application of finite element and constitutive models

116
1 NOTES FOR SHORT COURSE Application of Finite Element and Constitutive Models SOLID, STRUCTURE AND SOIL-STRUCTURE INTERACTION: STATIC, DYNAMIC, CREEP THERMAL ANALYSES By Chandrakant S. Desai 2012 Tucson, AZ, USA

Upload: hakhue

Post on 05-Jan-2017

228 views

Category:

Documents


3 download

TRANSCRIPT

1

NOTES FOR SHORT COURSE

Application of Finite Element and Constitutive Models

SOLID, STRUCTURE AND

SOIL-STRUCTURE INTERACTION:

STATIC, DYNAMIC, CREEP

THERMAL ANALYSES

By

Chandrakant S. Desai

2012

Tucson, AZ, USA

2

PREFACE

These notes present descriptions of static and dynamic finite element method, nonlinear

techniques used, various constitutive models (elastic, plastic, creep, thermal, and disturbance-

softening , procedures for determination of parameters for the constitutive models, parameters

for typical materials and interfaces, and program features for the DSC-SST2D code.

The DSC-SST2D based on the finite element method with the DSC model is considered

to be a general purpose finite element code for analysis of a wide range of problems involving

solids and interfaces or joints, subjected to thermomechanical static, cyclic (repetitive) and

dynamic loadings. The code permits a range of constitutive models for elastic, plastic, and creep

responses, microcracking leading to fracture, and fatigue and softening. As a result, the code can

be used for solutions in civil and geotechnical, mechanical and aerospace engineering,

engineering mechanics, and electronic packaging systems.

Although these notes mainly cover static problems, other codes are available for dynamic

two-dimensional analysis (DSC-DYN2D) and for dynamic three-dimensional analysis (DSC-

SST3D). Their brief descriptions are given below:

I. DSC-SST2D: Two-dimensional Computer code for Static, Dynamic, Creep and Thermal

analysis-Solid, Structures, and Soil-Structure Problems

1. Part I: Manual for Technical Background. The Notes for the Short Course herein have

been adopted from this manual.

2. Part II: User’s Guide

3. Part III: Examples Problems-Verifications and Applications

II. DSC-DYN2D: Two-Dimensional code for Dynamic and Static Analysis-Dry and

Saturated (Porous) Materials including Liquefaction

1. Part I: Manual for Technical Background

2. Part II: User’s Guide

3. Part III: Examples Problems-Verifications and Applications

III. DSC-SST3D: Three-Dimensional Computer code for Static and Coupled Consolidation

and Dynamic Analysis-Solid (Porous), Structures and Soil-Structure Problems:

1. Part I: Manual for Technical Background

2. Part II: User’s Guide

3. Part III: Examples Problems-Verifications and Applications

This manual (Part I) presents the descriptions of the DSC-SST2D code. The other two are

available in separate reports.

3

TABLE OF CONTENTS

TOPIC Page

Preface .......................................................................................................................................................... 2

Table of Contents …………………………………………………………………………..,,………… 3

Introduction ................................................................................................................................................. 6

Finite Element Method .............................................................................................................................. 7

Computational Algorithm ............................................................................................................... 8

Element Library ............................................................................................................................ 10

Constitutive Models ................................................................................................................................. 14

Nonlinear Analysis ........................................................................................................................ 16

Drift Correction ......................................................................................................................... 17

Continuous Hardening and HISS Models ................................................................................. 17

Program Features .................................................................................................................................... 19

Applied Forces ......................................................................................................................... 19

Initial or in situ Stresses ........................................................................................................... 20

Simulation of Sequences .............................................................................................................. 21

Addition of Material, or Placement or Embankment ................................................................ 21

Removal of Material or Excavation .......................................................................................... 24

Removal of Liquid (Water) or Dewatering ............................................................................... 24

Support Systems........................................................................................................................ 26

Mesh Change Option ................................................................................................................ 28

Boundary Conditions ................................................................................................................ 28

Dynamic Analysis ..................................................................................................................................... 28

Newmark Method ..................................................................................................................... 30

Wilson -Method ...................................................................................................................... 30

Mass Matrix ................................................................................................................................... 31

Absorbing Boundaries ................................................................................................................... 31

Cyclic or Repetitive Loading ......................................................................................................... 31

Creep Behavior .............................................................................................................................. 32

Material Parameters ................................................................................................................................ 32

Organization of Computer Program ...................................................................................................... 32

4

Appendix I: Constitutive Models .......................................................................................................... 33

Linear and Nonlinear Elastic Models ............................................................................................ 33

Linear Elastic Model……………………………………………………………………….33

Nonlinear Elastic Models ........................................................................................................ 33

Plasticity Models ..................................................................................................................... 34

Von Mises .................................................................................................................. 35

Mohr-Coulomb ........................................................................................................... 35

Drucker Prager ........................................................................................................... 35

Modified Cam-Clay ................................................................................................... 35

Cap Model .................................................................................................................. 37

Hoek-Brown Model ................................................................................................... 39

Hierarchical Single Surface (HISS) Models .............................................................. 39

Initial Values of and ...................................................................................................... 41

Interface/Joints Element………………………………………………………………….43

Cohesive and Tensile Strengths ........................................................................................... 44

Creep Models………………………………………………………………………………44

Viscoelasticplatic (vep) or Perzyna Model ............................................................................. 46

Multicomponent DSC or Overlay Models .............................................................................. 46

Specializations of Overlay Model ........................................................................................... 50

Number of Overlays and Thicknesses ............................................................................... 51

Layered Systems with Different Material Properties .............................................................. 51

Disturbance (Disturbed State Concept – DSC) Model: Microcracking,

Degradation and Softening .................................................................................................... 53

Speciaqlizations……………………………………………………………………………55

Thermal or Initial Strains ........................................................................................................................ 55

Elastic Behavior ...................................................................................................................... 55

Plane Stress ....................................................................................................................... 56

Plain Strain ........................................................................................................................ 56

Axisymmetric .................................................................................................................... 56

Thermoplastic Behavior .......................................................................................................... 57

Thermoviscoplastic Behavior .................................................................................................. 58

DSC Model .............................................................................................................................. 61

Cyclic or Repetitive Loading .................................................................................................................... 61

Unloading ............................................................................................................................... 63

Reloading ................................................................................................................................ 66

Cyclic Hardening ..................................................................................................................... 69

Appendix II: Elasto-plastic Equations .................................................................................................. 72

Appendix III: Drift Correction and DSC Computer Algorithm ........................................................ 74

DSC Computer Algorithm ....................................................................................................... 75

5

Appendix IV: Determination of Constants for Various Models ......................................................... 77

Elastic Constants ......................................................................................................................................... 77

Plasticity Constants ..................................................................................................................................... 79

Ultimate: , ........................................................................................................................... 79

Phase Change ........................................................................................................................... 81

Hardening ................................................................................................................................ 84

Nonassociative......................................................................................................................... 84

Cohesive and Tensile Strengths ........................................................................................ 86

Computer Code to Find Constants for 0- and 1-Models ..................................................................... 87

Viscoplastic and Creep Models, 0 + vp .................................................................................... 88

Mechanics of Viscoplastic Solution ........................................................................................ 88

Elastoviscoplastic: Overlay Models ........................................................................................ 92

Disturbance Model .................................................................................................................. 93

Cyclic Loading and Liquefaction............................................................................................................. 96

Cyclic or Repetitive Loadings, Unloading and Reloading ...................................................... 96

Initial Conditions ..................................................................................................................................... 98

Environmental Effects .............................................................................................................................. 98

Interface/Joint Behavior ........................................................................................................................... 98

Material Constants .................................................................................................................................... 99

Implementation and Applications ........................................................................................................... 99

Material Constants for Typical Materials: Soils, Rock, Concrete, Solders ................................ 101-107

References ........................................................................................................................................ 108-116

PART II: USER'S GUIDE ..........................................................................................................................

PART III: EXAMPLE PROBLEMS: VERIFICATIONS AND APPLICATIONS .............................

6

INTRODUCTION, FINITE ELEMENT METHOD,

CONSTITUTIVE MODELS, CONSTRUCTION SEQUENCES

INTRODUCTION

Nonlinear behavior of materials involving solids and interfaces can arise due to material

or geometric nonlinearity, or both. Material nonlinearity under mechanical, thermal and other

environmental loadings, can be due to several factors such as initial state of stress, stress path

dependent response, elastic, plastic and creep strains, change in the physical state defined by

change in the density, void ratio or water content, plastic yielding or hardening, microcracking

and damage leading to softening behavior.

Problems in solid and geomechanics can involve both types of nonlinearities. However,

in the current computer procedures, only material nonlinearity is considered with two-

dimensional (2-D) (plane stress, plane strain and axisymmetric ) and three-dimensional (3-D)

idealizations. The procedures and codes can be used for stress-deformation analysis of a wide

range of problems in solid, structural, geotechnical, and mechanical engineering and electronic

packaging involving “solid” materials, interfaces and joints. The loading can be static, cyclic and

repetitive and dynamic, and the material response can include elastic, plastic and creep

deformations, microcracking and damage leading to softening or degradation, fatigue failure, and

in microstructural instabilities like liquefaction. Typical examples are also presented. Part III of

the manual covers range of applications.

Realistic solution procedures for engineering problems require appropriate provision for

initial conditions, non-homogeneities and interaction effects. Conventional methods based on

classical theories of elasticity and plasticity may not be capable to handle the above factors.

7

Hence, the approach should be to adopt improved but simplified models that are capable to allow

for factors important for a given application. Very often it becomes necessary to resort to

numerical techniques so as to allow for these factors; the finite element method (FEM) is one of

the most powerful methods to solve engineering problems, and is used herein. The FEM code

involves the unified and general approach called the disturbed state concept (DSC), which allows

for hierarchical adoption of a wide range of constitutive models: elastic, elasto-plastic,

continuous yielding, elastoviscoplastic, and disturbance (damage), depending upon the need of

the user for specific application.

FINITE ELEMENT METHOD

In this part of the report, two-dimensional static idealization is considered. Two- and

three-dimensional static and dynamic analyses are covered in other manuals.

The finite element method has been discussed in detail in books such as Desai and Abel

(1972) and Desai (1979). The method presented here is based on the displacement approach for

2-D problems, which has been adopted in the computer code. For two-dimensional typical

element (Fig. 1), the displacement components at any point are written as

q N = u (1)

where {u}T = [u v] is the vector of displacement components u and v at a point in the x- and y-

directions, respectively, [N} is the matrix of interpolation functions, {q}T = [u1 v1 u2 v2 … un vn]

is the nodal displacement vector , and n denotes the number of nodes.

The strain-displacement and stress-strain relations are given respectively by

q B = (2)

and

8

C = (3)

where {} and {} are strain and stress vectors, respectively, [B] is the strain-displacement

transformation matrix, and [C] is the constitutive matrix.

By using the principle of minimum potential energy, the element equilibrium equations

are derived and then expressed in the incremental form as

Q = q k t (4)

where [k1] is the tangent element stiffness matrix, {Q} is the element nodal load vector, {Qr} is

the vector of unbalanced or correction loads, and denotes increment. The terms in Eq. (4) can

be expressed as

V d B C B = k t

T

V

t (5)

and

Sd T N + V d X N = Q T

S

T

V 1

(6)

and

dVBQ r

T

r (7)

in which X is the body force vector, T is the surface traction vector, r is the unbalanced

or correction stress vector, V is the volume of the element, and S1 is the portion of surface on

which surface loads are prescribed. Equations (5) and (6) are usually integrated numerically by

using Gauss quadrature methods.

Computational Algorithm

A nonlinear problem is analyzed as a series of “piecewise” problems by using

incremental techniques in which the tangent constitutive matrix {C1] is updated at each load

9

(-1,-1)

(-1,-1) (1,1)

(1,-1)

t

s

Local Coordinates

4

3

2 1

t

s

Y

X

Global Coordinates

(b)4-Node Isoparametric Element

(-1,-1)

(-1,-1) (1,1)

(1,-1)

t

s

Local Coordinates

8

1

7

1

6 5

1

4

3

1 1

1

2

1

t

s

Y

X

Global Coordinates

(a)8-Node Isoparametric Element

Figure 1. Two-dimensional Isoparametric Solid Elements

10

increment, Fig. 2. A mixed procedure (Figure 2) which combines both incremental and iterative

techniques has been adopted together with improved drift correction procedure(s). In this

procedure, after applying each load increment, iterations are performed until convergence is

reached. The convergence criterion employed is based on the ratio of the norm of unbalanced

load and sum of the norm of total load and norm of equilibrating load; details are given

elsewhere (Desai, et al., 1991).

Element Library

The computer program has the provision for the following types of elements:

(i) Solid elements

(ii) Interface/joint, and

(iii) Bar elements.

(i) Solid Elements

Either 4-noded or 8-noded isoparametric finite elements as shown in Fig. 1, or infinite

elements (not operational at this time) (Damajanic and Owen, 1984) as shown in Fig. 3, can be

used. Equations (5) to (7) are used to compute element stiffness matrix and nodal load vector,

respectively. The Gauss quadrature process allows 2 or 3 point integration rules, i.e., total 4 or 9

integration points.

(ii) Joint/Interface Elements

These elements are represented by a thin layer solid element (Desai, et al., 1984; Sharma

and Desai, 1992), or zero thickness Goodman element (Goodman, et al., 1968). They can be

either 4-noded or 6-noded elements (Fig. 4) corresponding to 4-noded or 8-noded solid elements.

The shear and normal responses found from special laboratory tests are used to define the

element stiffness matrix. The constitutive laws, discussed later, are written in terms of shear

11

Figure 2. Schematic of Incremental and Iterative Technique

Load

Q1

Q2

Q3

Displacement

12

0

Y

X

6

5

4

3

2

1

Global coordinate

s

t

Local coordinate

(a) Biquadratic singly infinite element

0

Y

X

3

2

1

Global coordinate

s

t

Local coordinate

(b) Biquadratic doubly infinite element

Figure 3. Two-Dimensional Infinite Elements

13

Figure 4. Joint/Interface Elements

y

Two-Dimensional

x

t

Body 2

Body 1

(8-noded)

Thin-Layer

Element

(4- or 6-

noded)

Body 1

Body 2

14

stress, , and normal stress, n. For the thin-layer solid element, the parametric study shows that

the ratio of thickness of interface element to its width of the order of about 0.01 yield satisfactory

simulation of the interface response simulated by using the thin-layer element with finite

thickness.

(iii) Bar Elements

Two types of bar elements, 2-noded linear, and 3-noded quadrilateral elements (Fig. 5),

have been used and provide compatibility with solid and joint elements. The element stiffness

matrix and computation of axial stress are given by Desai (1979) and Lightner and Desai (1979).

CONSTITUTIVE MODELS

A number of material models have been implemented in this program. They are:

(i) Linear elastic,

(ii) Nonlinear elastic (variable moduli or hyperbolic simulation),

(iii) Elasto-plastic conventional (von Mises, Drucker-Prager, Mohr-Coulomb, and Hoek-

Brown),

(iv) Elasto-plastic continuous yielding or hardening (critical state, cap),

(v) Hierarchical Single Surface (HISS) continuous yielding (0 and 1)

(vi) Viscoelastic plastic, and

(vii) Disturbed State Concept (DSC) models; details of this general and unified approach,

from which almost all of the above models can be derived as special cases, are given

later.

15

Figure 5. Bar Elements

Y

2

1

l

X

2-node bar element

2

Y

3

1

l

X

3-node bar element

16

Each of these categories may be used for solid, structural and geologic materials and

interfaces/joints, depending upon the material behavior and user’s judgment. However, the most

realistic models are considered to be those based on plasticity or viscoplasticity, in particular the

HISS models, as they include other plasticity models as special cases, and provide a number of

advantages and simplifications (Desai, et al., 1986 and Desai, 2001). The disturbed state concept

(DSC) allows for the above models as special cases, and includes microcracking, damage and

degradation or softening and stiffening or healing (Desai, 1994, 1995, 2001; Desai and Toth,

1996); stiffening is not included in this code.

Descriptions of the above models are given in Appendices I and IV.

Nonlinear Analysis

A nonlinear problem is solved by using incremental-iterative procedures with required

iterative (drift) correction and convergence schemes. The basic incremental stress-strain

equations are given by

d C = d t (8)

where {d and {d} = incremental stress and strain vectors, respectively, and [C1] is the tangent

constitutive matrix. In the case of piecewise linear approximation to nonlinear elastic behavior,

[C1] = eC1 will be composed of Et and t for solids, or knt and kst for interfaces and joints. For

elasto-plastic behavior

C C = C pt

ett (9)

where pC1 = tangent plasticity matrix (Appendix II).

The elastoplastic response forms a part of the creep or elastoviscoplastic and disturbance

(microcracking and softening) models in the DSC. Details of the models, elastoplastic, creep and

disturbance, and associated equations are given in Appendix I, together with the incorporation of

17

thermal and cyclic hardening effects. In all cases, a drift correction procedure is used with

respect to the drift of the yield surface during incremental loading. A brief description of the drift

correction procedure is given below.

Drift Correction: During each increment of loading, the stress must lie on or within the yield

surface (assuming unloading is elastic). If the increments are not very small, the stress state at the

end of an increment may not lie on the relevant yield surface leading to the problem of the drift

of the currently computed stress as shown in Figure 6. The initial stress state {A} at point A lies

on the previous yield surface, F ({A}, A) = 0, where is the hardening parameter (Appendix

I). During the next increment, yielding occurs and the state of stress moves to point B. The new

yield surface is given by F ({B}, B) = 0. Owing to the tendency to drift, the stress state

represented by point B does not necessarily lie on this new yield surface, Figure 6. This

discrepancy can be cumulative and, therefore, it is important to ensure that the stresses and the

hardening parameter, , are modified so as to lie on the yield surface.

Potts and Gens (1985) examined five different methods for drift correction. They

considered subincrements of strains for each increment, and concluded that the method which

considered hardening during drift correction gave improved results. This scheme is modified and

is described in Appendix III; it is incorporated in the program. Also incorporated is a modified

version of the scheme proposed by Ortiz and Simo (1986). Details of the modified schemes are

given by Desai and Wathugala (1987), Wathugala and Desai (1993).

Continuous Hardening and HISS Models

The classical plasticity models such as von Mises, Mohr-Coulomb and Drucker-Prager do

not allow adequately for the volumetric response, and for the existence of yielding before the

18

Figure 6. Schematic Showing Yield Surface Drift

J2D

J1

F({B},B)=0 F({A},A)=0

Drift

B

A

19

ultimate (failure) surface is reached. Hence, their use is often limited for evaluation of failure or

ultimate loads.

In the critical state and cap models, the continuous hardening or yielding parameter is

dependent only on the volumetric plastic strain, p

v . However, in the hierarchical single surface

(HISS) models, hardening is dependent on both volumetric and deviatoric plastic strain

trajectories, v and D, respectively. These models, including the viscoplastic and general

Disturbed State Concept (DSC), are described in Appendix I.

The critical state and cap models allow for yielding before failure, but do not allow for

(a) hardening due to plastic shear strains,

(b) possibility of dilation before peak stress,

(c) different strengths under different stress paths (e.g., compression and extension),

(d) nonassociative behavior for frictional materials, and

(e) involve multiple (two) yield surfaces, which can cause computational difficulties.

The HISS models that involve single continuous yield surface, removes the above

limitations, are considered to be general and more powerful. A perspective and comparison of

the HISS model with such other models as critical state, cap and Lade are given by Desai, et al.,

(1986), Desai and Hashmi (1989), Desai (1992), Desai (1994), Desai (2001).

PROGRAM (DSC-SST2D) FEATURES

The computer program has the following capabilities:

(i) Applied Forces

The program allows for three types of loads, as static, repetitive and dynamic:

a) Extenal loads – point loads and surface loads,

b) Prescribed displacements, and

20

c) Prescribed temperature.

External Loads: Point loads, constant or time dependent, are prescribed at nodes,

whereas the surface loads (constant or time dependent) in the form of distributed traction or

pressure acting on the element sides, are converted to the equivalent nodal loads in the program.

Thermal Loads: Temperature increments or time-dependent temperature is applied at

nodes.

For a linear elastic analysis, total load or temperature may be applied in a single

increment, but in the case of nonlinear analysis, the total load or temperature is applied in several

increments.

Displacements: The program has an option of prescribed displacements, at nodes.

Total displacements at the nodes may be applied in a single increment for linear elastic

analysis, whereas in the case of nonlinear analysis, they are applied in several increments.

(ii) Initial or in situ Stresses

A number of options are available for computing the in situ stresses (see Part II: User’s

Guide). For example,

a) Prescribed in situ stress: The in situ stress is calculated using the expressions

(Chowdhury, 1978)

so c n i y s K =

K =

n i sK + 1 y =

oy x

yox

2oy

(10)

21

where x, y, and xy are in situ horizontal, vertical, and shear stresses, respectively, is the unit

weight of soil, Ko is the in situ ratio (x/y), y is the depth to the point of stress, and is the

slope of the side of the structure or ground surface (Figure 7).

b) Computed in situ Stresses: A finite element analysis of a soil mass is carried out for

body forces only, assuming linear elastic behavior. The computed vertical stress y is kept the

same, and the horizontal stress x and shear stress xy are computed as

n i sK + 1

so c n i s =

K =

2o

xy x

yox

(11)

For horizontal surface, xy = 0.

Simulation of Sequences

(iii) Addition of Material, or Placement Embankment

Simulation of addition of materials, which is called embankment, or placement in the

sequential construction procedure is shown in Figure 8. For each layer (lift) of embankment

placed, the equivalent nodal forces due to gravity are computed. The Young’s modulus, E, of the

material in the added lift is set to a very small value (about one percent of initial E), which

simulates a very “weak” material. The incremental displacements and stresses are computed

during each lift cycle and are added to those from the previous cycle; iterations are performed (if

necessary) to obtain the equilibrium for each lift. The displacements of the new surface of the

embankment are set to zero. The horizontal stress in the newly placed lift is calculated as the

vertical stress times the in situ stress ratio, Ko.

Note that in the program, the sign of the element material numbers in a newly placed lift

are set to negative, which assigns small value of Young’s modulus to those elements. At the end

22

xy

xy

x

y

h

y v

(a) (b)

Figure 7. Initial Stresses for Inclined Surface

23

Figure 8. Addition of Materials or Sequential Construction-

Embankment

{o}

Initial Stresses

{i}={o}+{i}

Final Lift

{1}

First Lift

Stress Free Surface

24

of computations for the lift when equilibrium is reached, the sign of the element material

numbers is changed back to positive.

(iv) Removal of Material or Excavation

Figure 9 shows schematic of the simulation of excavation process, which is similar to

cut-outs in plates, and involves removal of material(s). The elements to be excavated (removed)

for each lift are deleted from the system and iterations are performed (if necessary) until

equilibrium is obtained. This will result in a “stress free” excavated surface.

The two key features of the program are:

a) Excavated elements are deleted from the initial and changing mesh.

b) Stress-free surface is established by applying equal and opposite forces on the

excavated surface and by satisfying the equilibrium equation, Eq. (4).

The above process was proposed by Goodman and Brown (1963) and Brown and King (1966).

(v) Removal of Liquid (Water) or Dewatering, Fig. 10

Dewatering causes compression or consolidation and can be modeled by using the

coupled-consolidation theory. However, in order to provide a simpler and economical

formulation, dewatering is approximated in the program by assuming uncoupled and

instantaneous response. The main effect accounted for is the increase in effective stress due to

change in the unit weight of the soil in the dewatered elements. This increase is equal to the body

force due to the weight of water within each of the elements which is dewatered. The equivalent

nodal forces are given by:

V d N = F T

W

V

(12)

where {F} is the element nodal force vector and w is the unit weight of water.

25

Figure 9. Removal of Materials or Sequential Construction- Excavation

{o} Initial Stresses

{i}={o}+{i} Final Lift

Stress Free Surface

Nodal Point

Forces

{1} First Lift

26

Note that Eq. (12) applies only to elements which were submerged earlier and are now above the

water-table due to the dewatering. Figure 10 shows the dewatering in which only elements 1, 2

and 3 have body force loads due to dewatering, and the remaining elements are affected

indirectly by the loading from these three elements.

(vi) Support Systems

Structural Supports or Tie-Backs: Installation of support system such as tie-backs, Fig.

11, can be considered similar to the prestressing of concrete beams, and introduces compressive

stresses to counteract extension and tensile stresses. The installation of tie-backs involves four

simulation steps: drilling/boring a hole (at an angle to the horizontal), placing the tie-back,

grouting the tie-back, and then tensioning the tie-back to provide the design compressive stress.

A tie-back usually consists of either steel cables or steel reinforcing rods or other

structural supports. In the case of geotechnical systems, only the last portion of the tie-back is

grouted to form an anchor, and the rest of it is usually encased in a sheath to prevent transfer of

load to the excavated face.

In the simulation of the tie-backs installation in the FEM procedure, the first two steps are

not considered, and the procedure followed is

1. Apply a force along the direction of the tie-back equal and opposite to the tension

force in the tie-back.

2. Solve for new displacements and stresses.

3. Add the bar elements which simulate the tie-backs.

4. Set the bar elements stresses to the initial tension in the tie-back.

The order of these steps may not follow the actual construction procedure. In the

construction procedure, the bar is placed first before the tensioning force is applied. If this is

27

Figure 10. Dewatering

Initial Water Level

Final Water Level

1 2 3

4 5 6

9 7 8

28

followed in the numerical procedure, bar elements will resist the tensioning forces, which is not

correct. The wrong and correct sequences are illustrated in Fig. 11.

(vii) Mesh Change Option

During any increment of the loading, the mesh can be changed, i.e., some elements can

be added or deleted, or some nodes added or deleted and/or material number of elements is

changed. This option is used to simulate embankment construction and excavation. The material

number may be changed in the case of dewatering.

(viii) Boundary Conditions

The prescribed boundary conditions (e.g., fixity) are imposed in such a manner as to

minimize the number of equations to be solved. This is achieved by not formulating equations

corresponding to degrees-of-freedom at nodal points where displacements are zero, because of

the boundary conditions.

DYNAMIC ANALYSIS

The finite element equations for dynamic analysis are given by

t Q = q K + q C + q M (13)

Where [M], [C ] and [K] are the mass, damping and stiffnesses matrices, respectively, {q} is the

vector of nodal displacements, {Q(t)} is the vector of time dependent nodal forces and the

overdot denotes time derivative.

The mass matrix can be consistent when it is evaluated from the expression resulting

from energy considerations, while it is evaluated as lumped when the mass is lumped at nodes

and appears only on the diagonals of the matrix (Desai and Abel, 1972).

Details of the frequency and time domain solutions for the dynamic equations are given

in Desai and Abel (1972) or in other texts on the finite element method. For the time domain

29

Figure 11. Schematic of Supports or Tie Backs

2P

Wrong Sequence Correct Sequence

2P

Step 1

Step 2

2P

Physical Problem

P

P

30

analysis, Equations 13 are integrated in the time domain, particularly for nonlinear analysis, by

using various time integration schemes such as Euler, Newmark Method, and Wilson’s -

Method. In the present code, Newmark and Wilson’s -methods are used. At time tn+1 = tn + t,

where t is the time step and tn is the previous time level at which quantities are known, Eq. (13),

are derived as

Q = q K *

1 + n

* (14)

where (i) for Newmark Method

K + C t

+ M t

1 = K 2

*

(15a)

q t 1 - 2

+ q 1 - + q t

C +

q 1 - 2

1 +

t

q +

t

q M + Q = Q

nnn

n

n

2

n

1 + n

*

(15b)

in which , are integration parameters in the Newmark’s scheme. For conditional stability: 2

0.5.

(ii) for Wilson -Method

KCt

Mt

K

3

6*

2 (16a)

q 2

t + q 2 + q

t

3 C +

q 2 + q t

6 + q

t

6 M +

Q - Q + Q = Q

nnn

nnn2

n1 + nn

*

(16b)

in which is a parameter, usually taken as 1.4.

31

It is often difficult to define the damping matrix [C]. Hence, approximate procedures are

sometimes employed; in one such method, the damping matrix is expressed as (Clough and

Penzien, 1993):

M + K = C Mk (17)

where k and M are constants adopted by the user.

In the case of cyclic material behavior, the hysteretic damping is included through the

tangent stiffness matrix, [K*], and it may not be necessary to include the damping in the

analysis.

Mass Matrix

The code allows for two options: consistent mass and lumped mass. The consistent mass

matrix is fully populated and is derived from the energy formulation. In the case of lumped mass,

the matrix is diagonal and the tributary masses are lumped at the element nodes.

Absorbing Boundaries

In dynamic analysis, the waves radiating from a structure are reflected back in the mesh

(body) from the artificial or discretized end boundaries. This can cause spurious errors in the

computed response. One way to reduce this effect is to select the end boundaries far enough such

that the waves are absorbed by internal damping of the material. However, if the end boundaries

are close to the structure, it is desirable to provide for the absorption of the waves at the end

boundaries. In this code, the viscous damping model proposed by Lysmer and Kuhlemeyer

(1969) is implemented. Since this model is not very efficient in absorbing surface waves, it is

advisable to extend the (lateral) end boundaries as far as possible away from the structure.

Cyclic or Repetitive Loading

32

Details of cyclic or repetitive loading involving loading, unloading and reloading and

cyclic hardening are given in Appendix I.

Creep Behavior

The code includes the general DSC model which allows for microstructural changes

leading to fracture, failure or liquefaction and available continuum models such as elastic, plastic

and creep. For the latter, viscoelastic (ve), elasticviscoplastic (evp), and viscoelasticviscoplastic

(vevp) models can be used (Desai, 2001).

MATERIAL PARAMETERS

Appendix IV gives details for the determination of material constants for the above

models, based on appropriate laboratory tests for solids and interfaces/joints. It also gives details

of the determination of initial hardening and yield surface based on in situ stresses. Further

details for the HISS and DSC are also discussed in various references. Desai, et al. (1986), Desai

and Zhang (1987), Desai (1994, 1995, 2001), Desai, et al. (1995), Katti and Desai (1995), Desai

and Toth (1996), Desai, et al. (1997).

ORGANIZATION OF COMPUTER PROGRAM

The computer program consists of a main program and about 65 subroutines. The

program is coded in FORTRAN 90. All storage is allocated at the time of execution, and if

desired, the storage can be readily adjusted to the minimum required for the problem to be

analyzed.

33

APPENDIX I

CONSTITUTIVE MODELS

This Appendix describes various constitutive models including the unified Disturbed State

Concept (DSC).

Linear and Nonlinear Elastic Models

Linear Elastic Model

It is simplest, but probably the least applicable model for the realistic simulation of

nonlinear behavior. Its main use can be for preliminary studies, and for limited situations

involving mainly the linear behavior.

The constitutive relation for the linear elastic case is given by

C = e (I.1)

where [Cc] is the elastic constitutive matrix, which, for linear elastic and isotropic material, is a

function of two elastic constants, Young’s modulus, E, and Poisson’s ratio, [Desai and

Siriwardane (1984); Desai (2001)].

Nonlinear Elastic Models

In the computer program, hyperbolic model proposed by Kondner (1963) and formalized

by Kulhawy, et al. (1969) and Duncan and Chang (1970) is included to represent the nonlinear

elastic behavior of solid or soil materials. The tangent modulus, Et and tangent Poisson’s ratio,

t, are given by (Desai and Abel, 1972)

n i s 2 + so c c 2

- n i s- 1 R - 1

p p K = E

3

31f

2

a

3

n

at

(I.2)

and

A - 1

p / g o l F -G =

2

a3

t

34

where

n i s 2 + so c c 2

- n i s- 1 R - 1

p p K

- d = A

3

31f

a

3

n

a

31

(I.3)

1 and 3 are major and minor principal stresses, respectively, c is cohesion, is the angle of

internal friction, pa is atmospheric pressure, Rr is failure ratio, n is modulus exponent, R is

modulus number, and G, F and d are Poisson’s ratio parameters.

A total of eight parameters, K , n , Rf, c, , G, F and d are required to compute Et and

t. If the Poisson’s ratio is assumed constant, five parameters, K , n , Rf, c, and are required.

For the joint/interface elements, the normal stiffness, kn, is often assumed constant (with

a high value) for compressive normal stress and the shear stiffness, ks, is represented by the

hyperbolid model; it is expressed as (Kulhawy, et al., 1969; Desai, 1974).

n a t + c

R - 1

p K = k

ana

*f

2

a

n

n

w

*t s

*

(I.4)

where and n are shear and normal stresses, respectively, ca is adhesion, a is angle of interface

friction, w is unit weight of water and K*, n

* and *

fR are constants. Thus, for the interface, six

constants, kn, K*, n

* ca and , are required.

Plasticity Models

Various plasticity models with relevant yield criteria swhave been incorporated in the

program. The details of these criteria can be found in Desai and Siriwardane (1984), Desai

(1994), Desai, et al. (1986), Desai (1995, 2001). Here, the expressions for the yield criteria are

presented with description of parameters. Compressive stresses are assumed positive.

35

1. von Mises yield criterion

0 = - J = F yD 2 (I.5)

where J2D is the second invariant of deviatoric stress tensor, Sij, and y is the yield stress in

simple tension or compression.

2. Mohr-Coulomb yield criterion

0 = so c c - 3

n i s n i s - so c J + n i s

3

J - = F D 2

1

(I.6)

where J1 is the first invariant of the stress tensor, ij, is the angle of internal friction, c is

cohesion, and is Lode angle given by

6

6 -

J

J

2

3 3 n i s

3

1 =

5 . 1D 2

D 31 -

(I.7)

in which J3D is the third invariant of deviatoric stress tensor, Sij.

3. Drucker-Prager yield criterion

0 = k - J - J = F 1*

D 2 (I.8)

where * and k are material constants, e.g., for plane strain conditions:

n a t 12 + 9

c 3 = k ,

n a t 12 + 9

n a t =

22

* (I.9)

4. Modified Cam-clay model (Schofield and Wroth, 1968)

0 = 1 - p

p +

p p M

q = F

oo

2

2

(I.10)

where po is the semi-major size of the ellipse, Fig. I.1, M is the slope of critical state (CS) line,

and p = (1 + 23)/3 and 2D31 J 3 q . If the critical state line is considered similar to

the Mohr-Coulomb failure envelope (Eq. I.6), then

36

Figure I.1 Yield Locus for Critical State Model

dp

p

Critical State Line

Mcs

A

M

vp

J1/3

q=3J2D

2po

37

n i s n i s- so c 3

n i s3 = M

(I.11)

The size of ellipse, po, is an exponential function of the hardening parameter v = plastic

volumetric strain p

v :

p x e p = p vo co (I.12)

where pco = initial value of po,

= hardening constant =

oe1,

eo = initial void ratio,

= compression index,

= swelling index, and

v = trajectory or volumetric plastic strain.

5. Cap Model

The Cap model proposed by DiMaggio and Sandler (1971) has been adopted here. It

consists of a failure envelope (Ff) and a Cap surface (Fc), Figure I.2, the expressions for which

are

0 = J - p x e - - J = F 1

///D 2f (I.13)

and

0 = L - J + L - X - J R = F 1

22

D 22

c (I.14)

where /,

/ and

/ are material parameters, and R, X and L refer to the geometry of the cap

(Figure I.2) which are related as

L - p x e - R + L = X /// (I.15)

The yielding (hardening) defined by the cap is function of the plastic volumetric strain,

p

v , which is denoted by the hardening parameter = p

v . The hardening rule is expressed as

38

Figure I.2 Failure and Hardening Surfaces in Cap Model

Drucker-Prager Surface

Ff

Fc

Rb

J2D

J1 L Z X

von Mises Surface

39

Z+ W

- 1 n D

1 - = X

(I.16)

where D and W are material parameters, and Z is related to initial cap.

6. Hoek-Brown Model Yield Criterion (Fig. I.3)

Hoek and Brown (1980) proposed a yield (failure) criterion for rock masses as

2c3c31 s+ m - - = F (I.17)

where 1 and 3 are major and minor principal stresses, respectively, c is uniaxial compressive

strength of intact rock material, and m and s are constants which depend upon the properties of

rock and upon the extent to which it has been broken before being subjected to stresses 1 and

3. The constant m has a finite positive value which ranges from about 0.001 for highly

disturbed rock masses to about 25 for hard intact rock. The maximum value of s is unity for

intact rock, and the minimum value is zero for heavily jointed or broken rock in which tensile

strength is reduced to zero.

In terms of stress invariants, Eq. (I.17) can be written as

0 = s- 3

J m - J

3

n i s + so c m +

s o c J 4 = F c

1D 2

c

2D 2

(I.18)

where is the Lode angle (Eq. 1.7).

7. Hierarchical Single Surface (HISS) Models (Desai, et al., 1986; Desai, 1995, 2001)

Advantages of the HISS model with respect to the foregoing models are listed in Chapter

1.

The two hierarchical models, isotropic hardening with associative behavior (0 model)

and isotropic hardening with nonassociative behavior (1 model), have been incorporated in the

program.

40

Figure I.3 Hoek-Brown Model

3

1

c

Uniaxial

compresssion

Tension t

RELATIONSHIP BETWEEN

PRINCIPAL STRESS AT FAILURE

Minor principal stress or confirming pressure 3

Majo

r p

rin

cip

al

stre

ss

1 a

t fa

ilu

re

Compression Uniaxial tension

Triaxial

compresion

41

The continuous yield function (Fig. I.4) in the HISS plasticity Model:

0 = S - 1 p

J +

p

J - -

p

J = F r

m

a

1

2

a

1

n

2

a

D 2

(I.19)

where , , m and n are material parameters, pa is atmospheric pressure, Sr is the stress ratio

5.1

232

27DD JJ , and is a yield or hardening function defined as (Desai, et al. 1986; Desai and

Hashmi, 1989):

1 / a = 1 (I.20a)

or

b + b - 1 b - p x e b =

D43

D21

(I.20b)

in which a1, 1, and b1 to b4 are material constants, 2/1p

ij

p

ij dd is the trajectory of or

accumulated plastic strains, including the volumetric plastic strain (v) and deviatoric plastic

strain (D) trajectories: 2/1

D ;3/ p

ij

p

ij

p

vv EdEd ; where p

ijE = tensor of deviatoric

plastic strains.

The plastic potential function Q is expressed as

S - 1 p

J +

p

J - -

p

J = Q r

m

a

1

2

a

1

n

Q2

a

D 2

(I.21)

where

r - 1 - + = voQ (I.22)

in which /vvr , o is value of at the beginning of shear loading, and is a nonassociative

parameter. Equations I.19 and I.21 are used for the nonassociative (δ1) model.

Initial Values of and

Solution for in Eq. (I.19) leads to (Desai, et al., 1991; Desai, 1995, 2001)

42

Figure I.4 Basic, 0, and Nonassociative, 1, Models

F/

90 FQ

Ca

Ca

J1

J

2D

(a) 0 model

F/

Q/

Ca

Ca

J1

J

2D

(b) 1 model

F

Q

43

p

J

S - 1 J

J - =

a

1

n - 2

m 21

D 2

(I.23)

The initial value of = o is obtained by substituting J1, J2D, Sr based on the initial state

of stress and the knowledge of the material constants , , n and m = -0.5. Then, the initial value

of =o is found from Eq. (I-20) as the values of a1, 1, etc. are known. The value of the

volumetric plastic strain trajectory v is found from (Desai, et al., 1991):

Q

Q 3

Q

=

j ij i

2 / 1

i i

vo

(I.24)

where Q is defined based on the initial stresses. Then,

2

v

2

o D - = (I.25)

In the case of isotropic or hydrostatic initial stress:

Interface/Joint Elements

The yield function and plastic potential function for the two-dimensional case are given

by [Desai and Fishman (1991); Desai and Ma, 1992; Desai (1995)]

0

22

a

n

n

a

n

a pppF

(I.26)

0

22

a

n

n

a

n

Q

a pppQ

(I.27)

0 =

/ a = =

p

J =

D

oo

/ 1

0

a

1

n - 2

o

1

44

where and n are shear and normal stresses, respectively, n and are related to phase change

and ultimate envelope, and and Q are hardening parameters for 0 and 1, respectively. A

simple form of hardening function is given by

in which p

rdu and p

rdv are the incremental plastic shear and normal relative displacements,

respectively, a and b are hardening parameters, and Q is similar to that in Eq. (I.22).

Cohesive and Tensile Strengths

The yield function in the HISS model is extended to include cohesive or tensile strengths

by transforming the stress tensor as (Fig. I.4)

j ij i*

j i R + = (I.28a)

where R is related to cohesive or tensile strength. Details are given in Appendix IV.

Here, R can be found from empirical relations (see Appendix IV). It can also be found as

/ c= R a (I.28b)

where ac is the intercept along J2D-axis (intersection of J2D-axis and ultimate yield surface)

and is related to the cohesive strength, and is related to the slope of the ultimate yield envelope,

Fig. I.4.

Creep Models

Various models including elastoviscoplastic (evp) by Perzyna (1966) have been used to

characterize the creep behavior, Fig. I.5 (Cormeau, 1976; Owen and Hinton, 1980; Desai and

Zhang, 1987; Desai, et al., 1995; Samtani, et al., 1995). Overlay model for creep has been

proposed in (Zienkiewicz, et al., 1972; Pande, et al., 1977; Owen and Hinton, 1980). A general

v d =

v d + u d =

/ a =

prv

pr

2 pr

2 2 / 1

b

45

Figure I.5 Schematic of Strain-Time Response Under Constant Stress

t2

h

Primary creep

a

b

c

d

e

f

g

i

t1

Secondary creep

Tertiary creep

Failure

Permanent set

Time

Str

ain

0

46

approach called Multicomponent DSC (MDSC) has been proposed by Desai (2001). If the strains

in the component overlays, Fig. I.6, is assumed to be the same, the MDSC model specializes to

the overlay model.

Viscoelasticplatic (vep) or Perzyna Model

MDSC model contains various versions, such as elastic (e), viscoelastic (ve),

elastoviscoplastic (evp), and viscoelasticviscoplastic (vevp). Figure I.7(b) shows the general

rheological representation of MDSC model, from which various versions can be extracted

(Desai, 2001). For instance, the evp, Perzyna type model is shown in Fig. I.7(a), which is based

on the following expression for viscoplastic strain rate vector, vp :

} {

Q } { p v

(I.29)

N

oF

F

(I.30)

where is the fluidity parameter, is the flow function, N is the power law parameter, and Fo is

the reference value (e.g., yield stress, atmospheric constant, etc.). For associative plasticity, F

Q.

Multicomponent (MDSC) or Overlay Models

In the overlay model (Fig. I.6), the behavior of a material is assumed to be composed of

those of several overlays, each of which undergoes the same deformation (strain) and provides a

specific material characterization. The total stress field is obtained as the sum of different

contributions from each overlay. By introducing a suitable number of overlays and assigning

different material properties (parameters) to each, a variety of special models can be reproduced,

as shown below.

47

The typical strain-time (creep) relationship under constant stress is shown in Fig. I.5. The

instantaneous elastic strain. o-a, is followed by a primary creep, a-b, during which, if unloading

Figure I.6 Rheological Overly Model and Elasticviscoplastic Models

E

y

(a) Viscoplasticity

E1,1 E2,2 Ek,k

F1( y1) F2( y2) Fk (yk)

1,

N1

2,

N2 k,

Nk

(b) Overlay Model

48

occurs, an instantaneous elastic recovery, b-c, is followed by delayed elastic recovery, c-d. If the

load is continued beyond the primary creep range, secondary creep (b-e) begins which is

accompanied by irreversible deformations. Unloading at any time during b-e leaves a permanent

deformation or set (strain). On continued loading, tertiary creep begins leading to failure.

The overlay model for the two-dimensional problem is illustrated in Fig. I.6. Each

overlay can have different thicknesses and material properties. The overlays do not experience

relative motion, or they are “glued” together. Therefore, the overlay models exhibit the same

deformation under given loading.

In the MDSC (overlay) model developed here, a number of units are arranged in parallel,

Fig. I.7. This results in different stress fields, {j}, in each overlay (j) which contributes to the

total stress field {} according to the overlay thickness, tj; hence,

t = jj

k

1 =j

(I.31)

in which k is the total number of overlays in the model, and

1 = t j

k

1 =j

(I.32)

The equilibrium equations for a (finite) element become:

Q = V d t B jj

k

1 =j

T

V (I.33)

in which {Q} is the load vector.

From Eq. (I.33), the element stiffness matrix is obtained as

V d B t C B = k jj

k

1 =j

T

V

(I.34)

49

where [Cj] is the constitutive matrix. This matrix will be different for each overlay, according to

the material properties.

Figure I.7 The Overlay Model in Two-Dimensional Situation (Pande, et al.,1997)

ti 1

50

The solution procedure (see later) is then identical to that of standard viscoplasticity (Perzyna

type) involving time integration, with stress being calculated for each overlay (Owen and Hinton,

1980). It should be noted that the viscoplastic strain in an overlay will be different due to

differences in threshold yield values and flow rates, but the total strains in all overlays are the

same.

Specializations of MDSC (Overlay) Model

The material parameters for elastic, viscous and yield characterizations are shown in Fig.

I-6. By adopting different values of the parameters, the overlay model can specialize to various

versions. For instance, consider the overlay model with two viscoplastic units; such a two-

overlay model is commonly adopted; Table 1 gives examples of specializations.

Table I.1: Specializations of MDSC (Overlay) Models

Specialization

Plasticity

Model

No. of

Overlays

Thickness

Parameters

Elastic (e)1

von Mises

1

1.0 E, , , N and very

high y

Viscoelastic (ve)2

von Mises

2

0.5, 0.5

E1, 1, 1, N1, y1 = 0;

E2, 2, 2, N2, y2 =

very high

Elastoviscoplastic

(evp)3 (Perzyna type)

Any

1

1.0

E, , , N, y or F

Viscoelasticviscoplastic

(ve vp)4

von Mises

Any

1

= 2

1

0.5

0.5

E1, 1, 1, N1, y1 = 0

E2, 2, 2, N2, y2 or F

1-4The following notes show resultant models with the specific choice of parameters.

51

Notes: 1Here, as σy is high, only the elastic spring will be operational because the dashpot slider

unit will be essentially not operational.

2Here, for overlay 1 as yl = 0, only the spring and dashpot will operate, as y2 > > , only the

spring will operate in overlay 2.

3Here, with one overlay, all units are operational.

4Here, the first overlay (with y1 = 0), leads to the spring and dashpot, and, in the second

overlay, all units are operational.

Number of Overlays and Thicknesses

Usually, two overlays are sufficient and the thickness of each overlay is prescribed as 0.5.

Layered Systems with Different Material Properties

When a problem with layered material (e.g., pavement) is to be analyzed, some materials

may behave as viscoelasticviscoplastic (vevp), and others are elastic or elasto-plastic, the

following procedure can be used:

(i) For the material with vevp response, two overlays (Table I.1) can be used.

52

(ii) For the elastic response, the material is considered with one overlay and infinitely large

yield strength (Table I.1).

(iii) For the elasto-plastic response of the material, one overlay is used and the fluidity

parameter, , is taken to be very small, approximately 1/600 of fluidity parameter prescribed for

the vevp material, and N = 1.

DISTURBED STATE CONCEPT (DSC)

The DSC is considered as the culmination of various models developed previously. It is

general and unified from which most of the other models can be obtained as special cases. Its

hierarchical nature allows formulation of general constitutive matrix in computer (finite element)

procedures; hence, a chosen model can be achieved by inserting material parameters for that

model, say, elastic or continuous yield plasticity.

The DSC has been covered in a number of publications (Desai and Ma, 1992; Desai,

1995, 2001; Desai and Toth, 1996; Katti and Desai, 1995; Desai, et al., 1998a,b). Hence, brief

description is given below.

In the DSC, a deforming material element is assumed to consist of various components.

For instance, for a dry material, it is assumed to contain two components: continuum or relative

intact (RI) and discontinuum or fully adjusted (FA) phases. These components interact and

merge into each other, transforming the initial RI phase to the ultimate FA phase. The

transformation occurs due to continuous modifications in the microstructure of the material. The

disturbance or microstructural changes act as a coupling mechanism between the RI and FA

phases.

The incremental constitutive equations for the DSC can be expressed as follows:

53

ic

cc

iia

dD

dCD

dCDd

1

(I.35a)

where a,i, and c denote observed, RI and FA states, respectively, {} and {} are the stress and

strain vectors, and dD the increment (or rate) of disturbance, D.

Degradation and Softening

The disturbance can be defined on the basis observed (laboratory and/or field) behavior

in terms of stress-strain, volumetric strain, pore water pressure, ultrasonic properties as P- and S-

waves, e.g., shear wave velocity (Desai, 2001). For instance, D can be expressed (Fig. I.8) as

ci

ai

D

(I-36a)

Disturbance can be expressed in terms of an internal variable such as accumulated deviatoric

plastic strain (D) or worki:

zDA

u eDD

1 (I-36b)

where Du, A, and Z are parameters determined by using Eq. (I-35).

The continuum or RI phase can be characterized by using models based on continuum

elasticity, plasticity or viscoplasticity. For instance, the constitutive matrix [Ci] can be defined by

the HISS plasticity or conventional plasticity model. The FA part can be modeled in various

ways by assuming that FA part (i) has no strength like conventional damage model by Kachanov

(1986), (ii) has hydrostatic strength like in classical plasticity, and (iii) has strength

corresponding to the critical state (Schofield and Wroth, 1968), at which the material deforms

without change in volume or density. For instance, if we assume that the FA part has only

hydrostatic strength, defined by bulk modulus, K, Eq. (I-35a) reduces to:

54

Figure I.8 Schematic of Elastoplastic and softening (DSC) Responses

Elastoplastic(virgin)

(a) Elastoplastic Response with Unloading and Reloading

Elastoplastic(i)

(b) DSC Softening with Unloading and Reloading

Softening: Observed(a)

Fully Adjusted(c)

D

55

iii

iia

SdD

ID

dCDd

3

1 (I-35b)

where {I} is the unit vector and {S} is the vector of shear stress components. Here, it is assumed

that the mean pressure p (= Ji/3 = ii/3) and the strains are the same in the RI and FA parts. In

that case, eq. (I-35a) can be written as

dCd DSCa (I-35c)

where [CDSC

] is the general constitutive matrix and dD = {R}T {d

i}, R is derived on the basis of

the adopted yield function (Desai, 2001). The constitutive matrix is given by

icT

ciDSC

R

CDCDC

1 (I-35d)

Specializations

If D = 0, that is, the material is considered as a continuum, Eq. (I-35a) reduces to

ii dCd (I-35e)

where [Ct] can be elastic, elastoplastic, or elastoviscoplastic model.

THERMAL OR INITIAL STRAINS

Thermal and mechanical (loading) cycles are available in the finite element code. The

implementation aspects for various characterizations and cyclic (loading-unloading-reloading)

are described below.

Elastic Behavior

In the case of elastic behavior, the effect of known temperature change causing initial

strains, are given below for various two-dimensional idealizations:

56

Plane Stress

0.0

T

dTT

dTT

xy

Ty

Tx

(I.37)

where is the coefficient of thermal expansion and dT is the temperature change = T – To, To is

initial (previous) temperature and T is the current temperature.

Plane Strain

dTET

T

dTT

dTT

Tz

xy

Ty

Tx

0.0

1

1

(I.38)

where E and are the elastic parameters.

Axisymmetric

0.0

T

dTT

dTT

dTT

rz

T

Tz

Tr

(I.39)

Then the incremental elastic constitutive relation is given by

TddC

dCd

e

ee

(I.40)

where [Ce] is the elastic (tangent) constitutive matrix, and {d }, [d

e} and {d (T)} are the

vectors of total, elastic and thermal strains, respectively.

57

If the parameters E and vary with temperature, they can be expressed in terms of

temperature as (Desai, et al., 1997; Desai, 2001):

TC

t

rT

TEE

(I.41a)

C

r

rT

T

(I.41b)

where Er and r are values at reference temperature, Tr (e.g., room temperature = 300 K), and cT

and c are parameters found from laboratory tests.

Thermoplastic Behavior

The normality rule gives the increment of plastic strain vector {dp(T)} as

TQTd p , ,

(I.42)

where Q is the plastic potential function; for associative rule, Q F, where F is the yield

function. Now, the total incremental strain vector {dt} is given by

TdTdTdTd pet (I.43)

where {d(T)} is the strain vector due to temperature change. Hence,

TdQ

dTd e

(I.44a)

and

dTIQ

TC

TCd

T

e

e

0

1

e

e

d

d

(I.44b)

where 0] 1 1[0

1 I for two-dimensional case and [1 1 1 0 0 0] for three-dimensional case.

Now, the consistency condition gives

0T , , dF (I.45)

58

Therefore,

dTT

Fd

Fd

FdF

T

(I.46)

Then, use of Eqs. (I.44) and I.46) gives

2/1

o

I d

QQFQTC

F

dTTCF

dTT

FTC

T

F

T

e

T

e

T

T

e

T

(I.47a)

Therefore,

dT

QQFQTC

F

Q

T

FITC

FQ

ITC

d

QQFQTC

F

TCFQ

TCd

T

e

T

oe

T

Toe

T

e

T

e

T

e

2/1T

2/1

I

(I.47b)

The parameters in the elastoplastic model, e.g., HISS-0. can be expressed as function of

temperature as

c

r

rT

TPTP

(I.48)

where P is any parameter such as E, , Eq. (I.40); , , R, n, Eq. (I.19); a1, 1, Eq. (I.20); Pr is its

value at reference temperature Tr, and c is parameter found from laboratory tests.

Thermoviscoplastic Behavior

The total temperature dependent strain rate vector, , is assumed to be the sum of

thermoelastic strain rate, )(Te , thermoviscoplastic strain rate, )(Tvp , and the thermal strain

rate due to temperature change dT, )(T , as

TTT vpe (I.49)

Here, the thermoviscoplastic strain is contributed by rheologic or creep and temperature effects.

59

With Perzyna’s (1966) viscoplastic theory, Eq. (I.29), Eq. (I.49) can be written as

TF

F

TFTT e

o

e

(I.50)

where and are temperature dependent fluidity parameter and flow function, respectively.

Then the constitutive equations are given by

T

F

F

TFTTC

o

e

(I.51)

Viscous or creep behavior requires integration in time. The thermoviscoplastic strain rate

is evaluated from Eq. (I.29) at time step n, Fig. I.9. Then the strain rate at step (n + 1) can be

expressed by using Taylor series expansion as (Desai, et al., 1995); Owen and Hinton, 1980)

IdTGdGT

IdTT

T

dT

TT

nnnnnvp

n

nvp

n

nvpnvpnvp

21

~

~

~

1

(I.52a)

where nd~ is the stress increment, dT

n is the temperature increment, and [G1]

n, [G2]

n denote

gradient matrices at time step, n.

The increment of viscoplastic strain, nvp Td )( , can be found during the time interval

tn = tn+1- tn, Fig. I.9, as

11

nvpnvp

n

vp TTtTd (I.53)

where 0- 1. For = 0, Eq. (I.53) gives the Euler scheme, for = 0.5 the Crank-Nicolson

scheme and so on. The present code allows for = 0 and 0.5.

Now, Eq. (I.51) can be written in the incremental form as

60

Figure I.9 Time Integration for Viscoplastic Strains

nvp

tn tn+1 t

tn

nvp

vp

n+1vp

61

TdTd

dTCd

nvp

nen

~ (I.54)

Use of Eqs. (I.52) and (I.54), leads to

TdTtG

tTd TCd

n

n

n

2

n

nvp

n~

evpn

~

(I.55)

where

1

1

n

neeevp tGTCITCTC

DSC Model

In the case of the DSC model, Eq. (I.35), the RI response can be simulated as elastic, Eq.

(I.40), elastoplastic, Eq. (I.47b), or elastoviscoplastic, Eq. (I.55), which include the temperature

dependence.

With the general DSC model, Eq. (I.35), the disturbance parameters, Du, A and Z, Eq.

(I.36b) can be expressed as functions of temperature, by using Eq. (I.48). Their values

determined from tests at different temperatures, which are used to define the function in Eq.

(I.48).

CYCLIC AND REPETITIVE LOADING

Cyclic and repetitive loading, involving loading, unloading and reloading, occur in many

problems such as dynamics and earthquakes, thermomechnical response such as in electronic

packaging and semiconductor systems, and pavements. If the simulated behavior involves

continuing increase in stress along the same loading path, without unloading and reloading, Fig.

I.10, it is often referred to as monotonic or virgin loading. The unloading and reloading are often

referred to as nonvirgin loading. Loading in the opposite side, i.e., negative side of the (stress)

response, is sometimes referred to as reverse (reloading) loading. Cyclic loading without stress

62

Figure I.10 Schematic of Loading, Unloading, and Reloading

Reloading(Reverse)

Unloading

Reloading

Loading

Virgin

A

Unloading

63

reversal is often referred to one-way, while with stress reversals, it is referred to as two-way. In

the case of degradation or softening, decrease in stress beyond the peak occurs, but it is

considered different from unloading.

For the virgin loading, the constitutive equations, Eq. (I.35), apply. For nonvirgin

loading, it is required to consider additional and separate, often approximate, simulations.

In the case of elastoplastic model (e.g., HISS-0), the simulated virgin response allows

for the effect of plastic strains and plastic hardening or yielding, Fig. I.11(a). In the case of the

softening behavior, the plasticity model can simulate the RI behavior, and the use of DSC allows

for the degradation, Fig. I.11(b).

Plastic deformations can occur during unloading and reloading, and can influence the

overall response, Fig. I.11. Although models to allow for such behavior have been proposed in

the context of kinematic hardening plasticity (Mroz, et al., 1978); Somasundaram and Desai,

1988), they are often relatively complex and may involve computational difficulties. Hence,

approximate schemes that are simple but can provide satisfactory simulation have often been

used; one such method implemented in the present code, is described below.

Unloading

As indicated in Fig. I.10, the unloading response is usually nonlinear. However, as a

simplification, it is often treated as linear. Here, both linear and nonlinear elastic simulations are

included. For the nonlinear case, of which the linear simulation is a special case, the procedure

proposed by Shao and Desai (1998a,b) is used. During unloading, the following incremental

stress-strain equation is used:

dCd UL (I.56)

64

Figure I.11 Unloading Interpolation Function for CTC and SS Tests

Ge

Gb

p

e

(b) Simple Shear Test

E

e

Eu

Eb

Current

1p 1

e

1

A

(a) CTC Test

1-3

65

where [CUL

] is the elastic constitutive matrix with variable elastic unloading modulus, Eu, Fig.

I.11, and the Poisson’s ratio, , is assumed to be constant. The modulus Eu is given by

pbu EEE

111 (I.57)

where Eb is the slope of the unloading curve (response) at the point (A) of unloading, Fig.

I.11(a), and Ep is the “plastic” modulus, which is evaluated by using the following equation:

2

22

1

K

D

b

D

aa

p

JJ

pKpE

(I.58)

where K1 and K2 are constants, pa is the atmospheric pressure (used for nondimensionalization,

and b

DJ 2 and J2D and the second invariants of the deviatoric stress tensor, Sij, at the start of

unloading (point A), and at the current state during unloading, respectively.

The values of K1 and K2 are found from laboratory tests. For triaxial compression CTC:

1 > 2 = 3) and simple shear (SS) tests, their values are derived as follows:

Triaxial Compression (CTC) Test

0.1113

1

22

2

bep

D

b

D

EE

JJK

(I.59a)

1

22

12

1

2

1

3

K

a

D

b

D

p p

JJ

KK

(I.59b)

where Ee is the elastic modulus (slope) at the end of unloading and p

1 is the “plastic” strain, Fig.

I.11(a).

Simple Shear (SS) Test, Fig. I.11(b)

The relation between the elastic (Young’s) and shear moduli (G) are given by

bb GE 12 (I.60a)

66

ee GE 12 (I.60b)

Substitution of Eq. (I.60) into Eq. (I.59) and replacing DJ 2 by (shear stress) and p

1

by 1 2/3 p , where p is the “plastic” shear strain, Fig. I.11(b), leads to

0.111

2

GGK

ep

b

(I.61a)

1

2

1

2

1

1

K

a

b

p pKK

(I.61b)

where b and are the shear stresses at the point of unloading, and during unloading,

respectively.

The values of p

1 and p are evaluated by using the following equations:

be

b

p

EE

JD 11

2

32

1 (I.62a)

and

be

bp

GG

11

2

(I.62b)

Reloading

Figure I.12 shows two cases of reloading, for the one-way and two-way. In both cases,

the following constitutive equation is used:

dCRdCRd eDSCa 1 (I.63)

where R is the interpolation parameter such that 0 R 1; R = 0 for the beginning of

reloading and R = 1 at the end of reloading. Thus, at the beginning of reloading, the behavior is

elastic, given by

dCd ea (I.64a)

at the end of reloading, virgin response resumes:

67

Figure I.12 Two Reloading Cases

Eb

Ebr

=Eb

E

Unloading

Reloading

A

B

(a)Reloading case 1: A B (one-way)

Unloading

Ebr

=Ee

Reloading(Reverse) Unloading

A

B’

(b)Reloading case 2: A B’ (two-way)

68

dCd DSCa (I.64b)

The elastic modulus, ER, for the two cases, Fig. I.12, is different. For case 1, the elastic

modulus at the start of reloading, Ebr

, is given by

bbr EE (I.65a)

where Eb is the unloading slope at the beginning of unloading, Fig. I.13(a). For case 2,

ebr EE (I.65b)

where Ec is the slope at the end of unloading, Fig. I.13(b).

The interpolation parameter, R , for both cases is found as

b

D

D

J

JR

2

2 (I.66)

where b

DJ 2 and J2D are the second invariants of the shear stress tensor at the beginning of the last

unloading and current level, respectively.

In computer (finite element) analysis, the reloading stress path may be between the above

two cases. Then, a parameter, S, is defined as an indicator of the direction of reloading:

d

b

Tb dS (I.67)

where -1 S 1, {b}, {} and {d} are the stress vectors before unloading, the current stress

vector and the next stress increment respectively. S = -1 indicates case 1 reloading, while S = 1

indicates case 2 reloading. Now, Ebr

is interpolated between Eb and E as

ebbr E

S

E

S

E 2

1

2

11

(I.68a)

Then, the modulus for reloading, ER, is found as

E

R

E

R

E brR

11 (I.68b)

69

where E is the elastic modulus of the material, which is often found as (average) slope of the line

joining the unloading and end of unloading points or the initial slope, Fig. I.13(a). Then at the

beginning of reloading when R = 0, ER = E

br, which ensures smooth transition from unloading to

reloading, Fig. I.12(b). At the end of reloading (R = 1), E-R

- = E, which ensures smooth transition

from reloading to the virgin loading.

Cyclic Hardening

In the case of elastoplastic behavior, there exists a yield surface (Fo) corresponding to the

initial or past state of stress experienced by the material before the present cyclic or repetitive

load is applied, Fig. I.13. When unloading occurs, the plastic strains can change (increase or

decrease), and hence, for the reloading after the unloading, the yield surface that defines the

elastic limit usually expands from Fo to the initial surface, Fi, corresponding to each cycle N (= 1,

2, …). As a result, the magnitudes of plastic strains decrease from one cycle to the next, which is

often referred to as cyclic hardening.

For a given load or stress (increment), the final or bounding surface, Fb, can be defined

by solving the incremental constitutive equations, (I-35). In the case of repetitive loading under

constant amplitude of load (stress), Fig. I.13(b), the maximum load (Pmax) will be the amplitude

of the load (stress). In the case of cyclic (one-way) loading, Fig. I.13(c), the bounding surface,

Fb, would change for each stress increase. Note that in the repetitive load analysis, here, the time

effects are not included.

Mroz, et al. (1978) proposed a model for cyclic hardening, which was adopted by

Bonaquist and Witczak (1997) for materials in pavement structures. The approximate (modified)

method for cyclic hardening implemented in the present code is similar, and is described below.

70

Figure I.13 Cyclic Hardening Under Repeated Loading

O

B

A

C

B’

F0 F1

F2

Fb

J1

J2D

(a)Cyclic hardening

B

A C

B’

O

(c)Loading-Unloading-Reloading

O B

A

Pmax

P

Time

(b)Repeated wheel load

71

For the given load or stress increment, two bounding surfaces are defined, Fo and Fb, Fig.

I.13, and the corresponding hardening functions and parameters are o and b, Eq. (I.20), and o

and b, respectively. Here, denotes the accumulated plastic strains:

2/1

pTp dd (I.69)

where {dp} is the vector of incremental plastic strains. Then the initial yield surface parameter,

i, for a given cycle, i, is expressed as

ohoicN

b

11 (I.70a)

where hc is the cyclic hardening parameter, determined from laboratory repetitive tests. It

controls the rate of expansion of the initial yield surface, Fi, at the end of unloading for a given

cycle, N. If hc = 0, no cyclic hardening occurs.

Bonaquist and Witczak (1997) considered repeated tests involving the same stress

(amplitude) to an initially unstrained material specimen, o = 0. Then, Eq. (I.70a) becomes

bhibcN

1 (I.70b)

or ch

b N

1

where is the plastic strain trajectory up to cycle N. Plots of normalized trajectory /b vs

number of cycles are used to find hc through a least square procedure. For the granular material,

hc = 1.06 was found (Bonaquist and Witczak, 1997).

With the above formulation, the value of i, Eq. (I.70) is used to evaluate the hardening

function, i, Eq. (I.20). It is used to define the elastoplastic constitutive matrix [Cep

] = [Ci], Eq.

(I.36e), the general DSC matrix [DDSC

], Eqs. (I-35e), and (I.63), when reloading occurs

72

APPENDIX II

ELASTO-PLASTIC EQUATIONS

The incremental total strain vector {d} is the sum of incremental elastic, {de} and

plastic, {p} strain vectors, i.e.,

d + d = d pe (II.1)

The incremental elastic strain is related to the incremental stress as

d C = d ee (II.2)

where {d} is the incremental stress vector and [Ce] is the elasticity matrix.

Using the theory of plasticity, the incremental plastic strain vector is given by the flow

rule

Q = d 'p (II.3)

where / is the scalar constant of proportionality.

The consistency condition is

dF = 0 (II.4)

Equations (II.1) to (II.4) are combined to obtain the incremental stress-strain relation

d C = d p e (II.5)

where [Cep

] is the elasto-plastic constitutive matrix.

The expression for / and [C

ep] are derived as

H -

Q C

F

d C

F

=

e

T

e

T

'

(II.6)

and

73

H -

Q C

F

C

F

Q C

- C = C

e

T

e

T

e

ep e

(II.7)

where H is the term due to hardening. For non-hardening yield function, H = 0 and for hardening

yield functions, H is defined as follows:

(i) Critical State and Cap model

F

F = H

(II.8)

where

F

F =

j ij i

2 / 1

F

(II.9)

(ii) HISS model (non-assdociative)

D Q

D

Q

F +

F = H

(II.10)

where

Q

Q =

j ij i

2 / 1

Q

(II.11)

and

Q

Q =

j i Dj i D

2 / 1

D Q

(II.12)

where D denotes deviatoric part. For associative model Q F.

The elastoplastic constitutive matrix [Cep

] represents the response of the material in the

relative intact (RI) state and forms a part of the general DSC matrix, Eq. (I.35c), when

disturbance (softening or degradation) is considered.

Derivations for creep and DSC models are given by Desai (2001).

74

APPENDIX III

DRIFT CORRECTION AND DSC COMPUTER ALGORITHM

Under a given stress increment, {d}, the stresses at point B do not lie on the yield

surface, Fig. 6 (in the main text), i.e., F ({B}, B) > 0, where is the hardening function. The

stress vector {B} and B are to be corrected so that F ({B}, B) 0. The method, designated as

“correction” method by Potts and Gens (1985) and modified by Desai, et al. (1991), is described

below.

The correction is carried out by an iteration procedure. At the nth iteration, the stresses

and hardening parameters are given by

Q C e/

1nn (III.1)

dnn 1 (III.2)

vvvn dn

1

(III.3)

DDDn dn

1

(III.4)

where

HC

F

e

T

n

Q

F

, 1-n1/ (III.5)

Fd / (III.6)

FVvd / (III.7)

FDDd / (III.8)

75

0in which

F

F =

j ij i

2 / 1

F

(III.9)

3 /

F =

i i

V F

(III.10)

F

F =

j i Dj i D

2 / 1

D F

(III.11)

For non-hardening, yield function, H = 0 in Eq. III.5 and for hardening yield function, H

is given by Eq. II.8 or Eq. II.10 of Appendix II.

The derivatives

QF, and

D

FF

, are evaluated at the stress point {n-1}. The

iterations are performed until the yield function is satisfied, i.e., F ({n}, n) 0 within the

tolerance of 10-6

or less. For the first iteration, {o} is taken as {B} and o as B.

DSC Computer Algorithm

According to Eq. (I.35a), the DSC incremental finite element equations are given by

~~

o~~~

Q - QdQqdkiDSC

(III.12)

where DSC

k~

is the nonsymmetrical stiffness matrix, i

qd~

is the vector of nodal increment

displacements, ~

Q is the applied load vector,

~oQ is the balanced load vector. Incremental

iterative solution of Eq. (III.12) involves negative definite stiffness matrix in the softening zone

(Desai and Toth, 1996). However, a number of approximate but simplified strategies can be used

(Desai and Woo, 1993; Desai, et al., 1999; Desai, 2001). One such scheme is to first solve for the

76

RI response by considering only the symmetric part of DSC

k~

that defines the RI behavior. Hence,

the following RI equations are first solved:

~

1~~

i

n

i

n

i

n Qdqdk (III.13)

where i

k~

is based on elastic, elastoplastic or other suitable model for the RI behavior, i

Qd~

, is

the vector of applied loads, and n denotes incfremental step. For elasticplastic model, the drift

correction will lead to convergent solution for incremental displacements, i

in

qd1~

, which in turn

can be used for computing the RI strains, i

ind

1~ and stresses,

i

ind

1~ , Fig. III.1. Then by

considering the observed and RI strains to be at the same level, i.e., ai

inin 11 ~~

, the observed

stress, i

in 1~ , is found by using Eq. (I.35) through an iterative procedure in which the

disturbance, Eq. (I.36), is found and updated. Details of the procedure are given in Desai (2001).

77

APPENDIX IV

DETERMINATION OF CONSTANTS FOR VARIOUS MODELS

Procedures for the determination of constants for the HISS-0 and 1 models are first

described below, Desai and Wathugala (1987), Desai (1990), Desai (1994). Brief details for

determination of constants for other models, elastoviscoplastic, and disturbance (softening or

degradation) are given later.

As stated before, the constants involved in the HISS models have physical meanings and

can be determined from uniaxial, shear, hydrostatic, triaxial (cylindrical) and multiaxial (cubical)

tests.

In fact, the constants can be estimated from One Compression and One Extension

Test.

If the angles of friction in compression and extension are assumed to be equal,

i.e., c = , then three compression test can be used to find constants.

For 0 and 1 models, computer code (see below) can be used to calculate the

constants.

Schematic plots required to find the constants and brief details are given below.

Elastic Constants, Fig. IV.1

Fig. IV.1. Elastic Constants

E

1-3

1

(a) E and

v

1

78

1. Find (average) E and from unloading slopes of (1 - v) vs. 1 and v vs. 1 curves,

Fig. IV.1(a)

2. For G and K use curves in terms of oct vs. oct and 3

1J vs. v, Fig. IV.1(b).

3. Relations between elastic constants and unloading (reloading) slopes (S) for different

stress paths are given in the following Table IV.1

Figure IV.1 (continued)

G

1

oct

S

1

S

2

S

3

3

2

2 ,

3

(b) Shear Modulus, G, and Bulk Modulus,

K

(c) Slopes in Stress-Strain

Curves

v

J1/

3

K

79

Table IV.1

Figure IV.2. Ultimate Parameters: and

i

i Ultimate(Asymptotic

) J2D

J1

Compressio

n

Extension

,

22 =23

1 HC

CTE

CTC

TE RTE

PL

RT

C

SSTC

Note: The elasticity parameters can be expressed as nonlinear functions of factors such

as shear stress and mean pressure.

Plasticity Constants

Ultimate: , (Fig. IV.2, IV.3)

Test E

CTC

RTE

3S1

2

2 S1

S2 + S3

S2 + S3

4 S1

32 S1 (S2 + S3 )

4 S1 +S2 + S3 )

TC

TE

CTE

RTC

3

22 (1+)( S1 + S3 )

3S1

2 (1+)( S1 +S2 + S3 )

SS

CTC (1>2 =3), and so on.

Where SI =(average) slope of the unloading/reloading curve, oct

VS i (i=1,2,3) plot, Fig Iv.1(c)

80

Figure IV.3 Ultimate Envelopes in Different Stress Spaces

E

S

C

J2D

J1

(a) Ultimate Envelopes in J2D - J1 Space

C=Compression

S=Simple shear

E=Extension

E

S

C

(b) Ultimate Envelopes in Mohr-Coulomb (-) Space

81

1. Find ultimate (asymptotic) stresses for given stress-strain curve under initial values of J1.

Ultimate value can be found by drawing an asymptote to the curve or by taking a value of about

5 to 10% higher than stress at peak.

2. Plot DJ 2 vs. J1 for ultimate values for compression, extension and/or simple shear

paths. At least two such points are needed. If the angle of friction is compression c =

angle of friction in extension E, only one point can be sufficient.

3. Use lease square fit to find and from F = 0 with = 0.

2

E

2

c

1

tan

1

tanmm

(IV.1

m

m

p

p2

2

)(1

)(1

(IV.1b)

where

E

e

E

c

c

c

E

cp

sin3

sin

3

2tan and

sin3

sin

3

2tan ,

tan

tan.

c, s, E and c, s, E, are shown in Fig. (IV.3).

Phase Change (Fig. IV.4)

Figure IV.4. Phase Change Parameter: n

1

1

v

Contraction to Dilation

F/J1=0

J2D

J1

=0.04

0.014

(a) In J1-J2D Space

82

1. Find the state of stress at which the volume change = 0 (i.e., 1J

F

= 0).

2. Find n by substituting the stresses in the following equation:

F

1

J

J-1

2 = n

s2

1

2D

(IV.2)

(at zero volume change)

The value of n can also be found from HC test by usng the following formula (Wathugala

and Desai, 1991)

dJ 2)-(n 3 = d J 1k k1-n

1 (IV.3)

where dJ1 and dkk are increments in the J1 vs. kk curve.

Although it may depend on factors such as initial density, an average constant value of n

can be often used. For dense sands, the value of n may be around 3.0, while for loose sands and

other materials such as rock and concrete, it would be higher, often of the order of 7 to 10.

Cohesive Materials (Soils): In the case of cohesive soils, usually the (undrained) stress path may

not reach the ultimate (asymptotic) curve, and failure can occur as the phase change or the

critical state line, Fig. (IV.5), is approached (Wathugala and Desai, 1991). Then, the parameter n

is found from

n

2 =

J

J2-n

1

1m

1a

(IV.4)

where Jlm = maximum value of J1 of a yield surface and Jia = intersection of the phase change line

and the same yield surface, Fig. (IV.5). Jlm can be obtained from the effective consolidation p/ as

p 3 = J 1m (IV.5)

n can also be found from the slopes of the phase change line, SPC, and the ultimate line

(curve), SUL as

83

Figure IV.5 Phase Change Parameter for Cohesive (Soil) Materials

2

1

n

2-n =

S

S

UL

PC (IV.6)

The values of and are found by least square or an optimization procedure from:

0.5)- = (m 1 = S+ 2-n

n

1

2J

Jrm

1-

m

12D

pc

(IV.7)

where the subscript pc denotes stress quantities at the phase change; a minimum of two such

values are needed for two stress paths such as compression and extension.

Hardening Parameters: (Fig. IV.6)

Yield Surface

J

2D

J1

J1m J1a

SUL

SPC

A

Phase Change Line

Ultimate Line

84

Figure IV.6. Hardening Parameters: a1 and 1

1. For a given stress increment find p

3

p

21 d ,d , pd based on unloading modulus. Then find

.

2. Substitute the state of stress in F = 0, from which find corresponding .

3. Plot ln vs. ln for different stress-strain curves. In many cases, the results will form a

narrow band. Then draw an average straight line. The slope gives 1 and the intercept

along ln gives a1, at ℓn = 0.

If the points are scattered, it may be necessary to express a1 and/or 1 function of factors

such as initial pressure and density.

▪ ▪

▪ ▪

▪ ▪ ▪

▪ ▪ ▪ ▪

▪ ▪

▪ ▪ 1

ln

a1

ln

i

i

di ppp

=∫( dij dij) 1/2

; F=0 ppp

ppp

Su

Nonassociative

85

4. For nonassociative parameter , find the (constant) slope, Su, Fig. IV.6, of the final

portion of the v vs 1 curve. Use Su in the following equation to find .

-

Z

Y

)r - (1 ) - (

1 =

vo

(IV.8)

where

2/32/3

D2D3

n

1

2/1

D2

2/32/3

D2D3

n

1

2/5

D2D311

p

2/12/3

D2D3

p1n1

2/32/3

D2D3

2

1

2/1

D2

2/32/3

D2D3

11

p2

1

2/5

D2D311

p

2/12/3

D2D3

p1

)JJ271(JJ32

)JJ271(

JJJ3S

)JJ271)(3(nJY

)JJ271(

JJ32)JJ271(

SJJJ3S

)JJ271)(3(J2Z

and

3 / = / = r

pv

vv

Cohesive and Tensile Strengths (Fig. IV.7)

If a material possesses cohesive and tensile strengths, the yield function F is shifted in the

stress space shown in Fig. (IV.7). Then the transformed stress tensor, ij, is expressed as

86

ft

1

2 2 =2 3

2 (2 +R)=2 (3+R)

R

2R

2 2 =2 3 * *

Uniaxial Tensile Strength

1=1+R *

Hydrostatic Axis

Ultimate Envelope

Ultimate Envelope

Compression

Extension

Fig. IV.7 Cohesive and Tensile Strengths

ijij*

ij R + = (IV.9)

where the term R is related to c and , Fig. I.4, and ij = Kronecker delta; R = 0 for

cohesionless materials. R can be related to the uniaxial tensile strength of the material, ft. An

empirical relation is given as (Salami and Desai, 1990; Lade, 1982):

f 1.014 R f 1.003tt

(IV.10)

Once R is known, *

ij is used in F* = 0 Fig. (IV.7), and the plasticity parameters ( and ) are

found based on the modified F.

For rocks, f1 can be found from the following expression (Hoek and Brown, 1980)

s4 + m - m 2

1 = f 2

ct (IV.11a)

where c = unconfined compressive strength, s1 = 1.0 for intact rock, and parameter m is found

from compression test results.

87

The value of R can also be obtained in a simplified procedure, as

/ c= R a3 (IV.11b)

where ac = the intercept of DJ 2 - axis with respect to the ultimate yield surface and is related

to the cohesive strength, and is related to the slope of the ultimate yield surface (line).

COMPUTER CODE TO FIND CONSTANTS FOR 0- AND 1 MODELS

Based on the information above, and in Desai (2001), the parameters can be found by

using EXCEL.

A computer code has also been prepared to evaluate the parameters for the 0-- and 1-

models. Here, the user needs to input available stress-strain data, and the constants are computed

and printed out.

Viscoplastic and Creep Models, 0 + vp: Figs. (IV.8), (IV.9)

For the viscoplastic model (Samtani and Desai, 1991; Desai, et al., 1995; Perzyna, 1966):

F

F =

Q

= ij

vpd =

j i

vp

N

o

ij

oF

F

(IV.12)

where = fluidity parameter and N is power law parameter.

Thus with cohesive/tensile strength, the number of constants for

o-model will be 7 + 1* = 8 (9 with Hoek Brown ft)

and for

1-model will be 8 + 1 = 9.

*If cohesive/tensile strength is included.

88

Mechanics of Viscoplastic Solution

Figure IV.8. Mechanism of Viscoplastic Behavior

From creep tests (on rock salt), general expression for axial strain, 1, is given by

(Hermann, et al., 1980)

T ) - ( t K = pN

31q

1 (IV.13)

t=0 t=∞ (a)

A

+ B

(c-i) J1

J

2D

A

+ B

(c-ii) J1

J

2D

A

+ B

(c-iii) J1

J

2D

A

+ B

(c-iv) J1

J

2D

plastic

(d)

plastic vp

(e)

= a1/ v1

(f)

A B

F=J2D-(-J1n+J1

2)(1-Sr)

m

F

(g) F=0

σ (b)

89

where t = time, T = temperature, 1 - 3 = d = stress difference, and q, N and p are parameters.

From creep tests, Eq. (IV.13) can be established by finding the constants using least square fit.

For a rock salt, average values q = 0.4 and N = 3.0 were found. Now, a general form of rate vp

is written as (Desai and Zhang, 1987)

F

F

F t K q = N

o

1-qp v

(IV.14a)

Then the fluidity parameters can be expressed as

dt t K t

1 = 1-qt

0 (IV.14b)

t = total time during creep test, Fig. (IV.9). Then Eq. (IV.14b) can be integrated numerically

over total time, t , Fig. (IV.9a), and the average value of can be found. For the rock salt =

5.06 x 10-7

(day-1

) was found based on 22 tests on rock salt (Desai and Zhang, 1987).

90

Fig. IV.9 (continued)

Point 1

Time, Seconds(105)

Axia

l S

train

,

1

t

Point 1 Point 2 Point 1

Time, Seconds(105)

d

1 psi=6.89kPa

(a) Typical Creep Test for Rock Salt(Hermann, et al.,1980)

Point 1

Time, Seconds(105)

0

Figure IV.9 Creep Parameters: and N

t

,N

(b) Schematic of Creep Curve

91

ln

ln(F/F0)

ln

N

(c) Evaluation of Creep Parameters

F/F0

Fig. IV.9 (continued)

92

In general, the creep parameters and N can be found from laboratory creep tests.

Equation (IV.12) is expressed as (Desai, et al., 1995):

~~

F

F

T

vpTvpN

oF

F (IV.15a)

Hence,

nF

FnNn

o

(IV.15b)

The values of F/Fo and are found from test data [Fig. IV.9(b)] for various stress increments

(levels). Then ℓn vs ℓn (F/Fo) are plotted, Fig. IV.9(c). The average slope gives the value of N

and the intercept when ℓn (F/Fo) = 0 gives the value of , the fluidity parameter.

Thus for 0 + vp model, the number of

constants = 7 (8) + 2 = 9 (10).

Elastoviscoplastic: MDSC (Overlay Models

The foregoing gives details of the viscoplastic model according to Perzyna’s theory,

which is a special case of the general elastoviscoplastic (vevp) model available in the code; it is

based on the overlay model (Appendix I) and provides four options: elastic (e), viscoelastic (ve),

elastoviscoplastic (evp-Perzyna) and general vevp model.

It is useful to note that the parameters in the elastoviscoplastic models are essentially the

same as elastic, plastic and viscous, Table I.1 (Appendix I). Hence, their determination follows

the same procedures as for elastic, plastic, viscous, etc., models.

93

Some of the advantages of the MDSC (overlay model) are:

1. It allows for four hierarchical options, Table I-1.

2. The parameters are the same as those required for various characterizations such as

elastic and elastoplastic, and creep.

3. The disturbance (DSC) model including microcracking, fracture and degradation

(damage) can be used directly with the evep models to characterize the relative intact (RI)

behavior. Thus, creep effects can be integrated with disturbance (or damage).

4. The parameters have physical meanings as they are related to specific deformation

states, and hence, the need for regression (which may lose the physical meanings) is minimized.

5. The model can allow implicitly for elastic, plastic and creep strains with

microcracking, damage (or degradation) in a single framework.

6. The implementation of the models in computer (finite element) procedures is straight

forward and standard, and includes the available convergence and rebustness characteristics

(Appendices II and III).

As a result, the MDSC (overlay) model can provide an integrated and unified approach

with compactness of parameters, and can lead to significant advantages and simplification

compared to the closed-form models (e.g., Schapery, 1969, 1984).

Disturbance Model: (Fig. IV.10)

Details are given in Desai and Ma (1992), Desai (1995, 2001), Katti and Desai (1995),

Desai and Toth (1996).

In this model, the intact behavior is represented by using the 0-model (7 or 8 constants).

It can also be simulated as linear or nonlinear elastic (Desai and Toth, 1996).

94

D

Peak

Relative

Intact Behavior(i)

Observed Behavior(a)

Fully Adjusted(c)

J2D

I2D

Ultimate(Du)

Ultimate

Figure IV.10 (a) RI, Observed and FA Responses and Disturbance

Figure IV.10 (b) Schematic for Determination of A and Z

ln(D)

Z

1

ln(A)

ln[-

ln(

)]

D

u-D

D

95

The behavior of the material part in the fully adjusted (FA) state can be simulated in various

ways (Desai, 1995); Desai and Toth, 1995): (i) it has no strength, like in classicalo continuum

damage model (Kachanov, 1986), (ii) as a constrained liquid with no shear strength but with

hydrostatic strength, or (iii) as critical state (Roscoe, et al., 1958) when the material can carry

shear stress reached up to that state for a given hydrostatic stress and deform at constant volume.

In the present code, the constrained liquid simulation is used.

For the disturbance and softening behavior, three additional constants, Du, A and Z, are

needed in the following equations for the disturbance, D.

A - p x e - 1 - D = DZ

Du (IV.16)

where Du = ultimate disturbance and A and Z are parameters.

Disturbance D can be defined approximately as (Fig. IV.10a):

J - J

J - J = D

c2D

i2D

a2D

i2D

(IV.17)

where a, i, and c denote observed, intact and fully adjusted responses, respectively. It can also be

found from other test data such as void ratio (or volume), effective stress or pore water pressure,

and nondestructive properties such as velocities (Desai and Toth, 1996; Desai, 1995; Desai, et

al., 1998).

Now, from Eq. (IV.16)

u

u

D

DD

D

Z

A- exp (IV.18a)

u

u

D

DDn

D

ZA- (IV.18b)

and

96

D

D - D n - n = ) A ( n + )( n Z

u

u

D (IV.18c)

The values of D and D (Eq. IV.17) are found for a number of points on the stress-strain

curve and a plot of Pn (D) versus ℓ n

u

u

D

DDn is obtained. Then the slope gives Z and

the intercept gives A, Fig. (IV.10b).

Thus for the DSC model, the number of constants:

Plasticity (0) model: = 7 (8) + 3 = 10 (11)

Disturbance: = 3

CYCLIC LOADING AND LIQUEFACTION

An anisotropic hardening model (2) in the context of HISS models for sands is available

in Somasundaram and Desai (1988). A similar model (*0) for clays is available in Wathugala

and Desai (1993). These models have been implemented in dynamic coupled finite element

procedures. However, the disturbed state concept (DSC) provides a relatively simple procedure

for including the cyclic behavior. Hence, the DSC model for soils, interfaces and solders (Katti

and Desai, 1995; Desai, et al., 1995; Desai, et al., 1997; Park and Desai, 1997; Shao and Desai,

1998a,b) is implemented in separate code that allows for static and cyclic behavior of solids,

geologic materials and interfaces; it also allows identification of instability and liquefaction

(Desai, et al., 1998b). This code (DSC_DYN2D) and its documentation can be available

separately.

Cyclic or Repetitive Loadings, Unloading and Reloading

The present codes are based on use of the 0 or 1 model for virgin loading; hence, the

parameters are the same as those above for the virgin loading. The unloading and reloading are

97

simulated by using special procedures described in Appendix I. The elastic parameters involve

slopes of unloading and reloading curves (Appendix I), and cyclic hardening involves parameter

hc; they are found from appropriate laboratory tests.

INITIAL CONDITIONS (Desai, 2001)

To introduce initial (stress) conditions, the values of and need to be found to establish

the starting conditions and the corresponding yield surface.

From Eq. (I.20), can be expressed as

J

)S - (1 1

2J

J - = 1

n-2

r

m

2D

(IV.19)

where the overdot denotes nondimensional quantity using pa.

For general initial stress conditions {o}, Eq. (IV.19) is used to find = o. Then 0 is

found from

) / a( = 11/

010

(IV.20)

For hydrostatic initial stress (x = y = z; xy = yz = zx = 0), Eq. (IV.19) reduces to

= v (D = 0), and 0 is found from Eq. (IV.20).

Environmental Effects

Fluid or Water

The DSC model has been developed for saturated porous materials, and is implemented

(as stated above) for dynamic and liquefaction analysis (Park and Desai, 1997; Desai, et al.,

1998b). Here, the effective stress approach is used. Separate codes (DSC-DYN2D and DSC-

SST3D) are available for this problem.

J = 1

n-2

0

98

The DSC model has been developed for partially saturated materials by incorporating

suction (or saturation); details are given in (Desai, et al., 1996; Geiser, et al., 1997).

INTERFACE/JOINT BEHAVIOR: Fig. (IV.11)

Same framework as for “solids”

Figure IV.11. HISS Model for Interfaces/Joints

F()

P(n)

R=Roughness

200 150 175 125 100 0 75 50 25

0

5 0

4 0

30

20

1 0

(b)

Ultimate

Phase Change

Smooth Interface

=0

Phase Change

200 1500 175 125 100 0 75 50 25

0

5 0

4 0

30

20

1 0

(a)

Ultimate

Rough Interface

=0

99

The procedures for finding material constants for interfaces/joints are similar to those for

solids, and are described in Desai (1994, 2001) and other references on joints and interfaces:

Navayogarajah, et al. (1991), Desai and Fishman (1991), Desai and Ma (1992).

MATERIAL CONSTANTS

A summary of material constants in various versions of the DSC/HISS models are given

below; the first four models and the overlay model are included in the present code.

Model

Constants for

0-model

Additional

Constants

Total

0-Associative

7 (8)*

--

7 (8)

1-Nonassociative 7 (8) 1 8 (9)

0 + vp Viscoplastic 7 (8) 2 9 (10)

0 + D Disturbance 7 (8) 3 10 (11)

0 0 + vp: Disturbance 9 (10) 3 12 (13)

Temperature 7 m 7 + m (depends on

how many parameters

are functions of T)

Elastoviscoplastic: Overlay Model See Table I.1

*8 Constants if R is included.

Material constants for typical materials and interfaces/joints are given at the end of this

Appenedix.

IMPLEMENTATION and APPLICATIOINS

Various versions described before have been implemented in static and dynamic

nonlinear finite element procedures. A computer subroutine for 0- and 1-models that the users

can implement in their specific codes is given by Desai, et al. (1991).

Some of the practical problems solved and validated are stated below:

Including Verification with respect to measured responses in Field and

Laboratory

Beams

100

Footings

Piles: Static, Dynamic, Saturated Soils

Single, Group

Retaining (Reinforced) Walls

Dams and Slopes

Tunnels

Building – Foundation Systems

Nuclear Power Plant Structures

Multilayer Systems

Railway Beds

1-D,. 2-D, 3-D

Pavements

Semiconductor Chip-substrate Systems

101

MATERIAL CONSTANTS for TYPICAL MATERIALS

Material Constants for Leighton Buzzard, Munich and

McCormick Ranch Sand, (0/1-Models) (Desai, 1990; Desai and Hashmi, 1989)

Material

Constant

Leighton

Buzzard

Munich

Sand

McCormick

Ranch Sand

Elastic

Constants

E

11500 (psi)

(79328 kPa)

0.29

16500 psi

(113685 kPa)

0.36

90000 psi

(620100 kPa)

0.30

Ultimate

State

Parameter

0.1021

0.36242

0.1051

0.747

0.0519

0.36

Phase Change

Parameter

n

2.5

3.2

4.0

Hardening

Constants

b1

b2

b3

b4

0.135

450.0

0.0047

1.02

0.1258

1355.0

0.001

1.11

4.88x10-4

714.0

0.004

1.04

Nonassociative

Constant

0.29

0.35

*Usually, the hardening function, Eq. (1.20a) is used. However, when the effect of hydrostatic

(HC) and proportional loading is significant, a mixed form of can be used (Eq. I.20b)

which for HC loading reduces to

Then b1 and b2 are found from HC tests and then b3 and b4 are found from shear tests.

b + b - 1 b - p x e b =

D43

D21

b - p x e b = v21

102

MATERIAL CONSTANTS FOR SOAPSTONE FROM DIFFERENT

STRESS PATH TESTS (0-MODEL) (Salami and Desai, 1990)

Ela

stic

ity

ENGLISH UNITS SI UNITS

K 449.51 ksi 3099.37 MPa

G 614.99 ksi 4240.4 MPa

E 1327.39 ksi 9152.4 MPa

0.0792 0.0792

Cohesive

and Tensile

Strengths

R 0.155 ksi 1.067 MPa

Pla

stic

ity

Ult

imate

m -0.50 -0.50

0.0468 0.0468

0 0.74922 0.74922

1 6.8410-4

6.8410-4

Phase

Change

n 7.0 7.0

Hard

enin

g 0.747 0.747

a1 1.21510-12

1.21510-12

Note: All constants, except where indicated, are nondimensional.

103

Material Constants for Rock Salt (ov-Model)

(Desai and Zhang, 1987; Desai and Varadarajan, 1987)

K 14,989 MPa

Elasticity G 8,143 MPa

E 20,685 MPa

0.27

Cohesive/Tensile Strength R 1.79 MPa

m -0.50

Ultimate 0.0945

0.995

1 0.00049

Plasticity Phase Change n 3.0

a1 1.80910-5

Hardening 1 0.2322

Non-

associate 0.275

Viscoplastic Fluidity =5.06 10-7

per day

Parameter N=3.0

104

MATERIAL CONSTANTS FOR PLAIN CONCRETE FROM

DIFFERENT STRESS PATH TESTS (0-MODEL) (Salami and Desai, 1990)

MATERIAL CONSTANT FOR PLAIN CONCRETE

ELASTIC

CONSTANTS

ENGLISH UNITS SI UNITS

K 487.86 ksi 3363.8 MPa

G 440.36 ksi 3036.3 MPa

E 1012.82 ksi 6983.4 MPa

0.154 0.154

CONSTANTS

FOR

ULTIMATE

YIELDING

=3R 1.1833 ksi 8.1589 MPa

0.1130 0.1130

0 0.8437 0.8437

1 3.9710-4

3.9710-4

n 7.0 7.0

CONSTANTS

FOR

HARDENING

1 0.4388 0.4388

a1 6.4010-12

6.4010-12

*Here is dependent on J1, given by =0-1J1

105

Material Constants for Plain Concrete

0 + D – Disturbance Model, Desai and Woo (1993)

Constant

Value

Units

Disturbance

Du

0.875

Z

1.502

A

668.0

Plasticity

0.750

0.0678

n

5.24

a1

4.6 x 10-11

1

0.83

R

1.50

MPa

Elasticity

E (Young's Modulus)

37,000

MPa

(Poisson's Ratio)

0.25

Note: No units indicates dimensionless constant

106

Material Constants for Solder (Pb/Sn)

Material parameters for various solders (e.g., Pb/Sn) are evaluated based on available test

data; they are reported, e.g., by Desai, et al. (1997, 1998a), and Desai (2001). The elastic, plastic,

creep and disturbance parameters for 40 Pb/60 Sn solder at strain rate sec/02.0 including

temperature dependence, Eq. (I.48), are given below.

Elastic and plastic constants for Pb-Sn solders at different temperatures sec/02.0

Temperature 208K 273K 348K 373K

Ultimate Parameter

0.00083 0.00082 0.00082 0.00081

State Change

Parameter, n

2.1 2.1 2.1 2.1

Hardening Parameter

a1(10-6

)

1

1(average)

8.3

0.431

0.615

2.93

0.553

0.615

1.25

0.626

0.615

0.195

0.849

0.615

Young’s Module, E

(Gpa)

26.097 24.105 22.455 22.005

Poisson Ratio, 0.38 0.395 0.408 0.412

Thermal Expansion

Coefficient, T(1/K)

(10-6

)

2.75 2.93 3.11 3.16

Yield Stress, Y 37.241 31.724 20.690 15.172

Bonding Stress, R

(MPa)

395.456 288.168 175.196 122.105

;a

,300

)(1

1300

5.5

300

;00082.0,300

)( 300

034.0

300

;MPa67.240R,300

R)(R 300

91.1

300

;GPa45.23E,300

E)(E 300

292.0

300

4.0,300

)( 300

14.0

300

;

6300

T

24.0

300

TT 100.3,300

)(

.

107

Viscous constants for Pb-Sn solders at different temperatures

Temperature 293K 313K 333K 373K 393K

Fluidity Parameter

0.5784 2.058 3.475 4.61 6.96

Exponent, N

average

2.655

2.67

2.645

2.67

2.667

2.67

2.448

2.67

2.74

2.67

sec/8.1,300

)( 300

185.6

300

Disturbance constants for Pb-Sn Solders at different temperatures

Temperature

223K 308K 398K 423K

Plastic strain range p

Disturbance, D

Z

0.103 0.307 0.04 0.082 0.022 0.102 0.036 0.039 0.097

0.733 0.870 0.521 0.603 0.700 0.591 0.661 0.701 0.722

A 0.056 0.072 0.188 0.130 0.500 0.146 0.197 0.202 0.170

Load Drop,

A

0.026 0.062 0.068 0.054 0.007 0.069 0.046 0.039 0.058

b 0.567 0.617 0.377 0.470 0.630 0.453 0.505 0.586 0.578

102.0A,300

A)(A 300

55.1

300

Z(average) =0.676

108

REFERENCES

Bonaquist, R.F. and Witczak, M.W. (1997). “A Comprehensive Constitutive Model for Granular

Materials in Flexible Pavement Structures,” Proc., 8th

Int. Conf. on Asphalt Pavements,

Seattle, WA, USA, pp. 783-802.

Brown, C.B. and King, I.P. (1966). “Automatic Embankment Analysis,” Geotechnique, Vol. 16,

No. 3.

Chia, J. and Desai, C.S. (1998). “Constitutive Modeling of Thermomechanical Response of

Materials in Semiconductor Devices with Emphasis on Interface Behavior,” Report to

NSF, Dept. of Civil Engng. and Engng. Mechs., The Univ. of Arizona, Tucson, AZ,

USA.

Chowdhury, R.N. (1978). “Slope Analysis,” Developments in Geotech. Engr., Vol. 22, Elsevier

Scientific Publishing Company, Amsterdam.

Clough, R.W. and Penzien, J. (1993).Structural Dynamics , McGraw-Hill, New York.

Clough, R.W. and Penzien, J. (1993). Structural Dynamaics, McGraw-Hill, Inc., New York,

USA.

Cormeau, I.C. (1976). “Viscoplasticity and Plasticity in Finite Element Method,” Ph.D.

Dissertation, Univ. College of Swansea, U.K.

Damjanic, F. and Owen, D.R.J. (1984). “Mapped Infinite Elements in Transient Thermal

Analysis,” Computers and Structures, Vol. 19, No. 4, pp. 673-687.

Desai, C.S. (1974). “Numerical Design Analysis of Piles in Sand,” J. of Geotech. Eng., ASCE,

Vol. 100, GT6, pp. 613-635.

109

Desai, C.S. (1979). “Elementary Finite Element Method, “ Prentice-Hall, revised version

published as Desai, C.S. and Kundu, T. (2001), Introductory Finite Element Method,

CRC Press, Boca Raton, FL.

Desai, C.S. (1989). Letter to Editor on “Single Surface Yield and Potential Function Plasticity

Models: A Review,” Computer and Geotechnics, Vol. 7, pp. 319-335.

Desai, C.S. (1990). “Modelling and Testing: Implementation of Numerical Models and Their

Application in Practice” in Numerical Methods and Constitutive Modelling in

Geomechanics, C.S. Desai and G. Gioda (eds), Springer-Verlag, Vienna.

Desai, C.S. (1992). Discussion to “Single-Hardening Model with Application to NC Clay” by

Lade, P.V., J. of Geotech. Eng., ASCE, Vol. 118, No. 2, pp. 337-341.

Desai, C.S. (1994). “Hierarchical Single Surface and the Disturbed State Constitutive Models

with Emphasis on Geotechnical Applications,” Chap. 5 in Geotechnical Engineering:

Emerging Trends in Design and Practice, Saxena, K.P. (ed), Oxford & IBH Publ. Co.,

New Delhi, India.

Desai, C.S. (1995). “Constitutive Modelling Using the Disturbed State as Microstructure Self-

Adjustment Concept,” Chapter in Continuum Models for Materials with Microstructure,

H.B. Mϋhlhaus (ed), John Wiley and Sons, U.K.

Desai, C.S. (2001). Mechanics of Materials and Interfaces: The Disturbed State Concept, CRC

Press, Boca Raton, Florida, USA.

Desai, C.S. and Abel, J.F. (1972). Introduction to the Finite Element Method, Van Nostrand

Reinhold, New York.

Desai, C.S., Basaran, C., Dishongh, T. and Prince, J. (1998a). “Thermomechanical Analysis in

Electronic Packaging with Unified Constitutive Model for Materials and Joints,”

110

Components, Packaging and Manuf. Tech., Part B: Adv. Packaging, IEEE Trans., Vol.

21, No. 1, pp. 87-97.

Desai, C.S., Chia, J.., Kundu, T. and Prince, J. (1997). “Thermomechanical Response of

Materials and Interfaces in Electronic Packaging: Parts I & II,” J. of Elect. Packaging,

ASME, Vol. 119, pp. 924-305.

Desai, C.S. and Fishman, K.L. (1991). “Plasticity Based Constitutive Model with Associated

Testing for Joints,” Int. J. Rock Mech. Min. Sci. & Geomech. Abstr., Vol. 28, No. 1, pp.

15-26.

Desai, C.S. and Hashmi, Q.S.E. (1989). “Analysis, Evaluation, and Implementation of a

Nonassociative Model for Geologic Materials,” Int. J. Plasticity, Vol. 5, pp. 397-420.

Desai, C.S. and Ma, Y. (1992). “Modelling of Joints and Interfaces Using the Disturbed State

Concept,” Int. J. Num. Analyt. Meth. Geomech., Vol. 16, No. 9, pp. 623-653.

Desai, C.S., Park, I.J. and Shao, C. (1998b). “Fundamental Yet Simplified Model for

Liquefaction Instability,” Int. J. Num. Analyt. Meth. Geomech,., Vol. 22, pp. 721-748.

Desai, C.S., Samtani, N.C. and Vulliet, L. (1995). “Constitutive Modeling and Analysis of

Creeping Slopes,” J. Geotech. Eng., ASCE, Vol. 121, No. 1, pp. 43-56.

Desai C.S., Shao, C. and Park, I. (1997). “Disturbed State Modelling of Cyclic Behavior of Soils

and Interfaces in Dynamic Soil-Structure Interaction,” Keynote Paper, Proc., 9th

Int.

Conf. on Computer Methods and Advances in Geomech., Wuhan, China.

Desai, C.S., Sharma, K.G., Wathugala, G.W. and Rigby, D. (1991). “Implementation of

Hierarchical Single Surface 0 and 1 Models in Finite Element Procedures,” Int. J. Num.

Analyt. Meth. in Geomech., Vol. 15, No. 9, pp. 649-680.

111

Desai, C.S. and Siriwardane, H.J. (1984). Constitutive Laws for Engineering Materials, Prentice-

Hall, Englewood Cliffs, NJ.

Desai, C.S., Somasundaram, S. and Frantziskonis, G.N. (1986). “A Hierarchical Approach for

Constitutive Modelling of Geologic Materials,” Int. J. Num. Analyt. Meth. in Geomech,,

Vol. 10, pp. 225-257.

Desai, C.S. and Toth, J. (1996). “Disturbed State Constitutive Modelling Based on Stress-Strain

and Nondestructive Behavior,” Int. J. Solids and Struct.Solids and Struct., Vol. 33, No.

11, pp. 1619-1650.

Desai, C.S. and Varadarajan, A. (1987). “A Constitutive Model for Short-Term Behavior of

Rock Salt,” J. of Geophys,. Res., Vol. 92, No. 11, pp. 11445-11456.

Desai, C.S., Vulliet, L., Laloui, L. and Geiser, F. (1996). “Disturbed State Concept for

Constitutive Modeling for Partially Saturated Porous Materials,” Report, Laboratoire de

Mecanique des Sols, Ecole Polytechnique Federale de Lausanne, Lausanne, Switzerland.

Desai, C.S., Wang, Z. and Whitenack, R. (1999). “Unified Disturbed State Constitutive Models

for Materials and Computer Implementation,” Plenary Paper, Proc., 4th

Int. Conf. of

Constitutive Laws for Engineering Materials, R.C. Picu and E. Krempl (Editors),

Rensselaer Polytechnic Inst., Troy, New York.

Desai, C.S. and Wathugala, G.W. (1987). “Hierarchical and Unified Models for Solids and

Discontinuities (Joints/Interfaces),” Notes for Short Course, Tucson, Arizona.

Desai, C.S. and Woo, L. (1993). “Damage Model and Implementation in Nonlinear Dynamic

Problems,” Int. J. Computational Mechanics, Vol. 11, No. 2/3, pp. 189-206.

Desai, C.S., Zaman, M.M., Lightner, J.C. and Siriwardane, H.J. (1984). “Thin Layer Elements

for Interfaces and Joints,” Int. J. Num. Analyt. Meth. in Geomech., Vol. 8, pp. 19-43.

112

Desai, C.S. and Zhang, D. (1987). “Viscoplastic Model for Geologic Materials with Generalized

Flow Rate, Int. J. Num. Analyt. Methods in Geomech., Vol. 11, pp. 603-620.

DiMaggio, F.L. and Sandler, I.S. (1971). “Material Model for Granular Soils,” J. Engg. Mech.

Div., ASCE, Vol. 97, EM3, pp. 935-950.

Duncan, J.M. and Chang, C.Y. (1970). “Nonlinear Analysis of Stress and Strain in Soils, J. Soil

Mech. and Found. Div., ASCE, Vol. 96, No. 5, pp. 1629-1653.

Geiser, F., Laloui, L., Vulliet, L. and Desai, C.S. (1997). “Disturbed State Concept for Partially

Saturated Soils,” Proc., Numerical Models in Geomech. Conf., Montreal, Canada.

Goodman, L.E. and Brown, C.B. (1963). “Dead Loud Stresses and the Instability of Slopes,” J.

Soil Mech. and Found. Div., ASCE, Vol. 89, No. SM3, pp. 103-134.

Goodman, R.E., Taylor, R.L. and Brekke, T.L. (1968). “A Model for the Mechanics of Jointed

Rock,” J. Soil Mech. Found. Div., ASCE, Vol. 94, SM3, pp. 637-659.

Herrmann, W., Waversik, W.R. and Lauson, H.S. (1980). “Creep Curved and Fitting Parameters

for Southeastern Near Mexico Bedded Salt,” Report SAND-80-0087, Sandia National

Lab., Albuquerque, NM.

Hoek, E. and Brown, E.T. (1980). “Empirical Strength Criterion for Rock Masses,” J. Geotech.

Eng. Div., ASCE, Vol. 106, No. G79, pp. 1013-1035.

Kachanov, L.M. (1986). Introduction to Continuum Damage Mechanics, Martinus Nijhoft

Publishers, Dordrecht, The Netherlands.

Katti, D.R. and Desai, C.S. (1995). “Modeling and Testing of Cohesive Soil Using Disturbed

State Concept,” J. Eng. Mech., ASCE, Vol. 121, No. 5, pp. 648-658.

Kondner, R.L. (1963). “Hyperbolic Stress-Strain Response: Cohesive Soils,” J. Soil Mech.

Found. Div., ASCE, Vol. 89, No. SM1, pp. 115-143.

113

Kulhawy, F.H., Duncan, J.M. and Seed, H.B. (1969). “Finite Element Analyses of Stresses and

Movements in Embankments During Construction,” Report No. TE-69-4, University of

California, Berkeley.

Lade, P.V. (1982). “Three-Parameter Failure Criterion for Concrete,” J. of Eng. Mech., ASCE,

Vol. 108, EM5, pp. 850-863.

Lightner, J.G. and Desai, C.S. (1979). “Improved Numerical Procedure for Soil-Structure

Interaction Including Simulation of Construction Sequences,” Report No. VPI-E-79.32,

Dept. of Civil Eng., Virginia Tech, Blacksburg, VA.

Lysmer, J. and Kuhlemeyer, R.L. (1969). “Finite Element Model for Infinite Media,” J. of Eng.

Mech., ASCE, Vol. 95, No. 4, pp. 859-877.

Mroz, Z., Norris, V.A. and Zienkiewicz, O.C. (1978). “An Anisotropic Hardening Model for

Soils and Its Application to Cyclic Loading,” Int. J. Num. Analyt. Meth. Geomech., Vol.

2, pp. 203-221.

Navayogarajah, N., Desai, C.S. and Kiousis, P.D. (1991). “Hierarchical Single Surface Model for

Static and Cyclic Behavior of Interfaces,” J. of Eng,. Mech., ASCE, Vol. 118, No. 5, pp.

990-1011.

Ortiz, M. and Simo, J.C. (1986). “An Analysis of a New Class of Integration Algorithms for

Elastoplastic Constitutive Relations,” Int. J. Num. Methods in Eng., Vol. 23, pp. 253-366.

Owen, D.R. and Hinton, E. (1980). Finite Elements in Plasticity: Theory and Practice, Pineridge

Press, Swansea, U.K.

Pande, G.N., Owen, D.R.J. and Zienkiewicz, O.C. (1977). “Overlay Models in Time-Dependent

Nonlinear Material Analysis,” Computers and Structures, 7, 435-443.

114

Park, I.J. and Desai, C.S. (1997). “Dynamic and Liquefaction Analysis Using Disturbed State

Concept,” Report, Dept. of Civil Eng. and Eng. Mechs., The Univ. of Arizona, Tucson,

AZ, USA.

Pender, M.J. (1980). “Elastic Solutions for a Deep Circular Tunnel,” Geotechnique, pp. 216-222.

Perzyna, P. (1966). “Fundamentals Problems in Viscoplasticity,” Adv. in Appl. Mech., Vol. 9,

pp. 243-377.

Potts, D.M. and Gens, A. (1985). “A Critical Assessment of Methods of Correcting for Drift

from the Yield Surface in Elasto-Plastic Finite Element Analysis,” Int. J. Num. Analyt.

Meth. in Geomech., Vol. 9, pp. 149-159.

Prager, W. (1958). “Nonisothermal Plastic Deformations,” Bol. Koninke Nederl., Acad. Wet.,

Vol. 8, (61/3), pp. 176-182.

Roscoe, K.H., Schofield, A. and Wroth, C.P. (1958). “On Yielding of Soils,” Geotechnique, Vol.

8, pp. 22-53.

Salami, M.R. and Desai, C.S. (1990). “Constitutive Modelling Influencing Multiaxial Testing for

Plain Concrete Under Low-Confining Pressure,” J. of Materials, ACI, Vol. 87, No. 3, pp.

228-236.

Samtani, N.C. and Desai, C.S. (1991). “Constitutive Modeling and Finite Element Analysis of

Slowly Moving Landslides Using Hierarchical Viscoplastic Material Model,” Report

(NSF), Dept. of Civil Engng. and Engng. Mechanics, The Univ. of Arizona, Tucson,

Arizona.

Samtani, N.C., Desai, C.S. and Vulliet, L. (1995). “An Interface Model to Describe Viscoplastic

Behavior,” Int. J. Num. Analyt. Meth. in Geomechanics, Vol. 20, No. 4, pp. 231-252.

115

Schapery, R.A. (1969). “On the Characterization of Nonlinear Viscoelastic Materials,” Polymer

Eng. Sc., 9, 295-310.

Schapery, R.A. (1984). “Correspondence Principles and a Generalized J Integral for Large

Deformation and Fracture Analysis of Viscoelastic Media,” Int. J. Fracture, Vol. 25, pp.

195-223.

Schofield, A.N. and Wroth, C.P. (1968). Critical State Soil Mechanics, McGraw-Hill Book Co.,

London.

Shao, C. and Desai, C.S. (1998a). “Application of Disturbed State Model for Cyclic Behavior of

Clay-Steel Interfaces,” Proc., 9th

Int. Conf. on Computer Methods and Advances in

Geomech., Wuhan, China.

Shao, C. and Desai, C.S. (1998b). “Implementation of DSC Model for Dynamic Analysis of

Soil-Structure Interaction Problems,” Report to NSF, Dept. of Civil Engng. and Engng.

Mechs., The Univ. of Arizona, Tucson, AZ, USA.

Sharma, K.G. and Desai, C.S. (1992). “Further on Analysis and Implementation of Thin-Layer

Element on Interface and Joints,” J. Eng. Mech. Div., ASCE, Vol. 118, No. 12, pp. 2442-

2462.

Smith, I.M. and Griffiths, D.V. (1988). “Programming for Finite Element Method,” John Wiley

& Sons, Chichester.

Somasundaram, S. and Desai, C.S. (1988). “Modelling and Testing for Anisotropic Behavior of

Soils,” J. of Eng. Mech., ASME, Vol. 114, No. 9.

Wathugala, G.W. and Desai, C.S. (1993). “Constitutive Model for Cyclic Behavior of Clays I:

Theory,” J. of Geotech. Eng., ASCE, Vol. 119, No. 4, pp. 714-729.

116

Zienkiewicz, O.C., Nayak, G.C. and Owen, D.R.J. (1972). “Composite and Overlay Models in

Numerical Analysis of Elasto-plastic Continua,” Int. Symp. Foundations of Plasticity,

Warsaw, Poland.

***12-25-12