This action might not be possible to undo. Are you sure you want to continue?
j j j j j j j j j j j j j j j
STRESSES IN PLATES AND SHELLS
STRESSES IN PLATES AND SHELLS
A. C. Ugural,
Ph.D.
Profe."sol' and Chairmall M l!l '/wllic:al Eugim"'ring DeplIrtmem Faide igh Dickillson University
McGrawHili Book Company
New York St. Lou is San Francisco Auckl~md Bogota Hamburg Johannes burg London Madrid Mexico Montreal N('w Delhi Panama Paris Sito Paulo Singapoft: Sydney Tokyo Toronto
This book was set in Times Roman. The editor was Frank J. Cerra; the production supervisor was Donna Piligra. Fairfield Graphics was printer and binder.
STRESSES IN PLATES AND SHELLS Copyright © 1981 by McGraw~HiIl, Inc. All rights reserved. Printed ill th!! Ullitt!d States of America. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by means, electronic, mechanical, photocopying, recording, or otherwise, with om the prior written permission of the publisher. 1234567890FGFG8987654321
Library of Congress Cataloging in Publication Data
Ugura), A C Stresses in plates and shells.
Includes bibliographical references and index. I. Plates (Engineering) 2, Shells (Engineering) 3. Strains and stresses, l. Title. TA660.P6U39 624.1'776 8013927 ISBN 0070657300
5 2.12 Effect of Shear on the Plate Deflection Circular Plates under a Concentrated Load at Its Center Annular Plates with Simply Supported Outer Edges Deflection and Stress by Superposition The Ritz Method Applied to Circular Plates on Elastic Foundation Asymmetrical Bending of Circular Plates Deflection by the Reciprocity Theorem Circular Plates of Variable Thickness under Nonuniform Load Problems 37 40 41 42 44 46 47 55 v .2 2.8 2.10 Introduction General Behavior of Plates Strain~Cllrvature Relations Stresses and Stress Resultants Variation of Stress within a Plate The Governing Equation for Denection of Plates Boundary Conditions Methods for Solution of Plate Deflections Strain Energy Methods Mechanical Properties and Behavior of Materials Problems ix xiii 1 I 2 4 6 10 12 14 16 20 22 25 27 27 2 Circular Plates 2.1 lot roduction 2.6 1.1 1.10 2.5 1.3 1.4 1.7 2.8 1.9 1.6 2.7 1.4 Basic Relations in Polar Coordinates The AsixYlTImetricai Bending Uniformly Loaded Circular Plates 27 31 32 36 2.9 2.3 2.CONTENTS Preface List of Symbols 1 Elements of Plate Bending Theory 1.2 1.11 2.
1 Plates of Various Geometrical Forms Introduction Method of Images Equilateral Triangular Plate with Simply Supported Edges Elliptical Plates Stress Concentration around Holes in a Plate Problems 4.8 5.5 6.2 4.5 5.5 5 5.6 5.2 13 3.4 4.\'i CO~TI':.10 \06 106 III 119 122 123 126 128 129 133 137 140 140 6 6.3 Plate Bending by Numerical Methods Introduction Finite Differences Solution of the Finite Difference Equations Plates with Curved Boundaries The Polar Mesh The Triangular Mesh Properties of a Finite Element Formulation of the Finite Elcmt:nt Method Triangular Finite Element Rectangular Finite Element Problems 106 5A 5. rs 59 3 3.3 6.2 6.4 6.3 4.1 6.9 5.10 3.1 5.6 6.11 3.4 3.2 5.1 3.5 16 3.9 3.7 Rectangular Plates Introduction Navicr's Solution for Simply Supported Rectangular Plates Simply Supported Rectangular Plates under Varioll~ Loadings Levy's Solution for Rectangular Plates Levy's Method Applied to Nonuniformly Loaded Rectangular Pl<ltt~ Rectangular Plates under Distributed Edge moments Method of Superposition Applied to Bending of Rectangular Plates The Strip Method Simply Supported Continuous Rectangular Plates Rectangular Plates Supported by Intermediate Columns Rcclangular Plates Oil Elastic Foundation The Ritl Method Applied to Bending of Rectangular Plates Problems 59 59 62 66 74 75 77 80 83 3.7 5.12 86 89 90 94 96 96 96 98 100 102 105 4 4.8 3.7 Orthotropic Plates Introduction Basic Relationships Determination of Rigidities Rectangular Orthotropic Plates Elliptic and Circular Orthotropic Plates Multilayered Plates The Finite Element Solution Problems 141 142 144 147 148 150 152 .
Strain.1 7.2 Bending Stresses Introduction Shell Stress Resultants 111 Shells 231 231 231 .6 10.6 Comparison of Small.8 Stress.6 7.1 8.6 9.5 10.CO:\TF0:TS \'jj 7 7.3 7.5 7.1 10.9 10.4 7.4 9. Buckling Application of the Energy Method The Finite Difference Solution Plates with Small Initial Curvature Bending to a Cylindrical Surface Problems 163 167 16S 172 174 8 8.:.2 Plates under Combined Lateral and Direct Loads Introduction Governing Equatio!l for the Deflection Surface J 53 153 153 156 159 7. and Displacement Relations Stress Resultants The Governing Differential Equations Simply SupP0ricd Rectangular Plate Subject to an Arbitrary Temperature Distribution Simply Supported Rectangular Plate with Temperature Distribution Varying over the Thickness Analogy between Thermal and Isothermal Plate Problems Axisymmctrically Heated Circular Plates Problems 111 185 185 186 187 188 189 190 191 193 197 10 Membrane Stresses 10.2 9.7 10. 10.7 Compression of Plates.10 215 217 218 220 224 226 11 11.and LargeDellection Theories The Governing Equations for Large Deflections Deflections by the Ritz Method The Finite Element Solution 177 179 181 184 Problems 9 Thermal Stresses 9.5 9.2 Large Deflections of Plates Introduction Plate Behavior When Dellcctions Are Large 174 174 175 8.1 J 1.4 Shells 198 198 199 General Behavior and Common Theories of Shells Load Resistance Action of a Shell Geometry of Shells of Revolution Symmetrically Loaded Shells of Revolution Some Typical Cases of Shells of Revolution Axially Symmetric Deformal ion Asymmetrically Loaded Shells of Revolution Shells of Revolution under Wind Loading Cylindrical Shells of General Shape Breakdown of Elastic Action in Shells Problems 202 203 205 10.1 Introduction III Plates 9.8 10.3 10.7 9.3 SA 8.5 8.3 9.
7 13.4 12. and Displacement Relations Compound Stresses in Shells Strain Energy in the Bending and Stretching of Shells Axisymmetrically Loaded Circular Cylindrical Shells A Typical Case of the Axisymm~trjcally Loaded Cylindrical Shells Shells of Revolution under Axisymmctrical Loads Governing Equations for Axisymmetrical Displacements Compari:.3 13. Tanks.5 12.9 11.11 Applications to Pipes.9 12. Moment.6 13.3 11.10 11.5 13.7 12..3 12.11 11.·iii ("():\TE~TS 11.2 297 297 299 301 303 303 304 307 309 313 References Answers to Selected Problems Index .on of Bending and Membrane Stresses The Finite Elcmcnt Representations of Shells of General Shape The Finite Element Solution ofAxisymmetrically Loaded Shells Prohlcms 2.4 13.6 12.8 Cylindrical Shells under General Loads Introduction Differential Equations of Equilibrium Kinematic Relationships The Governing Equations for Deflections A Typical Case of Asymmetrical Loading lnextensional Deformations Symmetrical Buckling under Uniform Axial Pressure Nonsymmetrical Buckling under Uniform Axial Compression Problems 296 Appendixes A A.7 11.12 Force.2 13.3 B Fourier Series Expansions Single Fourier Series HairRange Expansions Double Fourier Series Solution of Simultaneous Linear Equations Introduction The Gauss Reduction Method Rl B.2 12.1 13.4 11.2 A.1 12.6 11.5 11.10 12.~ 11.8 12.D 235 236 236 239 243 245 247 249 249 25:! 254 254 255 258 260 263 266 269 270 12 12. and Pressure Vessels Tntroduction Pipes Subjected to Edge Forces and Moments Reinforced Cylinders Cylindrical Tanks Thermal Stresses in Cylinders Thermal Stresses in Composite Cylinders Discontinuity Stresses in Pressure Vessels Cylindrical Vessel with Hemispherical Heads Cylindrical Vessel with Ellipsoidal Heads Cylindrical Vessel with Flat Heads Design Formulas for Co~vcntional Pressure Vessels Problems 272 273 274 275 278 278 279 280 282 284 287 290 293 13 13.l A.
both so useful in performing stress analysis in a more realistic setting. and science curricula. so important when computer·based solutions are employed. civil. tanks. including domes. and beams. the stress placed upon numerical methods is not intended to deny the merit of ciassical analysis which is given a rather full treatment. However. The volume attempts to fill what the writer believes is a void in the world of texts. Above all. Emphasis is given computer oriented numerical techniques in the solution of problems resisting mzalytical approaches. As sufficient material is provided for a full year of study. mechanical. The book offers a simple. Theories of failures are employed in predicting the behavior of plates and shells under combined loading. plates. the book may stimulate the development of courses in aduam:ed statics and structural analysis. A balance is presented between the theory necessary to gain insight into the mechanics and the numerical solutions. The reader is helped to realize that a firm grasp of fundamentals is necessary to perform the critical interpretations. the ix . Throughout the text.PREFACE The subject matter of this text is usually covered in onesemester senior and onesemester graduate level courses dealing with the analysis .{plates and shells. The text is intended to serve a twofold purpose: to complement classroom lectures and to accommodate the needs of practicing engineers in the analysis of plate and shell structures. The material presented is applicable to aeronautical. an eflort has been made to provide a visual interpretation of the basic equations and of the means by which loads are resisted in shells. comprehensive. chemical.. and pipes. The coverage presumes a knowledge of elementary mechanics of materials. astronautical. and methodical presentation of the principles of plate and shell theories and their applications to numerous structural elements. pressure vessels. and ocean engineering: engineering mechanics.
many of which are drawn from engineering practice. These fomlUlations usually necessitate the employment of tensor notation. often classical engineering examples are used to maintain simplicity and lucidity. The chapters have been arranged in a sequence compatible with an orderly study of the analysis of plates and shells. The book is organized so that Chap. Thc International System of Units (SI) is used. The expression defining the small lateral deflection of the midplane of a thin plate is formulated two ways. Recent publications dealing with shell theory include analytical presentations generally valid for any shell under any kind ofloading. The finite element method is applied to treat plates of nonuniform thickness and arbitrary shape as well as to represent shells of arbitrary form. numerical methods amenable to computer solution. the classical approaches requiring extensive mathematical background are not treated. and applications of the formulas developed and of the theories of failures to increasingly important structural members. 10 and 11 should be read before Chaps. and practical applications. 1 must be studied first. vector analysis. The former approach. is regarded as more appealing to the engineer. and is equally used in the case of thin shells. tion. Because of the introductory nature of this book. With regard to application. The theory introduced in this text is a special case of the above. The remaining chapters may be taken in any sequence except that Chaps. which requires less mathematical rigor but more physical interpreta. analyses of plates and shells made of isotropic as well as composite materials under ordinary and high temperature loadings. .author has attcmpted to provid~ the fundamcntals of theory and application necessary to prepare students for mon: advanced study and for professional practice. A variety of problems is provided for solulion by the student. This text oITers a wide range of fully worked out illustrative examples. approximately 170 problem sets. Numerous alternatives are possible in making selections from the book for two singlesemester courses. The first utilizes the fundamental assumptions made in the elementary theory of beams. 12 and 13. a multitude of formulas and tabulations of plate and shell theory solutions from which direct and practical design calculations can be made. Emphasized also are the energy aspecls of plate and shel! bending and buckling because of the importance of elJel'gy methods in the solution of many reallife problems and in modern computational techniques. Development of the physical and mathematical aspects of the slIbject is deliberately pursued. The volume may be divided into two parts. Chapters 1 to 9 contain the fundamental definitions and the analysis of plates. and a system of curvilinear coordinates. The equations governing shells arc developed only to the extent necessary for solving the mOre usual engineering problems. The author has made a special effort to illustratc important principles and applications with numerical examples. [lre given emphasis. The second is based upon the diITerential equations of equilibrium for the threedimellsiollal stress. The physical significance of the solutions. Chapters 10 to 13 deal with shells.
Stanek for her most skillful services in the preparation of tile manuscript. C. Fellster read the entire manuscript and made many corrections for which the author is most grateful. S. K. Uqural . and to Mrs. Dr. B. A. 7 and 8. H. who read Chaps. Lefkowitz.PRFf A(T xi Thanks arc due to thL' many studellts who ofkred constructive suggestions when drafts of this work were used as a text. He is indebted to Dr.
.
.\.. M... twisting moment per unit distance on axial plane of cylindrical shell N normal force per unit distance N" critical compressive load per unit distance xiii . . 11 M M* M x' M)' . twisting moment per unit distance on radial plane meridional bending moment per unit distance on parallel plane M. b area. numerical factor.LIST OF SYMBOLS It a..2)] elasticity matrix modulus of elasticity resultant external loading on shell element modulus of elasticity in shear acceleration of gravity (::e 9. dimensions. meridional bending moment per unit distance on parallel plan~ of conical shell M. outer and inner radii of annular plate flexural rigidity [D = E('/12(1 . numerical factors moment per unit distance.)I(1 + 1')] thermal moment resultant per unit distance bend ing moments per unit distance on x and)' planes twisting moment per unit distance on x plane radial and tangential moments per unit distance M. numerical factor modulus of elastic foundation. momentsum [M = (Mx + iH.\If xy M" M.81 m/s2) mesh width. axial load factor for slender mem bers in compression stiffness matrix of finite element stiffness matrix of whole structure integers. constant D [D] E F G g " k [k] [K] m.
.' i'x)" j'y! ~ i'. coefficient of thermal expansion. and zx planes II . . N'J radial and tengcntial forces per unit distance N'I' meridional force per unit distance on parallel plane N. (i r" r. I" W v" x. rectangular coordinates angle.xi" LlST Of SY~IH()LS . numerical factor angle. numerical factor shear strain./* N. pressure p* equivalent transverse load per unit area P concentrated force {Ql nodal force matrix of finite element Q. shear force per unit distance on x and . 1'1. temperature thickness distance measured along generator in conical shell displacements in x. tangential. weight distances."2 T s fI. y:. y.I' thermal force resultant per unit distance normal forces per unit distanc~ on x and J' planes shearing force per unit distance on x plane and parallel to y axis l'l~." Q.. axial. T\. to the axial plane of cylindrical shell meridional shear force per unit distance on parallel plane reactive forces radius polar coordinates radii of curvature o[ midsurface in xo and yz planes radii of curvature of midsurface in meridional and parallel planes. l". z (f.::x . and radial displacements in shell midsurface strain energy eJTective shear force per unit distance on ' and y planes radial and tangential effective shear [orces per unit distance work. cylinder geometry parameter [P' = 3(1 _. and z directions.pf! shear force per unit distance on parallel pIa Ile and perpendicular to meridional plane.p N \" N X. shear force per unit distance on plane perpendicular Q" R r r. N Xi' shear force per unit distance on axial plane and parallel to y axis of cylindrical shell Ny normal force per unit distance on parallel plane of conical shell p intensity of distributed transverse load per unit area. IV U ~.' l/a'('].(1.. principal radii of curvature surface forces per unit area.v planes Q" Q" radial and tangential shear forces per unit distancc Q. weight per unit volume or specific weight shear strains in the xy. y.
g <" f". y. z. angular nodal displacement curvature Poisson's ratio potential energy normal stress normal stresses on the x. shear strain in the 1'0 plane deflect ion. finite difference op~rator. variational symbol nodal displacement matrix of finite element norma] strain normal strains in x.SY\IBOLS X\ Ir!! t5 (0. and x directions shear stress on radial plane and parallel to the tangential plane ultimate stress in shear yield stress in shear angle. numerical factor change of curvature in shell .LIST Of. t x . [. and z planes radial and tangential normal stresses meridional normal stress on parallel plane principal stresses compressive stress at critical load ultimate stress yield stress shear stress shear stresses on the x. numerical factor. y. </> I. and z planes and parallel to the y. f. stress function. and z directions radial and tangential normal strains normal strain of the parallel circle and of the merid ian angie. 10" Ii K v n (J ar • ao <. y. t y ..: en l:" 1:".
j j j j j j j j j j j j j j .
and tank bottoms. plates in bending" or . unless otherwise specified. bulkheads.. 1 to 9. For clarity. '* n . . We are concerned with the relationships of external force or moment to strain. turbine disks. It is assumed.'!"il1 plates with small dellections. into equal halves by a plane parallel to the faces. however. the deflections and thicknesses of plates will be shown greatly exaggerated on some diagrams. A homogeneous material displays identical properties throughout. ' 'thil1 plates with large deflections. shells in bending. According to the criterion c~tt~ often applied to define a thil1 plate (for purposes of technical calculations) the ~f' Z' \ ratio of the thickness to the smaller span length should be less than 1/20. respectively. that plate and shell materials are homogeneous and iso. External forces acting on a plate may he con 1i£~. 8. and displacement. a simplification consistent with the magnitude of defommtion commonly found in plate structures. for which the thicknesses are much smaller than the other dimensions.CHAPTER ONE ELEMENTS OF PLATEBENDING THEORY l. When the properties are the same in all directions.~' tropic. Included among the more familiar examples of plates are table tops. the material is called isotropic ." In Chaps. side panels and roofs of buildings. we shall consider only small deflections of thin plates. This plane is called the midplane of the plate._ . street manhole covers.. " Plates may be classified into three groups. Thc plate thickness is measured in a direction normal to the midplane. for which it is common to divide the thickness t. The flexural properties of a plale depend greatly upon its thickness in comparison with its other dimensions. and'?ihick plates.l INTRODUCTION Plates and shells are initially flat and curved structural elements. With T~ the exception of Chap. " . stress. we treat plates. Many practical engineering problems fall into categories .
rederivation of some basic relationships and the governing equation are required. 5 and Sec.iial for this purpose (Sec. due to lateral loading. They arc attributable to fields such as gravity and magnetism. The components of displacement at a point. It may also be distributed arbitrarily over a finik area. and others).1a. are denoted by ".(' /o}'ces and hody forces. where the shape of the plate or loading configuration preciude a theoretical solution or where verification is sought. thin plates is based on the geometry of deformations. D. A number of . Since then a great many cases of platebending problems have been worked out: the fundamental theory (principally by Navier. 9 and 6. The detcmlination of plate shape and the selection of a material that is most efficient for resisting a given system of forces under specified conditions of operation is the design jimelion. in which the xy plane coincides with the midplane and hence the z deflection is zero. and" directions. are also efficient for this purpose. stated mathematically in this chapter. and surface forces. y. The . occurring in the x. respectively). When. BEHAVIOR OF PLATES Consider a loadfree plate. geometry. 1. It is often advantageous. Kirchhoff. 1. The approximate numerical and energy approaches (Chap. homogeneous. respectively. .9. the midsu~face at any point (xa' Ya) has deflection w (Fig. material properties.10). They may be stated as follows: (1) The deflection of the midsurface is small compared with the thickness of the plate.. 1. A surface force is of the concentrated type when its acts at a point. arc used to solve the bending problems of plates in the chapters to follow.e/ected referellces are supplied at the end of the text. These pertain to certain physical laws. and Levy) and numerical approaches (by Galerkin and Wahl. A basic understanding of material behavior and evaluation of the most likely modes oj)i. elastic.1 b). to the inertia of the plate.Utlre under anticipated conditions of service are essel. deformation takes place. Body forces act on volumetric elements of the plate.2 STHESSES 11\ PLATES A\:n SHULS sidered as sUI:fil(.2 GENERAl. respectively. The slope of the deflected surface is therefore very small and the square of the slope is a negligible quantity in comparison with unity. to employ experimental methods. To ascertain the distribution of stress and displacement for a plate subject to a given set of forces requires consideration of a number of basic conditions. and 11'. The literature related to plate and shell analysis is extensive. shown in Fig. 1. or in the case of motion. The first significant treatment of plates occurred in the 1800s. For the cases of nonuniformly heated and orthotropic plates. These conditions. 1. The rational design of plates and shells relies greatly upon their stress and deformation analysis to which this book is directed.fundamental assumptions oftbe smalldefiection theory of bending or socalled classical theory for isotropic. Chaps.
remains unstrained subsequent to bending.. This means that the vertical shear strains . are analogolls to those associated with the simple bel/dinll theory of beams. are negligible. This type of surface can be bent back to a plane without variation in the distances between any two points on the surface .. however.). the governing plate ~ equation can be derived in a concise and straightforward manner. (4) The stress normal to the midplane. (fi""~vs'J. 8.. Only nnder this limitation on the deflections will the squares of slopes be small compared with unity: hence the approximate expression used for the curvatures (Sec. it remains unstrained even for deflections that are equal or larger than its thickness but arc still small as compared with other dimensions of the plate. (J" is small compared with the other stress components and may be neglected.l (2) The midplane. An excep.' duces to one involving only two dimensions. In thick plates. resulting from transverse loading may also be omitted.. 1. the bending of plates is accompanied by:. \ . applies when a plate bends into a developable surface (e. The above assumptions.."'.ELE.g... ~ V"M Jw.. This supposition becomes unreliable in the vicinity of highly concentrated transverse loads. In the vast majority of engineering applications. : able form.'" and )'y. It is deduced therefore that the normal strain t.!. If the midsurface of a freely or simply supported and loaded plate has a develop.3) is sufficiently accurate. known as the Kirchhoff hypotheses. as in short.l!<f!'.. The deflection of the plate is thus associated principally with bending strains. .'. Because of the resulting decrease in complexity.. Consequently. . and assumptions (1) and (2) are inapplicable. Such plates are treated by means of a more general theory owing to the fact that assumptions (3) and (4) are no longer appropriate.2. (3) Planc@i19~initially normal to the midsurface remain plane and normal to that surface after the bending. the shearing stresses are important. faces of cones and cylinders). When the deflections are not small.: ':.\If. .· . tion. deep beams. adequate justification may be found for the simplifications stated with respect to the state of deformation and stress. Sec. a threedimensional plate problem re\1. sur."strain in the midplane.TS OF PLATEHE"iOl:\"G THEORY 3 (a) (b) Figure 1.
= x till Dx I I~' . Note that these expressions are also referred to as the kinematic relations.ty = 2c~ il2 w . gives 11=  ow Z ::ex ' ..4 STRESSFS IN PLATES .30) These formulas provide the strains at any point in the plate.'xz = \fV"" 1"' fh' e =  ')' c1y aw +flu ax cJz "fly· 0\1.. In a like manner.3a) represent the curvatures of the plate..n and }'. The curvature of a plane curve is defined as the rate of change of the slope angle of the curve with respect to distance along the curve.2. z).). integration of the expressions for . y) (a) indicating that the lateral deflection does not vary over the plate thickness.1) yields '.oy . (1. consideration is now given to the geometry of deformation.. the square of a slope may be regarded as negligible. y) and "o(x.:~ . the values of II and v on the midplane.. As a consequence of the assumption (3) of the foregoing section.3 STRAINCURVATURE RELATIONS In order to gain insight into the platebending problem. OW ox ti= ow . see thatEqs. Integrating Eq.la·f) /xy = au Dv 8y + ~~~ '}\z = where "/ij = }'ji(i.:.''1N!) SffELLS 1. the strain¥displacement relations l reduce to (. Because of assumption (1) of Sec. y. ~ eJ. Therefore .... (L2)iireconsistent wi~hasSnmpiiori (3fSubstitution of Eqs. (1.: =0 (l. y) (b) It is clear that 1I0(x. .. and the partial derivatives of Eqs. ~"""').' . we obtain w = w(x.2. A similar illustration appli. 1. + IIO(X..lh at s~ction "'" passing through arbitrary point A(".>.. Based upon assumption (2) of Sec. . treating the geometry of strain rather than the matter of cause and elTeet. (1. we conclude that Uo = Do = O. respectively.ld).2) The above expression for II is represented in Fig. \Ve ~I ( . l.":..2) into the firs'iihree of Eqs..'. (1.s for L' in the zy plane.iJx OJ' (1. y. Thus 11 = Z. (l.. 1. j = x. y) v= 2 ow oy + vo(x. y) represent.
11». The straincurvature relations. by means of Eqs.4 ) r~y = 1d~~) = "x...3) shows that a circle of curvature can be constructed similarly to Mohr's circle of strain. 1. = I oy a (8W) = Ky "by "xy (1..ELEME:.2 the curvatures K (kappa) at the midsurface in planes paral/elto the xz (Fig. yz. (1. Eqs. (K y " . (1. (l."x'y') o Kilt (aJ (b) Figure 1. ".2.lTS OF PLATEBE~DIi'G THEORY :. Figure 1. respectively . The last of these expressions is also referred to as the twist of the midplane with respect to x and)' axes.:. 1.3 . and x)' planes are.3 shows a plate element and a circle of . The local twist of a plate element is shown in Fig.x' where = Clearly.3b) An examination of Eqs. Figure 1.4) may be expressed in the form (1.4) are the rates at which the slopes vary over' the plate. The curvatures therefore transform in the same manner as the strains.3a) and (1..
the normal stresses" x. or su~race is usually identified by the axis normal to it.v(a. z).'y' \ set. 1. E 2(1 + 1') (1. and the directions of the stress components on the same surface. measured in the clockwise direction from the set of axes xy to the .n "yy. j = x.6) The double subscript notation for stress is interpreted as follows: the first subscript indicates the direction of a normal to the plane or face on which the stress component acts. A jace:plalle. elasticity. That is. and G represent the modulus of . (1. .g. If both the outer normal and the . v. Mohr's circle \ shrinks to a point. and the shear modulus of elasticity. (i.i From the circle and Eqs.6 STRESSES l\i PLATES A~J) SHELLS curvature in which 1/ and 1 represent perpendicular directions at a point on the midsurfacc.4) we have (see Prob. respectively. valid for an isotropic homogeneous material ex = B[IT x . This assertion is valid at any location on the midsurface. The curvature and the twist of a surface vary with the angle O. Repeated subscripts will be omitted in this text.5) The sum of the two curvatures in perpendicular directions at a point. (Fig.n lTv' and IT.)] B. Poisson's ratio. 1.. called the average curvature. This means that the curvature is the same in all directions. e. is thus invariant with respect to rotation of the coordinate axes. The prillcipal or maximum and minimum curvatures are indicated by 1(1 and 1\"2' The planes associated with these curvature::s arc calJed the priJlci· pal plalles of curvature. y. I Yx)'=(f " (x:: i x }' = B [<Ty 1 I l'(<Tx + cr. \ there is no twist in any direction. the second snbscript relates to the direction of the stress itself. . the x faces are perpendicular to the x axis. It is seen that when the two principal curvatures are the same.4a). The constants E. The connecting expression is G=~. + cr. stress and strain are related by the generalized Hooke's law.v(<Tx + <Tv)] ~I ry:: G where T. and 0'u are designated by ".4 STRESSES AND STRESS RESULT ANTS In the case of a threedimensional state of stress.1) L (1. 1. The sigll convention for stresses relies upon the relationship between the direction of an outward nomta[ drawn to a particular surface. = B[IT.)] G = tyz (a) e. The surlace is purely spherical at that point.j = Tj.
.'.. .'". = . the above appear in the form Ez " x = . (a) yields the following stressstrain relations for a thin plate ")' = . 1/2 dy dz = dy . = "" Ez "1'' (K.ELD1E::TS (IF I'LATERENDI?\G THEORY 7 /1 y x / +2 "! dx + f/2 1/2 Figure 1.7) Introducing the plate curvatures. (1..' (t.2x)' ". the stress is nega" tive. 1 " ' E + ""J (1. Substitution of t.v· Ez + VKxl = Ez . The stresses distributed over the thickness of the plate produce bending moments.8} Txy = l~KXY = 1+~: c2 w ox 8). and vertical shear forces.. Referring to Fig.. = 0 into Eqs.4 ~.3b) and (1. Eqs.i!°·.4). all stress components shown in Fig. = y. On this basis..L "t:f'+y (a) (/» stress component are in a positive (or negative) direction relative to the coordinate axes.~"'2" " 1_~. the stress is positive.:f 0y .4a are positive. r:2 (2 ZfJ x dz = Mx dy .1  Ez (0 .' We observe from these f()lmulas that the stresses vanish at the midsurlace and vary linearly over the thickness of the plate.2 2 )1' ay2 (l. 1. twisting moments. These moments and forces per unit length are also called sTress resultants. If'the outer normal points in a positive direction while the stress points in a negative direction (or vice versa). we have ( ' Z"x . IAu.'" (K + VK ) = _ 1 E_' zv' (':.
<.v'). Owing to this actiori. the top and bottom surfaces would be deformed into saddleshaped or antic/asric surfaces of curvature K. In fact. 1.5). expressions for the other resultants are derived \M..5 of Sec. and arc included in the derivation of the equilibrium equations (Sec..9a) fvl. A positive moment is one which results in positive stresses in the bottom half of the plate.. It is noted that if a plate element of unit width and parallel to the x axis were free to move sidewise under transverse loading. I IlvI" I I . _ Et' . all moments and shear forces acting on the element in Fig. (1.8) into Eq. a plate manijims greater stijfness than a beam by a factor 1/(1 ..11 ) is the . vertical forces Qx and Q.8 STRESSES 11\ PLATES ." ' .\:"U ~HF:E./G and 1'.flexural rigidity of the plate. and (1. The flexural rigidity would then be Et' /12..10) where D··_ ..2) (1. Substituting Eqs. upon derivation of the equilibrium equations. It is important to mention that while the theory of thin plates omits the elIect of the strain components y" = t. approximately 10 percent. . = t.91> ) The sign convention for shear force is the same as that for shear stress./G on bending. The vertical shear forces Qx and Q. ..12(1 . they are of the same order of magnitude as the surface loading and moments... as in the case of a beam. where lv/x)' = != I .2 (1. arc related to w. Accordingly.v)_·_fix iJy 02W (1. (1..9<1) we derive the following fomlUlas for the bending and twisting moments in terms of the curvatures and the deflection M= y D(1 .vx .5 are positive. The remainder of the plate prevents the anticlastic curvature however.._ \. 1.LS Similarly. 1. \ lvI... arc not negligible.
10) and by employing Eq..9a) and (1. are called tellsors of second order.' .. lAb) and "" are..8) by substituting Eqs.. after integration _ r"  J. or X }' cJy ex OT x : .\' (3 (IJ2) The maximum stresses occur on the bottom and top surfaces (at z = ±t/2) of the plate.. upon substitution of and 'Y' from Eqs. (1.. . through the usc of Hooke's law is not possible. (1. is readily determined by using the third of Eqs.12) that there is a direct correspondence between the moments and stresses. however.\.1) they are not related to strains. orr x " . 1..transfonnation equations for stress and moment are analogous. curvature.. moment (and moment of inertia)."( X} __ (~ . Hence) ... since according to Eq.2(1 _ .' The order refers to the number of subscripts needed to describe a component..2)4 E (1.2).11). which their components transform according to a certain law. It is observed Irom formulas (1. The Mohr's circle analysis and all conclusions drawn for stress therefore apply to the moments (sec Prob. a(Da + 8r. (b). The component (J.... =  Evij (t3 . (1.8). Mohr's circle is thus a graphical representation of a tensor transformation.. the shearing stresses '" (Fig. (1... In t his way we obtain 12M x}': r _ = .4z +:3 ex' + oy2 )(Cex' + il2W)] :')[(82 8 2(1 12 oy2 2 t 2 2 W (1. strain. (1.14) th~ The normal stress (J" thus varies as a cubic parabola over the thickness of : ... Determination of the stress components "" and '. These equations are l 'x.13) and integration 'n ".ELDtF:\TS OF PLATEB!:NDING THEORY '9 The stresses arc found from Egs.' ~+.ox 8y. The differential equations of equilibrium for a plate clement under a general state of stress serves well for this purpose.13) It is observed that the distribution of components Tn and "" over the plate thickness varies according to a parabolic law. Note that quantities such as stress. += 0  az (b) From the first two of the above expressions and Eqs..) d" ... (1.
\. With change of location. 1. a small quantity. 1. Fulfillment of these conditions establishes certain relationships known as the equations of equilibrium. the stress is measured in nelVtollS per square meter (N/m') or socalled pascal (Pa).5 VARIATION OF STRESS WITHIN A PLATE The components of stress (and thus the stress resultants) generally vary from point to point in a loaded plate.5 . Tn the illtel'llatiollal S). M xy + aMx )' dy ay Figure }. Note also that as the element is very small.14) are written in concise lorm in Sec. as for example. for the sake of simplicity the force and moment components may be considered to be distributed uniformly over each face. Consider an element dx dy of the plate subject to a uniformly distributed load per unit area p (Fig. varies in value relative to the positive x face. 1. 1. This variation with position may be expressed by a truncated Taylor's expansion M'+odx ux 8M. say M" acting on the negative x face. This stress. one of the moment components.2 is negligible.5).m or simply Nand N/m. following the development of equilibrium relationships. respectively.6. The moments and forces per unit length have units N·m. according 10 assumption (4) of Sec.'<D SHELLS plate. In the figure they are shown by a single vector.10 STRESSES I~ PLATES . in the load p cannot affect the accuracy of the result. The z directed shear stress components arc also regarded very small when compared with the remaining plane stresses. representing the mean values applied at the center of each face. We assume that inclusion of the plate weight. from upper left corner to the lower right corner. These variations arc governed by the conditions of equilibrium of statics. Equations (1.13) and (1. We shall eventually reduce the latter system of equations to a single relationship expressed in terms of moments.stel1l of Ullits (ST).
The condition that the sum of the: directed forces equal zero leads to . dx ily dv .\ (c) Finally. I ex OJ' + 1" Ix dy = () rrom which +.10) Q = D.1Q. introduction of the expression for Q.\ and y...DIN(.(O'W ) (ly ox' +oy' ".= D..= D(V'w) a \. _Q.15) This is the differential equation of equilibrium for bending of thin plates. Treating all the components similarly. _. from the equilibrium of moments ahout )' axis. + aM..W) = (1. ..I X 8Qx i dY + . (1.. such as the moment of p and the moment due to the change in Q.. Expressions for vertical shear forces Q.'....Q. from Eqs. ax aM..ELEMENTS OF PLATEBE. + . D·(V w) oy where (d) is the Laplace operatol'..v) a.....+1'= 8Qx 8Q.dx dy aM". (b) and (e) together with Eqs. from Eqs.\ is a fUllction of.._" .. may now he written in terms of deflection w.. (b) and (c) into (a) yields ... we have C:Nf x \' ilA1x _··+·_ ..... .. (b) Products of infinitesimal terms..16) i!. Q.·.. Similarly.=O cy ax .• and Q. (X dy .• and Q. the state of stress resultants shown in the figure is obtained. x (iJ w + .. dx dy = 0  } or aM x . THEORY 11 The partial derivative is used because . = 0 ex ity ".\1. ~+2:~+~·_·r= I' '''I 1'M "M ox' ax Dy oy' (1. oy 0 ax (a) The equilibrium of moments about x axis is governed by ._. ax ox' iJy' ox 8 2 2 f} Q. have been omitted..~.
Reduction of unknowns to one.6 THE GOVERNING EQUATION FOR DEFLECTION OF PLATES The basic differential equation ror the deflection of plates may readily be derived on the basis of the resulls obtained in the preceding sections. An alternate form of Eq. M).. Introducing into Eq.17a) This equation."t [PK.. plate problems are internally statically indeterminate." )" =3(!z [1 2t _ (~)2] t (1.\ PLATES .\ND SlIFLLS Since one equation (US) for three unknown moments ly/X' M. is not sullicient to obtain a solution.14) the stress components Ix:.4) . (a) is detem1ined by inserting the definition of curvatures from Egs. (1.. can also be written in concise form \j4W = D E. we have a K. x 2t [1.. To determine w.!y4 aw 4 a IV 4 a"w (1.10).18) Expression (1. (1.= p + 2 . When there is nO lateral load acting on the plate +~+=O ox 4 ~ ox2 iJy2 .(~~)21 t r. and kI. Mxy given by Eqs.. (YZ' and Ci z are as follows ! _= _"t_ ~q./iJl' deflection of thin plates..16) and (1.12 STRESSf<.l7) into Eqs. (1.17b) in which \j4 = \j2\j2 = (\j2)2.19) .15) the first expressions for Mx.S 1. it is required to integrate this equation with the constants of integration dependent upon the appropriate boundary conditions (discussed in the next section).. which roHows.l7) is the govemillg differential equation . 1.13) and (1. is made upon utilization of the momentdisplacement relations. first derived by Lagrange in 1811. (1. (1."tJ' .+a Kr 2 2 2 ox (Jx oy cy2 D (a) The above expresses the plate equilibrium in terms of the curvatures. (1.. Substituting Eqs. +2+=2 ax' iJx oy2 D iJ4 IV a'IV 84 IV oy" p (1..
(L17) for w. as in the case of a beam of rectangular section.20) Thus. (1. . In conclusion.22) are of the same form as the equation describing the deflection of a unifomlly stretched and laterally loaded membrane. (L14).= .b) The plate equation.~'2 + ay2 = t 2w  M oj) (L22a. then Eq. +" = \iJx oy D(l + v)V'w Letting M denote the moment fUllction or so called moment sum M = IvI x + My I +v = D V'w (1.' Hence. x aM Qy =. This reduction was first introduced by Marcus 2 Given the loading and the boundary conditions.10) is invariant. one can solve M from Eq.EUt. !llIlX = 3Q. V'IV = plD. is the solution of Eq. aM oy (b) and we may represent Eq. it is significant that the sum of the bendingmoment components defined by Eqs.. It can be demonstrated that Eqs. (1. In this way. (U2b) leads to IV..22a). lila:>. we obtain ~=~V\v=p Et 3 12(1 .000.21) the expressions for shear forces can be written Qx=a .p c'ivi DX" 82 M iJy' 6 2\\. (UI) yields Eq.. (U7) results from equating the stress normal to the plate to the surface loading per unit area at the upper face of the plate. (l. That is M.. occurs at z = 0. is thus reduced to two secondorder partial differential equations which are sometimes preferred.lFNTS or PLATEBE:'<[)lNG THEORY 13 The maximum shear stress.v') which with Eq. :2 t . + . using the formulas derived. and is represented by tbe formulas r .+ M y = D(I il'w 0211') + v) ( _._. The latter are discussed in Chap.__ Qx 2 t 0 ____ • r y. an analogy exists between the bending of a plate and membrane problems. serving as the basis of a number of experimental and approximate numerical techniques. 3 = .:. depending upon the method of solution to be employed.17) as follows .:."\:. (1. = p. )). We mention that an alternative derivation of Eq. 5. the key to determining the stress components. (1. from Eqs. (1.1 7). letting z = tl2 and fI.
n SHELLS 1. + (aM x. ex oy2 1 (1. + eM xy = D [a w + (2 ~~ 3 3 oy iJx ~ . 1. dy acts. That is v= . It is demonstrated below that these moments may be replaced by equivalent forces. The distrihution of stress in a plate must also be such as to accommodate the conditions of equilibrium with respect to prescribed forces or displacements at the boundary. on the righthand element. or some comhination. 1.\. Thus in an infinitesimal region of the edge shown within the dashed line.. The algebraic sum of these forces may be added to the shearing force Qx to produce an effective transverse force per unit length for an edge parallel to the y axis. while the lefthand element is subjected to [Mx.. These may be a given detlection and slope./i!y) dyj dy. Such a substitution causes an alteration of the distribution of stress and strain only in the immediate region of the boundary. The basic difference between the boundary conditions applied to plates and those of beams is the existence along the plate edges of twisting moments. we see that an upward directed force M x . Vx .. __ v) _8'". Yenan!'s principle l Wc now treat the boundary conditions for a rectangular plate with edges a and b parallel to the x and y axes.6. 1.\ Q X .6.14 STRFSSES IN I'LATES .6 . a twisting moment A1. + 8M ). the moments· are indicated as replaced by statically equivalent force couples. . It is seen that. ux x = D 0'11' 03". In the figure../oy) dy act. and a downward directed force M x .. or force and moment. . (1. one has VY = Q.. Consider two successive elemental lengths dy on edge oX = a (Fig. for an edge parallel to the x axis.). solution of Eg. + (aM".tY 3M XJ' +·d)' OJ' Figure 1..23b) y x M XY M .23a) Similarly. it can be shown that. ~ + 2 ~ v ~[oy' ( ) ox 2 Dy 1 (1.17) requires that two boundary conditions be satisfied at each edge.6). For a plate.7 BOUNDARY CONDfTfONS The differential equation of equilibrium which must be satisfied within the plate is derived in Sec. in accordance with St. as shown in Fig.
Owing to the symmetry of the uniform loading. this force must have the same magnitude and direction at all corners of the plate.26a) The first of these equations implies that along edge x = a. = 0.7b) At the edge considered. Clamped or Builtill Edge (Fig. We can now formulate a variety of commonly encountered situations. 1.7 .2). [secti~:t.6)..25) Simply Supported Edge (Fig. 1. since along these edges no twisting moment exists... The additional corner force for plates having various edge conditions may he determined similarly.7) are as follows. Hence w=O M = D ( '" x GX2 £l2\4' a "".c3) is due to /(I/"C""o/( a distribution of Ai".J results in (x=a. along an edge is statically equivalent to a distribution of vertical shear forces. ow/oy = 0. The boundar). the corners of the plate described tend to rise (Example 3.70) In this case both the deflection and slope must vanish.s (Fig. there may be concentrated forces Fe produced at the ('orners. 1.'igurc 1. At the corner (a.y=h) (124) The negative sign indicates an upward direction. Consider. 1. In addition to the edge forces described above.(:LFMENTS OF PLATL:~AEND(N(j THEOHY 15 Expression (1.?!·(!~. conditions which apply along the edge x = a of the rectangular plate with edges parallel to the x and y axes (Fig.. when two adjacent plate edges are fixed or Fee.}g.4. Thus. as an example.) + v~ = 0 8y2 2 (x = a) (1. the deflection and bending moment are both zero..0.. for instance.lPJ. a (b) (c) J. we have F. the case of a wli/iJl'miy loaded rectangular plate with "i"'J'jL. b) abovediscussed action of twisting moments (because A1". = Nt. if no anchorage is provided.x Plan view (a") r. That is w=O ow ax =0 (x = 0) (1.
of course. The support is not capable of resisting any shear force. 3. such as axisymmelrically loaded circular plates (Sec. nonhomogeneous boundary conditions.28) Some other types of boundary conditions may be treated similarly.16 Sn~ESSl'S TN PLATES A)'\[) SHELLS ill''. It follows that conditions expressed hy [q'. or displacement at the boundary. (1. The latter cases. choosing the acceptable series form is laborious and requires a systematic approach. The inverse method relics upon assumed solutions for w which satisfy the governing equation and the boundary conditions. Usually.2 and 3. a static boundary condition equates the internal forces (and moments) .7c) In this case the edge is free to move vertically. That is p\v a~~:3 (x = a) (1.4). in (1. (/. any other loading can be bandIed by infinite series (Sees.3). possible to have prescribed shear.25) to (1. The most powerful such method is the FOllrier series.7411' = pjD yields plate deflections only with considerable difficulty. rotation. but the rotation is prevented. It is common to attempt a solution by the inverse method. . Some cases may be treated by using polynomial expressions for w in x and y and undetermined coefficients. are expressed by replacing the zeros in Eqs. 1. (1.= all' ox 0 (x = a) (1.26) and (1. it is. .. 1. 4 ) . . in (1.28) with the specified quantity (Sec. in Eqs./?r' = O.26a) may appear in the following equivalent form w=o Free Edye Sucb an edge at x = a IS (x = 0) (1.25) both conditions are kinematic. Accordingly.28) the conditions arc mixed. once a solution has been found for sinusoidal loading. This approach otTers as an important advantage the fact that a single expression may apply to the emire surface of the plate. In addition to the homogeneous'boundary conditions described above. 9 . where. at tbe edges of the plate to the given external forces (and moments). the governing plate equation '1.27) Sliding Edge (Fig. 2.8 METHODS FOR SOLUTION OF PJ"ATE DEFLECTIONS Except for simple types of loadings and shapes.' "" . moment.26b ) free of moment and vertical shear force. It is observed that tbe boundary conditions are of two basic kinds: a geometric or kinematic boundary condition describes end constraint pertaining to deflection or slope. Thus .27) both arc static.
ELEM[NTS OF PLATEHEN DING TIIEOny 17
.~.I.:
Figure 1.8
Energy methods (Sec. 1.9) should be included in a treatment of general approach. These may be employed to develop solutions, otten in the form of
infinite series.
The role of the foregoing methods is twofold. They can provide" exact" answers where configurations of loading and shape are simple, and they can be used as the basis of approximate techniques through numerical analysis for more practical problems. Another approach to overcoming the difficulty involved in the solution of the governing equation is to use finite differences (Chap. 5). In this case. Eq. (1.l7) or Eq. (1.22) is replaced by finite difference expressions which relate the w (and M) at nodes that are removed from one another by finite distances. The resulting equations, however, serve only for numerical treatment. Calculation of plate deflection and stress is illustrated in the solution of the following pro blem.
Example 1.1 Determine the deflection and stress in a very long and narrow rectangular plate or socalled i~{illite strip (i.e., a ~ b). if it is simply supported at edges J' = 0 and y = b (Fig. 1.8). (a) The plate carries a nonuniform loading expressed by
. ~.
p(y) = PoSIll"b
. rry
(a)
where the constant Po represents the load intensity along the line passing through r = h12. parallel to the x axis. (b) The plate is under a uniform load Po. SOLUTION Clearly. the loading described deforms the plate into a cylindrical possessing its generating line parallel to the x axis. We thus have iJwlnx = 0 and 8'w/vx OJ' = O. and Eqs. (1.10) yield
su~face
M Expression (1.17) reduces to
)'
=
d'w D·, dy'
(b)
d4 w
P
d!l=Jj
(e)
This expression is of the same form as the beam equation. Hence. the solution proceeds as in the case of a beam.
18
STRESSES
,I\:
PLATES Ai'<D SHELLS
Note that since the bent plate surface is of developable type and the edges are free to move horizontally, the formulas derived in this example also hold for large deflections (II' ;:>: t but II' < b).
(a) Substituting Eq. (a) into Eq. (e), integrating, and satisfying the boundary conditions (1.26b) at y = 0 and y = b, we have
w=;r (
b)4 Po . ny
D slllh
(d)
.. The maximum stresses in the plate are obtained by substituting the above with v = 1/3 into Eqs. (1.12), (1.19), and (1.20)
CT x, nux = 0.2po
(~
r
CT", m.x = 0.61'0
(~
r( :
=
~, y = ~)
CT"mox = 1'0
(z = ~)
(z = 0, y
t",m" =
0.51'0(~)
= 0)
To gauge the magnitude of the deviation between the stress components, consider the ratios;
CT",!!"'" = 5(.t)2 ax. max b
'", mox = 2.5 ax. max b
(£)
If, for example, b = 20t, the above quotients are only io and ~ respectively. For a'thin plate, t/b < To, and it is clear that stresses CT, and '" are very small compared with the normal stress components in the xy plane. (b) Now Eq, (e) with I' = Po, upon integrating and satisfying II' = 0 and ow/ax = 0 at y = 0 and y = b, yields
II'
= Po b
24D b4
4(y4 _ 2,)'~
b'
+
r) b
(1.29)
This represents the deflection of a uniformly loaded and simply supported plate strip parallel to the y axis. 111e maximum deflection of the plate is found by substituting y = b/2 in Eq, (1.29), yielding 4 W m" = 51'0 b /384D. The largest moment and stress also occur at y = b/2, in the direction of the shorter span b. These are readily calculated by means of Eqs, (1.291 (Ll01 and (Ll2) as Po b2/8 and 31'0 b2/4t 2 respectively. It is observed that for very long and narrow plates the supports along the short sides have little effect on the action in the plate, and hence the plate behaves as would a simple beam of span b.
ELEMEJ".:TS OF Pl.ATE·UENDlNG THEORY
19
j'...t..
·u
~ ... ,
/~"/'tJ I / /
.11"
L! t::::t'==t;z::t=::t;:=:JI
Iy
Figure 1.9
I
A plate which transmits a constant bending moment is said to be in pure bending, We consider this simple and practical case in Example 1.2, As will become evident. in this situation the deformation pattern can be established from symmetry considerations alone,
Example 1.2 A rectangular bulkhead of an elevator shaft is subjected to uniformly distributed bending moments Mx = Mb and My, = M., applied along its edges (Fig, 1.9), Derive the equation governing the surface deflection for two cases: (a) M. l' M h , (1)) M. = Mo'
SOLUTION (a) Substituting Mb = M, and M. = My into Egs. (1.10), we obtain
alw
8~2

D(i~'~2)
Mbl'M a
 =
a2 w
ax cy
0
(e)
Integrating the above leads to
W=
Mb vA1 a 2 A1 a l'M b 2  .......  1"),x  2D(1 _ "~') Y 2D(1
+ c, x + (',y + C3
If the origin of xyz is located at the center and midsurface of the , deformedp/ate, the constants of integration vanish and we have
\1.1
lW b  vtv/a = . _. X2 2D(1  V'),


Ala·· vAl/)
2D(1  1")
y
2
(1.30)
(b) By letting M. = Me in Egs. (e), the result is
K.= K =
x
~
8'w
Y
ex'
=_..
Mb D(1 _ 1')
en
(1.31)
This reveals that there is a saddle point at the center of the plate, Integrating and locating the origin xyz as before, Eg. (I) leads to
IV
= _ _ ..
2D(1 ... 1')
lYf;, ....... (x'
... y')
It is clear that the above expression represents an anticlasCic surface. We note that in the particular case where 1vI. = M" Eq. (1.30) yields a paraboloid "f'revo/ution (see Prob. 1.8).
20
STReSSES IN PLATES A!\iD SHELLS
1.9 STRAIN ENERGY METHODS
As an alternative to the equilibrium methods, the analysis of defornlation and stress in an elastic body can be accomplished by employing ellergj' methods. These two techniques are, respectively, the Ilewconian and /agf"£/Ilgiall approaches to mechanics. The latter is predicted upon the fact that the governing equation of a deformed elastic body is derivable by minimizing the energy associatcd with deformation and loading. Applications of energy methods are effective in situa· tions involving irregular shapes, nonuniform loads, variable cross sections, and anisotropic materials. We shall begin our discussion of energy techniques by treating the case of loaded thin plates. . The strain energy stored in an elastic body, for a general state of stress, is given by' .
U=
2.1.1.1
1 '"
,.
(ax£x
+ u,r, + 0',0, + 'xyYx, + ',,1\, + '"Y,,) dx dy dz (1.32)
Integration extends over the entire body volume. Based upon the assumptions of Sec. 1.2, for thin plates 0'" l'x" 1'Y' can be omitted. Thus, introducing Hooke's law, Eqs. (1.7), the above expression reduces to the following form involving only stresses and elastic constants:
U=
... I 1 .1).1 [2E (0'; + a;  2vu"a,) + 2G '~y ] dx dy dz
(1.33)
For a plate of uniform thickness, Eq. (1.33) may be written in terms of deflection w by use of Eqs. (1.8) and (1.11) as follows
1·' D U= 2 .. ox'
A
II [(02W)' + (O'W)' + 28'w'8'w + 2(1 1 ily' ax' ily'
+ ily'
iJ'w v) (_._. )'] dx dy ilx ily
or, alternately
O'w)'  2(1  v) fa'w il'w ox2 ily'
( il'w ox iJy
)']1I dx dy
(1.34)
where A is the area of the plate surface. The second teInl in Eq. (1.34) is known as the gaussian clirvature. We observe that the strain energy is a nonlinear (quadratic) function of deformation .: or stress. The principle of superposition is therefore not valid for the strain . energy. In the case of a plate experiencing a temperature change, the strain energy that results from heating or cooling must also be included in the above expressions. 3 FOInluias (1.32) to (1.34) are useful i1l'the formulation of various energy techniques and the numerical finite element approaches. Next we review some commonly employed strain.energy methods based upon the potential energy and a variation in defoInlation of an elastic body.
ELEMENTS
nr
PLATEBENDI"KG THEORY
21
The principle of virtual work Suppose that an elastic body undergoes an w'hit/'UrI' incremental displacement or socalled virtual displacemel1t. This displacement need not actually occur and need not be infinitesimal. When the displacement is taken to be infinitesimal, as is often done, it is reasonable to consider the system of forces acting on the body as COIIStal1t. The virtual work done by surface forces T per unit area on the body in the process of bringing the body from the initial state to the equilibrium state is expressed
oW =
I (7~ ,lu + 7;, 0" + ~ ow) dA
(US)
Here A is the boundary surface and ,ll<, iiI!, ,)W are the x, y, z directed virtual displacements. The notation ii denotes the variatiol1 of a quantity. The strain energy iiU acquired by a body of volume V as a result of virtual straining is
tiU =
'V
f (0'" &,. +
a),
De,. + 0', oe, + 'x,. D)\,. + '., ,11'., + '" liy,,) dV
(1,36)
The total work done during the virtual displacement is zero: iiW iiU = 0, The principle of virtual work for an elastic body is thus represented oW=oU
(1.37)
The principle of minimum potential energy Inasmuch as the virtual displacements do not alter the shape of the body and the surface forces are regarded as const"nts, Eq, (1.37) can be written as follows:
oll = ,)(V  W)
=
0
(1.38)
In this expression II=VW
(1.39)
denotes the potential energy of the body, Equation (J.38) represents the condition of stationary potential energy of the system. It can be shown that, for stable equilibrium, the potential energy is a minimum, For all displacements satisfying given boundary conditions and the equilibrium conditions, the potential energy will assume a minimum value. This is referred to as the prillciple of minimum
potential energy.
pIx,
The potential energy stored in a plate under a distributed lateral load y) is
II =
JJJ (O'xl:x + O')'''y + 'x,lx)') dx dy dz v
I" .1,1 (M,K,
A
.1'( (pw) dx dy
A
(1.40)
For the case of constant plate thickness. the above may be written
II =
:1
+ My",. + M,,."'y) dx
" dy  .1.1 (pw) dx dy
A
(1.41)
A physical explanation of the terms of U in this expression is as follows, As 82 wj8x 2 = Kx represents the curvature of the plate in the xz plane, the angle
22
STRESSES I~ PLATES A:':D SHH.LS
corresponding to the moment M x dy equals  (:2W/CX 2) !lx. The strain energy or work done by the moments M x dy is thus 1M x k., dx dy. The strain energy owing to lvt,. dx and Mxv dy are interpreted similarly. The principle of potential energy, referring to Eg. (1.41), is expressed in the lorm: <In
= .... (( (M,
A
,IKx
+ M,. elK,. + M,y
OK,y) dx dy 
(.I' (p 011') tis dy =
A
0
(1.42)
The Ritz method The socalled Ritz method is a convenient procedure for determining solutions by the principle of minimum potential energy. The essense of this approach is described for the case of elastic bending of plates as follows. First choose a solution for the deflection IV in the form of a series containing undetermined parameters am" (n!, " = 1, 2, ... ). The deflection so selected must satisfy the geometric boundary conditions. The static boundary conditions need not be fulfilled. Clearly, a proper choice of the deflection expression is important to ensure good accuracy for the final solution. Thus, it is desirable to assume an expression for IV which is nearly identical with the true bent surface of the plate. Next, employing the selected Solulion, determine the potential energy n in terms of <1",,,. (This demonstrates that the am,,'s govern the variation of the potential energy.) In order that the potential energy be a minimum at equilibrium:
,=0, ""
an
call
(1.43)
The foregoing represents a system of algebraic equations which are solved to yield the parameters am,,' Introducing these values into the assumed expression for deflection, one obtains the solution for a given problem. In general, am" includes only a finite number of parameters, and the final results are therefore only approximate. Of course, if the assumed w should happen to be the" exact" one, the solution will then be "exact." Advantages of the Ritz approach lie in the relative ease with which mixed edge conditions can be handled. This method is among the simplest for solving plate and shell deflections by means of a hand calculation. The applications of the strainenergy techniques in the treatment of bending, stretching, as well as buckling problems of plates and shells, will be discussed throughout the text.
1.10 MECHANICAL PROPERTIES AND BEHAVIOR OF MATERIALS
The ordinary mechanical properties of materials, such as the static yield, ultimate strength, and modulus of elasticity, are adequate for use in most conventional analyses. Table 1.1 furnishes average properties for some common materials. Exact values may vary with changes in composition, cold working, and heat treatment. From experiments it is known that the Poisson's ratio)' (not
9 18 10.2 82.1 Average mechanical properties of materials Modulus of elasticity (GPa) Tension Shear Ultimate stress Yield stress (MPa) Tension [Comp.6 72.7 11.6 41 45 290 103 138 138 172 290 441 276 345 [20.7J 345 138 [552] 262 186 276 193 241 23.6 70 73 103 103 21.6 23.6 26.5 79 79 245 138 248 345 165 172 345 207 131 331 Steel Mild High strength Wood 77 77 2.4 117 103 45 200 200 714 25.2 Medium strength Copper Cast iron Magnesium alloy 41 41 16.Table 1.24 410550 483 [2870] 11.8 80.7 5.7 10.8 16.2 18.5 87 22.758.J Material Aluminum alloy Specific weight (kN/m') (MPa) Tension Coefficient of thermal Shear Shear expansion (10' per 0c) 6061·T6 2024T4 Brass Bronze Concrete 26.3 17.9 241 27.8 25.4 Structural N W .
a. rekrs to any action leading to an inability of a structural member to function satisfatorily..p).'P = ~. the material can experience only a small amount of yielding (less than 5 percent). whether brittle or ductile. Failure. Thus (1. however. prior to fracture. y1 Maximmn principal stress theory This theory predicts that a material fails by yi.:p = (iY/I and c. Some important variables associated with failure are the type of material. and an appropriate theory of failure consistent with the type of material.p = a. embodies an unusual integration of materials. generally. Yielding of plate (shell) elements. while brittle materials are weak in tensile strength. and the operational environment.. the shape and surface peculiarities. converting uuiaxial to combined stress data. respectively. by same yield stress in tension and compression..24 STRFSSES 1:\ PLATES A'D SHELLS listed in the lable) varies slightly for different metals over a relatively narrow range. is given in the following paragraphs 1 The discussions are limited to the case of plane stress. Ductile materials are weak in shearing strength. That is. When a metal undergoes an appreciable amount of yielding or pemlanent deformation. respectively.and shelllike structural elements. the material is considered as brittle. at the beginning of yielding or inelastic action ". It is noted that the analysis and design of plate or shell elements. Unless we are content to overdesign plate. must also be considered. arc the basis of elastic design criteria for ductile and brittle materials. as components of a missile or space vehicle.lding when the maximum (or minimum) principal stress exceeds the tensile (or compressive) yield strength. for instance. We shall refer to the ord inary characteristics of materials and takc \. it is necessary to predict the most probable modes of failure.5A proper design includes a prediction of the circumstances under which failure is likely to take place. i. The yield stress (or strain) determined in a simple tensile test is denoted by "~p (or "~p) and in a (or e. The formulas alone cannot predict at what loading failure will take place.45a) It is observed that if at and" 2 are of opposite Siy". When. In the case of materials possessing the simple compression test.e . We observe from the formulas derived that the breakdown of elastic action is likely to originate at certain locations of these members. The material strength. or initiation of fi'acture. configuration and rate of loading.44) Maximum shear stress theory This theory states that yielding will begin when the maximum shear stress in the material is equal to the maximum shear stress at yielding in a simple tension test. = 0.:p or (1.p = B ). 4. it is termed dl/ctile. having characteristics dependent upon environmental conditions. A brief discussion of some yielding theories ofJail lire. one tensile and the other . The algebraicaIly largest and the smallest principal stresses are designated "I and a2.3 for metals.
However. 1. The prediction of failure by fracture under combined stress may also be accomplished through the use of the maximum principal stress and maximum shear stress theories. (t. the maximum shear stress theory is in widespread use in some design codes because it is simple to apply and offers a conservative result.1 to 1. The various failure theories are applied in a number of chapters to follow. (1.44) and (1.46) Maximum distortion energy theory This theory states that failure by yielding occurs when." .ELE\fE\!TS OF Pl.(J"2)/2. Thus. (1. The critical values of the ultimate yield point stresses are usually determined in a tensile experiment. the distortion energy per unit volume in a state of combined stress will be equal to that associated with yielding in a simple tension test.).:~TEHENDl~G THEORY 25 cOlnpressive. where the failure of a specimen occurs by fracture.2  V1T2! \'(T11 = a~p = (J. provided the yield strengths (O'..2 A sheet of metal is transversely loaded so that tht' moment components at a critical location related to the set of axes x. 1..:p) are replaced by the ultimate strengths ((J.. one thcn obtains the yield conditions: (J ASb) Maximum principal strain theory This theory predicts that a material fails by yielding when the maximum (or minimum) principal strain exceeds the tensile (or compressive) yield strain. that is. we have la. ! 0. (1. the condition of the inelastic action is given by Eq.~) in Eqs. a. respectively. the maximum shear stress is (lTl .. and its employment in design practice is increasing. if (Jt and (J2 carry the same si(/11 and ()'3 is laken as zero.5) by employing Mohr's circle. upon application of Hooke's law.. Good agreement between the maximum principal stress theory and the experiment has been realized for brittle materials.p.47) It is noted that the maximum energy of distortion theory agrees best with experimental data for ductile materials.45). Accordingly. InterestinglY.1 Verify the result given by Eq. Show that the moment components M. PROBLEMS Sec!. This theory is generally accepted in design practice for brittle materials.45a). at any location in the body." M"" and M'~'J' associated with a new !'oct of axes x'y' indined a1 an angle of e" clockwise to the Xl' set .v are AI~" Mp and M.~p (1.5 1. The maximum principal strain theory is not supported by experimental results and is unsafe for ductile materials.
(b) Show that Eqs.~ + My) .5 The lateral deflection of a rectangular plate (Fig.>:.>:_v sin 20 M).3 ) 1.3 By means of.M~rsin 20 M.7 Determine whether the fonowing expression satisfies the boundary conditions of a simply sup· poned very long and narrow plate (Fig. Using v = 1.8 Consider the bending of the rectangular plate (Fig.. Explain why results are inconsistent. Sees.')hape. 1..} cos 20 + M. = !(M.26 STRESSES IN PLATES AN"D SHELLS (Fig. 1.My) sin 20 + M.Mohr's circ1e show that the gaussian curvature is equal to (P1. simply supported plate. 1.'O S 2 0 "'iiin 0 cos 0 sin 0 cos () 2sinOcosO 2sinOcosO ~:My ~ cos 2 (I . with built~in edges of lengths a and band SUbjected to a uniform load Po..30) and (1.3a) is subjected to uniformly distributed twisting moment M "y = M 0 along aU four edges. 1.2D(1 + v) Mo(x' + y') K=K= • ..3 +j. is given by X4 W==Co ( a 4 2~ x' + a2 b4 2. (121). 1M I (PUa) = 'HA1.). I == [ ~ M~_.2) by a Mohr's circle. = nnlb and x ~ O. (\. bI2): w Pb 2 = 3 21t Dn"'l 11 "" L 1 mt "3 sin ~ sin lXy{l 2 + ax)e'.8) It is seen that the first of the above expressions represents a paraboloid of revolution while the second implies that the surface is spherical.4 Demonstrate that at a corner of a polygonal. = !(M". (b) the maximum stress (Ix for 11 = 1...5....). where Co is a constant. (P1. y' y' ) "~_.. cos 20 (PUh) Represent Eqs.10 .c + My) + t(M... Detennine an expression for the deflection surface w.. in terms of M defined by Eq.]!!. in this case. alteruatively 1M .7) Here (J. for a = b.M)" 1.c . . (P\.sin 2 0 . """ 0 unless the corner is 90°.=L. 1. M".2 x' )(y. 1. (b) the maximum plane stress components (J.~). respectively: w= .. 1.6'0 1.c.=Mo: (u) Verify that. IM"i or.My) cos 20 .3) are expressed <. 1.8) carrying a concentrated load P at a point (0.j JIM. 1. 1.=M.c and tXl' at the center. the bending moments M 0 are uniform along the boundary and the twisting moments vanish.9) for the particular case in which M..u .. Then ascertain that the orientation of x'y' corresponding to the maximum and minimum or principal momellts is defined by tan 20 = _.9 Determine an expression of the strain energy for the plate described in Prob.6). .M).. obtain the corresponding expressions for: (a) the moment M. cos 2 0 sin 2 8 sin~ 0 {.t(M.. Detennine: (a) whether this def1~tion satisfies the boundary conditions of the plate.6 A square spacecraft panel (Fig. M. D(I M. + v) (P1.< 1. M. t.10) yield. even for a plate of any . .
2. 2. are usually circular in shape.10 and 2.2 BASIC RELATIONS IN POLAR COORDINATES In general.2 and 2. We shall discuss some of these in Sees. such as end plates and closures of pressure vessels. Sections 2. are employed. clutches. y) are related by the equations (Fig.4 to 2. The polar coordinate set (1'.la) X=l'cosf} y = l' sin f} x 27 . pump diaphragms. and turbine disks. developed in Sees. I n all cases the basic relationships in polar coordinates. polar coordinates are preferred over cartesian coordinates (used exclusively thus far) where a degree of axial symmetry exists either in geometry or loading.. Examples include a circular plate and a large thin plate containing holes. A number of circularplate problems have stress distributions that are symmetrical about the center.3.CHAPTER TWO CIRCULAR PLATES 2.1 INTRODUCTION In practice.9 and 2. Thus. 2. 2. 8) and the cartesian set (x. members that carry transverse loads. etc.12.11 deal with the situations involving asymmetrical loading. many of the significant applications fall within the scope of the formulas derived for circular plates.
10'.. ==cosO I' or ox ax x oy sin I' ._.._.. the chain rule together with the above relationships lead to . law r ao sin Ii (a) To evaluate the expression a'lVjOX'...r.......ATES A~J) SHELLS (a) (b) Figure 2....2 (d) ...2 .1 Referring to the above. oy' 0'" 2 Ii +2 ow cos 2 0 DO or I' or I' 2 OW sin 0 cos Ii a w cos 2 1:1 . (a)..28 STRESSES 1:\ Pl.=  a'lt' ax rJy a'w.80 1" 80' (b) 1" Similarly.2 80 1" (c) .J ("II') Dr ax r ill] AX = iJ'w cos' 1:1 _ 2_~'w ~in e cos e iJ\V sin' 0 +or 81" ao iJr " cos I' all' Sill 0 cos () (l'w sin' e + 2 .= . = tlr 0•• V I' = SIO (I ... this time applied to (!1I'/iJx rather than to IV: il2\~ = ax' O!!.t)2W sin e cos 0  +iJO'.+. 88 1:1 .. ..... = cos (I r Inasmuch as the deflection is a function of I' and 0..+ '..Sill a2 w a2 w. c0 2 cos 0 .... 0 cos (J + '' 8'11' S111 f!r 1 cos 20 l' ' v r ve Dwcos 20 ... = 00 X ~ ey.J _. ''. one uses Eq... = . (Ow.:2 oll'sin 0 cos (j  8'w sin (I }. sin 0 . '.r . ow ox or all' cos (j ...
Eq.: (3 (2..e" (I = O.'IV)] tar r itO r or ae r2 ae (2.' and the vertical shearing forces Q" Qo then have the same values as the moments M" M).~ T rO  . the laplacian opera (2. at the same point in the plate. (2.~.4a) Expression (1...:.2W..5. (h).o t3 (J Q = 12M"z t3 .23b) appears as 1. formulas for the plane stress components. ..v )(D . 1 be transformed from cartesian to polar coordinates_ Consider now the state of moment and shear force on an infinitesimal element of thickness t. and substituting the resulting expressions into Eqs.2) Q. described in polar coordinates (Fig. 111" M. tangential. (1.. = D····· (V'\V) (l}' ("i QQ= D. (1. (1..o = D [1.12).~: + ~_ ~~w + \.3) where M" Mo and M. (I» and (c) into tor becomes: ['1. are written in the following form a I' = . letting (I = 0 in Eqs. to simplify the derivatiolls.= D Mo= D M = i~'~' lor + v(!"'" r at r 0" r2 80 +1. from' Eqs. ae .10) and (1. 1.1 ) Determination of the fundamental equations of a laterally loaded plate in polar coordinates requires only that the appropriate relationships of Chap. M xy and the shears Q.x axis is taken in the direction of radius t. 2.~ 1 a ." Q.80 a'IV 1alV)] (10 (2Ab) describing the effective transverse force on the edge at Ii = (constant) and any r. Note that. Clearly the maximum stresses take place on the surfaces (at z = ±t/2) of the plate.1b).23a). = M o. Mo.: a (1 a. we have M. r/J 1 (12W . and (d). the. (c). 12M.J 2 ilr2 . an edge at I' = a and any 0 becomes v= Q + ~aM. for.2). twisting moments. oM.16). r I' r = (If) _Di~(V2\V)+1=V_~_(!Jl2w _LV.= Qo +.~' . i.CIRn :LAI~ PLA TF.. Thus.:Jij(V Il') + (I 8 2 v) 3r .( I .':'"')J 00 1'2 2 [~~_l. 12M. The radial..S 29 U pOll substitut ion of Eqs. The effective transverse force per unit length...i.:a(j(Vw) Similarly. are defined by Eqs. or ao (2. (d) of Sec.
1.1.~ or + . ..=O vDw r ow ·~O Dr a. (/» through (d) into Eg. They are listed in Table 2. = 0 . = D f "IV f ( 1V) l2 P (2. The boundary conditions at the edges of an annular circular plate of outer radius a and inner radius b may readily be written by referring to Egs. . (2.8) Ii = I *= " Arr + Bi r3 + qr"1 + Drr In r A*r" + B*r.' +D.· iJ.2 or ~'~ = 0 v....= 0 (2.6) and noting the validity of the resulting expression for all r and II.:= ro and any 0 \1'=0 iJw ~~O w=D (llw .28).9) Jo = AD + Bo r3 + Co In r + Do r' In r fj=A.T + .r. Table 2.: +? 'fiii' . D~ are constants. Upon introducing these expressions for f..6) and wp the particular solution of Eq. the complete solution is expressed (e) We assume the homogeneous or complementary solution to be expressed by the following series· W II = n~O I III cos lIe + L f~ sin II'" ro ro 118 (2. (1.1 Boundary conditions for circular plates Edge Clamped Simply supported Free Sliding At.n + 2 n n n 11 Here A... determined by satisfying the boundary conditions of the plate. and J~ are functions of r only. (1.3. It is noted that the inner (or outer) radius is represented by "0' Clearly. I 8'11'. one obtains the solution of Eq. and (2....17). 2.~i rill' ill" ( i)' 2 2 1V + 77). 3.rlnr (2.) .4).+~~() M.2). and I~ into Eq. (2. r 3 +C. (2. (2.+ ~2 802 1 a 1 8' )(3''''" 1all'. the governing differential equation for plate deflection in polar coordinates is derived: 4 V IV 1a 18 )(8 ~7)..5) Letting w" denote the solution of the homogeneous equation iJ2 ( f.· (2.: + ~:} >]0 2 8~:r + .7) I where /.:I +. leads to two ordinary differential equations with the following solutions (Prab.7). nomenclature employed in the table parallels that defined in connection with rectangular plates in Sec. Substituting Eq..30 STRESSES IN PLATES AND SHELLS Upon introduction orEqs.5)..r+B .11 + C* I'If + 2 + D*r. (2..7.6) in a general form.7) into Eq.25) to (1.
only M" M.lOa) The formulas for stress are readily obtained by substituting Eqs.2 into Eqs.2 dr 2 + v rdW) dr ao= _ ~(~ dw + vd'"'. (b). The moments and shear force.  1 va o) va. act on the circular plate element shown in Fig. 2.9ae) d w dW) d d dW)J Q = D.) 1 (2. and (d) of Sec. (1. (2.2) as follows M = r 'D(~'" dr 2 r £II' + rdr .. For this case.5)] now reduces to 1dr [1 V'w = (. = Co = 1'r(} = E (a.~ dr = 1 d dw) r dr ( r dr . (1. in an axisymmetrically loaded circular plate. (e).3 THE AXISYMMETRICAL BENDING The deflection II' of a plate will depend upon radial position I' only when the applied load and end restraints are independent of the angle O.8) and setting e equal to zero: rJ r =  1_ Ez (d'w .) (a) E(a o G T. dr' I'dI' dr' .7) with the result that e. are found from Eqs. and Q."d>l') dl" I' Mo= _D(~dll' + V~2W) 2 (2.. (2. dr D (2.+.CIRCULAR PLATES 31 2..( r .. dr dr' dr I'dI' dr The differential equation of the surface detlection [Eq. The situation described is the axisymmetl'ical bending of the plate.D. it is necessary to replace subscripts x by I' and y by 0 in Eqs.. . 1 Introducing the identity V'w d'w = 1 dw dri +. 2.1b.d~ + ~t£)(d2W + ~ dW) =!!..(d .11) v' rdr dr' To write Hooke's law in polar coordinates..
2. 2.4 UNIFORMLY LOADED CIRCULAR PLATES Consider the case of a circular plate of radius a under a uniformly distributed load Po. (2....m loading p = 1'0' the general solution of Eq. I d. except zero. The constants of integration (the c's) in this equation are determined for various particular cases described below.13) yield an infinite displacement 'at r = 0 for all values of c. Plate with clamped edge (Fig. dr ·1· ·1 'rp w = . In . the boundary conditions are w=o £I".. I..8). It is seen from a comparison of Eq.10) II' = II'h + II'p = c. The lateral displacement w is expressed by Eq.. (2. (2.. (2. (2.. tI... SatisfYing Eqs." In . r' + (4 + I' . therefore (.. + c. dr d. . (a).. 2.1 Oa) appears in the form 1£Llr3/1d(r1":)JII= D !' r£I.13) where the c's are constants of integration. .32 STRESSES IN PLATES AND SHELLS Eq. (2.13).r .. we obtain il!!HIHH!lor f..=0 dr (r = a) (a) The terms involving logarithms in Eq." 6~D (2.2 .2a) For this case.12) If the plate is under a unifo..a z (a) Po +a+l z (b) Figul"c 2. (2. = c.13) and the first of Eqs.10b) The deflection w is obtained by successive integrations when p(r) is given: 1· I·· I····· (2. that the homogeneous solutionj~ represents the case of axis ymmetricaI bending of circular plates.10) is (Prob. + c. = O. . D d. and c.. d.
16) . Plate with simply supported edge (Fig. At the edge (r = a). =  g'' poa 2 Mo= 8 vpoa 2 while at r = 0. Eqs.15) Mo = t~ [(I + v)a'  + 31')1"] The stresses from Eqs. 2.. (2.11) are a.. (2. (2. R PLATES 33 The dcfiection is then w = _ E~" (a' _ /. M. = {~ [(1 + v)a'  (3 (I + 1')/"] (2. The boundary conditions w=O 4 M.(3 + 1')1"] (2. and ".14) and (2. 2.14) The maximum displacement occurs at the center of the plate: Po (/4 W max = 641) (b) Expressions for the bending moments are calculated by means of Eqs. displacement must be finite at r = O. + C4 + 64D Poll = 0 C3 = .CIRCUL.Z 1 [(I + v)a' .' 3poz Algebraically extreme values of the moments are found at the center and at the edge. in Eq. The values of C. The curves are parabolas expressed by Eqs.14) and (2.9): M.3 is shown the variation of stress with the ratio r/a in uniformly loaded clamped circular plates (upper base line).. = M. Thus. (2.(I + 3v). (2. It is observed that the maximum moment occurs at the edge. wehave (~ :"~ (e) In Fig. 32D 1 + v +v .13) are therefore zero..')' 64D (2. = (1 + v)Po)a'/16. '] a" = 4(3 [(I + v)a .16) with v = 1/3.2h) As in the previous case.=O (r = a) poa' 3 yield t he following respective expressions cJa .. = 3: ".15) lead to M.
4 0. recommending its use in practice. a designer sometimes simplifies the model of the original clamped plate. 2t2 Base for /T t _:l. the distribution of moments can readily be obtained in the form M.2 0. Actually.18) M.:. To attain more accurate results. Based upon these considerations. (b) and (d). = ~~[(3 + v)a' . comparing Eqs. however. using instead a hinged plate. the effect of a definite amount of edge yielding or relaxation of the fixing moment can often be accommodated by the fonnulas for edge slope and the method of superposition. much simpler than that of a plate with restrained edges.. When the load resistance of a plate is limited by the large deflections. and a condition of true edge fixity is especially difficult to obtain.{! + 31')1'2] . a partially restrained plate exhibits deforinations nearly identical with those of a hinged plate.8 1.6 0. which occurs at r = 0. many support members tolerate some degree of flexibility.p_Cd_e_d. we see that the maximum deflection for a simply supported plate is about four times as great as that for the plate with a clamped edge.:g_e"~""'''I G. in geileral. The plate deflection is then w=l'oa~(/'4 _23+v~ +~~) 64D a 4 1 + va' 1+v The maximum deflection. As a result. = ~~ (3 + v)(a2  1") (2. 4 l / 4t 2 Base for supported edge o Figure 2. In addition.3 0.17) = Pou 4 5 + V 6415 1 + v (d) If v = 1/3.0 L ria poa' t2 from which C4 = poa 4 (S + 1')/64(1 + v)D. is thus Wrnax (2. the analysis of a simply supported plate is.34 STRESSES IN PLATES AND SffEllS fJ Oa l poa 2 t~C_la_lll. the results relatedta the simply supported case will be cOllsewative. 7 Given the deflection curve w.
Thus. 2.3 that the ratio of the maximum stresses lor the simply supported and the clamped plates is equal to 5f3.oz (3 4c ' + v)(a 2  ') 1' (2. and radius a.3. max = (fO. the corresponding ratio for a simply supported beam and a fixed beam is 1.19) = }POz[( ~4t' 3 + v)a . 2.3 as functions of radial location (for \' = 1/3). (To = 3/. while the deflection and stress will increase at the center.' thickness t.5 ~ (t)' (fyp (f) as the value of pressure differential governing the onset of the yielding action. (1. The plate is made of an isotropic material of tensile yield strength (Jr.S 35 The stresses are then (T. for uniformlY loaded ordinary plate structures of clamped edge.max ~8··· 3(3 + \')1'0 (a)' i (e) The stresses. are given by (12 = (ffJ.~a')(_poa')+ (_poa~)= (J2 4 t 4t 4t' 2 2 2 lP The above yields 1'0= 1. . 'J (l + 3v)r The maximum stress takes place at the center ofthe plate (I' = 0) and is given by err.47). = . in a simply supported circular plate are displayed (lower base line) in Fig.1nterestingly. ~poa2)' ( 4 t _ (_~1'. the maximum stress will be somewhat higher than obtained above.1 A circular clamped edge window of a submarine is subjected to uniform pressure dilTerential Po between the cabin and the outside.19).C1R(TL\R PLATF. Use the maximum energy of distortion theory to predict the loadcarrying capacity of the plate.ma~= ~ pou 4t 2 2 According to Eq. SOLUTION The principal stress components occur at the builtin edge and. We observe from Fig. It is noted however that owing to a small degree of yielding or loosening at a nominally clamped edge. Eqs. Example 2. (2.5. 2. referring to Fig. the stresses will be considerably lessened there. This is also valid for the plates with clamped edges of any other shape.
are neglected. .(a 2 . Q. W = _.2po/2m' = I'Po/2.v) (I .J . " 2~ 3(1 )2 In the case of a plate of radius 10 times its thickness. = ..17) and (b)]: I\' . Gt Consider.5 EFFECT OF SHEAH ON THE PLATE DEFLECTION Tn Sec.. 1. we have.20)./dl' referring to Eqs.2 and Example 1. We now demonstrate that the solutions for deflection of thin plates based upon the bending strains only. The deflection owino to the shear is II' = .v)D Po t ... f.. 3 Q. 2.. = p~~ [!:.. a measure of plate slenderness..20) The maximum deflection occurs at the center of the plate.r2) _ 16(1 . (1. 1 . The ratio of the shear deflection to the bending deflection at .~ 64D a 4 . the influences of the shear stress r" and the normal stress a. 2. Denoting the deflection of the midsurface due to the shear alone by 1\'. ). For thin plates (t/I' < OJ).2 1 + vva!:~. the shear strain. is obtained by addition of the deflections associated with bending and shear [Eqs. = 0 provide. This ratio is.1 we have ohserved that in the bending of plates.. into Eq. the above is small.2b). + ~"v + £12(14t'. we thus conclude that the bending theory yields a result of sufficient accuracy. 8Gt 31'0 (a 2 . PLATES AND SHELLS 2. upon integration.' '/l Qr dr (a) Here the vertical shear force per unit circumferential length.a' I ~) 1+ 3+ 2 (2. is expressed: a.~ I 2Gt 3 . = d"'. Introducing this value of Q. yield results of acceptable accuracy. for 1'= 1/3: p o t'a /16{1 . (a). a simply supported circular plate under uniform load Po (Fig.= dw.. . (2. for example.:. at a distance l' from the center.1')1) poa 4 (5 + v)/64{1 + 1')1) = .36 STRESSES It.2) 2 (b) The total d<11ectiol! 11'.
AI\ PLATES 37 2.P/2nr. in order that the deflection be finite at r = O. . 2. and below.22) The maximum deflection occurs at r = 0 and is given by W max 16.. C4 arc calculated for two particular cases.jj Pa' (e) t=a .4a) The boundary condition.6 CIIKULAR PLATES UNDEn A CONCENTnATED LOAD AT ITS CENTEn When a concentrated or a (nominal) point load P acts at the plate.9c) and (2. (2. 2. when introd uced into Eq. (a) and (b).CIRCU. Thus. r Figure 2. provides the following expression for the deflection w =i~jj (2r2 In ~ + a' .21). from Eqs. The value of '\ must be taken as zero. as ill Sec.4 .[ (a) (b) 1>.16"D The foregoing.. must equal . described Plate with clamped edge (Fig. =  P ···(2 In a + 1) 16"D Pa' c 4 .21) lead to IV = 0 and c2 a 2 In a + c 3 a 2 + C4 = 0 c.2 +. upon substitution into Eq. P 8nD c.21 ) The constants c" e 3 .C4 (2.rl) = (2. (2. iJwji'r = 0 at .a(2In a+ 1) + 2c.4. the constants of integration are C2 = _. a~ . we obtain 41> ~C2 =  I' P 2nr (b) Solving Eqs. The deflection surface of the plate is then represented W = c2 r2 In }' + ('3. (2. one can set Po = 0 in Eq.13). = a. = 0 (a) The additional condition is that the vertical shear force Q.21). The term involving '" must now be retained because of the very high shear forces present in the vicinity of the center. (2.
are 31': er.~ Pa 2 3 + v . Several other cases of practical importance can also be treated on the basis of the mathematical analyses described in the foregoing sections' For reference purposes Table 2. That is. .23) that (J.] (2.9) into the above gives three linear equations for C2' "" and "4' Solving for the constarits leads to the following relationships for deflection and stress: w = ~ [2.=·'·~ 1[[3 [. Analysis by a more elaborate method indicates.. r. Plate with simply supported edge (Fig. and era grow without limit as .(I + .2 lists the equations of deflection surfaces for three variously loaded circular plates (Probs. = r in Eqs.26) At r = 0.38 STnESSES l'i'" PLATES A:. 3 ttl r (J Q 3Pz = ". In I' + a ~.)".4b) In this case.71..2 16rrD (J.23). (2. can be obtained by replacing the original r. however.(I + . (2. the maximum finite stresses produced by the concentrated load can be calculated. If r. < 1. 2. 2..if) SHELLS The stresses corresponding to Eq... the deflection and the radial moment of the plate vanishes at the edge and the applied load I' = ~ 2rrI'Q. derived from Eq. ::: 1.5 to 2. is substituted for r in Eqs....·1 a r 1 (223) 31'" [ a "9 = i (I + v) Innt l' ~ v1 It is seen from Eqs.= 0 (M. (2. we have the maximum deflection: lot' mo< =~.71 the actual r..26). = 0 Q.11 J. may be used.24) The above applies to a plate of any shape for the cases in which 1'... Upon introducing the equivalent radius r. that the actual stress caused by I' on a very small area of radius r.)In .7). by a socalled equivalent radius re' The latter is given by the approximate formula: 7 (2.. (22l) and (2.~v (a 2 ~ 1'2)J 1 + \' (225) 31': a = .[(I 3 TCt +" ) In a r + 1 ~ . {wL.. (2. (2.22)..= I' 2rrr (d) Substitution of Eqs.) In .. O.. 16rrDI+v (e) To find the largest finite stress.
.· .. a~ 1_ I _ 3D mnlia . II ...(a' b) 1+.. .)' _ 2a _~20 0_~ I~4 0 In a_Ia l ~(~)' (~)'ln~J + (~)' 8D188a 0 4 2 "'roo> (.. a' 64D p a' I _I + ~~. · · · .2 Variously loaded circular plates Loading A. . Uniform load Po on inner portion II' w ~ ~Q..va' w(b)=.+2b 2 4D . a 0 2 4 a a 4 .''''''') . oS b) inn ~ I HUt b \ .... 8D I I 8 ~ (~)' + (~)' + ~ (~)' In ~ + ~ [I + (")' In ~ .0 I' oS b) 8 8 .. IS p. +")In..+ ~(~)' _I(b)' _ '(~)' In b) 80 8 8 . Uniform load Po on outer porrion /'0 Equation of elastic surface b (b)"]r '" ~ p. ~ II'(O)~ poa'l'.a 2 a 2 a a 1. .a ('b)' + 2 (b )' InI .b [..a' 1[1 + ~(~....'.(~)' JI 8 a a 2 LI. 2 t + v a P. +(b. "0 ~ rrut~ = w{O) = p 0__ ..3 (" )' + .0 2 . 1 B.. 1 h .bl' .Table 2.". <: b) I~·· a·~I b 1_ C:.'.( 211.. ( b') b) In·a ~ . Line load P! per unit length along 21rb w~ PI~ I 'P._ b a Q C. (I _ ~)' 64D 0 2 + p. ' (0 4D I+v a a~ (.I .) 1+.
2. Two typical cases of loading are illustrated below..9a). we discuss the bending of circular plates with a concentric circular hole.M. (2.5 .7 ANNOLAR PLATES WITH SIMPLY SOPPORTED OUTER EDGES In this seclion. The constants are evaluated by substitution of Eq. b' (I . Plate loaded by edge moments (Fig.!. The inner and outer radii of a plate are denoted by h and a.a' M.5a the boundary conditions are represented by w=O M. . In tI'  a'b' M. (e)...40 STRESSES IX PLATES AND SHELLS 2. at inner edge (Fig. (2....ge).b' (I + I')D M = _l>1.27) and (2..5a) To arrive a t a solulion for the del1cction we utilize Eg.. (2. per unit circumferential length.. uniformly distributed over the inner edge..I')D  r a (227) + ~/:::'I:l) a'b'(M.3).·' . Plate loaded by shear force Q.M. Inasmuch as Q. Expressions for deflection and moments a re then w = .M. is given by Eg.2 a' .= M. . respectively.b' .. M y =M2 (r = a) (r = b) (e) where the radial bending moment M.b' .) The stress components are calculated by means of Egs.a' . (b) into Eqs. socalled annular plates. This must be equal to 2"rQ" (a) (b) Figure 2. resulting in a total load 2"bQ. 2. (a) Three successive integrations of the above give w=±c1 r 2 +c21n r+C3 (b) Referring to Fig. a' _ b' I.5b) We now consider a plate under shear force Q..M. 2.a' + . = 0.
25).6a.= 0 111.:IArt PLATES 41 the total shear force at a distanc~ r from the center. the following expression for the plate deflection is obtained IV = 5?e!"b 1(1 4D 1 __ '~~) [2 +"__ 2(1 + v) (/' .+ a 2 a2 a r' It is observed that if the radius of the hole becomes infinitesimally small.2 = _ .29)..CIRCI.6 . however. by letting Ql = Pl2nb. the method of superposition may be used to good advantage to simplify the analysis_ The method is illustrated for the case of the annular plate shown in Fig_ 2.9c) becomes = QJ hT .1':[H: (r~.+ c.1) 4D C3 + '. (2. as expected.In . and Eg_ (2. reduces to Eq_ (2.8 DEFLECTION AND STRESS BY SUPERPOSITION The integration procedures discussed in the foregoing sections for determining the elastic deflection and stress of loaded plates are generally applicable to other cases of plates_ It is noted. (d). (2. 4 c /.29) +.28) is obtained by successive integrations of Eq.In ..=0 (r = a) (I' = b) (e) Upon substitution of the constants into Eqs.28) Qlh/. 2. that the solutions to numerous problems with simple loadings are already available_ For complex configurations of loads therefore. (In r .2 (d) The constants ('1> c" and outer and inner edges are determined from the following conditions at the w=o M. The plate is simply supported along its outer edge and is subjected to a Po jUJIDHliiJJlli (aJ (b J (c) Figure 2. b' ln~] a r 2b' 1 + v b rl . \Ve thus have Qr and Eg.In a I _b 2 1 v a (2. (2. b' In (bla) vanishes.b' (I.~) J The d isplaeement w W = 1)~) In r + ('..
0 'lUJ <:>%'i IP1m.6c (Prob. 7 .22 0. Shown in Fig. 0. itself assumed to experience elastic deformation.0 4.0 5.j.80 k 0.88 2.114 .0237 0. Inner and outer t'dges clamped ~U. Employing similar procedures.0 3.6c is an annular plate carrying along its inner edge a shear force per unit circumferential length Po b/2 and a radial bending moment.7..4 and 2.0 3.5 k .405 0. The solutions for each of the latter two cases are known from Sees.0 4." = k 2(p(){//t l ) Concentrated load: ~~'m~~ = k 3 (Pa 2 /Et·I ).54 2.b U""i i.' In all cases v. the deflection and stress at allY point of the plate in Fig.' c.5 2. We now consider the case of a plate supportedcontinllollsly along its bottom surface by a foundation. (2.18).44 Fd. 0. 2. 2.0064 0. 2.110 0.S Table 2. 2.062 0.830 1. Pn I JI' 2.3.976 1.6b and 2.414 0. B.13 uniformly distributed load Po at its surface.b')j16.0 4.0 5.824 0.273 0.08 2. a.0 0.0 5.6a can be found by the supe~position of the results at that point for the cases indicated in Fig.9 THE RITZ METHOD APPLIED TO CIRCULAR PLATES ON ELASTIC FOUNDATION In the problems discussed thus far. Inner and outeI" edges clamped alb 1. 1'0(3 + v)(a' .3 Variously loaded annular plates Uniform load: wma~ = k!(p~)aol/Et3).092 0. 2. is taken as v = 0.23 2. 2.179 0.42 STRESsES 11'\ PLATES A:":D SHELl.ti ~ . 1 FJ~I I ~ 2.234 1. support was provided at the plate edges. Table 2.813 0.664 0.u ~.14).19 0.0 0.703 0. The foundation reaction forces will be taken to be linearly proportional to the plate deflection at .0062 0. annular plates with various load and edge conditions may be treated.6b is a circular plate under a uniform load Po. I . Figure 2.nu = k 4 {Plt 2 ) A. OtHer edge supported alb 1. Hence.0329 0. Design calculations are often facilitated by this type of compilation.933 1.0 3.3 provides only the final results for several examples.71 1. defined by Eq. "m. k.5 k. and the plate was assumed to undergo no deflection at these supports.
:.Pe o Applying the minimizing condition. for example. from Eq.l')dWd'W... and airport runways. having the dimensions of force per unit surlacc area of plate per unit of deflection (e. (e) + [J 2  W.0 dl' 2 W I' (2.• f(dd.2(1'.dr . = 4clDrra 2 (1 + v) (e) The strain energy owing to the de/ormation of the elastic foundation is determined as follows o '0 The work done by the load is given by U2 = J .. but approximates closely many real situations.. (a) n (b) and the strain energy.!cJ/lI1dmiol1 or bedding COflstal1l or the foundation material. The above assumption with respect to the nature of the support not only leads to equations amenable to solution..1l) reduces to U = nD 1 I.. IJere l\' is the plate dc1kctioll and k is a constant. the expression for strain energy given by Eq. IT = U.CIRCULAR Pl.21t .g. 'vI." +..Id\l')2 .a 4 + ~da6) . If we retain. i.30)..e.l I'dI' . In this case of axis ymmetrica I bending.30) A solution can be assumed in the form of a series (a) in which en are to be determined from the condition that the potential energy of the system in stable equilibrium is minimum."o=Peo The potential enerlo'Y. only the first two terms of Eq. [I Co + (1/3) + P 3i~(l + V)/ka 4 j (j) . floor structures." I !kw r dr dll = trrk(c oa 2 2 + COC2a4 + !da 6 ) (d) W=P'(w). = 0. bridge decks. is thus n = 4clDrra 2 (1 + 1') + j: (e5a' + c oc. ' dr . aIT/ilc.' Examples of this type of plate include concrete slabs. We shall apply the Ritz method (Sec.ATES 43 any point. is then [J. we find that = rr:a. (P2.. 1. termed as the modulus (~r the ..9) to treat the bending of a circular plate of radius a resting freely on an elastic foundation and subjected to a center load P. (2. Paim).
(a) as shown in Fig. (a) 2. our concern was with the circular plates loaded axisymmetrically. center (..5) yields A = 1/192D. 1/ Plr5'cos 0 w =A··_··P a \ Introduction of the above into Eg. cos 0 a . Consider the case of a clamped circular plate of radius a and subjected to a linearly varying or hydrostatic loading represented by I' = Po + 1'1 ~... J (2. we obtain the maximum del1ection at the ~1:a2 [I + (I/:3)+32~(l + v)/".. 2. For the linear portion of the loading.4. For analysis of deflection and stress we must obtain appropriate solutions of the governing differential equation (2. is 4 w~ = l' o r /64D.7.. We now turn to asymmetrical bending.44 STRESSES IN J>LATFS AND SHELLS Theil by substituting Egs. The particular solution corresponding to Po. We thus have Wp 1'01'4 = 64D 1'1'" cos 0 + 192aD (c) o r "'~o+~ f a L Figure 2. 2. = 0): >I'm" = U) into (b). The boundary conditions are w=o  ow a. =0 (r = a) (b) where W = Hlp + W h .10 ASYMMETRICAL BENDING OF CIRCULAR PLATES In the foregoing sections. referring to Sec. (2..5).31 ) An improved approximation results from retention of more terms of the series given by Eq.7 .
and shear) must be finite at tbe center (r = 0).ATES 45 It is noted that the general method of obtaining the particular solution or Eq. Owing to the nature of p and wp' we take only the terms of series (2.64D 4 2p.. (2.4 (Levy's solution for rectangular plates). Hence.. follows a procedure identical with that described in Sec.!. 2. Tbe deflection w" and its derivatives (or slope.. moment. a solution exists for all 0 provided that poa 2 64D.5). = '192D (e) The deflection is theretore "' = 64D Po (a2 _ r2)' +.8) containing the function /0 and /.J'.32) The center displacement is (f) . The homogeneous solution IV" will be symmetrical in e.2)2 cos 0 I92Da (2.. given in Prob.20.. It follows that Bo = Do = B.7) vanishes.!: (a 2 _ . 3.+ Ao + Coa = 0 4 which upon solution. = 0 in expressions for f~ andr.a C. (e) and (d) yield two equations in the four unknown constants Since the term in each pair of parentheses is independent of cos 0.= 192D PIa' A. = D. (d) The conditions (b) combined with Eqs. (2. leads to poa AD = ....cmCULAR Pl. thusr~ in Eq.
(2. (1. and P 2 acting at the center and at r (any 0) of a circular plate with simply supported edge (Fig.' due to E. for example.ar(3 + \') ] cos . . [. The expressions for the bending and twisting moments are... = 1..32) reduces to Eg.. (1)".. respectively.32) and (2.. when the loading is uniform. M = Po 1'. the work done by P. 2.(1')1' dr diJ '0 '0 .1"(1 16' iJ = II + . = I I p(r.. ) . The method utilizes the reciprocity theorem together with expressions for deflection ofaxisymmetricaUy bent plates.  . 2. (a) as follows w. ".<> [(l2(1 + v) .. owing to displacement w" due to 1'" is equal to the work done by P. Betti and Lord Rayleigh. (2.(r) is the deflection at r due to a unit force at the center.1'(3 l'...: ) sin (I The case of the simply supported plate under hydrostatic loading can be treated in a similar way.£'. h'mn Egs.48 [1'3 (5 + v). we may write: (a) That is..2) M.25). According to the reciprocity theorem. The deflection at the center IV.. In the cases of fixed and simply supported plates. .8). of a circular plate with a nonuniform loading p(r..!) = _I~V p..0 (2.11 DEFLECTION BY THE RECIPROCITY THEOREM Presented in this section is a practical approach for computation of the cenler deflection of a circular plate with symmetrical edge conditions under asymmetricalor nonuniform loading. ra( I  ~. 16 [a2( 1 + v) ' + \')) . Consider. IJ) but symmetric boundary conditions may therefore be determined through application of Eg.3 (5v + 1) ..33) M.46 STRESSES I~ PLATES A0:D SHELLS We observe that.34) Clearly.22) and (2.. (2. For the sake of simplicity let P.ar(3v + 'lv)] 48 a I)] cos 0 (2. Iv(r) is given by the expressions obtained by setting l' = 1 in Eqs..211: .14) as expected. IV" = w" and IV" = "'(r). = 0 and Eq. the forces P. owing to displacement IVI2 due to 1'. . 1'.
9b. (I) of the preceding section.J~ (a) r.. it is unnecessary to determine an analytical expression for thickness as a function of radius.22) into Eg. 2. subjected to arbitrary lateral loading 9 Except for the requirement ofaxisymmetry. 2. 2._.9b.0 (" •0 ((Po + p. the general calculation procedure is outlined.. 2. 2.2)1' d" dO = poa~ 64D The above is identical with the value given by Eg.9a). Prior to illustrating the technique by means of a numerical example. A knowledge of the derivation of the basic relationships [Sec.10. as in Fig. it is necessary to match slopes and moments at the boundary between adjacent segments.C'IRClJl. and the division of the plate into small (finite) ring segments. Upon substituting 1'(1'. and pistons of reciprocating machinery. (2. ': cos 0)(2/'2 In " a I' + . For each element defined as in Fig. In order to accommodate the substitution of a series of constantthickness elements for the original structure of varying thickness. Ii) = Po + PI (ria) cos () and Eg. but that the thickness is taken as constant for each. r I . _ (b) Figure 2. (2.34).. reconsider the bending of the plate described in Sec.12(b)) is not essential in applying the method. As the method treats the plate as a collection of constantthickness disks.9 . the development of Sec.12 CIRCULAR PLATES OF VARIABLE THICKNESS UNDER NONUNIFORM LOAD In this section we discuss an approximate method for computing stresses and deflections in solid or anllllal circular plates of variable thickness. 2. Several applications immediately come to mind: turbine disks. 23 applies.' _. = '1~61Drr . we have lV. setting I' = 1 and integrating. Note that radial lengths of the segments need not be equal. clutches. no special restrictions are placed on the manner in which either the thickness or the lateral loading vary with the radial coordinate. Consider a circular plate (Fig. The boundary conditions are handled in the usual manner.AR PLA1TS 47 To illustrate the application of the approach.
The outer moments acting on an element are then (2.9).M.37) Ll. the average load on that element (Fig. = aIM"~ ... Bending moments The change in bending moment in proceeding from the inner edge to the outer edge of any element may be ascertained by rearranging Eqs. = 4(1 + ~) 2(T::::~) In p 1.fI') p.44) as follows: (2.' =  4(1 + p2 + '. the moments are either given or assumed.M.In p I+v 4" + .. tn+.48 STRESSES IN PLATES AND SHELLS (11) Calculation Procedure The expressions developed in Sec.12(b) may be so arranged as to facilitate the calculation process..M. Consider.r. 2. Ll.In p v) 2(1 . for each joint between adjacent elements.Mo Similarly. 2.39) It now follows that the moments at the inner dege of the next element are found from MrQ=Mrl+/lM.)" + 1 = (M o.36) The notation thus introduced is next applied in the determination of the following quantities. = a(M" . with this end in mind.M. Et' D= 12(! _ 1'2) 1.(I  (r) .) + PJJ (M").35) fi. a plate subdivided into a number of annular elements with the applied lateral loading on each element denoted Q.. (2.p2 I + p' g.v) (2.). n'/ _.p2 In addition. the moment increments corresponding to the interface between adjacent elements arc Ll.M. We shan apply the notation (J. =t(1 .! ~ I 3 3 tn (2. = .38) M.(I _ p') 8" p2 Iv .40) . = 0 (2..p' ).In p 2. (M o.)" + Ll.+l = (M.o = M" + Ll.M.. (2... "Q = I .. = ~ i+v In p  iv s.+ g.) + PoQ At the inner edge.
.37) to (2..Q Al~" = I'M." = A M.. beginning with the innermost element.)/M. Table 2.CIRCULAR PLATES 49 Table 2.. Seep 3. Superimpose the values found in steps 1 and 2 to obtain the general solution: Mr=M~+kM~ M!I= M~ + kM~ (2. M. Step 2. Boundary conditions The following steps are taken to satisfy the boundary conditions at the inner and outer edges of the disk: Step 1. When Egs.4... = AI".40) to obtain a particular solution (denoted by a single prime). (M~b ~ VM. (2.41 ) The constant k is calculated as indicated in Table 2. (2.J.h) .5 from the given boundary conditions at the outer edge of the disk..r.b)/(M~h ~ I'M.4 Assumed values at inner boundary of plate (A and 8 are any arbitrary values B of 0) Given inner houndary conditions Particular solution HOJnogl!1l0US solution M.5 Given outer boundary conditions Formula for kin Eq. (2. Apply Egs.=O AI~lI = B M~u= B AIr" Clamped Solid plate A1.41) M'b Clamped k k ~ =< (M" _..=B = M.40) are applied successively.. Begin the calculations with the appropriate inner boundary values specified in Table 2. Repeat step 1 with Q = 0 to obtain a homogeneous solution (denoted by a double prime). the moments at any intermediate edge may be found in terms of the moment at the inner boundary of the disk..37) to (2. M~'r= A iH."'" I'M:" M~" = A M~" = M." M~([ M.
(2._1/=5 I"f) ::: 360 102." 130 36 kN (on 360¥mrndia. Ll.42) where 111. Deflections The change in deflection.10 . use thl? values found in the final step to perform the calculations indicated in slep 1.. Stresses On the basis of the moments now known. 180 I" J 2 (on 260~lIlm~ diu.45): (2. circle) I t::: 20 . (2.42). circle) r::: 90 n=2 t = 25 11=1 t" 30 OJ "2.25 kN 9kN (on r = 290 r = 230 /". 150 kP3 (pressure) _. =~ (' (a) The stresses thus found will show a stepped distribution throughout the disk owing to the nature of the analysis.. (2.41). the bending stresses are calculated from: a. The actual distribution of stress may be approximated adequately by drawing a smooth curve through the calculated points. 6M. The results thus obtained should be the same as those already found at the conclusion of slep 3. circle) n=4 t" 16 1l~3 dia.w.50 STRESSES IN PLATES AND SHELLS To verilY the correclness of the final results. and Mo are given by Eqs. is for each element found by rearranging Eq. The total deflection at any point in the plate is thus computed by adding the increments given by Eq.08 o'~ Shearing force (kN) (a) (b) (e) Figure 2.
9).48800 0. Eqs. is next obtained by applying the expression for the outer edge (Table 2.35).10c.03271 0.CIRCULAR J>l. k.02024 0. Table 2.016 37790 0..00026 108. 2. 0.26035 ·0. The foregoing step is repeated for g = 0 to find the complementary solution (Table 2. The results of the complete calculation are presented in Tables 2. J.lOb. the above procedure remains applicable but requires the superposition of a number of determinate cases. (a) and (2.. Let E = 207 GPa and OJ.290 0.948 .130 0.290 0. ~ N • ).025 94450 0.00093 3.5): k = At. and tv1./D)IO' m/N 0. .00067 10.05254 0.37) through (2. All dimensions are in millimeters.17549 0.= SOLUTION The uniform lateral loading applied to each segment is shown in Fig. (2.6 Plate coefficients Eh':ment number. M~u = 100 (arbitrary).03215 0.48800 0.130 0.360 0.Q30 2080 0. Following this.912 4 0.02575 0.012 10720 0.10a.00022 395..230 0. Divide the plate into five segments as in Fig. (2.17554 0.M~b = 0 .03280 0.(595~~!. \Vhel1 indeterminate siluations af\.7)..03615 0.090 0...01411 0.03271 0.' encountered. .40 5 0.40) provide the particular solution (Table 2. M~u = M" = O.02924 0.9.020 49970 0.42) provide the final stresses and deflections (Table 2.6 through 2.01365 0. are then calculated using Eqs.01456 0.02354 ··0.22603 0.180 0.57813 ·0. The constant. The member is frce at the inner and outer edges. By assuming for the free inner edge.421l0 0.97 r. Example 2. (2.23920 ···0.~ )'Q (r.18550 0. 2. Eqs. Table 2. 2.21324 0. Symbol II Units m III 2 3 0.305 0. I " III Q II.16774 0.ATES 5] The foregoing proccdure~ apply to statically determinate prohlems.41).18176 0.8). II.04699 ·. .6 lists the various coefficients calculated on the basis of Eqs..19176 ·0.2 Calculate the stress distribution and the detlections of the steel disk shown in Fig.125 The final bending moments M.180 0.1l = 236.230 0.995 M~b 25.00031 34.01259 0.
49 2005.14 11012. M~o lUJ17 20.14 1481.66) 117.833 M.208 26.638 3396..48 23799.72 44.33 60.31 7965.4g 2993.657 5 27.965 26.833 29.639 25.9c) is  Q 2nrD (2.734 105..830 27~639 2 0 26. stresses.318 4313.148 ~ 5587.744 ~ 4 ~~~5587.o) x to~n (a. (a.71 5 282.732 6.035 100 73..730 2.833 46.406 2~470 32.5954. and deflectioll' Element number.89 282.) x lo~6 (a tl. Symbol M~.5 17652.778 1.)09 3.l 30. M~o M~j 11 2 3 4314.~F.256 2824. (i1w) x 10' (IV.89 4516.007 ~ 3312.) x 104 28. Symbol n 4 29.830 15.862 Table 2..41 11.770 0 26.642 3..499 (b) Basic Derivation The differential equation governing a circular plate of constant thickness.467 22228 67.S IN PLATES AND SHELLS Table 2.611 6267.8 Homogeneous solution Element number.182 9.. M..747 52.621 47~OO9 (a.778 34.548 439.078 6..682 1.036 158.) x 10' a 42.744 5 ~6267.999 4. (2.718 76.26t 1564.148 88. M.DJ5 30.7 Particular solution Element number.813 ~2264.318 100 M~" 122.522 18.533 3 2993.44 3493.769 13.43) ..600 .241 1821.094 41. n Symbol M.660 Table 2.49 0 636~53 M.296 . subjected to a constant shearing force Q = 2nrQ" from Eq..91g 135..813 0 135.73 1392.897 4 1481.9 Final moments.52 STREs.26 6305.) x 10· (.571 ···2520. Alro MOi 2 0 6305.125 13319 15.
Q (2 In d.v)] + T (I + v)c.. 1' ('2 Solving for ('. M0' .(1. To accomplish this... (b) is d'\\' = 1 .ATES 53 fntegrating this expression twice.c..v)] +2 1'. Moo = '2 (M" + M.. yields c. r .CIRCULAR Pl.y (M" r.9b) with the result for adjacent elements . where C1 = 1 D 1 Qr  8" c.. d 2 wjd.'i J I' .1) + ~ + ... (d) + (1 .. ~ Q .) + '2~ (Mo..91»...v) (1") 1'4n tv 8n r. Egs.. (d) and the bending moments M" and Mo. = . The derivative of Eg.y ro Continuity across the joint between adjacent elements is satisfied by eguating the radial moments M. (e) are substituted into Egs.) (e) Next.2 D 8" 1 I' e Co + 1) + '.v).(MOi + M. at the inner edge moments M". (b) and (e) into Egs_ (2. and slopes dwjdr at each side of the interface.. (2 In r .v)7 "2 Mo= [2(1 8" + (I + v)e.) Q . we obtain dl\' d.9a) and (2.. = of the element in terms of bending M o.9a) and (2. we have M. Q +(1 + v) In '(1 .2 ')." (c) Substituting Egs. + M" + l+v ('2 = [M~ =~~" ~  In r2n' Sf N M.. Q +(I+1')ln'+(Iv) 4n rv 8n (?) I. v) In.[2(1 Q Q + v) In I' + (1 ..J 2 /' (6 ) and ('2 are constants of integration which may vary from element to element.2 is eliminated between Egs.)  1 1 . at the outer edge of the same clement are found: M" = 2. (2.  I 1 1'2 en M.8. and c.  (1.
.=0 ) dr (il) This relationship is applied in the construction of Table 2.45) The boundary conditions at the inner and outer edges of the plate are related to the above expressions in the following manner. the above assumes the form D l..2 .11'" dw ~~ dr = .I'M. (e).. I 4n ~ri 8rr .rIO I'dI' + (2 "Yj "Y.~ (I Q ..2) .)" = (1.1'2)D" " Because the quantities expressions.~r.w = . the general solution of dw/dr in Eg.+1 (2. • WI dr I I' In r dr + .54 STRESSES IN PLATES J\i"iD SHELLS (M"" ~ df.r" 1 rill' + . . we obtain ~ D tJ... 2 2(1 + 1') (2. (2.~"). (b) over the radial length of an element: D I .\' )'/.dr r i .46) . MOi+Mrir..r.. J C I ..43) may be shown to be expressed dw = (dW)' + k(~\\')" dr dr til' (2.ro 1 (i) Performing the indicated operations. and A C2 from Eqs. The change in deflection w in proceeding from one edge to the other of the same element is found by first integrating Eq. By direct substitution. ~~ Q  i 2  fo (j) Introducing c. I .\l\! = .] + 21 I c (.44) also indicate that in the event the edge of a disk is fixed.[I" 2 In r + r2 .c. = (1 ~~~ r')/)" II (di\~\)'l tr" (/' 0 /I (AJ" . (i~\..44) and dwidr must be equal ill both of the above It then follows that the second of Egs.. I' I!. ( since dw .. (2. ro Q .44) divided by the first leads to (g) Expressions (2..I' 2 In I' 8n I 0 0 n 2r..4. In .
note that the particular and homogeneous solutions may be selected so as to satisfy the conditions imposed at the inner edge of the disk regardless orthe value of k.tress is limited to 90 MPa: (b) the maximum deflection of the cylinder head. 2.25 m radius and flat thin plate head is subjected to an internal pressure 7 MPa.:.4 The flat head of a piston is considered under a pressure (0 be a clamped circular plale of radius (I.1 10 2. and J. The constant k can thus be selected to satisfy the prescribed condition at the outer edge only.. A.)" ~~ where Po is constant. The same relationship applies to the moments given by Eqs.5. the double prime indicates the homoYl'Ill!oUS sollltion. and C of Table 2. assuming failure to take place according to the maximum principal stress theory.)' ~ Si6D  3 ('.)' + 2 ~ I (P2. (f). as indicated in Table 2.3 An aluminum alloy (6061·T6) flat simply supported disk valve of 0. B. Usc E = 200 GPa and v = 0. Let v = 0.4.3 and E = 200 GPa. Referring to Table 2. What is the factor of safety.8) .3.4) for Ihe resulting displacement.2 m diameter and 10 mm thickness is subject to a water pressure of 0. verify the results provided for the maximum deflections.2.030 m and is builtin at the edge.7 For the circular plates loaded as shown in the Figs. PROBLEMS St'('s.8 A simply supported circular platt! i!> under from the center to the edge: II rotationally symmetric lateral load which increases Show that the expression (Pc. 2.5 10 2. Derive the equation w = p a' [('. The yield strength of the material is 241 MPa. 22 A cylindrical thick. 2. repr':scllts a constant.ClRn'LAR PLATES 55 where the single prime denotes the particular solution.walled vessel of 0.8 2.1 A pressure control system includes Ii thin steel disk which is to clost' an eit'ctrical circuit by deflecting 1 mm at the center when the pressure attains a value of 3 MPa. Determine: (a) the thickness of the cylinder head if the allowable s. Calculate the required disk thickness if it has a radius of 0. 2. The head is f} = fin (. 2.5 MPa.
15 for a simply supported circular plate that is loaded only by a concentrated center force P. Cl = 0.. Set:s. 2. (P2.8).3. 2. and thereby transforming the resulting expression into an ordinary differential equation with constant coefficients.~ for j'l() /.7) into Eq. and V= 0.5b by setting a = 2h and \' = 0. . (2. Let k = 200 MPa/m. ) (P2. using the method of superposition. ) Solution of these equid imcnsionul equations can bc taken as: 1(' f.3. M J = 2M 2 . 2.9) (N ~ 1. taking the first three terms in Eq..15 An aircraft window is approximated as a simply supported circular plate of radius a.lOa}.] 2.9 Verify the result given by Eq. 2. Determine its maximum deflection. Take v = 0. Compute.9)./'. 2.12 Calculate the maximunl deflection w in the annular plate loaded as shown in Fig. Compare the result with that given in the table. (2. {2.'iultillg displacement.12 2. Use the Ritz method.2. (a) of Sec. (1.. E = 200 GPa. 2. . 2.S6 STRt:SSES IN 1'1An:S AND SHELLS where represent lhe re. assuming that a diametrical section of the bent plate i~ paraholic.5a by setting a = 2b.12).5 m. (2.9 to 2.3.3.13): (a) by integrating Eq.14 A pump diaphragm can be approximated as an annular plate under a uniformly distributed surface load Po and with outer edge simply supported (case A in Table 2..2 cO aWl'] rdrdO (P2J 1) 2. setting 1 = In r. 2. .13 Determine the maximum displacement in tht! annular plate loaded as shown in Fig. (2.(r) = d" r" andf! (r) = ".3). 2.11 Show that Eq.34) for the strain energy results in the following form in terms of the polar coordinates +2(1 1 I i"w ~v) ( . wherein {I" and bIt are constants and the ):s arc the roots of auxiliary equation of Egs. 2. [Him: Introduction of Eq. (b) by expanding Eq.3. and v = 0. 1. Employ the Ritz method.10 Verify the result given by Eg. 2.17 Determine th~ maximum deflection of a structural steel circular plate with free end resting on a gravelsand mixture foundation and submitted to a load P at its center. (2. t = 40 mm. 2.6) (N ~ !ead~ to 0. the maximum plate deflection for b = a/4 and v = 0.9. The window is su~ject to a uniform cabin pressure Po.16 Redo Prob.
CIR('UL,\H Pl._AlES 57
2.18:\ i,lmply 'iupported l.'irculal' plate described by
i~
loaded by u,:vlllmetrically distributed edgf
coupk~
AI,
III this case, it is observed thaI
=
/_"0
I
I, 1 ...
IU"COSIIO
(t
~
II)
\\'1'
=0 =
and thus
\1'
= \\'h
reduces to
II
I
"r/,\ ...
(A"r" +
C"I
J
"'2) cos nO
Verify that the resuiting deflection is
11'=
(P2.18)
2.19 Detennine the expression for the radial stress in the plate described in Prob. 2.18 by taking /1 = 0, 1. 2.20 The particular solution of Eq. (2.5), for an arbitrary loading p(r, 0) expanded in a Fourier series
pit. U) ~ 1',(1")
+
l: [P"(I") cos 110 + R.(t) sin 110]
Q'~
(a)
J
where
p.(,.) ~ / rI>', 0) cos ,,0 dO n .. r.
1 .'
(II
~
O. 1. ... )
R,,('") ~ is expressed in the general form: 6
1 ."
1t._~
I
p(t, 0) sin "U dO
(II
~
1,2 ....)
I\'JI =
Fo(r)
+ L [Fn(r) cos nO + Gn(r) sin nO
n""J
(Pl.20)
Here /'0(1'), F,,(r), <l.ud Gn{r) are functions of r. Demonstrate that substitution of Eqs. (lI) and (P2.20) into Eq. (2.5) leads to
(b)
Thus, for a prescribed loading p(r, (}), Eqs. (b) are solved for Fo, is then obtained from Eq. (P2.20).
F~,
and Gn . The particular solution
58
STRESSES I!'< PLATES II~D SHELlS
2.21 A simply Sllpporl<:d circular plate is subjected to hydrostatic loading. Employ the rc<:iprocit~ tbeor~m to find the maxim lim dcllcctioll. 2.22 A clamped circular plate carries a concentrated downward load P at a point located at 3 distance b from its center. Apply the reciprocity theorem to obtain the center deflection. 2.23 Redo Prob. 2.22 for the case of the plate with simply supported edge 2.24 For the turbine disk shown in rig. 2.100, compute the stress distribution and the deflectiollf., assuming the disk to be clamped at the inner edge and free at the outer edge.
CHAPTER
THREE
RECTANGULAR PLATES
3.1 INTRODUCTION
Jn this chapter consideration is given stresses and deflections in thin rectangular plates. As observed ill Chap. 1, the rectangularplate element is an excellent model for development of the basic relationships in cartesian coordinates. On the other hand, we shall see that rectangular plates in bending frequently lead to solutions in the form of series that are unsuited to hand computation of numerical values. That is, the deflections and moments are often described by unwieldy infinite series, and the summation of these series presents difficulties. Such computations are, of course, readily performed by digital computer. Rectangular plates are generally classified in accordance with the type of support used. We are here concerned with the bending of simply supported plates, clamped or builtin plates, plates having mixed support conditions. plates on an clastic foundation, and continuous plates. The latter often refer to structures consisting of a single plate supported by intermediate beams or columns. All cases are treated by relationships derived in Chap. L The strip method of Sec. 3,8 is a discussion of bending of rectangular plates based upon elementary beam theory.
3.2 NAVIER'S SOLUTION FOR SIMPLY sUPPORTED RECTANGULAR PLATES
Consider the rectangular plate of sides a and b, simply supported on all edges and subjected to a distributed load pix. y). The origin of coordinates is placed at the upper left corner of the plate as shown in Fig. 3.la.ln general, solution of the
59
60
srRf,SSFS 1:,\ PLAns ,~:\D SHELLS
r
b
f~a~  . , 
, , , ,
I
, , ~X ,
I I
1
I
I I L
,

,I
(b)
fy
(a)
I
Figure 3.1
bending problem employs the following FOl/riel' series (App. A) for load and detlection:
p(x. y) =
w(x. Y) =
L L Pm" sm
m=111""1
m=ln=l
(,(.
oc,
L Lam"
00
oo~
mnx a . mnx sm·a
•
. mcr sm ~b . nny sm b
(3.1a)
(3.1b)
where Pml! and amI! represent coefficients to be determined. This approach was introduced by Navier in 1820. The deflection must satisfy the diflerential equation (1.17) with the following boundary conditions (1.26h)
w=o
w=o
··i?:
a2 w _ ...
oyZ
(rw
,0
=0
=
(x = 0, x = a)
(a)
0
(y
= 0, )' =
b)
Clearly, these edge restraints arc fulfilled by Eq. (3.1b) and the coefficients am" must be such as to satisfy Eq. (1.17). The solution corresponding to loading p(x, y) thus requires determination of Pmn and amI!' As a physical interpretation of Eq. (3.1b), consider the true deflection surface of the plate to be the superposition of sinusoidal curves of m and n different configurations in the x and y directions, respectively. The coefficients am" of the series are the maximum central coordinates of the sine curves, and the m's and the n's indicate the number of halfsine curves in the x and y directions, respectively. For example, the term au sin (nx/a) sin (2ny/b) of the series is illustrated in Fig. 3.th. By increasing the number of terms in the series, the accuracy can, of course, be improved. We proceed by dealing first with a general load configuration, subsequently treating specific loadings. To determine the Fourier coe[ficiellts Pm", each side of Eq. (3.la) is mUltiplied by
JU,CTANGl'L;\R 1'1.\ ITS
61
and integrated between limits 0,
(I
and O. b:
I I pix, y) Sin  '0 ·'0 a
=
.Ii
,II
•
minx
.
JI'nv
SIll '
b
dx dv '
?.b.II,
'0'0
L L Pm" I I SIO
m"'"'IJI=l
,r...
~...
I1mx , nny , m'nx . /l'n)' ~ Slll  SIO   SIO 
a
h
a
h
ds dr
(b)
It can be shown by direct integration that
mrrx . m'1rx I S10 .... ... Sin   dx = JO
.<1.
'0
(I
a
la/2
(m of m') (m = m')
.0
I
,b,
mry .
S1l1 
b
n'njl sm ~. dv =
/0
Ib/2
b

(/1 + n') (/I = 11')
(3.2)
The coefficients of the double Fourier expansion are therefore p"", = ab· o
4
mnx . I I pix, y) sm ~ sm a
.b
.11 •
mry
'0
" dx dy b
(33)
Evaluation of am" in Eq. (3.1b) requires substitution of Eqs. (3.1) into Eq.
(1.17), with the result:
~ L L
w
/
m= I
n~ 1
a/
IIII
1
[("l7t)' + a
'
2 '
(/l7t)'("n)' + (m,)4) .... .a
b b
p"",I .. _./ Sin
Dab
~ Sill _._. = 0
InITx .. lin)'
This equatioli must apply for all x and y. We conclude therefore that
from which
I Pmn a = ...... _. __.__._ .... __..... _.. "''' n4 D [(m/a)' + (n/b)']'
(3.4)
Substituting Eq. (3.4) into Eg. (3.1b), the equation of the deflection surface of the plate becomes
W
= ',,
1
n
L L D ","1,,"1
00
oc,
2"'~2',
. mrr.'\ . I11tV ~ sm ~ [(m/a) + (II/h) ja h
Sill
Pm"
(3.5)
in which p"", is given by Eq. (3.3). It can be shown. by noting that Isin (lnnx/a) I <; I and Icos (mr)'/b) I <; 1 for every x and v and for HI and II, that series (3.5) is convergent. Thus Eq. (3.5) is a valid solution for bending of sirnply supported rectangular plates under various kinds of load ings. Application of the Navicr's method to severa1 particular cases is presented in the next section.
62
STRESSES 1;..: PLATES AND SHELL.';
3.3 SIi\IPLY SUPPORTED RECTANGLLAR PLATES UNDER VARIOUS LOADINGS
When a rectangular plate is subjected to a lIni(or/lllv distributed load p(x, y) = Po, the results of the previous section are simplified considerably, Now Eg, (3.3). alter integration, yields
Pm" =
or
4p
_".0.
lcnm
(1 _. cos nm)(l  cos nn) = ._,0 [1  (1)"')[1  ( I)")
nmll
41'
(/ll, n = 1,3, ... )
(a)
It is observed that because p"", = 0 for even values of III and ll, these integers assume only odd values. Introducing Pm" into Eq. (3.5), we have
w = l~)o
ff
,,6D ,,";;'
~ill~l1",,/aJ~in (II,,),/h) IIln[(m/lif + (n/b)']2
(111,11= 1, 3, ... )
(3.6)
Clearly, based upon physical considerations, the uniformly loaded plate must deflect into a symmetrical shape. Such a configuration results when II! and narc odd. The maximum deflection occurs at the center of the plate (x = a/2, y = h/2) and its value, from Eg. (3.6), is
16po
IV ",
m
u;;
(I.)
(
____
1)[(m+ll)f21t
=
nOD
~ ~ ;n;;lr;;i/a)Ft(i/;FFY
(b)
Note that in Eg, (3.6), sin ",n/2 and sin 11,,/2 are replaced by( _1)'mII;' and (I ),,,1,,2, respectil'ely. Introducing Eg, (3.6) into Egs, (1.10), the components of the moment are derived:
M, = 161'0
.
)'
,," '" , "'n[(m/a)' + (n/b}']'
ff
(",/a)2 ; v(i1/bY ... sin n'."~" sin
".'2:
b
(3.7)
a
M = _.
16po W «. v(m/a)2 + (n/b)' . . I1ny L L ~... sm mnx sm _....". '" " 11111[(mja)' + (ii/b)')' a b
~.
M
"Y
= 
16(1  \') L L _ ...........  . . cos III"X cos ......... . ~ .' I I11'Y ,,"ab "''' [(",/a)2 + (i1/b)2]2 .. a h
We observe that the bending moments M" and My are both zero at (x = 0, x = a) and (y = 0, y = h) respectively. However. the twisting moment M.,), does not vanish at the edges and at the corners of the plate. The presence of M xy causes a modification of the distribution of the reactions on the supports (Sec, 1.7). Recall, however, that SI. Venant's principle permits us to regard the stress distribution unaltered for sections away from the edges and the corners.
Figure 3. The maximum bending stress produced by the moment of Eq. (1.0534P o 02 while the first four terms result in (e) . \I'm" = 0. by application of Egs.3. (3. partiul.3) results in what is essentially the" exact" solution. 11 = 1. (h) is demonstrated by noting that retaining the first four terms (m = I.. Determine the reactions at the supports if the material is stored on the entire floor in such a way that the loading is expressed in the following approximate form . D ' It is thus observed from a comparison of the above that the series for the bending moments given by Egs.sin h a (d) Here Po represents the intensity of the load at the ccnter of the plate. und concentrated loadings arc discussed in the examples 3..7).281 Po a 2(t 2 Example 3. 111 = 3.7) does not converge as rapidly as that of Eg.1 A 'quare wallpanel is taken to be simply supported on all edges and subjected to a uniform pressure differential Po. p(x. The maximum bending moments. are determined by applying Eqs. moment. as shown in Fig. II = 1.00406po a 4 (D.. is determined to be "m" = 0.2 .4 = b: IV m " = 0.2. Determine the maximum deflection. (3. and stress.\\1Gl!LAR PLATES 63 Sit uations involving simply supported rectangular plates carrying sinusoidal.RFCr. found at the center of the plate. 11= I) of Eg. SOLUTIO"' The first term (Ill = I.l2). max = O. 3. y) ~ nx ny Po sin _. nnx = Al y. (e). The first term of the series yields Nl x .2 A rectangular warehouse Iloor slab of sides a and b is simply supported on all edges. (b).004161'0 D Very rapid convergence of Eg. (h) yields. for a <. 3. Example 3.2 and 3.
The minus sign indicates that they are directed upward as shown in the figure. cos . .... Substituting Eq.. Y ny en (a) a b The total load carried by the plate. dx dy .8) I·... V) sin 1[a~ (3.1. = R..y = .a sm .. is equal to . (e) !vI..10).J.. nx Sill ._.'( Q = ... (1.. dx + 2 I Rx dy '0 ....9)..'(/{I/a2 + I/b')' Q2 + 'b2 Sill I) Similarly. nx ./J .231». (...9) The edge support reactions thus also vary sinusoidally (Fig. neglecting its weight...5) in this case leads to W = .... try a b Introducing \\' from the above into Egs....IJ .. ny 4poab I I Po Sill ..~ for the edge X= a: Po (1 2. nJ' ~~. one has nb(I//'!. . n)' Rx =. setting R). 1t r ... {3. we have V).16) Po nx .. + b"i: 8111 !vi 1[X . respectively..v) ny cos h The shear forces are determined from Eqs. (1. . (1... Equation (1..23a) is applied to obtain the reactioll Rx = .= '' h .1) and letting Pmll = Po' Expression (3..nx.2).'(1/..tl ..."i/b'Y (~2 + 2:...tl '0'0 1r The reactions can act only vertically inasmuch as the edges sit on rollers.' + l/h')2(/1) cos Q' Q = x Po{1 . · S I I l na{l/a' + l/b ' ) a b Po nb(l/a' + l/b 2 ) SillCOS . Conditions of symmetry dictate that the reactions along the support at x = 0 and y = 0 are identical with those given by Eqs..8} and (3. (d) into Eq. from Eq. ~~ SIll t y = "'{I/(/> Po + 1/li)2 (" a' + b' I ). SIn . \Ie have: Po ' n4DO/a' + l/b 2 )2 n. 3..1[.64 STRESSES 11\ PLATES AND SHELLS SOlXTIO". for the edge y = b. =  (3....)..x . The sum of distributed reactions 2  '0 I R. SI11 III = /I = I. 1rV sm t.
. we have through the use of Eq. as already shown in Sec.~.) that the distributed reactions are larger than necessary to compensate for the p loading. dx dy from which Pmn = 4P n2/.7. 3. (i) Figure . supported freely by a rectangular frame..I'l I d I .. SID mITC ~. 41'0111>  81'.XI fe. after subSlitllling Eqs. l')  ~j' . S1l1 ~a  SIn b 1l1r:J·1 .j~._ Y = Yl' SOLUTION (a) Since I'(x.9) and integrating.. The reason for the corner reactions can be recognized intuitively.2). is subjected to center loading.. 1.10) The condition that the resultant of the laterally applied and the reactional forces in the plate equal to zero is thus satisfied (Fig. To prevent this. /tnt! S1I1 .qr a SIn .RECTAN"GlJLAR PLATES 65 is.br17. 1. _•. nnv Sill·· ".C'd I1mx!. Thus.3 Find the equations of the elastic surlace of a simply supported rectangular plate (Fig..(1 _.3) Pmn = .3) for two particular cases: (a) the plate is subjected to a load P uniformly distributed over a subregion 4cd.:·q and (3..24) (3. with contact made only at the middle of the sides. there \vill be a concentrated reaction Re = ~ Fe at each corner having a vallie (Eg. (3.. abcd·Y\···d P I ..tJ .. (g) and (I. the corners must be held down. When a rectangular plate. (b) the plate carries a (nominal) point load P at x = x.2+liPY (h) It is seen from Egs. mnx . it tends to deform in such a way as to have the corners rise. y) = P/4cd. b . (3.... 3. Example 3.
1l7fJ' b (3.5) Clearly.. (2... (3.01159Pa'jD and the error is thus 1.56 3. m = 5.l'db) sm·.. Eg. 3. Retaining the first nine terms of this series (m = I. Introduction of Eg.~117fx/a) sin (Ilrry/b) 4 L... 3.1b) results in the plate deflection: IV =. ..11) may be used with the method of superposition (Sees. 5. .11) reduces to . (i) into Eg. ..1'1 = h:'2j (' = ((/2.sin (I11r. 2 2' rr Dab '" " [(m/a) + (Il/b) ] ( 3.7) to determine the deflection of simply supported rectangular plates under various types of loadings. n = l.L. (i) appears in the form (Prob. Another advantage of the Levy's solution as compared with the Navier's methods is that instead of a double series one has to deal with a single . which occurs at the center.24). (j) into Eq.66 STRESSES J~ Pl. . When the load P is applied at the center of the plate (Xl = aj2.<.4 LEVY'S SOLUTION FOR RECTANGULAR PLATES It is seen in the foregoing section that the calculation of bending moments.. n = I./2..11) Eg. (3. D The "exact" value is percent.4) together with Eg. + (Il!h)']' . the foregoing re~ duces to Eg. using Eg. " =_~_ ~ ~(_I)[("'b)!2Jl sin . is from Eg.7) is not very satisfactory because of the slow convergence of the series. 2. L: L: sin (n"'xl/a). and U) Note that the load P is an idealization of the distributed load 1'(.1'1 = bj2).\TES AND :mEU.. 5) we obtain IV m" = Ll!'. (3. Inserting Eg. 3. (3.12): Wmax 4Pa 2 = n4D I L In II w oc I (m2 + n2):Z11 = (k) 1. (3.S d = 11. ror Xl = 0:2. (3. ..8 and 3. y) concentrated in a very small area of size defined by Eg. (b) When c and d are made to approach zero Eg..1h) gives the required detlection surface.. if the plate is square (a = b). An important approach which overcomes this difficulty was developed by M. 5. Itmx rr"Dab". = 0.a n4D 2 ['22 +~ 10' IVm " +1 18' +J 25' +~ 34 2 Pa +'j=001142 ' 50' . Levy in 1900. the maximum deflection. [(m/a)' a 4P 00 W S1l1 . (3. 3. (a).12 ) Furthermore. m = 3.
_ ~ 111'"1 L.\. 14' ')m _ 2()2d2{'~' + (". y) a solutiol1 \Vp must be obtained.£> IVII = ..\'O arbitrary sides at y = ±1>/2. (h) where the function 1..!. (3.18). a single expression can be derived for IV" that is valid for all rectangular plates having particular boundary conditions on the two opposite sides... m 0:= 1 sin .13) satisfies the conditions (1...Rf:CTA\:Gt. II .74". (1. ( .LAR PLATES 67 I· ... . 3. for each specific loading pix. at x = 0 and x = a) and arbitrill')' conditions of support on the remaining edges (at y = ±h/2) (Fig.) Isin (n. The total solution consists of the homogeneous solution l1'h or Eg. To complete the solution.  .. Figure JA series (App. (1.17): (a) Inasmuch as '. 3. (1. btr. We proceed with the description of the method by assuming that two opposite sides of the rectangular plate at x = 0 and x = a are simply supported as shown in Fig.. = 0 is independent of the loading. In this case.._J '. · ····a .13) Eg. p.. we must now apply to \V the boundary conditions on the fl. : II m1TY (3... Clearly..!X/a) IC05'(nmx/a ) .m (}') t.18) and the particular solution IV" of Eg.4.1 _.. and to satisfy Eg.1') must be obtained such as to fulfill the conditions of the supports at y = ±1>/2. A). x . Eg.13) into Eg. Substituting Eg. (b) appears as . The homogeneous solution is selected of the following general form "/. {.c . (1.'I '.4).~ 2 . dy4 "dy. m) r. The Levy's method is applicable to the bending of rectangular plates with particular boundary conditions on the two opposite sides (say. In generaL it is easier to perform numerical calculations ror single series than for double series.!~) 41m = 0 1I1n ../. : Ift". (.26b) along the x edges.18) yields •. " In order that this eguation be valid for any .
"".fm . mITy '0 + Bm cosh '::. 3.(. .. nmx J' cosh .. a + Bm nm.. (3. + dy2 (. y) in terms of a single Fourier series pix.18) (3. 3.:.x:' ( .1') = 1'" for which Eq... = Am Sinh . (119) yields..68 snO:::SSES I!\" PLATES A~D SHELLS Tile gelleral solution of the above is (see Proll. 0('.15) The homogeneous solution is therefore H'h = III~ 1. mrry Am smh .17). Simply Supported Rectangular Plate Under Uniform I. y) sin . (3. from Eq. ) (d) .17) where the k". . (Ll7) and noting the validity of the resulting expression for all values of x between 0 and a. we then obtain w. ~. y) wherein = l: p".. . to be determined later' for specified cases._..(y) sin OC mrc'( m"'"' 1 .. mITy a + Dm. .16) where . It is observed that (he boundary conditions (1.18) into Eq. a' a (3.. we find that ~4k".Q mnx '0 Substituting Eqs.. d)J 4 _ .' + (mY smh l' mrcy. upon integration 4p" Pm = .19) 1'..oading For this case pix..9) (3.)2 a ~~<".n". Let us also expand pix..::.(yj's are functions of y only.·V cosh .C • (3.1')=::: I pix..(y) Sill a<. . mn:v v smh ' a +D /U I1m y )..14) or by employing trigonometric identities .)\ a = Pm III D (e) Upon determination of a particular solution.26/» along the edges x = 0 and x = a are satisfied if the particular solution is expressed by the following single Fourier series IV p = m. The method is illustrated by can· sidering the following commonly referred to example. a (3..y cosh _ . k"" to this ordinary differential equation.. a mIT\' (3.·. (/ +C III.dx a a .17) and (3..Sill .l mnx l: k".1 n" Bm ) CII!> and Dill are constants..mrr (/11 = 1.. L: .
17) is therefore (3.200) This represents the deflection of a uniformly loaded. + 5..\. simply supported strip parallel to the x axis and may be rewritten in the following alternate form (see Example L1): wp =.05~.. The remaining edge conditions are lV=O Application of the above to w leads to two expressions. (1. a mnx (3. 3. Then..20) we have }~ = Bm '~':. (e) gives the unknown constants .17) and the simple support restraints at x = 0.. (3. These will be satisfied for all values of x when (e) in which cx.Cmy smh . ( L.. combining Eqs.. x = a. it must have the same values for + y and .21 ) This expression satisfies Eq.16) and (3. .. (e) becomes The particular solution of the above is k m = 41'" o"/m'n' D.»1=1.. 69 Then Eg..y (Fig.20b) The condition that the plate deflection must be symmetrical with respect to the x axis [i. Expression (3..e._ cosh mrry ~ a +..n = " 2il 11mb Solution of Eqs. (3..j{FCTA~GULAR f't·\lU.~ (x 4 24D'  2ax 3 + a 3 x) . (3.· ) sm m n /) 4 .16) if we let Am = Dm = O.3.a .4)] is satisfied by Eq. mny 4po .
00025 + ..'.::..~!~~·~·~c·~10?S·:h·~:~!~. arc given in Tahle 3. 2:.... (1 )'".... ) ~ 0.. I . anci 0. Since ~ m~~1..0.23) .24/J ) "2.. That is.3.22). and stresses of the plate can be derived by following a procedure similar to that described in Sec. I rn:' t I + . mum moments in the plate can also be put into the form (x Numerical values of the coefficients 01> = i.3 . y = 0) (3. An 1 mnp + ' ..24a) Expressions for the moments.~. edge forces. simply supported strip.23) The first term above represents the deflection Won" of the middle of a uniformly loaded. 3.00406j) ~~ It is observed that the result obtained. in the case of a square plate (a = band "".'n'D ~a' (0.68562 . even retaining only the first term of the series in the parentheses.~·~ cosh _:r:bm y .70 STRESSES !X PLATL:S A~D SHELLS The deflection surl.1 . (3.. the maximunl displacement is given by Wm " = 5~~ 384D .. ( 1)'m . Introducing the following notation into Eq.' ~29(3j we can write the following expression for the maximum dellection of the plate: (3. The second term is a very rapidly converging series. (3.. sinh 2 cosh:X1I! a 2:lmY ".ee of the plate IEq. will be accurate to the third significant figure. . For example. b 1 .11:2 n~' ... (3.1 for .. y ~ 0).m (3.. = nm/2).n!.L.~ m5 the maximum deflection of the plate is found to be 2 cosh (f) (J. The maxi. from Eq.22) SII1~ a The maximum displacement occurs at the center of the plate (x = ai'2. mnx (3..e 1)J may lhus he expressed w = 4jJoa'~ 7_ n5 D ') m""1:"3 ."" 5 . .L=Jtml~~~ am tanh .
m9 0. A window of a highrise building is approximated by a· rectangular plate with three edges simply supported and one edge clamped.16) and (3. I J.01282 0.20) '1'= m"'" 1.0375 0.25) \\'=0 \\'=0 ow ay fl 2 w =0 =0 (g) (y = b) oJ" _t:::.0479 2.1017 0. max decreases.~f~~~ )sin 111. _ t: : b x 1~ L Figure 3. ')2 . l l It is seen from the labJc that as bia increases \'v'rna'( and A1.:j. and .0 4.0479 0.0 3. Thus.)~ maY be summed only for odd integers of'n.0 0. The plate is under uniform windloading of intensity 1'0' Derive an expression for the deflection surface. mny .0406 omS4 various aspecl ralios bla of Ihe plate sides.1235 5. The general solution is obtained by combining Eqs. (3.01013 0. .53 0.1246 0. 3.0 0.5 and assume that the edge J' = 0 is clamped and the remaining edges are simplyslIpported." mny .00406 0. mnr ( Am smh ' + Bm cosh a + emy smh a a + DillY cosh Boundary conditions are represented by ~Jr +. Example 3.0464 0.5 .' (J' = 0) (3.01297 0.4. SOLUTION Let the uniformly loaded plate be bounded as shown in Fig. the deflection is s.\.RECTA:'TiL'LAR PLATES 7j Tahle 3.0122) O.1 bi(l 1.0 0. max increase while j\1y.vmmetl'ical about the line x = a/2.
Outline the derivation of the expression for the deflection surrace lV. In the case or a square plate (a = h) the center deflection and the maximum bending moment are found to be (Prab. are (l2w IV =0 0 ox' au.. on the two opposite arbitrary sides of the plate. the boundary conditions. I t .25). When Eqs.6 ..  2····· sinh /lm cosh {Jill .0028 Po.~ . ~.f!~I!. may be treated similarly as illustrated in the following example. // ///..h .10) w = 0..h2 Pm .6).. = m"I>!".~.1'  =0 0) (.. (/) f: ..' (x=O. £!!~!. (a) to H" leads to values of Alii' B"" ell(' and J)m: .. (3. (1.25) to (1.... . 3.5 The unirorm load Po acts all a rectangular balcony reinforcement plate with opposite edges x = 0 and x = a simply supported. x iu. SOLUTION For the situation described. a (/ cosh jim sinh {Jill .~~b 5 m n5 D cosh jim sinh jJm .1' "'\ Free Figure 3..~~~..Pm filii (II ) em:o::: __ .l~ ?5~.x=a) (. .(. 3.l_". Example 3. .~ (x = i' y = ~) (i) Situations involving other combinations of boundary conditions.1'=1» __ (l. and the fou~th edge y = 0 clamped (Fig.. the third edge y = h free.1I"l1t sinh Pill . an expression ror the plate deflection is established. D J1l1( 1/1 = ?eg.~.1' = (j) (k) w=o 0. Eqs.. a where /I".72 STKESSFS 1:: PLATES f\!'\'D SHELLS Applicalion of Eqs.(Jm nm (mn)' h cosh Pili . 8 m . (h) are inserted into Eq.27)..~. .~.."//' .
" = '.16) and (3. we obtain two equations which after solution yield B. x = (I.' = mnB. (8' If! + D'm J')emff}'la sin "a nlnx (3. It then follows that the elastic surface is given by 1 The caseS involving a clamped edge at y = 0 @1' a Fee edye at y = 0 may be treated in a like manner. (3.n. both of which satisfy the conditions (j). A~I and C:n should be equated to zero. I t l' Figure 3..RECTA~':GUI. (1. (3. (1.7).20). (3..20) with a suitable solution w" of Eg." and D~ in such a manner as to satisfy the boundary conditions on side y = O. .20) and (3.4Poa4 /n' Dm' and D. Substituting 2 lV = lV1 + wo_ into w = 0 and a . (k) and (I) to w" + IV p leads to definite values of the constants A"" B.3. Hence.. Applications of Egs. . respectively. Assume the edges x = a..25) and (1. It remains now to determine B.. of course. 3. It is observed from Eq.t. and y = 0 arc simply supported.. (120) and (126).14) that in order for the coefficients of W h .. . Example 3. respectively.'/2a. (1. (3.18).6 Derive an expression for the deflection surface of a very long and narrow rectangular floorpanel subjected to a uniform load of intensity Po (Fig. .. em.AR PLATES 73 ~l" . satisfies Eq. Dm· The deflection is then obtained by adding Ego.\'/cy 2 = 0 and setting y = O.26) The above. and its derivatives to vanish at y = ro. .27).18).16). applying Eqs.7 The particular and the homogeneous solutions are given by Egs. The boundary conditions of sides x = 0 and x = a are fulfilled by Eqs.. I I I r _t_ . SOLUTION The plate deflection can readily be obtained by superposing the solution for an infinite strip lV p given by Eg. the homogeneous solution w" can be represented by the following expression wIr = '"" L IT:! 1f!~1.
74 STRESSES 1:".2. dx a·o a (3.32) results in the displacements..12)..2.29) (3.+.4... ..PLATES AND SHILLS 3. Bounding the plate as shown in Fig. 3 2w/f1y2 = O. . Proceeding as in Sec.28 ) 2 mnx I pix) sm _ . 3. 1/1=1. . ..1:: mrrx 1'".21) then becomes 1r. Suppose now that the plate is square (a = b)..)' +C V m.2. w= rn=1. The moments and stresses are found by applying the usual procedure..<1 L ':. a (3.7.. 3. we obtain a4 l'V p = . mITX smh . . The total deflection expression (3..29) yields p. Slll mnx dx = Po x .31) where the constants Bm and em are to be determined from the conditions at y = ±hj2: w = 0. = mnbj2u. I~ ( B cosh In a 11l1r.\ only.D·· L m ... ~ m 11l1r... ( 1 _. the bending of a hydrostatically loaded plate (Fig. a". and assuming that tbe edges s = 0 and x = (I are simply supported...X 4 SIn ... following which Eq..17) as well as the simple support conditions (1.1. y = 0) is . Consider.2): Pm 2 = a'oa I·' .. ) (a) Equation (3.sm a m4 n. (1.2 . (3. The deflection occurring at the center (x = a/2. • Sill . we obtain in which. mny Pma'~).2 furnishes· the values of Pm for various types of load distributions (Proh.·(a) represents the deflection. 4 D a (3.. Tahle 3. 3. the loading is expressed by the Fourier series: p(x) = where Pm = . A ofTable 3..a 2po.26b) at x = 0 and x = u. Finally. Let us assume that the two arbitrary edges y = ±bj2 are also simply supported. as an example.5 LEVY'S METHOD APPLIED TO NONlJNIFORMLY LOADED RECTANGULAR PLATES Levy's approach is now applied to the treatment of bending problems of rectangular plates under nonuniform loading which is a function of. as before.30) a The above represents the deilection of a strip under load pix) and satisfies Eq._ mn r I +1 (Ill = 1. (3. m..Pm. Introduction of a given load distribution p(x) into Eq.4.32) together with Eq.
.. I . a result intuitively appreciated...:: I I x L The above result is onehalf the displacement of a simply supported rectangular plate under uniform load (Table 3... .. .t : ~ I h _ ~ 2po ( _ I)" .. ) r:o.. .RECTA~G[}LAR PLATES 75 Table 3. _... described by a Fourier sine serics (Fig.. 3.... XJ = 2pu sin I1mx. . " ". A ~PO . J. 2.:.8) f(x)= L m'" [ CC _ mn:x IvlmSIIl . _.'_J ....6 RECTANGULAR PLATES UNDER DISTRIBUTED EDGE MOMENTS Consider a simply supported rectangular plate subjected to symmetrically distributed edge moments at . mn a mire 1m ~ I.2.... 3.\' = ±b/2. .2 Variously loaded simply supported plates Geometry Type of loading and Expression for P'" Hydrostatic loading: p = Po·· a p. I I L _____ J (I ~i tJ' I~ Uniform load from (x J P'"  B e) to (x J + e) (/ = 4po Sln mn). ..:i I c Line load Po at x = P. 2•. ) r . a (a) .. 1 . 11111: (Ill = 1. a (I (Ill = 1.1 J. I . sm .. .. ) J 2. I I I I I L_ '.
(g) and (a) into Eq..21) with P" = 0. x = a) (e) (d) w=o (Y = V (.a 2 m a (l Substitution of Eqs.M m (b tanh ~ m cosh Cl:m 2 oc 0=../11 SIl1 . (e) leads to HI = I 00 C ( V ".. _ . 3. 2.+ emy.y smh mTtY) 111 a 2nD I!J sin (mTtx/a) I . nmx sinh . As before. (1. tanh rt. we have Bm cosh am + em ~ sinh from which In ft...:.l m Sin aa (J. 3.=1 III ... (f) 00 ( 111"'1 a a a This equation satisfies Eq. Sm mnx W = L Bm cosh ..a I1my (b) The boundary conditions are w=o (x = 0... = ± ~) + ~) . The condition (d) is fulfilled if the.y. = L J\. . Innx nmx I a em cosh 11.. . (3. setting Ct...= 2 Io f(x) sin a.S IN PLATES AND SHELLS lex) =: (ftlY)h12 Figure 3... S111 .76 STRESSJ.' 2nmD cosh am mTty. .. = 0 CXm b Bm = .. except that summation goes over In = 1. dx ' a· . s1I1h mny).) • m 11/= 1 c'" = The deflection is therefore II' = ....1 a a (333) .2.... cosh .8 Here Jvtm represents the unknown set of coefficients M"...17) and conditions (c) as already verified in Sec. (e) The solution next proceeds with an assumption of the deflection surface in the form of Eq. : nm)' .4. 2D It follows that 111""1 mn .. aMI1I ~'.C 2 + tan h Equation (f) now takes the form rnny b mnv) . cosh ~. = mTtb/2a... terms in parentheses in the above are set equal to zero..
each of which can be treated by the Navier's or the Levy'S approaches. 2. 3. ~ sm tn Inns =! ~~~J~ 8 D a interestingly. the general case can be obtained by combination of symmetric and antimetric situations..... first a given complex problem is replaced by several simpler situations.). we can set lanh am ~ C(m and cosh am ~ 1... The solution . (b For a square plate (a = b).3 . subjected to two equal and opposite moments at the ends..... (3. (b) gives Mm = 4Mo!mrr..10) ill the form 1\' = A1 (I (/2 0'(1368 "_..17) in the form of Eq. Consider.35) Cl m When a ~ b.. a (y = 0) (3.." 2/\1 0 " cr.16) and modifying the conditions (e) as follows: D .394M" M)' = 0.. Solutions for the symmetric [Eg. the bending of a plate under any lateral load..3:i) for H"In the case or uniformly distributed mOIT"lCnts. cosh .34) mrrv.'2 can be treated similarly by taking the solution of Eq. rrD 11/= I ~. 3.~1(Jh2 2nD f m=I... D JlIlx = 0. for example.  1 tanh am . and the antimetric moment distributions are useful in dealing with plates with various edge conditions (Sec.'~~' Slll _ mnx . The deflections obtained for each replacement plate are then superposed in such a way that the governing equation Irw = p/D and the boundary conditions arc fulfilled for the original case.1 = .34) and (1. (1. we have f(y) = j\1o_ and Eq. and the above reduces to W = . According to this procedure..8). (3.. (3.! momcnts and the stresscs are determined from expression (3.33)].. sin (m"x!a) . I...5)..ll'~b2 J'=b/2 Furthermore.j'2 cy il2W) ( = (M. J . ~ .256!1!Io The displacement occurring along the axis of symmetry is W Moab = ".7 METHOD OF SUPERPOSITION APPLIED TO BENDING OF RECTANGULAR PLATES The deflection and stress in a rectangular plate with any edge conditions and arbitrary loading can efficiently be determined by the method of superposition (Sec. deflection and bending moments at the center are determined by the use of Eqs.. J .33) then becomes \\.RECTI\KGLTLAn PLATES 77 Thl.v smh mn y ) (3.'TJj m·· I.. _I 'x m2 cosh ..7). 3. this result is the same as that for the eellter deflection of a strip of length b. Equation (3. m2 cosh am 2 a a . The particular case in which the plate is loaded by an antimerric moment distribution (M)')""hl2 = (M). with one edge clamped and the other edges simply supported (Fig..0_..tanh a".
Plate 2 also has all edges simply supported but in addition the two edges at y = ± b/2 are subjected to uni· formly distributed moments yet to be determined. ··)f:. a bending moment along edge y = 0 is applied of such a magnitude as to eliminate the rotations due to the lateral load... I X y (0) 'J' (iJ) I Iy Ci::]1 . from Eq. f: ."..) mnx a (b) wherein the coefficients M m are obtained so as to make the slope owing to these moments equal and opposite to that represented by Eq. 3.'... The . ..2 cosh am a b l. Then.9a).. Plate 1 has its edges simply supported and under uniform load Po.9b and 3. .. (3...9 begins with the assumption that all edges are simply supported.. the following bending moments are applied along side y = ±b/2 of plate 2: M.' s1I1h b . ...78 STRESSES IS PLATFS A~D SHELLS r .7 A rectangular plate has opposite edges at x = 0 and x = a simply supported.. we have WI = 4:~~4 _ ml...> .36) The rotation of the bent plate along side y = h/2 is then aWl oy = .  ~ . (a). mrrx tanh am(1 + "m tanh ~m)l sm _. The plate is subjected to a uniformly distributed load of intensity Po. n D 2p o a 3 L m= 1. (M y )hI2 (c) Figure 3. We wish to derive an expression for the deflection surface and the moments. [am m 1 .22). Example 3.. (:: . and the other two edges at y = ±b/2 clamped (Fig.y tan~_~~_+ b m 1 mnv. 2".3......... For plate I." u (a) In order to prevent this rotation and thus fulfill the actual conditions of the boundary of the initial plate. <0 3 .9('. 3. SOLUTION We shall proceed by superimposing the solutions of each of the two plates illustrated in Figs.. [1 _~ 2coshcx 2cosh ~~".= m""t L Mm sin·fS. The solution procedure for the following problem serves to further illustrate the method..<:.. l' + . . + rl~. S1l1"" nmx a (3...
nh am) m3 n 3 !X m .. (3. I oJ: I . :rrD m~'l.1'=0) 2' The maximum displacement._.\'=:1/.smh ~..[).= IVm " The values of bending moments for plate 2 is found by employing the usual procedure while those for plate 1 are listed in Table 3. .I) mn y .am tanh am cosh . To ascertain the bending moments for the original plate.. '" 111 1) ~ (XIII] . the results of the replacement cases are superimposed. we obtain M III = "f.~ a . mrry /11rrv) x (. (V=+h) . (3.a.v = 2' 0) The center deflection for plate 1 is (from Table 3. is thus w = w.oa 2 am . we find that the center dellection given by only the first term in the series is equal to ( X = ". It can be shown that the maximum moment occurs at the middle of the fixed edges and for a = h is given by My = 0.. For instance. (3..2.: .t. (X=" . we have .. which takes place at the center of a uniformly loaded square plate with two simply supported and two fixed edges.~ tanh tXm(a m ta~h r. when a = h.0697poa 2 = M m " . from which fiJI' sides . (a) and (e) into the above and solving.  Wl po a 4 = 0.1."Ja:m tanh am  1) (3.33). the deflection of plate 2 can be obtained by introducing Eq.:.. a (c) The requirement that the slopes for both plates at y value but be of opposite sign is satisfied if = ±h/2 have the same .00192.tanh amp_~a.'.37) into Eq..m .37) With the expression for Mm determined.33).' a a a (338) This series converges very rapidly and the first few terms will give a satisfactory result.1).. lvlm[tanh am(anr tanh am . .RECTANGL'LAR PLATES 79 detlection of plate 2 is given by Eq. tanh . 3. nlnx SIll .(1 + am tanh am) mS cosh". Upon inserting Eqs. Hence sin (mnx!a) "m ' tanh "".
4 and 3. the solutions lor other cases of practical importance may be found. 3. the plate is assumed to be divided into two systems of strips at right angles to one another.. In this socalled strip ". A discussion of a number of practical aspects of the edge fixity is found in Sec. __ .8 THE STRIP METHOD We now present a simple approximate approach..: x y (aj (b) Figure 3. the dellection and moment produced by the same loadinq decreases considerably.80 STRESSES Il'\ PLATES A:JD SHELLS D~nections and momcnts at any other points arl~ calculated in a likc manner. each strip regarded as functioning as a beam (Fig.ethod.3. due to H. 4 (d) AI" = M).. it will prove useful to introduce maximum deflection and moments of a beam with varions end conditions.10a). Grashof.2 2 b [21 i . that because this method always gives conservative values for both deflection and moment. Before proceeding to a description of the method.0513p oa2 = l'd on " (e) Upon comparison of the above with the results obtained in Examples 3. we observe that as the Humber ofbllilHn plate edges increases. \ . b['. = O. it is often employed in practice.4.. Note. _ L I a <I Q :: 1 ¥ : . for instance.. Expressions for such quantities for a beam of length L subjected to a uniform load p. 2. however. 3. are given in Table 3. Their numerical values are: = 000126 poa H'max· D . derived from the mechanics of materials. The method permits qllalitatir·e analysis of the plate behavior with ease but is less adequate. Employing similar procedures. A very efficient engineering approach 12 to design of the rectangular floor slabs is also based upon the strip method.10 . in obtaining accurate quantitative results.7. lor computing deflection and moment in a rectangular plate with arbitrary boundary conditions. the largest deflection also occnrs at the center and the largest bending moments are again fonnd at the middle of the fixed edges. . :. In the case of a clamped square plate. in general.
.. . Referring to the Eqs.39b) leads to b4 Po = Pu a4~:·b. ~ 1\' .3.lfw Wrn" = '1 . = 5p.24).. (3. (3.I'.. (3.4 +. b4 ''  M" =.. Values for these factors.. b) In the case of a plate with simply supported edges../ = k[ nl~~ = rti..40).)' and assume that the plate is divided into strips of spans (/ and b.....~ 1 ('U 192 £1 A1.4 Then Eq. 4 4 D 1 D(a + b ) The bending moments at the center are written {'J'\. Ala = kpi! MI>=O Consider a rectangular plate under a uniform load p. Pou b k2 "a4 + b 4 a2 [.41) p. and Eq.3 we again find that Pa a4 = 1'. Wb = 5p"b j384EI. _./we can represent the maximum deflection. we have: 4 4 4 W.RPCTA'\G[)LAR PLATFS 8] Tahle 3.. .. and Pb. Here k 1• k2' and k. occurring at the center of the plate. from Table 3.3 I\' = c 5 384 El flL~  ~.':.Z.42) and the bending moments at the edges are Mx = k 3 Pall 2 = P k'3. (3.4 2 4 (3. = k2Pua~ = . respectively. = .40) ~ Expressions (3. are constants dependent upon the side conditions of the plate. d...390) yields pa a = 1'. The deflection and moment at any other location in the plate can be expressed in a similar way. since from Table 3.+. h" a·~ 1)2 (14 (141) . referring to Table 3. carrying the uniform loads p..3. = w" 1'0 = Po + Pb (x = J' = 0) (3..a /384EI... .'b4 a4 (3. p a4 poa 4 b4 a_ = k . i i i I' ~ ! i. Pb = Pu ~.[.4.39a. b4 for Wa = lV b · The deflection and bending moments of th' c are obtained by means of Eqs..:~ M.pI. The loaded system of beams will he impossible to arrange in such a way as to compose the plate unless the following conditions are met: W..". (3... as follows \". are listed in Table 3.40) are also valid in the case of a clamped plate.
39a) and Table 3.. (d) in Sec. lOb). 3. b' .3 give /JIl Q4 ]92E1 From Eg. Eq.4/>4 w=_·_..4 that Wm " = 0.4 Edge (:ondition Simply supported Clamped " 5/384 1/384 "2 1/8 l/24 " 0 1/12 For example.7.' 42.. Example 3. we find from Eq..e of a uniformly loaded clamped sqllare plate (0 = b)..: PLATES AND SHELLS Table 3.384 D 19" (2a 4 + 5b 4 )D We also obtain the following maximum bending moments l\{. 3. simply supported and one edge built in is subjected to a uniform load Po (Fig.. The results for moments are crude approximations. (3.OOI3~D'· poa 4 (x = 0. in the ea. however.8 A rectangular plate with three edge.4 + 5/>4 0) and the bending moment at the fixed edge 5 poa'b" At r =8 2a4'+Sb4 (x = a) (e) . SOLUTION The plate is divided into simply supported and clamped simply supported beams. po a'b 4 (a) =128 2a4+ 5b 4 45 (x (y = ~L) (h) = 1 Po (l 4 M = .. y = 0) This is accurate to the fourth significant figure as compared with the value given by Eq. (3.41) and Table 3.. (3.82 STRESSES 1.39b) we then have 5h 4 p" = Po 2a4 +"5/>4 The plate deflection at the center is therefore 5 pb b 4 5 Po .= _. Determine the deflection at the center and the maximum bending moment of the plate. Thus.
Hence.\1 r is 11 percent greater than the values obtained by the bending theory of plates (Example 3. Intermediate supports are provided in the form of beams or columns.4).03571'0 a' lv! { = 0.1. A convenient way of looking at the problem is to draw a free body diagram of each rectangular panel as shown in Fig. Only a continuous plate with a rigid intermediate beam i.c. Consider the twospan simply supported continuous plate.l'L .OOJ72Pn((J.OS02po . moment . the intermediate beam represents a simple support to the plate. We shall assume that the beam does not prevent rotation of the plate. treated in this section.l In the' case 01 a sqllare plate (0 \I' = h).J I ..".~~===i ~ b ' I 2 c L.. it is termed continllous. In a continuo liS plate the several spans may be of varying length. i. 3.111>.~==~~= ! (a) (b) Figure .2 My = 0. The distributed moment along the common edge may be represented by a Fourier series (0) b I t ' 'Tr.IHOCTA:':C.08931'o a' The dellect ion at the center of the plate is 33 percent greater and the hend ing. the plate has zero delleetion along the axis of the supporting beam.1la)...9 SIMPLY SUPPORTED CONTINOOUS RECTANGULAR PLATES When a uniform plate extends over a support and has more than one span along its length or width. 3.\R PLATES Po.. half of which is subjected to a uniform load of intensity Po (Fig. 3. (a) to (e) red lice to = O..D IV! x = 0. Eqs.
for plate 2.'2 ".3. This continuity requirement is expressed ... .3. =0 2 wl 2 l 8 OX =0 w m~1.84 STRESSES 11" PLATES AND SI'II:LLS \Vhcn the set of coefficients A1m is determined. L Mm sin }'mY . 3. L /1.1 m sin Am Y (X2 = 0) We thus have eight equations (e) and (d) containing nine unknowns Am ..25) if y " replaced by x" x by y.. and W2 =0 D·. (Xl = 0) (e) w. the expressions ror thc boundary conditions can be handled with ease.5 and 3. The required additional equation is obtained by expressing the condi tion that the slopes must be the same for each panel at the middle support. The lateral deflection of each simply supported replacement plate may be obtained by application of Levy's method.. (x. Referring to Figs. respectively: w.11" we conclude from the symmetry in deflections that the general solution for plate I is given by Eq. . m= 1." Mm. That is where (/» Similarly.= a2 }VI ilx. setting Po = 0. = a) =0 = 0 ·=0 ax~ 82 w2 (X2 = a) (d) IV2 D= ax~ a2 H. 3 .. and a by />. . 00 [Em sinh A. (3. H.. and for a different set of constants is W2 = L: m=t. the deflection expressed in terms of the coordinates X2 and y..mX2 + Fill cosh AmX2 The boundary conditions for plate I and plate 2 are represented as follows..
.47)  .. the moment coefficients M". coth III A1m 2)'m D I ..'!'TI:S 85 Introducing Eqs.. In so doing. from Eq.45) leads to values for the constants as follows: 1 coth ... 4 = ..47) and (3.coth 2 ) ~ " In 2)'IJI D I'm = 0 (3.44) and (3.44) and (3." '1' f) I sll1h I'ma .RECTANGl'LAR PL. The foregoing approach may be extended to include the case of long rectangular plates with many supports...m ~. (3.45) then give the dellection of the continuous plate from which moments and stresses can also be computed..'~ (1 ".45) into the above we have Application of the edge conditions (e) and (d) to Eqs. Equations (3.48) H = ..48) available we obtain.. 'J ' m and Mm a E = ._.' It is noted that there are situations where the intermediate beams arc relatively flexible compared with the flexibility of the plate.. + 21'0 (1 + cosh I·ma) X41... but are functions of the hending and torsional stitTnesses of the supporting beams'" The design methods used in connect ion with continuous plates ut ilize the solutions derived in the foregoing sections and a number of approximations 1i ..44) and (3.Ia ./11 (3. "III a Having Eqs. csch Ama [Mm . 1. subjected to loading which is symmetric in y. (3. (3.. }'m" ItV!m m . _D 2}..46). The deflections and rotations of the plate arc not then taken as zero along the supports. an equation similar to that of the threemoment equation of continuous beams is obtained. (3.
+ Em _ .2c • . Bm cosh··_··. is the same. To attain a simplified expression for the lateral deflection it is assumed that: the plate is suhjected to a uniform load Po. • • • • • lal (h) Figure 3.4.J '. A solution can be obtained utilizing Levy's approach (Sec. 3. away !i'om the boundary of the plate.4).\: a (3.AR PLATES SUPPORTED BY INTER MEDIA TE COLUMNS In this section we consider the bending of a thin continuous plate over many columns.49) • • I> • .12 . The origin of coordinates is placecl at the center of a panel shown by the shaded area in Fig. We can therelore restrict our attention to the bending of one panel alone. the deflection may be expressed as a combination of that associated with a strip with uniform load and fixed ends y = ±b/2 and that associated with a rectangular plate. the column cross sections are so small that their reactions on the plate are regarded as point loads. 3. IV = wp + w" = 3~4D + P b* ( 1w 4V2)2 bi ( + Bo 1nny mny " L: 111""'2.10 RECTANGUl. sinh a a mn y ) a COS . and the dimensions of the plate are large as compared with the column spacing. mn. Accordingly. the columns arc equally spaced in mutually perpendicular directions. . Clearly.12a.86 STRESSES IN PLATES A~D SHELLS 3.. the maximum dellection occurs at the center of the panel.... and consider it as a unifomlly loaded rectangular plate (a x b) supported at the corners by the columns. That is.. 1. The loregoing set of assumptions enables one to assume that the bending in all panels.
(3049) into Eqs. cos _. The aboH" satisfies the boundary conditions for the rectangular panel along the x edges. referring to Eq.. 2 em 4 =  0'0' I .:Gl"L. (1..\R PI. From the conditions of symmetry of the bent panel we are led to conclude that the slope in the direction of the normal to the boundary and the vertical shear force vanish everywhere on the edges of the panel with the exception of the corner points.'...RECTA!'. .2 . Thus._.y = O. Inasmuch as (lW. can be represented by (e) Expression (c) and conditions (b).e and x = £1/2 + c or 2c (Fig. and Em are constants of WI!' yet to he determined. the second tern) in the parentheses above is zero. For purposes of simplifying the analysis. (e) we then determine the constants Em and Em.ATf'S 87 where Bu . Upon substituting Eq. Qy=O (o<x<i c) (b) I .12 0 (d) We are now in a position to represent the boundary conditions for the rectangular panel along the y edges: Qy= (e) +~) (J'=.12.= 0 thv ax (a) and Eqs.18).dx= 4 '". between x = a(2 .. the support forces are regarded as acting over short (infinitesimal) line segments. 3. Bm.pob( _1)".)....17) and (1.2( The shear force Q.":2 poah Q. lead to Co Poh = _. (AA). ..af2 mnx Q. The plate loading is transmitted to the columns by the vertical shear forces.dx = .
+ tanh (Xm sinh tanh tim where ~m = mnb/2a.0092 0.0131 0.n I .00358 0._. U) It foliows that the values of the constants are _ .49) together with Eqs..~ m Cl.10) and setting x = y = 0 one also determines the largest bending moments..4.0363 1. ... ~m + ~''')' (3.5 lists values" of the constants 0" O . The assumption made above.~.2 0. inay thus be dispensed with. The resulting expressions may be put into the following form (g) Table 3.0387 2.88 STRESSES J:\ PLATES AND SHr:LLS Finally..50) into Eqs.4 0.51) By inserting w from Eg.0)84 ...0411 S2 J . The moments near the columns as calculated by means of Eq.'3 ( tim Cf.!. pertaining to support force distribution in an infinitesimal length 2e.0331 O. An equivalent circular plate separated Table 3. _. Equation (3. (1. (3. .50) represent the deflection surface o[ the paneL The maximum deflection takes place at the centcr o[ the plate (x = y = 0) and is W =~ /J o h4./Il tanh ~ tani~T~~. we lind Bo such that the deflection H' is zero at the corners 11'=0 (x=i·J'=~) .(lO33? 0...00423 0..00292 0. .49) together with Eqs.50) p"o~ (_i)""( 2D . (3.5 0..O. representing the domain around the ends of the coiulllns..)3 "..0210 0. fnoJbD4 _ ('/1..5 hill () J 'A\ 1 l. It can now be assumed that each support reaction is distributed over a small concentric circular arca of the radius c....Bl 1._)J m . 0. and OJ for different values of the 2 aspect ratio b/a.. __ !"...49) are much larger than the moments some distance away.. (3... (3.OOSXI 0. max 384D (3...0 O._ .0149 1.0 0.
54) where Pm . (3./b) yields the equation of the plate deflection of the form (() if..kw. ..17) p  kw D (J. may thus easily be determined as illustrated in Sec. is shown in Fig. 3.. Hence. (1..la).53 ) This value of am" and Eq..126.. PmIJ .54) reduces to the solution for the rectangular simply supported plates [Eq.. ..1h) we substitute Egs. .. 2. sh('ar forces Qr around t11(' periphery. (3. (3. and rcactional force !'(lah al the inner fixed houndary.:.. . the momenl and stress can thus bc obtained at the supports (sec Sec.220 (Ir ) Referring to Fig. 3..\TES 89 from the panel and subj(:ctcd to uniform load Po.2. The problem is treated by Navier's approach described in Sec. W= »1""111""1 L L . 3. nnv ""D[(m/a)' + (n/h)']' + k ' a' 6 (3. the intensity of the foundation reaction q at any point of the plate is assumed proportional to the deflection II' at that point. 'GULAR l'L.3..1) into (3. 3.52) and the given boundary conditions... To evaluate am" in Eq. Consider the bending of a rectangular plate of sides a and b subjected to arbitrary loading p(x.12".52). sIn .8). (3. . 2. 3.. The plate is resting on an elastic subgrade and simply supported along its edges.. under any specific load.1). Inns .. As in Sec.. Hence.9. = n4Dnl~!a)' +'(~!b)'l' tk (3.HH'T . 3. Eq. The delleetion II' of a rectangular plate subjectcd to load p per unit suriilce area and reaction q mllst satisfy Eq. q = .v) (Fig. as in cases of nearly square panels. where k is the modulus of the foundation. (3...... The radius of this plate and the shear force per unlllcngth.]] RECTANGULAR PLATES ON ELASTIC FOUNDATION This section is concerned wilh the bending of the reclangular plalcs on clastic foundation. is given by Eq.Sin .5)].. It follows that Pmll a"". It is seen that for k = 0. Expressions for detkction and stress of a rectangular plate resting on an elastic foundation and at the same time simply supported at its edges. the load and the deflection are represented by Egs. (3. arc given byH c = 0..3). (3...
. y = b) Integration by parts of the last term in Eq. (1. The boundary conditions are w=o (x=O.13). The potential energy I1 therefore U .. W is  02W ( . (3. carrying a unifoml load of intensity Po (Fig. oX oy ox' . One can represent the work done by the lateral surf"ce loading p(x. x= a) W=O ow oy (a) = 0 (y = 0.34).90 STRESSES IN PLATES AND SHELLS 3. 1.'ox Jy )2]  I wPldx dy (3. 3.56) leads to z z 'r ··_·dxdy= 1·dx·II·dxdY 3 w cw ' 3 w 3w ...9. ..13 . The strain energy U associated with the bendiug of a plate is given by Eq.55) = where A is the area of the plate surface. y) as follows W=((ll'pdsdy A (3. Sec. 2 'S A A  a " b x 0 y Figure 3.56) The application of the method is illustrated by considering the bending of a clamped rectangular plate of sides a and b. Bw i)3w I ax <Jy ox oy ..12 THE lUTZ METHOD APPLIED TO RENDING OF RECTANGULAR PLATES We now apply the Ritz method to the bending of rectangular plates.(Ix ox av> .
57) A Assuming a detlection expression of the form W = L LG mn ro 00 ( m=II1=1 2mnx 21ln l' 1 .cos .· 2I1ny )] dx dy (3. Ih. W = Po. II (3.'2W iii:' 8~ay )'] dx dy = 8 w ("'W .' (02U' U = . [.57) yields U =. 2l1ny ) ) cos ~ n' + b2 from which COS 2nn\' b'. · .. (3.' ) a . by application of Eq. Introduction 01 the above into Eq. 1 cos '1"'. l (3.o' (l2Hl)' dx d.59) which is valid for I' I s. The work done by Po is. Hence 2 . (a). 2mnx)( 1 ' COSt.S5) .../1 ..2 c1x 2 0)'..RECTANGULAR PLATES 91 According to E'ls._. · .. [m' cos _.=1 L r3 (m)" + 3 (")4 + 2 (1Il)'(I1)'J a~n .60) 11I1nl =poab L L m= 1 n"" 1 00 From the minimizing conditions an/Gamn = 0..b J o r"' l:: o ro ro ~ am" ( 1 .( 1  2mnX)] I' I dx dl' U = 2n"abD 1 L m Im=1 .. ...l' (3.l._.b a a ".Ij ax' The bending strain energy therefore reduces to 0 D .58 ) the boundary conditions (a) are satisfied. it follows that which is valid for r I 11 and r I m.cos .. (h) become identically zero.  oc ·0 '0 moo· 1 11""" 2nmx L 4"a". I\ L I D . (3..cos .( 1 2 a a "' .) ( 1 .' +~. the Jirst two integrals in Eq..
3."D Po 162 (i. 021. in general. a". "11 = ['oa"/32. (e) gives a 11 o = " p a4 1 411A D 3 + 3(a/b)4 + 2(a/b) ··7 In the case of a square plate (a = b).00189p. "13 + 162a + 81 3 + 3 b + 2 b . taking only one term of the series.4 (a)'] P a ( b 4... Now Eq. = 0.4D.. 31 aD Simultaneous solution of the above.58) This is approximately 1..000201'1 .5 percent greater than the value. It should be noted that the result obtained. for a square plate (a = h) yields "11 = a22 0. based upon a more elaborate approach.4D 4 0 a)4 poa" h h h. Eg. given by Eg. a13. result in such accuracy when applying the Ritz method. (3. a". The maximum deflection takes place at the center of the plate and is obtained by substituting this value of {/ll into Eq. is remarkably accurate. (h (/11+ 2 (a)4 a + [243 + 3 (11)4 + 18 (11)2] a" + 1621133 = 4n4D a)4 [(a )4 (a )2] = 4. (c) results in 2a 11 + 2a" + 3 + 243 b 2 21 [ a)4 + 18· a31 + 162a33 = .7.... and ii".011841'.92 STRESSES IX PLATI:S A~IJ SHELLS Dropping all but the first term all. (d) of Sec.~.00268P1 a" = 0. So few terms will not. Let us express the deflection w by retaining seven parameters all' liu. a31 = 0.117741'1 a 1 2 = a21 "'3 = = 0.
. Example 3. the work done by the loading is W= P I I 00 oc.003011'. and supports a load P acting at a location x = x" y = y. (3. This is 3 percent smaller than the value determined from the solution of Eq. (J m=l II""] b (d) The condition (m/iJam . ). for simplicity. (d) of Sec. + 2 (111)'(17)'] amI! + r~l 2 (111)"Clmr ~~.4.cos .0027SP.. . 4n: abD! 3 ~4 = 0 together with Eqs. .Rl:TTA~GLTL\R Pl ATES 93 in which [11 = I}f)(/~ .17).. By retaining morc parameters in the series. 2m"x 1 am" ( 1 . Determine the maximum deflection of the plate. + I2 (17 )" a"'l. (e) valid for I' 1n and 1'1 111. The maximum deflection occurring at the center of the plate is.....000151' I in which P. and "".. from Eq. found to be Pa 2 wm a x · =000543 .) ( 1 .. 2nny.1'.13).. all' a".. the maximum deflection is found to be This result is exactly the same value given by Eq.h a b '" I ( 2111"X. Upon substituting these valLles into Eq. = 0. .cos 2mry. (3. we expect to improve the result. 3.._ . (1. Consider.006011'. _ . )(' Icos .59) and (d) now leads to I[ (In)" + r==I 3 (11)4 b 00 ... (3. (3... the case of a square plate (a = 1» with P acting at its center and retain seven parameters. O.·_.. 3.l2662P . a 12 = G21 = 0.55). = Pa'/D. (Figs. an a13 = a31 = a33 = 0.9 A rectangular portion (a x b) of a machine room floor has all its edges built in. a'2' all.7. SOLUTION Applying Eq.3 and 3.58).58)..1 cos _. a'3' aJ!. We then obtain the following values all = O.·D .. ) =0 b. ADT(+.
3. and a = 2b: (a) the maximum deflection.. If ( = a14.'1 Given a rectangular plate simply supported along its edges and subjected 10 a uniform load of intensity Po. and Y I = h/2.2. ~.14) and (3.6 A simply supported plate is subjected to loads PI and P2 at points (x = a/2.cntrated load P at its rl'lltcr (Fig.). the third edge free. 3.2). y = 0/4) and (x = 0.B Determine the equation of the elastic surface for a plate with h .12 Verify the results for Pm given in Figs. \' = 0. (3.2.·k is subjo:cled to a uniform J1rcs:. J = /)/4. = 0.n "'" Km('~"''' where K". A~sume that all othcr conditions arc the samc. simply supported on all edges.14) 3J5 Find the equation of the elastic surface for the plate loaded as shown in fig.15) by assuming a solution of tile form'/. \' the plate. A of Table 3.that would be produced when flo reaches its limiting value .1 A strucfural steel door 1. (b) the maximum dellection 11.7 A square steel plate 10 mm thick is built ill at the boltom of a ship uwwing 7 m~lcrs of water. respectively (Fig. 3. m are constants. The structure is filled with water lip to the upper edge level at . (h) the maximum stress in 3. 3. Determine the reactions for a simply supported rectangular plate with a com. •. Sees. determine the value of maximum deflection if b = 2a. 3. and the fourth edge clamped (Fig.!lhod and superposition.7 J. distributed over the subregion shown in Fig. 3.2.10 Verify the result fOf w given by Eq. 0.11 Redo Example 3. 3. find the maximum deflection and maximum stress in the plate. for a = b13. 3. Ohtain the center deflection It' by applying Navier's m. 3.. 3. Determine.U Determine the expression for shear force Q. for the plate descrihed in Prob..6).o opposite edges simply supported..\ J = e = a/2.5 Vt~rify the result given by Eq. Band C of Table 3. Xl = a/2.'. Determine. Take III = 11 = l.$ STRESSES I!'\ PLATFS A:\"IJ SHELLS PHOULEMS ~~t'c~.16 Determine lhc equation of the elastic surface for the plate k)aded as shown in Fig./2.9 Verify the rcsults givcn by Eqs. and m = 1 to compare the result for the maximum ddlcctio!l with that lis{t'd in Table 3.4 A concentrated load P is applied at the center of a simply supported rectangular plate (Fig.3). (i) of Example 3. The material properties are E = 200 GPa. are: P d~ w = O.4. using the Navier's approach and retaining only the first four term~ in the series solution: ((1) the limiting valu~ of Ilt) lhat can be applied to the plate without causing permanent deformation.3 and allowable yield strength (irf' = 240 MPa. Take the first two terms in the series solution.7. by rewining the first two knns of the series solution.14 A water level control structure consjsts of a vertically positioned simply supported plate.3 :1. C of Table 3.1.. Compare the result with those gil'en in Table 3. 3.. 33 A rectangular flat portion of a wind luntlel.3. and A./4). is under a uniform pressure Po.~)'!h)]ld. if 111= n = 1.' " .3..1 to 3. :.5 m long. B of Table 3. a = b. Y = 3(/. 3. 3. v~o) (P3. Let .4 to 3.\: = 0 (Fig. . Retain only the first two terms in the series solution and take {I = h. 3.1.9 m wide.10).1. the values for the deflection wand the hending moment M" at the center. 3. .6 if the edge y "'" 0 of the plate is built in. 3. 3..OO614··~ D ( x~~.\ by evaluating the limits of [sin (mlre/a)]lc and lsin (.10.urc Po. (j) of Sec. and 15 mm thil. The plate j~ regarded as simply supported. Show that by taking only· the first tefm of the series solution. :'L!. if a=/Jand x}=aJ2. 3.
if (/ =.<.2. 3. 3.29 Redo Prob. 3. 3. Let E = 20{) GPa.26 Derive the equation of the elastic surface of a simply supported rectangular plate on an clastic foundation (Fig.mel.sin . What should be the value of k in order to limit the maximum deflection to 0. The plate is fabricated of a material of tensile yield strength (Typ. b = 1. Take only the first term of the series solution.18 to 3.120).11. Compare the re. the side x = 0 clamped. = 0.:m).20 Employ the strip method to obtain the approximate center deflection and center bending moment for the simply supported rectangular plate under hydrostatic loading as shown ill Fig.8 In. 3. Assllme a solution of the form where C is an undetermined coefficient.12a).. selecting a factor of safety N. 3.11.9.22 Derive an expression for the moment f(y) distributed along the midsupport of the continuou~ plate shown in Fig. i)et<.~l ~~I " ~ L Lam" (' 1  cos '. 3.25 Determine the required thickness of a plate platform under uniform loading Po supported by columns at a and b = 2a distance apart (Fig.1.10. The coordinates are placed as shown in J7ig. 3.30 A uniformly loaded rectangular plate has two opposite sides built in and the other two sides simply supported. a b 211111:\) IIlt\. Take b = 40 and ~.RECL<\0. 3. =. 3. Referring to the equations derived in Sec. m~x at the support locations of a uniformly loaded large plate carried by equally spaced columns (a = b) (Fig. Design the platform.26 by employing the Ritz method."ults with those obtained in Example 3.3) if the plate experiences a central point load P. with two opposite sides y = ±bi2 fixed and the sides x = 0 and x = a simply supported.7. lOu MPa.. 3.2. for 111 = 2.5.21 Simply supported silo Hooring is loaded as shown in Fig. 3. Evaluate the deflection at the middle of the free edge.006 Jll. Obtain the maximum deflection and maximum bending moment M. for (/ = b.]2a).18 A rectangular plate. Take E = 200 CiP. determine the deflection at the center of plate I and plate 2. Assume a solution of the form W = ".19 Employ the strip method to determine the center deflection and the maximum hending moments for <t plate with two oppo.31 A llIliformly loaded rectangular plate has its edges y = 0 and y = h simply supported.26 by retaining in the series solution the first term only. is subjected to a hydrostatic loading as shown in Fig.12 3.11. A of Table 3.23 A uniformly loaded airport platform is supported by equally spaced columns (a = b) (Fig. 3. and Po = 20 MPa.! = 3 mm.. By employing the method of superposition derive an expression for the reactional moments M . 3. Retain only the first term of the series solution. according to: (0) maximum principal stress theory.28 A uniformly loaded rectangular steel plate rests freely 011 an elastic foundation of modulus k. Take a = b. Compare the results with those given in Table 3.3. along the clamped edges )' = :th/2. 3. a = 0.4 m. 3. . and the side x = a free. A of Table ..Ol'LAR PLATES 95 Tilt' allowable strt'SS for the pl<lte i:.. Sees.9a).1a. 3.b and I' = OJ. Use the equilibrium approach of Sec. 3.24 Determine the value of maximum stress (J x. Apply the Ritz method to derive an expression for the deflection surface. 3. (b) maximum principal strain theory. The coordinates are placed as shown in Fig. 3.~ite edges simply supponcd and the two edgfs cJampt. 3.J. 3. 3. 3.r"~x in each p.. and determine the coefficients Q mM employing the Ritz method.'ti (Fig.::rminc tilt' maximum platt' dimension and tlw maximum deflection. 3.27 Obtain the maximum stress fJ x in the plate loaded as described in Prob.054 "K. Specilic wClght of salt watcr '/ =:.la. 3.
CHAPTER FOUR PLATES OF VARIOUS GEOMETRICAL FORMS 4. 96 .5. but in some situations other geometrical forms are used. 4. Determination ofthis disturbed stress distribution is of considerable practical importance and is discussed briefly in Sec.2 METHOD OF IMAGES Certain problems of plates can be treated by arbitrary extension of the plate and/or introducing fictitious forces to produce the deflection forms sougbt. the bending of an isosceles right triangular plate with simply supported edges under a concentrated load P acting at arbitrary point A(x l .6) has proved very valuable. In developing approximate formulas for the deflections of polygonal plates. \.1 INTRODUCTION Structural members designed to resist lateral loading are generally circular or rectangular in shape. YI)' The plate is bounded as shown by solid lines OBC in Fig. 4. From the examples worked out so far. it becomes evident that a rigorous solution of the deflection for a plate with more complicated sbape is likely to be very difficult. Consider. a perturbation in stress takes place. the membrane analogy (Sec. When the uniformity of the crosssectional area of the plate is interrupted as in the case of a shipdeck or airplane fuselage with holes or windows.1. y) which satisfies V 4 w = p/D and the specified conditions at the boundaries. the bending problem of plates of any shape is solved if we derive the displacement function w(x. 4. as an example. We now deal with the problcm of plate boundaries of geometrical form ditTerent from those previously treated. As previously discussed.
.S i n (m' + Jll)..x.yIl for x" and (0 . Hence..3) together with the principle of superposition.. is Upon substitution of P for P. . from Eq. often referred to as the method troduced by Nadai.1) and (4. we obtain the deflection due to the force .P at points A and Ai[(a . The il1loges. (a . mnx ..PLATES OF VARJOl:S GEOMETRICAL FORMS 97 We begin the solution by utilizing Eq. the deflection may be determined of an isosceles right triangular plate for any kind of loading. For this purpose it is assumed that the triangular plate is one half of a simply supported square plate subjected to the forces P and . the square plate deflects in such a way that diagonal Be is a lIodal lille or fictitious support. sin (m1[... (3. (a .1'.. P at Ai: ~./o) sin (I11[Yl!a) .. a b (4.3) where w. (4. in Eq. . The deflection owing to the force P. By applying Eq. Point Ai is the mirror or image point of A with respect to diagonal BC Owing to the loading described.13 or .2) respectively.11). was inmethod described above. and "" are given by Eqs. respectively (Fig.2) The deflection surface of the triangular plate is then (4.. (4. 1l1ry .1). sm ... (3.)'.3. 4.1).).) for .'Ill. 3. (4. the deflection of DBC of the square plate is identical with that of a simply supported triangular plate DBC.11) derived in Sec.
4. With the expression for the deflection of the plate determined. shown in Fig. (4. that 11' 10'0. dl'. (1.../3 (a) x 2a '"Yr=O . we find. (4.4) This series converges very rapidly.)3 Therefore.10) and (1. the function which vanishes at the boundary (b) Figure 4.2 .  J) llll=1.= P. one can obtain the moments and the maximum stresses in the plate from Eqs..9B STRESSES IN PCATFS i\~[) SHELLS In t he particular case of a triangular plate under a uni!ormly distributed load of intensity p. 4. The equations of the boundaries are: x+=O 3 7~+l'"''~=0 " on BC on AC on AB x v.3 2a 3.3 EQlJILATERAL TRIANGULAR PLATE WITH SIMPLY SUPPORTED EDGES We now treat the case of a simply supported equilateral triangular plate ABC..fi 3.12).. from Eq...dx.4 r rr 6 f f 11"'2. after integration over the area of the plate.4.2..3).? .
<'.a(x' + y2) +.6) The largest moment takes place at the corners and acts on the planes bisecting the angles. = 1/' .na'] 4aD (4. [x'· 3xy2 .'(2  ).10M 0 /t 2 Equilateral triangular plate under uniform load Po In this case an expression for deflection of the form IV = 7&'.) ".:Mo 1 + I' + (1 .0 [x 3  3'.v )3X] . and the surface of a uniformly loaded membrane.3.Aa3](~a' _.TRICI\L fOH\IS 99 is the general expression for deflection of an equilateral triangular plate with simply supported edges.10) as follows: 1 ( I' 3x Mx = 2Mo [ 1 + vI .i\!". It follows that 3M" k=4aD The deflection is then IV = .5) and the center displacement is \110 =  A1oa 2 27D (x = . mox = (Y=ia) (d) 8. (l.35M0 The corresponding stress tT r . uniformly stretched over the same triangular boundary. To arrive at an analytical solution.] J M".22h) and set M = Mo. it is given by M )'.PLATES Of VARIOl'S GEO~IF. (4. are identical. Here k is determined lor two specific cases of loading as follows.7) .') (4.5) and (1. we introduce Eg (b) into Eg. mOX = 1. For \' = 0.I' = 0) (c) The expressions for the moments are determined by means of Egs.I')MoY 20 (4.1"  a(x' + 1") +. Equilateral triangular plate under uniform moment M 0 along its boundary It can be verified experimentally that the deflection surtilce of a plate loaded by uni· form moment along its boundary.a 3(1 .
These conditions are satisfied if we assume for deflection.100 STRESSES 1'" PLATES AND SHELLS satisfies Eq. 4.. Substituting Eq..._ . 4. 4. By employing Eqs. For example. we have a4 b4 k . (4....8) The largest moment takes place on the planes bisecting the angles of the triangle.. for .f Figure 4. y = 0) The maximum stress is given by "".4 ELLIPTICAL PLATES In this section. ... (1.02481'0(1' 2 (x = 0. subjected to a uniform load of intensity Po (Fig. m " = 0..129a. Therefore Eq.062a.155p o a'/t'.. _..2). 4 .8D 3a + 2a 2 h 2 + 3b 4 (e) x .02591'0a 0..Po .. It can be shown that the center moments are (4..=0 ilw (In IV = y' (lor a' + b' x .I7) and setting p = Po.. = OJ..3 .. (!.10) the bending moments may then be obtained.7) is the solution.. Two situations are treated below. the expression: k( 1 X2 a2  b2 y2)' (b) in which k is a constant. (b) into Eq.17) and the conditions of the simplY supported edges. consideration is given the bending of an elliptic plate with semiaxes a and b..... _ . (1..3)..m" = 0. m" = 0. Uniformly loaded elliptic plate with clamped edge The appropriate boundary conditions are \V=O . y = 0) (x = (e) At Y... =I) (a) where n is the normal to the plate boundary. at points along the x axis (Fig. we have Nl x .
..)' = lvl = .m" 8JJ[(3/h4) + (2/aW) + {3i( 4 )J (4. At the center (x = 0..12) (x = 0. ..10) By substituting Eg..(. we have My 8vDk lvl = .2 y b ±b) where k is given in Eg.10)..v) . ) a 1 lvl .. the moments are lvl x = 4Dk ( a2 I\' ) + b2 .4)..PLATFS OF V ..(I . . RfO{:S GrO:. For a = b the solution derived in this section reduces to the results obtained for a circular plate with clampcd edge (Sec. . 8Dk a2h 2 X)' . (d) \' +..IHRICAL FORMS 101 The required solution is then (4.. .y=O) (4.. (4.12) that the maximum moment occurs at the eXlremity t?f the minor axis. y = 0). (1. . . Egs.9) and the maximum deflection occurs at the center of the plate and is given by H' Po = .Y = 4Dk ( ... . the bending and twisting moments are obtained lvl x>.. (b) into Eg. (e). .. . (4. In the extreme case. x b2 ~ 8vDk . a b b. h At the ends of the minor and major axes.. = ...10) and the second of (d) appear as W = mox Po(2b )4 384D (e) Equations (e) correspond to the deflection and moment at the center of a fixedend strip of span 2b under uniformly distributed load.2 SDk (x= ±a.. 2. It is observed from Egs...
The technical literature contains extensive information on stress concentration factors presented in the form of graphs and tables 7 • 14 It is convenient to examine the stress concentration at the edge of a small hole in a very large plate (Example 4.70 0. expressions for deflection and hending moments may be ohtained.321 2.379 1.88 O. In some instances. (4.1 (//h W .lSg 0.· (l1J"h') 0. it is more usual to rely upon experimental methods.0 3. of the actual maximum stress to the nominal stress in the minimum section is defined as the stress concentration factor.3. = 0.26 0.02 0. Hence.1 A large. some final numerical results for the deflection and bending moments at the center of the plate. The solutions more often involve considerable difficulty." 4. reasonahly accurate results can also be expected. If the cross section changes gradually.13) where 0'" is obtained by applying the familiar flexure formula for beams. small as compared with the overall dimensions. Table 4.102 STI~ESSES IX PLATES A.0 4. where abrupt variation in the cross section takes place.206 1. On the other hand. the stresses in these regions can be analyzed by applying special theories which take into account the shear deformation. for \. as before. is subjected to ..465 Uniformly loaded elliptic plate with simply supported edge The boundary conditions are now given by \1.5 STRESS CONCENTRATION AROUND HOLES IN A PLATE The bending theory of plates applies only to plates of constant crosssectional area.222 0.\1D SHELL".0 1. 5.. (1/p.206 0.184 0.7) is also very efficient for this purpose. notches. Table 4. After routine hut somewhat more lengthy calculations than in the case of a fixed plate. thin plate containing a circular hole of radius a and width 2a. The results obtained for this case have been proven applicable to plates of any shape.433 2.1 provides.210 0.1).2 = 1 ) where. However. The finite element approach (Sec.5 1.58 0.. ordinary bending theory cannot predict the high values of stress which actually exist. Example 4. The condition referred to occurs in plates with holes.0 A(~ . 11 is a normal to the plate boundary. The ratio k. or fillets. 1.1 = 0 (for a' + X' v' i. b2 ) (Et 3 /P obl) M.
.8 1. .+=) MO 2. In the absence of any hole. (22). Determine the stress distribution around the hole.4 pure bending ililx = Mo and My = 0 (Fig.4). (a).C"' Q /' ) cos 2() J (e) .. and (2... L0. one has. r I! {JmJ~ x 8 \ \ . '::  /kb . are: M" = I" . o .. ... obtained by means of Eqs.4).. Guided by the expression for 11'" we assume the additional deflection of the form'! 2 2 11'.6 \ \ \ \ .."... SOLUTION We place the origin of coordinates at the center of the hole.2 = . ~ 2...PLATES OF VARJO[}S m.\' + (1 + v) Gas 20] (a) At r = a the radial moment and effective transverse force owing to the M 0.Ie..O\lETRIC'AL FOR~fS 103 J '~r .. the superimposed stresses ollly are zero at /' = ro..\ ~:~. 2a 3 4 5 6 7 Figure 4..' )/ ? Z! . .30) \V! M . . __ ?. 1M 0(1 + cos 20) I (b) cos 20 = ···M o a When a hole of radius a is drilled through the plate. '} ._..0 1.. 4. (1.. from polar form of Eq...~o~ [ A In /' M a 21) + ( B .) [1 4D(1 I' ..7 .=J! .....~ ' \': ... Mo0 . we superimpose on the original state of stress an additional state of stress so that: the combined radial moments and forces equal to zero at /' = a. 2 .  = ..
the dashed line represents a plot of the true stress concentration factor k" obtained from Reisner's theory which takes the shear deformation into account. mal( ·········M 0 3 + v 5 + 31' (4. For 2a < t. (b) and (d) into Eqs. it is evident that the maximum values of stress resultants occur at e = 1[/2 and 0 = 1[/4.16) where reduction in the crosssectional area due to small hole has been neglected./O'm. (4.3 + \' (4. The final deflection expression (w.15) into Eqs..{(1 . It follows that the tangential moment and shear force at the periphery of the circle (r = a) are M. = L8Af 0 fat. For I' = 1/3. Upon introducing Eqs.16) yields kh = 1. .20) and setting v = 1/3.14) 4Mo .14).v)B a + 6(1 .. (c) at r = a are determined as follows M" = ±M./t' and T". for example.. + IV.80..6). In Fig. the hole width is equal to the plate thickness the above quotient is 1/3.. Eq. (e) leads to the values of the constants..4.6(1 . Qo = .. (2. The moment and cflcctivc transverse force corresponding to Eq. the maximum stresses are found to be O'm" = IO.15) The stress concentration factor is therefore k _~+ 31' b . the ratio '".v)C] cos 2e} v" = 1 Mo[2(3 ." In both cases \' is taken to be 1/3.12) and (1. sm 20 (3 + v)a From Eqs. (4.8M.) is then obtained.. It is observed from the graphs that the stress concentration factor determined using the thin plate theory is quite crude. respectively: M 0. The ratio of these stresses is (f) When.sTRESSES IN PLATES AND SHELLS thus satislying both conditions stated above and £Oq. (1. = 0) (e) Substitution of Eqs. The intluence of shear upon the plate deformation is thus pronounced..1 + ~~2 = 0 (. (4. = M0 [I  2~1 : \~) cos 20 ] (4.viA + [4vB .104 . is larger.v)C] cos 20 (d) The conditions along the boundary of the hole are represented by v. 4.
.5 An elliptical plate built in at the edge is subjected to a linearly varying pressure Jl = Pox (Fig.'S.5) determine whether the above represents a possible solution.5 m wide.~ and the stress (1x (at x = a.!a)' :(y/b)'l' 24D (5/a') + (l/b') + (2/a'b') (P4.2). Sees.1 A simply supported wing panel in the form of an isosceles right triangle is subjected to a uniform load of intensity Prj (Fig. 4.3. PROBLEMS Sl. 0. obtain the bending moment M. find: (a) the twisting moment on the side AC.3 A manhole cover consists of a castiron elliptical plate 1.3. By means of Eq. the solutions offer unacceptably poor accuracy. at point . the other an elliptical shape with semiaxes a = 2c and 0 = c.0 m long. y = aI4): (a) the deflection.4(x "" a/4. Given un expression for the deflection of the form w ~ E<0. Taking v = 1/3.1).4 if the edges of the plates are simply supported.3). and 0. Determine the ratio of the maximum deflection and the maximum bending moment for the elliptical plate to those for the circular plate. Determine. y = 0) for (l = 2h. one having a circular shape of radius c. 4.4 10 4. as illustrated in the example. 4. Usc \' = 0. Assume that (a) the plate is simply supported at the edge and (0) the plate is built in at the edge. 4.4 Consider two plates. (P4.'t. both clamped CIt the edge and under a uniform load of intensity Po. (b) Ihe concentrated reactions at the corners. by retaining only the first term of the serics solution.!:.~~. Determine the maximum uniform loading the plate can carry when the allowable stress is limited to 20 MPa. 4.5).PLATES OF VARIOL'S GEOMETRICAL fORMS 105 The case of plates \vitb noncircular holes may also be treated hy thL' usc of the bending theory.1 to 4. 4.2 Consider an equilateral triangular plate with simply supported edges under a uniform moment Mo along its boundary (Fig. 4.. . For the reasons cited above.3 4. Take v = 0.02 m thick.6 Redo Prob. (h) the bending moments. 4. .5 4. 4. 4.
1 INTRODUCTION In the previolls chapters.6) and the finite element method (Sees.2 FINITE DIFFERENCES The method of finite differences replaces the plate differential equation and the expressions defining the boundary conditions with equivalent difference equations. 5. B). was presented in Sec.CHAPTER FIVE PLATE BENDING BY NUMERICAL METHODS 5. leading to an improved understanding of how the physical behavior relates to the numerical analysis. written for every nodal point within the plate.10).".2 to 5.7 to 5. This section is concerned with the fundamentals of finite differences. 106 . The solution of the bending problem thus reduces to the simultaneous solution of a set of algebraic equations.1. The use of numerical approaches enables the engineer to expand his or her ability to solve design problems of practical significance. Such calculations are commonly performed by means of a digital computer employing matrix methods. Among the most important of the numerical approaches are the method of finite differences (Sees. A less formal method. 5. 2. these analytical solutions are not always possible. 5. as distinct from the somewhat limited variety of shapes and loadings amenable to simple analytical solution. and one must resort to approximate numerical methods.12. however. Both techniques eventually require the solution of a system of linear algebraic equations (App. Tn some cases. Treated are real shapes and loadings. equilibrium and energy methods are employed in the solution of a number of platebend ing problems.
..) = (\'Y .. + '. . The second derivative can be written using the difference representation of the first derivative: d')' ... is called the first forward difference of Y at point ·x.'.1 a): dY ) = lim )..d '" h(t)" (53) A procedure identical to that used above will yield the higherorder derivatives.+1 .Y" Here l1y .first backward difFerence at Vy". The first central dif/'erellce oy" is thus "Y.f' 1:9 I T " ···~x Figure 5. is Vy" = Y. We shall hereafter consider only the central differences.. Ii "'x . 5. ~:hth (a) • . r 'r \11 .. = t(y..' ('l.. ..1 ) The . ( dx. !. ....Y.) = V(l1y. f:XI/+ I X L I j ~74 I 12 +++(bl 'r'.PLATE BENDI!\'G BY .. Over an increment I1x = h. L\.1 The finite difference expressions may be obtained from the definition of the first derivative with respect to x of a continuous function y = f(s) (Fig. = Y. O L1x The subscript Il denotes any arbitrary point on the curve.Yn = YII+1  dV Yn:':: II (_dx ) n (5.2) The central diflerences contain nodal points symmetrically located with respect to s". II.' )" '" I1(Vy. . .1'.... ..Y"_I '" h(~~)" (5. I 2 15 3 :I)jl . Ilf I "! \ " "/'(XI '1 " f"J ~·"'··r ~ t ' " J . the above expression represents an approximation to the derivative given by (~~ t '" 11:. ....'\"UMERICAl METHODS 107 .
 1 II'.108 STRESSES IN PI.1') of two variables.. based npon the definition of the partial derivaiives.. y)] = 1 II' (WI .lb).3) and (5. .n ) v.bx\v " 1 Here the subscripts x and y designate the directions in which the differences are taken.\ 2 d' l' ) /I (5..  ox w . By taking c"x = c"y = h.2IVo + W. 1 1 '" I. . + IV? . are written at a point 0 as follows !!W 1 ax ~ 211 [\\'(x + h.'1(.(w.2w(x.2wo + "'4) 0)' .1V8) .h."') = 2h2 (ox IV. The above expressions.IV4) 2h' .. and iJ'w 1 iJx' ~ 112 [w(x + ii. II + aI' () <')'1/1 (b) We now discuss the case of the function w(x.h)] = .011" CC/H/'O/ diff('fcl1ce at the above.2) into = J'1r j I  2YII + YII·· 1 ~ /1 ( ~I . 0)' ~ y)l = 2" (WI 2h 1  11'.iI.a:~ilv a'". (5.w(x .w(x V .. Eqliations (5.) (tI) iJ 2 w 1 .  U "()''11+ 1  0v .2 {. y) + IV(X .\. . 'n (J a(".ATES /!'xf) SHEJ LS The . 1 . 5. y) . ~ .4) The third and thc/c)lIrth celllral rlifiCrellces are ()3 Yn = (j(()2Yn) = MYn+ j  2YII + J'n. _ (I .l) = ()Yn+ 1 .4) lead to OW t}x ~. y) . II " . t ) ''111 )  iJ' 2 ).' (jAO. is .2 bYrl + bJ'n1 (a) and <4\.) (e) 1 ..."n! I  2 ().\>1' alter substitution or Eqs. For the purpose of illustration consider a rectangular plate. the plate is divided into a square mesh (Fig.1) and (5.) = 4h' (w.[IV(X )' + h) . . Ow ."..(. h' (w.
Similar formulas can be derived when the nodal points are not "!'ellly spaced.. (Wl0  2/] 2w. It is noted that the cellter point in each case is the node about which each operator is written. (a) and (b) in the x and y directions.PLATE BE'DING BY NU:\1ERICAL METHODS 109 The finite difference approximation ofth(.) + w. (II'. Laplace operator at the point () is thus V'w =1.1  4wo} (5.\V = . For reference purposes. respectively: (e) .. . some useful finite difference operators are represented in the form of coejficient pattel'l1S in Table 5.1. one can readily obtain the finite difference equivalent of the plate equations. '" '. + /] IV. + 211'4  W12) The mixed derivatives are also represented by (I) Having available the various derivatives as difference approximations.13 ely a\v 1 3 1 (j.5) The forlllulas for approximating the higherorder partial derivatives are developed by applying Eqs. + IV.
(5.) (g) I..:::c. (I.) + 'k.) The operator V'w given by Eq. 211 3 .2wo + 11'.1 Coefficient patterns for some finite difference operators ~I! {.) I.: (lV.) a'w =  1 (w. DxiJy ~ h iix k ii. . 1 '02 '" I' (w. That is.ox w. a2 \\. oy a2 w "'k"I (\\'.w.6) .) .II' 2hk (iixw.lV. one need 0111)' to replace" by k in the foregoing derivatives of w with respect to y.  2wo + ]V. _I 0 I x 'cI)~' \dx..2wo ox 1 + "") = .I~\:) ~ [J In the case of a rectangular mesh with 6" = hand 6y = k. for example: (1W 1 1 oy '" k' (w. . .) (5.l. ...~ lH SrRESSES J:\ PLATl:S AND SHELLS T"ble 5. h . .211'0 + \\'.5) becomes then V'w '" '~(w.11'6 4hk ' + "" . (. .
+ IV" .3 SOLUTION OF THE FINITE DIFFERENCE EQUATIONS We are now in a position to transform the differential equation of the bent plate into an algebraic equation.5). it . • + lV.22). I to the latter equations at point 0 leads to (5. 5. M and IV arc zero at the boundary. The application of the operator \7 2 of Table 5. one finds that the difference equation corresponding to Eq. Referring to the operator \7 4 of Table 5.17) is ["'9 + M'10 +"'11 + w12 + 2(\\'5 + 11'6 + W7 + Wg) 8(Wl + IV.8a. The finite ditTerence operators in any coordinate set are developed through transformation of the equations that relate the x and y coordinates to that set. + w. Difference operators in cartesian coordinates x and yare well adapted to the solution of problems involving rectangular domains. One of the noncartesian mcshes most commonly employed to cover the polygonal and irregular boundaries is the triangular mesh (Sec. If the plate shape is a parallelogram. 5.6 that Eq.6). (1. The polar mesh is lIsed in connection with shapes having some degree ofaxisymmetry (Sec.s shown in Sec. In the case of a simply supported plate. and one can solve the first set of equations independently of the second system to find all .PLATE BENOl'\G BY Nll\fJ:RICAL I\JFTJIODS III Unless otherwise specified. As an alternate approach to the plate·bending problem.4wo ) 112 =  Mo jj" Identical equations are written for the rcmaining nodes within the plate. 5Jb. such as 0 in Fig. Let us write this equation for an interior poil)t.1. The set of difference equations is then solved to yield the deflection. 1.4). skew coordinates. Solu· tion of the problem now requires the determination of those values of M and w satisfying the system of algebraic equations and the given boundary conditions. we sholl rel(:r to equidistance (in: = II} = h) nodal points. 5. it may often be more accurate and convenient to use coordinates parallel to the edges of the plate. When the plate has a curved or irregular boundary. 1 = Po D (5.I.7) Similar expressions arc written for every nodal point within the plate. the boundary conditions must also be converted into finite difference foml. (LJ 7) can be replaced by two second·order equations (1. In all cases. + IV 4 ) + 20\Vo] . At the same time. 5.h) (Wt + 1 IV. special operators must be used at nodal points adjacent to the boundary (Sec. the procedure for determining the deflections and the moments is the same as described in the following section.
= 1/ (M.7) instead of from Egs. SOLUTION The domain is divided into a number of small squares. With the values of M and II' available at all nodes.M 4 ) _. From the boundary conditions at points located on the boundary. (h) of Sec.112 STRESSES I~ PLATES AND SHELLS the values of kl within the boundary. at point 0: (5. it is necessary to solve both sets simultaneously.1 Use finite difference technigues to analyze the bending of a sguare plate (a x a) with simply supported edges subjected to a uniformly distributed load of Po . this indicated in the figure by the shaded area. one can derive expressions for the moments and shear forces ti'om Egs. .\}' = . 16 for example. The finite difference method can best be illustrated by reference to numerical examples. the values of M may be different at the edges.. (1.. Note that for plates with mixed boundary conclitions.8).10) The stresses are now readily detemlined through the application of Egs.1 to '1 2M = I' at the nodes 1.= 2M.lor mixed boundary conditions. . 1.4M. M and ware zero.=16 . Application of the operator '12 of Table 5. The deflection II' may then be obtained more advantageously by direct application of Eg. For plates with edges clamped Of jfee. The second sct is then solved.\') + 11'6 "'7 + 11'8) and Q.4M 2 p =  a2 ~6 p oa 2 4M.l0) and Egs. and 3 results in 2M 2 4M.9) M. 5. it is important to take into account any conditions of symmetry which may exist. + M. . I (5.6.20). This has been done in Fig.2. Thesc are.2a. It is seen that only oneeighth of the surface need Ihen be considered. We thus have" = a/4. [n labeling nodal points. (5. (5. Example 5. or . (IV 5 4112 D(l . (1.12) and (1. .
16D w.~ .00291 . we number the nodal points as shown ill Fig. leads to IIPoh4 2w . Let us subdivide the domain into 64 small squares.. poa 4 = 0 . .2 Simultaneous solution yields (a) Similarly. (a). D Deflection w. O.2b. = 8D 9poh4 F rom the above.D .79 percent less than the "exact" value given in Example 3.21) and (a) for v = OJ This value is approximately 4.4w = .4w.... at the center orthe plate is 0. = 0.PLATE BENDING BY l'i'UMERICAL METHODS 113 i"'.= II 1+IiI (b) j (a) I I T8 . each with h = a/8. from Eqs._2 . 7Pu h4 8D II. Taking into account the symmetry. p a4 w.5 percent less than the "exact" value.a i !_il"~I T a I a ..M/D together with Eqs. .00403 _0_._. one expects to improve the results._ a Figure 5. ±. (1. + w~ . = 0.0479poa 2 . 4W2 .1.00214"0W2 . By means of a finer network. 4W2 = . 5. . The bending moment at the center of the plate is. one obtains p a4 w. V 2 w = .
22 percent less than the "exact" value. at the inner nodal points.2W6)] 2 a 64 = 0.4ID = 0.= 0.0473po(/'. For example. at point 7..004055P oa4 /D = 0. we 20 simultaneolls equations for the 20 unknowns M and".0006631'0 0 41 D 0.0011 86p o a4/D 4 = 0. (5.M 8) .068881' 0 a 2 M. Thus for the situation described a small number of subdivisions of the plate yields results of acceptable accuracy for practical applications. Solving these.)· (Q. To calculate the moments at any other location we apply finite difference form of Egs.003770poa 4 /D 0. = 0'()4466po a' M 1 0 = 0. = 0. .06523poa' My = = 0. we have (Mx).w 9 )) 6·i = 0.003507Po.. for" = 0. On the basis of the results for It = a/4 and It = a/8. we obtain At 1.056641'0 a' M s = 0.03452P o a' 0.M.3.12 percent less than the exact value. M.4/D = = "'.002134Po(/4jD The deflection at the center "'10 is now 0.002937poa 4 /D 11'8 1\'9 IV 1 0 = 0.072781'0 a 2 11'6 and I\' 1 = 0. referring to Eg.10). which is 1.001515po. = 0. a still better approximation can be obtained by applying extrapolation approaches 1 ' or by reducing the net size It further. the twisting moment at point 6 is found to be We can obtain the shear forces at node 6.05377pou' M..027741'0(/' = 0.!? .)6 = 4 a = 0.0353p oa2 a = D[2w7 . For the center. (1.017781'00' "'16 = 0.002733p oa 4/D "" = IV.0384poa 2 In a like manner.10).001627Poa /D 11'.W4  + 1'(2W7 . = D[21V7  2W6 + 1'(2w 7 W9  W. (M. 1V 7 = 0. 0. Mx = My = 0. M4 = 0..1293P oa 1I . Writing the finite differhav~ ence equations at points I through 10.114 STI{ESSES IN PLATES AND SHELLS The values of :\4 and I\" are zero on the boundary.03291poa 2 M 7 = 0.0479poa 4 (M 3 . as follows (Qxl6 = (M7 .
4M3 Polt' . (1. (1. Equations (h) and (c) indicate that the values of w (and M) at the nodes immediately outside the boundary are equal to the values of IV (and M) at the points immediately inside the boundary along the same normal. SOLUTION The boundary conditions are (Fig.2. i x /1= ~a. and 3 respectively.:· I . + M. (5. + 2M2 .1 to V2M = Po at points 1. only the triangular shaded area is considered. I Figure 5. .23) and (1.17).24). Example 5. Owing to symmetry./c < V .4M. . t I .3 .=0 OW (lW ox oy (x = y = (b) That is. The support reactions are computed similarly through the use of Eqs. I 2 '\ .4M. 4 1"'<'< t·. we obtain at the boundary (e) where the prime denotes the first derivative with respect to x or y.: < 2M. Referring to Eq.5.V i .1. = Polz' 2M3 + M.3. I . we have 4M2 . (a) Applying the operator V2 of Table 5. the deflection and slopes are zero at the points on the border. 5.3): w=O ". with the corresponding labeling of nodes shown in Fig.' ~i '.22) and (b) application of Eq. = Poh' = .= .2 Determine the deflection and moment at various points of a clampededge square plate of sides a under a uniformly distributed load Po per unit surface area.PLATE BE!\OING flY :\UMERICAL METHODS f 15  L IL. 5. for the first central differenc~. Take h = a/4 and employ two approaches of solution: (a) application of Eqs (1.3).
21) that. the calculated deflection WI is 43 percent high. leads to 3211'2 + 811'. and 3.ll/D = 0= M6 Solving the above expressions simultaneously.037921'0. (1.·. referring to V'W = .1 to V'w = PolD at nodes 1. ~ 0.00180Po.. while at point 4 it is about 25 percent low.. . = (M.n SHELLS and.00143p o a /D.~ 4\\'2 = M 2 h'/D 21>'.116 STRESSES IN PLATES A."'/D 211'3 = M.00082poa 4 /D M 4 ='0.AI! D 411'.2 M 2 = 0... . "'2' and "" equal to those given by Eqs. 4 8w 1  16"'3 + 2611'2 = Po 4 4' (a)4 I' 0 4 2Wl .16w2 + 2411'3 = (a)4 4 Solution of the above set yields values of WI.. + 20Wl = I' (a)4 .4/D = OJJOI21poa 4 /D W. for v = 0.)4 = 0.2. 11' / D 2W2 = M. substituting" obtain M 1 = 0. the numerical value of moments at point 1 is about 6 percent high. (d). The accuracy may be improved by increasing the number of nodes or by applying'an extrapolation procedure.3: (M xlI = (M yh = 0. (b) Application of the operator V' of Table 5.0261591'0£12 M6=0 It follows from Eqs. about 13 percent higher than the series solution. respectively..02465poa2 (Mxl. =  M.4'1'1 =~A11h2/D 211'. we = 0.025101'0£12 Compared with the series solution for this problem. Similar calculations conducted for JJ = 8 result in a center deflection 4 WI = 0. +"'1 .411'. '1'3 = 0.013691'0£1 2 WI a/4.038621'0'" (d) M.02230poa' M 3 = 0.
. SOLUTION The boundary conditions are w=Q "8.=0 oy a2 w (x = O. (e).PLATE BENDl~G BY XCMERICAL METHODS 117 . The operator V4 (Table 5. and 3 (Fig. We wish to obtain deflections at points 1. 5. the accuracy of the solution can be improved.4b). we shall number the nodes as shown in the figure.4 Example 5.!I a 3 Figure 5.1) is applied to V411' = Po /D at points 1. 2.4a). 5.al / % //. .6W2 = PO 4 /D h 6"" + 12w2 6"'2 = PO II (D = 4 + 12"" 0 Simultaneous solution of the above. is approximated by a continuous plate with opposite edges y = ±a/2 clamped.3 A stockroom floor. x = 2a) (e) w=O Taking into account symmetry of the solution about the x axis and satisJYing the finite difference expressions (Sec.. to yield 2w.:. after setting II = a/3. 5.=0 Dw . and 3. . x = a. half of which is carrying a uniform loading.' /' /.. . remaining edges and the midspan simply supported (Fig. respectively. results in By decreasing the size of the mesh.2.2) corresponding to Eqs.
. It is seen Table 5. I I 2 0 r. 2 =0: (!2 W / oy 2 = 0: "'21 = W.. The finite difference formula corresponding to Eq. J L 27 : 2B ! 29 30 tv (a) 35 J2~. ~ 2(1 + W l7 + \I(W9 + v(w l6 W.~l)  2(1 + V)W I6 = 0 + v}w 14 = 0 . IO 13 21 I' " 22 2~ 17 18 70 24 31 25 37 1. 5.. Use the finite difference method with h = a/4 to obtain w at various points.. and the fourth edge built in (Fig. ~ 2)(1\'8  W25 + W 23  \VIO) + (6 I. 0 ~ ~ 7 8 /...17) is applied to all interior nodes and to nodes 16 and 24 on the free edge..W 17 ) = 0 2l')(W23 ~ W2S) = 0 16 and 24.l3' W34' W 2 3 = ~ W3~' W~ . SOLUTION The domain is divided into twelve squares of sides a/4 (Fig.. ..5 Example 5.5b). 16 10 u.... = w2 .. (1. 5. at nodes 16 and 24..... indicated in the mesh on the figure. Clamped at Conditions w = 0: Ws = W6 = ~\'7 = "'s = \\'9 = 0 (lw!cs = 0: WI~ = WI' WI'.. . : I </// ... we have w = O. The boundary conditions for each edge.(6  2v)(wl~ At x \VIS = 0.... Free at x=a ~ W l4 + (\. respectively.u + "'..5..' T 3a 4 ~ x S 6 . This application gives eight equations involving thirtysix nodes.. = W36 Continuity of zero curvature· (c 2 w/ax2 = 0) along edge y = 3a/4: w 30 = ~W32 = 0 Vx = 0.4 Consider the case of a uniformly loaded rectangular plate with two adjacent edges simply supported. Simply supported at x = 0 and y = 3a/4 = 0: "'12 = "'20 = "'n = W 2 1'( W:lI iPw/iJ.118 STRESSES TN PLATeS AND SHELLS 4 i.._.. are listed in Table 5.. At the nodes of the simply supported and the builtin edges. \\'15 = IV 3 • W l6 = W 4 Continuity of zero curvature (B2wj(JX 1 = 0) along x axis: Ws= w10=0 w y=o B..) I W3$ ~ W 17  \I'n) .5a).. (I .//////.. where the detlections are to be found.. the third edge free.2 Edge(s) A.r~ >~ 36 (b) Figure. replaced by the finite difference forms.2. respectively: WJ8 "'26 "' 13 = ~WII' = \~'29 = W:H1 = W 31 = 0 = \\'22 = ~W19 c. at nodes +W + H ~ W 22 + (v··· 2)(W I5 W 2 .
\= pl>h~ D Ii 1 W21( t 1 0 0 0 0 \\.25819N = 0.21.2 provides four more equations in terms of the same unknown values of w. Case C of Table 5.4 8 1 t9 8 1. . The one that will be discussed assumes that it is required to develop an irregular star using the actual boundary points.54558N IV 26 = 2.7 1 0 1 0 0 0 0 0 0 0 0 I 0 0 10 1 0 0 0 0 t 0 8 1 19 8 1. and 24.PLAT!' BF~DI.07433N W I7 W22 "'IS 11'23 "''' = 0.79839N = 0.1) does not apply to point 0 because of the unequal lengths of arms 01.27753N where N = Po 114/D and h = a/4.[ at nodes 13. Writing the finite difference expressions for Eg. 5. 16.46598N = 0. When at least one of the arms is not equal to 11.5. 14. the operator pattern is referred to as an irregular star.4 0. Dividing a portion of such a boundary into a square mesh (Fig.38943N = 0. we determine the bending moment at any node in the plate as illustrated in the previous examples. and 04.'24 ""} \\123 W 25 "'26 The above equations are solved IV13 W I6 yield \V15 = O.7 0 0 0 .4 0 0 0 0 1.6 0. 02.2. we have eight simultaneous equations containing twelve unknown values of 11' at nodal points 13 through 18 and 21 through 26. (U7) by applying the operator V' of Table 5.7 1 0 ··8 0 0 2.) 0 0 2. 15. From these values. 03.22.7 5.3): 20 8 21 1 0 0 0 H 2 8 1 0 8 21 0 8 1 2 8 21 s 0 2 1 0 0 2 ..30383N "'24 = 0. rather than those falling" outside. We thus have a total or thirtysix independent equations in terms of the thirtysix values of the deflections corresponding to each node.4 PLATES WITH CURVED BOUNDARIES We now treat the bending problems of simply supported plates having curved or irregular boundaries.'. There are available various approaches to such situations.6 1 0 0 2 0 0 0 j"' W"15 W 17 W IS Ir 14 wI(.4 1.(j BY NUl\1LQUCAI. \I[TIIODS '19 from the tahle that all edge conditions are represented by a total of tH'eJ/(Yeiilh.23.51951N IVI4 = 0.." associated with the continued regular mesh. The resuiting twelve independent expressions are represented in the following matrix form (for \' = 0.3 5. R 0 2 11 0 () 0 0 2 0 0 0 18 8 0 0 0 2 t 0 0 0 0 8 2 [) 0 0 0 0 0 8 2 0 I 0 0 2 8 () 2 0 8 19 8 2 8 1 8 ··8 2 5.6a) shows that the V 2 operator (Table 5.96226N = 0. expressions.5. and inserting the lV'S given in cases A and B of Table 5.63989N IV" = 0.45037N = 1.
.IV._hhtlh + "tl (b) h(w4 . 4. and 4: WI = Wo W2 W3 + a2hl + Q4hi a1h +a 3 h2 tl2 = HIO = = Wo  h + Q4h2 \. the expression for w can be approximated by (a) This equation. + Bw. (a) gives il2W) ('l' (X 0 = 2a.{wo .. + Cw.~ ) 2 "" Wo (5..o a 1 = .= 2a 2W3 4 (e) Next.. 0 '" 2w.{h + "4) h{w.120 STHESSES Ii\" PLATES A:\"D SHELLS .{h+h. ..{wo .. the above may be written h'{V'w)" '" Aw. we introduce Eqs. i 1.wo). For convenience.) ''~ At point 0 (x = 0.(i. referring to Fig._..(h + I. ..l1b) . 3.) Q3 W2) h'{w. y = 0). gives the following expressions for w at points 1. Eq..Boundary D 2 <i''+'\t'. 2W4  (2 +.. ". + ".. hh.wo) .' 4hh. .).) = hh.6 It is assumed that in the region immediately surrounding the node O.1 = +1'"+ . 5.ax' + a'w) J w ay' 2 (f."") _. (e).'4 Wo + at 114 + U3 hi a2 = ~ from which h'(w 4 .6<1.. (b) into Eqs.h.~ + 1 + ~..{1 + "(4) 2w. the operator V'w is obtained in the form h ..{w o . 2.11<1) in which = h.h..."'3) a .wo) .{I + . + DW4  {E + F)lI'o (5..+ h. il2"..._x h B (a) (b) Figure 5. Adding the resulting expressions.wo) + hi(".1. (oy' .lh..
22. Example 5.~ )11'0 (5. For situations involving other than simply supported plates with curved boundaries. the development of the finite difference expression that replaces V'w near the curved boundary is not simple.7 " .1. it is sufficient to find a solution in onefourth of the entire domain.044 m. In so doing. 3.7).m (d) IiI t b . At nodes 1. ' = POh.. Owing to the symmetry.6b. (5.41 r' 1112 111. 5.03 m.5 Determine the deflection and the bending moments at the center of a simply supported elliptical steel plate under uniform load of intensity Po (Fig. + . = 0. Let a = 0. Note that w = 0 and M = 0 at the boundary. The case where all of the arms hi' i = L 2. we obtain: . We can thus write six finite difference equations corresponding to Eq.~. 5. 5. 4. 2. I 2 3 '.26 1.06 0 0 I 4 0 0 1. while at 5 and 6 we employ the operator of Fig....12). are smaller than" may also be treated by following a procedure identical with that described in this section. 3. may be applied in the same manner as the standard form given by Eq.611. The equation of the ellipse together with the given data yields ". Compare the results with those given in Table 4. a somewhat different approach for approximating V'w is used. h 4 .34 2 0 0 0 4 1 0 2 0 2 4.05 m..7. In any case.1 m. (1. / / "] "" " / .0245 m. = 0. x Figure 5. and 4 the standard finite difference expression applies. . and h. b = 0.12) The finite difference equivalent forms of the Laplace operator. and h = 0.  h I. 11. Eqs.. M. = 0.11) and (5.20:.).11) is shown in Fig. (5.5). M 1M. (5.(.PLATE DENDI?<G BY NCMERICAL MFrHODS 121 The pattern associated with Eq.25 0 0 2 0 1 . presented in matrix foml as follows SOLUTION 4 1 0 1 0 0 2 4 1 0 1.15 In.
8a) to cover the domains of the circular plates. together with the values of moment given by Eqs.5 THE POLAR MESH It is convenient to employ a polar mesh (Fig.2 (e) Similarly.'L_'_ /4 D' D 11'.. (f).1.' The bending moments at the center.3.. results in six equations from which W.1.4 IV2 = 1.1 .~ rJ'w (a) Using the notation of Fig.853 Po /. by means of Eqs. in terms of the polar coordinates rand 0. The maximum deflection occurs at the center and.)] = 0. The laplacian.3) and (5. _.42Jpo /.. (e).1... from Eqs.263po 1>' = M m" The value of Mmn is 18 percent less than the result listed in Table 4. = 1.230po'" At 3 = 0. application of Eq. .9).7691'0/" M. (f) and (5. (1.2 M. = 0.066 Po /.. (5. = O. 1 8w or' r or +~ r' ao' 1 . = 1. 5_8a and by means of Eqs.520~···D PO/..282~.6 percent less than the value furnished in Table 4. 5. we obtain W 1 o = 1.038Et ' P b4 (il) This is 17. = 0.22 STRESSES IN PLATES A:':D SHELLS which whcn solved yields M 1 = 1. with v = 0.4).1): V2W=··~ 82 w + ..4 (f) IV.J84po h' M" = 1.22b) at nodes 1 through 6. 5.884Po/.674~D PO/. from Eq_ (2.2IV. . or 2h w4 ) where . we have  ow '" _1 (w.4 D 11'. is.0J8po /. are therefore (g) My = il 2 [2w 1 D  2w4 + 1'(211'.2 M6 = 0. = 1.= Vr .
8 The pattern for V2 in polar coordinates is presented by Fig.6 THE TRIANGULAR MESH Some plates have boundary configurations that can be covered with ease by the use of a triallgulm' mesh shown in Fig. the governing equations (1.13) x (a) (b) Figure 5.9 . 5. sin {3 y (5..9b the equations relating the cartesian coordinates to the triangular coordinates are: x = qI y = <].Pl. 5. + <]. Referring to the operator of Fig.90. In this case. 5.Sb. Referring to Fig.22) arc replaced by the finite differences in terms of the nodal values of a polar mesh. cos {3 sin a. + ql cos a. in which . = hN· Other operator patterns may be developed in a like manner.ATE BF. q " <]2' and q. 5. 5.."D1K(. it is necessary to express the governing plate equations in terms of the triangular coordinates.Sb. BY NUMERICAL METHODS 123 (a) (b) Figure 5. + q.
'. yields all' ow ax aw OJ' . 8y' and thus 0'11' = R[2 ! V2 w = .14) reduces to 2 _ 2 a'w I1w.124 STRESS):. rN PLATES AND SHELLS In which ~ and Ii are the angles between '1. where a = 60°. =ax oq. In the case of a commonly used equilateral triangular mesh.14) (il'w . + _. are then (Prob..z sin 2[3 (ql vq. Eq.a). . The chain rule. y) is given. (5. 5. written in the matrix fonn. (a).__ fix a'w a'w ay>  a'w +a'w aq~ ili = R '12 sin 2([3 .oy aql = ox (b) '.jq.13): 'a21V/aq~ r = )a'wjcJq~ I jO'w/iJqj. and eliminating a'w/ox we obtain "Y..13) and (a) can be employed to define the corresponding expression w(q. [1 cos ex cos' [J 2 2 sin rx cos ex sin 2 ex 2 sin p cos IJ. (e) into the second and the third.' =0 oy iJq. iJq3 OJ' = cos [J av .Z+[)2 +0' q.. and '1" and '13 and </" respectively. together with Eqs. (IX I ex O(h = cos =sm ex (f. where R = ! sin ct sin P sin (P . Eqs. When an expression for lI'(x. sin' p 0 0 ]1aa'w/ox'oy I w/ox 2 c'w/o).= i~~. and [J "" 120°.8m [J ihv .= aq. ax OJ' The second partial derivatives of the deflection.. (5. aq. q" q3) and its derivatives with respect to the triangular coordinates. The following partial derivatives are obtained from the above Dq. (a) = sin [3 rJ... +.' iJq3 aw cos [J + .a) .'.15) .)'W)' (5.' (e) Upon introducing the first of Eqs. q3 3 oqj l a'w 8'w a'w + O'Z' sin 2a J uq3 (5.
5.) M4 I (e) . Note that M = 0 and '" = 0 at the boundary. Figure 5. Other finite difference operators may be developed similarly. 3.10 The finite difference operator pattern for the ahove. directions.PLATE BEi"OING IlY NCMEIUCAL METHODS 125 Figure 5. the condition of symmetry is taken into account as shown in the figure.M. Operator V' of Fig. 5.6 Determine the center deflection of a simply supported skew plate under uniform loading of intensity Po (Fig.1) in the qjo q" q. (1.22) can then be readily written in terms of the triangular coordinates. 1 iW2. )1 .lpo h Ii1I 1' _~ 'I~: '1 1 I (d) .11 . 2. In labeling nodal points. The domain is divided into 32 small triangles. is given by Fig. 5. and 4./ I q.=_~~='M.10 is applied to Egs.q. The finite difference equivalent of Eq. M 'lIW'1 1. Use h = a(4.10. 2 D 1M.22) at nodal points I. obtained by application of the operator for the second derivative (Table 5. Example 5.11). resulting in the following two sets of expressions: SOLUTION [l and 4 5 2 2 1 6 2 0 4 1M _ . (1. . . \ n 2 1 5 2 2 6 0 o I~'I 6 "'4 .
We here discuss only the commonly used finite displacement approach wherein the governing set of algebraic equations is expressed in terms of unknown nodal displacements.00262p o a"ID.002R3po (/4.126 STRESSES 1:\ PLATE~ AND SHELLS From the above: WI = O. The plate. 17 .12a which is replaced by an assembly of triangular finite elements indicated by the dashed lines.2. it can be shown that lS the center detlection w. 6 is the determination of stilTness matrices for plate elements with non isotropic properties. The properties belonging to a discrete element will be designated bye. 1./D. For example. described in Sec. Illustrated in Chap. There are anum ber of finite element methods. By increasing the number of nodes~. The derivations are based on the assumptions of small def/ection theory.7 PROPERTIES OF A FINITE ELEMENT The powerful finite element method has developed simultaneously with the widespread use of digital computers and the increasing emphasis upon numerical methods. the plate is discretized into a finite number of elements (usually triangular or rectangular in shape). This approach permits the prediction of stress and deflection in a plate with a degree of ease and precision never before possible. In the finite element method. Hence equilibrium and compatibility must be satisfied at each node and along the boundaries between elements.wc expect to improve the result. = O. may have any nonuniform shape and loading. connected at their nodes and along hypothetic interelement boundaries. in general. The "exact" coefficient lor the solution is 0. seJecting " = a/8.12 . y (a) (/» (e) Figure 5. The literature associated with this method is voluminous.finite eiemel1l of an isotropic plate is defined. 18 To begin with. Consider the thin plate of Fig. after derivation of the stressstrain relations for the orthotropic materials. a number of basic quantities relevant to an individual.00256. 5. 5.
strain" displacemellt matrix of the form I ' ". It is. = . = [D]{e}.My . = or ( z2[D*] dz (e). = z[D']{r. from Eqs. of cOllfse.17a) or (~).'" .}. The stressgeneralized "strain" relationship. stress.PLATF BENDING HY N"UMEIUCAL ~1ETHODS 127 Displacement matrix The nodal displacements {6i.'.3)..18b) 1M). Strain.20a) ~ /t/2 {M). (5.v)j2 Concisely. Moments are connected to stresses by Eqs.8) is as follows o (I . The latter is expressed in the following general form (5. to be later determined for' a specific element. we define. for commonality in finite element analysis of all types of problems.. dz . ] {r. desirable that a displacement function be chosen such that the true displacement field be as closely represented as possible. a generalized . ) (5.19) Substitution of Egs.9): (S. (1. and elasticity matrices Referring to Eqs. (5. are related to the displacements within the clement by means of a displacement jimctiol1 {w}". (5. (1. ) xy 'I~)' Itxl 1= lax e ! ( 12 W 2 ' (5. (1.t/2 (5. This matrix is often referred to as shape jilllctioll./2 .16) where the braces indicate a column matrix and the matrix [P] is a function of position.v e I z{a). = [B]{b). 1M i I .18£1) {a}.206) .18) into the above yields the following momentgeneralized "strain" relations (M). (5.17b) in which [B] is also yet to be determined.).t/2 jMx I = x.
differ for anisotropic materials (Sec. Introduction of Eqs..16). = [D]({o} . equals [k]. A.1Sc) {M}. surface area of an element.21) The stressstrain relations. II 1 1 A A II •• n . Expression (S.20) into Eq.22a) where n. 2:: II {{DoB}{{M}. The thermal stress problems are treated in Chap.0_' (I' Dow) dx dy = 0 (S. 5.. Owing to the many causes (e. A' The element stUrness matrix [k]. is ~rr = 2:: "II (Mx ~"x + M). temperature changes) some initial strains may be produced in the plate. . 9. is the thermal strain matrix.g.) = 0 (a) . (5.'x). for instance.(c5j.{Q).22b) yields ±rr 1 (Doc5}. from Eq.42). respectively {er). . = JJ[Bl'[D][B] dx dy A (S. and (S. 1'2) ~ t' l v o 1 (5.roo)). (S. 1 I' Dow) dx dy = 0 (5.{EO})e (5. and the lateral load per unit surface area.23) .2:: .) dx dy . = [D*]{{E) . shrinkage. (1.([k]. and thus elasticity matrix.]28 STRESSES l~ PLATES A!\:D SHELLS The elasticiry matrix for an isotropic plate is therefore Et' [1 [D] = 12 [D*] = 12(1 _. respectively.8 FORMULATION OF THE FINITE ELEMENT METHOD A convenient approach for derivation of the finite element governing expressions and the characteristics is based upon the principle of potential energy.12a.22b) A' in which superscript T denotes the transpose of a matrix.7). The variation in the potential energy ~rr of the entire plate shown in Fig. and (5. 5.17). (S. and p represent the number of uniform thickness elements comprising the plate. the stress and moment matrices become. My + Mxy ~.20c) wherein rOO}.22a) may be rewritten as follows .'. 6. In the case of a transversely loaded and heated plate.
(a) leads to the expreSSIOn [1<1.12).". We now assemble Eqs.{Q\) = 0 (I> ) The above must be valid for all {L\(\}. 5. Generate lQ'(4. are independent and arbitrary.9 TRIANGULAR FINITE ELEMENT The triangular element can easily accommodate irregular boundaries and can be graduated in size to permit small elements in regions of stress concentration. (5. (5..23) in terms of the given element properties. determine element moment from {ML = [D){()}" and the element stress Irom Eqs.25) (L\W{[ K]{o) . each requiring a different type of general displacement function. (5.(( [PJi P dx d" . (a) to obtain {Q}e (5. (5.27) We observe that the plate stitfilesS matrix [K) and the plate I/odal force matrix {Q} are determined by superposition of all element stiffness and nodal force matrices.18) or (t . (5. (2) Determine {Q}. Generate [K) = L[k)e. Then.PLATE nE1\'DJ~G BY :\l. Eq. This procedure will be better understood when the characteristics of a certain element are derived.1 fp. Consider as . _ "{Q' l (3) Determine the nodal displacements from Eq.24 ) Since the changes in (l). A (5. Two most commonly used platebending elements.'nt lIodaljiJrce matrix [Q. (Q. The general procedure for solving a plate. " U[Bj'lD)f"o) dx liy + .(or shell) bending problem by the finite element method is summarized as follows: (1) Determine [k). are discussed in the following sections.26) [K) = " L [k)e 1 " IQ\ _ '\' l'Q'ff' 1 J L .26) by satisfying the boundary conditions: {oj = [Kr '{Q}. This yields the following go/wiling equatiolls for the entire plate: [K]{6} where = {Q} (5. it is used extensively in the finite element approach.f('.~H:R1Cr\L METHODS 129 i~ The eJemt..1' = for the element nodal force equilibrium.24) in terms of the applied loading. Because of this. from Eq. respectively.e = . from Eq.'. due to initial strain and transverse load.
{.x~ .\T + a6.\"y plane (Fig. 2Ym 0 m V' .. defining the displacement at any point within the element iim. =.iYj Xj 2y. (5.'~e numbering cOllvention of the nodes.I Xd'i Xi . Note the cou11l('rclock\\'i.130 STRESSES I!'. Observe that the number of terms in the above is identical with the number of nodal displacements of the element. leads to improved results. For practical engineering purposes.30) will provide the displacement at all locations in the plate...+ . 0 0 3x? •'\~ J i 0 0 1 0 0 x· J 0 Yj 1 0 YT 2Yj 0 °xj OJJ wm fJ. through (/9 are known. 0 a.IJ SHELLS flnitc clement modd a triangular plate eicment (. Oy. th~ an' ex (5. corresponding to a triangle with six nodes.28) The positive directions of the rotations are determined by the rillhthal1d rule as s hO\\o'11 in the figure. (2x mYm + y. Displacement function The nodal displacement matrix for the element is (5. = a i + (l2X + a 3 y + QlX2 + as.}I? . G9 Ym 3x.y.31a) . Wj 0 1 J'i I 0 .) (2x.). a rotation about the x axis (0.m a4 a5 a6 a..m 1 Xm 0 2xj Xl m 0 Yj XmYm Xm 3xf X:i (X.r m ) m ·v? .. but the analysis becomes more involved than described here.29) The displacement function. in most cases the accuracy of the solution based upon Eq.. .30) leading to a reasonably simple theoretical development. and a rotation about the y axis (O.5. + yi) 2 2 (x )y. The nodal displacements can be written as follows: Wi 1 0 0 Xj Ox.) (2x jYJ + y7) (X.). PlAI'ES A".). Xi 0 2X j x~ .J (5.12b). This function preserves the continuity of displacements but not the slopes along the element surfaces. (xf + 2xjY..111 coinciding with the . Rotations are related to the slopes as follows 0). (5..1'111 Ym 1 0 0 2xm Y.\'2 + a7x3 + 2 Qs(x . is chosen to be a modified thirdorder polynomial of the form ). When coefficients il. Each nodal displacement of the element has three components: a dellection in the z direction (w). 0 3y. Eq. Yi x. 3yf 0 • J v~ "t tl3 a2 3y" • J 0 0 8.x·vJ ) ) J.Ym + XmYI~) (x2 + 2xm.v +xy2) + Q9y3 (5.30) is acceptable.t.y. however.Yi + xi)'n (x~ + 2x. A displacement function of eighteen order.
(5.35(/ ) or l'1'\ = 'Je! [H]{a"J . and (5.PLAn' n[~DIW(l BY Nlj. OX}' . = [H][Cr I {.:j' 0. = i Q~n ~ = IQ.30 into Eg. External nodal forces The nodal forces owing to transverse surface loading may be obtained from Eg. by introducing Eg..0]. (5.. ).>. (5. We thus have = [B][.35).IL [B] = [H][C]l (5. (5. (5.3. . and [e] are given by Egs.34) {c}.2. the solution for the unknown constants is {a} = [C]l{()}e (5.31).36) The stiffness matrix Upon substituting [B] from Eg.37) Here the matrices [H]. x.24) or by physical intuition.3] Substitution of Eg. (5.\IER1CAL ~H:TIIODS 13] or concisely.33 ) where [L] = [I.I}.31) show that matrix [C] is dependent upon the coordinate dimensions of the nodal points.36) into Eg. The element nodal force matrix is represented by IQi Q . (5.32) Equations (5.~38) " i I . (5. respectively. _zm' {Q \. After multiplying the matrices under the integral sign. Y. Q. (5.1'1' . the integrations can be performed to determine the elemeht stiffness matrix. . QZ)' Q .16): (5.3Ib) From the foregoing.32) into (5.21). QBJ')' Q QOXm' QOym)\ (5 . x 2y + Xy'. Q. (:::1' I}. "y..35b) One can determine the generalized "strain "displacement matrix.35) {E}.17) gives (5.r.ftH]'[. (5.23) one has [k]. y2. = [[CT'lr(.= lo r~ ~ ~ 0 0 2 o o o o 2 o 2 6x o o o (5.o][H] dx dY)[Cf ' (5. [. The displacement function may now be written in the form of Eg. (5.
24) and (5. and Ihe moment per unit length about the yaxis (Fig. cm.75 ··0.083 0 0 0.75 0 0.25 0.417 0 0. the moment per unit length about the x axis.042 0 0.5 0 0 0 0 0.042 0 0 0 0.187 0 0. respectively. 0 0 0 III = 3).062 0 0 0 0.583 0 0.and Q.187 0 0.25 0.25 0.25 0 0.31) is obtained as = I.042 0 0.25 0 0. Example 5.5 0. j = 2.O42 0._: t I J~ ········4 J' Figure 5.031 0 0 0 0.083 0 0.33) {QL = [fer 11'Po . matrix 1 3 0 1 3 0 1 3 0 [el.25 0 0 0 0 0 0.13 where Q" Q"x. represents a portion of a thin elastic plate which is under uniform loading of intensity Po. shown in Fig. (5.J' denote the lateral force in the: direction.062 0.rr [Ll' dx dy (a) .03t 0 0 ~O.417 0 0 0 0.5 0 ·0. Determine the nodal force matrix. 5. The determination of the element nodal forces is demonstrated by the following longhand solution.25 [[Crt)' ~ 0.7 The element 123.132 STRESSES f!\! PLATES AND SHULS 3 'i'.12c). 5. Assume that the weight of element is negligible. SOLUTION Referring to the figure (where i defined by Eq.13.042 0.083 0 The nodal forces due to uniform surface loading Po are given by Eqs.5 0 0 0.25 0 0 0.~"~~<:~.25 0 0.25 0 0 0 0. 1 0 0 0 0 1 [C) = 1 4 0 0 0 I 1 0 0 0 0 1 1 0 1 0 0 0 1 16 1 0 0 8 1 0 1 0 0 0 0 0 1 4 4 1 0 0 1 2 0 1 64 2 0 0 48 1 0 2 0 0 0 0 0 1 20 24 9 0 0 1 From the above we have 0..042 0 0. (5..25 0.
It is clear that the finite element method. x 2.34) and tile limits of integrations into Eq.5.'" / a / r ! Figure 5. x'). 5. the digital computer must be employed to perform the necessary matrix algebra.'(3 + asx 2 y + a9x}" + awY' + a'l x'y + a12 . three nodal displacement components described in Sec. 0.14 . nonuniform load.53.. y'} £Ix dy This expression is readily integrated and then multiplied by [[Cr I]' to yield {QJ" = PII{1.2 p. 5.J '0 I I . one has [Q}. To ensure at least an approximate fulfillment of the continuity of slopes.\.40) ~. 1.10 RECTANGULAR FINITE ELEMENT Let us now consider the rectal/gular element shown in Fig. 0. = [[C] ')'1'0 • . even in the simplest of cases.0. however. or anisotropic materials..1'2.PLAT!: nENDIXG BY '.Yy3 (5._ x . /1 . (5. 0.\:2 + ClsXy + a 6 y2 + £1 7 .://=::::::::===.1. x.07.I 21 ".L~\fERICAL :\tETHODS 133 substituting Eg.9 are taken into account.20. + '.60. 5. may be determined similarly (sec Prob.14. For any problem of practical significance..120.I' . requires considerable algebra. 'y. (0).0. )".Yo.)'. Displacement function The element nodal displacement matrix is represented by (5. the finite clement approach offers a distinct advantage in the treatment of plates having irregular shapes..49.39) The following pOlynomial expression for the displacement of the element selected 1 7 WI' = ill ij/1ll1 is + (12 + Q3Y + 0 4 . Compared with other methods. /1. .31.15). isotropic.89. .71} The clement stiffness matrix [k].
v2)[R]I[k1] + [k 2) + v(k.c. (5. [D].!8 in Table 5.134 STRESSES r:. (5.43a) (5.3. )'2. the generalized "strain "displacement matrix is found npon inserting Eqs. (112: (5.. (5. upon introduction of Egs.'.S. xy') Inserting Egs.. xy. . a 12 x 12 matrix. It follows that El 3 J [k]'=180ab(i. and (5. (5. y3. (5. = (H]{a) As before.31).37).41) into Eg. (5AO) and (5. and [C] given by Egs. The unknown displacements.] to [k 4 ] and the matrix (R] are given 17 . (5.}_= [~ 0 0 0 0 0 0 2 0 0 0 0 ~ 0 2 0 6x 0 0 ~2y 2 0 4x 0 2x 4y 0 6)' ~6xy 0 0 _6x 2 ~Xy 1 ia" .28) into Eq. These coefficients in general provide rapid convergence and satisfactory accuracy.42) into Egs. y. (5. is obtained by introducing matrices (H]. (5. one has {I.S Nodal displacements. (5.43b) or (e).: PLATES A"\.42) The displacement function is expressed by Eq.17). (5. However. the matrix [L] in that eguation now has the form [L] = [1.43).] +i[k 4 ]r[R] I . External nodal forces The element nodal force matrix is where the Q's are defined in the previous section.1'2 __ ."D SHELl.41) in which (C]. x 2}'.43). The stiffness matrix The element stiffness matrix [k). are next found_ In concise form. Inversion of the ahove provides the values of the unknown coefficients Ql) . these arc (5. x 2 . strains. iJ 12} (5. 5. x'y.39). depends upon the nodal coordinates as in Egs.44) Explicit expressions of bendingstiffness coefficients [k.21). xy2.6.41) into Eg.33). and stresses may now be calculated employing the general procedure of Sec. .v I (5.
36) for a rectangular plate element (Fig.vI 15 ...30 = 6 84 8 6 (l 0 (I 20 0 30 10 \} 6 0 () . . " 84 " R \1 (J Symmdr:c . ("1'1 15 60 30 0 l(l 0 30 30 0 15 15 60 0 30 0 0 0 15 0 5 5 30 0 10 60 0 30 30 0 ·15 [k~J = 0 0 0 0 0 2<1 15 0 10 60 0 . 5..::!ril" (l 15 (l 0 0 0 () 0 0 (l 30 . 84 2 " " .15 (> o 0 15 0 30 0 20 0 0 0 " " " !l " 30 15 •.15 (l [kll:=. 6 0 84 6 2 " (l (l 0 0 () 15 0 IS [k. 0 00(/ ['I :. " IR) = [[') [ll) 10) 10]1 (0) I') [0] (0) [0] 10) ['I [OJ [OJ [0) [OJ ['I = [I 0] 0 0 .44) and (6. )5 0 l' (l J5 30 II " 0 . (1  6 6 « .~ " J5 " 0 0 iJ 1. 15 HI (l t5 15 Il 0 0 0 0 0 0 0 \) 60 30 0 30 15 0 30 20 Symmetric 6 84 6 0 .4 (. 60 30 0 30 30 0 0 0 20 15 15 60 0 30 30 0 n 15 0 0 0 15 (l 0 15 30 0 0 0 0 0 0 0 0 20 15 0 0 0 0 0 0 (I (I 0 0 0 0 0 () 0 10 Symmetric .) . ···6 .14).Table 5.3 Stiffness coefficients in Eqs.d 60 30 0 60 . (5.h r '> J 30 30 15 0 30 0 0 .' n 60 30 () 20 0 30 10 6 " 0 (I 60 [kJ~" (" .15 30 15 0 30 () 30 15 15 30 (I Symm. '" . 6 I) 6 :'. I. Il ." ···6 6 84 6 6 84 6 60 30 15 I) lO \} " 0 0 0 20 0 0 0 0 0 0 "6 6 where 2 \} " 0 6 \l 6 6 ..
we employ only one element per quarterplate.. Compute maximum value of 1\' if the plate is subjected to uniformly distributed load of intensity Po.. = {Wi' 0.8 Consider a square plate of sides II with two opposite edges and x = a simply supported and the remaining edges clamped x= (Fig. 0. 0. 0. I'LAT~S A1\.1 Po . O} {£I}.0.0. OJ The force·displacement relationship (5. 0. 0.3.26).15 Example 5. 0. A concentrated load Po(1 x 1)/4 = 1'0/4 is assigned to node I. we obtain = H'max= O.. 0.0.35 x 6))111'11 (0 + 20 + 0 + 035 x 8) 10x. together with the values of stiffness coefficients (Table 5. Take a = 2 m and \' = 0.15c). 5.D SHELLS I I.. 0_'2. 0.35 x 84) (0 . ~ ··t: I \ il . For the sake of simplicity in calculations. 0.28) are introduced along the lines of symmetry. is then readily reduced to the form IPo!41 I °1 £r' = lso7i=o.3).0. The boundary constraints permit only a lateral displacement \VI at node 1 and a rotation Ox2 at node 2 (Fig. SOLUTION Symmetry in deflection dictates that only one quarterplate need be analyzed.9) (60 [ + 60 + OJ x 30 + 0./ Sliding edges te) Nodal force and (a) Actual plate (b) Substitutt! plate dis plact'lll ents Figure 5.15(/). The substitute plate is shown ill Fig. provided that slic/illyedge conditions (1.35 x 6) °H'J (030+00. 0. 0.0.136 STRESSFS [. 5.. Nodal force and the displacement matrices are ° {QL = {Po/4.30 + 0. 0. 5.I From the above.3617~E 3 .151>.0.
3 Determine the /inite difference cquivalent of M . Take. Determine: (tI) the displacement II· at the nodttl point~ 1 through 9. CIt 5. for a rectanglliar mesh.The "exact" solution of thi' problem (see Example 3.5 Swept wing of an aircraft is approximated by a simply supported skew plate subjected to uniform loading Po (Fig. a nodal point () II and Vx at a nodal point 0 (Fig.6 Consider a uniformly loaded plate with two opposite sides ~imply supported.. M..I' . (IJ) the bending moments al nodal point 9..5. .1 Verify thaI the effective (Fig.5). in the form ~hear o forces arc represented.I) Referring to Table 5.3397Po/£(3 Wn1llX = It is apparent that the accuracy of the solution increases as the mesh is refined. .) O.J3 [W9  . PROBLEMS Secs. and 25 elements. 5. respectively: Wn1ax = O. 5.7) is 0. 16. 5. j) 11'"  2(3  V){W j  1\'3) + (2 . the results art!.5 Figure P5.2 Determine a finite difTercncc expression corresponding to \7 4 1\' = "ID (Fig.3355J'oEi'..1')(W.1b). and the fourth side free (Fig. tiS finite difference approximations at point Vx = .6). check the correctness orille reslllt for I~.1 to 5. 5. 5.1.1 __ = "and L1y = k. r~" Figure P5. When the quarterplate is divided into 4.3512Po/Et.3378Po/Et 3 H'max = 0. rectangular mesh with .3 S. P5.6 Free _. 5. P5.  \\'(>  \\'7 + WIl)] (PS.1h).1. \. 'I a _:0. Determine the deflection w at the nodal points 1 through 5.lh) for 5. 5. II ~ 3 5... the third side clamped.4 Calculate the maximum stress al the nodal point 22 for the plate shown in Fig.x = II and Ay == k.
t = J. _ _ _ = .ll Figure PS.1 0 Figure PS.12 . "" 11 6  5 Figure PS.9 Figure P5. 6 a/3 r '. L .138 STRESSES I~ PLr\TE:S AND SHELLS " / I t a " I T " J ..S a/3 1. a ___ 1. a t h. II 4' Ii.~ a ~~_. .1 .. ~ Figure PS. i. .. II"" _.7 Figure PS.. ~.
5. 5. 5.13 Verify tlll~ results given by Etjs.7 to 5.1.7 through 5. 5. in tcrms of E and v. (c) of Sec. lIsing only one clemen! per quarterplate.14 Redo Prob. described in Example 5. Compare the result with that givc::n in Table 3. th~ rectangular element of sides a .1Ild /1 (Fig.4 (0 5.!7 Calculate maximum deflection ofa simply supported and unirormly loaded 2 m x 2 m square plate. .PLATE HE:\Dl~G BY "". 5.15 Develop the stiffness matrix of the finite clement.6 llsing the chain rule. (5.9 if the inner edge is built in and the remaining edges are simply supported.16 Derin' the matrix [CJ given by Eq.10 5.MF:JUCAL METHODS J39 Sees. Sees. 5.7.12 Each of the \ariously shaped plaks ~howll in thL' figurcs is simply supported at <"Ill edges and carries a uniform loading of intensity Pu' Determine tht' deflection II' at the nodal points hlbeled on the mesh configurations describing the plates.41) for by locating the origin of xyz at the node II. 5.14). S.6 5. 5.t.
plywood. and gridwork. The latter consists of two systems of equally spaced parallel ribs (beams). and attached rigidly at the points of intersection. fillers in sandwich plate construction. A nonisotropic or anisotropic material displays directiondependent properties '9 Simplest among them are those in which the material properties differ in two mutually perpendicular directions. fiber reinforced composites. Presented in this chapter are the fundamental equations for the smalldetlection theory of bending of thin orthotropic plates. A material so described is orlhotmpic.. reinforced concrete.1 INTRODUCTION The plates analyzed thus far have been assumed to be composed of a single homogeneous and isotropic material.CHAPTER SIX ORTHOTROPIC PLATES 6. Orthotropic properties of several commonly employed materials are discussed as are applications to various orthotropic plates. 140 . mutually perpendicular. However. wood.'0 Examples include corrugated and rolled metal sheet. A number of manufactured materials are approximated as orthotropic. e. Topics on plates made from two or more different materials and numerical methods are also included. plates of anisotropic materials have important applications owing to their exceptionally high bending stiffness.g .
as this relation involving E. Ex.!me for both isotr. Eg. .2) are connected by' E = . An alternate representation of Egs. (6..ger applicable. v. shear '.' modulus of elasticity G is the s.2) :5 Here v.. elasticity.... (1. Equation (1.?pic and orthotropic materials. The same is true for the conditions of equilibrium of Sec.7) now assumes the following form: (J)' = EyD J• + Exyl. respectively.'. 1'y. the two sets of elasdcco~stanis' in Eqs.. E~ are the effective Poisson's ratios and effective moduli of. (6.1) and (6.I' = GyXY in which the moduli. The stresses arc obtained by introducing Eqs.1): (6." ./) is (6.. x E' l~'xvy (6.3) The straindisplacement relations (1. Clearly. Subscripts x and y reiate to the directions.3) arc based upon purely geometrical considerations and are unchanged for orthotropic plates. and G relies upon material isotropy. and G are all independent of one another.ORTHOTROPIC PLATES 141 6.6) is n%i.2 BASIC RELATIONSHIPS Solution of the bending problem of orthotropic plates requires reformulation of the Hooke'. E. ." E. and E~. 1. (1.. In particular.. law.2 _.1 ) ~" r1.5.x (6.4) .The.3) into (6.
D)'=JT (6. · +D .7) Q.1:.. (1. \\.5) 2 w Al.5.= where (6..5). as illustrated in Sees... 6.15) and (6.4 and 6. resulting in approximate expressions for the clastic constants.3 DETERMINATION OF RIGIDITIES Discussed briefly in this section is the determination of the rigidities of orthotropic material> used in plate structures. represent the flexural rigidities and the torsional rigidity of an orthotropie plate. The accuracy of these approximations is generally the most significant faclor in the orthotropic plate problem. We can obtain the vertical shear forces in the plate by substituting Egs.( D·· ..5: Qx = (6..142 srRESSFS IN PLATES AND SHELLS The formula. are M. Practical considerations often lead to assumptions. (1. inserting the foregoing into Eqs. The orlhotropic plate moduli and Poisson's ratios E~.6) The expressions for D" D. with regard to material properties. respectively. E~. (6..9). upon satisfying the given boundary conditions.. •J. for the moments. 6.. is expressed in the fom1 \~ (6.5) into Egs. (!X2 i)y2 i:'J2W alw) (6.) ..= ~2Gx\"""'''''''' . through the use of Egs. D x \" and G x.. (b) and (e) of Sec. and integrating the result ing expression.='. vY ' G (6. Dx ily c where 3E D =~ x 12 t 3 E. 1.9) which is solved for w.8) The governing differential equation of deflection for an orthotropie plate.1' .1 0) .
. _.. _ . Et J (/12.. s (shown as shaded) G~. Elastic modulus of thc plating n.~ B... D=..." 1· 1 E... Plate reinforced hy <:quidi~lant .1 Various orthotropic plates (ieometry A. ) I':" +I'. as in th~ case of isotropic materials... respcctively 1": Poi..il. Moment of inertia of Ihe stiffener cross s~ction with respecl to midplane of plating C.11 ) tJv x \\. x ".0 D> ~ EI . (6. J [._ ..: Poisson's ratio for concrete f.. H = . The plate rigidities are calculated from Eqs.: Elastic modulus of concrete and steel.~{. rl'spt"ctil'cly stiffeners +.. Jt'Spt'cli~eJy [aU.{ s EI~ l20 _ v~) D)" ~" £1. _.5(h/2s)J 'I J .ORTHOTROPlC 1'1...(1 + I') Dx . and (1).3).: Torsional tigidity of the plating E. Reinforced concrett' slab with x and y directed reinforcement ~ted haf$ Rigiditie> .. Platt: rt:inforc~"(\ by i1 set of equidistant ribs lJ "... Dxy 12(1 .'.···] E ." 1 (E <> E.Sh I 1 .)". .···!.>:. vi iz f..l+··.. ATfS 143 arc obtained hy tension and shear tests..~)" C: Torsional rigidity of one rib I: Moment of inertia about neutral axis of a T·scction of width.. I. .' ~ 1· \. 0. _ _ .. E': Elastic modulus of plating and stiffeners.1 1.....1\ 1'. (6. ). Moment of inertia of tht: slab (sted burs) about neutral axis in the section x = conslant and J" = constanl... E D~. Corrugated plate D x =_.< E'/ E.6)....' "" 0 where I = O.8): (6...) Table 6..~soll's ratio of plating s: Spacing between centcr/ines ofstilfeners 1. ..)." E.81 1 + 2.
nny a h (6.1 presents the rigidities for some commonly encountered cases. 21 Such a plate is lern1(~d an eql/ivale1ll or t. Table 6.8111 p(x. (3.1b). 6.. For example. as expected.4..14) and (3.9) yields the deflection and stress in t hat employed for isotropic to treat the case of a simply a nonuniform load p(x..10) experimentally. . (6. 22 It is noted that when E~ = E~ = E (and hence l'x = l'y = 1')..10) are approximated with Eqs.. it follows that the terms in the brackets must be zero. (6. leading to a "'" = ..3) into (3.144 SntESSES 1...12) 12(1 ~ .') G X)~ ~ .4/a 2b') + 'F.9)..4/h4)D~ x Sill .....14) The expression of the plate·deflection surface. (1.15) . to determine the constants of (6..1)./a) sin (n"y/b) dx dy . Pmlr{SUl . 3.Et' . Introduction of Egs. the bending stiffness of the ribs and the plating are combined and taken to be uniform across the replacement model. "~.. . 4 oc 00 I" /.2) supported rectangular orthotropic plate under (Fig. Subsequently.." '0'" (."4. 3. The lalter approaches consist of constructing an orthotropic plate with elastic properties equal to the average properties of components of the original plate.11) giving the rigidities.=/) Et 3 12(1 . is therefore IV = ai... (1I/ 4.... PLATES A:':D SHELLS When it is /lot possihle. the constants of (6. however.3) become Ex = £)' = Consequently.1) into (6. (6. m~. y) sin (mn. For reference purposes. Innx .. reduces to that of an isotropic plate [Eq.....~. __ x " ~l~~2 ~v £ (6. Eqs... in the case of a plate reinforced by ribs..4 RECTANGULAR ORTHOTROPIC PLATES The general procedure for the detemlination of rectangular orthotropic plates is identical with plates.v') (6.17)]. (6. 1nnx . y) m=II1"'1 I I I la oc 00 ("m+1[4 mll '4 a Dx + 2H ~2b2 + ~b'4 Dy €I m21l 2n 4 li4n+) I . p"'''. ll . __.. . £(' 24(1 + v) H=·····~. We now apply Navier's method (Sec.13 ) It follows that Eq. (6. D =D =_.4/a4)Dx + 2H(m'n2. by substitution of Eqs. one resorts to approximations deri\'ed using analytical techniques...Sin a b = 0 111ry Inasmuch as the above must be valid for all "and y..ransformed orthotropic plate.4(a' b2) + (n""4/b 4)D.4/a")D:'+ 2lf(m 2n 2..
It is given in Fig. 6.. (3. (6." D.1 for the case of evenly distributed nodes.16) When the material is isotropic the above reduces to Eq..ORTfIOTROPIC PLATES 145 For an isotropic material.a'b'f+(. i. (6. (6.+~D~.16) becomes \\' = 16po ~ ~ sin (lIl"xla) sin (IInYlb) DIT 6 . 'i.1J{ 211 2H )f::.+~D[i)~. For example.Z(I1::. We are led to conclude that the center deflection of the reinforced concrete plate (a x h) having rigidities D. referring to Sec. from Eqs.1 the pertinent coefficient pattern for finite difference expression of Eq. Deflection of orthotropie rectangular plates is also determined by following the same basic procedures that were prescribed in Chaps. is equal to that of an isotropic plate (lI. (6.4. D. x 0. (3. (6."lb4jDy] w ~ sin (mITx!a) sin (IlITyl") (6.[2) and (6.' mll(m' /ai + /1 2lhn' (/1/.:~4~(/~f. = H = D.1.[(m4Ia4)D. Based upon the notation (a) Eq.) of rigidity D. (3.5) and the stresses by applying the relationships of Sec. Referring to Table 5.1 J).'~..e" for Ax = Ay = h. II = 1. we can obtain the bending moments from Eqs. if the plate is made of a reillforced COllacte.5). . ) (b) which is of the same form as Eq. from Table 6.4.1 .~.9) of the orthotropic plate may readily be obtained. 3.1012Ff Figure 6.15): 161'0 W = "Teo' ~ ~ .6). and the above coincides with Eg. In the particular case of a rectangular plate under a uniformly distributed load Po. 3 and 5 for the Levy's solution and for the finite difference approach.. we have H = J DxD~. Having the expression for the deflection of the plate available. ~ D.6).. 3. 1. + 2H(m'n'. 211 rt.3 we can readily obtain from Eq..
. By decreasing the size of the mesh increment.1 111 \ 11 II (d) The simultaneous solution of Eqs.146 STREssES [1'\ PLATES AND SHELLS .1 A square orthotropic plate is subjected to a uniform loading of intensity Po. Eq. 2.718 4. .859 . .718 ii' 11'3 ( 21.718 20. = O. 13 . 3. D.5Do.718 8.OO19po a4lDo.859 19. SOLUTION For this case.859 2. Take D.OOI56poa4/Do.' .' ' .718]I 13.859 9. Applying Fig. Do 1 w.718 W II = E<~~~ ) 1 .859 ·.718 4.1 at the nodes I.248D o . 6.2). appears Po Do (e) Considerations of symmetry indicate that only onequarter (shaded portion) of the plate need by analyzed (Fig. .3. t ill = a . and 4. = O.718 J w2 i 17. ! t 4 I I Fignre 6. we obtain 18. Assume the plate edges are clamped and parallel with the principal directions of orthotropy.9). The conditions that the slopes vanish at the edges are satisfied by numbering the nodes located outside the plate surface as shown in the figure (See Example 5.718 8. using the finite differencc approach by dividing the domain into equal nets with h = a/4.859 6.13. = 0. the accuracy of the solution can be improved.~ ".. 6. the governing cxpression for del1ectioll.:>. (d) results in The center deflection. W. is about 24 percent more than the "cxact" value..2 Example 6." O. = Do. and ''x = v. If = 1. Determine the defiection. 2 I .2). 6. The values of ware zero on the boundary.429 17. (6..
= H = D. (h) The expression describing the deflected surface of the plate is then W=  Po ..5). The solution procedure follows a pattern similar to that described in Sec.) (c) When Dx = D. ~. 4.. Expressions for the moments may then be obtained from Egs.9) leads to an expression which is satisfied when k _ Po  8 3b"D x + 2a'b' H a b 4 4 . clamped at the edge and subjected to the uniformly distributed load Po (Fig. _ .. For a builtinedge orthotropic circular plate of radius a under uniform load. (6.a b" x' 8 3h*Dx + 2a 2 b2 H + 3a 4 D. Assume that the principal axes of the ellipse and the principal directions of the orthotropic material are parallel.5 ELLIPTIC AND CIRCULAH OHTHOTROPIC PLATES Consider an elliptic orthotropic plate with scmiaxes a and b.10). + 3a"D._.. 4. from Eq..17) This equation satisfies the boundary conditions for an elliptic plate with fixed edge.. Substitution of the above into Eq.~. Egs.17): IV =~l). we have.4."'~~""'.19) is identical with Eq. (2.4 The maximum deflection occurs at the center of the plate and is given by (6. 1 .ORTHOTROI'IC PLATES 147 6. respectively. 4.o. The result obtained above for an elliptic plate may readily be reduced to the case of a circular plate by setting b = a. (6.19) where D.18) reduce to Egs. (6. (4.9) and (4...17) and (6.3).. presented by Egs. Thus.. the deflection formula for an isotropic circular plate. (6.a' ( " (a) in which k is a constant to be determined. in the case of an isotropic plate.18) As anticipated. (6. Eq. = t{3D x + 2H + 3D. ... we Jet = IV k 1 . a' b' 4 y')2 (6. (6.~ (a2 _ 1'2)2 MD. (a) in Sec.(.14).
+ D. are expressed as follows . and portions of space vehicles) can often be satisfactorily approximated by considering a laminated plate.. yields M..<\TES AND SHELLS The bending moments and the twisling moment arc calculated by means of expression (6. (6.2(D.xy (6. for the ith layer (Fig.g ..148 S1RESSES I:\" Pl. = I:~" [(D.6 MVLTILA YERED PLATES Structures composed of an arbitrary number of bounded layers (e. in these structural assemblies.. Generally.• aircraft and marine windshields.3). . orientation of the principal axes.20) The stresses are then determined through application of the formulas given in Sec.y' + D..3). and the anisotropic properties.)")'x2)] I (6.(i)_ .)('" .3 . 6. We shall treat only the places consisting of isotropic layers. The straindisplacement relations (1. which when illlroduced into Eqs.2) .19)..4. each layer may possess a different thickness.5)... 6. 1..21 ) x Midplane Figure 6.
and summing the results..(i) Substituting strains defined by Eqs. dz (6.24)." I (J x I'i) Z .1 1 (tf  l/ t!. .8). [. t7}".3) the transformed flexural rigidity is given by D.24) The general method of deriving the governing equation for multilayered plates follows pattern identical with that described in Chap. and the isotropic material properties of each layer are known. transverse load p. (6.ORTHOTROPIC PLATFS 149 Hooke's law (1. (6.iJ I 1 2(1 + v. (6._. for the ith layer. solution of a multilayered plate problem reduces to that of a corresponding homogeneous plate. integrating over each layer. ._ \'.. (1._ ". (6._.22) rifJ = x)' . we obtain the stress resultants: I Mx I My' I hiXl" I ~ =2:: I . It can be shown that 23 the differential equation (1. irx I y Stresses defined by Eqs..) + ~": 1 ~l22J 1 '11+ 1 (6.~ = _.:. Layered plates of a symmetric structure about the midplane are of practical significance.. 1. The stress components in the ith layer may then be computed from Eq.7) now appears ali) .25) may be solved for w(x.21) into the above._~L..26) If boundary conditions.) 'X)' (6. Eq. 6. We observe that. y).23) ' '" . = ~ 3 [£ ~~. are (6. } + 1>.25) where the D.dXO] !!L .i j. is the transformed flexural rigidity of laminated plates. upon introduction of transformed rigidity. .17) now becomes a V4W= JL D.'.. All analytical and numerical techniques are tll\ls equally applicable to homogeneous and laminated plates.. For a plate of 211 + 1 symmetrical isotropic layers (Fig.
27a) 0 0 I x. they are as follows: hi. . is theretore E' .V shape and thickness. PI ATFS A?\I) SHELLS 6. 5 is applied for computation of displacement and stress in an orthotropic plate of arhilrClr.:: The elasticity matrix. J 1 . [ /'. solutions of orthotropic plate problems were limited to simple cases in which there was uniformity of structural geometry and loading.~. the finite clement approach of Chap. k. Let us consider a plate in which x' and . 6.4 and 6. The stress and generalized Figure 6. 6. the stressgeneralized" strain" relationship is given by Egs.1' 0 G (6. or. Written in the matrix form.27b) J a\ ! k = c[D*jl.' ... t:y (6.. (6. (S.. (6.l .\.21). from Eg.• 1'" = z .~~[ ~ " ."x l'}.28) o o The principal dircctions of orthotropy usually do lIot coillcide with the x and y directions. l'x \'y E' ~~[ ~ 0 0 G Bx \' .4).v' represent the principal directions of the material (Fig.~~ [ . When the principal directions of orthotropy are parallel to tbe directions of the x and y coordinates.S. t 'J.4 . succinctly I~ ax E'x . however. subjected to lIolllllli/arm loads.~ ". In this section.".2).: . For plates made of any nonisotropic material it is necessary to rederive the elasticity matrix [DJ.150 STRESSE'S I:'..~:::.7 THE FI!'iITE ELEMENT SOLVTlON Tn Sees..
i C..29) together with Egs. (5.' = I sin'IX [ 2 sin ex cos tI.34) for the elasticity matrix of the orthotropic plate in which the principal directions of orthotropy are not oriented along the x and .2).sin 2 a h'~y I {I.28) or (6. 1vx·i'y· "x.. referring to Egs.30) into the above. "x·E. we obtain {u} = o[TYfD*'][T]{s} = z[D*]{"J in which (6. l?y.v axes. For any particular .33) [DJ = !1 [TJ'[ D*'][T] L (6. sin' " sin (j' cos IX ] I Cx . y'.' (6. the transformation relating stress components in x. (6. from Eqs..:".ATf:S 151 .29a) o or o (6. 1 E.] and [R] are listed in Table 5. . Yxy at any point of the plate. = [T](e} (6. we obtain The coefficients [k. .30a) 2 sin r:x cos r:x cos 2 a . (6.3. (6. strain:' in the directions or these coord inates.\: . are written in the following matrix form I cx' li'x'Y' I COS' " .21) and (6.: . z' is written as {a} = [T]i{u'} (6.ORTHOTlWPIC 1'1. Similarly.31 ) Upon introducing Egs. With Eq. In the case of a rectallgular.~:. 1"x'v)'. \'r' (6.] to [k.33). the expression 3 (6. cos 2 ex sin IX cos (J.34). y. explicit expressions'B of stiffness matrices for orthotropic plate elements may be evaluated as outlined in the preceding chapter.'} Concisely. z to those in x'. orthotropic plate element.29b) Equations for transformation of the strain components l:x.32) [D*] = [TFID*'][T] We thus have.~.. arc related by .30b) where [1') is called the strain transformation matrix. (P1.
assuming that the principal x and y axes of the ellipse and the material coincide.3 Show that Eq. 1. and Ec = 21. E~ = 200 OPa.3 6. 6.1 A plate is reinforced by single equidistant stiffeners (Table 6.152 STRESSES t:. 3. 1. Assume the following properties: t = Ir:.4). Dx )'.1 to 6.1) arc specified. and 1= 12 X 10'1 m4.1. PROBLEMS Sees. 6.2 m. by taking n = III = 1.5 Determine the value of the largest deneclion in the plate described in Proh.. = 30 mm. y) is rJ.6 A simply supported square plate of sides a is subjected tn a uniform loud Po {Fig.7 Determine. 5. .ax = 1 mm. m = 11 = I. 6.34} for the strain energy appears in the following form in the case of orthotropic plates: (P6. 6. {1.15. What should be the value of Po for an allowable deflection \\'". Assume the plate is constructed of the material described in Prob. I. 6. t = 20 mm. S = 100 rum. consists of a Hat plate reinforced by equidistant steel stiffeners and is elliptical in form (Fig. = 2.1). Torsional rigidity of one rib is:' C = JG = O.3.1).12 A 5mm thick large plate is fabricated of an orthotropic material having the properties: E:. 6.2G = 13.2 Determine the rigidities of an ortholCopic steel bridge deck which may be approximated as a steel plate reinforced by II set of equidistant steel ribs (Table 6. 6. 0 = 2E~. Q = b.. s = 200 mm. and The process of arriving at solutions for the orthotropic plates is identical to that described in Sees. subjected to uniform lond Po.4 to 6..10 Redo Prob. Determine expressions for: (a) the deflection surface.246/"~(l.1).9 A steel clamped manhole cover.4.11 Derive expressions for the bending moments of an orlhotropic elliptical plate with builtin edge... v = OJ. 6.3). 10 mm. Compute the rigidities.8 to 5. The material properties are given in Prob.13). v = 0.7 6. 6. 6. 4.9 for a circular plate. 6. = 30 (Fig.3).. 6. if a = b.4 GPa. "'" lC)'/2. 6. Retain only the first term of the series solution.8 De~ivc an expression for the deflection of an orthotropic damped rectangular plate under a uniform load Pu (Fig.>: = Ir_.\. = 0. y') and the reference axes (. 6. Dy . Find the maximum deflection jf {/ = h.(Fig. Determine the elasticity matrix of the plate.4 A rectangular building floor slab made of a reinforced concrete material is subjected to a concentrated center load P .: PLATES A1\1) SHELLS orthotropic. 6. 3. The plate and stiffeners ace made of steel with E = 200 GPa. and E = 210 GPa. 6. Take b = 2a. 3. Retain Ihe first two terms of the series solution.2. 6. the center deflection of a simply supported square plate uniformly loaded by po. (b) the bending moment Mx. 6. . the appropriate values of the rigidit ics D" G x )' (Table 6.3) Sees. The plate is constructed from the material described in Prob. lise the Ritz method by retaining the first term of the series solution.1 and {/ = 2h = 4 m./2. material.10. ~'•. Compute the maximum displacement w. t = 0.6 GPa The angle between the principal directions of the material (x'. The plate edges can be assumed simply supported.()G.
2 hold. Attention will now be directed to situations in which lateral and illplane or direct force systems act at a plate section. and finite difference methods.CHAP fER SEVEN PLATES UNDER COMBINED LATERAL AND DIRECT LOADS 7. respectively. The latter forces are also referred to as the membrane jorces. and biaxial compression.2 is no longer valid. and assumption (2) of Sec. To begin.1 INTRODUCTION The classical stress analysis relations of the small deformation theory of plates resulting from a lateral loading have been developed in the preceding chapters. 9). or they may arise as a result of temperature changes (Chap. 7. This is followed by consideration of buckling stresses caused by inplane compression. upon application of equilibrium. the governing differential equations are modified to include the simultaneous action of the combined loading. and yet large enough so that the products of 153 . w is still regarded as small so that the remaining suppositions of Sec. 1. energy. pure shear. These forces may be applied directly at the plate edges. The problem of plates with small initial curvature under the action of combined forces is next discussed. The chapter concludes with consideration of a plate bent into a simple surface of practical importance.2 GOVERNING EQUATION FOR THE DEFLECTION SURFACE The midplane is strained subsequent to combined loading. 1. However.
which also act on the element. which are functions of x and y only. [12 2 and noting that for {3 small. from the equilibrium of N x dy forces. Referring to Fig. and that likewise. and N" = Ny.154 STRESSES IN PLATES A:. (. Writing cos {3 = (1 . Consider a plate element of sides dx and dy under the action of direct forces N x .x dX)dY cos {3' . 7. The sum of the x components of N" dx is treated in a similar way.). The other resultants due to a lateral load. N. The z component of the normal forces acting on the x edges is equal to (I> ) .. Eg.10) and (1. the stress resultants are given by Eqs. it is necessary to consider the z components of the inplane forces acting at each edge of the element. 7..2) To describe equilibrium in the z direction. {32/2 ~ 1 and cos {3 '>0 1.JD SHELLS (11) (/' ) Figure 7.1. = 1 .5.1a and b. + of!.. The top and front views of such an clement are shown in Figs..1.. The condition L: Fx = 0 then leads to 0i'l" + oN" ox oy Furthermore.sin 2 {3)"2 = (a) 1 t sin 2 {3 + . respectively. the condition = 0 (7. cos {3' '>0 1..16).1 the inplane forces or their derivatives and the derivatives of ware of the same order of magnitude as the derivatives of the shear forces (Qx and Q). Assume the body forces to be negligible. (a) reduces to (IN x /iJx) dx dy. as before. 1.. are shown in Fig.. = 0 oy (7. Thus.N xdy cos {3 ox in which {3' = {3 + (o{3/h) dx.1) L: F).. = ax 0 results in ~l!. we obtain ( N x+ ~l!.XJ'.
rx oy .)~\v + (o. .xdy fly ax For the forces in Figs_ 7. from I F. cjJ alV .l'jXl' + a~)')8W = ex 0)' ox Ox 0)' c7y 0 (c) It is observed from Egs.2) that the terms within the parentheses in the above expression vanish.dl. (b) and (e) of Sec.\1' + ~~~ dX) ex ox('X The z components of the shear forces N" on the x edges of the element are determined as follows.dx = cc.16) into Eg.1) and (7.dx [!X (IX ('Xl Neglecting higherorder terms.+. (1... are unchanged. and (7. and hence Egs.PLATFS l"':\~DFtt COMJlI:\'EJ) LATERAL AND DIRJ:CT LOADS ISS Ina. Example 7.211' II "II +.) 8x iJy (7 . dJ'~\~ + ex (N x + ?~x dX)dY(~.5. (1.2)... Egs.' + CV + p + N \ .3) Expressions (7.dy+. The slope of the deflection surface in the y direction on the x edges eguals ""'jay and awlc')' + (.. (1. subjected to combined lateral and direct forces.. Derive the eguation of the deflection surface.3) to determine the deflection surface of the plate.much '" /i and fi' are sm.1)..+ 1 N xYax a. (7. (e) yields 02W + 2NXJ' :.'. As the direct forces do not result in any moment along the edges of the element. Either Navier's or Levy's method may be applied to obtain a solution.2).l6).. = 0.y2\\> eN }>x i1W I d. .17) is now replaced by Eg. 7. and hellce . + (ON x + i1N..5.dx dv .3) are the governing differential equations for a thin plate. Introduction of Eqs.2 ox 8y2 . (7.1 A rectangular plate with simply supported edges is subject to the action of combined uniform lateral load Po and uniform tension N (Fig. we thus have iJQx l}. Eg_ (b) is therefore N. It is observed that Eg.+ N } . 12 ( W + '. 1..dx d)' (Ix cy ONx·OH> An expression identical to the above is found for the z projection of shear forces N yx acting on the y edges: N )'x ax oy  .2 wlex ?y) dx_ The: directed component of the shear forces is then N"':.v . (7. oQ) r2 w tJ2 w (32\1.1 and 1... sin /i '" /i " hl'ex and sin /1' " Ii'.
S1l1 .3): I' = ... (a) of Sec.. III /I 11111 a b (111. ) . ! ::: . (3. .. (3.' b' Upon comparison of Eqs.\:D SHELLS  !> . .._".. J' . . The lateral load 1'0 can be represented by (Sec.2 = . or the actual buck lillY load.\" = 0.15 2~'\. 3. (7.2.2) are identically satisfied.5) 7. expressed by Eqs. (7. and hence Eqs. 3.\" I : t " Figure 7.+""a' b' D n')' m )' (m.1 ~ _ ~) j. Platebuckling solutions using Eq.2 SOLUTION In this particular case. ':Y) I1lnx .1) and (7.. .5).3x 2 {!J~2 + t3y4 .. are not . When this is introduced into Eq.. are satisfied by assuming a deflection of the form given by Eq.= I. 11 = 1.SIn 1 b (7. it becomes unstable and begins to huckle at a certain critical value of the inplane force.~' . N x = N = constant and N.\.x/a) sin (:.156 STRESSES 1:': PtA rES A:. Buckling of plates is qualitatively similar to column buckling. ._. . (7. ~ I 'I ..3).Y/I>2.4) The conditions at the simply supported edges. i~' 13...!6. we obtain a m" = _.6) and (7. 3.lh). mr)! . a buckling analysis of the fonner case is not performed as readily as for the latter. SIn ~. .3) usually involve considerable difficulty and subtlety.4). (7. = N x. c4 w a4 w N alw 16po TC2D ~~ 00 00 1m. (/·. I'll /I mll[(m~+!. mt)' Inserting the above in Eq..~_~_)' + .3 COMPRESSION OF PLATES.·. N("')' 1 D na (7..P"~'·_·_·__1 N( na n Dmn r(111' +."Lfl ("".i " TC 16po I Im I Sill _.. we have 4 Dw a. .. .. n11tx .24 and the conditions that result in the lowest eigenvalue. ) The deflection is thus W = 16po 6 nD f f . we are led to conclude that the presence of a tensile (compressive) force decreases (increases) the plate deflection.:.._ 6 ""~ .' However.r + .. BUCKLING When a plate is compressed in its midplane.. 3.
2 A simply supported rectangular plate is subjected to uniaxial inplane forces N (Fig. I I o Figure 7. are poor in resisting compression. although quite capable of carrying tensile loadings. buckling or wrinkling phenomena observed in compressed plates (and sheUs) takes place rather suddenly and arc very dangerous. Usually.3 2 3 4 at all obvious in many situations.3). = N. 7. Determine the buckling load.. there is close correlation between theory and experimental data concerned with buckling of plates subjected to various types of loads and edge cond itions. Thin plates or sheets.7) . Example 7.1) and (7. = N = constant and N.4) is used to good advantage to ohtain the approximate buckling loads. the inplate load that results in an elastic instability.. mry m II a SJtl b (tn. the energy method (Sec. we thus take p = 0 in Eq. Fortunately. (7. For this case N. The governing equation for the displacement becomes SOLUTION DV·w+N.. For a plate.3).PLATES Ui'<DEl{ COMBINED L~TERAL ... 11 = 1. An illustration of plate behavior under compression loading is presented in the following example... This is especially true in plates having other than simply supported edges. . is indepeluiellt of the lateral loading. 7. and hence expressions (7.\ND DIRECT UMDS IS7 t. In the analysis of elastic stability. as in the case of a beamcolumn. J (7. =O GX 2 We assume the solution in the form lV (l2w (7...6) = LL IJ) 00 Jnnx.2) are satisfied identically.2. = 0. Q II!I1 . Often in these cases.
(7. indicating plate buckling.. Referring to the figure.!l.S11l . the magnitude of k depends upon I' only.. we set ~N" = 2n:1!. k = 434 and m = 2. while several halfwaves in the direction of compression can occur.3. Thus. 2.)( /. 4 are sketched in Fig. This provides the following N =_.10) m/.a Ill' T Jamn sm . the bucklillg mode given by Eg.158 STRESSES IN PLATES AND SHELLS which when substituted into Eq.. In the case of r = 1. ('" + In _ m' + 1) = ~'. when N attains the value given by the righthand side of Eg. To ascertain the aspect ratio.~. sin (ny/b). The corresponding critical load is N" = 4.8) that the minimum value of N occurs when II = 1.11) The variations of the buckling load factor k as functions of aspect ratio .. from Fig.5.8) can only be olle halfsille wa!)e... N.0 h ~ The nontrivial solution is n'D (~~~ + a2 b2 ~!.34n'D/b' under which the plate will buckle into two halfwaves in the direction of the loading._b2 minimum value of the criti 4n'D cr The corresponding critical stress.8). dr b' r from which cal load: 0 (7.9) for the critical load. is given by (7.)" m n'D =k'b2 (7. Clearly. a Nn 2 wry . at which the critical load is a minimum. the magnitude of N" and the number of halfwaves m for any value of the aspect ratio r can readily be found.8) Clearly. .r b2 (m r +. 7. when the simply supported plate buckles. for a specific m. for In = J. 73. one has am" " 0 and hence w 1 0.) + (r/m)]2 and. Here k = [(111/.)2 ~ ~~1_2_ = 2 a 0 from which (7.. It is observed from Eq.6) leads to L ~ ~ I'D1t4 L mn (Ill' + b "')' a 2 2" ~ Nrr . (7. (7. = 1. = a/b. 3. across the span. The resulting expression is thus n'D N cr = . for instance. lIm.\" . (7.
. unit displacement of the midplane in the y direction equals " = )' !(iJW)2 20r (7..4 APPLICATION OF THE ENERGY METHOD The principle of minimum potential energy may be employed to analyze plates under the action of lateral and inplane loading.remains unchanged..Jw dx )']!i2 = ex dx _ .. Work done by the direct forces may then be evaluated.'I)EH COi\lHI:"iED I. linear element AB is displaced to become A'B'..1 ((~W)' dx + .. 7.ATERAL AND DIRECT LOADS 159 It is also observed [rom Fig. and its horizontal projection owing to the displacement w is .. ~ (c7\\')2 : " 2 ex .. a lonil plate (h < a) with simply supported edges under a uniaxial compression tends to buckle into a llwnbcr of sqllare cells of side dimension h and its critical load for all practical purposes is given by Eq. dX' _ [ ('. Suhsequent to the belUling.12b) For the purpose of determining shear strains associated with plate bending....3 that a plate m times as long as it is wide will buckle in 1/1 halfsine waves. . 7. Consider element dx dy representing a point at the midplane of a plate (Fig.12£1) Similarly. consider now two infinitesimal linear elements 0 A and OB in the x and y . 2 ex . Inasmuch as midplane stressing does not occur.4a). i! L z (a) (6) Figure 7.PLATES U".4 . (7.< "". and the energy method applied readily to any particular problem. (7. We shall first develop expressions for the midsurface strains.) (a) ' j The midplane displacement per unit length (to a second approximation) in the x direction is therefore f = x . original element length dx . 7.10). Thus.
... due to displacement IV only....34): u '(OZW + illW)Z . They are now taken into account. PLATES AND SHELLS directions (Fig. However.(iJ w)Z J.v) foZwa'". (1. yields the midplane shear strain Yxy = ay ax  awow (7.. these elements Illove to the positions O'A' and 0'11'.120) Note that in previous discussion of plate bending. .. The platestrain energy owing to bending is given by Eg. The work done by the direct forces.41». .vN.14) Application of the basic expressions derived in the preceding paragraphs is illustrated in the discussion of a platebuckling problem which follows. respectively.1112. (ow )Z +2NXY8xcy 1dxdy ax oy x (7. It is further assumed that direct forces {Ire applied first (before lateral loads) and that they remaill unchanged d!lring plale bending...dx dy = '2 'oxi' ayZ ax2 ay' ax i!J' l D " 2 9\ (7. having direction cosines fl' 111 1 .. Due to z displacement .... since their products with direct forces may be of the same order of magnitude as the strain energy of bending.2(1. the strains given by Egs.l caused by inplane forces. then eguals 1 I" oW 011' W=z) [N (OW)' +N.. The latter assumption is widely used in the classical treatment of plates as well as beams and shells.2'2 . The plate . 7. This expression upon substitution of definitions (b) and (c).160 STI~ESSES I.. Ill' and '2.13 ) where A is the area of the plale. = 0 m.112.dx ow 111 = . given by ... 1 = Et (N x  vNy) I OJ' = Et (N" .= ex dx 0\' 1 ax (b) and I.  ~ 1~ 1 2 ~y ('!W)2 (c) Consider the following: }'X}' = " .. these strains are considered very small in comparison with B..12) were always omitted. (7. A'O'B' = SIn (" .
15) Example 7.d '0 Inserting Eq. m nab mnx mry (d) Clearly. 7. Hence. from Eg. sin 00 co _. ··~·I ~. . n 8w ow I I . SOLUTION Assume that the deflection surface of the plate is described by an expression of the form: W = L L am.D DJl{EC'T LOADS 161 i>:d . N x = Ny = 0. = S. (7.. Subsequently.m2_ . it is assumed that the plate undergoes some small disturbances and buckling takes place.. mrrx Sill .NDEn CO~1RI"ED L/\TERAL A0.5 is taken to be subjected to constant direct forces (N x • Ny. sin . i1 i 1L \' b ( i i its js~ J~ Figure 7.~.. buckling occurs if [i = W.2a m a ~.. (d) into this equation and observing that I ".3 Determine the buckling load of a simply supported plate under the action of uniform shearing forces N".":~ dx dy ~o ox 0): .Y ________ :~. we have the expression U=[iW=O for the potential energy of the plate. (7. the foregoing satisfies the boundary conditions at the simple supported plate edges.. Thus." "0 a cos . N x . The work done by S during the buckling of the plate is. pi prrx 1 0 (m ± p is an even number) (m ± p is an odd number) .') dllrini/ bending just before it buckles. the work done by the direct forces must be equal to the bendingstrain energy stored in the plate.5. W= S .13). That is. as shown in Fig. This means that the magnitude of these inplane forces is just equal to their critical values.PLA1ES L. At the time of transition from one to the other of these equilibrium forms. dx = '. no energy is gained or lost.
7' OC· 11I11pq (1m" llpq ( .T. from Eq.2 + b I~)' a2 00 2 .~ ~ a"(r/12 '::p'j(q2 _ /1 2) = 0 (7..c ':f. Upon application of Eq.2 )2 '. In so doing one has /1' +p )' Equation (7. (el) into Eq. only two parameters a'l and a22 are kept. (7.14) and integrating the resulting expression. is equivalent to requiring that n be a minimum.17) assumes the following convenient form .162 STRESSES IN PLATES AND SHEUS we ohtain W = 4S I I I I ·:.19) We thus have a system of linear equations in am." of which an approximate solution may be found by retaining a finite number of parameters a"".16) In order to ascertain the critical value of the shearing forces.17) in which p and q must be such that m upon the notation 1'= odd numbers. the coefficient am. This. Based a b A=32rb'S m/1pq (7. it can be demonstrated (Sec. is to be found such that S is a minimum. n ± q are .' + b' .am" rr4. If.15) now becomes (I) n = ~.... we have..9)..4S I I I I iXJ 00 ro 1nllpq am"a pq . 1.4D <Xl ~ mnpq _ /1') = 0 (7. (L43).18) expression (7. The strain energy associated with the bending of the buckled plate is found upon substituting Eq. m~pqn mnpq 2 2)( 1 2} (e) wherein In ± p and n ± p are odd numbers. (7.( "2)( m/I(Jq mpq~n 2 '} = 0 (7.a""'. for example.8S ~ ~ a pq (ni' ~p21({? ± p.Dah (nIl .. we thus have "'4.. (m 2 + /1'1")' 00 00 J.:4q~ ~j: a' 8 _ m II 11111 (".19): rI' 16}'(1 + r2}' a2 2 + '9(111 = 4 0 .
(I + .5 THE [<'INITE DIFFERENCE SOI.":EJ) LATERAL A~[) DIRECT LOADS 163 These equations han: the following determinant .PtA rES "l~~DER CO:>. 7. 7. as was observed..UTION The preceding sections have been concerned with rectangular plates subjected to uniform loads. (7. cr . The determination of the buckling load by means of analytical methods may be quite tedious and difficult.6a)..32b2 /"2 (7. = 1).2·1 with the solution . Example 7.20) This result ditTers by approximately 15 percent from the "exact" solution 24 for a square plate (..+ .4 Compute the buckling load of a simply supported square plate under uniform compressive loading per unit of boundary length (Fig. N 2 N 3 o 4 y (a) (b) (e) Figure 7. In order to enable the stress analyst to cope with the numerOUS compressed plates of practical importance. _.6 . .')2 S . Accuracy may be improved by retaining additional parameters. numerical techniques must be relied upon.. is thus 9n4/) (l + .')' 4 9 162(1 + . 5) to the governing difTerential equations of a plate under inplane loads. = + ..2 The magnitude of the critical load.__ ..... from Eq..IBI..1'4 9 i.18).'f =0 .9 (1 + r2j2 ... The following examples apply to the method of finite difTerence (Chap.
2.22b). shown by the shaded area in Fig. (7. 7. the conditions arc described by: w=o w=o Hence a2 w 0\1' C).22a) only. Dy2 =0 (on the boundary) and the problem is reduced to finding the solution of VM +·M =0 D V'IV = M o N (7. also leads to the trivial solution w = O. referring to Fig.\_ = IV.3) reduces to j\/ (7. for the trivial solution M = 0. The finite difference expression corresponding to Eq.·2 =0 0) (x= ±2 a) (. expression (7. is written at node 0 as follows: (7.6c. Upon substituting. On applying Eq. it is observed that Eq.25) The plate is now divided into a number of small squares.g.. (7.164 STRF.\" = = iV" = O.SSES iN PLATES AND SHELLS and I' Sot t:nOl\: The plate is subjected to constant inplane loads N.21 ) Along the edges. e.v = +2 (0 ) .24) with Na 2 K=D (7. It is concluded therefore that to treat the problem requires the solution of Eq. (7.23 ) where 11 represents the number of divisions of the sides. (7.22a) (7.2211 ) under the condition that IV = 0 and AI = 0 on the boundary.23) at nodes 1. For convenience.22a). 16. For IV = 0 along the boundary. in the computations we shall denote (7.66. . 7. Owing to conditions of symmetry one need only treat oneeighth of the plate.
= N xy = P = O.Df:R Cm. SOLUTION We have N x = N.. (7. = 2. approximated by a square plate with opposite edges y = ±a/2 builtin and the other sides simply supported (Fig.3) then becomes (7. from Eq. we ohtain. carries uniform compressive loads N. Example 7. The buckling load.75. three equations can he written. Tbese aft: n:presented in the [ollmving Illal rix form: =0 The ahove leads to a nontrivial solution if thc determinant is sct equal to zero.1 percent.IIIl~fJ) LATfc]{. .. and N. 7.\15 = kf(1 = 0 along the houndary. I i J . Equation (7...1\. the accuracy may he improved..7(1).25). J y ta) th) Figure 7.. Determine the critical buckling load. is thus This result is smaller than the" exact" solution 24 by 5.26) ..PLATf:S [. By increasing the number of subdivisions..7 .5 A portion of a missile launcher support fixture.\L A?\"J) DIRECT LOADS 165 3 with ll_l = .8284 and hence K = 18. In so doing.
wIl8(0+0+W 1 W 1) + 2011'1 + (0 . 7. 7.8) /1 = 4 and the antisymmetrical deflections. Note that the values of ware zero along the boundary. The finite ditTerence equivalents of Egs.26). (7. . K 9 2Wl  \I'd = 0 from which Of K = 54 = 5.7b. (b) are fulfilled by numbering the nodes located outside of the boundary as shown in Fig. employing h = a/n and K = Nu 2 /D.' PLATES A~D SHELLS Figure 7.8 The boundary conditions are \t. Thus. (e) It can be shown that using the critical load is (Prob. 7. (d) for the square plate under consideration.2. On applying Fig. the plate is snbdivided into nine small squares. we obtain the coefficient pattern for the finite ditTerence expression of Eg.' = 0 0 \t' = elw oy 0 = (b) As before.8. Fig.471..8 at point L we have (WI +11'1 +0+0)2(0+0+0. Proceeding with the finite ditTerence solution. 7.166 STRESSES Ii'. Antisymmetrical deflections in the directio1l of the compression result i/1 a smaller buckling load.
However. In order to take into account the extent of this influence.27).) a2~1.. mnx ..3).n 2 N 1~ D a SIll .27) it is observed that the influence of an initial curvature on the deflection is identical with that of a fictitious eqllivalent lateral load One concludes therefore that a plate experiences bending under direct forces only if it has an initial curvature.~T + 2N. provided that at any point the magnitude of the initial deflection 11'0 is small compared with the plate thickness I....' [!2W ·. respectively.:. Upon comparison of Egs.1 7) as in the case of !lat plates. Here WI is determined by solving Eg. (1. it is found that /.sm .  mnx . (d) . (h).~ 1 1 ( Cl ml . 7. m" = '''.···r G X . a sm b nny (a) II! I' Assume that the plate edges at x = 0 and x = a are subjected to uniform compressive forces N x = N (Fig. nny L L bmn sm . (1.\"V·~.PLATES UNDER ('OMHlNfD LATERAL A\:D DIRECT lOMJS 167 7.1 • II! " a h (c) is inserted into Eg. Consider..3) is modified as follows V4 wt = _. amnN n2 Dla'[m + (/l' a2 /mb')]' _ N . Ox (ly (7.x. The differential eguation (7. with Eq.' ay .27) where W = IVo + WI' This is the gOl. The total detlection w is obtained by the superposition of \\'0 and deflection IVI due to the lateral load. Eg. . as an example. nmx . 1( D P + Nx a2 l\' + N\.17) still holds. as might be anticipated on physical grounds._. and lateral loading are significantly affected by any initial curvature.. a simply supported plate for which the shape is described by Wo = lin loaded L La OC C(' • mll sm .17) and (7. It is noted that the left. (a) introd uced is V w 4 =. fJ._.{til initial curvature. the loadcarrying capacity and deformation of a plate under illplall(.6 PLATES wnIl SMALL INITIAL CURVATURE For a plate with an initial curuaWre subjected to the action of lareral load only.and the righthand sides of the above depend upon the change in the curvature and the total curvature of the plate.>el'lling d!lj'erential equation for deflection of thin plates with "". (7. the governing equation (1. mry sm a b (b) When the series solution w1 = .
3).. as descrihed in the alternate manner in Sec.8).. and the goverl1il1g equatiol1for deflection. it is seen that bm .. (b) is determined by substituting the above into Eq. To write the equation for a plate subjected to uniform tensile forces (Fig. which occurs at the center...:() SllFLLS The solution or Eq.. \\' = w(x).... When N reaches the critical value given by Eq. . (7.31) When axial force N x is zero. sm . Eg. (c) of Example 1.29) For m = .30) in which now Cl N = ... (7.1... reduces to (7...»1['_)] (7...... the deflection of an initially curved plate under simultaneous action of inplane forces N" N). n'D/a 2 [1 + (a'lb')]' Note that the deflection expression derived above is analogous to the relation for initially curved columns 1 From Eq. This means that the plate buckles.31)..3....... (d).... _ ..... 7.. and N". the foregoing agrees with Eq. (d) vanishes and w.c (7.m"l~ a b oc· . N l1'D/{r[m + (....")._ .2 a'.. grows without limit. is given by Wm:l x =Ia (11 1 (7.. 7. (c).... . If plate edges are 110t Fee to move horizontally.. a tension in the plate is .. may also be readily obtained... the maximum detlection of the plate.. By following an approach similar to that described above. ~ ....28) where ex = . _ . the denominator of Eq. Observe that the calculation of the plate deflection simplifies to the solution of Eq.....7 BENDING TO A CYLINDRICAL SURFACE Assume that a /ollg rectangular plate of uniform thickness t is bent into a cylindrical sUI}ace with its generating line parallel to the}' axis.. 7.= I.2). which is of the same fonn as the differential equation for deflection of beams under the action of lateral and axial forces. The total dellectioll is then W= \V "" amI! • mnx ..168 STRESSES '''' PLATES A". (7.. it is required only to change rite sign of N in the equations of the foregoing example. For this case._... sm . I11!Y o + n·'1 = L.~.... (and thus >I'd increases with increasing N.
~ ."D sinh 2' (h) Employing the identities cosh 2/. the length of which is large in comparison with its width. 7._" i Po .. sinh 2). A typical case in which bending to a cylindrical surface occurs is illustrated below.. .9 produced depe/UliJlf} "1'0/1 the magnitude of lateral deflection w.. cosh X cosh2 i. = ~'~'C"~' 1'0 L4 1 ~ cosh . The value of the axial tensile forces N x = N is such that horizontal movement of the edges is prevented. (7.. Employing the notation (7. of the C ~ ). Example 7. _ . 2 ' 16J 2 D 4 C. ~uTTL~>c _x .31).32) twice results in Eq. .. . + sinh 2. The tensile forces in the plate curry part of the lateral loading through membrane action. strip.9).PLATES l!~OER COMBll'\'ED LATERAL AND DIRECT LOADS 169 \. __ . (7. is subjected to a uniform loading of intensity Po. t·~.. and thus = 0 at x = 0 and x = L.. = cosh 2 . SOLUTION A strip of unit width removed from a plate of this type will be in the same cond ition as a laterally and axially loaded beam or socaJJed tierod (Fig.32) We observe that differentiating Eq. _ . Determine the lateral deflection and stresses in the plate. sinh L~ + C2 cos h 2h L IV (a) Since the defiections vanish at the ends.~>__ {J(. The problem then becomes complicated.6 A rectangular plate. = 2 sinh ).31) may be written D </x" = ~ Af(x) d2w (7. 2' 16). The bending moment at any section x of the strip is described by Af(x) = tPoLx ~ 1PoX2 ~ Nw In terms of this moment Eq. (7. = 1 + sinh' i... (732) appears as follows: ZAX w= C.oL .33) the solution of Eq. The longitudinal edges of the plate afe free to rotate but otherwise immovable.. // '\x r. ___ ../.L ·~~t} poL Iz Figure 7.
. 5A4 2 = }.1")/EI. It follows that the deflection curve of the strip is IV 1(. To determine N. It follows that !y'(I_... we find that at x L/2 d2~~. cosh'\ . from Eq. as defined by Eg. Hooke's law.V2)L Et ~ ((~iV)2 dx 2'0 dx (<I) .= D d.33).. i2().(}. (N . By = O. It is observed that the displacement and moment depend upon the quantity 2.('sech . _..I~(}. M m . .):"01' _ Pol~ Icosh All _ (2x/L)] . and hence the axial forces. 1 .. the maximum deflection and moment would be \1' m. (N .1'N ) Y I Et x yields Ny = vN x and f._ 384D 5l'oL4 (e) The effects of N upon displacement and moment are given byI..) where ..) which diminish rapidly with increasing }.35) where Jl' 24 .... ~ 2'0 dx ((<llV)2 dx o= = During bending the lateral contraction of the strip in the plane of the plate is assumed to be zero. (7.) I Et x )' _.12a).(Al and f. x = N(I .170 STHESSES 1:\ PLATES AND SHELLS and substituting Lqs. E " = .2) (7. The extension of the strip produced by the axial tensile forces is.I + . i'oi!x . (b) into Eq.~i =8 POL2.) =~ (1 .())=. = . (7..scch A) Interestingly. (a).liT Sjlj) (L .36) From Eg.34).vN.34) The maximum displacement occurs at midspan (x L/2): (7. we must consider the deformations. (7.x) = (7. if there were no tensile reactions at the ends of the strip. we obtain \\. in convenient form.
The term a. /(1 +~) ~. becomes NL' ~ = ~2jj (n Introducing Eg.....11) into the above we find that ~(I ' + a)) = .2)4]2 ''"'4' = 319.2) £1.= _ . and integrating. (d). (f) and (1. (. (7.01)'/12(J . from Eqs.37) becomes . and Eq.33) and (f)...01) 384x1.. (7. (7.. (e) into Eq. and a = 1.3. SJJ1 ITS ~ L (e) in which oJ designates the ollly. 384D (5PL4)2 (il) Consider. b = 1. wherein x is replaced by the present case are tensile.2 m..= 4i(lI:~)2 "lai Finally. inserting Eqs. We have D = 200 x J0 9 (O. and of dimensions L = 1. with /JJ = I. is then and the quantity i. from (g) For the present case. In any particular case.) 2 = 2 3 [5 X 70. as the axial forces ill Eq. (e). the above is solved for Egs.09) = 1.. 3ai 1 (7.94 (0. •_ o. 3 x(1 + IX).000(1.\HlI:.. and t = 10 mm.30).. a steel plate for which £ = 200 GPa and \' = 0. The quantity> in 0. for example. (7.r:n L\TERAL A:\"() DIRLeT LOADS 171 A good {l1)/l/'oxiIllOliol1 ror N is found hy selecting a deflection curve of the form 11' = ~ aI 1 + '" .8315 x 10" Equation (h) then yields a(l + ..PLATES VNDtR CU. under a uniformly distributed load p = 70 kPa. n = midspan deflection produced by the lateral load is given by Eg.37) IX.29).83xlO .. we have NL(I .0.
4) . (~.72 MPa Plates having other end conditions may be trealed similarly.1 A steel ship bulkhead is a rectangular plate of length a and width h.F . PROBLEMS Sees. = u xp + Uxb = 77. 7. show that the governing differential equation for deflection becomes . The platt.!. and 11 = 1. is subject to uniform lateral presslire Po and uniform tensik forces N along the four edges. The plate may be approximated as simply supported on all edges and subjected to uniform biaxial compression N.3 7.1 to 7.9079)' .0 0 om )2 2[1 :::scch (39079)1 = 9503 MP' (3. d The maximum stress in the plate is therefore (fm .4 If there are body force!. (7.s rrom which a=6. Derive a general expression for the deflection surface and determine the maximum deflection wand tllflximum stress rr. ('Ix :2'~ .9079 ~ The inplane tensile stress is now readily calculated as follows: 3(1 . m = 1. 7.03 = 172.2 A simply sllpponC(\ rectangular platc carries uniform tensile force N along sides. v.r) L = EX' .)212 (i) t 4 t 4 = ~ (70 0 )( 1.\'::=: 0 and x = (Fig. F .\ for a = b. = :. 7. PLATES "'in smu.2) and hydrostatic surface pressure described by p(x. cl y) ~ po~ . = 3.1895 and IT /. ~~~~:) 2)' J (P7.36) and the corresponding maximum bending stress is given by (f h = x ~M2"'" =. a Derive an expression for the deformed surface. 7.3).~ Po (~ ..3 A structural component in the interior of a spacecraft consists of a square plate of sides a and thickness f.172 STRESSES I. aCling in the midplane of the plate. assumed to be gimply supported. (7. Determine the buckling stres~ by using Eq. 7.69 + 95.)' ~OO~H)9~(:~~~?y no~ r = 77.69 MPa The maximum bending moment is given by Eq.2 .
10 A simply supported rectangular plait! with an initial delleclion defined by IfO . = ((0 sm nx ..11") .6 Rcdo Example 7. tty ~ (/ Sill .Il A long rectangular plate with edges damped carries a uniform load oritensity Po.7 7. 7.o dcnot.8 Verify the result given hy Eq. using the methOd of finite diJ1i.5 Redo Prob. An elemental strip of unit plate width.rED LATER. and r. 7.: the \ and r direclL'c! body force~ pef unit area of the midpl<lnc (If the plate Sees... Demonstrate that the differential equation of the deflection curve is expressed by (P7. 7. 7.2 employing the energy approach 7.5.1Ia) Obtain the following solution w= Po L't + ~~~ eoth .:NDFH cm.4 to 7.3 employing the i:ncrgy method. 7... 7.7 A simply supported square plate is subjected to uniform compression loads IV x = ~ N applied to Iwo opposite sides.rcnces with 11 = 3. Use n = 3.Uial pressure IV per unit of boundary I~nglh. 7. 7. except that now the ends arc fixed and bending moment~ Mo thus occur there.\L A!\:n OIl<F.'D  Ux L (P7.CT LOADS 173 IJere F.mr:.9 Apply the tinite difference approach to obtain the buckling load for a c1:. is similar to that shown in Fig. Determine the buckling load.unped square platc under uniform bi.. cosh 16}. Determine the maximum dispJacemenr for a = h. (d) of Example 7. bent into a cylindrical surface.9.PLATES l.  b is subject to uniform biaxial tensile forces N.
developable behavior implies the absence of any defomlation. Cylinders and cones have developable surfaces.CHAPTER EIGHT LARGE DEFLECTIONS OF PLATES 8. A developable surface completeJy recovers its original fiat shape and dimensions.and largedeflection theories. we must distinguish the cases of midplane deformation into "developable" or "nondevelopabJe" surfaces. For such situations. In some applications of thin plates. Because of these large displacements (x 2: t) the midplane stretches. while a sphere or a saddle is a nondevelopable surface. the tension 174 . the maximum deflection is equal to or larger than the plate thickness. 8. 8.2 PLATE BEHAVIOR WHEN DEFLECTIONS ARE LARGE In discussing large. This is followed by the application of the energy method to the solution of problems involving largeplate deflections. 8. The largedeflection theory of plates assumes that the deflections are no longer small in comparison with the thickness but are nevertheless small compared with the remaining plate dimensions. not predicted by the smalldeflection bending theory.cJeflection behavior.4. The final section presents a numerical treatment for the bending of plates experiencing large deflections..1 INTRODUCTION In previous sections of the text. for example. The general analytical solution of plate problems is formulated in Sec. an extended plate theory must be employed. and hence the inplane tensile stresses developed within the plate stiffen and add considerable load resistance to it. The largedisplacement behavior for plates of simple form is illustrated in Sec. The behavior described is also generally valid for plates of any other shape.2. Section 8.3 describes the differences between the small. accounting for the effects of large deflections. principally to give some idea of the additional loadcarrying action. however. relatively small plate deflection is assumed (w < I). When a plate bends into a cylindrical geometry.
8. and the classical formulas are valid until yielding impends or Wmax + a. where a is the smaller span length of the plate. Consider the behavior of a square or circular simply supported plate under a concentrated center load P (Fig. In practice. smalldisplacement range to the largedisplacement range. The distribution of stress at a section through points A and B of the plate is illustrated by the sketch shown in Fig. are caused by compressive stresses in a tangential direction near the plate edges. a central tensile and outer compressive stress region develops. < t does /lot hold in cases where a simply supported plate bends into a developable surface. On the other hand.1a). In the case of a simply supported plate undergoing a general large displacement. Thus. 8. For plates of ordinary proportions.1 in the midplane can be produced only if the end supports are immovable (Sec. which increase with the distance from an edge. the limitation IV".'~ ~ 'Compression (a) it) Figure 8. At this stage.AND LARGEDEFLECTION THEORIES It is now demonstrated that all relationships derived thus far for plates bending into nondevelopable surfaces are true in general 0111)' if the maximum deflection is small in comparison with the thickness of the plate.3 COMPARfSON OF SMALL.. the plate edges will he Fee of stresses in the direction normal to the boundary. The tensile stresses. the latter stresses often produce wrinkling or buckling ncar the edges of a simply supported plate. it goes through the pure bending.:. Clearly.LARGE DlFLECTlO"':.S or PLA US ] 75 ~"'.2. 8.lb. 8. while all other points within the plate will not be stressfree. For this purpose. owing to the chal1ge il1 plate geometr)" At higher loads. While the plate is being subjected to an increasing load. normal and tangential stresses may both occur at the edges of a builtin plate.3). are the direct tensile stresses utilized fully by carrying some of the lateral load ing. only when the plate edges are clamped and fixed. as mentioned in Sec. we . 7. the edge conditions have a pronounced effect upon the magnitude of the direct tensile stresses that may be developed within them. the relationship between P and the deflection w is 1/0 10l1ger lineal' (Sec.7). 1.
. 8. 10. consider the ease of a clampededge circular plate subjected to a uniform load of intensity Po (Fig. refer to Fig.c.. According to this method. The radial elongation produced by the deflection W is found from Eq. Integration of the latter expression for the slope at I' = a leads to: (b) Determination of the value of N in Eq.dr=I dr=. (2. from Eq. 8. theory Large ". The static equilibrium of verdcal forces is expressed "by 21rrN(dw/dr) = Po 1tr2 ...176 STRESSES 1. from which dw/dr = Po r/~N. .. ~ " or Po =.2).U(dW)' I . . The partial loads carried by the membrane and the bending actions in the plate are then added and equated to the actual plate loading.14): .. a4 64D W llIax To derive the membrane solution./ deflection the'ory '. where N denotes the col/stallt tensile force per unit length. (b) requires consideration of the midplane deformations. (7. :.1) are treated separately. 1. The bending solution for the maximum deflection occurring in the center is.·"(por)2 p~a3 / .2. PLATeS AND SHELLS 2 ... 2'0 dr 2'0 2N 24N .2 employ a simple approximate appI'olleil which is in good agreement with the exact theory for the circular plate.".0 Figure 8. As an example. o Willa". the bending solution and the membrane or very thill plate solution (Sec.~:  1 del1cctioll .120) \\' as follows i:_(: 1...
5_ a 10 3 l' a ~(\'''m".. .65'1 J (g) Clearly..i1g of" the midplane of the plate produced by large deflections owing to hending. and N" do not. as be/inT.3..' A modified form of Eg.:.)J' a U) An alternate form of Eq. (1 .. y~ 8. (7. (e) yields NJ = E I ..v P6a'lt 24 (d) Finally. We observe that the small·deflection theory is satisjactdry for "'m" < t/2 and that larger deflections prod uce greater error. (f).. .. It is important to note that the inplane forces N x • N.. depend only upon the external loading acting in the xy plane.'/64Dt are plotted in Fig.. elimination of N between Egs... For a thin plate element.LARGE DEFLECTIONS or PLATE:$ 177 The strain is thercfort' F. for example.'. respectively.= p~(f2 i4N~ (e) This strain is the same in all directions. The experimental data agrees well with the result given by largedeflection theory ./t and Po . the x and)' equilibria of .2. Hence. 8. When IV... there is a 65 percent error in the load according to the bending theory alone. by taking 4 l' = 0.. the first and the second bracketed terms represent the bending and the membrane solutions.. ~ 31va (IV.4 THE GOVERNING EQUATIONS FOR LARGE DKFLECTIONS . Upon apphcation of Eqs.3) may be employed in the analysis of deformation in a plate under transverse loading resulting in relatively large elastic deflection.v)Njt = Es together with Eg.. = t. They "are now effected also by the streu'I1. 1:. To illustrate the variation of load and deflection for a uniformly loaded circular plate with clamped edge.. (a) and (e) we have Po= 64D(IVm")' +~3.f':." )' a The load actually carried by the plate equals the sum of the partial loads resisted by the bending and the membrane actions. (b) and (d) results in the membrane solution: Po =8 . IVm.)2 + 0.. is poa 64i5i = "'t "'m" [ I (IVm.
3). E).x twice with respect to y. the following familiar relationship is obtained: (a) According to Hooke's law. we thus have the third equation in terms of Nx. 1. To ascertain the values of N. and w. a third expression will he developed from the midplane strain displacement relations as follows. and N. twice with respect to x. is id@tiCally. = ___ _ Gt IX)" (I> ) Upon introduction of Eqs.12). . y). any point in the midplane experiences x.1) and (7.178 STRESSES IN PLATES A!\O SHELLS direct forces are expressed by Eqs..1 ) As the strains arc evidently not independent of one another. related to the direct lorees as follows: Ny=t..}. +~ 2 E. ax OJ' (e) Substitution of the foregoing into Eqs.!~ )' 2 ax. (7. _ viJ'~) E ax' iJy2~:t:.\ and y. respectively (Sec.')2 oy (8. When the stretching of the midplane occurs during the hending of the plate.tullction cjJ(x.\) _0' 4' E (d) i'x y = ax ily . II +~. Thus. (7:1) and (7.. the following arc the genera/midplane straindisplacement relations owing to bending and stretching of the plate r = 'x ('ix 0.. Nr . {>' cjJ N x } ' = t . ('. (b) into Eq.and N".satisfiedby the stress . This expression. J\ and z directed displacements. by differentiating f.Ny. Then resultant strain components arc found by combining Eqs.. 110 = 11. (a).2).. (1.. (ihl. and YXY with respect to . V" = v. (b) yields 8 )' = !(O'q.}.vN xl 1 N x ).... f. in these two equations.) = ily a. iJ x{>l¢ .1) and (7.2). as well as Eqs.. = 1 Et (N...
8. (7. (d) into Eq.3) are the gOl'el"ning differential equatiol1s for large defiections of thin plates.!' Only as a result of recent progress in the development of numerical approaches has the general problem of plates been treated satisfactorily.y + N. (7. Plates of this type may thus be regarded as flexible membranes having no resistance to bending action.5 DEFLECTIONS BY THE RITZ METHOD" . and Eg.) results in ?"rj) ..</) 2 2 2 ex4' Dx (~yl cyl. satisfy the boundary cond ilions. and (l. 8.(04</> ..: \ LARGE D1'. Hence. . the normal and shear stresses. (8.31). c's.2 ) and introduction of Egs.1 (N. w = w(x).2) is satisfied identically.7).3) were introduced by von Karman in 1910.~_~t ~2W _.6. x ?y . upon application of Eqs.'.Cx + Nyf..1. In concluding this discussion. the midplane stresses are obtained through the lise of Eqs.S Of PLATES 179 Insel'ling Egs.2) and (8. .'~:1~ c:x ?x .FLEClIO\. Equations (8.9): VIII = 2" 1· .6.3 ) Expressions (8. )' E {. Once the stress function is known.~. (8.'¢ (.'7. (1. partial differential equations may be a formidable task. D t rJyl t)'2 . Determination of </> and w requires the solution of these equations which must. (e) into Eq. + . that is. . If the generating line of the cylindrical surface is parallel to the y axis.14). respectively.'". are determined.2) and (8. ('\1' .r. (c). and 8'¢l8x' and <"¢Ioy' are constants. consider the bending of the plate into a cylindrical surface (Sec. + ?4W = L !!~ + ~. In this section the discussion is limited to very thin plates..3) leads to ~~_:~ + :2 _~~~:. 8~y2 (8. 7. 1. of cOllrse.'. Unfortunately. where realislic problems are concerned..8).1 and 7. The strain energy associated with the stretching of the midplane of a membrane is given by (Sec.JY!. Eq. (. as in the case of small deflection.\T}'XY) dx dy (SAa) or (BAh) .y.. those in which deflections are many times greater than their thickness and on which only inplane or membrane forces do work. (1. Knowing the deflection lV.~. solving these coupled.c.i3).3) reduces to Eg. 7.simple regular shapes. nonlinear. Dy Z:2\1"_j (8. TIlis particular case of bending has already been treated in Examples 1.'5 We shall discuss the powerful finite clement solution of large deflections of thin plates in Sec._. A number of approximate solutions have been determined for the uniformly loaded plates of ..~4 + .~ ~_~_~.i~\~I +: Zy4 = 'I( ._.
.i..1 _~_ 4 . x Figun. + 2 ox iJ.O SHELLS Upon substituting the values of th~ strains from Eqs. a. )' is placed at the center of the plate.. !(~l')' ax ax I ux +Ct.t(~. .5) The Ritz method provides a simple approach to the determination of membrane deflections..· 8.l) into the ahovc.5)...3 L· . ('X C)' Ii: w~~r + 2~t~ ~.w . Ii.. Example 8. and IV are taken to be zero at the edges. Introducing Eqs.2b 2b 1/ = Co sin . 8.. as illustrated in the following example.i T b b J. the displacement may be represented by nx 1tV w=aocos cos ~:. The expressions (a) satisfy the boundary conditions.  . i I .. SOLUTION Assuming that the origin of coordinates x. + .')· 2]. (a) into Eq.!y + 21' [~II 'ZV +1 ux a..v 2 ~u (~I.1 What are the deflection and stress at the center of a very thin square plate of sides 2b due to a uniform load Po per unit ·surface area (Fig..180 STRESSES IN PLI\TF.25: Uno = f..~v(a\V)2 ilr ely.~ + (~.IV)' + (~~)' +('. integrating.2) .S A1'.. It is also observed that the requirement of symmetry of deformation is fulfilled by lI..1. +~ + I(ill\:)'2 (~w)']' ax + 8y. au o c. + cr. and w. = Co 11X 2b where ao and Co are unknown coefficients... cos b 11y 1fX ny (a) r... (8. v.. (S.. we find that for I' = 0. cos b 2b sin .3)? The displacements II.~ 156~~ b~ 17t a. thc strain energy is obtained in the following form: u = m 2(1 __ ...': ! dx dJ' (8..w ()v aw ow]l h 8}: ii.r .(35t + to)] (b) .... v.
462 (j~ ) b':. 8.1[]lh" yield two equations. if. ' . we must now extend the formulations of Sec.802hZ/PobIEi.7) The energy method can similarly be applied in the case of large deflections of plates having other boundary conditions and shapes.")2 OW ~': I ty ax ay f l (8." = nco!b = 0. The conditions (lnpllo = 0 and . The associated tensile stress is " = C=.. To begin with. = 0.{ii 3!"rEbi (8.)2 12 0.. 8. inplane strains of an element will be represented. The maximum lateral deflection occurs at the center and is given by = 0.>: '2 . 5. On this basis.. (8.8 to include the effect of midplane plate strains and corresponding inphme stresses.f' 2~~b~ cos 21~ !ly £1.802b 3 V Et 1· .. 0.. (a).. assume an initial system of inplane forces applied to the plate and regard these as constants during bending (Sec. On applying Eqs. The hehavior of plates experiencing large deflections is illustrated in Sec.:S or PLATES 1St The work dOllc by the uniformly distributed IO:ld is W = I I poa o cos 'h "b .F DEH.4).'/2 INJ.2.8) are given by The "generalized" Eqs.. Clearly. 7.~ (.462a6/b'.396. (5. I ~ .1). from Eqs. the solution of which results in the following values for the coefficients: II o = O.'\.6 THE FINITE ELEMENT SOLUTION The finite element method (Chap.9) The stress resultants are thus comprised of col/stallt direct forces and moments: \ N.147& (H6) IV".LA!U.IOa} .17): bendingstraindisplacement relations (8.J I I ''" 1\1.~. · ipo" Co = a~ 0. 5) is here applied to determining the large displacements and stresses in a plate of arbitrary shape under general loading.N'I = t '\ err !~x I IM. f} .fCTJO".1' ) nx IT\" (e) Potential energy n = u ~ W. ( = I I I Ox '1 Or T xy " e i Z dz {8.9' e \ M. as { i" = JC' J! (O>li . E (. the tensile strain at x = y = a is found to be: Ox = c.
40). I' C"'j' cr II (pw) dx dv '.~{M}..41) for the potential energy of plates is I ..'(! I . Inplane stresses {aJ and inplane strains {i:} are connected by the Hooke's law (8.t:~ )1' "'} (' .. " 1/2 (8.10).182 STRESSES l?\i PLATES AND SHELLS or .4 .I}. Rectangular finite element The bending properties of this element are developed in Sec. (8.16) . = [S](a} Applying Eq.! (1.13) where (8.11) Here [Dj = E 1~ \. = [0 0 8W 1 ( 0 ely e 2x 0 0 3x 2 0 x 2y 0 2xy x' )' 0 3y2 3x'). . [aj'1. we have (8.' Ioy I' <1w :. fWI or 0 ih. = I t'a'((' z d.18). (5. 1 (8.15a) {tI}.IOb) The bending stresses k} and the" generalized" bending strains {o} arc related by Eq.'(3 2xy 3"v' 2 J1al>' xy '" au} (8. (8. We now treat its inplane deformation properties.dx dy  ox.14) We shan proceed with the discussion by considering the case of a specific element. 2 . (5.15b) {OJ.2 represent the elasticity matrix jor plane stress.. (5.t{a'Je ~ '. t "..12 ) o I 011')' row'l 1 II {'}.II '. .42). 5. The derivatives (slopes) of 11'. Inasmuch as the direct and bending strains arc taken to be indepmdellt. dx dy + . = [Sj[ Cr lIb}. the expression (1. = [CH." IT = 2 A' .. arc from Eq.1\1'1.
(8. (8. 5. = [G](aL. 5. (8. Apply the procedure (steps 1 to 3) of Sec. The [k y ].16) . at the centroid) from Eq. into n =2 1" A .Q . 1/" ....J[sYi~ ~][S]dXdY)[CIJ (8.20) is known as the initial stress or geometric streSS matrix. = ". (5. = [k]. (5) Compute the geometrical stiffness [kG]' from Eq.. (8.21) [kGxl'=I[C'Y(. (3) Compute the inplane strains {tie ji'om Eq. Appropriate definitions of the smalldisplacement stiffness matrix [k]. The new term [kG]' = I[C 'l' OJ [S]'[ii][S] dx dY)[C I] A (8. .. [kG./.:.g.1...8.. are given by Eqs. and nodal force matrix {Ql.. dx dy + 2: [(5). respectively. (6) Compute the element total stiffness matrix from Eq.[kG.18) As in Sec.1. (8.[k G. .11): {ul. designates (8./ (I'll') dx dl' ·1 ' A (8.8 to obtain [k]" {Q}" and the hence small. wherein [kG. (8.}. (8.8).~r:..]' and [kG"]" may be written in a like manner. is therefore = [S][C]I upon introducing ). = [D]{t).22) Relationships for other components of the geometric matrix.23) and (5.]" for instance.Jeflection solution for the nodal displacements {Ii). The generalized procedure for solving a plate largedeflectiol1 problem is summarized as follows: (1) Assume the initial direct stresses {u} to be zero.13). is termed the total stiffness matrix of the element. expression.LAIWE lWfLE:CTIONS OF PLATES 183 in which [G] The potentialenergy Eq. Equation (8. (8..[k G)].21) in terms of the given element properties.1 {. (4) Compute the inplane stresses from Eq.16): {(i).19): [k 1 ].I.24)..20) may be expressed in the following convenient form [kG].19) These are modified equations 0/ equilihrium of nodal forces. it can be demonstrated readily that the principle of potential energy now yields (8.)' + ii. + [kd.]. Eqs.' [G]'[u][G] dx d)' [bj. (2) Compute the slopes at some representative point within the clement (e.. + !.17) It is observed that [G] is a matrix defined only in terms of the coordinates..
~4 (~~~ )']r dr 1 d.. 1\') is found upon replacement of w by tP in the above expression.3 Show that the system of equations (8..2 Apply Hooke'~ law and Eqs. is expressed by U = '!:~" f r(~~~)2 + . . (dW)' I\.~.. respectively.!) Here II and ware the /'.4) J' for a very (hin axisymmetrically landed circuhu plate.2 (P8.[~ + 1'du + . <p) + D and I.3) V " '" ~ where E P D L(w. 8.and zdirectcd displacements. due to the stretching of the midplane. the strainlargedisplacement rclations arc (P8.. PROBLEMS Sees.= iL(w. IS attained..l to 8.S (7) Repeal SICPS (I) 10 (4).1 Verify thaI for axisymmetrically bent thin plates.~~ (~~)2 + ~~~ + 2vu ~~ + .2) J' t dl' 2 d. 8. assumes the form: '1144.2) and (tU). each lime with" new I"rl" found by applying SlepS (5) and (6) until satisfactory convergence or the inplane ~trcsscs {iT}.1) to derive the following stressdispJacement relations for axisymmetricaUy loaded plates: a. The results determined by applying the finite element method to the large detlection orthin plates agrees weH with analytical solutions 1S It is interesting to note that some classical buckling problems of plates may also be treated by the approach described in this section. for axisymmetrically bent plates.184 ST[{ESSES IN PLAnS A~() SHELl.~. S.{w. w) (P8.6 8. dr til' r* r til' ell' dr m V· (PS. ~ .. 8.4 Verify that the strain energy. (P8.I! (~_~\_')2 +.
ex is expressed in meters per meter per degree Celsius.nentsarepi·ev"l1t<:<l ..'.iiifol. Here T and To are the final and the initial temperatures.. an experimentally determined material property. Coefficients of thermal expansion for common materials are listed in Table 1.i{i. there are produced normal strains but no th~. the modulus of elasticity E and coefficient of thermal expansion '" are treated as constants over the temperature ranges involved. a change in temperature 8T = T ~ To produces uniform linear strain in every direction.al siress"s.ifthe. = "'(ilT) (9. For homogeneous isotropic materials.owe~er.io~i.1 INTRODUCTION Solution for the deflection and stress in plates subjected to temperature variation requires reformulation of the stressstrain relationship.n. = O...1.". remains reasonably constant. thermalstresscswiiI occu" . [n this text..e.te . or .th~'pj. For isotropic materials. [n equation form. respectively. Stresses owing to the restriction of thermally induced expansion or contraction of a body are termed thermal scresses.jlg fr""lybecause gf therestrictionsplaced on the boun~ary even with a uniform'temperature. or if the material displays anisotropy even "iiih uliiform healing. When a free plate is heated unijormly.disphice.'leiTI]Jer. In SI units.1 ) where Ct. "I. fj'oni"()cc~rr. the thermal strains are expressed as c. i.expcriencesa.tuie·field. a change in temperature produces no shear strains.CHAPTER NINE THERMAL STRESSES [N PLATES 9. 185 . This is accomplished by superposition of the strain attributable to stress and that due to temperature. is termed the coefficient of thermal expansion. Over a moderate temperature change•.
the first terms represent the strain components in the midplane of the plate. and denoting 110 = 11 and "0 = 1'. = ··'~2· I ..3 into Eqs. with the exception of (2). y) represents the transverse deflection. In deriving the straindisplacement relationships. 9.3) An increase in temperature LI. = .2) " I XJ' From Eqs.. Substituting Eqs. the strains due to stress resulting from external forces: t. 1..2 STRESS.T)] (9.la) to (Lie).I' = i}y (9. and w(x..v It. (!.2)." = au (8y DU) + ax  20 ax'dy . . T is algebraically positive. (b) of Sec. 1 E 1 (ax .1'0.) + a(A T) . Al\"D DlSI'LACEi\IENT RELATIONS The total x and y strains. It is further assumed that straight lines.4) 1'. a state of plane stress is assumed. (9..'u f:. (1 + l')et(Ll. the total stress components are given by V" . xl + ". (. ar~ obtained by adding to the thermal strains. + ""x" ..2. STRAIN. Thus. we obtain the following expressions for the strains in terms of displacements: l\ = . 1. initially normal to the midsurface.(1'17) (9.. remain straight and normal to that surface after heating.x and [. apply. .)'w In the foregoing. = E (G y  . a~~2 (1V cu ?2\V s.{x' :. the assumptions of Sec.186 S'fRI:SSf:S J'\ PLATES A~D SHELl'.
The latter quantities are given by Eqs.!1:_) _ ~(A1] t 1. 7.5) Introducing Eqs. I IN:'.r. (9.3) and elimination of the displacement derivatives through the use of Eqs. (9.'o'1 . (9.I I ' u y . + _1\_'*_. (9.'l (ATJdz M' = aE • ~ri2 i .9) and (9.. the stress resultants are obtained: . d:: ·"I.I.4) into Eqs.. t 1 l' 1 l' rr }' =! (N.6): O"x = ~ (Nx t.E "t/2 I .At xy 12: .3) and (9.5.9).) + _12_2 (M + _1\_1_* ) _E:~(ATJ 3 t } Iv Y 1v 1~1 (9. The components of stress may now be determined by substitution of Eqs." I 0.8) r xy = "f N xy 1 + [3'.Z (Mx + . (1.'2IT''.1 and 1.6) Here the quantities N* = (I.THERMAL STRESSES l~ PL\TES 187 9. ~) t~. The inplane force components arc represented by ) Ny~' = jN. (1.x I (9.7) arc termed the themw! stress resultants.4) into (9.3 STRESS RESULTANTS The stresses distributed over the thickness of the plate result in inplaue fim'''s and mOlllents per unit length as shown in Figs.5). 1 (Ai)z dz (9.v t.
10) that it is possible to superimpose the deflections owing to the temperature alone with those owing to transverse loads alone. 1.I') in the upper half and tensile stress +~E(LiT)/(1 .\' (9./i)/'C('s and bend iJlg moments. respectively.9) and (9. (7. similar to that employed in Sec. y). <72 M * 1 ~ \' It is usual to denote (9. It is observed that Eqs. and bending stress in the case of nonuniform heating. (9. Eg..y. The twodimensional equilibrium and compatibility equations (in the sy plane) are employed to obtain the forces N as shown in the outline that follows./ellt transverse load.6 and the modilled moment resultants of Eqs.10) p* = . An example of the latter is a plate with its upper surhlce heated and lower surface cooled. (9. To derive the stress equations in polar coordinates.11 ) where p' is termed the equh'..1') in the lower half. (9. derived using the procedure described in Sec.fhr d{~flectiolJ \\'. one need only replace subscripts s by /' and )' by 0 in these expressions. are combined to yield I D l1"w = p .. related to the force resultants as follows: N x )' = (9. These expressions are identically satislled by the stress function </1(x.8) are of the same form as those associated with beams under compound loading. there will develop compressive stress . 11 2 M' 1 . y).9) reduces to " 1 v Dl1w= . The third term represents inplane stress in the case of uniform heating. i.9) For the case of a nonuniformly'heated free plate.188 STRESSES J:\ PLArES '\!'\[) SHELl. (9.8) permit the direct calculation 01" the stress components fot: u platt: of any cross section subject to all arbitrary temperature distribution.. \' (9.4..E(AT)/(1 . The first and second terms of Eqs. At some location (x. 8.e..4 THE GOVERNING DIFFERENTIAL EQUATIONS The gor:erniny eqllation.2).8) are due to the inplane .. the equation of compatibility is (a) .1) and (7.6). (9. p = 0. 9.S Equations (9.12) For a plate of constant thickness and negligible weight. We observe from Eqs. The differential equations of equilibrium in the plane of the plate are given by Eqs.112M' 1 ._1_.
' 9. the determination of the transverse detlection.2' Desigll data and other useful information dealing with thermal stresses are presented in the papers of Goodier 28 and others.= I) fl. the solution of a thermoelasticplate problem requires the solution of Eq. (1.5 SIMPLY SUf'POIUED RECTANGULAR PLATE SUBJECT TO AN ARBITRARY TEMPERATURE DISTRIBUTION This section deals with the deflection and stress owing to nonuniform heating of a simply supported rectangular plate (Fig. The boundary conditions at edge x = a of a rectangular plate (Fig. several methods are described.ATES 189 Tablr 9. useful in the solution of thermal belldillg problems of elastic thin plates. The second problem.25) through (1. is the direct cOllcern of plate theory. The approach introduced may be extended to all polygonal simply supported plates. y = b) (a) . 1..1 Edge Clamped 1\' = () Simply supported Free Al _x = (/ ill\' . and is not treated in this text.\ «2W "2. 3. In the sections to follow. are nOw expressed as shown in Table 9.1) =0 w= II' ° Mx =0 My = 0 (x = 0. The first problem is a concern of the theory of elasticity.13 ) What has been accomplished is the formulation of a plane stress problem in tlIermoeleasticity in such a way as to require the solutions of a single partial differential equation (which must. A number of references are available for those seeking a more thorough treatment"6. These two solutions may be obtained independently of one another and the stresses owing to each added if the total stresses are required.27)]. (9.L\L STRESSES I\: Pl. of course. Other kinds of boundary conditions may also be obtained by employing the stress resultants derived in this chapter and the procedure presented in Sec.TIIER/o.7). 1.7. x = a) (y = 0.12) into the above yields 'il"qJ + 'il'M' = ° (9.  I'}D Substitution of Eqs. In summary. (9. (9. fulfill the boundary conditions). [Eqs.9) for the deflection wand corresponding moments.) = AI'"  (I.1).1.13) for the midplane forces and Eq. We observe from the table that the boundary conditions for a simply supported and a freeedge thermoelastic plate are nonhomogeneous. The boundary conditions are represented by (Table 9.
190
STRESSES IN PLATES A:.:n SHELLS
From Etjs. (9.6) and (a), the following expression applies to the boundary:
. 1 fH* DV'w= _ ......... .. I\'
(b)
The governing equation, Eq. (9.10), is equivaleut to
D 17'",
+ ' ..
\1* = fix, y) 1  \'
17'/= 0
(c,d)
Equations (b) and (e) result il1/= O. which is an appropriate solution of Eg. (d). The problem at hand. as in Sec. 7.5. is thus represented by a secondorder differential equation
M* D 17'w =  .........1 ,'
(9.14)
which must of course satisfy the boundary conditions
1\'=0
(x=O,a;y=O,h)
(9.15)
The solution is obtained by the application of Fourier series ror load (moment) and deflection:
M* =
I
W
m=l
"p 1I~1 mit
~
'Jj
•
mIT>':
.
l11CY
SIn .......... Sin ........
a
b
(9.16a)
(9.16b)
HI
=
Ia mlt sm   8m b 111'" 1 11'" 1 a
L
00
•
mITX
.
11IT}'
The coefficients Pm" are. from Eq. (3.3):
Pm" =
4 I. 1M* SIn ab o·
,I>.'!
0 .
.
nt1L\:
II
.
nnr
h
_ . SIn _......
dx dy
(9.17)
Substitution of Eg. (9.16a), (9.17), and (9.16b) into Eq. (9.14) leads to
a = _1.... Pm" , "'" (1  ,')rr 2 D (m/a)' + (n/b)'
(9.18)
The det1ection w corresponding to the thermal loading M*(x. y) has thus been determined.
9,6 SIMPLY SUPPORTED RECTANGULAR PLATE WITH TEMPERATURE DISTRIBUTION VARYING OVER THE THICKNESS
The solution of a simply supported plate subjected to nonunirorm heating such that the temperature varies through the thickness only, T(z), can readily be obtained from the results of Sec. 9.5. In this case, the thermal loading At* is
THER~IAL STRESSES IN PLATES
J9J
cOllstant aod Eg. (9.17), after illlegratioo. leads to
4M* p"", =  [1  (1)"][1 (1)"]
1[211111 .
(a)
Substitution of Egs. (a) and (9.18) into Eq. (9.16h) yields the following expression for dellection:
l\'
= ... __. _   )',. _. __ ..... ...  ...... 
16M'
~
•. sin (mnx/a) sin (nnJ:/b)
(I  1')Dn
4
";;;";;'
1Il,,[(m/aV + (Il/by]
(111,11= 1,3, ... ) (9.19)
The bending moments and stresses in the plate may now be calculated from Eqs. (9.6) and (9.8), As already noted in Example 3.2, while the expression for deflection (9.19) converges very rapidly, the relationship for moments does not. An alternate solution 26 of Egs. (9.14) and (9.15), more suitable to the computation of moments, may be obtained by the usc of simple series for It' and M* rather than the double series as before (Sec. 3.4).
9.7 ANALOGY BETWEEN THERMAL AND ISOTHERMAL PLATE PROBLEMS
We now demonstrate that an analogy exists between the thermal and isothermal platebend ing problems, serving as a basis of a convenient procedure to determine the deflection. The analogy is complete ollly for the determination qf dellectioll. The thermal stresses are ascertained by adding «£(67)11  I') to the stress components u., and (J" of the isothermal solution. Plates with clamped edges The problem of the bending of builtin plates as a result of nonuniform thermal load requires the solution of Eq. (9.10) together with the specified boundary conditions given in Table 9.1. Note that the boundary conditions for a clamped edge do not involve expl,icitly the temperature, Thus, it is observed from a comparison of Eqs. (9.9) and (9.10), that the solution sought is identical with that for the same shaped clamped plate subject to the equivalent transverse load p*. The thermal problem is therefore reduced to an isothermal one, and the res nits and techniques of the latter case are valid for the problem under consideration. Table 9.2 provides a list of some examples. Table 9.2 Plates with clamped edges
Geometry
Rectangular
Loading (p*)
Uniform
Solution
Sees. 3.7 and 3.12 Sec. 2.4 Sec. 9.8 Sec.9.S Table 2.3
Circular (solid)
AnnuJar
Uniform
Radial Radial Uniform
192 snU':SSFS r:.: PLATES A:iD SHELL!;)
In the case a platC' tlrhirrarr .')I/(IIJ(, undergoing thermal variation f/Jroltyh tire (hie/dies:) only, we have V2;\J* = 0 and fl::: = o. According to the analogy.
\I'
or
or
= 0, and the corresponding stresses, from Eqs. (9.~n are
(0 )
N.\T Txr=
. t The first and the second terms represent the plane and the hendinostress components, respectively.
Plates with simply supported or free edges An analogy also exists between heated and unheated plates with other than clamped supports. In this case a modification of the edge conditions is required inasmuch as they contain the temperature. At a simply supported edge of the analogous isothermal plate, IV = 0 as beforc, but a bending moment M*/(l  v) must be assumed to apply. In a like manner, at a free edge of the analogous unheated plate, a force equal to (Nv[*lcx)/(l  ,,) must be applied. It is thus observed that a thermal solution can always be determined by superposition (~f various isothel'mal solutio1ls. Consider. for example. the bending caused by a nonuniform temperature distribution of a simply supported plate. The deflection of the plate is determined by adding the deflection of an unheated, simply supported plate subject to the surface load 1'*. to the deflection of an unheated plate carrying no transverse load but subject to the moment M*j(l  v) acting at its edges. The foregoing analogy is also useful in tbe experimental analysis of elastic heated plates. This is because it may be easier to test a plate at constant temperature with given transverse and edge loadings and then to impose upon it arbitrary temperature distributions.'9 Example 9.1 An aircraft window, which can be represented approximately as a simply supported circular plate, is subjected to unifoffil temperature T, and uniform temperature T, at the lower and the upper sLirfaces, respectively (Fig. 9.la). Determine the deflection and bending stress if the plate is free of
p*::: 0
il/*
M*
~lv
F
(a)
fht
(b)
A
+
CE===AJ
(e)
Iv
Figure 9.1
T!lm~'IAL SrRf.SSES E": PLATES
t93
stress at rye. Assume that linearly and lhal T, > '/~ .
th~
kmperature through the! hicklll'sS \'aries
SOLliTlOX The plate of Fig, 9.la is replaced by the plates shown in Figs, 9Jb and 9.1 c. The temperature difference between the faces is T)  T, and that between either face and the midsurlace is !J. T = !(T) + T,), Since for the present case M* is a function of z only, Egs. (9.11) give an equivalent loading p* = 0, The thermal stress resultant is, from Eq. (9,7),
11'1*
=
·xE ( ' f.HT) + Ii) + HT,
'1:2
l

'[i) ';2If
!,
dz =
c'!2C~ (T) 1
 T,)
 7;)
(9.20)
For the plate of Fig. 9. I h, the bending stress at the faces is
6,.
"E, = a" =  "E(t<. T) = '''' (7) _."'I  l' 2(1 _. ,,)
(b)
Also, there is no deflection of this plate (w = 0), Tn the case of the plate in Fig, 9.lc, introducing Eq. (9.l0) into Eq, (2.27) and setting b = 0, M) = 0, and Ai, = iV!' /(1  1'), we have
IV
=
2'i
a(a'  1")
, (7,  T,)
(921a,b)
M, =
M* M, = T~~ =
'12(1':::0 (T)
aEr'
 T,)
The bending stre>ses at the faces of the plate, upon introduction of Eqs, (9.20) and (921b) into (9.8), are as follows:
a, =
(T,
= 2 «(=:,;'j (T, 
~E
T,J
(e)
The resultant stress in the original plate is obtained by addition of the stresses given by Egs, (b) and (e):
This is the result expected, Equation (92Ia) leads to the relationship
"'",,, =2'1 (7;
rxa 2
 12)
(922)
for the maximum deflection of the original plate,
9.8 AXISYMMETRICALLY HEATED CIRCULAR PJ"ATES
Consider the bending of an axisymmetrically heated circular plate having simply supported or clamped edge conditions and subjected to temperatures varying with the /' and z coordinates, T(/', z), such that the equivalent transverse load
194
STRESSES IN PLATES AND SHELLS
1'* = p*(r). The expressions for moments, Eqs. (:~.9). for the situation described hecome
M,.=
_D(d2~ + \~'.':) _
d .. .. <I..
1
M*
v
M, = D(! ~'.v U r WI'
(9_23)
l'
+ "il"i")'
dr
__ J11* 1=
Thc plate detlection must satisfy the difTerential equation (2.10b)
~ ;" 1.. J;: [!:~, (r ~:~' )JI
with the boundar,l' conditions
I;
(9.24)
W=o W=o
dw
dr
 =
0
(9.25 )
D (~2", dr 2
+:' ~\\,) + !'!~_ =0
I'dr
1 v
Equations (9.25) refer to clamped and simply supported edges. respectively, They are derived by employing a procedure identical with that described in Sec, 1,7 together with Eqs. (9.23), The general solution of Eg, (9.24), referring to Eq, (2.13), may be expressed
w=c , ln
h +c,r'ln b +C3(1,2
r
r
b 2)+c,+wp
(9.26)
where the c's are determined from the boundary conditions, Note that inner radius b = 0 for solid plates, The outer radius of the plate is designated a (Fig, 2.5), The particular solution. denoted by w,' in Eq, (9.26), can be determined by four successive integrations of Eq, (9.24):
Wp
=
'b
I I r

.r
1
,,'I
1'3/ ~
,']
1
4"b
'bJ2'"b
I
.1'"21'
_'~ dr t
D
p*
dr2 dr3 d1'4
(9.27)
In the case of anannula .. plate under a temperature distribution such that p* = constant. Eq, (9.27) appears as follows (Prob, 9,9):
w =  p* [1'4 • MD
5b 4

4b 2 (b'
I' + 21.2 ) In  + 4/,',,2.
b
I
(9.28)
When b = 0, Eg, (9.28) yields the particular solution of a solid plate, For the sake of simplicity in the representation of the results, let
( .. =
a)
(9.29)
The values of constants are listed'6 in terms of the foregoing notation, (Prob, 9,6) in Table 9,3, Tn this table, represents the value of the thermal
At:
THERMAL STIU'SSES C'''; PLATES
195
Tahle 9.3
G':'Ol11erJ~
COJl~t:lnlS
in Ego
(9.~(')
A. Solid plate (damped at
.:dg~l"=aJ
CI=C~=O
H. Solid plulc
(suppor(~d
at
IC'dgc!"
_= Il)
C'. Allllular plntc (damped at inner r = Ii and outer r = a edges)
C,

_all'~l)11l2 _ /)2 .•. '21)2 III (1/1>)] .,. 2(n~ _ [,2)11';;>' = .....  ........  .........  .. . 
(a l
_
b I )l
_ 4.111[>2
In 2 (a/h)

(/\~.~IJf({/l _
b2 ) tn (a/b)] + w~(I·fb2
{/2 
211 z In (a/h)J
c.l
~"' ····~(;:;i~·i,if~·4~iPIr1·(~ibT··
stress resultant at ,. = u. Given a temperature distribution T(r, z), the deflection and moment in a solid or annular plate with simplY supported or builtin edges can thus be obtained from Eqs. (9.26) and (9.23). Plates with other edge conditions can similarly be treated. Example 9.2 Redo the problem discussed in Example 9.1 using the relationships developed in this section.
SOLUTION
As p* = 0, we have, from Eqs. (9.29), to row B of Table 9.3,
=
w:;') =
0 for alln. Referring
6111'
= £1 3 '
2(1=,;2)15
M:
where I'v1: is replaced by M*, because the thermal loading does not vary with radius. Equation (9.26) then yields
))' =
~~1" 3
£1
(a' _ ,.2)
This expression with Eg. (9.20) results in a solution of w, identical to that found in Example 9.1. Upon inserting Eqs. (9.21a) and (9.20) into Eqs. (9.23), we find that At, = M. = O. Applying Eqs. (9.8), we again obtain Or = a" = o.
3 then gives C1 = C2 = C3 = ('4 = 0 Hence. 9. PLATES AND SHELLS (0 ) (b) Figure 9.Eqs. (9.8). 1_ v (9. from Eqs. The thermal stress resultant At" is defined by Eq....23): M. = M" = O.30) and (9. To 1. 1. Eq.26) yields \\.7~) By means of Eqs.2 Example 9. __ . aEt 2 E~ = cr. linear temperature variation between its faces (Fig. Iv 2(11') (9. Row A of Table 9. (9.20) and the bending moments are found from .T2) .30) (1)) We now have I1T= (1. (b) the plate experiences a steadystate.31) may readily be found without any calculation by the use of analogy between heated and isothermal clamped plates.. SOLlTTION (a) have p' = The uniform temperature differential is I1T = T.2a). The displacement >1'= 0 as before./.)/2 (Fig. we have rJ. Assume that the plate is free of stress at 0"(' and that T1 > T2 . (9.2b).To..23) then lead to At. (9. The magnitude of the bending stresses at the plate surfaces.. is therefore a = r (fu £0:(111') = _. It is noted that inasmuch as p* = 0. . 9.:2 0 dz =0 . .(T1 . 9.8). We 0 and thus W~' = 0 for all /l.=0 (a) The thermal stress resultant is Ai" = a£(I1T) I . = E~(I1T) = .196 STRESSES 1. __. .2b).3 Determine the dellcctions and stresses in a circular plate with clamped edge for the following cases: (a) the plate is stressfree at temperature and uniformly heated to a temperature (Fig. = Mil = 12([~~ (T1 . ·/:2 Expressions (9.31 ) for the magnitude of the bending stress at the upper and the lower faces of the plate. (9. solutions (9.
Assume that the panel edges arc clamped.7 Determine the deflection wand the stress aT at the center of a simply supported ell'cular plate experiencing a temperature lleld A.\'2. 3. subject to a temperature field Az J .md the cooled (lower) surfaces are 54 and 24"('. 9. 9.8 Redo Prob.3. 3. 9.13). Sec.3 Determinc the maximum deflection II' and th\. 9.! maximum moment Alx in a simply supported square plate or sides (I. taking m = /I = 1 to obtain an t'xpression for [he deflection surface It.2 Derive expressions for the bending moments in a simply supponed square. where B is a constant.9 Determine the displacement II' at r = 2a of a hollow plate having inner (r = (I) and outer (r = 30) edges built in.70 GPa. The plate is subjected to a temperature distribution A.9a).2 X 10""6 per °c.plate of sides temperature distribution linearly varying over the thickness olily. Use E =. Assume a femperature distribution 8Z. respectivdy. 9.\' = (/.". .7 9. y) (Fig.6 Verify the re~uJts given in Cases A and R of Table 9.8 9. Arter a time al cruising speed. \' =.:\ where A is a constant./"2.0. The remaining edges at y = ±a/2 arc clamped (Fig.6 and 3. Employ the analogy between isothermal and thermal problems and apply the Ritz method..:r 3 . Compute the thermal slI'ess resultants for a linear temperature transition. ~ = 23. 9.5 A square plate is simply supported at . and t = 6 mm. 9. square plate of sides a. 9. the temperatures on the heated (upper) . where A is a constant 9. Use the analogy with the is{)tiJermaJ solutions given in Sees. where A is a conManl.THER\IAL STRESSES '" PLATES 197 PHom.1 An aluminum airplane wing panel is assumed to be stress~free at 20°C.EMS Sees.7.1 109. under an equivalent tralls\'ers~ tbermalloading p*(\:.3. Derive an approach to the evaluation of the center displacement II'. (I with 9.'(=0 and.7 for a plate clamped on all edges.3.4 Consider a built·jll edge. Retain only the first term of the series solutions.
To describe the shape of a shell. commonly applied theories. the plane bisec'iIng the shell thickness is called the midsur/ace. the general equations for thin shells are also applicable to curved plates. A' membrane. a membrane is a two198 .12) to shells of cqpstant . The analysis of shell structures often embraces two distinct. As in the treatment of plates. pipes. We now extend the discussion to curved sw:face stmcll/res termed thill shells.smallincomparison with the other two dimensions. but incapable of conveymg moments or shear forces. We shall limit our treatment (except Sees. either flat or curved~ is identified as a body of the same shape as a plate or shell. usually applies to a rather large part of the entire shell. Shells of technical significance are often defined as thil1 when the ratio of thickness t to radius curvature r is equal to or less than 1/20. airplane wings. automobile tires. Inasmuch as a curved plate can be viewed as a portion of a shell. missiles. our concern was with the analysis of thin flat plates. this ratio may be 1/1000 or smaller. the membralletheol'Y. roof domes.11 and 11. and a variety of containers.thickness. the exterior of rockets. The tirst of these. incandescent lamps.1 GENERAL BEHAVIOR AND COMMON THEORIES OF SHELLS Until now. For thin shells of practical importance. we need only know the geometry of the midsurface and the thickness of the shell at each point. Examples of shells include pressure vessels. factory or car sheds. Each of these has walls that arc curved. In other words. 11. caps.CHAPTER TEN MEMBRANE STRESSES IN SHELLS 10.
reinforced concrete. (2) Deflections are small compared with shell thickness. It is important to note that membrane forces are independent of bending' and are completely defined by the conditions of static equilibrium. elastic. the bel/dillg theory or gel/era! theory. metaL fabric.MEMBRANE STRf:SSES 1:': SHELLS 199 dimensional analog of a llcxible string with the exception that it can rcsi~l compression. Thus it penn its the treatment of discontinuities in the stress distribution taking place in a limited region in the vicinity of a load or structural discontinuity.resisting action of a shell differs from that of other structural forms is underscored by noting the extraordinary capacity of an eggshell or an . soap film. This hypothesis implies that strains I" and '. Tile second.g. however. Various relationships developed for bending theory in Chaps. B" owing to transverse loading may also be omitted. The bending theory generally comprises a membrane solution. but rather the analysis of stresses and strains owing to the edge forces or concentrated loadings. the membrane theory applies to all shells made of any material (e. corrected in those areas in which discontinuity etTects are pronounced. However. isotropic shells. remain plane and become normal to the de{imlled midsurface after the shell is subjected to bending. includes the etTects of bending. slope. Applications are also presented of the governing equations of the membrane theory to specific practical cases. That the Ioad. thin plates. For thin shells having no abrupt changes in thickness. The basic kinematic assumptions associated with the deformation of a thin shell as used in smalldetlection analysis are as follows: (1) The ratio of shell thickness to radius of curvature of the midsurface is small in compari. restricted to homogeneous.'" are negligible. As no material properties are used in the derivation of these forces.1). gridwork shell. however. 11 to 13 arc. or curvature. sandwich shell. (4) The zdirected normal stress. The former are also far simpier to calculate. and thin shells is illustrated by the unified set of assumptions (Secs. In this chapter. rI" is neglible.2 LOAD RESISTANCE ACTION OF A SHELL The common defonnational behavior of beams. 10. we consider only shells and loadings for which bending stresses are negligibly small. The goal is thus not the improvement of the membrane solution. Nonnal strain.on with unity. and plywood). which cannot be accomplished by the membrane theory only. 1. (3) Plane sections through a shell taken normal to the midsurface. information relative to shell membrane stresses is usually of much greater practical signHicance than the knowledge oftbe bending stresses.. The loadcarrying mechanisms of these members do not resemble one another. the meridional stresses are uniform throughout the wall thickness.2 and 10.
this despite their thinness and fragility. or . represented by an infinitesimal element. is £=(awr)= (11')"'E 1 pI' 2EI (b) = Associated with this strain. Based upon the symmetry of the shell and the loading. thus tfr = 1/50.1). subjected to a uniform pressure of intensity p (Fig. consider a part of a spherical shell of radius I' and thickness t.:n SHELLS FigUrl' 10. load is sustained by the midsurface.200 STRESSES IN PLATES A:. (A hen's egg has a radius along the axis of revolution I' = 20 mm and a thickness t = 0. from Hooke's law. the reduced circumference is 2".' r' = 1"(1 + . can develop inplane forces (thrusts) to form the primary resistance action in addition to those forces and moments existing in a plate or beam. The condition that the sum of vertical forces be zero is expressed: 2m' o N sin <Ii P"/'6 = 0 or N".. The compressive direct stress has the form (J= PI' 21 (a) The stress normal to the midsurface is negligible and thus the direct strain. being curved. To describe the phenomenon.4 mm. It is next demonstrated that the bending stresses play an insignificant role in the loadresisting action. the state of stress at a point. is as shown in Fig. in contrast to the case of plates.) The above behavior contrasts markedly with similar materials in plate or beam configurations under lateral loading. 10. 10..1 electric light bulb to withstand normal forces. The variation in curvature X (chi) is thus 2.).(1' + te). A shell. We note that.. isin¢ pro PI' 2 in which N is the inplane force per unit of circumference. asN is seen not tovary with</>. This relationship is valid anywhere in the shell.1.
ME~fBRANE STRESSES II\: SIII:I.S 201 Neglecting higherorder terms owing to their small magnitude) and introducing Eq. for example. Treated in greater detail in the sections which follow is the state of membrane stress in shells of various geomet ry. For the spherical shell under consideration we hal'e X = Xx = I. Consider. The foregoiug discussion relates to the simplest shell configuration. the above expression hecomes " x= . However. shearing stresses as well as the membrane and bending stresses will be present. For reference purposes.. The value k ~ 0. a concrete or masonry shell for which E = 20 GPa and t/a = 1/500.25 X 20 X 109 (s''''') = 10 MPa . such as described in Sees. We have an = 0.25 is often used. 13.7 and 13. and the value of constant k can be determined by rational analysis.I._a (n Here E is the modulus of elasticity. 13.tressing of the shell.)'.\')1' 2£1 (e) A relationship for the shellbending moment is derived from the plate formulas.8. Thus. Eqs.10) and (e) lead to !vi = D(Y.. It is noted that thin shells may be vulnerable to local bucklillfl under compression stresses. (1.I'  (1 . we now introduce the following relation 30 useful in the prediction of the critical stress an for local bucklillg of a thin shell: (J cr Et = k . We are led to conclude therefore that the applied Load is resisted predominantly by the illplalle . I_X + vY ) I_y = D(l  P I' ') . the conclusions drawn with respect to the basic action apply to any geometry and loading at locations away from the edges or points of application of concentrated load. and incorporated with an empirical factor in order to relate theoretical vallles to actual test data. 2Et = Hence. The problem of shell instabilit)' is treated in Chap. the bending stress is given by P 4 The ratio of the direct stress to the bending stress is (T (d) 2r 1 (e) "b It is observed that the direct or membrane stress is very much larger than the bending stress as (1/21') '" 1. (h). Should there be asymmetries in load or shape. .
are related inasmuch as "0 = 1'. Examples include the sphere. 10. and . are thus related to sides CD and Ae. The two radii 1'0 and 1'. Figure 10. compared with the llltimat~ strength of the concrete: of 21 JVIPa.2).2 shows that a point on the shell is conveniently located by coordinates 0. sin ¢ dll (a) It is assumed in the above description that the principal radii of curvature of the shell r 1 and /'. cylinder. at any point on the midsurface of the shell are the meridian plane and the paraliel plane at the point in question. Shell structures should thus be checked for the possihility of buckling in compressed areas as well as for yielding or fracture in those sections subjected to tensile forces (Example 1O. along with various relationships of differential geometry of a surface. dcmoll~ stratcs the importance of buckling analysis in predicting allowable load.4.1!>).2 . sin 'P (Fig. The midsurface of a shell of revolution is generated by rotation of a socalled meridian curve about an axis lying in the plane of the curve. and cone.'. The radii of curvature 1'. the radii are computed applying the equation which defines the shell shape. The planes associated with the principal radii of curvature /'.202 STRESSES J:\ PLATES AND SHELLS This. It follows that lengths of the curvilinear shell element are LAC = 1'0 dO = 1'." as illustrated in Example 10. The principal radius 1'.3 GEOMETRY OF SHELLS OF REVOUJTION Consider a particular type of shell described by a s//Iji/ce of revolution (Fig.2). are known constants. generates the shell surface in the direction perpendicular to the direction of the tangent to the meridian curve. 4" 1'0' and that elemental surface ABC D is defined by two meridians and two palYlllei circles or l'alYlliels. 1O. respectively. Figure 10. and /'. In the case of a radius of curvature which varies from point to point. 1O.
4 SYM:'IfETRICALLY LOADED SHELLS OF REVOLUTION In axisynunetrical problems involving shells of revolution.MEMflRA!"'E STRESSES lr...ro + Nor. producing the iollowing component in the z direction Nor. = 0."01'. Prescribed by the condition of symmetry. The zdirected component at each edge is then N~l'o dO sin (d</>!2). d8 d</> The force acting on the top edge of the element equals N. no shear forces exist and there are only two unknown membrane forces per unit length.4BCD cut from the shell of Fig. d</>. Neglecting higher terms. from L F. No and N d>' The governing equations for these forces are derived fronl two equilibrium con·· ditions.: SHELLS 203 10. d¢ dl! sin </> For the forces considered above. ii</>. The resultant in the direction of the radius of the parallel plane for both such forces is Nol'. respectively. d</> . The externally applied forces per unit surface area are represented by the components I>. 10.30 and 1O. and 1'. This force is nearly equal to N ~ /'0 dO d</>/2. the membrane forces and the loading display no variation with O. yielding the following magnitude of the resultant for both edges: N~ro dO d</> As the crosssectional area along each of the two sides of the element is 1'. we have (a) N. in the y and z directions.3 . sin</> + 1'. the force on these areas is Nor. The zdirected distributed load carried on the surface area of the element is p.l'or.2. the force on the bottom edge is also N.3h show two different views of the element . 1'0 dll. = 0 F (a) (b) (Cl Figure 10. dO. 1'0 dO. Description of the z equilibrium requires that the z components of the loading as well as of the forces acting on each edge of the element be considered. Figures 10.
N".1b).3c).. (10. (IO. The second term ofEq. Clearly.. . cos q.. t 1'2S1112¢. the lellhand side of the foregoing equation may be written ri1> ("oN.. 2"1'.1R and CD.) dq.= /'2 p. sm q. Olle of the basic rdations lor the axisymmetrically loaded shell is found as follows N. Recall that from conditions of symmetry. (e). 10. lorces N..:q. = 0. 1O. ro (102) It is noted that an equation of equilibrium which can be used instead of Eq. introduction of 2nc = F.204 STRESSES IN PLATES AND SHELLS This c~pressioll may he converted to simpler form by. (lO. dO acting on faces . (10.!..rj dq. + "oN.. (10.p. defined by Eq. cos q. :~.(N"l'o) .) and force N". = I'II'. By so doing. (Fig.' dividing hy fOr] and replacing I'(l b. (lO. while the third term is the loading component. drj! + cJ Y (e) Here constant c represents the effects of the loads which may be applied to a shell element (Fig. = d ') d .) sin q. sin + F = 0 and it follows that N~= F .) sin q. Substitution of Nfl from Eq. in the . that is.. I' j Nfl +.r. Dividing Eq. (l'oN. are constant around the edge.y f2 sin 1. determined through integration: N" = ... dli cos cf' + p. (b) by dO dq.I' direction is expressed . 1'0 . (b) is the component in the y direction of the radial resultant force Nol'j (/q. [1'1' 1 r2(p.111) The equilibrium of forces in the direction of the meridional tangent. Equilibrium of the vertical forces is therefore described by 2"1'. p. and = 1'2 sin q.11» We verify below that Eq. dO  d N"r j dq.sin' q.2).. = p.iip (N"I'... = p. results in the value of N". into Eq.3c). (10.1b) is an alternative form orEq. N. 1'0 dO = 0 (b) The first term represents the sum of normal forces acting on edges !lC and RD.sm q..  I' j l'21'.2) follows readily by isolating part of the shell intercepted by angle q.1a) into (10...COS (psin q. sm " . Here force F represents the res"lram of all externallvadillY applied to this free body.. leads Lo .l'j cos q.N. Thus.2) and multiplication of the resulting expression by sin q. i¢ (1'.. the equation of equilibrium orthe y d ireeted forces is now d iJ. + p sin q.
N.. Spherical shell For spherical shells one can set the mean radius a = 1'..1) appear in the form F N = ..vN) = .. </> 2na sin' (IOJ) 1> The simplest case is that of a spherical shell subjected to constant internal gas press"re 1'. membrane theory cannot everywhere fulfill the conditions of deformation. is therefore (T = N t I'a 21 (10. = 1'. is then a pa' b = . Inasmuch as any section through the center results in the identical free·body. Because of their freedom of motion in the: direction. = No = N. applying Hook's law. for the axisymmetrically loaded shells of revolution considered.1) arc sufficient to determine the socalled hoopji)l'c(' N'I and the meridional jc. In the latter case.r~ STRESSES l~ ~lIELLS 205 Equations (10. We now have I' = 1'" 1> = 90°..fllHA. However. from which the stresses arc readily determined. (10. (10." The action cited demonstrates the basic difference between the problem of a shell membrane and one of plane stress. from Eqs...4) In this typical case. 105 SOME TYPICAL CASES OF SHELLS OF REVOLUTION The membrane stresses in any particular axisymmetrically loaded shell in the form of a surface of revolution may be determined from the governing expressions of equilibrium developed in the preceding section.) . a compatibility equation is required. = (0) and can no longer serve as a coordinate on the meridian. and F = . it is clear that when a shell is subject to the actioll of concentrated surface loadings or is constrained at its boundaries.3). angle 1> is a constant (r. Then Eqs. The complete solution is obtained only by application of bending theory. Ncgativl.~IE~.4) where t is the thickness of the shell.(1 .. Treated in the following paragraphs are several common structural members. like a balloon.na'p. strains are produced such as to assure consistency with the field of stress and compatibility with one another.> algebraic results indicate compressive stresses.(N . Et 2EI (10. The expansion of the sphere. Instead we introduce .5) Conical shell (Fig. 10.Jrce N". The stress..
8)  sin ¢ . d¢ ..I' (p. Accordingly. along the generator. The membrane forces are then Ns .. 1'0 = seas ¢ 1'.No = p. are in the s and radial directions.s cot ¢ = PJo . the length of a meridional element ds = 1'1 d¢. = I s.7) An alternate form of Eg.sm ¢ where 1'" is the mean radius at the base.r _.. + p.:. . Hence ~. 2"1'0 F sm ¢ (lO.. lead to 1S I (N. (I0.s) d =  (Py + p. The sum of Egs.1. ro ~.. usually meaSllred Fan! the !'erlex.. = .. is N. d'S (N. (to.. the distance of a point of the midsurface. . cot ¢)s lis (lO. p. (10. = scot ¢ (b) These relationships...~ .. load components p. hoop and meridional stresses can be computed illdepellliently. (to. when introduced into Egs...' It is observed that gIven an external load distribution..6a. respectively. cot ¢)s The meridional force.).206 STRESSFS 1.6a) may be obtained from Egs.S d (lO.".6) yields .1 ds d d (a) Also. and p. upon integration of the above expression.2) and (10. coordinate s. h) ..s) . Clearly. PLATES Al'1J SHELLS Figure 10. N n = ·"'p..lb) and (b)..
9) in which x is measured in the axial direction. 0+po I"'::~ )f'. = . one can begin with the cOile equations.= _.1 Consider a simply supported covered lIlal'ket dome of radius a and thickness t. "a t (10.5 .5a). p and F = stresses: no'". carrying only its own weight p per unit area. for a dome of halfspherical geometry (Fig. Apply the maximum principal stress u . For a closedend cylindrical vessel under constal1[ illtel'llal pressure. Hence Eqs. (b) Assume that the hemispherical dome is constructed of 70mmthick concrete of unit weight 23 kN/m' and span 2a = 56 m.=N.++ ¢ 51°50' 90° ___ 21 (a) Ib) Figure 10.'E STRESSES I'\. setting radius a = 1'" = constant. (10.~IE\IRRA. the extension of the radius of the cylinder under the action of the stresses given above is (10..1 through 10. a. as demonstrated in Examples 10.l!. p= = p.11 ) Solutions of various other cases of practical significance may be obtained by employing a procedure similar to that described in the foregoing paragraphs. (a) Determine the stresses.~ +~._oS x F 2na (10. SHELLS 207 Circular cylindrical shell To ohtain the stress resultants in a circular cylindrical shell. and mean N.9) then yield the following axial and hoop a=x po 2t (III = . 1O.6. Example 10.10) From Hooke's law.Pr Equations (10..8) become q) = rr/2.'.
F= I p. Also check the possibility of local buckling.t( I + cos <p) a. (f) of Sec. I..208 STRESSES Il\! PLATES AND SHELLS theory to evaluate the shell's ability to re~isi failure by fracture. as a" < a o .644 MPa.13) These stresses are plotted for a hemisphere in Fig. The sign of a" on the other hand._. For q. and ventilation. a reinforcing ring is used to support the upper structure as shown in Fig. (e) Most domes are not closed at the upper portion and have a lantern. the stress level at which local buckling occurs in the dome is found to be IT" = 0.5a. and E = 20 GPa. The components o/the dome weight are I. a small tower for lighting.5a. The resultant of the . IO. = I' cos r/> (10.25(20 x 109)('87 = 12.5 MPa 00) H is observed that there is no possibility of local buckling. (10..2.pa!t at the edge. . When 4) is smaller than this value.644 MPa. = '0 211a'p(1 . 211a sin q. a dq. 10. a.pa/2t at the crown to .023!)a/t = 0. is . a~ is always compressive.5b). a" is compressive. > 51"50'. (e) Determine the stresses in a dome which is a truncated half sphere (Fig. changes with the value of q.x = 0 P. as shown in the figure. "0 is tensile. Clearly. If the tensile strength of the material is smaller than 0. Note also that concrete is weak intension and a different conclusion may emerge from consideration of failure owing to direct tensile forces. In this case. Let 2q. lO. Clearly..cos q. SOLUTIO)\. Note that no failure occurs as la. (cos =  <P l+~os·¢) (10. its value increases with <P from .) (c) Introduction into Eqs. even for large domes the stress lel'cl dlle to dead weight is . and F given by Eqs. 10..3) of p.I' = p sin (p [1.023 x 28 = 0.12) (a) Referring to Fig.0 be the angle corresponding to the opening and P be the vertical load per unit length acting on the reinforcement ring. an assessment of tensile reinforcement will be required to assure satisfactory design.1 < la".12) and (e). The second of the above equations yields a o = 0 for <p = 51"50'. The ultimate compressive strength or crushing strength or concrete 0u·' = 21 MPa. 10.far Fom the limit stress of the material. (10. the weight of that part of the dome subtended by q. Upon application of Eg. (b) The maximum compressive stress in the dome is a" = pal! = (O. and division of the results by I yields the membrane stresses: up . at least in compl·essioll.5b.
(COS 9 t </>0 _=_~o~l . Eg. ~I I f.... 10.cos </> :sin' ¢~ . r . Derive expressions for the hoop and meridional forces for two conditions of loading: (a) the shell carries its own weight I' per unit area (Fig.E STf~FSSES IN SHELLS 209 total load on that portion of the dome sllbtended by the angle is then 4..I'sin </>0 sin' </> t sin' 4' for the hoop and the meridional stresses.6b). t P sin .6 .' </> (10. (f)..ME~IBR/\'\. free to move as the shell deforms under loading. </)0 2na'p(cos (/1" . Note that the circumferential strain in the dome under the action of membrane stresses may be computed from Hooke's law This strain contributes to a change 1:" • a sin (J. we obtain "¢ =  t ap cos </>" . cos </» + .1 a'p sin </> d</> + P . Example 10.14) (Po " ='2'. F = 211 = I .6a). rnIL~=mpp. (b) the shen carries a snow load assumed to be uniformly distributed over the plan (Fig. 2n(/ sin </>" . ensures that no bending is produced in the neighborhood of the edge. in edge radius.cos 41) + lnP sin </>" Using Egs. The simple support shown in the figure.i. (!OJ). 1O. dF cos r ¢ A'f++" Py dF Pg cos ¢ dF raj (h) Figure 10.2 Consider a planetarium dome that may be approximated as an edgesupported trul1cated cOl1e.
(d) into Eq5.170) Similarly."'..6b). . = 0 at s = So leads to c = tp.16). = p cos So q. (10. The condition that N. q.. (10. + P. and wind load per surface area) are of the same m·der. cos 2 q. 5 cot q. It is interesting to note that the three typical loads (weight per unit surface area) snow load per plan area. 1'2 = scot 4) r/) = s cos cp p. Now the s.. and (10.. PsS I . we use a sketch of the forces acting on a midsurface element dF (Fig.s6 cot q.8).210 STRESSES 1:\ PLATES Al'\D SHELLS SOLUTIO" (a) Referring to Fig. (d) The weight of that part of the cone defined by s  is determined from F= or I p' 2nl'.l7h) N. + . 2 cot c q.and (Idirected stress resultants can readily be obtained. one has j1x= 0 P.16) yield N.15) (b) To analyze the components of the S/lOW load 1'. Eqs. c = O..(s' . (10.17) by t. (d).7).. (1O. Hence q. q. = p.16) From Eqs. sin q. =. sin q.) +c As no force acts at the top edge. so .6a. ds ''so . cos q. P 8 2 .. 2s sin (I N 0=  cos' 4) j1S sin q.I (I'. and (10. (d). (l0. For ordinary structures these might approximate 1500 to 2000 Pa. = . cos q. ds = .sf. sin s . Referring to this figure. sin q.56) cot 25 Upon dividing Eqs. we have N = ps····_f1 S cos' sin ¢ (10. p~ (S2 .)s ds + s c = _... 1O. Substituting F and Eqs.. the membrane stresses are obtained.8).5 2np I 5 cot q.S6 N . (10. .S F = np cos q. (10. 10.
= 2r.7 Example 10.. 10.. Introducing N.r [).. (fn is constant throughout the shell from the condition of Example 10.4 Figure 10.rpI b ~I Figure 10. (1O..~~ :\" ~. N ¢ sin q.2) or N ...ME~mRAxf STRESSES Ix SJlH. = I' and 1'.b) _ pa ..p and a (J.j..18) It is noted that symmetry..7). into Eq. r/'.2 pa(l'o + b) ii.LS 211 :. SOLUTION Consider the portion of the shell defined by brium of forces leads to 2nl'0 .. y Figure 10.la). setting p. /.[. _____ .8 represents the end enclosure of a cylindrical vessel in the form of a half ellipsoid of semiaxes a and h. = np(1'6 . = 1'(1'6 . The vertical equili q. Determine membrane stresses (T.b2 ) 21'0 sin q... Determine the membrane stresses resuiting from an internal steam pressure p..8 .2t pa (10..3 A shell in the shape of a tOI'US or doughllut of circular cross section is subjected to internal pressure l' (Fig. = we obtain N _ n 1'1'2(1'0  21'0 The stresses are then rr  • pa(l'o + h) 21'0t a n . .
21 cos CI.:.\2 + a 2 y2 = (l2}/ leads to y = ±h..~". ¢ = (n/2) + CI... Y j.21) .1) in terms of the principal curvatures. Introduction of Eqs.5 Analyze the membrane stresses in a thin metal cOl1tailwJ' of conical shape. (b) the shell is filled with a liquid of specific weight I (Fig. The magnitude of thL' derivatives of this expression are r' =''. = .~~ (J  t cos IX (10.. = a and Eqs./u2 . ~ . (f) From the first ofEqs.8). Consider two specific case.y 3 (e) Referring to the figure . tan ...: (a) the shell is subjected to an internal pressure p. = a 2 /b and Eqs. A ratio a/b = I..\: Expressions for the principal radii of curvature fl alld /"2 will he required." h" =~. The membrane forces can then be determined from Eqs.pm'!. Example 10.19) The load resultant is represented by F = npl'~ sin' ¢.J. = 1>' /a and /'. (e) into the familiar expression for the curvature. 1'(12 .1'.x1 /u. = (/. (10. which...~ ~ hx (J\/ c/... = 1'.9). supported from the top. = pro .. Clearly. (J pro .PI'._ • 2t It is observed that the hoop stress 17" becomes compressive for a2 > 2b'. yields 1'. (e) and (f). Expre"ions (10.~. (10.. [l + (i)')"'/y". Thus. after dividing by the thickne" t./a' . . the case of a sphere.20) At the el'OW/1 (top of the shell) 1'.'" . (10. l' 2t ~ (10.. SOLUTION (a) For this case. then become (J~ .20) reduce to cr=(j= . when substituted into the second of Eqs... 10. gives the radius 1'.~}: ).. 21>1 At the equator (base of the shell) 1'.lTIU:. 'I' =Y = . and F = .212 STRfSSES I\: PLATES A~D SHELLS SOl.i/b. p.. (10. the meridian stresses 17" are always tensile. II follows that rr . yields the lowest stress. The equation of the ellipse 1/._ h'y . (f)..20) appear as (j pa = ..\2 a..
10."~' (1O. and supported on a cylindrical pipe (Fig..10). 4t.8). _. (10..cos~ The load is equal to the weight of the liquid of volume acOdb. (J.Y + 1v) tan' IX introducing this value into the first of Egs. SOLUTION The loading is expressed .6 Delemline the membrane forces in a spherical storage tank filled with liquid of specific weight I'. occurs Examl. . t cos~ (1 a  . (1O. 2y/3) tan 2t cos C( Cf.\~. .s . .1.. (10.) According to the familiar laws of hydrostatics. = .)e 10.8). mil" = 3h 2 }' tan rJ.. That is.y)y tan ~ . At any arbitrary level y. the pressure is therefore: or a or I' = 1'.22a) Differentiating with respect to y and equating (0 zero reveals that the maximum value of the above stress occurs at y = "/2 and is given by a fI.f: STRESSES 10..(h . the second of Eqs. after division by t.. and dividing the resulting expression by t leads to (J = '.: SHELLS 213 Figure 10. F = rc.'(i! . = r(h .')"(h .. the pressure at any point in the shell equals the weight coIUIn11 u11it crosssectional area oJ the liquid at that point.~.\lHIB\{. The maximum value of this stress. max yh 1 tan 1". _ .9 (f..22b) Ct.j16t cos at y = 3h/4.y) (g) Employing becomes I'D = Y tan IX.
cos' </> ) '6.21·1 sn~r:ssl:S I!'' PLATES AND SHELLS · . (10.. )' 6 Icos</> 2 . (10.cos 2 1>(3 .( 5+ 2 cos' 1> .oP 2"a ){t  t cos' </>(1 . the sum of the vertical support reactions 4)'"a'/3.3) now yield i'a' N. No ="a' ( 1 . = 6 sin' ¢ [1 ~'a2 . the deformation associated with the membrane solution is no/ compatible with the continuity of the structure at support !l".=... ..21ra')H } cos' </>(1 .24) it is observed that both forces N.24) From Eqs..p and N" change values abruptly at the support (</> = 1>0).I cos 1>)] .' = 3 I )'a(1 . . Icos</> (10.¢ (10..23) and (10.10 Owing to this pressure..'a l (' Inserting the above into Eqs. the resultant force F for the portion intercepted by </> is: F = 2".. A discontinuity in No means a discontinuity of the deformation of the parallel circles on the immediate sides of the '111. That is F = '!na').2 cos 1>)) = 6' 1 . N. Thus. must also be taken into account in addition to the internal pressure loading.cos 1>} sin </> cos 1> d</> '0 .a' ( 2 cos' 1> ) 5 .6 cos</> .3).1 b i .1 + cos 1> 2 cos 2 rJ> ) Nn= 6 ).23) Equations (10. Figure 10..6 cos </> + T+~o. .i cos 1»] Equations (10.23) are valid for 1> > </>0' In determining F for </> > 1>".
owing to the tangcntial displacements. produced by the displacements v and IV is l' cos cp ~ IV sin 4'.11). the hoop strain is written as follows: "0 = . order infinitesimal terms arc neglected.6 AXIALLY SYMMETRIC DEFOHMATION We noll' discuss the displacements in symmetrically loaded shells of revolution by considering an element AB of length 1"1 d</> of the meridian in an unstrained shell. and a decrease in length w dcp produced by the radial displacement 11'. Inasmuch as the circumference of the parallel circle expands in direct proportion to its radius EO = ~ (1' cos 4' ~]V sin </» 1'0 (b) Recalling that 1"0 = 1'. is thus 1 dv 8<}> IV = r~· d.~ (a) The deformation of an element of a parallel circle may he treated in a like manner.<[ STRfSSES 1:\ SHILLS 215 10. In the analysis which follows. and higher.\IE/lfIIRA. It can be ShOWll that the increase in radius 1"0 of the circle..11 . sin cp. the small deformation approximation is employed. The deformation experienced by an cle. respectively (Fig.. The meridional strain c"" the total deformation per unit length of the element AB. ment of infinitesimal length 1"1 d</> may be regarded as composed of an increase in length (dr/dc/» dc/>.(v cot 1'2 1 1) ~ w) (e) Figure 10. 10. Let the displacements in the direction of the tangent to the meridian and in thc direction normal to the midsurface be denoted by rand w. AB is displaced to position A'B'. After straining.
Equation (10. and lTv are given by Egs.' (cot e/> =/(e/» l'= [rI(e/»de/> +cjsin e/> . (e). (a) and (c) leads to th~ following differential d¢ I .1". This equation has the solution d.. __.v cot e/> = E [ok..25) We observe that the symmetric d'eronnations of a shell of revolution may be determined by integrating Eq. one can readily obtain \l' from Eq. '" .13). we let d.Sa).. (d) gives de/> . (g) It is that follows + cos all (h) . l' = a' p(1 + ") [SIO ¢ In (I + cos e/» ."" (d) The strains are related to the membrane stresses by Hooke's Jaw I:.) ... + "1".25) is therefore SOLUTION d4) dr 1' cot a'p(1 + 1')( e/> = ~. Inserting this expression into Eq.7 Determine the displacements of the spherical roof dome supporting its own weight (Fig...~ In (1 EI I + cos CI.rr.LS Elimination of equation for r: l1' from Eqs.p = j. to.vll_·_·_L. (I). EI I + cos e/> necessary to choose c such that v = 0 at e/> = a (Fig. SOle/> (I) l' The constant of integration c is determined Irom a boundary condition. Example 10.25) when the membrane stresses are known.Sa)...h v cot q) = 1". It c = {/'p(l ~.va.(r. we obtain sin e/> . = a and stresses IT. .) (c) Introduction of the above into Eq. = 1".216 STRESSES 1:\ PLATES AND SHH. IO... (10. (10. j + CSill e/> . For the half sphere under consideration 1".cos 4) j+ cos¢ =f(e/» 2). (lTd. Next.. Once has been found. + vI"dl dv I (10.
uel"leclioll r is obtained and Eq. d¢ acting on the faces AB and CD of the element make an angle dO.12 . Horizontal components of the forces Nil • . one need not employ Eq.12). cos ¢ . 1'. referred (0 the unit area of the midsurface." as is always the ('(Ise for a thil1 shell (Sec. The surface load. (e) and Eg. (c) then yields II"... r .\' direction: Nil. not only do normal forces NfjJ and No act on the sides of an element. 1I . (e) directly give the solution. as l' = 0 tilere: the second of Eqs. d(l (b) The difference of the shearing forces acting on faces AC and ED of the element is expressed (e) Figure 10. It js noted that if the slIpport displacement \\' is to be determined. and thus have the following resultant in the. = N q. 10.l.. The xdirected forces are as follows. and p. 10.2).7 ASYMMETRICALLY LOADED SHELLS OF REVOLUTION In the bending of a shell of revolution under unsymmetrical loading. p. but shearing forces NrN ancl N¢o as well (Fig. The moment equilibrium requircs that Ny. (y).S 217 Upon substituting this yalw: of c into Eq.SSfS IN SHEI. has components Px. d¢ . The force (a) is owing to the variation of N". (q).MEMBRASE STRF.• .
conical..26) permit determination of the membrane forces in a shell of revolution with nonsymmetrical loading that may.3J . 10. The equilibrium of y. The wind load on shells is composed of pressure on lhe wind side and suction of the leeward side.0'" o tNo~ 1". dO def. is considered important.26n) To the expression governing the v equilibrium of the symmetrically loaded case (Sec. Assuming for the sake of simplicity that the wind acts in the direction of the meridian plane 0 = 0.26) upon following a procedure identical with that described in Sec. Components p" and p. Only the load compollent acting perpendie.. I" ) '" 0 + .261» (10.13 shows the distribution of the staticdesign wind load on a spherical dome. For purposes of illustration.. vary with 0 and $.. '~ I" oN ao'  N I" 0" cos (h P +p 1". Eq.8 SHELLS OF REVOLUTION UNDER WIND LOADING It is usual to represent dynamic loading such as wind and earthquake effects by statically equivalent or pseudoslatic loading adequate for purposes of design. 10.27) Tn the above. (10. and cylindrical shells may readily be deduced from Eqs.dar to the l1lidsw:face p.1"" = )' 0 (10.> produced by the difference in the shearing forces acting on the faces AB and CD of the element.5... Such a case is discussed in the next section. This distribution should be regarded as a rough approximation. 3 1. Fig.la) remains valid for the present case as well. lOA). We note that the governing equations of equilibrium for the spherical.26(") Equations (10. (IO. in general.. Inasmuch as the projection of the shearing forces on the z axis vanishes.[) SJIELLS The compon~nt of the external rore!.? is l)xfOl'l dO (4) (d) Thc x equilibrium condition thus reads: (I0. = 0 p. the components of wind pressure are as follows: px = 0 1'. p represents the static wind pressure intensity.and zdirected forces is therefore satisfied by the expressions:  c (N "". we must add the force "'. = p sin </' cos fl (10..218 STRESSES IN PL/\TES A}. are due to friction forces and are of negligible magnitude. 10.
Illustrated in the solution of the following problem. cos c/> cos 0 .'q..26) to obtain n. Figure 10. aN Or/! 8¢ (IoN.14 .pl'j sin () (1O. Example 10. (10.13 Proceed ing with the solution.'\'EMIlRAXT STRI:SSfS IN SflH. eliminates No. cos . The equations of equilibrium for shells of revolution IInder the action of wind pressure are then . = 0 iJq.) T. supported by a column at the vertex (Fig. (1'0 No. is the determination of membrane stresses. we substitute Eqs. +. q.) +'aif I'j + N o"l'j cos q.:~d¢ J dl'o + cot IP N.PI'. a shell having the shape of a circular cone.. 'i!li..ro + NOrj sin a aN Q (a) sin q. = pl'ol'j q.~ ( . Find tile /.8 Consider a mushroomlike shelter.28ac) N.' + (IN. 10. cos 0 The third expression..:~ To = ) 1') /1N t }<j} . /.I • j  NOI. = 0 N4.p/·o No = pro cos (} .'I'j sm q. when substituted into the first and second.LS 219  Wind Figure 10.14).27) into (10.
N. 0 3t S2 (1O. meridian.. .. (10.9 CYLINDRICAL SHELLS OF GENERAL SHAPE A cylindrical shell is fomled by moving a straight line. . 0 (d) The condition that the edge J' = L of the shell is free is fulfilled by setting c = ·pIl!3 in the above expression.2 ) G( 2s (1O.. (b). and N . we replace N" by N.28b).. Here s. parallel to its initial direction along a closed path.Sltl pL3_ S3 .SSlS 11': PLATES A. = stan . 1O.j 1'0 = s sin CI.29b) that meridian and shear stresses grow without limit at the top (s = 0).~~ ( L' 3. (I' L' ...29a) and (10. is a particular finite length. ps sin (f. (e).~D SHELLS hoop.220 .. SOl... becomes eN.. = us s ~. ([0. cos 0 (l0. Integration of the above leads to cr.28a).cos 2 . ..27) into Eq. A shell element is bordered by two adjacent . and shear stresses if' th~ shell is submitted described by Eqs..s' +psiTIIX)COS 0 3 S3 sin (J.'H[U. we obtain. the gel1erator. . It can be shown that the vertical resultant of forces N.15a is a cylindical shell of arbitrary cross section... [ (PS' 3 + c ). ([0. (10.+ ~.s (c) [n addition. We thus have T 9$ =  N(ls t = . = . the conical shell must be assumed to be fastened to the column along a circle corresponding to S" 10. To avoid infinite stresses. and (10. as expected at a point support.14. Depicted in Fig.29a) Equation (l0.UTlON (0 a wind pressure Referring to Fig. after integration N as =~.). and No" by No. 1'2 [0.2 _ . sm .. = ~' = ~ . Upon introducing these together with Eqs. transmitted in a paralle[ circle approaches the total load of the shell when s = St.27).29b) Expression (1O.29c) It is seen from Eqs..28c) then results in the hoop stress: No t .2S3 (To = L· .. S2 . (h) from which dro =sm a d. after substituting Eqs.29..
_. r dO = (IX {IN. Assume a llolllll1i(orm loading to act on the shell. 10. we obtain the equations of equilibrium of a cylindrical shell. 0. .S 221 r _ (a) (b) Figure to.o 0 No dx . 1O.o I eNo ex I' to Po (1O. dx + Px .7 the above expressions could he obtainecl directly from Egs..x' dO . I' dO = 0 Dividing by the differential quantities. The radial or normal component of the loading p" acts in the positive.26). It is observed that they are simple in structure and may be solved one by one. r dO = 0 c1x O'fl c1Nf) "' de . . spaced dx apart. dx . a free body diagram of a membrane clement contains the Iorces shown in Fig. inward direction.ISb. . dx . It follows that ~+""= aN."x + . dx NJ + ~' dx' r dO + p.'C8 = h r ae  Px As already mentioned in Sec.30ac) tlN 1 . dO + p.MFMHRA:\E STRESSES IN SIII'U. Then. The element so described is located by coordinates x and n.15 generators and two planes normal to the axial axis x.tN. The equilibrium of forces in the x. dx . (10. and r directions is represented by e "l!" dx ' r dO + ~t. The x and components of the externally applied forces per unit area arc labeled Px and Po and are indicated to act in the directions of increasing x and O.
. (0) dx +I.16 and filled to capacity with a liquid of specific weight i'.(O) "x=. Determine the membrane forces under two assumptions: (a) there is free spanning.~ I' ddli' . Example 10.222 STRESSES IN PLATES AND SHELl S Figure 10. SOLUTION At an arbitrary level defined by angle 0 the pressure is (1 ..(IJ) Nx = I' /. I n so doing.= . with expansion joints at both ends. (10. (10.31). 10.J(po+~a~n)dX+J. Subsequently. horizontal. (10. Substituting the above into Eqs.. (x = ± ~) (e) .31) " 1'( Px+~Te x+.16 For a prescribed loading.f. These functions arise because of the integration of partial derivatives.cos 0) Po = px = 0 (a) )'a where the minus sign indicates the outward direction. cylindrical conduit is supported as shown in Fig. we have N o = Prr N.'(1. are determined by integrating Eqs. II..(O) andf.30b) and (10JOc).(O) (a) The conditions for the simply supported edges are represented by xdf.30a). dO + f.a(1 . (b) both ends are rigidly fixed.cos 0) and the external forces are thus p. (0) are arbitrary functions of integration to be evaluated on the basis of the edge conditions.f' (0) 10N'·)d 2 .<oos 0 dx .ax sin 0 +(.: where..cos 0) NXII =  J)'a sin e dx + fdO) = ). N" and N.9 A long.' ( lX' (b) ="2' cos 0 . No is readily found from Eq. = . we have N o = 7. (10.
/2 where Nx =  ~ (I3 . 'L2 8 cos (I L dl l 2a dO 'If. no change occurs in the length of the generator: (" (N. (10. is written as follows: N" = '/a'(I . = 0 dO or (e) From the second of Egs.cos 0) N xo = yaxsin 0 N.MEMBRANE STRESSES l~ SHELLS 223 Introduction of Eqs..(8) = vl'a 2 (l .4x') cos G Note that the shear N xO and the normal N x forces. . LI =.cos 0) . (0) "I3 L df () =' cos 0 + .4x') cos e +fi(O) (il) N 8 = )'a'(l . (il) into Eg.~4 cos 8 .(0) = .f. we observe that c in Eg. (Ii) into (c) yields () = : . Eqs. then. (b) In this case.32). Through the application of Hooke's law. I Et .  I .. _' + li(O) 8 2a dB Adding and subtracting the above expressions provide..32ac) ~ (13 . sian provides the axial deformation. =  (I0.. This load is not present because the pipe is free of torque. (e) represents the value of the uniform load N. the above expres. (9) yields ··L' l. (b) and (d).) dx (f) Upon introducing the membrane forces.vN o) dx = () (g) " 1. .o at x = O. represent identical spanwise distributions with the shear force and the bending moment of a un!{orm[y loaded beam . (b).cos 0) Introduction of Egs. from Eqs..£2 U2 (N x  I'N. thus c = O.~8" cos 0 "I3 (Ii) d. respectively /. respectively.. The solution.
(10.08 ~! b~ 0. As already alluded to (Sec. Sketch the results.17).ATES AND SHELLS Hence. Eqs.10 BREAKDoWN OF ELASTIC ACTION IN SHELLS From a design point of view. is not zero. )cos (I 12 (1033) One finds that the circumferential strain. principal stress theory 0.32"). IJO). The tube is fabricated of a material having a yield point stress fIn" Determine the required wall thickness t remote from the ends. according to the various theories of failure.17 . (I0.cos 0) +. The detail of the distribution of supportreaction forces is obtained by application of the bending tbeory.02 stress theory 0 0.224 STRESSES IN Pl.04 t 0.32a). it is clear that for practical purposes. N x = v"a .08 0. energy _.10 0. This will be illustrated in the examples to follow for two particular shell structures. the dimensions that should be assigned to a loaded element depend upon thc failure theory held concerning the cause of the yielding.33). and (10.10 A circular cylindrical vessel with closed ends is subjected to internal pressure p (Fig. 10.04 a  0.of distortion theory 0. The membrane solution. some bending of the pipe occurs near the supports. '" (No . thus will agree very well with measurement at distances approximately a from the supports. 2(1 .02 Max.06 !{] D' Max. it is necessary to obtain the proper dimensions of a member which can sustain a prescribed loading without suffering failure by breakdown of elastic action. 10. (X2 2 =  Ii. Example 10.06 0.1 Fignre J0.vNx)!Et. 0. Because clamped edges inhibit any stich deformations at the ends.
.s theory. (1 Gyp . Detennine the proper wall thickness t using a safety factor N.47) and (a) yield t=  fi l'a 2 O'}.~) 2.max = "4t co~~ 3J1h2 tan IX yh' tan rt.9). the maximum principal stress theory leads to the least conservative result.ME~fBRASE STRES5FS IN SHELLS 225 SOLLTlON The hoop and the axial shell stresses are given by Egs. F ormlilas (1.10): 0'1 = (fa po = t 6')  = a _::::. (1.0: t (J2 = as =}' (a  (f) The largest values of the principal stresses are given by: (J 1 .P (e) The relationship between internal pressure and wall thickness as given by Eqs. :\ 2t l)(l (a) Maximam principal stress theory.46) and (a): (= PC!. and from Egs."v)y cos . On the basis of Egs. The tank is filled with a liguid of specific weight i' and is edgesupported. SOLUTION Expressions for the hoop and longitudinal stresses. As 111 = a yp 0'2 pa (b) arc of the saIlle sign.44) and (a). (10.45b) and (a): t=·· 20'yp pa (c) Maximum principal straill theory. jy Y"2t cos ~ Eqs. t Alaximum shear stre.'·. Apply (a) the maximum principal stress theory and (b) the maximum energy of distortion theory. (Y =~) 0'2.17 for v = 1/3. we obtain.p in compression (Fig. (10. (b) to (e) is represented in Fig. (1. (1. 10.1l A conical tank is constrllcted of steel of yield point stress Vyp in tension and iv. are (J 1 = (J 0 tan Cl = r(a . (d) Maximum energy of distortion theory. !i'om ') tanrt. It is observed that the maximum shear stress theory is the most conservative. From Egs. max = i6f~OS a (V = ~a) (g) . Example IO. 10.22).
and a compressivc ultimate strength (/"... + [ (ll .. t = 0..2t cos C( ...... (g) yield _~!:I' = !~9. jf therc is snow accumulation over the dome such that p. _ .47): [ tan IX]' tan .."(a .y))'. tan IX . It is constructed .ATFS ASD SHELLS (a) lHaxilllll11l I'nllcil'''{ siress IlwnI)' Equations (10. 10.226 STlU.~.225 '_ .'(a ...2 An observation dome of a pressurized aircraft is of ellipsoidal shape (Fig. the required wall thickness is the value given by the first of Eqs.' = 22 MPn.. ~y)y.I = . 10. 10.(Jyp cn (k) (J. cos (h) 0: (b) MaxilllulIl energy o/distortion theory Expressions (g) demonstrate that the principal stresses assume their maximum values at dilTerent locations.1 A spherical roof dome is subjected to a snow load p...52a Substitution of Eq.SSFS IN I'I.. _ .! tan O"'P [J. (f) are substituted into Eqs. The location at which the combined principal stresses is critical is ascertained as follows. = 2500 Pa. Hence. Eqs. ~~~ N 4tcos~ or I = ya1N tan C/..~ J2 t cos IX 2t cos IX . 0. (1. cn into (i) results in ~'a2 N tan (J.. (h).. the derivative of the foregoing with respect to y is set equal to zero to yield y = 0.5a).. lO.l')Y'.J [(ll [I t cos a tan 11. 2 Egs.6 10. (T. G)'p t = 0..2 . Determine the factor of safety according to the maximum principal strcss theory. Develop the expressions (J !> = 21 p. AS~S\lme that the dome is constructed of 8cmthick masonry having a span 2a = 70 ro.... cos CI.43) together with 3'1'a tan (J.125 j'(I~_~... (Fig..8)._16t cos !f..1 to 10.. p. J = u~' }P (i) Next. First.t) for the meridional and hoop stresses. the thickness based upon the maximum principal stress theory is 10 percent greater than that predicted by the maximum energy of distortion theory.250 _ . _.y))'_.a cos 2¢ 21 (PIO. PROBLEMS Sees. cos In this problem.
' (Fig.._ .10..3 A conical aluminum container of thickness 3 mm. (0) Demonstrate that the membrane stresses are given by (1 . Taking \' = 0.. ' . Derive the '0 (PIO..  :::: :'~'_. PlO.. in the cone and the No/> in the sphere.12 tn. .5 10. is filled with water (Fig.5 A supported truncated conical shell carries an upper edge load following expressions for the membrane stresses a = p .. Note that at the juncture of the two parts. _ (1/ + r){/"/ 41 (PIO. ·.'rial.. PlO.5) Figure PIO.Y{h..4 y ]0.(3h .6 The compound tank shown in Fig.15 m . The It:ngths of tht' 5nniaxcs arc u = 0.4 An t'dgesupporlcd hemispherical container is filled with a liquid of specific wcight . compute the locations measured vertically above the apex.I (PlO..y)t cos ox " • ~ sin ~ . Determine the limiting value of the prc<.6) The solution for the bottom part is governed by Eqs.c.2y)' .¢: _·_x y  _..3.._===.4) (b) Calculate the radial (w) and the circumferential (v) deformations of the shell. 10. ... 6t' C05 2 ~ ..9). apex angle (i = 45°. a ring must be provided to resist the difference in the horizontal components of the meridional forces N. and height 11= 3 m. ~ (l  L Figure P] 0. 10. ..sure differential th~' shell (:all resisl given a maximum stresS of 14 l\lPa. ~.5).4). Show that the hoop and the meridional stresses in the conical part are given by (' tan Cl au =. (P10.lnd /.~ = .. . PlO..~IEMBRANE STRESSES IN SHELLS 227 of 6nnnthiek phl~tle m'lh. for which (a) the hoop strain is zero and (b) the hoop strain is maximum.6 consists of a conical shell with a spherical bottom...=0..!i Cst sin ¢ p~ (Fig.4).
8 Verify that.:.6.p = . + 5  6 cos ¢ + i"+ e~:~"~ 2 cos 1> ) 2 O'. Derive the following expressions for the membrane forces N o = ···pssinrxtanrx pL2 _ SZ N~· $ (PIO. in represented by th~ spherical tank described in Example 10. 10.9 shows a compound tank comprised of a cylindrical shell and a spherical bottom.( 1 + ~6r I + cos ¢ l)(I  cos rfJ) (PIO. The lank is filled to a level h with a liquid of specific weight I'. Verify that the membrane stresses in the cylindrical part are given by ct(l=ybx (PIO.10 10.9 h 10. 10..9 Figure PlO. 10.9.. I t ! ti Figure PI 0.7 The mushroomlike shelter shown in Fig. Figure Pl0.6.6 10.9) I ±.228 STRESSES 17\ PLATES AND SHELLS ~l J' t l.11 Consider the tank of Prob. for example.14 is assumed to support only its own weight per unit surface area p.10 Determine the radial (IV) and the circumferential (v) deformations in the spherical tank described in Example 10. . Derive the following expressions for the membrane stresses in the bottom part 0'(1:= val (311 6t . r:t~ (~~! + 1 .. 'i~l:~s rfJ~) (PIO..7) 2SC050: 10.7 to 10. Sees.!l) Here the constant value C of the axial stress may be produced by the weight of a roof. the maximum shear stress is '( m~" oa' = !.S) 10.
_. cos (J 2px .1S) I' .(L2 _ 4x 2 ) cos {j 4at 10. Derive the following expressions for the membrane stresses Gil = . stn (J 2px .l6) 10.l2 A pipeline III the fonn nral) Op. 10.'11 ~'t'llIitil"(.. and thickness t..u pipe of radius d..!7). thickness t.'S T. smfJ 1 1 (PlO.17) .15 for the case of rigidly builtin cyJinder ends.". PlO.16).13 10. horizontal. Derive the following expressions for the membrane stresst. Determine the membrane stresses in the pipe if the ends nre assumed to be built in.S I O O (pJO.14 A horizontal circuJ.16 Redo Prob. be approximated as a semicircular crosssectional cylinder or socalled harrell. Refer to Fig.15 A simply supponed. 10.'{I!t/r (Fig. pa (Jo =:  . circular cylinder of radius a (Fig.l3) Figure PlO.cos 0 1 ~ po '.liaJl!/(:/ls IlHed wilh a liquid ofspecilic weight :'. ma. Ttf'= . 10. it is assumed that the conditions of simple sllpporl prevail..!lllIr . Al both ends oflhe shell. supporting its own weight f.16 for notation and aS$\1me that both ends art' fixed. __ . is filled with a gas at a constant pressure p.«1 = ' . PlO. Derive the rnJIowing expressions lor the membrane forces (PlU.<~ p = sint) nat (PlO. and length L carries its own weight p.MHlIlltl\:.17 Consider a thinwalled circular pipe of mean radius u. 10.13 The roof of an airer<lft hangar.r STRf:SSES TK SHELLS 229 IO. length L..l2) Note that () is now measured from the horizontal axis. loaded as a cantilever (Fig..13). 10. Verify that now the membrane stresses are represented by 2px t . Develop the following expressions for the membrane stress!!:. 1 (PIO.
5 MPa (Fig.17 to. assuming failure to occur in accordallc\.06 tn.230 STRESSES IN PLATES AND SHELLS ~12" i~ ~0.4 m. lO. .18 A circular toroidal shell is subjected to internal pressure 1) = 1. \ ol"'f·_ Figure PIO. Assume that the yield strcss (1yp = 200 MPa. and ( = I mm.' with the energy of distortion theory. Wbat is the faclor of safety.7). Thc dimensions of th~ torus arc (/ = 0. b = 0.
11. the resultant forces and moments representing the internal stresses.2 SHELL STRESS RESULTANTS In deriving an expression for the stress resultants. P. that is. [n Secs. to that membrane theory cannot. This theory also fails to predict the state of stress at the boundaries and in certain other areas of the shell. shear forces. We shall begin by deriving the basic relationships between the stress resultants and the deformations for shells of general shape. The origin of a cartesian coordinate system is located at a 231 .l INTRODUCTION It was observed in Chap.5 are developed the relationships for the stresses and strain energy under an arbitrary loading. in all instances. and the first solutions involving shellbending stresses date back to only 1920. Stress analysis of cylindrical shells under general loads is postponed until Chap. The complete bending theory is mathematically intricate. one proceeds as in the case of plates. and IV which define the geometry or kinematics of deformation of a shell. 13. consider an infinitestimal element (Fig. considering membrane forces. after applications to various common structural members are presented in Chap. This element is defined by two pairs of planes. To develop the governing dilTerential equations for the midsur(ace displacemen ts 1/. 12. These shortcomings are avoided by application of bending theory. With the exception of Sec.4 and 11.1 a). and moments to act on the shell structure. we shall. limit consideration to the most significant practical case involving rotationally symmetrical loading. in this chapter.CHAPTER ELEVEN BENDING STRESSES IN SHELLS ll. I 1. normal to the midsurface of the shell. 11. lUI. pro vide solutions compatible with the actual conditions of deformation.
• (1 w .1b is thus N. The stresses acting on the plane faces of the element are" x' "y. the lengths measured on the midsurface.:: ds... and using the true arc length given above. Letting N x represent the resultant normal force acting on plane face yz per unit length."2 ~ dz = .'. M) M a (1 K) I1vI!.. as shown. Becaus..1_"2 aAl  ZKy) dz Expressions for the remaining stress resultants per unit length are derived m a similar manner. respectively.1 ) 21)x) .r.e of shell curvature.x 1 .'I _ I. .) I . T x." ..AI   Z~y II x .. 1.(rx  z) (' 1 .'''" ) ..= x ds. I" '. = t........ The complete set shown in Fig.. r ds z) x x where /'x and r" arc the radii of principal curvatures.lf2 j "x(1 . in the xz and )'2 planes. with the x and y axes tangent to the lines of principal curvature.l .'2 a.: PLATES AND SIIELLS c A T t tOI \ I \ \ r. / (11. = 't:2 I. 1 . Il.232 STRESSJ:S I:". N. this becomes N... / \ I I \ / ! I I (a) tb) Figure 11.. ( •) Cancelling arbitrary length ds.:K. and T . 1 I Nxy N yx Qx I Q.... .>12 ( I  _ ) .i2 Ix)' . '.IvI. dz r. but rather .. . we have N x ds. the arc lengths of an element located a distance z from the midsurface are not simply ds x and ds y .. '. and z perpendicular to the midsurface.1 corner of the element. ".
On this basis) N .lC as in the treatment of plates. = O.He not gencrally equaL nor are twisting moments NIx}. and 111 y. = N yx and A/ x . and consequently. sides mn and m'n' are straight lines. the zdirected stress is neglected. According to our assumption. (Jx. however. 11.l ' = wI. Hooke's law is then written (a) !. . is given by (b) .ty and N rx .}' = {'X}..RENDJ'\." and T x . The figure shows the midsurface stretched and side mn rotated with respect to its original configuration. MOMENT. noting that. 10. by assumption (3) of Sec. :/rx and :/1")' may be·ne{/leeted in comparison with unity. t is small relative to i'x and 1'. fT..:).\. 112.:' This is because in general r x i r. Consider the deformed shell element of Fig. G Let us first deteImine the strains appearing in the above expressions.G STRESSl:S IN SHEl.3 FORCE.LS 233 The sign convention is the sa..\ . ofa fiber of length lr' located in the xz plane a distance z from the midsurface.• The stress resultants are thus described by the same expressions as apply to thin plates. AND DISPLACEMENT RELATIONS To relate the stress resultants to the shell deformations. fT.\. for which we are concerned. The unit elongation B.I •. must be evaluated in terms of strains.. It may be concluded thal even though I:q = Ty:n sbearing rorces N.. For thin shells.1.
z(Xx + VXy)] (11.. (c) into Eq. ll.xO"'y + "X~X )] =. (b). Referring to Fig. I'~. designates the twist of the midsurface. I . t <Ii I'x... and into (a).XJ' . Because for the case under analysis. Let lxyO denote the shearing strain of the midsurface. 11...JL [r . "x.(X T xv (.x may be omitted. and ds x . (e) The nomenclature parallels that used in connection with The distrihution of shear strain )'xy is next evaluated.0 UpOIl cur"ature is negligible. PLATES i\~D SHELLS Here I:1l j is the elongation experienced by II. we obtain en E a x = 1' ["xo  \' + V"yO .2) al ' 1 __ v2'Y 0 + vr·. For the)' direction. (XI'O ~ ?X)G _4 . (d)... Upon substitution of Eqs.la) and 1\. it can be demonstrated that the influence of '. the length of the midsurface fiber.. ) = r y HyO  ZXI' .0. and referring to the third of Eqs.3a) for plates.. U) Here Xx. . we have . a similar expression is obtained. it represents the efl·ect of the rotation of the shell elements about a normal to the midsurface. Substituting Eqs. Clearly. we have in which rx is the curvature prior to deformation. The unit elongation at any distance normal to the midsurface is thus related to the midsurface stretch and the change in curvature associated with deformation. (e). the radius of curvature after deformation. Introducing the foregoing considerations. In addition. zj. the above expression becomes (d) where X represents the change of curvature of the midsurface. Owing to the rotation of edge AB relative to 0: about the x axis (Fig. (l.2 (r) where 0xO represents the xdirected midsUiface unit deformation.234 STRESSES /:.
when Eq.. What remains are the menlbrane forces N x' Ny and N X.x' 11. Equations (11 3) are the cOIlstiwth'e equations for shells. 12 and 13. Cases involving bending moments are treated in the sections which follow. 10." O'y' and T X)' within the shell is lineal'. and the second terms. methods for determining membrane forces were discussed for shells of various shapes._ = 3Q)..) defines the ffexural rigidity of the shell." and M x). and in Chaps. = M)" now vanish. G STRESSES IN SHELLS 235 Finally. = (113) i(i+0 /'.D(X. In Chap. the same as for a plate. Knowing the stress resultants.'fy =/".I' = NJ. and shells. ) 2/ (1 _<l. (11. bending stress. (112) with the result that (11. as for plates or beams. as in case of plates.".5) Their values.4) The first terms above clearly describe membrane stress.fjr:\"DI:.2) is introduced into [q... + 1'Xx) N" = Ny.5)..4) and (11. the analysis of stress is vastly simplified as M. The fundamental stress relationships are thus identical for beams. the stress resuhants become My = .3) into Eqs.~)' t' (11.o M. plates. we substitute the strains and deformations of Eqs. We observe that distribution of the stress components 0'. .xvo Et Here D = Et"il2( 1 . that the vertical shearing stresses are governed by a parabolic distribution: r . (11.4 COMPOUND STRESSES IN A SHELL We are now in a position to express the compound stresses in a shell produced by the forces and moments. (11. Should the actual conditions be such as to permit bending to be neglected. as hefore. one can readily compute the stress at any point within a shell through the application or Eqs. negkc(ing terms and :. arc smal! in comparison with the other planestress components. It can be verified. (I LJ). For this purpose.
The bendingstrain energy. plane forces and is given by Urn = ~ JJ (Nx"x" + Ny")'o + Nx)')'x)'o) dx dy A (11. Owing to symmetry. M ".O)' .")(ZxXy . upon replacing the curvatures Eg. No and M".32) through (1. upon application of the appropriate stresses and strains. + 1..3) into the above expression leads to the following form involving the strains and clastic constants: Et Uno = 2(1 _ vll.. 11.7) and (11. only three of the six equilibrium . Subject to the foregoing simplifications.)2.7) where A represents the surface area of the shell.5 STRAIN ENERGY IN THE BENDING AND STRETCHING OF SHELLS Equations (1.. N" and Qx' Furthermore. = to JJ [(X. The strain energy plays an important role in treating the bending and buckling problems of shells (Chap. an element cut from a cylinder of radius a will have acting on it only the stress resultants shown in Fig.P Xy . 11..) vanish.:.6 AXISYMMETRICALLY LOADED CIRCULAR CYLINDRICAL SHELLS Pipes.i [(e xo + B.8) permit the energy to be evaluated readily for a number of commonly encountered shells of regular shape and regular loading.X.8a) Introduction of Eqs. XX)" is found to be K_'r> K y . (1.\' lJl U. That is.A 2(1 . the circumferential force and moment.o)] dx dy (1I. ~\'J and the normal stress (0.34). The membrane energy is associated with mid surface stretching produced by the in. 13).Sb) Expressions (11. lead to a strain. V)(B'OByo'" !. /I and w. As in the bending of plates. do not vary with O. boilers.). The components of strain energy of a deformed shell are the hendillystrain energy lIb and the membranestrain energy Um . and various other vessels under internal pressure exemplify the axisymmctrically loaded cylindrical shell. (11.2(1 A . 1\x. The circumferential displacement l' thus vanishes and we need consider only the .1.236 STRESSES 1:': PLAT!'S A~D SIIELLS 11.y)] dx dy (11. we assume that the transvcrSl: shearing strains C'x.\ and z displacements.3: N". tanks.34) by the changes in curvature x.energy expression for the shells.
90. a dl! . 11. dx = 0 (b) Equations (a) and (b) are.HEND1NG STltFSSES IN SHELLS 237 /x adO f \ z~ M.c) It is interesting to note that Eq. (11.9(') is a statement of the basic beam relationship: the shearing force is the first derivative of the bending moment.3. de + jJ.': dx x dQ dx I I I I . =  r p" dx + (' (e) .90) the axial force N. a dO x Qx . (11. No / / dO 11/ V < Figure 11. Equilibrinm in the x (axial) and z (radial) directions now requires that dN I .t' dx .b.3 ' a equations of the shell element remain to be satisfied. I I IN. I Q + _. . dx .(xx ' a dJ +Px'a dO . a dB = Equilibrium of moments about the y axis is governed by (a) 0 ~dfoi1= dx . through cancellation of like terms. rewritten dN x . Suppose also that the external loading is as shown in Fig. d X= 0 I dx ~. is N. a dO + N" dx . From Eq. (11.+ Px= 0 dx :/.
13h) A more convenient form of this expression is d 4 .9b) and (11..a dO ..10a). cannot be determined from Eqs. and AJ.v2 ) = 4aijj = ~2.E."+VI:xl= I . v' a .'''')' + Et2 IV _ dx 2 a Et N vN x .d dx' (D ci.l d'". B = x dx dll fo =  (a . Mo= .12) where D is the flllxural rigidity of the shell. Mx= D J. (11. ~'JO = w a (d) Applying Hooke's law.2 dx Et (dll . Employing Eqs..13a) For a shell of constant thickness.IV) dll . (I 1. p. (11.v' _ . Pr dx' + 4f1 W .13a) becomes D ·_+wv .9b) and (11..1 I) The bending moment displacement relations are the same as for a plane bent into a cylindrical surface. = 0 Finally.v2 ·N dx Ee x + v· \V a Then..10b) fI 4 Et 3(1 . I . Clearly..11). from Hooke's law and Eqs.ge) and eliminating Qx. That is.(..10a) from which du 1 . from symmetry.238 STRESS!:. given by Eq. (1.. I . = ° (11. (11.. (I 1.' l' dx du) (11.. the circumferential force is found to be NO=~(r. Because v = 0.11) and (11..'aD = D Here (11.LS where (' is a constant of integration. + vvo) = T~:.i" (11.w 4 vN.l4) ... hecause d'wjd y 2 = 0.. the unknown quantities Qx' N Q . we have N x = I _ v2 (v x Ee ..S I~ PLATES AND SIIH..12)..9c) alone and it is therefore necessary to examine the midsurface displacements.v~) (11. Eq.·"p =0 2 4 d4 w dx a a r (11. we have 2 d'M.'Mx (11. the strain·displacement relations are. (d)... when the above expression is combined with Eqs. . the following is obtained: dx' + ~ No + p.
15b) is given by Here ('1> ("2... upon integration. by addition and subtraction of 4m 2 {32.10) simplify to ~=l'· dl{ dx HI a d4H~ .. determined on the basis of the appropriate boundary conditions.13) and (Il.7 A TYPICAL CASE OF THE AXISYMMETRICALLY LOADED CYLINDRICAL SHELL This section deals with the bending problem of a cylinder with length very large compared with its diameter. We thus have In = ±{3(1 ± i) It follows that Let fix) represent the particular solution "'p' It is noted that the results of membrane theor}' can ahvays be considered as the particular solutions l?l the equations of belldillY Iheory (Sec. 11. directly yields lI. subjected to a load P uniformly distributed along a circular section (Fig.10/)) or (11.. The homogeneous solution of Eq. 12. expression C4 are constants and ". 4m2/f = O. (IU5b) may therefore be written W= e/lX(C l cos{3x + C 2 sin lix) + e/lX(C. (1\.. C" C 4 are arbitrary constants of integration. It also represents 1 the equation of a beam of fluxural rigidity D.. m3 • 1114 are the roots of the + 4{34 = ° This equation may be written. Pr J) (11.BE!'<DfSG STRESSES IN SHELLS 239 and whcr~ geometric purmlloer fJ has the dimension of CI.4). When an axial load does nol exist. Equations (11. (I 1.cos{3x + C 4 sin{!x) +/(x) (11.4). Inasmuch as there is no .15a. and Eqs. resting on an elastic foundation and subject to loading 1'..15b) is an ordinary differential equation with constant coemcients. 1/1 2 + 2{32 = ±2m{3. c 3 ..4 111 1• ni 2 .lOa) represent the YOl'erninli d isp/acelllelit conditions for a symmetrically loaded circular cylindrical shell. b) The first of tilese. N. the socalled infinite cylinder. 11.+ 41fH' = dx" . = 0. Section 11. Hence. Expression (11. the reciprocal of length. as (/Il' + 2{3')2 .16) where C l' C 2.7 serves to illustrate application of the theory. The general solution of Eq.
12) and (11. The solution may be written as follows: w = eJ1X(C.r(x) = 0 in Eq. cos fix + C.' [.) 1 We observe that the deflection attelluotes with dis/(Jllce as an exponentially damped sille wave of wave length./2 sin ({3X + ~.20) The conditions applicable immediately to the right of tile load are d\v Qx = D dx3 =. cos fix + C 2 sin fix) + eP"(C 3 cos fix + C.=..f1x (11. (11. distributed over the surface of the shell and N.240 STRI:SSl:S 1:\ PlATES A.17).18) Inasmuch as N x = 0.18) and setting x = 0. = 0. sin fix) (11.89.19) From Egs.. C. o a (11.[) SHELLS f' p Figure 11. p 8{3D The displacement is therefore II' = 8{33D (sill {3x + cos {3x) Pe. ..9(')./ai.4 pres<ure 1'. the deflection and all derivatives of w with respect to x must vanish...=C..2 p dw dx = 0 (a) These describe the respective requirements that each half of the cylinder carry onehalf the external load and that the slope vanish at the center owing to the symmetry.16). (11. (11. (11.2 x d2 w dx Mo= (11. boundary conditions for the right half are deduced from the fact that as x 4 00. we set .10a) and (11.3: 2rr!{J '" 4. for l' = 0. We thus have w = e'PX(C.21) This result may be expressed IV = ~~:.17) Owing to shell symmetry.11) yield Etw N = _ . These conditions are fulfilled if C 3 = C. (11. Eqs. Introducing Egs. (a) into Eg._. M = D _ . sin fix) (11. = 0 in Eq.
(flx) = eP'(cos fix .020 0.038 0.067 0.006 0.(fix) 0.(flx) N.064 ·0.031 ·0.012 0.6 I.155 0.172 '.356 0.0 J.109 1.sin fix) 1 f2(flx) = e{" sin fix = .'G STRESSES I~' SHELLS 241 Table II.0 5.002 0.128 o.(P.012 0.065 0.196 0. _ j: I'P 4f1 3 V· (n.056 .8 ·0.iR 0.000 0.H 2.024 0.006 o.802 0. (1121) into Eq. =  2.OOS 0.199 0.22) ".2 4. Table 11.(fix) 0.102 0..4 2.020 0.0 om J 0.2 l1.  'ili>!' = 4jj' I.049 ·0. (fix) (11.5 6.010 l1.201 0.009 0.208 0.(fix) f.005 0.8 5.243 0.508 0.009 O.4 3.(flx) = e"X(cos fix +.041 0.) 0.009 0.001 0.002 OWl 0. =  At. /4 (fix) P .009 0.123 0.! lix 0.143 0. (ILl9) and (11.({Jx) fix 3.037 0.041 0.O. I cos lix = .2 1. The term px is dimensionless and is usually thought of as expressed in radians.123 0.163 ().001 0.965 0.000 0.061 0. = :fI /.2f1 j / ' .000 0. Substituting Eq.090 0.025 0.:DIr.012 OJ)(J4 0.313 0.000 OJI04 0.6 3.640 0.4 4.1l0 0.008 0.(flx) EtP 8f13Da!.002 0.6 4.002 ··0.O.057 4. p I\' = 8f1'D!.007 0.OO7 1.012 ~O.032 0.078 65 7.024 O.878 0.Oll ··0.(flx) =  ..024 2.043 0.007 0.0 4. 1" I ~r:.026 ·0.453 0.OIS ·0.6 2.001 The following notations are used to more conveniently represent the expressions for deflection and stress resultants: !.0 ··0. 1 1 (l1.2f1f'I f.OOJ 0.310 0.111 ·0.4 0..(fI') = e j..038 0.067 0.20). ({h) I.2 0.763 0.001 0.635 0.23) M " = .617 0.0 3.sin fix) = {if.002 0001 OJ)(JS 0.003 0.O08 0.000 0.179 0.042 0.199 0.) Q.0 1."2pf' = 1.281 0.016 O.BF'..o·n 0. f.(px) 0.037 0.014 ·0014 f.202 0.006 [.4 1.Q(16 0.. = .1 furnishes numerical values of these functions for various values of fix.056 0.001 0.390 0.161 0.002 ··0.0 [.000 0.285 (.~61 !.fJOO 0. h.003 0.2 2.6 0$ 1.(fix) 1.001 0.322 0.8 0.
onc takes .. determined by applying Eqs. Because of this. The maximum deflection and moment occur at x = 0. loaded at midlength.4) and (11.23) to be Willa" = Sjj3jj = 2£i P Po'p Mmax = p 4/J (11.23) together with the principle of superposition permits the determination of deflection and stress resultants in long cylinders under any other kind of loading.. = P dx is llw = *p~1 fdPx) . Referring to Table 11.4. (11. SOLUTION Through the use of the first of Eqs. (11. Example 11. max = . it is concluded that bending is of a local character.24) The largest values of bending stress are found at x = 0 and z = til. 3P 2fJt2 ~.5). (11.1. A shell of length L = 2n1P. (! 1. found from Eqs. (11.22) and hence in Eqs. Therefore.23): a x. in the opposite direction to that shown in Fig. 11. in most engineering applications. (11. it is observed that each quantity in Eqs. will experience maximum deflection and bending moment nearly identical with those associated with a long shell. decreases with increasing px.242 STRESSES IN PlATES ANI) SHELLS These expressions are valid for ' 2: O. (b) The foregoing are the maximum axial and circumferential stresses in the cylinder. For the lell bulf of tbe cylinder. the ejfect of the cOllcelltrated loads may be neglected at loeatiollS for which x> (niP)..23) the deflection llw at point 0 owing to load P. 11. Application of Egs. respectively.1 A very long cylinder of radius a is subjected to a uniform loading p over L of its length (Fig.23). Derive an expression for the deflection at an arbitrary point 0 within length L.
The third term is due to the shear forces Q" . represented by an infinitesimal element in Fig. 10. cone. the values of l:'Uih) and f. ifb and c are large.b 11'=1 '0 Inserting into the above /. at the point at which b = c. In a like manner. and circular cylinder (Sec. Note that.fib cos ({Jh) . U.N"I'. Oil faces AC and BD of the element.el " cos ({Je)] or II' = 2Et pa' [2 f'({Jb) f. Conditions of symmetry dictate that only the resultants Q.({Je)] (e) The maximum deflection of the cylinder occurs at midlength of the distrihuted load.8 SHELLS OF REVOLUTION UNDER AXISYMMETRICAL LOADS Let us consider a body in the yelleral form of a shell of revolution subjected to rotationally symmetrical loads.23).1'.HENDI. we define the state of stress resultant acting at a point of such shells. applying the last four of Eqs. 10. The sphere.ej STfOOSSES 10. we obtain after integration w= ~ pa 2 2Et [2 . d¢.. and N.e . 1'0 dO. (fix) from Eqs.4. (10. d¢ + 1'. (l1. The development of the equilibrium equations of shell element ABeD proceeds in a manner similar to that described in Sec. 11.r. we can obtain the expressions for the stress resultants at O.2).6) are typical simple geometries in this category.22). These faces form all angle d¢ with one another.·o dO = 0 (a) The first two and the last terms of the above are already specified by Eg. and that the normal forces No and the bending moments Mo cannot vary with O. exist. 11.~. To begin with. The notations for the radii of curvature and the angular orientation are identical with those of membrane theory (Fig. M. dO) d¢ .. SHE!. ro dO .2).~ Q• . N •. (11. .6.i (N._ M o.({Jc) are quite small and the deflection will approximately equal pa 2 /Et. The cOlldition that summation of the ydirected forces be equal to zero is fulfilled by ( tjJ d I. LS 243 The displacement at point () produced by the entire load is then . dO d¢ cos ¢ ' .
we obtain the equations of equilibrium: d dq./'.l'o) = (! 1. dq.Mo' 1'.. dO = 0 (e) for the equilibrium of the forces around the x axis. the third is the resultant of the moments Mer.' cos q.p + ror.25) d¢ (M".) .l'o) .+ p.Nol'j cos ¢ . The terms of Eg. in Egs. Dropping the factor de dq. (e) are described as follows: the first is the increment of the bending moments M~l'o dO. common to all terms.ro d + Nor. (a) to (e). Note that the two moment ('ectors Mer./'oQ.1'.l'o dO. cos ¢ form an angle dO with one another and thus have a resultant expressed by the last terms.Q.. 1'" dO to Eg. = 0 0 N. sm ¢ + dq. dq. 1'0 = 0 ..6 The condition of equilibrium in the: direction may readily be obtained by adding the increment of the shear forces Q~ . dq. dq. ... cos q. lOA. N. the second represents the moment of the shear forces Q. d(Q. dO sin q. + p. + 11> (Q..l'. dq. .ro dO dcp + N e/'. .244 STRESSES 1:\ PLATES AND SHELLS Figure 11.l'o . dq. That IS. acting on the side faces AB and CD of the element are not parallel.M"I'. d</)I'" dll = 0 (b) Finally. (a) of Sec. (N". Their horizontal components Mer. one writes the expression d4> (M~' 1'0 dO) d¢  d Q~' 1'0 de· 1'.I'.p.I'" dO) dq.
BE!'.'DI;\G STRESSES I' $HELl.S
245
The governing equations for tht>.
COm111011
shdls of revolution subjeckd to
axi.sYll1l1lclricalloads may be derived from the above expressions.
Conical shell (Fig. 10.4) For this case it '/' = constant (I', = (X). Thus.
1'2 =
is observed in Sec. 10.5 that
M,j,=M,
8
cot
4,J
Employing these. the equations of equilibrium (11.25) assume the form:
~~~ ds (N ~s)  N,,I =
S
d
 p).s
d No + ds (Q~,s) cot ¢
=
p,s cot ¢
()
(11.26)
d.i: (M,s) 
d
Q,s
+ M" =
Spherical shells Denoting by a the radius of the midsurface of the shell, we have 1', = 1'2 = II and 1'0 = s sin ¢. The equilibrium conditions (11.25) then sitnrJify to
d¢ (N"
d
sin ¢)  N" cos ¢  Q. sin ¢ = p,.a sin <I'
N. sin ¢
+ N" sin ¢ + d~ (Q. sin
¢) = p,(I sin ¢
= ()
(11.27)
d~ (,''11. sin ¢) 
M, cos ¢  aQ. sin ¢
Cylindrical shells (Fig. 11.3) We can use Eg. (11.26) for the cone, letting s = x = 1'2 tan ¢, ¢ = n/2, and 1'2 = a. By so doing, one obtains expressions which are identical with Eqs. (11.9) of Sec. 11.6. Interestingly, the first expression of Eqs. (11.26) agrees with the corresponding expression of membrane theory. It is noted that by cancelling the terms involving shear forces and moments, Egs. (11.26), (11.27), and (11.9) reduce to the conditions of membrane theory of conical, spherical, and cylindrical shells, respectively.
11,9 GOVERNING EQUATIONS FOR AXISYMMETRICAL DISPLACEMENTS
In the preceding section it is observed that thrce equilibrium conditions (11.25) of an axisymmetrically loaded shell of revolution contain five unknown stress resultants N., N" Q,,, M", and lvI,. To reduce the number of unknowns to
246
STRESSES Il\: PLAUS .<\'.;0 SHELLS
(a)
(n)
Figure 11.7
three, relationships involving the forces (N </>' No), the moments (AI q" M 0)' and the displacement components (v, w) are developed in the paragraphs which follow. The membrane strains and the displacements at a point of the midsurface are connected by Eqs. (a) and (b) of Sec. 10.6:
f~·
I dv
IV
.• 
I'
'"'
d'"
r
The forceresultant strain relations (11.3) then lead to
N
•  1  ,,' r,
,
",,= 
l'
cot ¢ .
W
(11.28)
1'2
1'2
_~t .. [..1.
(dl) _
d¢
w) +..". (v cot ¢ 1'2
W)]
.
No
=
T~I;;2
U; (t, cot ¢  w) + f~ (5~  w) J
(11.29)
Identical expressions for M </> and AI n can be obtained by considering the variations in curvature of a shell element (Fig. 11.6). For this purpose we examine the meridional section of the shell element (Fig. 11.7). The rotations of the tangent of the top lace AC consist of: a rotation with respect to a perpendicular to the meridian plane by an amount v!r , owing to the displacement" of point A to point .4' (Fig. 11.7a); a rotation about the same axis by dll'/(r, d¢) produclxl by the additional displacement of point B with respect to point A (Fig. 11.7b). The total rotation of the upper edge is therefore
+1',
v
d~v
r, d¢
(a)
The top and bottom faces of the clement initially makes an angle dO with one another. The rotation of the bottom face BD is then
BEi\DJNO STRESSES I~ SHELLS
247
The variation or curvature or thl' mcridian. Ihe angular nuiatioll divided by t!J~ length "1 dej) or the arc, is thu:.;
x,;, = },
;!~
(/, +I',d~~j,)
[0
(! 1.30)
It is observed that, owing to the symmetry of deformation, each of the lateral edges AB and CD of the shell element also rotates in its meridian plane by an angle defined by Eq. (a). It may be verified that the unit normal to the right face
of a shell element has a ydirected component equal given by
d</> . cos </>. Thus, the
rotation of face CD in its own plane has a component with respect to the yaxis
 (". + ~dW_) cos 4' dO
1', 1',
d</>
Dividing this rotation by length
X, =
I
1'0
dB, we have the change of curvature:
=
(t'
1',
+ _dw )COS </>
1',
d¢
(t
1',
+ .~.... )cot4:
1',
1'0
d</>
(11.31)
1'2
Finally, inserting Eqs. (11.30) and (11.31) into Eqs. (11.3), the momentdisplacement relations are obtained:
M¢ = D
If; ~j, (;~ + ;:,~J¢) + U: + ,:;!'h)
1': 1', 1', 1', 1', 1',
cot </> J
M" = .... D [( ':
+~d,,' )C'?.lt +! .5.1 (" +~.) 1 dt/) d</> I't d</>
(l1.32)
Now Eqs. (11.25) togetller with Eqs. (11.29) and (11.32) lead to three expressions in three unknowns: I" IV, and Q". Furthermore, using the first of the resulting three equations, the shear force Q. can readily be eliminated in the last two. The expressions (11.25) are thus reduced to two equations in two unknowns: v and lV, These governing equationsIor displacements, usually transformed into new variables,32 are employed to treat the shellbending problem. An application of the equations derived above is presented in the next
section.
ILIO COMPARISON OF BENDING AND MEMBRANE STRESSES
The equations for the stress resultants (11.29) and (11.32) may be used to gauge the accuracy of the membrane analysis discussed in Chap. 10. To determine the bending moments, which were omitted in the membrane theory, the expressions for the displacements developed in Sec. 10.6 are introduced into Egs. (11.32). The bending stresses are then obtained from Eqs. (I I A). Upon comparison of
248
STRESSES W PLATES AND SHELLS
the stress magnitudes so (iL'termined with those or the membrane stresses, COIlc1usions may be drawn with respect to the accurflcy of the membrane theory, as
in the following example.
Example 11.2 Consider the spherical dome supporting its own weight described in Example 10.1. Assume that the supports are as shown in Fig. lO.5a. (a) Determine the bending stress. and (b) compare the bending
and membrane stresses,
SOLUTTON
(a) The displacements v and w. from Egs. (g) and (e) of Sec. 10.6 together with Egs. (10.13), are
v=
:,2 P~t+ 1') (I+~OS a
2
):dc;S4>
)
+ In      sinrjJ l+cos"
I+COS(/»
(a)
\\' = v cot
a 1 + l' rjJ   p (       cos rjJ Er 1 + cos </>
Introduction of the above into Eqs. (11.32) yields the following expressions for the moments;
M
=M¢ "
pt 2 2 + v =   cos rjJ
12 1  \'
(I"
(b)
The magnitude of the bending stress
11" = 2
at the surface of the shell is then
p 2
I cos
I'
+ \'
4'
(e)
(b) The value of membrane stress
(J
(J
from Eqs_ (10_13) is
~.
ap = ...... t(l + cos </J)
The ratio of the bending stress to the membrane stress.
 =   (I + cos (/» cos rjJ
(J
"b
2+1' t 2(1  1') a
("b)
(j
(ti)
l'
Ims a maximum value at the top of the shell (rjJ = 0), For
t = 3.29a
= OJ,
(e)
max
For a thin shell (a> 20r) the above ratio is small. It is thus seen that membrane theory provides a result of sufficient accuracy. The values of the forces No and N. may be determined by introducing Eq. (b) into Eq. (11.27). However, it can be verified that (see Prob. 11.8) these closer approximations for the forces will differ little from the results (10.13).
BENDING STl~ESSI:S IN SHELLS
249
We are led to conclude from the foregoing example that. for thin shells, the stresses (and the displacements) as ascertained from the membrane theory
have values of acceptable accuracy. The results will of course be inaccurate if expansion of the shell edge is prevented by a support. In the latter case, the forces exerted by the support on the shell produce bending in the vicinity of the edge. The local stresses caused by concentrated forces discussed in Sec. 11.7, will be treated in considerable detail in Chaps. 12 and 13.
11.1 1 THE FINITE ELEMENT REPRESENTATIONS OF SHELLS OF GENEUAL SHAPE
The factors which complicate the analysis of shell problems may generally be reduced to irregularities ill the shape or thickness of the shell and nonuniformity of the applied load. By replacing the actual geometry of the structure and the load configuration with suitable finite element approximations (Sec. 5.7), very little sacrifice in accuracy is encountered. Consider the case of a shell of variable thickness and general arbitrary shape. There are a number of ways of obtaining an equivalent shell whicb will not significantly compromise the elastic response. For example, one can replace the actual shell with a series of curved or flat (straight) triangular elements or finite elements of other form, attached at their edges and corners. Whatever the true load configuration may be, it is then reduced to a series of concentrated or distributed forces applied to each finite element. When a shell of revolution is subjected to nonuniform load. the usual finite element approach is to replace a shell element with two fiat elements, one subjected to direct force resultants and the other to moment resultants. The a pplied load may be converted to uniform or concentrated forces also acting on the replacement elements. Inplane and bending etTects may then be analyzed separately and superposed. Hence. a shell element may be developed as a combination of a memhralle element and a plate element of the same shape. The shell is thus idealized as an assemblage of flat elements. Curved elements have been proposed to secure an improved approximation of shells. but analysis employing them is more complex than is the case using straight elements. In the general treatment ofaxisymetrically loaded shells given in the next section. the latter clements arc considered.
11.12 THE FINITE ELEMENT SOLUTION OF AXISYMMETRICALL Y LOADED SHELLS
An axisymmetrically loaded shell may be represented by a series of conical jilistra (Fig. 11.8a). Each element is thus a ring generated by the straight line segment between two parallel circles or "nodes." say i and j (Fig. 11.8b). The
250
STJ{ESSES rJ'..: PLATES AND SH£iLLS
f
h
.;.....
~
!~/'.~.""~• , I
~~..J&='C=+~==..=..;;...:!/
z
(a)
L __..... r
!
(b)
Figure 11.8
thickness may vary from element to element. As before, the displacement of a point in the midsurface is specified by two components v and I\' in the meridional and normal directions. respectively. Referring to Fig. 11.81>, the "strain"· displacement relations, Eqs. (11.28), (11.30), and (11.31), simplify to
( I
IEl
= 'I' x, I' = '/
XO
I E, 180
I I
.
. (I\'
cos'" + " sin", )/r . 2 2 _d wlds I'  (dIV/ds) sin "'/r
dv/ds
I
{I 1.33)
The stressresultant strain relations, Eqs. (11.29) and (11.32), are then
or
{N" N,. M" Mol
=
[DJ{8}
(11.34b)
Here [DJ is the elasticity matrix for the isotropic axisymmetrically loaded shell. Three displacements are chosen at each node (Fig. 11.81». The element nodal displacement matrix is thus
{(51,
= j~;~.:~JJ j~~i=
(11.35)
where ii, lV, and fJ represent the axial movement, radial movement, and rotation, respectively. The displacements within the element, expressed in standard form, are
{I} =
(11.36)
)cos<Plr : 2(13s.2S..(1 + s.)cos <Plr 6(1+2s. contain six constants: lV = Cl.(3~2Iil):s~(1+s!)h 0 : 0 :.sn cos ¢II' 6(1 .) sin ¢il''': 0 +s. the coordinate s of the nodal points is substituted in the displacement functions (11. On evaluating I' and w at the nodes i and j.I) Denoting in Egs. I The following general expressions.I (I .38).)/11 s. and finally obtain' 7 lwf /ul_llS.s.. 0 : 0 (11.J.IS)j I (11.s. I = lCOS<P ~sin ¢ cos 4> smq) 0 0 0JI[..) sin <PI.(2 .38) To determine the values of the "'5.39) (a) wherein s. we can relate them to Egs.l[A]. w. = {e) = [Plio}.) sin <p/rh : (1 + 4s.) 1 d1.(1 .BE!'.) sin ¢II' : "s1(1 : 1 (11. This will generate six equations in which the only unknowns are the coefficients. (11.I dS)'l . + 2slJ cos <PI" : hs..:.. : (l [ I": o o : : 3s.2s.)/h' 6s. we can solve for '" to <7... . employed for in.3s\)/h 6s.39) the two·bysix matrix by (:1 in which [Pl.3 7) 0 I{I.'! > = [}.3sj) sin ¢lr 'C 1 o sl(32s. > ( (d':~. The matrix [N] is a fUllction of position yet to be developed.(1 .)!.33) then lead to [B]{e}.1. we have = [P11~.35) through usc of a transformation matrix. (11. I Oll (dll.. ~ [[P.. [BJPJ]{O}.3 + tX.42a) (11. (11..40) Eqs.] = .s. for example.6 in terms of the nodal displacements v" .]{b 1 1 (1 I.l[.3s. +.l].l1 [A~l{O}. [Pj][."DI?\iCi STHESSI!S [!'.lJ]{S}. (O.=1." SHELLS 251 These are to be dctermined from [()~·e and position s: slope and displacement continuity arc maintained throughout the dement. At node i.41) [B. = [[B.S + 0:582 + 0:6S3 (! 1..: 0 :t3Iii+2~::sdl~2s!+snh:O:s. ..2 : 2(2 . (11.42b) . By so doing. I " IV} ) I(dwlds).: 0 : I .
11.3. Use \' = 0. In the axisymmetrically loaded shells of revolution. fabricated of a material of 200 MPa tensile yield strength and 300 MPa compressive yield strength. M" M" arc ignored and the expressions developed in this section are considerably reduced in complexity.5 A long steel pipe of 0. if only the membrane theory solution is required.05 III apart (Fig.43 ) Clearly. PROBLEMS Sees. (11. 11. We then determine the strains from Eqs.060 m in diameter and 2 mm thick is subjected to a uniform line load P distributed over the circumference of the circular cross section at midlength. 11.1 A long steel cylinder 0.3. the stress resultants from Eqs. radius /' must be expressed as a function of S prior to integration of the above.3 and 0)/.5. Employing the energy of distortion theory.8 may now be applied to obtain the solution of the shell nodal displacements.252 STRESSES I~ PLATES A:\. (11. 11. (5. matrix for the dement is given hy Fq. 11. Clearly. Let \" =.41). For the midlength between the loads obtain. "concentrated" or "nodal" forces are actually loads axisymmetrically distributed around the shells.~(l' 11.3 A long circular pipe of diameter tl = 0. (0) the radial contraction and (b) axial and hoop stresses at the outer surface. 11. Determine the value of P required according to the following theories of failure: (0) maximum shear stress and (b) maximum principal stress. Determine the maximum values of (a) axial stress 0:". (11. by taking v = 0.5 m and wall thick ness I = 5 mm is bent by a load P uniformly distributed along a circular section (Fig. = 210 MPa. predict the value of the maximum load that can be applied to the cylinder without causing the elastic limit to be exceeded. what is the deflection at this point? 11. = 2rch I [B)'[D)[B)r dS '0 . 5. 11.5. The pipe is.4).4).4 Consider the cylinder loaded as shown in Fig. Steps 1 to 3 of the general procedure described in Sec.34).1 I (11.2 If point 0 is taken to the right of the loaded portion of the cylinder shown in Fig. = ([B)'fD)lB)dA 'A (b) Here the area of the element equals elA = 2rrr tis = 2nrh dS 1 (c) Expression (b) then appears as [k). and the stresses from Eqs.23) in the form [k).D SHPLLS The stiffnes.0.75 m in diameter and 10 mOl thickness is subjected to loads P uniformly distributed along two circular sections 0. (b) tangential stress a(h and (c) shear stress <.4).7 11.1 to t 1. . the quantities I" XII' /3.
(10. Take \' = OJ. tin. . 11.26) by two expressions containing N. (11. Compare tht' results with those giH!1l by Eqs. 11. 11.K Determine the forces N ~ and N of.OI:'<G STRESSES IN SHELU.7 Represent the equations of equilibrium of a conical sheJi (11.27). N u • M" and M(I.13)... expn:~siolls by eliminating shear 11. in a $pherical dome carrying its own weight f' using Eq.12 11. 253 Sees.BP'''.6 Reduce the dirferential equations or equilibrium (11.10 and Eqs. (h) of Sec.8 to 11.25) to force Q".
1 INTRODUCTION There are many examples of thin shells employed as structural components.7 to 12.:'1Y expressions presented in this text result from severely limiting assumption used in their derivation.3 apply to uniform and reinforced cylinders. dome. and rigorous performance requirements present difficult design challenges. 10 we observe cases in which. as needed. We now discuss the general analysis of stress and defomlation in several typical members.ainty as to applicability and restrictions of use.4. In Chap. 12.10) under uniform pressure apply to illternal pressure.CHAPTER TWELVE APPLICATIONS TO PIPES. TANKS. Sections 12. That is. 12. A comparison of the bending and membrane stresses for tanks with a vertical axis is made in Sec. when a vessel such as a vacuum tank is to be constructed to withstand external pressure. consideration is given the membrane stresses taking place over the entire wall thickness. or tower is subjected to a particular loading. The discussions of Sees.6 deal with uniform and composite heated shells. the stresses thus determined are valid only if the pressure is not significant relative to that which would cause failure by elastic instability. degree of caution is necessary when employing formulas for which there is unc'.. However. '>.2 and 12. extreme temperature. AND PRESSURE VESSELS 12. 254 . when a cylinder. They pertain equally to cases of extemal pressure if the sign of p is changed. Particular cognizance should be taken of the fact that high loadings.5 and 12. All the expressions derived for thin cylindrical vessels (Sees. liquid tank. The relatively simple form of r . 12. applying equations derived for plates and shells as well as the principle of superposition.
1 .7 that the ends of a long cylinder are not affected appreciably by central loading. into Eq.10) Consider a long pipe loaded by uniformly distributed forces Q! and moments M! along edge x = 0. of special significance in engineering) is usually divided into 1011" and short classifications. Discussed below is the bending of circular cylinders subject to edge forces and moments.18): (12. TA:\. It is clear from the discussion in Sec.1 ) QJ f~. each end can be treated independently. (11. Constants C! and C. A long cylinder loaded at the middle (Fig. 12. 11. Introduced in the Jast section of this chapter are some relationships employed in the design of plates and shelllike structures. L (a) '1 QJ (b) Figure 12. AND I'RESSLlRI' VESSELS 255 the compressive str('sses produced by this pressure in the shel1 must he lower than the critical stresses at which buckling of the walls might OCCUI'.18) can be determined by applying the following conditions at the left end of the shell (x = 0) (aJ (x= 0) The results are (b) The deflection is now found by substituting C! and C. Long pipes (Fig. however. 12.'S TO PIPES. 11. of Eq. (t 1.APPLICATIOI". of length L > (niP). therefore.4) is defined as of length L> (2nI13).2 PIPES SUBJECTED TO EDGE FORCES AND MOMENTS The circular cylinder or pipe. For shorter cylinders. the influence at the ends of a central force is substantial.KS.
11.(fJx)] ~\~ =L [2fJM 1 .(fJx). d': D L [2(JM .5) Four constants of integration require evaluation. the following expressions for end deflections and end slopes are obtained (12.17). Expression (I 1. (12. cos fJx sinh fJx + C4 cos fJx cosh fJx (12. To accomplish this. Short pipes (Fig.2) The deflection is positive radially inward. (11.1. 12. may also be dealt with by applying the general solution (11. of length L < (2nlfJ). The accompany ing slope under the loadings is (x = 0) (123) Finally.7.22). ((Jx)  Upon substituting the above into Eqs. become negligibly small for x > (nip). one obtains expressions for Nil' M. After routine but somewhat lengthy algebraic manipulations. Altogether.1) yield expressions for the derivatives.(fJx) + 0.256 STRESSES 11\ PLATES A!'\lJ SHELLS The maximum displac~lllel1t OCClirs outward at the loaded section and is given by (x = 0) (12.6) .4 ) dx' = dx =  fJD [fJM Ifl(fJ") + Qd2(f3x)] QIJ3(fJX)] 1 .17) is rewritten in the following alternate form in terms of hyperbolic functions: II' = C 1 sin fJx sinh [Ix + C2 sin (Jx cosh fJx + C.. 11.lb) The bending of this type of shell loaded along both edges. functions f. two boundary conditjons at each end may be applied. Now opposite end conditions interact. (fJx)] dx 2(J'D d'2 w (12. leading to four equations with four unknown constants C 1 through C4 .4. and Qx' The stresses are then found through the use of the dcfinitions given in Sec. successive dilferentiations of Eq. we have W = . using the notation of Eqs. As already observed in Sec. hence the minus sign.(fJx) through f:. 1.f:.20)." M".2)3[j [fJM . (11.19) and (11. and thus the results employing them.1.(fJx) + Q.J.
738 0.803 03170 0.689 2.0601 O.0 0.0300 L008 O.026R 0. AND PRESSt:l{F VEiSSELS 257 Table 12.El 2 (e) f:i}r:}.674 1.4 0.0 1.033 0. in the case of long shells h.1 lists values of ".005 1.""KS.(/iL)..2c.(/iL) as a function of /iL Interestingly.3).2 5.09() 4.0580 1. (/iL).000 0.APPLICATIO:'.6000 0.(PL) h.9770 1.cos /iL sinh f3L + sin /iL Table 12.502 0. TA.2b) is.200 0.0068 0.0 0. (10. (a) the ends are built in (Fig_ 12. 12.sin /iL " (flL) =2 sinh /iL + sin /iL (12.017 1.0 1.050 5.1 PL h.6 1.500 where h (flL) = cosh flL + cos f3L 1 sinh flL + sin /iL sinh liL .".(fJL) 0.6).2a) and the axial force N x = 0: (h) the ends are simply supported.1 A long cylindrical shell is subjected to uniform internal pressllre [1.S TO PIPES. SOLUTION (a) The problem may be treated as the sum of the cases shown in Figs_ 12.2b and 12. and ".\ = !'.4 2. as expected.+ (al (b) + (el Figure 12. The radial expansion owing to f! (Fig. and Eqs_ (1 Z. from Eq.7) ".2 .(/iL) " I. . reduce to Eqs.(/iL) = cosh /iL .(fJL) h. (/iL) '" h . (12. Determine the stress and deformation of the pipe for two sets of conditions.925 3. ".0 0. Example 12.2) and (12.lOO 0.1670 1.(/iL) " ".11).893 0.100 0.
2) and (12.2pTjj (PM. (c): .20). = .3). 2j4(PX)]  .3) together with Eq. + Qrl = 0 from which p Q. = 0 along the edges.. Upon introduction of Q.4) and (e).)I1s for the edgeloaded pipe (Fig. + Qrl = I Et pa 2 1 2f1'D (2f1M. 12. The expression for deflection..2f.2c and Eqs.8) The axial moment. 12.  EIV a vax (f) Clearly. (d). the minus sign indicates a negative shear force.4). and airplane fuselages. The stiffness.. P (d) In Eqs. 12. (b) Since in this case M. using Eqs. (12. = O. . . p a2 Et (12. 12. we have M. (g) and M. from the first of Eqs.(px) . (10 = . the first and second terms in the above represent the membrane and the bending solutions.3 REINFORCED CYLINDERS Cylindrical shells are often stilTened by rings or socalled colla.. given by Eg. size. is Mx and <J x = = 2p 2 [(. (11.s. respectively._ .2c. is then W=  4ft4V [(3(PX) p p _.' submarines. (12."'I'D SHELLS The boundary conditi. = 0 into Eq.2) then yields Q.. an expression may be derived for the displacement w.2c) is rcpresented as [ollows.258 STRESSES I:'\i PLATES . applying Eqs. We shall discuss the bending of a long pipe reinforced by equidistant rings and subjected to a uniform internal pressure of intensity p (Fig. Examples are found in pipelines. 12.. (c).6). acting as shown in Fig. (12.(px)] (e) 6M x /t 2 The circumferential stress. (12. referring to Fig. Equation (12.2p 3 D (j (q) where" is given by Eg. = _ .
given by Eg. Ql = P/2 and Ml is found by setting the slope dwjdx = 0 at an end point. In the problem under consideration. 12. reaction forces P per unit circumferential length of pipe are produced between each ring and the shell.6).\D PRFSSt:R[ VESSELS 259 Figure 12.(flL) = .(flL) (a) Then.. one assumes the stiffening rings to be very rigid in comparison with the shell. the radial displacement would then be b 1 . ~fla2 El froll1 which [h (fJL) _ hHflL).I I = P. Referring to Eg.S. applying the first of Egs.. If there were no ring.j 1 2h. (12.6).10). we obtain Er [4f3M [ "3(f3L) or M1 = f3'a' Ph. (12.11). the pressure would produce a.3 and collar spacing are important in the analysis. (10. constant hoop stress pajr. It is presumed that each ring has small Cl'osssectiOlwi dimensions compared with shell radius a and does not resist shell rotation. to the expansion b owing to 1'. (10. produced by the forces P.a. Et pa 2 Inasmuch as the rings are present. That is.9) .. TA:\I\.~(h) Ph.. (f3L)] = 0 4jji. I h1(f3L)j Pfl "lUlL) .' E! . To solve for P.. The portion of the shell between two adjacent rings may be represented by the cylinder shown in Fig. one eguates the pipe deflection under the ring. cylinder deflection is zero under the collars.2/13(flL) / = 6 1 Er ~i = P (12.1b. From the second of Egs. /\:. Cylinders with collars which prohibit dellection For this case.APPLICATIONS 10 PIPES.
p a ( 1 . we obtain At.({IL).260 STRESSJ:S IN PLATES A~D SHELl.12) for the moment under a collaI'. The radial elongation of the shell is thus . respectively. Included among these are: tanks with nonrigid or rigid bottom plates resting on the ground.10) Cylinders with collars which resist deHeclion A second type of reinforced shell is encountered when the rings arc relatively fiexible compared with the flexibility of the shell.9) into Eg..\.4 CYLINDRICAL TANKS Many cylindrical tanks may be treated using a procedure similar to that described in the foregoing sections. 12. (12.9) is replaced by el 2 = .. In this situation. (1124) which was dilTerently derived.:. . lind h. ='2jP "2(fiL) p (12.S Note tilat when /JL is large."v 2  Et  (b) For this case. tanks with an elastic roof and . The internal pressure acting at the ends creates an axial extension pal2Et of the pipe. 2Et . This agrees with Eq. in Eq. (a) and simplifying. Now 0.6b to obtain an equation used to evaluate the force I' under a ring: 1'/1 lit (fJL)  ! '2 ". The interface forces I' increase the inner radius of the ring by 60 = (~~)a = :~ in which A and Pa are the area of the ring cross section and the ring tensile loree. the detlection of the shell is /l01 zero under the collaI'. (12. = ~. and the above reduces to .1 I). the ends of the shell are assumed to be open. p is replaced by p(l. (12.. h. Upon introducing the value of P li'orl1 Eq..10) we also have At ' 2 p2 =~12(~L)(I'_~t) A (12.11) Inserting I' = P . the functions II 1 (fIL). Cylinders with closed ends In the two cases discussed above.(Pt/A) into Eq. (12.(fJL) approach unity.10) and (12. A variation of the deflection and the stress in the pipe' occurs when the ends arc closed.') 0.(PL) I hhfJL) J = P A 1'1 (12.iJ P/ia" .1'/2) in Egs.
l3a). TANKS. Thus. II'fJ.. !2.sm/J'  y(h . in concrete water tanks it is usually common to choose a linear variation of thickness increasing from top to boltom.x)a' Et (e) .cos/ix+C. The solution is thus rendered more manageable. }J:. Flexural rigidity D and thickness t must no longer be regarded as constant.  sin fix] .' (Fig... from Eq.. p" is replaced by . tanks constructed of plates of several different constant thicknesses (Fig.'(h . locating the coordinates in the figure~ we have \V = 0 dw . ...x).6 (aJ Figure f2. =0 dx (x = 0) (a) The differential equation is. . We here consider a cylindrical tank of uni/orm thickness entirely filled with liquid of specific weight. ..1!). one must deal with a linear differential equation of fourth order and variable coefficient. The tank bottom is assumed to be built in and the top is open. The particular solution ofEq. (b) is found to be . requiring lengthy manipulation. A~D I'gESSL'RE nSSELS 261 ...] +e [C.APJ. At lhe lower edge. PI2. {3""3. (ll.. (b) where the outward pressure acting at any point on the wall. (11.4a).>LJCATro\"s TO PIPES.J 5/. The general solution is thus IV = e1IX[C 1 cos fix Ox + ('. In the case of tanks of variable wall thickness. ben..4 (bJ bottom. Thc physical conditions indicate that the upper edge is free to deform and thaI no force exists there (N x = 0). d"lI' dx' + 4/' " 11' = .). However.(h x)a'IEt.x) ·0.(h..2a. solution of the problem requires integration of Eq.. but as functioIls of axial distance x.
xl px + (1).(PX) J No =  a EIW = l'ah 1 .262 STRESSES IN PLATES A ':\:D SHELLS If t is small relativt to a and II.15) . [cos 1 . after integration. We next evaluate the stress resultants from Eqs. We now satisfy the remaining conditions. (I l. Mx = . (a) into Eq. = 0 in the above expression.. provides the axial displacement 1I = Uo "0 where the constant.14) Mo= vMx The maximum bending moment occurs at the bottom of the tank. a longitudinal slice of unit width of the cylinder Illay be considered infinitely long. (e).13) where .(1 .f.12) together with (11.' (12. Equation (11. is llo = O.j~(px) .f4({JX)  [x (12.Ph p. 11. Using Table 111. .D~~~ = ~f32~2[)11 rUPx) + (1 .C ) dx ' 1 +Et i'02 = 0 from which The radial deflection of the tank is then IV =  ).pi. (hv. .13): (1 )fi(px) J I .10a) and (12. from 1/(0) = 0. (11.= P(C .jfi )f.h 2 e . as is usually tilt cast. Sill px JI I or IV =  'Ia h Et [1 _:'h: .22).15a).~)ji(flx)J .a'''] x 'Et 11 .h \1. C 3 = C.l 1) and (11. It is already observed in Sec.({Jx) and j~({Jx) are defined by Eq.7 that. because w must be finite for all x. at x = 0: (12.. the deflection at any point is easily determined from the above equation. Upon substitution of Eqs. Ph I v ~ dx + a .
A::D PRESSURF VESSELS 263 The slress resultant No.\ = 0. ... (e) and (f). The figure shows that membrane theory is valid at sections away from fixed end of a long. is from E'1. however.. and treating the terms I/Ph and . llIax = t' i'ha (e) wp = We observe that the force resultant corresponding to the particular solution. the differences in the results given by the theories are pronounced in the lower half of the shell. heat exchangers. theory.'(h . (12. we have (J. is found from 1''1.'(h _.TIO'\...Ai'PUC.y.' as small compared with unity. respectively.<s. TANKS.S TO PIPES.. "i(h . is a maximum at . Based upon membrane 1 (T'l." maximum stress is in fact A plot of Eqs. where the dashed and solid lines denote the stressresultant distribution according to the membrane and bending theories.4b.max = fi yfJa t (f) fi times larger than that predicated by membrane theory.i'a_' Dh (1 3 t Et _~ ) Ph Introducing D = Et'/12{1 . it is observed that the tr.14) and (d) is presented in Fig. x): No = '..')/at into the above.. because N. Thermal stress and deformation in thin cylindrical shells is a major factor in the design of structures such as boilers. = 0 and /I.x)a 2 /Et. Comparing Eqs. 12. NIJ assumes its maximum value No. This conclusion is also valid for other cases.. according to membrane theory. pressure vessels.14): N. m!lX = 6Mx2 m" = I=~'.300) by substituting /".x)a The above agrees with Eq. In any case. membrane theory applies reasonably well to the upper portion of the tank.5 THERMAL STRESSES IN CYLINDERS Thermal stresses are developed whenever the expansion or contraction that would normally be produced by the heating or cooling of a member is restricted. 12. (10..\ = 0. thin cylinder. and ." According to bending theory. = . = . a EIWp = . (12.~.2) and p' = ~3(1 . For a comparatively thickwalled and short cylinder. the maximum stress is therefore = 't'!la. (d) which was derived by applying membrane theory.("  x)a max (d) At . 0' .J.
(12. 12. = 2a. at points F===OZJ'x Figurt' 12. 12. > T.T. ['LATI:S AND SHELl.. at the outer surface and the uniform temperature 7. 2J.264 STl~ESSES I!\.3). i. for example.2) and (12. ao:(!l T) (12.J2(PX)] By applying Egs. + Qrl = 1 2/3'D (2PM 1 + Ql) = or M 1 = 2aa(!l T)fJ2 D Q.2a).16) At. consideration is given two representative examples of temperature fields under which thermal stresses occur in cylindrical shells. For this case. free radial expansion of thc pipe (Fig. (px)] _.1. Examine. = 1 4arx(!lT)p 3 D (b) Expressions for the deflection and the moment are then derived as illustrated in Example 12. In the shell described. Unless otherwise specified. following it procedure similar to that described in Example 12. These situations may be analyzed through application of the Egs. we thus have . In this section.. 12.S Iluclear piping. T. as positive.p2 D(!lT)[J. at the inner surface (Fig.e.. Uniform temperature distribution A uniform temperature change in a cylinder with clamped or simply supported ends causes local bending stresses at the edges. is described by Ii = aa(!l T) aa(!lT) 0 (a) Referring to Fig.16).(Px) .2jFi5 (PM.5 .. we shall also take !l T = Tl .2c.21».5). The temperature gradient through the thickness is assumed to be linear. a long cylinder with fixed ends. 12.1: l\' = ao:(!l T)[J. Radial temperature gradient Consider a long cylinder of arbitrary cross section subjected to the uniform temperature T. (12.(px) _. the stresses at any point can readily be determined as in isothermal bending. under a unifonn temperature change !l T (Fig.
) _j:~(t. m" = ~...='12(1 . This explains why thermal cracks initiate at the ji'ee edge of cylinders.17). 9. for l' = OJ.tJnped~edge circular plate (Sec. It is observed from Fig. T) '" 2. Next we discuss the stress distribution at the ed(Jes of a circular cylinder of uniform thickness with Fee ends.. and (12./. = 2\/. Observe that this stress is larger than the axial stress.30). from Eg.3(1 . TA:":l\. [ 13(1 ..S. Expressions for stress and deformation in cylindrical shells with clamped or . (12.19) The stress defined by Eq. Eqs.19). is 25 percent larger than the stress at points in the pipe away from the ends.=a..2c that the stresses given by Eg..1J_ .APPLICATIONS TO PIPES. a moment equal to but opposite in sign to that defined by Eq. (12. 12. T)t' 12(11') (12.)3(11')' + .20)... M x vE'X(t. (9.... (12. in order for an edge to be Fee.T)t' M.1') Ex(t. (e) must be applied there. T) (12.) can be determined by applying Eq.19) and (J 1.~~t. The hoop and axial stresses. T)t ' (e) Clearly.2 Therefore. arc thus <r.v) N. one obtains the deflection and stress resultants applicable to the free edge (x = 0): IV = = ipijj ~.17) mllst be balanced near the edge by distributed moments kl..Ti(1 .=G= ±i(lv) Ea(t.SSFlS 265 remOle jrom the ('lids...17) Note that the outer surface will be in tension and the inner surJace in com pression. substituting the resulting expression for IV into Egs. from which ..l8d) provide the resultant thermal hoop stress occurring at the free end. That is.18ad) M. Thus. the thermal stress is the same as in a cJ:. I  v'] (12. (J •.8)./1=.4). (12. The deflection from such a moment (+ M. (12.18c).v) Ex(t.. A1\'f) PRESSURE "F.. By so doing. (11.
6 THERMAL STRESSES IN COMPOSITE CYLINDERS The foregoing section and Sec.(I!.mi!. + O'. 12. the free motion of allY shell having different material properties is restricted by components adjacent to it. 9.22a~f) Jad 1_ I (f. He. Equation (12. In the development which follows.. T)] (12. T2)) I I[e. ."bWI:.5).3) give (f = I :. T's) (Fig.20) In the case of a cylinder subjected to a temperature lI"adiellt. used in design computations.' If a freeedged multishell cylinder undergoes a .T.)] I IX.31. .Generally. both axial and hoop stresses take place. we thus have (f 0 = E[e." PLATES AND SHELLS simply supported ends.)] 1 (12. = Nh t(a b2 + "b.2 1= 1 '. Each may have a ditferent thickness and be subjected to different uniform or variable temperature differentials.T.T..60). thinwalled shells.)A" (a) N.v. can also be obtained by modifying the cylindrical shell equations appropriate to isothermal bend ing.!'la(1!. 12)) I Ea Jab21_ Eb lab3{l:::'\~ J["o . and c refer to the individual components.(1!.)] 1 where the subscripts a.IX.266 STR[SSJ:S I!'. The axial forces corresponding to each layer are next ascertained from the stresses given by Eqs..z)A.(I!. each component shell is assumed homogeneous and isotropic. . = i(O'.\[ to .?rential temperature gradients. 1 j[s."b(1!.21) permits stress detennination as follows: Ja. h.2 provide the basis for design of composite or compound multishell cylinders. each under different temperature gradients (the I!. and ax = a o = (f (Sec. On applying Egs. 12.)A.  1='.a(1!.22) N. I J . T)] (12.~ ['" . Expressions (9.  ~(I!.wl11 temperature change I!. constructed of a number of concentric.7')]1 I(f. = I(O'" + O'. are then produced in the cylinder walls. (12. Only a hoop stress ao and circumferential strain c.2). 12. is a system of nondimensional curves J5 based upon these solutions or the finite clement met hod.1  E..:'I[so  !'l. as well as for shells under axial or cirCllmfl.21 ) Consider a compound structure comprised of three components. (9.
+ AcE.22) sLlch stresses as may be necessary to satisfy the boundary conditions._. Note that near the ends there will usually be some bending of the composite cylinder.3.= 18.. T)b + A. = 398 mm. H. and c. TA!'iKS.9 X 10. and 1'.2 The cylindrical portion of a jet nozzle is made by just slipping a steel shell over a brass sheJI (Fig.6 per·C Vs = OJ .' Steel tube (a) (b) Figure 12.. /'. T)a + AbEb~b(1'. and (b). 12.1): E" = 103 GPa E. = 200· C I'.E. The stresses at the inner and outer surfaces of each shell may be calculated readily upon inserting Eq. (12..22). assuming for the strain is obtained: Co Va = Vb = VC' the following expression = .22). = 200 GPa ~. Example 12. as shown in the previous section.1'.. (a). (I'..\<. (I'.. what stresses will develop in the two materials'l The properties of brass and steel are as follows (Table 1. The radii of· the tubes are 1'.t 6T. are the average of the temperature dinercntiaIs at the boundaries of elements a. (12. = 400 mm.23) into Eqs.g. (1223) Tn the above.. = 392 mm.6 Here the A's represent the crosssectional area of each shell. = 11. AND PRESSUn£ VE$SELS 267 6T. T)b' and (I'. and the total thermal stresses will be obtained by superimposing upon Egs. b. as illustrated in Examples 12.  Au Ea~a(1'._". respectively [e.APPLICATI(P"..S TO PIPES. . (I'.~. T)a = ttl'. A. T2 = 150·C I'.T3 = 40'C must be maintained. T.(I'. T).A~~1E 6T. (12.72)).6 per"C "b = 0.6b). + 1'.. + AbE.. If the uniform temperature differentials at the boundaries 1'.3 a._ 6T"'I7fi.2 and 12. T). T).E. (12.7 x 10.i:. The condition that the sum of the axial forces be equal to zero is satisfied if (h) From Egs.".
25) cOlltracting.~O' )1894 . (b).006 x 103 x 109 The expansion of the interfacial radius is then M. A sketch of the stresses throughout the wall thickness is given in Fig. respectively.9(40) 1 1 167/ for the stresses at the outer and inner surfaces of each shell..[t(L.:+:.(h[}(LlT.. tli anel A~ = 27ft]." = t.t:.754 mm. (.'1::'oT 11894 .24).\_T. the following values are obtained )a. \VhCfC flo and t~ arc the wall thicknesses of tbe brass and steel tuhes.26) ..6 (175) + 0. (12.iD SHELLS SOU_:TIO~ We have A" = 2m'].23).7 x 10. SOLUTION For the case under consideration the composite cylinder is (12. 12. It is observed that N.)) +lbEh(. .. (12.3 1189418..~ ./10. roc.1381 MPa 1".3 (0.894 x 10. + tj.\T.20) then appears as or (12.. Determine the hoop stress which develops in each component upon cooUng. Equation (12.002 x 200 x 10' x 11.1_200x 103J189411.002 x 200 x 109 + 0.24) and thus 0.7(200)1_/1271 MPa 1"" (.1_ ~()3.754(10') m or 0.22) and (12. Example 12.894 x 10" 0..9(150) 1_ /.3 Reconsider the preceding sample problem for the case in which a brass tube just slips over a steel tube with each shell "n(formly heated and the temperature raised by !':.:\ T.ll tsE. On applying Eq.117(150) (.005) +t(138 + 167)(0.(106 x 103 x 10" x 18.6b.E" (12. + Nh= H127 +40)(0.a.18.lL.398) = 0.9 x jO"h(95) = 1..015)= 0 as required according to Eq. Through the use of Eqs. The radial growth at other 10' cations may be found in a like manner.1 40 1 and 1".. = 1.268 STRESSES IN PLATES A. respectively.
7 DISCONTINUITY STRESSES IN PRESSURE VESSELS Ever broadening use of vessels for storage.10 can provide only a few basic relationships with all active area of contemporary pressure vessel analysis.6 can therefore be employed to obtain an approximate solution applicahle at the juncture of vessels having spherical.Ci.28) (12.. is now expressed: tfl = . .. and "'T is a negative quantity because of cooling. . 11.7 that the bending is ofa local character.tsE. 5. (12. .27) Hence the hoop stresses. expressions for circular plates and cylindrical shells are utilized. and conical type vessel heads can be assumed as /leal'll' cylindrical in shape. elliptical. + thEh~hl (12.E. T/INKS.. and experimental techniques for obtaining appropriate operating stresses.27). after simplification are found to be Eh ('" T)(C<b .24) and (12.. and environment has given special emphasis to analytical. from Eqs. 12. 12.) (12.11). Any incompatibility of deformation at the joint produces bending moments and shearing forces..2(.29) Thus the stresses in the brass and steel tubes are tensile and compressive respectively. !\ND PRESSL~RE VESSELS 269 The hoop strain. industrial processing. as "b> ". elliptical.£hl£. pressure... The finite element method (Sec..APPLICATlU:':S Hl PIPES.25) and (12. The narrow region at the edge of spherical. (12. 11.T + (t.).8 to 12.7) has gained considerable favor in the design of vessels relative to other methods. Additional details and a list of references relevant to stresses in pressure vessels are to be found in a variety of publications. the discontinuity stresses become negligibly small within a short distance.'·36 The ASME unfired pressurevessel code furnishes information of practical value for end design (Sec. referring to Eqs.. The stresses owing to this bending and shear are termed discolltinuity stresses. It is observed in Sec. or conical ends. numerical. and power generation under unique conditions of temperature. Cylindrical shell equations of Sec. Sections 12. ln the case of flatend vessels.~ + tbEb ("'T)[t. "') . Hence.E. A discontinuity of the membrane action in a vessel occurs at all points of external restraint or at the junction of the cylindrical shell and its head or elUl possessing difTerent stiffness characteristics.
if the tube and its ends are disjointed (Fig. Owing to the action of p the tube and its ends tend to expand by different amounts.7a.7b). . . Because at some distance away from the joint. through application of Eq.31) It follows that.I . = (1 ."\..11): 15.7 . membrane theory yields results of sumcient accuracy. Eg. is .= . ~ . (10. having hemispherical ends. the differential radial extension owing to the membrane stresses would be Ii = ~.v) _ t pa 2 (12..5).270 STRESSES IN PLATES 1\'...8 CYLINDRICAL VESSEL WITH HEI\USPHEHlCAL HEADS Consider a cylindrical vessel of radius a. (IO..)£(2 .: (l2.v) 2Et pa 2 {I 2.32) . 12. extension of the radius of the spherical heads owing to N... a . as shown in exaggeration by the dashed lines in the figure.lO): N = pa x 2 No = 1''' (a) For the spherical ends. (lOA) yields N= pa 2 (b) The growth in cylinder radius produced by membrane forces N x and No is expressed by Eg.0 SHELLS 12. .'. 12.I. .. under uniform internal pressure of intensity p. for the cylindrical part of the vessel we have. ' . (10._ _ _ _ _ _ _ _ _ _ _ _ _ _ _ I  (a) (b) Figure 12..30) Similarly.'.. from Egs. Fig.
34). and Eq.£/' (/3x) QI atp (12. from Eq....2) and (12.lCATlO:':S TO PIPES. (12.. TA~KS. . and hcnding moments AIl' both uniformly distributed along the circumference. yields Ql =2Et pa'/3'D p 8/3 (c) The displacement..33) and the bending moment is therefore M. is of such magnitude as to result in deflections at the edges of the cylinder and its head whose sum is equal to 6. By means of Eqs. (a)..11) and (11.) = . AxD l'RESSL:Rr vrss[[. as dMxldx = 0 at this point. Applying Eqs. As . = it + t~3/f~1'. we have flo =~) [1 .3..4).lame..32).8.. (11.. This stress is composed of the membrane solution component palt. forces Ql produce equal deformations at the edges of each component at the joint. A{1. this discontinuity o/" displacement is prevented by shearing rorces Q.vM x: (J() Substituting Eqs....6) and a bending stress resulting from the moment At (j = .fi(C.D d~2 =  d 2 \V If f. [ a t = L293't (12. Hence..1. we thus have This.34) into the above. The highest sness occurs on the outer face at ' = "/4/1 of the tube.35 ) Hoop stress also occurs at the outer surface of the cylinder.(fix)1 (d) .ilready mentioned. (12. j 1. it is found that  ax".Ji iA/3x) + 4.4). (12.. In practice.. (12. these discontinuity effects create bending stresses in the vicinity of the joint. the thicknesses of the cylinder and its ends are usually the .s 271 In [Ill actual vessd. I n this case. for I' = 0. (/3. is then (12.33) and (12. the conditions of continuity at the junction arc satisfied if JIll ~ 0 and Q.34) The maximum moment takes place at x = "/4/3. a circumferential stress produced by denection w (see Sec./3(~~::':'v2/. = ./i(~) ap Ew 6\· = . :.14).APPl.. together with definitions (1.
38) .30). (12.36). 12.272 STRESSES IN PLAnS AND SHELLS For )' = 0.032 t ap (fix = 1.~5) (12. has the largest magnitude. is most critical in the design of the vessel./3(Iv'/2 4" 11'8 ~ ap . Note that the last stress. lll<lX = 1.) = EI 1.35).32) reveals that the extension at the joint of an ellipsoidalended vessel is greater by the ratio a2 /b 2 than that for a vessel with hemispherical heads. Hence.172 2. (12. (12.m" = CT Olllall _ ap ft· + 1. Eqs. Referring to Eqs.8) follows the same pattern as employed in the foregoing section. We are led to conclude therefore that the hnap stress [Eq. (" ) = 2.0.VO". the shearing force and the discontinuity stresses also increase in the identical proportions. and (d) of Sec. t 3al' . the maximum stress. 12. (10.l.8 . ap (12. referring to Table Il.37) A comparison of Eqs. (12. (12. the maximum stress in the spherical ends is always smaller than the value given above. is fOllnd to be (f O. as before. In the particular case of a vessel for which a = 2b.8 yield the following expressions for maximum axial and the hoop stresses in the vessel: O"x. Figure 12.36) Inasmuch as the membrane stress is lower ill the heads than in the tube.9 CYLINDRICAL VESSEL WITH ELLIPSOIDAL HEADS The general procedure for the determination of the stress in a vessel with ellipsoidal ends (Fig. of the two components is thus o=i~~(~) (12.3.37) and (12. The radial increment oe in the cylindrical portion is given by Eq. 12.2 b' 2 2 v) 2 (a) The radial extension 0 = Oe .20) the increment at the edge of the ellipsoidal member is found to be a pa ( a b'="E(O"o..36)).
. referring to Eq. the slope (/. from Eqs.10 CYLINDRICAL VESSEL WITH FLAT HEADS Cylindrical vessels are often constructed with flat ends as shown in Fig. I (Xh= pa 3 8D(1 + v) ah = cc.ah = (/.v) (d) Equation (12..39) We next consider the radial displacement. Therefore. In this case.9. (12. due to Q. TANKS. according to Eq. =  2P'D (PM. and bending moments M. = 2/]2 D (2PM. and M 1 of the cylindrical portion at the joint is. A solution may be obtained by assuming that the heads bend into spherical surfaces subsequent to the application of the uniformly distributed loading p.3): a.30): 15.2) provides the increment in the cylinder radius at the joint: 15. (2.( v) D 1+'c " aMI (a) Similarly. + Qd 1 (b) The condition that the edges of the two parts rotate the same amount is satisfied by ah . (2 . 12.9 . aQ. (12.27). AND PRESSURE VESSELS 273 12.. to make the displacements compatible. The internal pressure will producc extensions in thc tube and its heads as observed in the cases of the vessels previously discussed. + Qd 1 (e) l':2z2=====~ rio Figure 12. The extension of the cylinder radius owing to p is. at the joint there must act shearing forces Q. (12. = 2Et. is expressed by 15 h = . Etll (1 .v) I pa 2 (e) It can be verified that the radial displacement of the circular plate head owing to Q. respectively. For a circular plate head the edge slopes a owing to p and the edge moment M" are.APPLICATIONS TO PIPES. each head is a circular plate.17) and (2.
formulas for computing the thickness of shell and end. The following symbols are employed in the table: p = internal pressure (lb/in') r = inside radius of shell or hemispherical head (in) 2 <J. The following factors and a host of others contributing to an ideal vessel design are described by the code: approved techniques for joining the end to the shell.11 DESIGN FORMULAS FOR CONVENTIONAL PRESSURE VESSELS It is observed in Secs. Table 12.v)p 4 k 2 + 2(2 .40).v)p 3k + 2(2 .[(co'l _V7)p":. The results are presented in the following form: Q.2C:i3k"'+.40) Finally. When a vessel is properly designed and constructed.4+~( 1+':1' )""']k+'7:( 1_:v'. M =  p 3 k 3 + 2(2 .2 lists the minimum required thicknesses.1')(1 + v) pa' . 2 (12. 12.8. however. (12. For some of the shells and the end types discussed in Chaps.V)p4 + (1 + v)]k + (f::' v2 )p 4 2p 3 k 3 + (1 .274 STRESSES IN PLAns AND SHELLS The compatibility of the radial displacements of the edges of the two parts requires that (12. maximum allowable stress values. these stresses are reduced greatly and it becomes unnecessary to consider them." maximum allowable stress (lb/in ) e = lowest joint efficiency D = inside diameter of conical head or inside length of the major axis of an ellipsoidal head (in) . and M. The AS ME code for unfired pressurevessels" furnishes formulas for calculating the required minimum thickness of the shells and the ends.c)p 8k p _ (12. . are based solely upon the membrane stresses.8 and 12. types of closures. materials in combination. 12.9 that the discontinuity stresses at the joint of a cylindrical shell and its head is quite a bit larger than the membrane stress in either portion. temperature ranges.42) The displacement and the discontinuity stresses may now be obtained by following a similar procedurc to that described in Sec. 12. All the formulas. 10 through 12. except those associated with the case of flat heads.39) and (12.1')(1 + 1') po 2p 3 k' + [(1 . using Egs. the solution is completed by determining Q.C.41) where 4 3(1 _ v')a' k=a{3= t. but the shape and proportions of the end and the manner of attachment to the shell are specified so as to avoid high discontinuity stresses. corrosion.
12.2p Wholly spherical shell t~ P" 2rre . for practical considerations: (a) the minimum distance from the joint at which the pipe diameter attains its completely expanded value.1 A long steel pipe is subjected to a circumferential load Ql at its end (fig. closures.'· Note that the preceding is only a partial description of the code specifications which.0.APPLICATIONS TO PIPES.5 m. (b) the maximum bending stress.1 to 12. Determine: (a) the least distance from the end at which no deflection occurs.6 MPa. In its present status. The juncture of the two parts is assumed to be rigid (i. reference should be made to the current edition of the code.O.4 12.:".2 Geometry Cylindrical shell Required thickness (in) 1= p" (Je .2Cp t~ P' Conical head pD 2 cos a(ae .5 for circular plates welded to the end of the shell It is mentioned that the required wall thicknesses for tubes and pipes under internal pressure are determined in accordance with the rules for shell in the code.O".1a).e.g. The radius and the thickness of the pipe are a = 0. Find.O.08 m and I = 3 mm.5m radius and lOmm wall thickness is subjected to an internal pressure of 3. and branch connections of piping systems. (b) the allowable value of Q. TANKS. c = 0. The pipe is joined to a pressure vessel. Assume E = 200 GPa and \' = 0.3.2 A steel pipe of 0. if complied WIth. . For the complete requirements. 12. e. provide assurance that discontinuity stresses may be neglected.~ pD 2(Je .2p " = half the apex angle of the conical head d = diameter of flat head (in) c = a numerical coefficient depending upon the method of attachment of head.6p) Flat head Ellipsoidal head t=dR t ~ . PROBLEMS Sees. 12. AND PRESSURE VESSELS 275 Table 12..6p Hemispherical head t ~ 002ac_'. if diametrical expansion at the end is limited to 15 x 10. w = dw/dx = 0 at the end of the pipe). the code is applicable when the pressure does 110t exceed 3000 Ib/in 2 Pressures in excess of this amount may require special attention in the design and construction of the vessels. Usc E = 200 GPa and" = 0.O.3.
9 Redo Prob. 12.025 m 2 • What is the maximum bending stress in the pipe if the collars are constructed of a very rigid material.11 Figure P12.8 A narrow ring is given a shrink fit onto a long pipe of radius a and thickness t.56. Let v = 0. P12. 12. (c) of Sec.05 m from the end. The tank is filled to the top with water (specific weight 9. (12.walled bronze pipe of radius a and wall thickness t has a bronze collar shrunk over it at its midlength.7 m. setting IX = PL/2: w max = _ pL4 64Da4 (1 _ 2cos 0': cosh a ) cosh a + cosh 2a (P12.4 12.6 A circular steel pipe is reinforced by collars spaced L apart (Fig.5 When the cylinder described in Prob. Derive the following expression.7 A long thin.4).6 m. The crosssectional area of the ring is A.5) as the homogeneous solution.276 STRESSES IN PLATES AND SHELLS 12.4) reduces to Eq. what is the shrink·fit bending stress in the pipe? 12.10 Compute the maximum bending and membrane stresses in a cylindrical tank wall with clamped base (Fig.OSm radius and 3mm walllhjc~lless is tilled with hot water under 4 MPa uniform pressure.7 m. Determine the deflection and the hoop stress at a distance of L = 0.jat3.3 A long steel boiler tube of O. 12. Verify that if the crosssectional area of the collar is A = 1. Employ Eqs. 12. The tube is attached to a "header'" in a simply supported manner at its end. t 2 <tk. and t = 20 mm.4 A short cylindrical shell subjected to uniform internal pressure is simply supported at both ends (Fig. The crosssectional area of the collar is 0.3). The upper half portion of the tank is constructed of sheet steel of thickness t2 and the lower half of sheet steel of thickness II' Determine the values of the discontinuity moments and shearing forces along the joint at nln. 12.6 for a ring made of a relatively flexible material as compared with the flexibility of the shell.3 m. Let E = 200 GPa and v = 0. Use v = 0.2. t = 10 mm.l1 shows a water tower of radius a and height" is filled to capacity with a liquid of specific weight '/.81 kN/m 3 ).4 is long.3. The dimensions are a = 2. The cylinder is under an internal pressure p. occurring at the midlength of tile shell. and v = OJ. 12. the decrease of shell diameter equals the expansion of collar diameter at the time of the shrinkage. If the outer radius of the pipe is greater by 0 than the inner radius of the ring.4 MPa.3.40). 12. a = 0. Use L = 1. 12.~I h Figure P12. 12.3.11 . p = 1. 12. show that Eq. h = 3. What conclusion can be drawn from this result? 12. (P12.4) The foregoing represenls the maximum displacement. Figure P12.
12. (b) the ASME unfired pressurevessel code form ula.18 A spherical steel vessel is rated to operate at an internal pressure of up to 2 MPa. a = 0. .2 (Fig. = lOOOe and T2 = 300"C at the inner and the outer surfaces.5 m.APPLICATIONS TO PIPES. 12. t = 10 mm. 12.6 m. 12.3.3. and \' = 0. b = 0. respectively. 12.6 Ill. 12. and I' = 0.3. e = 0.6b) for uniform temperature differentials equal to tlTI = tlT2 = AT) = 134"C. 12. a = 0.. 12. ~b= 19 x 10.and v. IT = 100 MPa. Use E" = 100 GPa. A:\D PRESSURE VESSELS 277 Sees.=O. and \' = 0. Determine the minimum required thickness of vessel according to: (a) the membrane theory.17 Obtain the maximum discontinuity and 'membrane stresses in the Ilathead vessel (Fig. TANKS.8 for p = 1 MPa. Assume that the ends of the tube are simply supported and that E.8.11 12. 12. tc = th = 10 mm. ~ 120 GPa I'~ = OJ What are the maximum thermal stresses. 12.13 A long brass cylinder of mean radius (J = 0.12 A long copper tube of mean diameter 0. The dimensions and strength of the shell are r = 0.14 Compute the stresses in the composite cylinder described in Example 12. 12.9) for p = 1 MPa.6 per"C.16 Determine the maximum discontinuity and the membrane stresses in the vessel of Fig. t = 10 mm. Calculate the maximum stress if the ends of the shell are assumed to be free.6 m.7 for p = 1 MPa.3.3 m.51012.5 III and thickness t = to mm has the uniform tempcratures T. a = 0.60 m and thickness { = 5 mm is heatcd in oil to 120"'e above room temperature 20°e.15 Determine the maximum discontinuity and the membrane stresses in the vessel shown in Fig.
This requires first that the equations governing the stress variation within the shell be developed (Sec.6).3 and 13. 278 . the approach applied earlier in Chap. and material properties of a shell. the critical loads must also be considered. 13.4). 13. described in the last two sections. circumferential.2). 13. In order to develop the relationships involving the appliedloading.5). internal stress resultants. that the deformation causing strain be related to the stress resultants through the appropriate stressstrain relationships as well as to the loading (Secs. To complete the analysis of shell deformation and stress.1 INTRODUCTION This chapter is devoted to methods for determining the deflection and stress in circular cylindrical shells under arbitrary radial. ll • 39 The former is valid when no midsurface straining of the plate occurs owing to a particular type of loading (Sec. 13. because axial buckling is synonymous with collapse of the shell. and deformations. that the governing equations for deflection be solved by satisfying the edge conditions (Sec. The critical load is particularly significant under axial compression. Buckling action underscores the difference in physical behavior of a thin shell under compression and tension. and finally. The latter applies to thin shells of very large radius or small depth. The inextensionalshell theory and the shallowshell theory are simplified shell theories. crosssectional. II is again employed. Special cases of the abovedescribed general procedure form the basis of thc theories of plates and beams.CHAPTER THIRTEEN CYLINDRICAL SHELLS UNDER GENERAL LOADS 13. or axial loads. second.
aiJM. the force and moment intensities vary across the elements.a = 0 (l3. yields the expressions oMo + a oM x • iJx ae _ aQ. To simplify the diagrams. leads to an identity and provides no (a) (b) Figure 13.1 . + p.a = 0 In a like manner.e) ax The sixth equilibrium equation. the notation N.1ac) aQ. 11. y. now requires that oNx iJN" a. 13.Qo + p..+ '(1.CYLIKDRICAL SHELLS l:NDER GENERAL LOADS 279 13. An element separated from a cylinder of radius a will now have acting on it the internal and surfaceforce resultants (Fig.6. Clearly. respectively. 13. = 0 (13. = new information. is employed to denote N x + (oN. L M.+ px a = 0 ox u iJN 0 iJN xO iJe + aax ./ox) dx.1a) and moment resultants (Fig. Equilibrium of forces in the x.ld. and z directions. after dividing by dx dO. relative to the x and y coordinate directions. elc. this time taking into account all stress resultant and surfaceloading components.2 DIFFERENTIAL EQUATIONS OF EQUILIBRIUM To establish the differential equations of equilibrium for a circular cylindrical shell. summation of moments.+iJMox_aQ 0 0(1 x 0. ax + N.1b). we proceed as in Sec. iJO + a iJQ.
(13.a = a ox x 0  EPM {/ 82Mx 0 These three expressions contain six unknown stress resultants.a _! (aD + ax2ae ) (13.2. 13. It is therefore necessary to identify three additional relationships. (13.3df) aw ax The terms of X.f . iJO 1 iJ2Mo + a aNx~ _ !~Mo _ aMx(j + p a ax a ao = 0 ox .2). which is now ds' long. It is then displaced to position A'B'. based upon the deformation of the shell as discussed in the next section. Kinematic expressions relating the midsUiface strains to the displacements are thus: e=x iJx au 1 av c·=. .3ac) ·1 \0(V The components Cx and Yxo are the same as in the case of the plates except that adO is replaced by dy.ao + aa'. stretching of the midsurfacc and change in curvature of the shell element were neglected.. (:.ao w a (13.280 STRESSES IN I'LATES A!'\D SHELLS The transverse shears Qx and Qo may be eliminated from the above expressions. 11. 10. at which buckling of shell may take place.a0 2 + 2.2a).6 by setting r 1 = a. As already noted in Sec. we may begin as in the analysis of plates and shells of revolution.Changes in curvatures. Solid line AB oflength ds represents the side view of the midsurface prior to deformation.3 KINEMATIC RELATIONSHIPS In studying the geometry of strains and curvatures of a nonsymmetrically deformed cylindrical shell.._7". are determined by considering the deformation of a circumferential element (Fig. Collection of the equilibrium equations and rearrangement then yields a~IV~ + a~xo + p ax ae a= x 0 aN. (a) of Sec. and N xy arc small in comparison with their critical values.+ No + p.. Thus.. are expressed by Xx = a2w ax' Xo = a' 1 (aD a2 w) ae + 08' Xxo . Circumferential strain Co is obtained from Eq.2ac) Y x . for thin shclls: Ny" = N ox and Mxo = Mox. Xx and Xo. having a curvature in the circumferential direction: ao/as = aO/6 ao)= l/a. Its curv~r9 then becomes ao' as'~ dO + (a'w/os') (aw)de ds ~1Y) (l. the results obtained hold ollly if the forces N x> Ny. 13. We note that in the derivation of Eqs. and twist Xx.
+ds} ds ds 2 ' (b) <l"'P. (l3.3e). \ \1 we can therefore replace w in the above expression by av/ao to obtain Eq.3f). Now consider a similar element at point B. the second element experiences the following angular displacement about the y axis: dO ox 00 ax aw a2 w (a) The same element also experiences the following angular displacement in the plane tangent to the shell ox + 8v 8(ov/ox) dD 011 u Neglecting a higherorder infinitesimal term. has a component with respect to the y axis _ iJv dO ax (b) . additional explanation is required. in Eqs.:lS2. With respect to the terms of the Xx. Owing to displacement w.:. as indicated at point A in Fig. /d 2 11' J: \ ' \ ! . Neglecting the etTect of circumferential midsurface straining. During deformation. SHELLS CNDER GENERAL LOADS 281 \ de B' ds ~dlV (a) . a2J'a Figure 13. this element rotates through an angle .2 Hence.':! ~. owing to angle dO.::. B . it is found that the latter rotation.'.. representing the twist of each midsurfacc shell element. Note that an increase in circumferential length dv owing to the radial displacement w is only w dO.2b.ow/ax about the y axis and through an angle av/ox about the z axis. Here the sign convention used is based upon the righthandscrew rule.JC . (13. CYLINDRICAl. 13.
= 2(1 + v) ax +. and w: a'u 1 . In so doing.3f). there is derived a set of three expressions in three displacements u.4aI) D[:. (13. This is accomplished by first introducing Egs.5ac) p.a' .12 ax4 + a' ax' ae' + a4ae4 t' (a'V 1 a'v) .= Mx a'w D [ ax' (av a'w)] + a' 08 + ae' v (13.(1 ..2).'" + v ax ] au I_v' aGO a Et (av lOll) N x .. we have the following stressresultant displacement relations N = x ~ [au I _ v' ilx + v(~ av .~ )dO This expression yields the result given by Eg. upon substituting Eqs. v."')] a ae a ~ [~av .a' iJli + 12a' ax' 00 I a'w) t' [ iJ'v I a'v] P (1.] .12a' ax' ao + a' ae' =  (13. (13. Mx= M.4) into Egs.= _D(I_V)~(av + ax ae a'w) a ax Then..4 THE GOVERNING EQUATIONS FOR DEFLECTIONS The expressions governing the deformation of cylindrical shells subjected to direct and bending forces can now be developed.va'u I + v a'v yaW Px(1.ae N = .v') ~ . ax = Et I + v iJ'u I . (a) and (II) that the total angle of twist between the clement at A and B is equal to Xx. (13.V iJ'v 1 iJ'v 1 iJw t' ( iJ'w ~ iJx iJe + 2~ iJx' + a' iJli' .(:~ +:~~')+v:.Y Et 4 4 v au 1 av w t' (a 4 w 2 aw I a w) aax + a' ali .3) into the constitutive eguations (11.v') ax' + 2a' ae' + ~ ax ae .v') + a' ae' + 12a' (1 . + . 13.v) ilx' + a' 08' = . (13.282 STRESSES IN PLATES AND SHELLS We observe from Egs..2).a de =  (a~2.
( 1 .. (13.w) 4 p.6 and Chapter 12.t' +au +1.5) or (13. which in many applications.~ 1+ v iFu 1 va'u 1+ v iJ'v ~ ax ao + 2ax' + a' 00' . Based upon the simplifications which lead to Eqs. (13.\DRICAL SHELLS l '.v 4 a' u 1 a'v vaw ax = 1 ow 4 Px(1 Et v') (13. (13.+.6)..9).2 for the thin sheJls (t <i' a).CYU.7) . Several examples of such a case are given in Sec. By so doing.a' 00' ax' 9 (13.. Approximate relations Equations (13. yield numerical results of negligible practical importance.5) contain a number of terms.4).3 and 13.32 The last two terms on the lefthand side of Eq. (13.(1 _ ..lor the displacements in thinwalled circular cylindrical shells under general loading. 11.w . By dropping these terms of small magnitude. (11...') a 00 Et When a circular cylindrical shell is loaded symmetrically with respect to its axis. Such quantities are of the same order as those which have already been neglected in Secs. the following simplified set of gouel'lling equations for displacements of circular cylindrical shells are obtained ax' + 2a' 00' + ~ ax 00 . the effect of displacements It and von the bending and twisting moments must be regarded as negligible.' ax Et [ou + v ~ 00  (1 au a w)J N = ~ (~av _.a' 00 = a'u 1. as expected the solution (13.5b) and the last term on thc lefthand side of Eq. the simplified elastic law is then N x = 1 _ . much time and effort is saved in the solving for the displacements.. 11.". From Eqs.JDElt GE:\ERAL LOADS 283 These arc the fJOL'erllillg equatiolls .5c) are of this type.= ..) Et "au .6) can readily be reduced to the form given by Eqs.6) p.(a w 2 a ax a' ao a' 12 aw 4 ax4 a' iJx' 00' 1 +4 a. ll .'" + " ax OU) 1 _ .' a ae a Et (au 1 au) N xo = 2(1 + v) ax + ~ ao a'w v a'w) Mx = D (ax' + a' iJO' 1 a'w O'W) Mo= D (.
5 A TYPICAL CASE OF ASYMMETRICAL LOADING A pipe supported on saddles at intervals or at the edges and filled partially with liquid is a typical case of axially unsymmetrically loaded cylindrical shells. ..) <J. in each particular case. A particular example of such a solution is illustrated in the next section.8) in which the angle 8 is measured as shown in Fig..284 STRI:SSES IN PLATES AND SHELLS We observe that the prohlem of circular cylindrical shells reduces.3a. Both ends of the tank may be taken as supported by end plates so that.9) _  ~== =co _ __ ~~~~:~ IL ... at x = 0 and x = L: V= 0 w=Q Mx=O (a) It is observed that the expressions (a) and the conditions of symmetry of deformation are satisfied by assuming the following displacements 1I cos nO cos ' m=111:0 L <Xl mnx v = L L: bm~ sin nO sin . The upper part (a.3a). . For the lower part (0. Consider a circular cylindrical tank filled to a given level with a liquid of specific weight y (Fig. j f t (a) (b) t a Figure 13. 13. 8 . a). = p = ya(cos 8 .) mnx (13. 8 .3 ..6)..m=lll""O L = Q mll 00 L I <J. 13.5) or (13. rr) of the tank carries no load.. to the solution of a set of three differential equations (13. 13..cos (X) px = py = 0 (13. The latter set is often preferred for practical applications. p.
I3.1 Calculate the maximum values of lateral deflection and stress in the tank filled with liquid (0: = rr).notation )..cm . 3..[2m2 n 2 + (I bm.8) and (13. shown in Fig. L = 25 cm. (13.7) and (13.2). .=_ W 2 2 (13.R GENERAL LOADS 285 Here the angle ~ defines the liquid level. = 8ta(cos 0: sin /1a .10) and using the above notation are.(h + a cos 0) and the coefficients are Pmn = 0 4)'h PmO 4ya Pm! = Inn = mn (m = I. = _ P Lt "m:".[(1 .a .12) for any m and 11. bm ".3vAmn . A. 3. Cl  cos a) Pm! =  2ya nm (20! . upon substituting (13..I ) ' Cl (m =_ I.2vAInrr = 0' am.1la) PmO = 2 4. l3. ) 11  2. can be determined in each particular case upon application of Eqs.'11 = 0 _'1')1.\2 . Example 13. lib) If the cylinder is/illed to capacity (Fig.. (13.CYLINDRICAL SHELLS LNDF. .(l + v)AInl1n + cm . determined in the usual manner (Sec. are of the fOlm P.. (13.. 1'=0./1 cos /10: sin eI) mmr 2(112 .bm. by setting .3.3).C.. therefore: am ..b.(i + v)Aml1rr . .v)m 2 rr2 + 222112]. t = 7 em. is given by expressions (13.11) or (b). The stress resultant and displacement at any cylinder location may then be ascertained using Eqs.3A211 + C_ . It is usefuHo expand the foregoing load into a Fourier series: fJ =  I I m ~oo . Inasmuch as Pm.". 3.sin 20:) (13.= L a ry= t 2L The governing equations (13.L mnx (13. SOLUTION As a = h. the parameters am".+ M2(m 2rr2 + 22/12)2] [3.6).10) The coefficients Pm .2/12] a. . (SIll mn .12) Cm . the loading is expressed P = ).3b. ) (b) We now introduce the . Use: a = h = 50 cm.8).3b)... n Pmn cos 118 sm .
It is observed in the foregoing example that the coefficients am".525 2(IO') 57.00503 1212 6. diminish very rapidly.8) suffice to yield fairly accurate results. 5 and 11 = 0..lt (e) For 11 = 1. (13.1 .7)' Through the use of Eqs. 8 = O. (13. On applying the third of Eqs. force.71' NI cm 2 as the maximum axial and tangential stresses in the tank.26 0. from Eqs.8) and (13. in the case of longer and thinner shells am". .66. bm and em" ". and moment are found at x = L/2.4).m" = 1763.) _ 3 7 49 . Nlcm N N Mx.7)' 6(1763. . .mox =5657.8).00122 for 11 = 0.7): N x.704 0.5 and n = 0.742 0.mox =  2481.526 49.21' Nlcm = 1607.286 STRESSES IN PLATES AND SHELLS Table 13. bro" and em" diminish rather slowly.1 for m = 1.m"  _ 1607. We conclude therefore that only a few terms in the series (13.51' M.21' _ 6(5657. The numerical values' o of the coefficients are furnished in Table 13..768.0432 0. the expressions for the coefficients become more complicated.1051 0. These will be computed by taking m = 1. (11..12) and (b) yield vZ. and it is necessary to calculate a somewhat large number of coefficients in order to obtain the deformation and stress resultants with sufficient accuracy. we then obtain = _ 2481.18 0. Eqs.7 E y em In a like manner.3. m = 1.88 0. 1. The largest values of the lateral deflection.. 3. 1.1073 0.1 m I 3 5 2(10') llmozr Cmo 21 2(JO') ami z. 3. However. 2(IO') b nol 2(IO') Zt CmlZ"c 1183 6.7). mox N'.7).51') = 10472 0" x. Y N/cm "O. we have Wmax = CIO + C II  C30  C31 + Cso + CS1 = 11.00499 . (13. max 7 49 2 . together with the given data.
13b) oe Ot' (b) The conditions (13. 13.6 INEXTENSIONAL DEFORMATIONS Associatcd with the il1extensional d4ormatiol1s of shclls. the midsurface inplane strain components given by Egs.13a) that" depends on leads to W= (a) e only.. (13. we shall dcal only with the inextcnsional deformation of circular cylindrical shells.=O We also observe from Egs. Hence. there is considerable bending caused by the changes in curvature.. e) act and lengthening of the horizontal diameter occur.\:DIHCAL SJlFLLS UNDER GENERAL LOADS 287 13. However. (13.=N. we obtain Nx=N. Deformations of these types are thus described as inextensional.13) are thus fulfilled by assuming the displacements as follows: 11 (13. Such cases include: a cylinder subjected to loads without axial symmetry and confined to a small circumferential portion (Fig.CYU. a cylinder with free ends under variable pressure p{x.4ac). l3. a simplified shell thcory is useful under certain conditions. In inextensionalshell theory.O) where g" p{x. The inextensional theory is often preferred when shell structures resist loading principally through bending action. The conditions o!illextensibility are therefore 0" = 0 ox 10.4b).13ac) Upon introduction of the above in Egs. but 110 stretching of midsurface length. In both situations shortening of the vertical diameter along which P or p{x. and Egs.4 . The illextel1siollal shell theory is applicable to a variety of shell forms. 0) de '" 0 (Fig.3ac) are taken to be zero. (13.+ov=0 a oe ox (13.14) (a) (b) Figure 13. (l3.4a).
3) into Eq.16).(b" sin nl! + b" cos nl!) n 5" sin nO) 00 1 _ v.17) Inserting the displacements wand v from Eqs. = a(a . a". (13. + (ll")'] + tL'n'[b.18) Omission of the term n = 1 in the above expression is explained as follows. cos 0  Q.1. + (b")'] . all cross sections of the cylinder deform alike. b". (13. sin 0 + a cos 0) . = a 2: (a" cos n8 //= 00 Q" sin 0) (13. 11. is then D('[(OV G02 u = 2a4.16ac) x 2: nib" sin nO + 5" cos nO) n=1 In the above.( [xt + 2(1  v)x~. According to expressions (13. (e). sin 0) w.v)a'[b. a represents the radius of cylinder.17) takes the form U = nDL _ J. = x w. A shell experiences inextensional deformations under a prescribed load in such a manner that a minimum strain energy is produced. (n' a3 I)' {a'n'[a.15ac) 1 W.288 STRESSES IN PLATES AND SHELLS with U. = 0 V. Since expressions (13.5): U = D "2. and b" are determined for each loading situation. The constants a". + 2(1 + (b")']) (13. one has "1 = a(a .4a). =  2: (b" cos nl! 11=1 00 (13.14) for a cylindrical shell of length 2L (Fig.]a dO dx (e) The strain energy of a cylindrical shell undergoing inextensional deformation. = a 2: n(a" sin 8 + Q" cos I!) /1=1 00 and '" = a 00 n=1 2: .. upon substituting Eqs. the required expression for the bendingstrain energy of the shell is given by (Sec. Eq. (13.14) yield o'w/ox' = Xx = 0.15). For n = 1. 13. The displacements varying along the length of the shell are given by the last two equations of (13. J 00 + o'w)' + 2(1 00 v)a' (O'W + ox a d8 dx ox 00 Ov)'] (13.
)2] + 2(1 . It is clear that the terms involving an and bn in Eqs.W is therefore rr = nDL 2. 13..: In(n 2 1 nDL •  3 + fsmne (1319) 1)2 (n' ..1)2 [tn' 13 + 2{1 . (1.c=~ n • a2 p n(n2 _ 1)2nDL (n = 2. are w(O) and win).14) become. a =  _ 02(1 + cos IIn)P • 211nDL(n2 . in this case.2 A circular cylinder of length 2L is loaded by two. nOn[1 + cos nn]lp Applying Eqs. Equation (13.)a2] I cos ne . The same conclusion holds for dispacements L/" v" and w. equal and opposite concentrated forces a distance c from the midspan (Fig.. The potential energy rr = u . and are readily evaluated: v.1) 2 nDL[tn 213 + 2(1 ..16c) become zero at these points.v)a 2 ] nca'P a. Po.{n'[a'(an)' a + W(0.v)a'] . (13. . = 0.. The remaining terms with the coefficients an and On in w contribute only to the work prod uced by the forces. 3.. ) V and = _ (n' . from conditions coefficients a.. The ends of the shell are free. Derive the equations describing the elastic surface. n2 cx nex I. = 0 for n = II = I..15c) and (13. Pa 3 w = nDL ~ <Xl I 1 In(/1' _ Tj' + (/1 2 _ 1)2[tn213 + 2(1 _ .).: 11=2 00 (n' _ 1)2 .40. = O.18) is useful in determining the components of shell displacement as is next observed. 4. SOLUTION The radial displacements at x = c. Equations (13.' nDL 00 / ~ 00 ac cos nO (n 2 _ 1 )2 [tn2 13 + 2(1 .1)'[tn 213 + 2(1 .~. nan[1 + cos nn] + c. Example 13. ) (n = 2. .1)' o• = or nca'(1 + cos IIn)P 2nDL(112 .43). 4.. .V)a2] Pa v=2. in the direction of the loads. arr/a.. = 0 and orr/v.v)a 2] ii=c.CYLINDRICAL SHELLS UNDER GENERAL LOADS 289 The foregoing describes the rigidbody displacement of a circle in its plane and hence does not contribute to the strain energy of the shell.v)a 2(onfl + laJ.
9 may be applied.137 2DL The displacements at any section of the cylinder may be found similarly. bending. as ascertained from w(1l/2) + w(31l/2).. we describe the fundamental concepts of the energy approach by considering the symmetrical buckling of a thinwalled circular cylinder under uniform axial compression. equals 2Pa3 00 (_I)(n/2)+ 1 Pa3 (e) Wh = 1lDL ~ (n' _ I)' = 0. lin are uniquely determined by the applied load. a more general case of buckling is discussed which employs the equilibrium approach. the methods described in Sees.. the edge bending due to displacements (13. is w.149 2DL (d) At the center cross section the horizontal increase. In the section following. When a circular cylindrical shell is under uniform axial compression. this increase in energy must be equal to the work done by the compressive load owing to axial straining and bending as the cylinder deflects due to buckling actiOn. Hence. the vertical reduction of the cross section at the center.8 and 1. At the start of buckling. Thus. To check the stability of the elastic equilibrium of compressed shells. (11. in inextensionalshell theory. (13..32 a critically important area of engineering design. axisymmetrical buckling of the shell may take place at a particular value of compressive load (Fig.19) by applying Eqs. it is observed that thin shell structures are often subjected to compressive stresses at particular areas. At the critical value of load. (d) and (e). from w(O) + w(1l).4). and axial compression. bl!' an. 1. In this section.5). We thus provide only an introduction to shell instability. the critical load is that for which oU = oW (a) . When the loads are applied at midlehgth (c = 0).4).290 STRESSES IN PLATES AND SHELLS where II = 2. the shellstrain energy is increased by the following: midsurface straining in the circumferential direction.24. 13. 110 boundary conditions can be fulfilled at the free edges of the sheIL However. The expressions for the moments can be obtained from . It is observed in the above solution that all coefficients an.7 SYMMETRICAL BUCKLING UNDER UNIFORM AXIAL PRESSURE In numerous examples of the preceding chapters.. 13. We shall determine the critical value of compressive forces N per unit circumferential length. The stresses are then determined from Eqs. the general solution (13.19) is of local character and does not have a pronounced effect on the accuracy of the results given by Eqs. = 1lDL ~ 2Pa 3 00 1 (112 _ Pa 3 1)' = 0. 4.
CYU7"DRICAI SHELLS UNDER GENERAL LOADS 291 1. and e92 are the axial and circumferential midsurface strains after buckling. E exl E = 1' (ex. =Xxy =0 (g) . Referring to Sec. YXY =%. From Hooke's law. It is therefore assumed that the axial compressive force does /lot change during the buckling.Et N where t is the thickness of the shell. 1nn:x L (b) Here Co is a constant.·~i I i Figure 13. the latter strain is expressed C0 2 = Ve x 1  w a= vex! + mnx a 8m L Co . + V£. As the cylinder deflects. (d). the radial deflection satisfying the boundary conditions is w= CoSIO .6. 1. The axial midsurface strain before buckling is given by ex! = .S This is the principle of virtual work (Sec. L is the length of the shell.L~.9).) v (c) or (d) in which ex. If the cylinder is simply supported at both ends.. it becomes shorter in length. and III is the number of halfsine waves in the axial direction. (e) Inserting the above into Eq. (f) Owing to the axisymmetrical deformation. 11.
(e) to (Il).v2) '" l.. is applied to find the critical value of N. 2 2 4 4 (j) The criterion for buckling.7) and (11.mnx n m I.}< bU = 2ntEvc.eJi!te!!h'_~q. Eqs.d '0 . the length at which the minimum buckling load has effect: 4 Lw = " aI 2( 1 . The corresponding minimum buckling load is (13. We thus determine..5).21 ) The above is the length of halfsine waves into which the shell buckles (Fig. the increase in strain energy during buckling must be determined. (i) and (j). it is required that (13.L Co sm L. the work done is bW = 2nN [ \' /' '0 nm:~' I. It follows that nEIL ( y. (11.5 am"D 2U For the above to be valid for any co.ex l) and the bendi~g_ of_£h~_ll€cllerlllo!}ul. After addition of bU.I"D SHELLS and the change in curvature in the axial plane is (i1) As N remains constant during buckling.. are introduced into Eqs.' dx + / L 2'0 Q ''( mrr Co  Co mnx)2 J sin . 13.dx L L (i) The components of this work are associated with the change of the axial strain (e x2 . we have ~ f·'].22) . for v = 0..292 STRESSES IN PLATES A:.~. oW = bU. For this purpose. (b)~ Next..+ .8).dx + n cf.72jai 1 (13. together with Eqs. The lowest critical load is of greatest interest and is found by setting the derivative of N" with respect to L equal to zero for m = I.."sin . and iiUm.20) As for columns and plates.EtL + cO2 naLD 2 " a r.3. for each value of m there is a unique bucklingmode shape and a unique buckling load.
(13. as is done for plates and columns. These components. + a aM.o ax + aN x ax' iJ 2 v . When the compression load N exceeds Non given by Eq.(iJ 2V/iJ 2X) dx .e. 13. We are led to conclude that the straight column is unstable when N > N. iJO a +a iJN. yields a y component N .o_0 iJN. are then IF. ox ae ' . the straight cylinder is in stable equilibrium. If N < No. unintentional deviations from the assumed initial state of the structure.. The solution Co = 0 therefore represents an unstable or a stable configuration according to whether N is greater or smaHer than Ncr.v = Et 0. However. IF.1). = iJQ. iJN.24) ao ao 0+· aQ=O aM. = 0 0 0 iJQ. 7. small. assuming p. a iJx + iJe  8N.. = 0..2. the system is unstable as the work done by N is greater than the increase in strain energy. it is found that the force N.. except N" are very small.'· i.(iJ 2 w/iJ 2 X) dx· a dO in the expression = o." If N = N" the cylinder is in neutral equilibrium for small displacements. cylindrical shells often buckle into axially nonsymmetrical forms. oM.23) for the critical stress for axially symmetric buckling of a simply supported circular cylindrical shell under uniform axial loading. (13.605a (13. ax + ae + aN. _ aQ.Q. after division by dx dO. a dO in the ·expression = 0 and a z component N. This case of instability is considered here through application of the equilibrium method.CYLI~DRICAL SHELLS lJ~l)ER GFNERAL LOADS 293 We therefore have the relationship (fee = a v I Et 2) 3(1 . This approach is based upon the assumption that all surface forces. a2 w (13. the product formed of these forces and the derivatives of displacement are neglected. The equations of equilibrium. are added to Eqs. owing to initial impel/eetions.1) for shell displacements to include axial load effects. We shall first modify the governing equations (13.8 NONSYMMETRICAL BUCKLING UNDER UNIFORM AXIAL COMPRESSION The preceding section concerned simple buckling of an axially loaded cylinder.22). = py = p. ax2 + No = aM. Thus. Following a procedure similar to that given in Sec.
a2 00' = 0 1 + V iJ'u 1 . II' and their derivatives.. + ~ 00 20 = 2 (13..a a'v a4 1 iJ4 w w a4 w) 1iJ'v (~ ao' + a iJx' iJO + a' ax4 + ~ iJ04 + 2a iJx' ao' = 0 (1.294 STRESSES Il\" PLATES Al'<D SHELLS The shear forces Qx and Q. (13. J (13.27)  (1.4) to express all stress resultants in terms of u. In "0 doing.v 1 a2 u iJx' + 2. .ax' + a2 ao' .aq ax'+ v ax + ~ao . we have a aN . It is noted that Eqs.27) are satisfied for the particular case in which the solution is expressed in terms of constants c 1 and c2 . + oN...26) the differential equations of the buckling problem are expressed as follows: a2 u 1 + v a'v vaw 1 .v a v) iJ 2 v 1 a v 1 iJ'w + ( a' iJ0 2 + a' aO + ax' ao + 2..ax 2 . 13.25) a2 w 0 We now employ the elastic law (13. v. a solution is obtained for the axisymmetrical buckling treated in Sec.ax ao + 2.7. may be eliminated from the above equations.V a'v 1 iJ 2 v 1 ow 2. when one assumes v = 0. and u and w to be functions of x only.o = 0 ax ao aN'+aaN xO + aN a v _oMxo_!aMo=o 2 ae ax x ax ex a ae a2 M i'PM 1 a2 M aN x ax' + N. + aax'" + 2 ax .. Also.q ax' = 0 a2 w au 1 av w .. Taking the compressive force N x = N to be positive and introducing the dimensionless parameters N(I .v2 ) q= Et (13. v=o This represents the cylindrical form of equilibrium wherein the compressed cylinder uniformly expands laterally.iJx ao .~ ax + 2.a' iJO 2 2 aJw 1 .
Hence. as unknowns. S mnx 00 00 A = mna L (13. (13. 2 +mn b [(1 . dq/d'l = 0.)(1 + ~)A2 2 + (I +rJ. (13. bm " and cm . .q).. This is attributable to the raet that the edge conditions have only a minor influence on the magnitude or the critical load provided that the shell length is not small (say L > 2a).28) can be used irrespective of the type of edge supports. (13.CYLI:JORICAL SHELLS UNDER GENERAL LOADS 295 The general solution of Eqs.l1q ) 2 A2l (13.30) + cm.2 + aW + 112)2] = 0 The foregoing represent three linear equations with buckling amplitudes a. and we shall therefore neglect the terms containing the square or these quantities.L (13. this expression reduces to Eq.28) v= = LLb m . Observing that the minimum value of q takes place when 22 and n 2 are large numbers. Usually a and q defined by Eqs. the buckling condition of the shell.. 2 To determine the minimum value of q. When we introd uce the notation '" /I .31) becomes q = a'l + (1 .29) and substitute the solution (13. v). is determined by setting equal to zero the determinant of coefficients..n + bmn 2.V)A 2 1 . (13.20). the results obtained from Eqs.28) into Eqs. The minimizing condition.27).2) (n 2 + 22)2 q= Et" = IX A2 + (112 + A2)2 (I _ 1')22 (13.. • mn cos nO mnx L Sin .V 2) amn (A + 2. W C1 L For long cylindas. we let 'I = (n 2 + A Y/A 2 Equation (13.[n + an(n 2 + A2)] = 0 am..~ L.27). the trigonometric functions drop out entirely. The nontrivial solution.+ emn VA amn = 0 n(1 + vjJ. and we find that n(1 . if the origin of coordinates is placed at one end of the shelL can be expressed by the series CI U =  va 00 x 00 + C2 + L I a 00 c£. leads to 'I = )(1 . the expanded determinantal equation then results in N(I . mll Sin m " nmx n8 cos .1')/'1. + bm. (13...31) We see that ror n = 0. the result obtained for symmetrical buckling...[1 + a(112 + 22)] + cm . + " " emil SIn 11e' I n .[1 ..v2 )/a.26) are much smaller than unity.
3.2 m diameter and 12 mm thickness is used in a structure as a column. (13. constructed of long steel piping of O. Use r = 5 mm. . many of which are difficult to analyze. Assume that all the other data are unchanged. what is the critical stress? 13.2 During a stage of firing.O) Equation (13. which is ·based upon the smalldisplacement theory.605Ka Et (13. Note that the critical stress depends upon the material properties. What axial load can be tolerated without causing the shell to buckle? Use E = 200 GPa. for shell structures with end restraints other than those of simple support.23).l. is in satisfactory agreement with the tests for cylinders having Lja < 5. based upon the coefficients 42 K = 1 .32).1 Detennine the maximum deflection wand stress in the tank described in Example 13.5~mlong steel pipe of 1. Eq. initial imperfections. and radius.4 A water tower support. For example. E = 210 POa. The value of critical stress defined by Eq. (13.8m diameter. as in the cases of plates. v = OJ. for v = 0.296 STRESSES IN PLATES AND SHELLS The corresponding critical stress is therefore (J" = t.32). while it is independent of cylinder length. which takes into account the squares of the derivatives of the deflection w. PROBLEMS Sees.3 A 2. thickness. This is especially true.3.j'~3('71~_~v"fi2) Ncr Et (13.0. 13. becomes (J" = 0. often does not agree with expcrimental data.e. thin cylindrical missile casing of 1 m radius and 25 mm thickness. (13. This discrepancy is explained by applying the largedeflection theory of buckling. is to carry an axial compression load of 450 kN. 13.1 to 13. To relate the theoretical value obtained in this section to actual test data.32). What should be the minimum thickness of the shell? Assume that v = 0. is loaded in axial compression.901(1 .33). There are many other kinds of shellbuckling problems.p = 400 MPa. and u). which was derived in a different way for axisymmetrical buckling. and a host of additional factors. and O'yp = 300 MPa. a long. 13.33) The result.32) This coincides with Eq. (13.= a. it is necessary to incorporate an empiricalJactor in Eq.8 13. IF E = 200 GPa.
and make use of the orthogonality relations: mnx IL cos L.0 x (A.APPENDIX A FOURIER SERIES EXPANSIONS A.2n.I SINGLE FOURIER SERIES The Fourier series are indispensable aids in the analytical treatment of many problems in the field of applied mechanics. Here L is a nonzero constant. 4n:.cos . .I) by cos (mnx/ L). To determine the coefficients a" we multiply both sides of Eq. for sin x there are periods 2n.. mcx ) (AJ) in which Qo. The Fourier expansion corresponding to J(x) is of the form: 10 J(x) = 2" + '~. n I.L mtrx sin L L d . L) and determined outside of this interval by J(x + 2L) = J(x) is said to be periodic. The representation of periodic functions using trigonometric series is commonly called the FOllrier series expansion.. integrate over the interval of length 2L. . A functionJ(x) defined in the interval (L. of period 2L.2) ( sin L L mrx 0 sin :nnx dx = 1 L " \L m'!" m=1l 297 . cos L 1l1rx + b. .L cos .. (A. 00 ao ( a.p and bn are the Fourier coefficients.!n m=11 for all m. a.L 0 "nx d 1 L x=\L Il7rx m. Sill L .. for example.
=0. = L 1 nnx L/(x) cos L dx . L) results in b" = L L/(x) sin L .J (0) sin n 1l 1 .n sin x sin I1X dx o =0 b. I . (A.3). ) (A. = . . J . j Y. = n and b. Upon application of Eq.0 1 nx dx + 1 j 1t· . ) (A.L 1 rlnX dx (n = 1.n '0 sin 2 xdx=2 .n sin x cos x dx = 0 0 = .L (..298 STRJ::SSES IN PLATES AND SHELLS Iv) 1 .3b) Example A.1 (0) cos nx dx + .3a) Similarly. Figure A.1 sin x cos nx dx n'n n·o = _ ~(cos nn + cos nn _ _ 2_) 2n 1 ..n 2 ) 1 (/1 +1) a.l) by sin (mnx/L) and integration over the interval ( . A. 1 =1 " . . (A. The period 2L = 2n and L = n.n a.!.1..L. mUltiplication of both sides of Eq..I In so doing.I Develop the Fourier expansIOn of the periodic function defined by f(x) = 0 if n < x < 0 f (x) = sin x if 0 < x < n SOLUTION The function is shown by solid line in Fig. we obtain a..2.0 1 .n 1 + 11 1 _ /1 2  1 + cos 111' n(1 .
1.0 f( il) ("ltU)(du) .2. this integral is therefore 1 L Because 'L /.e. we substitute x = .. . . = 11 1 y.=/ f(x) cos L dx+.J(x) cos '0  Ilrrx d L x (n = 0. and in particular X. for an even.. We can rewrite Eq. . Substitution of Eq. nnu L du (e) Here.. in the figure. i. and x = implies u = 0. respectively.. the approximating curves approach the graph of J(x) .... is thus (x) = ~ . L).cos ( . the evaluation of the Fourier coefficients can be simplified by employing a halfrange series over the interval (0. the coefficients a..=1t +2 sin x 2 cos 2x 31t are sketched by dotted and dashed lines.3a) as 1 . 2.+(.u and hence dx = duo Inasmuch as x = . for 11 = 0. y axis. (a) then yields.L implies u = L..2 HALFRANGE EXPANSIONS In numerous applications.0 mrx 1 . (a) In the first integral above. (A.=cos ° (b) f( u) = f(u) expression (b) becomes nit" ) nnu L . ) (A./ fill) cos L· o 1 . J(x)=J(x) for all x.3 3 15 + .FOURIER SERIES EXPANSIO~S 299 The required Fourier series.4) .I.. Viewed geometrically.L mrx a. For example. Assume first that J(x) is an even function. a graph of even function is symmetric with respect to the vertical. the dummy variable of integration u can be replaced by any other symbol./ f(x) cos L dx L '0 L'l.. x 2 and cos x arc even functions. periodic functionf(x). 1. (e) into Eq.: Q" 2 = L /'1.L . A. It + sin x 2 _ ~ 11 cos 6x cos 8x (COS 2" + cos 4x +35. ) The first and the first three terms 1 y. L cos . We observe that as the number of terms increases.
. at the points of discontinuity. < 2) in (a) sine series and (b) cosine series.. .. X . The graph of an odd function (e.g. ' . A. A. 1. ) (d) It is observed that the odd function is discontinuous x = 2 and x = . 3. (A. (a) We shall treat f(x) as an odd function of period 2L = 4.2. 2. x and sin x) is skewsymmetric. (A. The halfrange single series expansion of various common loadings is listed in Table 3. . /4 / 2 / / / 0 2/4 / / 2 0 4 (a) (b) Figure A..2.. SOLUTION The function is indicated by the solid line in Fig. 2n. A procedure identical with that described above yields 2 b" =L ··0 I f(x) sin . For the sine series a" = 0.. The extended function is shown in Fig.. ) (A. f(x)= f(x) = O.X _ ./ x / / /'.. 4 (.3b).2. it is readily verified that b" In the case of an odd function. and the sine series (d) converges to zero. Example A.5) gives 2.300 STRESSES IN PLATES AND SHELLS f(x) / f(x) / / / / / 2 / . 2 / / /'.. 2 bn = 2 x '0 I .l...2 By using Eg. nnx dx L (n = 1.2 Determine the halfrange expansion of the function f(x) = x (0 < .5) and a" = o. . nx sm T . the mea" value of the function... SIn  "nx dx 2 =  ~ cos nn 4 117[ Hence 00 f(') = L n=1 4 "nx cos nn sin Ill[ 2 =..2 sm 2""" +"3 sm 1. .2a by the dashed lines. 4 A / / . and Eg. for all values of x.
7) . multiplied by a halfrange sine series in y.(cos 2 2 4 n n '17<  1) (n =f 0) ao =.lSIONS 301 (b) Now the definition affix) is extended to that of an even function of period 2L = 4 (Fig.. we can expand pix. That is pix.3 DOUBLE FOURIER SERIES The idea of a Fourier series expansion for a function of a single variable can be extended to the case of functions of two or more variables. (3. however. A. . Note that the latter representation converges more rapidly than the former.. h" = 0 and 2 When 0 . J/m" sm 00 co co . using for the period of expansions 2a and 2b. y) = I Pm(Y) sin nmx In""! a (A.. Eq. our concern is only with the interval 0 < x < 2.2h). For instance. respectively.FOURIER SERIES EXPA:.2a and A. A.2 "0 a" = 2.6) with 00 Pm{Y) = n= 1 I Pm" sin nny b (a) Treating Eq.. a mnx. (A.2 x dx "0 = 2 It follows that fix) = 1 + I <X) II"" 1 4 22 (cos nn Iln  1) cos 2 mrx 8 ( nx 1 3nx 1 = 1 .5) is applied to yield Pm{Y) = . A. y) into a double Fourier sine series.2b. respectively. (A.1a): pix. Eq. As/(x) is even. pix. x cos 11 = I1nx 2 dx = . y) = m2:..a "0 a mnx y) sin dx a (A. sm b mry The above represents a halfrange sine series in x.6) as a Fourier series wherein y is kept constant. 2 .cOS+'cos+cos n2 2 32 2 52 (e) The series (d) and (e) represent the extended function shown in Figs.
. (c) This agrees with Eq.2 which was derived differently. for Eq.302 STRESSES IN PLATES AND SHELLS Similarly. (a): Pm . the results can be obtained for cosine series or for series having both sines and cosines. = Ii 2 j'b 0 Pm(Y) sm . y) sin sin .Q.7).. .3) of Sec. together with (A.b mnx Il1[Y Pm" = b p(x. The doublesine series expansion of some typical loadings is illustrated in Sec. 3. then leads to 4 _.dx dY b a ·0"0 a I I . (3.3. Similarly. 3.l11ry b dy (b) Equation (b).
2. A system of linear equations GllX 1 aUxl + a l2 x 2 + . .I) can be written in the matrix form (B.... + a l 1/xl/= C 1 + anX2 + .2) equal to Eqs.I).. + a2l/x" = C2 (B. (i. the terms of each row of the coefficient matrix [A] must be mUltiplied by the unknown terms of the column matrix {X}. 303 .1 INTRODUCTION Simultaneous algebraic equations are encountered frequently in the numerical treatment of plate and shell problems. The quantities au and c. (B. (B. In order that Eqs..2) or concisely [A){X} = {e} wherein the rule for the matrix multiplication is evident from the original equations. /1) are treated as constants. ..j = 1.APPENDIX B SOLUTION OF SIMULTANEOUS LINEAR EQUATIONS B.
involving a minimum number of computational steps. 1A 1 =1 o.I X2 1/4 x 31 3/21 0 1 . .43 applied to facilitate the solution of simultaneous equations. Most methods require a knowledge of the basic theory.. and the behavior of matrices.2). The development is based upon the assumption that the determinant of the coefficient matrix is nonzero. operation. Note that the same matrix row operations that are performed on [A] must be performed on {e). To accomplish this divide the first row by (pivot element a 11) 20 subtract . This is the simplest practical procedure. In the next section is described a common elimination approach. by means of digital and analog computers and desk calculators. some are well adopted for hand computations while others are more suitable for electronic computers. B.. (8. consider an 11 x n system of equations (B..2 THE GAUSS REDUCTION METHOD The object of the Gauss reduction or Gauss' method is to eliminat~ the unknowns systematically from the simultaneous equations. and Xn = Sn· As a first example.3) or Xl = Sl' X2 = S2' .304 STRESSES 1. consider the problem of determining the unknowns in a 3 x 3 set of equations: 20 2 [ 5 14 1 4 311xll114f o jX 2 = j3 1 X3 (a) 2 The Gauss reduction can be carried out in the following steps: Step 1 Transform the first colum to the unit matrix. When these operations have been properly completed. the unit matrix [I] will appear on the left. and the column matrix of the solution IS} on the right lO That is. To describe the method. PLATES AND SHELLS There are many techniques 16.2 times the first row from the second row subtract five times the first row from the third row The result of performing these three operations on both sides is [~ 7/10 2/5 1/2 3/2 7/10 3/10 ] lXd~= J8/5. Numerous modified forms of the Gauss reduction process exist.
we obtain o [o 1 0 1 0 3/8] 3/4 1/8 IX'i lX 2 = X3 7/2 ( 4 1/2 Step 3 In this case. Consider. 1 o o 1 0 I X2 r = 1 21 1/ 41 The successive transformations have now been completed. One observes that a value for x.delimination process. while the other elements of the second column are transformed to zero. yields (b) [j o 1 2 3 1 3 11 6 . the case of a 4 x 4 set: 1 2 3 1 2 4 356 [ 4 2 1 The lirst step of the operations. Thus.3/4 times the third row from the second row In so doing. can be ascertained (without applying the operations of step 3) by back substitution. one has 1 0 0] lx' I [ 0 1 lx. values of X2 and x. For this purpose divide the second row by 2/5 subtract . = 2. X2 = 1. on both sides we must divide the third row by 1/8 subtract .7/10 times the second row from the lirst row subtract 1/2 times the second row from the third row Carrying out the above operations on both sides. Thus. We thus have x.SOLUTION OF SIMULTANEOUS LINEAR EQUATIONS 305 Step 2 Similar operations are performed with the second row. having been found at step 2 by the described Jorwa.3/8 times the third row from the lirst row subtract . = 4. the third element of the third row is the pivot element and the third column must be transformed to a unit matrix. each calculation involving only one unknown. this element must be transformed to unity. Inasmuch as the until matrix appears on the left. Now the second element of the second row is the pivot element. the matrix on the right is the solution. as a second example. and x. described in preceding example.
= 2. successive transformations to unity of the second. = 0. third. X2 = 1. the interchahge can always occur between the equation with zero pivot element and an equation below it. interchanging the second and the third equations. the second column must be reduced to unity. and x. one can continue the elimination: 2 1 3 [! o 3 1 6 11 It is noted that at the beginning of any step. This is impossible because '22 (current pivot) is zero. and fourth columns will develop 3 [!l 1 3 1 7 1 4 2 1] xd = x2l X31 112 4 2 x. [ go o ~~ :.306 STRESSES IN PLATES AND SHELLS Next.] 0 3~ [ o~o ~ g g] ~~f='1 if 0 1 0 0 0 1 X3 X3 x. . However. Then. 0 1 We thus have X. = 1.
130134. J. Marcus. Timoshenko and J. . Smith. 17. G. Peterson. Springer. Ketter and S.). A. P. 23. A. Wiley. PrenticeHall. Szilard. C.. Johns. M. Williams. D. 24. R. 1925. 19.REFERENCES 1." Proc. 110. 8. Sokolnikoff and R. Huffington. J. F. Prawell. 10. McGrawHili. Air Force Flight Dynamics Lab. 1949.. "Stresses in Laterally Loaded Disks of Nonuniform Thickness... S. Pister and S. 1960. Gawain and E. S. Wiley. Numerical Methods in Engineering. Thill Plate Design for Transverse Loading. M. C. 22. McGrawHill. 1968. Ootober. R. 13. G. S. 16. D. Formulas/or Stress alld Strain. 1962. 1967. Spartan Books. Bernhart. Stability. B. "Theoretical Determination of Rigidity Properties of Orthotropic Stiffened Plate. Mech. Nadai. D. 1975. Gordon & Breach. B. Structllral Design of Missiles alld Spacecraft. P. J. Abraham. G. W. E. 1971.. 4. Durry. 1962. Fiugge (ed. McGrawHill. M. 6." Prod. P. K. L. 1956. 11. Eng. 1969. S. L. Proceedings of the Second Conference on Matrix Methods ill Structural Meclianics. Weiner. McGrawHili. 1981. 1959. Miner and J. Theory and Analysis of PlatesClassical and Numerical Methods. 2. pp. Die Theorie elastischer GelVebe. 1974. 3.. Wiley. Vinson and T. 2d ed. H. 1969. Wiley. 27. Elastic Bending of Layered Plates. AFFDL~TR68150. Theory of Plates and Sheils. McGraw~ HilI. K. Theory ofTlJermal Stresses. 1974. 1932. Wiley. McGrawHill. Jr. R. Elsevier. C. Handbook of Engineering Mechanics. Elsevier. B. 1961. 1965. D. 25. Analysis of Plates. H." J. Wiley.. R. Aalami and D. ASCE. R. J. Advanced Strength and Applied Elasticity. L. Lekhnitskii. L. Ugural and S. T. 1950. Chou. McGrawHill. and Vibrations.1966. Gere. Boley and J. S. 21. Roark and W. Berlin. 1955. 307 . 15. Die elastischen Platten. I. 26. Thermal Stress Analysis. Theory olElas!ic Stability. C. 84. W. Stiffened PlatesBending. M. Appl. B. December.). Fenster. Baron. 7. Troitsky. 14. H. Mech. Salvadori and M. Pergamon. Langhaar. McFarland. S. RedhefTer. WoinowskyKrieger. 1975. Handbook of Engineering Materials. 1968. Stress Concentration Design Factors. Wright~Patterson AFB. Modern Methods of Engineering Computation. Dong. Mathematics of Physics and Modern Engineering. W. 20. Energy Methods in Applied Mechanics.. Eng. A. 18. Anisotropic Plates. Timoshenko and S. The Finite Element Method in Engineering Science. 1976.. PrenticeHall. S. L. 1975. H. 9. Young. 23. 1972. and W. Zienkiewitcz. McGrawHili. 12. Div. D. Composite Materials and Theil' lise in Struclures. Seastone (eds. J. s. B.
. R. J. 1962. 33. O. L. pp. D. 1977. J. N. T. "Stability of Thin Walled Tubes Under Torsion. 18. Wiley. W. Gould. J. Design of Piping Systems. 1974. H. Mag.308 STRESSES IN PLATES AND SHELLS 28.. 38. Engi1leeri1lg Design. 1956. . Quick Way to Calculate Thermal Stresses in Cylindrical Shells. Vinson. Goodier. 1953. Elementary Statics oJShells. Deformation of Thin Cylindrical Shells Subjected to Internal Loading. Van Nostrand Reinhold. Almroth. IIi. 20(1). 42. A. J. L. Lin. "Similarity Laws for Stressing Heated Wings. 7. 39. Wiley. Heyman. 1977. ) 32. 36. J. O. 30. McGrawHili. 7477 of lJesign Data and MefhOtis. Theory and Design of Modern Pressure Vessels. P. 1974. Pfluger. Des.tIall. ASME Boiler and Pressure Vessel Code. A. Aeru. September 21. H. Tabakrnan and Y. 1961. Shoup. 29. H. 1975. 37. Plates.. 40." pp. Wiley. pp." N AC A Rep. E.. Strrlctural Mechanics: The Behavior oj Plates arul Shells. 34. 31. Fliiggc.. 1933. 41. Buckling oj Bars. . The M." Phi/. ASME. and Shells. Prentice~. A Practical Guide to Computer Methods Jor Engineers. F. 479. ser. 43. Tsien. . Lexington Books. 1964. Brush and B." J. Harvey. Sci. Static Analysis oJ Shells.1978. 1979. Section VIII. "Thermal Stresses. Kellogg Co. Donnell. McGrawHili. January." Mach. 1934. Stresses ill Shells. Equilibrium oj Shell Structllres.W. J. 10991116. Springer. 35. Wojtaszak.. Oxford University Press. 1. S. Faupel. H. 1953. D.
1] CHAPTER 3 3.1(a) 3.59 mm " m = 0.5 O"x = 0".1.15 2.16 2. = (Pb'/16nD)[2 In (bla) + (alb)' .00328 Pa'lD (b) O".12 2.ANSWERS TO SELECTED PROBLEMS CHAPTER 1 1.(ap)'] Mx=t(2MXOo~) DC 6( a +b U = .6 1. ab 0.1.3) Wm " Wm " =Pa'18nD(ln a .9 Wm " = 0.m"= 0. = Etco 132(1 .836 Pic' 309 .=4Plnr' + 0..3 2.m.3) w.22 N= 3.4(a) Po = 126.v)][xy .·.0929 ) CHAPTER 2 2.9 W = [MoID(1 .1 t = 1.v)a' r xy = 0 1.1625 a' b' 4 4 O"x.0237M 1 a'lEt' = poa'148(ln a .396 kPa W (b) Wm " = 11.7(a) 1.28 mm 2.
03 mm WZ.36kN'm H = 17. e/Wmal(.10 Po = 0.011 MPa/m Wm " = 0.y) Po = 56.10 5.6 (b) Po = 78.0055 poa /D 0.706 x 10.00283 poa 4 /D = W m " = 0.3(a) 4.68 kPa = 2.4 5.310 STRESSES Il'\ PLATES AND SHELLS 3.483 x 10.21 3.480a(Npo/u yp )1/Z k = 1.28 3.6 3.5 kN' m Dxy=O 6.00697 a'(P l + P.5 5.7 5..25(a) 3.9 6. c R = 2Mo/J3 4. = 110.= 0.0102 poa 4 /D = 0.6 6.6 po = 6.17 3.496a(Npo/u yp )1/Z (b) t = 0.25 kN'm 6.1 Dx= H= 146.m" = 0.31 t = 0.003211 poa 4 /D poa 4 /D = 0.6 pO .2 M AC = (M o/2a)(J3x .5 kN'm Dx= 19.12 U m" = 2.26 kPa Wmax .43 kN'm Dy = 1346.14 po(a/t)Z W3 W5 W5 = = Wmox = 0.107/a4 = MPa m m Wmox Wm " 0.271 CHAPTER 5 5.0112 poa 4 /D CHAPTER 4 4.m" = Um" Wmal( 4 = 1.04506 Wm " Wz = Wm " = 0.03335 poa 4 /D CHAPTER 6 6.24 \V.2 D.1986 po(a/t)' 3.)/D a = 679 mm Wl.
5(b) O'x=308.965P Pa CHAPTER 12 12.195P Pa 0'.93 mm (b) Q.0447 "..10 = uo/[1 + (Na 2 /2.815 m (b) 1.189 kN/m 10.9 M* = 146.14 10.p = 896 kPa 2.347.2 Dla 2 N" = 4. m" = 19605...407 m "0 = pair N= 2.18 N = 40.7 N" = 2.8 N M x.1 10.3 12.75 MPa .6 x 10.3{a) 10..P/D 7 11.6 L = 18.2 /l..I{a) 12.2Dla 2 N" = 6. = .0045(1 + v)aBa'r' CHAPTER 10 10.5{a) P = 96. = 42.86. m .1 9.9 7.=543.16 N N* = 380 kN/m N x = Ny = 542.893 kN/m 2w = 218 X (b) P = 121.6 m 0'.2D)] CHAPTER 9 9..2 Dla' Wm .555 kPa 129.3 7.371 kN/m w = 0.6 aAEt' W =  0. 7.4 "x = vpalr CHAPTER 11 11.08 kN/m 11. = X 10.2 10.1 Pm•• = 81..08.A~SWERS TO SELECTED PROBLEMS 311 CHAPTER 7 7.3 9.86 kN/m Mx = My = 208.3{a) 11.
246 kN . b = 339.16 "rn" = l.18(a) t = 5 mm (b) t = 6.13 12.312 STRESSES IN PLATES AND SHELLS 12. mox =  5655.68 MPa = 60 MPa 12.27 mm CHAPTER 13 13.761.5 MPa "O.3 W m" = 164.5(P/t")[a 2 t' /3(1 .8 12.m = 67.1")]''4 "'.12y N/em 2 P" = 74.1 13.2y/E em "x. mox "0.
192. table. 174 by edge moments. 47 uniformly. 176. 46 Boundary conditions. 217 Average curvature. 19 pure. 30. to anticlastic surface. 44 at center. 177 by reciprocity theorem. 17. 249 Bending of plates. 20 polar coordinates in. 46 by superposition. 41 differential equation for. 199 Asymmetrically loaded shells. 184 on elastic foundation. 16 in using finite differences. 47 Clamped edges. 100 rectangular plates with. stress concentration around. 199 Betti. 140 Annular plates. 15 for sliding edges. boundary conditions for. 27 with circular hole. 2 methods in. 118 Buckling (critical) load. 167 by lateral load. 16. 33. 194 deflections of. table. 26 Bending (general) theory of shells. 203. 35. of circular pia tes. 40 with clamped edges. 32. of classical plate theory. 34 for simply supported edges. 2 of classical shell theory. 30 for free edges. 37 boundary conditions for. 18. 102 Circular plates. unit. 168 to developable surface. 19 under combined loading. 31. 37 non uniformly.INDEX Anisotropic plates. 8. 42 under linearly varying load. 44. 36 elongations. 38 thermal stresses in. 8 Approximate investigation of bending. 191. 30. 2732 to paraboloid of revolution. 3. 32 with simply supported edges. 40 Anticlastic surface. 75 effect of shear on. 90 313 .30. 6 Axisymmetrically loaded shells. 283 of rectangular plates. 19 to spherical surface. 159 having initial curvature. 15 expressed in polar coordinates. large. 47. due to. 153 to cylindrical surface. asymmetrically. 80 Assumptions. 16 Boundary conditions: practical aspects of. 290 Circular hole. 4. for builtin (clamped) edges. 193 elliptical plates with. 189 types of. 16 in thermal bending. 37. 243. 189 expressed in finite differences. 44 loaded. 176 of cylindrical shells. 193 of variable thickness. 15.. E. 156. 35. 118 circular plates with. 32.
173 differential equation for. conical. 100 with simply supported edges. uniformly loaded. 257 Cylindrical tanks. supported by beams. 189. 239 partial pressure. 289 equivalent radius for. in pressure vessels. 218. 272 Ellipsoidal shells.18. 42 rectangular plates on. 63 uniformly loaded. 293 classification of. 77 Developable surface. 100 of laterally loaded plates. 3. 163. plates supported by. 247. plates under.174 Differential equation. 234. 59 under concentrated load. 211 Elliptic plates. 83 supported by columns. 127 nodal. 46 by superposition. 74 of infinite length. 242 pressure. 280 theory of. 13 orthotropic. 289. 17. 263 composite. of finite element. 278 a typical case of. 41. 193 of elliptical plates. 188 for bending of cylindrical shells. 247 Displacement. 292 Curvature. 99 under concentrated load. 13 large (see Large deflections) limitations regarding. 72. applied in bending and buckling. 220 thermal stresses in. 6. 200. 168 with simply supported edges. 17. 20 principal. 167 Cylindrical bending of plates. 17. 154. formulas for pressure vessels. 255 buckling of. 290. 282 reinforced by rings. 90 with simply supported edges. simply supported. 17. 245 Continuous rectangular plates. 89 Elasticity matrix. circular plate under. 37. 239 of general shape. 236 a typical case of. 32. 96 uniformly loaded. 90. 207. 20 of triangular plates. for rectangular plates. 278 under uniform. 159 in calculating large deflections. 166 for cylindrical shells. for bending of plates. energy methods in calculating. 11. 239 governing equations for displacements. 102 Energy methods. 153 Concentrated load. 248 Elastic foundations. 4 average. 168 with clam ped edges.26 partially loaded.314 INDEX Columns. 93 Conical shells. initially. 284 axisymmetrically loaded. 142 subjected to thermal gradient. 127 matrix. 42. 209 spherical. 93 due to temperature change. 260. 96 bent by edge moments. 205. 86 Combined loads. 6 change in shell. 71. 65 under sinusoidal load. line load. bending of. 179 . of circular plates. 232 Curved plates. 220 kinematic relations of. 62. 165. 258 asymmetrically loaded. 244. 191. 127 Discontinuity stresses. 37 cylindrical shell under. 17. 20 of cylindrical sheils. 212. 168 with large deflections. 128 for orthotropic materials. 44 due to temperature change. 287 geometric parameter for. 11 under combined loads. 3 Deflection: of rectangular plates. 279 for bending of shells of revolution. 150 Ellipsoidal ends of vessel. 86 Critical load. 99 by reciprocity theorem. 278 membrane. 12. 190. 38 rectangular plate under. 219. 274 Dome. 155 to cylindrical surface. 126 function. circular plates on. inextensional. general (bending). 156. 66. 169 Cylindrical shells. 12. 292 of plates. 266 under general loads. 100 with clamped edges. 192. 191 under hydrostatic load. 269 Design. 179 under lateral loads. 78 (See also Rectangular plates) Deflection. with clamped edges. 68 with mixed edge conditions. bent by edge forces and moments. bending of plate to. 2 differential equations for. 280 gaussian. 284 Deflection. 255 deformation of. 66. 209.
of plates. 111 operators used in formulating. 189 expressed in finite differences.. 16. 8 transformation equations for. 123 Finite element. of plate. 273 Flexural rigidity. Mohr's circle. 282.. plate of. method of. 297 Free edge. 23 Membrane equation. 287 Infinite length. 26 with simply supported edges. 17. 13. 225 Finite difference equations. 188 Failure. 176 Large deflections: comparison of theories of small and. for simply supported plates. L. B. 293 Initial strains. G. 18. 5 Moments. 270 Images. 173 with built~in edges. ItO for orthotropic plates. table. 232 represented in terms of displacements. 181 for small deflections. 2 Gauss' method. J. 140 basic relationships for. displacemen ts. 179 differential equations for. L. in bending of plates. 151 triangular. 148 transformed rigidity of. G. 96 Inextensional deformation. properties... modulus of. hypotheses. 217 Membrane stress. 13 Multilayered. for plates. M. comparison of bending and. 17 under uniform load. 26 Initial curvature. 3 shear force. 12 Large deflections. 169. 202 Methods. 220 in the form of surface of revolution. 179 finite element solution of. 249 Flat ends of vessel. square. H. formulation of. table. 66. 181 theory. 149 Navier. 174 Local stresses. energy method in. 175 calculation of. 169.. in bending of plates. 126 rectangular. strain energy in. under concentrated load. 17. cylindrical shells. to cylindrical surface. 140 Hemispherical ends of vessel.. 129 Orthotropic plates. 173 Levy. 169 under sinusoidal load. 266 plates. 127 forces. 199 cylindrical. properties of. 30. 122 for deflection of rectangular plates. 80 Gridwork system. 142 of shells. III Membrane forces in shells. 128. 235 Forces at boundary of plates.. 174 approximate method for. 304 Gaussian curvature. 128 for shells. 167 Initial imperfections. 203. 66 Marcus. 129 Finite element method. 24. 38 Long rectangular plates. 128 Kirchhoff. 20 Midplane of plate. 144 Equivalent transverse load. for rectangular plates. 2 Navier's solution. 17. bending of plates with. 118 Galerkin. boundary conditions for. 283 Momentsum. 59 Nodal. 60. 2 of shell.INDEX 315 Equivalent orthotropic plate. 198 Modulus of foundation. 173 under concentrated load. 2 Kirchhoff. 15 Lagrange. 163 for circular plates.26. 43. 133 for large deflections. 180 Meridian plane of shell. 247. 2 Levy's solution. 152 . 119 for skew and triangular plates. for buckling of rectangular plates. 141 bending of. 15 Foundation. 145 for plates with curved boundaries. H. 8 of orthotropic material. 203 Membrane theory of shells. 17. 26 of shells. application of. theories of. I Midsurface. C. of cylindrical shells. 20 Grashof. 43 Fourier series.26 bending of. 13 Material. of plates. 247 Membranes. deflection of. 133 orthotropic. of curvatures. 16.
57 for large deflections. 20 expressed in polar coordinates. 258 Ritz. 151 partially loaded. 189. 148 orthotropic. 219. 68 Ring reinforcing. 142 Parallel plane of shell. 63 variously loaded. 8. 235 Stress resultants. 191 under uniform load. of plates. 13 sign convention for. 127 Stress concentration factor. 280 Strain matrix. 248 Strain energy in bending.211 flexural rigidity of. in bending of plates. 86 deflection calculation. 159. 147 properties of. of isotropic plates. 93 continuous~ 83. 269. 46 Rectangular plates. 102 rectangular plates with. 202 Plates. 272. 26 Long (see Long rectangular plates) with mixed support conditions. 245 constitutive equations for. 90 under concentrated load. 38 elliptical plates with. 118 circular plates with. on deflection. of plates. boundary conditions for. 72.S. 19. 128. 199 local buckling in. 152 of shells. 74 of infinite length. 4. 31 strain energy expressed in. 203. 232 Pure bending of plates. 273. and stretChing. 189 expressed in finite differences. 65 of semiinfinite length. 6 planes of. 179 of orthotropic plates. 16 Spherical dome. 66. 74 thermal stresses in. 218. 62.316 INDEX Orthotropic plates: circular. 209. 62. Lord. 71. 30. 293 conical. 236 Strain energy in buckling. 78 multilayered. 144. 77 . boundary conditions for. 19 Radius of curvature. 234. 21 ellipsoidal. 6 Stress matrix. 59 Skew plates. 41. 198 general procedure for. 245 Simply supported edges. energy method of calculating. 127 Stress. 63 Reciprocity theorem. 207. by energy method. 190. 31 Pressure vessels. 18. 200. 59 under sinusoidal load. 231 Stress. 215. 207. table. 6. 234 Rayleigh. 160 of shells. 274 Principal curvature. 235 general behavior and theories of. classification of. 2. 11 effect of. 278 of general shape. 217 axisymmetrically loaded. 8 in shells. 80 (See also Deflection) on elastic foundation. 73 with simply supported edges. 46 Reactions at boundary of plates. 90 by finite element method. 89 under hydrostatic pressure. compound. deflection by. 1 general behavior of. 174 Polar coordinates. 232 change in. in shell. in plates. 133 by finite difference method. 235 cylindrical (see Cylindrical shells) deflections of. 33. asymmetrically loaded. 75 with clamped edges. 290. 12. 36 Sheils. 205. 144 rigidities of. 17. 9. 146. 202. of cylindrical shells. thermal (see Thermal stress) Stiffness matrix. 160 Strain in shell. 57 stresses expressed in. 147 elliptic. 125 Sliding edges. 243. 205 of tocus. 249 breakdown of elastic action in. 27 differential equation for deflection in. 80 Superposition. 205. 270. 22 Semiinfinite rectangular plates. method. 249 in the form. 15. 278 Shearing forces. 73 Shallow shells. 212. 201 spherical. 129 Strip method. 211 load resistance action in. 29. 30. 143 rectangular. 292 Strain in plate. 59 bent by edge moments. theory of. of space of revolution (see Surface of revolution) some typical cases of. 224 compression of cylinddcal. 102 Stress in plate. 202. 111 by strip method. 5. 213. 7.
effect of. plates under. 191 Thermal stress resultants. 72 simply supported. 90 continuous. 260. 205 Tanks. 192. 185. 190. 193 III rectangular plates. 211 Torsional rigidity. 36 on stresses around hole. compound. with simply supported edges. 32 elliptical.212 cylindrical. 71. 227. 198 Thick plates. of orthotropic plate. 217 axisymmetrically loaded.245 membrane forces in. 100. 1 Thin shells. 5. 24. 2 Wind pressure. 204 particular cases of. 243.228 conical. 263 composite. 3. analogy between. on conical shelter. 187 in cylindrical shells. shells having form of. 191. 21 von Karman. 47 Virtual work. A. displacements of. 99 under uniform load. M. 86 with mixed supports. 284 spherical. 191 Thermal strain. 83. 224 Thermal and isothermal problems. 104 on bending of plates. 186. 187 Thin plates. 219 on spherical dome. 215. 99 Twisting moment. 68 triangular. 179 Wahl. 142 Transverse shear. 185. 98 under concentrated load. 218 . on deflection of plates. 62. 102 rectangular. 189.249 asymmetrically loaded. 18 Triangular plates. Th. 8 Twist of surface. circular. 96 subject to edge moments. principle of. 213 Theories of failure. 249 equations for determining. circular plates.. 266 ". circular plates of.INDEX 317 Surface of revolution. clamped.. 1 Toroidal shells. 202. 99 Variable thickness. 186 Thermal stress. 234 Uniform load. 203.
j j j j j j j j j j j j j j j j j j j .
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue reading from where you left off, or restart the preview.