u n d e r point load
A simply supported isotropic spherical shell panel under central point load is analyzed for its large displacement response using 4 x 4Q4 and 2 x 2Q9 meshes in a quarter of the shell. The geometric and material parameters of the shell are shown in Figure 10.8.5. Figure 10.8.6 shows the response, including the postbuckling range (calculated using the modified Riks Wernpner method). The figure also includes the results of Bathe and Ho [53].
Center deflection, u, (in)
Figure 10.8.4: Maximum deflection versus the load magnitude for a simply supported orthotropic plate.
Symmetry line: uo = 4x= 0 Simply supported:
uo = U o = w o = GX = 0
Simply supported:
E = lo4 psi, v = 0.3, R = 100 in., a
= b = 30.9017 in.
Figure 10.8.5: Geometry and boundary conditions of the spherical shell panel analyzed.
17.5 15.0
1
Bathe and Ho I531
0 9node elements
0.0
0.5 1.0 1.5 2.0 2.5 3.0 Transverse deflection, w d h
3.5
Figure 10.8.6: Loaddeflection curves for a simply supported spherical shell panel under central point load (see Figure 10.8.5 for the geometry and boundary conditions).
Isotropic cylindrical shell panel u n d e r point load
An isotropic shallow cylindrical shell panel hinged along the longitudinal edges and free a t the curved boundaries and subjected to a point load is analyzed (see Figure 10.8.7a). A quadrant of the shell is modeled with 2 x 2Q9 mesh of continuum shell elements. The structure exhibits snapthrough as well as snapback phenomena, as shown in 10.8.7b. The solution obtained by Crisfield [46] is also shown in Figure 10.8.7b to be compared with the present results.
S i m p l y supported composite spherical shell panel u n d e r u n i f o r m load
A simply supported laminated spherical shell panel under uniform load was analyzed for its large displacement response with 2 x 2Q9 mesh of continuum shell elenlents in a quadrant of the shell. The geometry and material parameters used are: n = b = 50 in., h = 1 in., R = 1,000 in., El = 25E2, E2 = 106 psi, G12 = G13 = 0.5E2, Ga3 = 0.2E2 psi, z 4 2 = 0.25. The effect of edge boundary conditions and symmetry conditions on the nonlinear response is investigated using BC1 and BC3. The effect of slight difference in the boundary conditions is very significant on the deflection response, as shown in Figures 10.8.8a and 10.8.8b for twolayer crossply (0/90) and (45145) angleply laminates, respectively.
642
MECHANICS OF LAMINATED COMPOSITE PLATES A N D SHELLS
Sympetry line: u , = & = 0 Simply uo = U o
h
J
\
R
uo=~o=w,=Q)y=O Simply supported:
Symmetry line: uo = g), = 0
E = 3103 Nlmm2, v = 0.3 = 2540 mm, a = 254 mm, B = 0.1
6
5 1'0 1'5 20 2'5 Center deflection, w, (mm)
Figure 10.8.7: Geometrically nonlinear response of a shallow cylindrical shell. (a) Geometry and finite element mesh. (b) Loaddeflection curves.
+
Ref. 82
6
(a) Crossply laminates
0.0
0.5
1:0 1.5
2.0 2.5 w O (in)
3.0
3.5 4.0
(b) Angleply laminates Figure 10.8.8: Geometrically rlorlliriear response of a shallow cylindrical shell. (a) Loaddeflection curves for (0190) laminates. (b) Loaddeflection curves for (45145) laminates.
Ninelayer crossply (0/90/0/90/ . .) simply supported spherical shell panel
A crossply spherical shell laminated of nine graphiteepoxy material layers with the
material properties
and subjected to uniform transverse load. The same geometry as that in the last problem ( a = b = 50 in., h = 1 in., R = 1,000 in.) is used. A quadrant of the shell was modeled using 2 x 2Q9 mesh of continuum shell elements and simply supported (BCl) boundary conditions. The loaddeflection curve obtained with the modified RiksWempner method is compared with that obtained by Noor and Hartley [75] in Figure 10.8.9. Note that the laminated shell exhibits softening first and then stiffening and does not have a limit point. This response is similar to that in Figure 10.8.7b with the same boundary conditions.
10.8.6 Closure
This completes the nonlinear finite element analysis of laminated plates and shells using continuum shell element. Additional examples involving stiffened shells can be found in [21,23,78,79].
10
1
Noor and Hartley [751 ' 'Present solution
a = b = 50 in., h = 1in., R = lo3 in.
I
I
I
I
I
1
I
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4
Center deflection, ZO h
Figure 10.8.9: Loaddeflection response of a simply supported (BCl), ninelayer (0/90/0/90/. . .), laminated spherical shell panel under uniform load.
10.9 Postbuckling Response and Progressive Failure
of Composite Panels in Compression
10.9.1 Preliminary Comments
The classical lamination theory, in which the transverse shear effects are neglected, is often used to analyze laminated composite structures. Because of low moduli and strengths in transverse directions compared to that of inplane directions, composite laminates may fail due to transverse stresses. Indeed, it is found that composite laminates loaded in compression fail due to high interlaminar stresses (see [83,84]). Therefore, shear deformable plate and shell elements are needed to provide information regarding the throughthickness strength of composite structures. Insight gained by using these elements may aid in the characterization of failure modes of composite panels. In this section we present a case study of the postbuckling response of two graphiteepoxy panels loaded in axial compression. The study makes comparisons between the experimentally obtained and analytically determined postbuckling response of corrlposite panels (see Engelstad, Reddy, and Knight [84]).
10.9.2 Experimental Study
The postbuckling and failure characteristics of flat, rectangular graphiteepoxy panels, with and without holes, and loaded in axial compression have been examined in an experimental study by Starnes and Rouse [83]. The panels were fabricated from commercially available unidirectional Thornel 300 graphitefiber tapes preimpregnated with 450°K cure Narmco 5208 thermosetting epoxy resin. Typical lamina properties for this graphiteepoxy system are
E2 = 12.0 GPa (1,890 ksi) G12 = 6.4 GPa (930 ksi), y = 0.38, hk = 0.14 mm (0.0055 in.) z El
= 131.0 GPa (19,000 ksi),
(10.9.1)
where hk denotes ply thickness. Each panel was loaded in axial conlpression using a 1.33 MN (300 kips) capacity hydraulic testing machine. The loaded ends of the panels were clamped by fixtures during testing and the unloaded edges were simply supported by knifeedge restraints to prevent the panels from buckling as wide columns. A typical panel mounted in the support fixture is shown in Figure 10.9.la. Most panels exhibited postbuckling strength and failed along a nodal line of the buckling mode in a transverse shear failure mode, as shown in Figure 10.9.1b (see [83]). However, a different failure mode was observed for some of the 24ply panels with holes. These panels failed along a transverse line passing through the hole, and failed soon after buckling. Here we analyze two panels, denoted C4 and H4 (see Figure 10.9.2) in [83]. The finite element results are compared with the experimental results of Starnes and Rouse [83]. Panel C4 is 50.8 cm by 17.8 cm (20.0 in. long and 7.0 in. wide), 24ply 45/02/ f 45/02/ f 45/0/90), (orthotropic). Panel C4 was observed in laminate, (f the test to buckle into two longitudinal halfwaves and one transverse halfwave. The second panel, Panel H4, is a 50.8 cm by 14.0 crn (20.0 in. long by 5.5 in. wide) 24ply laminate (f 45/0/90)3, (quasiisotropic). A 1.91 crn diameter (0.75 in. diameter) hole is located 19.1 cm (7.5 in.) from one of the loaded edges and along
(a) Typical panel with test fixture
(b) A transverse shear failure mode
Figure 10.9.1: (a) Typical panel with test fixture (load frame). (b) Failure mode (from Starnes and Rouse [83]).
22)] where (6A) is the vector of incremental nodal displacements. 9CR. (10. Panel H4 was observed in the test to buckle into four longitudinal halfwaves and one transverse halfwave with the hole located near the buckle crest of the second longitudinal halfwave. 10.2: Geometry and finite elements meshes of the C4 and H4 composite panels used in the postbuckling study. and it is denoted here as the ninenode ChaoReddy element [21]. ([KL].8.8. and {F) is the force vector [see Eqs.b.19a.8.C4 specimen model (b) H4 specimen model k 1 Figure 10. which are based on continuum formulation of a laminated shell. the panel centerline.9.21) and (10. [KNL])are the linear and nonlinear parts of the stiffness matrix. The final incremental equations of equilibrium for an element are of the form [see Eqs.9.d)]: .3 Finite Element Models Finite element models of such panels were developed in Section 10. (10.8.
2a shows the model used for the C4 specimen. [c']~ material coordinates. In the maximum stress criterion.85881). Hence. (10. which has the form [see Eqs.3). we use Gauss quadrature in the surface directions of the shell. The amplitude of each mode was selected to be 15% of the total laminate thickness. The finite element model of Panel H4 is different.14.9. In order to proceed beyond the critical buckling point in the analysis of each panel. an initial geometric imperfection.8. This model has four "rings" of elements around the hole with each ring subdivided into 16 elements. [C] is the constitutive elasticity matrix. To evaluate the integrals in Eq. was assumed in the finite element analysis. This allows efficient progress past the critical buckling point.8. 10.9. failure is assumed to occur if any one of the following conditions are satisfied: . but explicit integration in the thickness direction.44)] where QtJ are the plane stressreduced elastic coefficients in the material coordinates and 0 is the fiber orientation angle. respectively. due to the presence of the hole. Thus the thickness direction integration for matrices [KL]and [KNL]gives the following laminate stiffnesses: ' Here (.4 Failure Analysis The maximum stress and TsaiWu failure criteria are used (see [13. Figure 10. [BL] and [BNL]are linear and nonlinear straindisplacement transformation matrices. P is the number is the constitutive matrix for the kth lamina in the principal of laminae. but does not affect the results in the postbuckling range.43) and (10. The total numbers of ninenode quadrilateral elements in the finite element models of Panels C4 and H4 are 72 and 124.In these equations. (10.9. typically the same shape as the first linear buckling mode. and {R) is the external load vector. is the thickness coordinate of the bottom of the kth lamina. as shown in Figure 10.2b. All matrix elements refer to the deformed state and are measured with respect to the original undeformed configuration.9. the finite element model of Panel C4 consists of 12 ninenode quadrilateral elements along the panel length. The finite element model used in Reference 84 consisted of six elements per buckle halfwave in each direction. [ S ]and {s) are the matrix and vector of second PiolaKirchhoff stresses.
In reality. then the longitudinal modulus El a t that point is reduced to zero. g 2 . if the a1 stress exceeds the longitudinal tensile strength XT. At each load step. F. To model this effect. laminate failure occurs due to propagation of damage as the load is increased.. YT. 13. 3. After nonlinear iterative displacement convergence is achieved. (a) identify the maximum value of Hi.e. If failure occurs (i. For example.2 . z ) coordinates at the middle of each layer at each Gauss point. ZT) are the lamina normal strengths in tension (T) along the (1. When (al. engineering material properties are updated as failure progresses. For the TsaiWu criterion. respectively. 2. (XT. for the maximum stress criterion. 1. 3) directions. Thus H1 corresponds to the modulus E l . 0 3 ) are the normal stress components. T) are the shear strengths in the (23. a progressive failure approach is used in the nonlinear finite element analysis. and H6 t o G23. they should be compared with (Xc. H2 to Ez. An outline of the steps used in the analysis is given below. which are normal strengths in compression (C) along the (1. and (R. g3) are compressive. 2. H to G23. calculate stresses in the global (x. Yc.Wu criterion is given by where gi denote the stress components referred to the principal material coordinates. 2. and [Dl of the laminate. AS a consequence of this reduction. 4. 3) principal material directions. Transform the stresses t o the principal material coordinates. Gauss point stresses are used in the selected failure criterion. Zc). S . 12) 0 planes. then the following expressions are used to determine the failure mode: The largest Hi term is selected as the dominant failure mode and the corresponding 4 modulus is reduced t o zero. respectively. a modification of the lamina properties was made at that Gauss point. F 2 I ) . y.where (a1. Compute the failure index. gfj) are shear stress components. (a4. The Tsai. Hs t o G13. If failure occurred at a Gauss point.05. . which results in reduced stiffnesses [A]. if failure occurs.[B].
The postbuckling response exhibits large outofplane deflections (nearly three times the panel thickness. (b) outofplane deflection wg near a point of maximum deflection.. proceed to the next load step. the largest value is well below the material allowable values: XT = 1400 MPa (203 ksi) in tension and Xc = 1138 MPa (165 ksi) in compression. end shortening uo. see Figure 10. and (c) recompute laminate stiffnesses and restart the nonlinear analysis a t the same load step (i..9. is large.These experimental and finite element results agree well up to failure of the panel. 10.1Pc. normalized by the analytical end shortening u . the transverse shear stress distributions were also obtained by integrating the equilibrium equations. . Figure 10.1Pc. 5. wherein the inplane stresses were computed using the constitutive relations. The end shortening of the panel is monitored as in a compression test. These are all shown as functions of the applied load P. normalized by the laminate thickness h for P = 2.. Although a. The stresses were determined using the constitutive relations for both the inplane and transverse components. normalized by the theoretical buckling load PC..3a). In addition. normalized by the panel thickness h (Figure 10.9. Figure 10.4b contains a photograph of the Moir4 fringe pattern from Reference 83 corresponding to the outofplane deflections observed during the testing of Panel C4 at the same load.9.4a contains a contour plot of the outofplane deflections generated frorn the finite element analysis at an applied load of 2.(b) reduce the appropriate lamina moduli at that Gauss point.9.5 Results for Panel C4 Comparison between test results from Reference 83 and finite element results from Reference 84 for Panel C4 are shown in Figure 10. The contour plot of a.9. The figure shows (a) .3.9.3b) and high longitudinal strains from front and back surfaces (nearly three times the analytical buckling strain. stress through the thickness direction z . see [84]) that high compressive axial stresses occur along the longitudinal edges of the panel. The failure load is defined to be that load for which the panel undergoes large end shortening for small increments of load.. and (c) the longitudinal surface strains e near a point of maximum outofplane deflection. normalized by the analytical buckling strain e. at buckling (Figure 10.9. see Figure 10..3~).9.5 shows the distribution of the maximum a.1 P. the normal stress is nearly uniform across the panel.3b).6 contains the distribution of the normal stress a.9. If no failure occurs. At the buckling load..9.. indicates (not shown here. It is clear that the 0" layers carry the largest transverse shear load. Figure 10.e. return t o Step 1). Figure 10.. Stress distributions in each layer of the laminate were calculated using the nonlinear finite element results in order to determine the failure loads. over the entire panel in this 0' ply for an applied load of 2... in the third layer of the laminate (a 0" ply) a t panel midlength for three values of the applied load. These results indicate that the outofplane deflections from both test and analysis halfwaves with a bucklingmode nodal line a t panel midlength.
0 Maximum deflection.5 4. e l e .0 a.5 Strain.0 0.5 3.5 1.0 Figure 10.0 2. w o / h 2.5 0.0 0.0 0.0 Y 2.5 2..0 2.5 (a) E n d shortening 1.5 1.5 2.0 1 1 1 1 cd a d 0 Test i 0.p l a n e deflection I I 1 0.0 u t . 3.5 (c) Surface strains I I l l [ T ' 2.5 1.0 1. u o l u.9.0 0.0. . .0 E n d shortening.0 1.5 0.3: Postbuckling response characteristics of panel C4.01 1 1 1 1 I 3. 0. E m h h L 1.o f . 2 a .0 1.5 1.
.4: Comparison of experimental (Moirk) and analytical outofplane deflection patterns for panel C4. 9.5: Transverse shear stress. z l h Figure 10.0 Thickness coordinate.8 1.. of panel C4. Figure 10. a.0 0.4 (a) Contour plot of the analytic:a1results [84 Photograph of Moire fringe patte:rn [831.4 0.9.9.0 1 i 1 111 r r I I I TlT l l r r ~ T ~ ~ r ~ r ~ v ~ ~ ~ ~ ~ ~ ~ ~ l ~ ~ ~ ~ r 0. distribution through the thickness .6 0.2 0.
distributions in the third layer from the surface (0" ply) of panel C4.6: Axial stress..(a) Contour plot of axial stress distributions (b) Stress distributions across panel midlength Figure 10... .7: Transverse shear stress.. a. (a) Contour plot of shear stress distributions (b) Stress distributions across panel midlength Figure 10. distributions in the third layer from the surface (0" ply) of panel C4. CT.9.9.
9.1Pc. entire panel in this 0" ply for an applied load of P = 2. . the transverse shear stresses o . XT = 80. This is attributed to the presence of stress interaction terms in the TsaiWu criterion failure index. over the .7 shows the distribution of the transverse shear stress in the third layer of the laminate (a 0" ply) at panel midlength for three values of the applied load. In addition to the strengths already mentioned.0 MPa (27. indicating the panel failure due to transverse shear stress.9. we obtain Exr =  1 awo 2 (4. using the maximum stress and TsaiWu failure criteria.9. A contour plot (not shown here) of the distribution of the transverse shear stress a. Figures 10. The peak values of the transverse shear stress u .0 ksi).8a and 10.11) into the strain in Eq. Xc = 189.  auo + ax + 4.9. P = 2. and the dashed curves denote the transverse shearing stress distributions obtained from the equilibrium equations. indicates that high transverse shear stresses occur along the bucklingmode nodal line. Figures 10. leads to the conclusion that the transverse shearing strain ex. redistribute towards the edges of the panel. transverse compressive strength.) ax The quantity (outofplane deflection gradient) is largest along a bucklingmode nodal line and the quantity (related to the membrane strain) is largest along the panel edges. After buckling. respectively.9. At some point in the analysis a dramatic change in slope indicates an inability of the panel to support additional load. T = 69.1Pc.4 ksi). This location is identified as the failure load. This failure mode can be further explained through a close examination of the GreenLagrange strain component in conjunction with the displacement field of the firstorder shear deformation theory Substituting of the displacements from Eq. A similar examination of the other transverse shearing strain ey.8a and 10..8b present the progressive failure results for Panel C4.8b show that the TsaiWu criterion estimates the experimental failure more closely than the maximum stress criterion. approach the material allowable value of S = T = 62 MPa (9 ksi) for . (10.9 MPa (11.0 MPa (10. is the dominant one.9. At the buckling load.10) and noting that $y is zero along a bucklingmode nodal line. (10. and inplane shear strength. Both methods give very similar results. the peak transverse shear stress occurs near the center of the panel. The solid curves represent the transverse shear stress distributions obtained using the constitutive relations. the other allowables used are transverse tensile strength.Figure 10.9. % .7 ksi).
Element distortion around the hole could be another contributing factor. u o1 u.0 2.. and lack of restraint of the model around the hole. These occur because of zero energy modes....8: Progressive failure results of panel C4. Figure 10.... respectively.0 5.39Pc......90Pc.. i ! 0.. It is necessary to use full integration to alleviate this problem... ..9...9.. Figure 10.0 8.0 / .9a contains a comparison of end shortening obtained numerically and experirnentally..0 4...9. 9 ~ show the longitudinal surface strains e (both top and bottom surfaces) across the panel at the hole for a load of 0. These results are in good agreement with experimental results from Reference 83.5 1. it would predict spurious modes..0 3.0 E n d s h o r t e n i n g ..0 6.0 Ultimate failure Firstply failure Progressive failure 0 Test Test failure 10.. E 1.0 6.0 E n d s h o r t e n i n g ...6 Results for Panel H4 Panel H4 was analyzed to investigate deformation and failure of a panel with a hole. and the mesh should be sufficiently refined so that element locking effects are negligible.. and 1.0 8..5 2 Ld a Progressive failure 0 0 ..0 ( b ) TsaiW u failure criterion 0. Figures 10.0 0.9b and 1 0 ..0 12.9. a.. a 2.0 1.0 4... It should be noted that if uniformly reduced or selectively reduced integration were used in the analysis of this panel. 9 .0 Test Test failure I (a) Maximum stress failure criterion 2... u o1 ucr 7. An imperfection of 1% of panel thickness times mode 1 was used to proceed into the postbuckling range. .. 10. k * a.
u o / a 0.6 0.t kl m a a 0.000 0.2 0.8 0.001 (a) End shortening 0.1 0.0 Figure 10.9.5 ylb 0.3 0.003 0.9: Postbuckling response characteristics of panel H4.001 1 1 I I I I 1 I l l 7 0.90 0.005 1 1 i 1 1 1 1 r i i I I I I I I j I r 71 (b) Surface strains a t P I P .7 0.004 E n d shortening.0 0. .9 1.002 0. .4 0. = 0.
a transverse shear mechanism develops along nodal lines away from the hole.008 0.006 0. Simultaneous firstply failure occurs due to a.For panel H4.. Progressive failure results are shown in Figures 10.000 / 0 Progressive failure Test Test failure 1 / 1 1 1 1 1 1 1 1 I I I I 1 1 1 1 1 1 1 ( b ) TsaiWu failure criterion 0.9.10a and 10. The TsaiWu criterion is.015 0.010 E n d s h o r t e n i n g . Thus the failure mode is not a dominant transverse shear mode as for Panels C4 and C10. components around the hole edge. l a 0. the peak stress approaches only 48.001 0. u. u. in better agreement with experimental results.10: Progressive failure results of panel H4.0 ksi) a t the experimental failure load. once again. At this load the inplane shear stress approaches its allowable around the hole. but a more complex interacting mode with a dominant inplane shear component.000 0.018 a 2 0 0. I I 1 ( a ) Maximum s t r e s s failure criterion 0.9. once again.002 0. and a. I 1 1 1 ' 1 I 1 1 $ 1 I l l I I I Firstply failure Progressive failure Test failure 1 1 1 1 1 1 I l l 1 1 1 1 .. We close this section with a comment that the case study presented in this section brings out the importance of interlaminar stresses.009 0./a 0012 Figure 10. References 13 and 84 contain additional results of postbuckling and progressive failures.003 0.006 0.000 0. However.012 E n d s h o r t e n i n g .004 0.10b. .3 MPa (7.9.
Explain why. Problems 10. the von KBrmdn nonlinear formulations of laminated plates using the classical and firstorder shear deformation theories of plates and Sanders theory of shells are developed. For studies on damage and failures in composites.1 Consider the nonlinear differential equation  d ( U ) =( x ) 0<x < 1 Show that the finite element model is given by [ K e ] { u e ) {F") = (2a) 10.10 Closure The objective of the chapter was to introduce the concept of geometric nonlinearity. z) L() 2 dw 3 dx . The development of continuum shell element is also presented. transient and buckling response and ultimate failure of laminated plates and shells. UZ = 0.4. nonlinear transient response of laminated plates. and the tangent stiffness matrix coefficients of the FSDT element are derived. Numerical results of the nonlinear analysis using the FSDT plate and shell elements as well as continuum shell element are presented to illustrate the influence of symmetry boundary conditions on the nonlinear response. effect of the geometric nonlinearity on the response of antisymmetric crossply plate strips.2 Compute the tangent coefficient matrix for the finite element model of Problem 10. The NewtonRaphson iterative method of solution is discussed. the initial guess for u should be nonzero. present the continuum shell finite element.1. and study the influence of geometric nonlinearity on bending. w ( 5 . develop finite element models of the von KBrmBn nonlinear plate and shell theories.Z ) = WO(X) (1) Show that the von KArmBn nonlinear strains are given by and that the total potential energy associated with a laminated beam is .10. = ?o.3 Consider the displacement field of the EulerBernoulli beam theory (see Example 1. the reader may consult [13. 10.921101. and postbuckling response and progressive failure analysis of laminated panels under compressive load. In particular. Note that in an iterative solution of the problem.1): U l(z.
.he force and moment res~dt. are t.3 are  whcre N. b is the width. Assume finite element interpolation of ug and wg in the form and . and A'[.where y = qb is the distributed transverse load. show that iriterpolat.ants Show that the force arid moment resultants for symmetrically larrlinatcd beams can be expressed in terms of the displacements as 10. and A = b h is the area of cross section of the beam. ZA < x < 2 3 : Define the secondary variables P.5. p are the Hermite cubic .$. 10. In particular.ion functions. and O = 2.in terms of the displacements.5 Use t..6 Develop the nonlinear finite element model of a laminated beam using the weak forms given in Problem 10. 10. are t.4 t o derive the following weak forms for a beam finite element. .he equations of equilibrium of Problem 10.. arid Q.lle linear Lagrange interpolation furictions..4 Show that the EulerLagrarige equatioris associated with the displacement field of Problem 10.
the element stiffness matrix is unsymmetric.b) can be written in matrix form as 10. hence.Note: [Kl2IT# [K21].7 Use matrix notation and express the total potential energy functional for an element. as IIe({A}) =2 [ B L ] T [ DB L I ~ X ) [ ] + 1(1'" X A [BL]'[DI B N I ~ ~ ) [ . Equations (3a.
and Mzx the force and moment resultants are . u3(x. = u..3) with the von KBrmBn nonlinearity are given by where N.4. Q . (2) Z) + Z4. Z) = 0.8 Use the principle of the minirrlum total potential energy and show that where 10. )= w..10...10 Beginning with the displacement field u1(5. t (x) show that the equations governing the Timoshenko beam theory (see Example 1. (x). u2(x.9 Show that the tangent stiffness matrix of the EulerBernoulli beam element is where and 10.
AK2 dx dx dx 10.11 Show that the weak forms of the Timoshenko beam theory with the von KBrmAri nonlinearity are given by 10.A (dWO  dx )  d$.13 Evaluate the direct stiffness coefficients [KaP] ( a .and 4..12 Assume that the generalized displacements (uo.. 2 . wo. wo. are approximated by mX) of the Timoshenko beam theory and show that the finite element model is of the form KZ? 23 = 1 J z B E.Show that the force and moment resultants can be expressed in terms of the displacements as 10. . d: dx @ dx dx + X A I:" dQ2 dQ2 G.P = 1 . 3 ) of the nonlinear Timoshenko beam finite element assuming linear but equal interpolation of uo..
(8. show that the finite element model is of the form for a = 1 . Derive the finite element model associated with the governing equations developed in Problerri 10. d y = a2 d&. . and d z = d<. In particular. .2. Derive the equations of motion of Sanders nonlinear shell theory. A m : The tangent stiffness coefficients are Compute the tangent stiffness coefficientasfor the shell finite element of Section 10. 5 and I = 1.16.7.23) according to the Sanders' [70] nonlinear shell theory are where arid d l = crl d E l .14 Compute the tangent stiffness matrix coefficients associated with the nonlinear Timosheriko beam finite element. The nonlinear straindisplacement relations associated with the displacement field in Eq.2... . . 5 . My. . 2 .NONLINEAR ANALYSIS OF PLATES AND SHELLS 663 10. j . and QFJ for a = 1.2. and define the coefficients N . ..6. .
15.. 5. 12. W. 30. CRC Press. J. P.. 179183 (1976). 101(EM3). C.. 2. 17.. 11. Noor. Reddy. B. B. "Nonlinear Analysis of Unsymmetrically Laminated Plates. 0 ." Journal of Mechanical Engineering Science.. Chia. Finite Element Analysis of Composite Laminates. 705718 (1988). 19972003 (1971). C. Bennett. K. and Savoia. and Prabhakara. "Nonlinear Vibration of a Rectangular Plate Arbitrarily Laminated of Anisotropic Material. K. M. "Simple Finite Elements with Relaxed Continuity for Nonlinear Analysis of Plates." in Instability of Continuous Systems. Khot. 3. and Kennedy. N." Journal of Applied Mechanics. "Buckling and Postbuckling Behavior of Composite Cylindrical Shells Under Axial Compression. J.. .1. N.. B. Leipholz. 13. 19. and Berke. K. "PostBuckling Behavior of Stiffened CrossPly Cylindrical Shells. 61. "Large Deflection and Large Amplitude n e e Vibrations of Laminated CompositeMaterial Plates. Ochoa. 18(4). Zaghloul. ASCE. S. J. 265'281. A.." A I A A Journal. G. Zaghloul. Chandra. (1979). 331340 (1981). 169185 (1975).). and Reddy. 10. J.. IUTAM Symposium. M. J. L'GeometricallyNonlinear Transient Analysis of Laminated Composite Plates.. N. and Sawamiphakdi. N. B. and Hartley. 47(4). 509514 (1976). and Miravete." Fibre Science and Technology. 21482154 (1992). "Large Amplitude Flexural Vibration of Cross Ply Laminated Composite Plates. J. R. "Effect of Shear Deformation and Anisotropy on the NonLinear Response of Composite Plates. 40." Journal of Sound and Vibration. 392398 (1971). R. The Netherlands (1992). Iu. 4. and Reddy. 9. 9981000 (1994). "Large Deflection Vibration of CrossPly Laminated Plates with Certain Edge Conditions. UK.References for Additional Reading 1. 452458 (1973). Barking.). Chang." Journal of Applied Mechanics. Reddy." Computers and Structures." Symposium on Computational Methods i n Nonlinear Structural and Solid Mechanics.. "Effect of Transverse Shear on Nonlinear Vibration and PostBuckling of Antisymmetric CrossPly Imperfect Cylindrical Shells.. 14. 9.. A. 8.. J. B. N. S. M. C. 13. 42. W. N. 16. SpringerVerlag. J. Holister (Ed). A. (Eds. "Nonlinear Behavior of Symmetrically Laminated Plates. J. N." Journal of Applied Mechanics. L. 7. Applied Science Publishers. 229235 (1970)." Developments i n Composite Materials . Washington.. V. Hinton et al. 6. D. and Kennedy. Berlin. A. "Large Deformation Analysis of Laminated Shells by Finite Element Method. S. 621629 (1983). Y. 18. N. 5556 (1977)." Proceedings of the Third International Conference on Finite Element Methods. J.. Venkayya.. N. "Buckling and Postbuckling of Initially Imperfect Orthotropic Cylindrical Shells Under Axial Compression and Internal Pressure. (1980). 234236 (1975). S.. and Chao. C. T. V." A I A A Journal. Savoia. Australia. and Raju. 155202 (1982)." in Recent Advances i n Nonlinear Computational Mechanics. FL (1995)." A I A A Journal. Chandra. Y. A. 0 .C. and Chia. Reddy. N. 243263 (1975).. "Nonlinear Vibration of Simply Supported AnglePly Laminated Plates." A I A A Journal. "Analysis of Layered Composite Plates Accounting for Large Deflections and Transverse Shear Strains. "The Layerwise Shell Theory for PostBuckling of Circular Cylindrical Shells. Bert. Practical Analyszs of Composite Laminates. J. S. Reddy. Reddy. H.. "Large Deflection of Unsymmetric CrossPly and AnglePly Plates. Y. E . Reddy." International Journal of Solids and Structures. 30.. Boca Raton. 8. 20." Journal of Engineering Mechanics Division.. Khot. Kluwer. (ed. 8. 21(4). J.
C. 285~300(1986). J . J. N. NJ (1996) 32. N. W. Hintori (Eds. 27. "Formulation of Large Deflection Shell Analysis by Assumed Stress Finite Element Method. J.. Materials and Dynamics. .. H. Boston." Computer Methods i n Applied Mechanics and Engineering. . W. J. Reddy. 104(2). 3157 (198(i). 193239 (1977).). (Eds.. J . E. T . 353386 (1975). 36. (Eds. N. "A Mixed Finite Elerrient for the Analysis of Laminated I'lat.. J . K.! Ramm. and Reddy. Wunderlich. "A Plate/Sliell Element for Large Deflections and Rotations.." in Formulation." in Formulations and Computational algorithm^ zn Finite Element Analysis. "The HybridStress Model for Thin Plates.ods i n Engineering.. 39. Stricklin. Pian.rithms i n Finite Element Analysis. Boston. N.. and Reddy. MA (1977).. 33. Tillerson. Putcha. N.). Dvorkin. 34.sand Coinputatronal Algorithms i n Finite Element Analysis. J... Hughcs and E.J. MIT Press.21." A I A A J o ~ ~ r n a28(11). "Nonlinear Analysis of Composite Laminates Usirig a Generalized Laminate Plate Theory. "On DiscreteKirchhoff and Isopararrietric Shell An Assessment." Advances i n Computer Methods 271 Structural Mechanics and Design. A. J.ed Plates by Using a Mixed Element Based on a Refined Plate Theory.). .. G.  37. J . Boston. and Wilson. Huntsville. L. J. 27(1). Bathe. Reddy. Barbero. Horrigrnoe. Liao. (Eds. T. Oden et al. University of Alabama in Huntsville. W. Oxford University Press.. J.s and Algo. Ramm.es.J. "On Refined Computational Models of Composite Laminates. and Reddy. 1991 2007 (1984). J. N. and Munir: N." Computers and S t r u c t ~ ~ r e s . .." I n t e r n a t i o ~ ~ a l Journal for Numerzcal Meth.. U . N. N. E. J . S. S. T . E. J .). "Incremental Variational Principle and Finite Element Models for Nonlinear Problems. P. 3139 (1983). K . Elernents for Nonlinear Analysis (1983). Bathe et al. K . 301324 (1972). W. MIT Press. and Ho: L. L. S. 8998 16.. N. . Yuceoglu et al. (Eds. A.. 22." Journal of Sound and Vihmtion. R.J. and Reddy. 201217 (1976). "Static Geometrical and Material Nonlinear Analysis." International Journal for Numerical Methods i n Engzneering. Reddy.ite Element Procedures. "Analysis of Laminated Composite Shells Using a Degenerated 3D Element. Pineridge Press.. MIT Press. 29. N. H. "ContinuumBased Stiffened Coniposite Shell Element for Gcornetrically Nonlinear Analysis. et al.. C. 20. "Finite Element Formulations for Large Deformation Dynamic Analysis. 9. "Ir~cremental Formulations for Geometrically Nonlinear Problems. Putcha.1994 (1990). arid Pian. 361382 (1989). (Eds.: "Improvement of Plate and Shell Finite Elements by Mixed Formulations. Oxford. Von Riesemanri~ W..Hall. N. l. and Reddy. and Haisler. AL. MA (1977). S. 15. T . E. Bathe. W. L. "On Mixed FiniteElement Forrriulations of a HigherOrder Theory of Cornpositc~ Laminates. MA. 529538 (1986). 27. 23. UK. Fin. "A Refined Mixed Shear Flexible Finite Elerrient for the Nonlinear Analysis of Laminated Plates. and Bergan. Chao. A n Introduction t o Nonlinear Finite Element Analyszs. Bathe. H. 26. 24. 3 ." A I A A Journal. E. 38. 1987. Lee. 12391260 (1980).. 95101 (1989). ASME AD06.). I." Computers and S t n ~ c t u r e s 22(2). J. 28. . 2934 (1978). K .). J.. "Stability and Natural Vibration Analysis of Laminat. 25. G. K . H. 7. 30. Prentice." International Journal for Numerical Methods in Engineering. K . and Boland. Putcha. Englewood Cliffs." International Journal for Numerical Methods i n Engineering. Spilker.." in Admnces i n Aerospace Structures. and Reddy. R . Bathe. E. R.. 31.. N." A I A A Journal." in Form~u1ation. P. N. Bathe et al.for Plate and Shell Structures. 1 6 ." Finrte Element Methods . UK (2004). 35.
W. . "Discrete Approximations Related to Nonlinear Theories of Solids. 14. 52. "Vibration Characteristics of Functionally Graded Cylindrical Shells Under Various Boundary Conditions. T. 1462.. 57.. N. "A Note on Symmetry Considerations in the Transient Response of Unsymmetrically Laminated Anisotropic Plates. Gong. Ramm." International Journal of Impact Engineering.... J.. C. T. and Reddy. 13... Lam." International Journal of Solids and Structures. 60.. C. J. K. Loy. and Reddy. T. C. R.. N.." Computers and Structures." International Journal of Solids and Structures. G." European Journal of Mechanics. 61. 185199 (1999). and Kitipornchai. Liao. Sun... Wempner.. 55. J. L. 34(6). 131140 (1987). "The Elastic Response of Functionally Graded Cylindrical Shells to Low Velocity Impact. Park. 125(11). J. "An Incremental Approach to the Solution of Snapping and Buckling Problem. and Reddy.. 6389 (1981). "A Fast Incremental/Iterative Solution Procedure That Handles SnapThrough. Loy. "Analysis of Asymnietric Composite Laminates.. N. NACA (1948). C. J." Report No. J. C. 46. N.40. W. N. C." Journal of Engineering Mechanics. D. 53. Praveen. M.." International Journal of Solids and Structures. N. 5562 (1981). and Chin.. and Heyliger. "Bending of Rectangular Plates with Large Deflections. 805815 (1990). 35(33). 663684 (2000). Wunderlich et al." Applied Acoustics." International Journal for Numerical Methods i n Engineering. L.. J.. 737. G. and Reddy. (Eds. 529551 (1979)." Computers and Structures... N. 7. N. 56.. "Analysis of Anisotropic. 15. 49. "A Mixed Updated Lagrangian Formulation for Plane Elastic Problems." International Journal for Numerical Methods i n Engineering 47. A. 20.. N. 9(4). W . L. Reddy.. P. "Thermoelastic Analysis of a Functionally Graded CeramicMetal Cylinder. Y. Riks." A I A A Journal. 15811599 (1971). 50. 18. K." Computers and Structures. 501510 (1967). Crisfield. 44574476 (1998). M.. 1259 1267 (1999). C. Batoz. and Ho. Praveen. 175194 (1984)." Nonlinear Finite Element Analysis i n Structural Mechanics. G.. J. N." International Journal of Mechanical Sciences. N." Z A M M . 26(6). 58. 54.714718 (1988). 59. G. Y. Berlin. Bathe. S. G.). 47. Reddy. 309324 (1999). "Analysis of Functionally Graded Plates. M. 673681 (1981). 397417 (1999). 42. H. Stiffened Composite Laminates Using a ContinuumBased Shell Element.. and Stanley. "Influence of Large Amplitudes on Flexural Vibrations of Elastic Plates." Report No. A. Y. "Strategies for Tracing the Nonlinear Response Near Limit Points. "Axisymmetric Bending of Functionally Graded Circular and Annular Plates." Journal of Applied Mechanics. J. Lkvy. J. 22(4). Reddy. 41. and Reddy. C. E. and Reddy. K. and Dhatt. 278290 (1986). "A Curved C0 Shell Element Based on Assumed NaturalCoordinate Strains. 51. Lam. S.. "A Simple Effective Element for the Analysis of General Shell Structures. "Bending of Rectangular Plates with Large Deflections. 43. K. 12621266 (1979). 45. Chin." Internationul Journal for Numerical Methods i n Engineering. K. Yamaki. 41(3). 111129 (2000). "Vibration of Functionally Graded Cylindrical Shells. 48. J. S. Lam. 44." Journal of Composites Technology and Research. SpringerVerlag.. 41. N. "Nonlinear Transient Thermoelastic Analysis of Functionally Graded CeramicMetal Plates. "Incremental Displacement Algorithms for Nonlinear Problems. Pradhan. 13. Reddy. 53.. J. NACA (1942). S. Wang.. N. E. Wang. T. C.
Y." European Journal of Mechanics.. "A Geometric and Material Nonlinear Plate and Shell Element . 38. R. J. 62. 67. 95 101 (1989). Bathe. "ThreeDimensional Solution of Smart Functionally Graded Plates. 841860 (2001). and Shen. N. 234241 (2001).. Rcdtly. N.ical Sciences. J . 11. J. N.. and Reddy." Co7np. H . Bergan. 79. J. "Frequency Correspondence Between Membranes and Functionally Graded Spherical Shallow Shells of Polygonal Planform. Noor." A I A A Journal. "Nonlinear Analysis of Laminated Shells Incl~~dirig Transverse Shear Strains. 977987 (2002). W~inderlich(Eds. Woo. G.. Reddy. 70. P. Stanley. A.). J ." Internatzonal Journal of Mechanical Sciences. 440 441 (1985). Yarig.S. H. Q.S. I.Jr. "A SolidShell Transition Element for Geometrically Nonlinear Analysis of Laminated Composite Structures. 65. J. 353386 (1975). Liew. J . P. N. 68. C. and Cheng. 1 5 . E." Corn. and Engelstad. K. 2136 (1963). 579 6 0 2 (2002).. 967 985 (2002)..: "Nonlinear Thermoelastic Analysis of Functionally Graded Plates Using the ThirdOrder Shear Deformation Theory. J.urnal of Applied Mech. 63. "A Rectangular Laminated Anisotropic Shallow Thin Shell Finite Elmielit. H.. and Reddy. J . SpringerVerlag (1986). 2348 (1980). P. and W. llFinit. 64. 68." . 74.Journal of Sound and Vibmtion. 3 . C. Shen." Composite Science and Technology. A. J.. Liao. L. Liao.61. Z.. J." Jo. 18431854 (1988). J . 27(1). and Clierig. "A ContinuumBased Stiffened Composite Shell Element for Geometrically Nonlinear Analysis. J . Rao." Internatio7~alJournal of N o n . Q. Structures. Ng. 71. . 72. "Nonlinear Analysis of Functionally Graded Plates and Shallow Shells." A I A A Jourr~al. 1295 1309 (2001). "Nonlinear Bending Response of Furlctiorially Graded Plates Subjected to Transverse Loads and in Thermal Environments. "Computational Procedures for Postbuckling for Composite Shells" in Finite Element Methods for Nonlinear Problems. 26.uter Methods i n Applied Mechanics and Engineering. "ThreeDimensional Thermomechanical Deformatioris of Functiorially Graded Rectangular Plates. W. 561584 (2002). N. and Hartly. 75.. 7.anzcs.. Kreja. Z.s. and Reddy. 255(3). 359385. T . Bathe. 73. 615 626 (1977). N.. J . G. . "Nonlinear Shell Analysis via Mixed Isoparametric Elements. 44(5). 11231 142 (1997). Bathe."Vibration Characteristics and Transient Response of ShearDeformable Functionally Graded Plates in Thermal Environments.. L. J . Lam." Ircternational Journal of Computational Engineering Science (to appear). M." International J o u r ~ ~ a l of Machan. 20(5). L. Rarrirn. S. S." I. M. Slien. 76. K. 21(1).. N.. 9.. .: "Finite Elernents Based on a FirstOrder Shear Dt:formation Moderate Rotation Shell Theory with Application to the Analysis of Composite Mechanics. N. "Nonlinear Theories for Thin Shells.L i ~ ~ e a r 77.. A. 44. 32(6). Y. K. Reddy.J..nternational Journal for Numerical Methods i n Engineering.. C. Q. 38. K. N. 78. S.pubws and Stmctures. L. Schmidt." Quarterly of Applied Mathematic. E.. and Chandrashckhara. 1333 (1978). K... K. and Felippa. Reddy." International Journal of Solids and Structures. and Retldy. K. Sanders . : "Postbuckling Analysis of Axiallyloaded Functionally Graded Cylindrical Shells in Therrml Environments. arid Wilson.uters and Structures. and Cheng. A/Solids. Reddy." Comp. "Dynarr~icStability Analysis of Functionally Graded Cylindrical Shells Under Periodic Axial Loading." International Journal for N~rmericalMethods i n Engineersrcg. Z... and Meguid S. 62.23(3).e Element Formulations for Large Deformation Dynamic Analysis." International Journal of Solids and Structures. 69. 74097421 (2001). and Bolourchi. J. 66. Aliaga...
S." Journal of Reinforced Plastics and Composites. 61.. 22. C. 11. Jr. First Ply Failure. J ." International Journal of NonLinear Mechanics. J." Composites Science and Technology." Journal of Composite Materials.. H... and Lo." A C I . 90. J. 1 5 . J ." AIAA Paper No. Jr. and Rouse. 371393 (1987). "A Failure Criterion for Laminates Governed by Free Edge Interlaminar Shear Stress. Zaghloul. 30(8). Reddy. and Dakshina Moorthy. W.Journal. 3. Theocaris. 100. 227255 (1992). 539561 (1964)." Computers and Structures. N. "Postbuckling Response and Failure Prediction of GraphiteEpoxy Plates Loaded in Compression. 101.. and Reddy. N. and Laws.. S. Dvorak. Azzi... "Positive and Negative FailureShears in Orthotropic Materials. 4062 (1984). N..80. N.. F. 15101512 (1992). 8399 (1985). 97. Reddy. F." Journal of Reinforced Plastics and Composites." A I A A Journal. C. S. Hill. S. "A Comparative Study of Failure Envelops in Composite Laminates. B. Jamison.. C. J. N. S. 834855 (1987). and Kennedy. Y. W. 86. and Sun. Tsai. M. and Pandey. "Flexural Failure Analysis of AnglePly Laminates of GFRP and CFRP. Joo. R. Engelstad. F. N. 2. Reddy. "A Theory of the Yielding and Plastic Flow of Anisotropic Metals. K. N. J . 4349 (1980). 82. Soni. N." Composite Materials.. R. 84. "Phenomenological Anisotropic Failure Criterion. Jr. 3255 (1992). A. N. 1 .. 81. Y. M. "The Role of Microdamage in Tensile Failure of Graphite/Epoxy Laminates. Wu. Reddy. G.. "A Survey of Macroscopic Failure Criteria for Composite Materials. Y. Stiffness Reduction and Crack Opening Profiles. V. J. Scordelis. 810543 (1981).. S. "Analysis of Progressive Matrix Cracking in Composite Laminates 1 . N. Turvey. 529538 (1986). E. 2106.. P. 629 649 (1995). 3442 (1983). 87. D. N. Series A. 98. 92.. 85.. "An Initial Flexure Failure Analysis of Symmetrically Laminated CrossPly Rectangular Plates. Chang. "Factors Influencing Nonlinear Static Response Prediction and TestAnalysis Correlation for Composite Panels. "A Refined Mixed Shear Flexible Finite Element for the Nonlinear Analysis of Laminated Plates.. J. "Nonlinear Progressive Failure Analysis of Laminated Composite Plates. A. 16. J. 1325 (1994). "Postbuckling and Failure Characteristics of Selected Flat Rectangular GraphiteEpoxy Plates Loaded in Compression.. 21. Putcha.. "Transverse Matrix Cracks in CkossPly Laminates: Stress Transfer. 88. and Chang." Proceedings of the Royal Society.. 227248 (1998). 96. "A Progressive Damage Model for Laminated Composites containing Stress Concentrations. 5. 95. 99.2113 (1992). 93." Composite Structures. J . A. 25." Journal of Composite Materials." International Journal of Solids and Structures. S. G.." Ezperimental Mechanics. Starnes." Composites Science and Technology. S. 42. 83." Computers and Structures." Journal of Reinforced Plastics and Composites. 29. N. 89. D. K." Journal of Applied Mechanics. 94. N. 130(34). Reddy. N. 451 463 (1980). and Knight. Praveen.. 281297(1948). 193. 44. S. J . 309329 (1987). 24. "Nonlinear Behavior of Symmetrically Laminated Plates. M. "A firstPly Failure Analysis of Composite Laminates. 283288 (1965). "Anisotropic Strength of Composites. Turvey. S. 26(10)... G. and Tsai. 21. K. and J . Reddy." Journal of Strain Analysis. 30(5). "Computer Analysis of Cylindrical Shells. G." Acta Mechanica. 234236 (1975)." Journal of Composite Materials. P.. Knight. 91. 353431 (1974). R. "Linear and NonLinear failure analysis of Co~nposite Laminates with Transverse shear. 2. W. J. and Reddy.. K. T..
B. H . The Netherlands (1994). G." Acta Matallurgica. 107. "A Progressive Failure Model for Corrlposite Laininates Containing Openings. Hiltner. M. and Osman. Tan.. Delft. Vol." Journal of Composite Materic~ls. Disscrtat.. UK. Turvey. 110. "Effects of Shear Deforrriation on the Onset of Flexural Failure in Symmetric CrossPly Lamirlatcd Rectangular Plates.. (Ed.26(15). "Exact and Approximate Linear and Nonlinear Initial Failure Analysis of Larnir~atedhlirldlin Plates in Flexure. P. 108. Chen. 556577 (1991). 4. F. J. F. and Engblom. Amsterdam." Composite Structures. Latievczc. 105. J . 25. London. A. 104. 43. 103. B. Ph... J.. G . ? "The Mechanics of Matrix Cracking in BrittleMatrix Fiber Cornposites. Damage Mechanics of Composite Matwlals. Elsevier. and Baer. H . 28.. 87 102 (1987).po. UK.).)." Composrtes Science and Technology.J. N. Marshall.." Corn. 257267 (1992).. hlarshall (Ed. arid Le Dantec.5.. 20132021 (1985). Ochoa. Numerzcal Anatysis of Failure Mechanisms i n Fibre Metal Laminates. I . I3. 109. Cox. Delft University Press. 0 . Hashagen. 9. C. "Darnage and Failure Mechanisms of Continuous Glass Fiber Reinforced Polyphenyltme Sulfide." Composite Struct~rrcs. of 106. 141 146 (1987). hlarshall (Ed. I . "Damage Modelling of the Elementary Ply for Laminated Composites. Turvey. Talreja. S. Pipes. E. London. . Y. and Evans. 22892306 (1992). D. Elsevier. A. "Analysis of Progressive Failure in Composites.102. G. D.. E.ion. .rm~l Composite Materials. R.sites Science and Technology. The Netherlands (1998). Elsevier. Composite Materials Series edited by R. 0." Jou.). 133371 (1989). 33(11).
.
2. such theories should be used only when necessary. Figure 11. including the classical laminate theory and firstorder shear deformation laminate theory as special cases.1 shows the kiriernat. However. Analytical as well as finite elerrlent results of this thirdorder theory are developed and numerical results are compared with those of the classical and firstorder theories. . In principle. and they adequately describe the kinematic behavior of rnost laminates. may not require shear correction factors. This avoids the need for shear correction coefficients used in the firstorder theory. However. 11. they involve higherorder stress resultants that are difficult to interpret physically and require considerably more computational effort. theories higher than third order have not been attempted.ics of deformation of a transverse normal on edge y = 0.ThirdOrder Theory of Laminated Composite Plates and Shells 11. w ) as cubic functions of the thickness coordinate. Higherorder theories can represent the kinematics better. The reason for expanding the displacements up to the cubic term in the thickness coordinate is to have quadratic variation of the transverse shear strains and transverse shear stresses through each layer.26] that contains other lowerorder laminate theories.1 Displacement Field The thirdorder plate theory to be developed is based on the same assumptions as the classical and firstorder plate t. and can yield more accurate interlaminar stress distributions.2 A ThirdOrder Plate Theory 11. Here we present the original thirdorder shear deformation laniinate theory of Reddy [25. it is possible to expand the displacement field in terms of the thickness coordinate up to any desired degree. except that we relax the assumption on the straighhess and normality of a transverse normal after deformation by expanding the displacements (u.heories. There are many papers on thirdorder theories (see [I321) and their applications [33521. the displacement fields of these theories are related (see Reddy [49]). Although many of them seen1 to differ from each other on the surface. 11.2. due to the algebraic complexity and cornputational effort involved with higherorder theories in return for marginal gain in accuracy.1 Introduction The classical laminate plate theory and the firstorder shear defornlatiorl theory are the simplest equivalent singlelayer theories. Therefore.
.) and (A.1: Deformation of a transverse normal according to the classical. 0. Clearly. (Q. firstorder.Figure 11..2. and thirdorder plate theories.) are functions to be determined.. A. Consider the displacement field where (@. &). we have .
1) now can be expressed in terms of uo. (11. and the theory derived using the displacement field (11. The numher of dependent unknowns can be reduced by imposing certain conditions.5). 5 ) . wo. we have which in turn requires.There are 9 dependent unknowns.26.2.2.26]: Expressing the above conditions in terms of strains.49] based on the displacenlent field (11. 4 using the relations in Eq.2. Suppose that we wish to impose tractionfree boundary conditions on the top and bottom faces of the laminate [25. Next.2. ?I".j = 4 . 4 and . The weakform finite element models based on the theory require Cointerpolation of all 9 dependent unknowns.1) will result in 9 secondorder partial differential equations. we shall derive a thirdorder theory [25. . for arbitrary Qij ( i .4): . Thus we have The displacement field (11.
7) yields the strains where (c2 = 3cl and cl = 4 / 3 h 2 ) 11. and the virtual kinetic energy SK are given by .3 Equations of Motion The equations of motion of the thirdorder theory will be derived using the dynamic version of the principle of virtual displacements. (3.2 Strains and Stresses Substitution of the displacements (11.11.5) into the nonlinear straindisplacement relations in Eq.2. virtual work done by applied forces 6V. The virtual strain energy 6U.3.2.2.
Substituting for SU. integrating by parts & in Q0 of t o relieve the virtual generalized displacements.2. we obtain the following EulerLagrange equations: .Svo. The same definitions hold for the stress resultants with a hat.11)(11. and using the fundamental lerrirna of calculus of variations. and any differentiation.2. noting that the virtual strains can be written in terms of the generalized displacements using Eqs. (3.14).2.13) where Oo denotes the midplane of the laminate.6 .14) (11. Suo.2. which are specified. (11.Swo. a and P take the symbols x and :y.2.dxdy (11. 6V.4.15) In Eq. and (11.5).2.2. (11. (11.7ac).13) into the virtual work statement in Eq. and SK from Eqs.
2 .25) Secondary Variables : Nnn.. Nns.where Map= M a p  clPap ( a .2.  c2Rn ( a = 4 . 6 ) . Mns  cl [(13iio J ~ & c116. Vn. Mnn.2. Pnn. us. "o) ax + + (1360 + J ~ * . 5 ) (11.2. = Q.P = 1 .] The stress resultants are related to the strains by the relations .24) (11.21) The primary and secondary variables of the theory are a w0 Primary Variables : un. wo. where ds (11.n.c116a~) "o n. 4n. K . Q.
The transformed coefficients Qij are related to Qsj by Eq. dwO/dz. a Fij.30~~) of the order 3 x 3 and those in (11. (11. Note that the matrices in (11. This completes the development of the Reddy thirdorder laminate theory. Hij for i. therefore.. Additional stiffness coefficients are defined by Note that the stiffnesses Eij.. are (11.The stiffnesses in Eq. and (k Dij were given in terms of the layer stiffnesses Qij ) and layer coordinates z k + l and z k in Eqs. j = 4.).5) introduced in the present thirdorder theory.2.2. A simplified thirdorder theory may be deduced from the general thirdorder theory presented here by omitting the higherorder stress resultants (P.4) contains. j = 4. The coefficients Aij.5. Many of these theories were developed for only isotropic plates. However. j = 1 . differential. despite their different looks.3. 6 and Dij and Fij for i .4.19).. 2 .. The displacement field in Eq.3813) and (3. Note that the equations of motion of the firstorder theory are obtained from the present thirdorder theory by setting cl = 0. not an algebraic relationship..PZy) but keeping the higherorder stress resultants (R. EZj.2.30b) are defined for i . R.Pyy. the thirdorder plate theories reported in the literature.2. (8. .2. Bij. (11.1. j = 1 .30b) are of order 2 x 2. the classical plate theory can be .2.5. which is a obtained from this theory only by replacing 4 with cp. as shown in Table 11.3 HigherOrder Laminate Stiffness Characteristics Since a detailed discussion of the laminate stiffnesses was presented in Section 3. We recall that t h e planestressreduced stiffrlesses Qii in the material coordinate system are given in terms of the engineering constants as where the subscript 1 refers to the fiber direction and 2 to the direction transverse to the fiber. are equivalent. as special cases.. 6 and those in Eq.e. The resulting theory is not consistent in energy sense. Even for moderately thick laminates the contribution can be small. brief discussion is presented here for additional laminate stiffnesses (i.30a) are defined for i . and. they are expected to contribute little to thin laminate solutions.49a). Therefore. (3.2. the displacement fields used by other researchers to derive a thirdorder plate theory. + 11. 2 .Fijand so on of the thirdorder theory involve fourth or higher powers of the thickness.
). = u.  Relationship with 4. = u. . pa = 4. +f (z)p.30alb) become &] thickness and for i.5): u.. Vlasov [ 5 ] . Levinson [211. 0. = u. j = 1 .+z$.. Bhimaraddi [27]. c3f (z)O. we have .. f  (4. and Hence. = u. &a = 3 2 Schmidt [16] ZU:.26] Murthy [22]  zu. (11.) Krishna Murty [20] u.. Reference Displacement Field and variables1 u.3.2. 2 .clz3(4.. the nonzero stiffnesses of Eqs.  3 Uo . + u.zu:.2. s $a = 4. u s = u:.+u:. + ~ 1 .1 SingleLayer Plates Single Isotropic Layer For a single isotropic layer of material constants E and u [ G = h . Jemielita [14].. (11.2.Table 11. 11. ) % = U: + f (z)$. =  & (4. +i( z ) ~ .1: Relationship of the displacements of other thirdorder theories to 0 the one in Eq. 6 . Reddy [49] u. Reddy [25.. .
Q55 0 1{ i F } y . The nonzero stiffnesses of Eqs. = 0) The plate constitutive equations for the higherorder stress resultants become (and similar equations hold for N's and M ' s ) { = 4 4. ..The plate constitutive equations (11. the stiffnesses can be expressed in t e r m of the Qij and thickness h. (11. g {":I YTZ + [Q. . A11 [vAll 0 vA11 All 0 0 0 +'~ll ] ($1 YXY Single Specially Orthotropic Layer For a single specially orthotropic layer.2.30qb) becorne (Bij = E 2.28) for the thirdorder theory become { = Nq.2.3.
e.4. 4 . (11.and 6 [see Eq.3. except that the plate stiffnesses are given by Eq.3.1I)].10) and (11. Symmetric Laminates with Multiple Isotropic Layers When isotropic layers of possibly different material properties and thicknesses are arranged symmetrically from both a geometric and a material property standpoint. the resulting laminate will have the following laminate constitutive equations for the thirdorder theories: .11).3. j = 1 . in general. the stiffnesses are expressed in terms of the coefficients Cij and thickness h. (2. (11. The nonzero higherorder stiffnesses are (Bij = 0) for i.3. have the same form as the generally orthotropic singlelayer plates [see Eqs. the principal material coordinates do not coincide with those of the plate).Single Generally Orthotropic Layer For a single generally orthotropic layer (i. the relations between strains and resultants can be further simplified. For certain special cases of symmetric laminates. 5 .13). as explained next.3.3a)l. (11. The plate constitutive equations are the same as in Eqs. the stiffnesses are expressed in terms of the transformed coefficients Qij and thickness h.2 Symmetric Laminates The force and moment resultants for a symmetric laminate. 11. The plate constitutive equations are The higherorder thermal stress resultants for this case are given by Single Anisotropic Layer For a single anisotropic layer.10) and (11.3.. 2 .
.3. this class of antisymmetric laminates have the feature that Fls = F26 = H16 = H26 = 0. (11. about the midplane of the laminate do not exhibit coupling between bending and extension.24) with The thermal stress resultants for this case are given by Symmetric Laminates with Multiple Specially Orthotropic Layers A laminate of multiple specially orthotropic layers that are symmetrically disposed. (3. The coupling stiffnesses Bij and EiJ are not zero.where the laminate stiffnesses Fij and HiJ are defined by Eqs.16).24) with The thermal stress resultants have the same form as those given in Eq. The laminate constitutive equations are again given by Eqs.3.3.17) 11.13)( 11. (11. (11.3. where the laminate stiffnesses Fij and Hi. both from a material and geometric properties standpoint. The relations between the stress resultants and the strains are .2.3 Antisymmetric Laminates Due to the antisymmetry of the lamination scherne but symmetry of the thicknesses of each pair of layers.5. are defined b y Eqs.
t ) = O (11. and all other Eij = Dq5 = F4.4.2.20) are very similar in form to the firstorder shear deformation theory. = O.1): ~) ) ) ~ 0 ( x .21) Ez2 = . t. wO(x. t= 0 $ z ( ~ . 5 = F45 = I3 ~ = = I7 = 0 (11. t = O ) 7 Y. = O VO(O. 26.t)=O. ~ o ( x . = O7 &(O.5. t ) = 0 . 11. y . The L&y solutions are presented in Section 11. it is possible to develop the Navier solutions of simply supported antisymmetric crossply and angleply laminates using the thirdorder theory (see References 25.4 The Navier Solutions 11. In fact. we develop the Navier solutions of antisymmetric angleply and crossply laminates. w o ( a . $ l c ( ~ .5 = 13 = 1 5 = I7 = 0 (11.3~1) .1 Preliminary Comments The equations of motion of the thirdorder theory of Reddy presented in Eqs. the stiffnesses E16 and Ez6 go to zero as the number of layers in the laminate increases. Y. b . the coupling coefficients El1 can be shown to approach zero as the number of layers increases. w 0 ( 0 . t. . (11. and finite element models are discussed in Section 11.5 = 0 For regular antisymmetric crossply laminates.22) D4. For antisymmetric crossply laminates the following stiffnesses are zero: AI6 = A26 = A45 = B I 6 = B 2 = DI6 = D26 = II = 0 ~ E16 = E 2 = F16 = F26 = H16 = H26 = 0 4 .E l l . t ) = O .6. b .4.4. b .2) The SS1 boundary conditions for the thirdorder shear deformation plate theory are (see Figure 11.5 = F45 = F16 = F26 = HI6 = Hz6 = Ell = Ez2 = E I 2 = EG6= 0 For a fixed laminate thickness.O.3. t= O. w O ( x . 0 7 = 0 . v ~ ( aY. t . and 29).16)(11.4.4. O .1) For antisymmetric angleply laminates the following stiffnesses are zero: A I 6= Ell = Elz = A45 = B I 1 = B I 2 = B22 = &j6 = D16 = DZ6= = E22 = E66 = F16 = F26 = H16 = HZ6= 0 4 5 II = 0 = F4. ~7 t.3.2. y . This completes the development of the thirdorder theory of Reddy.Antisymmetric CrossPly Laminates For antisymmetric crossply laminates the coupling stiffnesses have the following properties: (11. Antisymmetric AnglePly Laminates For antisymmetric angleply laminates the stiffnesses can be simplified as (11. In the next section.
at x = O and x=a Figure 11. TG at y=0 and y=b Figure 11. The SS2 boundary conditions for the thirdorder shear deformation plate theory are (see Figure 11.2) In the following sections.4.4.2: Simply supported (SS2) boundary conditions for antisyrnnietric angleply laminates. we present the Navier solutions of crossply laminates for the SS1 boundary conditions and antisymrnetric angleply laminates for the SS2 boundary conditions.4.1: Simply supported (SS1) boundary conditions for antisymmetric crossply laminates. .
(11. Ymn) the of Navier solution of crossply laminates are governed by . t ) = ( n=l m=l 0 0 0 0 C C Umn(t)cos a x sin P:y Vmn t )sin a x cos /3y ( (11..2.16)(11. t ) = xx where a = m ~ / and /3 = nrlb.y. Wmn.3a. Xmn.5) and (11.1) hold. (11. (11.y.y.4.4.5a) vo ( x .4.20) will show that the Navier solution exists only if the laminate stiffnesses are such that the conditions in Eq. The transverse load q is also expanded in double a Fourier sine series ~ ~ = ~ ab ( la /' o t b o ) q ( x .b) are satisfied by the following expansions: uo x .2. The coefficients (Urn.6) into Eqs. t ) sin a x sin dxdy Substitution of Eqs.4. Vmn.4.11A . 2 Ant isymmetric CrossPly Laminates The boundary conditions in (11.
buckling.13)l The ordinary differential equations (11.7) can be specialized to static bending analysis. (6.The thermal resultants are defined by [see Eqs.3.7) in time can be solved for transient response using the Newmark integration procedure described in Chapter 7. and natural vibration. . Equation (11.11) through (6.4.4.3.
In the absence of thermal effects they are given by . PUmn + a v m n +aymn (11.3. (6.The inplane stresses in each layer can be computed from the equations [see Eqs.16) + where y = 4/h2.3. (11.29) and (6.30)] where m=l n=l ( (REn (Rgyn + zS$% + c1z3~$:) sin a x sin py + zSgn + c 1 z 3 ~ $ ~i)n a x sinpy s ~+ Z2~ C~ ~ ~ 2 ~ ~ 2cos PY S + cos %) PXm. Note that the transverse shear stresses are layerwise quadratic through the thickness.4.4. The transverse shear stresses can also be determined using the equilibrium equations of 3D elasticity. Y 01%~) a x cos 0y sin (Xmn+ a W) cos a x sin py .15b) (pxmn + a y m n + 2aPWmn) The transverse shear stresses from the constitutive equations are given by 0 0 0 0  m=l n=l (.
q&) have the same expansions as in Eqs.20) [${A) [ ~ ? { d >F ) ={ + with the following coefficients The mass and coefficients with hat and overbar are the same as those defined in Eqs.4.5~e). 1 1KrLn(t) a x sin Py cos OC OC (11.9). I ) = y.Substituting the expansions in Eqs. .4. (1l.b) into Eqs.4. (11.3 Antisymmetric AnglePly Laminates The simply supported (SS2) boundary conditions in (11.2.19b) and (wo.4. (11.2.4.8) and (11.4.4.b) are satisfied by uO(x. and thermal effects are not considered.16)(11.&.7a) (11.4a. (11.19a.4.4. we obtain equations of the form in (11.20).
23~) The transverse stresses determined from the equilibrium equations of 3D elasticity are where .) cos a x cos Py + + + + (11. a2wmn) a x sin Py sin (BYmn p2 Wmn) a x sin Py sin .(px. v COS COS + 1 (11. cos a x cos py ..The inplane stresses in each layer can be computed from the equations QU.4.(@Urnn aVmn) sin ax sin Py p. aYmn 2aPW....23a) &g YXY =I . m=l n=l 0 0 0 0 CC { (ax..4.
and classical laminate plate theory (CLPT) for the following two problems (see Reddy [25]): 1.4 GPa).4. A fourply (0/90/90/0) square (alb = 1) laminate with layers of equal thickness and subjected to sinusoidally distributed transverse load. thirdorder shear deformation plate theory (TSDT).1 and 11. The material properties of a ply are assumed to be Material 1: El = 25 x lo6 psi (175 GPa).4 Numerical Results Bending Analysis Tables 11.26) The following nondimensionalized quantities are reported in the tables: a b h (&) The origin of the coordinate system is taken at the lower left corner of the plate. 2. .11.2 x lo6 psi (1.25 (11.5 x lo6 psi (3. ~ 1 = 2113 = 2 0. A threeply (0/90/0) square (alb = 1) laminate with layers of equal thickness and subjected to sinusoidally distributed transverse load.4. firstorder shear deformation theory (FSDT).5 GPa) G23= 0.2 contain nondimensionalized center deflections and stresses obtained with 3D elasticity theory (ELS).4. E2 = 10"si (7 GPa) Glz = GI3 = 0.4.
Table 11. FSDT alh Variable CLPT t 3D elasticity solution of Pagano [53]. $ The second line corresponds to stresses computed from 3D equilibrium equations.4.1: Nondimensionalized center deflections and stresses in simply supported (SS1) threelayer (0/90/0) square laminates under sinusoidally distributed transverse load.4. square 4 ELS t TSDT FSDT ELS TSDT FDST ELS TSDT FDST ELS TSDT FDST 10 20 100 CLPT t 3D elasticity solution of Pagano and Hatfield [54]. .2: Nondimensionalized maximum deflections and stresses in simply supported (SS1) symmetric crossply (0/90/90/0) laminates under sinusoidally distributed transverse load.Table 11. 4 Equilibriumderived stresses.
The fact that no shear correction coefficients are needed in the thirdorder theory makes it more convenient to use. respectively. symmetric crossply laminate (0/90/90/0) under sinusoidally distributed load. In general.4% in TSDT and 11.4 and 11.5 show distributions of nondimensionalized [a = ~ ( h ~ / ~ ~ a ~ ) normal stresses . Figure 11.(h/qoa) and ay.(h/qoa) are shown as functions of thickness in Figures 11. Plots of constitutively derived and equilibriumderived transverse shear stresses . the thirdorder theory underpredicts deflection by 3% while the firstorder theory underpredicts by about 12. Material 1). a1 h 45 50 Figure 11.5% for a / h = 4. the errors are 2.3 contains plots of nondimensionalized center deflection. symmetric crossply laminate (0/90/90/0) under sinusoidally distributed load (Material 1. Figures 11. for a square. w = ' ~ u ~ ( ~ ~ h ~ / q ~ a ~ ) . versus sidetothickness ratio a / h for Problem 2 (a square. a / h = 10). the equilibriumderived transverse shear stresses compare more favorably with the elasticity solution than those obtained from the constitutive equations for equivalent singlelayer theories. The thirdorder theory predicts a cubic variation whereas the classical and firstorder theories predict linear variation of the stresses. for a / h = 10.4.7.4.6 and 11. predicted by the classical. .@ .3: Plots of nondimensionalized center transverse deflection versus sidetothickness ratio of a symmetric crossply (0/90/90/0) laminate under sinusoidally distributed load (Material I ) . Compared to the elasticity solution.8% in FSDT.The stresses 0 5 10 15 20 25 30 35 40 Sidetothickness ratio.@ and eyY maximum ] . The errors are much less at lower values of a l h .From the results it is clear that the thirdorder theory gives more accurate results for deflections and stresses when compared to the firstorder shear deformation plate theory with K = 516. firstorder.4.4. It is known that the shear correction factor K depends on the lamina properties and the stacking sequence. and thirdorder plate theories through the thickness of a square symmetric crossply laminate (0/90/90/0) under sinusoidally distributed load (Material 1.4. a / h = 4 and 10). = oy.4. = aX.
2 0. firstorder.4.2 0..4 0. distributions predicted by the classical.6 0. firstorder. alh=4 0. Figure 11.Figure 11.6 0.5: Comparison of normal stress avl/distributions predicted by the classical..8 FSDT. .4 0..0 0. +TSDT. and thirdorder plate theories.. alh=4 Stress. . and thirdorder plate theories.8 0.4: Comparison of center normal stress a.4.
z) 0....... . .....20 0...04 0.. FSDT (C) TSDT (E) TSDT (C) ........00 0. .. ..4..... CLPT (E) FSDT (E) .....30 0. as functions of thickness coordinate.........0... ..10 0. + CLPT (E) (E) FSDT (C) TSDT (E) TSDT (C) (El: equilibriumderived (C): constitutivelyderive' 0.. ..16 Stress.60 Figure 11.08 0... (a/2.........z) 0.... . 5..........20 Figure 11. FSDT ..7: Plots of constitutively derived (C) and equilibriumderived (E) transverse shear stresses ag..b/2....... G (O.. .4...... as functions of thickness coordinate. .00 0. + (El: equilibriumderived (C): constitutivelyderivec 0..40 Stress..6: Plots of constitutively derived (C) and equilibriumderived (E) transverse shear stresses a.12 0....50 0.
3) square laminates under sinusoidal loading are presented in Figure 11. The thirdorder theory correctly satisfies vanishing of transverse shear stresses at the top and bottom of the laminate. bendingstretching coupling is negligible for laminates with six or more layers.) antisymmetric angleply laminates under sinusoidal loading with different fiber orientations.and sixlayer (01 .9 for various ratios of moduli (GI2 = G13 = 0.4.3) square laminates under sinusoidal loading (SS1) for a / h = 10.4 contains a comparison of the maximum deflections of two. The effect of coupling between bending and extension is increasingly more pronounced with an increasing degree of material anisotropy.5411 0. The deflections of antisymmetric angleply (45145).6382 0.5052 0.4635 TSDT 1. 2 4 2 = 0.3: Maximum deflections.4688 0. because the displacement field in TSDT is derived to satisfy these conditions a priori. a h FSDT TSDT FSDT 1.4635 CLPT 1. and a / h = 10.6E2. the coupling terms cannot be neglected. Figure 11.4. .6354 0. Table 11.4.5060 0.25. G23 = 0.4. (n = 1. G23 = 0. while the stresses computed using constitutive equations are always discontinuous for all equivalent singlelayer theories due to the continuity of the transverse shear strains through thickness of the laminate.) square plates subjected to sinusoidally distributed transverse load. . G12 = G13 = 0. The following material properties are assumed Material 2: . (n = 1. G x lo2.0636 Table 11.8 shows the effect of material anisotropy on the deflections of antisymmetric crossply (0/90). As noted earlier.4687 0.2E2.derived using the equilibrium equations are continuous through thickness because they are made to satisfy the interface continuity conditions (in determining the constants of integration).5E2.5473 0. .5E2.8 / .4. . The results predicted by FSDT and TSDT are very close. hence the deflections of a twolayer laminate without accounting for the coupling are virtually the same as those obtained for the sixlayer laminate (see Table 11. of simply supported (SS1) antisymmetric crossply (0/90/0/90/. 7/12 = 0.25.3). it is interesting t o note that the firstorder theory overpredicts deflections for the twolayer case and underpredicts for the sixlayer case when compared to TSDT. a / h = 10). Even at low modulus ratios.
[Z] TSDT 1 Figure 11. 0 0 5 10 15 20 25 30 35 40 El432 Figure 11. 0 0 0 .1 ..4. .9: The effect of material anisotropy on the nondimensionalized deflections w of simply supported (SS2) antisymmetric angleply (451 .3) square laminates. 3 ) square laminates.4. (n = 1.45).8: The effect of material anisotropy on the nondimensionalized deflections w of simply supported (SS1) antisymmetric crossply (0/90). (n = 1 .
a  h Source n=2 n =6 n =2 n=6 n=2 n=6 4 10 TSDT FSDT TSDT FSDT TSDT FSDT TSDT FSDT TSDT FSDT CLPT 20 50 100 Natural Vibration Fundamental frequencies. 30°.) antisymmetric angleply square plates (SS2) with 6 = 5".B26. G23 = 0.10. of simply supported laminates are presented for the following three cases: 1.4.) are the largest at this particular fiber orientation for a given number of layers. .5 contains the nondimensionalized fundamental frequencies w = w ( a 2 / h ) J m of the first laminate as a function of modulus ratio E1/E2 (G12 = G13 = O.25) for a / h = 5 and 10. 2 4 2 = 0.E26.. . . 3. Table 11.4.4. w x lo2. Fourlayer (0/90/90/0) symmetric crossply square plate (SSI).5E2..6 contains natural frequencies of twolayer and sixlayer antisymmetric angleply laminates. Table 11.El6.4: Maximum deflections. 6 = w l l ( a 2 / h ) J m .As in the case of antisymmetric crossply plates. Two. TSDT does not require shear correction factors.4. The rotary inertias are included in all cases and theories. Two.) square plates subjected to sinusoidal loading. Table 11. . .) antisymmetric crossply square plates (SS1). The fundamental natural frequencies 0 of antisymmetric crossply laminates (Material 2) as functions of the sidetothickness are presented in Figure 11. . of simply supported (SS2). the coupling causes a significant reduction of the plate stiffness. with the most critical case being for 6 = 45".and sixlayer (0/90/. All these results indicate that there is no significant difference between the predictions of FSDT and TSDT.and sixlayer (8181. 2. This observation can be explained by the fact that the magnitudes of the bendingstretching terms (B16. . and 45". antisymmetric angleply (616/ .6E2. . However.
(n = 1 .0 10 20 30 40 50 60 70 80 90 100 Sidetothickness ratio.5: Nondimensionalized frequencies w of (0/90/90/0) laminates as functions of modulus ratio.4. a l h Figure 11. crossply t Approximate 3D solutiori of Noor [55] .4.10: Plots of nondirnerlsionalized fundamental frequency versus sidetothickness ratio for crossply (0/90). 3 ) square laminates. Table 11.
4. Table 11. Table 11. The results are also compared with approximate 3D elasticity results obtained by Noor [57].4. . a  h CLPT FSDT TSDT .4. In Figure 11. 3 = wll  f &.7: Nondimensionalized uniaxial buckling loads. The buckling loads of the same laminate as a function of modulus ratio E 1 / E 2 are presented in Table 11.) antisymmetric crossply laminates are shown as a function of the sidetothickness ratio. of simply supported (SS1) symmetric crossply (0/90/90/0) square plates.4.11. a  h Source TSDT FSDT CLPT TSDT FSDT CLPT TSDT FSDT 4 10 20 CLPT 50 TSDT FSDT CLPT TSDT FSDT CLPT 100 Buckling Analysis The uniaxial critical buckling loads of a fourlayer (0/90/90/0) crossply plate (Material 2) with various sidetothickness ratios are compared in Table 11.) square plates. . . .4. Both TSDT and FSDT give very good results and the difference between them is not very significant. the buckling loads of twolayer and sixlayer (0/90/0/.8. antisymmetric angleply (818/ .4. of simply supported (SS2).6: Nondimensionalized fundamental frequencies.9 contains critical buckling loads for the twolayer and sixlayer antisymmetric angleply laminates. N = N~~ 6.7.Table 11.
4. ul2 = 0. natural vibration.1 Preliminary Comments The L6vy type solutions for bending.5.6E2. .9: Nondimensionalized uniaxial buckling loads. and buckling of rectangular laminates of crossply constructions have been developed for the thirdorder theory of Reddy (see [34.25. the reader may consult references at the end of the chapter.48. of symmetric crossply (0/90/90/0) square plates (G12 = G13 = 0. N = N.42.41.50521).5E2. a l h = 10).5 L&y Solutions of CrossPly Laminates 11. 0 = 5" Source 5 10  0 = 30" 0 = 45" n =6 n =2 n=6 n=2 7~=2 n=6 TSDT FSDT TSDT FSDT TSDT FSDT TSDT FSDT TSDT FSDT CLPT 20 50 100 11. the governing equations appropriate for the antisymmetric crossply laminate construction are given by . For the static case.. El /E2 CLPT FSDT TSDT ELS [57] Table 11.44.51]).Table 11.&. . For additional results. G23 = 0. N = N. .4..a2 of simply supported (SS2) antisymmetric angleply (01O/ .) square plates.8: Effect of material anisotropy on the uniaxial buckling loads. Here we present the solutions for static bending of crossply laminates (see Khdeir and Reddy [50.
4.. and "d 01  ax.alh Figure 11.3) square laminates subjected to uniaxial compressive load (Material 2). ( U ~ / E ~ ~versus sidetothickness ~ ) ] ratio for antisymmetric crossply (0/90).11: Critical buckling load [N = N. . (n = 1. 8 2 XI = x and 22 =y .
1: The coordinate system (a12 5 x a/2. Figure 11. in conjunction with the statespace concept can be used to determine bending solutions of crossply laminated rectangular plates with two parallel edges simply supported and other two having arbitrary combination of boundary conditions. clamped.5. 0 5 y 6) and boundary conditions used on the simply supported edges for the Lkvy solutions of rectangular crossply laminates using the thirdorder shear deformation theory. Suppose that the edges y = 0. b are simply supported.sin a b The displacement quantities are represented as where /3 = m73lb. = P..11.5. b : uo = wo = $hz = N. while the remaining ones (x = fa / 2 ) may have arbitrary combinations of free..4) A sinusoidal distribution of the transverse load is considered 732 73y q(x.5. < < . y) = q0 cos . =My. =0 (11. The generalized displacements may be expressed as products of undetermined functions and known trigonometric functions so as to identically satisfy the simply supported boundary conditions at y = 0 .1). and simply supported edge conditions (see Figure 11.2 Solution Procedure A generalized Lkvy type solution.5.
5.1) results in the following five differential equations: where primes denote the derivative with respect to x.6) into Eqs. (11.Substitution of (11A. and the coefficients Ci are defined by where .
In order to reduce the system of equations (11.5.7) to a statespace form, the components of the state vector Z(x) are defined as follows:
Z1 = U,, Z2 = U:,, Z3 = V , Z4 = Vh, Z5 = W,, Z6 = Wk, Z7 = Wi, Z8= W ) Z q = X l n , Z1O=X:rL, Zll =Y,, Z12 =Y:, i" (11.5.11)
Using the definitions (11.5.1I ) , the systems of equations (11.5.7) may be converted to the firstorder differential operator form Zt=AZ+r where the matrix [A] is a 12 x 12 matrix (11.5.12a)
and the load vector r is
{T)
= (0,
fF sin a x , 0, f? cos a x , 0, 0, 0, f;
cos a x , 0,
fr
sin a x , 0, f 5T cos a x ) T (11.5.13)
The solution to Eq. (11.5.12a) is given by
Here K is a column vector of constants to be determined from the edge conditions, and
where n = 12, X i denote the distinct eigenvalues of [A], and [S]denotes the matrix of eigenvectors of [A]. The boundary conditions for simply supported (S), clamped (C), and free (F) at the edges x = fa / 2 are
where the stress resultants Pap and R, of the thirdorder theory are defined in Eq. (11.2.14).
11.5.3 Numerical Results
The nondimensionalized center transverse deflections and stresses of twolayer and tenlayer crossply laminates under sinusoidally distributed transverse load are presented in Table 11.5.1 for various boundary conditions and sidetothickness ratio of a / h = 5. The nondimensionalized variables used are
For the thick plates considered in this case, there is a significant difference between the results predicted by TSDT and FSDT; FSDT slightly overpredicts deflections and underpredicts stresses.
Table 11.5.1: Nondimensionalized deflections and stresses of antisymmetric
crossply square plates for various boundary conditions ( a l h = 5, E1/E2= 25, GI2 = GIs = 0.5E2, G23 = 0.2E2, ~ 1 = 0.25). 2
No. uf Theory Variable
SS
SC
CC
FF
FS
Layers
2
TSDT
FSDT
CLPT
10
TSDT
FSDT
CLPT
Tables 11A.2 and 11.5.3 contain deflections and stresses in crossply laminates subjected to a sinusoidally distributed temperature field (see [50])
T ( z ,y, z) = zT1 cos a x sin By
The following nondimerlsionalized variables are used:
(11.5.20)
The difference between the results obtained with TSDT and FSDT is insignificant for sidetothickness ratios greater than 5. Additional numerical results based on the Lkvy solution technique for TSDT are presented along with the finite element results in Section 11.6.3. For additiorlal results the reader may corlsult the references at the end of the chapter. In the next section we develop the displacement finite element model of the Reddy thirdorder theory.
Table 11.5.2: Nondimensionalized center deflections
Laminate
blh
of crossply square plates subjected to sinusoidally distributed temperature distribution.
Theory TSDT FSDT CLPT TSDT FSDT CLPT TSDT FSDT CLPT TSDT FSDT CLPT TSDT FSDT CLPT TSDT FSDT CLPT SS
0 / 9 0 / ... 10 layers
5
TSDT FSDT CLPT TSDT FSDT CLPT
11.6 Finite Element Model of Plates
11.6.1 Introduction
Recall from Eq. (11.2.24) that the primary variables of the thirdorder theory developed in Section 11.2 are (u,, us, wo, wo,, = aw0/an, &, q5s), where (u,, u,) denote inplane normal and tangential displacements, and (&, 4,) are the rotations of a transverse line about the inplane normal and tangent. A displacement finite element model based on Eqs. (11.2.16) through (11.2.20), with (uo,vo , wo, W O , ~$ dy) as the primary variables, is called a displacement , , finite element model (see Phan and Reddy [30]), and it requires the Lagrange interpolation of (uo,vo, q5,, &) and Hermite interpolation of wo. A conforming element will have eight degrees of freedom (uo, vo, wo, W O , ~W O , ~W O , , ~ , 4,, 4,) , , whereas a nonconforming element will have (uo,vo, wo, wo,, , &, 4,) seven degrees of freedom per node.
Table 11.5.3: Nondimensionalized thermal stresses of crossply square plates
subjected to sinusoidally distributed temperature.
Stress
Laminate
b/h 5
Theory
SS
SC
CC
FF
FS
FC
a,,
019010
TSDT FSDT CLPT
TSDT FSDT CLPT TSDT FSDT CLPT TSDT