Draft
Lecture Notes in:
FRACTURE MECHANICS
CVEN6831
c VICTOR
E. SAOUMA,
Dept. of Civil Environmental and Architectural Engineering University of Colorado, Boulder, CO 803090428
May 17, 2000
Draft
ii
Victor Saouma
Fracture Mechanics
Draft
Contents
I PREAMBULE 3
1 1 1 1 1 4 5 5 5 5 6 7 8 8 9 9 10 12 14 15 1
1 FINITE ELEMENT MODELS FOR for PROGRESSIVE FAILURES 1.1 Classiﬁcation of Failure Kinematics . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Continuum Mechanics Based Description of Failure; Plasticity 2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1.1 Uniaxial Behavior . . . . . . . . . . . . . . . . . . . . . . . 2.1.2 Idealized StressStrain Relationships . . . . . . . . . . . . . 2.1.3 Hardening Rules . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Review of Continuum Mechanics . . . . . . . . . . . . . . . . . . . 2.2.1 Stress . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.1.1 Hydrostatic and Deviatoric Stress Tensors . . . . 2.2.1.2 Geometric Representation of Stress States . . . . . 2.2.1.3 Invariants . . . . . . . . . . . . . . . . . . . . . . . 2.2.2 Strains . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.2.1 Hydrostatic and Deviatoric Strain Tensors . . . . 2.2.2.2 Invariants . . . . . . . . . . . . . . . . . . . . . . . 2.3 Yield Criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4 Rate Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5 J2 Plasticity/von Mises Plasticity . . . . . . . . . . . . . . . . . . . 2.5.1 Isotropic Hardening/Softening(J2 − plasticity) . . . . . . . . 2.5.2 Kinematic Hardening/Softening(J2 − plasticity) . . . . . . .
. . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . .
3 Continuum Mechanics Based Description of Failure; Damage Mechanics
II
FRACTURE MECHANICS
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3
1 1 2 3 6 8
4 INTRODUCTION 4.1 Modes of Failures . . . . . . . . . . . . . . . . . . . . . 4.2 Examples of Structural Failures Caused by Fracture . 4.3 Fracture Mechanics vs Strength of Materials . . . . . . 4.4 Major Historical Developments in Fracture Mechanics 4.5 Coverage . . . . . . . . . . . . . . . . . . . . . . . . . .
Draft
ii 5 PRELIMINARY CONSIDERATIONS 5.1 Tensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1.1 Indicial Notation . . . . . . . . . . . . . . . . . . . . . . . . 5.1.2 Tensor Operations . . . . . . . . . . . . . . . . . . . . . . . 5.1.3 Rotation of Axes . . . . . . . . . . . . . . . . . . . . . . . . 5.1.4 Trace . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1.5 Inverse Tensor . . . . . . . . . . . . . . . . . . . . . . . . . 5.1.6 Principal Values and Directions of Symmetric Second Order 5.2 Kinetics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.1 Force, Traction and Stress Vectors . . . . . . . . . . . . . . 5.2.2 Traction on an Arbitrary Plane; Cauchy’s Stress Tensor . . E 51 Stress Vectors . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.3 Invariants . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.4 Spherical and Deviatoric Stress Tensors . . . . . . . . . . . 5.2.5 Stress Transformation . . . . . . . . . . . . . . . . . . . . . 5.2.6 Polar Coordinates . . . . . . . . . . . . . . . . . . . . . . . 5.3 Kinematic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3.1 Strain Tensors . . . . . . . . . . . . . . . . . . . . . . . . . 5.3.2 Compatibility Equation . . . . . . . . . . . . . . . . . . . . 5.4 Fundamental Laws of Continuum Mechanics . . . . . . . . . . . . . 5.4.1 Conservation Laws . . . . . . . . . . . . . . . . . . . . . . . 5.4.2 Fluxes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.4.3 Conservation of Mass; Continuity Equation . . . . . . . . . 5.4.4 Linear Momentum Principle; Equation of Motion . . . . . . 5.4.5 Moment of Momentum Principle . . . . . . . . . . . . . . . 5.4.6 Conservation of Energy; First Principle of Thermodynamics 5.5 Constitutive Equations . . . . . . . . . . . . . . . . . . . . . . . . . 5.5.1 Transversly Isotropic Case . . . . . . . . . . . . . . . . . . . 5.5.2 Special 2D Cases . . . . . . . . . . . . . . . . . . . . . . . . 5.5.2.1 Plane Strain . . . . . . . . . . . . . . . . . . . . . 5.5.2.2 Axisymmetry . . . . . . . . . . . . . . . . . . . . . 5.5.2.3 Plane Stress . . . . . . . . . . . . . . . . . . . . . 5.6 Airy Stress Function . . . . . . . . . . . . . . . . . . . . . . . . . . 5.7 Complex Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.7.1 Complex Airy Stress Functions . . . . . . . . . . . . . . . . 5.8 Curvilinear Coordinates . . . . . . . . . . . . . . . . . . . . . . . . 5.9 Basic Equations of Anisotropic Elasticity . . . . . . . . . . . . . . 5.9.1 Coordinate Transformations . . . . . . . . . . . . . . . . . . 5.9.2 Plane StressStrain Compliance Transformation . . . . . . . 5.9.3 Stress Functions . . . . . . . . . . . . . . . . . . . . . . . . 5.9.4 Stresses and Displacements . . . . . . . . . . . . . . . . . . 5.10 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
CONTENTS 1 1 2 3 5 5 6 6 7 7 9 9 10 11 11 11 12 12 14 15 15 16 17 17 18 19 19 21 21 21 22 22 22 23 24 25 26 28 29 29 31 31
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Tensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
III
LINEAR ELASTIC FRACTURE MECHANICS
33
1 1
6 ELASTICITY BASED SOLUTIONS FOR CRACK PROBLEMS 6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Victor Saouma
Fracture Mechanics
Draft
CONTENTS 6.2 6.3 6.4 6.5 6.6 Circular Hole, (Kirsch, 1898) . . . . . . . . . . . . . . . . . . . . . . . . . . . . Elliptical hole in a Uniformly Stressed Plate (Inglis, 1913) . . . . . . . . . . . . Crack, (Westergaard, 1939) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.4.1 Stress Intensity Factors (Irwin) . . . . . . . . . . . . . . . . . . . . . . . 6.4.2 Near Crack Tip Stresses and Displacements in Isotropic Cracked Solids V Notch, (Williams, 1952) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Crack at an Interface between Two Dissimilar Materials (Williams, 1959) . . . 6.6.1 General Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.6.2 Boundary Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.6.3 Homogeneous Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.6.4 Solve for λ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.6.5 Near Crack Tip Stresses . . . . . . . . . . . . . . . . . . . . . . . . . . . Homogeneous Anisotropic Material (Sih and Paris) . . . . . . . . . . . . . . . . Assignment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
iii 1 4 7 10 12 14 17 17 18 19 21 21 24 25 1 1 2 11 12 12 13 14 14 15 1 1 1 1 4 5 5 6 1 1 2 5 5 6 8 8 10 10 11 12 12
6.7 6.8
7 LEFM DESIGN EXAMPLES 7.1 Design Philosophy Based on Linear Elastic Fracture Mechanics 7.2 Stress Intensity Factors . . . . . . . . . . . . . . . . . . . . . . 7.3 Fracture Properties of Materials . . . . . . . . . . . . . . . . . . 7.4 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.4.1 Example 1 . . . . . . . . . . . . . . . . . . . . . . . . . . 7.4.2 Example 2 . . . . . . . . . . . . . . . . . . . . . . . . . . 7.5 Additional Design Considerations . . . . . . . . . . . . . . . . . 7.5.1 Leak Before Fail . . . . . . . . . . . . . . . . . . . . . . 7.5.2 Damage Tolerance Assessment . . . . . . . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
8 THEORETICAL STRENGTH of SOLIDS; (Griﬃth I) 8.1 Derivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.1.1 Tensile Strength . . . . . . . . . . . . . . . . . . . . . . . . 8.1.1.1 Ideal Strength in Terms of Physical Parameters . 8.1.1.2 Ideal Strength in Terms of Engineering Parameter 8.1.2 Shear Strength . . . . . . . . . . . . . . . . . . . . . . . . . 8.2 Griﬃth Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.2.1 Derivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 ENERGY TRANSFER in CRACK GROWTH; (Griﬃth II) 9.1 Thermodynamics of Crack Growth . . . . . . . . . . . . . . . . 9.1.1 Brittle Material, Griﬃth’s Model . . . . . . . . . . . . . 9.1.2 Critical Energy Release Rate Determination . . . . . . . 9.1.2.1 From LoadDisplacement . . . . . . . . . . . . 9.1.2.2 From Compliance . . . . . . . . . . . . . . . . 9.1.3 Quasi Brittle material, Irwin and Orrowan Model . . . . 9.2 Energy Release Rate; Equivalence with Stress Intensity Factor 9.3 Crack Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.3.1 Eﬀect of Geometry; Π Curve . . . . . . . . . . . . . . . 9.3.2 Eﬀect of Material; R Curve . . . . . . . . . . . . . . . . 9.3.2.1 Theoretical Basis . . . . . . . . . . . . . . . . . 9.3.2.2 R vs KIc . . . . . . . . . . . . . . . . . . . . . Victor Saouma
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
Fracture Mechanics
Draft
iv 9.3.2.3 9.3.2.4 10 MIXED MODE CRACK PROPAGATION 10.1 Analytical Models for Isotropic Solids . . . . . . . . . . . . 10.1.1 Maximum Circumferential Tensile Stress. . . . . . . 10.1.2 Maximum Energy Release Rate . . . . . . . . . . . . 10.1.3 Minimum Strain Energy Density Criteria. . . . . . . 10.1.4 Observations . . . . . . . . . . . . . . . . . . . . . . 10.2 Emperical Models for Rocks . . . . . . . . . . . . . . . . . . 10.3 Extensions to Anisotropic Solids . . . . . . . . . . . . . . . 10.4 Interface Cracks . . . . . . . . . . . . . . . . . . . . . . . . 10.4.1 Crack Tip Fields . . . . . . . . . . . . . . . . . . . . 10.4.2 Dimensions of Bimaterial Stress Intensity Factors . . 10.4.3 Interface Fracture Toughness . . . . . . . . . . . . . 10.4.3.1 Interface Fracture Toughness when β = 0 . 10.4.3.2 Interface Fracture Toughness when β = 0 . 10.4.4 Crack Kinking Analysis . . . . . . . . . . . . . . . . 10.4.4.1 Numerical Results from He and Hutchinson 10.4.4.2 Numerical Results Using Merlin . . . . . . 10.4.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
CONTENTS Plane Strain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 Plane Stress . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 1 2 2 3 4 6 8 9 13 15 16 17 19 19 20 20 22 24
. . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . .
IV
ELASTO PLASTIC FRACTURE MECHANICS
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
29
1 1 2 2 4 4 7 8 1 2 2 3 4 1 1 2 3 3 5 8 9
11 PLASTIC ZONE SIZES 11.1 Uniaxial Stress Criteria . . . . . . . . . . . . 11.1.1 FirstOrder Approximation. . . . . . . 11.1.2 SecondOrder Approximation (Irwin) . 11.1.2.1 Example . . . . . . . . . . . 11.1.3 Dugdale’s Model. . . . . . . . . . . . . 11.2 Multiaxial Yield Criteria . . . . . . . . . . . . 11.3 Plane Strain vs. Plane Stress . . . . . . . . .
12 CRACK TIP OPENING DISPLACEMENTS 12.1 Derivation of CTOD . . . . . . . . . . . . . . . 12.1.1 Irwin’s Solution . . . . . . . . . . . . . . 12.1.2 Dugdale’s Solution . . . . . . . . . . . . 12.2 GCTOD Relations . . . . . . . . . . . . . . . . 13 J INTEGRAL 13.1 Genesis . . . . . . . . . . . . . . . . . 13.2 Path Independence . . . . . . . . . . . 13.3 Nonlinear Elastic Energy Release Rate 13.3.1 Virtual Crack Extension . . . . 13.3.2 †Generalized Derivation . . . . 13.4 Nonlinear Energy Release Rate . . . . 13.5 J Testing . . . . . . . . . . . . . . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
Victor Saouma
Fracture Mechanics
Draft
CONTENTS 13.6 13.7 13.8 13.9 Crack Growth Resistance Curves . . . . . . . . . . . . . . † Load Control versus Displacement Control . . . . . . . . Plastic Crack Tip Fields . . . . . . . . . . . . . . . . . . . Engineering Approach to Fracture . . . . . . . . . . . . . 13.9.1 Compilation of Fully Plastic Solutions . . . . . . . 13.9.2 Numerical Example . . . . . . . . . . . . . . . . . 13.10J1 and J2 Generalization. . . . . . . . . . . . . . . . . . . 13.11Dynamic Energy Release Rate . . . . . . . . . . . . . . . 13.12Eﬀect of Other Loading . . . . . . . . . . . . . . . . . . . 13.12.1 Surface Tractions on Crack Surfaces . . . . . . . . 13.12.2 Body Forces . . . . . . . . . . . . . . . . . . . . . . 13.12.3 Initial Strains Corresponding to Thermal Loading 13.12.4 Initial Stresses Corresponding to Pore Pressures . 13.12.5 Combined Thermal Strains and Pore Pressures . . 13.13Epilogue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
v 9 11 12 16 18 27 29 30 32 32 33 33 35 37 38
V
SUBCRITICAL CRACK GROWTH
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
39
1 1 2 2 3 3 4 4 4 4 5 5 5 5 6 6 6 7 8
14 FATIGUE CRACK PROPAGATION 14.1 Experimental Observation . . . . . . . . . . . . . . . . . . 14.2 Fatigue Laws Under Constant Amplitude Loading . . . . 14.2.1 Paris Model . . . . . . . . . . . . . . . . . . . . . . 14.2.2 Foreman’s Model . . . . . . . . . . . . . . . . . . . 14.2.2.1 Modiﬁed Walker’s Model . . . . . . . . . 14.2.3 Table LookUp . . . . . . . . . . . . . . . . . . . . 14.2.4 Eﬀective Stress Intensity Factor Range . . . . . . 14.2.5 Examples . . . . . . . . . . . . . . . . . . . . . . . 14.2.5.1 Example 1 . . . . . . . . . . . . . . . . . 14.2.5.2 Example 2 . . . . . . . . . . . . . . . . . 14.2.5.3 Example 3 . . . . . . . . . . . . . . . . . 14.3 Variable Amplitude Loading . . . . . . . . . . . . . . . . . 14.3.1 No Load Interaction . . . . . . . . . . . . . . . . . 14.3.2 Load Interaction . . . . . . . . . . . . . . . . . . . 14.3.2.1 Observation . . . . . . . . . . . . . . . . 14.3.2.2 Retardation Models . . . . . . . . . . . . 14.3.2.2.1 Wheeler’s Model . . . . . . . . . 14.3.2.2.2 Generalized Willenborg’s Model
VI
FRACTURE MECHANICS OF CONCRETE
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11
1 1 2 2 3 4
15 FRACTURE DETERIORATION ANALYSIS OF CONCRETE 15.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15.2 Phenomenological Observations . . . . . . . . . . . . . . . . . . . . . 15.2.1 Load, Displacement, and Strain Control Tests . . . . . . . . . 15.2.2 Pre/PostPeak Material Response of Steel and Concrete . . . 15.3 Localisation of Deformation . . . . . . . . . . . . . . . . . . . . . . .
Victor Saouma
Fracture Mechanics
Draft
vi 15.3.1 Experimental Evidence . . . . . . . . . . . . . . . . . . 15.3.1.1 σCOD Diagram, Hillerborg’s Model . . . . . . 15.3.2 Theoretical Evidence . . . . . . . . . . . . . . . . . . . . 15.3.2.1 Static Loading . . . . . . . . . . . . . . . . . . 15.3.2.2 Dynamic Loading . . . . . . . . . . . . . . . . 15.3.2.2.1 Loss of Hyperbolicity . . . . . . . . . 15.3.2.2.2 Wave Equation for Softening Maerials 15.3.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . 15.4 Griﬃth Criterion and FPZ . . . . . . . . . . . . . . . . . . . . . 16 FRACTURE MECHANICS of CONCRETE 16.1 Linear Elastic Fracture Models . . . . . . . . . . 16.1.1 Finite Element Calibration . . . . . . . . 16.1.2 Testing Procedure . . . . . . . . . . . . . 16.1.3 Data Reduction . . . . . . . . . . . . . . . 16.2 Nonlinear Fracture Models . . . . . . . . . . . . . 16.2.1 Hillerborg Characteristic Length . . . . . 16.2.1.1 MIHASHI data . . . . . . . . . . 16.2.2 Jenq and Shah Two Parameters Model . . 16.2.3 Size Eﬀect Law, Baˇant . . . . . . . . . . z 16.2.3.1 Derivation . . . . . . . . . . . . 16.2.3.2 Discussion . . . . . . . . . . . . 16.2.4 Carpinteri Brittleness Number . . . . . . 16.3 Comparison of the Fracture Models . . . . . . . . 16.3.1 Hillerborg Characteristic Length, lch . . . 16.3.2 Baˇant Brittleness Number, β . . . . . . . z 16.3.3 Carpinteri Brittleness Number, s . . . . . 16.3.4 Jenq and Shah’s Critical Material Length, 16.3.5 Discussion . . . . . . . . . . . . . . . . . . 16.4 Model Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
CONTENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 4 7 7 10 11 11 13 13 1 1 2 3 3 6 6 6 6 7 7 10 11 12 12 13 14 14 14 15
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Q . . . .
. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . .
17 FRACTURE MECHANICS PROPERTIES OF CONCRETE 17.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17.2 Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17.2.1 Concrete Mix Design and Specimen Preparation . . . . . . . 17.2.2 Loading Fixtures . . . . . . . . . . . . . . . . . . . . . . . . . 17.2.3 Testing Procedure . . . . . . . . . . . . . . . . . . . . . . . . 17.2.4 Acoustic Emissions Monitoring . . . . . . . . . . . . . . . . . 17.2.5 Evaluation of Fracture Toughness by the Compliance Method 17.3 Fracture Toughness Results . . . . . . . . . . . . . . . . . . . . . . . 17.4 Speciﬁc Fracture Energy Results . . . . . . . . . . . . . . . . . . . . 17.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17.6 Size Eﬀect Law Assessment . . . . . . . . . . . . . . . . . . . . . . . 17.7 Notation and Abbreviations . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
1 . 1 . 2 . 2 . 4 . 5 . 5 . 6 . 7 . 9 . 11 . 11 . 13
Victor Saouma
Fracture Mechanics
. . . . . .6 Order of Singularity . . . . . . . .3 MacroScale Correlation Analysis . .7. . . . . . . .1 Fracture Testing . . . . . . . . . .
. . . . . . . . 19.3. . . . . . . .3 Quarter Point Singular Elements . .3 Equivalent Domain Integral (EDI) Method 21. . . . . . . . . . . . . . .1. . .2. 7 . . . . . . . . . . .
. . .2 Single Virtual Crack Extension. . 18. . . . . . . . . . . . .1 Fracture of Concrete . . . . . . . . . . . . 20 ENERGY RELEASE BASED METHODS 20. . . . . . .
. . . .1 Energy Release Rate J . . . . . . . . . . . .8 Numerical Evaluation . . . . .2. . . 13 1 1 1 2 3 3 4 1 1 4 5 5
. . 19. . . . . . . Decomposition
. .3 Fractals and Fracture . . . . . 19. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . 8 . .2. . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . 20. . . .
. .1. . . . . . . . .1 Energy Release Rate . .
. . . 19. . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. .
. . . . .1. . . .4 Correlation of Fractal Dimensions With Fracture Properties . . . . . . . Displacement 21 J INTEGRAL BASED METHODS 21. . . . . .
. 18. . . . .1 Numerical Evaluation . . . . . . . . . . . . . .4 Review of Isoparametric Finite Elements . . . . . . . . . . . . . . . . 18. . . . . 11 . . 21. . . . . . . .
. . . . . .3 Numerical Determination of Fractal Dimension . 19. . . . . . . .
. . . . . . .9 Historical Overview . . . . . . . . . . 19. . . .1 Isotropic Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20. . . . . . . 18. . . . . . . . . . 21. . . . . . .
. . . .2 Fractal Geometry . . . . . . . . . . . . .1. . .2 Experimental Procedure . . .1 Mode I Only . . . . . 5 . . . . . . . . . . . . . .2 Anisotropic Case . . . . .
. . . . . . . 18. . . . . . . . . . 7 . . . 3 . . . . . . . . . . . . . . . . . . . . 20. . . .1 Spatial Variation of the Fractal Dimension . . . . . .
vii 1 1 1 1 3 4 6 6 7 8 10 10 14 14 16 21
VII FINITE ELEMENT TECHNIQUES IN FRACTURE MECHANICS 23
19 SINGULAR ELEMENT 19. . . . . . . . . . . . . . . . . .
. . . . . . . .1. . . . . . . . . . . . . . . . .1 Introduction . . . . . . . . . . . 19. . . . . .2 Correlation Between Fracture Toughness and Fractal Dimensions 18. . . . . . . . . . . . FRACTURES and SIZE EFFECTS 18. . . . . . . . . . . .
. . 18. .2 Virtual Crack Extension. . . 20. . . . 1 . . . .
. . . . . . . . . . . . . .
. .
. . . . . . 6 . . . . . .1 Two Virtual Crack Extensions. . . . . . . .5 How to Distort the Element to Model the Singularity . .3. . . . . . . . .2 Mixed Mode Cases . 18.3 Computation of Fractal Dimension . . . . . . . . . . . . . . . .
. . .10Other Singular Elements . . 20. . . . . . . . . . . . . . . . . . . . . . . . . . .2. . . . . . . . . . . . . . .2.7. . . . . . . . . . . . .Draft
CONTENTS 18 FRACTALS. . . . . . . . . . . . . . . . . . . . . .4 Fractals and Size Eﬀects . . .
. . . . . . . . . . .2 Proﬁle Measurements . . 2 . . . . . . . . . . 19. . . . . . . 1 . . . . . . .5 Conclusions . 18. . . . . . . . 18. . .
. . . . . . . . . . . . . . . 18. . . . . . 18. . . . . . . . . . . . . . . . . . . .1. . . . . . . . . . . . . . . . . . . . . .3.
. . . . . . . . . . . . . . 19. 1 . . . . . . . . 9 . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3. . . . . . .
. . . . . . .2 Displacement Extrapolation . . . . . .
Victor Saouma
Fracture Mechanics
. . . . . . . . . . . . . . . . . . . . . . . . . . . 19. . . . .7 Stress Intensity Factors Extraction . . . . . . . 18. . . . . . . . .2 Mixed Mode SIF Evaluation . . . . . . . . . . .1 Introduction . .
. . . 18. . . . . . . . . . . 19. . .
. . . . .
. . . . . . . . . . . . . . . . 22. . .
. . .
. . . . . . . . . . . . . . .3 Validation . . . . . . . . . .2. . . 24. . 24. . . .
CONTENTS . . . . . . . . .3. . . 23. . . . . . . . .2 Biaxial interface test. .2 Extraction of SIF . . . . . . . . . . . . . . . . .8 Field Equations for Thermo. . . . . .
. . . . . . . . . . . . . . .2. . . . . . . . . .
. . . . .
. . . . . . . . . . . . . . . . . . . 24. . . .3. . . . . . . . . . . . . . .6 Initial Stresses Corresponding to Pore Pressures . . . . . . .1 Interface element formulation. . . . . . . . . . . . . . . . . . 21. . .
. . .1 Griﬃth criterion and ICM. . . . . . .2 Computational Algorithm . . 24. . . . . . . . . . 10 . . . . . . .3. 24. .3 Finite Element Implementation . . . . . . . . . . . . . . . . . .3 Nonlinear solver. . .5 Notation .3. . . . . . . . . . .1 Relation to ﬁctitious crack model. . . . . . . . . . . .4. .2. . . . . . 7 . . . . .1 2D case . . . . . . . . . . 23. . 21. . . . . . . . . . . . . . . . . . . .1 Direct shear test of mortar joints. . . . . .3. . . . . . . Fictitious. . . . . . . . . . . . . . . . . . . . . . . . .3 Penalty Method Solution . .
. . . . . . . . . . . . 23. . . . . . Victor Saouma
. . .3. . . .2 3D Generalization . . . . . . . . . . . . . . . . . . . . . .6 Line search method. . . . . . . . .3. . . . . . . . . . . . .4 Body Forces . . . . . . .1 General Formulation . . . . . .1. . . . . .2. . . . . . . . . . . . 21. . . . 24. . . . . . .
. . . .2. .
. . . . . . . . . . . . 22 RECIPROCAL WORK INTEGRALS 22. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . .1 LoadCMOD . . . . . . . . 24. . . .
. 23. . . . . . . . . .1 Introduction . . . . . . . . . . .
. . .2. . . 24. . . 8 . . . . . . 24. . . . . .5. . . . . . . . . . . .and PoroElasticity . .
.1 J Components . . . . . 21. .
.
. . . . . . . . . . . . . .
. . . . . . . . . . . 10 . . . . . .
. . . . . . . . 23. . . . . . . . . . . 10 1 . . . . . . . . . . . . . . . . . .2. . . .3 Parametric Studies . . . . . . . .2 σ and u Decomposition . . . . . . . . . . . . 22. . . . . .
. . . . . . . . . . . . . . . . . . . . . . 8 . . .
. . . . . . . . 22.7 Combined Thermal Strains and Pore Pressures . . . . . . .Draft
viii 21. . .
. .
. . . . 5 . . . . . . . . . . and Eﬀective Crack Lengths 23. . . . . . . . .
. 24. . . . . . . . . . . . . .4 SecantNewton method. . . . . . . .3. 23. .
23 FICTITIOUS CRACK MODEL 23.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4 Conclusions . . 22. . . . . . . . . . . . . . . . . . . . . . . . . . . . 22. . . . . . . . . . .5. . . . . . . . . . . . . .2 Volume Form of the Reciprocal Work Integral . . . . . 24. . . . . . . . . . . . . . . . . . . . . . . . . .
Fracture Mechanics
. . . . .3. . . . . . . 24. . . . . . . . .2 Real. . . .3 Surface Tractions on Crack Surfaces . . . .3. . . . . . . . . . . .1. . . . . . . 5 . . 23. 24. . . . . . . . 24 INTERFACE CRACK MODEL 24.
. 10 . . . . . . . . . 10 . . . . . .5 Initial Strains Corresponding to Thermal Loading 22. . . . . . . . . 7 . . .
. .
. . . . .
. . . . .2 Discretization of Governing Equations . 24. . . . . . 1 . . . . . . .
. . . . . . . . . . . . . . . . . . . . . 23. . . . . . . . . . . . .
. . .
. . .1 Introduction .
. . . . . . . . . . . . . . .
. . .
. . . . . . . . . . . . . . . . . . . .
.4 Mixed Mode Crack Propagation . . . .3.2 Constitutive driver. . .
. . . . . .
. . . . . . . . . . . . 24. . .
. . .5 Examples and validation . . .2 Criterion for crack propagation. . . . . . . . 24. . . . . . . . . .
. . . . . . .
. . . . .
. . .5. .3. . . . . . . . . .4. . . . . . . . . . . . 23. . . . . . . . . . . . . . 23.3 Modiﬁed Iosipescu’s beam.1 Weak Form of Governing Equations . . 22.3. 11 1 1 2 2 4 6 8 10 10 11 14 17 17 1 1 2 8 9 9 11 16 18 18 20 21 21 23 24 24 24 28
. . . . . . .5 Element secant stiﬀness.4 IncrementalIterative Solution Strategy . . .2 Interface Crack Model . . . . . . . . . . . . . . . . . .3. . . . .
. . . . . . . . . . .4 Anchor bolt pullout test. . . .6 Conclusions . . . . . . . . . . . . . . . . 41
Victor Saouma
Fracture Mechanics
. . . . . . . . . . . . . .Draft
CONTENTS
ix
24. . . . . . . . . . . . . . 37 24. . . . . . .5. . . . .
Draft
x
CONTENTS
Victor Saouma
Fracture Mechanics
.
. . . Single Edge Notch Tension Panel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2 6. . . . . . . . . 1 . . . . . . . . . . . . . . . . . . .3 2. . . . . . .1 4. . . . . .4 2. . . . . . . . . . . . . . . .1 7. . . . . . . . Elliptical Hole in an Inﬁnite Plate . . .5 2. . . . . . . . Cauchy’s Tetrahedron . . . . . . 2 . . . .3 6. . . . . . . . . .2 2. . . . . . . . . . . . . . . . . . . . . . . .2 4. . . . . . . . . . . . . .3 5. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cartesian Coordinates Curvilinear Coordinates . . Failure Envelope for a Cracked Cantilevered Generalized Failure Envelope . . 1 2
. . . Independent Modes of Crack Displacements Plate with Angular Corners . . Typical StressStrain Curve of an Elastoplastic Bar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Plate with Angular Corners . . . . . . . . . . . .4 6. . . . . . . . . . . . . . 4 . . . Flux Through Area dS . . . . . 2 . . .8 6. . . . . . . . DiscreteSmeared Crack Models . . . . . . . . . Coordinate Systems for Stress Transformations Circular Hole in an Inﬁnite Plate . . . . . .7 5. .4 5. . . Isotropic Hardening/Softening . . . . .2 2. . . . . . .5 6. . . . . . . . . . . . . . . . . . . .8 2. . . . 18 . . . .2 5. . Transversly Isotropic Material . . . . . Column Curve . . . . . . . . . . . . . . . . . .Draft
List of Figures
1. . Compact Tension Specimen . 14 . . . . . . . . .6 5. . . . . . . . .4 7. . . . .7 2. . . . . . . . . . .
. . . . 6 . . . .1 6. . . . . . .1 5. . . . . . . . . . . . . . . . . . . . . . Stress and Strain Increments in ElastoPlastic Materials . . . . . . . . . . . . . . . 2 . . . . . . . . . . . . . .2 7. . . . . 3 . . .3 4. . . . . . . . . . . . . . . . . . . . . . . . . Cracked Cantilevered Beam . . . . . . . . . . . . .6 7. . . . . . . . . . . .5 Kinematics of Continuous and Discontinuous Failure Processes . . . . . . . . . Kinematic Hardening/Softening . . 7 . . Double Edge Notch Tension Panel Three Point Bend Beam . . . . . . . . . . . . .4 5. . . . . . . . . . . .1 1. . . . . . . . . . . . . . . . . . . . . . . . . . 11 . Beam . . . . . . . . . . . . . . . . . . . . . .1 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . HaighWestergaard Stress Space . . . . . . . . . Equilibrium of Stresses. . . . . . . . . . . . .5 5. . . . . . . StressStrain diagram for Elastoplasticity . . . . . . . . . . . . . . . . . . . . . . . . Middle Tension Panel . . . 14 . . . . . . . . . . . . . . . . . . . . . . 3 4 5 5 8 8 9 16 18 26 27 29
Stress Components on an Inﬁnitesimal Element Stresses as Tensor Components . . . . . . . . . . . . . . . . . . . 5 . . . . . . . . . . . . . . . . . . . . . . . . . .9 4.6 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Crack in an Inﬁnite Plate . . . . . . . . . . . . . . . . .3 7. . . . . . . . . . . . . . . . Elastoplastic Rheological Model for Overstress Formulation Bauschinger Eﬀect on Reversed Loading . General Yield Surface. . 15 . . . . . . . . . . . . . . . . . . . . . 10 . 2 3 3 4 4
. . . . . . . . . . . . .
. . . . . . .
10. . Approximate Solutions for Long Cracks Radiating from a Circular Hole in an Inﬁnite Plate under Tension . . . . . . . . . . . . . .3 11. . . . . . . . . . . . . . .5 Locus of Fracture Diagram Under Mixed Mode Loading . . . . .10Geometry of kinked Crack. . . . . . . . . . . . . . 2 2 3 5 3 5 6 6 8 9 11 14 15 16 2 4 6 7 7 9 12 14 15 17 18 21 23 23 25 2 3 5 5
Energy Transfer in a Cracked Plate . . . . . . . . . . . . . . Graphical Representation of the Energy Release Rate G . . . . . . . . . . . . . . . . . . . . . . . . . . . .Draft
ii 7. . . . . . Two Opposite Point Loads acting on the Surface of an Edge Crack . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Stress Strain Relation at the Atomic Level . . . . . . . . . . . 14 . . . . .9 7. . . . . . . . . . . . . . . . . . . . . . . . . . . . (Hutchinson and Suo 1992) . . 6 . . . . . . . . . . . 10. . . .9 9. . (Hutchinson and Suo 1992) . . .1 Mixed Mode Crack Propagation and Biaxial Failure Modes . . . . . . . . . . . . . . . .1 9. . . . .7 9. . . . . . . . . . . . . . . . . . . .3 8. . . 8 . . . . Determination of Gc From Load Displacement Curves . . . . . . 10. . . . . . . . .
. . . . . 10. . . . . . . . . . . . . . . . . . . . .1 11. .14 8. . 10. . . . . . . . Corner. . . . . . . . . . . . . 10. . . . . . . . .4 9. . . (Hutchinson and Suo 1992) 10. . . . . . . . . . . . . . . . 8 . Eﬀect of Geometry and Load on Crack Stability. . . . . . . .8 7. 10. . . . . . Two Opposite Point Loads acting on the Surface of an Embedded Crack . . . . . .8 9. . . . . . . . . . . . .11 7. KI for DCB using the Compliance Method . . . . . . . . . 10. . . . .10 7. . . . . . . . Growth of Semielliptical surface Flaw into Semicircular Conﬁguration . . . . . . . . . . . . . 11. . . . . . . . . . . . . . . . .2 11. . 10. . . . . . . . . . .7 Angles of Crack Propagation in Anisotropic Solids . . . . . . . . and Surface Cracks . . . . . . . . .
5
. . .3 9. . . . . . 7 . . . . . . . . . . . . . . . .6 7. . . . . . . . . . . . . . . . . . . . . . . . . . Radiating Cracks from a Circular Hole in an Inﬁnite Plate under Biaxial Stress Pressurized Hole with Radiating Cracks . .11Schematic variation of energy release rate with length of kinked segment of crack for β = 0. . . . 10.9 Geometry and conventions of an interface crack. . . . Embedded. . . . . . . . . (Gtoudos 1993) R Curve for Plane Strain . . .4 9. . .1 8.13 7. . . . . . . . . . . . 11 . . . . . . . . . . . . . . . . . . . . . . . . . . 10. . . . . . . . . . . . . . . . . . . and Newman’s Solution . . . .2 8. .14Finite Element Mesh of the Plate Analyzed . . . . . . . . . . 10. . . .12Conventions for a Crack Kinking out of an Interface. . . . . . . . . . . . . . .12 7. .3 Sθ Distribution ahead of a Crack Tip . . .13Geometry and Boundary Conditions of the Plate Analyzed . . . SecondOrder Approximation of the Plastic Zone Dugdale’s Model . . . . . . . .2 Crack with an Inﬁnitesimal “kink” at Angle θ .6 Fracture Toughnesses for Homogeneous Anisotropic Solids . . . . . . . . . . . . . . .15Variation of G/Go with Kink Angle ω . . .4 FirstOrder Approximation of the Plastic Zone . 10. . . . . .10
LIST OF FIGURES
Approximate Solutions for Two Opposite Short Cracks Radiating from a Circular Hole in an Inﬁnite Plate under Tension . . . . . . . . . . . . . . . . .6 9. . . 10. . . . . . . . . . . . . . . . . . Plastic Zone Ahead of a Crack Tip Through the Thickness . . . 6 . . . . . . . . . . . . . .7 7. Uniformly Stressed Layer of Atoms Separated by a0 Energy and Force Binding Two Adjacent Atoms . . . .
Victor Saouma
Fracture Mechanics
. . . 9 . . Variable Depth Double Cantilever Beam . . . . . . . . . . . . . . .2 9. (Hutchinson and Suo 1992) . Experimental Determination of KI from Compliance Curve . . . . . . . . . . . Point Load on a Crack . . . . . Inﬂuence of Atomic Misﬁt on Ideal Shear Strength .4 Angle of Crack Propagation Under Mixed Mode Loading . . . . .8 Failure Surfaces for Cracked Anisotropic Solids . . . . . . . . . . . . . . . .5 9. . . R Curve for Plane Stress . Elliptical Crack. . . . .
. . Barenblatt’s Model . . . . . . . . . .8 11. . . . (Anderson 1995) . . Stages of Fatigue Crack Growth . .13Eﬀect of Plasticity on the Crack Tip Stress Fields. . . . . . . . . . . . .6 15. . . . .3 Virtual Crack Extension Deﬁnition of J . . . . . . . . . . . . . . . . . . . .2 Closed Contour for Proof of J Path Independence . . . . . Normalized Mode I Plastic Zone (VonMyses) . . 10 Localization of Tensile Strain in Concrete . . . . . . . . . 8 Load Displacement Curve in terms of Element Size .3 15. . . . . .1 Crack Tip Opening Displacement. . . . . . . . . . . . 13. . . . . . . . . . .7 15. . . . . . . . . . . . . . . .5 ElasticPlastic versus Nonlinear Elastic Materials . . . . . . . . .19Circumferentially Cracked Cylinder . 13. . . . . . . . . . . . . . . . . . . . . . . . . . 13. . . . . . . .18Axially Cracked Pressurized Cylinder . . .7 Experimental Derivation of J . . . . . . . . . . . . . . . . . 13. . . . . . . . . . . . .2 14. . . . . . . . . . . . 14. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Plastic Zone Size Across Plate Thickness . . . (Anderson 1995) . . . . . . . . . . . . . . . . . . . 13. . . . . . . . . . .4 Arbitrary Solid with Internal Inclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12. . . . . . . . . .7 15. . . . . . . . . . 2 StressStrain Curves of Metals and Concrete . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13. . . . . . . . . . . .10J. . . . . . .1 14. . . . . . . . . .1 J Integral Deﬁnition Around a Crack . . 13.4 15. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .14Compact tension Specimen . Plane Strain . . .5 15. . . . . . 13. . . . . . . . . . . . . (Anderson 1995) 13. . . . . . JR versus Crack Length. . . . . . . . . . . . . . . . . . . .9
LIST OF FIGURES Eﬀect of Plastic Zone Size on Dugdale’s Model . . . . . . 13. . . . . . . . . . . . . . Around a Circular Path .10Plate Thickness Eﬀect on Fracture Toughness . 13. .3 14. . . . . . . . . . . . . . . . . . . . . 6 StrainSoftening Bar Subjected to Uniaxial Load . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13. . . . . . . . . . .1 15. Repeated Load on a Plate . . . . .01) . . . Grwoth . . . . . . . . . . . . . . . . . . . . . . . . . 11 1 2 1 3 4 6 8 8 10 10 12 13 13 15 16 18 20 20 23 23 26 30 1 2 2 4 7 7 8
12. . . . . . . . . . . . . . . . . . . Forman’s Fatigue Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13. . . . . . (Anderson 1995) . . . . . . . 13. . . . . .
iii 6 7 8 9
. . . . . . . . . . . . . . . . . .16Single Edge Notched Specimen . . . . .
Test Controls . . . . . . . . . .2 15. . . .7 11. . . 13 Griﬃth criterion in NLFM. . . . .20Dynamic Crack Propagation in a Plane Body. . . 14 Fracture Mechanics
Victor Saouma
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Retardation Eﬀects on Fatigue Life . . . (Anderson 1995) . . . . . . . . . . . . . . . . . 13. .9 SN Curve and Endurance Limit . . . . . . . .5 14. . . . . . . . . . . . 5 Concrete Strain Softening Models . .8 J Resistance Curve for Ductile Material. . . . . 13. . . . . . . . . . . . . . . . . . . 3 Caputring Experimentally Localization in Uniaxially Loaded Concrete Specimens 4 Hillerborg’s Fictitious Crack Model . . . . . . . . . . . . . . . . . .2 Estimate of the Crack Tip Opening Displacement. . . . . . Stress and Plane . . . . . . . 13. 11.Draft
11. . . . . . . . . . . . . . . . . . . . . . . . . . . (Kanninen 1984) . . . 10 . . . .9 J. .15Center Cracked Panel . . . . . . . . . . . . Cause of Retardation in Fatigue Crack Yield Zone Due to Overload . . . . . . . . . . . . . . . . .5 11. 13. . .6 14. . (Anderson 1995) . .11Normalize RambergOsgood StressStrain Relation (α = . 13. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4 14. .6 11. . . . . . . . . . . . (Anderson 1995) . . . . . . . . . . . . .8 15. . . 13. . . . . . . . . .6 Nonlinear Energy Release Rate. . . . . . . . . . . . . . . . .12HRR Singularity. . . . . . . . . Plastic Zone Size in Comparison with Plate Thickness. . .17Double Edge Notched Specimen . . . . . . . . . . . . . . . . . . . . . . . . . .
. . 3 . . . . . . . . .2 16. . . . ceramics. . . Dashed lines indicate adjustable sidFixed grid boundaries. . . . . . . . . . . . . .4 17. . . . . . . . .8 KIc versus D for concrete (this study). . . *Block diagram of the experimental system. D) KIc versus D for concrete (this study). . . . . . . . . . . . . . . . 1991). .9 Size eﬀect for WS specimens for da =38 mm (1. . 1991). . .4 16. . 10 . . . . . . .3 19. . J. . . . . . .
6 8 9
10
15 17
18 19 2 3 6 7
Victor Saouma
Fracture Mechanics
. . 1991) . . . . . . . . . 1991) . . E. . . u 17. .. . . . . . . . 1991). . . . . . . . . . . . 5 . . . . . . . .6 17. . . and Saouma. . . .1 (A) Straight line initiator. . . . . . . . . . . . . . . . . . . . . Number of occupied boxes (N ) is plotted versus box size. . Crack Mouth Opening Displacement (CMOD) gage straddles the initial notch. . .9 Variation of L(S) in terms of S . . . . . . . . . .5 in) (Br¨hwiler.. . . . . . . . 1988). . . . . . *Mean fracture toughness values obtained from the rounded MSA WS specimen tests. . . .
LIST OF FIGURES . . . . . *Wedge ﬁxture and line support. . . . . . . . . . . . . . 18. . . . . . . . .7 GF versus D for concrete (this study).3 17. . . . . . . . B. . . . Singular Element (QuarterPoint Quadratic Isoparametric Element) . . . . . . . . . and alumina (Mecholsky and Freiman. . . .7 ServoControlled Test Setup for Concrete KIc and GF . . . . . . 18. . . . . . .1 19. (C) Modiﬁed Koch curve. . Broz. . . . . . . . Isoparametric Quadratic Finite Element: Global and Parent Element . . . . . . . . . and silicon (Tsai and Mecholsky (1991) . . . . . . . . . . . 18. . . . . . . . . . . . . 12 2
18. . Flexible grid boundaries.1 17. 11 . . . . . . . . . . . .5 Plot of box counting method applied to the proﬁle of a typical fractured concrete specimen. . . . . . . . . . eﬀective crack length curve. . . CMOD curve for a “Large” specimen. ceramics. . .5 17. . . . . . . .4 Stress Intensity Factor Using Extrapolation Technique . . . . . . . . . . . ..2 19. . . 4 . . . . . . . . . . . . . . . . . vertically. . . . . . . V. . 19. . . . . and alumina (Mecholsky and Frieman. . .
. . . . . *The three stages of the fracture toughness vs. . . and silicon (Tsai and Mecholsky (1991) .3 16. . . . . . . . . polystyrene (Chen and Runt. . and diagonally. . . . . . . . . . 18. . . . . . 18. . . . . . . . . . . . . *Typical AE record for a “Large” WS specimen test. . . . . . . . . . . . . . . . . .Draft
iv 16. . . . . . 17. Slope of line ﬁt to data is the fractal dimension (D). . . (B) Quadratic Koch curve. . . . . . . Finite Element Discretization of the Crack Tip Using Singular Elements . . . .6 AB) GF and KIc versus D. . . Flint (Mecholsky and Mackin. . . . . . . . . horizontally. *Typical LoadCMOD Curve from a Concrete Fracture Test Eﬀective Crack Length Deﬁnition . . . . . . . . . . . Size Eﬀect Law .5 17. . Inelastic Buckling . . . . . .J. .1 16. . . . . . . . . . . . . . . . . . . *Typical PSP vs. . . . . . . . . . . C) GF versus D for concrete (this study). . . . . . . . 18. . .4 Typical grid overlying an object. . . . .8 *Mean speciﬁc fracture energy values obtained from the rounded MSA WS specimen tests. . . . . . . . . . . . ceramics. . . . . . . . . . . . . . . . . . and alumina (Mecholsky and Frieman. . . . . . . . . Flint (Mecholsky and Mackin. . . . . . . . . . . . . . . . . . . . . . . . . 3 4 5 5 6 7 8
*Wedgesplitting specimen geometry. 10 . . . . . . . . . . . . . . . . . . . ceramics. . . . . . . . and triadic Koch curve. fractal generator. . . . .2 17. . . . . . 1989). . . 1988). . 18. 18.3 Orientations of measured proﬁles over the fractured surface. . . . . . . and alumina (Mecholsky and Freiman. . . polystyrene (Chen and Runt. . . . . . . 1989). .E. . .2 Frontal view of wedgesplittingtest specimen showing forces applied to specimen by lateral wedge loading (FS ) between two circular pins located near top of specimen on either side of the vertical starting notch. . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . .3 23. . . . . . . . . . . . . . . . . . 24. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .19Direct shear test on mortar joint. . . . . . . . . ˇ 24. . . . . . . . . .5 23. . . . . . . . . . . . . . .5 Inner and Outer Surfaces Enclosing a Tube along a Three Dimensional Crack Front . . . . 24. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21. . . . . . . . . . .6 Bilinear softening laws.5 Displacement Correlation Method to Extract SIF from Quarter Point Singular Elements . . . . . . . . . . . . . .M. . . . Victor Saouma
Fracture Mechanics
. . . . .1 Contour integral paths around crack tip for recipcoal work integral . . . . . . . . . . . . . . . . .11Deﬁnition of inelastic return direction. . . . . . . . . . . . . . . . . . J. . . . . . . . . . Γ0 . . . . . and Slowik. 24. 20. . . . . . . . . . . .2 Simply connected Region A Enclosed by Contours Γ1 . 24. 24.2 Displacement Decomposition for SIF Determination . . . . . . . . . . . . . . . . . and Cervenka. . . . . . .7 Stiﬀness degradation in the equivalent uniaxial case. . . . . . . . . . . . . . . . . . . . .4 Interpretation of q in terms of a Virtual Crack Advance along ∆L. . . . . . . . . 24. . (Anderson 1995) . . . . . . . . . . . . . . . . 24. . . . . . . . . .
24. . . . . . and Chandra Kishen. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .15Line search method.. 24. . . . . . . . . . . . . . . . . . . . . . .1 Crack Extension ∆a . . . . . . .9 Local coordinate system of the interface element. . 24.E. . . . . . . . . . . . . . . . . . . Numerical Predictions vs Experimental Results for Wedge Splitting Tests Real. . . . . . .13Sheartension example. . . . . . . . .12Inﬂuence of increment size. .4 23. . . . . . . . . . . . . . . . .10Algorithm for interface constitutive model. .14Secant relationship. .8 Interface element numbering. 8 19. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Γ+ . . . . . . . . . .2 23. . . 21. . . . . . . . . . . . . . . . . . . . V. . . (Anderson 1995) . .6 Nodal Deﬁnition for FE 3D SIF Determination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24. . . . J. . . . 21. . . . .1 23. . . . . . . .21Nonlinear analysis of the mixed mode test. . . . . . . . . . . . . . . . . Wedge Splitting Test. . . . . . 21. . . . . .1 Numerical Extraction of the J Integral (Owen and Fawkes 1983) . . . 24. 24. . . and Eﬀective Crack Lengths for Wedge Splitting Tests . . . . (Saouma V. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1994) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2 Wedge splitting tests for diﬀerent materials.1 Mixed mode crack propagation. . . . . . . . . . . . . . 24. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .20Experimental setup for the large scale mixed mode test. . . . . . . . . . . . . Fictitious. . . . . . . . . . . . 24. . 24. . . . . . . . . . . . . . . . . . . . 24. . . . 24. . . . . . . .5 Failure function. . . . . . and Γ− . . . . . . . . 21. . . . . . . . . Eﬀect of GF on 50 ft Specimen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24. . . . . . . . . . . . . . . . and FE Discretization . . . . . . . . . . . . . . . .16Griﬃth criterion in NLFM. . . . .7 Body Consisting of Two Subdomains . 10 20. . . . . . . 24. . . . . . Eﬀect of s1 on 3 ft Specimen . . . . . . 2 4 2 6 8 8 9 2 3 11 12 13 14 15 16 2 3 4 4 5 6 8 10 11 12 14 16 16 18 21 22 25 25 26 27 29
22.Draft
LIST OF FIGURES
v
19. . . . . . . . . . . . . . . . Eﬀect of wc on 50 ft Specimen . . . . . . . . . . . .18Schematics of the direct shear test setup. . . . . . . . . . . . . .3 Surface Enclosing a Tube along a Three Dimensional Crack Front. . . . . . . . . . . . .17Mixed mode crack propagation. . . . . . . . (Anderson 1995) . . . . . . . . . . . . . . . . . . . .6 23. . . . . . . . . .4 Interface fracture. . . 23. . . . . . . .3 Interface idealization and notations. . . . . . . . . . . .
. 24. .28Crack paths for Iosipescu’s beam. . 24. . . . . . . . . .23Crack propagation in Iosipescu’s beam. . . . . . . . .33Load displacement curve for test I. . . . . . . . . . 6). . . . . . . . . . .32Crack patterns. . . . . 24. . .31Crack propagation for anchor bolt pull out test II. . . . . .34Load displacement curve for test II. . .4. . . . . . . . . . . . . . . . . . . . . . . . . 24. . . . . . . . (Steps 1 & 3). . . . . . . . . . . . . . . . . . . . . . . .26Meshes for crack propagation in Iosipescu’s beam (Steps 1. . . . . . . . .5). . . . . . . .27Iosipescu’s beam with ICM model. . . . . . . . 24. . . . . . . . . . h = 50 x 100 mm. 24. . . . . . . . . . . . . . . . . .30Crack propagation for anchor bolt pull out test I. . . . . . . . . 24. . . . . . . . . .
Victor Saouma
Fracture Mechanics
. . . .25Multiple crack propagation in Iosipescu’s beam (Step 5). . . (Increment 11 & 39 in Step 24. . . . . 31 32 33 34 35 36 36 37 38 39 40 40 41
24. . . . . . . . . . . . . . . .24Multiple crack propagation in Iosipescu’s beam (Steps 3. . . . . . . . . . . . . . . . 24. . . . . 24.29Large Iosipescu’s beam. . . . . 24. . . . . . . . . . . . . . . . . . . . . .Draft
vi
LIST OF FIGURES .3. . . . . .22Crack propagation in Iosipescu’s beam. .4). 24. . . . . . . . .
. . . 24 10. . 13. 1981) . . . . 10 16. . (Kumar et al. . . . . 13. . .1 Material Properties and Loads for Diﬀerent Cases . . . . . . . . . . . . 1981) . . . . . (Kumar et al. . . . (Kumar et al. . (Kumar et al. . . . . . . . . . 13. . . 26 12. 12 16. . 12
10. . . German and Shih 1981) . . . 1981) 13. . . . . . . . .7 F and V1 for Internally Pressurized. . . . . . .1 Comparison of Various Models in LEFM and EPFM . . . . . . . V1 . . . . . . . . . . . . . .9 F . . . . . . . . . . . . . .1 Strain Energy versus Fracture Energy for uniaxial Concrete Specimen . Axially Cracked Cylinder. . . . . . . . . 2 15 19 21 22 24 25 26 28 28
15. . . 1981) . . . . . 13. . . . . . . . . . . . . . . . . . . . .2 Analytical and Numerical Results . . . . . .4 Column Instability Versus Fracture Instability . . . . . . . . . . .1 Eﬀect of Plasticity on the Crack Tip Stress Field. . .2 hFunctions for Standard ASTM Compact Tension Specimen. . . . . . . . . . . . . . . . . . . . Axially Cracked Cylinder. . . 5
Number of Elastic Constants for Diﬀerent Materials . . . . . . . . . . . . . (Anderson 1995) . . . . . . . . . . . . . . .3 7. . . . . . . . . .45C − Ni − Cr − Mo Steel . . . . . 13. . . . . . . . . . . . . . .1 5. . . . . . . . . . . . . . . . . . . . and Internal Pressure . . . . . . . . . . 12 . . . . 7 . . C Factors for Point Load on Edge Crack . . . . . . . . 9 . . . . . . . . . . . . . .1 7. . Approximate Fracture Toughness of Common Engineering Materials Fracture Toughness vs Yield Stress for . . . 15 17. 15 16. . . . (Kumar et al. . . 13. . . . 17. . . . . . . . . 1
. . . . . 23 10. . . . .3 Plane Stress hFunctions for a CenterCracked Panel. . . . . .6 hFunctions for an Internally Pressurized. . . . .2 Summary relations for the concrete fracture models. . . . . (Kumar. . . . . . . . . . . . . . . . . . 2 3
. . . . . . . .2 Experimentally obtained material properties of the concrete mixes used. . . 27 Summary of Elasticity Based Problems Analysed .1 6. . . . . . . . . . . . . . . . . . 13. . . . . . . . . . . . . (Kumar et al. . . . . . . . . . (Kumar et al. . . . . . . 1981) . . . . . . . . . .3 Numerical Results using Sintegral without the bimaterial model . . to Biaxial . . .1 7. . . . .4 hFunctions for Single Edge Notched Specimen.Draft
List of Tables
4. . and V2 for a Circumferentially Cracked Cylinder in Tension. . . . . . .8 hFunctions for a Circumferentially Cracked Cylinder in Tension. . . . . . . . . . .2 7. . .5 hFunctions for Double Edge Notched Specimen. . . . . . Newman’s Solution for Circular Hole in an Inﬁnite Plate subjected Loading. 13. . . . . . . . . . . . . . . . . . . . . . . . . . .1 Size Eﬀect Law vs Column Curve . . . . . . . 1981) .3 When to Use LEFM or NLFM Fracture Models .1 Concrete mix design. . . . . . 1981) .
. . . . . . . . . . F
18. . . . . . . . . and Saouma. . . . . . . . . . . . . . . . . . . . . . E. . . . . . . . . . . . . 12 3 7 7 7 9 9 11 12 13 16 20 21
Wedgesplitting specimen dimensions. . . . 8 . 18. CHECK Mapped proﬁle spacing. . . . . . . . .3 17. . Test matrix. . . . . . . . Concrete mix design .3 Material properties for ICM for Iosipescu’s test. . .7 17. V. . . . . . .11Ampliﬁcation factors for fractal surface areas with D = 1. . . . . . . . . . . . 18. . Maximum size aggregate) . . . . . . . . . . .8 Fractal dimension for various proﬁle segments and distances from centerline in specimen S33A . . . . .. . . . . . . . . . . . . . . . . . . . .6 Computed fractal dimensions for various synthetic curves . . . . . Broz. . . . . . . . . . . . . . . . . . . . . . . . . . . . .12Comparison between “corrected” G∗ and Gc values based on Swartz Tests (1992). Fracture energy values obtained from the CJWS specimens. . . . Size Eﬀect Law model assessment from the WS test program (average values)(Br¨hwiler. . . . . . J. . . . . . . .5 17. . . . . . . 28 24. . . . . . .6 17. . . .2 Material properties for direct shear test. .9 Comparison between D. . . Fracture toughness values obtained from the CJWS specimens. . . . . . . . 3 . .E. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 . . . . . orientation. .5 Computed fractal dimensions of a straight line with various inclinations . . . Summary of fracture toughness data obtained from the WS tests. . 10 . 18. . . . . . . . . . and GF for all specimens (MSA. . . . . . . . . . . . . . . . .7 Fractal dimension D versus proﬁle orientations .. . . . . . u
Fractal dimension deﬁnition . . . . . . . . . . . . . . . . . 18. and resolution for the two specimen sizes investigated . . . . . . .Draft
ii 17. . . . . . . . . . . . . . .9
LIST OF TABLES . . . . . . . . . . . . . . .J. . . . . . . . . . . . . . . . . . . . . . . . . . .2 18. . . . . . . .10Linear regression coeﬃcients between GF and KIc with D . .1 Material properties for direct shear test. . . . . . . . .1 18. . .4 17. . .1 . . . . 29
Victor Saouma
Fracture Mechanics
. . . . . . . . . . . . . 1991) . . . . . . KIc . 18. . 27 24. . . . . Range and resolution of the proﬁlometer (inches) . 18.3 18. . . . . . 8 . . . . . . . . . . . 18.8 17. . . . . . . . .4
24.. Summary of speciﬁc fracture energy values obtained from the WS tests. . . 18. 4 . . . . . . . .
Kirch. Elasticity Elasticity. Day 14 19 21 26 29 2 4 9 11 16 18 23 26 2 4 9 11 16 18 30 1 6 8 13 15 20 22 27 29 COVERAGE Intro. Merlin Theoretical Strength Theoretical Strength Energy Energy MERLIN Mixed Mode Plastic Zone Size CTOD.Draft
LIST OF TABLES COVERAGE Mon. Feb. Coverage Overview. Methods Num. Williams LEFM LEFM. 11 12 13 14
Victor Saouma
Fracture Mechanics
. J J Fatigue Fatigue Fatigue Concrete Fatigue Concrete Concrete EXAM Concrete Num. Examples BiMaterial. Methods Experiment Anisotropic
1
10 Apr. Hole Crack. 1 2 3 4 5 6 7 8 9 Mar. Jan. Griﬃth Notch.
Draft
2
LIST OF TABLES
Victor Saouma
Fracture Mechanics
.
Draft
Part I
PREAMBULE
.
Draft
.
where we have a discontinuity of both strain and displacements. Whereas continuity of displacements is maintained. .1 Classiﬁcation of Failure Kinematics
1 We can distinguish three separate modes of failure according to the degree of C(dis)continuity in failure processes. + uε =ε
C Continuity
. u += . + uε =ε
Figure 1.Draft
Chapter 1
FINITE ELEMENT MODELS FOR for PROGRESSIVE FAILURES
1. .
111111111 000000000 111111111 000000000 111111111 000000000 111111111 000000000 111111111 000000000 111111111 000000000 111111111 000000000 111111111 000000000 111111111 000000000 111111111 000000000
+ 
111111111 000000000 111111111 000000000 111111111 000000000 111111111 000000000 111111111 000000000 111111111 000000000 111111111 000000000 111111111 000000000 111111111 000000000 111111111 000000000 111111111 000000000 111111111 000000000
+ 
111111 000000 111111 000000 111111 000000 111111 000000 111111 000000 111111 000000 111111 000000 111111 000000 111111 000000 111111 000000 111111 000000 111111 000000 111111 000000 111111 000000 111111 000000 111111 000000 111111 000000 111111 000000
+ 
1
0
1
C Continuity
. .
. + uε =ε
C Continuity
.1: (?). u += .1: Kinematics of Continuous and Discontinuous Failure Processes Diﬀuse failure: corresponds to C 1 continuity of motion. Localized failure: corresponds to a C 0 continuity of motion. we do have a discontinuity in strains. . . Fig. . Discrete failure: corresponds to C −1 continuity of motion. u += . We do have a continuity of both displacement (increments) and strains across a potential failure surface. 1.
solutions rooted in continuum mechanics (plasticity and damage mechanics) have historically addressed problems with C 1 and C 0 continuity. Fig.1. or within the framework of fracture mechanics. of Elements
Figure 1. 3 With reference to Fig. This includes Plasticity theory Damage mechanics
Victor Saouma
Fracture Mechanics
. 1. Damage) Constant Number of Elements/Nodes
Fracture Mechanics Increase No. whereas fracture mechanics has focused on problems with C −1 continuity. we have two class of solutions
Continuum mechanics based solutions. in which the solid is assumed to be inherently continuous.Draft
2
FINITE ELEMENT MODELS FOR for PROGRESSIVE FAILURES
2 The numerical simulation of progressive failure in solids has been historically addressed within the frameworks of either computational plasticity (and more recently damage mechanics). 1. of Elements Decrease No. 4 In the context of ﬁnite element analysis of failure processes diﬀerent strategies can be adopted.2:
11 11 1 1 11 00 00 0 0 00 11 1 1 11 00 0 0 00 11 1 1 11 00 0 0 00 11 11 1 1 11 00 00 0 0 00 11 11 1 1 11 00 00 0 0 00 11 1 1 11 00 0 0 00 11 11 1 1 11 00 00 0 0 00 11 11 1 1 11 00 00 0 0 00 11 1 1 11 00 0 0 00 11 1 1 11 00 0 0 00 11 11 1 1 11 00 00 0 0 00 11 11 1 1 11 00 00 0 0 00 11 1 00 0 11 00 11 11 1 1 11 00 00 0 0 00 11 11 1 1 11 00 00 0 0 00 11 00 1 11 0 00 11 00 1 11 0 00 11 11 1 1 11 00 00 0 0 00 1 1 11 0 0 00 1 1 11 0 0 00 11 11 1 1 11 00 00 0 0 00 11 1 1 11 00 0 0 00 11 11 1 1 11 00 00 0 0 00 11 11 1 1 11 00 00 0 0 00
Smeared
Embedded
Discrete
Lattice
11 11 00 00
Continuum Mechanics (Plasticity.2: DiscreteSmeared Crack Models
Smeared Crack Embedded Crack Discrete Crack Lattice Model
5
Broadly speaking.
whereas discrete cracks are more suitably analysed within the framework of fracture mechanics.1 Classiﬁcation of Failure Kinematics
3
Fracture mechanics based solutions. where a physical discontinuity is assumed to exist from the onset.
6 Smeared and embedded cracks formulations are both based on continuum mechanics formulations. INSERT FMC1 here
Victor Saouma
Fracture Mechanics
.Draft
1.
Draft
4
FINITE ELEMENT MODELS FOR for PROGRESSIVE FAILURES
Victor Saouma
Fracture Mechanics
.
the response is linearly elastic.Draft
Chapter 2
Continuum Mechanics Based Description of Failure. In this plastic range.1
σ σy2 σy1 σy0
A0 A1
A2
O
O1
O2
ε1
p
ε1 ε2
p
e
ε ε2
e
Figure 2.
9
.1
7
Introduction
Uniaxial Behavior
The typical stressstrain behavior of most metals in simple tension is shown in Fig. 2.
At point A0 .1.1: Typical StressStrain Curve of an Elastoplastic Bar
8 Up to the A0 . and eventually fails at C.1
2. Plasticity
2. from A0 to C the material becomes plastic and behaves irreversibly. the stiﬀness decreases progressively. the material has reached its elastic limit. and unloading follows the initial loading path. O − A represents the elastic range where the behavior is load path independent.
Thus only part of the total strain ε1 at A1 is 1 recovered upon unloading. The new yield point in compression at B corresponds to stress
σ
σyA σy0 2σy0
O C A0
A
2σy0 ε
σyB
B
σy0
Figure 2. when the material is loaded in compression after it has been loaded in tension. that is the elastic strain εe .
E 1+ ε
σy (ε p )
σ
Figure 2. 2. A complete unloading would leave a permanent strain or a plastic strain εp . 2. 1
The Rheological model for elastoplasticity. Plasticity
10 Unloading from any point between A0 and C results in a proportionally decreasing stress and strain along A1 O1 parallel to the initial elastic loading O0 A0 . has only one spring element and a friction element.3. Fig. This phenomena is called Bauschinger eﬀect.
Victor Saouma
Fracture Mechanics
.Draft
2
11
Continuum Mechanics Based Description of Failure.3: Bauschinger Eﬀect on Reversed Loading
σB which is smaller than σy0 and is much smaller than the subsequent yield stress at A.2: Elastoplastic Rheological Model for Overstress Formulation
12 In case of reversed loading.2. the stressstrain curve will be diﬀerent from the one obtained from pure tension or compression. Fig.
but also on the entire loading history.1)
˙ = ˙e + ˙p if σ ≤ σy (elasticity).2)
(2. Fig.4) (2. it is not possible to express the stressstrain relationship in terms of total strain or stress.
σ
Et A
1 B
dσ dε
e
dε
p
E 1
dε
ε
Figure 2. then. stress history and deformation history.4.e. One is the Total Formulation of Plasticity (Deformation Theory) and the other is the Rate Formulation of Plasticity (Rate Theory). i.5.1 Introduction
3
13 It is thus apparent that the stressstrain behavior in the plastic range is path dependent. 14 Since there is not a one to one relationship between stress and strain in the plastic state.3) (2. 16 The deformation theory is a total secanttype formulation of plasticity that is based on the additive decomposition of total strain into elastic and plastic components
= while the rate theory is deﬁned by
e
+
p
(2. Fig.Draft
2. 2. Thus for elasticplastic materials only an incremental relationship between stress and strain increments can be written.4: Stress and Strain Increments in ElastoPlastic Materials
15 There are two theories corresponding with elastoplasticity.6)
Victor Saouma
Fracture Mechanics
.5)
(2. In general the strain will not depend on the the current stress state. then ˙ = ˙e = if σ > σy (plasticity). 2. ˙ = ˙e + ˙p ˙= where ET = σ ˙ σ ˙ σ ˙ + = E Ep ET EEp E + Ep σ ˙ E
(2.
9a) (2. and plastic ﬂows begins when the yield stress is reached ε = ε = σ for σ < σy0 E σ + λ for σ = σY 0 E (2.8a) (2.7b)
ElasticLinearly Hardening model.8b)
ElasticExponential Hardening where a power law is assumed for the plastic region σ for σ < σy0 E ε = kεn for σ = σY 0 ε = Victor Saouma (2.Draft
4
17
Continuum Mechanics Based Description of Failure.
Note that
Hardening ET = = 0.1. Perfectly Plastic < 0.2
18
Idealized StressStrain Relationships
There are many stressstrain models for the elasticplastic behavior under monotonic loading:
ElasticPerfectly Plastic where hardening is neglected.9b) Fracture Mechanics
. Plasticity
> 0. Softening
σ σy E ε εe εp
Figure 2. where the tangential modulus is assumed to be constant ε = ε = σ for σ < σy0 E 1 σ + (σ − σy0 for σ = σY 0 E Et (2.5: StressStrain diagram for Elastoplasticity
ET
2.7a) (2.
18) −p 0 σhyd = −pI = 0 0 0 −p Victor Saouma Fracture Mechanics
.
2.12) (2.2.11) κ= Wp = εp = σdεp dεp Plastic Work Plastic Strain (2.1
2. 20
First we deﬁne a hardening parameter or plastic internal variable.1.17)
1 1 1 σ = −p = (σ11 + σ22 + σ33 ) = σii = tr σ 3 3 3 then the stress tensor can be written as the sum of two tensors:
Hydrostatic stress in which each normal stress is equal to −p and the shear stresses are zero. then
t c σy (κ) − σy (κ) = 2σy0
(2.2.3
2.14)
Kinematic hardening rule states that the diﬀerence between the yield stresses under tension t c loading and under compression loading remains constant.13)
κ.1. If we denote by σy and σy the yield stress under tension and compression respectively. tangent modulus Et and plastic modulus Ep with the hardening parameter κ. κ =
21
a hardening rule expresses the relationship of the subsequent yield stress σy . σ = σ(κ) (2.1
22
Review of Continuum Mechanics
Stress
Hydrostatic and Deviatoric Stress Tensors
If we let σ denote the mean normal stress p (2. The most commonly hardening rules are: Isotropic hardening rule states that the progressively increasing yield stresses under both tension and compression loadings are always the same.2
2. which is often denoted by √ κ = εp = dεp dεp Equivalent Plastic Strain (2.15) (2.10)
Hardening Rules
19 A hardening rule describes a speciﬁc relationship between the subsequent yield stress σy of a material and the plastic deformation accumulated during prior loadings.Draft
2.16)
or alternatively σ − c(κ) = σy0 where c(κ) represents the track of the elastic center and satisﬁes c(0) = 0. The hydrostatic stress produces volume change without change in shape in an isotropic medium.2 Review of Continuum Mechanics RambergOsgood which is a nonlinear smooth single expression ε= σ σ +a E b
n
5
(2. −p 0 0 (2.
FIg. as the coordinates. and every point on this axis has σ1 = σ2 = σ3 = p. a/ 3). σ3 )
σ2
π/2
σ1
Figure 2.6: HaighWestergaard Stress Space
24 The decomposition of a stress state into a hydrostatic.p)
Hy
dr
os
tat ic a
xi s
σ3
ξ
deviatoric plane P( σ1 . 0. s2 . 1/ 3. 2. Considering an arbitrary stress state OP starting from O(0. pδij and deviatoric sij stress components can be geometrically represented in this space.
Vector NP represents the deviatoric component of the stress state (s1 . σ3 ). the vector OP can be decomposed into two√ components ON and NP. and axis Oξ is called the hydrostatic axis ξ.20) 3
26
Victor Saouma
Fracture Mechanics
. σ2 .
(s
O
π/2
1 . The former is along the direction of the unit vector √ √ (1 3.s 2 . Any plane perpendicular to the hydrostatic axis is called the deviatoric plane and is expressed as σ1 + σ2 + σ3 √ =ξ (2.p. σ2 . and NP⊥ON.2
Continuum Mechanics Based Description of Failure. 0) and ending at P (σ1 .1.s 3)
ρ
N(p.Draft
6 2. s3 ) and is perpendicular to the ξ axis. This stress representaion is known as the HaighWestergaard stress space.2. s11 − σ s12 s13 s22 − σ s23 s = s21 s31 s32 s33 − σ Geometric Representation of Stress States
(2. and σ3 . Plasticity
Deviatoric Stress: which causes the change in shape. σ2 . a threedimensional stress space can be constructed.19)
23 Using the three principal stresses σ1 .6. 25
Vector ON represents the hydrostatic component of the stress state.
2 Review of Continuum Mechanics
7
and the particular plane which passes through the origin is called the π plane and is represented by ξ = 0.Draft
2. The vector NP lies in a meridian plane and has ρ= s2 + s2 + s2 = 1 2 3 2J2 (2. They are therefore invariants of the stress state. 29
If we examine the deviatoric stress invariants.31)
J2 = −(σ(1) σ(2) + σ(2) σ(3) + σ(3) σ(1) ) (2.1. the cubic equation takes the form λ3 − J1 λ2 − J2 λ − J3 = 0 (2.23)
where the symbols J1 .28)
In terms of the principal stresses.24)
2 σ31
+
2 σ12
(2.21)
27 The projection of NP and the coordinate axis σi on a deviatoric plane are shown in Fig. s11 − λ s12 s13 s21 s22 − λ s23 s31 s32 s33 − λ
= 0
(2.3
Invariants
28 The principal stresses are physical quantities. J2 and J3 denote the following scalar expressions in the stress components: J1 = σ11 + σ22 + σ33 = σii = tr σ + J2 = −(σ11 σ22 + σ22 σ33 + σ33 σ11 ) + 1 1 2 1 (σij σij − σii σjj ) = σij σij − Iσ = 2 2 2 1 2 (σ : σ − Iσ ) = 2 1 J3 = detσ = eijk epqr σip σjq σkr 6
2 σ23
30
(2.30)
Victor Saouma
Fracture Mechanics
.27) (2. those invariants can be simpliﬁed into J1 = σ(1) + σ(2) + σ(3) J3 = σ(1) σ(2) σ(3) (2.22c is expanded.29) (2.22c)
σrs − λδrs  = 0 σ − λI = 0
When the determinant in the characteristic Eq.22a) (2.25) (2. Any plane containing the hydrostatic axis is called a meridian plane. whose values do not depend on the coordinate system in which the components of the stress were initially given.22b) (2.26) (2. 2.2. COMPLETE HERE AND CHECK
2.
if we deﬁne 1 1 e = tr E 3 3 (2. while the spherical or hydrostatic strain 1 e1 represents the volume change. s11 − σ s12 s13 s22 − σ s23 s = s21 s31 s32 s33 − σ
(2.38)
We note that E measures the change in shape of an element. while the spherical or hydrostatic strain 1 e1 represents the volume change. the lagrangian and Eulerian linear strain tensors can each be split into spherical and deviator tensor as was the case for the stresses. 3 Victor Saouma Fracture Mechanics
. if we deﬁne
1 1 e = tr E 3 3 then the components of the strain deviator E are given by 1 1 Eij = Eij − eδij or E = E − e1 3 3
(2.37)
then the components of the strain deviator E are given by 1 1 Eij = Eij − eδij or E = E − e1 3 3 (2.33)
We note that E measures the change in shape of an element.Draft
8
Continuum Mechanics Based Description of Failure.2
2.2.32)
(2.36)
33
Similarly. −p 0 0 (2.35) −p 0 σhyd = −pI = 0 0 0 −p Deviatoric Stress: which causes the change in shape. Hence. Plasticity
2.34)
then the stress tensor can be written as the sum of two tensors: Hydrostatic stress in which each normal stress is equal to −p and the shear stresses are zero. The hydrostatic stress produces volume change without change in shape in an isotropic medium. Hence.2.2.1
Strains
Hydrostatic and Deviatoric Strain Tensors
31 The lagrangian and Eulerian linear strain tensors can each be split into spherical and deviator tensor as was the case for the stresses. 3
32
If we let σ denote the mean normal strain p 1 1 1 σ = −p = (σ11 + σ22 + σ33 ) = σii = tr σ 3 3 3 (2.
the material behaves elastically. The elasticplastic behavior of most metals is essentially hydrostatic pressure insensitive.39)
where the symbols J1 . J3 ) = 0 f (ρ.49b) (2. the stress state can be uniquely deﬁned by either one of the following set of variables f (σ1 .2.2
34
2.49a) (2.2. those invariants can be simpliﬁed into J1 = e(1) + e(2) + e(3) J2 = −(e(1) e(2) + e(2) e(3) + e(3) e(1) ) J3 = e(1) e(2) e(3) (2. and the yield surface can generally be expressed by any one of the following equations. In biaxial or triaxial state of stresses. thus the yield criteria will not depend on I1 . ρ.50b)
Victor Saouma
Fracture Mechanics
. the elastic limit is deﬁned mathematically by a certain yield criterion which is a function of the stress state σij expressed as
f (σij ) = 0
(2. σ2 .46) (2. the elastic limit is obtained by a welldeﬁned yield stress point σ0 . J2 and J3 denote the following scalar expressions in the strain components: J1 = e11 + e22 + e33 = eii = tr E J2 = −(e11 e22 + e22 e33 + e33 e11 ) + + 1 1 2 1 (eij eij − eii ejj ) = eij eij − J1 = 2 2 2 1 2 (E : E − IE ) = 2 1 J3 = detE = eijk epqr Eip Ejq Ekr 6 e2 23
35
(2.3 Yield Criteria Invariants
9
Determination of the principal strains (E(3) < E(2) < E(1) .45) (2. on it it begins to yield. Within it.3
Yield Criteria
36 In uniaxial stress states.41) (2.50a) (2. this surface is called the yield surface.44)
In terms of the principal strains. λ3 J1 λ2 − J2 λ − J3 = 0 (2. f (J2 .48)
For isotropic materials. J3 ) = 0 f (ξ.Draft
2.47)
2.40) e2 31 + e2 12 (2.49c)
those equations represent a surface in the principal stress space.43) (2. θ) = 0 (2. θ) = 0 (2.42) (2. σ3 ) = 0 f (I1 . J2 .
39
Internal variables deﬁne hardening and softening by describing motions and deformations of the yield surface. into elastic and plastic components as ˙ = ˙ e + ˙ p. equal to zero it has yielded entering the plastic range. This is further illustrated in Fig.
= 0.7. > 0.
42 Back to Eq.52)
38
In order to deﬁne the plastic strain rate ˙ p . q)
< 0.
σ F<0 F>0 F=0
Figure 2.
elastic plastic . ˙ (2. we need to ﬁrst determine if plasticity has occurred.4
Rate Theory
37 Using Flow Theory of Plasticity we decompose the strain rate ˙ . 2.
40 41 Assuming the Yield Function F = 0 is satisﬁed then the rate of change distinguished between plastic loading and elastic unloading is
˙ F (σ.7: General Yield Surface. (2. 2. ˙p = λ (2. not permitted
(2.54) ∂σ m
Victor Saouma
Fracture Mechanics
.53)
If the yield function is less than zero then the state of stress and internal variables is in the elastic range.51)
The total strain rate ˙ is deﬁned kinematically in terms of the straindisplacement matrix B by redeﬁning the rate of change of the displacement ﬁeld u as a function of time (velocity vector ˙ u) as ˙ = B ue . Plasticity
2.Draft
10
Continuum Mechanics Based Description of Failure. we deﬁne the Plastic Flow Rule as ˙ ˙ ∂Qp = λm.51 and focusing our attention to the plastic strain rate ˙ p . and if the yield function is greater than zero the stresses are no longer following the deﬁned yield function and this state is not permitted. This is done by introducing a Yield Function F in terms of stresses σ and internal variables q.
2.56a) (2. 2. ∂t ∂F ∂F ˙ ˙ :σ+ : q = 0.55)
˙ Therefore.59)
(2.60a) (2. we diﬀerentiate F (σ. the following KuhnTucker condition is always true ˙ ˙ λF = 0.58a) (2. we use the tangential stressstrain relation as follows: ˙ ˙ σ = E : ˙ = E : [˙ − ˙p ] = E : [˙ − λm] (2.Draft
43 44
2.60c enables use to deﬁne the plastic multiplier as ˙ λ= n:E: ˙ Hp + n : E : m (2. q) with respect to time so
˙ F ˙ F or
= =
∂F = 0.61). In the case of plastic loading. As a result.the plastic multiplier must be equal to zero (λ = 0) in order for the rate of change Yield Function to be less than zero (F < 0). a Consistency Condition is deﬁned to enforce that the rate of change of stress and internal variables satisfy the yield condition having the plastic ˙ loading (F = 0).62)
48
substituting Eq. we get ˙ σ = E : [˙ − m( n:E: ˙ )] Hp + n : E : m (2. Qp = F and when nonassociated ﬂow is used. (2.60c)
46
Eq.
11
The associated ﬂow is referred to as a normality of plastic behavior.60b) (2.61)
47
Finally.56b)
∂F ∂F ∂q ˙ ˙ ˙ F = :σ+ : λ = 0. Qp = F . ∂σ ∂q
(2.57)
(2.63) Fracture Mechanics
Victor Saouma
. ?? as ˙ ˙ ˙ F = n : σ − Hp : λ = 0 then by substitutions ˙ F ˙ F ˙ F ˙ = n : E : [˙ − ˙p ] − Hp λ = 0 ˙ ˙ = n : E : [˙ − λm] − Hp λ = 0 ˙ ˙ = n : E : ˙ − λn : E : m − Hp λ = 0 ∂F ∂σ ∂F ∂q : = − ∂q ∂λ
(2. When associated ﬂow is used.4 Rate Theory ˙ where λ is the plastic multiplier and Qp is the plastic ﬂow potential. the plastic multiplier.58b)
(2.
45 After establishing the Plastic Flow Rule. ∂σ ∂q ∂λ Introducing the normal n and the hardening parameter Hp n = Hp we can rewrite Eq.
Draft
12
49
Continuum Mechanics Based Description of Failure; Plasticity
Rearranging we ﬁnd ˙ σ = [E −
E :m⊗n:E ]: ˙ Hp + n : E : m
(2.64) (2.65)
or ˙ σ = E ep : ˙ where E ep is the elastoplastic material operator (modulus of elastoplasticity) deﬁned by E ep = E − E :m⊗n:E Hp + n : E : m (2.66)
50
Note that for associated ﬂow, n = m and for nonassociated ﬂow, n = m.
2.5
J2 Plasticity/von Mises Plasticity
51 For J2 plasticity or von Mises plasticity, our stress function is perfectly plastic. Recall perfectly plastic materials have a total modulus of elasticity (ET ) which is equivalent to zero. We will deal now with deviatoric stress and strain for the J2 plasticity stress function.
1. Yield function: 1 2 1 F (s) = s : s − σy = 0 2 3 2. Flow rate (associated): ˙ ˙ ep = λ ˙ 3. Consistency condition (F = 0): ∂F ∂F ˙ ˙ ˙ :s+ :q=0 F = ∂s ∂q ˙ ˙ since q = 0 in perfect plasticity, the second term drops out and F becomes ˙ ˙ F =s:s=0 Recall that ˙ ˙ ˙ ˙ s = 2G : ee = 2G : [e − ep ] ˙ ﬁnally substituting ep in ˙ substituting s back into (2.70) ˙ ˙ ˙ F = 2Gs : [e − λm] = 0 ˙ and solving for λ s: ˙ ˙ λ= s:s (2.73) ˙ ˙ ˙ s = 2G : [e − λs] (2.71) (2.72) (2.70) (2.69) ∂Qp ˙ ∂F = λs ˙ =λ ∂s ∂s (2.68) (2.67)
(2.74)
Victor Saouma
Fracture Mechanics
Draft
2.5 J2 Plasticity/von Mises Plasticity 4. Tangential stressstrain relation(deviatoric): ˙ ˙ s = 2G : [e − ˙ then by factoring e out ˙ s = 2G : [I 4 − Now we have the simpliﬁed expression ˙ ˙ s = Gep : e where Gep = 2G : [I 4 − ˙ s:e s] s:s
13
(2.75)
s⊗s ˙ ]:e s:s
(2.76)
(2.77)
s⊗s ] (2.78) s:s is the 4th order elastoplastic shear modulus tensor which relates deviatoric stress rate to deviatoric strain rate. 5. Solving for E ep in order to relate regular stress and strain rates: Volumetric response in purely elastic ˙ tr (σ) = 3Ktr (˙ ) altogether 1 ˙ ˙ ˙ σ = tr (σ) : I 2 + s 3 ˙ ˙ σ = Ktr (˙ ) : I 2 + Gep : e 1 ˙ σ = Ktr (˙ ) : I 2 + Gep : [˙ − tr (˙ ) : I 2 ] 3 1 ˙ σ = Ktr (˙ ) : I 2 + Gep : ˙ − tr (˙ )Gep : I 2 3 2 ˙ σ = Ktr (˙ ) : I 2 − Gtr (˙ )I 2 + Gep : ˙ 3 2 ˙ σ = KI 2 ⊗ I 2 : ˙ − GI 2 ⊗ I 2 : ˙ + Gep : ˙ 3 and ﬁnally we have recovered the stressstrain relationship 2 ˙ σ = [[K − G]I 2 ⊗ I 2 + Gep ] : ˙ 3 where the elastoplastic material tensor is 2 E Ô = [[K − G]I 2 ⊗ I 2 + Gep ] 3 (2.87) (2.80) (2.81) (2.82) (2.83) (2.84) (2.85) (2.79)
(2.86)
Victor Saouma
Fracture Mechanics
Draft
14
Continuum Mechanics Based Description of Failure; Plasticity
s3 Fn+1= 0 for Hp > 0 (Hardening) Fn = 0 for Hp = 0 (Perfectly Plastic) Fn+1= 0 for Hp < 0 (Softening) s1 s2
Figure 2.8: Isotropic Hardening/Softening
2.5.1
52
Isotropic Hardening/Softening(J2 − plasticity)
In isotropic hardening/softening the yield surface may shrink (softening) or expand (hardening) uniformly (see ﬁgure 2.8). 1. Yield function for linear strain hardening/softening: F (s, 2. Consistency condition:
p ef f )
=
1 o 1 s : s − (σy + Ep 2 3
p 2 ef f )
=0
(2.88)
˙ ∂F ∂ q ˙ ∂F ˙ ˙ λ=0 :s+ : F = ˙ ∂s ∂q ∂ λ from which we solve the plastic multiplier ˙ λ= ˙ 2Gs : e 2Gs : s +
2Ep o 3 (σy p ef f )
(2.89)
+ Ep
2 3s
:s
(2.90)
3. Tangential stressstrain relation(deviatoric): ˙ ˙ s = Gep : e where Gep = 2G[I 4 − Gs : s + 2Gs ⊗ s
2Ep o 3 (σy
(2.91)
+ Ep
p ef f )
2 3s
:s
]
(2.92)
Note that isotropic hardening/softening is a poor representation of plastic behavior under cyclic loading because of the Bauschinger eﬀect.
53
Victor Saouma
Fracture Mechanics
Draft
2.5 J2 Plasticity/von Mises Plasticity
15
Fn+2 = 0
s
3
Fn+1 = 0 Q P
S
O R Fn = 0
s1
s2
Figure 2.9: Kinematic Hardening/Softening
2.5.2
54
Kinematic Hardening/Softening(J2 − plasticity)
Kinematic hardening/softening, developed by Prager [1956], involves a shift of the origin of the yield surface (see ﬁgure 2.9). Here, kinematic hardening/softening captures the Bauschinger eﬀect in a more realistic manner than the isotropic hardening/ softening. 1. Yield function: F (s, α) = 1 2 1 (s − α) : (s − α) − σy = 0 2 3 (2.93)
2. Consistency condition (plastic multiplier): ˙ λ= where C = Ep and C is related to α, the backstress, by ˙ ˙ ˙ α = C e = λC(s − α) For perfectly plastic behavior C = 0 and α = 0. 3. Tangential stressstrain relation (deviatoric): ˙ ˙ s = Gep : e where Gep = 2G[I 4 − 2G(s − α) ⊗ (s − α) ] (s − α) : (s − α)[2G + C] (2.97) (2.96) (2.95) ˙ 2G(s − α) : e (s − α) : (s − α)[2G + C] (2.94)
(2.98)
Victor Saouma
Fracture Mechanics
Draft
16
Continuum Mechanics Based Description of Failure; Plasticity
Victor Saouma
Fracture Mechanics
Draft
Chapter 3
Continuum Mechanics Based Description of Failure; Damage Mechanics
Draft
2
Continuum Mechanics Based Description of Failure; Damage Mechanics
Victor Saouma
Fracture Mechanics
Draft
Part II
FRACTURE MECHANICS
Draft
.
In fact. Buxbam. The next section will provide a brief panoramic overview of the major developments in fracture mechanics. a cracked structure can be designed on the sole basis of strength as long as the crack size does not exceed a critical value. Rosenﬁeld. Such a simpliﬁcation is not new. By analogy. Should the crack size exceed this critical value. However. In the next section we will discuss some examples of well known failures/accidents attributed to cracking. an unsafe design may result as instability (or buckling) is overlooked for slender members. then
. tensile instability (necking) 5.Draft
Chapter 4
INTRODUCTION
In this introductory chapter. Then.
4. If column strength is based entirely on a strength criterion. elastic instability (buckling) 2. gross plastic deformation (yielding) 4. we shall start by reviewing the various modes of structural failure and highlight the importance of fracture induced failure and contrast it with the limited coverage given to fracture mechanics in Engineering Education. using a simple example we shall compare the failure load predicted from linear elastic fracture mechanics with the one predicted by “classical” strength of materials. Honton and McMillan 1983). Buhr. and are the least well understood. fracture often has been overlooked as a potential mode of failure at the expense of an overemphasis on strength. the chapter will conclude with an outline of the lecture notes. Finally. and proper design procedures have been developed to resist them.1
Modes of Failures
The fundamental requirement of any structure is that it should be designed to resist mechanical failure through any (or a combination of) the following modes: 1. large elastic deformation (jamming) 3. fracture Most of these failure modes are relatively well understood. Fisher. and ﬁnds a very similar analogy in the critical load of a column. Thus failure curves for columns show a smooth transition in the failure mode from columns based on gross section yielding to columns based on instability. fractures occurring after earthquakes constitute the major source of structural damage (Duga.
about 4 percent of the gross national product. or more recently fatigue fracture of bulkhead in a Japan Air Line Boeing 747 4. examples of fracture failures include: • Mechanical. which resulted in the recall of 637 of them 5. scientists. on the basis of those two theories (strength of materials and fracture mechanics).2
Examples of Structural Failures Caused by Fracture
Some wellknown. and rails 2. rock cutting in mining 2. one could draw a failure curve that exhibits a smooth transition between those two modes.
1 When high strength rolled sections were ﬁrst introduced. after some spectacular bridge girder failures. Again. or marine 1. hydraufracturing for oil. and engineers (metallurgical. which exploded in midair during the ﬁfties. 2. fracture of the Glomar Java sea boat in 1984 6. gas. 4. axles. fatigue fractures found in the Grumman buses in New York City. and classical.
Victor Saouma
Fracture Mechanics
. However. 1983) stated: [The] cost of material fracture to the US [is] $ 119 billion per year. and civil). “Biting” of candies (!) Costs associated with fracture in general are so exorbitant. fracture of the Liberty ships during and after World War II 3. it has become a ﬁeld of research interest to mathematicians. such as the Comet airliners. fracture of airplanes. fatigue crack that triggered the sudden loss of the upper cockpit in the Air Aloha plane in Hawaii in 1988 • Civil engineering 1. 3. In light of the variety. mechanical. aeronautical. fracture of train wheels.1
4. and complexity of problems associated with fracture mechanics.Draft
2
INTRODUCTION
a fracturebased failure results. and geothermal energy recovery 3. aerospace. that a recent NBS report (Duga et al. in many cases fractures are manmade and induced for beneﬁcial purposes Examples include: 1. Costs could be further reduced by as much as $ 28 billion per year through fracturerelated research. The costs could be reduced by an estimated missing 35 billion per year if technology transfer were employed to assure the use of best practice. there was a rush to use them. fractures of bridge girders (Silver bridge in Ohio) fracture of Statfjord A platform concrete oﬀshore structure cracks in nuclear reactor piping systems fractures found in dams (usually unpublicized)
Despite the usually wellknown detrimental eﬀects of fractures. it was found that strength was achieved at the expense of toughness (which is the material ability to resist crack growth).
and KIc . based on this ﬁrst approach.12 πa
2
This example is adapted from (Kanninen and Popelar 1985). Eq. a cantilevered beam of length L.3)
2. width B. Fracture toughness is a measure of the material ability to resist crack growth (not to be confused with its tensile strength.4)
where KI is a measure of the stress singularity at the tip of the crack and KIc is the critical value of KI . height H. Fig. KI is related to σmax through: √ (4. is a material parameter (analogous to σy ). KI is a structural parameter (analogous to σmax ).5) KI = 1.2 governed failure.12σmax πa where a is the crack length. the maximum load which can be safely carried is:
SOM Pmax =
BH 2 σy 6L
(4.Draft
4. In applying a diﬀerent approach. Rather. which is associated with crack nucleation or formation). the structure cannot be assumed to be defect free. we consider a simple problem.2) σmax ≤ σy Thus. or (4.1)
1. 4. and subjected to a point load P at its free end. the maximum load that can be carried is given by: BH 2 KIc FM √ (4. failure is governed by: KI ≤ KIc (4.1: Cracked Cantilevered Beam is given by 6P L BH 2 We will seek to determine its safe loadcarrying capacity using the two approaches2 .1 Maximum ﬂexural stress
Figure 4. σmax = (4.3 Fracture Mechanics vs Strength of Materials
3
Fracture Mechanics vs Strength of Materials
In order to highlight the fundamental diﬀerences between strength of materials and fracture mechanics approaches. an initial crack must be assumed. 4. Thus. for the strength of materials approach in the linear elastic fracture mechanics approach (as discussed in the next chapter). one based on fracture mechanics. Based on classical strength of materials the maximum ﬂexural stress should not exceed the yield stress σy .
Victor Saouma
Fracture Mechanics
.6) Pmax = 6L 1.3
4.
call for the following remarks:
2. we can generalize our preliminary ﬁnding by the curve shown in Fig. z
Victor Saouma
Fracture Mechanics
. on the upper left failure is governed by linear elastic fracture mechanics.1. a column failure load is governed by either the Euler equation for long columns.2. 4 This curve will be subsequently developed for concrete materials according to Baˇant’s size eﬀect law. we should emphasize that size eﬀect is not unique to fractures but also has been encountered by most engineers in the design of columns.3 and 4. and on the upper right failure is triggered by a combination of fracture mechanics and plasticity. which is similar to Fig.P. depending upon its slenderness ratio.4. Eq. On the basis of the above. 4. Column formulas have been developed. as seen in Fig. whereas fracture failure is caused by the presence of a crack. or the strength of materials for short columns. 1984) for the failure of small or large cracked structures on the basis z z of either strength of materials or linear elastic fracture mechanics. Both equations are in terms of
BH 2 6L
INTRODUCTION
The two equations. 4.3 but also of crack size a. This last zone has been called elastoplastic in metals.6 governing the load capacity of the beam according to two diﬀerent approaches. 4. Similarly compressive strength of concrete is known to be slightly aﬀected by the cylinder size.Draft
4 1. as shown in Table 4. The fracture mechanics approach is not only a function of an intrinsic material property. In all other cases. We thus identify four corners: on the lower left we have our usual engineering design zone. or plasticity.
3 We will see later that KIc is often a function of crack length. 3.4 Finally. 4. 4. In fact. a perfect material is assumed. Also note that column instability is caused by a not perfectly straight element.2: Failure Envelope for a Cracked Cantilevered Beam On the basis of this simple example.
Figure 4. where factors of safety are relatively high. As will be shown later.2. and nonlinear fracture in concrete.3. Z. similar transition curves have also been developed by Baˇant (Baˇant. where failure stress is clearly a function of the crack length. The strength of materials approach equation is a function of a material property that is not size dependent. on the bottom right we have failure governed by yielding. we can schematically represent the failure envelope of this beam in Fig.
4: Column Curve
Approach Strength of Material Column Instability Fracture
Governing Eq.3: Generalized Failure Envelope
Figure 4.3 Fracture Mechanics vs Strength of Materials
5
Figure 4.1: Column Instability Versus Fracture Instability
Victor Saouma
Fracture Mechanics
.Draft
4. σ=P A σ= σ=
π2 E ( KL )2 r K √c πa
Theory Plasticity Euler Griﬃth
σy
KL r
Imperfection Dislocation Not Perfectly straight Microdefects
KIc
Table 4.
Thus. = 1 + 2 a ρ
1/2
(4. (RAE) decided that Griﬃth should stop wasting his time. and ρ is the radius of curvature)5 . 3. who was not orginally interested in the strength of cracked structures (fracture mechanics was not yet a discipline). and he was transferred to the engine department.Draft
6
INTRODUCTION
4. Inglis (Inglis 1913) extended the solution for stresses around a circular hole in an inﬁnite plate to the more general case of an ellipse. Following an investigation. respectively. With his assistant Lockspeiser.
5
Note that for a circle. During crack extension. They found that the strength increased rapidly as the size decreased. The second major contribution made by Griﬃth was in deriving a thermodynamical criterion for fracture by considering the total change in energy taking place during cracking. After Griﬃth’s work. potential energy (both external work and internal strain energy) is released and “transferred” to form surface energy. the subject of fracture mechanics was relatively dormant for about 20 years until 1939 when Westergaard (Westergaard 1939a) derived an expression for the stress ﬁeld near a sharp crack tip. but rather in the tensile strength of crystalline solids and its relation to the theory based on their lattice properties. Griﬃth was then working at the Royal Aircraft Establishment (RAE) at Farnborough. England (which had a tradition of tolerance for original and eccentric young researchers). Asymptotic values of 1.600 and 25 Ksi were found for inﬁnitesimally small and bulk size specimens. Unfortunately. Griﬃth’s ﬁrst major contribution to fracture mechanics was to suggest that internal minute ﬂaws acted as stress raisers in solids. It should be noted that this problem was solved 3 years earlier by Kolosoﬀ (who was the mentor of Muschelisvili) in St Petersbourg. in reviewing Inglis’s early work.4
Major Historical Developments in Fracture Mechanics
As with any engineering discipline approached for the ﬁrst time. 4. which is approximately equal to E/10 where E is the Young’s Modulus (Kelly 1974).F. and was testing the strength of glass rods of diﬀerent diameters at diﬀerent temperatures (Gordon 1988). On the basis of those two observations. Inglis’s early work was followed by the classical studies of Griﬃth. a German Engineer by the name of Kirsch showed that a stress concentration factor of 3 was found to exist around a circular hole in an inﬁnite plate subjected to uniform tensile stresses (Timoshenko and Goodier 1970). resulting in a ﬁre. While investigating the unexpected failure of naval ships in 1913. 2. it is helpful to put fracture mechanics into perspective by ﬁrst listing its major developments: 1. however history remembers only Inglis who showed that a stress concentration factor of S.C. thus strongly aﬀecting their tensile strengths. one night Lockspeiser forgot to turn oﬀ the gas torch used for glass melting. In 1898.
Victor Saouma
Fracture Mechanics
. 5. a stress concentration of 3 is recovered.7)
prevails around the ellipse (where a is the half length of the major axis. Griﬃth determined that the presence of minute elliptical ﬂaws were responsible in dramatically reducing the glass strength from the theoretical value to the actually measured value.
who introduce his J integral in 1968 in probably the second most referenced paper in the ﬁeld (after Griﬃth).S. and 10 fractured into two parts. and loading condition. P. (b) He altered Westergaard’s general solution by introducing the concept of the stress intensity factor (SIF).
Victor Saouma
Fracture Mechanics
. This brief overview is designed to make detailed coverage of subsequent topics better understood when put into global perspective. Thus a new meaning was given to cracks in cementitious materials. who around 1963 utilized the crack opening displacement (COD) as the parameter to characterize the strength of a crack in an elastoplastic solid. (ii) fracture of laminates and composites. 11. and Petersson.Draft
4. This resulted in what is sometimes called the modiﬁed Griﬃth’s theory. Nonlinear considerations were further addressed by Wells. 9.P. are below their critical value. (iii) numerical techniques. A. fracture mechanics was still a relatively obscure and esoteric science. and by Rice. Hillerborg (Hillerborg.4 Major Historical Developments in Fracture Mechanics
7
6. L. M. fracture mechanics concepts must be used. more than any other single factor. In 1979 Baˇant and Cedolin (Baˇant. or surround it by a corrosive environment. In 1976. Other major advances have been made subsequently in a number of subdisciplines of fracture mechanics: (i) dynamic crack growth. taken separately. it introduced a path independent contour line integral that is the rate of change of the potential energy for an elastic nonlinear solid during a unit crack extension. Paris in 1961 proposed the ﬁrst empirical equation relating the range of the stress intensity factor to the rate of crack growth. In either case the original crack length. After the war. the large number of sudden and catastrophic fractures that occurred in ships during and following World War II gave the impetus for the development of fracture mechanics. George Irwin. and Mod´er. and Cedolin.000 welded ships constructed during the war. He made three major contributions: (a) He (and independently Orowan) extended the Griﬃth’s original theory to metals by accounting for yielding at the crack tip. Of approximately 5. Up to this point. This form of crack propagation is driven by either applying repeated loading (fatigue) to a crack. who was at the U. 10. made use of Griﬃth’s idea. and that classical strength of materials models would yield results that are mesh sensitive. 1976) introduced e the ﬁctitious crack model in which residual tensile stresses can be transmitted across a portion of the crack. 1979) showed that for the z z objective analysis of cracked concrete structure. Subcritical crack growth was subsequently studied. Another major contribution was made by Erdogan and Sih in the mid ’60s when they introduced the ﬁrst model for mixedmode crack propagation. 12. with 150 of these being seriously damaged. and others. and thus set the foundations of fracture mechanics.E. Z. (iv) design philosophies.000 suﬀered structural damage. Naval Research Laboratory. over 1. However. 8. (c) He introduced the concept of energy release rate G 7.
such as concrete and rock. and more speciﬁcally fatigue crack growth will be covered in chapter seven. and the second on Williams’s classical paper. Fracture of cementitious material. and William’s solution will be extended to cracks along an interface between two dissimilar materials. chapter two will provide the reader with a review of elasticity. First we will derive expressions for the size of the plastic zone ahead of the crack tip. First. we shall review some of the major models currently investigated. with emphasize on both metallic and cementitious materials. will be studied in chapter ten. Subcritical crack growth. twelve. In particular we shall revisit the major equations needed to analytically solve simple problems involving elliptical holes or sharp cracks. Chapter six will extend the simple mode I crack propagation criteria to mixed modes (where a crack is simultaneously subjected to opening and sliding) by discussing some of the major criterions. Knott (Knott 1976). then we will deﬁne the energy release rate G. Finally. The last chapter. and derivation of the J integral will then be covered in chapter eight. we shall examine some of the fracture testing techniques. L. we shall thoroughly examine the theoretical strength of crystalline materials and contrast it with the actual one.
Victor Saouma
Fracture Mechanics
. Z. and Cedolin. This chapter. Barsom and Rolfe (Barsom and Rolfe 1987). and show how it can be used in some practical design cases. followed by methods to extract the stress intensity factors from a ﬁnite element analysis and evaluation of J integral will be presented. In chapter nine. Chapter four will formalize the Linear Elastic Fracture Mechanics approach. Through Westergaard’s solution. First techniques of modelling the stress singularity at the crack tip will be examined. we shall introduce the concept of stress intensity factors. For more detailed coverage. First Inglis solution of a circular and elliptical hole will be presented. A complementary approach to the stress based one. Those solutions will be presented in detail in chapter three. and to appreciate the value of numerical based solutions which will be discussed later. Cherepanov (Cherepanov 1979). a recent book by Baˇant (Baˇant.Draft
8
INTRODUCTION
4. The ﬁrst one is based on Westergaard’s solution. ElastoPlastic fracture mechanics. the reader is referred to the numerous excellent books available. With the rigorous derivation of the stress ﬁeld ahead of a crack tip performed.5
Coverage
Following this brief overview. In this extensive chapter. and discuss the duality between the stress based and the energy based approaches. Numerical techniques will then be discussed in chapter eleven. and Anderson’s (Anderson 1995). is an important one to understand the mathematical complexity of solutions of simple crack problem. Kanninen (Kanninen and Popelar 1985).P. such as Broek (Broek 1986. then the problem of a sharp crack in an inﬁnite plate will be solved using the two classical methods. will focus on numerical techniques for cementitious materials. and examine some applications. Broek 1989). mathematically the most challenging. z z 1991) covers (among other things) some of the issues related to fracture of concrete. will be presented in chapter ﬁve which discusses Energy Methods in linear elastic fracture mechanics. then criteria for crack growth presented. Also covered in this chapter will be the solutions of a crack in a homogeneous anisotropic solid based on the solution of Sih and Paris.
ideally.
. which essentially exists to operate on vectors v to produce other vectors (or on tensors to produce other tensors!). 11
We hereby adopt the dyadic notation for tensors as linear vector operators u = T·v or ui = Tij vj u = v·S where S = T
T
(5.1b)
12 Whereas a tensor is essentially an operator on vectors (or other tensors). which can only provide limited background to a solid fracture mechanics course.1
Tensors
10 We now seek to generalize the concept of a vector by introducing the tensor (T). Most often.
7 8 Accordingly. but not all. We designate this operation by T·v or simply Tv. of the material in this chapter will be subsequently referenced. students have had a graduate course in Advanced Strength of Materials. 13 Tensors frequently arise as physical entities whose components are the coeﬃcients of a linear relationship between vectors. 9 It should be noted that most. an introductory course in Continuum Mechanics should be taken prior to a fracture mechanics. independent of any particular coordinate system yet speciﬁed most conveniently by referring to an appropriate system of coordinates.
5. this preliminary chapter (mostly extracted from the author’s lecture notes in Continuum Mechanics) will partially remedy for occasional deﬃciencies and will be often referenced. it is also a physical quantity.Draft
Chapter 5
PRELIMINARY CONSIDERATIONS
Needs some minor editing!
Whereas. this is seldom the case.1a) (5.
Alternatively. 18
Hence. Victor Saouma Fracture Mechanics
(5. If there is one letter index. The following rules deﬁne indicial notation: 1.2)
other ﬁrst order tensors aij bj .
A force and a stress are tensors of order 1 and 2 respectively. For instance: (5. 4 for him). and the resulting tensors summed. 2.3) a1i xi = a11 x1 + a12 x2 + a13 x3 3. A Tensor of order zero is speciﬁed in any coordinate system by one coordinate and is a scalar.1
Indicial Notation
16 Whereas the Engineering notation may be the simplest and most intuitive one. δij uk vk . D11 D22 D13 Dij = D21 D22 D23 D31 D32 D33 other examples Aijip . A tensor of order one has three coordinate components in space. A repeated index will take on all the values of its range. εijk uj vk • Second order tensor (such as stress or strain) will have two free indices. it often leads to long and repetitive equations. the summation was over a repeated index (i if in our example).Draft
2
15
PRELIMINARY CONSIDERATIONS
14 A tensor is classiﬁed by the rank or order.1. this so called indicial notation is also referred to Einstein’s notation. In general 3D space the number of components of a tensor is 3n where n is the order of the tensor. hence it is a vector. he decided that there is no need to include the summation sign there was repeated indices (i). Einstein got tired of writing the summation symbol with its range of summation below and above (such as n=3 aij bi ) and noted that most of the time i=1 the upper range (n) was equal to the dimension of space (3 for us. 3
(5. and thus any repeated index is a dummy index and is summed over the range 1 to 3.5)
. 17 While working on general relativity. the tensor and the dyadic form will lead to shorter and more compact forms.4)
a1
19
a1 a2 a3
=
2 a 3
a
i = 1. and that when the summation involved a product of two terms.
5. An index that is not repeated is called free index and assumed to take a value from 1 to 3. For instance: ai = ai = assuming that n = 3. Fikk . Tensor’s order: • First order tensor (such as force) has only one free index: ai = ai = a1 a2 a3 (5. that index goes from i to n (range of the tensor). Hence.
10a)
21
Using indicial notation.1 Tensors
3
• A fourth order tensor (such as Elastic constants) will have four free indices. Derivatives of tensor with respect to xi is written as .2
22
Tensor Operations
The sum of two (second order) tensors is simply deﬁned as: Sij = Tij + Uij (5.
5. i.1. when expanded would yield: x1 = c11 z1 + c12 z2 + c13 z3 x2 = c21 z1 + c22 z2 + c23 z3 x3 = c31 z1 + c32 z2 + c33 z3 Similarly: Aij = Bip Cjq Dpq (5.9) (5.i
∂vi ∂xi
= vi.j
∂Ti. For instance: (5. we may rewrite the deﬁnition of the dot product a·b = ai bi (5.8a)
A11 = B11 C11 D11 + B11 C12 D12 + B12 C11 D21 + B12 C12 D22 A12 = B11 C11 D11 + B11 C12 D12 + B12 C11 D21 + B12 C12 D22 A21 = B21 C11 D11 + B21 C12 D12 + B22 C11 D21 + B22 C12 D22 A22 = B21 C21 D11 + B21 C22 D12 + B22 C21 D21 + B22 C22 D22 (5. there is one free index p thus there are three equations.j. For example:
∂Φ ∂xi
= Φ.6)
20 Usefulness of the indicial notation is in presenting systems of equations in compact form.
4.Draft
5.i
∂vi ∂xj
= vi.13)
Victor Saouma
Fracture Mechanics
.7) xi = cij zj
this simple compacted equation.k
(5. thus each equation has nine terms.12)
we note that in the second equation.j ∂xk
= Ti. there are two repeated (dummy) indices q and r.11)
and of the cross product a×b = εpqr aq br ep (5.
19a) (5. thus producing a tensor of order two less than that to which it is applied.18)
in any rectangular system. For example: 2 → 0 Tij → Tii .16a) .17b) (5.sn . 4 → 2 (5.19c) (5..14)
In a contraction.16c)
vi Tjk = Sijk
26 The inner product is obtained from an outer product by contraction involving one index from each tensor.s .j (5.k Bi
→ A
i
.17d)
k
ai Ejk → ai Eik = fk Eij Fkm → Eij Fjm = Gim A
i .sm 3 → 1 Eij ak → Eij ai = cj . For example
25
ai bj A
i .17c) (5.19b) (5.
28
The following innerproduct axioms are satisﬁed: T:U = U:T T : (U + V) = T : U + T : V α(T : U) = (αT) : U = T : (αU) T : T > 0 unless T = 0 (5.k
(5.17a) (5. mp Ampr → Ampr = Bq .15) .k Bi
=D
27
The scalar product of two tensors is deﬁned as T : U = Tij Uij (5.16b) (5.19d)
Victor Saouma
Fracture Mechanics
..Draft
4
23 24
PRELIMINARY CONSIDERATIONS
The multiplication of a (second order) tensor by a scalar is deﬁned by: Sij = λTij (5. r Amr → Amr = B. 5 → 3 qs qr The outer product of two tensors (not necessarily of the same type or order) is a set of tensor components obtained simply by writing the components of the two tensors beside each other with no repeated indices (that is by multiplying each component of one of the tensors by every component of the other). For example
ai bj
→ ai bi
(5.k Bj
= Tij = C
i. we make two of the indices equal (or in a mixed tensor. 2 → 0 ui vj → ui vi . we make a subscript equal to the superscript).
20)
But we also have ui = T ip v p (again from Eq.4
31
Trace
The trace of a secondorder tensor.1a) in the barred system. equating these two expressions we obtain T ip − (aj aq Tjq )v p = 0 (5.23)
By extension. higher order tensors can be similarly transformed from one coordinate system to another.1 Tensors
5
5.3
29
Rotation of Axes
The rule for changing second order tensor components under rotation of axes goes as follow: ui = aj uj i = aj Tjq vq i = aj Tjq aq v p p i From Eq.26)
. ?? From Eq.24a)
(5. denoted tr T is a scalar invariant function of the tensor and is deﬁned as tr T ≡ Tii Thus it is equal to the sum of the diagonal elements in a matrix.
30
If we consider the 2D case.24b)
T
T xx T xy 0 = AT T A = T xy T yy 0 0 0 0 cos2 αTxx + sin2 αTyy + sin 2αTxy 1 = 2 (− sin 2αTxx + sin 2αTyy + 2 cos 2αTxy 0
(5. 5. Victor Saouma Fracture Mechanics (5.1. 5.25)
5. From Eq.24c)
1 2 (− sin 2αTxx + sin 2αTyy + 2 cos 2αTxy sin2 αTxx + cos α(cos αTyy − 2 sin αTxy
0
2
0 0 0
(5. using sin 2α = 2 sin α cos α and cos 2α = cos2 α − sin α.1a From Eq.24d) alternatively.1. this last equation can be rewritten as
T xx yy T xy
T
Txx cos2 θ sin2 θ 2 sin θ cos θ 2 2θ cos −2 sin θ cos θ Tyy sin θ = − sin θ cos θ cos θ sin θ cos2 θ − sin2 θ Txy
(5. ?? (5. ??
cos α sin α 0 A = − sin α cos α 0 0 0 1 T Txx Txy 0 = Txy Tyy 0 0 0 0
(5.22) (5.21) i p hence T ip = aj aq Tjq in Matrix Form [T ] = [A]T [T ][A] i p Tjq = aj aq T ip i p in Matrix Form [T ] = [A][T ][A]
T
(5.Draft
5.
34) 1 (Tii Tjj − Tij Tij ) (5.29)
and the direction ni is called principal direction of Tij .6
Principal Values and Directions of Symmetric Second Order Tensors
33 Since the two fundamental tensors in continuum mechanics are of the second order and symmetric (stress and strain).28)
If the direction is one for which vi is parallel to ni .1. a vector given by the inner product
vi = Tij nj
(5. or Tik Tkj = δij and Tik Tkj = δij
5.33)
the roots are called the principal values of Tij and IT IIT IIIT Victor Saouma = Tij = tr Tij (5.1. this can be rewritten as (5.Draft
6
PRELIMINARY CONSIDERATIONS
5. 34 For every symmetric tensor Tij deﬁned at some point in space. (T11 − λ)n1 + T12 n2 + T13 n3 = 0 T21 n1 + (T22 − λ)n2 + T23 n3 = 0 T31 n1 + T32 n2 + (T33 − λ)n3 = 0 To have a nontrivial solution (ni = 0) the determinant of the coeﬃcients must be zero. the inner product may be expressed as Tij nj = λni (5. we examine some important properties of these tensors. there is associated with each direction (speciﬁed by unit normal nj ) at that point. Since ni = δij nj . Tij − λδij  = 0 (5.31a)
35
Expansion of this determinant leads to the following characteristic equation λ3 − IT λ2 + IIT λ − IIIT = 0 (5.5
32
Inverse Tensor
An inverse tensor is simply deﬁned as follows T−1 (Tv) = v and T(T−1 v) = v (5.30) (Tij − λδij )nj = 0 which represents a system of three equations for the four unknowns ni and λ.32) (5.27)
−1 −1 alternatively T−1 T = TT−1 = I.35) = 2 = Tij  = det Tij (5.36) Fracture Mechanics
.
(σ21 . x2 and x3 faces (It should be noted that those tractions are not necesarily normal to the faces. In other words.g.
36
7
It is customary to order those roots as λ1 > λ2 > λ3
For a symmetric tensor with real components. and they can be decomposed into a normal and shear traction if need be).
37
5. σij where the 1st subscript (i) refers to the direction of outward facing normal. the three principal directions are mutually orthogonal. t1 σ11 σ12 σ13 t2 (5.38) σ = σij = σ21 σ22 σ23 = t σ31 σ32 σ33 3
41
In fact the nine rectangular components σij of σ turn out to be the three sets of three vector components (σ11 . σ33 ) which correspond to the three tractions t1 . Fig 5.1 is a second order cartesian tensor. and the second one (j) to the direction of component force. second and third invariants respectively of Tij . 5.
40 The traction vectors on planes perpendicular to the coordinate axes are particularly useful.2. σ23 ). dF = ρbdV ol.
A stress. σ22 . e.1
38
Kinetics
Force.2 Kinetics are called the ﬁrst. tdS = i tx dS + j ty dS + k tz dS (5.37)
S
S
S
S
Most authors limit the term traction to an actual bounding surface of a body. stresses are nothing else than the components of tractions (stress vector).Draft
5.
39 The surface force per unit area acting on an element dS is called traction or more accurately stress vector.2. t2 and t3 which are acting on the x1 . it requires the secondorder tensor with all nine components. σ13 ).
42
The state of stress at a point cannot be speciﬁed entirely by a single vector with three components. the stress vector at that point on any other arbitrarily inclined plane can be expressed in terms of the ﬁrst set of tractions. σ12 . gravity.2
5. (σ31 . and use the term stress vector for an imaginary interior surface (even though the state of stress is a tensor and not a vector). the principal values are also real. When the vectors acting at a point on three such mutually perpendicular planes is given. Fig.
43
Victor Saouma
Fracture Mechanics
. surface forces: are contact forces acting on the free body at its bounding surface. Traction and Stress Vectors
There are two kinds of forces in continuum mechanics
body forces: act on the elements of volume or mass inside the body. Those will be deﬁned in terms of force per unit area. electromagnetic ﬁelds. σ32 . If those values are distinct.
Draft
8
X3
PRELIMINARY CONSIDERATIONS
σ33 σ σ31 σ13 σ σ 11 ∆X2
X1
12 32
σ 23
21
∆X3
X2
σ
σ 22
∆X1
Figure 5.2: Stresses as Tensor Components
Victor Saouma
Fracture Mechanics
.1: Stress Components on an Inﬁnitesimal Element
X3
X3
V3
σ33 t3 σ31 σ13 σ σ 11 t1
12
σ 32 σ 23 t2 σ 22
X2
V1 X1
V
V2 X2
σ
21
(Components of a vector are scalars)
X 1 Stresses as components of a traction vector
(Components of a tensor of order 2 are vectors)
Figure 5.
and will be obtained without any assumption of equilibrium and it will apply
X2
t
1
B
∆
*
1
S
n
t
*
3
∆
S3
* tn ∆ S
O
C
h N
A
X1
t 2 ∆ S2
*
X3
ρb
*
in ﬂuid dynamics as well as in solid mechanics.2.39)
We have thus established that the nine components σij are components of the second order tensor.2
Traction on an Arbitrary Plane.3. 5.Draft
5. t1 .
45 This equation is a vector equation. and the corresponding algebraic equations for the components of tn are
tn1 tn2 tn3 Indicial notation tni dyadic notation tn
46
∆V
*
Figure 5. Cauchy’s stress tensor. This will be done through the socalled Cauchy’s tetrahedron shown in Fig.2 Kinetics
9
5.40) Fracture Mechanics
Victor Saouma
.3: Cauchy’s Tetrahedron
= = = = =
σ11 n1 + σ21 n2 + σ31 n3 σ12 n1 + σ22 n2 + σ32 n3 σ13 n1 + σ23 n2 + σ33 n3 σji nj n·σ = σT ·n
(5. t2 and t3 . Cauchy’s Stress Tensor
44 Let us now consider the problem of determining the traction acting on the surface of an oblique plane (characterized by its normal n) in terms of the known tractions normal to the three principal axis. Example 51: Stress Vectors if the stress tensor at point P is given by
t1 7 −5 0 t2 σ = −5 3 1 = t 0 1 2 3
(5.
0).52) Fracture Mechanics IIσ = −(σ(1) σ(2) + σ(2) σ(3) + σ(3) σ(1) ) (5.3
Invariants
47 The principal stresses are physical quantities.51)
Victor Saouma
.48) (5. 6). 0.2. 48 When the determinant in the characteristic Equation is expanded.47) (5. 0) and C(0. Solution: The vector normal to the plane can be found by taking the cross products of vectors AB and AC: e1 e2 e3 −4 2 0 −4 0 6 (5.42)
7 −5 0 −5 3 1 = 0 1 2
−9 7
5 7
10 7
(5. They are therefore invariants of the stress state.Draft
10 N = AB×AC =
PRELIMINARY CONSIDERATIONS
We seek to determine the traction (or stress vector) t passing through P and parallel to the plane ABC where A(4. 0. whose values do not depend on the coordinate system in which the components of the stress were initially given. the cubic equation takes the form
λ3 − Iσ λ2 − IIσ λ − IIIσ = 0
(5.50) (5.41a) (5.44)
where the symbols Iσ .41b)
= 12e1 + 24e2 + 8e3 The unit normal of N is given by n= Hence the stress vector (traction) will be
3 7 6 7 2 7
6 2 3 e1 + e2 + e3 7 7 7
(5.43)
and thus t = − 9 e1 + 5 e2 + 7 7
10 7 e3
5.46) (5. B(0. IIσ and IIIσ denote the following scalar expressions in the stress components: Iσ = σ11 + σ22 + σ33 = σii = tr σ IIσ = −(σ11 σ22 + σ22 σ33 + σ33 σ11 ) + 1 1 2 1 (σij σij − σii σjj ) = σij σij − Iσ = 2 2 2 1 2 (σ : σ − Iσ ) = 2 1 IIIσ = detσ = eijk epqr σip σjq σkr 6
2 σ23
49
(5. those invariants can be simpliﬁed into Iσ = σ(1) + σ(2) + σ(3) IIIσ = σ(1) σ(2) σ(3) (5. 2.45)
2 + σ31
+
2 σ12
(5.49)
In terms of the principal stresses.
58)
5.57)
52
For the 2D plane stress case we rewrite Eq.53)
then the stress tensor can be written as the sum of two tensors: Hydrostatic stress in which each normal stress is equal to −p and the shear stresses are zero. σ11 − σ σ12 σ13 = σ21 σ22 − σ σ23 σ31 σ32 σ33 − σ
σdev
(5.25
σ xx yy σ xy σxx cos2 α sin2 α 2 sin α cos α 2 2α = cos −2 sin α cos α σyy sin α − sin α cos α cos α sin α cos2 α − sin2 α σxy
σ
(5.23.4
50
Spherical and Deviatoric Stress Tensors
If we let σ denote the mean normal stress p 1 1 1 σ = −p = (σ11 + σ22 + σ33 ) = σii = tr σ 3 3 3 (5.Draft
5. 5.2.5
51
Stress Transformation
From Eq.2 Kinetics
11
5. −p 0 0 (5.6
Polar Coordinates
It is often necessary to express cartesian stresses in terms of polar stresses and vice versa.59b) (5.2. 5.55)
5. This can be done through the following relationships σxx = σrr cos2 θ + σθθ sin2 θ − σrθ sin 2θ σyy = σrr sin θ + σθθ cos θ + σrθ sin 2θ σxy = (σrr − σθθ ) sin θ cos θ + σrθ (cos θ − sin θ)
2 2 2 2
(5.54) −p 0 σhyd = −pI = 0 0 0 −p Deviatoric Stress: which causes the change in shape.59a) (5.59c)
and σrr = σxx + σyy 2 1− a2 r2 + σxx − σyy 2 1+ 3a4 4a2 − 2 r4 r cos 2θ
Victor Saouma
Fracture Mechanics
.2. the stress transformation for the second order stress tensor is given σ ip = aj aq σjq in Matrix Form [σ] = [A]T [σ][A] i p σjq = aj aq σ ip i p in Matrix Form [σ] = [A][σ][A]
T
by (5.22 and 5. The hydrostatic stress produces volume change without change in shape in an isotropic medium.56) (5.
60a) 1+ 3a4 r4 cos 2θ (5.62c)
54
The Eulerian ﬁnite strain tensor can be written as
∗ Eij =
1 2
∂uj ∂uk ∂uk ∂ui + − ∂xj ∂xi ∂xi ∂xj
or E∗ =
1 (u∇x + ∇x u − ∇x u·u∇x ) 2 K+Kc K c ·K
(5.62a) (5.1
53
Kinematic
Strain Tensors
The Lagrangian ﬁnite strain tensor can be written as Eij = 1 2 ∂uj ∂uk ∂uk ∂ui + + ∂Xj ∂Xi ∂Xi ∂Xj 1 or E = (u∇X + ∇X u + ∇X u·u∇X ) 2 J+Jc J c ·J
(5.62b) (5.60c)
σxx − σyy 2
σxx − σyy 2
5.64c)
56
Alternatively these equations may be expanded as
Victor Saouma
Fracture Mechanics
.61)
or: E11 = ∂u1 1 + ∂X1 2 1 ∂u1 + E12 = 2 ∂X2 ··· = ··· ∂u1 2 ∂u2 2 + + ∂X1 ∂X1 ∂u2 1 ∂u1 ∂u1 + + ∂X1 2 ∂X1 ∂X2 ∂u3 2 ∂X1 ∂u2 ∂u2 ∂u3 ∂u3 + ∂X1 ∂X2 ∂X1 ∂X2 (5.63)
55
Expanding ∂u1 1 − ∂x1 2 1 ∂u1 ∗ + E12 = 2 ∂x2 ··· = ···
∗ E11 =
∂u1 2 ∂u2 2 + + ∂x1 ∂x1 ∂u2 1 ∂u1 ∂u1 − + ∂x1 2 ∂x1 ∂x2
∂u3 2 ∂x1 ∂u2 ∂u2 ∂u3 ∂u3 + ∂x1 ∂x2 ∂x1 ∂x2
(5.64b) (5.3.3
5.60b) sin 2θ + σxy 1 − 3a4 2a2 + 2 r4 r cos 2θ (5.Draft
12 +σxy 1 + σθθ = σxx + σyy 2 −σxy 1 + σrθ = − 3a4 r4 3a4 4a2 − 2 r4 r 1+ a2 r2 sin 2θ − sin 2θ 1− 3a4 2a2 + 2 r4 r
PRELIMINARY CONSIDERATIONS
(5.64a) (5.
70)
∂w 1 ∂u 2 ∂v 2 ∂w 2 + + + ∂z 2 ∂z ∂z ∂z ∂u ∂u ∂u ∂v ∂v ∂w ∂w 1 ∂v + + + + 2 ∂x ∂y ∂x ∂y ∂x ∂y ∂x ∂y 1 ∂w ∂u ∂u ∂u ∂v ∂v ∂w ∂w + + + + 2 ∂x ∂z ∂x ∂z ∂x ∂z ∂x ∂z 1 ∂w ∂v ∂u ∂u ∂v ∂v ∂w ∂w + + + + 2 ∂y ∂z ∂y ∂z ∂y ∂z ∂y ∂z 1 (ui. and the resulting strains are referred to as the Almansi strain which is the preferred one in ﬂuid mechanics. If the strains are given. The strains have been expressed in terms of the coordinates x.66) (5.j + uj. in the Lagrangian coordinate which is the preferred one in structural mechanics. Eulerian coordinates x . In most cases the deformations are small enough for the quadratic term to be dropped.i + uk.e.67) (5. we note that:
(5. y . 5. 6. the resulting equations reduce to
Victor Saouma
Fracture Mechanics
. i. We deﬁne the engineering shear strain as γij = 2εij (i = j) (5. Alternatively we could have expressed ds 2 − ds2 in terms of coordinates in the deformed state.72)
2.j ) 2
εij = From this equation. z .Draft
5. 4. then these straindisplacements provide a system of (6) nonlinear partial diﬀerential equation in terms of the unknown displacements (3).e. εik is the GreenLagrange strain tensor. i. y.65)
2
+
+
(5.71)
1.i uk.3 Kinematic ∂u 1 + ∂x 2 ∂v 1 + ∂y 2 ∂u ∂x ∂u ∂y
2
13 ∂v ∂x ∂v ∂y
2
εxx = εyy = εzz = εxy = εxz = εyz = or
+
2
+
2
∂w ∂x ∂w ∂y
2
(5. z in the undeformed state. 3.69) (5.68) (5.
the strain is often expressed through the linear operator L ε = Lu (5. this would yield 9 equations in total. j = 1 and l = 2): ∂ 2 γ12 ∂ 2 ε11 ∂ 2 ε22 2 + ∂x2 = ∂x ∂x ∂x2 1 2 1 Victor Saouma (5.Draft
14 εxx = εyy = εzz = γxy = γxz = γyz = or εij = which is called the Cauchy strain
57
PRELIMINARY CONSIDERATIONS ∂u ∂x ∂v ∂y ∂w ∂z ∂u ∂v + ∂x ∂y ∂w ∂u + ∂x ∂z ∂w ∂v + ∂y ∂z
(5. but due to symmetry. there are 6 independent ones) for determining (upon integration) three unknowns displacements ui .76) (5. It can be shown (through appropriate successive diﬀerentiation of Eq. and there must be some linear relations between the strains.77) (5.75) (5.i ) 2
(5.74) (5. In 2D.73) (5.80)
or
εxx εyy ε
∂ ∂x 0 0 zz = ∂ εxy ∂y ∂ εxz ∂z
0
∂ ∂y
0
∂ ∂x
0
∂ ∂z L
εyz ε
ux uy 0 uz ∂ ∂x
∂ ∂z ∂ ∂y u
0 0
(5.2
Compatibility Equation
If εij = 1 (ui.81)
0
5.82)
In 3D.78)
1 (ui.79)
In ﬁnite element.k + uk.i ) then we have six diﬀerential equations (in 3D the strain tensor has a 2 total of 9 terms.83) Fracture Mechanics
. however only six are distinct.j + uj. Hence the system is overdetermined.3. this results in (by setting i = 2. ∂xj ∂xj ∂xi ∂xk ∂xi ∂xj ∂xj ∂xk (5. ??) that the compatibility relation for strain reduces to: ∂ 2 εjj ∂ 2 εjk ∂ 2 εij ∂ 2 εik + − − = 0.
For that we need to wait for the next chapter where constitututive laws relating stress and strain will be introduced. In its most general form. and yet we will not have enough equations to determine unknown tensor ﬁeld.. Only with constitutive equations and boundary and initial conditions would we be able to obtain a well deﬁned mathematical problem to solve for the stress and deformation distribution or the displacement or velocity ﬁelds.86c) Fracture Mechanics
t t
−3 −1
. and several other tensors which describe strain at a point. These diﬀerential equations of balance will be derived from integral forms of the equation of balance expressing the fundamental postulates of continuum mechanics.. momentum and energy. Piola Kirchoﬀ). . it yields: ∂σ22 2 ∂ 2 σ22 ∂ 2 σ11 ∂ 2 σ21 ∂ 2 σ11 −ν + −ν = 2 (1 + ν) ∂x2 ∂x2 ∂x2 ∂x2 ∂x1 ∂x2 2 2 1 1
15
(5.
63 Hence.4.) a.4 Fundamental Laws of Continuum Mechanics (recall that 2ε12 = γ12 ). The dimensions of various quantities are given by
dim(a) = dim(AL−3 ) dim(α) = dim(AL dim(A) = dim(AL Victor Saouma
−2 −1
(5.
61
5.
59
We have also obtained only one diﬀerential equation.86a) ) ) (5.Draft
5. In general. we will derive additional diﬀerential equations governing the way stress and deformation vary at a point and with time. When he compatibility equation is written in term of the stresses. and to modify A which is the quantity of interest. those tensors will vary from point to point and represent a tensor ﬁeld. we read the previous equation as: The input quantity (provided by the right hand side) is equal to what is lost across the boundary.4
58
5. A is the rate of volumetric density of what is provided from the outside.
60 In this chapter. linear momentum. energy.85)
Rate of variation
Exchange by Diﬀusion
where A is the volumetric density of the quantity of interest (mass.
In this chapter we shall derive diﬀerential equations expressing locally the conservation of mass. They will apply to any continuous medium.1
Conservation Laws
62 Conservation laws constitute a fundamental component of classical physics.86b) (5. and α is the rate of surface density of what is lost through the surface S of V and will be a function of the normal to the surface n. that was the compatibility equation. such a law may be expressed as
d dt
V
AdV +
S
αdS
=
V
AdV
Source
(5.84)
Fundamental Laws of Continuum Mechanics
We have thus far studied the stress tensors (Cauchy. A conservation law establishes a balance of a scalar or tensorial quantity in volume V bounded by a surface S.
5.
5. and energy.
68
We can generalize this deﬁnition and deﬁne the following ﬂuxes per unit area through dS:
Victor Saouma
Fracture Mechanics
. we deﬁne the volume ﬂux as Volume Flux =
S
v·ndS =
S
vj nj dS
(5.
The ﬂux across a surface can be graphically deﬁned through the consideration of an imaginary surface ﬁxed in space with continuous “medium” ﬂowing through it. momentum.4: Flux Through Area dS then the ﬂow is in the negative direction). Hence. we need to deﬁne the concept of ﬂux across a bounding surface. constitute what is commonly known as the fundamental laws of continuum mechanics. equilibrium and symmetry of the stress tensor.87)
where the last form is for rectangular cartesian components. the resulting diﬀerential equations will provide additional interesting relation with regard to the imcompressibiltiy of solids (important in classical hydrodynamics and plasticity theories). and the ﬁrst law of thermodynamics. If we assign a positive side to the surface. and take n in the positive sense.
65 The enunciation of the preceding three conservation laws plus the second law of thermodynamics. Fig.4.4 (If v·n is negative.2
Fluxes
66 Prior to the enunciation of the ﬁrst conservation law.Draft
16
64
PRELIMINARY CONSIDERATIONS
Hence this chapter will apply the previous conservation law to mass.
67
v
vdt
n dS
vn dt
Figure 5. then the volume of “material” ﬂowing through the inﬁnitesimal surface area dS in time dt is equal to the volume of the cylinder with base dS and slant height vdt parallel to the velocity vector v.
4 Fundamental Laws of Continuum Mechanics
17
Mass Flux Momentum Flux Kinetic Energy Flux Heat ﬂux Electric ﬂux
=
S
ρv·ndS =
S
ρvj nj dS
S
(5. provided Newton’s Third Law applies.4. Equation of Motion
70 The momentum principle states that the time rate of change of the total momentum of a given set of particles equals the vector sum of all external forces acting on the particles of the set.91) (5.88) (5. so that the density in the neighborhood of each material particle remains constant as it moves.4
Linear Momentum Principle.3
Conservation of Mass. This equation shows that the divergence of the velocity vector ﬁeld equals (−1/ρ)(dρ/dt) and measures the rate of ﬂow of material away from the particle and is equal to the unit rate of decrease of density ρ in the neighborhood of the particle.96b)
V
dvi ∂Tij + ρbi − ρ dV ∂xj dt
= 0
or for an arbitrary volume dv dvi ∂Tij or ∇T + ρb = ρ + ρbi = ρ ∂xj dt dt Victor Saouma (5.92)
=
S
ρv(v·n)dS = 1 2 ρv (v·n)dS = S2 q·ndS =
S S
ρvk vj nj dS 1 ρvi vi vj nj dS S2
= = =
qj nj dS Jj nj dS
J·ndS =
S S
5.97) Fracture Mechanics
.4.96a) (5.Draft
5.89) (5. The continuum form of this principle is a basic postulate of continuum mechanics. then the continuity equation takes the simpler form
69
∂vi = 0 or ∇·v = 0 ∂xi this is the condition of incompressibility
(5. If the material is incompressible.95)
Then we substitute ti = Tij nj and apply the divergence theorm to obtain ∂Tij + ρbi dV ∂xj =
V
ρ
V
dvi dV dt
(5.93)
The vector form is independent of any choice of coordinates.
tdS +
S V
ρbdV =
d dt
ρvdV
V
(5.94)
5. Continuity Equation
∂vi dρ dρ +ρ + ρ∇·v = 0 = 0 or dt ∂xi dt (5.90) (5.
j + ρbi = 0 where ρ is the density.
Victor Saouma
Fracture Mechanics
. Writing the summation of forces. or the linear momentum principle.5
Moment of Momentum Principle
73 The moment of momentum principle states that the time rate of change of the total moment of momentum of a given set of particles equals the vector sum of the moments of all external forces acting on the particles of the set.5 with the assumption of equilibrium (via Newton’s second law) considering an inﬁnitesimal element of dimensions dx1 × dx2 × dx3 . bi is the body force (including inertia). will yield
72
Tij.4. or more simply equilibrium equation.Draft
18
71
PRELIMINARY CONSIDERATIONS
which is Cauchy’s (ﬁrst) equation of motion. 5.
σ σyy δ yy d y + δy
(5.98a)
We note that these equations could also have been derived from the free body diagram shown in Fig.99)
+ τyx
δ τ yx y d δy σxx + δ σxx d x δx
dy
σ xx τ xy τ yx σyy dx + τxy
δ τ xy x d δx
Figure 5. Cartesian Coordinates
5. this equation yields: ∂T11 ∂T12 ∂T13 + + + ρb1 = 0 ∂x1 ∂x2 ∂x3 ∂T21 ∂T22 ∂T23 + + + ρb2 = 0 ∂x1 ∂x2 ∂x3 ∂T31 ∂T32 ∂T33 + + + ρb3 = 0 ∂x1 ∂x2 ∂x3
(5.5: Equilibrium of Stresses. When expanded in 3D.
5. the ﬁrst principle states that the time rate of change of the kinetic plus the internal energy is equal to the sum of the rate of work plus all other energies supplied to. hence if we really need to evaluate this quantity.j dV + ρbi vi dV = ρvi dt V V V Applying the divergence theorem. If mechanical quantities only are considered.101) vi Tji.4. First Principle of Thermodynamics
The ﬁrst principle of thermodynamics relates the work done on a (closed) system and the heat transfer into the system to the change in energy of the system. and the SI unit is the Joule. in the absence of distributed couples (this theory of Cosserat will not be covered in this course) we postulate the same principle for a continuum as (r×t)dS +
S V
(r×ρb)dV =
d dt
(r×ρv)dV
V
(5.100)
5.103)
where u is the internal energy per unit mass or speciﬁc internal energy.97. electromagnetic. the principle of conservation of energy for the continuum may be derived directly from the equation of motion given by Eq.). by heat transfer through the boundary. 5.
76
dU dW dK + = +Q dt dt dt
(5. we need to have a reference value for which U will be null.
77 If both mechanical and non mechanical energies are to be considered. 78 For a thermomechanical continuum. We shall assume that the only energy transfers to the system are by mechanical work done on the system by surface traction and body forces.5 Constitutive Equations
19
Thus. similarly dim u = L2 T −2 with the SI unit of Joule/Kg.6
75
Conservation of Energy.102)
this equation relates the time rate of change of total mechanical energy of the continuum on the left side to the rate of work done by the surface and body forces on the right hand side. This is accomplished by taking the integral over the volume V of the scalar product between Eq. 5. or removed from the continuum per unit time (heat.Draft
74
5. it is customary to express the time rate of change of internal energy by the integral expression
d dU = dt dt
ρudV
V
(5.5
Constitutive Equations
ceiinosssttuu Hooke. etc. We note that U appears only as a diﬀerential in the ﬁrst principle. chemical. 1676
Victor Saouma
Fracture Mechanics
. dvi dV (5. The dimension of U is one of energy dim U = M L2 T −2 .97 and the velocity vi .
due to the symmetry of both σ and ε.110b)
85
Hooke’s law for isotropic material in terms of engineering constants becomes σij εij = = E ν E ν δij εkk Iε εij + or σ = ε+ 1+ν 1 − 2ν 1+ν 1 − 2ν ν 1+ν ν 1+ν σij − δij σkk or ε = σ − Iσ E E E E (5. k.108) (5. 2.105)
In terms of Lame’s constants.110a) (5.109)
λ =
84
Similarly in the case of pure shear in the x1 x3 and x2 x3 planes. 3. there are at most 36 9(9−1) distinct elastic terms.104)
The (fourth order) tensor of elastic constants Dijkl has 81 (34 ) components however. Hooke’s Law for an isotropic body is written as Tij = λδij Ekk + 2µEij 1 λ δij Tkk Tij − Eij = 2µ 3λ + 2µ or or T = λIE + 2µE −λ 1 IT + T E= 2µ(3λ + 2µ) 2µ (5. 2
80
For the purpose of writing Hooke’s Law. m = 1.111) (5.107)
83
In terms of engineering constants: 1 E = λ+µ λ .112)
86
When the strain equation is expanded in 3D cartesian coordinates it would yield:
1 = E
εxx εyy εzz γxy (2εxy ) γyz (2ε ) Victor Saouma yz γzx (2εzx )
1 −ν −ν 0 0 0 −ν 1 −ν 0 0 0 −ν −ν 1 0 0 0 0 0 0 1+ν 0 0 0 0 0 0 1+ν 0 0 0 0 0 0 1+ν
σxx σyy σ zz (5. j. we have σ21 = σ12 = τ all other σij = 0 τ 2ε12 = G and the µ is equal to the shear modulus G.Draft
20 Ut tensio sic vis Hooke. l = 1. 6
(5.113) τxy τyz Fracture Mechanics
τzx
. 3 (5.ν = µ(3λ + 2µ) 2(λ + µ) E νE . 5. 2.µ = G = (1 + ν)(1 − 2ν) 2(1 + ν) (5. (5. the double indexed system is often replaced by a simple indexed system with a range of six:
81
62 =36
σk = Dkm εm
82
k.106) (5. 4. 1678
79
PRELIMINARY CONSIDERATIONS
The Generalized Hooke’s Law can be written as: σij = Dijkl εkl i.
5.116)
where E is the Young’s modulus in the plane of isotropy and E the one in the plane normal to it. replacing into Eq.5.5 Constitutive Equations If we invert this equation. µ corresponding to the shear moduli for the plane of isotropy and any plane normal to it. 5.1
90
For problems involving a long body in the z direction with no variation in load or geometry. ν corresponding to the transverse contraction in the plane of isotropy when tension is applied normal to the plane. E a44 = − 1 µ
(5. we can express the stressstrain relation in tems of εxx εyy εzz γxy γyz γxz = = = = = = a11 σxx + a12 σyy + a13 σzz a12 σxx + a11 σyy + a13 σzz a13 (σxx + σyy ) + a33 σzz 2(a11 − a12 )τxy a44 τxy a44 τxz ν .114 we obtain
σxx σ E yy = σzz (1 + ν)(1 − 2ν)
τxy
(1 − ν) ν ν (1 − ν) ν ν 0 0
0 0 0
1−2ν 2
εxx εyy γ
xy
(5. E
a13 = −
(5.114)
5. Plane Strain
5.5. and µ is shear moduli for the plane of isotropy.2
89
Special 2D Cases
Often times one can make simplifying assumptions to reduce a 3D problem into a 2D one. E a33 = − 1 .115)
and a11 =
1 .117)
Victor Saouma
Fracture Mechanics
.5. we obtain
σxx σyy σ zz = τxy τyz
21
1−ν ν ν E ν 1−ν ν (1+ν)(1−2ν) ν ν 1−ν 0
τzx
εxx 0 εyy εzz γ (2ε ) 1 0 0 xy xy G 0 1 0 γyz (2εyz )
0 0 1
γzx (2εzx )
(5. then εzz = γyz = γxz = τxz = τyz = 0.Draft
87
5. Thus.2. E
a12 = −
ν .1
Transversly Isotropic Case
88
For transversely isotropic. ν corresponds to the transverse contraction in the plane of isotropy when tension is applied in the plane.
119)
γrz
5.Draft
22 5.120a) (5. ∂x2 2
σ22 =
∂2Φ .114 we obtain:
σxx
εxx 1 ν 0 1 σyy = 0 εyy ν 1 1 − ν2 τ 0 0 1−ν γxy xy 2 1 ν(εxx + εyy ) εzz = − 1−ν
(5.2. substituting into Eq.118d)
92
The constitutive relation is again analogous to 3D/plane strain
σrr σ E zz = σθθ (1 + ν)(1 − 2ν)
τrz
1−ν ν ν ν 1−ν ν ν ν 1−ν ν ν 1−ν 0 0 0
0 0 0 0
1−2ν 2
εrr ε zz εθθ
(5.5.118c) (5.
2
(5. Airy has shown that we can deﬁne such a function Φ(x) such that:
σ11 =
∂2Φ .2.120b)
5. 5. we can use a polar coordinate sytem and εrr = εθθ = εzz = εrz = ∂u ∂r u r ∂w ∂z ∂u ∂w + ∂z ∂r (5.118b) (5.121)
95
In polar coordinates: σrr =
1 ∂Φ r ∂r
+
1 ∂2Φ r 2 ∂θ 2
σθθ =
∂2Φ ∂r 2
∂ σrθ = − ∂r
1 ∂Φ r ∂θ
(5.5.3
Plane Stress
93 If the longitudinal dimension in z direction is much smaller than in the x and y directions. then τyz = τxz = σzz = γxz = γyz = 0 throughout the thickness.6
Airy Stress Function
94 In elasticity problems we seek a function in terms of the spatial coordinates which can satisfy both the equilibrium and the compatibility equations.118a) (5.2
91
PRELIMINARY CONSIDERATIONS
Axisymmetry
In solids of revolution. Again. ∂x2 1
∂ Φ σ12 = − ∂x1 ∂x2 .122) Fracture Mechanics
Victor Saouma
.
then from Eq.127)
Victor Saouma
Fracture Mechanics
. and stresses at the tip of a crack (Westergaard). 100
First we deﬁne the complex number z as: z = x1 + ix2 = reiθ (5.126b)
If f (z) = α + iβ where α and β are real functions of x1 and x2 . A solution to Laplace’s equation is referred to as a harmonic function. x1 and x2 are the cartesian coordinates.123)
where ∇2 is the Laplacian operator.126b we have:
102
∂f (z) ∂x1 ∂f (z) ∂x2
= =
∂α ∂x1 ∂α ∂x2
∂β + i ∂x1 = f (z) ∂β + i ∂x2 = if (z)
i
∂β ∂α +i ∂x1 ∂x1
=
∂α ∂β +i ∂x2 ∂x2
(5.98a) that the equilibrium equation is automatically satisﬁed. the satisfaction of the compatibility equation (Eq.Draft
5.124)
98 Thus. f (z) one which derivatives depend only on z. 5.7
Complex Variables
99 As will be shown in the next chapter.
97
In polar coordinates ∇2 ∇2 Φ = 1 ∂2 1 ∂ ∂2 + 2 2 + ∂r 2 r ∂r r ∂θ 1 ∂2Φ ∂ 2 Φ 1 ∂Φ + 2 2 + ∂r 2 r ∂r r ∂θ (5.126a) (5. the Airy stress function will enable us to solve elasticity problems provided we can come up with the right choice for Φ which satisﬁes the natural boundary conditions (stresses).126a and 5. and f (z) is analytic. however we need to extend Airy stress functions to complex variables in order to analyze: 1) stresses around an elliptical hole (Inglis).7 Complex Variables
23
It can be shown (by direct substitution of these equations in Eq. 5.
We further deﬁne an analytic function. and r and θ are the polar coordinates.
5.84) further requires that Φ must be such that
∇2 ∇2 Φ ≡
∂2 ∂2 + 2 ∂x2 ∂x2 1
∂2Φ ∂2Φ + ∂x2 ∂x2 1 2
=
∂4Φ ∂4Φ ∂4Φ +2 2 2 + =0 ∂x4 ∂x1 ∂x2 ∂x4 1 2
(5. 5.125)
where i =
101
√ −1. Applying the chain rule ∂ f (z) = ∂x1 ∂ f (z) = ∂x2 ∂z ∂ ∂z f (z) = f (z) = f (z) ∂z ∂x1 ∂x1 ∂z ∂ ∂z f (z) = f (z) = if (z) ∂z ∂x2 ∂x2 (5. we can use Airy stress function with real variables to determine the stress ﬁeld around a circular hole.
96 For linear elastic isotropic materials.
∂x1 ∂x2 ∂α ∂β =− ∂x2 ∂x1 (5.135)
and by separation of real and imaginary parts we can then solve for σ22 − σ11 & σ12 . then its conjugate function is deﬁned as: ¯z f (¯) = α − iβ (5. Fracture Mechanics
Victor Saouma
.
107
If f (z) = α + iβ and both α and β are real. 5.133)
109
Substituting Eq. we can determine the stresses σ11 + σ22 = 4Reψ (z) z σ22 − σ11 + 2iσ12 = 2[¯ψ (z) + χ (z)] (5. Hence we can rewrite Eq. and α and β are conjugate harmonic functions.
110
Displacements can be similarly obtained. and then add them up we obtain
104
∂2α ∂2α + =0 ∂x2 ∂x2 1 2 which is Laplace’s equation. then with respect to x2 .121. ﬁrst with respect to x1 .131)
provided that both ψ(z) (psi) and χ(z) (chi) are harmonic (i.134) (5.132)
Note that conjugate functions should not be confused with the conjugate harmonic functions.
5.129)
Similarly we can have: ∇2 (β) = 0 (5.7. 5.128)
If we diﬀerentiate those two equation.130)
Hence both the real (α) and the immaginary part (β) of an analytic function will separately provide solution to Laplace’s equation.
105
or
∇2 (α) = 0
(5.133 into Eq.e ∇2 (ψ) = ∇2 (χ) = 0) analytic functions of x1 and x2 .131 as:
108
Φ = Re[¯ψ(z) + χ(z)] z
(5.1
106
Complex Airy Stress Functions
It can be shown that any stress function can be expressed as Φ = Re[(x1 − ix2 )ψ(z) + χ(z)] (5.Draft
24
103
PRELIMINARY CONSIDERATIONS
Equating the real and imaginary parts yields the CauchyRiemann equations: ∂β ∂α = . ψ and χ are often refered to as the KolonovMuskhelishvili complex potentials. 5.
Similarly lines of constant β represent confocal hyperebolae which intersect the ellipse at right angle.
113
Recalling that cosh α = 1 α (e + e−α ) 2 1 α (e − e−α ) sinh α = 2 eiβ = cos β + i sin β (5.143)
= 2e
2iθ
[¯ψ (z) + ψ (z)] z
Individual stresses are obtained by separating the real from the immaginary components.Draft
5.138a)
114
Separating reals from immaginary parts we obtain x1 = c cosh α cos β = x2 c cos β(eα + e−α ) 2 c = c sinh α sin β = sin β(eα − e−α ) 2 (5.134:
117
¯ z σαα + σββ = 2[ψ (z) + ψ (¯)] = 4Reψ (z) σββ − σαα + 2iσαβ
118
(5.137b) (5.
We next seek to solve for x1 and x2 in terms of α and β.139) (5. 5.
116
In terms of complex potentials.142) (5.136) where c is a constant.141)
Thus for a constant value of α this represents the equation of an ellipse in the x1 − x2 plane.6.8
112
5. we obtain x2 x2 1 2 + = c2 cosh2 α sinh2 α (5. Fig. Fracture Mechanics
Victor Saouma
. The relationship between z and p is given by z = c cosh p (5.8 Curvilinear Coordinates
25
Curvilinear Coordinates
111 A variable z = x1 + ix2 in the cartesian coordinate system can be expressed as p = α + iβ where α and β are coordinates in a curvilinear system. it can be shown that the stresses are given in an analogous way as in Eq.137a) (5. 5. 5.140)
115
If we eliminate β from those equation.137c)
we substitute those equations into Eq.136 c α+iβ e + e−α−iβ x1 + ix2 = 2 c α = e cos β + ieα sin β + e−α cos β − ie−α sin β 2 = cos β (eα + e−α ) +i sin β (eα − e−α )
2 cosh α 2 sinh α
= c(cosh α cos β + i sinh α sin β)
(5.
ε1 ε 2 ε
3
ε4 ε5
=
a11 a12 a13 a11 a13 a33
2(a11 − a12 ) a44 a44
ε6
σ1 σ2 σ3 σ4 σ5 σ6
(5. In the most general case this would yield 36 independent constants aij . If it has three mutually orthogonal planes of elastic symmetry.6: Curvilinear Coordinates
5.9
119
Basic Equations of Anisotropic Elasticity
εi = aij σj (5. If the the material 2 has one plane of elastic symmetry. then we would say that it is orthogonally anisotropic or orthotropic. and we will have a16 = a26 = a36 = a45 = 0. Fig.7. 5. (Lekhnitskii 1981).104 is
where the indices i and j go from 1 to 6.146)
Victor Saouma
Fracture Mechanics
.Draft
26
PRELIMINARY CONSIDERATIONS
Figure 5. however by virtue of symmetry (aij = aji ) this reduces to 21 (7)(6) . 5. then there would 13 independent constants.144)
An alternate form of Eq. thus there will be 9 independent constants.145)
121
If the material is transversally isotropic then it will have 5 independent constants.
120
ε1 ε 2 ε
3
ε4 ε5
=
a11 a12 a13 a22 a23 a33 a44 a55 a66
ε6
σ1 σ2 σ3 σ4 σ5 σ6
(5.
147b) (5. only 9 are independent because the following relations (5.1: Number of Elastic Constants for Diﬀerent Materials
123
In terms of engineering constants for an orthotropic solid we would have εx = εy = εz = γyz = γxz = γxy = 1 ν21 ν31 σx − σy − σz E1 E2 E3 ν12 1 ν32 − σx + σy − σz E1 E2 E3 ν13 ν23 1 − σx − σy + σz E1 E2 E3 1 τyz µ23 1 τxz µ13 1 τxy µ12 E1 ν21 = E2 ν12 (5. 3D 2D 36 9 20 9 12 5 12 5 12 5 Number of Indep. of the 12 elastic constants. Coeﬀ.Draft
5.148a) Fracture Mechanics
Victor Saouma
.147f)
however.147a) (5. 3D 2D 21 6 13 6 9 4 5 4 2 2
Table 5.147c) (5. Class of Material General Anisotropy One plane of Symmetry Orthotropic Transversely Isotropic isotropic Number of Non Zero Coeﬀ.7: Transversly Isotropic Material
122
The total number of coeﬃcients for diﬀerent materials is summarized in Table 5.147d) (5.1.147e) (5.9 Basic Equations of Anisotropic Elasticity
27
Figure 5.
For transversely isotropic solids in 3D.Draft
28
PRELIMINARY CONSIDERATIONS E2 ν32 = E3 ν23 E3 ν13 = E1 ν31 (5. Denoting by E the Young’s modulus in the plane of isotropy and E the one in the plane normal to it.).150)
126 Thus we have ﬁve elastic constants.152)
129 Note that after transformation. Transversely isotropic have through every point a plane in which all directions are equivalent with respect to elastic properties (such as in laminates. a transversely isotropic material will not have anymore 5 nonzero coeﬃcients. E a33 = − 1 . but 9 in (3D).8 then the compliance matrix is given by [C ] = [Γ]T [C][Γ] (5. E a13 = − ν . ν corresponding to the transverse contraction in the plane of isotropy when tension is applied normal to the plane. and z in terms of the principal elastic constants (as opposed to constants in equations for an arbitrary system of coordinates). quartz.148c)
124 Note that the preceding equations are written for the principal directions of elasticity.149e) (5. a number of others are transversely isotropic. 5.9.149c) (5. Fig. roller compacted concrete.. we would have ν corresponds to the transverse contraction in the plane of isotropy when tension is applied in the plane. 125 Whereas very few natural or manmade materials are truly orthotropic (certain crystals as topaz are). E a12 = − ν . and µ is shear moduli for the plane of isotropy. we have
εx = a11 σx + a12 σy + a13 σz εy = a12 σx + a11 σy + a13 σz εz = a13 (σx + σy ) + a33 σz γxy = 2(a11 − a12 )τxy γyz = a44 τxy γxz = a44 τxz and a11 = 1 .149d) (5.149b) (5.149f)
(5. x.
5.1
127
Coordinate Transformations
If the elastic constants are to be determined for an arbitrary orientation 12.148b) (5.151)
128
where [Γ] is the usual second order tensor transformation for rotation in a plane. shist.. E a44 = − 1 µ
(5.
Victor Saouma
Fracture Mechanics
. µ corresponding to the shear moduli for the plane of isotropy and any plane normal to it. etc. y. cos2 θ sin2 θ 2 sin θ cos θ 2 cos2 θ −2 sin θ cos θ [Γ] = sin θ 2 θ − sin2 θ − sin θ cos θ cos θ sin θ cos
(5.149a) (5.
9. Eq.155)
134
For isotropic material this equation reduces to: ∂4Φ ∂4Φ ∂4Φ +2 2 2 + =0 ∂x4 ∂x ∂y ∂y 4 (5.153) bij = aij − a33
5.156)
Victor Saouma
Fracture Mechanics
.131.9 Basic Equations of Anisotropic Elasticity
29
Figure 5. 5. where ψ and χ were the Muskhelisvily complex potentials. 133 It can be shown that using the Airy stress function deﬁned in Eq. y) can be written as Φ(x. y) = 2Re[Φ1 (z1 ) + Φ2 (z2 ] (5.9.121 and combined with the compatibility Equation (Eq.2
Plane StressStrain Compliance Transformation
130 If we consider εi = aij σj for plane strain and εi = bij σj for plane stress then it can be shown that ai3 aj3 (5. 5.82) for anisotropic solids we obtain (neglecting body forces)
a22
∂4Φ ∂4Φ ∂4Φ ∂4Φ ∂4Φ + (2a12 + a66 ) 2 2 − 2a16 − 2a26 + a11 4 = 0 ∂x4 ∂x∂y ∂x ∂y ∂x∂y 3 ∂y
(5.3
131
Stress Functions
The stress function Φ(x.154)
where Φ1 (z1 ) is an arbitrary function of z1 = x + s1 y and Φ2 (z2 ) is an arbitrary function of z2 = x + s2 y
132 Note the analogy with φ = Re[(x1 − ix2 )ψ(z) + χ(z)] derived earlier for isotropic cases.Draft
5.8: Coordinate Systems for Stress Transformations
5. 5.
Two cases are possible: (a) Roots are all diﬀerent (b) Roots are pairwise equal 5.123 as derived earlier. All roots are complex or purely imaginary for an ideally elastic body with a11 = 0. 2a12 + a66 = 0. then a16 = a26 = 0 and we have E1 E1 − 2ν1 )s2 + =0 (5. purely imaginary and unequal roots Victor Saouma Fracture Mechanics
.161) s4 + ( µ E2 and (a) s1 = βi & s2 = δi.160b) (5. They characterize the degree of anisotropy for plane problems.160c) (5.157)
So s1 and s2 are complex parameters of ﬁrst order of plane stress (or strain). From it we can judge how much a body diﬀers from isotropy.158a) (5. 5. 6.
135 136
PRELIMINARY CONSIDERATIONS
The characteristic equation of this homogeneous partial diﬀerential equation is a11 s4 − 2a16 s3 + (2a12 + a66 )s2 − 2a26 s + a22 = 0 By energy considerations.Draft
30 which is Eq. Only exceptions are: (a) a22 = a26 = 0 ⇒ 2 roots equal to zero (b) a22 = a26 = 2a12 + a66 = a16 = 0 ⇒ all four roots are zero (c) a11 = a16 = 0 ⇒ 2 roots are inﬁnite (d) a11 = a16 = 2a12 + a66 = a26 = 0 ⇒ 4 roots are inﬁnite. 3. & a22 = 0 2. Leknitskii.159a) (5. (Lekhnitskii 1981) has shown that: 1.160d) (5.160a) (5. For isotropic material α = 0 β = 1 s1 = s2 = i ¯ s1 = s2 = −i ¯ (5. If a material is orthotropic and x and y coincide with 1 and 2. Two of the roots are conjugates of the two others: if we let s1 = α1 + iβ1 s2 = α2 + iβ2 then ¯ ¯ s3 = s1 s4 = s2 then β1 & β2 are both positive and β1 = β2 4.
In addition we have s1 = s2 = 8.166) (5.168) υ = 2Re[q1 φ1 (z1 ) + q2 φ2 (z2 )] where pi = a11 s2 + a12 − a16 si .163a) (5.4
137
Stresses and Displacements
dΦ1 dΦ2 and φ2 (z2 ) = dz1 dz2 and using the deﬁnition of stress functions.163c) (5. this chapter has provided the mathematical foundation required to develop the solutions of some “simple” problems which will be presented in the subsequent chapter.169) (5. 2) i qi = a12 si + as22 − a26 i (5. complex and equal roots (c) s1 = α + βi. the stresses are φ1 (z1 ) = σx = 2Re[s2 φ1 (z1 ) + s2 φ2 (z2 )] 1 2 σy = 2Re[φ1 (z1 ) + φ2 (z2 )] τxy = −2Re[s1 φ1 (z1 ) + s2 φ2 (z2 )] and the displacements are u = 2Re[p1 φ1 (z1 ) + p2 φ2 (z2 )] (5.167)
If we deﬁne:
(5.163b) (5.
Victor Saouma
Fracture Mechanics
.170a)
5.10
Conclusion
138 In summary.164)
(5.Draft
5.165) (5.162a) (5. (i = 1.162b)
(5. Invariants for orthotropic bodies are I1 = a11 + a22 + 2a12 = I2 I3 I4 1 1 2ν12 + − E1 E2 E1 1 4ν12 = a66 − 4a12 = + µ12 E1 1 1 = a44 + a55 = + µ13 µ23 ν13 ν23 ν31 + ν32 = a13 + a23 = −( + )=− E1 E2 E2 s2 cos ψ − sin ψ cos ψ + s1 sin ψ s1 cos ψ − sin ψ cos ψ + s2 sin ψ
31
(5.9.10 Conclusion (b) s1 = s2 = βi. s2 = −α + βi 7.163d)
5.
Draft
32
PRELIMINARY CONSIDERATIONS
Victor Saouma
Fracture Mechanics
.
Draft
Part III
LINEAR ELASTIC FRACTURE MECHANICS
.
Draft
.
1 Introduction
7 This chapter will present mathematically rigorous derivations of some simple elasticity problems. 1898)
Analysing the inﬁnite plate under uniform tension with a circular hole of diameter a. and subjected to a uniform stress σ0 .1: Summary of Elasticity Based Problems Analysed
. All the theoretical basis required to follow those derivations have been covered in the previous chapter. The ﬁrst one is a thick cylinder subjected to uniform radial pressure (solution of which is well known from Strength of Materials). whereas the ones around the hole should be written in polar coordinate system.
Problem Circular Hole Elliptical Hole Crack V Notch Dissimilar Materials Anisotropic Materials
Coordinate System Polar Curvilinear Cartesian Polar Polar Cartesian
Real/Complex Real Complex Complex Complex Complex Complex
Solution Kirsh Inglis Westergaard Willimas Williams Sih
Date 1898 1913 1939 1952 1959 1965
Table 6.
6. the second one is a thick cylinder subjected to both radial and shear stresses which must be compatible with the traction applied on the rectangular plate. A summary of problems to be investigated is shown in Table 6. (Kirsch.1.2
8
Circular Hole. Fig. 6. 10 We will solve this problem by replacing the plate with a thick tube subjected to two diﬀerent set of loads.
9 The peculiarity of this problem is that the farﬁeld boundary conditions are better expressed in cartesian coordinates.Draft
Chapter 6
ELASTICITY BASED SOLUTIONS FOR CRACK PROBLEMS
6.1.
3a) (6. the boundary conditions are given by: σyy = τxy = 0 (6.121) that σxx = ∂ Φ .4)
thus the stress function becomes Φ = Ar 2 + Br 4 + C 1 + D cos 2θ r2 (6.5)
16 Next. Thus. and D. 5. this would would suggest a stress function Φ of ∂y 2 2 .6)
17 Next we seek to solve for the four constants of integration by applying the boundary conditions. 5. Using Eq. away from the hole.
11
Recalling (Eq. 5. 5. the presence of the circular hole would suggest a polar the form Φ = σ0 y representation of Φ.
12 13
2
Since sin2 θ = 1 (1 − cos 2θ). C. Alternatively. we could simplify the stress function into 2 Φ = f (r) cos 2θ (6.122. B. we must determine the constants A.3b)
4 1 d d2 − 2 + 2 dr r dr r (note that the cos 2θ term is dropped)
15
The general solution of this ordinary linear fourth order diﬀerential equation is f (r) = Ar 2 + Br 4 + C 1 +D r2 (6.Draft
2
σo
ELASTICITY BASED SOLUTIONS FOR CRACK PROBLEMS
y
τr
θ
σrr
σrr b σo x a b a
σrr τr
θ
b
a
θ
θ
θ
I
II
Figure 6.1) σxx = σ0 .1: Circular Hole in an Inﬁnite Plate First we select a stress function which satisﬁes the biharmonic Equation (Eq. We will identify two sets of boundary conditions:
Victor Saouma
Fracture Mechanics
.124) yields 1 ∂2 1 ∂ ∂2 + 2 2 + ∂r 2 r ∂r r ∂θ 1 ∂2Φ ∂ 2 Φ 1 ∂Φ + 2 2 + ∂r 2 r ∂r r ∂θ 4f 1 df d2 f − 2 + 2 dr r dr r = 0 = 0 (6. substituting y = r sin θ would result in Φ = σ0 r 2 sin2 θ.123).2)
14
Substituting this function into the biharmonic equation (Eq. and the farﬁeld boundary conditions. From St Venant principle. the stresses are given by
σrr = σθθ = τrθ =
1 ∂Φ r ∂r ∂2Φ ∂r 2 ∂ − ∂r
+
1 ∂2Φ r 2 ∂θ 2
= − 2A + = 2A + 2A + =
1 ∂Φ r ∂θ
4D r 2 cos 2θ 2 + 6C cos 2θ 12Br r4 2 − 6C − 2D sin 2θ 6Br r4 r2
6C r4
+
(6.
6.9c) (6.9d.11a) (6.10b)
Upon substitution in Eq.6 the four boundary conditions (Eq. only the last two equations will provide us with boundary conditions. 6.9b) (6.11b) (6.8a and 6.9d)
Where state I corresponds to a thick cylinder with external pressure applied on r = b and of magnitude σ0 /2.8b) (6.2 Circular Hole. Outer boundaries: around an inﬁnitely large circle of radius b inside a plate subjected to uniform stress σ0 . and 6. 1898)
3
1. 2. Around the hole: the stresses should be equal to zero: (σrr )r=a = 0 ✛ (σrθ )r=a = 0
✛
(6.11d)
= 0 = 0
Victor Saouma
Fracture Mechanics
.11c) (6.10a. Hence.7)
yielding (recalling that sin2 θ = 1/2 sin 2θ. 6. (Kirsch.10a) (6. into state I and II: (σrr )I r=b = (σrθ )I r=b (σrr )II r=b (σrθ )II r=b 1 σ0 2 = 0 1 σ0 cos 2θ ✛ = 2 1 σ0 sin 2θ ✛ = 2 (6.Draft
6.10b) become
18
6C b4 6C 2A + 6Bb2 − 4 b 6C − 2A + 4 a 6C 2A + 6Ba2 − 4 a − 2A +
4D b2 2D − 2 b 4D + 2 a 2D − 2 a +
= =
1 σ0 2 1 σ0 2
(6.8c)
For reasons which will become apparent later.8a) (6. the stresses in polar coordinates are obtained from Strength of Materials σrr σrθ σrθ σθθ = cos θ − sin θ sin θ cos θ σ0 0 0 0 cos θ − sin θ sin θ cos θ
T
(6. and cos2 θ = 1/2(1 + cos 2θ)). 6. it is more convenient to decompose the state of stress given by Eq. 1 (σrr )r=b = σ0 cos2 θ = σ0 (1 + cos 2θ) 2 1 σ0 sin 2θ (σrθ )r=b = 2 σ0 (1 − cos 2θ) (σθθ )r=b = 2 (6.9a) (6.9c.8b. 6.
into Eq.e.18b)
x1 β=0 = a = c cosh α cos β = c cosh α0 ⇒ a = c cosh α0 x2 β=π/2 = b = c sinh α sin β = c sinh α0 ⇒ b = c sinh α0 Victor Saouma
Fracture Mechanics
. 5. The elliptical hole is itself deﬁned along α = α0 .8. These stresses are obtained from Strength of Materials yielding for this problem (carefull about the sign) σrr = σθθ = σ0 2 σ0 2 1− 1+ a2 r2 a2 r2 (6.13a) (6. Fig. 5. 4 D= a2 σ0 2 (6. 6. we deﬁne a and b as the major and minor semiaxes respectively. and as we go around the ellipse β varies from 0 to 2π. 23
Alternatively.15) (6.18a) (6.17) which for θ = σθθ = −σ0 . and with b much greater than a (Eq.16)
σθθ r=a = σ0 (1 − 2 cos 2θ) (6. 6.Draft
4
19 20
ELASTICITY BASED SOLUTIONS FOR CRACK PROBLEMS
a b
Solving for the four unknowns. at the edge of the hole when r = a we obtain σrr = 0 σrθ = 0
π 2 3π 2
(6.140 we obtain (6.6.13b)
21
Thus.8a and 6. 6. we obtain: a4 σ0 . an inﬁnite plate). both σrr and σrθ are equal to the values given in Eq.
= 0 (i.14b) (6. 1913)
Next we consider the problem of an elliptical hole in an inﬁnite plate under uniform stress.14a) (6. and gives a stress concentration factor (SCF) of 3.14c)
3a4 r4
1 σ0 cos 2θ 2
σrθ = − 1 −
3a4 2a2 + 2 r4 r
1 σ0 sin 2θ 2
22 We observe that as r → ∞. Adopting the curvilinear coordinate system described in sect.9a and 6. we must superimpose the one of a thick cylinder subjected to a uniform radial traction σ0 /2 on the outer surface. we obtain σrr = σθθ = σ0 2 σ0 2 1− 1+ a2 r2 a2 r2 + 1+3 − 1+ a4 4a2 − 2 r4 r 1 σ0 cos 2θ 2 (6. 4 B = 0.3
24
Elliptical hole in a Uniformly Stressed Plate (Inglis.
25
Thus substituting β = 0 and β =
π 2
in Eq. 6. and taking A=− σ0 .11a. substituting Eq.
6.8b respectively.2. For θ = 0 and θ = π.12)
C=−
To this solution. 6.9b.139 and 5.
The maximum value of σββ occurs at the end of the ellipse where β = 0 or π.19a) (6. For those points we have cos 2β = 1.20b)
1 π 4χ(z) = −σ0 c2 (cosh 2α0 − cosh π) p + e2α0 − cosh 2 p − α0 − i 2 2 where p = α + iβ
28
Since σαα = 0 for α = α0 .21)
thus diﬀerentiating (σββ )α=α0 =
29
sinh 2αo − 1 + e2α0 cos 2β σ0 cosh 2αo − cos 2β
(6.143 ¯ z σαα +σββ = 2[ψ (z) + ψ (¯)]
0
(6.23) Fracture Mechanics
. We then seek to apply the boundary conditions in an analogous manner as we did for the circular hole problem:
26
1. substituting into Eq. and since the tangent to the ellipse is now parallel to x2 we have: (σββ )β=0. Around the elliptical hole (α = α0 ) we have σαα = σαβ = 0 Inglis found that the following complex potentials satisfy the boundary conditions. Thus.20a) (6.2: Elliptical Hole in an Inﬁnite Plate respectively. 5.22)
But we need to express σββ in terms of a and b rather than α0 . and are periodic in β (period of 2π)
27
(6.19b)
4ψ(z) = σ0 c
1 + e2α0 sinh p − e2α0 cosh p
(6.3 Elliptical hole in a Uniformly Stressed Plate (Inglis.135 we obtain σ11 + σ22 = 4Reψ (z) = σ0 z σ22 − σ11 + 2iσ12 = 2[¯ψ (z) + χ (z)] = σ0 2. we can solve for σββ from Eq. 5.142 and 5.Draft
6. 1913)
x2
5
σο
α = αo x
2b
1
2a
σο
Figure 6.π = σ22 = α=α0 Victor Saouma sinh 2α0 − 1 + e2α0 σ0 cosh 2α0 − 1 (6. At inﬁnity we have σ22 = σ0 and σ11 = σ12 = 0.134 and 5.
27a) (6.π = σ0 1 + 2 α=α0 a ρ (6.25)
The following relations cosh 2α0 = 2 cosh2 α0 − 1 sinh 2α0 = 2 sinh α0 cosh α0 when combined with cosh α0 =
a c
(6.18a and 6. Combining Eq.π = σ0 1 + 2 α=α0 a b (6. and we hall rewrite this expression in terms of a and b. 6. substituting into the above equation gives: 1 (6. sin t = 0 and cos t = 1.Draft
6 we obtain
ELASTICITY BASED SOLUTIONS FOR CRACK PROBLEMS
30 We now have a solution for the stress at the tip of the ellipse. 34 Alternatively. 3 dx2 + dx2 1 2 (6. and that for a degenerated ellipse. Thus.e a crack there is an inﬁnite stress.
Victor Saouma
Fracture Mechanics
. 6.31)
From this equation. Substituting into Eq.29) ρ2 = (dx1 d2 x2 − dx2 d2 x1 )2
Using a parametric representation.26a) (6.23 (σββ )β=0.28)
33 We observe that for a = b.18b with (6.30) ρ2 = 2 2 (a2 sin2 t + b2 cos2 t)3 a b At the tip of the ellipse x2 = 0 and x1 = ±a.27b)
32
Finally.26b)
and sinh α0 = sinh 2α0 = cosh 2α0 =
b c
yield 2ab c2 a2 + b2 c2 (6. i. we can deﬁne x1 = a cos t and x2 = b sin t. we note that the stress concentration factor is inversely proportional to the radius of curvature of an opening. substituting those two equations into Eq. we recover the stress concentration factor of 3 of a circular hole.24) cosh2 α − sinh2 α = 1
c2 = a2 − b2
31
(6. and ρ becomes 2 equal to ρ = ba . 6.28 (σββ )β=0. let ρ be the radius of curvature of a parametric curve. However it is expressed in curvilinear coordinates. From analytical geometry.
33c)
Note that we should not confuse the Airy stress function Φ with the complex function φ(z). (Westergaard.
39 Let us verify that Φ satisﬁes the biharmonic equation. and the later by Williams.33a)
σ22 =
1
∂2Φ ∂x2 1
¯ = Reφ(z) + x2 Imφ(z) ∂ ∂x2 ∂ ¯ = Reφ(z) + x2 Imφ(z) + Imφ(z) ∂x1 ∂x1 ∂x1
(6.33b) (6. 5.126a that ∂x1 f (z) = f (z). From Inglis solution. Denoting by φ (z) and φ (z) the ﬁrst and ¯ ¯ second derivatives respectively. (Westergaard.
Westergaard’s solution. we have
∂Φ ∂x1
=
∂ ∂x2 ∂ ¯ ¯ ¯ Re φ + x2 Imφ(z) + Imφ(z) ∂x1 ∂x1 ∂x1
0
(6.4
35
Crack. however neither the nature of the singularity nor the stress ﬁeld can be derived from it. 1939)
Just as both Kolosoﬀ (1910) and Inglis (1913) independently solved the problem of an elliptical hole. The ﬁrst one was proposed by Westergaard. the second has the advantage of being extended to cracks at the interface of two diﬀerent homogeneous isotropic materials and be applicable for V notches. Fig.
Victor Saouma
Fracture Mechanics
. and ∂ recalling from from Eq. 1939)
x2
7
σ0
σ0 2a σ0
x1
σ0
Figure 6.3: Crack in an Inﬁnite Plate
6. Taking the ﬁrst derivatives. and φ(z) and φ(z) its ﬁrst and second integrals respectively of the function φ(z).
36 Let us consider an inﬁnite plate subjected to uniform biaxial stress σ0 with a central crack of length 2a. 6. Whereas the ﬁrst one is simpler to follow. there are two classical solutions for the crack problem. we know that there would be a theoretically inﬁnite stress at the tip of the crack. (Westergaard 1939b) starts by assuming Φ(z) as a harmonic function (thus satisfying Laplace’s equation ∇2 (Φ) = 0).4 Crack.32)
is a solution to the crack problem1 .
37 38
Westergaard has postulated that ¯ ¯ Φ = Reφ(z) + x2 Imφ(z) (6.Draft
6.3.
substituting into Eq.36b is satisﬁed. Equilibrium: ∂σ11 ∂σ12 + ∂x1 ∂x2 ∂σ22 ∂σ12 + ∂x2 ∂x1 Let us consider the ﬁrst equation ∂ ∂σ11 = ∂x1 ∂x1 ∂ ∂σ12 = ∂x2 ∂x2 ∂2Φ ∂x2 2 ∂2Φ ∂x1 ∂x2 = ∂ [Reφ(z) − x2 Imφ (z)] ∂x1 (6.
Victor Saouma
Fracture Mechanics
.98a in 2D:
43
1. it can be shown that the biharmonic equation is satisﬁed. 6. thus Φ is a valid solution. we have from Eq. it can be shown that σ12 = − ∂2Φ = −x2 Reφ (z) ∂x1 ∂x2 (6.37d) (6.34b) (6.36a then we do obtain zero.37b) (6.126b that if (z). 5. it can be shown that Eq.34c) (6.
42
If we want to convince ourselves that the stresses indeed satisfy both the equilibrium and compatibility equations (which they do by virtue of Φ satisﬁying the biharmonic equation). 5. diﬀerentiating with respect to x2 .37e) = 0 = 0 (6. 6.34d)
41
Similarly.36a) (6. Similarly. and recalling from Eq.123.36b)
= Reφ (z) − x2 Imφ (z) ∂ [−x2 Reφ (z)] ∂x2 ∂ Imφ (z) = −Reφ (z) + x2 ∂x1 = −Reφ (z) + x2 Imφ (z) =
If we substitute those two equations into Eq. we obtain ∂Φ ∂x2 σ11 = ∂2Φ ∂x2 2 ∂ ∂x2 ¯ ∂ ¯ ¯ (Reφ(z) + [x2 Imφ(z) + Imφ(z)] ∂x2 ∂x2 ∂x2 ¯ ¯ = −Imφ(z) + x2 Reφ(z) + Imφ(z) ∂ ∂x2 = x2 Reφ(z) + Reφ(z) ∂x2 ∂x2 = −x2 Imφ (z) + Reφ(z) =
∂ ∂x2 f (z)
=
(6.35)
Having derived expressions for the stresses and the second partial derivatives of Φ.34a) (6.33d)
Similarly.Draft
8
40
ELASTICITY BASED SOLUTIONS FOR CRACK PROBLEMS = Reφ(z) + x2 Imφ (z) (6.37c) (6. 5.37a) (6.
2. Along the crack: at x2 = 0 and −a < x1 < a we have σ22 = 0 (traction free crack).43)
Victor Saouma
Fracture Mechanics
.39a) (6.38b)
and are obtained by integration of the strains which were in turn obtained from the stresses. At inﬁnity: at x2 = ±∞. displacements are given by ¯ 2µu1 = (1 − 2ν)Re φ(z) − x2 Imφ(z) ¯ 2µu2 = 2(1 − ν)Imφ(z) − x2 Re φ(z)
9
(6. we also need to have σ22 = 1− 0a . and σ22 = Re
σ0 1−
a2 x2 1
(6. this is extended to σ22 = 2
x1
1− a2
x 1
σ22 = 0 when x2 = 0 and −a < x1 < a. As a check we compute 2µε11 = ∂u1 ∂x1 = (1 − 2ν)Reφ(z) − x2 Imφ (z) = (1 − ν) [Reφ(z) − x2 Imφ (z)] −ν [Reφ(z) + x2 Imφ (z)]
σ11 σ22
(6. we expect σ22 → σ0 as x1 → ∞. we must determine φ such that the boundary conditions are satisﬁed.Draft
6. For reasons which will become apparent later.40)
(6.39b) (6.
44 45 Next. for symmetry. 1939) 2. thus the function φ(z) should become imaginary along the crack. we generalize our problem to one in which we have a biaxial state of stress applied on the plate. However.41)
this shows that Eε11 = σ11 − ν(σ22 − σ33 ). and σ22 to be greater than σ0 when  x1 −a > (due to anticipated singularity predicted by Inglis). Hence:
1. 6. σ22 = σ0 We note from Eq. (Westergaard.33d that at x2 = 0 σ22 reduces to (σ22 )x2 =0 = Reφ(z)
46
(6.42)
Furthermore. and we simply determined the stresses in terms of it.39d)
= (1 − ν)σ11 − νσ22 Recalling that µ= then Eε11 = (1 − ν 2 )σ11 − ν(1 + ν)σ22 E 2(1 + ν)
(6.38a) (6. and for plane strain. Compatibility: In plane strain.4 Crack.39c) (6. ε33 = 0 ⇒ σ33 = ν(σ11 + σ22 ) and Eε11 = (1 − ν 2 )σ11 − ν(1 + ν)σ22 So far Φ was deﬁned independently of the problem. and veriﬁed that the biharmonic equation was satisﬁed. thus a possible choice for σ22 would be σ σ0 .
50) ΦII (z) = −x2 ReφII (z) ⇒ φII = 2 1 − a2 z
and for the same crack but subjected to antiplane shear stresses (mode III) ΦIII (z) = σ13 1−
a2 z2
(6. we can derive expressions for the stress ﬁeld around a crack tip in a plate subjected to far ﬁeld shear stresses (mode II as deﬁned later) using the following expression of φ τ ¯ (6.51)
6. and 6.44)
If we perform a change of variable and deﬁne η = z − a = reiθ and assuming η 1. then the ﬁrst term of Eq. we substituting x2 = r sin θ into the second term
49
x2 Imφ = r sin θIm
σ0 2
a = σ0 2(reiθ )3
θ θ 3θ a sin cos sin 2r 2 2 2
(6.1
53
Stress Intensity Factors (Irwin)
Irwin2 (Irwin 1957) introduced the concept of stress intensity factor deﬁned as:
2 Irwin was asked by the Oﬃce of Naval Research (ONR) to investigate the Liberty ships failure during World War II.4. φ(z) = σ0 1−
a2 z2
(6.34d.47) (6.Draft
10
47 48
ELASTICITY BASED SOLUTIONS FOR CRACK PROBLEMS
Thus from Eq.48) (6. 52 Using a similar approach. however this should not aﬀect the stress ﬁeld close to the crack tip. 6. 6. 6. 6. the uniaxial case may be reproduced by superimposing a pressure in the x1 direction equal to −σ0 . and a iθ = cos θ + i sin θ.33d can be rewritten as recalling that e Reφ(z) = Re ≈ Reσ0 σ0
η2 +2aη η2 +a2 +2aη
≈ Re
σ0
2aη a2
a ≈ Reσ0 2η
a ≈ σ0 2reiθ
a −i θ e 2 ≈ σ0 2r
θ a cos 2r 2
(6.46)
50
Combining the above equations.
Victor Saouma
Fracture Mechanics
.42 we have (note the transition from x1 to z).49)
51 Recall that this was the biaxial case.33d.45a)
Recalling that sin 2θ = 2 sin θ cos θ and that e−iθ = cos θ − i sin θ. just as thirty years earlier Inglis was investigating the failure of British ships.35 we obtain σ22 = σ0 σ11 = σ0 σ12 = σ0 θ 3θ θ a cos 1 + sin sin + ··· 2r 2 2 2 θ 3θ θ a cos 1 − sin sin + ··· 2r 2 2 2 θ θ 3θ a sin cos cos + ··· 2r 2 2 2 (6. with Eq.
(6.4: Independent Modes of Crack Displacements
54
From Eq.
Figure 6. • Tearing Mode.
θ θ 3θ KII sin cos cos σ12 = √ 2 2 2 2πr
II f22
(6. but the deformations are skew symmetric about the x − y and x − z planes.52)
where σij are the near crack tip stresses.48 and 6.4 Crack. 6. as shown in Fig. but the deformations are symmetric about the x − y plane and skew symmetric about the x − z plane.θ=0 K σ III 23
(6. 1939)
KI σ22
11
√ KII = lim 2πr σ12 r→0. 6.55)
Victor Saouma
Fracture Mechanics
. I: The two crack surfaces are pulled apart in the y direction. III: The crack surfaces slide over each other in the zdirection.49 with θ = 0. (Westergaard. II: The two crack surfaces slide over each other in the xdirection. • Shearing Mode. 6.53a)
55
Thus stresses and displacements can all be rewritten in terms of the SIF
σ22 12 σ 23
σ
II III f I (θ) f11 (θ) f11 (θ) KI 1 11 I (θ) f II (θ) f III (θ) =√ KII f22 22 22 2πr f I (θ) f II (θ) f III (θ) K III 12 12 12
(6. but the deformations are symmetric about the x − z and x − y planes.Draft
6.4: • Opening Mode.54a)
i. we have √ 2πrσ22 KI = √ a = 2πrσ0 2r √ = σ0 πa where r is the length of a small vector extending directly forward from the crack tip.47.e. and Ki are associated with three independent kinematic movements of the upper and lower crack surfaces with respect to each other.
As we shall see later. One of the underlying principles of FM is that unstable fracture occurs when the SIF reaches a critical value KIc .
3
Victor Saouma
Fracture Mechanics
. 4. the near crack tip (r a) stresses and displacements are always expressed as:
Pure mode I loading: σxx = σyy = τxy = σzz KI (2πr) KI (2πr) KI
1 2
cos cos sin
3θ θ θ 1 − sin sin 2 2 2 3θ θ θ 1 + sin sin 2 2 2 θ 3θ θ cos cos 2 2 2 θ θ κ − 1 + 2 sin2 2 2 θ θ κ + 1 − 2 cos2 2 2
(6. previous equations are exact in the limit as r→0 2.e. the spatial distribution is a function of elastic constants. i.56g)
1 2
(2πr) = ν(σx + σy )τxz = τyz = 0 KI r 2µ 2π
1 2
1 2
u = v =
cos sin
KI r 2µ 2π w = 0
1 2
√ Note that in certain literature. 3. k = √ f (g)σ a is used.56c) (6. Note that this polar distribution is identical for all cases. for anisotropic cases. Tada “Stress Analysis of Cracks”. Distribution of elastic stress ﬁeld at tip can be described by KI . Paris and Irwin 1973).2
Near Crack Tip Stresses and Displacements in Isotropic Cracked Solids
56 Using Irwin’s concept of the stress intensity factors.56d) (6. (Tada.56a) (6. and Cartwright & Rooke.
6.56b) (6. (specially the one of Lehigh University). which characterize the strength of the singularity at a crack tip. 7. KII and KIII .56f) (6. The SIF is the measure of the strength of the singularity (analogous to SCF) √ 5. 6. “Compendium of Stress Intensity Factors” (Rooke and Cartwright 1976). crack geometry. instead of K = f (g)σ πa. KIc or fracture toughness represents the inherent ability of a material to withstand a given stress ﬁeld intensity at the tip of a crack and to resist progressive tensile crack extensions.4.56e) (6. SIF are additives. and loading. K = f (g)σ πa where f (g) is a parameter3 that depends on the specimen. Since higher order terms in r were neglected.Draft
12
ELASTICITY BASED SOLUTIONS FOR CRACK PROBLEMS
1.
60c we can write the stresses in polar coordinates θ K √ I cos 1 + sin2 2 2πr θ K √ I cos 1 − sin2 2 2πr θ K θ √ I sin cos2 2 2 2πr θ 2 θ 2
Pure mode I loading: σr = σθ = τrθ = Pure mode II loading: σr = σθ = τrθ = Victor Saouma θ 3 3θ K 5 √ II − sin + sin 4 2 4 2 2πr θ 3 3θ KII 3 √ − sin − sin 4 2 4 2 2πr θ 3 3θ KII 1 √ cos + cos 2 4 2 2πr 4 (6. and 5.4 Crack.Draft
6.58d) (6.59a) (6.58c) (6. (Westergaard.60c) Fracture Mechanics (6.59b) (6. 5.59c)
.60b. 1939) Pure mode II loading: σxx = − σyy = τxy = σzz KII (2πr) KII
1 2 1 2
13
sin
θ θ 3θ 2 + cos cos 2 2 2
(6.
Using Eq.57b) (6. 5.57f) (6.57a) (6.58b) (6.60a.58e)
cos
(2πr) = σy = σz = τxy = 0
1 2
w =
KIII 2r µ π u = v=0
sin
θ 2
where κ = 3 − 4ν for plane strain.57d) (6. and κ =
57
3−ν 1+ν
for plane stress.57e)
sin cos
(2πr) KII
θ 3θ θ cos cos 2 2 2 3θ θ θ 1 − sin sin 2 2 2
(2πr) = ν(σx + σy ) KII r 2µ 2π
1 2
1 2
τxz = τyz = 0 u = sin
1 2
θ θ κ + 1 + 2 cos2 2 2 θ θ κ − 1 − 2 sin2 2 2
(6.57g) (6.57c) (6.60a) (6.58a) (6.57h)
KII r v = − 2µ 2π w = 0 Pure mode III loading: τxz = − τyz = σxx
cos
KIII
1 2
sin θ 2
(2πr) KIII
1 2
θ 2
(6.60b) (6.
61)
where F (θ.5 making an angle 2γ.124 1 ∂2 1 ∂2Φ 1 ∂ ∂2 ∂ 2 Φ 1 ∂Φ + 2 2 + 2 2 =0 + + (6.Draft
14
ELASTICITY BASED SOLUTIONS FOR CRACK PROBLEMS
P
r 2γ α
θ
Figure 6. For γ = 0 we recover the crack problem of Westergaard. Williams 1957) proposed the following solution
Φ(r.65b)
Victor Saouma
Fracture Mechanics
. and m(λ) is yet to be determined.61 into the biharmonic equation (Eq. 6.62) ∇2 ∇2 Φ = ∂r 2 r ∂r r ∂θ ∂r 2 r ∂r r ∂θ Note that the problem he originally considered was not a crack.124) gives ∂ 2 F (θ. λ) + 2(λ2 + 1) + (λ2 − 1)2 F (θ.63)
61
Substituting F (θ. Williams (Williams 1952.5
V Notch. 5. Eq. λ) with em(λ)θ . λ) ∂ 4 F (θ. this equation reduces to em(λ)θ (λ − 1)2 + m(λ)2 (1 + λ)2 + m(λ)2 = 0 (6. λ) = em(λ)θ . Fig. λ) = 0 ∂θ 4 ∂θ 2 (6. but rather a plate under tension with angular corners.65a) (6. 5. 6.64)
62
The roots of this equation are m(λ) = ±i (λ − 1) m(λ) = ±i (λ + 1) (6. θ) ≡ r λ+1 F (θ.5: Plate with Angular Corners
6. (Williams. to the biharmonic equation. 1952)
58 Using the method of separation of variables in 1952.
59 60
Substituting Eq. λ)
(6.
F3 and F4 .5 V Notch. F2 . 5. λ) F3 (θ.70b)
[−A(λ − 1) sin(λ − 1)θ − B(λ + 1) sin(λ + 1)θ
+ C(λ − 1) cos(λ − 1)θ + D(λ + 1) cos(λ + 1)θ]
The boundary conditions are next applied by considering a plate with a central crack. λ) F2 (θ. Hence we select as solution
F1 (θ. Substituting Eq.66)
and ﬁnally F (θ.λ)
(6.70a) (6. λ) is a real function then the solutions of the diﬀerential equation 6. λ) will be a linear combination of F1 . which implies that
= 0 = 0
(6. 1952)
15
63 Since F (θ. thus Φ(r.Draft
6.68b)
σθθ = r λ−1 λ(λ + 1)F (θ) σrθ = r Substituting σθθ = r λ−1 λ(λ + 1)[A cos(λ − 1)θ + B cos(λ + 1)θ + C sin(λ − 1)θ + D sin(λ + 1)θ] σrθ = −λr
λ−1 λ−1
(6.71b)
F (α) = F (−α) = F (α) = F (−α) = 0
ω sin(λ − 1)α 0
(6.122
64
σθθ = σrθ we obtain:
∂2Φ ∂r 2 ∂ = − ∂r
(6.
cos(λ + 1)α 0 sin(λ + 1)α 0 0 sin(λ − 1)α 0 ω cos(λ − 1)α cos(λ + 1)α
A 0 B 0 =0 sin(λ + 1)α C
(6. λ) F4 (θ.72)
cos(λ − 1)α
0 where ω =
λ−1 λ+1 . (Williams.68a) 1 ∂Φ r ∂θ (6.61 into Eq. we seek to determine the stresses in polar coordinates in order to apply the boundary conditions.69a) (6. applying the following 4 boundary conditions along the crack edges
65
σθθ θ=±α σrθ θ=±α where α + γ = π.73)
D
Victor Saouma
Fracture Mechanics
. θ) = r λ+1 [A cos(λ − 1)θ + B cos(λ + 1)θ + C sin(λ − 1)θ + D sin(λ + 1)θ]
F (θ.69b)
[−λF (θ)]
(6. λ)
= = = =
cos(λ − 1)θ cos(λ + 1)θ sin(λ − 1)θ sin(λ + 1)θ
(6.71a) (6. 6.67)
Next.62 are also real functions.
λ) = An cos n n − 1 θ + Bn cos +1 θ 2 2 n n − 1 θ + Dn sin +1 θ +Cn sin 2 2
(6.79a) (6. i.67 F (θ.69a and 6. the above equation is written in terms of the summation and the diﬀerences of the stress equations.74 for the symmetrical loading. 6.
Let us denote by λn the eigenvalues of λ which are solution of Eq. 6. and C and D to unsymmetric loading (mode II). Substituting into Eq.78b)
(6.76a) (6. 6.
Victor Saouma
Fracture Mechanics
. and ξn solutions for the antisymmetric loading:
69
sin 2λn α + λn sin 2α = 0 sin 2ξn α − ξn sin 2α = 0
(6.78a) (6.75b)
70
For the case of a crack. 6.e α = π solution of the characteristic equation4 is sin(2πλn ) = 0 sin(2πξn ) = 0 (6. · · · (it can be shown that n = 2 gives rise to a 2 rigid body motion contribution). 3.77a)
71 We observe that the previous expression can be simpliﬁed by noting that for each eigenvalue λn and ξn there is a relationship between A and B. 4.74)
68 We observe from Eq.79b)
4 Note that the solution for λn and ξn for other angles can not be obtained algebraically. and between C and D in Eq.67 that the coeﬃcients A and B correspond to symmetric loadings (mode I).75a) (6.76b)
which has solutions λn = n with n = 1. 6. thus yielding a block diagonal matrix.Draft
16
ELASTICITY BASED SOLUTIONS FOR CRACK PROBLEMS
66 Note that whereas Eq.69b are expressed in terms of the four constants (A. a numerical technique must be used. C. For symmerical loading we have
An cos(λn − 1)α + Bn cos(λn + 1)α = 0 An ω sin(λn − 1)α + Bn sin(λn + 1)α = 0 and for antisymmetric loading we have Cn sin(ξn − 1)α + Dn sin(ξn + 1)α = 0 Cn ω cos(ξn − 1)α + Dn cos(ξn + 1)α = 0
(6. This would lead (after some simpliﬁcations) to:
sin 2λα ± λ sin 2α = 0
(6. B.73. and D). 67 It can readily be seen that A and B are independent of C and D. and that for this homogeneous equation a nontrivial solution would exist if and only if the determinant of the system of linear equations vanishes to zero.
59a and 6. 6. 6.5 V Notch. 6. (Williams. Eq.83b.82a) (6.59b and 6. From this we observe that
b1 = a1 =
K √I 2π KII √ 2π
(6.Draft
72
6.81)
The stresses are obtained by substituting σrr = σθθ = σrθ = bn √ r bn √ r bn √ r an θ 1 3θ 5 +√ cos − cos 4 2 4 2 r θ 1 3θ an 3 cos + cos +√ 4 2 4 2 r an θ 1 3θ 1 +√ sin + sin 4 2 4 2 r 5 θ 3 3θ − sin + sin 4 2 4 2 θ 3 3θ 3 − sin − sin 4 2 4 2 θ 3 3θ 1 cos + cos 4 2 4 2 (6.83b) (6.84b)
76 This solution can be compared with Westergaard’s solution by comparing Equations 6. 1952) Thus we can deﬁne an = bn = Bn cos(λn − 1)α ω sin(λn − 1)α =− =− An cos(λn + 1)α sin(λn + 1)α Cn sin(λn − 1)α ω cos(λn − 1)α =− =− Dn sin(λn + 1)α cos(λn + 1)α
17
(6.83c)
75
Finally.60b with Eq.59c and 6. 6.80a) (6.60a with Eq.60c with Eq. and Eq.83a.83a) (6.86)
Victor Saouma
Fracture Mechanics
. 6.80b)
these ratios are equal to 1/3 and −1 respectively for α = π and λ = 1/2.82c)
74
These equations can be further simpliﬁed into σrr = σθθ = σrθ = θ θ an 5 bn θ √ cos 1 + sin2 + √ − sin + r 2 2 r 4 2 θ θ an 3 bn θ √ cos 1 − sin2 + √ − sin − r 2 2 r 4 2 θ a θ 3 3θ bn θ 1 √ sin cos2 + √n cos + cos r 2 2 r 4 2 4 2 3 3θ sin 4 2 3 3θ sin 4 2 (6.84a)
−bn r ξn [(κ + ξn cos 2α − cos 2ξn α) sin ξn θ − ξn sin(ξn − 2)θ] v = Re an r λn [(κ − λn cos 2α − cos 2λn α) sin λn θ + λn sin(λn − 2)θ]
+bn r ξn [(κ − ξn cos 2α + cos 2ξn α) cos ξn θ + ξn cos(ξn − 2)θ]
(6.82b) (6. it can be shown that the displacements will be given by u = 1 2µ 1 2µ Re an r λn [(κ + λn cos 2α + cos 2λn α) cos λn θ − λn cos(λn − 2)θ] (6.85) (6. and F (θ) =
73
1 3 an sin θ + sin θ + bn 2 2
3 1 1 cos θ + cos θ 3 2 2
(6.83c for n = 1.
1959)
General Function
We shall now consider the problem of a crack at the interface between two dissimilar isotropic materials.6: Plate with Angular Corners
6.Draft
18
ELASTICITY BASED SOLUTIONS FOR CRACK PROBLEMS
Ε 1 . we rewrite Eq. on the free edges (at θ = ±π). ν1
Ε 2 . ν2
Figure 6. 6.90c)
F (θ) [+A2 (λ − 1) sin(λ − 1)θ + B2 (λ + 1) sin(λ + 1)θ
−C2 (λ − 1) cos(λ − 1)θ − D2 (λ + 1) cos(λ + 1)θ] = 0 Victor Saouma
Fracture Mechanics
. 6.87)
Hence: Fi (θ. on the free edges (at θ = ±π).69a σθθ θ=π = r λ−1 λ(λ + 1)[A1 cos(λ − 1)π + B1 cos(λ + 1)π +C1 sin(λ − 1)π + D1 sin(λ + 1)π] = 0 σθθ θ=−π = r or F1 (π) = F2 (−π) = 0 • zero stresses. λ) where the subscript i refers to material 1 and 2.61 as Φi (r.6
6.88)
6.90b) (6.1
77
Crack at an Interface between Two Dissimilar Materials (Williams. 6. θ) ≡ r λ+1 Fi (θ. Fig.90a) (6.6. λ) = Ai cos(λ − 1)θ + Bi cos(λ + 1)θ + Ci sin(λ − 1)θ + Di sin(λ + 1)θ (6. Thus from Eq. Zak and Williams 1963). Accordingly. σθθ . σrθ . From Eq. (Williams 1959.69b σrθ θ=π = −λr λ−1 [−A1 (λ − 1) sin(λ − 1)θ − B1 (λ + 1) sin(λ + 1)θ +C1 (λ − 1) cos(λ − 1)θ + D1 (λ + 1) cos(λ + 1)θ] = 0 σrθ θ=−π = −λr = −λr
λ−1 λ−1 λ−1
(6. 6.6.89b)
λ(λ + 1)[A2 cos(λ − 1)π + B2 cos(λ + 1)π
−C2 sin(λ − 1)π − D2 sin(λ + 1)π] = 0
(6.6
78
(6.89a) (6.2
79
Boundary Conditions
Boundary conditions for this problem are: • zero stresses.
89b.90c.6 Crack at an Interface between Two Dissimilar Materials (Williams. and αi ≡ we obtain
1 [−(λ + 1)F1 (0) + 4A1 (1 − α1 )] = 2µ1 1 −F1 (0) − 4C1 (1 − α1 ) = 2µ1
1 [−(λ + 1)F2 (0) + 4A2 (1 − α2 )] (6.. will lead to 8 homogeneous linear equations (Eq.
Victor Saouma
Fracture Mechanics
.
5
The original paper states: . 6. After some algebraic simpliﬁcation. 6.6..
For the homogeneous case α1 = α2 and k = 1.94a) 2µ2 1 −F2 (0) − 4C2 (1 − α2 ) (6.92)
• Continuity of displacements (ur .76b for a crack in one material: n n = 1. 1959) or F1 (π) = F2 (−π) = 0 • Continuity of σθθ at the interface. the previous equation reduces to cot2 λπ = 0 or sin2 λπ = 0 thus we recover the same solution as the one of Eq. 6.94b) 2µ2
6. θ = 0 A1 + B1 = A2 + B2 • Continuity of σrθ at θ = 0 along the interface (λ − 1)C1 + (λ + 1)D1 = −(λ − 1)C2 − (λ + 1)D2
19
(6. and the lowest eigenvalue controls.89a.95)
µ1 µ2 .93a) 2µi 1 λ r {−Fi (θ) − 4(1 − αi )[Ci cos(λ − 1)θ − Ai sin(λ − 1)θ]} (6. 6. C1 ..90b. C2 and D2 . 6. 81 A nontrivial solution exists if the determinant of the 8 equations is equal to zero. 3.94b) in terms of the 8 unknowns A1 .92. 6. . B1 . 6.94a. (6. uθ ) at the interface.Draft
6. A2 .91. 6.91)
(6.. 6.96) λ= 2 Note that we exclude negative values of n to ensure ﬁnite displacements as the origin is approached. D1 . This determinant5 is equal to
cot2 λπ + where k =
82
2k(1 − α2 ) − 2(1 − α1 ) − (k − 1) 2k(1 − α2 ) + 2(1 − α1 )
2
=0
(6..93b) 2µi
νi 1+νi
where µ is the shear modulus.3
Homogeneous Equations
80 Applying those boundary conditions.. B2 . Using the polar expression of the displacements ui = r ui = θ 1 λ r {−(λ + 1)Fi (θ) + 4(1 − αi )[Ci sin(λ − 1)θ + Ai cos(λ − 1)θ]} (6. 2.
99a) (6.101) (6. or cot λπ = ±iβ
84
(6.102d) (6.98)
To solve this equation.99e)
85
Substituting in Eq.95 leads to cot2 λπ = −β 2 .99a lead to cot λπ = = u(1 − v 2 ) − iv(1 + u2 ) u2 + v 2 (tan2 λr π + 1) tanh λj π tan λr π(1 − tanh2 λj π) −i 2 λ π + tanh2 λ π tan r tan2 λr π + tanh2 λj π j Re(cot λπ)=0 (6. we deﬁne β= 2k(1 − α2 ) − 2(1 − α1 ) − (k − 1) 2k(1 − α2 ) + 2(1 − α1 ) (6.100a) (6. we obtain sin 2λr π = cos 2λr π = sinh 2λj π = cosh 2λj π = 2u 1 + u2 1 − u2 1 + u2 2v 1 − v2 1 + v2 1 − v2 (6. we use the following trigonometric relations. 6. 6.100b) sin 2x − i sinh 2y cosh 2y − cos 2x 2u 1 + u2 1 − u2 1 + u2 2v 1 − v2 1 + v2 1 − v2 (6.102a) (6.102c) (6.99c) (6.103b)
Victor Saouma
Im(cot λπ)=±β Fracture Mechanics
.103a) (6.102b) (6.97)
and thus.99d) (6.95.99b) (6. 6. Eq.Draft
20
83
ELASTICITY BASED SOLUTIONS FOR CRACK PROBLEMS
Noting that there can not be a real solution to Eq. (Abramowitz and Stegun 1970) cot z = sin 2θπ = cos 2θπ = sinh 2θπ = cosh 2θπ = where u = tan λr π v = tanh λj π where we have assumed a complex value for λ λ = λr + iλj Then.
. If from 6. and accordingly from Eq. 1.105)
We note that for this case. 2 1 = ± tanh−1 β π β+1 1 log = 2π β−1 (6.6 Crack at an Interface between Two Dissimilar Materials (Williams.109b) (6. 2.106) (6. (6. 2. 3. as the eﬀect of dissimilar materials has been accounted for and is embedded in λ.107a) (6.
6. .5
Near Crack Tip Stresses
88 Now that we have solved for λ..Draft
86
6.95 ﬁnally leads to Re(cot λπ) = 0 Im(cot λπ) = ±β we thus have two equations with two unknowns. Alternatively.103b tan λr π = 0 then λr = n = 0. λ)
G(r)
we note that we no longer have two sets of functions.6. 1. λj → 0 as α1 → α2 and k → 1 in β.107b) (6.109c)
log
Victor Saouma
Fracture Mechanics
.4
87
Solve for λ
Let us solve those two equations. We rewrite Eq. ?? σrr = σθθ = σrθ =
6
Recall that tanh
−1
x=
1 2
1 ∂ 2 Φ 1 ∂Φ = r −2 G(r)F (θ) + r −1 G (r)F (θ) + r 2 ∂θ 2 r ∂r ∂2Φ = G (r)F (θ) ∂r 2 1 ∂Φ 1 ∂ 2 Φ − = r −2 G(r)F (θ) − r −1 G (r)F (θ) r 2 ∂θ r ∂r∂θ
1+x 1−x
(6.109a) (6.. 6.104b)
6.. 6. from Eq.
89
The stresses will be given by Eq. 6. .108) Φ(r) = r λ+1 F (θ.6. 1959)
21
Thus.104b 1 λj = ± coth−1 β π 2.87 as (6. 6. Eq.104a cot λr π = 0 ⇒ tan λr π = ∞ and6 : λr = λj 2n + 1 n = 0. we need to derive expressions for the near crack tip stress ﬁeld. 3.104a) (6.107c) (6. Two sets of solutions are possible: 1.
However.110b becomes r iλj = eiλj log(r) = e = e
iλj [log(r)+i2kπ iλj log(r)−2kπλj −2kπλj iλj log(r)
(6.112d) (6.116)
we need to replace λ by λr + iλj . ﬁrst we recall the following relations. · · ·. λ) = A cos(λ − 1)θ + B cos(λ + 1)θ + C sin(λ − 1)θ + D sin(λ + 1)θ (6.112a) (6. and since z = arg z = 2kπ.Draft
22
90
ELASTICITY BASED SOLUTIONS FOR CRACK PROBLEMS
Therefore.115b) (6. log(r) = log r + i2kπ and Eq. 6.112b) (6. (Abramowitz and Stegun 1970) sin(x + iy) = sin(x) cosh(y) − i cos(x) sinh(y) cos(x + iy) = cos(x) cosh(y) + i sin(x) sinh(y) Victor Saouma (6. First we note that G(r) = r λ+1 = r λr +1+iλj r
iλj
(6.110a) (6.110a becomes G(r) = r λ+1 = r λr +1 r iλj = r and for λr = 1/2 G(r) = r 2 [cos(λj log(r)) + i sin(λj log(r))] ReG(r) = r cos(λj log(r)) 1 3 cos(λj log(r)) + λj sin(λj log(r)) G (r) = r 2 2 1 3 3 λj − λ2 cos(λj log(r)) + + G (r) = r − 2 j 4 2 2
3 2 3
(6. F (θ). Eq. ±2.88 F (θ.112e)
= e = e
e
−2kπλj
[cos (λj log(r)) + i sin (λj log(r))]
91
Now.114b)
λr +1
[cos(λj log(r)) + i sin(λj log(r))]
(6.110b) x2 + y 2 = √ r 2 + 0 = r. G (r) and G (r) in terms of λ = λr + iλj . Hence. which was deﬁned in Eq.114a) (6. 6.115c) sin(λj log(r)) (6. for k = 0.115d)
92
Back to F (θ).113)
and accordingly. 6.117b) Fracture Mechanics
. we have
= e
iλj log(r)
recalling that log(z) = log z + i arg z.117a) (6. we must solve for F (θ).115a) (6. ±1. and k = 0. r iλj becomes r iλj = cos(λj log(r)) + i sin(λj log(r)) (6.111)
(6.112c) (6.
125b) +b cosh λj θ − sin + λj sinh λj θ cos 2 2 2 2 2 2 (6. we now can determine (6. 1959)
23
Re {sin [(λr ± 1) + iλj ] θ} = sin(λr ± 1) cos(θ) cosh λj θ Re {cos [(λr ± 1) + iλj ] θ} = cos(λr ± 1) cos(θ) cosh λj θ
(6.125a) F (θ) = f (θ)g(θ) + f (θ)g (θ) 3θ 1 θ θ 3θ 3 cos + cos + sin + λj sinh λj θ sin = a cosh λj θ 2 2 2 2 2 2 3θ 3 θ θ 3 3θ − sin + 3 cos (6. 6. σrθ = 0. 6.121)
96
Similarly at θ = ±π. 6. θ)] = r
λr +1
cos(λj log(r)) cosh λj θ (6. that is g1 (−π) = g2 (π) or C = −D = −a (6.
93
6.119b) [A cos(λr − 1)θ + B cos(λr + 1)θ + C sin(λr − 1)θ + D sin(λr + 1)θ]
g(θ)
Re [Φ(r. or g1 (−π) = g2 (π) or A = 3B = b (6. σθθ = 0.118a) (6.124a) 3θ 3 θ 3 − sin + b − sin 2 2 2 2 (6. Eq.122)
97
From those two equations we rewrite Eq. Thus.124b) 3θ θ + sin 2 2 + b 3 cos 3θ θ + cos 2 2 (6.6 Crack at an Interface between Two Dissimilar Materials (Williams.116 Re [F (θ)] = cosh λj θ
f (θ)
(6.109c F (θ) = 0.120)
95 Applying the boundary conditions at θ = ±π. from Eq.119a)
(6.123)
98
99
Victor Saouma
Fracture Mechanics
.120 g(θ) = a sin We now determine the derivatives f (θ) = λj sinh λj θ 3θ 1 θ 3 cos + cos g (θ) = a 2 2 2 2 Thus. 6.109b F (θ) = 0.118b)
Substituting those relations in Eq.Draft
thus.119c)
[A cos(λr − 1)θ + B cos(λr + 1)θ +C sin(λr − 1)θ + D sin(λr + 1)θ]
94
For λr =
1 2
g(θ) = A cos
3θ θ 3θ θ + B cos − C sin + D sin 2 2 2 2
(6.
127b) j 4 2 2 3 cos(λj log(r)) + λj sin(λj log(r)) F (θ) (6.109c and 6. the second derivative F (θ) is determined (6.109a to determine the stresses σrr = r − 2 cos(λj log(r))F (θ) + r − 2
1 1
σθθ = r − 2 σrθ = r − 2
1
1
3 cos(λj log(r)) + λj sin(λj log(r)) F (θ) (6.154 such that they satisfy the boundary conditions of the problem under consideration.
103
u1 = KI v1 = KI w1 = 0 u2 = KII v2 = KII w2 = 0 u3 = 0 v3 = 0 w3 = KIII Victor Saouma
1 1 2r 1 Re s1 p2 (cos θ + s2 sin θ) 2 − s2 p1 (cos θ + s1 sin θ) 2 (6.128c) 1 1 2r 1 Re p2 (cos θ + s2 sin θ) 2 − p1 (cos θ + s1 sin θ) 2 (6.127c) cos(λj log(r))F (θ) + 2
6. 5. (Sih.Draft
24
100
ELASTICITY BASED SOLUTIONS FOR CRACK PROBLEMS
Similarly.128g) (6.128d) π s1 − s2 1 1 2r 1 Re q2 (cos θ + s2 sin θ) 2 − q1 (cos θ + s1 sin θ) 2 (6.126b) j 2 2
101
We can now substitute in Eq.128h) 2r c44 c55 − c2 45 π
−1 2
1
Im (cos θ + s3 sin θ) 2
(6.127a) 2 3 3 λj − λ2 cos(λj log(r)) + + sin(λj log(r)) F (θ) (6. For an inﬁnite plate with a central crack in an anisotropic body the derivation for the stress functions was undertaken by Sih.109b.128b) π s1 − s2 (6. 6. Paris and Irwin.128f)
(6.128i) Fracture Mechanics
.126a) F (θ) = f (θ)g(θ) + 2f (θ)g (θ) + f (θ)g (θ) 3θ 1 θ 3θ 1 θ 9 3 − sin cos + cos + 2λj sinh λj θ = a cosh λj θ − sin 4 2 4 2 2 2 2 2 3θ θ +λ2 cosh2 λj θ sin + + sin j 2 2 3θ 3 θ 3θ 3 θ 9 3 − cos − sin + 2λj sinh λj θ − sin b cosh λj θ − cos 4 2 4 2 2 2 2 2 θ 3θ + 3 cos +λ2 cosh2 λj θ cos (6.128e) π s1 − s2 (6.128a) π s1 − s2 1 1 2r 1 Re s1 q2 (cos θ + s2 sin θ) 2 − s2 q1 (cos θ + s1 sin θ) 2 (6.7
102
Homogeneous Anisotropic Material (Sih and Paris)
To analyze an anisotropic body with with a crack. 6. we need to derive the two stress functions Φ1 and Φ2 in Eq. Paris and Irwin 1965). This solution is the “counterpart” or generalization of Westergaard’s solutions.
An important observation to be made is that the form of the stress singularity r −1/2 is identical to the one found in isotropic solids. in general complex.
109
6.8 Assignment
25
where s1 and s2 are roots. 2. and the roots of interests are taken such that βj > 0. 5.130)
105 After appropriate substitution. and pj = a11 s2 + a12 − a16 sj j qj = a12 sj + a22 − a26 sj (6.132c)
107 For inplane loadings.132a) (6. for plane skewsymmetric loading: σx = σy = σxy = K 1 √ II Re s1 − s2 2πr K 1 √ II Re s1 − s2 2πr K 1 √ II Re s1 − s2 2πr s2 2 (cos θ + s2 sin θ) 2 1 (cos θ + s2 sin θ) s1 (cos θ + s1 sin θ)
1 2 1 2 1
− − −
s2 1 (cos θ + s1 sin θ) 2 1 (cos θ + s1 sin θ) 2 s2 (cos θ + s2 sin θ) 2
1 1 1
(6.Draft
104
6. in the anisotropic case they will also depend on the material elastic properties and the orientation of the crack with respect to the principal planes of elastic symmetry (through s1 and s2 ).157 where sj = αj + iβj for j = 1.132b) (6.129) (6. of Eq.131a) (6.
108
It should be noted that contrarily to the isotropic case where both the stress magnitude and its spatial distribution are controlled by the stress intensity factor only.8
Assignment CVEN6831 FRACTURE MECHANICS
Victor Saouma
Fracture Mechanics
.131c)
106
and. these stresses can be summed to give the stresses at a distance r and an angle θ from the crack tip.131b) (6. it can be shown that the cartesian stresses at the tip of the crack for symmetric loading are
σx = σy = σxy =
K s1 s2 √ I Re s1 − s2 2πr K 1 √ I Re s1 − s2 2πr K s1 s2 √ I Re s1 − s2 2πr
s2 (cos θ + s2 sin θ) s1 (cos θ + s2 sin θ) 1 (cos θ + s1 sin θ)
1 2 1 2 1 2
− − −
s1 (cos θ + s1 sin θ) 2 s2 (cos θ + s1 sin θ) 2 s1 (cos θ + s2 sin θ) 2
1 1 1
(6.
1
Derivation of SIF
Due Date: Feb. 15. determine the stress intensity factors at the crack tip. (KP = √P ) πa φ= Hint: (a) At crack tip (b) z = reiθ
η a
<< 1.Draft
26
ELASTICITY BASED SOLUTIONS FOR CRACK PROBLEMS
Spring 2000 Victor E. When the vessel is subjected to an internal pressure p.
σz σθ
σθ
β
σz t R
2a
σz
σθ
2. The stress function for a crack subjected to splitting forces P.
= η+a Fracture Mechanics
Victor Saouma
. Calculate the expressions of σ22 at x2 = 0 and derive an expression for KI .133) πz z 2 − a2 where z = x1 + ix2 and P is a load per unit thickness. as shown below
x2
a P
x1 P 2a
is given by: Pa √ (6. A cylindrical pressure vessel of radius R and thickness t contains a through crack of lenght 2a oriented at an angle β with the circumferential direction. 2000 1. Saouma
HomeWork No.
5. Using Mathematica. 4.
σy c 2(a .Draft
6.135)
Based on those expressions.
Victor Saouma
Fracture Mechanics
. program either: (a) Westergaard’s solution for a crack subjected to mode I and mode II loading.8 Assignment 3.134) (6.c)
σy c
Using Mathematica and the expressions of KA and KB from the previous problem. The stress intensity factor of the following problem:
x2
27
x P B P a a A x1
is given by: P KA = √ πa P Kb = √ πa a+x a−x a−x a+x (6. determine the stress function Φ. determine an expression for the stress intensity factors for this case. (b) Williams solution for a crack along dissimilar materials. Barenblatt’s model assumes a linearly varying closing pressure at the tip of a crack. and results from the previous problem.
Draft
28
ELASTICITY BASED SOLUTIONS FOR CRACK PROBLEMS
Victor Saouma
Fracture Mechanics
.
Flaw size: 1 . The design variables are:
Material properties: (such as special steel to resist corrosive liquid) ⇒ Kc is ﬁxed.
. Finally.
9
Thus a crack will propagate (under pure mode I). at the point of incipient crack growth: √ (7.1) KIc = βσ πa
10 11 Thus for the design of a cracked. a.
1
In most cases. as only. Design stress level: (which may be governed by weight considerations) ⇒ σ is ﬁxed. the engineer would have to decide what design variables can be selected. structure. or potentially cracked.
7
First we shall examine how is linear elastic fracture mechanics (LEFM) eﬀectively used in design examples. under the assumptions of linear elastic fracture mechanics (LEFM).1
Design Philosophy Based on Linear Elastic Fracture Mechanics
One of the underlying principles of fracture mechanics is that unstable fracture occurs when the stress intensity factor (SIF) reaches a critical value KIc .Draft
Chapter 7
LEFM DESIGN EXAMPLES
Following the detailed coverage of the derivation of the linear elastic stress ﬁeld around a crack tip. we now seek to apply those equations to some (pure mode I) practical design problems. Hence. also called fracture toughness. whenever the stress intensity factor KI (which characterizes the strength of the singularity for a given problem) reaches a material constant KIc . a refers to half the total crack length. two of these variables can be ﬁxed.
8
7. then we shall give analytical solutions to some simple commonly used test geometries. followed by a tabulation of fracture toughness of commonly used engineering materials. and the third must be determined. and the introduction of the concept of a stress intensity factor in the preceding chapter. this chapter will conclude with some simple design/analysis examples. KIc represents the inherent ability of a material to withstand a given stress ﬁeld intensity at the tip of a crack and to resist progressive tensile crack extension.
Finally. it is essential that the crack length a be properly known. simpliﬁed inspection would result in larger crack size assumptions. occasionally.2
12
Stress Intensity Factors
As shown in the preceding chapter. it should be mentioned that whereas in most cases the geometry is ﬁxed (hence β).
12 13 Thus. 1973). and those assumptions are dependent upon the crack detection methodology adopted.
π sec πa = 1 as anticipated.
Because of their importance. a number of simple problems have been solved and their analytic solution is found in stress intensity factor handbooks. The most commonly referenced ones are Tada.
7. Fig.2) √ σ πa
=
1 + 0. the third one is ﬁxed. 7.2
a W
3
(7.256
a W
2
+ 12. In most cases it is not. (Rooke and Cartwright 1976). equal to the smallest one that can be detected.152
β
(7. expressions of SIF of commonly encountered geometries will be listed below:
14
Middle Tension Panel (MT). Thus assumptions must be made for its value. a simpler inspection procedure would result in a larger minimum crack size than one detected by expensive nondestructive techniques. Paris and Irwin’s (Tada et al.
Once two parameters are speciﬁed. This explain the interest developed by some mathematician in solving fracture related problems.3)
We note that for W very large with respect to a. W Fracture Mechanics
Victor Saouma
. Fortunately. and Murakami (Murakami 1987)
13 In addition. The presence of a crack.Draft
2
LEFM DESIGN EXAMPLES
Figure 7.1 KI = sec
β
πa √ σ πa W a W − 1. In return. analytic derivation of the stress intensity factors of even the simplest problem can be quite challenging. must be assumed. specially in conjunction with fatigue life predictions. and Roorke and Cartwright.1: Middle Tension Panel In assessing the safety of a cracked body. there is the possibility to alter it in such a way to reduce (or maximize) β. increasingly computer software with preprogrammed analytical solutions are becoming available.
d.56 a W
2 L W
= 2.15 − 3.23 a W + 10.3 KI = 1.5)
Three Point Bend (TPB).12 − 0.7)
We note that this is not exactly the equation found in the ASTM standard. Fig.7 − 104.) Standard Test Method for PlaneStrain Fracture Toughness of Metallic Materials KI = 16. Fig.43 a W − 4.12.2: Single Edge Notch Tension Panel
Figure 7.42
a W
4
√ σ πa
(7. but rather an equivalent one written in the standard form.5 used in ASTM E399 (Anon. 7.6 a W + 370 a W
β 2
− 574
a W
3
+ 361
a W
4
P √ πa BW
σ
(7.
1) is grater than one and is
Double Edge Notch Tension Panel (DENT).93 W + 2.99 −
a W
1− 2 1+
a W a 2W
a 2.4 3 KI =
a W
1.3: Double Edge Notch Tension Panel Single Edge Notch Tension Panel (SENT) for KI = 1. Victor Saouma Fracture Mechanics
.7
a 2 W
PS BW 2
3
1−
a W
3 2
(7.6)
Compact Tension Specimen (CTS).46
a W
3
√ σ πa
(7.79
β
a W
2
+ 15.2 a W
3
− 21. 7. 7. n.4)
β a We observe that here the β factor for small crack ( W approximately 1. Fig. Fig. 7.Draft
7.74
+ 30.12 + 0.2 Stress Intensity Factors
3
Figure 7.
5: Compact Tension Specimen
Victor Saouma
Fracture Mechanics
.Draft
4
LEFM DESIGN EXAMPLES
Figure 7.4: Three Point Bend Beam
Figure 7.
and assume that we have a central crack with an eﬀective length aef f = 2a+D .7 we obtain: KI = σ π = a+D 2 (7.9)
1+
β
Similarly. in this case. β remains equal to 3. if we had only one single crack radiating from a hole.10)
D √ 1 + σ πa 2 2a
β
Victor Saouma
Fracture Mechanics
.36σ πa
Long Crack D 2a + D. 7.6.2 Stress Intensity Factors
5
Figure 7. Fig. whereas for long crack. we know that there is a stress concentration factor of 3.12. thus 2 KI = σ π = 2a + D 2 D √ σ πa 2a (7.Draft
7. 7.36. for a crack radiating from this hole. and for an edge crack β = 1. Fig. we consider two cases
a Short Crack: D → 0.8) = 3. and thus we have an approximate far ﬁeld stress of 3σ.6 thus √ KI = 1. 7. then we will present the exact one: Approximate: For a plate with a far ﬁeld uniform stress σ.12(3σ) πa √ (7. Fig.6: Approximate Solutions for Two Opposite Short Cracks Radiating from a Circular Hole in an Inﬁnite Plate under Tension Circular Holes: First let us consider the approximate solution of this problem. we can for all practical purposes ignore the presence of the hole. for short crack.
7. (WHY?) Victor Saouma Fracture Mechanics
. Fig. Fig. We note that for the pressurized hole only. using Newman’s solution β is given in Table 7. 7. hence we would have a stable crack growth.1 Pressurized Hole with Radiating Cracks: Again we will use Newman’s solution for this problem. and distinguish two cases: Pressurized Hole Only: or λ = 0.7: Approximate Solutions for Long Cracks Radiating from a Circular Hole in an Inﬁnite Plate under Tension
Figure 7. We also note that KI would be the same for a pressurized crack and borehole.8 √ KI = βσ πa where.8: Radiating Cracks from a Circular Hole in an Inﬁnite Plate under Biaxial Stress Exact: Whereas the preceding equations give accurate results for both very short and very large cracks.12) (7. KI decreases with crack length. as it would have been for an unpressurized hole but an identical far ﬁeld stress.1.13) (7.11)
For both cases. in the intermediary stage an exact numerical solution was derived by Newman (Newman 1971). β is given in Table 7.Draft
6
LEFM DESIGN EXAMPLES
Figure 7.9 2pR KI = β √ πa Pressurized Hole and Crack: or λ = 1 √ KI = βp πa (7.
0135 .0775 .9768 1.1725 .9670 1.0495 .4325 0.3979 .9029 .0077 .1149 1.1899 1.6025 .3058 .99267 1.0904 1.7585 .0168 .9976 .04 1.2134 1.7494 1.8723 .4514 .50 3.15 1.00 4.6898 .6701 .1: Newman’s Solution for Circular Hole in an Inﬁnite Plate subjected to Biaxial Loading.0161 .8259 1.6025 1.2457 1.08 1.7981 .01 1.9976
β Pressurized Hole λ=1 λ=0 .20 1.9322 .20 2.4183 .9855 .1476 1.2405 1.00
β Biaxial Stress λ = −1 λ = 1 λ = 0 0.3058 .9: Pressurized Hole with Radiating Cracks
Victor Saouma
Fracture Mechanics
.00 2.4485 .8723 1.40 1.8957 . and Internal Pressure
Figure 7.4897 .6898 1.0571 .2319 .2350 1.80 2.5551 1.0358 .9855 1.9768 .30 1.3256 0.02 1.7929 .Draft
7.7494 .6262 .0395 1.7104 .9670 .9242 1.5551 .2188 .0409 .9029 1.2208 .9764
Table 7.2188 .0649 1.8677 .9851 .10 1.9566 .50 1.9513 1.9250 .99267 .0178 1.9242 .4958 1.0252 .06 1.0536 .5971 .9154 .6082 .7843 .4183 .8264 .8259 .7053 .1746 .7971 .3334 .5688 .60 1.7929 1.8400 .9358 .2 Stress Intensity Factors
7
a R
1.4958 .9513 .25 1.0336 .0582 .
4 + 1.14) (7. 7. as it provides the Green’s function for numerous other ones.11: Two Opposite Point Loads acting on the Surface of an Edge Crack Point Load Acting on Crack Surfaces of an Embedded Crack: The solution of this problem.10 and the subsequent one.2 Embedded Elliptical Crack A large number of naturally occurring defects are present as embedded. (Irwin 1962) proposed
Victor Saouma
Fracture Mechanics
. is of great practical importance.3 (7.Draft
8
LEFM DESIGN EXAMPLES
Figure 7.16) KI = πa 1 + x 2 a
a
where C is tabulated in Table 7. Fig.10: Two Opposite Point Loads acting on the Surface of an Embedded Crack
Figure 7.
A KI =
P πa P πa
a+x a−x a−x a+x
(7. surface or corner cracks (such as ﬁllet welding) Irwin.11 is C 2P x 2 −0. Fig. 7.15)
B KI =
Point Load Acting on Crack Surfaces of an Edge Crack: The solution of this problem.
80.9 > 0.9
C 1 1. Corner.07 1.6 0.Draft
7. and Surface Cracks
Victor Saouma
Fracture Mechanics
.01 1.60.12: Embedded.11
Table 7.7 0.8 0.03 1.2 Stress Intensity Factors
9
x a
< 0.70.2: C Factors for Point Load on Edge Crack
Figure 7.
θ =
the stress intensity factor is maximum: √ σ πb = =σ Φ0 πb Q (7. At the end of the major axes.21)
√ then we retrieve the solution KI = σ πa of a through crack.22)
(KI )θ= π 2
4.1 + 0.23)
Thus an embedded elliptical crack will propagate into a circular one “pennyshaped”.17)
where Φ0 is a complete elliptical integral of the second kind Φ0 = = 1− a2 − b2 sin2 θdθ a2 (7. Newman and Raju (Newman and Raju 1981) developed an empirical SIF equation for semielliptical surface cracks.464 a b t
2
1 1. At the end of the minor axes. with x = a cos θ and y = b sin θ: b2 sin2 θ + 2 cos2 θ a
1 4
√ σ πb
(7. Using the results of three dimensional ﬁnite element analysis. 7.24)
Fracture Mechanics
.17 reduces to KI = 2. and to surface cracks emanating from circular holes.17 was given by Cherepanov (Cherepanov 1979) KI = sin2 θ + for 0 ≤
b a
b a
2
1 4
cos2 θ
√ σ πb
(7.
π 2
3.20)
≤ 1. If a = ∞ & θ =
π 2
2 √ σ πa π
(7. 7.19)
0
Q
An approximation to Eq. 7.Draft
10 1 KI (θ) = Φ0
π 2
LEFM DESIGN EXAMPLES
the following solution for the elliptical crack.35
(1 − sin θ)2
(7.65 − 2
cos θ + sin θ
1 + 0. The equation applies for cracks of arbitrary shape factor in ﬁnite size plates for both tension and bending loads. Fig. quarter elliptical corner cracks. This is perhaps the most accurate solution and is almost universally used: K = σ πb M1 + M2 b a Victor Saouma
2 2
√
b t
2
2
+ M3
1 4
b t
4
b 1 + 1. If a = b then we have a “pennyshape’ circular crack and Eq.13. Surface Cracks Irwin’s original solution has been extended to semielliptical surface ﬂaws.
This solution calls for the following observations: 1.18) (7. θ = 0 the stress intensity factor is minimum (KI )θ=0 = σ π ba Φ0
2
(7.
08 a For very long cracks
b a
√
b 1 + 1.
Metallic Alloys: 16 Testing procedures for fracture toughness of metallic alloys are standardized by various codes (see (Anon. An exhaustive tabulation of fracture toughnesses of numerous alloys can be found in (Hudson and Seward 1978) and (Hudson and Seward 1982).d. Equation 7.29)
7.0 and b ≤ 0.26) b a
24
M3 = 0.13σ πb 1 − . Although there is not yet a standard for fracture toughness of concrete. −
(7. BS 5447.65 +
+ 14 1.09 M2 = 0. but there has been much recent interest in its applicability to both concrete and rocks.3
Fracture Properties of Materials
15 Whereas fracture toughness testing will be the object of a separate chapter. Furthermore. For shallow cracks b t b K = 1.) and (British Standards Institution.25.65 − 2
(7. Concrete: Fracture mechanics evolved primarily from mechanical and metallurgical applications.46 t
+ 11.13: Elliptical Crack.5 − 0. Victor Saouma Fracture Mechanics
. For a approximately equal to 0.13σ πb 1 + 3. Equation 7. London 1977)).24 reduces to of θ.Draft
7. 1985) has proposed a standard procedure for determining GF . Hillerborg (Anon.464 a
1 1. a subcommittee of ASTM E399 is currently looking into a proposed testing procedure for concrete. n.5
b t
4
(7. and Newman’s Solution M1 = 1.8. K is roughly independent t 1. provided b b 0 < a ≤ 1.54
−1
(7.28)
1.24 reduces to
2
√ b K = 1. we shall brieﬂy mention the appropriate references from where fracture toughness values can be determined.13 − 0.2 + b a b a b a
(7.89 0.3 Fracture Properties of Materials
11
Figure 7.25)
−1
− 0.27)
Newman and Raju report that this equation is accurate within ±5 percent.
4.45C − Ni − Cr − Mo steel. with a decreasing fracture toughness with increase in yield stress.099Mpa m
7.45C − Ni − Cr − Mo Steel Rock: Ouchterlony has a comprehensive review of fracture toughnesses of numerous rocks in an appendix of (Ouchterlony 1986). Table 7.3 provides an indication of the fracture toughness of common engineering materials.5 1 3
Table 7. Is this reasonable?
Victor Saouma
Fracture Mechanics
. The speciﬁed design σ stress is 2y . Pressure Vessel Hardened Steel Aluminum Titanium Copper Lead Glass Westerly Granite Cement Paste Concrete Nylon
LEFM DESIGN EXAMPLES KIc √ ksi in 49 190 20 2030 70 100 18 0.4: Fracture Toughness vs Yield Stress for . Table 7. To save weight. The smallest crack size (2a) which can be detected is approximately .12 in.3: Approximate Fracture Toughness of Common Engineering Materials Yield Stress Ksi 210 220 230 240 290 300 KIc √ ksi in 65 60 40 38 35 30
Table 7. Medium Carbon Steel. an increase of tensile strength from 220 ksi to 300 ksi is suggested.4
7.Draft
12 Material Steel.30) 1ksi in = 1.7 16 0.1
Examples
Example 1
Assume that a component in the shape of a large sheet is to be fabricated from . and a proposed fracture toughness testing procedure can be found in (Ouchterlony 1982). and that √ √ (7. √ Note that stress intensity factors in metric units are commonly expressed in Mpa m.4.
σy should not be raised to 300 ksi. .127 + 3. Independent tests gave 40.07 in. with a potential factor of safety of one against cracking (we can not be sure 0.07 2
−1
− 0.126
2
. .65 + = 4.126
4
.24. Hence.126 in.07 . First we use the exact solution as given in Eq.0 inch long by (probably) . Afterwards.4 Examples
13
√ √ At 220 ksi KIc = 60 ksi in. Thus.K. the total crack length is larger than the smallest one which can be detected (which is O.89 0. there will be a two fold increase in weight. cracks approximately 4. for the second case the total critical crack size is approximately ﬁve times smaller than the minimum ﬂaw size required and approximately eight times smaller than the ﬂaw size tolerated at the 220 ksi level. in.07 2. The question is whether the cracks were critical for the 610 psi pressure? 15 in For a cylinder under internal pressure. the hoop stress is σ = pD = 610 2lb 2(.13 − 0.3 π. We observe that since the design stress level is approximately half of that of the weaker alloy.07 1 + 1.464 2
1 1.2 + = 3.
7.Draft
Thus.126) in = 2t in 36. with a = 2.2
Example 2
A small beer barrel of diameter 15” and wall thickness of .06π 100% that there is no crack of that size or smaller as we can not detect it).4. 7.65 − 2
Fracture Mechanics
.3 ksi. Alternatively. and at 300 ksi KIc = 30 ksi in.127 M2 = 0.09 = 1.189 .0255
We observe that for the ﬁrst case.126” made of aluminum alloy exploded when a pressure reduction valve malfunctioned and the barrel experienced the 610 psi full pressure of the CO2 cylinder supplying it with gas.07 + 4.994 Substituting √ K = 36. we should have a √ decrease in design stress (since KIc and acr are now set) KIc = σd πavis ⇒ σd = √KIc = πavis √ 30ksi in √ = 69 ksi.).0947 .07 2
24
.07 inch deep were discovered on the inside of the salvaged√ pieces of the barrel (it was impossible to measure their depth).994 . 310 psi = 36.07 2. This can be used as the far ﬁeld stress (neglecting curvature).
7.07 1. the design stress will 2 √ σ 1 be given by σd = 2y and from KIc = σd πacr ⇒ acr = π KI c σy
2
Yield Stress σy 220 300
Design Stress σd 110 150
Fracture Toughness KIc 60 30
Critical Crack acr .0127
Total Crack 2acr . and t = .247 M3 = 0. Finally. −
.54
−1
+ 14 1. ksi in for KIc of the aluminum alloy.5 − 0. upon substitution we obtain: M1 = 1. b = . if we wanted to use the ﬂaw size found with the 300 ksi alloy.247 Victor Saouma .
126
4
7.07 2.14: Growth of Semielliptical surface Flaw into Semicircular Conﬁguration . This in turn will result in a sudden depressurization. Leak Before Fail: In this case.07 .1
16
Additional Design Considerations
Leak Before Fail
As observed from the preceding example.35
√ = 44.85 > KIc .
2
1 4
0+1
1 + 0.Draft
14
2a=2t t
LEFM DESIGN EXAMPLES
Semicircular final crack (leak)
2c Initial Elliptical Crack
Figure 7. crack growth occur. 7. as this would usually be immediately noticed and corrrected (assuming that there is no leak of ﬂamable gas!).126
2
+ 11.13)(36.5.07 . Two scenarios may happen.5
.46 = 60. pressurized vessels should be designed to ensure a leak before fail failure scenario.
17
Hence.07) 1 + 3.5
7. and this will stop any further crack gowth.07 . Fig. and there is a sudden and unstable crack growth.
Victor Saouma
Fracture Mechanics
.14 Breakthrough: In this case critical crack conﬁguration is reached before the crack has “daylighted”. and the crack “pierces” through the thickness of the vessel before unstable crack growth occurs. for long cracks we would have obtained: K = (1.1 + 0.2ksi in
.126
2
(1 − 1)2
This is about equal to the fracture toughness. many pressurized vessels are subject to crack growth if internal ﬂaws are present.3) π(. Note that if we were to use the approximate equation.
2
7.Draft
7. or Residual Strength Diagram 2. it should be noted that leak before break assessment should be made on the basis of a complete residual strength diagram for both the part through and the through crack. Determine crack growth with time. It can also be applied to establish a fracture control plan. resulting in Crack Growth Curve. This will result in a plot of crack size versus residual strength.5 Additional Design Considerations
15
18 Finally. Determine the eﬀect of cracks on strength. or stress combination.
Victor Saouma
Fracture Mechanics
.5. Various ratios should be considered
Damage Tolerance Assessment
19 Fracture mechanics is not limited to determining the critical crack size. or damage tolerance analysis with the following objectives:
1. load.
Draft
16
LEFM DESIGN EXAMPLES
Victor Saouma
Fracture Mechanics
.
(Griﬃth 1921). hence the surface energy is stronger.1. and diamonds are approximately .1.1
Tensile Strength
Ideal Strength in Terms of Physical Parameters
We shall ﬁrst derive an expression for the ideal strength in terms of physical parameters. This derivation. The total energy which must be supplied to separate atom C from C’ is (8. we shall develop an expression for the theoretical strength of perfect crystals (theoretically the strongest form of solid). 8. The reason why we do not notice it is that solids are too rigid to be distorted by it.Draft
Chapter 8
THEORETICAL STRENGTH of SOLIDS. and in the next section the strength will be expressed in terms of engineering ones. Fig. most solids. When the surface of a liquid is extended (soap bubble. he had postulated that this can be explained by the presence of internal ﬂaws (idealized as elliptical) and then used Inglis solution to explain this discrepancy. (Kelly 1974) is fundamentally diﬀerent than the one of Griﬃth as it starts at the atomic level. When insects walk on water it sinks until the surface energy just balances the decrease in its potential energy.
From watching raindrops and bubbles it is obvious that liquid water has surface tension. insect walking on liquid) work is done against this tension.1.14 respectively. and the factor of 2 is due to the fact that upon separation. we have two distinct surfaces. As such. the chemical bonds are stronger than for liquids.
1
. and 5.0.
8.
In this section. and energy is stored in the new surface.1
8. 1. Surface energy γ is expressed in J/m2 and the surface energies of water.1
Derivation
9 We start by exploring the energy of interaction between two adjacent atoms at equilibrium separated by a distance a0 . (Griﬃth I)
7 We recall that Griﬃth’s involvement with fracture mechanics started as he was exploring the disparity in strength between glass rods of diﬀerent sizes. For solids.
8
8.1) U0 = 2γ
where γ is the surface energy1 .077.1.
Draft
2
THEORETICAL STRENGTH of SOLIDS. (Griﬃth I)
Figure 8.2: Energy and Force Binding Two Adjacent Atoms
Victor Saouma
Fracture Mechanics
.1: Uniformly Stressed Layer of Atoms Separated by a0
Solution I: Figure 8.
thus F = 0 at a = a0 .Draft
8. we would have
2γ = U0 = =
x dx λ 0 2πx λ λ theor 2 σmax [− cos ( )] 0 2π λ
theor σmax sin 2π −1 1
λ 2
(8.2) σ = σmax sin 2π λ
theor and the maximum stress σmax would occur at x = λ .1 Derivation
3
Figure 8. then the σ − ε curve will be as shown in Fig. Hence. it would appear that the sine curve would be an adequate approximation to this relationship. if we F deﬁne the stress as σ = a2 . we have F = dU .6)
Also for very small displacements (small x) sin x ≈ x.2 reduces to
theor σ ≈ σmax
x 2πx ≈E λ a0 E λ a0 2π
(8. Fig.3: Stress Strain Relation at the Atomic Level Force being the derivative of energy. we get
theor σmax ≈
Eγ a0
(8.1. 8. Furthermore. and from Eq. thus Eq. Hence. 8. From this 0 diagram. then the strain would be equal to ε = a0 .6. da and is maximum at the inﬂection point of the U0 − a curve.7)
elliminating x. the slope of the force displacement curve is the stiﬀness of the atomic spring and should be related to x E. x theor (8. 8. 8.5) (8. The energy required to separate 4 two atoms is thus given by the area under the sine curve. If we let x = a − a0 .2.9)
Victor Saouma
Fracture Mechanics
.
theor σmax ≈
(8.4)
= ⇒λ =
2πλ λ theor σmax [− cos ( ) + cos(0)] 2π 2λ 2γπ theor σmax
(8.3) (8.3. 8.8)
Substituting for λ from Eq.
1.14)
Thus this would be the ideal theoretical strength of steel. let us consider steel which has the following properties: γ = 1 m2 . since as a ﬁrst order approximation a ≈ a0 then the surface energy will be γ≈ Ea0 10 (8.8 would yield Ea theor (8.15) σmax ≈ a0 π
10
Alternatively combining Eq. we can simplify this approximation to:
theor σmax =
Eγ a0
(8. the strain energy per unit area due to σ (for linear elastic systems) is U σ = 1 σεao 2 = Eε U= σ 2 ao 2E (8.2 Ideal Strength in Terms of Engineering Parameter
We note that the force to separate two atoms drops to zero when the distance between them is a0 + a where a0 corresponds to the origin and a to λ . 8. Thus. 8.Draft
4
THEORETICAL STRENGTH of SOLIDS.17)
12
However.18) Fracture Mechanics
Victor Saouma
. if we take a = λ or λ = 2a. 8.16)
11
(8.16 × 1010 2 m E ≈ 6
(8.13) (8. Thus from Eq.9
J Example: As an example.6 with λ = 2a gives a≈ Combining those two equations γ≈ E a0 a π
2
γπ
theor σmax
(8. For our theoretical strength. U = 2γ ⇒
theor (σmax )2 a0 2E
theor = 2γ or σmax = 2
γE a0
Note that here we have assumed that the material obeys Hooke’s Law up to failure. then the total surface energy of two new fracture surfaces is 2γ.11)
which is the same as Equation 8. (Griﬃth I)
Solution II: For two layers of atoms a0 apart.10)
If γ is the surface energy of the solid per unit area. 8. since this is seldom the case. 2 2 combined with Eq.9 we would have: theor σmax ≈
(2 × 1011 )(1) 2 × 10−10 N ≈ 3. and a0 ≈ 2 × 10−10 m. N E = 2 × 1011 m2 .12) (8.1.
we can assume that the shear stress is
theor τ = τmax sin 2π
x λ
(8. Griﬃth was exploring the theoretical strength of solids by performing a series of experiments on glass rods of various diameters.9 will ﬁnally give E theor σmax ≈ √ 10 (8. combined with Eq.21)
Because we do have very small displacement. What happen if we slide the top row over the bottom one.4: Inﬂuence of Atomic Misﬁt on Ideal Shear Strength
13
This equation.
8.23)
8. He observed that the tensile strength
Victor Saouma
Fracture Mechanics
.19)
which is an approximate expression for the theoretical maximum strength in terms of E. we can elliminate x from
theor τmax sin 2π
x λ
2πx x = γG = G λ h Gλ theor ⇒ τmax = 2πh
(8.4 γxy = x/h. Again.Draft
8.2 Griﬃth Theory
5
h
x
Figure 8. Fig. 8. then
theor τmax
E 12(1 + ν)
E 18
(8. and G = E/2(1 + ν).2
14
Shear Strength
Similar derivation can be done for shear.2
Griﬃth Theory
17 Around 1920.22a)
16
If we do also assume that λ = h. 8.
15
(8.20)
and from basic elasticity τ = Gγxy and.1.
we obtain
theor act σmax = 2σcr
a a0 macro
(8. 500 a0
2
act σcr =
E2 100.Draft
6
18
THEORETICAL STRENGTH of SOLIDS. by extrapolation to “zero” diameter he obtained a theoretical maximum strength of approximately 1. a Asssuming ρ ≈ a0 and since 2 a0 1. Griﬃth postulated that the theoretical strength can only be reached at the point of highest stress concentration.000
=
E √ 100 10
(8.24)
act σ is the stress at the tip of the ellipse which is caused by a (lower) far ﬁeld stress σcr .9. and his “strike of genius” was to assume that strength is reduced due to the presence of internal ﬂaws.
Griﬃth had thus demonstrated that the theoretical strength could be experimentally approached. The left hand side is based on a linear elastic solution of a macroscopic problem solved by Inglis. 000 psi. The right hand side is based on the theoretical strength derived from the sinusoidal stressstrain assumption of the interatomic forces.
23
Finally. this equation would give (at fracture)
act σcr =
Eγ 4a
(8. 8. and ﬁnds its roots in microphysics.000 psi. 2.26)
Macro
22
From this very important equation. for an ideal plate under tension with only one single elliptical ﬂaw the strength may be obtained from
21
theor act σmax = 2σcr
micro hence.2. assuming an elliptical imperfection.28)
Victor Saouma
Fracture Mechanics
. and from equation 6. 000a0
act σcr = γ = Eγ 4a Ea0 10 E ao act σcr = 40 a a = 2.
8.27)
24
As an example. we observe that 1. and on the other hand for very large diameters the asymptotic values was around 25. he now needed to show why the great majority of solids fell so far below it. (Griﬃth I)
1 (σt ) of glass decreased with an increase in diameter. furthermore. and that for a diameter φ ≈ 10.000 psi. equating with Eq..31
theor act σmax = σcr 1 + 2
a ρ
(8. he came across Inglis’s paper.25)
a = ao
Eγ a0 Micro
(8. 20
Hence. and accordingly the farﬁeld applied stress will be much smaller. let us consider a ﬂaw with a size of 2a = 5.000 in. σt = 500.600.1
Derivation
19 In his quest for an explanation.
29)
Therefore at failure
act = σcr theor σmax =
theor σmax 200 E 10
act σcr ≈
E 2.Draft
25 26
8.000
=
30.30)
which can be attained. or a factor of 100. 000a0 in γ ≈ Ea0 this is enough to lower the theoretical 10 E √ fracture strength from √10 to a critical value of magnitude 100E 10 .000
Victor Saouma
Fracture Mechanics
. 000 = 15 ksi
(8. For instance for steel
E 2. Also
theor act σmax = 2σcr act σ theor = 2σcr = 10−6 m = 1µm max = 1˚ = ρ = 10−10 m A a ao
a ao
27
10−6 act = 200σcr 10−10
(8.2 Griﬃth Theory
7
Thus if we set a ﬂaw size of 2a = 5.000 2.
(Griﬃth I)
Victor Saouma
Fracture Mechanics
.Draft
8
THEORETICAL STRENGTH of SOLIDS.
the Griﬃth model for elastic solids. is one based on energy transfer (or release). This dual approach will be developed in this chapter. 11 Thus. and the subsequent one by Irwin and Orowan for elasticplastic solids. 10 Hence. Hence energy is neither created nor consumed. 9 Griﬃth’s main achievement. sum of the kinetic energy and the internal energy) is equal to the sum of the rate of work done by
15
.. which occurs during crack propagation. in providing a basis for the fracture strengths of bodies containing cracks. but rather because a certain energy criteria was met.e. 12
It should be noted that this is a global energy approach. (Griﬃth II)
7 In the preceding chapters.
14
The ﬁrst law of thermodynamics states The timerate of change of the total energy (i.1
Thermodynamics of Crack Growth
13 If we consider a crack in a deformable continuum aubjected to arbitrary loading. Since only the change of energy is involved.Draft
Chapter 9
ENERGY TRANSFER in CRACK GROWTH. On this basis. show that crack propagation is caused by a transfer of energy transfer from external work and/or strain energy to surface energy. 8 An alternative to this approach. a criteria for crack propagation. was his realization that it was possible to derive a thermodynamic criterion for fracture by considering the total change in energy of a cracked body as the crack length increases.
9. we have focused on the singular stress ﬁeld around a crack tip. which was developed prior to the one of Westergaard which focused on the stress ﬁeld surrounding a crack tip. (Griﬃth 1921). then the ﬁrst law of thermodynamics gives:
The change in energy is proportional to the amount of work performed. any datum can be used as a basis for measure of energy. It will be shown later that for linear elastic solids the two approaches are identical. Griﬃth showed that material fail not because of a maximum stress. based on the strength of the singularity was ﬁrst developed and then used in practical problems.
21 This equation represents the fracture criterion for crack growth. U the total internal strain energy (elastic plus plastic). Griﬃth’s Model
For a perfectly brittle material. 9.5)
that is the rate of potential energy decrease during crack growth is equal to the rate of energy dissipated in plastic deformation and crack growth. The left hand side represents the energy available for crack growth and is given the symbol g in honor of Griﬃth. and for a unit thickness we can replace A by a. then Q and K are equal to zero. It indicates that the work rate supplied to the continuum by the applied external loads is equal to the rate of strain energy (elastic and plastic) plus the energy dissipated during crack propagation. Since all changes with respect to time are caused by changes in crack size.6)
the factor 2 appears because we have two material surfaces upon fracture. it is often referred to as the crack driving force. It is very important to observe that the energy scales with a2 . whereas surface energy scales with a. then we can rewrite the ﬁrst law as ∂W = ∂a ∂U e ∂U p + ∂a ∂a + ∂Γ ∂a (9. two limiting cases can be considered.1
20
Brittle Material.Draft
2
16
ENERGY TRANSFER in CRACK GROWTH. we can write ∂A ∂ ∂ = ∂t ∂t ∂A (9. 9. 18
Thus Π = Ue − W ∂U p ∂Γ ∂Π = + − ∂a ∂a ∂a (9. It will be shown later that this can have serious implication on the stability of cracks. and on size eﬀects.1)
where K is the kinetic energy. They will be examined in conjunction with Fig. and Q the heat input to the system. Γ the surface energy.4) (9.
19
9. and is a material constant (related to the toughness). we can rewrite Eq. (Griﬃth II)
the external forces and the change of heat content per unit time: d (K + U + Γ) = W + Q dt (9.3 as G = −
def
∂W ∂U e ∂Γ ∂Π = − = = 2γ ∂a ∂a ∂a ∂a
(9.1 in which we have a crack of
Victor Saouma
Fracture Mechanics
.3)
17 This equation represents the energy balance during crack growth. Because G is derived from a potential function.2)
and for an adiabatic system (no heat exchange) and if loads are applied in a quasi static manner (no kinetic energy).1. The right hand side represents the resistance of the material that must be overcome for crack growth. W the external work.
Here there is both external work (9.Draft
9.7) (9.12) ∆W = P1 (u2 − u1 ) and a release of internal strain energy.13) Fracture Mechanics
. so the decrease in potential energy is the same as the decrease in stored internal strain energy. an increase in crack length from a to a + da results in a decrease in stored elastic strain energy.8) (9.11)
Fixed Load: P2 = P1 the situation is slightly more complicated. ∆U = 1 1 P2 u1 − P1 u1 2 2 1 (P2 − P1 ) u1 = 2 < 0 (9.1: Energy Transfer in a Cracked Plate length 2a located in an inﬁnite plate subjected to load P . hence Π2 − Π1 = ∆W − ∆U 1 1 = − (P2 − P1 )u1 = (P1 − P2 )u1 2 2 (9.1 Thermodynamics of Crack Growth
3
Figure 9. and in each one the change in potential energy as the crack extends from a to a + da will be determined: Fixed Grip: (u2 = u1 ) loading. ∆U . Two diﬀerent boundary conditions will be considered.10) (9. under ﬁxed grip there is no external work (u2 = u1 ).9)
Furthermore. Thus the net eﬀect is a change in potential energy given by: Π2 − Π1 = ∆W − ∆U Victor Saouma (9. Griﬃth assumed that it was possible to produce a macroscopical load displacement (P − u) curve for two diﬀerent crack lengths a and a + da.
as the crack extends there is a release of excess energy. the decrease in strain energy (and potential energy in this case) for the ﬁxed grip would be 1 (9. In either case. and the other half released.23)
The diﬀerence between the two sides of the inequality will appear as kinetic energy at a real crack propagation. this energy is released from the strain energy. Energy Release Rate per unit crack extension = Surface energy dΠ = 2γ da Victor Saouma
(9. Under ﬁxed grip conditions.14) (9. 2
23
At the limit as ∆a → da.17) du = u2 − u1
then as da → 0. external work is produced.15)
Thus under ﬁxed grip conditions there is a decrease in strain energy of magnitude 1 u1 (P1 −P2 ) 2 as the crack extends from a to (a + ∆a). the energy released is consumed to form surface energy. (Griﬃth II) 1 = P1 (u2 − u1 ) − P1 (u2 − u1 ) 2 1 P1 (u2 − u1 ) = 2 (9.22)
26 In summary.18) dΠ = udP 2 and for the constant load case 1 (9.16) (9.19) dΠ = P du 2
24
Furthermore.Draft
4
22
ENERGY TRANSFER in CRACK GROWTH. whereas under constant load. deﬁning the compliance as u = CP du = CdP (9.20) (9. 27
Thus a criteria for crack propagation would be dΠ ≥ 2γda (9. Under ﬁxed load condition.24)
Fracture Mechanics
. half of it is consumed into strain energy. there is a net decrease in potential energy of magnitude 1 P1 (u2 − u1 ). we deﬁne: dP = P1 − P2 (9.21)
25
Then the decrease in potential energy for both cases will be given by dΠ = 1 CP dP 2 (9.
n i+1 (9.24.29b) (9. and for incipient crack growth.3 using Westergaard’s solution.2.27)
This equation derived on the basis of global fracture should be compared with Eq.11 derived from local stress analysis.n
(9. 9.1
29
Critical Energy Release Rate Determination
From LoadDisplacement
With reference to Fig.29a) (9.
9.2: Determination of Gc From Load Displacement Curves
28 Using Inglis solution.28)
where OAi Ai+1 = (OAi Bi ) + (Ai Bi Bi+1 Ai+1 ) − (OAi+1 Bi+1 ) 1 1 1 Pi ui + (Pi + Pi+1 )(ui+1 − ui ) − Pi+1 ui+1 = 2 2 2 1 (Pi ui+1 − Pi+1 ui ) = 2 Thus. this reduces to
2 σcr πada = 2γda E
(9.30)
This equation will be rederived in Sect. 9. Griﬃth has shown that for plane stress inﬁnite plates with a central crack of length 2a1 2 πaσcr dΠ = (9. 8. the critical energy release rate will be given by G=
1
(9. 9.2 The energy released from increment i to increment j is given by G= OAi Ai+1 a − ai i=1.26)
or σcr = 2E γ πa (9.Draft
P
9.2
9.25) − da E note that the negative sign is due to the decrease in energy during crack growth.
Victor Saouma
Fracture Mechanics
.1.1 Thermodynamics of Crack Growth
P Ai P 0000000000000 1111111111111
5
x
A1 x A2
a2
x A3
a3
a4
x
A4 x A5
a5
u
i0000000000000 1111111111111 1111111111111 0000000000000 1111111111111 0000000000000 1111111111111 0000000000000 1111111111111 0000000000000 1111111111111 0000000000000 1111111111111 0000000000000 1111111111111 0000000000000 1111111111111 P 0000000000000A 1111111111111 0000000000000 i+1 i+1 1111111111111 0000000000000 1111111111111 0000000000000 1111111111111 0000000000000 1111111111111 0000000000000 1111111111111 0000000000000 1111111111111 0000000000000 1111111111111 0000000000000 1111111111111 0000000000000 1111111111111 0000000000000 Bi 1111111111111 0000000000000 B 1111111111111 0000000000000 i+1 1111111111111 0000000000000 i i+1
1 0 1 0 1 0 1 0 1 0 1 0 1111111111111 0000000000000 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0u O u
a1
u
Figure 9.29c)
1 Pi ui+1 − Pi+1 ui 2B ai+1 − ai i=1. Combining with Eq.1.
4: KI for DCB using the Compliance Method Victor Saouma Fracture Mechanics
. a low modulus plate material (i.32)
Thus we can use an experimental technique to obtain G and from G = 9. B is the thickness. high strength aluminum) can be used to increase observed displacement.3: Experimental Determination of KI from Compliance Curve With regard to accuracy since we are after K which does not depend on E. Fig.3
31
to get KI .
Figure 9.2
30
ENERGY TRANSFER in CRACK GROWTH. let us consider the double cantilever beam problem. and u = CP .4. (Griﬃth II)
From Compliance
Under constant load we found the energy release needed to extend a crack by da was 1 P du. then GBda = 1 1 P d(CP ) = P 2 dC 2 2 (9.
32 33
As an example.Draft
6 9.e. (where u.31)
at the limit as da → 0. 9. then we would have: G= 1 P2 2 B dC da
2 KI E
(9. copliance and the load respectively). 2 If G is the energy release rate. Fig.1.
Figure 9.2. C and P are the point load displacement.
36) (9. hence we would have an unstable crack growth. 9. Had we had a beam in which B increases with a. Fig. in which K remains constant as a increases was proposed by Mostovoy (Mostovoy 1967) for fatigue testing.35)
1 3a2 + 3 h h
36
Substituting in Eq.33)
ﬂexural Taking ν =
1 3
shear
we obtain C = dC da = 8 EB 8 EB
a 0
3x2 1 + dx 3 h h
(9.Draft
34 35
9.38)
Thus the stress intensity factor will be K= √ 2P GE = B 1 3a2 + 3 h h
1 2
(9. such that
38
1 3a2 + = m = Cst h3 h then K=
(9.32 G = = = 1 P 2 dc 2 B da 1 1 P 2 8 3a2 + 2 3 2 EB h h 4P 2 3a2 + h2 EB 2 h3 (9. 9.
Victor Saouma
Fracture Mechanics
.42) B Such a specimen.39)
37
Had we kept G in terms of ν G= 3 4P 2 3a2 + h2 (1 + ν) EB 2 h3 4 (9.41)
2P 1 m2 (9.40)
We observe that in this case K increases with a.5.34) (9.1 Thermodynamics of Crack Growth From strength of materials: C= 24 EB
a 0
7
6(1 + ν) x2 dx + h3 EB
a 0
1 dx h
(9.37) (9.
and this energy is provided by either release of strain 2
More rigorous estimates will be made in Chapter 11. This process will thus dissipate mechanical energy into heat. there will be a small plastiﬁed zone at the tip of the crack.43)
40 This was explained by plastic deformation.
Victor Saouma
Fracture Mechanics
.
To approximate the size of the plastic zone2 .44)
It is evident that we have a larger plastic zone on the surface than the interior.5: Variable Depth Double Cantilever Beam
9. (Griﬃth II)
Figure 9. the surface energy γ would have to be substituted by a much larger value αγ. Irwin and Orrowan Model
39 Whereas Griﬃth’s model has been experimentally validated on perfectly brittle material such as glass. Orowan. The elastic solution predicts an inﬁnite stress.
9. it failed when applied to quasibrittle ones such as mild steel. Thus. the critical load would be
σf =
2E αγ πa
(9. we apply VonMisses criteria on both the interior and the surface (where σ3 = 0):
41
Interior Surface
2 : (σ1 − σ2 )2 + (σ2 − σ3 )2 + (σ1 − σ3 )2 = 2σy 2 + σ 2 + σ 2 = 2σ 2 : (σ1 − σ2 ) y 2 1
(9. where α can be as high as a 1.2
Energy Release Rate. has shown that for quasi brittle solids.000.Draft
8
ENERGY TRANSFER in CRACK GROWTH. permanent plastic deformation is left behind it.1. Irwin has shown that it is possible to approximate α. For thin plates. Thus Irwin’s criteria can indeed be applicable only for thin plates.3
Quasi Brittle material. Energy is needed to create new surfaces. in actuality since no material can withstand such a stress. Equivalence with Stress Intensity Factor
42 We showed in the previous section that a transfer of energy has to occur for crack propagation. When the crack goes through this plastic zone.
6.56f)
σyy = v =
θ 3θ K θ √ I cos 1 + sin sin 2 2 2 2πr θ r θ KI sin κ + 1 − 2 cos2 2µ 2π 2 2
(9.25.47)
(where µ is the shear modulus). 6. Setting θ = π.
Our derivation instead will be based on Westergaard’s solution.6: Graphical Representation of the Energy Release Rate G
45
This energy change is given by: G= 2 ∆a
a+∆a 1 a
2
σyy (x)v(x − da)dx
(9. this results in: G= Victor Saouma
2 KI E
(9. and after and simplifying. or a combination of strain energy and external work.2 Energy Release Rate.Draft
44
9. His original solution was erroneous. whereas the 2 in the denominator is due to the linear elastic assumption. Fig. 9. It remains to quantify energy in terms of the stress intensity factors.
46 47 Upon substitution for σyy and v (with θ = π) from the Westergaard equations (Eq.
Figure 9.
43 In his original derivation. Thus.48) Fracture Mechanics
.46) (9. the energy released during a colinear unit crack extension can be determined by calculating the work done by the surface forces acting across the length da when the crack is closed from length (a+ da) to length a.56b and 6. Griﬃth used Inglis solution to determine the energy released. 9. Eq. however he subsequently corrected it. Equivalence with Stress Intensity Factor
9
energy only.45)
We note that the 2 in the numerator is caused by the two crack surfaces (upper and lower).
and on the material resistance. 9.54) ∂A2 = 0 neutral equilibrium
and the potential energy is Π = U − W .53) G= 2 a22 2a22
9.55c) = E
Ue = = Victor Saouma
Gda 1 σ 2 πa2 2 E
(9. the total energy consumed over the crack extension will be:
da da
dΠ =
0
Gdx =
0
σ 2 πada σ 2 πa dx = E E
(9. 9. 1965).
49
Finally. and stable when it is a minimum.55d) (9. 9.52)
50 Sih. Hence. a suﬃcient condition for crack stability is.1
Eﬀect of Geometry.6. (Sih et al.55a) KI = σ πa 2 KI (9.48 for anisotropic materials as a11 a22 a11 2a12 + a66 + K2 (9.25 derived earlier by Griﬃth. 9. Π Curve
52 From Eq.51)
we note that this is identical to Eq. (Griﬃth II)
E =E E = E 1 − ν2
plane stress plane strain
(9. developed a counterpar to Equation 9.3
51
Crack Stability
Crack stability depends on both the geometry. Fig.55b) G = E σ 2 πa (9. (Gtoudos 1993) 2 (Π + Γ) < 0 unstable fracture ∂ > 0 stable fracture (9.Draft
10 where and
48
ENERGY TRANSFER in CRACK GROWTH.6 yields √ (9.49) (9. crack growth is considered unstable when the energy at equilibrium is a maximum.3.50)
√ Substituting K = σ πa we obtain the energy release rate in terms of the far ﬁeld stress G= σ 2 πa E (9.7. then the potential energy of the system is Π = −U e where Eq.
9.
53 If we consider a line crack in an inﬁnite plate subjected to uniform stress. Paris and Irwin.55e) Fracture Mechanics
.
If we plot Γ. From beam theory
Ue =
Ed3 h2 8a3
(9.
σ P d
h
111111111 000000000 111111111 000000000 111111111 000000000 000000000 111111111 000000000 111111111
a 2a
Γ.54. we observe from Fig.3 Crack Stability
11
and Γ = 4a (crack length is 2a). From Eq.Draft
9. a wedge of thickness h is inserted under a ﬂake of mica which is detached from a mica block along a length a.
54 If we now consider the cleavage of mica. a crack in a linear elastic ﬂawed structure may be characterized by its: Fracture Mechanics
Victor Saouma
. (Gtoudos 1993)
9. 9. Π and Γ + Π.7.56)
and the surface energy is Γ = 2γa. Fig. 9.(Γ+Π)
Γ=4 γ a
U
e
+Γ Γ=2γ a
ac
a
STABLE
UNSTABLE
Figure 9.57)
Again. which corresponds to stable equilibrium.7: Eﬀect of Geometry and Load on Crack Stability. then we observe that the total potential energy of the system (Π + Γ) is maximum at the critical crack length which corresponds to unstable equilibrium. the equilibrium crack is ac = 3Ed3 h2 16γ
1/4
(9. R Curve
As shown earlier.7 that the total potential energy of the system at ac is a minimum. 9.2
55
Eﬀect of Material.(Γ+Π)
Τ+Π ac a Ue Π=−σ 2πa/2E
2
Γ. The energy of the system is determined by considering the mica ﬂake as a cantilever beam with depth d.3.Π.Π.
energy release rate determined from its global transfer of energy accompanying crack growth Thus for a crack to extend. 9. two criteria are possible: 1.61)
Which is Eq. or fracture toughness.2 R vs KIc
58 Back to Eq.58)
we can rewrite it as G = R ∂U e ∂W − G = ∂a ∂a ∂U p ∂Γ + R = ∂a ∂a (9. (Griﬃth II)
1.62)
Victor Saouma
Fracture Mechanics
. and the second to energy consumed during crack propagation.
2 σcr πada = 2γda E dΠ
(9.1
57
Theoretical Basis ∂U e ∂U p + ∂a ∂a
Revisiting Eq. 9. 9.3.3 ∂W = ∂a
+
∂Γ ∂a
(9.2.59b) (9. 9.
59
This equation can be rewritten as
2 σcr πa ≡ R E 2γ Gcr
(9. crack instability will occur when for an inﬁnitesimal crack extension da.Draft
12
56
ENERGY TRANSFER in CRACK GROWTH. the rate of energy released is just equal to surface energy absorbed. 9. 2.59a) (9.3.2. stress intensity factor determined from the near crack tip stress ﬁeld 2.27 as originally derived by Griﬃth (Griﬃth 1921). Compare the stress intensity factor K with a material property called the critical stress intensity factor KIc .60)
or σcr = 2E γ πa (9.52.59c)
where R represents the rate of energy dissipation during stable crack growth. Compare the energy release rate G with a material property called the critical energy release rate GIc . The ﬁrst part corresponds to plastic deformation.
thus must be a straight line. Alternatively.Draft
and as thus
60
9. and for thin plates we do not have constant R due to plane stress conditions. If we increase the stress from σ2 to σ1 . 63 The duality between energy and stress approach G > Gcr = R.
61 Critical energy release rate for plane stress is found not to be constant. and is constant. the crack will extend. the R curve is constant and is equal to GIc . 9.
For plane strain problems. However. G = σEπa . the critical energy release rate is deﬁned as R (for Resistance) and is only equal to a constant (Gcr ) under plane strain conditions.66)
and the critical energy release rate is R = Gcr =
66
K2 dΠ = 2γ = Ic da E
(9. we observe that contrarily to the Westergaard/Irwin criteria where we zoomed on the crack tip. Pu and Underwood 1974). 9. 64 Whereas the Westergaard/Irwin criteria can be generalized to mixed mode loading (in chapter 10).66. G is always a linear function of a. where plane strain is approached for thick plates. it would have extended at σ2 . thus KIc is not constant. 62 Using this energetic approach.67)
Criteria for crack growth can best be understood through a graphical representation of those curves under plane strain and plane stress conditions.8 From 2 Eq.
68
Victor Saouma
Fracture Mechanics
.64)
In general.3 Crack Stability √ σcr πa =
13
2E γ = KIc
2 KIc E
(9. we raise the G value from B to A. Had we had a longer crack a2 . and GIc correspond to plane strain in mode I which is constant.65) G = GI + GII + GIII = E 1−ν 65
From above. if the crack size is a1 .63)
Gcr = R =
(9. and we will instead use K1c and G1c .2. Hence. At A. Using Fig. KIc . a global energy change can predict a local event (crack growth). 9. or K > KIc . should we assume a colinear crack extension under mixed mode loading. the shape of the Rcurve depends on the plate thickness. the energy release rate at a stress σ2 is represented by point B. we have the energy release rate given by G= σ 2 πa E (9. then 1 − ν2 2 K2 2 (KI + KII + III ) (9.3. (Hussain. the energy release rate for mixed mode loading (where crack extension is not necessarily colinear with the crack axis) was not derived until 1974 by Hussain et al. should also be noted.3 Plane Strain
67 For plane strain conditions.
4
Plane Stress
71 Under plane strain R was independent of the crack length.R Constant Grip σ R=G Ic
∆a
Figure 9.ν) 1 E
2 2 σ πa G=(1. (Griﬃth II)
G. under σ1 at point A.8: R Curve for Plane Strain
69 Alternatively. At H crack extension occurs along HN. its G value increases from O to H (note that LF and MH are parallel). and the crack goes fromH to K. had we had a crack of length a2 loaded from 0 to σ2 . this increase is at a lower rate than the increase in R dR dG < da da thus the crack will stabilize and we would have had a stable crack growth. the G line is given by LF (really only point F). under plane stress R is found to be an increasing function of a.R
σ πa G=(1. and then fracture occurs. then G = R and the crack propagates by a small increment ∆a and will immediately stop as G becomes smaller than R. So by loading the crack from 0 to σ2 . (point C) then G > R and the crack extends to a + ∆a. 9. Fig.ν) 2 E
G. point B. However. 2.9 72
If we examine an initial crack of length ai : 1. (9.
Finally.Draft
14
2 2
ENERGY TRANSFER in CRACK GROWTH. If we increase σ1 to σ2 . we can plot to the right ∆a. G increases from O to F. and to the left the original crack length ai .2. At a stress σ2 . however. G increases to H. further increase of the stress to σ1 raises G from F to H.3. G may increase linearly (constant load) or as a polynomila (ﬁxed grips). 3.68)
Victor Saouma
Fracture Mechanics
. thus there is no crack extension.R
K
N
A σ
1
C
R=G Ic σ2 σ1 σ
H R=G Ic F ∆a L a1
2
σ2
B a1 a2 a a2
σ Constant Load G. G < R. if we now increase σ1 to σ3 .
70
9. it should be noted that depending on the boundary conditions. On the other hand.
76 77 Alternatively. If the plastic zone is small compared to the crack (as would be eventually the case for suﬃciently long crack in a large body). 9. for a thick plate it would be predominantly plane strain. Fig. Finally. and the surface energy expressed in terms of R. excess energy is transformed into kinetic energy. R is ﬂat since the surface energy γ is constant.
73
From this simple illustrative example we conclude that Stable Crack Growth: G>R Unstable Crack Growth: G > R
dG da dG da
< >
dR da dR da
(9. but it grows faster than R thus we would have an unstable crack growth. Nonlinear material would have a small plastic zone at the tip of the crack. 75 Some materials exhibit a ﬂat R curve. while other have an ascending one. then not only is G equal to R.3 Crack Stability
G. then R would approach a constant value.54 where the potential energy has been expressed in terms of G. if we increase σ1 to σc . For thin sheets. the load is predominantly plane stress.
74 Finally. For ideaally brittle material.10. we note that these equations are equivalent to Eq.Draft
9.69)
we also observe that for unstable crack growth. 9.
The thickness of the cracked body can also play an important role. The shape of the R curve is a material property. The driving force in this case must increase. Hence a plane stress conﬁguration would have a steeper R curve.9: R Curve for Plane Stress 4.R F D R
15
C σc ai σ3 σ 2 H σ1 B A ∆a
Figure 9.
Victor Saouma
Fracture Mechanics
.
10: Plastic Zone Ahead of a Crack Tip Through the Thickness
Victor Saouma
Fracture Mechanics
. (Griﬃth II)
Figure 9.Draft
16
ENERGY TRANSFER in CRACK GROWTH.
Thus. 8 Whereas under pure mode I in homogeneous isotropic material. so far the only fracture propagation criterion we have is for mode I only (KI vs KIc .1 (10. Fig. for pure mode I problems. KII . yet we usually measure only mode I fracture toughness KIc (KIIc concept is seldom used). for the general mixed mode case.
11 In the absence of a widely accepted criterion for mixed mode crack growth. KII . σy ) = 0 Such an equation may be the familiar VonMises criterion. If the stress intensity factors are in such a critical combination as to render the crack locally unstable and force it to propagate. with respect to the crack axis.
9
Once again.2)
and would be analogous to the one between the two principal stresses and a yield stress. Thus. σ2 . The angle of incipient propagation. crack propagation is colinear.1)
The determination of a fracture initiation criterion for an existing crack in mode I and II would require a relationship between KI .3) Fyld (σ1 . θ0 .Draft
Chapter 10
MIXED MODE CRACK PROPAGATION
7 Practical engineering cracked structures are subjected to mixed mode loading. three of the most widely used criterion are discussed below. 10. KIc ) = 0
(10. we seek to formultate a criterion that will determine:
1. fracture initiation occurs if: KI ≥ KIc (10. and KIc of the form
10
F (KI . thus in general KI and KII are both nonzero.
. 2. in all other cases the propagation will be curvilinear and at an angle θ0 with respect to the crack axis. and GI vs R).
1. Subsequently the actual algorithmic implementation of those models will be presented. and Sih. and one for anisotropic ones.C. the maximum circumferential tensile stress theory. at its tip in a radial direction 2.1
Maximum Circumferential Tensile Stress.6) (10. 1963) presented the ﬁrst mixedmode fracture initiation theory.59c and 6.5)
this equation has two solutions: trivial (10. written in polar coordinates.Draft
2
MIXED MODE CRACK PROPAGATION
Figure 10. i. G.1
Analytical Models for Isotropic Solids
12 First four models for the mixed mode crack propagation will be presented. 6.7) Fracture Mechanics KI sin θ0 + KII (3 cos θ0 − 1) = 0 Victor Saouma
. when σθmax reaches a critical material constant
15 It can be easily shown that σθ reaches its maximum value when τrθ = 0.
10. F.4) (10. It is based on the knowledge of the stress state near the tip of a crack. Replacing τrθ for mode I and II by their expressions given by Eq.e at an angle θ0 such that τrθ = 0 3. in the plane perpendicular to the direction of greatest tension. three for isotropic materials.
13 Erdogan and Sih (Erdogan. 14
The maximum circumferential stress theory states that the crack extension starts: 1.60c
τrθ
θ K θ 3 3θ K θ 1 √ I sin cos2 + √ II cos + cos 2 2 2 4 2 2πr 2πr 4 θ0 ⇒ cos [KI sin θ0 + KII (3 cos θ0 − 1)] = 0 2 = θ0 = ±π
(10.1: Mixed Mode Crack Propagation and Biaxial Failure Modes
10.
6.12)
10.11) (10.2
18
Maximum Energy Release Rate
In their original model. 1974).13)
Fracture Mechanics
. (Erdogan.2:
KI (θ) KII (θ) Victor Saouma
=
4 3 + cos2 θ
1− 1+
θ π θ π
θ 2π
KI cos θ + 3 KII sin θ 2 KII cos θ − 1 KI sin θ 2
(10. θ) will establish for the general mixed mode case the duality which is the basis of fracture mechanics: the equivalence in viewing fracture initiation from either a global energy balance or a local stress intensity point of view.1 Analytical Models for Isotropic Solids Solution of the second equation yields the angle of crack extension θ0 tan 1 KI θ0 1 = ± 2 4 KII 4 KI KII
2
3
+8
(10.1. 20 This (insurmountable) problem was solved in 1974.60b)
θ0 KI θ0 σθ = √ cos 1 − sin2 2 2 2πr
θ0 3 3θ0 KII 3 − sin +√ − sin 4 2 4 2 2πr
(10. KI (θ) and KII (θ) in terms of the stress intensity factors of the original crack KI and KII .C. 10.9)
must reach a critical value which is obtained by rearranging the previous equation √ θ0 3 θ0 − KII sin θ0 KI cos2 σθmax 2πr = KIc = cos 2 2 2 which can be normalized as θ0 KI θ0 3 KII − cos3 cos sin θ0 = 1 KIc 2 2 KIc 2 (10. 1963).”
19 Finding G(δ. G.59b and 6. and Sih.Draft
10. Fig. Fundamentally. Hussain et al. Erdogan and Sih noted that: “If we accept Griﬃth (energy) theory as the valid criteria which explains crack growth. θ)).8)
16 For the crack to extend. have solved for the stress intensity factor of a major crack with an inﬁnitesimal “kink” at an angle θ.10)
17 This equation can be used to deﬁne an equivalent stress intensity factor Keq for mixed mode problems
Keq = KI cos3
θ0 θ0 3 − KII cos sin θ0 2 2 2
(10. by Hussain. the maximum circumferential tensile stress. Evaluation of G(δ. F. (Hussain et al. θ) poses insurmountable mathematical diﬃculties. σθ (from Eq. Pu and Underwood. then the crack will grow in the direction along which the elastic energy release per unit crack extension will be maximum and the crack will start to grow when this energy reaches a critical value (or G = G(δ.
1.15)
The angle of crack propagation θ0 is found by maximizing G(θ). thus. when this minimum reaches a critical value. along which the strain energy density at a critical distance is a minimum (i.
Victor Saouma
Fracture Mechanics
. formulated by Sih (Sih 1974).
24 The minimum strain energy density theory. then this critical value can be determined by setting KII = 0 and Gcr = E .14)
2
1− 1+
θ π θ π
θ π
[(1 + 3 cos
22
2 θ)KI
2 + 8 sin θ cos θKI KII + (9 − 5 cos2 θ)KII ]
(10.17)
For pure mode II (KI = 0). crack propagates along path of minimum resistance). ∂G(θ) ∂θ 2 G(θ) ∂ ∂θ 2 = 0 < 0 (10.16) (10. 9. substituting in Eq. 10.2o If crack extension occurs when G reaches a critical value which is the same scalar quantity 2 KIcr for all cases.e. Eq.15
23
1 4 3 + cos2 θ0 1 + 3 cos2 θ0
2
1− 1+ KI KIc
θ0 π θ0 π 2
θ0 π
+ 8 sin θ0 cos θ0
KI KII 2 KIc
+ 9 − 5 cos2 θ0
KII KIc
2
=1
10.Draft
4
MIXED MODE CRACK PROPAGATION
Figure 10.3
Minimum Strain Energy Density Criteria.65
G(θ) = yielding G(θ) = 4 E 1 3 + cos2 θ
2
1 E
2 2 KI (θ) + KII (θ)
(10. postulates that a fracture initiates from the crack tip in a direction θ0 .2: Crack with an Inﬁnitesimal “kink” at Angle θ
21 Those expressions of KI (θ) and KII (θ) were then substituted into Irwin’s generalized expression for the energy release rate (assuming colinear crack growth). it is found that θ0 = 75.
18)
where µ is the shear modulus (often referred to as G).23)
(plane stress)
(10. We note that it is also possible to decompose the strain energy density into two components a volumetric one and a dilational one:
29
dW = dV Victor Saouma
dW dV
+
D
dW dV
(10.25)
κ = 3 − 4ν (plane strain) and µ is the shear modulus. 10.19)
for plane stress.
27 Using Westergaard’s solution for a cracked inﬁnite plate and substituting the stress into Eq.Draft
25 26
10. Direction of fracture initiation (in 3D) is toward the point of minimum strain energy density factor Smin as compared to other points on the same spherical surface surrounding that point ∂S ∂θ ∂2S ∂θ 2 = 0 > 0 (10. this equation reduces to: 1 κ+1 dW 2 = (σx + σy )2 − 2(σx σy − τxy ) dV 4µ 4 where κ = 3 − 4ν plane strain.28)
V
Fracture Mechanics
. In two dimensional problems.29. we obtain dV
1 S(θ) ∂W 2 2 = a11 KI + 2a12 KI KII + a22 KII = ∂V r0 π r0 where a11 = a12 = a22 = where κ=
3−ν 1+ν
(10.26) (10.1 Analytical Models for Isotropic Solids The strain energy density dW per unit volume dV is ν 1 2 1 dW 2 2 2 2 2 = (σx + σy + σz ) − (σx σy + σy σz + σz σx ) + (τ + τyz + τzx ) dV 2E E 2µ xy
5
(10.
28
This model is based on the following assumptions: 1.22) (10.20)
1 [(1 + cos θ) (κ − cos θ)] 16µ sin θ [2 cos θ − (κ − 1)] 16µ 1 [(κ + 1) (1 − cos θ) + (1 + cos θ) (3 cos θ − 1)] 16µ
(10. Fracture initiation is assumed to occur when Sθmin reaches the maximum critical value Scr . dW . and κ =
3−ν 1+ν
(10.27)
2.21) (10.24) (10.
Its direction coincides with the direction of maximum distortion while Smin coincides with maximum dilation.
Victor Saouma
Fracture Mechanics
. Whereas this may raise some fundamental questions with regard to the model. and then the criteria are assessed for local fracture stability. and Smax is associated with yielding. the angle of crack propagation θ0 is ﬁrst obtained. 2. 10.31)
32
Thus.4 and 10.Draft
6 where the dilational part is given by dW dV =
D
MIXED MODE CRACK PROPAGATION
1+ν 2 2 2 (σx − σy )2 + (σy − σz )2 + (σz − σx )2 + 6(τxy + τyz + τxz ) 6E dW dV 1 − 2ν (σx + σy + σz )2 6E
(10. thus θ0 = 0 and Scr = (S(θ))min = S(θ = 0) = a11 KIc Scr = =
2 2(κ − 1)KIcr 16µπ (κ − 1) 2 KIcr 8πµ
(10.5. 10. In applying σθmax . Fig. we need to deﬁne another material characteristic r0 where to evaluate σθ .29)
and the volumetric by =
V
(10. we note the following 1.4
33
Observations
With reference to Fig. Algorithmically. the fracture locus predicted by this theory is given by: 8µ KI a11 (κ − 1) KIc
2
+ 2a12
KI KII 2 KIc
+ a22
KII KIc
2
=1
(10.32)
10.1.30)
We note that Smin is associated with brittle fracture.3
30
Figure 10.3: Sθ Distribution ahead of a Crack Tip
31
If we set KII = 0. results are independent of the choice for r0 .
2
0.2
0.3
0.5: Locus of Fracture Diagram Under Mixed Mode Loading
Victor Saouma
Fracture Mechanics
.8
0.4 KI/KIc
0.4
σθ max Sθ min Gθ
max
0.8
1.2
0.0
Figure 10.4
0.8
0.9
1.)
30
max
σθ max Sθ min Gθ
max
30 20 20 10 10
0
0
1
2
3
4
5 KII/KI
6
7
8
9
10
0 0.1 Analytical Models for Isotropic Solids
7
60
50
40
θ (deg.0 0.0
0.6
0.Draft
80 70 60 50
10.4: Angle of Crack Propagation Under Mixed Mode Loading
1.0
Figure 10.0
0.7
0.0
0.6
KII/KIc
0.1
0.)
40
σθ max Sθ min Gθ
θ (deg.5 KII/KI
0.6
0.
(Advani and Lee 1979) proposed the following KII 2KIc
2
+
KI KIc
=1
(10. failure of the structure (crack reaching a free surface) will occur. (b) With a decrease in the SIF (and the energy release rate G). All models can be represented by a normalized fracture locus. That is. thus resulting in a global instability. Thus during its KI KII trajectory a crack will most often be in that portion of the normalized KIc − KIc space where the three theories are in close agreement. A crack will always extend in the direction which minimizes KII . 10. the SIF pair will return to within the locus. 6. It can be argued whether all materials must propagate in directions of maximum energy release rate.
Victor Saouma
Fracture Mechanics
. (Awaji and Sato 1978) proposed KI KIc
u
+
KII KIIc
u
=1
(10. If the pair of SIF is inside the fracture loci.2
Emperical Models for Rocks
34 Many researchers have proposed emperical relations for the mixed mode crack propagation of cracks. then the crack is locally unstable and will continue to propagate in either of the following ways: (a) With an increase in the SIF (and the energy release rate G). For each model we can obtain a KI eq in terms of KI & KII and compare it with KIc 9. specially for composite materials. There is a scale eﬀect in determining the tensile strength ⇒ σθmax 7.33)
36
Awaji & Sato.
10. then that crack cannot propagate without suﬃcient increase in stress intensity factors.
KII KI
11. 5. 35
Advani & Lee. If outside. Near the crack tip we have a near state of biaxial stress 8. a crack under KI mixedmode loading will tend to reorient itself so that KII is minimized. Sθmin theory depends on ν
MIXED MODE CRACK PROPAGATION
4. all three theories give identical results for small ratios of and diverge slightly as this ratio increases. For all practical purposes. Sθmin & σθmax depend both on a ﬁeld variable that is singular at the crack tip thus we must arbitrarily specify ro (which cancels out). due to a stress redistribution.34)
where u ≈ 1.Draft
8 3.6 is a material constant
37
Many other emperical relations have been derived. 12.
6. then it can be shown (Lekhnitskii 1981) that the roots of Eq.157. This can be readily seen from Eq. as shown in Fig. for a homogeneous transversely isotropic solid with elastic constants E1 .3 Extensions to Anisotropic Solids
9
Extensions to Anisotropic Solids
38 The maximum circumferential tensile stress theory. two values are needed to characterize the brittle behavior of the crack.6: Fracture Toughnesses for Homogeneous Anisotropic Solids
40 If the applied load and material properties are aligned symmetrically with reference to the crack. G12 . originally developed for isotropic solids. whereas a parasitic crack sliding displacement occurs in anisotropic materials.6 along the principal planes of elastic symmetry. has been extended to the anisotropic case by Saouma et al.131c where a nonzero σxy may result from a pure mode I loading at θ = 0. as the elastic modulus. if the material is orthotropic and the crack is aligned with one of the principal planes of elasticsymmetry. and 1 2 µ12 . a pure mode I displacement occurs in isotropic materials. However.
However.
41
si = αi + iβi fall into one of three categories: 1. 5.
2
2
1
1
(a) Determination of K Ic
2
2
(b) Determination of K Ic
1
1
1 β
(c) Determination of K Ic
β
Figure 10.36) (10. E2 .
The fracture toughness. KIc and KIc . Ayari and Leavell 1987). 10.Draft
10.3
39
10.35)
(10. Case I: α1 = α2 = 0 β1 = β2 Victor Saouma
(10.37) Fracture Mechanics
. (Saouma. is uniquely deﬁned for an isotropic material.
6. respectively.45)
(10.42) KIc = KIc cos2 θ + KIc sin2 θ
42 43 This is equivalent to a scalar and vectorial characterization of the fracture toughness for isotropic and anisotropic solids.44. for a crack arbitrarily oriented with respect to direction 1.
(10. σxy are the cartesian stresses at a point at which polar coordinates are r and θ with respect to the crack tip.47) (10. It should be noted that a possible remedy to the need of performing two separate fracture toughness tests is to conduct only one for the deter1 mination of KIc and then postulate that the ratio of the fracture toughness in both directions is equal to the ratio of the elastic modulii:
2 1 KIc = KIc
E1 E2
(10.43)
44
The maximum circumferential stress is expressed in terms of σθmax where: σθ = σx sin2 θ + σy cos2 θ − 2σxy sin θ cos θ (10.40) (10.41)
β Finally.48)
B = (s2 sin θ + cos θ) 2 C = (s1 sin θ + cos θ)
46 For the isotropic case the angle of crack propagation θ0 is obtained by simply maximizing σθmax but a diﬀerent approach should be followed for the anisotropic case where the angular
Victor Saouma
Fracture Mechanics
.39)
α1 = −α2 = 0 β1 = β2 resulting in zero values of KII . it can be readily β 1 2 seen that the spatial variation of KIc in terms of KIc and KIc will be the same as the one for a tensor of order two. KIc would be a function 1 2 of KIc and KIc .132c.46) (10. Recalling the stress intensity factor deﬁnition from Eq. and 10. σy . After some algebraic manipulation this yields:
KI KII Re [A (s1 B − s2 C)] + √ Re [A (B − C)] σθ = √ 2πr 2πr where A = 1 s1 − s2
3 3 2
(10. Case III:
MIXED MODE CRACK PROPAGATION
α1 = α2 = 0 β1 = β2
(10.44)
and σx .38) (10. Case II: 3.52.
45 This model could be readily extended to anisotropic bodies by ﬁrst combining equations 6. thus: β 1 2 (10. This theory assumes that crack growth will start from the crack tip in the direction along which the tangential stress σθ is maximum and the shear stress is zero.Draft
10 2.
C. 10.46. 50
In this study the following assumptions were made: 1. E1 /E2 was varied from 0.25 to 4 and KII /KI from 0 to 10. note that an attempt to recover the isotropic case from this model would result in a singularity in evaluating A from Eq. 2. Erdogan and Sih’s original formulation (Erdogan.49 along with the proper values for KI and KII . and Sih. If the left hand side of Eq.3 Extensions to Anisotropic Solids
11
variation of the fracture toughness (and thus critical tensile strength) is not constant.49 is greater than 1.50. G. to simplify the performance of a nondimensionalized analysis. 10. F. Fracture Mechanics
Victor Saouma
. it is then substituted for θ in equation 10. then the crack will propagate along the direction θ0 .49 or its normalized equivalent:
Re [As1 B − s2 C] + Max cos2 θ +
1 KIc 2 KIc
KII KI Re[A(B−C)]
sin2 θ
(10. Appendix I of (Saouma et al. in order to remove the dependency on µ12 .
51
Furthermore.45:
θ
KI Re [A (s1 B − s2 C)] + KII Re [A (B − C)] σθ =1 max = 1 2 σθ KIc cos2 θ + KIc sin2 θ
(10.Draft
10. 10. This singularity occurs because for the isotropic case the roots of the characteristic equation 2a are known to be all identical and equal to i (Lekhnitskii 1981).51)
52 Finally. algorithmically the angle of crack propagation θ0 is found by maximizing 10. 49 In an attempt to evaluate and compare the above model for a wide range of E1 /E2 a computer program was developed and a graphical representation of a parametric study was generated.
53 54
In this parametric study.
48 Finally. The crack tip is subjected to a mixed mode loading. The crack is aligned with direction 1 assumed to correspond to one of the two major planes of elastic symmetry (corresponding to an orthotropic case). 1963) can be recovered from this extension.52)
This approximation of G12 (which reduces the number of independent elastic properties to three) was observed by Batugin and Nirenburg (Batugin and Nirenburg 1972) in conjunction with their tabulation of elastic properties of 15 diﬀerent anisotropic rocks. and in light of the potential application of these models to fracture of anisotropic rocks it was assumed:
G12 =
E1 E2 E1 + E2 + 2µ12 E2
(10. 1987) shows that by applying Hospital’s rule. it was assumed:
1 E2 KIc 2 = E KIc 1
(10.50)
when θ0 is determined from Eq.49)
47 Thus. In this σθ max is obtained by setting θ = 0 and equating K to case one should maximize σmax where σθ I
θ KIc in equation 10.
7: Angles of Crack Propagation in Anisotropic Solids
Victor Saouma
Fracture Mechanics
.Draft
12
MIXED MODE CRACK PROPAGATION
Figure 10.
A crack once initiated within an interface can either be stable or propagate in an unstable manner.4 Interface Cracks
13
55 First the angles of crack propagations were determined. the angle of crack propagation is smaller than the one predicted for the isotropic case. its physical 1
This section was written by Kishen Chandra.8 and we note that:
56
1.7. For E1 smaller than E2 anisotropy accelerates the crack extension. Fig. and almost pure mode II cases (KII /KI = 10).Draft
10. if the crack has to kink. depends on the relative toughnesses of the interface and the ones of the adjacent materials. the stress intensity factor for a bimaterial interface is complex. and compare those two values with G0c and Gc (material properties) respectively. For E1 less than E2 .
10. 10. the energy release rate G if the crack was to kink.
58 59 The method presented is very similar to the one used for by the Maximum Energy Release rate. 10. For E1 greater than E2 . The angle of crack propagation predicted by the original isotropic theory is recovered by this generalization. The criteria for an unstable crack to propagate or kink is described in this section. Such crack growth is exceedingly important for composite materials. Fig. We determine the energy release rate G0 if the crack were to propagate along the interface. As mentioned earlier. For E1 greater than E2 .6. 3. the angle of kink with respect to the interface will be considered. 5. that is. the angle of crack propagation is about 20 degrees. 6. The isotropic case is recovered for E1 = E2 . For predominantly mode II cases. 4. It can also branch out. The unstable crack can propagate along the interface or kink into one of the materials. as opposed to 67 degrees for the isotropic case. For E1 greater than E2 . THose angles. and it approaches 85 degrees for the most extreme case. the angle of crack propagation is greater than the one predicted for the isotropic case. the angle of crack propagation surface is nearly ﬂat. Noncolinear crack propagation can take place under pure mode I if E1 /E2 is less than 0. Also.
Victor Saouma
Fracture Mechanics
. and can be extended to concrete/rock crack interfaces1 . and E1 smaller than E2 . call for hte following observations:
1.
1 Then the normalized crack failure surface with respect to KIc is plotted with respect to E1 /E2 . it can propagate along the interface and then kink. anisotropy retards the crack extension. 2.4
Interface Cracks
57 Interface crack propagation is rapidly gaining wide attention. 2. 3. Whether a crack is forced to remain inside an interface or branch out. 60 This section is introduced with the crack tip stresses at the bimaterial interface.
Draft
14
MIXED MODE CRACK PROPAGATION
Figure 10.8: Failure Surfaces for Cracked Anisotropic Solids
Victor Saouma
Fracture Mechanics
.
?? r iλj = cos(λj log(r)) + i sin(λj log(r)). it is customary to replace KI and KII by K1 and K2 . may conveniently be expressed in complex form (Hutchinson and Suo 1992) (K1 + iK2 )r iε (10.1
10.
Victor Saouma
Fracture Mechanics
. Thus K1 and K2 cannot be decoupled to represent the intensities of interfacial normal and shear stresses as in homogeneous fracture. Equation 10. K1 and K2 are components of the complex stress intensity factor K = K1 +iK2 2 . (Hutchinson and Suo 1992)
61 Considering the bimaterial interface crack in Fig. When ε = 0.53 shows that the stresses oscillate3 heavily as the crack tip is approached (r → 0). These dimensions are used to obtain a relationship between the homogeneous and the bimaterial stress intensity factors.Draft
10.9 for tractionfree plane problems the neartip normal and shear stresses σyy and τxy .56) α= ¯ ¯ E1 + E2 ¯ ¯ where E is the plane strain Young’s modulus. α and ε vanish when the materials above and below the interface are identical. Dunders deﬁnes an additional mismatch parameter α: ¯ ¯ E1 − E2 (10.9: Geometry and conventions of an interface crack.55) β= 2[µ1 (1 − ν2 ) + µ2 (1 − ν1 )] in which µ and ν are the shear modulus and Poisson’s ratio. With these relations. respectively. and subscripts 1 and 2 refer to the materials above and below the interface. E = E/(1 − ν 2 ).4. and ε is the oscillation index given by
ε=
1−β 1 ln[ ] 2π 1+β
(10.54)
where β is one of Dunders’ elastic mismatch parameters (Dunders 1969).
2 3
Note that for interface cracks. 10. Furthermore. the relative proportion of interfacial normal and shear stresses varies slowly with distance from the crack tip because of the factor r iε .We note that β. which for plane strain is given by µ1 (1 − 2ν2 ) − µ2 (1 − 2ν1 ) (10. Recall Eq. respectively. an analytical expression for the energy release rate of the crack kinked into one of the materials is then derived in terms of the energy release rate of the crack advance in the interface.4 Interface Cracks
15
dimensions are brieﬂy explained.
Crack Tip Fields
y
µ1 δy
υ1
Ε1 θ
1
x
δx µ
2
υ
2
Ε
2
2
Figure 10.53) σyy + iτxy = (2πr) √ where i = −1.
The energy release rate.60) K = KI + iKII = σ L
Thus it is clear that dimensionally.9. They have proposed to systematically take β = 0 in the analysis of fracture specimens and subsequent application of experimental data in failure predictions. on the interface are given by Eqn. 10. KI and KII such that. however. which would be discussed later.
10. the dimensions of the stress intensity factor are √ (10. 10.4. Usually ε is very small and the eﬀect of nonzero β is of secondary importance.2
Dimensions of Bimaterial Stress Intensity Factors
65 The tractions ahead of the crack tip.
KI + iKII √ 2πx
(10. is generally exceedingly small compared to the crack tip plastic zone and may therefore be neglected (Hutchinson and Suo 1992). the homogeneous and bimaterial stress intensity factors diﬀer by L−iε . Therefore K will necessarily have the dimensional form √ σ L (10. The zone of contact.Draft
16
MIXED MODE CRACK PROPAGATION
62 The crack ﬂank displacements. for plane strain are given by (Hutchinson and Suo 1992.57)
where δy and δx are the opening and sliding displacements of two initially coincident points on the crack surfaces behind the crack tip.54. It may be noted from Eqn. G.58)
64 The complications associated with nonzero ε and β may be circumvented by the approach to bimaterial interface crack problems proposed by He and Hutchinson (He and Hutchinson 1989).59) K = K1 + iK2 = iε L
where L is a length quantity such as crack length or ligament length.10.
σy y + iτx y = ahead of the tip (x > 0. y = 0).53. 10. for plane strain is given by (Carlsson and Prasad 1993)
63
G=
2 2 ¯ ¯ (1/E1 + 1/E2 )(K1 + K2 ) 2 2 cosh (πε)
(10. 10. as ε is an oscillatory term as shown in Eqn.61)
Victor Saouma
Fracture Mechanics
.
67 Considering the kinked crack of length a as shown in Fig. This will be discussed later in more detail. Carlsson and Prasad 1993)
¯ ¯ 4 (1/E1 + 1/E2 )(K1 + iK2 ) √ iε rr δy + iδx = √ (1 + 2iε) cosh(πε) 2π
(10.57 that crack face interpenetration is implied when ε = 0. as shown in Fig. for extension of the crack along the interface.
66
Alternatively for a homogeneous material. the singular stress ﬁeld at its tip is the classical ﬁeld with the conventional stress intensity factors. 10.
K2 . 71
The energy release rate G of the kinked crack (a > 0) is given by.
(10.4 Interface Cracks
y
17
µ1
υ1
Ε1
1 r θ ω y 2 x
µ
2
υ
2
Ε
2
a
x
Figure 10. as4
¯ KI + iKII = c(ω. (Hutchinson and Suo 1992). was proposed by Hutchinson et.58. and c and d are complex valued func) tions of ω. the energy release rate Go of the interface crack advancing in the interface is related to K as shown in Eqn. G=[ (1 − ν2 ) 2 2 ](KI + KII ) 2µ2 (10.64)
10. 10.3
Interface Fracture Toughness
70 In plane strain. 10.67)
4 Note the analogy between this expression and Eq.
Victor Saouma
Fracture Mechanics
. β)Ka−iε (ω.Draft
10. β. The coeﬃcients c and d values are tabulated in He and Hutchinson (He and Hutchinson 1989). for the Maximum Energy Release Rate criterion.65) + ] Go = [ 2 µ1 µ2 4 cosh (πε) ¯ where K denotes the complex conjugate of K.62)
where (¯ denotes complex conjugation for convenience. (Hutchinson and Suo 1992)
68 From dimensional considerations. β)Kaiε + d¯ α. al. α.10: Geometry of kinked Crack.63) (10. is c = d = 1 (exp−iω/2 + exp−i3ω/2 ) 2 1 (exp−iω/2 − expi3ω/2 ) 4 (10. and may be rewritten in the form (Malyshew and Salganik 1965) ¯ (1 − ν1 ) (1 − ν2 ) KK (10.4.. 69 A relation between the intensity factors of the kinked crack in the homogeneous material and the ones at the interface. for the equality between KI . a length quantity must be used to relate those two sets of stress intensity factors (Hutchinson and Suo 1992).62 gives G=[ (1 − ν2 ) ¯ ][(c2 + d2 )K K + 2Re(cdK 2 a2iε )] 2µ2 (10.66)
72
Combining this equation with Eqn. An approximations of these coeﬃcients (He and Hutchinson 1989). 10.13 for the SIF at the tip of a kinked crack as developed by Hussain et al. KII and K1 . α.
ε is zero when β = 0 regardless the value of α.54.Draft
18
73
MIXED MODE CRACK PROPAGATION
This expression can be reduced further by writting K as (He and Hutchinson 1989). L is the inplane length quantity characterizing the speciﬁc interface crack problem when a = 0. From Equations 10.67 and 10. contact will inevitably occur if ε = 0 when a is suﬃciently small compared to L (He and Hutchinson 1989). K = K1 + iK2 = Keiε L−iε (10. therefore ε = 0.
γ G = q −2 Go [c2 + d2 + 2Re(cd exp2i¯ )]
(10. when β = 0. since this will open up the crack at the kink.68 and using the real angular quantity γ as the measure of the loading combination. Contact between crack faces is less likely for the kinked crack (a > 0. the stress intensity factors. The oscillatory behavior of the interface crack ﬁelds and the adependence of G only appear when β = 0. (He and Hutchinson 1989). When a/L becomes
Victor Saouma
Fracture Mechanics
. a sensible approach to gaining insight into interfacial fracture behavior. 10. while avoiding complications associated with the oscillatory singularity. 10. would be to focus on material combinations with β = 0.70) (10. ω > 0) loaded such that KI and KII are positive.65. According to He and Hutchinson (He and Hutchinson 1989). suﬀers contact between the crack faces within some small distance from the tip.59. 10. This is the case of similar moduli across the interface (α = β = 0). 10. as predicted by the elastic solutions of the crack ﬁelds (Comninou 1977.72)
(K2 /K1 )
74 When ε = 0.
75 76 The interface crack with a = 0. KI and KII and G are independent of a. Rice 1988).68)
where by 10.71) (10. Nevertheless.
G
GU G*
GL
a L
Figure 10.11 as predicted by Eqn.69 when ε = 0.11: Schematic variation of energy release rate with length of kinked segment of crack for β = 0. (Hutchinson and Suo 1992)
77 The dependence of G on a for a given kink angle is shown qualitatively in Fig.69)
where (1 − β 2 ) 1/2 ] (1 + α) γ = γ + ε ln (a/L) ¯ q = [ γ = tan
−1
(10.
From Eqn.
. with γ = γ.76) (10. because once nucleated. 10.75)
G coincides with G given by Eqn.Draft
10. G = q −2 Go [c2 + d2 + 2Re(cd exp2iγ )] (10. governs the shear stress on the interface and the relative shearing displacement of the ﬂanks. τxy ) = (δy .78)
The interface stress intensity factors K1 and K2 play precisely the same role as their counterparts in elastic fracture mechanics for homogeneous.3. 10. G should be relevant if there exists cracklike ﬂaws emanating from the interface whose lengths are greater than the zone of contact. which are found to be (He and Hutchinson 1989) GU = q −2 Go [c + d]2
−2
(10. δx ) = G = (K1 . 10.1
79
Interface Fracture Toughness when β = 0
When β = 0 (and thus ε = 0) Eqns. The mode 1 component K1 is the amplitude of the singularity of the normal stresses ahead of the tip and the associated normal separation of the crack ﬂanks.53. the kinked crack has an energy release rate which rapidly approaches G as it lengthens (He and Hutchinson 1989).
80 81 When β = 0.79)
For the case of a ﬁnite crack in an inﬁnite plane.74)
GL = q
Go [c − d]
2
and which depend on K1 and K2 only through Go . the measure of the relative amount of mode 2 to mode 1 at the crack tip is taken as (Hutchinson and Suo 1992)
γ = tan−1 (K2 /K1 )
82
(10.57 and 10. From a physical standpoint.80) Fracture Mechanics
Victor Saouma
. That is. 10.4 Interface Cracks
19
suﬃciently small. while the mode 2 component. G should be relevant in testing for kinking if the fracture process zone on the interface is large compared to the contact zone of the idealized elastic solution.4. G oscillates between a maximum GU and a minimum GL .
∞ ∞ γ = tan−1 (τxy /σyy )
(10. K2 ) (2πr) 1 4 1 √ √ [ ¯ + ¯ ] r(K1 .69. then more attention must be paid to the adependence of G and to the consideration of the contact. K2 ) 2π E1 E2 1 1 1 2 2 [ ¯ + ¯ ](K1 + K2 ) 2 E1 E2 (10. given by Eqn.e. If it is not. Contact between the crack faces will invalidate the prediction for G fron Eqn.69 when a/L is in the range where oscillatory behavior occurs.69. For values of a/L outside the oscillatory ¯ range G approaches G . respectively become: (σy .73) (10.
78
10. 10.77) (10. In any case G should play a prominent role in necessary conditions for a crack kinking out of an interface. when ε = 0.58. i. isotropic solids. K2 .
79.4.
88
For a choice of L within the zone of dominance of the Kﬁeld. does not occur. the deﬁnition reduces to Eqn.79 when β = 0. and a choice based on a material length scale. It is useful to distinguish between a choice based on an inplane length L of the specimen geometry. the tractionfree line crack solution for the displacements implies that the crack faces interpenetrate at some point behind the tip.3.
89 90 The choice of the reference length L is somewhat arbitrary.
Interface Fracture Toughness when β = 0
85 When β = 0. The measure of the proportion of ”mode 2” to ”mode 1” in the vicinity of the crack tip requires the speciﬁcation of some length quantity since the ratio of the shear traction to normal traction varies (very slowly) with distance to the tip when β = 0. Both of these features have caused conceptual diﬃculties in the development of a mechanics of interfaces. since Liε = 1 when ε = 0. such as crack length. a generalized interpretation of the mode measure is the most important complication raised by the oscillatory singularity. deﬁne γ as
γ = tan−1 [
Im(KLiε ) ] Re(KLiε )
(10. such as the size of the fracture process zone or a plastic zone at fracture. Eqn. while the latter is advantageous in interpreting mixed mode fracture data. a mode 1 crack is one with zero shear traction on the interface a distance L ahead of the tip. The former is useful for discussing the mixed mode character of a bimaterial crack solution. the deﬁnitions of mode 1 and mode 2 require some modiﬁcation. 10.82)
where K = K1 + iK2 is the complex stress intensity factor. When ε = 0. First.
As noted by Rice(1988) (Rice 1988).
Victor Saouma
Fracture Mechanics
.83)
Moreover. the decoupling of the normal and shear components of stress on the interface and associated displacements behind the crack tip within the zone dominated by the singularity. a deﬁnition of a measure of the combination of mode is made that generalizes Eqn. independent of material fracture behavior. Noting the stress distribution (10.Draft
20 10. In addition.
86 87 Let L be a reference length whose choice will be discussed later.81)
84 Where Γ(γ) is the toughness of the interface and can be thought of as an eﬀective surface energy that depends on the mode of loading. 10.82 is equivalent to γ = tan−1 [( τxy )r=L ] σyy (10. as explained by Hutchinson (Hutchinson and Suo 1992).2
MIXED MODE CRACK PROPAGATION
83 The criteria for initiation of crack propagation along the interface when the crack tip is loaded in mixed mode by γ is G = Γ(γ) (10. and a mode 2 crack has zero normal traction at that point.53) on the interface from the Kﬁeld. When β = 0. and the approach explained here is along the lines of one of his proposals (Hutchinson and Suo 1992). 10.
The analysis provides the relationships among KI and KII for the kinked crack and K1 and K2 for the interface crack as dependent on the kink angle ω and the material moduli. The energy release rate of the kinked crack G is also related to the energy release rate of the interface crack. if the lower material into which the crack kinks is relatively stiﬀer (α < 0). i.
10. the parent crack is loaded with a complex interface stress intensity factor K = K1 + K2 with mixity γ deﬁned by Eqn.82 relative to some reference length L. as shown in Eqn. The more compliant is the material into which the crack kinks. Roughly speaking.e.12 shows a semiinﬁnite crack lying along the interface with its tip at the origin.12. for all combinations of loading.69.4
Crack Kinking Analysis
91 Fig. Conversely. A parametric study is conducted using these relations and the qualitative features which emerge are listed below.4 Interface Cracks
y
21
1
x ω a G
t
2
Figure 10. 10. the larger is α. Victor Saouma Fracture Mechanics
.. with all factors being equal. as shown in Eqn. switching the signs on α and β.12: Conventions for a Crack Kinking out of an Interface. the larger is the energy release rate.4. the energy release rate for a crack kinking into the stiﬀ material can be less than the interface release rate Go . 10.
• Bimaterial Problem with β = 0 1. the authors have carried out analysis of a crack kinking out of an interface with the aim of providing the crack mechanics needed to assess whether an interface crack will tend to propagate in the interface or whether it will advance by kinking out of the interface.Draft
10. This suggests that under conditions when the compliant material is tough and the stiﬀ material and the interface are each relatively brittle with comparable toughnesses (as measured by a critical value of energy release rate). α > 0 implies that material 1 is stiﬀer than material 2 and vice versa.4. 2. i.62.4. 10. 10. For deﬁniteness. Prior to kinking (a = 0). then the energy release rate is reduced. Go . Negative γloadings with upward kinking can be analyzed by exchanging the materials.e. γ is taken to be positive with kinking down into material 2 as shown in Fig. If the diﬀerence in the elastic moduli of the two materials are relatively large. (Hutchinson and Suo 1992)
10. the crack will tend to be trapped in the interface for all loading combinations. 10.1
Numerical Results from He and Hutchinson
92 In their paper (He and Hutchinson 1989).
Draft
22
MIXED MODE CRACK PROPAGATION
3. 6. for nearly all γ ¯ ˆ except near γ = π/2 where it becomes about 2 deg.5. ¯ ˆ • Bimaterial Problems with β = 0
93 As discussed in connection with Eqn. 11. This shows that the crack kinks into more compliant material 2 with ﬁnite angle for γ = 0. 10. ω and ω are signiﬁcantly diﬀerent.5.5.75 (He and Hutchinson 1989). Plots of G /Go as a function of ω are found in (He and Hutchinson 1989) for (α = 0.25). From the plots (He and Hutchinson 1989) showing the direction ω corresponding to ˆ the maximum energy release rate (i. where dG/dω = 0 or at ω = 0.2
Numerical Results Using Merlin
93 Merlin is a three dimensional ﬁnite element program having the capabilities of performing fracture mechanics based analysis. 4.e. β = −0. The direction ω corresponding to KII = 0 is sometimes suggested as an alternative ¯ to ω as the kink direction. For more negative values of α than 0. 8. which indicate that the eﬀect of β on ω is relatively weak. A comparison between ω and ω is shown in He and ˆ ¯ ˆ Hutchinson (He and Hutchinson 1989). Curves of ω associated ˆ with the maximum value of G versus γ are also shown in (He and Hutchinson 1989). If the stiﬀ material is even less tough than the interface. ω increases smoothly as γ increases from 0 to 90 deg. meaning K2 = 0. β = 0. but not necessarily by kinking. In order to check the accuracy of the numerical results with those of the analytical ones. For example. the curves are quite similar to thwe curves obtained with the same values of α and with β = 0.69. G is not independent of a when ε = 0. depending on γ. the following analysis is performed.67. when material 2 is more stiﬀer than material 1. ˆ
10. for α = 0 and α = ±0. the range of γ in which Gmax occurs at ω = 0 becomes signiﬁcant.e.5. 9. the direction of maximum energy release rate is greater than zero for α > 0. however. and 60 deg. The diﬀerence between the two directions is also very minor for α = ±0. for the crack kinking into the more compliant material. but G approaches G for all but very small a. would not necessarily satisfy KII = 0.5. while KII = 0 at values of ω near the local maximum of ˆ G which occurs for ω between 45 deg. the maximum of G also occurs at ω = 0 where γ is in the vicinity of 90 deg. For suﬃciently negative α.25) and (α = −0. In this range of γ. suggesting that the crack may smoothly curve out of the interface. When γ = 0. The Stern and Becker integral method has been implemented in it to determine the stress intensity factors for bimaterials. i. there exists a range of γ in the vicinity of γ = 0 for which the maximum occurs at ω = 0 7. whichever gives the larger G) as a function of the loading angle γ for various values of α.
Victor Saouma
Fracture Mechanics
. 10. the diﬀerence between ω and ω is less than 1 deg. when α = −0. In the homogeneous case when α = 0. it is seen that for positive α i.4. Although the β values in these examples are quite large. For α.4. When α is negative. more negative than 0. the largest energy release rates occur when ω is small approaching zero.e. the crack may leave the interface. Such a path. ˆ 5. the maximum occurs at ω = 0 for all γ.
σxx for material 2 was varied for diﬀerent cases as tabulated in Table 10. consisting of two isotropic materials is analyzed.13: Geometry and Boundary Conditions of the Plate Analyzed .3. Since the analytical results for this problem were available for plane stress condition in (Lin and Mar 1976). The material in the upper portion of the plate is identiﬁed as material 1 and the material in the lower portion is identiﬁed as material 2. the tensile stress σxx for material 1 was ﬁxed at 1 psi. The crack. used for bimaterial interfaces as compared to the conventional stress intensity factors KI and KII used in the case of homogeneous interface. and the ﬁnite element mesh is shown in Fig. Symmetry of the plate about the centreline perpendicular to the crack and the material interface is considered in the ﬁnite element model.1. is on the interface between the two materials.13.4 Interface Cracks
23
94 A square centrecracked plate of dimensions 20 X 20 inches. which are as follows:
96
K1 =
5
σ[cos(ε log 2a) + 2ε sin(ε log 2a)] + [τ [sin(ε log 2a) − 2ε cos(ε log 2a)] √ a (10. The stress intensity factors K1 and K2 5 was obtained in (Lin and Mar 1976) using the expressions derived by Rice in 1965 (Rice and Sih 1965).14.
95
σyy
Material # 1
σxx1
1" 20" Notch
Material # 2 σxx2
σyy 10"
Figure 10. The plate is subjected to tensile stresses in all directions. The tensile stress in the ydirection. 10. seven diﬀerent cases with varying moduli ratios were considered in this analyses. The geometry and the natural boundary conditions are shown in Fig. of length 2 inches. This mesh was generated using the preprocessor developed for Merlin. 10. The elastic modulus E2 for material 2 was kept ﬁxed at 1 psi and the poisson’s ratio of material 1 was ﬁxed at 0.
It is worth mentioning at this point that the modulus of elasticity of concrete is in the range 2 × 106 − 4 × 106 psi and for rock 5 × 105 − 10 × 106 psi.84) cosh πε
K1 and K2 as explained earlier are the real and imaginary parts of the complex stress intensity factor K = K1 + K2 .
Victor Saouma
Fracture Mechanics
. Generally in the case of dams the ratio of the modulus of elasticity of concrete to that of rock is somewhere around 3.Draft
10. σyy in all the cases was 1 psi and in the xdirection.
98
β 0.581 E02 6.30
α 0.1: Material Properties and Loads for Diﬀerent Cases
Victor Saouma
Fracture Mechanics
.335 E02
Table 10.82 0.285
ε 0.91 0.30 0.98 0.143 0.000 0.280 0.000 123.99 0.0 4.579 E02 7.000 10.711 E02 9.38 0.36 0.37 0.000 3.352 E02 9.Draft
24
MIXED MODE CRACK PROPAGATION
Figure 10.000
ν2 0.244 100.00 0.
Case 1 2 3 4 5 6 7
E1 /E2 1.31 0.30 0.14: Finite Element Mesh of the Plate Analyzed .227 0.35 0.159 E02 7.600 1000.30 0.00 0.50 0.53 0.234 0.35 0.30
σxx2 1.208 0.30 0.000 22.
0
25
Table 10.872 0.3.85) cosh πε
K2 =
97 To have a consistent deﬁnition of K1 and K2 as deﬁned in earlier chapters.25 0.281 0.352 0.761 0.77 5.15 1.22 1.69 along with the approximate expressions of c and d given by Eqns.86 0.875 0. 10.871 0.80 0.
Victor Saouma
Fracture Mechanics
. The stress intensity factors obtained in this case are the conventional ones KI and KII .27 1.747 0.21 1. It is seen that the diﬀerences between the analytical and numerical results are very small.760 0.760 0. 10.386 0. the above expres√ sions have to be multiplied by the factor π cosh πε.0057 % diﬀerence K1 K2 22.2: Analytical and Numerical Results Case 1 2 3 4 5 6 7 Analytical K1 K2 1. the fracture energy G is calculated using Eqn.841 0.00 1.82 0.58. This implies that the use of the conventional Sintegral to obtain the bimaterial stress intensity factors may lead to erroneous results.01 — 5.15.869 0.60 97 26.76 20.00 — 13.2408 1.525 0.761 0.750 0.750 0.873 0. 99 Plots showing the variation of G/Go versus kink angle ω for the loading angle considered in the analyses are shown in Figs.27 1.0 6.26 Numerical K1 K2 1.770 0.23 1.21 1.747 0.22 1.265 0.770 0.0047 1. 10. 10.0041 1.70 96 23.24 1.0 6.02 96 27.0072 1.0 6.0 6.63 and 10. 10.765 0.773 0.2.24 1. which is used in the case of homogeneous interface is shown in Table 10. Using the modiﬁed values of K1 and K2 .79 and in the case of the numerical solution γ is obtained using the relation γ = tan−1 (τ /σyy ).Draft
10.0158 1.24 1.312 0.26 1. These analytical results along with the numerical results obtained using Merlin is tabulated in Table 10.76 0.10 98
Table 10.00 1. respectively.773 0.26 % diﬀerence K1 K2 4.13 1.765 0.4 Interface Cracks Case 1 2 3 4 5 6 7 Analytical K1 K2 1. These plots are obtained using Eqn.22 1.0083 1.872 0.3: Numerical Results using Sintegral without the bimaterial model τ [cos(ε log 2a) + 2ε sin(ε log 2a)] − [σ[sin(ε log 2a) − 2ε cos(ε log 2a)] √ a (10.70 98 28. It is seen that the percentage diﬀerence between these values and the theoretical stress intensity factors K1 and K2 are quite high.16 1.26 Numerical KI KII 1.40 93 25.64.275 0. 98 The results of the stress intensity factors obtained by the Sintegral method and without the use of the bimaterial model. where τ and σyy are the shear and normal stresses at the crack tip.13 1. The loading parameter γ in the case of the analytical solution is obtained from Eqn.
4 0.3 1. ν1= 0. ν1 = 0. ν1= 0.2 0.9 deg.) 135
E1/E2 = 123.0.3 deg.8
0.7 deg.3 1. Numerical γ = 16.0 0.0
0
0.0
0
45 90 Kink Angle ω (deg)
135
E1/E2 = 100.
0.0 Analytical γ = 6. Numerical γ = 12.3.2 1.4 0. Numerical γ = 12.5
45 90 KInk Angle ω (deg.3.0.2 0.4 0.
0.4 0.2
1.Draft
26
1.8
Fracture Mechanics
.0 Analytical γ = 7.2 deg.0
1.8
MIXED MODE CRACK PROPAGATION
E1/E2 = 3.6
G/Go
Analytical γ = 7.3.25.2 1.0 0.7 deg.8 E1/E2 = 10. ν1 = 0.1 deg.8 135
G/Go
0. ν1 = 0.3.2 1. 0 45 90 Kink Angle ω (deg.6 0.6 0. ν2 = 0. Numerical γ = 13. Numerical γ = 3.3 1. ν1 = 0.6 0.2 0. Numerical γ = 15. 0 45 90 Kink Angle ω (deg. 0 45 90 Kink Angle ω (deg) E1/E2 = 100.6 0.8 deg.0.0 Analytical γ = 8.2
E1/E2 = 1000.8
G/Go
0. ν2 = 0.) E1/E2 = 22. ν1 = 0.2 1.2 0.0.3
G/Go
0.) 135
G/Go
0.5 deg.0.2
0. ν2 = 0.0
0.4 deg. ν2 = 0.5 deg. ν2 = 0.)
135
0.8 135
G/Go
0.2
0.0 0. 0 45 90 KInk Angle ω (deg.0 1.2 1.4 Analytical γ = 8.0 Analytical γ = 4.6
0. ν2 = 0. ν2 = 0.6.35 1.3.0
Victor Saouma
0.1 deg.0 0.35 1.3
1.3.3.5 deg.4
0.
e. crack kinks in material i . 0 ≤ γ ≤ γmax . 2. When there is no dissimilarity in the elastic properties of the materials across the interface. 8. there will still be a loading range. Hence the level of Gic required to prevent kinking out of the interface will depend on the interface toughness Goc at the loading angle γ applied. Gmax = maximum energy release rate in material i
1. 6.5
10. Conditions for crack kinking and propagation: • • • • Gmax < Gic Go < Goc Gmax ≥ Gic Go < Goc Gmax < Gic Go ≥ Goc Gmax ≥ Gic Go ≥ Goc no crack propagation. the direction of kinking associated with the maximum energy release rate and with KII = 0 are virtually the same.Draft
10. When the crack has penetrated well into material i. Let Goc = critical fracture energy of the interface. When the fractured interface has some roughness. 9. 5. The condition for propagation in the interface is Go = Goc and that for propagating in any one of the two materials is G = Gic .
3. a criterion based on KII = 0 is expected to hold. When Gic is comparable to Goc . If Gic is suﬃciently large when compared to Goc . Gic = critical fracture energy of material i.e. A choice of criterion for the initial kinking step will have to be guided by experiment. 7.4. i. while for γ > γmax . Goc might depend on the loading parameter γ.
Victor Saouma
Fracture Mechanics
. crack kinks and propagates i. crack deﬂection into the core is a likely scenario because of favourable conditions and large crack driving force.4 Interface Cracks
27
Summary
100 The results for the kinked crack can be used to access whether an interface crack will propagate in the interface or whether it will kink out of the interface. crack propagates along the interface. Crack deﬂection analysis based on the crack kinking and maximum principal stress theories indicate that global mode I loading generally favors interfacial crack propagation. For global mode II loading. 4. the crack branches. the interface crack will kink into material i. the crack will never kink into material i. such that the crack stays in the interface.
Draft
28
MIXED MODE CRACK PROPAGATION
Victor Saouma
Fracture Mechanics
.
Draft
Part IV
ELASTO PLASTIC FRACTURE MECHANICS
.
Draft
.
estimates of the fracture process zone will be separately discussed.Draft
Chapter 11
PLASTIC ZONE SIZES
7 it was shown in chapter ?? that. then our linear elastic assumptions are correct.
10
The evaluation of the plastic zone for plastiﬁed materials can be determined through various levels of approximations:
11
1. LEFM is not applicable (thus it would be incorrect to use a K or G criterion) and a nonlinear model must be used. Multiaxial yield criteria Each one of them will be separately reviewed. and a fracture process zone for cementitious materials and ceramics. 8 If this zone is small compared to the crack size. if not. 1 Due to the intrinsically diﬀerent behavior of concrete compared to metals. whereas the next chapter will develop criterions for crack growth.
. What is the size of the plastic or process zone? 2. Clearly. Uniaxial stress criteria (a) ﬁrst order approximation (b) second order approximation (Irwin) (c) Dugdale’s model 2. What are the criteria for crack growth?
This chapter will answer the ﬁrst question by focusing on metals1 .1
Uniaxial Stress Criteria
12 First we shall examine criteria in which only the uniaxial stress state (σyy normal to thee crack axis) and we shall consider three models of increasing complexities.
11. all materials have a ﬁnite strength. thus there will always be a small plastiﬁed zone around the crack tip. under linear elastic fracture mechanics assumptions. 9
Thus there are two important issues associated with nonlinear fracture: 1. This “damaged” zone is referred to as a plastic zone for metals. the stress at the crack tip is theoretically inﬁnite.
Draft
2
PLASTIC ZONE SIZES
11.2. we started with a stress distribution which satisﬁed equilibrium. In this model the force (area under the stress curve) which was “eliminated” is simply redistributed to satisfy equilibrium requirements. with reference to Fig. 11. In the ﬁrst model. and “cut down” all stresses exceeding σyld .2
SecondOrder Approximation (Irwin)
14 In 1960. Irwin (Irwin 1960) developed a secondorder approximation for the plastic zone based on the stress redistribution occurring at the crack tip. Hence.56b) σyy = KI (2πr)
1 2
cos
3θ θ θ 1 + sin sin 2 2 2
(11. 6.2)
σ σyld
2
(11.3)
Figure 11. As a result equilibrium was no longer satisﬁed. 11.4)
0 ∗ rp
∗ σdr − σyld rp
K ∗ √ dr − σyld rp 2πr 0 ∗ √ rp σ πa ∗ √ dr − σyld rp 2πr 0 Fracture Mechanics
Victor Saouma
.1.1)
to the yield stress σyld for θ = 0.1: FirstOrder Approximation of the Plastic Zone
11.
The simplest estimate of the size of a process zone is obtained by equating σy (Eq. Fig.1 σyy = or
∗ rp = 2 1 KI a = 2 2π σyld 2
KI
∗ 2πrp
= σyld
(11.1
13
FirstOrder Approximation.1. we have:
B = σyld δ A = = =
∗ rp
(11.
9)
∗ ∗ ∗ Note that rp = 2rp and that we can still use rp but with aef f = a + rp .3. thus this simpliﬁes into
∗ ∗ ∗ δ + rp = 2rp ⇒ δ = rp
(11.6)
16
∗ From Eq. rp =
a 2
σ σyld
2
.5)
15
Equating A to B we obtain:
∗ ∗ σ 2arp − σyld rp = σyld δ ∗ ∗ σyld (δ + rp ) = σ 2arp ∗ (δ + rp )2 =
2aσ 2 ∗ r 2 σyld p
(11.8)
rp
=
∗ 2rp
rp =
1 KI = 2 π σyld
σ σyld
2
a
(11.7) (11.1 Uniaxial Stress Criteria
3
Figure 11. thus we can consider ∗ which would result in: an eﬀective crack length of a + rp
17
∗ Kef f = f (g)σ π(a + rp ) = f (g)σ π(a +
K2 ) 2 2πσyld
(11.2: SecondOrder Approximation of the Plastic Zone
∗ rp
= = σ
σ
0
a −1 ∗ r 2 dr − σyld rp 2
∗ a 1 rp ∗ 2r 2 0 −σyld rp 2
∗ ∗ = σ 2arp − σyld rp
(11.10)
Victor Saouma
Fracture Mechanics
.Draft
11. 11.
Draft
4
18
PLASTIC ZONE SIZES
For linear elastic fracture mechanics to be applicable. 2.13) (11.00025) = 56.
Dugdale (Dugdale 1960) assumed that the actual physical crack of length 2a is replaced by a total eﬀective crack of length 2c.12)
Since
rp a
is very small. hence LEFM is applicable.008) = 55. such that: 1. a far ﬁeld applied stress σapp = 350 MPa.1.11)
11.0033) = 66M P a m
(11. 2(385)2
(11.008) = ≈ . 11.0033m = 3.
25 The solution to this model is found by ﬁrst considering a point load P applied at a distance x from the crack center. Fig.008) = .3
24
Dugdale’s Model. A constant stress σyld is applied over ρ where a < x < c causing (a negative) Kρ .3mm.3. c is selected in such a manner that K = 0 or Kremote = −Kρ .00025m = .4M P a m √ √ Kapp = σ πa = 350 π(.
11.5M P a m (11.25mm 2 2σyld 2(1. then
∗ rp =
(350)2 (. 400)2
(11. 23
If yield stress was decreased by heat treatment to 385 MPa.17) Fracture Mechanics
. where the stress intensity factors are given by:
KA = KB = Victor Saouma
P √ πa P √ πa
a+x a−x a−x a+x
(11. and the plate has a yield stress σyld = 1.14)
22 We note that there is only 2 percent diﬀerence between those two solutions. 11. Using the 1st order approximation:
∗ rp =
21
σ2 a (350)2 (.008 + .008 + . Keﬀ ≈ Kapplied √ ∗ Keﬀ = σ π(a + rp ) = 350 π(. as in Fig.2.1
19
Example
Considering an inﬁnite with central crack of size 2a = 16 mm.
20
First we seek to determine the size of the plastic zone and the eﬀective SIF. 400 MPa.16)
and in this case LEFM may no longer be applicable.15)
and
√ Kef f = 350 π(. we must have: LEFM ⇔ K ≈ Kef f (11.4.1. where c = a + ρ.
4: Point Load on a Crack
Victor Saouma
Fracture Mechanics
.1 Uniaxial Stress Criteria
5
Figure 11.Draft
11.3: Dugdale’s Model
Figure 11.
11. or ρ → ∞. Using the Cosine Taylor’s expansion on Eq. 11. as a 1 a = =1− c a+ρ 2! πσ 2σyld
2 σ σyld
→ 0. Fig.21)
From Eq.23 should be compared with Eq.21. a → 0.19 and solving.5: Eﬀect of Plastic Zone Size on Dugdale’s Model 2.23) a) previously obtained.6. we obtain: σyld Kρ = √ πa
c a
a+x + a−x
a−x a+x
dx
(11.20 with Eq. 11. we can rewrite:
4
+
1 4!
πσ 2σyld
−
1 6!
πσ 2σyld
6
(11. we obtain: a π σ = cos c 2 σyld
29
(11. we observe 1. 11. 11.5 c
Figure 11. see Fig.19)
28
On the other hand.
Victor Saouma
Fracture Mechanics
. 11. the stress intensity caused by the remote far ﬁeld stress is given by: √ (11. and replacing in the above equations.21.9 (rp =
30 Finally it should be mentioned that similar equations have been derived by Barenblatt (Barenblatt 1962) on the basis of a linear stress distribution.
Eq. 11. 11.20) Kremote = σ πc
Equating Eq. As σ → σyld .22)
Neglecting all but the ﬁrst two terms and solving for ρ ρ= π2 8 σ σyld
2
a=
π KI 2 8 σyld
σ σyld 2
(11.Draft
6
26
PLASTIC ZONE SIZES
Assuming dP = σyld dx.18)
27
Integration of this equation results in Kρ = 2σy a c arccos π c (11.
or σ3 = 0 for plane stress.2
31
Multiaxial Yield Criteria
All the previous models have restricted themselves to θ = 0 and have used uniaxial yield criteria. 6.27)
(11.
33 With those stress expressions.56c σ1 = σ2 σ3 for plane strain.24)
where the stresses were obtained in Eq. but the size of the plastic zone can be similarly derived from a multiaxial yield criterion. Substituting the principal stresses (with r = rp ) into this equation and solving for rp yields • For plane strain: rp (θ) = 1 KI 3 sin2 θ + (1 − 2ν)2 (1 + cos θ) 2 4π σyld 2 (11.29) σe = √ (σ1 − σ2 )2 + (σ2 − σ3 )2 + (σ3 − σ1 )2 2 2
θ K θ √ I cos 1 + sin 2 2 2πr θ KI θ cos = √ 1 − sin 2 2 2πr = ν(σ1 + σ2 )
(11.2 Multiaxial Yield Criteria
7
Figure 11. we would obtain: 1 1 (11.6: Barenblatt’s Model
11.25) (11. Using the VonMises criteria.
32
The principal stresses at a point with respect to the crack tip are given by: σ1.56a.56b. any yield criteria could be used.26) (11.Draft
11. 6.28)
and yielding would occur when σe reaches σyld .30)
Victor Saouma
Fracture Mechanics
. and 6.2 = σx + σy ± 2 σx − σy 2
2 2 + τxy
(11.
In general. the expected plastic (or more appropriately process) zones are shown in Fig.025. as no stress redistribution has been accounted for.32) Fracture Mechanics
.5 −0. 4. 3.6 −0. and for diﬀerent mixed mode ratios.3 −0.Draft
8
34 35
PLASTIC ZONE SIZES
• For plane stress: rp (θ) =
1 KI 3 1 + sin2 θ + cos θ 2 4π σyld 2
(11. for the plastic zone sizes: 1.7.9 1. p The ratio of the plastic zone size to the plate thickness B must be much smaller than unity for plane strain to prevail.6 0.7 −0. 37
r
Plane Strain Victor Saouma
KIc ⇔ rp < .3 0.1 0.3)
Figure 11. The plastic zone for plane strain is much smaller than the one for plane stress (by a factor of (1 − 2ν)2 ).0 −0.31)
For the DruckerPrager model.7: Normalized Mode I Plastic Zone (VonMyses) 2.1 Plane Stress Plane Strain (ν=0.7 0. This is only a ﬁrstorder approximation.5 0.
0. Fig. 11.3 0.7 0. Similar shapes could be derived if both KI and KII expressions were used. Plane Stress
36 Irrespective of a plate thickness.1 −0.3
Plane Strain vs.3 −0.2 −0. It has been experimentally shown that this ratio should be less than 0.1 0.5 0. 11.8.4 0. Diﬀerent sizes would be obtained from alternative models (Tresca’s would be larger than VonMises’s). there is a gradual decrease in size of the plastic zone from the plate surface (plane stress) to the interior (plane strain).
11.2 0.1 0.4 −0.025B
(11.
2 2 Finally. it should be noted. Plane Stress
9
Figure 11.
38
2
. 11. the plate thickness should increase
39 Furthermore. and the maximum shear stress τmax is equal to approximately 45 degrees from the crack plane.Draft
11. Fig. that fracture toughness KIc can only be measured under plane strain conditions.
Plane Stress: σz = 0.9.3 Plane Strain vs. the diﬀerent stress ﬁelds present at the tip of the crack under plane stress and plane strain will result in diﬀerent deformation patterns.10
40
Victor Saouma
Fracture Mechanics
.8: Plastic Zone Size Across Plate Thickness
KI We also observe that since rp is proportional to σyld as either the SIF increase or the yield stress decrease. This is best explained in terms of the orientation of the planes of maximum shear stress for both cases.
σx 2
and occurs at
Plane Strain: In this case we have σy < σz < σx . and the maximum shear stress is equal to σx −σy which is not only much smaller than σx but occurs on diﬀerent planes. once again. Fig. 11.
Plane Stress and Plane Strain
Victor Saouma
Fracture Mechanics
.9: Plastic Zone Size in Comparison with Plate Thickness.Draft
10
PLASTIC ZONE SIZES
σ
τ τ max
y
σz
x
σ
σx
σy
σ
z
Plane Stress
y
σ
τ τ max
x
σx
σ
σz
σy
σ
z
Plane Strain
Figure 11.
Draft
11.10: Plate Thickness Eﬀect on Fracture Toughness
Victor Saouma
Fracture Mechanics
. Plane Stress
11
Figure 11.3 Plane Strain vs.
Draft
12
PLASTIC ZONE SIZES
Victor Saouma
Fracture Mechanics
.
in those cases where LEFM is not applicable. 8 In many cases it is found that LEFM based criteria is either: too conservative and expensive as it does not account for plastiﬁcation at the crack tip.
Figure 12. 10
But ﬁrst let us note the various stages of ductile fracture: 1.1: Crack Tip Opening Displacement. Under LEFM assumptions. Slow (stable) crack growth 4. the crack tip opening displacement (CTOD) is zero. 12.Draft
Chapter 12
CRACK TIP OPENING DISPLACEMENTS
7 Within the assumptions and limitations of LEFM we have two valid (and equivalent) criteria for crack propagation: 1) K vs KIc which is a local criteria based on the strength of the stress singularity at the tip of the crack. Fig. and 2) G vs GIc (or R) which is a global criteria based on the amount of energy released (or consumed) during a unit surface crack’s propagations. Crack initiation 3.1. Blunting of an initially sharp crack. an alternative criteria for crack growth in Elasto Plastic Fracture Mechanics (EPFM) is sought. however in elastoplastic material due to blunting it is diﬀerent from zero. Unstable crack growth
. 9 Thus. and/or invalid based on calculations ∗ of rp where LEFM assumptions are checked. (Anderson 1995) 2.
a local criterion based on the crack tip opening displacement (CTOD).
13
References (Cottrell n.d. Fig.1 1. However. 2. Local Vector K CTOD Global Scalar G J
LEFM EPFM
Table 12. Firstorder approximation based on a ﬁctitious crack.
14 15
There are two approaches to determine the CTOD: 1. JIC .K. whereas the J integral is more commonly used in the United States.1: Comparison of Various Models in LEFM and EPFM Historically the CTOD was ﬁrst proposed as a valid criteria for crack propagation on the basis of Cotterell and Wells work in the early 60’s at the British Welding Institute.Draft
2
11
CRACK TIP OPENING DISPLACEMENTS
Again two approaches are currently in use. This has formed the basis of current “R6” speciﬁcations for ductile failure in the U. 12.56f. Secondorder approximation based on Dugdale’s model.1
Derivation of CTOD
Irwin’s Solution
16 The vertical displacement of a point next to the crack tip due to mode I loading is given by Eq.
12
Under LEFM the crack tip opening displacement is clearly zero. 2. Table 12.) and (Wells 1963) have introduced the concept of the crack tip opening displacement to characterize elastoplastic fracture.1. This criterion is still the one primarily used in the United Kingdom. (Anderson 1995) Victor Saouma Fracture Mechanics
.1
12.2: Estimate of the Crack Tip Opening Displacement.
12. then the crack tip will blunt resulting in a nonzero crack tip opening displacement (CTOD). a global criterion based on the quasistrain energy release rate (J integral). when the material is allowed to yield.2
Figure 12. 6.
3)
and using κ =
3−ν 1+ν
for plane stress we obtain CTOD =
2 4 KI π Eσyld
(12..1)
If we substitute θ = ±π we obtain the upper and lower displacements of the crack face.
the CT OD can be approximated by CT OD =
Derivation of this equation can be found on p. For x = a this reduces to v(a) =
20
c 4 aσyld log π E a
(12.2) COD = 2v = µ 2π
∗ If we determine the crack tip opening displacement a distance rp away from the crack tip using Irwin’s plastic zone correction from Eq.9)
or CTOD = note that for small
1
K2 π2 σ2 1+ + . 2 Eσyld 24 σyld
K2 Eσyld . Kanninen (Kanninen 1984) has shown that the crack opening along the crack is given by1 : √ √ √ √ x c2 − a2 + c2 − x2 x c2 − a2 + a c2 − x2 2 aσyld √ √ log √ + log √ (12..7)
we would then obtain
(12.1. 203 of (Anderson 1995)
Victor Saouma
Fracture Mechanics
.10)
σ σyld .1 Derivation of CTOD KI r v= 2µ 2π
1 2
3 θ θ κ + 1 − 2 cos2 2 2
sin
(12.21.8)
21
using the series expansion of log sec: CTOD = 8 aσyld 1 π E 2 π σ 2 σyld
2
+
1 12
π σ 2 σyld
4
+ ···
(12.2
Dugdale’s Solution
19 Using Dugdale’s solution. π σ a = cos c 2 σyld CTOD = 2v = π σ 8 aσyld log sec π E 2 σyld
(12. Hence the crack opening is given by r κ+1 KI (12. and due to symmetry their sum corresponds to the crack opening displacement.9
18
∗ rp =
2 1 KI 2 2π σyld
(12.
(12. 11.5) v(x) = π E a c2 − a2 − c2 − x2 x c2 − a2 − a c2 − x2
for 0 ≤ x ≤ c. 11.Draft
17
12.4)
12.6)
Combining this equation with Dugdale’s solution for c from Eq.
12)
Victor Saouma
Fracture Mechanics
.
Recalling also that the energy release rate G is given by G = approximate equation CT OD = and its counterpart CT ODcr = R σyld G σyld
we obtain the following
(12.Draft
4
CRACK TIP OPENING DISPLACEMENTS
12.11)
(12.2
22
GCTOD Relations
K2 E .
Γ is a closed contour followed counterclockwise. as shown in Fig.
.1: J Integral Deﬁnition Around a Crack
8 Whereas Eshelby had deﬁned a number of similar path independent contour integrals.1
7
Genesis
Eshelby (Eshelby 1974) has deﬁned a number of contour integrals that are path independent by virtue of the theorem of energy conservation. and dΓ is the element of the arc along the path Γ.Draft
Chapter 13
J INTEGRAL
13.2)
where w is the strain energy density.1) J= ∂x Γ with w=
0 ε
σij dεij
(13. he had not assigned them with a particular physical meaning.1. 13. The twodimensional form of one of these integrals can be written as: ∂u wdy − t dΓ (13.
t ds u
ds
Ω
Γ
y x
Figure 13. t is the traction vector on a plane deﬁned by the outward drawn normal n and t = σn. u the displacement vector.
9a) (13.11)
which is identical to the second term of Eq.2.10a)
14
Hence. 13.7)
The strain is given by Eq. ny and nj are direction cosines. ∂ ∂w = σij ∂x ∂xj ∂ui ∂x (13.i dΩ ∂ui ∂x
(13.8)
Substituting ∂w ∂x 1 ∂ ∂ui σij 2 ∂x ∂xj ∂ ∂ui = σij ∂xj ∂x = + ∂ ∂x ∂uj ∂xi (13. 13.6. Eq. then dx = −ny dΓ.j + uj.3)
and assuming Γ to be deﬁned counterclockwise. ?? εij =
12
1 (ui. we need to show that the former is indeed equal to zero for a closed path.2
Path Independence
9 Before we establish the connection between Eshelby’s expression for J.5)
we obtain J=
Ω
11
∂ ∂w − ∂x ∂xj
σij
dxdy
(13. and the energy release rate J.4)
10
Invoking Green’s theorem vi ni dΓ =
Γ Ω
vi.Draft
2
J INTEGRAL
13.9b)
13
On the other hand. 13. Substituting J=
Γ
wnx − nj σij
∂ui ∂x
dΓ
(13. the ﬁrst term in the square bracket becomes ∂w ∂εij ∂εij ∂w = = σij ∂x ∂εij ∂x ∂x (13.i ) 2
(13. and dy = nx dΓ and ti = nj σij where nx .
15
Thus the integrand of Eq.
J=
Γ
wdy − ti
∂ui dΓ ∂x
(13. we have ∂ ∂xj σij ∂ui ∂x = σij ∂ ∂xj ∂ui ∂x + ∂σij ∂ui + ∂xj ∂x
0
(13. Fracture Mechanics
Victor Saouma
.3 vanishes and J = 0 for any closed contour.6)
Substituting the strain energy density.
3. and in the absence of body forces. Along paths Γ2 and Γ4 .
y
Γ 4 Γ 2
x
Γ 3
Γ 1
Figure 13. the change in potential energy is dΠ da =
Ω
dw dΩ − da
ti
Γ
dui dti + ui dΓ da da
(13. We shall proof that when J is applied along a contour around a crack tip.1
19
Virtual Crack Extension
Considering a twodimensional crack surrounded by a line Γ which encompasses an area Ω. we will now exploit this to proove that around a crack.3
18
Nonlinear Elastic Energy Release Rate
Let us now establish the connection between the two previous interpretations of J.2 if we consider the closed path Γ = Γ1 + Γ2 + Γ3 + Γ4 in which Γ1 and Γ3 are arbitrarily chosen contours.
13. the contributions to J from Γ2 and Γ4 vanish. Taking into account the diﬀerence sense of integration along paths Γ1 and Γ3 we arrive at the conclusion that the values of J integrated over paths Γ1 and Γ3 are identical. it represents the change in potential energy for a virtual crack extension da.12)
For a virtual crack extension. the path independence of J is assured.Draft
17
13. Under quasistatic conditions. provided that the stresses and displacement gradients are continuous. Obviously J = 0 over Γ in order to satisfy compatibility conditions. the traction vector ti = 0 and also dy = 0. 13.2: Closed Contour for Proof of J Path Independence
13. and the one associated with the energy release rate. J is nonzero and is independent of the path.13a)
Victor Saouma
Fracture Mechanics
. the potential energy is given by Π=
Ω
20
wdΩ −
ti ui dΓ
Γ
(13. the one mathematically deﬁned by Eshelby.
With reference to Fig. Two slightly diﬀerent derivations are presented. Because these two paths were arbitrarily chosen.3 Nonlinear Elastic Energy Release Rate
3
16 Having shown that indeed J = 0. Consequently.
3.Draft
4 =
Ω
J INTEGRAL
dw dΩ − da Γu
0
ti
dui dti dΓ − + ui da da
Γt
dui dti ti da + ui da dΓ
0
(13. 13. 13.15)
Ω
Γt
26
We can also rewrite:
∂w ∂εij ∂ ∂w = = σij ∂a ∂εij ∂a ∂xj
∂ui ∂a
(13. the coordinate axis also moves.
22 23
For a crack extension. Fig. the second term inside the square bracket will be zero along Γt because the traction is constant during crack growth. Since Γu is zero along the path.16)
27
From the divergence theorem ti ∂ui dΓ = ∂a =
Ω
Γt
Ω
σij
∂ ∂xj ∂w ∂a
∂ui ∂a
dΩ
(13.13b dΠ = da ∂w ∂w − dΩ − ∂a ∂x ti ∂ui ∂ui − ∂a ∂x dΓ (13.14)
for a crack extension along a ( ∂x = −1). the ﬁrst one with prescribed displacement (Γu ) and a second one with prescribed traction (Γt ).17b)
Victor Saouma
Fracture Mechanics
. ∂a
25
Substituting into Eq.
y
x
Γ
Ω
Figure 13.17a) (13.13b)
We have dcomposed the contour path into two parts. we maintain a closed contour integral along Γt .3: Virtual Crack Extension Deﬁnition of J
24
Thus we can write
∂ ∂ ∂x ∂ ∂ d = + = − da ∂a ∂x ∂a ∂a ∂x
(13.
21
Furthermore.
13.20)
which is the same as Eq. multiply both sides by −1 and note that nx dΓ = dy − dΠ da =
Γt
=
Γt
∂ui dΓ ∂x ∂ui wdy − ti dΓ ∂x wnx − ti
(13.16. ??. We deﬁne the initial boundary conditions as such Surface initial boundary conditions: (a) Displacements u0 on Γu i (b) Tractions t0 on Γt . No inertia. the ﬁrst terms in each of the two integrals in Eq.21) Γ = Γt + Γu
31
1.18)
We now apply the divergence theorem.19) (13.
Victor Saouma
Fracture Mechanics
. Due to ti and ui . let us thoroughly derive the expression for the J integral.3. and we are left with
dΠ = da
Γt
ti
∂ui dΓ − ∂x
Ω
∂w dΩ ∂x
(13. We will derive an expression for the energy released during a unit surface crack extension based on the following assumptions:
1. no initial stresses 4.
13.15 cancel out. Plane stress or plane strain 7.4 (13. Linear or nonlinear elastic solid 3. the J integral is equal to the energy release rate for linear and nonlinear elastic material under quasistatic conditions. Mode I loading 8. 13.Draft
29
13. 2D stress and deformation ﬁeld 6. No thermal loading 5.2
†Generalized Derivation
30 Following the brief preamble.3 Nonlinear Elastic Energy Release Rate
5
28 Hence. 13. the hole will grow and will have a new boundary. from this Equation and Eq. i
0 Internal stress and strain: σij and ε0 ij
2. Stress free crack Considering an arbitrary solid with both traction and displacement boundary conditions and an internal inclusion. Fig. or body forces. Henceforth. Homogeneous body 2. ∆Γ and volume ∆V .
25) (13.26)
=
1 2
∆V
1 2
∆Γ
t0 du∗ dΓ i i
for the case of linear elasticity 6.
4. thus i i −∆Π =
∆V 0 σij ε0 dV − ij 0
1 2
∆Γ
t0 ∆ui dΓ i
Energy Release Rate G
(13. we reduce t∗ on ∆Γ to zero.u∗ =ui +∆ui i i t∗ =t0 . Holding the loading on Γu and Γt constant.22)
(13. u*i ∆v ∆Γ Γ t
σij .Draft
6
t
*
J INTEGRAL
o t
t i.27)
Victor Saouma
Fracture Mechanics
.u∗ =u0 i i i i
t∗ du∗ dΓ i i
(13. However to prevent growth we will apply t∗ = σij nj on ∆Γ in order to maintain initial i 0 εij 0 and σij (Recall Irwin’s derivation of G).23)
5. εij Γ u
o
o
Figure 13.24)
∆V
∆Γ
strain energy
=
∆V
w ε0 dV −
0 σij ε0 dV − ij
t∗ =0. For a sharp crack and linear elastic systems during crack extension ∆V = 0 and during crack extension and t∗ = t0 . The total reduction in potential energy during this process was −∆Π = w ε0 dV − t∗ du∗ dΓ i i (13.4: Arbitrary Solid with Internal Inclusion
0 3. Thus we go from i t∗ = 0 i u∗ = 0 i This will result in a new internal state
0 σij + ∆σij ε0 + ∆εij ij
to
t∗ = 0 i u∗ = 0 i
(13.
3 Nonlinear Elastic Energy Release Rate 7.1 vanishes and
J=
Γc
wdy
(13. 9.Draft
13. so the integral may be made to depend only on the crack tip singularity in the deformation ﬁeld. So for a linear elastic material − ∂Π = G.29)
wdy − t
∂u dΓ ∂x
(13. as Eq. we can write − d dΠ =− dA dA wdV − t∗ du∗ dΓ
dΓ
7
(13. which implies ∂a J =G (13. The utility of the method rests in the fact that alternate choices of integration paths often permit a direct evaluation of J.2 to shrink to an arbitrarily small contour Γc surrounding the crack tip.32)
where Π is the potential energy.35) KI = EJ1 /(1 − ν 2 ).30)
It is identical to the previous expression but we do not restrict ourselves to go around ∆V .28)
dV
For unit thickness dV = dxdy and dA = dx dΠ = dA 8. 13. (plane strain)
Victor Saouma
Fracture Mechanics
. but around any closed path. 34 In linear elastic analysis.1 yields √ EJ1 . 13. 13. Let us now deﬁne J=
Γ
dΓ
wdy −
t
dΓ
∂u dΓ ∂x
(13.
33 An arbitrarily small curve Γ may then be chosen surrounding the tip. and under pure mode I. J =− ∂Π = ∂a wdy − T ∂u dΓ ∂x (13. (plane stress) (13. We can take Γ1 to be any external contour away from the crack tip and Γ2 to be identiﬁed with dΓ (increase in volume).34)
so that J is an averaged measure of the strain on the crack tip.31)
r
J =−
∂Π ∂a
(13. Back to sharp crack but not necessarily linear elastic systems. then the second term of Eq. Rice’s J can be used for the computation of stress intensity factor KI . This integral was ﬁrst proposed by Eshelby who had deﬁned a number of contour integrals which are path independent by virtue of energy conservation theorems.33)
32 Allowing Γ1 in Fig.
36) J =− da for a unit thickness crack extension. which relates the total strains to stresses.4
Nonlinear Energy Release Rate
35 Whereas LEFM is restricted to linear elastic materials.Draft
8
J INTEGRAL
13. Whereas the loading behavior of both materials is identical. we consider the load displacement curves curve of a notched specimen made of nonlinear elastic material for two diﬀerent conditions.6: Nonlinear Energy Release Rate. 13. 13. Fig. is applicable. and the deformation theory of plasticity.6 during crack extension from a to a + da:
P 000000000000000000000000000000000000000 111111111111111111111111111111111111111
* dU=dU
d∆ dP
111111111111111111111111111111111111111 000000000000000000000000000000000000000 111111111111111111111111111111111111111 000000000000000000000000000000000000000 111111111111111111111111111111111111111 000000000000000000000000000000000000000 111111111111111111111111111111111111111 000000000000000000000000000000000000000 a 111111111111111111111111111111111111111 000000000000000000000000000000000000000 U* 111111111111111111111111111111111111111 000000000000000000000000000000000000000 da 111111111111111111111111111111111111111 000000000000000000000000000000000000000 a+ 111111111111111111111111111111111111111 000000000000000000000000000000000000000 111111111111111111111111111111111111111 000000000000000000000000000000000000000 111111111111111111111111111111111111111 000000000000000000000000000000000000000 U 111111111111111111111111111111111111111 000000000000000000000000000000000000000 111111111111111111111111111111111111111 000000000000000000000000000000000000000 111111111111111111111111111111111111111 000000000000000000000000000000000000000 111111111111111111111111111111111111111 000000000000000000000000000000000000000 111111111111111111111111111111111111111 000000000000000000000000000000000000000
∆
Figure 13. Thus. For nonlinear elastic solids. ??. most metals have a nonlinear stressstrain curve. Fig. Rice deﬁned J as the energy release rate in nonlinear elastic materials: dΠ (13. as long as there is no unloading. (Anderson 1995)
Victor Saouma
Fracture Mechanics
.
36 37 Following a similar approach to the one of sect. they
σ
Nonlinear Elastic
ElasticPlastic
ε
Figure 13.5: ElasticPlastic versus Nonlinear Elastic Materials diﬀer when they are unloaded. we can assume a nonlinear elastic behavior.5 illustrates the uniaxial stressstrain behavior of elasticplastic and nonlinear elastic materials.
∂A ∆ thus.39)
P
From Fig. the energy absorbed by the 1 specimen. the load. J could be determined from the slope of the tangent to the curves. a similar one applies to the J integral. 41 This is accomplished by testing a number of identical specimens. crack initiation and Fracture Mechanics
44
Victor Saouma
. However.8 where following blunting.6
42
Crack Growth Resistance Curves
Just as the energy based criteria for crack propagation in LEFM was simply stated as G > R. J could be obtained from a single test.7. 13.5 J Testing Load control we have ∆ = 0 and Π = U − W = U − P ∆ = −U ∗ U∗ =
def P
9
(13.37)
∆dP
0
(13. Thus for constant load JP = Constant Displacement. P = 0. 13. many materials with high toughness do not necesarily fail catastrophically at a particular value of J. The rising R curve is illustrated in Fig.Draft
where
13. and J∆ = − dU da (13. at various ﬁxed displacements. Hence. Hence. Hence. an LEFM would grossly overestimate the material resistance to crack growth. Since J = − B ∂U . 13. 40 For nonlinear material. E
13. we resort to the energy deﬁnition of J. the diﬀerence between JP and J∆ is 1 dP d∆ which is vanishingly small com2 pared to dU . Whereas we could exploit the contour line integral deﬁnition of J to determine it from a single test by attaching an array of strain gages.
43
Rising R curve is associated with with growth and coalescence of microvoids. but with slightly diﬀerent crack lengths.
13. in those cases. and the crack length. Fig.5
J Testing
39 For linear elastic materials. the evaluation of J is straightforward since it is equal to G. and hence we can not have a simple relation between J. these materials display a rising R curve. the principle of superposition does not apply.6. The area under each P − ∆ curve is U . this is ususally impractical. Instead. which in turn is directly related to the stress intensity factor K.
38 Therefore J for load control is equal to J for displacement control and for linear elastic 2 materials J = G = K . We then plot U in terms of a.38)
is the complimentary strain energy.40)
∆
dU ∗ da
(13.
7: Experimental Derivation of J
JR
Initiation
Crack Blunting
Crack Extension
Figure 13.8: J Resistance Curve for Ductile Material. (Anderson 1995)
Victor Saouma
Fracture Mechanics
.Draft
10
P a 1 a 2 a 3< a 4 < < a1 a2 a3 a4 U ∆ 4 ∆ 3 ∆2 ∆
1 dU/da
J INTEGRAL
U
∆1 ∆2 ∆3 ∆4 P ∆ J ∆
a1
a2
a3
a
4
a
a
a 1 a2 a3 a4
∆
Figure 13.
a tearing modulus as TR = and the applied tearing modulus as Tapp = E dJ 2 σyld da (13. We deﬁne. The total displacement is given by
49
∆T = ∆ + Cm P where Cm is the compliance of the system.7
47
† Load Control versus Displacement Control
Crack propagation can occur under either load or displacement control. we can take a similar derivative of J(a. JIc can be speciﬁed at the onset of initiation.42) E dJR 2 σyld da (13.47)
∆T
52
from Eq.
The crack stability can be discussed within the context of the test stiﬀness (or its inverse the complinace). and keeping ∆T constant dJ da Substituting
∂P ∂a
=
∆T
∂J ∂a
+
P
∂J ∂P
a
dP da
(13.46)
dividing both sides by da.43)
13.
50
(13.45)
51
Similarly. 13. P ) dJ = ∂J ∂a da +
P
∂J ∂P
dP
a
(13.48)
a
Victor Saouma
Fracture Mechanics
.Draft
45
13.
48 Unstable crack propagation will occur if the rate of change of J exceeds the rate of change of R. An arbitrary deﬁnition of a critical J.45 yields dJ da =
∆T
∂J ∂a
−
P
∂J ∂P
a
dP da
Cm +
P
∂∆ ∂P
−1
(13. crack growth would occur if J Tapp Tapp = JR ≤ TR Stable crack growth > TR Unstable crack growth (13. the rate of change of J depends on the loading condition.41)
46
Hence.7 † Load Control versus Displacement Control
11
propagation occurs.44)
Diﬀerentiating this equation yields d∆t = ∂∆ ∂a da +
P
∂∆ ∂P
dP + Cm dP = 0
a
(13. Whereas in both cases J (and G) is the same.
Cm = ∞.
56
Starting with J=
Γ
wdy − t
∂u dΓ ∂x
(13.10 we substitute dΓ = rdθ. (Anderson 1995)
13. and is smaller than in the previous case. we had a √r stress and strain singularity. 13. 13.
Figure 13. 13. Fig.9: J. and may thus lead to unstability. we now seek to determine the corresponding ones. and for n = 1 we have a linear elastic response. and thus the second term vanishes. Fig.11. r If we now consider a material with the following (uniaxial) power law hardening model (RambergOsgood) which is often used to curveﬁt stressstrain data:
58
ε εyld
=
σ σ +α σyld σyld
n
(13.8
55
Plastic Crack Tip Fields
1 Whereas under LEFM assumptions. Fig.
Victor Saouma
Fracture Mechanics
. and for n = ∞ we would have an elastic perfectly plastic one. α is a dimensionless constant. for displacement control. y = r sin θ.
dJ da
is always positive and
Alternatively. and dy = r cos θdθ π ∂u w cos θ − t rdθ (13.9. JR versus Crack Length.50) J= ∂x −π
57
but J should be independent of r by virtue of path independence hence both w cos θ and t ∂u ∂x should be proportional to 1 .Draft
12
53 54
J INTEGRAL
dJ da
For load control. x and y respectively. Cm = 0 corresponds to an inﬁnitely stiﬀ system.49)
Taking the contour around a circle of radius r.51)
n is the strain hardening exponent. for EPFM.
5
2.5
0.5
σ/σyld
1.001 2.10: J.Draft
13.01)
Victor Saouma
Fracture Mechanics
.8 Plastic Crack Tip Fields
13
r2 r1
Figure 13. Around a Circular Path
Ramberg−Osgood Stress−Strain Relation
α=.11: Normalize RambergOsgood StressStrain Relation (α = .0
0
1
2 ε/εyld
3
4
5
Figure 13.0
1.0
n=1 n=15 n=10 n=7 n=5
0.
we know that the energy must be proportional to σε ∝ 1 ⇒x+y =1 r
(13. Rice.62)
66 For elastic perfectly plastic material n = ∞ and the stress ﬁeld is nonsingular while the strain ﬁeld has a singularity of the form r −1 .Draft
14
60
J INTEGRAL
59 In the vicinity of the crack tip.55)
62
Furthermore.57) (13.60)
D r 1+n
1
65
For linear elastic solids n = 1 and these equations reduce to the familiar ε(r) = σ = C r2 D r2
1 1
(13. 13. the plastic strain is dominant (the elastic one is negligible) and σ n ε =α (13.53) (13. (Hutchinson 1968).56.59) (13.58)
64
(13.
67
Victor Saouma
Fracture Mechanics
. we would have σ = ε = c1 rx c2 ry
1 r
(13.56)
63
Solving Eq.51 c2 r y εyld =α c1 rx σ
n
⇒
yld
c3 1 = c4 nx ⇒ y = nx ry r
(13.
n 1 Those singularities 1+n and 1+n are often referred to as the HRR singularities after Hutchinson.52) εyld σyld
Denoting by x and y the order of the stress and strain singularities.55 and 13. and Rosengren.54)
61
From above. 13. from Eq.61) (13. we obtain x = y = Thus in general we would have ε(r) = σ(r) = C r
n 1+n
1 1+n n 1+n
(13.
63) (13.64)
ε˜ (θ. the stress is proportional to √r and this area is called the Kdominated region. 2πr
70
1 √ r
in
the elastic region.8 Plastic Crack Tip Fields It can be shown that there is a relationship between J and the crack tip σ − ε ﬁeld: σij = σyld εij = ασyld E EJ 2 ασyld In r EJ 2 ασyld In r
1 1+n
15
σij (θ. n) ˜
n 1+n
(13.
Figure 13. If the crack grows outside it. we should note that at the crack tip we have two singularities. For the crack to be J controlled.e. (Anderson 1995)
Small Scale Yielding: In this zone both K and J characterize this zone. 13. If we have monotonic loading. 13. n) ij
˜ ˜ where In is an integration constant which depends on the σ−ε curve. those two zones must be embedded within a zone of J dominance.12: HRR Singularity. and the later
. (Anderson 1995) Fig.Draft
68
13. From Fig. a J dominated region occurs in Victor Saouma Fracture Mechanics
.13. i. the stress components increase or decrease at diﬀerent rates. At a short distance 1 from the crack tip (relative to L).
71
Small scale yielding Elastic Plastic Conditions Large Scale Yileding
Large Strain Y Y Y
J Dominated Y Y N
K Dominated Y N N
Table 13.1: Eﬀect of Plasticity on the Crack Tip Stress Field. The material directly in front of the crack also violates our assumption of of proportional loading. J also characterizes the σ and ε singularities in EPFM just as K did in LEFM (σij = Finally.12. we observe that material behind a
propagating crack unload elastically (as opposed to nonlinear elastic). then the measured R curve is no longer uniquely characterized by J. where L represents a characteristic dimension of the structure. the ﬁrst one
1 r 1+n
1
√K . and table 13. and σ & ε are dimensionless I functions of n and θ (analogous to fij in LEFM which also depend on the stress state (plane stress/strain)
69
Thus.1 schematically illustrate the eﬀect of plasticity on crack tip stresses.
the HRR is approximately valid. 1981). ﬁnite strain region occurs within approximately 2δ from the crack tip where the large defomration invalidates HRR.
13.9
Engineering Approach to Fracture
72 The solution of a plastic problem involves the determination of the J integral. c: Kdominated zone
a b c
a
b
a
Figure 13. but K no longer. Finally. n)
Victor Saouma
Fracture Mechanics
. J becomes size and geometry dependent. This usually involves a ﬁnite element analysis. Large Scale Yielding: here the size of the ﬁnite strain zone becomes signiﬁcant relative to L and there is no longer a region uniquely characterized by J.65a) (13. (Anderson 1995) the plastic zone where the elastic singularity is no longer valid. (Kumar et al. then a simpliﬁed engineering approach can be followed.13: Eﬀect of Plasticity on the Crack Tip Stress Fields. n) (13.65c) (13.Draft
16
J INTEGRAL
a Large Strain Region b: Jdominated zone. we can write J = Je (ae ) + JP (a.
73
General Solution: For RambergOsgood material. ElasticPlastic: J is still valid. n) ∆c = ∆ec (ae ) + ∆P c (a.
If such a capability is not available. and a ﬁrst order approximation of J is required.65b) (13. Inside the plastic zone.65d)
δ = δe (ae ) + δP (a.
ε Consider a nonlinear or fully plastic incompressible body and with ε if the plastic deformation can be described by J2 . Elastic Solution: The elastic solution can be written as Je = f1 δe ∆ce K2 a P2 = I W E E a P = f2 W E a P = f3 ( ) W E (13.
3 2
(13. The ﬁeld quantities increase in direct proportion to the load or displacement parameter raised to some power dependent on n.9 Engineering Approach to Fracture
17
where: Je (ae ).Draft
13. For example if the traction is Ti = P Ti and P is a loading parameter.
For such a material Ilyushin (Il’yushin 1946) has shown that the solution of the boundary value problems based on the above equation. δp (a. and ∆ec (ae ) are the elastic contributions based on an adjusted crack length ae . n) (13. deformation plasticity theory with power law and isotropic hardening. n)
.66b) (13.67c)
We note that P is the generalized load per unit thickness. n). n). and that δ is the crack opening displacement. n) are the plastic contributions based on the material hardening exponent n. and ∆pc (a. Jp (a. εp . furthermore. then σij εij = P σij (xi .66a) (13.67a) (13. and involving a single load or displacement parameter which is increasing monotonically has two important properties: 1. then the small strain constitutive relation is given by ¯ 3 σ sij εij = α( )n−1 εy 2 σy σy where Sij and σe = respectively. We note that P is the generalized load per unit thickness.69b) (13.67b) (13. and that δ is the crack opening displacement.66c)
for plane stress β = 2 and for plane strain β = 6. εe e p .69c) Fracture Mechanics
n
ui = αεy Victor Saouma
ui (xi . Fully Plastic Solution: Considering a fully plastic incompressible cracked body. where ae = a + φry ry = φ = 1 n − 1 KI βπ n + 1 σy 1 1 + (P/Po )2
2
(13. δe (ae ).69a) εij (xi . n) = αεy P σy P σy
n
(13.68)
Sij Sij are the stress deviator and teh vonMises eﬀective stress. ∆ is the loadpoint displacement.
1. n)(P/P0 ) ∆cp = αεy ag3 (a/W )h3 (a/W. 1981).2 W
P
a W
b
Figure 13. Since the integrand of J involves the product of σ and u gradients. εij (xi . n) are functions of xi and n and are independent of P . then then the fully plastic J will be proportional to P n+1 and we can write: Jp = αεy σy bg1 (a/W )h1 (a/W. analysis and are tabulated (±5%) in (Kumar et al. α is from Eq.14. 2.9.14: Compact tension Specimen
Plane Strain P0 = 1.70d)
n+1
where P0 coresponds to the limit load based on σy and b is the ligament length (W − a).455βbσy Plane Stress P0 = 1.25 W P
1. Fig. n)(P/P0 )n+1 δp = αεy ag2 (a/W )h2 (a/W. but are independent of P .Draft
18
J INTEGRAL
where σij (xi .071βbσy β = gi = 1 Victor Saouma 2 a b
2
(13.7.70c) (13.1
Compilation of Fully Plastic Solutions
Compact Tension Specimen: Table 13.70a) (13. Those functions can be obtained from F.74) Fracture Mechanics
a a +4 +2−2 −1 b b
.E.73) (13. n)(P/P0 )
n n
(13. 13.70b) (13.
13. Since σ and ε increase in the same proportion. n) and ui (xi . 13. n)(P/P0 ) δtp = αεy bg4 (a/W )h4 (a/W. n).52 ε/εyld = α(σ/σyld )n The dimensionless functions (h1 − h4 ) depend upon a/W and n and possibly other geometric parameters. fully plastic solution based on deformation plasticity is also the exact solution to the same problem posed for incremental or ﬂow theory.71) (13.72) (13.
284 1.132 0.614 0.988 0.20 0.314 0.99 0.37 1.03 0.71 0.06 0.51 1.32 2.25 8.76 1.75 2.585 0.801 0.95 0.54 2.485 0.18 0.45 4.16 6.59 0.494 1.2: hFunctions for Standard ASTM Compact Tension Specimen.37 1.180 0.00 1.00 0.461 1.07 0.41 1.48 1.977 3.40 9.35 3.347 0.36 1.90 8.530 0.27 1.919 0.875 3.9 10.909 0.75 1.02 1.76) (13.568 0.20 5.95 3.03 0.02 0.20 0.86 1.47 0.73 1.77) (13.393 0.93 n=5 n=7 Plane Strain 1.47 1.484 2.310 0.85 4.431 1.63 1.03 0.29 0.85 2.55 2.514 0.22 1.23 17.26 0.12 0.52 1.37 4.47 1.120 0.6 10.377 1.540 4.686 0.985 n = 16 1.5 7.18 2.24 4.08 2.67 4.436 1.695 0.511 3.39 6.42 1.686 0.216 0.620 1.25 0.94 9.494 0. Fig.97 0.32 5.08 5.59 2.48 3.314 0.774 0.16 1.52 4.63 3.32 12.59 1.41 0.130 5.601 4. (Kumar et al.78 2.77 1.833 2.25 4.41 1.66 1.15 12.050 6.857 0.05 12.72 8.18 1.793 0.45 3.23 n=2 2.9 Engineering Approach to Fracture
19
∆ corresponds to the crack opening displacement at the load line and δ is the crack mouth opening displacement.79 Plane Stress 1.92 2.710 0.933 n = 20 1.440 1.370 0.70 7.48 0.12 2.36 0.61 17.666 0.54 0.470 5.775 2.602 1.83 1.71 6.92 0.80 2.023 0.91 1.03 0.683 0.58 1.28 10.68 0.17 0.64 4.05 1.44 0.788 0.26 10.29 0.00 8.81 1.78 11.Draft
a W
13.7 8.5 8.746 0.59 0.24 3.94 1.33 1.68 2.65 1.936 0.77 7.19 0.99 1.368 0.75 2.59 2.38 0.266 0.29 4.73 1.729 5.09 2.45 3.57 3.395 2.39 n = 10 1.85 1.746 0.64 1.7 7.250 2.869 0.12 1.749 0.57 1.490 1.74 7.440 1.76 7.55 12.33 6.61 5.80 1.39 4.29 1.21 1.270 n = 13 1.5 8.248 0.78 1.01 3.03 4.31 0.89 2.33 10.51 1.423 1.45 1.23 0.814 0.6 9.44 1. 13.35 2.77 0.38 0.57 14.10 0.28 1.853 0.89 1.56 3.717 1.824
=
1 4
a W
=
3 8
a W
=
1 2
a W
=
5 8
a W
=
3 4
a W
≈1
a W
=
1 4
a W
=
3 8
a W
=
1 2
a W
=
5 8
a W
=
3 4
a W
≈1
Table 13.9 9.15 1.110 0.26 3.27 7.74 3.373 1.80 1.10 2.57 5.74 2.506 1.40 7.236 0.80 0.67 1.42 1.356 1.448 0.21 3.86 2.15 √ Plane Strain P0 = 4bσy / 3 Plane Stress P0 = 2bσy g1 = g4 = a/W g2 = g3 = 1
(13.459 0.32 1.928 0.92 1.43 1.31 1.3.42 2.09 1.
n=1 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 2.665 0.60 9.14 0.51 5.887 0.95 1.370 0.52 1.17 1.298 1.52 0.24 0.938 0.443 1.75) (13.92 1.9 0.91 2.00 0.176 0.94 5.42 3.460 12.18 5.95 1.83 n=3 1.120 0.998 0. 1981) Center Cracked Panel: Table 13.098 0.33 0.25 11.80 1.14 0.97 3.983 0.60 7.620 0.238 0.686 0.43 2.901 4.540 1.79 1.78)
Victor Saouma
Fracture Mechanics
.30 3.11 0.69 4.79 1.73 1.14 1.46 1.23 6.57 1.345 0.97 0.37 1.34 1.575 1.
Draft
20
J INTEGRAL
P
L X2 a X1 b W L
P
Figure 13.15: Center Cracked Panel ∆ corresponds to the average loadpoint displacement deﬁned by ∆= 1 2W
W W
[u2 (x1 , L) − u2 (x1 , L)]dx1
(13.79)
δ is the crack opening displacement at the center of the crack. Single Edge Notched Specimen: Table 13.4, Fig. 13.16.
σ ∆ /2 b W a L L ∆ /2
Figure 13.16: Single Edge Notched Specimen Plane Strain P0 = 1.455βbσy Plane Stress P0 = 1.072βbσy β = 1+ a b
2
(13.80) (13.81) − a b (13.82) (13.83) (13.84)
g1 = g4 = a/W g2 = g3 = 1
∆ is the loadpoint displacement at the centerline of the specimen, and δ is the crack mouth opening displacement. Double Edge Notched Specimen: Table 13.5, Fig. 13.17.
Victor Saouma
Fracture Mechanics
Draft
a W
13.9 Engineering Approach to Fracture
21
n=1 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 2.80 3.05 0.303 2.54 2.68 0.536 2.340 2.350 0.699 2.21 2.030 0.803 2.12 1.71 0.844 2.070 1.350 805 2.08 0.889 0.632 2.80 3.530 0.350 2.54 3.100 0.619 2.340 2.710 0.807 2.210 2.340 0.927 2.12 1.970 0.975 2.07 1.550 0.929 2.08 1.030 0.730
n=2 3.61 3.62 0.574 3.01 2.99 0.911 2.62 2.39 1.06 2.29 1.860 1.07 1.96 1.320 0.937 1.73 0.857 0.70 1.64 0.428 0.400 3.57 4.09 0.661 2.97 3.29 1.01 2.53 2.62 1.20 2.20 2.01 1.19 1.91 1.46 1.05 1.71 0.97 0.802 1.57 0.485 0.452
n=3 4.06 3.91 0.84 3.21 3.01 1.22 2.65 2.23 1.28 2.20 1.60 1.16 1.76 1.04 0.879 1.47 0.596 0.555 1.40 0.287 0.291 4.01 4.43 1.00 3.14 3.30 1.35 2.52 2.41 1.43 2.06 1.70 1.26 1.69 1.13 0.97 1.46 0.685 0.642 1.31 0.31 0.313
=
1 8
a W
=
1 4
a W
=
3 8
a W
=
1 2
a W
=
5 8
a W
=
3 4
a W
=
7 8
a W
=
1 8
a W
=
1 4
a W
=
3 8
a W
=
1 2
a W
=
5 8
a W
=
3 4
a W
=
7 8
n=5 n=7 Plane Strain 4.35 4.33 4.06 3.93 1.30 1.63 3.29 3.18 2.85 2.61 1.64 1.84 2.51 2.28 1.88 1.58 1.44 1.40 1.97 1.76 1.23 1.00 1.10 0.96 1.43 1.17 0.707 0.52 0.701 0.52 1.11 0.89 0.361 0.25 0.359 0.25 1.14 0.98 0.181 0.13 0.182 0.14 Plane Stress 4.47 4.65 4.74 4.79 1.55 2.05 3.20 3.11 3.15 2.93 1.83 2.08 2.35 2.17 2.03 1.75 1.59 1.57 1.81 1.63 1.30 1.07 1.18 1.04 1.41 1.22 0.785 0.61 0.763 0.62 1.21 1.08 0.452 0.36 0.45 0.36 1.08 0.97 0.196 0.15 0.198 0.15
n = 10 4.02 3.54 1.95 2.92 2.30 1.85 1.97 1.28 1.23 1.52 0.799 0.796 0.863 0.358 0.361 0.642 0.167 0.168 0.814 0.105 0.106 4.62 4.63 2.56 2.86 2.56 2.19 1.95 1.47 1.43 1.43 0.871 0.867 1.01 0.474 0.478 0.867 0.262 0.263 0.862 0.127 0.127
n = 13 3.56 3.07 2.03 2.63 1.00 1.80 1.71 1.07 1.05 1.32 0.664 0.665 0.628 0.250 0.251 0.461 0.114 0.114 0.688 0.084 0.084 4.41 4.33 2.83 2.65 2.29 2.12 1.77 1.28 1.27 1.30 0.757 0.758 0.853 0.383 0.386 0.745 0.216 0.216 0.778 0.109 0.109
n = 16 3.06 2.60 1.96 2.34 1.71 1.64 1.46 0.89 0.888 1.16 0.564 0.565 0.458 0.178 0.178 0.337 0.081 0.081 0.573 0.068 0.068 4.13 4.00 2.95 2.47 2.08 2.01 1.61 1.13 1.13 1.17 0.666 0.668 0.712 0.313 0.318 0.646 0.183 0.183 0.715 0.0971 0.0973
n = 20 2.46 2.06 1.77 2.03 1.45 1.43 1.19 0.715 0.719 0.978 0.466 0.469 0.300 0.114 0.115 0.216 0.0511 0.052 0.461 0.0533 0.054 3.72 3.55 2.92 2.20 1.81 1.79 1.43 0.988 0.994 1.00 0.557 0.560 0.573 0.256 0.273 0.532 0.148 0.149 0.630 0.0842 0.0842
Table 13.3: Plane Stress hFunctions for a CenterCracked Panel, (Kumar et al. 1981)
Victor Saouma
Fracture Mechanics
Draft
22
n=1 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 4.95 5.250 26.60 4.34 4.760 10.30 3.88 4.540 5.14 3.40 4.450 3.15 2.86 4.370 2.31 2.34 4.320 2.02 1.91 4.29 2.01 3.58 5.15 26.10 3.14 4.67 10.10 2.81 4.47 5.05 2.46 4.37 3.10 2.07 4.30 2.27 1.70 4.24 1.98 1.38 4.22 1.97 n=2 6.93 6.47 25.80 4.77 4.56 7.64 3.25 3.49 2.99 2.30 2.77 1.54 1.80 2.44 1.08 1.61 2.52 1.10 1.57 2.75 1.27 4.55 5.43 21.60 3.26 4.30 6.49 2.37 3.43 2.65 1.67 2.73 1.43 1.41 2.55 1.13 1.14 2.47 1.09 1.11 2.68 1.25 n=3 8.57 7.56 25.20 4.64 4.28 5.87 2.63 2.67 1.90 1.69 1.89 0.91 1.30 1.62 0.68 1.25 1.79 0.765 1.37 2.14 0.988 5.06 6.05 18.00 2.920 3.700 4.360 1.940 2.630 1.600 1.250 1.91 0.871 1.105 1.840 0.771 0.910 1.81 0.784 0.962 2.08 0.969
a W
J INTEGRAL
=
1 8
a W
=
1 4
a W
=
3 8
a W
=
1 2
a W
=
5 8
a W
=
3 4
a W
=
7 8
a W
=
1 8
a W
=
1 4
a W
=
3 8
a W
=
1 2
a W
=
5 8
a W
=
3 4
a W
=
7 8
n=5 n=7 Plane Strain 11.50 13.5 9.46 11.1 24.20 23.6 3.82 3.06 3.39 2.64 3.70 2.48 1.68 1.06 1.57 0.94 0.923 0.51 0.928 0.51 0.954 0.50 0.417 0.21 0.697 0.37 0.081 0.42 0.329 0.17 0.769 0.47 1.03 0.69 0.435 0.26 1.10 0.92 1.55 1.23 0.713 0.56 Plane Stress 5.30 4.96 6.01 5.47 12.70 9.24 2.120 1.53 2.530 1.76 2.190 1.24 1.370 1.01 1.690 1.18 0.812 0.525 0.776 0.510 1.09 0.694 0.461 0.286 0.755 0.551 1.160 0.816 0.478 0.336 0.624 0.447 1.150 0.798 0.494 0.344 0.792 0.677 1.54 1.27 0.716 0.592
n= 10 16.1 12.9 23.2 2.170 1.910 1.500 0.539 0.458 0.240 0.213 0.204 0.085 0.153 0.167 0.067 0.233 0.296 0.125 0.702 0.921 0.424 4.14 4.46 5.98 0.96 1.05 0.63 0.677 0.762 0.328 0.286 0.380 0.155 0.363 0.523 0.215 0.280 0.490 0.211 0.574 1.04 0.483
n = 13 18.1 14.4 23.2 1.55 1.25 0.97 0.276 0.229 0.119 0.090 0.085 0.036 0.064 0.067 0.027 0.116 0.146 0.062
n = 16 19.9 15.7 23.5 1.11 0.875 0.654 0.142 0.116 0.060 0.039 0.036 0.015 0.026 0.027 0.011 0.059 0.074 0.031
n = 20 21.2 16.8 23.7 0.712 0.552 0.404 0.060 0.048 0.025 0.012 0.011 0.004 0.008 0.008 0.003 0.022 0.027 0.011
3.29 3.48 3.94 0.615 0.656 0.362 0.474 0.524 0.223 0.164 0.216 0.088 0.248 0.353 0.146 0.181 0.314 0.136
2.60 2.74 2.72 0.40 0.419 0.224 0.342 0.372 0.157 0.0956 0.124 0.0506 0.172 0.242 0.100 0.118 0.203 0.0581
1.92 2.02 2.00 0.23 0.237 0.123 0.226 0.244 0.102 0.0469 0.0607 0.0247 0.107 0.150 0.062 0.067 0.115 0.0496
Table 13.4: hFunctions for Single Edge Notched Specimen, (Kumar et al. 1981)
Victor Saouma
Fracture Mechanics
Draft
13.9 Engineering Approach to Fracture
P, ∆/2
23
L x2 x b W L 1 a
P, ∆/2
Figure 13.17: Double Edge Notched Specimen Plane Strain P0 = (0.72W + 1.82b)σy √ Plane Stress P0 = 4bσy / 3 β = g1 = g4 = 1 g2 = g3 = W/a − 1 1+ a b
2
(13.85) (13.86) (13.87) (13.88) (13.89)
−
a b
∆ is the loadpoint displacement at the centerline of the specimen, and δ is the crack mouth opening displacement. Axially Cracked Pressurized Cylinder: Table ??, Fig. 13.18.
Ro
Ri
p
p
a
b W
Figure 13.18: Axially Cracked Pressurized Cylinder
Victor Saouma
Fracture Mechanics
Draft
24
n=1 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 0.572 0.732 0.063 1.10 1.56 0.267 1.61 2.51 0.637 2.22 3.73 1.26 3.16 5.57 2.36 5.24 9.10 4.73 14.2 20.1 12.7 0.583 0.853 0.0729 1.01 1.73 0.296 1.29 2.59 658 1.48 3.51 1.18 1.59 4.56 1.93 1.65 5.90 3.06 1.69 8.02 5.07 n=2 0.772 0.852 0.126 1.320 1.63 0.479 1.83 2.41 1.05 2.43 3.40 1.92 3.38 4.76 3.29 6.29 7.76 6.26 24.8 19.4 18.2 0.825 1.050 0.159 1.23 1.82 0.537 1.42 2.39 1.04 1.47 2.82 1.58 1.45 3.15 2.14 1.43 3.37 2.67 1.43 3.51 3.180 n=3 0.922 0.961 0.200 1.38 1.70 0.698 1.92 2.35 1.40 2.49 3.15 2.37 3.45 4.23 3.74 7.17 7.14 7.03 39.0 22.7 24.1 1.02 1.23 0.26 1.36 1.89 0.77 1.43 2.22 1.30 1.38 2.34 1.69 1.29 2.32 1.95 1.22 2.22 2.06 1.22 2.14 2.16
a W
J INTEGRAL
=
1 8
a W
=
1 4
a W
=
3 8
a W
=
1 2
a W
=
5 8
a W
=
3 4
a W
=
7 8
a W
=
1 8
a W
=
1 4
a W
=
3 8
a W
=
1 2
a W
=
5 8
a W
=
3 4
a W
=
7 8
n=5 n=7 Plane Strain 1.13 1.35 1.14 1.29 0.372 0.57 1.65 1.75 1.79 1.80 1.11 1.47 1.92 1.84 2.15 1.94 1.87 2.11 2.43 2.32 2.70 2.37 2.79 2.85 3.42 3.28 3.46 2.97 3.90 3.68 8.44 9.46 6.64 6.83 7.63 8.14 78.4 140. 36.1 58.9 40.4 65.9 Plane Stress 1.37 1.71 1.55 1.87 0.504 0.82 1.48 1.54 1.92 1.91 1.17 1.49 1.34 1.24 1.86 1.59 1.52 1.55 1.17 1.01 1.67 1.28 1.56 1.32 1.04 0.88 1.45 1.06 1.44 1.09 0.979 0.83 1.30 0.96 1.31 0.97 0.979 0.84 1.27 0.97 1.30 0.98
n = 10 1.61 1.50 0.911 1.82 1.81 1.92 1.68 1.68 2.20 2.12 2.01 2.68 3.00 2.48 3.23 10.90 7.48 9.04 341.0 133.0 149.0 2.24 2.38 1.41 1.58 1.85 1.82 1.09 1.28 1.41 0.845 0.944 1.01 0.737 0.790 0.809 0.701 0.741 0.747 0.738 0.775 0.779
n = 13 1.86 1.70 1.30 1.86 1.79 2.25 1.49 1.44 2.09 1.91 1.72 2.40 2.54 2.02 2.66 119.0 7.790 9.4 777.0 294.0 327.0 2.84 2.96 2.18 1.59 1.80 2.02 0.97 1.07 1.23 0.732 0.762 0.809 0.649 0.657 0.665 0.630 0.636 0.638 0.664 0.663 0.665
n = 16 2.09 1.94 1.74 1.89 1.78 2.49 1.32 1.25 1.92 1.60 1.40 1.99 2.36 1.82 2.40 11.3 7.14 8.58 1570.0 585.0 650.0 3.54 3.65 3.16 1.59 1.75 2.12 0.873 0.922 1.07 0.625 0.630 0.662 0.466 0.473 0.487 0.297 0.312 0.318 0.614 0.596 0.597
n = 20 2.44 2.17 2.29 1.92 1.76 2.73 1.12 1.05 1.67 1.51 1.38 1.94 2.27 1.66 2.19 17.4 11.0 13.5 3820.0 1400.0 1560.0 4.62 4.70 4.73 1.59 1.70 2.20 0.674 0.709 0.830 0.208 0.232 0.266 0.020 0.028 0.032
0.562 0.535 0.538
Table 13.5: hFunctions for Double Edge Notched Specimen, (Kumar et al. 1981)
Victor Saouma
Fracture Mechanics
Draft
13.9 Engineering Approach to Fracture 2bσ √ y 3Rc = Ri + a a = W = 1
25
p0 = Rc g1 g2
(13.90) (13.91) (13.92) (13.93)
δ is the crack mouth opening displacement. Note that in the elastic range: 2√ 2pR0 πa a Ri KI = 2 − R2 F W , R R0 0 i 2a 8pR0 a Ri , δe = 2 − R2 )E V W R0 (R0 i where the dimensionless functions F and V are tabulated in Table 13.7.
n=1
a W a W a W a W
(13.94) (13.95)
n=2
= = = =
1 8 1 4 1 2 3 4
h1 h2 h1 h2 h1 h2 h1 h2 h1 h2 h1 h2 h1 h2 h1 h2 h1 h2 h1 h2 h1 h2 h1 h2
6.32 5.83 7.00 5.92 9.79 7.05 11.00 7.35 5.22 5.31 6.16 5.56 10.5 7.48 16.10 9.57 4.50 4.96 5.57 5.29 10.80 7.66 23.10 12.10
a W a W a W a W
= = = =
1 8 1 4 1 2 3 4
a W a W a W a W
= = = =
1 8 1 4 1 2 3 4
n=3 = 1 5 7.93 9.32 7.01 7.96 9.34 9.03 8.72 7.07 10.37 9.07 6.97 6.01 5.54 2.84 3.86 1.86 W 1 = 10 Ri 6.64 7.59 6.25 6.88 7.49 7.96 6.31 6.52 11.6 10.7 7.72 7.01 8.19 3.87 5.40 2.57 W 1 = 20 Ri 5.79 6.62 5.71 6.20 6.91 7.37 5.98 6.16 12.80 12.80 8.33 8.13 13.10 5.87 7.88 3.84
W Ri
n=5 11.50 9.49 9.59 7.26 5.61 3.70 1.24 0.56 8.76 7.65 8.08 6.40 6.47 4.29 1.46 0.71 7.65 6.82 7.47 6.01 8.16 5.33 1.90 1.01
n=7 13.1 10.6 9.71 7.14 3.52 2.28 0.83 0.26 9.34 8.02 7.78 6.01 3.95 2.58 1.05 0.37 8.07 7.02 7.21 5.63 4.88 3.20 1.23 0.45
n = 10 14.94 11.96 9.45 6.71 2.11 1.25 0.493 0.129 9.55 8.09 6.98 5.27 2.27 1.37 0.787 0.232 7.75 6.66 6.53 4.93 2.62 1.65 0.883 0.24
Table 13.6: hFunctions for an Internally Pressurized, Axially Cracked Cylinder, (Kumar et al. 1981)
Circumferentially Cracked Cylinder: Table 13.8, Fig. 13.19. P0 = Victor Saouma
2 2 2πσy (R0 − Ri ) √ 3
(13.96)
Fracture Mechanics
Draft
26
W RI W RI W RI
J INTEGRAL
= = =
1 5 1 10 1 20
a W
F V1 F V1 F V1
= 1 8 1.19 1.51 1.20 1.54 1.20 1.54
a W
= 1 4 1.38 1.83 1.44 1.91 1.45 1.92
a W
= 1 2 2.10 3.44 2.36 3.96 2.51 4.23
a W
= 3 4 3.30 7.50 4.23 10.40 5.25 13.50
Table 13.7: F and V1 for Internally Pressurized, Axially Cracked Cylinder, (Kumar et al. 1981)
σ
σ
Ri Ro
a W
b
σ
σ
Figure 13.19: Circumferentially Cracked Cylinder
Victor Saouma
Fracture Mechanics
Draft
13.9 Engineering Approach to Fracture Rc = Ri + a a g1 = g4 = W g2 = g3 = 1 δ is the crack mouth opening displacement. In the elastic range KI √ = σ πaF 4σaV1 4σaV2 a Ri , W R0
a Ri W , R0
27 (13.97) (13.98) (13.99)
(13.100) (13.101) (13.102)
δe = ∆ce =
E
a Ri W , R0
E
where the functions F , V1 , and V2 are tabulated in Table 13.9.
13.9.2
Numerical Example
From (Anderson 1995), Consider a single edge notched panel with W = 1 m, a = 125 mm. Determine J in terms of the applied load assuming plane stress conditions, neglect plastic zone correction. Assume: σyld = 414 MPa, n = 10, α = 1.0, E = 207, 000 MPa, εyld = σyld /E = 0.002. 1. From Eq. 13.81 b = W − a = 1000 − 125 = 875 mm 125 a = = 0.143 b 875 β = = 1+ a b
2
(13.103a) (13.103b) (13.103c) (13.103d) (13.103e) (13.103f)
−
a b
1 + (0.143)2 − 0.143 = 0.867
P0 = 1.072βbσy = (1.072)(0.867)(414) MPa(875) mm(25) mm = 8.42 MN
2. From Table 13.4, for a/W = 0.125, and n = 10, h1 =4.14. Thus the fully plastic J is given by Eq. 13.70a Jp = αεy σy bg1 (a/W )h1 (a/W, n)(P/P0 )n+1 = (1.0)(0.002)(414, 000) kPa(0.875) m = 2.486 × 10−8 P 11 where P in in MN, and Jpl in in kJ/m2 . 125 P (4.14) 1000 8.42 MN (13.104a)
10+1
(13.104b) (13.104c)
Victor Saouma
Fracture Mechanics
17 5.67 1.91 1.59 2.30 0. (Kumar et al.08 3.09 9.11 3.50 1.55 1.79 3.45 0.546 0.26 1.56 2.94 5.369 3.10 3.00 4.23 6.08 2.76 1.05 1.47 1.40 4.45 5.91 4.71 0.69 7.49 2.77 5.68 4.57 2.82 7.87 3.79 3.9 5.10 5.330
a W
=
1 8
a W
=
1 4
a W
=
1 2
a W
=
3 4
a W
=
1 8
a W
=
1 4
a W
=
1 2
a W
=
3 4
a W
=
1 8
a W
=
1 4
a W
=
1 2
a W
=
3 4
Table 13.03 7.44 3.78 4.07 6.95 5.37 0.8 5.77 1.36 2.51 3.46 1.03 2.48 0.31 0.366 11.43 2.07 4.23 2.17 6.1 10.39 4.64 5.56 9. 1981)
Victor Saouma
Fracture Mechanics
.93 5.30 0.39 4.58 757 5.8: hFunctions for a Circumferentially Cracked Cylinder in Tension.61 3.07 7.555 5.23 7.93 2.Draft
28
n=1 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 h1 h2 h3 3.26 W 1 = 20 Ri 5.12 7.673 4.49
n = 10 11.22 1.82 0.19 9.94 7.00 5.18 2.62 6.23 1.16 7.36 1.30 6.18 4.82 6.69 6.54 4.55 6.725 0.96 4.88 4.30 3.1 10.53
W Ri
J INTEGRAL
n=5 7.68 0.13 5.18 1.65 6.40 0.09 3.59 4.57 1.86 4.94 6.29 6.56 1.43 0.98 1.04 W 1 = 10 Ri 5.04 6.84 1.19 1.01 4.12 5.31 2.36 n=2 n=3 = 1 5 5.13 2.79 4.39 1.808 0.110 0.90 2.21 3.10 0.48 2.55 5.56 6.79 1.52 2.31 4.64 5.25 1.78 2.686 7.0 4.50 9.41 6.99 9.12 3.46 4.30 0.93 6.35 1.28 3.57 1.22 5.68 6.80 6.85 2.13 7.88 1.07 0.772
n=7 8.47 0.63 6.22 6.36 11.00 2.79 1.67 1.99 4.07 2.69 2.548 4.13 6.60 2.54 7.46 1.560 0.09 5.96 6.07 2.97 2.33 4.17 4.49 6.35 6.94 1.59 0.70 1.1 11.34 1.52 0.57 2.74 1.18 1.40 4.63 5.36 1.07 2.89 5.775 7.67 1.78 4.73 1.36 5.26 2.325 0.638 0.59 2.59 3.71 0.316 0.09 5.78 3.90 5.51 3.30 4.59 3.04 4.733 1.99 1.
025)2 m2 (1.23
− 21. V1 .59 0.105c)
1.770)2 = 4.32 1.55 0. 1981) 3.56 a W
2
1.89 5.18 1.29 1.49 4.15 3.76 0.885 2.43 0. (Kumar et al.16 1.99 2.486 × 10
2 −8
(13. and Jel in in kJ/m2 .04
a W
= 3 4 2.67 2.72 2. The total J is J = Jel + Jpl = 4.0) m(207.584P 2 (0.26 1.9 Engineering Approach to Fracture
a W
29
a W
W RI
=
1 5
W RI
=
1 10
W RI
=
1 20
F V1 V2 F V1 V2 F V1 V2
= 1 8 1.105b) σ πa (13. and the Elastic J is given by Jel = KI = =
2 KI E
(13.Draft
13.32
a W
= 1 2 1.117 1.82 2.22
= 1 4 1.09 2. 000P 2 (0.67 0.9: F .12 − 0.22 1.4.584P + 2.26 1. and V2 for a Circumferentially Cracked Cylinder in Tension.81 0. 7. KI is given by Eq.19 1.61 2.49 0.106b)
1000 900 800 700 JElastic JPlastic J
J Integral [kJ/m ]
2
600 500 400 300 200 100 0
0
1
2
3
4 5 Applied Load [MN]
6
7
8
9
Victor Saouma
Fracture Mechanics
.74
Table 13.03 3.76 1. 000) MPa
where P in in MN.106a) P
11
(13.42
a W
4
√ (13.105a) a W + 10.74
a W
3
+ 30. 4.255 1.743 1.84 0.36 1.
116)
where H= E
E 1−ν 2
plane strain plane stress
(13. Hence. the Ti are the components of traction. there exists more than one pair of stress intensity factors (or none). this gives J1 = J2 =
Γ
Γ
wdy − t
∂u dΓ ∂x ∂u wdx − t dΓ ∂y
(13.
74 Rice has shown that in linear elastic fracture. J can be written as: 2 2 2 (13. 13.115 and 13.108) (13.113) (13.109)
where w is the strain energy density.110) Jk = {wnk − ti ∂xk
which is also path independent.115) (13.116 is the intersection of a circle and a hyperbola.
75 Knowles and Stenberg (1972) noted that this can be considered as the ﬁrst component of a vector ∂ui }dΓ (13. respectively.112)
76
Hellen and Blackburn (1975a) showed that J = J1 − iJ2 (1 + ν)(1 + κ) 2 2 (KI + KII + 2iKI KII ) = 4E (13. and the contour of integration is around the tip.
Victor Saouma
Fracture Mechanics
.10
J1 and J2 Generalization.111) (13.107) (1 − ν 2 )(KI + KII ) + (1 + ν)KIII = EJ1
where J1 is expressed in indicial notation by: J1 = =
Γ
Γ
∂u dΓ ∂x ∂ui wn1 − ti dΓ ∂x1 wdy − t
(13. and in terms of the stress intensity factors.114)
77 Thus the values of energy release rates (J1 and J2 ) for crack extensions parallel and perpendicular to the crack. ni are the direction cosines of the normal to the contour.Draft
30
J INTEGRAL
13. will be given by:
J1 =
2 2 KI + KII H −2KI KII J2 = H
(13.117)
Note that solution of Eq. When written in tensorial notation.
13. Whereas the loop can have any shape. the time rate of change of the strain energy equals the integral over A of the time rate of change of the strain energy density W (εij ) less the ﬂow of this density through the loop Γ0 .11
78
Dynamic Energy Release Rate
If we consider a body containing a propagating crack. We consider the area A bounded by the outer boundary Γ. and the inner loop Γ0 . the energy release rate can not be expressed by a pathindependent contour integral because if a wave front intercepts one contour.
82
Thus.119b) (13. but not the second one. (Kanninen 1984)
13.
Victor Saouma
Fracture Mechanics
. Therefore. then two diﬀerent values would be represented by only the contour integral.
˙ U = lim
Γ0 →0 A
˙ W (εij )dA − lim
Γ0 →0 Γ0
W (εij )Vn dΓ
(13. region A is time dependent.Draft
13. (Kanninen 1984).
79 80 The rate of work P done by the tractions on Γ is equal to the rate of increase of the strain ˙ ˙ energy U and kinetic energy T in region A and the ﬂux F of energy into the crack tip region:
˙ ˙ P =U +T +F where P U T =
Γ
(13.11 Dynamic Energy Release Rate
X2
31
Γ Γ
a(t) V 0 X1
Ω
Figure 13.118)
ti ui dΓ W (εij )dA 1 ρui ui dA ˙ ˙ 2
(13. dW
1
= σij dεij
(13. Fig.120)
where Vn = V n1 = V cos θ is the component of the cracktip velocity normal to Γ0 and θ is the angle that the outward unit normal to Γ0 makes with the x1 direction.119a) (13.20
We seek to determine the energy released to the crack tip and thus consider a vanishingly small loop Γ0 surrounding the crack tip1 . it must remain ﬁxed relative to a coordinate system attached to the crack tip. the traction free crack faces.20: Dynamic Crack Propagation in a Plane Body. Hence.119c)
= =
Γ0 →0 A Γ0 →0 A
lim
lim
81 Since the loop Γ0 moves with the crack tip.121a)
in contrast to the static case.
?? thus
J = lim or
Γ0 →0 Γ0
(w + t)δ1j − σji
∂ui nj dΓ ∂x ∂ui dΓ ∂x
(13.j − σij.122 reduces to F = lim
Γ0 →0 Γ0
1 ˙ ˙ W + ρui ui Vn + ti ui dΓ ˙ 2
(13. then Eq.127) J= V
86 87 If we have a moving coordinate system x − y attached to the crack tip where x = X − a(t) and y = Y .125)
This is the rate at which energy is lost from the body due to ﬂux through Γ0 . and the kinetic energy density t (not to be confused with the traction).126)
In an increment of time dT .124)
and the equation of motion u σij.j = ρ¨i and the divergence theorem. On the other hand.130)
This equation applies to all types of material. Thus the energy release rate is F (13.j = (σij ui ). close to the crack tip the displacement changes rapidly with position (at a ﬁxed time) and the ﬁrst term dominates in all cases. the crack extends by da = V dt and the energy expended is F dt.j ui (13.j dA − lim ˙ ρ¨i ui dA − lim u ˙
W (εij )Vn dΓ
(13. 13.j ˙ ˙ = =
Γ0 →0 A Γ0 →0 A
(13.121c) (13.121b)
Γ0 →0 Γ0
lim
σij ui.Draft
32
J INTEGRAL ˙ W ˙ U ˙ T = σij εij = σij ui. Victor Saouma Fracture Mechanics
.121d)
lim
Γ0 →0 Γ0
1 ρui ui Vn dΓ ˙ ˙ 2
83
Substituting into Eq.
85
F = lim
Γ0 →0 Γ0
[(w + t)V δ1j + σji ui ] nj dΓ ˙
(13. This last equation can be rewritten in terms of the strain energy density w. 13.122)
84
We then use ˙ ˙ ˙ σij ui. the second term is equal to zero.128) ui = −V ˙ ∂x ∂t under steady condition. then ∂ui ∂ui + (13.129)
J = lim
Γ0 →0 Γ0
(w + t)dy − σji nj
(13.118 we obtain F = lim
Γ0 →0 Γ0
1 W + ρui ui Vn dΓ + ˙ ˙ 2
Γ
ti ui dΓ − lim ˙
Γ0 →0 A
(σij ui.j + ρ¨i ui )dA ˙ u ˙
(13.123) (13.
12 Eﬀect of Other Loading
33
Eﬀect of Other Loading
This section is taken from (Reich 1993)
Surface Tractions on Crack Surfaces
J integral was ﬁrst extended by (Karlsson and B¨cklund 1978) to account for the eﬀect of a surface tractions on the crack surfaces involves simply extending the deﬁnition of the contour path to include the crack surfaces. Unfortunately.2
91
Body Forces
The most straightforward method to account for body forces in the J integral is to add correction terms in the form of volume integrals such that path independence is maintained.1 dΓ) −
ˆ tj uj.1. the integrand of volume form of the standard Ji integral (i.i )exp = 0.e. For path independence to be maintained in the presence of body forces the volume integral shown above must be subtracted from the standard line integral form of the Ji integral
92
Ji =
Γ
(W ni − tj uj. Therefore. The resulting form of the J integral was then given as J =
Γ
(W dx2 − tj uj. Victor Saouma Fracture Mechanics
.i )exp is the gradient of that portion of W that depends explicitly on xi . the strain energy density W is retained in the integrand of the integral accounting for the applied ˆ surface tractions ti
90
Ji = where Γt = Γ+ ∪ Γ− .132)
13. but instead is Ji =
Ω
(bj uj.i ) dΩ
(13.i ) dΓ +
Γt
ˆ (W ni − tj uj. When body forces are present σij.i = (σjk uj. t t
Γ
(W ni − tj uj.12
13. Equation ??) does not vanish.12. therefore.1
88
13.
89 The strain energy density W is absent from the integral over Γt since the crack surfaces are coincident with the negative x1 axis and there is no need to integrate in the x2 direction. (Karlsson and B¨cklund 1978) oﬀered no proof showing that path independence a was maintained and a signiﬁcant portion of their discussion was devoted to the ﬁnite element implementation of this procedure.k .133)
where bj is the body force vector.
Proof of path independence for this procedure was later provided by (Atluri 1982) using the contour paths shown in Figure 22.i ) dΓ
(13. The proof was given for the Ji integral.131)
where Γt is that portion of the upper and lower crack surfaces between the points where Γ ends ˆ and begins and ti is the applied surface traction vector.i ) dΓ −
Ω
(bj uj.134)
which corresponds to the form proposed by (Atluri 1982).i ) dΩ = 0
(13. where (W.12.Draft
13.i ).j = −bi and W.1 dΓ
Γt
(13. This means that the strain energy density W is a function of position and that (W.
140)
96 Some changes are also required in the deﬁnition of W.137) W = Cijkl εij εkl 2 and the resulting constitutive law is
94
¯ σij = Cijkl εkl = Cijkl (εkl − α T δkl ) ¯ which is identical to Equation 22.135)
and the net eﬀective stresses σij being deﬁned as ¯ σij = ¯ ∂W ∂ εij ¯ (13.53. εjk and ε0 .k ] dΩ σ
(13. respectively.k + (uk.j ] (13.i ). the Ji integral must also be deﬁned in terms of σij rather than the total stresses σij .136)
0 It must be noted that in the absence of initial stresses σij that σij = σij .12.i ) dΓ
(13.i = but ∂ εjk /∂xi is deﬁned as ¯
∂W ∂ εjk ¯ ∂ εjk ∂xi ¯
(13.i ] = [(uj.i ).141)
∂ε0 ∂εjk ∂ εjk ¯ jk = − ∂xi ∂xi ∂xi
(13. Therefore. where σij are the ¯ total stresses.i due to the change in the deﬁnition of W .139)
¯ ¯ where ti = σij nj are the net eﬀective surface tractions. ∂εjk /∂xi jk is deﬁned as 1 1 ∂εjk = [(uj.i ).143) ∂xi 2 2 Victor Saouma Fracture Mechanics
.Draft
34
J INTEGRAL
13.138)
Since the deﬁnition of the strain energy density W is now in terms of the net eﬀective stresses ¯ σij . The chain rule is still applied to obtain
W.i − (¯jk uj. For a linear ¯ elastic material W is deﬁned as 1 ¯ ¯ (13. W will be deﬁned in terms of the net strains εij and the net eﬀective stresses σij as ¯ ¯
W =
0
εij ¯
σij d¯ij ¯ ε
(13.
(13. The ¯ appropriate form of the Ji integral is
95
Ji =
Γ
¯ (W ni − tj uj. On application of Green’s theorem to the line integral form of Ji the resulting volume integral form is Ji =
Ω
[W. σij can be replaced by σij throughout this discussion. Looking forward to the presentation of the extension in which thermal strains and pore pressures will be considered in combination.i + (uk.k ).3
Initial Strains Corresponding to Thermal Loading
93 In order to apply the Ji integral to problems in thermoelasticity the strain energy density W must ﬁrst be redeﬁned.142)
reﬂecting the contributions of the total and thermal strains.j ).
144)
Recalling that σjk = σkj and substituting the above expressions for ∂εjk /∂xi and ∂ε0 /∂xi jk into Equation 13. However.i ) dΩ = 0 ¯
(13.i ).i in the integrand ¯ of the volume form of the Ji integral Ji = −
Ω
(α σjj T.i )exp = 0.k .147 and 13.i ). The expression for W.k .i ). Moran and Nakamura 1986) if the net eﬀective stresses are replaced by the total stresses.i clearly includes the σjk (uj. the volume integral form of Ji .k = 0 and W.
13.145)
and inserted into Equation 13. in the absence ¯ σ σ of body forces σjk.12 Eﬀect of Other Loading and ∂ε0 /∂xi is deﬁned as jk ∂ε0 jk = α T.i can be rewritten as ¯ W.
99 Including body forces along with the thermal strains at this point is actually quite simple.141 the expression for W.k term that also results from (¯jk uj. ¯ This means that W is function of position due to the presence of the thermal strains ε0 and jk that (W. To maintain path independence the volume integral shown above must be subtracted from the standard line integral form of Ji yielding
98
Ji =
Γ
¯ (W ni − tj uj.i δjk ∂xi
35
(13.i ) dΩ = 0 ¯
(13.i = σjk [(uj. the strain energy density W must also be redeﬁned for problems in poroelasticity. with W being deﬁned in terms of the total strains εij and the eﬀective stresses σij as
W =
0
εij
σij dεij
(13.4
Initial Stresses Corresponding to Pore Pressures
101 As was the case when the Ji integral was applied to problems in thermoelasticity.i ).148)
100 To maintain path independence this volume integral must be subtracted from the standard line integral form of Ji to obtain
Ji =
Γ
¯ (W ni − tj uj.k = −bj in the presence of body forces the integrand of the volume form of ¯ the Ji integral will also include a term involving bj
Ji = −
Ω
(α σjj T.12.i ) dΓ +
Ω
(α σjj T. leaving only a term involving T.i ) dΩ ¯
(13.k − α T.133.i − bj uj. in this case the redeﬁnition is more cosmetic in nature.i δjk ] (13.147)
which corresponds to the form proposed by (Shih. Therefore.i − bj uj.i = (¯jk uj.140.i ) dΩ ¯
(13.150) Fracture Mechanics
Victor Saouma
. Recalling that σjk.149)
which is quite clearly a simple combination of Equations 13.i ) dΓ +
Ω
(α σjj T.146)
where σjj is the trace of the net eﬀective stress tensor.Draft
97
13.
On application of Green’s theorem to the line integral form of Ji the resulting volume integral form is
105
Ji =
Ω
[W.151)
It must be noted that this deﬁnition of W is identical to the one given in Equation ??. which is in fact what is being fractured.i ).155)
where ti = σij nj are the eﬀective surface tractions.k = p.156.k (13. since initial strains and stresses were not considered in Equation ??. albeit slightly misleading. ˆ This form of the Ji integral is valid so long as the only source of applied surface tractions ti ˆi = 0. The appropriate form of the Ji integral is
104
Ji =
Γ
(W ni − tj uj. the volume integral form of Ji . the Ji integral must also be deﬁned in terms of σij rather than the total stresses σij .i ) dΓ
(13.154) σij = Cijkl εkl + σij = Cijkl εkl − p δij
which is identical to Equation 22.62. This would also seem to be in agreement with of the classical interpretation of the eﬀective stresses (Terzaghi and Peck 1967).k δjk and W.j uj. σij = σij and this notation was valid. where ti are ˆ on the crack surfaces is from ﬂuid in the crack. However. leaving only the term involving p.i dΩ = 0 (13. where σij was used in lieu of σij . since in this case σij = 0 and t the applied eﬀective surface tractions. as is dictated by the principle of eﬀective stress 0 (13. For a linear elastic material W is deﬁned as 1 (13.k in the integrand of the volume form of Ji integral Ji = − Victor Saouma p.k .Draft
36 and the eﬀective stresses σij are deﬁned as σij = ∂W ∂εij
102
J INTEGRAL
(13.158) Fracture Mechanics
Ω
Ω
. In the absence of body forces σjk.i = (σjk uj.k δjk uj.157) W. where σij are the only stresses acting on the skeleton of the porous material.153)
103 The stressstrain relationship deﬁning the total stresses σij is obtained by simply adding 0 expressions for the initial stresses σij and the eﬀective stresses σij .i ).i dΩ = − p. Since the deﬁnition of the strain energy density W is now in terms of the eﬀective stresses σij .k ] dΩ
(13.i − (σjk uj.156)
106 Based on the expression for W given in Equation 13.i ).i = ∂εij ∂xi
and inserted in Equation 13.i can immediately be written as ∂W ∂εij = σjk (uj.152) W = Cijkl εij εkl 2 resulting in the following constitutive law σij = Cijkl εkl (13.150 and the previous proofs W.
i + p.133.i ) dΓ +
Ω
p. which is typically the case for problems in poroelasticity.64.j uj.i dΩ
(13.
111 Since the deﬁnition of the strain energy density W corresponds to that given for problems in thermoelasticity the form of the Ji integral and the gradient of the strain energy density W. Therefore.j in the integrand of the volume form of Ji integral
Ji = − Victor Saouma
Ω
(α σjj T.
13.i ). as it was in the ¯ σ same situation for problems in thermoelasticity.5
Combined Thermal Strains and Pore Pressures
110 In the presence of both thermal strains and pore pressures the strain energy density W is ¯ deﬁned in terms of the net strain tensor εij and the net eﬀective stress tensor σij . This means that W is a function of position due the presence of the pore pressures and that (W.135 for problems in thermoelasticity.k = p. just as it was ¯ in Equation 13. 112 However.k δjk rather than σjk. as deﬁned by Equation 13.161)
which is quite clearly a simple combination of Equations 13.12 Eﬀect of Other Loading
37
where p. and the expression for W.j ) uj. the volume integral form of Ji is given by Equation 13.i is given by Equation 13. this stressstrain relationship is valid only so long as all initial stresses correspond to pore pressures.i dΩ = −
Ω
(bj − p. where the eﬀective body forces bj are deﬁned in Equation 22.138. is actually quite simple at this point.i dΩ
(13.
107 To maintain path independence the volume integral shown above must be subtracted from the standard line integral form of Ji yielding
Ji =
Γ
(W ni − tj uj.i and p. the integrand of the volume form of the Ji integral can be written in one of two ways
Ji = −
Ω
bj uj. is substituted into the principle of eﬀective stress to obtain a stressstrain relationship deﬁning the total stresses σij
0 σij = Cijkl εkl + σij = Cijkl (εkl − α T δkl ) − p δij ¯
(13. and W.159)
which is analogous to the form proposed by (Shih et al.160)
with the latter form being somewhat more illuminating than the former.12.66. in the absence of body forces σ ¯jk.i = (¯jk uj. 1986) for problems in thermoelasticity. Of course.k = 0.i dΩ = 0
(13.140.
109 To maintain path independence this volume integral must be subtracted from the standard line integral form of Ji to obtain
Ji =
Γ
(W ni − tj uj.159 and 13.k .162)
which is identical to Equation 22.141.i will also correspond to those for problems in thermoelasticity.j uj.163) Fracture Mechanics
. leaving terms involving both T. Recalling that σjk.139.j is the gradient of the pore pressures. The resulting constitutive law.i )exp = 0. when thermal strains and pore pressures are considered in combination the line integral form of Ji is given by Equation 13.i ) dΩ = 0 ¯
(13.j ) uj.k = −bj in the presence of body forces.Draft
13.
108 Including body forces along with the pore pressures.i ) dΓ −
Ω
(bj − p.
13. Many problems in plasticity can be treated as such using deformation theory of plasticity (i. This means that W is a function of position due the presence of the pore pressures and that (W. if unloading occurs (during crack growth for instance).i ) dΓ +
Ω
[α σjj T.j uj.i ] dΩ = 0 ¯
(13. If surface tractions are applied in the crack the contour path for Ji integral must be expanded to include the upper and lower crack surfaces.i − (bj − p.Draft
38
J INTEGRAL
where T. To maintain path independence the volume integral shown above must be subtracted from the standard line integral form of Ji yielding Ji =
Γ
¯ (W ni − tj uj.65.k = −bj in the presence of body forces. and b) contour integral.i is the gradient of the temperatures and p. then incremental theory of plasticity should be used 3.i ] dΩ ¯
Γ
Γt
Ω
(13.i )exp = 0. Recalling that σjk.166)
which is quite clearly a simple combination of Equations 13.164 and 13.i ) dΓ + [α σjj T. citejextension: Jˆ = JKe + JKd + JKt + JKp + JKb K (13. Γ+ and Γ− .i ) dΩ ¯
(13.133.i − (bj − p. as an epilogue we note 1. however.165)
With the latter form again being the more illuminating of the two forms shown. J is valid for nonlinear elastic systems.j ) uj. nonlinear elastic systems). the integrand of the volume form of the Ji integral can be written as
Ji = −
Ω
(α σjj T.168) Fracture Mechanics
Victor Saouma
. In general. this was done in 1971 by Brobery. Several attempts are continuously made to extend J while still retaining its essential features: a) Energy release rate.13
115
Epilogue
Finally.64.i ] dΩ ¯
(13.164)
which is a simple combination of the extensions for thermal strains and pore pressures.i + p.i ) dΓ + ˆ (W ni − tj uj.i ) dΓ +
Ω
(α σjj T. When Rice derived the J integral in 1968 he did not propose it as a criteria for crack growth.167) ˆ where Γt = Γ+ ∪ Γ− and tj are the applied eﬀective surface tractions.i − (bj − p. respectively.e.
113 Including body forces along with the thermal strains and pore pressures is again a simple procedure at this point.j ) uj.i − bj uj.j is the gradient of the pore pressures. One of them. 2.j ) uj. where the ¯ eﬀective body forces bj are deﬁned in Equation 22. as deﬁned in Equation t t 22. t t Ji = ¯ (W ni − tj uj.
114 To maintain path independence this volume integral must be subtracted from the standard line integral form of Ji to obtain
Ji =
Γ
¯ (W ni − tj uj.i ) dΩ = − ¯
Ω
[α σjj T.
and Kanninen.Draft
13. Barnes. and body forces respectively. 1983.
Victor Saouma
Fracture Mechanics
. ∗ An Elasto Plastic Finite Element Investigation of Crack Initiations Under Mined Mode Static and Dynamic Loading by Ahmad.13 Epilogue
39
for elastic. dynamic. thermal. in Elasto Plastic Fracture ASTM STP 803. plastic.
Draft
40
J INTEGRAL
Victor Saouma
Fracture Mechanics
.
Draft
Part V
SUBCRITICAL CRACK GROWTH
.
Draft
.
fatigue life prediction was based on S − N curves. 9 Thus an important question that arises is “how long would it be before this subcritical crack grows to reach a critical size that would trigger failure?” To predict the minimum fatigue life of metallic structures. 14.3 1.2 between σmin and σmax . 14. 8 As in many structures one has to assume the presence of minute ﬂaws (as large as the smallest one which can be detected). The loading is usually caused by vibrations. crack propagation will occur. 14.
.1: SN Curve and Endurance Limit Diagram) using a Strength of Material Approach which did NOT assume the presence of a crack. Historically. This stage is diﬃcult to capture and is most appropriately investigated by metallurgists or material scientists. we would observe three distinct stages. or is subjected to a corrosive environment.1 (or Goodman’s
Figure 14. an understanding of the rate of fatigue crack propagation is required. Fig.
14. Fig. The application of repeated loading will cause crack growth. Fig. and to establish safe inspection intervals.1
10
Experimental Observation
If we start with a plate that has no crack and subject it to a series of repeated loading. Stage 1 : Micro coalescence of voids and formation of microcracks.Draft
Chapter 14
FATIGUE CRACK PROPAGATION
7 When a subcritical crack (a crack whose stress intensity factor is below the critical value) is subjected to either repeated or fatigue load.
Draft
2
FATIGUE CRACK PROPAGATION
Figure 14.3: Stages of Fatigue Crack Growth and compared to stage II and III it is by far the longest.2
Fatigue Laws Under Constant Amplitude Loading
12 On the basis of the above it is evident that we shall be concerned with stage II only. It is important to recognize that it is an empirical
Victor Saouma
Fracture Mechanics
. fatigue crack growth can take place under:
1. Stage III : Crack has reached a size a such that a = ac . Stage II : Now a micro crack of ﬁnite size was formed. Furthermore. 14 All of these relationships indicate that the number of cycles N required to extend a crack by a given length is proportional to the eﬀective stress intensity factor range ∆K raised to a power n (typically varying between 2 and 9). Constant amplitude loading (good for testing) 2. (K << KIc ). 2.
14.
11
Thus we shall primarily be concerned by stage II. and crack growth occurs after each cycle of loading. 3.1
Paris Model
15 The ﬁrst fracture mechanicsbased model for fatigue crack growth was presented by Paris (Paris and Erdogan 1963) in the early ’60s. Variable amplitude loading (in practice)
13 Empirical mathematical relationships which require the knowledge of the stress intensity factors (SIF). have been established to describe the crack growth rate.
14.2: Repeated Load on a Plate
Figure 14. Models of increasing complexity have been proposed.2. its SIF’well belowKIc . thus rapid unstable crack growth occurs.
Unfortunately. A remedy to this problem is the usage of numerical methods. the crack shape.3)
or N= dN = da C [∆K(a)]n (14.4)
Thus it is apparent that a small error in the SIF calculations would be magniﬁed greatly as n ranges from 2 to 6. Fig. analytical expressions for K are available for only few simple cases. Slow increase in crack growth at Kmin ≈ Kth
thus it was modiﬁed by Foreman (Foreman. C the intercept of line along dN and is of the order of 10−6 and has units of length/cycle.
17
However. it is essential to properly determine the numerical values of the stress intensity factors.1)
vs ∆K. given by da = C (∆K)n dN which is a straight line on a loglog plot of √ ∆K = Kmax − Kmin = (σmax − σmin )f (g) πa
da dN
(14.
16
Equation 14.1 can be rewritten as : ∆N = ∆a C [∆K(a)]n
af ai
(14.4 C(∆K)n da = dN (1 − R)Kc − ∆K 14.2 Fatigue Laws Under Constant Amplitude Loading
3
law based on experimental observations.6) dN (1 − R)(1−m)
Victor Saouma
Fracture Mechanics
. and load are in such a combination that an analytical solution for the SIF does not exist and large approximation errors have to be accepted.2. 1973).
18
14.2.2
19
Foreman’s Model
When compared with experimental data. of which the ﬁnite element method has achieved greatest success. Increase in crack growth rate as Kmax approaches KIc 2.5)
20 Walker’s (Walker 1970) model is yet another variation of Paris Law which accounts for the Kmin σmin stress ratio R = Kmax = σmax n da ∆K =C (14. boundary conditions.2)
da a is the crack ength. Kearney and Engle 1967). Most other empirical fatigue laws can be considered as direct extensions. it is evident that Paris law does not account for: 1. 14. and n is the slope of the line and ranges from 2 to 10. N the number of load cycles. or reﬁnements of this one. in most practical cases.2. and
(14.1 Modiﬁed Walker’s Model (14. Because of the sensitivity of N upon ∆K. Thus the stress analyst has to use handbook formulas for them (Tada et al.Draft
14.
5
14. the table lookup method extracts directly from the experimental data base the appropriate coeﬃcients.1
Examples
Example 1
An aircraft ﬂight produces 10 gusts per ﬂight (between takeoﬀ and landing).1 R = . an eﬀective stress intensity factor is given by (Broek 1986):
∆KIeﬀ = ∆KI cos3
θ0 θ0 − 3∆KII cos sin θ0 2 2
(14.5. Thus to properly use a fatigue law. In principle each of the above discussed mixedmode theories could yield a separate expression for the eﬀective stress intensity factor. and the angle of crack growth θ0 .2. in general a crack will be subjected to a mixedmode loading resulting in both ∆KI and ∆KII .Draft
4
FATIGUE CRACK PROPAGATION
Figure 14. 26 For the case of maximum circumferential stress theory.3
21
Table LookUp
Whereas most methods attempt to obtain numerical coeﬃcients for empirical models which best approximate experimental data. The aircraft is made up of Victor Saouma Fracture Mechanics
.3
R = . In a “roundrobin” contest on fatigue life predictions. ∆K is directly read (or rather interpolated) for available data.4: Forman’s Fatigue Model
14.7)
14.2. however.2.
14.
22
This method is based on the availability of the information in the following table:
da dN
R =1
23
∆K R = . Each gust has a σmax = 200 MPa and σmin = 50 MPa. this model was found to be most satisfactory (Miller and Gallagher 1981).2.4
24
Eﬀective Stress Intensity Factor Range
All the empirical fatigue laws are written in terms of ∆KI . It has two ﬂights per day.
25 One approach.4
For a given
da dN
and R. an eﬀective stress intensity factor is sought. consists in determining an eﬀective stress intensity factor ∆Kef f in terms of ∆KI and ∆KII .
583)
= 316 months ≈ 26 years!
14. How long would it be before the crack will propagate to its critical length? √ 2 √ ER Assuming K = σ πa and Kc = ER.
14. Improve the inspection so as to reduce tha ssumed initial crack length amin . If a longer lifetime is desired. 2.3
14. 4. The smallest detectable ﬂaw is 4mm.8)
da = C[∆K(a)]n
da C (σmax − σmin )n ((πa) 2 )n
(∆σ)n . 3.1 mm thus ai = .3
Example 3
Rolfe and Barsoum p.Draft
14.5 )3 1 + √.004 ]
a−1.9)
= = 10. so as to increase the critical crack length ac at instability.4×10−3
=
4×10−3
(5 × 10
−11
da = 1064 ) (200 − 50)3 (πa)1. 428 cycles
C −2128a−.428) cycles × 10 ﬂight × 1 ﬂight × 30 month ≈ 17.0084 .1
28
Variable Amplitude Loading
No Load Interaction
Most Engineering structures are subjected to variable amplitude repeated loading.5 years.5 da
(14. C = 5 × 10−11 cycle .2. Employ a diﬀerent material with higher KIc . the following questions arrise: Victor Saouma Fracture Mechanics
. then ac = σ2Kc π = σ2 π or
max max
ac = ⇒N =
af ai
(70 × 109 )(15 × 103 ) = 0.0084
day 1 1 thus the time t will be: t = (10. E = 70GP a. 14. however.004
1
8.38 month cycle 2 day ≈ 1.261263. then we can: 1.1mm 1 1 t = 2128[− √.2.3 Variable Amplitude Loading
5
m kJ aluminum which has R = 15 m2 . Reduce the maximum value of the stress σmax .428 (189. Reduce the stress range ∆σ.2
27
Example 2
Repeat the previous problem except that more sophisticated (and expensive) NDT equipment is available with a resolution of . and n = 3. Thus.0084 + √.5 .0084 = .5.5
((πa).4mm (200 × 106 )2 π
af ai
(14. 583cycles t=
1738 10.0084m = 8.0001 ] = 184.004
(∆σ)3 1 2128[− √. most experimental data is based on constant amplitude load test.5.3.
Draft
6 1.”
30
Is this correct? Why? Under which condition overload is damaging! Retardation Models
14. Accurate “block by block” numerical integration of the fatigue law ∆a = C(∆K)n ∆N solve for a instead of N . This phenomena is called Retardation.10)
n
(14.2
14. 2. (14.5. Upon load release. it would be relatively easy to establish a crack growth curve by means of a cyclebycycle integration.3. 33 During loading. However crack growth under variable amplitude cycling is largely complicated by interaction eﬀects of high and low loads.3.11)
where ∆Krms is the square root of the mean of the squares of the individual stress intensity factors cycles in a spectrum. thus “Aircraft that logged some bad weather ﬂight time could be expected to possess a longer service life than a plane having a better ﬂight weather history. NH • Without high wind related gust load.
Victor Saouma
Fracture Mechanics
. 32 A high level load occurring in a sequence of low amplitude cycles signiﬁcantly reduces the rate of crack growth during the cycles applied subsequent to the overload. Root Mean Square Model (Barsoum)
FATIGUE CRACK PROPAGATION
2. Fig.12)
14.3.2
31 Baseline fatigue data are derived under constant amplitude loading conditions.2. 14.2. Is there an interaction between high and low amplitude loading?
da = C(∆Krms )n dN ∆Krms =
k 2 i=1 ∆Ki
(14. but many structural components are subjected to variable amplitude loading. If interaction eﬀects of high and low loads did not exist in the sequence. the surrounding material is elastically unloaded and a part of the plastic zone experiences compressive stresses.1
29
Load Interaction
Observation
Under aircraft ﬂight simulation involving random load spectrum: • High wind related gust load. the material at the crack tip is plastically deformed and a tensile plastic zone is formed. How do we put the two together? 1. NL
NH > NL .
Victor Saouma
Fracture Mechanics
. 14.Draft
14. the process of growth is steady state.7. 36 When the load history contains a mix of constant amplitude loads and discretely applied higher level loads. the larger the zone of compressive stresses. there is no observable direct eﬀect of the residual stresses on the crack growth behavior. 14. As the crack propagates through this perturbed zone under the constant amplitude loading cycles. Fig. the patterns of residual stress and plastic deformation are perturbed. aoL is the crack size at which the higher load excursion occurred.
38 Thus there is retardation as long as the current plastic zone is contained within the previously generated one.2.6: Cause of Retardation in Fatigue Crack Grwoth
14. the crack growth rate returns to its typical steadystate level.3. If the load is repeated in a constant amplitude sense.3 Variable Amplitude Loading
7
Figure 14.14)
in which rpi is the current plastic zone size in the ith cycle under consideration.2. After the crack has propagated through the perturbed zone. that the plastic deformations occurring at the crack tip remain as the crack propagates so that the crack surfaces open and close at nonzero (positive) levels.
Figure 14. in essence. ai is the current crack size.1
Wheeler’s Model
37
Wheeler (Wheeler 1972) deﬁned a crackgrowth retardation factor Cp : da da = Cp dN retarded dN Cp = (14. however.6.13)
linear m
rpi aoL + rpoL − ai
(14.5: Retardation Eﬀects on Fatigue Life
34 The larger the load. it grows slower (the crack is retarded) than it would have if the perturbation had not occurred. rpoL is the plastic size generated by a previous higher load excursion. 35 Measurements have indicated. Fig. and m is an empirical constant.
17)
φ=
1−
Kmax.2. At that time.i and Kmin.i oL − 1 s
w oL KR = KR = Kmax 1 −
ai − aoL − Kmax. Thus. the stress intensity factor KI is replaced by an eﬀective one:
eﬀ KI = KI − KR
(14.7: Yield Zone Due to Overload 14.
40
Equation 14.i − KR
(14. and therefore its maximum and minimum levels (Kmax.th Kmax.15)
in which KR is equal to:
w KR = φKR
(14.i rpoL
(14. rpoL is oL the yield zone produced by the overload. the retardation eﬀect is sensed by the change in the eﬀective stress ratio calculated from:
41
Reﬀ =
eﬀ Kmin. ai − aoL = rpoL and the reduction becomes zero. are reduced by the same amount (KR ).i ).i is the maximum stress intensity for the current cycle. Kmax is the maximum stress intensity of the overload.
Victor Saouma
Fracture Mechanics
. aoL is the crack size at the occurrence of the overload.i eﬀ Kmax.18)
and ai is the current crack size. This equation shows that retardation will occur until the crack has generated a plastic zone size that reaches the boundary of the overload yield zone. and Kmax.19)
because the range in stress intensity factor is unchanged by the uniform reduction.2.Draft
8
FATIGUE CRACK PROPAGATION
Figure 14.i
=
Kmin.i − KR Kmax. Engle and Wood 1971).16) (14.2 Generalized Willenborg’s Model
39 In the generalized Willenborg model (Willenborg.15 indicates that the complete stressintensity factor cycle.3.
which is the overload (shutoﬀ) ratio required to cause crack arrest for the given material. the crack growth increment ∆ai is: ∆ai = da = f (∆K.20)
43 In this model there are two empirical constants: Kmax.Draft
42
14. which is the threshold stress intensity factor level associated with zero fatigue crack growth rate.3 Variable Amplitude Loading Thus. and S oL .th . for the ith load cycle.
Victor Saouma
Fracture Mechanics
. Reﬀ ) dN
9
(14.
Draft
10
FATIGUE CRACK PROPAGATION
Victor Saouma
Fracture Mechanics
.
Draft
Part VI
FRACTURE MECHANICS OF CONCRETE
.
Draft
.
however their capabilities are seriously challenged under tensile stresses. such as dams. it has been mostly applied to metallic materials. Hence.
However for unreinforced structures. the apparent inability of the constitutive models to properly and eﬃciently model tensile cracking is a major handicap. much was initially borrowed from the wealth of information and research previously undertaken in conjunction with metals by metallurgists or mechanical engineers. which required the alteration of existing models. it quickly became evident that by its very heterogeneous nature concrete has some unique fracture characteristics. relatively few research has been undertaken to properly understand and characterize fracture models. As such for unreinforced concrete structures prone to tensile cracking.
.Draft
Chapter 15
FRACTURE DETERIORATION ANALYSIS OF CONCRETE
15. this chapter will exclusively focus on the phenomenological aspects of concrete fracture from a material point of view. 9 By now fracture mechanics is universally acknowledged as a viable tool of analysis for investigation of concrete cracking. However. even the simplest constitutive models appear to perform reasonably well under compressive regimes. For the most part.
10 11 Whereas an inappropriately large number of papers have been published on computational models for concrete (some of which will be discussed in a separate chapter). And after many years of development on numerous (plasticity based) constitutive models for concrete.1 Introduction
7 It is ironic that although the foundation for fracture mechanics was laid by Griﬃth in the early thirties for brittle materials (such as concrete) (Griﬃth 1921). tensile cracking remains its undisputed “Achille’s heel”. This apparent inability to properly model tensile cracking is of minor importance in reinforced concrete structures in which the steel “takes over” the tensile stresses. Although there have been some pioneering eﬀorts to apply fracture mechanics to concrete (Kaplan 1961). it was not until the midseventies that a number of researchers from the academic community focused their attention on various aspects of this application. 8 In applying fracture mechanics to concrete. a fracture mechanics based model (rather than a plasticity based one) should be used.
when the tensile strength is reached.
15. Vienna (International Conference on Fracture and Damage of Concrete and Rock 1988). whereas we shall focus exclusively on concrete. Fig. For all materials. the reader is directed to the proceedings of the international conferences/workshops on Fracture of Concrete: Lausanne (Wittmann. or of strain (measured by a strain/clip gage or other instruments).. there is a gradual release of strain energy which is then transferred to surface energy during crack formation. In this case. (Carpinteri and Ingraﬀea 1984). Displacement. and Strain Control Tests
Before proceeding with the coverage of linear and nonlinear fracture mechanics of concrete. there is a sudden and abrupt brittle failure. 13 Finally. except that the feedback is provided by (“strategically positioned”) strain gage or a clip gage or an arbitrary specimen deformation Victor Saouma Fracture Mechanics
. thus the sudden failure can be explosive. of displacement (as measured by an internal displacement transducer). Strain Control: is analogous to displacement control. Zurich (Wittmann 1993). For softening material there will be a postpeak response with a gradual decrease in stress accompanying an increase in displacement. Mazars and Saouma 1994) conferences. and (Sih and DiTommaso 1984). 15. Modern testing equipment can be programmed to apply a predetermined rate of load (as measured by a load cell).2
Phenomenological Observations
14 We start by discussing some of the general phenomenological aspects associated with fracture testing and response of concrete. it is essential that the reader understand the diﬀerence in material and structural response when a specimen is subjected to uniform tensile load. z to RILEM reports. The strain energy accumulated in the specimen is suddenly released once the ultimate load of the specimen is reached. F. Displacement/Stroke Control: the crosshead applies an increasing displacement to the specimen. special ACI publications (446 1991). most of the presentation in this chapter applies also to ceramics and ﬁber reinforced ceramic. Dungar and Morris 1991) and Cahmberry (Bourdarot. Breckenridge (Baˇant 1992).Draft
2
FRACTURE DETERIORATION ANALYSIS OF CONCRETE
12 For additional information.1: Test Controls Load Control: the crosshead applies an increasing load irrespective of the specimen deformation or response. 1983). H. More specialized coverage of dam fracture can be found in the proceedings of the Boulder (Saouma.2.
15.1
Load P ∆ Displacement ε Actual Programmed Strain
t
t
t
Figure 15. Houston (Shah and Swartz 1989).1
15
Load. displacement or strain control during testing. Evanston (Shah 1984).
both obtained in a straincontrolled test of an uncracked or unnotched specimen. Not only is it not of practical usefulness. Victor Saouma Fracture Mechanics
. since a strain cannot be properly deﬁned. Subsequently. (b) Concrete response in the postpeak zone is most interesting. as illustrated in Fig. This will be discussed in the subsequent section. The descending branch of the concrete response in Fig.
σ σ f’ .2.Draft
16
15.1%. a stresscrack opening displacement is a more appropriate model.and postpeak response of metals and concrete. internal microcracking induces a nonlinear response up to a peak stress ft and a corresponding strain of approximately 0. Under load control only the prepeak response can be measured.
17 Both stroke and strain controlled tests require a closedloop experimental set up.2% 25% Steel ε ε
c
ε
. due to internal dislocation plastic deformation with strain up to 25% may result to be followed by strain hardening.1% .15. Prepeak: (a) Metal exhibits a linear elastic response up to the yield stress σyld .2: StressStrain Curves of Metals and Concrete 1.
15. which is usually expensive and was not widely available until very recently.2 Phenomenological Observations
3
(not necessarily corresponding to the loading direction). but also it is largely overshadowed by necking. 15.2 is an idealization of the average material response. Subsequently. and an approximate strain of 0. A more accurate description should account for the localization of the induced cracks. 2.01%. Postpeak: (a) Metals response in the postpeak zone is not yet well understood. To accomplish this test a clip gage or a strain gage has to provide the feedback signal to the testing equipment in order to accordingly adjust the stroke.2
18
Pre/PostPeak Material Response of Steel and Concrete
As an introduction to concrete fracture. as it can exhibit additional strains.6f’ c . Thus away from the localized crack there is an elastic unloading. (b) Concrete exhibits a linear response up to approximately 0.01% Concrete
Figure 15. let us compare the pre. and at the crack. Most concrete fracture tests are conducted under strain control with a gage located at the mouth of the crack providing the specimen response.6ft .2.
we pose the next question.3. and 3 snapback. At this location.
24
Victor Saouma
Fracture Mechanics
. and instead we should characterize the crack by its crack opening displacement COD which will then be plotted along with the stress (σCOD). 15. and ε = u/L.3
15. In either case.
If we were to homogeneize by taking σ = F/A.
F. u
1 F 2 1 3 F F
u
Elastic Unloading Elastic Softening
u 2 3
u
Snapback
Figure 15. transducer 1 will record elastic unloading. and having attributed the ductile response of metals to internal dislocation. loaded uniaxially by a tensile load F . Fig. 2 strain softening. we can record elongation by an LVDT (linearly variable displacement transducer) mounted as in conﬁguration 1. the area under the stressstrain curve represents the energy that can be absorbed by a unit fracture surface and thus concrete will exhibit a more brittle response than metals.
23
This capability of transmitting stresses across a crack under controlled displacement is a characteristic of softening materials. 2 or 3. Having experimentally observed that the macroscopic response of concrete exhibits a softening response.Draft
4
19 20
FRACTURE DETERIORATION ANALYSIS OF CONCRETE
Postpeak responses can be obtained only under displacement or strain controlled tests. and cracking is initiated. it is clear that we can not provide a unique deﬁnition of the strain across the crack. Why is concrete softening?
15.1
Localisation of Deformation
Experimental Evidence
21 If we consider a concrete specimen. the strain can no longer be deﬁned.3: Caputring Experimentally Localization in Uniaxially Loaded Concrete Specimens
22 Once the peak load is reached.3.
A. but drastically diﬀerent results can be obtained from diﬀerent geometries. the softening curve is often indirectly determined by testing notched specimens. characterized by: (a) peak stress at its tip equal to the tensile strength of concrete (b) decreasing stress distribution from ft at the tip of the ﬁctitious crack to zero at the tip of the physical crack It should be noted that along the FPZ. True or physical crack across which no stresses can be transmitted. Hillerborg’s Model
5
25 From the previous discussion. Fig.
In what is probably the most referenced work in the nonlinear fracture of concrete literature. Along this zone we have both displacement and stress discontinuities.3 Localisation of Deformation σCOD Diagram. Hillerborg (Hillerborg. P. the crack is composed of two parts. 1976) presented in 1976 a very e simple and elegant model which has been previously described qualitatively. M.1
15. In this model.Draft
15.3. Hordijk. 2.
Victor Saouma
Fracture Mechanics
. However. and Mod´er. Fictitious crack. and testing machines. and Petersson. we have displacement discontinuity and stress continuity. or Fracture Process Zone (FPZ) ahead of the previous one. The exact charachterization of the softening response should ideally be obtained from a uniaxial test of an uncracked specimen. it has been found (Li and Liang 1986.4:
26
Figure 15.E. sizes. it is clear that concrete softening is characterized by a stresscrack opening width curve (and not stressstrain).1. Hence.4: Hillerborg’s Fictitious Crack Model 1. 15. Reinhardt and Cornelissen 1989) that not only are those tests extremely sensitive.
a value of 1. 4.P. Z. In addition. 3. (Br¨hwiler. For gravity dams.Draft
6
FRACTURE DETERIORATION ANALYSIS OF CONCRETE
27 This model is among the most widely used in nonlinear fracture mechanics ﬁnite element analysis.
1. GF : or fracture energy. few “engineering” structures have been analyzed. 1984) found that GF may z z be predicted (with a coeﬃcient of variation of about 16%) from the following empirical equation: da (15. Broz. Baˇant and Oh (Baˇant. da is the aggregate size in inches. A topic of much research lately has
Stress
f’_t
s_1 G_F w_1
Crack Opening
w_2
w
Figure 15. E. 15. is recommended. Br¨hwiler and Boggs 1991). however due to the computational complexity. The area under the curve.. Note that for arch dams. Shape of the softening diagram (σ − COD). (a) The ﬁrst part has been associated with (unconnected) microcracking ahead of the stressfree crack (b) The second part with bridging of the crack by aggregates 2.5: Concrete Strain Softening Models
been the experimental determination of the fracture energy GF . laboratory tests could be u performed on recovered cores to obtain a better indication of GF . and the resulting shape Victor Saouma Fracture Mechanics
.1) GF = 0.0214(ft + 127)ft 2 Ec where Ec and ft are in pounds per square inch.5. 5. is a measure of the energy that needs to be spent to generate a unit surface of crack.35 × 10−3 kip/in. With reference to Fig. this value could u probably be increased on the basis of laboratory tests. and in general a bilinear model for the strain softening should be used. Also. 1988). There is an inﬂection point in the descending branch. By analyzing numerous test data. (Saouma. termed the fracture energy GF (not to be confused with Gc or critical energy release rate).
Mihashi. F. 1988). Gopalaratnam and Shah 1985. this tensile strength can not be taken as zero... Fracture Mechanics
29
Victor Saouma
. and Simonin.2. Fig.3 Localisation of Deformation
7
of the softening diagram (Cedolin. H. 15. Test results (Br¨hwiler and Wittmann 1990) show that the u fracture properties generally increase with increasing loading rate. For dynamic analysis.3) (15. Jeang and Hawkins u 1985.. and the fracture energy GF . This simple model can be uniquely deﬁned in terms of the tensile u strength ft .2
15. Rokugo. As ft is seldom determined experimentally.7)
w1 = 0.4ft GF w1 = 0.5) (15. Dei Poli and Iori 1987..3.4)
whereas for structural concrete. (Mindess and Young 1981).... ??. and Simonin. Rokugo. In order to assess the relevance of the exact value of GF and the softening curve shape on numerical simulations.H.Draft
15. otherwise there will be no fracture process zone. 6. (Wittmann. P.. u P. H. In (Br¨hwiler and Wittmann 1990). Mihashi. F. Duda 1990. 1988). F. In lieu of a direct tension test. K. a ﬂexural test can be performed under strain control. 1988.3. it is assumed to be 9% of fc . Br¨hwiler. we have a linear stress strain relationship σ = Eε. and the fracture energy GF could still be determined from the area under the load and corresponding displacement curve. Wittmann.2) (15. However. 7.. Br¨hwiler.5.. and Simonin.6) (15.. Br¨hwiler. P. Fig. E. dynamic compressive preloading leads to a reduction of the fracture properties at both quasistatic and high loading rates. K.. Giuriani and Rosati 1986). Mihashi.1
28
Theoretical Evidence
Static Loading
Let us consider a simple bar loaded in uniform tension and made up of m elements.
15. the fracture properties of dam concrete depend on both rate of loading and preloadings.75 w2 = 5
GF ft
where ft is the uniaxial tensile strength. it was found that the optimal points for concrete with 1” maximum size aggregate are: s1 = 0. Petersson 1981.H. E. K.8 ft GF w2 = 3 ft (15.H. three diﬀerent set of fracture experiments are analysed using the average reported fracture energy. the corresponding values are: s1 = ft 4 GF ft (15. Prior to reaching the peak stress ft . Rokugo. Within the context of a nonlinear fracture mechanics analysis. The shape of the softening diagram is assumed to be the bilinear one proposed in (Wittmann. u H. E.
Upon failure of this element. it fails.12) Fracture Mechanics
Victor Saouma
. At that point the load carrying capacity of the bar is exhausted. in the postpeak zone the displacement will be
32
u = εf =
εe = ε = ⇒ε =
L L εf + (m − 1) εe m m ft σ − ft + E h σ E u L E − h σ − ft σ + E Eh m
(15. and for linear softening ft (15. the concrete softens and the peak strain is εu . 31
The postpeak stress is given by σ = ft + h(ε − ε0 ) (15.10e)
33
The slope of the stressdisplacement curve is u ˙ = σ ˙ E−h 1 + E Ehm εu ε0 (15.10d) (15.11)
34
If we deﬁne n=
(15. the other ones will simply unload elastically.8)
for degrading (softening) material h (softening modulus) is less than zero.9) h=− εu − ε0 where ε0 = ft /E. Thus. when this element reaches its own tensile strength (lower than ft ). We next assume that one of the elements is weaker than the other m−1 ones.6: StrainSoftening Bar Subjected to Uniaxial Load
30 Once the peak load is reached.10a) (15.10c) (15.Draft
8
1 0 1 0 1 0 1 0 1 0 1 0
FRACTURE DETERIORATION ANALYSIS OF CONCRETE
σ
ft
Weak Element
111 000 111 000 111 000 111 000 111 000 111 000
m elements
σ
m= m=n E
h 1 m=2 m=1
L/m
L
1
ε0
ε u /2
εu
ε
Figure 15. Thus.10b) (15.
000.10e reduces to ε= σ n(ft − σ) + E mE
9
(15. As a result we would have: • Loss of local material stability • Loss of structural stability • Loss of ellipticity • Mesh dependence
To illustrate this. let us consider a concrete member subjected to uniaxial load with the following properties: ft = 300 psi. in all cases the maximum displacement will be umax = wmax = .02 inches and the corresponding load will obviously be equal to zero. For specimens larger than 200 inches. GF = 1 300 × .
37
In all cases the peak load will be Pmax = 300A where A is the cross sectional area (since results are independent of A we shall assume A=1). 3.000 Ksi.3 Localisation of Deformation then h = −E/(n − 1) and Eq.02 = 200 inches. Physically.13 and 15. For a 200 inch specimen. 15. Specimens less than 200 inch long do exhibit a softening branch which can be experimentally obtained only through displacement controlled tests.02 = 3lb/in.02 in. which can only be hypothesized but not experimentally obtained. For m = 1 we have the softening curve reproduced.e. however for m = n we do have a brittle failure. .
39
The displacement corresponding to the peak load will be equal to wmax if L = wmax = u . 15. wmax =. there is a snapback. then from Eq.7 illustrates the load displacement curves in terms of the length L.13)
and the postpeak slope is ˙ 1 n ε = − σ ˙ E mE
35
(15.
38
Similarly.
Victor Saouma
Fracture Mechanics
.000 = . snapback behavior. i. 2 300 From these properties. the peak strain is εu = 3. 15. we note that depending on the number of element used (m).14)
From Fig. and for m < n we would have a strain decrease. Physically. This is impossible to achieve experimentally. The corresponding displacement will be given by u = εu × L.14 we note that we would have an elastic unloading. we do have a perfectly brittle response. 2. ??.Draft
15. Let us now consider the load displacement curve (P − u) of this member for various lengths L. For m = 2 half the bar localizes.0001. Fig. this is impossible as it would imply failure occured without dissipation of energy. there is a wide range of possible responses.
36
If we use an inﬁnite number of elements (m → ∞). E =3. that implies that the localization zone can not absorb the elastic energy released by the elastic part of the bar.0001 This ﬁgure calls for the following observations:
40
1.
Draft
10
FRACTURE DETERIORATION ANALYSIS OF CONCRETE
300.0
200.0
Load (Lbs)
100.0
L=10 in. L=100 in. L=200 in L=300 in. L=400 in.
0.0 0.000
0.010
0.020 0.030 Displacement u (inch)
0.040
Figure 15.7: Load Displacement Curve in terms of Element Size L
U A
10 .15
100 1.5
200 3.0
300 4.5
400 6.0
Table 15.1: Strain Energy versus Fracture Energy for uniaxial Concrete Specimen 4. Large specimens could still be tested, however rather than using the entire specimen length as “gage length” (or displacement/stroke control test), we would have to use a smaller gage length (through strain control tests) across a potential crack. Thus a notched specimen should be used. In summary, the load displacement curve of this simple test is clearly size dependent. The larger the specimen, the larger the stored strain energy which would be released to form surface energy.
41
An energetic interpretation of this ﬁgure would lead us to compare the elastic energy U = = 1 300 × .0001AL = .015AL at peak load with the fracture energy GF = 3. As long 2 as the strain energy is smaller than the fracture energy, then we do have structural softening; if the two energies are equal, then we would have a perfectly brittle response; ﬁnally, if the elastic energy exceeds the fracture energy, then we would have a sudden failure with snapback. Table 15.1 tabulates those energy values.
42
1 2 ft u AL
Again we observe that U = GF for L = 200 inches. Thus, as long as the energy released A can be transformed into fracture energy, we do have a stable conﬁguration. However if the accumulated strain energy being released (including not only the one stored in the specimen, but also in the experimental setup) is greater than the one which can be absorbed to create
43
Victor Saouma
Fracture Mechanics
Draft
44
15.3 Localisation of Deformation new surface energy (cracks), then we do have instability.
11
This simple exercise can be generalized to the analysis of simple tensile structures in which elements are either perfectly brittle and/or softening.
45 Finally, and as a minor remark, for a successful postpeak experiment the total strain energy (of the specimen and of the testing frame) should be less than the fracture energy. Hence, to avoid snapbacks the testing frame should be as stiﬀ as possible.
15.3.2.2
Dynamic Loading
Adapted from (Sluys 1992) 15.3.2.2.1 Loss of Hyperbolicity
46
We can write the dynamic equilibrium of a system along the x axis as ρ ∂σxz ∂σxx ∂σxy ∂2u + + = 2 ∂t ∂x ∂y ∂z (15.15)
For a uniaxial state of stress, and with σxx = f (εxx ), this equation reduces to c2 ∂2u ∂2u = 2 2 ∂x ∂t (15.16)
where c2 = 1/ρ ∂f and is the square of the speed of the wave propagation. Equation 15.16 is the ∂ε classical wave equation of motion (hyperbolic equation). D’alembert’s solution of this equation is of the form u = f (x + ct) + g(x − ct) (15.17) and a real solution exists as long as ∂f /∂ε ≥ 0. Thus, in a softening regime where ∂f /∂ε < 0, the speed of wave propagation becomes complex and the problem is thus illposed. We thus have a loss of hyperbolicity. 15.3.2.2.2 Wave Equation for Softening Maerials
47
Considering Equation of Motion: ρ ∂σ ˙ ∂x ∂u ˙ Kinematic Equation: ε = ˙ ∂x Constitutive Equation: σ = f ε = ∂2u ∂t2 (15.18a) (15.18b) (15.18c)
where u is the velocity v, and the strain can be decomposed into an elastic and inelastic ˙ component. (15.19) ε = εe + εi and thus ˙ σ = f εi ˙ Victor Saouma (15.20) Fracture Mechanics
Draft
12
48
FRACTURE DETERIORATION ANALYSIS OF CONCRETE
where f denotes diﬀerentiation with respect to the inelastic strain, Taking εe = σ/E, the x derivative of the preceding equation yields ˙ ˙ ∂σ ˙ ∂x ∂ εi ˙ ∂x ∂ ε ∂ ε˙e ˙ = f[ − ] ∂x ∂x ˙ ∂2u 1 ∂σ ˙ ] = f[ 2− ∂x E ∂x f E ∂2v = E + f ∂x2 = f (15.21a) (15.21b) (15.21c) (15.21d)
⇒
∂σ ˙ ∂x
Substituting this equation into Eq. 15.18a we obtain the wave equation for a onedimensional strainsoftening element
49
f + E ∂2v ∂2v −f =0 c2 ∂t2 ∂x2 e
(15.22)
where ce = E/ρ is the linear elastic longitudinal wave velocity. This equation should be contrasted with Eq. 15.16 for the elastic solids. One way to solve this second order partial diﬀerential equation (linear only if f is constant) is to consider the variation of the ﬁrst derivative of the velocity v with respect to both x and t
50
∂v ∂t ∂v d ∂x d
= =
∂2v ∂2v dx dt + ∂t2 ∂x∂t ∂2v ∂2v dt dx + ∂x2 ∂x∂t
(15.23a) (15.23b)
51 Combining the last three equations together yields a system of three second order diﬀerential equations ∂2v ∂t2 (E + f )/c2 0 −f e ∂2v dt dx 0 =0 (15.24) ∂x∂t ∂2v 0 dt dx 2
∂x
52
The characteristic determinant is (E + f )/c2 0 −f e dt dx 0 0 dt dx = E+f dx2 − f dt2 c2 e (15.25)
If the determinant is equal to zero, then a solution for the curve in the x − u − t plane should coincide with the characteristic direction. dx = ±ce dt f E+f (15.26)
53 In a wave equation, the characteristic (±dx/dt) coincides with the wave speed (±c). In case of softening (f < 0) and if we have snapthrough (f > −E), then the characteristics and
Victor Saouma
Fracture Mechanics
Draft
15.3 Localisation of Deformation
13
therefore the wave speeds are imaginary. This results in a loss of hyperbolicity into ellipticity whenever softening occurs.
54 We thus have a domain governed by hyperbolicity, and another one by ellipticity where the wave can not propagate. Spatial interaction between the two is impossible, and we have an illposed problem. 55 A strainsoftening material is thus nondispersive. In a dispersive media, harmonic waves with diﬀerent frequency propagate with diﬀerent wave speeds. 56
Assuming that we have a single linear harmonic wave propagation v(x, t) = Aei(kx−ωt) (15.27)
where ω is the angular frequency, k = 2π/λ the wave number. Substituting Eq. 15.27 into 15.22 we obtain f k (15.28) ω = ce E+f ω (15.29) k it is now apparent that in a softening material c = cf is independent of k and is therefore nondissipitative, thus all the waves would have the same imaginary speed. cf =
57
The wave velocity being equal to
For proper description of localisation, wave propagation must be dispersive.
15.3.3
Conclusion
In summary, Fig. 15.8 illustrates the localization of strain in a plate subjected to uniform tensile deformation.
P=30 Elastic Loading 716 715 714 713 100 50 y 0 −10 10 0 x 50 y 0 −10 10 0 x P=50 P=70
3500
430 429 428 427 100
1200 1000 800
2500 3000
600 100 50 y 0 −10 10
Strain
2000
0 x
Localized Failure
1500 Elastic Unloading 1000
Peak Load Localization of Damage
P=80
P=90
P=100
1500 1000
2000
4000
500
Elastic Loading
1000 500 0 100 50 y 0 −10 10 0 x 0 100 50 y 0 −10
2000
0 10
0 100 10 0 x 50 y 0 −10 10 0 x
0 −10 x 60 50 40 30 20 10 0
100
90
80
70
Localization and Elastic Unloading
Imposed Displacement
Figure 15.8: Localization of Tensile Strain in Concrete
Student Version of MATLAB
Student Version of MATLAB
Victor Saouma
Fracture Mechanics
Draft
14
FRACTURE DETERIORATION ANALYSIS OF CONCRETE
15.4
Griﬃth Criterion and FPZ
58 Let us consider a cohesive crack with both normal and tangential tractions in a thin plate subjected to far ﬁeld stresses, and let us assume the crack is to be under general mixed mode conditions, Figure 24.16. 59 To verify if the nonlinear model satisﬁes Griﬃth criterion, it is necessary to compute the energy released by a unit crack propagation. The Jintegral provides a method to evaluate the energy release rate. The Jintegral is a path independent integral and in twodimensional is given by: ∂u (15.30) t J = (W nx − ˆ )dΓ ∂x Γ
σ τ
111111111111111111 000000000000000000τ 111111111111111111 000000000000000000 111111111111111111 000000000000000000 111111111111111111 000000000000000000σ 111111111111111111 000000000000000000 1111111111111111111 0000000000000000000 11 11 00 00 11 1 00 0 Γ 1 1 0 0
0
Γ
τ σ
Figure 15.9: Griﬃth criterion in NLFM.
60 Due to its path independent character it is possible to evaluate the Jintegral along the crack surfaces. ∂v ∂u ˆ ∂u ds = +σ τ dx (15.31) t J(Γo ) = − ∂x ∂x ∂x Γo FPZ
61
Applying Leibnitz rule for the diﬀerentiation of deﬁnite integrals1 the Jintegral is equivalent J(Γo ) =
FPZ
to:
d dx
u
τ du
0
dx +
FPZ
d dx
v
σ dv
0
dx
(15.32)
62 The expressions in parentheses represent the surface energies dissipated in mode I and II at every point along the fracture process zone normalized with respect to crack surface. 63
Hence, we deﬁne:
0
u
τ du = q II (x),
0
v
σ dv = q I (x)
(15.33) (15.34)
J(Γo ) =
1 d dt b a
f (x, t)dx =
b ∂f a ∂t
FPZ
dq II (x) dx + dx
FPZ
dq I (x) dx = GII + GI = Gc c c dx
(x, t)dx; In other words, diﬀerentiation and integration can be interchanged.
Victor Saouma
Fracture Mechanics
Draft
64 65
15.4 Griﬃth Criterion and FPZ
15
where GII and GI is the energy dissipated by a unit propagation of the cohesive crack in mode c c II and I respectively. It should be noted that in general, GII and GI are not equivalent to GII and GI , but are c c F F rather functions of these and the stress state along the interface. However, it is possible to consider two special cases for pure mode I and II cracks. In the case of pure mode I crack, the Jintegral is equal to: J(Γo ) =
FPZ
d dx
v
σ dv
0
dx =
0
wσ
σ(v) dv = GI F
(15.35)
66
Similarly, in the case of pure mode II crack, the Jintegral is equal to: J(Γo ) =
FPZ
d dx
u
wτ
τ du
0
dx =
0
τ (u) du = GII F
(15.36)
where wσ and wτ is the critical crack opening and sliding respectively for which normal and tangent stresses can no longer be transferred across the crack.
67
The following conclusion, can be drawn based on the basis of the previous discussion: 1. It was shown that a unit extension of a cohesive crack model dissipates energy whose amount depends on the softening laws used by the model. The amount of dissipated energy also depends on the loading conditions in FPZ. In pure mode I and mode II loading, speciﬁc fracture energies GI and GII are dissipated respectively. If the structural system F F cannot provide these energies, the crack would not propagate. 2. In the limiting case, when the dimensions of the analyzed problem increase, the cohesive crack gives identical results as LEFM. 3. In ﬁnite element implementation, errors are introduced due to discretization errors. In large structures, ﬁne mesh would be necessary at the crack tip to model the fracture process zone. If the FPZ is not modeled adequately, the Griﬃth criterion for crack propagation is violated, and erroneous results will be obtained.
Victor Saouma
Fracture Mechanics
Draft
16
FRACTURE DETERIORATION ANALYSIS OF CONCRETE
Victor Saouma
Fracture Mechanics
Draft
Chapter 16
FRACTURE MECHANICS of CONCRETE
Needs some serious editing
16.1
7
Linear Elastic Fracture Models
The simplest fracture mechanics model for concrete is clearly the one based on linear elastic fracture mechanics (LEFM). However its validity to cementitious (concrete, rock) materials has been seriously challenged by the lack of experimental evidence suggesting a unique size independent fracture toughness value. This size dependence has been largely attributed to the presence of a process zone ahead of the crack tip, and to the prior extraction of fracture toughness value from notched rather than cracked test specimens (only a crack has a process zone ahead of it).
8 Should we rely on cracked test specimens to extract fracture toughness values, than (as will also be discussed below in the Two Parameter Nonlinear Model of Jenq and Shah) more consistent values for KIc are obtained. The generally accepted procedure to extract LEFM parameter from fracture test uses the Compliance Method. 9
In a fracture test of concrete, in which the fracture toughness KIc is sought, one needs: 1. A measure of the crack length, a. Since the exact crack length is impossible to measure in concrete, we will replace a by an eﬀective (or elastically equivalent) crack length aef f 2. Corresponding load, P . 3. Relationship between stress intensity factor, crack length, and load. In general those √ quantities are related by an expression of the form KI = βP πa, where β is a correction factor accounting for geometry, boundary conditions and load (but not size).
10 The simplest way to determine crack length would be to approximate it to the notch length. However, as indicated above, this would yield erroneous results as such an analysis would not account for the presence of the process zone laying ahead of the crack tip and this will cause the results to be size dependent. Thus the presence of the process zone would have to be indirectly taken into account.
compliance.
will be sought. In doing so the ﬁnite element calibration yields a normalized compliance Cn of the form (16. This is accomplished by ﬁrst performing a ﬁnite element calibration study in which the relationships between crack length.e. but not essential.1. 16.1) Cn = C.3)
We note that in Eq.
1
Victor Saouma
Fracture Mechanics
. a ﬁnite element calibration must ﬁrst be performed. and stress intensity factors are determined. Fracture toughness and crack length. 16.
The compliance is taken as the ratio between the Crack Mouth Opening Displacement (CMOD) (corresponding to the clip gage which will be used in the experiment) to the measured load (corresponding to the one measured by the testing machine load cell).
13 Thus a series of linear elastic analysis with diﬀerent crack lengths (starting with the initial notch) are ﬁrst conducted.E
Hence. 14 Since we assume linear elastic behavior of the specimen. we use the CMOD. Following the analyses.1
12
Finite Element Calibration
As shown later a direct relationship between: 1. Note that we need not measure the true compliance (i. Specimen compliance and crack length. If they are not determined. From each analysis we determine the compliance1 and the stress intensity factors in terms of crack length. However. and through linear regression the following relationships are numerically approximated:
15
a KI
a(Cn ) KI (a)
(16. by selecting a unit force in the ﬁnite element calibration we can readily equate the displacement to the compliance. than the inverse of Eq.2 must also be determined in the form of: Cn Cn (a) (16. since the fracture toughness is independent of the Elastic Modulus. we seek to determine an elastically equivalent eﬀective crack size.
16. to determine numerically the SIF. but conceptually we could use any other displacement/strain as a measure of the compliance. displacement corresponding to the applied load. Furthermore.2) (16. this method is not as accurate as the direct evaluation of K. As for most specimen geometries such analytical relationships are not available. 2. In most cases.4)
and the derivative of the compliance is taken to measure G.2 we have the crack length in terms of the computed normalized compliance.Draft
2
FRACTURE MECHANICS of CONCRETE
11 Instead. the compliance can be readily normalized by the Elastic Modulus (thus yielding material independent calibration). the Elastic Modulus is usually taken as 1. but one which relates an experimentally measured displacement resulting from an applied load. Furthermore it is desirable. aef f which without FPZ would yield the same CMOD (or compliance) as the true crack. the results are tabulated.
16. a series of unload and reload is performed in order to determine the experimental compliance of the cracked specimen for diﬀerent crack lengths3 . 16.
2
Victor Saouma
Fracture Mechanics
. The load is then reapplied and the response would then extend into the postpeak regime.3
Data Reduction
Eﬀective Young’s modulus The eﬀective crack length determination entails measurement of the specimen compliance (inverse of the stiﬀness or displacement over load).1: ServoControlled Test Setup for Concrete KIc and GF
17 During the test. 18 The CMOD is ﬁrst gradually increased. Throughout the descending branch of the structural softening.1.2
16.1 Linear Elastic Fracture Models
3
Testing Procedure
16 Fracture toughness tests should be conducted under strain control where the strain is a measure of the crack mouth opening displacement2 in a servocontrolled manner. The compliance determination in turn requires the concrete Young’s modulus. 3 Unload reload are not required for the experimental determination of GF as this would simply be the area under the PCMOD curve. thus eﬀectively reducing the load. 16. Fig. Following a linear part. a plot of the load versus crack mouth opening displacement should be observed.2. This is associated with slow crack growth and thus the formation of a FPZ turning the notch into a crack. the CMOD is decreased.1. At around the peak load.1. Fig. a nonlinear prepeak response is ﬁrst observed. The elastic modulus can be determined from:
To conduct such tests under displacement control might result in undesirable snapbacks if the testing equipment is not rigid enough.Draft
16.
Figure 16.
This method assumes that in the ﬁnite element calibration this relationship was determined using an appropriate numerical technique. Fig. There are two approaches to compute KI : Numerical SIF Calculation In this method we simply use Eq. it will be indirectly computed as an eﬀective value. Victor Saouma Fracture Mechanics
. we can compute the corresponding eﬀective crack length aef f from Eq.3.3 to determine the stress intensity factor (thus the fracture toughness). 16. Using the slope of the initial part of the LoadCMOD curve. Henceforth. 2. 16. we deﬁne the eﬀective crack length (or elastically equivalent length). aef f as the notch length yielding the same compliance (or stiﬀness) as a crack with a FPZ. Experimentally measured through compressive cylinder tests. we ﬁnally seek to determine the corresponding stress intensity factor. 000 fc . Having determined Cn . Cexp from the slope of the load CMOD during an unloading/reloading cycle in the specimen post peak response.2 (after substituting a by aef f ). and critical load Pcr have been determined. this compliance can be normalized to yield Cn = Eef f . Stress Intensity Factor By now the eﬀective crack length aef f .1 Eef f is determined as Eef f = Cn Cexp (16.5)
Eﬀective Crack Length Since the crack length can not be directly measured. 16. Indirectly measured from the specimen compliance.2: *Typical LoadCMOD Curve from a Concrete Fracture Test 1. 16. Thus in a test we can determine the experimental compliance. the last method yield the most reliable result. The ACI approximate equation E = 57. Since the ﬁrst method is only an approximation. The crack compliance in turn is experimentally measured and has a PZ.Draft
4
FRACTURE MECHANICS of CONCRETE
Figure 16. We note that the notch compliance is obtained through the ﬁnite element calibration and does not account for PZ.Cexp . Having previously determined Eef f . the experimental compliance of the notched specimen is measured and using Eq. 3. and the second one measures the compressive modulus while an average between tensile and compressive value is more representative.
9. and each subsequent one will give a KIc corresponding to a diﬀerent crack length.1
Nonlinear Fracture Models
Hillerborg Characteristic Length
19 The characteristic length.2 Nonlinear Fracture Models
5
Figure 16.
16. the initial unload reload will give Eef f . we obtain: G= E P 2 ∂C (16.4. C. Since the compliance is numerically approximated by Eq. KI = In summary. Thus more than one fracture toughness value can be obtained from a test whereas only one value of GF can be obtained.Draft
16.8) 2B ∂a This apparently simple and “elegant” method calls has a major drawback. compliance. 16. 1976) in 1976 is a material property which gives an indication of the material
Victor Saouma
Fracture Mechanics
. than the Compliance Derivative Method can be used to determine them.3: Eﬀective Crack Length Deﬁnition Compliance Derivative Method If the SIF were not derived from the ﬁnite element analysis. A. 16. proposed by Hillerborg (Hillerborg. its derivative is very likely to be a very poor numerical approximation resulting in gross error when Eq.2. Recalling from Eq. and Mod´er.E.6 with G = EI where E = E for plane stress E and E = 1−µ2 for plane strain. 16.2
16.32 we have: P 2 ∂C (16. Combining Eq.8 is determined. P. crack length.7) 2B ∂a Recalling that the experimental compliance is related to the normalized one through Cn = ECexp .6) 2B ∂a where P. a. M. Eq. and B are the applied load. and assuming plane stress condition. lch . and e Petersson. 16. and specimen K2 width respectively.7 can be rearranged into: KI = P 2 ∂Cn (16.
1989) showed that LEFM u d may not be valid for d/lch < 25. we have two conditions that are simultaneously satisﬁed: KI
S = KIc
(16. and is deﬁned as: rp =
FRACTURE MECHANICS of CONCRETE
lch =
EGF ft 2
(16.9)
where. From this diagram the following three stages can be identiﬁed:
1.Draft
6 brittleness. increasing KI with increasing crack length a).2. M. and the CTOD refers to the notch opening (and not crack) opening. and not the result of some prior crack nucleation/extension.2. Jenq and Shah (Shah and Jenq 1985) have introduced a twoparametermodel..
16. 1976) and (Hillerborg.e. During the second stage. Linear elastic response up to a load corresponding approximately to Pcr . Analysis of three point bending beams. During this stage the CTOD (crack tip opening 2. to characteristic length. At the peak load. This process zone formation has also been referred to as slow crack growth. (Hillerborg.
Victor Saouma
Fracture Mechanics
. displacement) is zero as predicted by LEFM.12)
CT OD = CT ODc
where the right hand side values are now considered material properties. That is. where d is the beam depth. where d is the specimen height. and Roelfstra. P. d. A. lch . 11.E. E. A. and that a ratio of lch = 25 would result in a 5% diﬀerence between the load carrying capacities predicted by LEFM and FCM. is the dominant parameter for describing size eﬀects using the Fictitious Crack Model (FCM). P. A. 1983) e Hillerborg has shown that the ratio of structural dimension. let us ﬁrst consider the load CMOD (crack mouth opening displacement) of a notched specimen.10)
20 In (Hillerborg. This is caused by the formation of the process zone ahead of the crack tip (the existing crack being prenotched or precast. As a result of this microcracking the crack tip starts to open in a fashion similar to the blunting of sharp cracks in metals due to yielding. This equation should be compared with Irwin’s plastic zone size Eq.1. 3. and for unstable geometries (i. To introduce this model.. ft is the tensile strength. signiﬁcant inelastic deformation takes place. and Mod´er. the 2. 2. (Br¨hwiler. for which a process zone ﬁrst has to be developed). and Petersson. d LEFM is considered to be applicable for large values of lch .11) (16. 1985a) and of compact tension specimens. induced LEFM KI is less than KIc .2
Jenq and Shah Two Parameters Model
21 To circumvent the problems associated with nonlinear fracture.9
2 1 KI GE = 2 2 π σyld σyld
(16. and combing the eﬀective crack length proposed by Irwin (one which takes into account the plastic zone) with the CTOD criteria.E. and GF is the speciﬁc fracture energy. E is Young’s Modulus.1
MIHASHI data
16.
Draft
23
16. whereas in Hillerborg’s approach the SIF concept is entirely discarded and replaced by a ﬁctitious load.
The ﬁrst most important contribution proved that fracture mechanics criteria must be used. which is again normalized with respect to d2 . In most cases this entails a ﬁnite element calibration.
4
nda a . the area of the crack zone5 . nda a. 5 The size eﬀect law as presented. Based on an extensive test program. was derived:
Q=
E · CT ODc S KIc
2
(16.2 Nonlinear Fracture Models
7
22 Thus in this simple. the distribution of which requires at least two parameters. Z. d2
Victor Saouma
Fracture Mechanics
. a uniﬁed approach z to concrete strength through the so called Size Eﬀect Law was presented (Baˇant. and CT ODc is evaluated at the notch. E.
16.P. For large Qvalues the material exhibits a high brittleness. when combined with Young’s Modulus. as the strength approach is incorrect.
This model has been shown to yield results that are specimen size insensitive. 2Ec n
(16. d 2.
24 25 Finally. the S critical values. were found to be sizeindependent material properties.2.13)
The Qvalue is related to the brittleness of the specimen. and elegant model we implicitly account for softening by introducing a second parameter (CTOD). α2 = where da is the size of aggregate. or for a concrete 2 E specimen of thicknesss b. Q. 1984). and as originally published. and area of the cracked zone (if a blunt crack model is 2 assumed). The total strain linear elastic energy U is proportional to 1 σ Vol. This model requires an analytical expression for KI and the CT OD along the crack for each geometry considered. clear. α1 = a .1
26
Size Eﬀect Law. is derived for a crack band model.3. thus the nonzero value for CT ODc .14)
where the energy release rate is a function of: 1. and depth d:
27
U=
1 2 2 σ bd f (α1 . A minor drawback of this model is that it requires an analytical expression of the COD and KI along the crack for the geometry considered. termed the critical material length. a single fracture parameter describing the material. z One of the premises of this model is that the total potential energy release U caused by fracture is a function of crack length a. The model distinguishes between notch and crack length. which in most cases can only be obtained through a ﬁnite element calibration. Baˇant z
Derivation
In what is perhaps the second most important contribution of Baˇant4 . ξi ).2. α2 .3
16. the crack length a normalized with respect to specimen dimension d. KIc and CT ODc .
24 and then into 16.23)
32 At this point we introduce GF written in terms of stressstrain for a crack band (note the diﬀerence with GF written for a discrete crack written in terms of stressCOD). we obtain:
1 2 2 ∂U = σ bd ∂a 2Ec n
∂f ∂α ∂f ∂α2 1 + ∂α1 ∂a ∂α2 ∂a
f1 f2
(16. 16. we obtain: da 1 2 σn bf1 d 1 + λ0 2Ec d = nda 1 − Ec Et ft 2 b 2Ec (16. The physical implication of this is that a crack band of 3 times the average aggregate size propagates rather than a sharp discrete crack.18) (16.
33
Substituting Eq.24)
z where wc is the crack width. 16.23.Draft
8
28
FRACTURE MECHANICS of CONCRETE
For the crack to propagate we must satisfy ∂U = GF b.16 and 16.16) (16.19) (16.26) Fracture Mechanics
Victor Saouma
.17)
30
Thus substituting Eq.15 we obtain: ∂U = ∂a = =
1 2 2 2Ec σn bd f1 d
+
f2 nda d2
(16. through regression analysis. Baˇant determined.17 into 16.21)
Eq. Thus if we take the partial derivative ∂a of U with respect to a in Equation 16.14.
GF = wc 1 −
Ec Et
ft 2 2Ec
(16.25)
and n = 3. 16. that wc = nda (16.20)
1 2 2Ec σn b (f1 d 1 2 2Ec σn bf1 d
+ f2 nda )
f2 n da f1 d
1+
31
Introducing λ0 as: λ0 = n
f2 f1
(16.25 into 16.22) (16.20 becomes: GF b = = ∂U ∂a da 1 2 σn bf1 d 1 + λ0 2Ec d (16.15)
29
The partial derivatives of α1 and α2 with respect to a are: ∂α1 ∂a ∂α2 ∂a = = 1 d nda d2 (16.
(16.Draft
34
16. calculation of the constants B and λ0 analytically z z or numerically is too diﬃcult.P. 38
Finally deﬁning λ = β = d da λ λ0 (16.16. λ0 da
(16. Et
(16.28)
36
λ0 f1 da 1 − σn = f t f2 λ0 f1 da 1 + or: σn = where 1 B=√ f2 1− Bft (1 +
d λ0 da )
Ec Et d λ0 da
(16. and they are best obtained through statistical regression analysis of test data.21) and rearranging the denominator as: 2
f1 d 1 + λ0 we can reduce Eq.2 Nonlinear Fracture Models Finally we solve for the nominal stress σn : nda 1 − σn = f t
Ec Et
9
.34)
where ft∗ is termed the sizereducedstrength and is a characteristic of the entire structure and not only of the material.
Victor Saouma
Fracture Mechanics
.30)
Ec .27 to:
da d
= λ0 f1 da
d +1 . B 1+β (16. 1984). 16.33)
we can rewrite Bf σn = √ t 1+β and we can deﬁne: ft∗ = √ σN ft = .32) (16.35) (16.31)
37 As noted by Baˇant (Baˇant.27)
a f1 d 1 + λ0 dd
35
f Substituting nda = λ0 f1 da (from Eq. Z.29)
(16.
4: Size Eﬀect Law 2. geometrically identical specimens. For structures with very large size compared to aggregate size. σN (16.
Figure 16. Again for small structures (small β). with σn versus size d. The size eﬀect law can be plotted on a loglog scale. Then Eq. 5. yielding ft strength criterion governs. the value of λ ∗ ft .
λ0 λ . To assess the size eﬀect law. 3. From statistical regression analysis. the intercept a. Thus Eq. 16. and the classical λ0 in Eq.
Despite its general appeal.Draft
10 16.35
and we see that “For very large concrete structures. and the 1 slope b can be determined. such as dams reduces to ft∗ ft (or large rock masses). this model calls for the following comments: 1.3. B = analysis. and then B = √a and λ0 = a b
Victor Saouma
Fracture Mechanics
. 16. 16. Eq.2. 16. 6. many of which involved uncracked (initial) structures.35 may be neglected in comparison to 1.35 can be cast in the form: Y = a + bλ = 1 2 . Eq.
σN ft
and can be determined from plastic limit
λ λ0
4. as shown in Fig. it has been experimentally validated with numerous tests.4.35 represents a gradual transition from the strength criterion for small structures to linear elastic fracture mechanics for large structures. i. Furthermore. 16.36)
1 1 where a = B 2 and b = B 2 λ0 . but with diﬀerent sizes must be tested.35 asymptotically approaches the size eﬀect of linear elastic fracture mechanics”.2 Discussion
40
FRACTURE MECHANICS of CONCRETE
39 Undoubtedly the size eﬀect law is a very elegant generalized model for concrete fracture. 16. For structures of a small size relative to the size of aggregate. It attempts to provide a uniﬁed mathematical model for concrete cracking by merging two diﬀerent approaches. In general.e for small λ.
1.
and on a dimension d characteristic of
Victor Saouma
Fracture Mechanics
. 9. ft . L. and. 16. Table 16. (Carpinteri 1986). was developed by Carpinteri to assess the existence of either one of the two failure mechanisms (Carpinteri 1982a).1.5. The dimensionless brittleness number. and Cedolin. (Carpinteri 1982b).Draft
16. then a strength criterion must be used. s.5: Inelastic Buckling
16.1. The point of intersection of the two asymptotes corresponds to β = 1.2. This parameter depends on the fracture toughness. (2) fracture collapse in terms of LEFM. If λ0 = β is less than 0.
8. At no stage did we have to introduce (explicitly or implicitly) any LEFM equation (in the sense of a stress singularity at the crack tip). It can be shown. 16. (Baˇant. 11. then a LEFM criterion is to be used. the square component is again not linked to LEFM. Note that those are arbitrary guidelines. the tensile strength. Z. KIc .P.4
41
Carpinteri Brittleness Number
In fracture testing. and if β is greater than 10. There is a strong analogy between the size eﬀect and column buckling. • expression for GF in terms of the square of ft in Eq. The square root relationship between ft∗ and d (through λ) comes from: • an appropriate choice of the normalizing parameter for the crack length a in deﬁning α1 and α2 . Although it is correct to directly relate GF to the area under the uniaxial stress strain curve (a measure of the energy required to produce a unit surface).24. 10. 1991) that the fracture energy GF can be z recovered from the Size Eﬀect Law 12. Fig.
σ
σ
Euler Buckling (Linear Elastic)
σ=πE
kL r
2
( )
σ
yd ET < E
Effect of Residual Stresses
2
σ
Gross Yielding Inelastic Buckling Proportional Limit
yd
σr
Proportional Limit E
ε
kL r
Figure 16.2 Nonlinear Fracture Models
11
λ 7. failure may be caused by diﬀerent mechanisms: (1) ultimate strength collapse (SOM).
1: Size Eﬀect Law vs Column Curve the specimen (or structure) geometry being considered: s= KIc √ ft d (16.e. can be traced back to the fundamental equation governing linear elastic fracture mechanics (Hillerborg 1985a): √ (16. this section will analytically show the similarity between them. lch .1
45
Hillerborg Characteristic Length. respectively.J. V.3
Comparison of the Fracture Models
43 All models considered in this chapter are used to assess the brittleness and by consequence the applicability of LEFM6 . and Saouma.
They showed in a numerical analysis that. Broz.E. over limited ranges. fracture mechanics is applicable since the specimen is more brittle. for small brittleness numbers.37)
42 For large brittleness numbers. formulations describing identical physical phenomena. The fracture toughness
6
This section is taken from (Br¨hwiler. Similar studies have been reported by Planas and Elices (Planas and Elices 1988). J. Because the models originate from diﬀerent.3. which is obtained from the nominal stress at failure.. based on size eﬀect alone. is deﬁned by the critical brittleness number. E. and then all subsequent ones are found to be a variation of this model. Conversely. and β is a correction factor accounting for specimen geometry. s0 . boundary.
16.. small specimen dimension d. lch
Hillerborg’s characteristic length. tensile softening of concrete and its size eﬀects. and the interpretation of tests through Fracture Mechanics (FM) is not valid. The boundary between SOM and FM applicability. but related.. SOM governs. i.e. and loading conditions.
44
16.Draft
12 σcr = σy 1 −
σy 4π 2 E
FRACTURE MECHANICS of CONCRETE Column Buckling Euler Equation Slenderness ratio Plastic failure Inelastic stresses Inelastic Buckling Column Equation (SSRC)
KL rmin 2
Size Eﬀect LEFM Size Plastic failure Cohesive stresses NLFM Size Eﬀect law (Baˇant) z
Bf σn = √ t 1+β
Table 16.. First the relationship between Hillerborg’s model and Griﬃth’s LEFM model is rederived. i. 1991) u
Victor Saouma
Fracture Mechanics
..38) KIc = β · σcr πacr where σcr and acr are critical stress and crack length. existing fracture models may be thought as approximations of one another in a mathematical sense and experimentally indistinguishable.
β z
48 The intersection between the SOM and LEFM approaches in the size eﬀect diagram (Fig.47 are combined into one equation.43) σN M = σN
deﬁnes the point where the brittleness number.44) (16.40)
Since concrete has a low Poisson’s ratio ν.45 and 16. 16. plane stress is primarily considered.42)
Eq. 16. or: EGF = β 2 πacr = lch ft 2 (16.48)
2 Because LEFM is assumed valid. yielding: ft 2 d0 2 = KIc
49
(16.4) where LEF SOM (16.
16.3. Gc =
2 KIc E
(16.
16. may be approximately proportional to the length of the fracture process zone. β is equal to 1.Draft
where. Thus Eqs. 16.47)
C BD
2
= K2
(16. 16. the strain energy release rate Gc becomes equal to GF . leads to:
d0 ft 2 d0 = = K2 GF E lch Victor Saouma
(16.39 can be rewritten as:
2 EGF = β 2 σcr πacr
(16.49) Fracture Mechanics
. From KIc = D P max √ d t·d (16. for a given specimen geometry.45)
KIc P max = √ td D d0 and combining with σN P max = t·d C
(16.46)
we obtain
Bft P max = td C Eqs. or d = λ0 da = d0 . which.38 and 16. and for “large” concrete structures in which LEFM is applicable.2
Baˇant Brittleness Number.39)
E =
46
E
E (1−ν 2 )
: for plane stress : for plane strain
(16.42 shows that the critical crack size is of the same order of magnitude as the characteristic length lch .39). substituting KIc with GF · E (Eq.3 Comparison of the Fracture Models
13
KIc in turn can then be related to the critical energy release rate Gc as derived by Irwin (1957).41)
47
Hillerborg’s characteristic length is then obtained with σcr = ft . 16.
At the tip of the real crack (or notch) in the FCM.5
53
Discussion
Table 16. 1985b) and (Baˇant 1985).52)
where. and assuming that KIc = E · GF .3.2 summarizes the preceding analysis of models and indicates that all models are related to Hillerborg’s characteristic length.3. to ﬁctitious crack width. and Eq. Therefore. σ. w curve. the following expression is obtained:
s2 = or
lch EGF = 2 d ft d lch d
(16. Since lch is a comprehensive parameter to describe material fracture.4
Jenq and Shah’s Critical Material Length. the brittleness number β used to assess the applicability of LEFM in the Size Eﬀect Law can be related to Hillerborg’s characteristic length. A. 16. Q
52 In the Fictitious Crack Model. 16. diﬀerences in experimental results Victor Saouma Fracture Mechanics
. w.Draft
14
FRACTURE MECHANICS of CONCRETE
50 Thus. z
16. 16.50)
s=
(16. CT ODc . relating the tensile stress. or:
GF = Cft · wc
(16.53: Q=
2 With KIc = GF E. according to the TwoParameter Model.52 can be rewritten as: CT ODc = Combining Q (Eq. C is a constant characterizing the σ vs. E. 1988). and thus related to Griﬃth’s model. CT ODc is set equal to wc . Eq. when LEFM is valid.55)
16. lch .13) with Eq.3. Comparison between the Size Eﬀect Law and the Fictitious Crack Model showed that both models yield good agreement in the prediction of the size eﬀect (Hillerborg. 16. wc can be interpreted as the critical crack tip opening displacement. the u d/da ratio in the Size Eﬀect Law may be replaced by d/lch (Br¨hwiler. Although relations among all the models have been demonstrated.37).49 indicates that d/lch has a similar signiﬁcance to the ratio between structural dimension d and maximum aggregate size da in the Size Eﬀect Law. 16. s
2 Squaring the expression s of the brittleness number (Eq.53)
E 2 G2 F KIc 2 C 2 ft 2
(16. Q is ﬁnally related to lch :
GF Cft
(16. and wc is the crack width for which σ = 0..54)
Q=
lch EGF = 2 2f 2 C C t
(16.3
51
Carpinteri Brittleness Number. GF . the area under the softening law.51)
16. is deﬁned as the speciﬁc fracture energy.
the results of the comparisons indicate a uniﬁed approach could possibly be postulated.4
Model Selection
54 In light of the diversity of models. Accordingly we should select: 1. An NLFM for an arch dam which cracking is caused by foundation settlement. Under those conditions nonlinear eﬀects may be negligible compared to the ones induced by LEFM. the eﬀect of the process zone is smaller than for a “short” one.Draft
16. Engineers should exercise great care in selecting an appropriate model. than only a prepeak response is of importance. K2 Gc = EIc and Gc = GF . Stability7 : In a structure subjected to imposed load rather than imposed displacement nonlinear eﬀects are not be as negligible in a stable structure as they are in an unstable one.4 Model Selection
15
Table 16.
7
Victor Saouma
Fracture Mechanics
. no matter how small or large the crack is. in an unstable one the stress intensity factor KI increases with crack length. Validation of such an approach in the laboratory would prove diﬃcult due to specimen size considerations. Model selection should account for:
1.e.2: Summary relations for the concrete fracture models. Loading Load Control.3: When to Use LEFM or NLFM Fracture Models
A stable structure is one in which KI decreases with crack length. Type of loading: If a structure is subjected to a load rather than imposed displacements (such as foundation settlements).. Author Year Parameter Relation Hillerborg 1976 lch lch =
EGF ft 2
Carpinteri 1982 s s=
lch d
Jenq & Shah 1985 Q Q=
lch C2
Baˇant z 1984 d0 d0 = K2 lch
are due to the initial assumptions of very large concrete structures. Crack size: In a “large” crack.
16. Thus Table 16. i. 2. 3. Unstable Load Control. Stable Displacement Control “Small” Crack LEFM NLFM NLFM “Large” Crack LEFM LEFM LEFM
Table 16.3 provides some guidance for model selection. An LEFM model for a dam in which cracking is caused by ﬂooding. where LEFM is valid. 2. Yet. Thus LEFM is more applicable for the large ones.
Thus ultimately the type of analysis to be undertaken hinges on both technical and economical considerations.Draft
16
FRACTURE MECHANICS of CONCRETE
55 In all cases. however in many (but not all) cases the complexity and expenses associated with such an analysis yield results very close to the ones obtained from a linear elastic ones.
Victor Saouma
Fracture Mechanics
. it should be emphasized that ideally a nonlinear analysis is to be undertaken.
4in. Large scale double cantilever beams with a length of 350cm (138in. An investigation using four diﬀerent specimen geometries ranging in size from 100cm (39.Draft
Chapter 17
FRACTURE MECHANICS PROPERTIES OF CONCRETE
Adapted from (Saouma. and tapered double cantilever beams with a height of 210cm (83in. and speciﬁc fracture energy. 1984) for fracture testing of concrete. Several experimental investigations have been performed on large concrete members to determine both “objective” fracture properties and minimum dimensions for test specimens.) (Rossi et al. Other large concrete fracture specimens include three–point bend (TPB) beams with a depth of 80cm (31.) have been used at the “Laboratoire Central des Ponts et Chauss´es” in Paris since 1976 (Chhuy et al. 38mm (1. compact tension (CT) specimens with heights ranging from 30cm (12in. Broz. e 1979. In this chapter.) and 76mm (3. GF .) to 144cm (56.1
Introduction
Concrete dams are often referenced as a potential application of Fracture Mechanics.5in. 91cm (36in.0in.) and 152cm (60in. 19mm (0.5in) (Hilsdorf & Brameshuber 1985). since the fracture parameters often appear to be size and aggregate dependent. The fracture properties obtained from small concrete specimens may be size and geometry dependent because the fracture ligament of the specimens is too short to achieve a steady crack propagation regime. representative of mass concrete used in dams are examined.7in. 1988).). From each set of experiments. Most of the research undertaken on the fracture of concrete has concentrated on laboratory testing of small specimens constructed of structural concrete. Br¨hwiler and Boggs 1991) u
17. a basic understanding of the fracture processes occurring in concrete has been developed. Based on these test results.) to 150cm (59in. However.) (Wittmann et al. testing and data evaluation procedures developed for relatively small specimens are extended to a series of large concrete Wedge Splitting (WS) specimens with heights of 31cm (12in. Three diﬀerent concretes with maximum aggregate sizes. the experimentally determined values cannot be reasonably extrapolated to mass concrete structures such as dams. yet. are determined and discussed. KIc .). 1990).75in.) is presented in Wittmann & MetzenerGheorghita (1985). The inﬂuence of aggregate shape and geological origin on the fracture properties are
. A test series of WS specimens used to study the fracture behavior of the cold–joint between lifts in a dam is also presented. little work has been performed to evaluate fracture properties of dam concrete.). Rossi et al. values for fracture toughness.).
390 2.
FRACTURE MECHANICS PROPERTIES OF CONCRETE
17.
CONCRETE MIX DESIGN (no entrained air. Bureau of Reclamation. w/c ratio= 0. and 76mm (3in.1: Concrete mix design.180)
water cement sand 19mm MSA (0.2.55) maximum aggregate size mix 19mm (0.) cylinders for the 76mm mix were used for the determination of the concrete material properties. are summarized in Table 17.320) (1.75in. fc .550) (1. and the splitting tension test (Brazilian) described in ASTM C49685 (Anon. 38mm (1. and tensile splitting strength.S.0in.650) (1. of each concrete mix were obtained prior to performing each wedgesplitting (WS) specimen test.75in. in general.5in. 17.5in.080) (4. and maintenance of many federally owned dams in the western United States. The test specimen shape used for the experimental program was the wedgesplitting geometry.420 2. Based on the presence of such large aggregates in a dam concrete mix. For each aggregate size.) MSA mix.) 38mm MSA (1.75in. performed according to ASTM C39 (Anon. ft. 1983). which consisted of quarried subangular aggregates.) in diameter.) × 30cm (12.0in.) 38mm (1.5in.).sp .0in.Draft
2 also considered. shown in Fig.) total
Ten standard 15cm (6.0in.1.0in.1
Experiments
Concrete Mix Design and Specimen Preparation
Unlike structural concrete. The majority of the mixes were batched with rounded river aggregates.0in.) cylinders for the 19mm and 38mm MSA mixes.170) 978 609 491 (1. Details of the mix designs used for the construction of all specimens are shown in Table 17.0in. which had been previously used in research programs for the testing Victor Saouma Fracture Mechanics
.2
17.5in. a concrete mix similar to those used in actual gravity dams was obtained from the U. equal to or larger than 76mm (3.) 76mm MSA (3.) Weight Weight Weight kg/m3 (lb/yd3 ) kg/m3 (lb/yd3 ) kg/m3 (lb/yd3 ) 163 139 102 (275) (233) (174) 297 252 187 (500) (424) (315) 918 782 695 (1.030) (827) 0 609 491 (1. three MSAs were selected for testing: 19mm (0.) 76mm (3. 1985).). except for a batch of the 38mm (1. Table 17.) × 60cm (24.1.030) (4. The compressive strength. construction.).030) (827) 0 0 491 (827) 2. The mean values obtained from unconﬁned compression tests.2.5in. in dam concrete mixes the MSA is.) in diameter. in which the maximum size aggregate (MSA) typically does not exceed 38mm (1. and ten 30cm (12.480 (4. which is the agency responsible for the design.
9 (2. a sharpened 3.2mm (0.2: Experimentally obtained material properties of the concrete mixes used.41 (349) Eef f .96 (574) 2.5in.5in.67 (388) 3.
MSA 19mm (0. and to predict fracture properties of dam concrete with intermediary values to ensure a continuity of results.5 (12. cm (in.6 times greater than a commonly used threepoint bend (TPB) beam geometry (RILEM 1985) of equal volume. The large fracture area compared to the aggregate size and specimen volume makes the WS geometry well suited for testing large MSA and specimens under laboratory conditions.0) 41 (16. Specimen construction consisted of wooden formwork built for the placing of the concrete. the concrete was placed in two separate lifts.0) 41 (16.0in.) subangular 76mm (3.3.) fc . MPa (ksi) 18.8 (3.25) 3 (1. and two hollow steel sleeves ﬂanking the plate.) 38mm (1.710) 24.0) 15 (6. the Victor Saouma Fracture Mechanics
. MPa(psi) 2. which is a region of inherent weakness.Draft
17. Linsbauer & Tschegg 1986.5(1.0)
a0 4.
Specimen Dimensions. The ﬁrst provided the 19mm and 38mm MSA mixes and the other the 76mm MSA concrete mix. The WS specimen has a large fracture area compared to u the concrete volume.0) 91 (36.317) 18.0) 23 (9.1: *Wedgesplitting specimen geometry.2 Experiments
3
Table 17.0) 6 (2. The dimensions of the WS specimens tested in the experimental program are summarized in Table 17.) w t s D 31 (12.125in. was studied through a slight oﬀset between the initial notch and the level of the concrete in the ﬁrst lift. with the exception of the coldjoint (CJ) specimens.sp .0) 61 (24. Two diﬀerent ready–mix suppliers delivered the concrete.600) 36.81 (407) 2.900 (2.0) 91 (36.0) 30.200 (3. MPa (psi) 25.500 (2.6 (5.) thick steel plate mounted inside each form to obtain the vertical notch normal to the load line.0) 8 (3)
Specimen Type “Small” “Medium” “Large”
H 31 (12.) 38mm (1. Br¨ hwiler u 1988.2 (6. approximately 4. The tendency of the crack to propagate into the coldjoint.6 (3. The specimen dimensions and aggregate sizes were selected to enable a comparison Table 17.000 (2.75in.0) 8 (3) 152 (60.360) 16. Br¨hwiler & Wittmann 1990).3: Wedgesplitting specimen dimensions.400)
Figure 17. In all specimens.460) 23.0) 107 (42.0)
with results previously obtained by other researchers in the academic community.610) 16. For the CJ specimens. with the notch plate initially in a horizontal orientation. and cured at room temperature in the laboratory for at least 28 days prior to testing. the concrete was placed with the notch in the vertical direction.0) 15 (6.0)
h 20 (8.740) ft.0) 152 (60. Three days after the ﬁrst lift was placed.75) 15. of small concrete specimens (Hillemeier & Hilsdorf 1976.
Draft
4
FRACTURE MECHANICS PROPERTIES OF CONCRETE
surface was sandblasted to remove surface laitance and ensure a good bond along the joint. Grade 16 silica sand. was then ﬁxed and the WS specimen was raised by the actuator such that the wedges were forced between the bearing rings on each side. driven by an air pressure of 100 psi. Table 17.4: Test matrix.) 76mm (3.2. which acts as
Figure 17.
Victor Saouma
Fracture Mechanics
. a line support along the bottom face of the specimen. and tested using a procedure identical to the standard WS specimens experiments without the coldjoint.2.5in. the specimens were turned upright. After curing for 28 days. the WS specimens were centered on a steel rod. such that the notch was in a vertical orientation. The 38mm and 76mm MSA concrete mixes used for the CJ specimens were identical to the mix designs used during the standard WS testing program.0in.2.2: *Wedge ﬁxture and line support. also shown in Fig. The concrete mixes associated with each specimen size and the number of specimens in each series are presented in the test matrix of Table 17. As a result of the horizontal displacement imposed by the wedges. and needle bearing rings were mounted on the protruding ends. the surface was continuously water cured but was saturated surface dry (SSD) when the second lift was placed. 17. Two steel rods were inserted through steel sleeves in the specimen. The wedgeloading apparatus.4. After sandblasting.75in. 17. the specimen was subjected to a pure Mode I loading condition. A ﬂexible connection between the wedgeloading ﬁxture and the testing machine load cell was used to account for any small misalignments and evenly distribute the induced load. was used in the blasting process.) 3 3 3 3 3 3 3 3 3 3
17.
Specimen Type “Small” “Medium” “Large” “Medium” “Large” (coldjoint) Aggregate Shape rounded rounded rounded subangular rounded Number of Tests maximum aggregate size mix 19mm (0.2
Loading Fixtures
For testing. imposing a horizontal displacement on the specimen at the point of contact. The line support ﬁxture was mounted to the actuator of a closedloop servohydraulic materials testing system. The sandblasting operation and the placing of the second lift were performed on the same day. shown in Fig.) 38mm (1.
processed.3: *Block diagram of the experimental system. were performed.2.Draft
17. or ﬂexibility.3
17. ﬁltered. which were used to monitor the change in the specimen compliance. WS specimen experiments. The wedge ﬁxture is a statically determinate beam. is equal to 15o . The experiment was performed until the specimen was split into two halves.0µm/sec. the primary deformation monitored was the crack–mouth opening displacement (CMOD).) A typical PSP vs. and the testing machine for the WS specimen tests are shown in a simpliﬁed block diagram in Fig. the PV splitting force is given by PSP = 2·tanα . the data acquisition system. both linear and nonlinear responses in the ascending prepeak branch and a descending postpeak branch can be observed. the microcracking associated with fracture process zone (FPZ) formation and with transient releases of elastic energy due to localized aggregate fracture and bond failure were monitored with an acoustic emission (AE) sensor mounted on the surface of the specimen. the vertical load and the CMOD were monitored and recorded using a standard data acquisition system. Unload/reload cycles. they were reduced by using hardened steel inserts along the inclined wedge surface and needle bearings with a low coeﬃcient of friction. In the
Figure 17. In the PSP vs.3. and therefore. The sensor used during testing was sensitive in Victor Saouma Fracture Mechanics
. PV .4
Acoustic Emissions Monitoring
In several WS tests.4: *Typical PSP vs. The AE sensor was mounted on the face of the specimen in a direction such that the sensor was normal to the predicted fracture plane.4.
17. CMOD
Figure 17. was directly obtained from the load cell of the testing machine. curve. The vertical force induced by the wedge ﬁxture. 17. α. and related to the behavior of the material throughout the experiment. The closedloop servohydraulic test machine used a programmed constant rate of CMOD (1. a piezoelectric crystal is stimulated by the stress waves and produces electrical signals. During the test. Inside the sensor. The signals are then ampliﬁed.2. where the wedge angle.2 Experiments
5
Testing Procedure
The experimental setup of the electronic controller. CMOD curve representative of the WS specimens is presented in Fig. (Frictional forces were neglected. or 40µin/sec) as the feedback control to obtain stable crack growth with structural postpeak response. 17. The CMOD was measured on the top surface of the specimen by a clip gage mounted over the mouth of the initial notch. CMOD curve for a “Large” specimen.
5.2. and (2) little or low AE activity. A typical record of the AERMS voltage signal for each corresponding load step vs. is observed in AE testing of composite materials. The data shown in the curve can be divided essentially into three
Figure 17. Section III: The AE bursts increase in amplitude and number. and for the stress intensity factor.1)
where β = ef f and f (β) is the geometry function for the WS specimens (Saouma et al. testing time is shown in Fig 17. and improper ﬁt of the specimen halves. The electrical signal was then ampliﬁed. was determined using the compliance method. Further applications of this nondestructive evaluation method were performed in the ﬁeld testing program as presented in Saouma et al. With each AE burst.Draft
6
FRACTURE MECHANICS PROPERTIES OF CONCRETE
the 30 kHz range. This eﬀect occurs primarily in the low CMOD regime. the observed low AE activity and liftoﬀ of the RMS signal from a baseline value indicates microcracking and the formation of the process zone at the notch tip. indicating microcracking associated with the formation of the process zone. 1989a. both the splitting load and Young’s Modulus E were taken as Victor Saouma Fracture Mechanics
.5: *Typical AE record for a “Large” WS specimen test. Based on the observed readings. H Broz 1989). interpreted as crack extension or aggregate/matrix/bond failure. and the signal was ﬁltered by a bandpass ﬁlter with a bandwidth of 45–135 kHz. The AE monitoring and evaluation techniques were used only as a qualitative aid to the test. and is not visible in Section III. a = a(C). sections: Section I: The specimen is in the ascending branch of the PSP vs. In the analysis.5
Evaluation of Fracture Toughness by the Compliance Method
The fracture toughness. A ﬁnite element calibration analysis was ﬁrst performed to determine a relationship between crack length and compliance. converted into a DC voltage by the processing unit. the material response was evaluated as (1) large AE bursts or continuous activity. The Kaiser Eﬀect which is characterized by the absence of detectable acoustic emissions until previously applied stress levels are exceeded. CMOD curve.
17. KIc . KI in terms of crack length: KI =
a
PSP √ πaef f · f (β) H ·t
(17. a rise in the RMS signal was recorded with an XY plotter. (1990). and fed into a trueRMS (rootmeansquare) voltmeter to eliminate the eﬀects of constant background noise. Section II: The peak load is reached and large AE spikes resulting from crack extension can be observed. indicating continuous crack formation and growth during the descending branch. The lack of the Kaiser Eﬀect in Section III during the unload/reload cycles may have been caused by crushed small concrete particles.
and the
Figure 17. This eﬀect. From the analysis of the experimental data the following observations are made: 1. and compare with the normalized (numerical) compliance Cn corresponding to a0 . CMOD record. Stage II is obtained from the post peak data falling within a plateau. aef f curve that typically exhibited three stages (Fig. 17.3 Fracture Toughness Results
7
unity. For each successive postpeak unload/reload cycle:
i (a) Determine the experimental compliance. Stage III demonstrates a sudden decrease in the KIc values. 17. and singular elements were used for SIF determination. from Eq. From the initial ascending branch of the PSP vs. (1990). determine the eﬀective Young’s C Modulus Eef f = C in (Fig. determine the experimeni tal initial specimen compliance Cexp .5 and 17. KI . determine the fracture toughness.4).6).6. for the fracture toughness were determined.
exp
2. Victor Saouma Fracture Mechanics
. larger residual CMOD (caused by the inability of the crack to perfectly close) was observed.3
Fracture Toughness Results
Reduction of the experimental data resulted in a KIc vs.4). 17.6: *The three stages of the fracture toughness vs. Cexp . as the slope of the line obtained by connecting the initial unloading point to the ﬁnal reloading point of each cycle (Fig. From those two values. but the assumption was increasingly violated as the test progressed.Draft
17. independent of specimen size. vertical load components induced by the wedge ﬁxture on the specimen were included. Since the relative standard deviations in the vertical columns of Table 17. i (b) Normalize Cexp with respect to Eef f and then determine aef f . eﬀective crack length curve.e. prepeak unload/reload cycles are performed. i. at ﬁrst negligible. eventually becomes predominant as the crack grew down the ligament. KIc . the stress intensity factor. values increase monotonically up to a plateau. “objective” values. A possible explanation for this phenomena is based on the inherent assumptions of the data reduction method. 17. (c) From aef f and the Pcr corresponding to the unload point.1. 17. The plateau is synonymous with steady crack propagation and was also observed by Rossi et al. In Stage I.
17. Only the fracture toughness values associated with Stage II were considered in the statistical analysis.5 are all within 20%.7 and summarized in Tables 17. As the eﬀective crack length increased. A perfectly linear elastic system was assumed in the ﬁnite element calibration (resulting in zero residual CMOD upon unloading). section of the curve is included for reference only. The mean values with the respective relative standard deviations are plotted in Fig. The following general procedure was then followed to interpret the experimental data: 1. Since this stage corresponds to the formation of the FPZ.
) rel. σ (%) 5.9 1.5: Summary of fracture toughness data obtained from the WS tests.71 (643) 0.0 2.09 (991) rel.9 15.03 (941) 16.86 (783) rel.5in.64 (582) 0.8 1.02 (929) 1.50 (457) 0.52 (476) 0.8 3.98 (893) rel.1 76mm Mean 0. σ (%) 9.) rel.0in.0 1.62 (567) (3.Draft
8
FRACTURE MECHANICS PROPERTIES OF CONCRETE
Figure 17.75in.16 (1.1 Mean 1. σ (%) 20.3 12.3 12.03 (941) 1.34 (1.
√ Fracture Toughness.7: *Mean fracture toughness values obtained from the rounded MSA WS specimen tests.01 (923) 0.0in.3 Mean 1.4 Mean 1.
Table 17.) 76mm (3.2 2.060) 12.99 (909) 1.7 15. MSA A B C 38mm Mean 0.) 38mm (1.03 (941) 15.05 (957) rel.72 (656) (1.93 (843) 13.23 (1.220) rel. MN/m3/2 (psi in) specimen no.3 Summary 0. KIc .5in.7 Mean 1.120) 0.3 12.2 
Table 17.5 16. σ (%) 13. KIc . MN/m3/2 (psi in) 19mm (0. σ (%) 4.) Mean 1.
Type “Small” “Medium” “Large” Summary “Medium” (subangular) √ Fracture Toughness. σ (%) 12. σ (%) 6.7
Victor Saouma
Fracture Mechanics
.6: Fracture toughness values obtained from the CJWS specimens.9 13.05 (953) 0.
GF . 6.8 and plotted in Fig. The GF values determined from the experiments are shown Tables 17.8. aggregate failures were more predominant in the subangular specimens.6). had: (a) Peak loads 50% lower than comparable specimens constructed with no “lift” joint. the fracture toughness values also appear to be independent of aggregate sizes. (c) At peak load. The CJ specimens. In the analysis of CJ fracture surfaces. 1990) of the concrete surfaces yielded a constant fractal dimension for all the rounded aggregate specimens. which are 1020 % higher than the values obtained for the 19mm MSA concrete mix. speciﬁc fracture energy can also be obtained directly from ﬂexuralbased specimen geometries. 5. the eﬀective crack length of the subangular MSA specimens was found to be smaller than the aef f of the rounded MSA specimens.4 Speciﬁc Fracture Energy Results
9
2. The previous observation was further evaluated through a test series of mediumsize specimens constructed with subangular andesite aggregates (as opposed to the rounded aggregates used in all previous specimens). Alternatively. Br¨hwiler & Wittmann u u 1990). 3. (b) Aggregate debonding was prevalent in fracture surface of the rounded specimens. Thus. divided by the ligament area (Br¨hwiler 1988.7 and 17. a relationship between the fractal dimension and fracture toughness was observed. GF was calculated as the area under the curve of PSP vs. Based on the relative standard deviation of the horizontal rows. (1990). The 38mm and 76mm rounded MSA specimens yielded approximately equal values of GF .
17. is deﬁned as the total energy required to break a specimen into halves. such as the WS specimen. and no descending portions of the fracture toughness curve (as illustrated in Fig. each test specimen yields one single value of GF . Based on the analysis of the experimental results the following comments can be made: 1.4
Speciﬁc Fracture Energy Results
The speciﬁc fracture energy. The experiments reported here were conducted in the laboratory environment under unconﬁned conditions. which were used to quantify the response of liftjoints found in dams. A fractal analysis (GamalElDin et al.34 MN/m3/2 (1.Draft
17. (b) Fracture toughness values approximately 3060% lower than for specimens without lift joints.220 psi in). however. normalized by the projected fracture area. corresponding CMOD. In the WS specimens. The trends in the results of the test series can be summarized as follows: √ (a) The experiments yielded a fracture toughness value of 1. (c) Smaller residual crack opening displacements following each cycle of unload/reload. indicating the presence of a smaller process zone. 4. which is 30% higher than the value obtained from the rounded aggregate specimens. Victor Saouma Fracture Mechanics
. 17. the eﬀects of conﬁning stresses on the apparent fracture toughness are presented in Saouma et al. GF is ideally obtained from the area underneath the loaddeformation curve in a direct tension test. 17.
48 × 10 ) 237 (1.7 
Table 17.31 × 10−3 ) 19.70 × 10−4 ) (4.5in.75in.7 226 (1.21 × 10−3 ) 15.0 11.5 6.5in.35 × 10−3 ) 18.Draft
10
FRACTURE MECHANICS PROPERTIES OF CONCRETE
Figure 17.2 173 (9.58 × 10−3 ) 13.0in. σ (%) Mean rel.37 × 10 ) 230 (1.90 × 10−4 ) (5.) 206 (1. σ (%)
Summary 226 (1.1 7.
Speciﬁc Fracture Energy. σ (%) Mean rel.2 211 (1. σ (%) Mean rel.29 × 10−3 ) 12.1 (3.60 × 10−4 )
Victor Saouma
Fracture Mechanics
.1 −3 −3 189 (1.7 80.88 × 10−4 ) 223 (1.55 × 10−4 ) (7.) 76mm (3.29 × 10−3 ) 15.45 × 10−3 ) 15.29 × 10−3 ) 29.8 
Type “Small” “Medium” “Large” Summary “Medium” (subangular)
Mean rel.4 (1.) (5.0in.18 × 10−3 ) 238 (1.1 13. N/m(kip/in) specimen no.36 × 10−3 ) 9.85 × 10−4 ) 76mm 133.
Speciﬁc Fracture Energy. σ (%) Mean rel.5 254 (1.5 15.5 277 (1.8: *Mean speciﬁc fracture energy values obtained from the rounded MSA WS specimen tests.6 138. GF .7: Summary of speciﬁc fracture energy values obtained from the WS tests. GF .27 × 10−3 ) 226 (1.59 × 10−4 ) (4.8: Fracture energy values obtained from the CJWS specimens.8 98. N/m(kip/in) 19mm (0.) (7. MSA A B C 38mm 99.1 −3 259 (1.0 85.) 38mm (1.08 × 10 ) 240 (1.
Table 17.
Debonding of rounded aggregates may result in a FPZ which is larger than the process zone caused by the fracture of subangular aggregates. the aggregate’s strength. the √ unconﬁned KIc of concrete is a material constant.6
Size Eﬀect Law Assessment
More speciﬁcally. (1990).9 it appeared that aggregate size does not play a dominant role in assessing the nominal strength of a concrete structure.. From Table 17. The evaluated fracture parameters. The fracture properties for CJWS specimens exhibited higher variations in results. 1991). more energy was required to fracture the subangular aggregates than to debond the rounded ones. KIc and GF . were analyzed. 5. J. 2.E. V. 4. For the coldjoint specimens: (a) The GF values were approximately 4070 % lower than the standard WS specimens. KIc and GF . as well as.
Victor Saouma
Fracture Mechanics
. or shape. but diﬀerent da .5
Conclusions
1. 4. (Br¨ hwiler. Based on the concrete mixes tested. The lower fracture energy values were attributed to the inherent weakness of the coldjoint between specimen halves. (b) The subangular aggregate shape inhibitted debonding. an assessment of the size eﬀect law in terms of: 1) testing three concrete mixes (with diﬀerent aggregate sizes) in identical specimen sizes. and were lower than for specimens without coldjoints.. and more energy was required to break the subangular aggregate than to debond rounded aggregate. E. the Size Eﬀect Law assumes avariation of the nominal strength in terms of the maximum aggregate size.04 MN/m3/2 (941 psi in). are independent of specimen size and are “objective” only if certain minimum dimensions are exceeded. 3. are practically independent of aggregate size and depend on the angularity. and ﬁnally 3) determination of the fracture toughness through an extrapolation of the size eﬀect law. Broz. 3.J.5 Conclusions
11
2. which were obtained from two separate local readymix suppliers. A similar size dependency for GF for relatively small specimens with ligament lengths of less than 20cm was observed by Rossi et al.Draft
17. the test data for specimens of “Medium” size.. The fracture parameters. and using the compliance method for fracture toughness evaluation. To assess this dependency. (b) The higher GF values were correlated to visibly rougher surfaces at the coldjoint. For the two types of aggregates used in this experimental program.
17. and Saouma. The experiments performed on the specimens constructed with the subangular aggregate mix indicated: (a) GF values were 25% higher than the values obtained from the specimens constructed from the rounded aggregate mixes. 2) The same concrete mix on three diﬀerent specimen sizes. u In its original formulation.
17. 6. equal to 1.
68 1. Broz.93 ksi in) obtained using the compliance method (CM) (Saouma. V.4 a0 /d
max Fsp kN (kip) 24..E. V.J.9: Size eﬀect for WS specimens for da =38 mm (1.18 0.1 123.5
0.372 1..J. V.J.9 113.22
σN MPa (psi) 3.54
However.) 44. and Saouma. the fracture toughness for the WS specimens for d0 = 1570 mm is √ SEL CM found to be KIc = 1. Also. Alternatively..40 ksi in). (Br¨hwiler.05 2.81 1.) 19 38 19 38 76 38 76 a0 mm (in.5 in) (Br¨hwiler.9: Size Eﬀect Law model assessment from the WS test program (average values)(Br¨hwiler..E..E. E. J. J. u and Saouma.5 64. E.54 MN/m3/2 (1. E. Broz.0 69.. from “small” laboratory test values of a geometrically identical “inﬁnitly large” one.1 17.
Figure 17. Fig. the SEL covers a range of sizes of about 1:20. 1991) that as u SEL d → ∞ KIc = Y Bft d0 Using this equation.372
FRACTURE MECHANICS PROPERTIES OF CONCRETE
Table 17. the Size Eﬀect Law can be rearranged in a formulation to predict the fracture SEL toughness KIc .68 1.8
0.21 1. It can be√ shown. we determined that the shape and geological composition of the aggregates are far more important parameters. and Saouma.20
304. 1991) u
da mm (in. 17. which is 51% higher than the value of KIc √ 3/2 (0. when geometrically similar specimens with identical concrete mix were considered then indeed the size eﬀect law proved to be a valid tool to determine structure’s nominal strength.) 248 248 762 762 762 1.. Broz.. J. The diﬀerence between KIc and KIc appears to be large a u priori however it can be explained by the two diﬀerent approaches used to evaluate fracture toughness from experimental data. 1991) Finally.65 1.. = 1.Draft
12
Specimen “Small” “Small” “Medium” “Medium” “Medium” “Large” “Large” Size d mm (in. Broz. and Victor Saouma Fracture Mechanics
.1 63.45 152.02 MN/m SEL CM Br¨hwiler and Boggs 1991).9.
7 Notation and Abbreviations therefore.sp GF H h KI KIc AP KIc P PV PCR PSP s t w AE ASTM CJ CMOD CT FPZ LEFM LVDT MSA NLFM RMS SOM SSD TPB WS
17.7
a a0 aef f Cexpi Cexp Cn D E Eef f fc ft ft. extrapolation to inﬁnite size may not be valid.
13
Notation and Abbreviations
Crack Length Initial Notch Length Eﬀective Crack Length Initial Experimental Compliance Experimental Compliance Normalized Compliance Diameter of Loading Rods Young’s Modulus Eﬀective Young’s Modulus Concrete Compressive Strength Concrete Uniaxial Tensile Strength Concrete Splitting Tensile Strength Speciﬁc Fracture Energy Specimen Height Specimen Ligament Length Stress Intensity Factor Fracture Toughness (Critical Stress Intensity Factor) Apparent Fracture Toughness Vertical Force Splitting Force Corresponding to Unload/reload Cycle Splitting Force Distance from Top of Specimen to Point of Load Application Specimen Thickness Specimen Width Acoustic Emission American Society of Materials Testing ColdJoint Crack Mouth Opening Displacement Compact Tension Fracture Process Zone Linear Elastic Fracture Mechanics Linear Variable Diﬀerential Transformer Maximum Aggregate Size NonLinear Fracture Mechanics Root Mean Square Strength of Materials Saturated Surface Dry Three Point Bending Wedge Splitting
Victor Saouma
Fracture Mechanics
.Draft
17.
Draft
14
FRACTURE MECHANICS PROPERTIES OF CONCRETE
Victor Saouma
Fracture Mechanics
.
1
Introduction
Fracture of Concrete
There is not yet a standard for concrete fracture testing although several have been proposed (RILEM. By deﬁnition. cracktip opening displacement CT OD.1. natural objects (such as rivers. which characterize the material resistance to cracking) and the roughness of the fracture surface as characterized by the fractal dimension D. coupled with our bias toward Euclidian geometry. This increases the total length of the fractal curve by four thirds
. is deﬁned. Fig. For the triadic Koch curve. and at least three additional parameters that characterize the concrete strain softening curve.Draft
Chapter 18
FRACTALS.
18. or discontinuous.1. which describes a transformation on the initiator. As the transformation operation is repeatedly applied to the ﬁgure. fragmented. 1983) . This chapter addresses a technique that uses fractal geometry to determine whether there is any surface indication of the direction of crack propagation in concrete and whether there is a correlation between fracture properties (KIc and GF . Concrete fracture properties include fracture toughness KIc . which represents an initial geometric form. handicapped us in studying and properly modeling such objects until the advent of fractal geometry (Mandelbrot. the triadic Koch curve. then a generator. Hence. an initiator. an inherent property of fractal objects is the statistical replication of patterns at diﬀerent scales. brittleness number β. 1985).. we note that at each step the number of line segments is increased by a factor of 4 and that the length of each new line segment generated is one third the length of line segments in the previous generation. The apparent randomness of these irregularities. fracture energy GF . FRACTURES and SIZE EFFECTS
Adapted from (?)
18. 18.1A illustrates the generation of a selfsimilar synthetic fractal curve.1
18. is applied. First. a fractal curve is generated. clouds.2
Fractal Geometry
Although most manmade objects have linear or smoothly curvilinear shapes. mountains. a magniﬁed part of a fractal object is statistically identical to the whole. characteristic length lch . The implications of the fractal nature of the cracked surface on the “true” fracture or surface energy are then discussed. and fractures) are commonly rough.
1: (A) Straight line initiator. and triadic Koch curve. (C) Modiﬁed Koch curve. fractal generator.
Victor Saouma
Fracture Mechanics
.Draft
2
FRACTALS. (B) Quadratic Koch curve. FRACTURES and SIZE EFFECTS
Straight line initiator
q ✔ ✔ ✔❚ 1/3✔ ✔ ❚ ✔ ✔ ❚ ✔ ✔ q ❚ ✔ ❚ ✔❚ ✔ ❚
Fractal generator
✔❚ ✔ ❚
❚ ❚ ✔ ✔
✔ ✔ ❚ ❚
✔❚ ✔ ❚
Fractal ﬁgure (ﬁrst iteration)
q
q
q
q
1/3
1/3
q
1/3 (A)
Length of subparts (S) of fractal generator segments
q ✔ ✔ r ✔ ✔ ✔ ✔ q ✔ ✔
r
q
❚
❚ ❚ ❚
r
q
q
r
q
r
q
r
q
r
q
q
r
q 2 q
r
r (C)
q 2 q
r
r
q
(B)
Figure 18.
if the extrusion is fractal. Note that no fractional lengths can be taken.
18. D = 1. if the triadic Koch surface is uniformly extruded along the z direction. a divider (that is ruler) of given length S is used to determine the total length of the curve.2619. 4. and 16 .3
Numerical Determination of Fractal Dimension
Although the fractal dimension D of synthetic curves can be readily determined analytically.1A. 18. 18.2) ln N . Thus.1: Fractal dimension deﬁnition at each step. The total length will eventually tend to inﬁnity. Alternatively.1. A fractal is a set for which the HausdorfBesicovitch dimension (D) strictly exceeds the topological dimension (DT ). as the number of iterations increases. 9. For the curve shown in Fig. oﬀered the following tentative deﬁnition of a fractal. the curve length L is given by SN where N is an integer.1)
Note that the Euclidian deﬁnition of dimension. the dimension of arbitrary curves requires the use of numerical methods. whereas the HausdorfBesicovitch dimension is 1. Hence. we subdivided line segments into four parts. 1988) provided that we have two independent fractal sets. 1983). Mandelbrot (1983) showed that.1A the topological dimension is 1.2619 + 1 = 2. 3 . the number of subdivisions N will 1 vary from 1.2619 (Feder. ln s (18. for a set S that is the product of two independent fractal sets S1 and S2 the fractal dimension of S equals the sum of the fractal dimensions of S1 and S2 . Mandelbrot (1983). is recovered if a line is subdivided into equal parts without changing its length. then D is greater than 2. (18. The fractal dimension D is deﬁned as D= An alternative form of this equation is N = sD .1.1 Introduction Step 1 2 3 N 1 4 16 S 1
1 3 1 9
3 s 1 3 9 Length 1
4 3 16 9
Table 18. thus N = 4 and we used S = 1/3 or s = 3.Draft
18. Hence. This gives a fractal dimension of D = ln 4 = 1. for a given ruler length. Thus. the fractal dimension of the surface is S = S1 + S2 = 1. and Mandelbrot. In our previous example. that is the topological dimension. Since N = S in this case. fractal surfaces have DT = 2 and fractal dimensions D > 2. 1987. This operation is repeated for a ruler of
Victor Saouma
Fracture Mechanics
. 3.2619. Three of the most commonly used methods for measuring the fractal dimension of surface roughness are reviewed here.2619 for the curve shown ln 3 in Fig. 16. and the corresponding 4 lengths will be 1. as a rule of thumb. s = S where S is the scaling factor from 1. The results of 9 this operation are tabulated in Table 18. Ruler Method: In the ruler method (Aviles and Scholz.
Indeed.5)
18. Victor Saouma Fracture Mechanics
. which for our proﬁles is approximately 8 × 10−2 meters. (Barton et. Brown (1987. For each contour line of diﬀerent elevation over the fracture surface. We expect the box method to exhibit a crossover length when applied to selfaﬃne data sets such as our fracture surface proﬁles. each having a diﬀerent cell size. ranging from clustering of cocitations in scientiﬁc papers (Van Raan. It has been used in numerous ﬁelds. Fracture roughness proﬁles in rock have been shown to be selfaﬃne (Brown 1987. unpublished).1. If a linear relationship is found. a sequence of grids. FRACTURES and SIZE EFFECTS
smaller and smaller length S.
2
(18. In principal. is strictly applicable to selfsimilar data sets. in this study we limited the largest box size to 8 × 10−2 m.al. 1 = S · ( )D S
S 1−D
(18. which is deﬁned as the ruler length beyond which the ruler method will always yield a fractal dimension of approximately one.3)
S·N
L
ln L = (1 − D) ln S.. 1990) to earthquake prediction (Levi. as well as the ruler method. 1988).4)
Box Method: The box method is used in this study to measure the fractal dimension of the fracture roughness proﬁles. fracture and fraction (fractals have fractional dimensions) have the same latin root fractus. 1988) A = L D−1 . and the fractal dimension is given by N 1 = ( )D S = sD . The fractal dimension is then be obtained from (Feder. which have yielded conﬂicting conclusions. The box method. 1988) shows that the crossover is the length where the standard deviation of amplitude is equal to the wavelength. which means to break and to create irregular fragments. then the object is said to be fractal within the speciﬁed range of cell sizes and the fractal dimension is the slope of the line. the method is simple. the Slit Island method is applicable to fractal surfaces (2 < D < 3). 1990). The roughness and irregularities of fractured surfaces make them ideal candidates for a fractal analysis. Slit Island Method: Whereas the preceding two methods measure the fractal dimension of fracture proﬁles (1 < D < 2). is placed over maps of the fracture traces and the number of cells intersected by fracture traces counted.Draft
4
FRACTALS.4
Correlation of Fractal Dimensions With Fracture Properties
Fractal analysis is a simple and powerful tool for quantifying complex physical quantities. as pointed out by Cahn (1989). Therefore. Brown’s analysis shows that application of the ruler method to selfaﬃne data sets yields the correct fractal dimension only when the maximum ruler length used is less than the crossover length. The number of grid cells intersected by the proﬁle N is then plotted on a loglog scale with respect to 1 the inverse of the grid cell size s = S .
(18. the perimeter L and the area A are determined and then plotted on a loglog scale. This potential correlation has been the subject of numerous studies.
later on. Discrepancy between results was recently pointed out by Cahn (1989). with fractal dimensions very close to one. Recently. or cement pastes. Mecholsky and Freiman (1991) showed that there is a spatial variation of D along the length of the crack reﬂecting crack initiation and. Mecholsky et al. E is the elastic modulus. should be adjusted to reﬂect the rough nature of the surface rather than the ﬂat nominal surface. Similarly. Similar results were subsequently obtained for single crystal silicon (Tsai and Mecholsky. it should be pointed out that it has been recently hypothesized that there is a universal value (2. 1991) and for polystyrene fracture surfaces (Chen and Runt. Termonia and Meakin (1986) showed that the synthetic fracture surfaces are also fractal and that the fractal dimension has a universal value of D = 1. and a0 a parameter having the units of length. Finally. and polycristalline ceramics). rocks. 1989).3) for the fractal dimension of fractured surfaces (Bouchaud et al. (1990) showed through a simple theoretical kinetic model which simulates crack propagation. Chelidze and Gueguen (1990) showed that the fractal dimension increases with the surface energy (and hence the fracture toughness) of rocks. This would then systematically result in fracture toughness values √ (KIc = Gc E ) that would increase with fractal dimension. Underwood and Banerji (1986).1 < D < 2. where D∗ is the fractional part of D. and an increase between fractal dimension and fracture toughness. Issa et al. Most recently. did not detect a linear fractal curve and constant D but rather a reverse sigmoidal curve and variable D. microbranching. hydrated cement paste is fractal and has a fractal dimension of about 3. They report a fractal dimension of 2. Peng and Tian. (1984) showed that not only are cracked metallic surfaces (300grade maraging steel) fractal but their fractal dimension is inversely proportional to the impact energy determined by the Charpy impact test (itself proportional to fracture toughness). a fractal approach was applied to the fracture of rocks and soil. Saouma et al.22 and increases with aggregate sizes. that the resulting fracture surface is fractal and that its dimension depends on the material’s elastic constants. (1992). Young and Crawford (1992) determined that fracture proﬁles are fractal. More recently. using AISI4340 steel specimens.02. Although most researchers have found that fractured surfaces are fractal.117 for ﬁne mortar. Mandelbrot et al. (1990) showed that cracked concrete surfaces are fractal. Mecholsky and Mackin (1988) correlated the fracture toughness of Ocala chert (ﬂint) with the fractal dimension and determined that one increased with the other.Draft
18. fractal dimensions and other physical properties. Through simulation of a numerical model for fracture based failure of a polymeric.087 for cement paste and 2. based on twodimensional surface studies.27 ± 0.1 Introduction
5
In the ﬁrst study of fracture surfaces. All of the reported studies concern themselves with metals. More recently. Heping (1987) postulated that the surface area of a crack for which the critical energy release rate Gc = dU dA is determined. polymers. Lange et al. on the microscopic level. the J integral ceases to be an invariant. Winslow (1985) showed through Xray scattering techniques that. For fractures in soils. and only limited research has been reported in the fractal analyses of cracked concrete surfaces. who noted that the relationship between fracture roughness as quantiﬁed by fractal dimension and fracture toughness is diametrically opposite for ductile fracture (maraging steel) and for brittle fracture (chert. (1993) investigated the correlation between fracture roughness and fracture toughness. Mosolov (1991) has shown that for fractal cracks. 1 1
Victor Saouma
Fracture Mechanics
. (1989) concluded that the toughness increases with the roughness for six diﬀerent alumina materials and ﬁve glassceramics having diﬀerent microstructures and that
2 the fracture toughness is directly related to the fractal dimension KIc = Ea0 D∗ 2 . found that the fractal dimension D varies from 2.1 to 2.
2). (1992).0µm/sec for the CMOD. Through this conﬁguration.54 cm. as well as the eﬀect of coldjoints.44 N/m. thick and with maximum aggregate sizes up to three inches1 . who tested both ductile and brittle materials and reported a universal fractal dimension of D = 2. a series of wedge splitting (WS) tests was performed on specimens 3 and 5 ft long (Fig. (1991) .910 Mpa m. This hypothesis was more recently strengthened by the ﬁnding of M˚ aløy et al.
18. 1 lb/in.
Since all experiments and data reduction were performed using U. 16 in. FRACTURES and SIZE EFFECTS
Figure 18. 1990) .006.1
Experimental Procedure
Fracture Testing
As part of a study on the applicability of fracture mechanics to the failure analysis of concrete dams by Saouma et al.11 ± 0. 1 ksi in=0.=2.48 cm.48 Pa.=175. Finally. the following conver√ √ sions factors should be considered: 1 in. as given in Table 18. In the WS experiments.Draft
6
FRACTALS. customary units.2: Frontal view of wedgesplittingtest specimen showing forces applied to specimen by lateral wedge loading (FS ) between two circular pins located near top of specimen on either side of the vertical starting notch.2. The eﬀects of both aggregate size and the type of aggregate (crushed subangular as opposed to river rounded) were studied. unload/reload cycles were performed to monitor change in specimen compliance from which the eﬀective crack length aef f was then calculated. Crack Mouth Opening Displacement (CMOD) gage straddles the initial notch.
1
Victor Saouma
Fracture Mechanics
. 1ft=30. During the test both vertical load and CMOD were monitored and recorded.2. 1 psi=689. which also provided the feedback control to achieve a constant rate of 1. stable crack growth with structural postpeak response was achieved.2
18. 18. the primary deformation measured was the crack–mouth opening displacement (CMOD).S.
orientation.0 in.0 in. 2. 4.
Table 18. 175/ in.3: Range and resolution of the proﬁlometer (inches)
18.001 in.0 in. 15. 192/ in. 160 /in. and resolution for the two specimen sizes investigated
Victor Saouma
Fracture Mechanics
. 4. the precision shown in Table 18. for each specimen proﬁle.005
X Y Z
Table 18. Finally. Two separate computers are used.171 2.Draft
18. which.0 in.2: Concrete mix design Range 24 60 5 Precision .0 in.5 in. Movements of the two bushing sets are controlled by twoworm gear and steppedmotor assemblies such that the LVDT can travel along any preprogrammed path in the X − Y plane.2 Experimental Procedure Contents Water (lbs/yd3 ) Cement (lbs/yd3 ) Water/Cement ratio Sand (lbs/yd3 ) Gravel (lbs/yd3 ) Maximum size aggregate 1.2
Proﬁle Measurements
A fully automated proﬁlometer (Saouma et al. 15.0 in. 160/in. 4. The entire system is placed on a rigid steel base over the specimen. SPECIMEN SIZE Proﬁle 0◦ length 90◦ ±45◦ Resolution 0◦ of 90◦ readings ±45◦ Spacing 0◦ between 90◦ proﬁles ±45◦ 3 ft 16.4: CHECK Mapped proﬁle spacing.66 0.4.0 in. the ﬁrst to control positioning of the LVDT. groundsteel rods which are rigidly attached to a second bushing. 192/in. 5 ft 26. 4.66 1. 3. are mounted at right angles with respect to the ﬁrst set of rods. 4.0 in.67 in. 280 208 424 315 0.2.. 17. which scans the surface by means of a motorized computercontrolled placement of a linear variable diﬀerential transformer (LVDT). The second bushing traverses a second pair of rods. 1990) .0013 . was used to determine crack proﬁle with an average resolution of 0.0 in.318 1.0021 .481
7
Table 18. the second to record and store its reading (following reception of a signal from the ﬁrst one). in turn.0 in.054 2.0 in. The LVDT is seated on a composite bushing that slides along two parallel. 20. 175/in.0 in.
a reference base line was ﬁrst deﬁned by simply connecting the ﬁrst and last proﬁle points.025.3.
18.2. The curves included straight lines at diﬀerent orientations to the box grid. and diagonally. and 8th generation triadic Koch curves (Fig. The box method can be applied to both selfaﬃne and selfsimilar proﬁles with the limitations stated above. whereas in the ﬂexible grid method only the ﬁrst one is ﬁxed. 18. two strategies were considered ﬁxed and ﬂexible grid boundaries. horizontally. 4th and 5th generations of the quadratic Koch curves (Fig. In our implementation. and (3) tolerance used in determining the number of intersected cells. 18. and hence it was adopted. 5th . In the method of ﬁxed grid boundaries the two end points are ﬁxed.1A).3
Computation of Fractal Dimension
Following preliminary tests. 18. (Fig 18. and 5th and 6th generation modiﬁed Koch curves (Fig. FRACTURES and SIZE EFFECTS
Initial Notch
✲ ✻ ✻ ✻ ✒✻ ❅ ❅ ✒ ❅ ✒ ❅ ✒ ❅ ❅ ❅ ❅ ❅ ❅ ❅ ❅ ✲ ❅ ❅ ❅ ❅ ❅ ❅ ❅ ❅ C L ❅ ❅ ❅ ❅ ❅ ❅ ❅ ❅ ✲ ❅ ❅ ❅ ❅ ❅ ❅ ❅ ❅ ❅ ❅ ❅ ❅ ❅ ❅ ❅ ❅ ❘ ❅ ✲ ❘ ❅ ❘ ❅ ❘ ❅ ✲
X (crack propagation direction)
Figure 18. 6th . vertically. The resolution of the two algorithms was determined by testing against synthetic fractal curves.5 and 18. it was determined that the box method yielded more consistent results than the ruler method.
Victor Saouma
Fracture Mechanics
. that is a count of the number of cells intersected by the proﬁle. it was concluded that fractal dimensions can at best be determined with an accuracy of ±0. then grids with a logarithmic size distribution ranging from 2 × 10−4 to 10−1 (from four times the proﬁle precision to the proﬁle length divided by four) were deﬁned. measurements along the four principal orientations (at 45◦ increments) were recorded in order to assess potential geometric anisotropy in fractal dimensions as shown in Fig. With regard to the initial grid deﬁnition.1B.1BC).3: Orientations of measured proﬁles over the fractured surface.Draft
8 Y
✻
FRACTALS. 18. (2) distribution and number of grid sizes.6. Based on the results shown in Tables 18. These test problems underscore the complexities in determining fractal dimensions of stochastic curves.4). Although the implementation of the box method may a priori appear to be quite simple. three factors inﬂuence its reliability: (1) orientation of the grid with respect to the proﬁle.
000 1.062 1.177 401 Flexible grids D Grid size 1.057 1.0◦ 10.000 1.007 1.000
40 points 1.0◦ 25.161
Table 18.262 1.068 1.020 1.017 1.0◦ Grid size 300 points 1000 points 1.486 500 1.307 500 1.188 93 1.017 1.030 1.006 1.190 500 1.2 Experimental Procedure
9
✚ ✚ ❅❍ ✚ ❍❍ ✚ ❍ ✚ ❈ ✚ ✟ ❈ ✟✟ ✟ ❈ ❈ ❈
❈
❈ ❈
✚ ✚ ❅❍ ✚ ❍❍ ✚ ❍ ✚ ❈ ✚ ✟ ❈ ✟✟ ❈ ✟
❈ ❈
❈
❈
❈
A
B
Figure 18.017 1.296 500 1. Angle of inclination 0◦ 5.477 500 1.Draft
18. Flexible grid boundaries.313 64 1.0◦ 15.007 1.0◦ 45.000
Table 18.4: Typical grid overlying an object.305 500 1.008 1.000 1.0◦ 30.078 1.500 1.009 1.017 1.017 1.0◦ 40.186 500
Koch curve type Triadic Triadic Triadic Quadratic Quadratic Modiﬁed Modiﬁed
Generation 5th 6th 8th 4th 5th 5th 6th
Theoretical Value 1.048 1.6: Computed fractal dimensions for various synthetic curves Victor Saouma Fracture Mechanics
.008 1.0◦ 20.0◦ 35.019 1. Dashed lines indicate adjustable sidFixed grid boundaries.008 1.5: Computed fractal dimensions of a straight line with various inclinations Fixed grids D Grid size 1.377 15 1.410 17 1.000 1.008 1.477 93 1.080 1.013 1.285 647 1. B.062 1.
FRACTURES and SIZE EFFECTS
for a Typical Fractured Surface 3.5 3.5). and (3) specimen and aggregate sizes and aggregate angularity. Having conﬁrmed the fractal nature of the cracked surfaces. three important questions arise: (1) what is the spatial variation of the fractal dimension.11 proposed for fracture surfaces by M˚ aløy et al.0
2.3..5 2.2 to 10−2. Number of occupied boxes (N ) is plotted versus box size. (2) is there a correlation between fractal dimension and fracture properties for concrete. Slope of line ﬁt to data is the fractal dimension (D). (D = 2.
18. The division by two to obtain the proﬁle fractal dimension from a surface fractal dimension was discussed previously.0 Log of Box Size (in meters)
3. 18. (1991)) and with the universal value of 2. assuming 2 that we do have two independent and orthogonal fractal sets as partially supported by Table 18. and is justiﬁed by the fractal nature of orthogonal proﬁles having almost identical fractal dimensions (as will be discussed following).0
2. D versus Proﬁle Location and Length: Variation in fractal dimension as a function of proﬁle location and proﬁle length was studied for a series of proﬁles oriented parallel to the
Victor Saouma
Fracture Mechanics
.5
2.3
Fractals and Fracture
The fracture surfaces of our specimens are fractal for grid sizes ranging from 2× 10−3.2 = 1.Draft
10
Log (N)
Box Sizes vs Number of Occupied Boxes FRACTALS.0
1. (1992) . (2) the orientation of the proﬁle with respect to the direction of crack propagation.1 m (Fig.5: Plot of box counting method applied to the proﬁle of a typical fractured concrete specimen.5
Figure 18. (1992).1 compares exactly with the one determined experimentally for concrete by Issa et al.1
Spatial Variation of the Fractal Dimension
The spatial variation of the fractal dimension was investigated with particular attention to the (1) location and length of the proﬁle within the width of the crack front. The average fractal dimension of approximately 1.1.7 and by Long et al. and (3) how do the results for concrete compare with other materials?
18.
114 1.7 2.5 1.5 1.110 1.094 1.107 1.064 1.109 1.104 1.9 1.096 1.123 1.085 1.127 1.098 1.097 1.088 1.101 1.105 1.1 1.109 1.3 Fractals and Fracture
11
Specimen
Proﬁle direction 0◦ 90◦ +45◦ −45◦ 0◦ 90◦ +45◦ −45◦ 0◦ 90◦ +45◦ −45◦ 0◦ 90◦ +45◦ −45◦ 0◦ 90◦ +45◦ −45◦ 0◦ 90◦ +45◦ −45◦
A
B
C
A
B
C
Proﬁle Distance from Centerline 1 in.123 1.111 1.1 2.0 0.5in.112 1.105 1.084 1.147 1.3 1. 3 in.088 1.090 1.098 1.9 0.8 1.094 1.073 1.8
Table 18.109 1.127 1.101 1.087 1.111 1.087 1.120 1.100 1.073 1.3 1. MSA 1.087 1.090 1.117 1.112 1.106 1.111 1. MSA 1.123 1.116 1.090 1.Draft
18.0 2.0 3.096 1.2 1.105 1.100 1.100 1. 5 in.7: Fractal dimension D versus proﬁle orientations
Victor Saouma
Fracture Mechanics
.7 2.122 1.092 1.104 1.7 1.096 1.106 1.109 1.096 1.2 1.108 1.094 1. 3ft specimens of 1.097 1.165 1.104 1.107 1.113 1.071 1.128 1.088 1.1 1.102 1.097 1.115 1.109 1.118 1.130 1.115 1.106 1.118 1.099 1.089 1.084 1.098 1.097 1.113 1.100 1.087 1.101 1.0 0.089 1.115 1.1 1.133 1.069 1.097
1.083 1.1 0.126 1.118 1.098 1.085 1.094 1.5 1.092
Average
σD %
1.127 1.106 1.107 1.106 1.129 1.123 1.097 1.096 1.092 1.100 1.111 1.148 1.101 1. 7 in.105 1.113 1.102 3ft specimens of 3.0in.107 1.108 1.115 1.1 0.123 1.115 1.098 1.125 1.
150 1. 7 in. D versus Specimen Size. 5 in. D versus Orientation of Proﬁle: It has been speculated that the direction of crack propagation might aﬀect the value of D.099 1.111 1.106 1.083 1. (1991) found that there is such a correlation for metals.119
First half 1.Draft
12 Distance from specimen centerline 1 in.093 1. This was conﬁrmed by visual inspection of the cracked surfaces.121 1.124
Table 18. implying a reduction in fractal dimension D as the proﬁle moves from the centerline to the edge.8).7). it was observed that in most cases D is greater over the ﬁrst half than over the second half. 2.110
FRACTALS.082 1.8: Fractal dimension for various proﬁle segments and distances from centerline in specimen S33A direction of fracture propagation on a single fracture specimen because it was postulated that: 1. it may be caused by the presence of a fully developed process zone in the ﬁrst half as opposed to the second one. FRACTURES and SIZE EFFECTS Zone of analysis Second Third Fourth quarter quarter quarter 1. We concluded by examination of the results (Table 18.113 1.084 1. When the fractal dimensions of the same specimen were determined over diﬀerent parts of the length (Table 18.082 1. Assuming a linear relationship between fractal dimension D and the proﬁle’s distance x from the centerline. 3 in. which showed that the center of the cracked surface appears to be rougher than the edge.118 1.100
Second half 1. If this is correct. D = a + bx (18. the lower the fractal dimension everywhere on the fracture surface.073 1.120 1. It should be noted that Long et al.118 1.and 5ftlong specimens (Table 18.126 1.128 1. Aggregate Size.080 1. that there is no apparent correlation between the fractal dimension and the proﬁle orientation.119 1. Even though the diﬀerence is quite small (more supporting experimental data is needed to fully address this point). This may result from the correlation between D and KIc (discussed Victor Saouma Fracture Mechanics
.075 1.6) it was observed that in most cases b is negative. Diﬀerent D values across the thickness might be obtained due to the near plane stress condition on the edge of the specimen and the near plane strain condition at the center.9). and Aggregate Type: Comparing the fractal dimension D obtained from 3. First quarter 1. In most cases D is greater in the ﬁrst quarter than in the fourth one. then it should be possible through postmortem investigation of a cracked surface to determine the direction of fracture propagation. Diﬀerent D values along the length may be obtained due to the lack of a fully developed process zone in the early part of the crack surface. there is a clear indication that the longer the specimen size.
076 1.128 1. 5.105 1. MSA (subangular basalt aggregate) 1.5in.083 1.3 1.17 909.096 1.36 1.138.073 1.3 Fractals and Fracture
13
Specimen S32A S32B S32C Average S33A S33B S33C Average SS32A SS32B Average S52A S52B S52C Average S53A S53B Average CJ52B CJ53A CJ53C Average
√ Distance from centerline Fractal dimension GF .5 1.060 1.080 1.064 1.0 1.094 1.3 1.0 1.004.072 1.113 1.9 1.082 1. 6.59 494
Table 18.2 1.085 1.049 1.29 976 3ft specimens of 1. 5ft specimens of rounded 1.40 1.073 1.63 1.73 1.090 1.052 1.112 1.061 1.058 1.087 1.085 1. MSA 1.118 1. Maximum size aggregate)
Victor Saouma
Fracture Mechanics
.21 901.057 0.117 1.9 1.098 2.8 1.42 1.083 1. KIc [psi in] 1 in.17 1.5 1.064 1.103 1.048 1.088 1.065 1. 13.107 1.109 3. 5ft specimens of rounded 3.28 812.097 0.3 1. 7.059 1.084 1.084 1.8 1.050 1.070 1.123 1.027 1. 3 in.6 0.0in.1 1.087 1.113 1.050 1.091 1.094 1.074 1.0in.46 457.27 908 3ft specimens of rounded 3.073 1.082 1.085 0.064 1.050 1. 10.100 1.073 1. 12.046 1.48 1.121 1.096 1.043 1.065 5ft cold joint specimens 1. 6.56 567.76 643.045 1.8 1.077 1.107 1.064 1.062 1.9 1.077 1. MSA 1.070 1.059 1. Average σD % [lb/in] Average σK % 3ft specimens of rounded 1.096 1.103 1.6 1.64 1.35 893.115 1. 7 in.073 0.074 1.8 1.Draft
18.274.7 1.3 1.073 1.100 0.049 0.098 1.037 1.081 1. MSA 1.078 1.0 0.096 1.9 Not applicable 1.6 1.5in.067 1.9: Comparison between D.5 1.120. 3. 11.147 1. 2.097 1.9 1.112 1.096 1.5 1.070 1.051 1.166.113 1.096 1.9 1.4 1. 6. KIc .089 1.164.118 1.107 1. 3. MSA 1.058.7 1.1 0.26 862.070 1. 5 in.137.5in.082 1.051 1.57 1. 3.064 1. 9.096 1.056 1.099 1.1 1. 6.130 1.206.069 1.055 0.066 1.059 1. and GF for all specimens (MSA.063 1.065 0.069 1.082 0.
This conﬁrms two earlier ﬁndings: (1) fracture toughness values are independent from aggregate sizes (Saouma et al..7) (18.
18.Draft
14
FRACTALS. debonding the aggregate from the mortar and yielding a rougher surface obtained through less energy.8)
Because of the low goodness of ﬁt.3
MacroScale Correlation Analysis
Noting that the range of variation for both D and KIc is indeed small. Table 18. Alternatively.5in.
18. KIc and GF . subangular aggregates yielded smaller D values than rounded ones.9 shows that for 3ftlong specimens having 1.3. the crack circumvented rounded aggregates while breaking subangular ones. We ﬁnd no correlation between D and the aggregate size. Table 18. This is because of the weak bond between the two layers of concrete that results in relatively smooth surfaces. and KIc tested higher for concrete than for the coldjointed specimens (Table 18. Mecholsky and Freiman (1991) reported a much wider Victor Saouma Fracture Mechanics
.3.87 − 5. If coldjoint specimens are excluded. χ2 = 0. The fracture plane went around rounded aggregate.6A and 18. Table 18. If a correlation could be found. great care should be exercised in using the above equations. as well as their respective averages and standard deviations.258 √ psi in. that shows bigger specimens yield slightly higher fracture toughness and smaller D values than smaller specimens. 18. (1991). χ2 = 0.9). (1991). The former would yield rougher surfaces whereas the latter would yield tougher ones. and Figs. a lower fractal dimension is synonymous with a higher fracture toughness because there is mostly aggregate rather than bond failure. Excluded are the results obtained from cold joints because these yield both lower fractal dimensions and lower fracture toughness due to the poor interface bond in those specimens. 220 − 4.10 summarizes this correlation between fracture properties and fractal dimensions. 766D lb/in. D increases with a decrease in fracture toughness and energy. by measuring roughness of the fracture surface in a forensic study. respectively. Linear regression between D and both the speciﬁc fracture energy GF and the fracture toughness KIc resulted in the following expressions: GF = 7.95D KIc = 6. This analysis is possible because the specimens were originally tested for both fracture toughness KIc and fracture energy GF as part of a diﬀerent research study of Saouma et al. 1984) is not dependent on aggregate sizes z as reported in Saouma et al. It should be noted that D.6B illustrate the variations of GF and KIc versus D. maximum size aggregate (MSA). This correlates with our observation that during fracture. FRACTURES and SIZE EFFECTS
in the next section).2
Correlation Between Fracture Toughness and Fractal Dimensions
We now turn our attention to the possible correlation between fractal dimension D and fracture properties (Table 18. The fracture plane went through this subangular strong aggregate leaving a relatively smooth surface. in regular specimens. Specimens prepared with subangular basalt aggregate had generally higher GF and KIc and lower D.9 summarizes fractal and fracture analyses for all specimens. GF . This can be explained by the nature of the fracture surfaces.9).267 (18. 1991) and (2) the sizeeﬀect law (Baˇant. a comparison with ceramics and alumina was performed. one could determine post mortem important fracture properties.
0 Concrete Alumina Glass ceramic 0.06 Experimental point GF=7.40
18.20 0.08
1.40 0. 1989). and alumina (Mecholsky and Frieman. ceramics. and alumina (Mecholsky and Freiman.10
1.11
1.10 Fractal Dimension (D)
1. D) KIc versus D for concrete (this study).0 1. ceramics.60 0. 1991).Draft
1. KIC)
1200.0 400.0 800. GF)
1.00 1.0 1. Ceramics
4.0
1.12
Fracture Energy vs Fractal Dimension Concrete.60 1. 1991).6: AB) GF and KIc versus D.80 1. 1988).11
1.0 1400.2204.09 1.0
1.20 Fractal Dimension (D)
1. Flint (Mecholsky and Mackin.95 D
1.0
Fractal Dimension vs Fracture Toughness
Various Materials
2.00
1.0 1.10
1.3 Fractals and Fracture
15
Fracture Energy vs Fractal Dimension
Concrete 1600.0
Fracture Toughness (Mpa m )
3.0
Fracture Energy (lb/in.0
4. GF)
1/2
3.07
1.08
1.0 600. Alumina.80 0.0
Concrete Alumina Glass ceramics Ocal chert (flint) Polystyrene Silicon
2.875.07
1.30
0.10 Fractal Dimension (D)
1.760 D
Fracture Energy (lb/in. C) GF versus D for concrete (this study).30
Figure 18.20 Fractal Dimension D
1.12
1.0 200.00 0.0 0.0
Fracture Toughness vs Fractal Dimension
Concrete
Fracture Toughness (psi sqrt[in]. polystyrene (Chen and Runt. and silicon (Tsai and Mecholsky (1991)
Victor Saouma
Fracture Mechanics
.09 1.0 1000.06 Experimental point KIc=6.00
1.20 1.
As an example.0 in.2 11.447 0.009 0. more speciﬁcally.6 11.2 13.0 in. FRACTURES and SIZE EFFECTS Proﬁle distance from centerline a b χ2 Sx.34 3.Draft
16 1. we conclude that this is caused by the heterogeneous nature of concrete and. for very large specimens the order of singularity of stress is − 1 .0 in. 5.605 6. the 2 total length of the proﬁle goes from 1.y %
GF = a + bD 5.0 in.160 0.953 KIc = a + bD 4066 2766 4396 3087 1510 444 5507 4147 6220 4766
0. these quantities depend on either the size of the specimen or of the ruler used. 18. we note that only concrete has a negative slope (decrease in fracture toughness with increase in fractal dimension). and those values are plotted along with the ones obtained from the present investigation (Fig.258 0. 7. A comparison of the fractal dimension and the fracture toughness in concrete determined in our investigation.4 12. In a fractal analysis.6D.4
Fractals and Size Eﬀects
The size eﬀect law (Baˇant.8 13.26 5. 3. however. 18. by the nature of the aggregate cement bond. 3. to 4 . Singularity of results is achieved for very small ruler or very large specimen sizes. consider the synthetic fractal curve of Fig. Average
FRACTALS.87 5.26 1. overall.212 0. 5. From this ﬁgure. to 16 as the ruler length decreases from 1. 18.1A.183 0. 1984) and a fractal analysis bear many similarities: z LogLog plot: In both cases the experimental investigation must be carried over at least two orders of magnitude of size (even though this is rarely done in size eﬀect studies).174 0. and may be explained by the diﬀerent nature of the fracturing process in concrete aggregate. 7.511 0.0 in. is plotted in Fig.4 14.31 4.1 12. D increases with KIc . Size/Scale dependency: Neither a fractal investigation nor a size eﬀect study yield a unique value for length/area or strength.0 in. There is no sharp discontinuity between the concrete and ceramic results.029 7.440 7. and the order of the singularity is 1 − D.2 13.0 in. all others have a positive slope.267
12. As indicated earlier. This observation on the mesoscale clearly contradicts our earlier ﬁnding on the microscale.5
Table 18. to 3 9
Victor Saouma
Fracture Mechanics
.0 in.10: Linear regression coeﬃcients between GF and KIc with D range of values. Alternatively.
18. as well as the same characteristics in other materials reported in the literature.003 0. an inﬁnitesimally small ruler should yield an inﬁnite proﬁle length.7 10.6).419 0. Average 1.
20 Fractal Dimension (D)
1.0
3.0
1.4 Fractals and Size Eﬀects
17
Fracture Energy vs Fractal Dimension Concrete.00
1. 1991)
Victor Saouma
Fracture Mechanics
. GF)
2.7: GF versus D for concrete (this study).10
1. ceramics.0 1.30
Figure 18.0
Fracture Energy (lb/in.0 Concrete Alumina Glass ceramic 0. and alumina (Mecholsky and Freiman. Alumina. Ceramics
4.Draft
18.
FRACTURES and SIZE EFFECTS
Fractal Dimension vs Fracture Toughness
Various Materials 4.0 1. ceramics.00
1.0
Fracture Toughness (Mpa m )
1/2
3. and alumina (Mecholsky and Frieman.8: KIc versus D for concrete (this study). 1989). Flint (Mecholsky and Mackin. polystyrene (Chen and Runt.20 Fractal Dimension D
1.0
0.30
Figure 18.Draft
18
FRACTALS. 1991). 1988).0
Concrete Alumina Glass ceramics Ocal chert (flint) Polystyrene Silicon
2.10
1.0
1. and silicon (Tsai and Mecholsky (1991)
Victor Saouma
Fracture Mechanics
.
4 a=9.6 L=36
19
Total Length (L(S))
10
3
10
2
10
1
10
6
10
5
10
4
10 10 Yardstick Length (S)
3
2
10
1
10
0
Figure 18.
For the 3ftlong specimens where L was measured with a ruler of length (S) approximately L equal to 0.. the correct proﬁle length LD is given by S∗ (18.2 a=14.103 a=22. this entails testing of various similar specimens
2 Ironically.9)
1 3.9. 18. From Eq.
S→
(18.11)
where P SEL and P F are selected properties obtained from the size eﬀect law and from a fractal analysis. a fractal dimension (D) of 1.4 we obtain the following relationship between the ruler length 9 S and the total length L(S) in terms of the fractal dimension D. From this ﬁgure it is apparent that practically any proﬁle length (hence surface area) can be obtained by choosing the appropriate ruler length2 .9: Variation of L(S) in terms of S to 1 .4.6.04. Rewriting Eq. (18. D=1.1 in.33. In the former. the larger the computed stress.
Victor Saouma
Fracture Mechanics
. L(S) = aS 1−D .103 was determined.4. respectively. S0 where L0 is the length measured with S0 . This ﬁgure also illustrates how a small variation in fractal dimension D results in substantially larger L(S). the smaller the crack tip element size.Draft
18. Hence.4 Fractals and Size Eﬀects
10
4
Effect of Yardstick on Total Length
a=28. In determining the correct surface area. 18. 1. reference is made to the surface area obtained through measurements with a ruler of size S0 .103 = 28. D=1. D=1.11−1.7. 18.2. Also shown are L(S) for various hypothetical fractal dimensions of 1.9. D=1.10) L∗ = ( )(1−D) L0 . We hypothesize that
d→∞
lim P SEL = lim P F .40 and the variation of L(S) is shown in Fig. this singularity is reminiscent of the one associated with the recovery of (singular) crack tip stresses in terms of the ﬁnite element mesh size used. and 1. a = S 1−D = 36 .
0059 and 0.13. 18.460 2 Ksi.0029 in. which would increase the surface area and thus reduce GF .12 2. although justiﬁed within the context of a ﬁnite element analysis where the roughness of the cracked surface is not accounted for. but how do we deﬁne the cracked surface area? Traditionally. Alternatively.0008 in).929 1.11 illustrates the ampliﬁcation factor for a surface area on the basis of Eq. Table 18. 4.929 Ksi in and E= 2. .0004 in.) and that S ∗ is a threshold or characteristic smallest length scale such as the size of an atom. and this approach.00040. 000 = 0.62 2. 18.76 2. FRACTURES and SIZE EFFECTS S∗ “Macro”: sand #100 # 200 . Hence. 1991): GF =1. As pointed out by Heping (1987).0008 in. with subsequent extrapolation to inﬁnity through the size eﬀect law.25 .02
Table 18. Eq. S ∗ can be determined on the basis of a “microscopic” scale associated with the approximate size of the calcium silicate hydrate (CSH).1 in. fracture energy GF should be decreased such that S0 (18. S0
(18.37 lb/in. which is approximately 1020µm (0. which is much lower than 2.12) d→∞ E For 1. For a surface area.18 4. with respective sizes of 0. the following overall average √ values were determined (Saouma et al. recall that GF is deﬁned as the energy required to create a unit surface area.. or crystal.63 3. however may not yield the correct material GF .12 we determine Gc = . KIc =.79 3.Draft
20 S0 1.35 lb/in. (18. 18.03 “Micro”: CSH 20µm 10µm . In order to validate the former hypothesis we consider the fracture energy GF for which the following relation holds K2 lim GF = Gc = Ic .11: Ampliﬁcation factors for fractal surface areas with D = 1. . 2.0029 in.10 becomes A∗ = = S ∗ × S ∗ (1−D) A0 S0 × S0 S ∗ 2(1−D) A0 .5in.78 3. The total energy experimentally consumed is determined from the area under the loaddisplacement curve.14) G∗ = ( ∗ )2(1−D) GF F S Victor Saouma Fracture Mechanics
.1 . maximum size aggregates in 3ftlong specimens.16 3. For concrete S ∗ can be determined either “macroscopically” as the size of#100 or #200 ASTM sieves.1
FRACTALS.0059 in.44 1.460 GF .13)
Recall that S0 is the size of the ruler with which L0 was determined (in our case 0. molecule. the roughness of the surface should be accounted for. in the latter it entails careful analysis of just one specimen using various rulers lengths.22 2. This quantity is referred to as a monomers by Feder (1988) and is associated with the physical structure of the surface. the nominal surface area has been used (RILEM 1985). To explain this discrepancy through fractal interpretation. Using Eq. .
it should be mentioned that whereas G∗ is quite sensitive to the assumed S ∗ values.45 lb/in ) . Finally.293 0.30 0. 3. F .203 0.241 0. we reexamine tests results of Swartz and Kan (1992).30 Ec psi × 106 4.30
18. Fracture surface proﬁles in concrete are fractal.727 0. fracture toughness increases with a decrease in fractal dimension in concrete.0004 Note that G∗ is closer to Gc than GF is.03 KIc √ psi in 922 1.30 NC.308 GF lb/in. Subsequently.824 0.06 to 1. This approach is to be contrasted with the Euclidian analysis of multiple specimen sizes through the size eﬀect law.08 4. 6. Using the same correction factor as the one employed above
.64 NP. 5.12.46 5.1) 1.225
Table 18.952 0. who used a diﬀerent fractal analysis technique.12.74 5. 0. 4. (1992). it was concluded that 1.1 . however. Victor Saouma Fracture Mechanics
. For the concrete studied here.266 1.37 = 0.1 2(1−1.5 5.1)
= 0.Draft
Specimen NC. who investigated the inﬂuence of aggregate/paste bonding on the mode I fracture = ( strength.and 5ftlong specimens having a maximum size aggregate of 3 in.64 0.19 0.280
21
G∗ F lb/in.419 0. F the choices of this later parameter is entirely justiﬁed from a purely material point of view. The fractal dimension of cracked concrete is insensitive to the orientation of the roughness proﬁle.
we obtained the results shown in Table 18.315 0.565 0.
18.64 HC.64 0.12: Comparison between “corrected” G∗ and Gc values based on Swartz Tests (1992).30 0.331 .187 0.286 0. were ﬁrst tested for fracture toughness KIc and fracture energy GF .5
Conclusions
Wedge splitting tests on 3.570 0. F As an additional evaluation of this approach.0004 2(1−1.523 1.30 HC.189 0. 0. proﬁle measurements along the cracked surfaces (with diﬀerent lengths and orientations) were performed and their respective fractal dimensions D determined. great care should be exercised in generalizing this ﬁnding. From this investigation. Using the proposed approach in which the cross sectional area is adjusted by the fractal dimension. The fractal dimension of the monotonically cast specimens is in the narrow range of 1.273 0.679 Gc lb/in.54 6. and the physical nature of the fracturing process should be closely examined. 0.025. 2. a fracture parameter was approximated from a single specimen size through fractal analysis.5 Conclusions
Aggregate type Crushed Limestone Crushed Quartzite Crushed & Polished Limestone Crushed & Polished Limestone Crushed Quartzite Crushed Limestone Water/cement 0. Fractal dimensions numerically determined by the box method have a resolution of ±. Fractal dimensions near the center of the fracture surface are slightly larger than those computed on the sides of the same surface.64 0.64 NP.206 980 1. These values are close to one independently measured by Issa et al.
There appears to be a linear relationship between fracture surface roughness as deﬁned by fractal dimension and fracture energy for concrete and ceramics. 8. FRACTURES and SIZE EFFECTS
7. 9. Some fracture mechanics parameters can potentially be recovered from a fractal analysis of a single small specimen without resorting to increasingly large specimen sizes as is required by the size eﬀect law. There are strong analogies between the fractal analysis and the size eﬀect law.Draft
22
FRACTALS.
Victor Saouma
Fracture Mechanics
.
Draft
Part VII
FINITE ELEMENT TECHNIQUES IN FRACTURE MECHANICS
.
Draft
.
in this chapter we shall use three criteria:
1. 3. Hence numerical techniques should be used. the stress intensity factors were determined by equating the numerically obtained displacements with their analytical expression in terms of the SIF. Finally. it would necessitate to have a very ﬁne mesh at the crack tip to approximate the stress singularity with nonsingular elements. 1973)) ones are only crude approximation. or the handbook ((Tada et al. either there is no analytical solution. the Ph. Whereas Boundary Element Methods are increasingly being used.2
Displacement Extrapolation
9 This technique was the predominant one prior to the serendipous discovery of the quarter point singular element.D. some of the more recent methods are partially covered in (Anderson 1995). Techniques through which the SIF can be computed a post priori following a standard ﬁnite element analysis via a special purpose postprocessor.
11
. surface tractions in 2D and 3D respectively.
8 Numerical methods for fracture mechanics can be categorized in many diﬀerent ways.1
7
Introduction
For most practical problems. it was recognized that unless singular elements could be used. Techniques in which the SIF are directly evaluated as part of the augmented global stiﬀness matrix. that is the r − 2 stress ﬁeld at the tip of the crack is properly represented.
In early ﬁnite element studies of LEFM.
19. For an overview of early ﬁnite element techniques in ﬁnite elements the reader should consult (Owen and Fawkes 1983). Those in which the singularity is modelled. thesis of Reich (Reich 1993) and of Cervenka (?) contain some of the major extensions of modern techniques to include thermal load.Draft
Chapter 19
SINGULAR ELEMENT
19. they are far behind in sophistication the Finite Element Methods which will be exclusively covered in this chapter.
1
2.
10
Following a liner elastic analysis. body forces. (Aliabadi and Rooke 1991).
1
12
Figure 19.
Barsoum (Barsoum 1974) and Henshell and Shaw (Henshell and Shaw 1975) independently demonstrated that the inverse square root singularity characteristic of linear elastic fracture mechanics can be obtained in the 2D 8noded isoparametric element (Q8) when the midside nodes near the crack tip are placed at the quarter point.
19. the midside nodes adjacent to the crack tip must be shifted to their quarterpoint position.3
Quarter Point Singular Elements
14 This section discusses the easiest and most powerful technique used in ﬁnite elements to model a stress singularity. in order to model a stress singularity without altering the ﬁnite element code.
15 16 Thus. A similar approach could be used using stresses rather than displacements. 19. this section will:
Victor Saouma
Fracture Mechanics
.1: Stress Intensity Factor Using Extrapolation Technique
13 Usually. and the predicted SIF values aould be the one extrapolated from the crack tip. Since then this element became known as the quarterpoint element. However this is likely to yield less accurate predictions. and then the following equations would be used: u v u v = KI = KII (2κ − 1) cos θ − cos 3θ 2 2 (2κ + 1) sin θ + sin 3θ 2 2 θ −(2κ + 3) sin 2 − sin 3θ 2 (2κ − 3) cos θ + cos 3θ 2 2
(19. they would be plotted with respect to their distance from the crack tip. θ is taken to be equal to π.1)
After the SIF would have been determined for each point.Draft
2 4G 4G 2π r
2π r
SINGULAR ELEMENT
Such a correlation will be performed along a radial line emanating from the crack tip. 17
In light of the simplicity and accuracy achieved by this element. Fig.
Draft
19.4 Review of Isoparametric Finite Elements 1. cover a brief review of the isoparametric element formulation 2. show how the element can be distorted in order to achieve a stress singularity 3. determine the order of the stress singularity 4. provide a brief review of all the historical developments surrounding this element
3
5. discuss the eﬀect on numerical accuracy of element size, order of integration, and local meshing around the crack tip 6. brieﬂy mention references to other singular elements
19.4
Review of Isoparametric Finite Elements
18 In the isoparametric ﬁnite element representation, both the internal displacement and coordinates are related to their nodal values through the shape functions:
Figure 19.2: Isoparametric Quadratic Finite Element: Global and Parent Element x y {d} = u v
8
=
i=1 8
Ni 0 0 Ni Ni 0 0 Ni
xi yi ui vi
(19.2) (19.3)
=
i=1
where the Ni are the assumed shape functions. For quadratic isoparametric serendipity elements (Fig. 19.2) the shape functions are given by:
19
Ni = Ni = Ni = Victor Saouma
1 (1 + ξξi ) (1 + ηηi ) (ξξi + ηηi − 1) , i = 1, 3, 5, 7 4 1 1 − ξ 2 (1 + ηηi ) , i = 2, 6 2 1 (1 + ξξi ) 1 − η 2 , i = 4, 8 2
(19.4) (19.5) (19.6)
Fracture Mechanics
Draft
4
20 21
SINGULAR ELEMENT
In Fig. 19.2, xi , yi are the nodal coordinates, ui , vi are the nodal displacements. As the strain is the derivative of the displacement, we will need later to deﬁne ∂N and ∂N . ∂x ∂y N has been deﬁned in Eq. 19.4  19.6 in terms of the natural coordinates ξ and η. Thus the chain rule will have to be invoked and the inverse of the jacobian will be needed. In this case, the jacobian matrix is: [J] = =
∂y ∂x ∂ξ ∂ξ ∂y ∂x ∂η ∂η 8 ∂Ni i=1 ∂ξ xi 8 ∂Ni i=1 ∂η xi
(19.7)
8 i=1 8 i=1 ∂Ni ∂ξ yi ∂Ni ∂η yi
(19.8)
22
The inverse jacobian is then evaluated from: [J]−1 = =
∂ξ ∂x ∂ξ ∂y ∂η ∂x ∂η ∂y ∂y ∂η − ∂x ∂η
(19.9) − ∂y ∂ξ
∂x ∂ξ
1 DetJ
(19.10)
23
The strain displacement relationship is:
8
{ }=
i=1
[Bi ] di
(19.11)
where [Bi ] is the strain matrix given by:
∂Ni ∂x [Bi ] = 0
∂Ni ∂y
0
∂Ni ∂y ∂Ni ∂x
(19.12)
where the following chain rule is invoked to determine the coeﬃcients of [B]:
∂N ∂x ∂N
∂y
= [J]−1
∂N ∂ξ ∂N
∂η
(19.13)
24 Finally, it can be shown that the element stiﬀness matrix of an element is given by (Gallagher 1975), (Zienkiewicz 1977):
1
1
[K] =
−1 −1
[B (ξ, η)] [D] [B (ξ, η)] detJdξdη
(19.14)
where the natural coordinates ξ and η are shown in Fig. 19.2 and [D] is the stressstrain or constitutive matrix.
25
The stress is given by: {σ} = [D] [B] ui vi (19.15)
Victor Saouma
Fracture Mechanics
Draft
19.5
19.5 How to Distort the Element to Model the Singularity
5
How to Distort the Element to Model the Singularity
26 In Eq. 19.15, if the stresses are to be singular, then [B] has to be singular as the two other components are constants. Consequently, if [B] is to be singular then the determinant of J must vanish to zero (Eq. 19.7) at the crack tip. 27 Now considering a rectangular element of length L along its ﬁrst side (123, in Fig. 19.2), we can readily see that both oﬀdiagonal terms ( ∂y and ∂x ) are zero. Thus, for the determinant ∂ξ ∂η of the jacobian to be zero we must have either one of the diagonal terms equal to zero.
It will suﬃce to force have:
28
∂x ∂ξ
to be zero. Making the proper substitution for
8
∂x ∂ξ
at η = −1 we
∂x ∂ξ
=
η=−1 i=1
Ni xi
1 [−1 + 2ξ + 2ξ + 1] (0) 4 1 + [1 + 2ξ + 2ξ + 1] (L) 4 1 + [−1 + 2ξ − 2ξ + 1] (L) 4 1 + [1 − 2ξ + 2ξ − 1] (0) 4 1 + (−2ξ − 2ξ) (x2 ) 2 L 1 + (−2ξ + 2ξ) 2 2 1 + (1 − 1) (L) 2 1 + (−1 + 1) (0) 2 1 1 (2 + 4ξ) L + (−4ξ) x2 = 4 2 =
(19.16)
29 After simpliﬁcation, and considering the ﬁrst corner node (where η = ξ = −1), we would have: L ∂x + 2x2 = 0 (19.17) ξ=−1 = 0 ⇔ (1 − 2) ∂ξ η=−1 2
x2 =
L 4
(19.18)
Thus all the terms in the jacobian vanish if and only if the second node is located at L instead 4 of L , and subsequently both the stresses and strains at the ﬁrst node will become singular. 2
30
Thus singularity at the crack tip is achieved by shifting the midside node to its quarterpoint position, see Fig. 19.3.
31 32
We should observe that instead of enforcing
∂y ∂η
∂x ∂ξ
along edge 13 to vanish at the crack tip, we
could have enforced Victor Saouma
along edge 17 to be zero at the crack tip. Fracture Mechanics
Draft
6
SINGULAR ELEMENT
33 A similar approach will show that if node 8 is shifted to its quarterpoint position the same radial strain variation would be obtained along sides 17. However, along rays within the element emanating from node 1 the strain variation is not singular. The next section will discuss this issue and other variation of this distorted element in more detail.
Figure 19.3: Singular Element (QuarterPoint Quadratic Isoparametric Element)
19.6
Order of Singularity
34 Having shown that the stresses at the ﬁrst node are singular, the obvious question is what is the degree of singularity. 35
First let us solve for ξ in terms of x and L at η = −1 (that is, alongside 123):
8
x =
i=1
Ni xi 1 (1 + ξ) (1 + 1) (ξ) L 4 L 4
1 L 1 − ξ 2 (1 + 1) + 2 4 1 ξ (1 + ξ) L + 1 − ξ 2 = 2 x ⇒ ξ = −1 + 2 L =
36
(19.19) (19.20)
Recalling that in isoparametric elements the displacement ﬁeld along η = −1 is given by: 1 1 u = − ξ (1 − ξ) u1 + ξ (1 + ξ) u2 + 1 − ξ 2 u3 2 2 (19.21)
37
we can rewrite Eq. 19.21 by replacing ξ with the previously derived expression, Eq. 19.20): u = − + + 1 −1 + 2 2 x L 2−2 x u1 L
1 x −1 + 2 2 2 L x x −4 4 u3 L L
x u2 L (19.22)
Victor Saouma
Fracture Mechanics
Draft
38 39
19.7 Stress Intensity Factors Extraction This complex equation can be rewritten in the form: u = A + Bx + C x L
1
7
(19.23)
We thus note that the displacement ﬁeld has had its quadratic term replaced by x 2 , which means that when the derivative of the displacement is taken, the strain (and stresses) are of the form: 4 4 4 1 1 −1 3 2 √ √ − + − u1 + u2 + √ u3 (19.24) x =− 2 2 xL L xL L xL L Thus the strength of the singularity is of order 1 , just as we wanted it to be for linear elastic 2 fracture mechanics !
40
19.7
41
Stress Intensity Factors Extraction
A number of techniques (including the ones discussed in the subsequent section) can be used to determine the SIF when quarterpoint elements are used, Fig. 19.4 but by far the simplest
Figure 19.4: Finite Element Discretization of the Crack Tip Using Singular Elements one to use and implement is the one based on the nodal displacement correlation technique.
42
This technique, ﬁrst introduced by Shih et al. (Shih, de Lorenzi and German 1976), equates the displacement ﬁeld in the quarterpoint singular element with the theoretical one. This method was subsequently reﬁned by Lynn and Ingraﬀea (Lynn and Ingraﬀea 1977) who introduced the transition elements, and extended by Manu and Ingraﬀea to threedimensional isotropic problems (Ingraﬀea and Manu 1980).
This method was ﬁnally extended to full threedimensional anisotropic cases by Saouma and Sikiotis (Saouma and Sikiotis 1986).
43
19.7.1
Isotropic Case
44 For the quarterpoint singular element, in two dimensions, and with reference to Fig. 19.5, the displacement ﬁeld is given by:
Victor Saouma
Fracture Mechanics
Draft
8
SINGULAR ELEMENT
Figure 19.5: Displacement Correlation Method to Extract SIF from Quarter Point Singular Elements u v = uA + −3uA + 4uB − uC = v A + −3v A + 4v B − v C
r L r L
+ 2uA + 2uC − 4uB + 2v A + 2v C − 4v B
r L r L
(19.25) (19.26)
where u and v are the local displacements (with x aligned with the crack axis) of the nodes along the crack in the singular elements. On the other hand, the analytical expression for v is given by Eq. 6.56f with θ = 180, yielding: κ+1 r (19.27) v = KI 2G 2π
45
Equating the terms of equal power ( 1 ) in the preceding two equations, the 2 and we obtain: 2G 2π −3v A + 4v B − v C KI = κ+1 L
46
√
r term vanishes, (19.28)
47 If this approach is generalized to mixed mode problems, then the two stress intensity factors are given by:
KI KII
=
1 2G 2κ+1
2π L
0 1 1 0
−3uA + 4 uB − uD − uC − uE −3v A + 4 v B − v D − v C − v E
(19.29)
48 Thus it can be readily seen that the extraction of the SIF can be accomplished through a “postprocessing” routine following a conventional ﬁnite element analysis in which the quarterpoint elements have been used.
19.7.2
49
Anisotropic Case
Following a similar procedure to the one previously described, for the anisotropic case,1 Saouma and Sikiotis (Saouma and Sikiotis 1986) have shown that the three stress intensity
1
Anisotropic modeling is important for either roller compacted concrete dams or layered rock foundations.
Victor Saouma
Fracture Mechanics
Draft
19.8 Numerical Evaluation factors can be evaluated from:
9
KI
II K III
K
= [B]−1 [A]
2π L
(19.30)
where [A] is obtained from the displacements of those nodes along the crack in the singular quarterpoint wedge element, as shown in Fig. 19.6:
[A] = 2v − v + 2v − v + v + 1 η (−4v + v + 4v − v ) + 1 η 2 (v + v − 2v ) C E F D B C E F F C D 2 2 B 2wB − wC + 2w E − wF + wD + 1 η (−4wB + wC + 4wE − wF ) + 1 η 2 (wF + wC − 2w D ) 2 2
2uB − uC + 2uE − uF + uD +
1 η (−4uB 2
+ uC + 4uE − uF ) +
1 2 η 2
(uF + uC − 2uD )
(19.31)
and [B] is obtained from the analytical solution to the displacements around the crack tip in homogeneous anisotropic solids:
i 1 Re s1 −s2 (q2 − q1 ) D [B]−1 = 1 Re s −i (s1 q2 − s2 q1 ) D 1 −s2
−i s1 −s2
Re Re
(p2 − p1 )
1 D 1 D
0 0
1
i s1 −s2
(s1 p2 − s2 p1 ) 0
(19.32)
0
(c44 c55 −c2 ) 45
19.8
50
Numerical Evaluation
Based on an extensive paramteric study, Saouma and Schwemmer, (Saouma and Schwemmer 1984), have formulated the following recommendations for the use of the singular quarter point element: 1. Use a 2 × 2 (reduced) integration scheme. 2. Use at least four (in pure mode I problems), or eight (in mixed mode problems) singular elements around a crack tip. 3. Have the internal angles of all the singular elements around the crack tip approximately equal to 45 degrees. 4. Unless an excessively small l/a ratio is used, little improvement is achieved by using transition elements. 5. For problems with uniform nonsingular stress distribution, little improvement is achieved by using a small l/a. 6. For problems where a nonsingular, stress gradient is expected, l/a should be less than 0.5.
51 Note that the above recommendations remain valid in mixed mode crack propagation studies. Although initially they cannot be directly extended to cases with high KII , it is well known KI that the crack tends to propagate in a direction that minimizes (but not completely eliminates) KII where the above recommendations would be valid.
Victor Saouma
Fracture Mechanics
Draft
10
SINGULAR ELEMENT
Figure 19.6: Nodal Deﬁnition for FE 3D SIF Determination
Victor Saouma
Fracture Mechanics
the singularity is along the edges and diagonal only. respectively). This led to the introduction of the transition element. when inserted around the singular elements. in which he has shown that the error in estimating the SIF is one order of magnitude smaller than the discretization error in locating the quarterpoint. one can control the point where singularity is to occur (between the corner node and inﬁnity.
56
Lynn and Ingraﬀea (Lynn and Ingraﬀea 1977) generalized the concept of the quarterpoint singular element and showed that by varying the placement of the side node. but that. whereas it is omnipresent inside the triangular element. Ying (Ying 1982) showed that in his investigation Hibbitt erroneously concluded that the strain energy of a rectangular quarterpoint element is singular. In a recent paper. Yamada et al. 53 Barsoum (Barsoum 1974) and Henshell and Shaw (Henshell and Shaw 1975) independently demonstrated that the inverse square root singularity characteristic of linear elastic fracture mechanics can be obtained in the 2D 8noded isoparametric element (Q8) when the midside nodes near the crack tip are placed at the quarter point. (Yamada. Such a discrepancy is caused by the fact that a collapsed 8node quadrilateral does not actually degenerate into an NIT (Newton 1973). Ezawa and Nishiguchi 1979) extended the concept of the 8node isoparametric element to the variable numbering element. which. Vasilakis and Pu 1981) extended the concept of quadratic transition elements to cubic transition elements. Hence the exact location of the midside node is not crucial as long as the discretization error is small.Draft
19.9 Historical Overview
11
Historical Overview
52 Because quarter point have been so popular (and still are) in light of their simplicity. and that in the collapsed quadrilateral element the singularity prevails along the two sides only. Barsoum (Barsoum 1977) showed that smallscale yielding (characterized by 1/r stress singularity) could be modelled. 54 Barsoum (Barsoum 1976b) then showed that the triangular element formed by collapsing one side of the Q8 led to far better results than the rectangular element. Also discussed by Ying is the error associated with the location of the quarterpoint of a singular element. Freese and Tracey (Freese and Tracy 1976) showed that the natural isoparametric triangle (NIT) and the collapsed quadrilateral perform equally well. This concept was subsequently extended to plate bending and shell fractures by Barsoum (Barsoum 1975). than there is a substantial deterioration in the SIF calculation from the collapsed quadrilateral. Also by having multiple independent nodes (of a collapsed Q8). (Hussain.
57
Victor Saouma
Fracture Mechanics
.
Hibbitt (Hibbitt 1977) proved that the singular rectangular element has a singular stiﬀness whereas the triangular one does not. as previously known. and no change in the NIT. He also demonstrated that a variety of stress singularities (1/n) can be achieved by isoparametric elements with a polynomial approximation of order n. Again Hussain et al. between quarterand midpoint. Hussain and Lorensen 1977). (Barsoum 1976a).9
19. He attributed this diﬀerence to the better results achieved by the triangular element as reported by Barsoum. If the side opposite to the crack tip is curved. this section will provide an overview of the major extensions to this simple method to determine SIF. 55 The extension of the quadratic isoparametric quarterpoint element to cubic isoparametric was proposed by Pu et al. (Pu. resulted in improved SIF calculations as the l/a (element length over total crack length) ratio decreased. This was further elaborated on by Barsoum’s discussion of the previous work (Barsoum 1982).
usually based on few analyses. This dependency of the SIF on the l/a ratio was subsequently further described by Lynn and Ingraﬀea. it is important to quantify the discretization error associated with the quarterpoint singular element. 62 Few of the above papers discussed the numerical accuracy achieved by the quarterpoint element in any detail or attempted to provide recommendations for its best usage.2 to 0. This source of error would still hold even if transition elements were used.03 in both two. Another detailed investigation of the optimum quarter pointelement size was carried out by Ingraﬀea and Manu (Ingraﬀea and Manu 1980). (Fawkes. Three sources of modeling errors associated with the quarterpoint elements were discussed: (a) those associated with the type of quarterpoint element. on the appropriate l/a ratio to be used. An optimum ratio of l/a=0. standard shape functions. analytical solutions. and it was found that the aspect ratio eﬀect was relatively unimportant.and threedimensional analysis. (b) the sides of the element should not be curved. (b) those associated with the conﬁguration of the element boundary. straight or curved. and concluded that “it is clearly impossible to recommend a particular crack tip element size suitable for all situations. Owen and Luxmoore 1979).25 was reported. and (c) those associated with the location of the quarterpoint node.Draft
12
SINGULAR ELEMENT
58 In his dissertation. which simulates the discrete crack propagation inside a ﬁnite element mesh with automatic remeshing.and threedimensional analyses. This study revealed that the best results were achieved by the analytic element. In this study. triangular or rectangular. 63 It is clear that such a numerical evaluation of this increasingly popular element is long overdue.20 to 1 percent for l/a = 0. On the other hand. where l/a was varied from 0. 59 An assessment of the quarterpoint elements (in pressure vessel fracture analysis) was oﬀered by Barsoum (Barsoum 1981) for both two. 60 In a recent paper.03. Ingraﬀea (Ingraﬀea 1977) investigated both the eﬀect of the l/a and aspect ratio of the singular elements. by using a singular element that is too small. elements based on the use of distorted shape functions. Harrop (Harrop 1982) qualitatively discussed the optimum size of quarterpoint crack tip elements. Thus any singular element that is too large cannot represent a structure nonlinear (and nonsingular) stress variation. Also one of the major arguments expressed against the use of this element is the SIF dependency on l/a. Furthermore. in light of a new generation of computer programs (Saouma and Zatz 1984).” 61 An assessment of various crack tip singular elements for use with isoparametric elements was discussed by Fawkes et al. and (c) perturbation of the quarterpoint node by e leads to an error in calculating the stress intensity factor of ge2 where g is the ratio of the crack tip element to crack length. The errors in the twodimensional results varied from 8 percent for l/a = 0. Barsoum indicated that (a) triangular elements are to be preferred over the quadrilateral ones (whether collapsed or not). and the next best group was the one using the distorted shape function. At best there have been conﬂicting indications. It should provide the stress analyst (who could be using a general purpose ﬁnite element program) with guidelines for the mesh preparation around the crack tip and some idea about the level of accuracy to be expected from the analysis. but the region of the mesh representing the stress singularity also decreases.
Victor Saouma
Fracture Mechanics
. the error in representing the ﬁnite stress term decreases. superposition process. He indicated that the singular element can represent the stress singularity and a constant ﬁnite stress term only. and hybrid techniques were evaluated using a test problem of both single and combined mode fracture. He thus pointed out that some crack tip element size has to be optimum.
33)
uij ki 2x − L x u3 L
(19. Kobayachi and Nakagaki 1979) 3. Although this method has yielded some very good results. two problems with known analytical stress intensity factors are considered. through appropriate and careful interpretation of those results. A number of parameters will be varied during the executions. It is anticipated that. and numerical values of the stress intensity factors will be compared to the exact ones.
Victor Saouma
Fracture Mechanics
. From the numerous analyses. but one should be aware of the following alternative formulations: 1.10
65
Other Singular Elements
For the most part this section discussed only the quarter point element. a data base will be created and graphically represented. which account for the singular terms:
4 4 4
ui =
k=1
fk uik + KI Q1i ¯
k=1
¯ fk Q1ik Qij =
+ KII Q2i
k=1
¯ fk Q2ik
(19. in which the shape functions are enriched with additional terms.10 Other Singular Elements
13
64 In the present study. guidelines for the mesh preparation around the crack tip could be achieved along with an approximate evaluation of the resulting accuracy.34) (19. exact ﬁeld modeling (Rao.
19. (Apostal 1974) 2.Draft
19. hybrid elements by Atluri (Atluri.35)
u = 1+
2x −3 L
x u1 + 4 L
x x −4 u2 + L L
66 In this method the stress intensity factors are treated as primary unknowns. This method was ﬁrst proposed by Benzley when the quarterpoint element (discussed in the next chapter) was not yet known. and are thus obtained during the global stiﬀness matrix decomposition. Raju and Krishna Murty 1971). its main drawback is that the ﬁnite element program has to be altered in order to implement it. enriched elements by Benzley (Benzley 1974).
Draft
14
SINGULAR ELEMENT
Victor Saouma
Fracture Mechanics
.
20.1
8
Mode I Only
Energy Release Rate
Recalling that the strain energy release rate G is given by: G=− K2 ∂U = I ∂a E ∆U ∆a (20. determine the total strain energy from either one of the following approaches: (a) U = ut Ku where u is nodal displacement. Fig. is a simple introduction to the second.1. Determine G from G
9
(U +∆U )−U (a+∆a)−a
=
∆U ∆a
=
2 KI E
Note that: 1. Increase the crack length from a to a + ∆a.
20. For an initial crack length a.1)
a simple algorithm for the SIF calculation emerges: 1.1. The former one. we will exploit the deﬁnition of the energy release rate G to derive a simple method of determining the stress intensity factors.1
20.Draft
Chapter 20
ENERGY RELEASE BASED METHODS
7 In this class of solutions. This procedure requires two complete separate analyses. 2. although of academic relevance only.
. and K is the global structural stiﬀness matrix. (b) U = ut P where P and u are the externally applied nodal load and displacement. 3. We shall distinguish between mode I and mixed mode cases. and reanalyze. respectively.
we obtain: G=−
1
∂[K] ∂{P } 1 u {u} + u 2 ∂a ∂a
(20.4)
It has nevertheless been shown that slightly improved results can be obtained if quarterpoint elements are used.2
Virtual Crack Extension.
10 As previously noted.Draft
2
ENERGY RELEASE BASED METHODS
∆a
Figure 20.3)
12
Noting that the ﬁrst term in the last equation is zero. 2 This same technique is now given two diﬀerent names: stiﬀness derivative (Park) and virtual crack extension (Hellen).
20.1 3.1: Crack Extension ∆a 2.1.2)
and −G = ∂Π ∂a =
∂ u ∂a
[K]{u} +
1 2
u
∂[K] ∂a {u}
−
∂ u ∂a
{P } − u − u
∂{P } ∂a ∂{P } ∂a
u = − ∂∂a ([K]{u} − {P }) + 1 u 2 0
∂[K] ∂a {u}
(20. thus both Park (Parks 1974) and Hellen (Hellen 1975) have independently proposed a modiﬁcation of the preceding approach2 . 4. Usage of an overrelaxation solver (such as the GaussSeidel) can reduce computational time for the second analysis in which the global stiﬀness matrix is only slightly altered.
Victor Saouma
Fracture Mechanics
. 11
The potential energy Π is given by: Π= 1 u [K]{u} − u {p} 2 (20. the stiﬀness matrix is only slightly perturbed in the second analysis (associated with a+∆a). The stress singularity need not be modelled. This technique is restricted to mode I only.
Only the portion of the stiﬀness matrix associated with the elements surrounding the crack tip needs to be perturbed. Such a “discrimination” is possible if we take into account the expressions of the energy release rate obtained by Hellen and Blackburn (Hellen and Blackburn 1975b) in which:
J1 = G1 = J2
2 2 KI + KII K2 + III E 2µ −2KI KII = G2 = E
(20. we note that: 1.2. The method can easily be generalized to threedimensional problems. 2. and E = 1−ν 2 for plane strain. determine the total strain energy U . and determine G2 . than the energy release rate is directly related to the derivative of the stiﬀness.5) (20.6)
E where E = E for plane stress. (b) Along θ = π .2 Mixed Mode Cases
3
13 Thus if the load is unaltered during the crack extension. For an initial crack length a. the algorithm for SIF extractions in mixed mode problems using virtual crack extensions is as follows:
1. Better results are obtained if singular elements are used. 4. 3. This method is restricted to mode I loading only. Extend the crack length from a to a + ∆a: (a) Along θ = 0. 2. we now generalize them to mixed mode loading. G1 and G2 are associated with π virtual crack extensions at θ = 0 and θ = 2 respectively. 5. 2 3.Draft
14
20.2
Mixed Mode Cases
15 Having presented two simple techniques for the SIF extraction for pure mode I cases. Solve for the two SIF from:
Victor Saouma
Fracture Mechanics
.
Finally.1
Two Virtual Crack Extensions. It can be shown (Parks 1974) that this technique is equivalent to the determination of the J integral.
20.
20.
16 A major limitation of the preceding methods is that in its present form it does not distinguish the mode I from the mode II components in the energy release rate. and determine G1 .
17 Based on this decomposition of G.
we can decompose the nodal displacements into two local components: (20.Draft
4 KI = KII =
G1 −G2 α
ENERGY RELEASE BASED METHODS s± s∓
8G2 α 8G2 α
s2 + 2 s2 + 2
(20.10) (20. we have a technique in which only one analysis is required.7) (20.2
Single Virtual Crack Extension.8)
where s = 2
18
and α =
(1+ν)(1+κ) E
An alternative to this technique is to use G(θ) = G1 cos(θ) + G2 sin(θ).2: Displacement Decomposition for SIF Determination u1 v1 u2 v2
{∆1 } = {∆2 } = Victor Saouma
= =
1 2 1 2
u+u v−v u−u v+v
(20.
With reference to Fig.2.9) ∆ = ∆1 + ∆2
20
where
y y’ x’ P y’ x’
x’ P’ y’
x
Figure 20. and use two distinct values of θ. In the single solution displacement decomposition method of Sha (Sha 1984) and Ishikawa (Ishikawa 1980). Displacement Decomposition
19 A major disadvantage of the preceding techniques is that at least one complete ﬁnite element analysis is required. which are not necessarily 0 or π .11)
Fracture Mechanics
.2. followed by either two separate ones or two virtual crack extensions. 2
20. 20.
we can determine (following one single analysis):
1 1 ∂[K] 1 ∂{P 1 } ∆ {∆ } + ∆1 2 ∂a ∂a 1 2 ∂[K] 2 ∂{P 2 } {∆ } + ∆2 G2 = − ∆ 2 ∂a ∂a G1 = −
(20.Draft
20.15)
We emphasize that the saving of one analysis (or virtual crack extension) is made possible through the constraint of having a symmetrical local mesh around the crack tip.13)
22 Because propagation is now assumed to be colinear.12) (20.14) (20.2 Mixed Mode Cases
5
21 Noting that better results are achieved if singular elements are used around the crack tip. we can determine the two stress intensity factors from
KI KII
23
= =
E G1 E GII
(20.
Victor Saouma
Fracture Mechanics
.
Draft
6
ENERGY RELEASE BASED METHODS
Victor Saouma
Fracture Mechanics
.
ds is the element of arc along path Γ. First let us restrict ourselves to the more general case in which isoparametric elements are used. than the numerical evaluation of J will be relatively simple. and is as follows:
11
1. and w is the strain energy density. Now let us start from the basic deﬁnition of J: J=
Γ
wdy − t ·
∂d ds ∂x
(21.
The algorithm for the J calculation closely follows the method presented in (Owen and Fawkes 1983). 9 In this chapter we shall present the algorithm to evaluate J on the basis of its contour line integral deﬁnition. its extension to Ji is quite straightforward. let us assume that the element connectivity is such that the path is along ξ = ξcst . Because the stresses are most accurately evaluated at the gauss points. the J integral is equivalent to G and we have: G=J =− ∂Π = ∂a (wdy − t · ∂d ds) ∂x (21. However. as in Fig. and hence care must be exercised in properly determining the J integral along a path passing through them. the path can be conveniently chosen to coincide with ξ = ξcst and/or η = ηcst . 21. For the sake of discussion. We note that for corner elements the integration will have to be performed twice along the two directions.1
7
Numerical Evaluation
Within linear elastic fracture mechanics. Evaluation of J according to the ﬁrst approach is identical to the one of G and has been previously presented.1)
r
8 Thus it is evident that we do have two methods of evaluating J: the ﬁrst one stems from its equivalence to the energy released rate. Whereas derivation will be for J integral only. We note
. most standard ﬁnite element codes only provide Gauss point stresses.2)
where t is the traction vector along n. d is the displacement vector.Draft
Chapter 21
J INTEGRAL BASED METHODS
21. 10 If the stresses were to be determined at the nodes. 2. and the second one from its deﬁnition as an integral along a closed contour.1. which is normal to the path.
1: Numerical Extraction of the J Integral (Owen and Fawkes 1983) that the crack is assumed to be along the x axis. The arc length ds and dy are given by: ds = dy = dx2 + dy 2 = ∂y dη ∂η ∂x ∂η
2
+
∂y ∂η
2
dη
(21.7)
7.3)
(21.2. 21.4) σx n1 + τxy n2 τxy n1 + σy n2 (21. The traction vector is given by: ti = σij nj ⇒ t = 4. 3.8)
Fracture Mechanics
.5)
6. If it is not. The strain energy density w is: w = = 1 (σx εx + 2τxy γxy + σy εy ) 2 ∂u ∂v ∂u ∂v 1 [σx + τxy ( + ) + σy ] 2 ∂x ∂y ∂x ∂y u v (21. The displacement vector is: d= 5.6) (21. Next we can evaluate part of the second term of J: t· Victor Saouma ∂u ∂v ∂d = (σx n1 + τxy n2 ) + (τxy n1 + σy n2 ) ∂x ∂x ∂x (21. Let us now determine each term of Eq.Draft
2
J INTEGRAL BASED METHODS
η
x
ξ=ξ cst
3 2 x5
x x
6
x x
9
x
8
x
ξ
7
x
1
4
n y
Gauss Point Numbering Sequence
x
Figure 21. stresses and displacements would ﬁrst have to be rotated.
which is a unit vector normal to the contour line at the Gauss point under consideration. 12.2 to obtain the contribution to J from a particular Gauss point within an element. 8.
1 1 ∂u σx + τxy ∂x −1 2
Je =
∂u ∂v + ∂y ∂x
w
+ σy
∂v ∂y
∂y ∂η
dy
− (σx n1 + τxy n2 )
∂u ∂v + (τxy n1 + σy n2 ) ∂x ∂x
t· ∂d
∂x ∂η
1
2
+
ds
∂x 2 ∂y dη ∂η
(21. 10. ηq )Wq
(21. τxy are readily available at the Gauss points.9)
=
−1
Idη
9. we have:
N GAU S
Je =
q=1
I(ξp . ∂y . Finally we are left to determine n1 and n2 (components of n). we substitute in Eq.
∂Ni ∂x [B] = 0
∂Ni ∂y
0
∂Ni ∂y ∂Ni ∂x
(21. Stresses σx . For instance ∂u = ∂Ni {ui } ∂y ∂x ∂x ∂Ni where the ui are the nodal displacements and ∂x is the cartesian derivative of the shape function stored in the [B] matrix:
∂u ∂u ∂v ∂x . Another term not yet deﬁned in Eq. and ∂v are obtained through the shape function. σy . ∂x . Since the integration is to be carried out numerically along the path (using the same integration points used for the element stiﬀness matrix).11)
where i ranges from 1 to 8 for quadrilateral elements. 11. Having deﬁned all the terms of J.1 Numerical Evaluation
3
where n1 and n2 are the components of n.10)
where Wq is the weighting factor corresponding to ηq and N GAU S is the order of integration (2 or 3).Draft
21. Victor Saouma Fracture Mechanics
.12)
13.
This term is actually stored already in
∂y ∂ξ ∂y ∂η
∂x ∂ξ ∂x ∂η
(21.9 is the Gauss point Jacobian matrix: [J] =
∂y ∂η . 21. 21.
and subsequently determine KI and KII from Eq. 20. 0 ∂η ∂η ∂ξ ∂η ∂ξ
D2
(21. ∂y ∂ξ . or:
∂x ∂η
∂x ∂ξ
i
j
∂y ∂η ∂y ∂ξ
k 0 0
(21. ?? we obtain J1 = J2 = Victor Saouma
2 K 2 + KII K2 ∂d }ds = I + III ∂x H µ −2KI KII ∂d }ds = {wdx − t · ∂y H
{wdy − t ·
(21. 0.18)
where N =
2 2 D1 + D2 and all terms are taken from the Jacobian matrix.
21.2
12
Mixed Mode SIF Evaluation
In subsection 21.15)
This leads to: C= 0.19)
combining with Eq. we can now return to the original plane and deﬁne D = C×A ⇒ D= ∂y ∂y ∂x ∂x ∂y ( − ). at best only KI may be determined. In this section.
13
First let us redeﬁne the two contour integrals according to (Knowles and Sternberg 1972) as: Jk = {wnk − t · ∂d }ds ∂xk (21. ∂η ∂η ∂ξ ∂η ∂ξ
D1
∂x ∂x ∂y ∂y ∂x ( − ).
0 0
(21.
J INTEGRAL BASED METHODS
(a) Deﬁne two arbitrary vectors: A along ξ = ξcst and B along η = ηcst such that:
∂y ∂η .
∂x ∂y ∂η ∂ξ
−
∂y ∂x ∂η ∂ξ
(21.8.13) (21. we shall generalize the algorithm to extract both J1 and J2 through a postprocessing for our ﬁnite element analysis.17)
(d) The unit normal vector is now given by:
n1 D1 N
n=
2 0
n
=
N 0
D2
(21. Once again the outlined procedure is based on the method outlined in (Owen and Fawkes 1983).1 we have outlined two procedures to extract the J integral from a ﬁnite element analysis.21) Fracture Mechanics
. Based on this technique. (b) Now we deﬁne a third vector. ∂x ∂ξ .Draft
4 At = B =
t ∂x ∂η .20) (21.14)
Note that we have deﬁned the threedimensional components of those two vectors.16)
(c) With C deﬁned. which is normal to the plane deﬁned by the preceding two: C = A × B.
we will start by evaluating the energy release rate. we then apply Green’s theorem.26) Fracture Mechanics
.
The essence of the method consists in replacing the contour integral. we have to perform only one analysis. Green’s theorem is invoked. we shall derive an alternative expression for the energy release rate.1
21.3. The method is really based on Rice’s J integral.Draft
where
14 15
21. and only subsequently we shall derive expressions for the SIF. We adopt the expression of J derived for a propagating crack (thus determined around a path close to the crack tip).
The procedure to determine J1 and J2 will be identical to the one outlined in 21. Eq. Having deﬁned a closed path. and replace a contour integral by a surface integral.130 J = lim (w + T ) δ1i − σij ∂uj ni dΓ ∂x1 (21. However.G.3
Equivalent Domain Integral (EDI) Method
16 In this section. by a closed integral (outer and inner) while multiplying the expression of J by a function q equal to zero on the outer surface and unity on the inner one. it is recognized that evaluation of J in 2D involves a line integral only and a line integral plus a volume integral if body forces are present.3 Equivalent Domain Integral (EDI) Method
5
H=
E
E 1−ν 2
plane strain plane stress
(21.25)
Γ0 →0
Γ0
where w is the strain energy density.1. in this method.23) (21.22)
We note that the original deﬁnition of J is recovered from J1 .1 and previously presented with the addition of the following equations: dx = −n2 ds ∂x dx = − dη ∂η
(21. 13.3. the line integral is replaced by a surface integral (and a volume integral for body forces). For 3D problems. (deLorenzi. and J will be evaluated through a volume integral in 3D and a surface integral in 2D. 18 Again as for the previous case.
19
21.24)
21. H. T is the kinetic energy 1 ∂ui ∂ui T = ρ 2 ∂t ∂t Victor Saouma (21. this method is quite attractive.1
20
Energy Release Rate J
2D case
Recalling the expression for the energy release rate of a propagating crack. Thus computationally. 1985). Contrarily to the virtual crack extension method where two analyses (or a stiﬀness derivative) had to be evaluated. 17 Recognizing that surface integrals may not be easily evaluated in 3D.
by considering the contour shown in Fig.28)
σij
∂uj − wδ1i qmi dΓ − ∂x1
Γ+ ∪Γ−
σ2j
∂uj qdΓ ∂x1
(21. this J integral is path independent only if Γ is within an elastic zone. we will be rewriting an alternative form of this equation.27)
Unlike the conventional J integral. Thus. For quasistatic cases (T = 0).2: Simply connected Region A Enclosed by Contours Γ1 . For linear (or nonlinear) elastic material J can be evaluated along either one of those two contours. and is thus applicable to all types of material models. Γ0 . and Γ− are respectively the upper and lower crack surfaces along the contour. (Li. and Γ+ . we can rewrite Eq. and Needleman. and q is an arbitrary but smooth function which is equal to unity on Γ0 and zero on Γ1 . m1 = 0 and m2 = ±1 on Γ+ and Γ− ). As such. F. 21.25 around the following closed contour
24
Γ∗ = Γ1 + Γ+ + Γ− − Γ0 yielding (and assuming that the crack faces are traction free) J=
Γ∗
(21.Draft
6 and δ the Kronecker delta. let us construct a closed contour by connecting inner and outer ones. and mi = −ni on Γ0 .29)
where mi is the outward normal to Γ∗ (thus mi = ni on Γ1 . 21. and Γ− . A. Z. (Anderson 1995) Γ0 is the inner vanishingly small contour. Victor Saouma Fracture Mechanics
. The outer one Γ1 is ﬁnite. while the inner one Γ0 is vanishingly small.
Γ Γ
1
mi
x2
Γ Γ

+
0
mi A*
ni
x1
Figure 21. F. and Shih. However. but only the inner one gives the exact solution in the general case. is the outer ﬁnite contour. C. Γ+ . 1985). 23 This equation is not well suited for numerical evaluation as the path would have to be along a vanishingly small one where the stresses and strains could not be determined. if it is taken within the plastic zone than it will be path dependent.
22 This equation is derived from an energy balance approach. the contour path for this equation can not be arbitrarily selected.
21
J INTEGRAL BASED METHODS
An alternative form of this equation (Anderson 1995) is J = lim
Γ0 →0
Γ0
(w + T ) dy − σij ni
∂uj dΓ ∂x
(21.2 where Γ1 .
and q = 1 on the inner one. the second term is equal to zero. 21.25 and 21.29 v.
30 We note that deLorenzi (deLorenzi. it is evident that the second term of Eq.n =
Γ A
∂vx ∂vy + ∂x ∂y
dxdy
(21.37)
Hence.39) G= ∆A ∂x1 ∂xi
Victor Saouma
Fracture Mechanics
.30)
we obtain J = = ∂ A∗ ∂xi
A∗
∂uj − wδ1i q dA ∂x1 ∂uj ∂q ∂ ∂uj σij − wδ1i + σij ∂x1 ∂xi ∂xi ∂x1 σij
(21. in the absence of crack surface tractions. 21. Eq.
Applying the divergence theorem to Eq.G.35)
substituting εij = we obtain 1 ∂ ∂w = σij ∂x 2 ∂x
29
1 2
∂uj ∂ui + ∂xj ∂xi ∂ ∂x ∂uj ∂xi ∂ ∂xj ∂ui ∂x
(21.33)
however from equilibrium we have ∂σij =0 ∂xi
28
(21.36)
∂ui ∂xj
+
= σij
(21.38)
A∗
This expression. 21. is analogous to the one proposed by Babuska for a surface integral based method to evaluate stress intensity factors. (Babuska and Miller 1984).32)
where A∗ is the area enclosed by Γ∗ .3 Equivalent Domain Integral (EDI) Method
7
25 Note that since the integral is taken along the contours.34)
Furthermore. H.29 are identical.Draft
26
21. by explicitly specifying q = 0 on the outer one.32 vanishes and that we are left with J= σij ∂ui ∂q − wδ1i dA ∂x1 ∂xi (21.31) − ∂w q dA ∂x1 (21. 1985) has shown that the energy release rate is given by ∂uj ∂∆x1 1 σij − wδi1 dA (21.
27
Let us show that the second term is equal to zero: ∂ ∂xi σij ∂uj ∂x1 = σij ∂ ∂xi
∂w ∂x
∂σij ∂ui ∂uj + ∂x1 ∂xi ∂x1
0
(21. the derivative of the strain energy density is ∂w ∂εij ∂εij ∂w = = σij ∂x ∂εij ∂x ∂x (21. Furthermore.
and x3 tangent to the crack front. we observe that the two expressions are identical for q = ∆x1 . We now deﬁne a weighted average J over the crack front segment of length ∆L as
33
¯ J∆L =
∆L
J(η)qdη
r0 →0
(21.1. In this context it was merely a mathematical device.3 we deﬁne a local coordinate system such that x1 is normal to the crack front. (Anderson 1995) For an arbitrary point.42)
(21. the J integral is given by Eq.25. we shall generalize to 3D our previous derivation. q is the weight function previously introduced. 21. S0 is the vanishingly small surface area of the tube.4 illustrates an incremental crack advance over ∆L where q is deﬁned as ∆a(η) = q(η)∆amax and the corresponding incremental area of the virtual crack is ∆Ac = ∆amax Victor Saouma q(η)dη
∆L
(21. 21.38 with 21. and thus q can be interpreted as a normalized ∆a virtual displacement. 21. From Fig.39. (Anderson 1995).
Figure 21.43) Fracture Mechanics
.3.3: Surface Enclosing a Tube along a Three Dimensional Crack Front. x2 normal to the crack plane. Thus comparing Eq. 21. We now consider a tube of length ∆L and radius r0 that surrounds the segment of the crack front under consideration. we have replace a contour integral by an equivalent area integral to determine J.40) ∂uj qni dS ∂x1 (21. q can be again interpreted as a virtual crack advance and Fig.2
32
3D Generalization
In this section. 21. In summary.Draft
8
31
J INTEGRAL BASED METHODS
for a unit crack growth extension along x1 .41)
=
lim
S0
wδ1i − σij
where J(η) is the pointwise value of J.
3 Equivalent Domain Integral (EDI) Method
9
∆ a max
q ∆ a max ∆L
Figure 21. this expression of J can not be numerically determined for a vanishingly small radius r0 . Note that this equation is the 3D counterpart of Eq. 21. as such and as in the previous 2D case. (Anderson 1995) As in the previous case.45)
and S+ and S− are the upper and lower crack surfaces respectively.44)
S ∗ = S1 + S+ + S− − S0
(21.4: Interpretation of q in terms of a Virtual Crack Advance along ∆L.29 which was written in 2D. S0 and S1 the inner and outer tube surfaces. 21.Draft
21. this equation reduces to a volume integral Fracture Mechanics
Victor Saouma
.5. we deﬁne a second tube of radius r1 around the crack front. Fig.5: Inner and Outer Surfaces Enclosing a Tube along a Three Dimensional Crack Front ∂ui − wδ1i qmi dS − ∂x1 ∂uj qdS ∂x1
¯ J∆L = where
S∗
σij
S− ∪S+
σ2j
(21.
35
Applying the divergence theorem.
34
Figure 21.
2
Extraction of SIF
36 From Eq. P.46)
A1 ∪A2
and q must be equal to zero at either end of ∆L that is on A1 and A2 . (Nikishkov. N. S. (Nikishkov. N.2. P.49)
2µG3
where.3. and Atluri. KII and KIII . P. G.47)
37 Note that k = 1. N. (Nikishkov and Vainshtok 1980) KI KII KIII = = =
1 ∗ 2 E √ 1 ∗ 2 E √
√
(J1 − J2 − G3 ) + (J1 − J2 − G3 ) −
(J1 + J2 − G3 ) (J1 + J2 − G3 ) (21. S. Again there are two approaches. G. 21. G2 and G3 known we need to extract the three stress intensity factors KI . Hence we shall generalize this equation and write it as (Nikishkov. 1987) (ignoring the second and third terms)
¯ Jk ∆L =
V∗
σij
∂q ∂ui ∂q −w ∂xk ∂xj ∂xk
dV
(21.50) for plane strain and E ∗ = E
which is a weighted value of E such that we retrieve E ∗ = for plane stress.48)
With G1 .
38
21. N.
21.Draft
10 ¯ J∆L = +
V∗
J INTEGRAL BASED METHODS ∂w ∂ ∂uj ∂q − wδ1i + − + ∂x1 ∂xi ∂x1 ∂xj ∂ui wδ1i − σij δ1i qdA ∂x1 ∂ui ∂x1
σij
σij
q dV (21. and Atluri. P. S.1
39
J Components
Based on the solution by Nikishkov. G. In (Nikishkov. G. The third term will also be equal to zero because q is arbitrarily selected to be zero at each end. and Atluri. S. However. 1987) it is shown that in the absence of nonelastic (thermal and plastic) deformations the second term would be equal to zero. 2 only thus deﬁning G1 = J1 and G2 = J2 .3. 1987) have shown that G3 has a similar form and is equal to
GIII =
V∗
σ3j
∂u3 ∂q ∂q − wIII ∂x1 ∂xj ∂x1
dV
(21.46 it is impossible to extract the 3 distinct stress intensity factors. and Atluri.
Victor Saouma
Fracture Mechanics
. 1987) E∗ = E 1 + 1 − ν2 ν 1+ν ε33 ε11 + ε22
E 1−ν 2
(21.
3 Equivalent Domain Integral (EDI) Method σ and u Decomposition
11
As for the solution by Shah.2
40
21. x3 ) = ui (x1 .51)
similarly the stresses are decomposed as {σ} = σ I + σ II + σ III σ11 + σ11 σ11 − σ11 σ +σ σ −σ 22 22 22 22 σ +σ 0 33 1 1 33 =2 + σ12 − σ12 2 σ12 + σ12 σ23 − σ23 0 σ31 − σ31 0 where ui (x1 . we can decompose the displacement ﬁeld as {u} = uI + u1 + u1 =1 2 u2 − u2 u +u 3 3 uII + uIII u1 − u1 +1 2 u2 + u2 0
0 +1 0 2 u −u 3 3
(21. −x2 . x3 ) = σij (x1 . x3 ) σij (x1 .56)
41 Whereas this method may be diﬃcult to use in conjunction with a 3D ﬁnite element mesh generated by triangularization (due to the lack of symmetry around the crack front).52)
(21. x2 . it has been succesfully used by Cervenka (1994) in conjunction with a unit volume integration in the ˇ FE code MERLIN (Reich. Cervenka and Saouma 1997).53) (21.
Victor Saouma
Fracture Mechanics
.55)
σkj
∂uk ∂q ∂q − wk ∂x1 ∂xj ∂x1
dV
(21. −x2 .3. x3 ) and the stress intensity factors are then determined from KI = where Gk =
V
∗
+
1 2
0 0 σ33 − σ33 0 σ23 + σ23 σ31 + σ31
(21.2.Draft
21.54)
√
E GI
KII =
√
E GII
KIII =
√
2µGIII
(21. x2 .
Draft
12
J INTEGRAL BASED METHODS
Victor Saouma
Fracture Mechanics
.
and ti are the displacements and surface tractions. Γ+ and Γ− .Draft
Chapter 22
RECIPROCAL WORK INTEGRALS
Chapter adapted from (Reich 1993)
22. Γ and Γ . corresponding to the crack surfaces complete the t t deﬁnition of Γ. respectively. respectively.
10
Γ is an arbitrary surface deﬁned in the counterclockwise direction around the crack tip but far away from it. For a solid free of body forces and initial strains and stresses the reciprocal work theorem is deﬁned as ˜ ti ui dΓ = ˜ (22. ˜
To apply the reciprocal work theorem to a cracked solid the simply connected region Ω must be deﬁned such that the singularity at the crack tip is avoided. associated with one equilibrium ˜ state and ui .1) ti ui dΓ
8
Γ
Γ
where Ω is any simply connected region within the solid and Γ is the contour of that region. t
13
. as is shown in Figure 22. The reciprocal work theorem deﬁnes the relationship between two equilibrium states for a solid.
11 12 Γ is a circle of radius centered on the crack tip that is deﬁned in the clockwise direction around the crack tip completely inside Γ.1. Γ+ is deﬁned on the upper crack surface between Γ t and Γ and Γ− is deﬁned on the lower crack surface bewteen Γ and Γ.
Another pair of surfaces. that begin on one crack surface and end on the other.1
7
General Formulation
In addition to conservation laws. associated with ˜ another equilibrium state.
9 The equilibrium state deﬁned by ui and ti is called the primary state and the equilibrium ˜ state deﬁned by ui and ti is called the complementary or auxiliary state. a form of Betti’s reciprocal work theorem (Sokolnikoﬀ 1956) can also be exploited to directly compute stress intensity factors (Stern 1973). This is accomplished by deﬁning a pair of surfaces. ui and ti are the displacements and surface tractions.
Draft
2
y
RECIPROCAL WORK INTEGRALS
14 Naturally. Ω is the region inside this closed path through the solid.
Γ
+ Γt
crack surfaces
Γε
n crack
tip
Γt

ε
x
Ω
n
Figure 22. respectively.5)
Victor Saouma
Fracture Mechanics
. for the elastic i state required to insure that boundary conditions on ui and σij are satisﬁed. (Stern 1973) determined that
18
Γ
˜ (ti ui − ti ui ) dΓ = ˜
Γ
˜ i (ts ui − ti us ) dΓ + o(1) i ˜
(22.
16
This expanded expression is then rewritten in the form of Somigliana’s identity to obtain ˜ (ti ui − ti ui ) dΓ + ˜ ˜ (ti ui − ti ui ) dΓ = 0 ˜ (22.2) ti ui dΓ + ti ui dΓ
15
Γ
Γ
Γ
Γ
in which the contributions from Γ and Γ are clearly separated. ue and σij are the displacements and stresses.3)
Γ
Γ
17
The displacements ui and the stresses σij for the primary state can be decomposed into ui = us + ue + u0 i i i s e σij = σij + σij (22.
˜ ˜ Recognizing that the product u0 ti has no contribution to the integral since the tractions ti i are self equilibrating due to the lack of body forces and taking into account the orders of the displacements and stresses in the various elastic states.4)
s where us and σij are the displacements and stresses. respectively.1: Contour integral paths around crack tip for recipcoal work integral
Assuming that Γ+ and Γ− are traction free the deﬁnition of the reciprocal work theorem can t t be rewritten as ˜ ˜ ti ui dΓ + ˜ ti ui dΓ = ˜ (22. and u0 are the i displacements of the crack tip. Since the material inside Γ is not included in the deﬁnition of Ω the singularity at the crack tip has been excluded. for the singular elastic state i e at the crack tip.
25 This choice for A normalizes the integrand for Γ involving the singular elastic state and the auxiliary singular state (Stern et al. 1975).9) c1 KI + c2 KII = i
Γ
as was shown in Equation 22.8)
with the o(1) term going to zero as
is decreased.6)
Based on this relationship a singular elastic state us and ts can be assumed. 1976). and the ˜ value I can be determined from the auxiliary singular state and far ﬁeld displacements and tractions. in polar coordinate system. are given by Westergaard as:
28
ur − u0 = r Victor Saouma
1 4µ
r 2π
1 2
(2κ − 1) cos
3θ θ − cos KI 2 2 Fracture Mechanics
. 24
The value of the complex constant A for the auxiliary singular state is determined to be A = 2µ (2π) (1 + κ)
1 2
(c1 + i c2 )
(22.L.
22 23 The procedure for constructing an auxiliary singular state will be outlined here using the homogeneous isotropic medium for this discussion. ui and ti .
27 The stress intensity factors. a crack in a homogeneous orthotropic medium (Stern and M. the displacements and the stresses. For the isotropic case. but since the integral is evaluated well away from the crack tip this is of no concern (Stern 1973). When the integral is evaluated using ui and ti obtained from the numerical method the constants associated with the coeﬃcients c1 and c2 are the stress intensity factors.1 General Formulation
3
As is decreased the elastic singular state us and ts becomes more dominant and the o(1) i i terms can be ignored allowing the integrals over Γ and Γ to be related in the following manner I = lim
→0 Γ
˜ i (ts ui − ti us ) dΓ = − i ˜
Γ
˜ (ti ui − ti ui ) dΓ ˜
(22. 26
Having determined A. in the neighborhood of the crack tip. computed using a suitable numerical method. Becker and Dunham 1976). Once the singular elastic state has been assumed. and a crack on the interface between dissimilar isotropic media (Hong and Stern 1978).Draft
19 20
22. the auxiliary singular state is constructed by taking λ as the negative of the value used in the singular elastic state. the product of this integral is I = c1 KI + c2 KII + o(1) (22. KI and KII .
21 Perhaps the most attractive feature of this approach is that the singularity at the crack tip need not be rigorously modeled in the numerical method used to obtain ui and ti . an auxiliary i i ˜ singular state ui and ti can be constructed from the assumed singular elastic state.7)
where c1 and c2 are arbitrary constants.6.
Auxiliary singular states have been constructed for a crack in a homogeneous isotropic medium (Stern. λ = − 1 in this case. can therefore be directly related to the integral over Γ ˜ (ui − u0 ) ti + ui ti dΓ ˜ (22. This of course means that the strain 2 energy for the auxiliary singular state is unbounded at the crack tip.
17)
uθ = ˜
θ θ 3θ 3θ + 3 sin − cos c1 + (2κ − 1) cos c2 2 2 2 2 θ θ 3θ 3θ − 3 cos − sin c1 + 7 sin c2 2 2 2 2 θ θ 3θ 3θ + 3 cos + sin c1 + sin c2 2 2 2 2 θ θ 3θ 3θ + sin − cos c1 − 3 cos c2 2 2 2 2
(22.18) (22. and 1 (22.13) 2 2 2 2
1 2
1 3θ 3θ θ θ + sin KI + cos + 3 cos KII + O r − 2 2 2 2 2
(22.19) (22.12) 2 2 2 2
3 cos sin
1 3θ 3θ θ θ + cos KI − 3 sin + 3 sin KII + O r − 2(22.14)
where u0 and u0 are the radial and tangential components. 1976): ur = ˜ 1 2(2πr) (1 + κ) θ θ 3θ 3θ − 3 cos − sin (2κ + 1) cos c1 + (2κ + 1) sin c2 2 2 2 2 1 2(2πr) (1 + κ) −(2κ − 1) sin σr = − ˜ µ 2(2πr 3 ) 2 (1 µ 2(2πr 3 ) 2 (1
1 1 1 2 1 2
(22.15) KI = lim (2πr) 2 σθ θ=0
r→0
KII = lim (2πr) 2 σrθ θ=0
r→0
1
(22. respectively.
Victor Saouma
Fracture Mechanics
.11) 2 2 1 3θ 3θ θ θ 5 cos − cos KI − 5 sin − 3 sin KII + O r − 2(22.Draft
4 − (2κ − 1) sin uθ − u0 θ = 1 4µ r 2π
1 2
RECIPROCAL WORK INTEGRALS
1 3θ θ − 3 sin KII + O r 2 2 2
(22.20)
7 cos + κ) cos + κ) µ
σθ = − ˜
σrθ = − ˜
2(2πr 3 ) (1 + κ)
1 2
3 sin
(22.16)
are the usual stress intensity factors.10)
−(2κ + 1) sin
3θ θ + sin KI 2 2
− (2κ + 1) cos σr = σθ = σrθ = 1 4(2πr) 2 1 4(2πr) 1 4(2πr)
1 2 1
1 3θ θ − 3 cos KII + O r 2 (22. The auxiliary solution to be used in the reciprocal work relation is based on Williams solution (Stern et al.21) where c1 and c2 are arbitrary constants. of the displacements r θ u0 of the crack tip.
the evaluation of the contour integral in terms of traction and displacement takes the form:
I
= −
π
C
((u − u0 ) · ˜) − u · t)ds t ˜ (22. meaning that the reciprocal work integral can also be extended include initial stresses without modiﬁying the auxiliary solution. 3. the description of the extension for body forces was rather superﬁcial. 1976). it is not always obvious how the reciprocal work theorem should be
Victor Saouma
Fracture Mechanics
. Unfortunately. we obtain: I = c1 KI − c2 KII (22.2 Volume Form of the Reciprocal Work Integral
5
29 Now.
22. Substitute into Eq. 22. for the case of thermal strains it was clearly shown that there is no need to consider thermal loading in the auxiliary state. However. being limited to a footnote. 2.L. Bastero.Draft
22.24 yield KI . on the inner circular boundary.
33 In addition to demonstrating the reciprocal work integral for cracks in homogeneous isotropic (Stern et al. Perform a linear elastic ﬁnite element analysis. Atkinson and Bastero 1991).2
Volume Form of the Reciprocal Work Integral
34 The ﬁrst step to be taken when formulating extensions to the reciprocal work integral is the deﬁnition of the reciprocal work theorem accounting for the applied loads in the two equilibrium states. 22. ?? now reduces to: c1 KI − c2 KII = u − u0 · ˜ − u · t ds t ˜ (22. Stern also proposed extensions for treating body forces (Stern et al. The components of c2 in Eq. Extract u and t (displacements and traction) from the analysis. 1975). 4.22)
=
−π
[σr (ur − u0 )σ˜ (uθ − u0 ) − σr ur + σrθ uθ ]rdθ ˜ ˜ ˜ r rθ θ
30
When the two solutions are substituted into the preceding equation. Atkinson and MartinezEsnaola 1989). Bastero and MartinezEsnaola 1988. Bastero and Miranda 1986. sharp notches (Atkinson.23)
31
Thus it can be readily seen that Eq.24 along with the auxiliary solution. and on the interface between dissimilar isotropic materials (Hong and Stern 1978). and cracks in coupled poroelastic media (Atkinson and Craster 1992). More recent developments include the treatment of dynamic crack propagation (Atkinson. 5. 1976) and thermal strains (Stern 1979).24 yield KII . The components of c1 in Eq.24)
C
32
From this equation an algorithm for the SIF determination emerges: 1. Unfortunately. homogeneous orthotropic (Stern and M. 22. and no example problems were presented for either of these extensions.
j = 0. ti . as
Γ
ti ui dΓ + ˜
Ω
bi ui dΩ = ˜
Γ
˜ ti ui dΓ +
Ω
˜i ui dΩ b
(22.j and ˜i = −˜ij. Recalling from ˜ ˜ b σ the equilibrium equation that bi = −σij. and bi is referred to as the primary state and the b equilibrium state deﬁned by ui .j dΩ ˜
(22. Applying Green’s theorem (Kreyszig 1979) to convert the line integrals to volume integrals yields
39
ti ui dΓ = ˜
Γ Ω
σij. surface tractions.j ui dΩ + ˜
Ω
σij ui. and ˜i is referred to as the auxiliary state.j ui dΩ ˜ (22.j ui dΩ = ˜ σij. and ˜i are the displacements.25)
where ui .Draft
6
RECIPROCAL WORK INTEGRALS
deﬁned to account for the applied loads.j = 0 and σij. and ni and collecting terms. Knowing that the line integral form of the reciprocal work theorem can be rewritten in ˜ the volume integral form shown in Equation 22. the appropriate deﬁnition of the reciprocal
40
Victor Saouma
Fracture Mechanics
. 36 (Sokolnikoﬀ 1956) deﬁned the reciprocal work theorem relating two separate equilibrium states for a solid.
37 The equilibrium state deﬁned by ui .j dΩ ˜ (22. ti .j dΩ = ˜
Ω Ω
σij ui. both including body forces. surface tractions. and body forces. and bi are the displacements. ui . and Γ corresponds to the entire surface of the solid.j ui dΩ + ˜
and the reciprocal work theorem clearly simpliﬁes to σij ui.j dΩ ˜
Ω
˜ ti ui dΓ =
Γ Ω
σij. ti . and body forces. ˜ ˜ b respectively.
35 It will be shown here that the line integrals in the reciprocal work theorem can be converted to volume integrals using Green’s theorem and that the form of the integrand for the volume integrals is such that the appropriate form of the reciprocal work theorem can be determined quite simply. ti . σij .26) ti ui dΓ −
Γ
Ω
Γ
Ω
˜ where σij and σij are the stress tensors for the two equilibrium states. allowing the line integrals to be written in a form compatible with Green’s theorem
ti ui dΓ = ˜
Γ Γ Γ
(−σi2 ui dx1 + σi1 ui dx2 ) ˜ ˜ (22.29.29)
In the absence of body forces the volume integrals are not included in the deﬁnition of the reciprocal work theorem and the expression shown above is still valid since σij. Ω corresponds to the volume of the solid.27) (−˜i2 ui dx1 + σi1 ui dx2 ) σ ˜
Γ
˜ ti ui dΓ =
˜ ˜ by expanding ti and ti in terms of σij . for one equilibrium state. for the other equilibruim state.
38 Adopting a counterclockwise path around Γ the expressions relating dΓ to dx1 and dx2 given in Equation ?? are still valid. particularly when they are the result of initial strains or stresses.j the reciprocal work theorem can be rewritten as ˜ ti ui dΓ − ˜ σij. respectively.28) σij ui.
quadratures based on sampling points that coincide with the nodal locations should be avoided since ui is singular at the crack tip. ˜
Victor Saouma
Fracture Mechanics
. However.33)
Γ
ˆ Provided that ti is not expressed in powers of r less than − 1 .31)
ˆ where Γt = Γ+ ∪ Γ− and ti is the applied surface traction vector on the crack surfaces in the t t primary state.3
22.32) ti ui dΓ +
Γ
Γt
Γ
Clearly.3 Surface Tractions on Crack Surfaces
7
work theorem to account for initial strains and stresses in the primary state can be obtained quite easily. This is accomplished by simply writing the volume form of the reciprocal work theorem such that there is a direct relationship between the stresses and displacements in the primary state.
Surface Tractions on Crack Surfaces
41 The extension to the reciprocal work integral to include the eﬀect of surface tractions on the crack surfaces in the primary state parallels the approach proposed by (Karlsson and B¨cklund a 1978) for the J integral.Draft
22. the limit exists and the stress 2 intensity factors are deﬁned as
47
c1 KI + c2 KII =
Γ
˜ [ ti (ui − u0 ) − ti ui ] dΓ − ˜ i
ˆ˜ ti ui dΓ
Γt
(22. the integrand of the integral over Γ is identical to that for the case of a primary state free of surface tractions on the crack surfaces. 42 For a primary state free of body forces with surface tractions on the crack surfaces the reciprocal work theorem is deﬁned as
ti ui dΓ = ˜
Γ Γ
˜ ti ui dΓ
(22. when the integral over Γt is evaluated i using numerical integration techniques.30)
43 This expression can be rewritten such that a separate integral is given for each portion of the contour path
Γ
ti ui dΓ + ˜
Γ
ti ui dΓ + ˜
Γt
ˆ˜ ti ui dΓ =
Γ
˜ ti ui dΓ +
Γ
˜ ti ui dΓ
(22.
44 This expression for the reciprocal work theorem can be rewritten in the form of Somigliana’s identity as ˜ ˜ ˆ˜ (ti ui − ti ui ) dΓ + ˜ (ti ui − ti ui ) dΓ = 0 ˜ (22.34)
where u0 are the displacements of the crack tip.
45 46
The value I is then deﬁned as I = − ˜ (ti ui − ti ui ) dΓ − lim ˜
→0
ˆ˜ ti ui dΓ
Γt
(22. which means that Equation 22.5 still holds and the solution for the auxiliary singular state is still valid.
53.5
53
Initial Strains Corresponding to Thermal Loading
For problems in thermoelasticity the constitutive law deﬁnes net eﬀective stresses σij in ¯ 0 . when the integral over Ω is evaluated i using numerical integration techniques. which are. in turn. However.Draft
8
RECIPROCAL WORK INTEGRALS
22. is deﬁned by the displacements ui . ¯
Victor Saouma
Fracture Mechanics
. as is shown in Equation 22.5 still holds and the solution for the auxiliary singular state described is still valid. the eﬀective stresses σij are then directly related to the displacements ui and should be used in the deﬁnition of the reciprocal work theorem rather than the net eﬀective stresses σij .39)
where u0 are the displacements of the crack tip.36)
50
This expression can be rewritten in the form of Somigliana’s identity as ˜ (ti ui − ti ui ) dΓ + ˜ bi ui dΩ + ˜
Ω Γ
Γ
˜ (ti ui − ti ui ) dΓ = 0 ˜
(22. the integrand of the integral over Γ is identical to that for the case of a primary state free of body forces. which means that Equation 22. The expression for the reciprocal work theorem can be rewritten such that a separate integral is given for each portion of the contour path
49
ti ui dΓ + ˜
Γ Γ
ti ui dΓ + ˜
Ω
bi ui dΩ = ˜
Γ
˜ ti ui dΓ +
Γ
˜ ti ui dΓ
(22. the limit exists and the stress 2 intensity factors are deﬁned as c1 KI + c2 KII =
Γ
˜ [ ti (ui − u0 ) − ti ui ] dΓ − ˜ i
bi ui dΩ ˜
Ω
(22. σij are the result of εij . 52
The value I is then deﬁned as I = − ˜ (ti ui − ti ui ) dΓ − lim ˜
→0
bi ui dΩ ˜
Ω
(22.4
Body Forces
48 For a primary state with body forces but free surface tractions on the crack surfaces and initial strains and stresses the reciprocal work theorem is deﬁned as
ti ui dΓ + ˜
Γ Ω
bi ui dΩ = ˜
Γ
˜ ti ui dΓ
(22. terms of the total strains εij and the thermal strains εij
54 σ ¯ij can be decomposed into eﬀective stresses σij and thermal stresses σij .37)
51 Clearly.35)
where bi is the body force vector. It should be noted that since the line integrals are deﬁned over Γ this form of the reciprocal work integral could also account for surface tractions on the crack surfaces. ˜
22.38)
Γ
Provided that bi is not expressed in powers of r less than − 1 . quadratures based on sampling points that coincide with the nodal locations should be avoided since ui is singular at the crack tip. Therefore.
44)
ˆ where ti is the applied eﬀective surface traction vector.i is the gradient of the temperatures.j = αCijkl T. which. This expression for the reciprocal work theorem can be rewritten in the form of Somigliana’s identity as
61
Γ
˜ (ti ui − ti ui ) dΓ + ˜
Γt
ˆ˜ ti ui dΓ +
Ω
bi ui dΩ + ˜
Γ
˜ (ti ui − ti ui ) dΓ = 0 ˜
(22.40)
Ω
Ω
56 The relationship between the line and volume integral forms of the reciprocal work theorem can be readily obtained by applying Green’s theorem to the volume integral with σij in the integrand
Γ
ti ui dΓ = ˜
Ω
σij.60. it is clearly evident that a volume integral is required to complete the deﬁnition of the reciprocal work theorem.45)
62 Clearly.5 still holds and the solution for the auxiliary singular state described in Section ?? is still valid.Draft
22.42)
where T.41)
where ti = σij nj is the eﬀective surface traction vector.j ui dΩ + ˜
Ω
σij ui. as deﬁned in Equation 22.j dΩ ˜ (22. Therefore.
Victor Saouma
Fracture Mechanics
.5 Initial Strains Corresponding to Thermal Loading
9
55 The volume form of the reciprocal work theorem for a primary state that includes thermal strains is σij ui.j dΩ ˜
(22. the form of the reciprocal work theorem in which the line integrals on Γ have been separated is
Γ
ti ui dΓ + ˜
Γ
ti ui dΓ + ˜
Γt
ˆ˜ ti ui dΓ +
Ω
bi ui dΩ = ˜
˜ ti ui dΓ +
Γ Γ
˜ ti ui dΓ
(22.
60 Recalling that the natural boundary conditions are deﬁned in terms of the total stresses. which means that Equation 22. as deﬁned in Equation 22.i δkl )˜i dΩ = u
Γ
˜ ti ui dΓ
(22.
59
A more general form of the reciprocal work theorem would be ti ui dΓ + ˜ bi ui dΩ = ˜ ˜ ti ui dΓ
Γ
(22. which in this particular case does not include a true body force vector bi . Recalling from the equilibrium equation that σij. much like the eﬀective body force vector bi .j dΩ = ˜ σij ui.43)
Γ
Ω
where bi is the eﬀective body force vector. does not include a true applied surface traction vector in this case. the integrand of the integral over Γ is identical to that for the case of a primary state free of initial strains.58. the appropriate form of the reciprocal work theorem for a primary state with thermal strains but no body forces is
58
Γ
ti ui dΓ − ˜
Ω
α (Cijkl T.
57
Therefore.j δkl in the absence of body forces. ti = 0 on the crack surfaces.
σij can again be decomposed.62. quadratures based on sampling points that coincide with the nodal locations should be avoided since ui is singular at the crack tip.46)
Provided that the temperature T is not expressed in powers of r less than 1 .Draft
10
63
RECIPROCAL WORK INTEGRALS
The value I is then deﬁned as I = − ˜ (ti ui − ti ui ) dΓ − lim ˜
→0
Γ
Γt
ˆ˜ ti ui dΓ − lim
→0
Ω
bi ui dΩ ˜
(22. Since εij is deﬁned in terms of the displacements ui . Recalling from the equilibrium equation that σij.j δij in the absence of body forces.i ui dΩ = ˜
Γ
˜ ti ui dΓ
(22. which does not include a true body force vector bi in this case.
67
Therefore.44
Victor Saouma
Fracture Mechanics
. but in this 0 instance the constituent stresses are σij and the initial stresses σij corresponding to the pore pressures. as was shown in Equation 22. Equation 22.i is the gradient of the pore pressures.
70 Recognizing that the general form of the reciprocal work theorem accounting for initial stresses is identical to that accounting for initial strains (i.
66
Therefore. the limit exists 2 and the stress intensity factors are deﬁned as c1 KI + c2 KII =
Γ
˜ [ ti (ui − u0 ) − ti ui ] dΓ − ˜ i
Γt
ˆ˜ ti ui dΓ −
Ω
bi ui dΩ ˜
(22. it is clearly evident that a volume integral is required to complete the deﬁnition of the reciprocal work theorem. the eﬀective stresses σij are then directly related to the displacements ui and the reciprocal work theorem is again deﬁned in terms of the eﬀective stresses in the primary state.
69
A more general form of the reciprocal work integral is ti ui dΓ + ˜ bi ui dΩ = ˜ ˜ ti ui dΓ (22. deﬁnes the total stresses σij in terms of the total strains εij and the pore pressures p. as deﬁned in Equation 22.
As was the case for problems in thermoelasticity.40 and 22.48)
where p.49)
Γ
Ω
Γ
where bi is the eﬀective body force vector. which is obtained by substituting the constitutive law deﬁning the eﬀective stresses σij into the principle of eﬀective stress.j = p. Equations 22. i
64 However.43).41 also apply when the primary state includes initial stresses.64. when the integrals over Γt and Ω are evaluated using numerical integration techniques. ˜
22.e. the appropriate form of the reciprocal work theorem for a primary state with pore pressures but no body forces is
68
Γ
ti ui dΓ − ˜
Ω
p.6
Initial Stresses Corresponding to Pore Pressures
65 The stressstrain relationship for poroelasticity. Equations 22.47)
where u0 are the displacements of the crack tip.
Draft
22.
22. respectively. the restrictions imposed on the power of r for ti .47 apply for initial stresses as well. the applied eﬀective surface traction vector is deﬁned by Equation 22.47.47
c1 KI + c2 KII =
Γ
˜ [ ti (ui − u0 ) − (ti + p ni + α T Cijkl ni δkl ) ui ] dΓ ˜ i ˆ (ti + p ni + α T Cijkl ni δkl ) ui dΓ ˜ (bi − p.i − α Cijkl Ti δkl ) ui dΩ ˜ (22. the stress intensity factors for a primary state which includes any combination of surface tractions on the crack surfaces. and p are still in eﬀect.50)
− −
74
Γt
Ω
ˆ Naturally. it is quite clear that Equation 22.
72 73 Therefore. 76 The general stressstrain relationship obtained by substituting the constutive law into the eﬀective stress principle is 0 (22. as are the restrictions on the choice of numerical integration techniques.
ˆ Due to the general deﬁnitions of the applied eﬀective surface traction vector ti and the ˆ eﬀective body force vector bi . body forces. the restrictions on the choice of numerical integration techniques are also still in eﬀect.43.46 to 2 exist. the cases of a primary state with true surface tractions ti on the crack surfaces and true body forces bi are also addressed by Equation 22. and initial strains and stresses are deﬁned by Equation 22. However.51) σij = Cijkl (εkl − ε0 ) + σij kl
Clearly.
77 78 The thermal strains for an isotropic material are deﬁned in terms of the temperature T and the coeﬃcient of thermal expansion α as
ε0 = α T δij ij Victor Saouma
(22. Naturally. in the absence of initial stresses σij = σij and in the absence of both initial strains ¯ and stresses σij = σij .
Combined Thermal Strains and Pore Pressures
71 Recalling that in the absence of initial strains and stresses that the total stresses σij and the eﬀective stresses σij are equivalent. T .69 into Equation 22.8
Field Equations for Thermo.7 Combined Thermal Strains and Pore Pressures
11
through 22.7
22.52) Fracture Mechanics
. bi .and PoroElasticity
75 In thermo or poroelasticity the thermal strains and pore pressures are usually treated as initial strains and initial stresses. The relationship between the stress intensity factors and the reciprocal ˆ work theorem is obtained by substituting the expressions for ti and bi deﬁned by Equation 22.43 also deﬁnes the reciprocal work theorem for solids that are free of initial strains and stresses.65 and the pore pressure p rather than the temperature T must be expressed in a power of r greater than 1 in order for the limits in Equation 22.
In the classical interpretation of the behavior of a porous material (Terzaghi and Peck 1967). respectively.i δkl (22.55)
Adopting the standard form of the eﬀective stress principle the equilibrium equation and natural boundary conditions. the resulting constitutive law for thermoelasticity is σij = Cijkl (εkl − α T δkl ) ¯ The thermal stresses σij are deﬁned as σij = α T Cijkl δkl ¯ and the net strains Eij are deﬁned as ¯ Eij = εkl − α T δkl
80
(22.57)
and may be rewritten in terms of the temperature gradient vector T.54 and the assumption of a homogeneous material.
82
ˆ The applied eﬀective surface traction vector ti is deﬁned as ˆ ˆ ti = ti + σij nj (22. Therefore.61) σij = −p δij
83
(22.51.i bi = bi − α Cijkl T. Substituting this expression for the thermal strains into Equation 22.56)
ˆ where bi is the eﬀective body force vector and ti is the applied eﬀective surface traction vector.j (22.53)
(22.j + bi = 0 ˆ σij nj − ti = 0 (22.60)
where p is the pore pressure deﬁned using the compression positive sign convention.54)
(22. but in the sign convention for standard solid mechanics tension is considered to be positive.Draft
12
79
RECIPROCAL WORK INTEGRALS
where δij is the Kronecker delta. can be rewritten in terms of the eﬀective stresses σij.59)
and may be rewritten in terms of the temperature T ˆ ˆ ti = ti + α T Cijkl ni δkl based on the deﬁnition of the thermal stresses given in Equation 22. and δij is the Kronecker delta. the minus sign corrects the discrepancy in the sign conventions.58)
based on the deﬁnition of the thermal stresses given in Equation 22. the initial stresses corresponding to a pore pressure are deﬁned as 0 (22.
81
The eﬀective body force vector bi is deﬁned as bi = bi − σij. Pore pressures are typically deﬁned using the sign convention for soil mechanics in which compression is positive. Victor Saouma Fracture Mechanics
. the pore pressures p act only in the voids of the material and the eﬀective stresses act only on the skeleton of the material.54.
can be rewritten in terms of the eﬀective stresses
σij.53 and 22.i − α Cijkl T.
When the equilibrium equation and natural boundary conditions are written in terms of σij ˆ the eﬀective body forces bi and applied eﬀective surface tractions ti are deﬁned as
92
bi = bi − p.j + bi = 0 ˆ σij nj − ti = 0 or the net eﬀective stresses σij ¯ σij.Draft
85
22.63)
ˆ where bi is the eﬀective body force vector and ti is the applied eﬀective surface traction vector.51.
When thermal strains and pore pressures are considered in combination the constitutive law is deﬁned as a simple combination of Equations 22.8 Field Equations for Thermo.i is the pore pressure gradient vector.i (22.69) Fracture Mechanics
.j + bi = 0 ¯ ˆ σij nj − ti = 0 ¯ (22. respectively.j + bi = 0 ˆ σij nj − ti = 0
(22.62)
is obtained by substituting the expression for the initial stresses into Equation 22. ˆ ti = −p ni on these surfaces. excess pore pressures resulting from dilatant behavior in the skeleton of the material are not considered.
The stressstrain relationship for poroelasticity σij = Cijkl εkl − p δij (22.and PoroElasticity
13
84 It must be noted that the pore pressures p being considered in this discussion and throughout the remainder of this chapter are the steady state pore pressures.64)
ˆ where p.62
89
σij = Cijkl (εkl − α T δkl ) − p δij
90
(22.
87
The eﬀective body force vector bi is deﬁned as bi = bi − p. The applied eﬀective surface traction vector ti is deﬁned as ˆ ˆ (22. respectively.i δkl ˆ ˆ ti = ti + p ni + α T Cijkl ni δkl Victor Saouma
(22.68) (22.65) ti = ti + p n i
88 Since σij = 0 on surfaces exposed to hydrostatic pressures but no other surface tractions. can be rewritten in terms of either the eﬀective stresses σij σij.
86 Adopting the standard form of the principle of eﬀective stress the equilibrium equation and the natural boundary conditions.66)
The equilibrium equation and natural boundary conditions.67)
91 The ﬁeld equations deﬁned in terms of σij are obtained by adopting the standard form of the principle eﬀective stress and the ﬁeld equations deﬁned in terms of σij are obtained by adopting ¯ the alternate form of the principle eﬀective stress.
Victor Saouma
Fracture Mechanics
.64 and 22. respectively).e. when the equilibrium equation and natural boundary conditions are written in ˆ terms of σij the eﬀective body forces bi and applied eﬀective surface tractions ti are identical ¯ to those for poroelasticity (i.65. Equations 22.Draft
14
RECIPROCAL WORK INTEGRALS
93 However.
a series of three wedgesplitting (WS) test specimens are analyzed.
9
23. 91. A.
To demonstrate the accuracy of the computational algorithm. and Petersson. The crack paths are deﬁned by interface elements that initially act as constraints enforcing the bond between adjacent subdomains. Carpinteri and Valente 1990. Gerstle and Xie 1992). The computed results are compared to the envelopes of the experimental response for each specimen size. Specimen sizes are 31. 8 A load scaling strategy. 1976). The computational algorithm treats the structure as a set of subdomains bonded along assumed crack paths. but change state to function as standard interface elements as the crack propagates. Dahlblom and Ottosen 1990. which allows for load controlled analyses in the postpeak regime. and 5 ft). The magnitude of the cohesive stresses on the crack surface are determined by a softening law that relates the stress to the relative displacement of the crack surfaces through the fracture energy.. Roelfstra and Sadouki 1986. P. Constraints are enforced on the global system of equations using a penalty approach. in Numerical Models in Fracture of Concrete.
7 An incremental formulation for the Fictitious Crack Model (FCM) will be presented. Cervenka and Saouma. Plizzari. is used to enforce stress continuity at the tip of the Fracture Process Zone (FPZ). Gopalaratnam and Ye 1991. and Mod´er. In e the FCM the zone of microcracking and debonding ahead of the crack front is modeled as a cohesive stress that acts to close the crack. M.Draft
Chapter 23
FICTITIOUS CRACK MODEL
Originally published as: Implementation and Validation of a nonlinear fracture model in a 2D/3D ﬁnite element code by Reich. Material properties for the concrete are taken as the mean values of the observed experimental results for all specimen sizes. Wittman Ed. Balkema (1993). Many implementations of the FCM have been reported (Ingraﬀea and Gerstle 1984.E.1
10
Introduction
The most commonly implemented nonlinear fracture model for concrete using the discrete crack approach is the FCM (Hillerborg. Bocca. but none of the implementations based on a discrete
11
. 3. and 152 cm (1.
Draft
2
FICTITIOUS CRACK MODEL
crack approach claim to be based on the standard incremental formulation normally associated with nonlinear analyses. Only the implementation by Dahlbom and Ottosen (Dahlblom and Ottosen 1990). which is based on a smeared crack approach.4a) (23.2)
σdΩ −
Ω
δuT bdΩ −
δuT ˆ = 0 tdΓ
Γt
(23. Surface tractions on the interface are due to bonding of the subdomains tbm or to cohesive stresses in the FPZ tcm .
23. an incrementaliterative solution strategy based on the modiﬁedNewton algorithm that includes load scaling and allows for load control in the postpeak regime will be discussed.5b) Fracture Mechanics
.3)
where δ = Lδu = Lu σ = D (23. The penalty method solution for the mixed system of equations will be discussed. In either case. Deﬁning the volume of the body as t
Ω = Ω1 ∪ Ω2 and the surface of the body subject to prescribed surface tractions as Γt = Γt1 ∪ Γt2 . uses an incremental formulation. the Principle of Virtual Work for the body is δ
Ω T
(23. Ω1 and Ω2 that intersect on a surface ΓI without penetration. stress continuity on ΓI requires that
15
tb2 = −tb1 tc2 = −tc1 Victor Saouma
(23.1
Weak Form of Governing Equations
14 Figure 23. an incremental solution algorithm for the FCM based in the discrete crack approach will be presented and its performance evaluated by comparing the computed response of WS test specimens against known experimental results.4c)
Within each subdomain of the body Ωm the Principle of Virtual Work must also hold. The weak form of the system of mixed equations will be derived from the Principle of Virtual Work.2. but additional integrals are required to account for the work performed by the surface tractions tIm on the interface ΓI . particularly the automatic selection of the penalty number.4b) (23.1)
(23. The weak form equations will then be discretized for solution using the ﬁnite element method.1 shows a body consisting of two subdomains.5a) (23.
23.2
13
Computational Algorithm
Treatment of the structure as a set of bonded subdomains results in a system of mixed equations with the unknowns being displacements and surface tractions on the interface between the subdomains.
12 In this chapter. Each subdomain may be subject to body forces bm or to prescribed surface tractions ˆm on Γtm . Finally.
2 Computational Algorithm
t2
3
t1 ΓI Γt
c 2
Ω ΓI
2
Γt Γu
Γu
2
1
Ω1
b
1
Figure 23.11) Fracture Mechanics
.7b)
ΓI
δuT tI2 dΓ = − 2
ΓIb
δuT tb dΓ − 2
ΓIc
Both tb and tc are unknown. the Principle of Virtual Work for subdomains Ω1 and Ω2 is written as
17
δ
Ω1
T 1 σ 1 dΩ T 2 σ 2 dΩ
− −
Ω1
δuT b1 dΩ − 1 δuT b2 dΩ − 2
Γt1
δuT ˆ1 dΓ − 1t δuT ˆ2 dΓ + 2t
ΓIb ΓIb
δuT λdΓ − 1 δuT λdΓ + 2
ΓIc
δuT tc dΓ = 0 1
(23. u1 ΓIb and u2 ΓIb .9)
δ
Ω2
Ω2
Γt2
ΓIc
δuT tc dΓ = 0 (23.1: Body Consisting of Two Subdomains
16
Deﬁning the interface surface as ΓI = ΓI b ∪ ΓI c .7a) (23. but as tb acts on the bonded.10) 2
On ΓIb the displacements for the two subdomains.4c and including the external work performed by the surface tractions on the interface surface.8) λ = tb Substituting λ into Equations 23.6)
where ΓIb is the bonded interface surface and ΓIc is the interface surface subject to cohesive stresses. or constrained. (23.4a23. Victor Saouma
(23. interface it will be treated as a Lagrange multiplier (23.Draft
23.3 and 23. This condition can be written as a constraint in the strong form
18
u2 ΓIb −u1 ΓIb = 0. must be equal. the external work on the interface is written as
ΓI
δuT tI1 dΓ = 1
ΓIb
δuT tb dΓ + 1
ΓIc
δuT tc dΓ 1 δuT tc dΓ 2
(23.
13c) (23.2.10 and 23.13a) (23.13c23. The following weak form δλT (u2 − u1 )dΓ = 0 (23.13d)
Num and Nλ are standard shape functions in that for each node there is a corresponding shape function whose value is one at that node and zero at all other nodes.
23.5a23.16)
23
Recognizing that Km =
Ωm
BT Dm Bm dΩ m
(23. Each subdomain Ωm is discretized for displacements um such that nodes on Γtm and ΓI are included in the vector of discrete displacements um .2
Discretization of Governing Equations
19 Discretization of Equations 23.Draft
4
Γi
FICTITIOUS CRACK MODEL
but a weak form is required to be compatible with Equation 23.923.4c must be expressed in terms of the discrete displacements and virtual displacements using Equations 23.5b.12 will be presented as if each subdomain were an element (Zienkiewicz and Taylor 1989). To discretize the integral deﬁning the virtual strain energy. The nodes at which the surface tractions due to bonding λ on ΓIb are discretized are at the same locations as those for the displacements.4a23. 20 Displacements um within the subdomains Ωm and the surface tractions λ on the bonded interface ΓIb are deﬁned in terms of their discretized counterparts using shape functions
um = Num um λ = Nλ λ δum = Num δum δλ = Nλ δλ
(23. the stresses and the virtual strains deﬁned in Equation 23.14a) (23.15)
the virtual strain energy can be written as δ
Ωm T m σ m dΩ
= δuT m
Ωm
BT Dm Bm dΩum m
(23.13a23.17) Fracture Mechanics
Victor Saouma
.13d
21
δ
m
= LNum δum
(23.13b) (23. The number of nodes on ΓI in Ω1 is equal to the number of nodes on ΓI in Ω2 . the extension to multielement subdomains is straightforward and will be omitted from this discussion.13b and 23. For each node on ΓI in Ω1 there is a node on ΓI in Ω2 with the same coordinates. (23.12)
was chosen for the constraint equation as it makes the system of mixed equations symmetric.14b)
σm = Dm LNum um
22
Deﬁning the discrete straindisplacement operator Bm as Bm = LNum .
Equation 23.13c23.23) Qm = u
ΓIb
and the load vector for the cohesive stresses as fc m =
Γtc
NTm tc dΓ u
(23.22a) (23.2 Computational Algorithm
5
is the standard stiﬀness matrix for the ﬁnite element method.21)
26 To discretize the external virtual work due to surface tractions on the interface.19a) (23.22b)
δuT tc dΓ = δuT m m
27 Deﬁning the operator matrix for the load vector due to surface tractions on the bonded interface as NTm Nλ dΓ (23. the surface tractions and the virtual displacements must be expressed in terms of the discrete surface tractions and virtual displacements using Equations 23.24)
the external work due to surface tractions on the interface is
ΓIb
δuT λdΓ + m
Γtc
δuT tc dΓ = δuT (Qm λ + fcm ) m m
(23.13d
ΓIb Γtc
δuT λdΓ = δuT m m
ΓIb Γtc
NTm Nλ dΓλ u NTm tc dΓ u
(23.13d
Ωm
δuT bm dΩ = δuT m m δuT ˆm dΓ = δuT mt m
Ωm
NTm bm dΩ u NTm ˆm dΓ u t
(23. the sum of the internal virtual work and the external virtual work is
Ωm
δuT bm dΩ + m
Γtm
δuT ˆm dΓ = δuT fm mt m
(23.16 can be rewritten as δ T σm dΩ = δuT Km um (23.19b)
Γtm
Γtm
25
Recognizing that fm =
Ωm
NTm bm dΩ + u
Γtm
NTm ˆm dΓ u t
(23.Draft
23.20)
is the standard applied load vector for the ﬁnite element method. the displacements and the virtual surface tractions must be expressed in terms of the discrete displacements and the discrete virtual surface
Victor Saouma
Fracture Mechanics
.18) m m
Ωm
24 Discretization of the integrals for the internal virtual work due to body forces and the external virtual work due to prescribed surface tractions simply involves expressing the virtual displacements in terms of the discrete virtual displacements using Equation 23.13c23.25)
28 To discretize the weak constraint equation.13b and 23.13a23.
25 into Equation 23.10 and rearranging terms. leaving m K1 u1 − Q1 λ = f1 + fc1 K2 u2 + Q2 λ = f2 − fc2 (23. leaving QT u2 − QT u1 = 0 2 1 (23.28. it is now possible to deﬁne the discrete system of mixed equations. 23. Substituting Equations 23.13d δλT u1 dΓ = δλ δλT u2 dΓ = δλ
T ΓIb T ΓIb
NT Nu1 dΓu1 λ NT Nu2 dΓu2 λ
(23. and 23.29a23.24.29b)
31
As δuT appears in both sides of Equation 23.29a) (23.2.13b and 23. which can be written in matrix form as
K1 0 −QT 1
0 K2 QT 2
−Q1 u1 f1 + fc1 Q2 u2 = f2 − fc 2 λ 0 0
(23. The discrete system of mixed equations is deﬁned by Equations 23.31.31)
T
as the discrete constraint equation.13c23.30b)
32
In a similar fashion. it can be eliminated.26b)
29
Recognizing that QT = m
ΓIb
NT Num dΓ λ
(23.30b and 23.26a) (23.13a23.Draft
6
ΓIb ΓIb
FICTITIOUS CRACK MODEL
tractions using Equations 23.21. the weak constraint equation can be rewritten as δλT (u2 − u1 )dΓ = δλ (QT u2 − QT u1 ) = 0 2 1
T
(23.27)
is the transpose of the operator matrix for the load vector due to surface tractions on the bonded interface deﬁned in Equation 23.32)
23. the discrete Principle of Virtual Work is written as δuT (K1 u1 − Q1 λ) = δuT (f1 + fc1 ) 1 1 δuT (K2 u2 2 + Q2 λ) = δuT (f2 2 − fc 2 ) (23. δλ can be eliminated from Equation 23.923.30a23.29b.18. Reducing the system of mixed equations to a singleﬁeld equation decreases the number of unknowns that must be solved for and simpliﬁes the use of direct solution methods.
Victor Saouma
Fracture Mechanics
.30a) (23.3
Penalty Method Solution
33 The penalty method (Zienkiewicz and Taylor 1989) was chosen for the solution of the discrete system of mixed equations because it reduces the problem to that of a singleﬁeld.28)
ΓIb
30
Having deﬁned the discretized form of all integrals in the governing equations.
It is now
(23.33.Draft
23. When the eﬀect of K1 and K2 is signiﬁcantly diminished the computed displacements away from the interface.2 Computational Algorithm
7
34 Direct solution methods can be used with the system of mixed equations.
39
Victor Saouma
Fracture Mechanics
.34)
Substituting Equation 23. the total number of unknowns would change as the crack propagates.35)
37 The selection of a good penalty number is a rather diﬃcult task. The goal is to select a penalty number that yields an acceptable error when the computed displacements are inserted in the constraint equation without sacriﬁcing the accuracy of the displacements away from the interface. The author’s experience is that a penalty number selected using
α=
max(diag(Km )) × 106 max(diag(Qm QT )) m
(23. but interlacing of the equations is required to avoid singularities (Wiberg 1974). but the character of the system of equations changes as the eﬀect of K1 and K2 is diminished. 35
To obtain the penalty form of the system of mixed equations. If the penalty number is too small the computed displacements will yield a substantial error when inserted into the constraint equation 0 (23.34 into Equation 23. which are not included in the constraint equation.37)
yields very good results for the class of problems being considered. will lose accuracy due to round oﬀ errors.32 is rewritten as
K1 0 −QT 1
0 K2 QT 2
−Q1 u1 Q2 u2 1 −αI λ
f1 + fc 1
=
f − fc 2 2 0
1 αI
(23.36) Q2 u2 − Q1 u1 =
38 As the penalty number is increased the error approaches zero. a singleﬁeld penalized stiﬀness matrix equation is obtained (K1 + αQ1 QT ) −αQ1 QT 1 2 T −αQ2 Q1 (K2 + αQ2 QT ) 2 u1 u2 = f1 + fc 1 f2 − fc 2 (23. Interlacing a system of mixed equations with an ever changing number of unknowns would certainly create major bookkeeping problems in a ﬁnite element code. Equation 23. Another troublesome aspect related to the use of direct solution methods with the system of mixed equations is that since crack propagation is simulated by the release of constraints on the interface.33)
where α is the penalty number. Penalty numbers selected in this fashion result in computed values of u1 and u2 on the interface that tend to be identical for the ﬁrst ﬁve or six digits when the penalized stiﬀness matrix is assembled in double precision. α should be suﬃciently large that possible to express λ in terms of u1 and u2 λ = α(Q2 u2 − Q1 u1 )
36
is close to zero.
42
Assuming that the applied loads are proportional. At the beginning of the ﬁrst iteration both ∆un and ∆β n are zero. On that portion of the interface where constraints have been released.43)
where ∆β n is the incremental load factor at the beginning of iteration n and δβ n is the correction to the incremental load factor for iteration n. However. The incremental displacements for a generic increment are deﬁned as (23.42)
and ∆un is the incremental displacement vector at the beginning of iteration n and δun is the correction to the incremental displacement vector for iteration n.. At the beginning of each load increment i.40)
The modiﬁedNewton algorithm (Zienkiewicz and Taylor 1991) is used to solve for incremental displacements due to the applied incremental loads. At zero load. some form of automatic load scaling must be included in the solution strategy.Draft
8
FICTITIOUS CRACK MODEL
23. as the magnitude of the applied loads that causes the strength criteria to be satisﬁed exactly is not known a priori. the incremental load factor is deﬁned as ∆β n+1 = ∆β n + δβ n (23.
The use of a strength criteria to detect the onset of crack propagation requires that the magnitude of the applied loads be such that the surface tractions at a node on the constrained interface are precisely equal to the maximum allowable stress. The incremental load factor for increment i is ∆βi and the applied incremental load vector is ∆βi f = ∆βi The load factor at the end of increment i is βi+1 = βi + ∆βi
44
f1 f2
(23. the load factor is βi and the applied load vector is
43
βi f = βi
f1 f2
(23. In this case. equality is required between the normal surface traction and the uniaxial tensile strength.39)
(23.2. As load is applied.e.
45
In a similar fashion.38)
The value of βi is zero at the beginning of the ﬁrst increment. cohesive stresses act until the relative displacements of the unconstrained interface surfaces become large enough to dictate otherwise. surface tractions on the constrained interface violate a strength criteria and the corresponding constraints are released. the entire interface is constrained (i. 41
In this solution strategy crack propagation occurs after every increment.4
IncrementalIterative Solution Strategy
40 An incrementaliterative solution strategy is used to obtain the equilibrium conﬁguration for each crack length. Victor Saouma Fracture Mechanics
. a load factor β can be used to scale an arbitrary set of applied load vector f of some arbitrary magnitude. fully bonded).41) ∆un+1 = ∆un + δun where u= u1 u2 (23.
(K1 + αQ1 QT ) −αQ1 QT 1 2 T −αQ2 Q1 (K2 + αQ2 QT ) 2
n fc = n fc 1 n −fc2
(23.51)
Having shown how the load factor is implemented within the incrementaliterative solution strategy.49)
Since the K−1 f term does not change throughout the course of the iterative process it can α be deﬁned as a constant value for the increment
48
δuT = K−1 f α
(23.
47
Recognizing that
n rn = βf + ∆β n f + fc − pn
(23.50)
49 The displacement vector δuT is commonly called the tangent displacement vector (Crisﬁeld.A. At this point.46)
is the load vector due to cohesive stresses on the interface at the beginning of iteration n. 1981).45)
(23. ∆λ is the incremental n surface traction vector at the beginning of iteration n. M.Draft
46
23.2 Computational Algorithm Displacement corrections are computed by solving
n Kα dun = (βf + ∆β n f + dβ n f + fc − pn )
9
(23. and
nelem
pn =
i=1 Ωei
BT D( + ∆ n )dΩ
(23. the last detail left to explain is the procedure for computing δβ n such that the strength criteria is exactly satisﬁed. the total surface tractions for iteration n must be expressed in terms of its various contributions λ
n+1
= λ + ∆λ + δλr + δβ n δλT
n
n
n
(23.44)
where Kα = is the penalized stiﬀness matrix.48)
is the residual force vector at the beginning of iteration n. δλr is correction to the incremental
Victor Saouma
Fracture Mechanics
.52)
where λ is the surface traction vector at the beginning of the increment. the iterative displacement correction can be deﬁned as
δun = δβ n δuT + K−1 rn α
50
(23. Equation 23.44 can be written in a more compact fashion as dun = K−1 (δβ n f + rn ) α (23. Since the surface tractions on the constrained interface are used to determine the magnitude of the applied load.47)
is the reaction vector for the state of stress at iteration n.
Figure 23. and a bilinear curve for the softening behavior was systematically used with the inﬂection f point deﬁned by s1 = 4t . In those tests. 1.1
55
LoadCMOD
The selected material properties were the average values reported in (Saouma. and 5 ft) with maximum aggregate sizes ranging from 19.75.54)
where (n)i is the normal vector at node i and ft is the uniaxial tensile strength. Fig.H. 38.. and 3) fracture energy GF . 3. 1988). input prameters were intentionally kept to a minimum. ft ft Rokugo. 2) Poisson’s ratio ν. analysis of previously tested wedgesplitting test specimens was undertaken. wedgesplitting specimens of 31. and 3 in) were tested under crack mouth opening displacement control.. numerical predictions are shown in comparison with the range of experimental results. and Simonin. H.55) δβ n = min (δλT )i (n)i
52
n
n
Provided that the cohesive stresses on the interface are treated as forces and no stiﬀness matrix is assembled for those interface elements. u
56 For each analysis. δλr and δλT are deﬁned as = α(QT δun2 − QT δun1 ) 2 r 1 r = α(QT δuT 2 − QT δuT ) 1 (23.53a) (23. F. E. 1997) by practicing engineers. Br¨hwiler u and Boggs 1991) for the three specimen sizes with the 38 mm maximum aggregate size concrete.. K.3
54
Validation
In order to assess the algorithm just presented. the iterative load factor correction is deﬁned as n n ft − (λ)i + (∆λ )i + (δλr )i (n)i (23.3. 23.75 GF . Br¨hwiler. P.
53 The use of stiﬀness matrices for the interface elements subject to softening is avoided because their presence in the global stiﬀness matrix will eventually cause it to become nonpositive deﬁnite. ∆λ . described in (Saouma. this solution strategy allows for load control in the post peak regime.53b)
δλT
51
The strength criteria is applied to λ
n+1
on a nodebynode basis such that
n+1
max((λ
)i (n)i ) = ft
(23. Mihashi.Draft
10 δλr
n
FICTITIOUS CRACK MODEL
surface traction vector due to the residual load vector rn for iteration n. In order to facilitate the use of Merlin (Reich et al. and included: 1) Young’s modulus E. 91. 3) tensile strength ft . and 152 cm (1.. Broz. Br¨hwiler u and Boggs 1991).
23.5. Recognizing that λ. The tensile strength ft was taken as 0.3. and 76 mm (0.2. and w2 = 5 GF as recommended in (Wittmann.9fc . Broz. w1 = 0. and δλT is the surface n traction vector due to the tangent displacement vector δuT .
23.
Victor Saouma
Fracture Mechanics
. and δλr are ﬁxed for iteration n.
Fictitious. Whereas the ﬁrst two are directly obtained from the nonlinear ﬁnite element analysis. Material properties are the same as used for the preceding analyses. which is a scaled up version of the 5 ft specimen. it may be inferred that the size of the process zone tends to remain constant.
23.
Victor Saouma
Fracture Mechanics
. it can be readily observed that very good engineering comparaison is achieved between numerical predictions and experimental results. we note that the eﬀective length approximation appears to be a valid engineering assumption for the larger specimen sizes. and FE Discretization From these ﬁgures. ﬁctitious and eﬀective cracks were determined.4 we observe that as the specimen size increases the relative size of the fracture process zone (diﬀerence between the total crack and the true crack length) decreases as anticipated. Finally. Finally.
57 58 Also shown are the results of an analysis of a hypothetical 50 ft specimen.2: Wedge Splitting Test.3.Draft
23. Following the nonlinear analysis. the third one was obtained by a separate procedure. 60 Observing those three diagrams. the lengths of the real. Figure 23. using the two analyses. the normalized stress intensity factor and the compliance ( CM OD ) P were recorded. a series of linear elastic analyses with increasing crack lengths were performed using a unit load.3 Validation
11
Figure 23.2
Real. From the 5 ft and 50 ft specimens. and Eﬀective Crack Lengths
59 For each of the preceding analysis. the eﬀective crack length was determined as the crack length in the linear elastic analysis which would yield the same compliance as the one obtained from the nonlinear analysis. For each crack length.
Thick. = 1 in.4
0.8
CMOD [mm]
CMOD [mm]
c)
5 foot Specimen
Experimental Envelope Numerical
d)
50 foot Specimen
120
30 Numerical
Splitting Force [KN]
80
Splitting Force [KN]
20
40
10
Thick. = 9 in.
0 0.2
0. = 16 in
20
0 0.2
1.4
0.0
0.2
0. = 16 in.3: Numerical Predictions vs Experimental Results for Wedge Splitting Tests
Victor Saouma
Fracture Mechanics
.0
0.8
1.6
0
0
1
2
3
4
CMOD [mm]
CMOD [mm]
Figure 23.1
0.4
0.Draft
12
a) 1 foot Specimen
Experimental Envelope Numerical 20 80
FICTITIOUS CRACK MODEL
b)
3 foot Specimen
Experimental Envelope Numerical
Splitting Force [KN]
Spliiting Force [KN]
60
10
40
Thick.0
0.5
0 0.3
0.6
0.
Thick.
2
W = 54 in
0.0 0.0 0.2
0.8
Effective Crack Length a_eff/W
Effective Crack Length a_eff/W
c)
5 foot Specimen
True Crack Effective Crack Total Crack
d)
50 foot Specimen
True Crack Effective Crack Total Crack
0.
0.6
Crack Length a/W
0.2
0.6
0.4
0.8
Effective Crack Length a_eff/W
Effective Crack Length a_eff/W
Figure 23.2
0.6
W = 9. and Eﬀective Crack Lengths for Wedge Splitting Tests
Victor Saouma
Fracture Mechanics
.6 0.6
0.0
0.2
0.0 0.2
W = 28 in.
0.8
0.0 0.8
0.
0.75 in.4
0.8
Crack Length a/W
0.0 0.2 0. Fictitious.8 0.6
Crack Length a/W
0.6
0.8
Crack Length a/W
0.6 0.4
0.2 0.4 0.4 0.Draft
a)
23.4
0.0
0.4
0.4: Real.3 Validation
13
1 foot Specimen
True Crack Effective Crack Total Crack
b)
3 foot Specimen
True Crack Effective Crack Total Crack
0.8
0.4
W = 540 in.0 0.
6
c_1 = 0.8
CMOD [mm]
Effective Crack Length a_eff/W
c)
G_F = 260 N/m
d)
G_F = 300 N/m
0. Softening Law: Whereas GF for dam concrete was found not to exceed 300 N/m.75
10
0.5 we observe that despite the range of GF .6
0. Hence.8
Effective Crack Length a_eff/W
Effective Crack Length a_eff/W
Figure 23. some basic parametric studies were undertaken to assess the eﬀect of certain parameters.2
0
1
2
3
4
0.8
0.2 0.3. the loadcmod
30
Spliiting Force [KN]
20
Crack Length a/W
G_F = 240 N/m G_F = 260 N/m G_F = 300 N/m
0. G_F = “large” b) Thisa) wasFracture Energy Effect size selected as it best approximates what may be perceived as240 N/m structures such as dams.2
0.5: Eﬀect of GF on 50 ft Specimen curves.6
0.2 0.2 0.4
0.6
0.4
0. as well as crack lengths are practically identical.2
0.
Eﬀect of GF : First the eﬀect of GF on the response of the 50 ft specimen was investigated.54 to 1.6
Crack Length a/W
True Crack Effective Crack Total Crack
0. this conﬁrms what has long been known that is the relative importance of ft is much greater than that of GF . From Figure 23.4
0.3
Parametric Studies
61 Finally. it was speculated that the large aggregate sizes may result in larger wc or critical crack opening displacement beyond which no tensile stresses can be transmitted across the crack.4
c_2 = 5.8
Crack Length a/W
0.6
0. Hence a series of analyses in which wc was varied from 0.4
0. while w1 and GF were Victor Saouma Fracture Mechanics
.8
True Crack Effective Crack Total Crack
0.4
0.8
True Crack Effective Crack Total Crack
0.0
0 0.Draft
14
FICTITIOUS CRACK MODEL
23.6 mm.
were undertaken.2 0. Furthermore.4
0.6
0.2
0. and a very limited one on the postpeak response.Draft
30
23. We currently have no explanation for these results. the eﬀective crack length becomes smaller than even the true crack.2 0.6.6
0.4
G_F = 240 N/m
0 0.7.8
0. Interestingly we note that the peak load decreased with the increase in wc . there is a slight increase in both total crack and true crack.3 Validation
15
Crit. we note that for large wC values.6
0.6 mm
0. it was found that it also has limited eﬀect on the crack length.1 mm
d)
w_c = 1. However. Similarly.8
Effective Crack Length a_eff/W
Effective Crack Length a_eff/W
Figure 23. These results
Crack Length a/W
w_c = 0.2
0.8 True Crack Effective Crack Total Crack
Spliiting Force [KN]
20
0.2
0
1
2
3
4
0.6: Eﬀect of wc on 50 ft Specimen may be attributed to the substantial drop of s1 (from 4t ) required to maintain a constant GF or area under the softening curve.4
0.54 mm w_c = 1.4
0. which indicates that this parameter has practically no eﬀect on the peak load.6
0.2 0. Crack Opening Effect w_c = 0.60 mm 0.6
10
0.6
Crack Length a/W
True Crack Effective Crack Total Crack
0.
f f
Victor Saouma
Fracture Mechanics
. The results of these analyses are shown in Figure 23.10 mm w_c = 1.54 mm b) kepta) constant.8
CMOD [mm]
Effective Crack Length a_eff/W
c)
w_c = 1.4
0. Eﬀect of s1 : Whereas in the original analysis s1 was taken as 4t . the eﬀect of its variation was investigated through Figure 23. as wc increases.8
Crack Length a/W
0.8
True Crack Effective Crack Total Crack
0.4
0.
8
Crack Length a/W
Effective Crack
0.0 0.0
0.4
0.Draft
16
a)
80 s_1 = f’_t/5 s_1 = f’_t/4 s_1 = f’_t/3 2.5
0 0.6
0.0
G_F = 240 N/m G_F = 240 N/m
20 0.8
Effective Crack Length a_eff/W
Figure 23.8
0.4
0.2
0.6
0.6
0.5
s_1 = f’_t/5 s_1 = f’_t/4 s_1 = f’_t/3
40
1.4
0.5
b)
Softening Bilinear Law
Spliiting Force [KN]
60
Stress [Mpa]
1.4
True Crack
0.2
0.0
0.0
FICTITIOUS CRACK MODEL
3 foot Specimen
2.6
CMOD [mm]
Crack Opening [mm]
c)
3 foot Specimen
Total
s_1 = f’_t/5 s_1 = f’_t/4 s_1 = f’_t/3
Crack
0.7: Eﬀect of s1 on 3 ft Specimen
Victor Saouma
Fracture Mechanics
.2
0.2 0.
5
m Ωm Ω b bm Γtm Γt ˆ t ˆm t u um δu δum L
m
Notation
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = Subdomain index. total nodal Lagrange multiplier vector at
δ δ m D σ σm ΓI ΓI b ΓI c tIm tbm tcm tb tc λ Num um Nλ λ
Victor Saouma
Fracture Mechanics
.4 Conclusions
17
Conclusions
62 In summary.Draft
23.
23. Robustness of the algorithm was. a new framework for linear/nonlinear fracture mechanics analysis of concrete was presented. validated by the numerical duplication of experimental results with standard material properties and no parametric optimization. either 1 or 2 Subdomain m Domain of body Body force vector Body force vector for subdomain m Surface of subdomain m subject to prescribed surface tractions Surface of body subject to prescribed surface tractions Prescribed surface traction vector Prescribed surface traction vector for subdomain m Displacement vector Displacement vector for subdomain m Virtual displacement vector Virtual displacement vector for subdomain m Diﬀerential operator to obtain strain vector from displacement vector Strain vector Strain vector for subdomain m Virtual strain vector Virtual strain vector for subdomain m Elastic stress strain matrix Stress vector Stress vector for subdomain m Interface surface Bonded interface surface Interface surface subject to cohesive stresses Surface tractions on interface surface Surface tractions due to bonding on interface surface of subdomain m Surface tractions due to cohesive stresses on interface surface of subdomain m Surface tractions due to bonding on interface surface Surface tractions due to cohesive stresses on interface surface Lagrange multiplier vector Shape functions for displacements in subdomain m Nodal displacement vector for subdomain m Shape functions for Lagrange multipliers on bonded interface Nodal Lagrange multiplier vector.
This model is currently being extended to include the eﬀect of a pressurized true crack and FPZ and to perform mixed mode analysis and 3D analyses. to some extent. Future analyses will focus on its application for the nonlinear analysis of cracked concrete dams.4
63
23.
Draft
18 Bm Km fm Qm = = = = = fc m I = α = f = i = βi = = ∆βi n = ∆un = = δun ∆un+1 = ∆β n = = δβ n ∆β n+1 = u = = Kα β = n = fc = pn n = r = δuT = δun r ∆λ n δλr
n
FICTITIOUS CRACK MODEL
= =
= δλT n+1 = ∆λ n = = ft
beginning of increment Straindisplacement matrix for subdomain m Stiﬀness matrix for subdomain m Applied load vector for subdomain m Operator matrix for load vector due to Lagrange multipliers on bonded interface surface of subdomain m Load vector due to cohesive stresses on interface surface of subdomain m Identity matrix Penalty number Arbitrary applied load vector Increment number. element number. or node number Total load factor at beginning of increment i Incremental load factor for increment i Iteration number Incremental nodal displacement vector at beginning of iteration n Correction to incremental nodal displacement vector for iteration n Incremental nodal displacement vector at end of iteration n Incremental load factor at beginning of iteration n Correction to incremental load factor for iteration n Incremental load factor at end of iteration n Nodal displacement vector at beginning of increment Penalized stiﬀness matrix Load factor at beginning of iteraction n Load vector due to cohesive stresses on interface at beginning of iteration n Reaction vector at beginning of iteration n Residual load vector at beginning of iteration n Nodal tangent displacement vector for increment Correction to incremental nodal displacement vector due to residual loads for iteration n Incremental nodal Lagrange multiplier vector at beginning of iteration n Correction to incremental nodal Lagrange multiplier vector due to residual loads for iteration n Nodal Lagrange multiplier vector due to tangent displacement for increment Incremental nodal Lagrange multiplier vector at end of iteration n Normal vector Uniaxial tensile strength
Victor Saouma
Fracture Mechanics
.
The more complicated combination of sheartension is investigated using large biaxial tests of concreterock interfaces. Chapter 23.
24. and assumes that there are no sliding displacements nor shear stresses along the process zone. direct shear tests on mortar joints are used to test the model performance in the shearcompression regime. it is indeed correct that a crack is usually initiated in pure mode I (i. The applicability to mixed mode cracking in homogeneous concrete is tested using experiments on modiﬁed Iosipescu’s shear beam and anchor bolt pullout tests. but physically unrealistic. even for mixed mode loading. deﬁnes a relationship between normal crack opening and normal cohesive stresses.Draft
Chapter 24
INTERFACE CRACK MODEL
This chapter was chapter 6 of the PhD thesis of Cervenka
7 This chapter discusses the nonlinear modeling of concrete using a discrete crack fracture mechanics based model. the crack may curve due to stress redistribution or nonproportional loading. However. and extended to account for the inﬂuence of water pressure inside the crack. Several examples are used to validate the applicability of the proposed interface crack model. The model is a generalization of classical Hillerborg’s ﬁctitious crack model. which can be recovered if shear displacements are set to zero. The most popular model simulating this behavior is Hillerborg’s ﬁctitious crack model (FCM) described in Section ?? and Figure ??. it is desirable to incorporate these shear eﬀects into the proposed crack model. opening mode) in concrete. In a previous work. a fracture process zone (Section ??) exists ahead of the crack tip.1. it is well accepted that the weakest part of the structure is the damfoundation interface. 11 Finally for concrete dams. the classical FCM model was implemented by (Reich 1993) for mode I crack propagation. It addresses two important issues: mixed mode fracture in homogeneous materials and interface fracture. Therefore. which is also the location of highest tensile stresses and lowest tensile
. First. during crack propagation. and signiﬁcant sliding displacements develop along the crack as schematically shown in Figure 24.e. A new threedimensional interface crack model is derived.
9 10 The classical FCM model. This assumption is only partially valid for concrete materials. Based on experimental observations.
In concrete materials.1
Introduction
8 The assumption of singular stresses at the crack tip is mathematically correct only within the framework of linear elastic fracture mechanics.
and Prat. as described in Chapter ??.E. Taylor. and Cervenka. 1992). V. T. R.Draft
2
INTERFACE CRACK MODEL
t
σ
τ
u
Figure 24. Baˇant..C..
17
ˇ Experimental data (Saouma V..1: Mixed mode crack propagation. R.
24. Given the scope of this work. The model should be capable of simulating the behavior of rockconcrete and concreteconcrete interfaces.C. Hence. and Chandra Kishen. 14 Interface elements were ﬁrst proposed by (Goodman. Z.C. Baˇant... 1991) and (Carol. I. z 1992). the two major objectives of this chapter are: (1) Modiﬁcation of the FCM model to account for shear eﬀects along both the fracture process zone and the true crack. we can develop a general model which addresses both objectives. I. The presented model is a modiﬁcation of the one ﬁrst proposed by (Carol.E.. within the framework of a discrete crack implementation. P. strength. (2) Development of an interface model based on fracture mechanics to simulate cracking along rockconcrete interfaces..P. it is necessary to address this problem. yet simple enough so that all its parameters can be easily derived from laboratory tests.. Therefore.H. T. R. numerous interface constitutive models have been proposed for a wide range of applications such as rockjoints (Goodman. and then it will be used to simulate cracking both in homogeneous concrete and along a rockconcrete interface.C. Since then. can be visualized as an interface between two identical materials. Taylor. and J. and z Prat.C. P. J.2
16
Interface Crack Model
The objective is to develop a physically sound model. and Brekke. R.
12 13 The FCM model. and Slowik.C..E. 15 In the following section an interface crack model will ﬁrst be proposed. and Brekke.P. 1968) to model nonlinear behavior of rock joints. Fracture Mechanics
Victor Saouma
. 1968) masonry structures (Lotﬁ 1992) and concrete fracture (Stankowski 1990) (Feenstra P. Z.
5 3 1.0 CMOD [mm]
1.M.7 N/m
1. and Chandra Kishen. the objective is to deﬁne relationships between normal and tangential stresses with opening and sliding displacements. V. (6) Under compressive normal stresses neither the shear and nor the normal stiﬀnesses decrease to zero.0
Vertical Load [kN]
24.
18
The major premises upon which the model is developed are: (1) Shear strength depends on the normal stress. Solid Concrete GF = 158. (5) There is a zero normal and shear stiﬀness when the interface is totally destroyed. i. is caused by crack formation.e. Solid Limestone GF = 41. two faces of the interface come to contact.Draft
2.5 2. the rockconcrete contact is idealized as an interface between two dissimilar materials with zero thickness. 1994) J. (7) Irreversible relative displacements are caused by broken segments of the interface material and by friction between the two crack surfaces.0
ˇ Figure 24. LimestoneConcrete GF = 39. and both tangential and normal stiﬀnesses become nonzero. In addition. J.
19 20
Figure 24.0 0. The notation used in the interface model is illustrated in Figure 24.4 illustrates the probable character of the fracturing process along an interface. should a compressive stress be introduced in the normal direction following a full crack opening.2) that the decrease in tensile strength is not abrupt. (9) The dilatancy vanishes with increasing sliding or opening displacements. and Slowik.5
0. (2) Softening is present both in shear and tension. Thus. J. but is rather gradual.
Victor Saouma
Fracture Mechanics
.. along which the energy of the system is gradually dissipated.0
0.2: Wedge splitting tests for diﬀerent materials. (8) Roughness of the interface causes opening displacements (i.2.M. 1994) on rockconcrete interfaces show (Figure 24.0
0. (4) Reduction in strength.5
1. (Saouma V. which depends on the compressive normal stress. (3) There is a residual shear strength due to the friction along the interface. softening.8 N/m 3. and Cervenka.9 N/m 2.E. This is caused by the presence of the fracture process zone.5
2. dilatancy) when subjected to sliding displacements.2 Interface Crack Model
3
2
1
1. In the present model.e.
4: Interface fracture. τ
Material 2
Interface
Interface Model
u z.
τ σ
True Crack
Fracture Process Zone
Intact Interface
Figure 24.
Victor Saouma
Fracture Mechanics
. σ
Material 2
ux . σ
u y. τ 2
u x. τ 1
Figure 24.3: Interface idealization and notations.Draft
4
INTERFACE CRACK MODEL
Material 1 Material 1
u y.
23
F = F (c.2 Interface Crack Model In the proposed model the strength of an interface is described by a failure function:
2 2 2 F = (τ1 + τ2 ) − 2 c tan(φf )(σt − σ) − tan2 (φf )(σ 2 − σt ) = 0
5
(24. Z. The plastic displacements are assumed to be caused by friction between crack surfaces and the fracturing displacements by the formation of microcracks.1)
where: • c is the cohesion. z 1992).. irreversible) displacements up and fracturing displacements uf .Draft
21
24.2) Fracture Mechanics
.C. Baˇant. • σt is the tensile strength of the interface.
22 The shape of the failure function in twodimensional case is shown in Figure 24. I. ui = up + uf 2 2 2 uieﬀ = ui  = (ui + ui + ui )1/2 x y z Victor Saouma
(24.5: Failure function.e.5. and it corresponds to the failure criteria ﬁrst proposed by (Carol.. φf ). • τ1 and τ2 are the two tangential components of the interface traction vector.P. • φf is the angle of friction. and Prat. P.
τ
1 1 tan( φf ) Initial Failure Function
tan( φf )
c σt σ
Final Failure Function
Figure 24. σt = σt (uieﬀ ) u = ue + ui . c = c(uieﬀ ). The inelastic part can subsequently be decomposed into plastic (i. The inelastic displacement vector is obtained by decomposition of the displacement vector u into an elastic part ue and an inelastic part ui . The evolution of the failure function is based on a softening parameter uieﬀ which is the norm of the inelastic displacement vector ui . • σ is the normal traction component. The general threedimensional failure function is obtained by mere rotation around the σaxis. σt .
25 The critical opening and sliding corresponding to zero cohesion and tensile strength are denoted by wσ and wc respectively.P. P. but rather is the energy dissipated during a shear test with high conﬁning normal stress.6: Bilinear softening laws.Draft
6
24
INTERFACE CRACK MODEL
In this work both linear and bilinear relationship are used for c(uieﬀ ) and σt (uieﬀ ). and they are determined from the condition that the area under the linear or bilinear softening law must be equal to GI and GIIa respectively.
σt 0 s1σ
σt
I GF
c c0 s1c wσ uieff w1c GF
IIa
uieff wc
w
1σ
Figure 24.e. wσ w t ∀ uieﬀ > wσ σt (uieﬀ ) = 0 wσ =
2GI −(s1σ +σt0 )w1σ F s1σ
(24. I. wc w ieﬀ ∀ uieﬀ > wc c(u ) = 0 wc =
2GIIa −(s1c +c0 )w1c F s1c uieﬀ wσ ) 2GI F σt0
(24. Z. This parameter was ﬁrst introduced by (Carol.4)
where GI and GIIa are mode I and II fracture energies.C. w1σ are the coordinates F F of the breakpoint in the bilinear softening laws for cohesion and tensile strength respectively.3) bilinear for cohesion
σt (uieﬀ ) = σt0 (1 − σt (uieﬀ ) = 0 wσ =
∀ uieﬀ < wσ ∀ uieﬀ ≥ wσt linear for tensile strength
bilinear for tensile strength
σt (uieﬀ ) = σt0 + uieﬀ s1σ −σt0 ∀ uieﬀ < w1σ w1σ ieﬀ −w 1σ σt (uieﬀ ) = s1σ (1 − u σ −w1σ ) ∀ uieﬀ ∈ w1σ . Baˇant. The F F signiﬁcance of these symbols can be best explained through Figure 24.6. w1c and s1σ . and Prat. c(uieﬀ ) = c0 (1 − c(uieﬀ ) = 0 wc =
uieﬀ wc ) 2GIIa F c0
∀ uieﬀ < wc ∀ uieﬀ ≥ wc linear for cohesion
∀ uieﬀ < w1c c(uieﬀ ) = c0 + uieﬀ s1c −c0 w1c ieﬀ −w 1c c(uieﬀ ) = sc (1 − u c −w1c ) ∀ uieﬀ ∈ w1c . the area under a F τ ux curve). 1992) z in their microplane model. s1c .. It should be noted that GIIa is not the pure mode II fracture energy (i.. This representation seems to be more favorable to the pure mode
26
Victor Saouma
Fracture Mechanics
.
which is a relative measure of the fractured surface. γ can be determined from a pure mode I test through: γ = up ui (24. In the equivalent uniaxial problem the normal inelastic displacement is set equal to uieﬀ . which corresponds to the ﬁnal shape of the failure function in Figure 24. 24. σ = {τ1 . (Figure 24.8)
where up is the residual displacement after unloading and ui is the inelastic displacement before unloading. Then. a GIIa test requires a large F normal conﬁnement.
It is assumed.3 (Alvaredo and Wittman 1992).
32
The stressdisplacement relationship of the interface is expressed as: σ = αE(u − up ) (24. 1 . Alternatively. γ is usually assumed equal to 0.7 into Eq. if GII is used. Furthermore.5 and is given by: 2 2 (24. F F The residual shear strength is obtained from the failure function by setting both c and σt equal to 0.
31 For concrete.6) D = Ao Kno where Kno is the initial normal stiﬀness of the interface.
30
Experimentally. the F whole shearcompression region of the interface model would be an extrapolation from the observed behavior.Draft
27
24.2 Interface Crack Model
7
II fracture energy GII . the secant normal stiﬀness can be determined from:
29
Kns =
σ σt (uieﬀ ) σt (uieﬀ ) = e = u − up u + up + uf − up σt (uieﬀ )/Kno + (1 − γ)uieﬀ
(24.6.10)
where: (a) σ is the vector of tangential and normal stress at the interface.7).7). The determination of GII would require a pure shear test without F F conﬁnement. the evolution of the damage parameter D is deﬁned by formula:
D = 1−
σt (uieﬀ ) σt (u ) + (1 − γ)uieﬀ Kno
ieﬀ
(24. that the damage parameter D can be determined by converting the mixed mode problem into an equivalent uniaxial one (Figure 24. Then.5) τ1 + τ2 = tan2 (φf ) σ 2
28 Stiﬀness degradation is modeled through a damage parameter.11) Fracture Mechanics
. D is related to the secant of the normal stiﬀness Kns in the uniaxial case: Af Kns = 1− (24. and is therefore easier to accomplish. Ao and Af are the total interface area and the fractured area respectively.9)
which is obtained by substituting Equation 24.2 (Dahlblom and Ottosen 1990) or 0. D ∈ 0. which is extremely diﬃcult to perform. σ}T Victor Saouma (24.7)
where γ is the ratio of irreversible inelastic normal displacement to the total value of inelastic displacement. τ2 . whereas the second approach represents an interpolation between the upper bound GIIa and the lower bound GI . Thus.
7: Stiﬀness degradation in the equivalent uniaxial case.
Victor Saouma
Fracture Mechanics
. a linear relationship is assumed: φd (uieﬀ ) = φd0 (1 − φd (uieﬀ ) = 0
uieﬀ udil )
∀uieﬀ ≤ udil ∀uieﬀ > udil
(24.e. The dilatancy of the interface is given by dilatancy angle φd . In compression. (c) E is the elastic stiﬀness matrix of the interface. Kto 0 0 0 E = 0 Kto 0 0 Kno
33
(24. and it is related to the damage parameter D.12)
It should be noted that α can be diﬀerent from 1 only if the normal stress σ is positive (i. the damage parameter D is activated only if the interface is in tension. which implies that no dilatancy is considered in the elastic range.2. that the oﬀdiagonal terms in the elastic stiﬀness matrix E of the interface are all equal to zero. and there is full contact between the two crack surface. the crack is assumed to be closed. (b) α is the integrity parameter deﬁning the relative active area of the interface. The activation of D is controlled through the fraction σ+σ 2σ . In the proposed model.14)
where udil is the critical relative displacement after which. and φd0 is the initial value of the dilatancy angle. which is again assumed to be a function of uieﬀ . the interface does not exhibit the dilatancy eﬀect any more. which is equal to one if σ is positive. The dilatancy is introduced later after the failure limit has been reached through the iterative solution process. α =1− σ + σ D 2σ (24.1
34
Relation to ﬁctitious crack model. the interface is in tension).
It is possible to prove that the proposed interface crack model (ICM) reduces to Hillerborg’s ﬁctitious crack model in the case of zero sliding displacements. and is zero otherwise. In other words.
24.13)
It should be noted.Draft
8
INTERFACE CRACK MODEL
σ
σi K no u K ns
σ G I F u i= u u = γu
p i ieff
u
i
Figure 24.
19) B = −B 1 T .
24. since for zero z sliding displacements.15 becomes a function of the interface opening only:
Kno →∞
lim
σ = σ(ui ) = σ(uz ) = σ(COD) z
(24.13. · · · . Constitutive driver for the computation of internal forces. −B p+1 T .1
36
Interface element formulation. 3. Interface element formulation. the interface stresses develop only along the σaxis in the σ × τ1 × τ2 space (Figure 24. Nonlinear solution algorithm on the structural level. matrix B is equal to: (24.18)
¯i ¯i where u+ and u− denote the element nodal displacements in the local coordinate system of the interface on the upper and lower interface surface respectively.1 (FCM a special case of ICM. The total opening uz of the interface is given by: z uz = σ(ui ) z + ui z Kno
If the limiting case of Kno equal to inﬁnity is considered. Then. then ui becomes equivalent to uz . The implementation of a nonlinear model into a ﬁnite element code consists of three major subtasks: 1.
24. and the stress in the interface is given by: (24.3 Finite Element Implementation
9
PROOF 24. After the tensile strength σt is reached.5).16)
which is precisely the deﬁnition of Hillerborg’s ﬁctitious crack model. · · · . uieﬀ is equivalent to ui . softening starts. Given this deﬁnition. +B p+1 T
Victor Saouma
Fracture Mechanics
. and z the normal stress in Equation 24.3
35
Finite Element Implementation
The ﬁnite element implementation of the interface crack model previously presented will be discussed in this section. 2. +B 1 T .Draft
24.17)
where E is the interface material stiﬀness matrix.) We assume that all shear displacements are zero.3.
Standard interface elements are used in this work.15) σ = σ(ui ) z Normal traction σ is now a function of the normal inelastic displacement ui only. and B is the matrix relating element nodal displacements ue to slidings and openings along the interface:
1/2Nen
u=
i
¯i Ni (¯ + − u− ) = Bue ui
(24. The element stiﬀness matrix is computed using the well known relation: Ke =
Ae
B T EB dA
(24. given by Equation 24.
23) v 1 = ∂ x .8 for several two. Finally.8: Interface element numbering.and threedimensional interface elements. p is the order of the interface element. 37 This deﬁnition of matrix B corresponds to the element numbering shown in Figure 24. η) Bi = 0 0 Ni (ζ.21)
en Subscript i is a node numbering index on one element surface ranging from 1 to N2 . where Nen is the total number of element nodes.
38 The transformation from global to local coordinate system of the interface element is accomplished through the transformation matrix T . (24. η) corresponding to node i. ζ and η are the natural coordinates of the interface element. η) 0 0 0 0 Ni (ζ.22)
The rows of the transformation matrix T are formed by vectors v i deﬁned by following formulas: ∂x ∂x ∂x ∂ζ ∂ζ × ∂η v2 = v3 × v1.Draft
10
INTERFACE CRACK MODEL
where submatrix B i is a diagonal matrix of shape functions Ni (ζ. In threedimensional case it has the form: Ni (ζ. v3 = ∂ x ∂ x  ∂ζ   ∂ζ × ∂η  Victor Saouma Fracture Mechanics
.20)
and in twodimensional case it is given by: Bi = Ni (ζ) 0 0 Ni (ζ) (24. η)
(24.
3 1
4 2 11 5
4 1
5 2
6 6 3 4 1 2 15 16 7 3 9 1 8 10 6 12 4 3 5
12 7 1
8 7 4 6 2
5 6 8 10 2 9 4 3
14 13 5
1
2
11 3
Figure 24. and is equal en to ( N2 − 1). which in general threedimensional case is:
vT 1 T T = v2 vT 3
(24.
26) ∆F = Fn+1 − Fn = 0 Because the failure function is assumed to be satisﬁed for state n.27) Fracture Mechanics
Victor Saouma
. the problem can be stated as follows: For a given stress state σ n . softening parameter uieﬀ and displacement increment ∆un .2
40
Constitutive driver. which is the euclidean norm of the inelastic displacement vector.
v1 =
∂x ∂ζ .25)
These two conditions are equivalent to an incremental form of the consistency condition (Equation ??): (24.
The mathematical theory of plasticity is used in the development of the constitutive driver for the interface crack model. ∂x  ∂ζ 
v 2 = {−v1y . uieﬀ ) = 0 n+1
(24.3 Finite Element Implementation
11
39 The twodimensional case can be recovered from the two preceding formulas by deleting the last row in matrix T and considering the following deﬁnition of vectors v i . On the constitutive level in the sense of ﬁnite element implementation. v1x }
(24.8. the failure criterion must be satisﬁed:
41
Fn (σ n . plasticity theory is used to describe the evolution of the failure function based on the softening parameter uieﬀ .24)
Local coordinate systems deﬁned by these transformations are shown in Figure 24. In n+1 both states n and n + 1.9.
y’ x’ ζ
z’ η ζ
y’
x’
Figure 24.3.
24.Draft
24. uieﬀ ) = 0 n
∧
Fn+1 (σ n+1 . n determine a new stress state σ n+1 and corresponding value of softening parameter uieﬀ . The inelastic displacements are subsequently decomposed according to Equation 24. Thus. 44
The elastic predictor is given by: σ e = σ n + E∆un (24. it is necessary to also ensure the satisfaction of the failure function at state n + 1.
42 43 In this work. plastic and fracturing eﬀects can be separated.9: Local coordinate system of the interface element.
F =0
0
σe σ
0
τ
0
l
F1= 0
l=E m
i
i
1
0
Kn Kt
tan( φd) σ1
l
1
c0 c1 σ1 σ0 t t
Q =0 Q1 = 0
σ
Figure 24. The inelastic corrector returns the trial stress state back to the failure surface: σ n+1 = σ e − ∆λ E m (24. we substitute the expression for the new stress state σ n+1 .28)
where ∆λ is the inelastic multiplier and m is the direction of the inelastic displacements.Draft
12
INTERFACE CRACK MODEL
where σ e are the trial tractions outside the failure surface if a totally elastic behavior is considered. Inelastic multiplier ∆λ is determined from the failure condition at state n + 1. the geometrical interpretation of this condition is the determination of an intersection of a line emanating from point σ e in the direction Em with the moving failure surface (Figure 24. uieﬀ ) = 0 n+1 (24. expands or shrinks depending on the softening introduced through uieﬀ .30)
49 To this equation.10: Algorithm for interface constitutive model. F = 0.10 for a twodimensional case. the failure function is equal to:
2 2 2 2 F = (τ1 n+1 + τ2 n+1 ) − 2 c tan(φf )(σt − σn+1 ) − tan2 (φf )(σn+1 − σt ) = 0
(24. This is schematically shown in Figure 24. 46 The failure surface.10).
47 The increment of the plastic multiplier ∆λ is computed by solving a quadratic equation obtained by considering the particular form of the failure function 24.29.1 in Equation 24. which are equal to (Equation 24. 48
For this case.31) Fracture Mechanics
. Fn+1 (σ e − ∆λ E m.29)
45 In the three dimensional space σ × τ1 × τ2 .28):
τ1n+1 = τ1e − ∆λKto m1 = τ1e − ∆λl1 τ2n+1 = τ2e − ∆λKto m2 = τ2e − ∆λl2 σn+1 = σe − ∆λKno m3 = σe − ∆λl3 Victor Saouma
(24.
we can identify three major steps to the proposed algorithm: 1. when the normal m cannot be constructed. when the return direction m can be determined on the basis of Q. ∆λ > 0 ∧ ∆λ = min(∆λ1 . l2 and l3 are components of vector l indicating the direction of inelastic return in the stress space. m is deﬁned by connecting the trial tractions σ e with the origin of the σ × τ1 × τ2 space (Figure 24.10).35)
For the deﬁnition of m. For this case. l1 .2 = 2A
2 2 2 A = l1 + l2 − µ2 l3 B = 2µ2 σe l3 − 2l1 τ1e − 2l2 τ2e − 2cµl3 2 2 2 C = τ1 e + τ2 e − 2cµσt + µ2 σe + µ2 σt
13
(24.38) (24. we must distinguish between the case. Inelastic corrector simultaneously satisfying the following two equations: Fn+1 (σ e − ∆λ Em.3 Finite Element Implementation The result of this substitution is a quadratic equation with roots: √ −B ± B 2 − 4AC ∆λ1. uieﬀ ) = 0 n+1 uieﬀ = uieﬀ + ∆λ m n+1 n (24.Draft
50
24.34)
The direction of inelastic displacements m is deﬁned as the normal vector to the plastic potential Q (Figure 24. and they are related to the direction of inelastic displacements m through the stiﬀness matrix E. and the pathological case of the apex of Q.
l = Em
53
(24. Elastic predictor: σ e = σ n + E∆un 2.37)
Victor Saouma
Fracture Mechanics
.11):
54
m =
τ1 /Kto
τ /Kto 2 σ/K
no
if
τ  σ
≤
1 Kto tan φd Kno
∧ σ>0 (24. ∆λ2 ) (24. which is deﬁned using the dilatancy angle φd (uieﬀ ) as:
2 2 Q = τ1 + τ1 − (
Kn tan φd )2 σ 2 = 0 Kt
(24.36)
m =
2 2 τ1 + τ2 tan φd
τ1 τ2
otherwise
55
At this stage.33)
52 In the previous equations.32)
where
51
The required solution must satisfy the following conditions.
the inelastic displacements due to friction and microcracks development are separated.39)
56 In the fracturing corrector.2.Draft
14
Case (2)
INTERFACE CRACK MODEL
F =0
τ
σe
l=E m
Case (1)
1
Kn Kt
tan( φd)
c
σe
l=E m
Q =0
σt
σ
Figure 24.11: Deﬁnition of inelastic return direction. Fracturing corrector: E s = αE up = u − E −1 σn+1 s (24.1 and is shown schematically on Figure 24.10. The evolution of damage parameter D is deﬁned by converting the mixed mode problem into an equivalent uniaxial case as described in Section 24.
The complete algorithm of the interface constitutive driver is described in Algorithm 24.9. This separation is controlled by the damage parameter D deﬁned by Equation 24.
57
Victor Saouma
Fracture Mechanics
. 3.
It demonstrates the accuracy of the proposed algorithm as the normal stresses are always correctly evaluated for the given displacements. This is clearly an unstable problem on the structural level. n – Elastic predictor: σn+1 = σ n + E ∆un – Inelastic corrector: ∗ uieﬀ = uieﬀ n n+1 ∗ Do · Evaluate return direction m · Determine dλ such that F (σ n+1 − dλEm. uieﬀ and ∆un n • σ n+1 = σn + αE ∆un • if F (σ n+1 .) • Input: σn . and the second KuhnTucker condition (Equation ??) is not satisﬁed causing the model to fail on the constitutive level. a simple example of two
Victor Saouma
Fracture Mechanics
.
58 59 The robustness of the interface constitutive driver can also be tested by analyzing an example of shear under ﬁxed tension in the normal direction. To investigate the behavior of the proposed interface model in the sheartension regime. In these formulations. a negative value of dλ is obtained. and the specimen fails in an unstable manner.12 illustrates a simple example of direct tension test with the proposed interface crack model. which could occur in some plasticity based models in which the consistency condition is used to determine the increment of the plastic multiplier dλ. However.Draft
24. This is a direct consequence of the geometrical approach during the inelastic corrector phase of the algorithm and the displacement based softening. uieﬀ n+1
15
Figure 24. uieﬀ ) = 0. uieﬀ ) > 0 n – Update σ n . uieﬀ ) = 0 n+1 · uieﬀ = uieﬀ + dλm n+1 n+1 · σ n+1 = σ n+1 − dλEm ∗ While dλ < ε – Fracturing corrector: α = 1 − σ+σ D(uieﬀ ) n+1 2σ E s = αE up = u − E −1 σn+1 s • Output: σ n+1 . This example also illustrates the ability of the proposed model to simulate Hillerborg’s ﬁctitious crack model.1 (ICM constitutive driver. the algorithm should not fail on the constitutive level.3 Finite Element Implementation ALGORITHM 24. since following certain sliding displacement the tensile strength is lower than the originally applied normal stress. This is clearly an undesirable feature since a snapback on the constitutive level could occur even in otherwise stable problems. and there is no loss of accuracy when larger increment steps are used. This implies that on the structural level the solution cannot converge. due to the snapback on the constitutive level. and ∆un such that F (σ n .
Draft
16
Normal Stress σ [psi]
INTERFACE CRACK MODEL
120. In the ﬁrst increment tensile tractions are applied on the upper element.0
22 increments 11 increments 3 increments
60. a ﬁnite analysis including this material formulation involves a system of nonlinear equations. The resulting loaddisplacement curve is shown in Figure 24.
The proposed interface crack model is clearly a nonlinear material formulation.3.005
0. convergence on the structural level could no longer be achieved.02
0. by the NewtonRaphson method.0
0.0
u
1.0 0.0 0. 2. due to the unstable nature of the problem. Such system can be solved. for instance. however no problems on the constitutive level are observed. The incremental tangent stiﬀness matrix for the proposed material formulation can
Victor Saouma
Fracture Mechanics
. quadrilateral elements connected by an interface element is analyzed.
24.
1.00
0.06
0.0
100.08
Figure 24.015 CSD [mm]
0.
60 We observe the tendency for snapback behavior when CSD ≈ 0.010
0.5
0.020
0. To exploit the full NewtonRaphson method a tangent stiﬀness matrix would have to be computed at each iteration.0
40.04 Displacement uy [in]
0.025
0.0
20.000
0. and therefore. and in the subsequent increments horizontal displacements are prescribed for the upper element.5
Shear stress τ [MPa]
σ=const.0
0.0245 mm. After this point.3
61
Nonlinear solver.13.13: Sheartension example.030
Figure 24.12: Inﬂuence of increment size.0
80.
(2) Use of secant interface stiﬀness on the element level while preserving its positiveness and symmetry.A. This would imply the need to store the full stiﬀness matrix on the structural level. 1991).41)
where n is the normal vector to the failure surface passing through the trial stress state σ e (Equation 24. This means that it is not possible to use the initial structural stiﬀness throughout the whole iterative process. it can be expected that the initial stiﬀnesses of the interface elements are very large. the tangent matrix E T becomes negative.3 Finite Element Implementation be computed from the incremental stressdisplacement relationship: ∆σ n = E ∆un − ∆λEm
17
(24.
65
In this work two approaches are suggested to mitigate this problem: (1) Use of secantNewton method to accelerate the convergence on the structural level.44)
where Niter is the number of iterations in the inelastic corrector part of Algorithm 24.e. as it would result in an excessive number of iterations.Draft
24.
On the other hand.1. Following similar arguments leading to equation 24. interface elements). the incremental stressdisplacement is given by a sum:
∆σ n = E ∆un −
Niter i=1
(∆λi Emi )
(24. and therefore. This is clearly not an eﬃcient approach. In addition. and in some cases. the new stress state is computed using the iterative process described in Algorithm 24.
62 From this equation it is possible to derive a formula for an incremental tangent material stiﬀness matrix E T : (24. they represent penalty numbers modeling a rigid contact.43)
63 In this particular case.40)
when multiply the last term by a fraction which is equal to unity: ∆σn = E ∆un − ∆λEm nT E∆un nT E∆un (24. only small portions of the structural stiﬀness matrix will be unsymmetric. Therefore. the incremental tangent stiﬀness is computed by the following expression:
E T = E I −
Niter i=1
Emi nT E i ∆λi T ni E∆un
(24.
66
Victor Saouma
Fracture Mechanics
.27). and a method for solving unsymmetric and nonpositive system of equations would have to be adopted. which is introduced in the softening regime of the interface model.45)
64 During softening. Both methods are supplemented with the linesearch technique of (Crisﬁeld.42) ∆σn = E T ∆un
where: E T = E I − ∆λ
EmnT E nT E∆un
(24.43. since only few elements will be aﬀected by the nonlinear behavior (i. M. the matrix becomes also unsymmetric due to the dilatancy.1.
In this method.
69 In this work.3. it is possible to determine its secant form
Victor Saouma
Fracture Mechanics
.
F
r
(i) Ko
Ks
r
(i1)
r r (i1) (i)
du * (i1)
du * (i)
u
Figure 24.A.Draft
18
INTERFACE CRACK MODEL
24.14. M. W.C.48)
where K is the structural stiﬀness matrix.47)
where dui is the iterative update of the displacement vector computed in iteration i by solving: dui = K −1 r i (24. B and C are given by (Davidon. 1968) rankone quasiNewton update is used. and coeﬃcients A.C. W.47).49)
24.5
Element secant stiﬀness.3. but rather the vector of iterative displacement corrections is updated to satisfy the secant relationship.46)
For onedimensional case. u∗ and u∗ . and r i are residual forces at iteration i. the meaning of this formula is illustrated by Figure 24. The stared symbols. (Davidon.
du∗ du∗ i−1 i = ri ri − ri−1
68
(24.46) on the element level. 1991). 1968): C= u u r u u r r A = 1 − C. and the corrected iterative update of the displacement vector in iteration i is equal to:
du∗ = Adui + Bdu∗ + Cdui−1 i i−1
(24. Considering the diagonal form of the material stiﬀness matrix E. B = −C
(du (d ∗ +d i −d i−1 )T i i−1 ∗ +d T i −d i−1 ) ( i − i−1 ) i−1
u
(24. 24. it is not necessary to recompute the structural stiﬀness matrix at each iteration.4
SecantNewton method.
70 It is also possible to employ the secant formula (Eq.14: Secant relationship. represent the displacement vector updates based on the secantNewton i i−1 corrections (Equation 24.
67 The secantNewton method is described in detail in (Crisﬁeld.
κ ≈
elem Kmax interf Kmin ace
≈
Ehmax 2 (Ehmin )/t ×
10−8
≈ 1014
(24. and it would be unrealistic to have a ratio of largest to smallest element of the order of 104 . we consider an extreme case of K interf ace of the same order as E (i. This ratio should be suﬃcient in most practical problems. For double precision data type. m is usually stored in 52 bits. the interface stiﬀness will be is reduced to 10−8 times its original value.53). Alternatively. The system will become illconditioned when: s ≤ 0
73
(24. maximal and minimal element sizes. t ≈ problem unit).
i Kt1 = i Kt2 = i Kn =
i−1 ∆ui xn+1 −∆ux n+1 i−1 i τ2n+1 −τ2 n+1 i−1 i σn+1 −σn+1 i ∆uz n+1 −∆ui−1 n+1 z i−1 i τ1n+1 −τ1 n+1
19
∆ui n+1 −∆ui−1 n+1 y y
(24.
72
The loss of accuracy due to ﬁnite precision arithmetic is given by: s = p − log(κ) (24. since the interface stiﬀness can be estimated from: E (24. In this work the shear and normal stiﬀnesses cannot be less than 10−8 times their original value. Then after cracking. which corresponds to approximately 16 signiﬁcant digits.m × β e (24.50)
71 To preserve the positiveness of the material stiﬀness matrix a minimal value for shear and normal stiﬀnesses must be speciﬁed.55)
e.53)
where p is the number of signiﬁcant digits in the computer representation of real numbers and s is the accuracy of the solution. This number is based on the assumption that the ratio of the lowest elastic modulus to the largest interface stiﬀness is below 10−4 . However. and it is possible to estimate the condition number of the system using the elastic modulus. This should be adequate for the types of of problems under consideration in this work. it should be kept in mind that this is a worst case scenario. the ratio K interf ace ≈ 10−4 corresponds to the assumption of interface thickness being equal to 10−4 times a unit length of the problem. Thus. as was assumed in Equation 24. the element sizes were assumed to be in the range of the order 10−2 .54)
A real number f is internally represented in a computer memory by three integers m. 102 . Therefore.e.52)
In the formula.51) Kn = t
E where t is the interface thickness.Draft
24.
Victor Saouma
Fracture Mechanics
.52. the accuracy after decomposition is in the worst possible scenario equal to 2 (Equation 24. which is of course an unacceptable level of accuracy.3 Finite Element Implementation from the stress and displacement corrections in each iteration. The mantissa m gives the number of signiﬁcant digits. β and f = .
75 The fundamental principle behind the line search method (Crisﬁeld.
74 Numerical experiments showed. and the iteration process diverges. 1991) is to determine a scaling factor ω. However. after the iterative correction
79
Victor Saouma
Fracture Mechanics
.15.58)
Then an approximation of ω can be evaluated using the following formulas based on the linear interpolation between s(0) and s(1).
Π(ω) = Π(ui−1 + ωdui ) = Π(ω) +
∂Π(ω) ∂u(ω) δω ∂u(ω) ∂ω
(24.59)
A more accurate value of ω can be determined through recursive applications of this formula. 1991). the resulting high residual forces attempt to slide the interface backwards.56)
The functional Π(ω) would be stationary if the last term is equal to zero. M. the residual forces acted along the same direction as the iterative displacement correction. the backward sliding is too large. which may not be in equilibrium with the rest of the ﬁnite element mesh. The excessive sliding in turn introduces large dilatancy eﬀects and high compressive stresses in the normal direction in the subsequent iteration. ω = −s(0) s(1) − s(0) (24.60)
78
Graphically. such that the functional of total potential energy is stationary.
77
s(1) = r(ui−1 + dui )dui
(24.A.56 is equivalent to: ∂Π(ω) ∂u(ω) δω = r(ω)dui δω = 0 ∂u(ω) ∂ω
76
(24. (Crisﬁeld. and their scalar product s(0) was positive. Such ω can be approximately computed from s(ω) for ω equal to zero and one. for the current iterative displacement correction.57)
If we introduce a new symbol s(ω) representing the scalar product of vectors r(ω) and dui . M. that the partial derivative of total potential energy Π(ω) with respect to displacements u(ω) is equal to the vector of residual forces r(u). Thus.3. It can be shown. but since the stiﬀness of the interface is underestimated. Due to this. These high compressive stresses and the frictional properties of the interface combined with the excessive slidings will cause large shear stresses. Originally. s(0) = r(ui−1 )dui . the line search is illustrated in Figure 24. the last term of Equation 24. This problem can be solved by combining the previously discussed secantmethods with line searches.A.
We observe that it corresponds exactly to the divergence problem previously described.6
Line search method. that often the diagonal approximation of the secant stiﬀness underestimates the true stiﬀness of the interface and allows for excessive interface sliding.Draft
20
INTERFACE CRACK MODEL
24. ωi+1 = ωi −s(0) s(ωi ) − s(0) (24. then the objective is to ﬁnd a scalar multiplier ω such that s(ω) is equal to zero.
4. a crack initiation criterion can be based on tensile stresses. what the line search method is able to recognize and correct. where the stress based criteria are not applicable. and let us assume the crack is to be under general mixed mode conditions. In the context of discrete crack analysis. It can be readily veriﬁed that the Griﬃth energy based criterion is also satisﬁed in the nonlinear fracture mechanics through an appropriate softening law.
is considered. The Jintegral is a
Victor Saouma
Fracture Mechanics
.Draft
24.
82 Let us consider a cohesive crack with both normal and tangential tractions in a thin plate subjected to far ﬁeld stresses. The line search method can be implemented in the context of various load control techniques. M.4
81
Mixed Mode Crack Propagation
In most engineering problems. and this is exactly. 83
The Jintegral provides a method to evaluate the energy release rate.A. The criteria for LEFM analysis were discussed in Section ?? of Chapter ??.4 Mixed Mode Crack Propagation
21
s( ω) s(0) ω s(1)
Figure 24. 1991). this is accomplished by modiﬁcations of the initial mesh. necessary to establish appropriate criteria for crack initiation and propagation. To verify if the nonlinear model satisﬁes Griﬃth criterion. as they are inﬁnite at the crack ti.
80
24. and therefore. must be determined during an analysis. The implementation of line searches in the context of the arclength method is discussed in (Crisﬁeld. and energy control is conducted through an appropriate softening diagram. which is based on crack step control mechanism. Figure 24. the crack path is not known a priory.1
Griﬃth criterion and ICM. It is. and can be therefore easily used for nonlinear fracture mechanics analyzes using the FCM model.16. This is to be contrasted with LEFM. it is necessary to compute the energy released by a unit crack propagation. the residuals have opposite orientation with respect to the iterative displacement update dui . (Reich 1993) implemented the line search method with an indirect displacement control technique.15: Line search method. and s(1) is negative. This indicates that the displacements should be smaller. In the nonlinear fracture mechanics analysis.
24. therefore.
Draft
22 J =
Γ
INTERFACE CRACK MODEL
path independent integral and in twodimensional is given by: (W nx − ˆ t ∂u )dΓ ∂x (24.
84 Due to its path independent character it is possible to evaluate the Jintegral along the crack surfaces. but are rather functions of these and the stress state along the interface. It should be noted that in general. dq II (x) dx + dx
∆x
0
σ d∆y = q I (x)
(24.62) t J(Γo ) = − ∂x ∂x Γo ∂x FPZ
85
Applying Leibnitz rule for the diﬀerentiation of deﬁnite integrals the Jintegral is equivalent J(Γo ) =
FPZ
to: d dx
∆x 0
τ d∆x
dx +
FPZ
d dx
∆y 0
σ d∆x
dx
(24. we deﬁne:
∆x
0
τ d∆x = q II (x). the Jintegral is equal to: J(Γo ) =
FPZ
d dx
∆y
wσ
σ d∆y
0
dx =
0
σ(∆y ) d∆y = GI F
(24. it F is possible to consider two special cases for pure mode I and II cracks.61)
σ τ Γ τ σ Γo τ σ
Figure 24.66)
Victor Saouma
Fracture Mechanics
.16: Griﬃth criterion in NLFM.
87
In the case of pure mode I crack. However.64) (24.63)
86 The expressions in parentheses represent the surface energies dissipated in mode I and II at every point along the fracture process zone normalized with respect to crack surface.65)
J(Γo ) =
FPZ
FPZ
dq I (x) dx = GII + GI = Gc c c dx
where GII and GI is the energy dissipated by a unit propagation of the cohesive crack in mode c c II and I respectively. Hence. ∂∆y ∂∆x ˆ ∂u ds = +σ τ dx (24. GII and GI are not equivalent to GII c c F and GI .
in which interface elements are placed along the crack. When they are found to exceed the tensile strength of the material. If the structural system cannot provide these F F energies.2. when the dimensions of the analyzed problem increase. a nonlinear analysis is performed. speciﬁc fracture energies GI and GII are dissipated respectively. A crack is initiated when a maximal principal stress σ1 exceeds the tensile strength of the material. From the updated boundary representation. and from author’s experience. a new mesh is again generated and the problem is reanalyzed from the beginning. The amount of dissipated energy also depends on the loading conditions in FPZ. the crack step size should be: L (24. the Griﬃth criterion for crack propagation is violated. A crack of certain length ∆a is inserted into the boundary representation of the model in the direction perpendicular to the direction of the maximal principal stress. and new crack surfaces are inserted into the boundary representation of the problem. errors are introduced due to discretization errors. In this manner the ﬁnite element model is adaptively modiﬁed until the structure is fully cracked or the prescribed loading level is reached. The exact solution is approached as this length tends to zero.68) ∆a ≤ 10 where L is maximal dimension of the problem.17. The following conclusion can be drawn based on the basis of the previous discussion: (1) It was shown that a unit extension of a cohesive crack model dissipates energy whose amount depends on the softening laws used by the model.
89
24. this is however not feasible. ﬁne mesh would be necessary at the crack tip to model the fracture process zone. (3) In ﬁnite element implementation. and erroneous results will be obtained. the analysis is interrupted. the cohesive crack gives identical results as LEFM.67)
where wσ and wτ is the critical crack opening and sliding respectively for which normal and tangent stresses can no longer be transferred across the crack. In pure mode I and mode II loading.2
Criterion for crack propagation. and the maximal principal stresses at crack tips are monitored. and is shown graphically in Figure 24. This process is described by Algorithm 24.Draft
88
24. In large structures.4. the crack would not propagate. If the FPZ is not modeled adequately.4 Mixed Mode Crack Propagation Similarly. Then. in the case of pure mode II crack. (2) In the limiting case. the Jintegral is equal to: J(Γo ) = d dx
∆x 0 wτ
23
FPZ
τ d∆x
dx =
0
τ (∆x ) d∆x = GII F
(24.
Then.
91
Victor Saouma
Fracture Mechanics
. a new mesh is generated. and the length of the new crack ∆a is arbitrarily selected by the user.
90 In this work a stress based criterion is used for crack initiation and propagation.
INTERFACE CRACK MODEL
ALGORITHM 24.1) Nonlinear ﬁnite element analysis. (5) If maximal principal stress exceed ft .1
Direct shear test of mortar joints.E.M.. 100 and 50 psi).20. 1994). which corresponds to a pure shear test. In these experiments.
96 The interface constitutive model described above is used to analyze the biaxial test specimen in Figure 24. The experiments were performed at the University of Colorado in Boulder (Shing. The remaining interface parameters were obtained from the experimental loaddisplacement curves.2 (Mixed mode crack propagation.
Table 24.5. V. (2) Generate ﬁnite element model. Interface stiﬀnesses Kt and Kn were determined by data ﬁtting in the elastic region.
94 95 Figure 24. Analyses are performed for these conﬁnement values and also for zero value.Draft
24 (1) Input: Boundary representation.19 shows a good correlation between the experimental and analytical loaddisplacement curves. Finite element model.
Victor Saouma
Fracture Mechanics
. Manzouri. These specimens were tested at the University of Colorado by (Saouma V.
24. J. (4.)
(4.1 summarizes the selected material properties. a mortar interface between two brick was tested at three diﬀerent levels of compressive conﬁnement (150.1) Add new crack surfaces of length ∆a to the boundary representation in the direction perpendicular to σ1 . and Slowik. (6) Output: Boundary representation.2
Biaxial interface test. Schuller and Amadei 1994) . a nonlinear analyzes of a direct shear test with various levels of normal conﬁnement are presented. J. (3) Do (3.5.
93 In this section. ˇ and Cervenka.5
Examples and validation
92 Number of problems with known experimental or analytical results are analyzed in this section to validate the proposed interface mode for simulating discrete fracturing along interfaces and in homogeneous concrete.2) Goto Step 2.
24. Atkinson. and Chandra Kishen. (4) While: maximal principal stresses < ft .
24.
25"
u
16.
Victor Saouma
Fracture Mechanics
.
Principal Stresses higher then tensile strength
Figure 24.Draft
24.18: Schematics of the direct shear test setup.
FE Model
Updated Boundary Rep.
σ F
2.86"
Figure 24.17: Mixed mode crack propagation.5 Examples and validation
25
Initial Boundary Rep.25" 2.
Victor Saouma
Fracture Mechanics
.3 0. Shear Displacement [inches] Rel.Draft
26
INTERFACE CRACK MODEL
20
Experiment 31 Analysis
σ = 150 psi
Experiment 28 Analysis
σ = 100 psi
Shear Force [kips]
10
0
10
20 20
Experiment 15 Experiment 25 Analysis
σ = 50 psi
Analysis
σ = 0 psi
Shear Force [kips]
10
0
10
20 0.2 0.1 0.0 0.2 0.3 0.19: Direct shear test on mortar joint.2 0.3 Rel.1 0.2 0.1 0.0 0.1 0.3 0. Shear Displacement [inches]
Figure 24.
2 in Maximal dilatant displacement udil
vertical Load F
simulated hydrostatic load
Crack
50 kN 50 kN
E=23 GPa
E=38 GPa
simulated gravity load
135 kN
Strain gages
135 kN
F/2
Crack Crack opening sliding
F/2
Figure 24.0 deg Friction angle φf 20.
Victor Saouma
Fracture Mechanics
.0045 in 40. Material Parameter Values Brick Modulus of elasticity E 800 ksi Poisson’s ratio ν 0.0 deg Dilatancy angle φD 0.0 lbf/in Speciﬁc mode II fracture energy GIIa F Stress at break point of bilinear softening law s1 12.025 in Crack sliding at break point of bilinear softening law cw1 0.0 psi Cohesion at break point of bilinear softening law c1 0.Draft
24.5 psi Crack opening at break point of bilinear softening law sw1 0.5 Examples and validation
27
Table 24.3 lbf/in Speciﬁc mode I fracture energy GI F 3.2 Mortar Interface Shear stiﬀness Kt 600 ksi/in 200 ksi/in Normal stiﬀness Kn 50 psi Tensile strength ft Cohesion c 80 psi 45.1: Material properties for direct shear test.20: Experimental setup for the large scale mixed mode test.
2: Material properties for direct shear test.55 106 [Psi] Poisson’s ratio ν 0. The material properties are listed in Table 24. however.05 lbf/in F Ratio of irreversible displacement γ 0.36 106 [Psi] Poisson’s ratio ν 0. and hence only the aspects of the analysis directly related to the use of interface crack model are discussed in this section. More analysis and experiments are necessary to determine the reason for this discrepancy.0 deg Friction angle φf 45.3. This is documented on an analysis of the modiﬁed Iosipescu’s beam from Section ??. φf = 30 [deg]. The comparison of the numerical results with experimental observations are shown in Figure 24. GI = 61 [N/m]. the material properties were optimized to yield comparable results with the experimental ones (σt = 0.68 [MPa]. In the prepeak region. The numerical results show a good agreement with the experimental ones in the peak and postpeak region. 102
The model is supported at the two top plates and tractions are applied on the two bottom Fracture Mechanics
Victor Saouma
.15 0.3
Modiﬁed Iosipescu’s beam.887 106 [Psi/in] 4. 101 The label ICM indicates the material properties for the interface crack model.18 Concrete 2 Modulus of elasticity E 3. GII = 400 [N/m]).35 lbf/in Speciﬁc mode I fracture energy GF Speciﬁc mode II fracture energy GIIa 1.5 in Maximal dilatant displacement udil Softening law for tensile strength linear Softening law for cohesion linear
24. the initial stiﬀness of the specimen is overestimated.2. F F
We observe that GI is comparable with the GF measured in wedge splitting tests on concrete.21.Draft
28
98
INTERFACE CRACK MODEL
97 In this ﬁrst study.453 106 [Psi/in] Normal stiﬀness Kn 120 psi Tensile strength ft Cohesion c 220 psi 60. which is used to simulate the fracture process zone. φd = 40 [deg].
99
Table 24. boundary conditions and material properties of the problem were discused in detail in Section ??. Material Parameter Values Concrete 1 Modulus of elasticity E 5.18 Interface Shear stiﬀness Kt 1.
100 The presented interface crack model (ICM) can also be used to model discrete cracks in homogeneous material.5. The geometry.68 [MPa]. c = 0.0 deg Dilatancy angle φD I 0. F limestone and concretelimestone in Figure 24.
0 deg.00
0.
Table 24.30
Figure 24.0 MPa 50.10 0.01 m Maximal dilatant displacement udil Softening law for tensile strength bilinear (Wittmann & Br¨hwiler) u Softening law for cohesion linear
Victor Saouma
Fracture Mechanics
.5 Examples and validation
29
500
400
300 Experiment Analysis Experiment Analysis
200
100
0 0.20 CMSD [mm]
0. Friction angle φf 45.3: Material properties for ICM for Iosipescu’s test.0 deg.22 ICM Shear stiﬀness Kt 15.6 CMOD [mm]
0.3 0.0 N/m Speciﬁc mode I fracture energy GI F Speciﬁc mode II fracture energy GIIa 700.0
0.21: Nonlinear analysis of the mixed mode test.15 Steel plates Young’s modulus E 200 GPa Poisson’s ratio ν 0.4 0. Dilatancy angle φD 70.2 GPa/m 20.8
0.0 GPa/m Normal stiﬀness Kn 2.8 MPa Tensile strength ft Cohesion c 5.2
0.Draft
Vertical load [kN]
24. Material Parameter Values Concrete Young’s modulus E 35 GPa Poisson’s ratio ν 0.0 N/m F Ratio of irreversible displacement γ 0.
Later it curved down. third mesh regeneration) and the ﬁnal mesh when the crack reached the bottom edge.
104 105 In this ﬁgure. two cracks were found to violate the stress criterion: the notch crack and the right bottom crack. the bottom cracks were unloaded. which was determined previously by the linear elastic fracture mechanics analysis was used. This process is repeated until the whole beam is cracked.24 and 24.23 also show the positions on the loaddisplacement curve corresponding to each step and the normal and shear stresses along the crack.3.e.3 degrees with respect to the horizontal axis. and can be compared to experimental data and classical ﬁctitious crack analysis. and reached the bottom edge of the beam exactly at the right corner of the right bottom plate. The magnitude of the tractions is such that the total force on the left bottom steel plate is equal to 1/11 F and on the right bottom plate is 10/11 F. during this analysis. The crack was initiated at the notch at an angle equal to −37. and the crack path was determined automatically during the analysis. the crack path. The ﬁnite element meshes that were used at steps 1. A new ﬁnite element mesh is generated. In the following nonlinear analysis. Therefore. Six diﬀerent ﬁnite element meshes were necessary in this analysis. In the ﬁrst one.Draft
30
INTERFACE CRACK MODEL
plates. the crack patch must be known or assumed. step 4).4 and 5. the mesh corresponding to the third step (i. a second analysis was performed.e. In the subsequent step (i. Then new crack faces are inserted into the boundary representation of the problem in the direction perpendicular to the direction of the maximal principal stress. for the case of ICM the crack path was determined automatically
Victor Saouma
Fracture Mechanics
. tensile principal stresses were observed along the bottom edge between the two bottom supports that were higher than the tensile strength. However.27. and some of them are shown in Figures 24. The applied forces are scaled to satisfy a prescribed value of the relative sliding. due to the stress redistribution.23: the initial one.
The ﬁnal loaddisplacement curves from both analyzes are shown in Figure 24. This displacement control scheme is described in detail in (Reich 1993). Then. The FCM label means that the classical Hillerborg ﬁctitious model. therefore. Figures 24. where F is the total force applied in the vertical direction. The analysis is controlled by the relative sliding at the notch mouth.22 and 24. the label ICM indicates the results from this analysis. when the interface crack model was used.25. after second remeshing). there are no interface elements in the ﬁnite element mesh. Altogether. Single crack analysis: Initially. again new crack faces are inserted into the boundary representation. ﬁve remeshing steps were necessary in this analysis. In this case. This means. They are inserted into the model after the maximal principal stresses at the notch reach the tensile strength.e. and the analysis is restarted from the beginning. a single crack was propagated starting at the notch. that at each increment a unit vertical force F is applied. and only the notch crack propagated. The analysis is stopped when the maximal principal stress at the crack tip is larger then the tensile strength.22 and 24.4 and 5 are illustrated in Figure 24. as it was implemented by (Reich 1993) was used. in which totally ﬁve cracks were considered. therefore four additional cracks were inserted into the boundary representation at each node where the stress criterion was violated.26. tensile stresses larger then tensile stress were observed along the bottom edge of the beam. This ﬁgures show the shaded contour areas of maximal principal stresses on the deformed shapes at steps 3. Multiple cracks analysis: This analysis was performed in an analogous way to the single crack one. and three of them are shown in Figures 24.
103 Two analysis were performed. In other words. however. and a new mesh is generated. but in the second step (i.
10
0.0
10.5 Examples and validation
31
3.05
0.0
σ/τ [MPa]
1.0
20.20
Distance from crack mouth [mm]
CSD [mm]
Figure 24.0
50 40
2.20
Distance from crack mouth [mm]
CSD [mm]
3.0
3.0 0.0
0.15
0.Draft
24.10
0.0 0.00
0.0
F [kN]
σ τ
30 20 10 0 0.0
1.22: Crack propagation in Iosipescu’s beam.05
0.
Victor Saouma
Fracture Mechanics
.0
2.0
1.0
F [kN]
4.0
1. (Steps 1 & 3).0
σ τ
30 20 10 0 0.0
50 40
2.0
σ/τ [MPa]
1.0
30.00
0.0
0.15
0.
0
0.00
1.0
1.10
0. (Increment 11 & 39 in Step 6).15
0.00
1.0
INTERFACE CRACK MODEL
σ/τ [MPa]
σ τ
F [kN]
0 20 40 60 80 100
40 30 20 10 0 0.0
σ/τ [MPa]
σ τ
F [kN]
0 20 40 60 80 100
40 30 20 10 0 0.
Victor Saouma
Fracture Mechanics
.20
Distance from crack mouth [mm]
CSD [mm]
Figure 24.0
0.23: Crack propagation in Iosipescu’s beam.05
0.05
0.0
0.0
1.0 50 2.15
0.Draft
32
3.20
Distance from crack mouth [mm]
CSD [mm]
3.0
0.10
0.0
50
2.
5 Examples and validation
33
Figure 24.Draft
24.
Victor Saouma
Fracture Mechanics
.24: Multiple crack propagation in Iosipescu’s beam (Steps 3.4).
25: Multiple crack propagation in Iosipescu’s beam (Step 5).Draft
34
INTERFACE CRACK MODEL
Figure 24.
Victor Saouma
Fracture Mechanics
.
5 Examples and validation
35
Figure 24.
Victor Saouma
Fracture Mechanics
.4.3.Draft
24.26: Meshes for crack propagation in Iosipescu’s beam (Steps 1.5).
Theoretically. the two loaddisplacement
107
Victor Saouma
Fracture Mechanics
. Clearly. there is almost no diﬀerence between the diﬀerent crack paths. The unsuitability of LEFM for this problem size is clearly demonstrated by the loaddisplacement curves in Figure ??. LEFM and ICM) should become equivalent. ∆a = 20 mm ICM.10 CMSD [mm]
0.29. which shows the load displacement curves for a specimen ﬁfty times larger than the original one. ∆a = 10 mm LEFM. ∆a = 20 mm.0
40. but for FCM the crack path was determined by a previous LEFM analysis (Section ??). ∆a = 10 mm.Draft
36
F [kN]
INTERFACE CRACK MODEL
FCM (crack path from LEFM) ICM (crack path from LEFM) ICM (single crack) ICM (multiple cracks) Experiment (Schlangen 1993)
50. as the problem size increases. ∆a = 30 mm LEFM. Clearly. where the maximal peak load of about 40 kN is overestimated by almost 100 %. which implies that although LEFM is not applicable for this problem size.
106 The crack paths determined by LEFM and ICM analyzes are compared in Figure 24. prediction.15
0. it can still be used for crack pattern
60
40
20
y [mm]
0
LEFM.27: Iosipescu’s beam with ICM model.0 0.e. 4 cracks
20
40
60
0
50
100 x [mm]
150
200
Figure 24.28. during the nonlinear analysis.0
20. the two approaches (i.05
0.20
Figure 24. 3 cracks ICM.0
30.28: Crack paths for Iosipescu’s beam.0
10. This is demonstrated in Figure 24.00
0.0
0.
7
0.). but when the crack approached the support.1
0. Eight remeshing steps were necessary in this analysis. and as for specimen I. but nonlinear fracture mechanics is used. The fracture process zone is modeled using the interface crack model developed in this chapter. and second at the bottom edge. which indicates that the selected crack increment ∆a of 50 mm was too large.2
0.
112 113 Two cracks were considered in this analysis. but below the
Victor Saouma
Fracture Mechanics
. Originally the crack propagated at an angle of about 20 degrees. same problem is reanalyzed. LEFM is not applicable for this problem size as is documented on the resulting loaddisplacement curves (Figure ??).
109
The problem geometry is shown in Figure ??. which shows the crack patterns and shaded areas of maximal principal stresses at remeshing steps 2.31.30. it sharply curved down. and continued at about 55 degrees. and material properties are listed in Table ??. and the crack tip “oscilates” around the correct path.
110 111 This analysis exhibits a “zigzag” crack pattern. it ﬁrst propagated in an almost horizontal direction (1020 deg. This example can also be considered as a proof that the ICM model
4e+04
3e+04
LEFM FCM (crack path from LEFM) ICM
2e+04
1e+04
0e+00 0.5.Draft
F [kN]
24.5
0.29: Large Iosipescu’s beam.
The results for specimen type II are shown in Figure 24. ten diﬀerent ﬁnite element meshes were used. Altogether.4
108
Anchor bolt pullout test. For both geometries the loads were overestimated if LEFM was used. The ﬁrst crack proved to be the dominant one.6
0.3
0.
The ﬁnite element meshes and crack patterns for specimen type I are illustrated in Figure 24. Subsequently a secondary vertical crack developed below the vertical support causing the ﬁnal failure.0
0.8
Figure 24.5 Examples and validation
37
curves are almost identical.9 and 10. In this section. h = 50 x 100 mm.4 CMSD [mm]
0. One started at the top edge of the anchor head.
In Section ?? the anchor bolt pullout experiments were analyzed using the linear elastic fracture mechanics (LEFM).
24. approaches LEFM as the structural size encreases.6.
Draft
38
INTERFACE CRACK MODEL
Figure 24.30: Crack propagation for anchor bolt pull out test I.
Victor Saouma
Fracture Mechanics
.
31: Crack propagation for anchor bolt pull out test II.
The crack paths for both specimens are plotted in Figure 24.32. and continued at the angle of approximately 45 degrees. The experimental curves are adopted from (Shirai 1993) and (Slowik 1993) and the numerical curves correspond to the best results reported by (Shirai 1993).
Figure 24. the crack again sharply turned downward. It should be noted however that the experimental crack patterns in this ﬁgure are only approximate. a secondary vertical crack eventually developed under the support.Draft
24. since no quantitative data about the exact crack patterns are reported in the literature. Again.
Victor Saouma
Fracture Mechanics
.5 Examples and validation
39
support.33 and 24. These ﬁgures show comparison of this analysis with experiments and numerical simulations by other researchers.34 for specimen I and II respectively.
114 115 The loaddisplacement curves are shown in Figure 24. and they show a good agreement with the experimentally observed ones.
20
0.30 0.32: Crack patterns.70
Figure 24.
2.33: Load displacement curve for test I.0
0.5
Experiments Analysis ICM Other Smeared Models Other Discrete Models
0.10
0.50
0.0
1.40 Displacement [mm]
0.5
F/(bdf‘t)
1.00
0.Draft
40
900 350 800 300 700 250 600 200
INTERFACE CRACK MODEL
y [mm]
400 LEFM ICM Experiment
y [mm]
500
150 LEFM ICM Experiment
300
100
200 50 100
0
0
100
200
300
400
500
0
0
25
50
75
100 125 150 175 200
x [mm]
x [mm]
Figure 24.
Victor Saouma
Fracture Mechanics
.60
0.5
2.0 0.
0
Experiments (Nakajima) Analysis.Draft
F/(bdf‘t)
24.34: Load displacement curve for test II. and the capability to capture size eﬀect.0
4. 118 The robustness of the interface constitutive algorithm is ensured by fully exploiting the geometrical meaning of the predictorcorrector algorithm. 119 Due to the generality of the proposed model.
24.0 0. it can be used for nonlinear fracture mechanics of both homogeneous materials and interfaces. an interface crack model was developed and applied to modeling of discrete fracture of concrete. which implies the satisfaction of Griﬃth’s criterion for crack propagation. the model approaches the linear elastic fracture mechanics solution.0
0.30 0. when shear eﬀects must be considered.0
41
6. ICM Other Smeared Models Other Discrete Models
2.40 Displacement [mm]
0.20
0.50
0. that for large structures.6
116
Conclusions
In this chapter.00
0.60
0. 120 To model mixed mode crack propagation.
It was showed.
117 The proposed model is an extension of Hillerborg’s ﬁctitious crack model into a more general case of mixed mode fracture. the model was combined with the automated mesh generation to adaptively modify the ﬁnite element mesh to capture curve crack trajectories. rockconcrete interfaces and masonry joints.70
Figure 24.10
0.
121
Victor Saouma
Fracture Mechanics
.6 Conclusions
8.
Anderson. Atkinson. J.: 1995.: 1992. 637–646. Kluwer Academic Publishers. RILEM Draft Recommendation (50–FCM): Determination of the Fracture Energy of Mortar and Concrete by Means of Three–Point Bend Tests on Notched Beams. 55. Thermomechanical failure criteria for rock media. Numerical Fracture Mechanics. and Stegun. 231– 263. Technical Report ACI 446. S. and Bastero. R. Alvaredo.: 1991. Fundamentals and Applications.: 1986. Fracture Mechanics. pp. PhD thesis. M. On the use of betti’s reciprocal theorem for computing the coeﬃcients of stress singularities in anisotropic media. Anon. Stress analysis in sharp angular notches using auxiliary ﬁelds.: 1974.: 1992. MI. Development of an Anisotropic Singularity Finite Element Utilizing the Hybrid Displacement Method. E399–74. The application of invariant integrals in diﬀusive elastic solids. M. I. and MartinezEsnaola. Apostal. with body forces. Applied Mathematics Series. inertia. models and deformation of material properties. Atluri. A.. Detroit. I. Philosophical Transactions of the Royal Society of London. CRC Press. International Journal of Engineering Science 29(6). Pathindependent integrals in ﬁnite elasticity and inelasticity. J. M. State University of New York. Engineering Fracture Mechanics 31(4).
.: 1979. C. 727–741. Fracture mechanics of concrete: Concepts. Engineering Fracture Mechanics 25(1). Series A 399(1653). and arbitrary crackface conditions. C. S.1R91. Pathindependent integrals in fracture dynamics using auxiliary ﬁelds.. Germany. K. F. Crack formation due to hygral gradients.: 1991. Anon. Handbook of mathematical functions. Bastero. Advani.Draft
Bibliography
446. Atkinson. C.: 1985. and Miranda.: 1991. J. Abramowitz. ASTM American Society of Testing Materials. Technical report.d. No.: 1970. 287–290. and Rooke. A. D.. Third International Workshop on Behaviour of Concrete Elements Under Thermal and Hygral Gradients. Weimar. Materials and Structures 18. J.: n. National Bureau of Standard.: 1988. Aliabadi. Atkinson. Bastero. 341– 364. C. Engineering Fracture Mechanics 16(3). American Concrete Institute.: 1982. Standard Test Method for Plane Strain Fracture Toughness of Metallic Materials. and Wittman. 53–62. Atkinson. T. and Lee. C. 19–25. and Craster.
Babuska.. 55–125. and Nirenburg. Engng.: 1987. Structural Mechanics in Reactor Technology Conference. 10. Mat. Barsoum. Barsoum. 1111–1129. PrenticeHall. Num. Engng. 18. J. Engineering Fracture Mechanics 34(2). 7–11. Bastero. I. Meth. Tech. T. M. Meth.. Aviles. Fracture & Fatigue Control in Structures. S. 25–37. and Scholz.: 1976a.: 1962. Paris. J.. AIAA/ASME/SAE 15th Structures. Use of pathindependent integral a for mixedmode fast crack propagation in uncoupled thermoelastic media. 325–335. Denver. Atkinson.: 1976b. Int. Barsoum. of Engrg. Las Vegas. Int. The postprocessing approach to ﬁnite element method– part ii. Batugin. boxdensity. and Sato. 551–564.: 1977. Barsom. Applications of Fracture Mechanics. J. num. C. 167–169. Barsoum. R. Dimension– a computer program that computes the fractal dimension of lines or points in a plane using the box. Meth. S. and Rolfe. Barsoum. Barenblatt. Barton. Jour. 603–605. The mathematical theory of equilibrium crack in the brittle fracture. H. Int. A. Technical report. Int. boxrotate. Advances in Applied Mechanics 7. second Ed.
Victor Saouma
Fracture Mechanics
. S. Approximate relationship among the elastic constants of anisotropic rocks and anisotropy parameters. R. R. J.: 1974. Iskopaemykh 7(1). J. G. 100. Application of an assumed displacement hybrid ﬁnite element procedure to two dimensional problems in fracture mechanics. FizTekhn. Nevada. Unpublished. J. and Samuel. 331–344. Engrg. and Miller. boxﬂex. Application of quadratic isoparametric ﬁnite elements in linear fracture mechanics.: 1979. J. J.: 1989.: 1989. 20. Engng. 11. R. C. num. A. Engng. num. 85–98. R.: 1975.Draft
2
BIBLIOGRAPHY
Atluri. Letter to the editor. Fract. A degenerete solid element for linear fracture analysis of plate bending and general shells. Combined mode fracture toughness measurement by the disc test. 1420–1422. and perimeterarea methods. Fract. R. J.: 1981. Probl. J. Structural Dynamics and Materials Conference. Kobayachi. 92(B1). 175–182. CO. Barsoum. S. C. United States Geological Survey.: 1978.: 1982. 11. and Nakagaki.: 1984. R. Int. 10. Razrabotki Polezn. An assessment of the quarterpoint elements in pressure vessel fracture analysis. Res. Triangular quarter point elements as elastic and perfectly plastic crack tip elements. On the use of isoparametric ﬁnite elements in linear fracture mechanics. C. Awaji. Schutter. Fractal analysis applied to characteristic segments of the san andreas fault. 10. J. R. Barsoum. and MartinezEsnaola. Further application of quadratic isoparametric ﬁnite elements to linear fracture mechanics of plate bending and general shells.: 1987. Meth. Meth. Geoph. Int. num.. Int. J.: 1972.
: 1974.P. Cardiﬀ.: 1986. Stability of Structures.: 1985.. J.J. P. metal. Brown. J.: 1989. Engng.: 1988.. and Roelfstra. rock. and Plan`s. Fractal dimension of fractured surfaces: a e universal value?. E. 4th Edition. L. of Engineering z Mechanics. ASCE 110(4). on Recent Developments in the Fracture of Concrete and Rock. J. Doctoral Thesis No 739. Size eﬀects in the mixed mode crack propagation: Softening and snapback analysis. Br¨hwiler. Bouchaud. J. ASCE 105. CO. University of Wales. BS 5447. Baˇant. Balkema.P. Lapasset. Proceedings. E.. A.. Z. Size eﬀect in blunt fracture: Concrete. Baˇant. Br¨hwiler.E. Representation of singularities with isoparametric ﬁnite elements. London: 1977. A note on the description of surface roughness using fractal dimension. Carpinteri. Lausanne. Baˇant.
Victor Saouma
Fracture Mechanics
. S. Num. Meth. and Valente. and Cedolin. Fracture model evaluation of dam concrete.. Brown. P. Broz. 297–315. Oxford University Press.: 1990. Z. Fracture Mechanics of Dam Concrete Subjected to QuasiStatic and Seisu mic Loading Conditions. Bocca. E. E. Blunt crack propagation in ﬁnite element analysis. Elementary Engineering Fracture Mechanics. V. Broek. Proceedings of the First International Conference on Fracture Mechanics z for Concrete and Concrete Structures. L.: 1991. Failure of Dam Concrete Subjected to Seismic Loading u Conditions. and Cedolin. F. and Saouma. D.: 1987. Engineering Fracture Mechanics. Int. The Practical Use of Fracture Mechanics. S.: 1988. (in German). 73–79. z Baˇant. Swiss Federal Institute of Technology. Engineering Fracture Mechanics 35(1–3).: 1990. Mazars. Br¨hwiler. Benzley. of z the Engineering Mechanics Division. (eds): 1994. 235–251. and Saouma. Int. Tensile Softening Properties and Structural Response u of a Concrete Member. Conf. in ’Dei Poli Anniversary Volume’ edited by L. Martinus Nijhoﬀ Publishers. E. E. 286. D. 518–535. (ed.. Elsevier. Proceedings of the International Workshop on Dam Fracture and Damage. Europhysics Letters 13(1). 159–170. Z. Z. Br¨hwiler. 8.: 1979. u ASCE. British Standards Institution.: 1989. Cedolin. Breckenridge. Journal of Civil Engineering Materials 4. 335–338.: 1984. S.): 1992. J. V. Kluwer Academic Publishers. 537.: 1990. G. Comment on hillerborg’s comparison of size eﬀect law with ﬁctitious crack z model..Draft
BIBLIOGRAPHY
3
Baˇant. S. Politecnico di Milano pp. Bourdarot. Geophysical Research Letters 14(11).. Correction to: A note on the description of surface roughness using fractal dimension. Broek. Z. 1095–1098. J. and Wittmann. Geophysical Research Letters 15(11).: 1991.E.
Sci. Martinus Nijhoﬀ Publishers. ASCE 108. L.A. 833–848. Carpinteri. C. of Engineering Mechanics. 55–76. W. John Willey & Sons. Variance algorithm for minimisation. Abstr.: 1981. M. 334–335.C.: 1992. Notch sensitivity in fracture testing of aggregative materials. Tensile behavior of concrete. 223–225. J. 55–62.: n.. (eds): 1984.: 1993. 299–304. Engineering Fracture Mechanics 21. Carpinteri. 27(3). Stresscrackopening relation and size eﬀect in highstrength concrete. 201–202. ASCE.A. Fracture Mechanics of Concrete Structures. Chen. and Ottosen.: 1982b.: 1991. Fractal dimension and fracture. G. pp. ASME. O. and Gueguen. I. Engineering Fracture Mechanics 16. Nonlinear Finite Element Analysis of Solids and Structures. 10. A. Chelidze. J. Engineering Fracture Mechanics 44.: 1990. Cottrell. J. Evidence of fractal fracture. Crisﬁeld. 129–143. & Geomech. Fracture Mechanics of Concrete: Material Characterization and Testing. H. Vol 1. Crisﬁeld. and Ingraﬀea. 631–636.P. deLorenzi.: 1987. M. Dei Poli. Journal of Engineering Mechanics 116(1).d. Comninou. Fractal analysis of polystrene fracture surfaces. Mechanical Damage and Crack Growth in Concrete. Journal of Structural Division. A. Rock Mech.: 1985. Nature 338. 35–46. Dahlblom. Z. Proc. A. Y. Min. S.: 1989. Journal of Applied Mechanics pp. CO. A.: 1989.: 1977. Baˇant.G. Carpinteri. Polymer Communications 30..: 1990. N. A. A. Computer J. and Prat. 406–410. Interfacial fracture of sandwich beams. R. Martinus Nijhoﬀ Publishers. Cedolin. H.C. Mechanics of Brittle Fracture. 467–481. I.: 1990. and Prasad. A. and Iori..: 1986. S. 581–590. Computers and Structures 13(13). A fast incremental/iterative solution procedure that handles ‘snap through’. P.
BIBLIOGRAPHY
Carlsson. Cherepanov. and Runt. Duda. Carpinteri. Darmastad Concrete 5. Application of fracture mechanics to concrete structures.
Victor Saouma
Fracture Mechanics
. T. The interface crack. Energy release rate calculations by the ﬁnite element method.: 1982a.: 1968. M. Carol. Microplane type constitutive models for disz tributed damage and localized cracking in concrete structures. Smeared crack analysis using a generalized ﬁctitious crack model.Draft
4 Cahn. S. Elsevier. McGrawHill. Int. Davidon. Breckenridge. L.: 1979.
and J.Draft
BIBLIOGRAPHY
5
Duga. and Rosati. R. G. The natural isoparametric triangle versus collapsed quadrilateral for elastic crack analysis. 768–770. Dugdale.: 1967. Journal of Applied Mechanics 36. Eng.: 1974. J. V.: 1988. N. Engineering Fracture Mechanics 40(6). Div.. Int.J. Studi e Ricerche... The Science of Structures and Materials. Griﬃth.: 1991. i: Models and stability analysis. Trans. and McMillan. (in Italian). Edge . Mech. Gallagher.: 1983. J.. B. G. Goodman.: 1969.: 1991. Plenum Press. Numerical analysis of crack propagation in cyclicloaded structures.bonded dissimilar orthogonal elastic wedges under normal and shear loading. and Sih. 459–464. and Engle. Softening response of plain concrete in direct tension. D. Journal of Basic Engineering 85. 416–434. Phil. R. 100–108. S. Corso di Perfezionamento per le Costruzioni in Cemento Armato 8. Fawkes.C. ii: Applications. A. Behaviour of concrete elements under tension after cracking. Finite Element Analysis Fundamentals. V. T. J. Eshelby.. 650–652.: 1985. Kluwer Academic Press.. and Shah. E. and Xie. C. S. Scientiﬁc American Library. Prospect of Fracture Mechanics Nordhoﬀ pp. Gtoudos. R. Fractals. J. The economic eﬀects of fracture in the united states. 65–82.C. Numerical characterization of the nonlinear fracture process in concrete. A Model for the Mechanics of Jointed Rocks. The phenomena of rupture and ﬂow in solids. A. 12. 163–197.: 1992. A. Victor Saouma Fracture Mechanics
. R.. Calculation of energy release rate. 8. Phys. Eng. Erdogan. W. Rosenﬁeld..H. An assessment of crack tip singularity models for use with isoparametric elements. Soil Mech. Freese.. Mech. National Bureau of Standards..: 1975. R. Fract. 69–84. Numerical study on crack dilatancy.: 1921. 117(4). Gerstle.: 1968. and Luxmoore.: 1963. J. D. 519–527. J. R. W. R. M. J. Taylor. and Brekke. B. J. Fisher. Dunders. J. Buxbam. Yielding of steel sheets containing slits. E. P. S. F..: 1976. Prentice Hall.E. Kearney. V. An Introduction.: 1993. Fracture Mechanics. Buhr. Feenstra P. Englewood Cliﬀs. of Basic Engineering 89. London A221. ASCE. Giuriani. J. Gordon. Owen. On the crack extension in plates under plane loading and transverse shear. E. ACI Journal 82.. d.C. Feder. 143–159. Mech. Foreman. S. 733–769. Gopalaratnam. Soc. P. and Ye. Fract.: 1979. Journal of Engineering Mechanics 118(2). D.: 1960. R. Honton. A. A. 991–1006. Gopalaratnam. Fem modeling of ﬁctitious propagation in concrete. 310–323.: 1988. Roy. Technical Report SP6472.: 1986. Sol. S. and Tracy. 11.
and Stern. R.).: 1989. Meth. T. and Shaw. Strain energy release rate for a crack under combined mode i and mode ii. A. Kinking of a crack out of an interface. Heping. X. Whittman (ed. of Fracture. W. C. He. J. Methods in Eng. Int. P. J.
Victor Saouma
Fracture Mechanics
. and Seward. Hillerborg. Lausanne. The Netherlands. International Conference On Fracture Mechanics of Concrete. The optimum size of quarterpoint crack tip element.: 1976. Journal of Elasticity 8(1). Hussain. 495–509. ASTM.Draft
6
BIBLIOGRAPHY
Harrop. Swartz (eds).: 1987. S. H. W. International Journal of Fracture 11(4). J. Reinhardt. Int. International Journal of Fracture 41. Hellen. Hibbitt. Meth. S. M.: 1985a. G. The calculation of stress intensity factors for combined tensile and shear loading. 14. and Petersson. J. pp.: 1978.: 1977. and Cornelissen. Int. The calculation of stress intensity factor for combined tensile and shear loading. Int. K. A. H. S. and Mod´er.: 1975b. T. 605–617. Int. Engng. Y. 291–296. Some properties of singular isoparametric elements. and Hutchinson. NewYork. Num.: 1982. Conference on Fracture Mechanics of Rock and Concrete. M. J. 605–617. 17. RILEM Materials and Structures 18(106). The fractal eﬀect of irregularity of crack branching on the fracture toughness of brittle materials. T. Henshell. F. 180–184. International Journal of Numerical Methods in Engineering 9(3). and Blackburn. 270–278. 223–249.E.H.: 1975. RILEMSEM Int. Wittmann (editor). W. Shah and S. D. A compendium of sources of fracture toughness and fatigue crack growth data for metallic alloys.: 1975. Part II 20. Hudson.. of Fracture. num. J. S. Fracture mechanics parameters of concrete from uniaxial tensile tests as inﬂuenced by specimen length. Engng.: 1974. Fract. Int. Analysis of One Single Crack. 267–274. J. and Underwood. The computation of stress intensity factors in dissimilar materials. A Compendium of Sources of Fracture Toughness and Fatigue Crack Growth Data for Metallic alloys. num. Pu.: 1989. A. M. 21–34. H. The Theoretical Basis of a Method to Determine the Fracture energy GF of Concrete. STP 560 pp. Analysis of crack formation and e crack growth in concrete by means of fracture mechanics and ﬁnite elements. Fracture Mechanics of Concrete. Hillerborg.: 1978. L. Crack tip elements are unnecessary. A.C. Hudson. Cement and Concrete Research 6(6). The Theoretical Basis of a Method to Determine the Fracture energy GF of Concrete. Hong. in S. C. Elsevier. C. SpringerVerlag. and Seward. Hillerborg. 1101–1103. On the method of virtual crack extension. J. in F. 773–782. 11.: 1975a.: 1983. and Blackburn. 2–28. Vol. Hellen. Hellen. M.: 1985b. Hordijk.: 1982. 11. Amsterdam. K. Hillerborg. R59–R117.. D. Journal of Applied Mechanics 56.
L. J. Hingham. Conf. Fracture. a review. H.: 1962. Meth. G. G. Vienna. Structures and mechanics report. Inglis. 15(10). and Pu. Hutchinson. M.: 1985. Crack propagation and the fracture of concrete. C. Int. C. Kanninen. Correlation between crack tortuosity and fracture toughness in cementitious materials. Int. Irwin.: 1984. Kanninen. Int. J. University of Colorado.: 1981. Irwin. A. A. Ingraﬀea.: 1946. Victor Saouma Fracture Mechanics
. A. Discrete Fracture Propagation in Rocks: Laboratory Tests and Finite Element Analysis. W. Prikadnaia Matematika i Mekhanika. in S. J. To be Published in Engineering Fracture Mechanics. A. Advanced Fracture Mechanics. G.: 1992. Trans. J. Kaplan. Journal of the American Concrete Institute 58(5). 63.. Inst. A.: 1977. 1397–1406. M. Mech. University of Washington. Issa. Jeang. Naval Architects 55. Mech. and Chudnovsky.: 1980. Mixed mode cracking in layered materials. PMM 10. 219–241.). 16.: 1984. L. 1427–1445. WA. Journal of Mechanics and Physics of Solids 16. Swansea.: 1957.Draft
BIBLIOGRAPHY
7
Hussain. J. Ingraﬀea. Transactions ASME. Plastic zone near a crack and fracture toughness. Proceedings of the NATO Advanced Workshop on Application of Fracture Mechanics to Cementitious Composites. and Hawkins. 591–610.: 1968. Department of Civil Engineering. Il’yushin. Appl. p. Engng. on Numerical Methods in Fracture Mechanics. Meth. Mass. 171–209. M. Singular behavior at the end of a tensile crack tip in a powerlaw hardening material. Engng.: 1960. 347–356. N. num. International Journal of Fracture 60. A. C. Irwin. Ishikawa. and Popelar. Proc. PhD thesis. Stress intensity factor computation in three dimension with quarterpoint elements. and Gerstle.. Analysis of stresses and strains near the end of a crack traversing a plate. J.: 1993. Hutchinson. The crack extension force for a part through crack in a plate. 63–191. Shah (ed. and Manu. 97–105. M. 13–31. Quadratic and cubic transition elements. A ﬁnite element analysis of stress intensity factors for combined tensile and shear loading by only a virtual crack extension. Martinus Nijhoﬀ.: 1913. M.: 1985. pp. Nonlinear analysis of concrete fracture. 3rd Int. Hammad. J. The theory of small elasticplastic deformations. Vasilakis. International Conference on Fracture and Damage of Concrete and Rock: 1988. Nonlinear fracture models for discrete crack propagation. Advances in Applied Mechanics 29. Num. J. 7th Sagamore Conf. M. Applications of fracture mechanics to ﬁber composite materials and adhesive joint. and Suo. W. Ingraﬀea. Z. Stresses in a plate due to the presence of cracks and sharp corners. Appl..: 1961. Seattle. Transactions ASME. Oxford University Press. F.: 1980.
second edn.: 1977. PhD thesis. Int. F. E. Lynn. Freeman. and Shih. Y. A. D. C. J. 44. Lekhnitskii.. Li. Journal of Physics. John Wiley and Sons. and B¨cklund. L. San Francisco. and Ingraﬀea. 44. Oxford University Press.
BIBLIOGRAPHY
Karlsson.: 1979. Arch. V.: 1983. J. J. 17–19. Journal of Engineering Mechanics. and Freiman. 187–211. ASCE 112(6). num. Finite element analysis of stress intensity factors for cracks at a bimaterial interface. R.: 1984. W.: 1974. A. J. Passoja. 566–586. Strong Solids.Draft
8 Kelly.: 1972.: 1993.: 1992. B. The strength of adhesive joints using the theory of cracks. Meth. M. and Needleman. S. Rat. E. Fractal character of fracture surfaces of metals. fourth edn. and Paullay. D. Knowles. 721–722. Li.: 1985..: 1986. 405–421. C. S. Moscow (English Translation).
Knott. C. University of Colorado. Jennings. Malyshew. Studies on the fractal dimension of a fracture surface formed by slow stable crack propagation. Butterworths. Levi.H. Advanced Engineering Mathematics. Journal of Fracture 12. B. W. A. L. and Shah. M. and Liang. Lange. Long. 11. E. On a class of conservation laws in linearized and ﬁnite elastostatics. J. Fracture processes in concrete and ﬁber reinforced cementitious composites. Victor Saouma Fracture Mechanics
.: 1981..: 1976. Mir. Lotﬁ. H. Electric Power Research Institute. Int.: 1990. S. and Sternberg. Knowles. German.: 1978. H. and Salganik. D. Journal of the American Ceramic Society 74(12). E. 4 th Edition.: 1981. A. Int. B. Relationship between fracture surface roughness and fracture behavior of cement paste and mortar. 3136–3138. Journal of Fracture 1. Arch. Relationship between fractal geometry and fractography. J. Mandelbrot. and Shih. 114–118. The Fractal Geometry of Nature.: 1991. A comparison of methods for calculating energy release rates. Engng. A. Anal.: 1965. J. Rat. and Stenberg.. Physics Today pp. On a class of conservation laws in linearised and ﬁnite elastostatics. Journal of the American Ceramic Society. 521–531. Theory of Elasticity of an Anisotropic Body. V. International Journal a of Fracture 14(6).: 1976. Boulder. Mech. P. R311–R314. Suqin. and Lung. Lin. Nature 308. Mech. Transition elements to be used with quarterpoint cracktip elements. Mandelbrot. Anal. Kumar. K. Q. and Mar. 187–211. Mecholsky. Z. Kreyszig. B. An Engineering Approach for ElasticPlastic Fracture Analysis. C. Jintegral at loaded crack surfaces. 1031–1036.: 1972. F. Fundamentals of Fracture Mechanics. Finite Element Analysis of Fracture of Concrete and Masonry Structures. EPRINP1931. Are fractures fractal or quakes chaotic. Engineering Fracture Mechanics 21(2).: 1991.
ASTM STP 738. 696–701.: 1981. I. 7. ASTM. Meth. T. Fracture toughness testing of rock. pp. and Erdogan.: 1992. Fractal j integral in fracture. Engng.: 1987. in S. Quantitative analysis of brittle fracture surfaces using fractal geometry. J. and Gallagher. A. and Fawkes. Soc. and Mackin. (2 Vols). num. Hinrichsen.: 1973. VA. Y. Journal of Materials Science Letters 7. V. SveDeFo. Soviet Technical Physics Letters 17(10). Swansea. 661–681. P.: 1967. Paris.: 1982. Amer. Bucci (eds)..: 1986. Sweden. International Society for Rock Mechanics. Int. J. J. Fatigue Crack Growth Measurements and Data Analysis.K. of Basic Engineering.Draft
BIBLIOGRAPHY
9
Mecholsky. e. Mostovoy. G. and R. J. Technical report. 579–581. J.: 1983. R. Physical Review Letters 68(2). J. by the ‘equivalent domain integral’ method. Suggested methods for determining fracture toughness of rock material. A. and Atluri. Owen.: 1963. 205–251.: 1971. U. Fractal analysis of fracture in ocala chert. J. S. Pergamon Press. F. F. J.: 1981. K. PrenticeHall. An emperical stress intensity factor equation for surface cracks. E. S. M. Newton. Engineering Fracture Mechanics 15. Hudak Jr. and Vainshtok. F. National Aeronautics and Space Administration.: 1981. Method of virtual crack growth for determining stress intensity factors ki and kII . G. K. P. Engineering Fracture Mechanics. M˚ aløy.: 1987. Nikishkov. Newman. Swedish Detonic Research Foundation. Pineridge Press. a. N. Use of crackline loaded specimens for measuring plane strain fracture toughness. Technical report. Hampton. 1801–1821. Degeneration of bricktype isoparametric elements. 213–215. Inc. Ceram. Technical Report NASA TN D6376. 185–192. An analysis of several fatigue crack growth rate (fcgr) descriptions. D. Strength of Materials 12. Langley Research Center. Passoja.: 1989. Nikishkov. Murakami. F. An improved method of collocation for the stress analysis of cracked plates with various shaped boundaries.. 698– 700. International Journal of Numerical Methods in Engineering 24(9). Mosolov.: 1991. and FeinbergRingle. Concrete..: 1980. of Materials 2(3). and Raju. Jour. Stockholm. 60–65. Mindess. Calculation of fracture mechanics parameters for an arbitrary threedimensional crack. and Young. Miller. Ouchterlony. ASME. Stress Intensity Factors Handbook. A. Experimental measurements of the roughness of brittle cracks. Philadelphia. and Roux. Newman.: 1988. S. 1145–1147.
Victor Saouma
Fracture Mechanics
. D. Mecholsky. Hansen. 72(1). A critical analysis of crack propagation laws. Commission on Testing Methods. Ouchterlony.
E. Compendium of Stress Intensity Factors. Eﬀect of aggregate and specimen size u on fracture properties of dam concrete. Meth. Raju. Rice. ˇ Reich. An automated ﬁnite element procedure for fatigue crack propagation analyses. On the Marriage of Mixed Finite Element Methods and Fracture Mechanics: An Application to Concrete Dams. Technical report. University of Colorado.: 1986. J.. J. Pu. R.: 1997. V. Engineering Fracture Mechanics 27(2).: 1991. D.: 1971. J. ASCE. P. and Saouma.: 1974. and Sih. P.: 1981.: 1988. and Sikiotis.. London. Victor Saouma Fracture Mechanics
. H.: 1988. and Elices. G. Mixed mode crack propagation in homogeneous anisotropic solids. plasticity. EPF Lausanne. Applied Mechanics pp. Sizeeﬀect in concrete structures: mathematical approximations and experimental validation.edu/~saouma/Merlin. V. Boulder.: 1986. and Schwemmer. Technical report. and linear and nonlinear fracture mechanics. and D. Planas. Journal of Physics A: Mathematical and General 23.. J.: 1993. G. France . Crack growth and development of fracture zones in plain concrete and similar materials.Draft
10
BIBLIOGRAPHY
Parks. Journal of Materials in Civil Engineering 3(3). Technical Report TVBM 1006. and Cartwright. D. J. 321–333. HMSO. Rooke. Plane problems of cracks in dissimilar media. Tech. and Sadouki. Num. 418–423. Cervenka. and Leavell.: 1977. V. E. a threedimensional ﬁnite element program based on a mixediterative solution strategy for problems in elasticity. Broz. Journal of Applied Mechanics 55.: 1984. Fracture/1: Theory and applications. Lund. A. Fracture. EPRI. and Zatz. International Journal of Numerical Methods in Engineering 20. Elastic fracture mechanics concepts for interfacial cracks. J. D. D. Saouma. 204–218. 171–184.colorado. and Boggs. Reprints.U.. Br¨hwiler. R. S. Engineering Fracture Mechanics 25(1). Engineering Fracture Mechanics Journal 20(2).: 1965. Sweden. PhD thesis. Saouma. Cachan (F). Roelfstra. 115–121. Numerical evaluation of the quarter point singular element. 1629–1641. The fractal nature of a fracture surface. A powerful hybrid method in ﬁnite element analysis. V. Peng. J. D. Workshop on ’Strain Localization and Size Eﬀect due to Cracking and Damage. V. M. 98–103. K. I. M.: 1984. 462–476. ASME. pp.: 1990. Engng. ARLCBTR77047.. Int.: 1987. Rep. J. V. Saouma. and Krishna Murty. W. Reich.. T. and Lorensen. I. Int. http://civil. R. Saouma. 3257–3261. Collapsed 12 node triangular elements as crack tip elements for elastic fracture. Palo Alto. Stress intensity factors in anisotropic bodies using singular isoparametric elements. Merlin. H. Hussain. M.. A. Rao.S. A stiﬀness derivative ﬁnite element technique for determination of crack tip stress intensity factors.: 1976. Petersson. Rice. Saouma. Lund Institute of Technology. Ayari. Laboratory for Building Materials (LMC).
Institut f¨r Baustoﬀe. A. (eds): 1984.. and Slowik. Chicago. Fractal Characterization of Cracked Concrete Surfaces. 12. National Conference on Earthquake Engineering. Conference on Fracture Mechanics of Rock and Concrete. M. Engineering Fracture Mechanics Journal. PhD thesis. Evaluation of grout injection techniques for unreinforced masonry structures. Application of Fracture Mechanics to Cementitious Composites. de Lorenzi. Dungar. Electric Power Research Institute (EPRI). Modeling and Computer Analysis. and Cervenka. Int..: 1993.: 1985. M. International Journal of Fracture Mechanics. Fracture. Baˇant (ed. and German.: 1990. (eds): 1989. Proc. T. (eds): 1991. R. Crack extension modelling with singular quadratic isoparametric elements. McGrawHill. Mixed mode fracture of rockconcrete interfaces. W. RILEMSEM Int. Z¨rich. Technical report.: 1994. J. Sluys. A two parameter fracture model for concrete. N. (ed. G. Sih.S. SpringerVerlag. and Chandra Kishen. Czech Republic.: 1986. New York. T. Strain energy factors applied to mixed mode crack problems.: 1956. P. S. S. V. Beitr¨ge zur Experimentellen Bestimmung Bruchmechanisher Materialpaa rameter von Betonen. and Jenq. Shah.: 1994. Vol. J. Energy release rate along a 3d crack front in a thermally stressed body. Shah. Fract. G.. ˇ Saouma V. Int. 647–651. Sih.: 1993. Japan. R. Boulder. Sih. Shih. and Nakamura. 79–102.. Atkinson. Schuller. C. CA.. C.E. and Swartz. Journal of Engineering Mechanics Division.. Slowik. N. and DiTommaso. Shih. G. 3.). Int. USJapan Workshop on Fracture and Size Eﬀect in Concrete. ETH. S. P.. Sokolnikoﬀ. pp. G. in Z. Jci round robin analysis in size eﬀect in concrete structures. B. J. Delft University of Technology. Switzerland. and GamalElDin. J. Wave Propagation. ASCE. Martinus Nijhoﬀ Publishers. Palo Alto. J. Moran. of the Fifth U. u u Research Report No. J. G. Sha. Northwestern University.): 1984.M. S. NATO Workshop. I.: 1976. Proceedings Dam Fracture. and Morris. H. V. P. Y. and Irwin. NY. and Amadei.. Saouma. Localisation and Dispersion of Softening Solids. Manzouri. V. Prague. III. Shing. Fracture..: 1984. On the virtual crack extension technique for stress intensity factors and energy release rate calculation for mixed fracture modes. L. USEurope Workshop on Fracture z and Damage of QuasiBrittle Materials: Experiment. D. Paris. Barton. On cracks in rectilinearly anisotropic bodies.: 1992. Int. Fracture 30. Mathematical Theory of Elasticity.. NewYork.E. Shirai. B. Shah.: 1965. S. 851–860. C. Fracture Mechanics of Concrete: Structural Application and Numerical Calculation.: 1974.Draft
BIBLIOGRAPHY
11
Saouma.
Victor Saouma
Fracture Mechanics
.
Rosenfeld (ed.: 1975. A.: 1974. G. 181–186. Eﬀects of Environment and Complex Load History on Fatigue Life. pp.. Mech. Formation of fractal cracks in a kinetic fracture model. S. and Meakin. The eﬀect of stress ratio during crack propagation and fatigue for 2024t3 and 7075t6 aluminum.: 1992. 91–95. Y.: 1972. P. J. pp. Fractals in fractography. Materials Science and Engineering 80. S. M. Victor Saouma Fracture Mechanics
.: 1939a. Wiberg. Boulder.: 1991. 161–171. Elsevier Applied Science. The calculation of stress intensity factors in anisotropic materials by a contour integral mehtod. Paris. Numerical Simulation of Progressive Failure in Particle Composites.. Journal of Elasticity 9(1). K. 1–15. pp. Timoshenko. Termonia. K. R. International Journal of Numerical Methods in Engineering 8(2). Wheeler. J. and Goodier. J. Recent Advances in Engineering Science. Nature 320. 359–368. A.: 1973. J. Tsai. O. Fractal dimension of cocitation. 1240. Bearing pressures and cracks. British Welding Research Ass. New York. Hellertown.E.: 1976.: 1990. Stern. P. Bearing pressures and cracks. ASME. E.Draft
12
BIBLIOGRAPHY
Stankowski. Journal of Material Research 6(6).: 1973. and M. Boston. Swartz. Stern. The numerical calculation of thermally induced stress intensity factors.: 1970. and Dunham. Computational Fracture Mechanics. Spectrum loading and crack growth. and Kan. Y. E.: 1986.L. H.. A boundary integral representation for stress intensity factors. MA. Volume 7. Terzaghi. S. PhD thesis. NY. M.: 1939b. Stern. Fracture Mechanics of Concrete Structures. N. New York. T.: 1979. A contour integral computation of mixed mode stress intensity factors. E. Appl. Matrix structural analysis with mixed variables. and Mecholsky. 2nd edition. Tada. R. 429–431. 626. The Stress Analysis of Cracks Handbook. Report M13. and Peck. Theory of Elasticity. Y. Westergaard. Appl. Del Research Corporation.: 1970. 125–132. The inﬂuence of aggregate/paste bonding and strength on mode i fracture mechanics properties of concrete. John Wiley & Sons. H. J. 167–194. of Basic Engineering 94. Underwood. ASTM STP 462. Becker. H. Mech.. Wells.: 1990. Westergaard. 437–442. Application of Fracture Mechanics at and Beyond General Yielding. Scientiﬁc Publishers. NY. and Banerji. M.). University of Colorado. in M. PA. and Irwin. International Journal of Fracture 12(3).: 1967. (invited review). M. Soil Mechanics in Engineering Practice. Stern. Van Raan. McGraw Hill. Walker. Nature 347.: 1963. Fractal fracture of single crystal silicon.: 1986.
R. J. F. Reconsiderations on singularity or crack tip elements. Wittmann. Int. (ed. of Applied Mechanics. Fracture Mechanics of Concrete. 291–295. The analysis of fracture proﬁles of soil using fractal geometry. M. Mihashi. and Simonin. Technical report. Wittmann. Zienkiewicz. 14. Wittmann. McGrawHill. ASME.: 1959. The Finite Element Method. of Applied Mechanics. H. Engle. Zienkiewicz. Australian Journal of Soil Research 30. M. F.: 1979. Basic Formulation and Linear Problems. The Finite Element Method. Zienkiewicz. Balkema. Proceedings of the ﬁrst Bolomey Workshop. Numerical Models in Fracture Mechanics of Concrete..: 1977.: 1971.: 1989.Draft
BIBLIOGRAPHY
13
Willenborg.: 1952. 21–32. C. and Wood.. J. Materials and Structures 21. London.): 1983.
Victor Saouma
Fracture Mechanics
. Ying.. 199–204. Jour. and Taylor. 4th ed. Bulletin of the Seismological Society of America 49(2).: 1985.: 1957. Y.. Vol. A.: 1992.. Meth. Cement and Concrete Research 15. Williams. O.: 1991. and Crawford. The Finite Element Method. L.: 1982. McGrawHill. Jour.. M. J. On the stress distribution at the base of a stationary crack. Solid and Fluid Mechanics. 1. Zak. 109–114. Rokugo. Dynamics and Nonlinearity. H. and Taylor. A crack growth retardation model using an eﬀective stress concept. Y. Engng. L. F. A note on the singularity and the strain energy of singular elements. E. Fracture Energy u and Strain Softening of Concrete as Determined by Means of Compact Tension Specimens. and Nishiguchi. num.): 1993. Air Force Flight Dynamics Laboratory Report. London. 526–528. D. 1992. P. num. ASME 19(4). K. Proceedings of a Conference in Lausanne.: 1963. London. C. The stresses around a fault or crack in dissimilar media. 2. Vol. Int.: 1988. R. R. 817–824. Elsevier Science Publishers Company Inc.. O. M. L. 18. 4th ed. of Applied Mechanics. Zurich. I. Williams. Br¨hwiler. Winslow. McGrawHill. Crack point stress singularities at a bimaterial interface. Yamada.. AFFDLTM711FBR. R. and Williams. Williams. 1525–1544. Engng. Meth. J. ASME 24(1). (ed. Young. Ezawa. The fractal nature of the surface of cement paste. I. 31–39. Stress singularities from various boundary conditions in angular corners of plates in extension. third edn. O. L. Jour.