'irJ s-74-/~

4, 7
MISCLANEOUS PAPGR S-74-18

A SIMPLE ELASTIC CONSTITUTIVE EQUATION FOR GRANULAR MATERIAL

M, M, Ai-Husspkini

ip?

"f"

June 1974 Sponsored by Assistant Secretary of the Army (R&D), Department of the Army
Condwd by

U. S. Army Engineer Waterways Experiment Station Soils and Pavements Laboratory Yicksburg, Mississippi
APPROVED FOR PUIBUC RELEASE. DISIRIBUTIONI UNLUMITED

Best Available Copjo56 S

L4c Q,

Destroy this report when no longer needed. Do not return it to the originator.
4 ,

The findings in this report are not to be tonstrued as an official Department of the Army poSition urh|ess so designated by other outhorized documents.

S.SACEW[$

+925~o
~

.52M3a
. ..

. .~ ..

.......

...

!....

MISCELLANEOUS PAPGR-S-74-18
C

A SIMPLE ELASTIC. CONSTITUTIVE

EQUATION FOR GRANULAR MATERIAL

by

M. M. Al--lussaini

June 1974 Sponsored by Assistant Secretary of the Army, (R&D), Department of the Army

Project No. 4A061101A9ID

Conaucted by U. S. Army Engineer Waterways Ixperiment Station Soils and Pavements Laboratory Vicksburg, Mississippi
&sUYMf VIIKSHUR . MISS,

APPROVED FOR PUBLIC RELEASE; DISTRI8UTION UNLIMITED

FOREWORD
The study reported herein was initiated by the U. S. Amy Engineer Waterways Experiment Station (WES), Vicksburg, Mississippi, 4A061101A91D, and was "In-House

funded under( Department of the Army Project No.

Laboratory Independent Research Program (ILIR)," sponsored by the Assistant Secretary of the Army (R&D). This work was ,accomplished during the period March 1972-December 1973 under the general supervision of Dr. Fo C. Townsend, Chief, Soils Research Facility; Mr. and Mr. J. P. Sale, C. L. McAnear, Chief, Soil Mechanics Division; This

Chief, Soils and Pavements Laboratory, WES.

report was prepared by Dr. M. M. Al-Hussaini and was reviewed by Mr. W. C. Sherman, Jr., Chief, Earthquake Engineering and Vibrations

Division, and Dr. B. Rohani of the Soil Dynamics Division prior to its publication. appreciated. Directors of WES during the conduct of the study and the preparation and publication of this report were BG E. D. COL G. H. Hilt, CE. Technical Director was Mr. Peixotto, CE, and Useful suggestions and comments by Dr. Rohani are greatly

F. R. Brown.

iii

CONTENTS Page
FOREWORD . . . . . . . . . . . . . . . . . . . . . . . . . . . . CONVERSION FACTORS, BRITISH TO METRIC UNITS OF MEASUREMENT . . . iii vii

SUMMARY ................ PART I:

........................... ..................... ........................ ................... .................... ........

....

ix 1 1 2 3 4 5 11 17 20 27 27 29 38 38 38 39 40 43 43 44 45 46

INTRODUCTION .................

Background ................... Purpose of the Study .............. Scope of the Study ................ PART II:

REVIEW OF ELASTIC CONSTITUTIVE MODELS ....

...................... Linear Models .................. ......... Nonlinear Models Using Functional Forms ... Constitutive Models with Shear and Bulk Moduli ........ Higher Order Elastic Material Models ... ........... PART III: DEVELOPMENT OF THE CONSTITUTIVE MODEL .......... ................. .................. 6 ...... .y .... ............. ............. ................. . .. .................. Total Strain Deformation ....... The Constitutive Model ........ PART IV: Material Material Material Material PART V: Parameters a. and Parameters a and Constant X'. ........ Parameter p.........

... ... .... ... .... ... ... ..... .... .......

EXPERIMENTAL DETERMINATION OF MATERIAL CONSTANTS . . .

APPLICATION OF THE CONSTITUTIVE EQUATION ....

.... Hydrostatic State of Stress ...... ............... Uniaxial State of Strain ....... ................. .... Interpretation of Volumetric Deformation of Granular ........................ ... Material .............. Cylindrical State of Strain ........... ............... Predicted and Experimental Correlation for the Cylindrical State of Strain ................. .................... Plane Strain State .............. .................... PARE VI: TABLES 1-6 PLATES 1-35 CONCLUSIONS AND RECOMMENDATIONS .... ........... ... . ... LITERATURE CITED ....................................

47 48
49 51

CONVERSION FACTORS,

BRITISH TO MMRIC UNITS OF MSASUREMET

British units of measurement used in this report can be converted to metric units as follovs: Mu.tii pounds per square inch 'Byy o.6894757 To. Obtain newtons per square centimeter

vii

SUMMARY
In the past or until recently the majority of stress-deformation and stability analyses have been restricted to ideal material behavior. Such idealizations in material properties and geometrical conditions Realmay lead to divergence between observed and predicted behavior.

istic stresp and deformation analyses of homogeneous earth masses :or soil-structure interaction problems using numerical techniques such as the finite element and finite difference methods require the formulation of a constitutive model for the soil and structural materials. A literature review made in this study indicated that most pro-. cedures used in modeling soils are based on the theory of elasticity and curve fitting. (This study is limited to constitutive models which are based on theory of elasticity.) Linear, bilinear, trilinear, and hyperbolic models provide, under special conditions, good agreement between observed and predicted soil behavior. Unfortunately, these models lack sufficient experimental and theoretical verification to be qualified as constitutive models. A general constitutive model should predict or reproduce soil behavior under any state of stress and not be restricted to the state of stress from which it is derived. Constitutive models based on higher order elastic continuum are probably the only hope for generating truly representative material models. However, the procedure used in obtaining the needed prameters for such models is very difficult if not impossible unless some simplified assumptions are made. A nonlinear elastic constitutive relationship was developed for two granular materials: crushed Napa basalt and Painted Rock Dam material. The behavior of the material was assumed to conform with Cauchy elastic material (i.e., the state of stress is only a function of the state of strain); also, the tensorial dilatancy, which contributes to volume expansion of the material, was ignored. Previous laboratory data obtained from hydrostatic compression, triaxial compression, and plane strain shear tests on both crushed Napa basalt and Painted Rock Dam material were used to obtain the needed parameters. The resulting constitutive model was used to predict the stress-strain relations for the uniaxial state of strain (i.e., K. tests), and the predicted curves were compared with laboratory Ko data. The results showed that there is a qualitative agreement between the data predicted by the model and those observed in the laboratory. However, the quantitative agreement between the predicted and observed data needs to be improved. ix

The proposed constitutive relationship accounts for nonlinear pressure-volumetric strain behavior, nonlinear shearing stress-strain behavior, and the effect of superimposed hydrostatic pressure on the behavior of soils. The constitutive relationship, however, does not account for shear-dilatancy phenomena often observed during laboratory testing of soils. Therefore, this constitutive model should not be expected to predict the exact behavior of the material. A more complicated constitutive equation which includes tensorial nonlinearity might significantly improve the accuracy of the model. In such a case, however., more experimental work is required to 'evaluate the additional unknown parameters which are needed to develop the constitutive model.

A SI14PLE ELASTIC CONSTITUJTIVE EqUATIOY FOR GRANULAR MATERIAL

PART I:

INTRODUCTION

Background .1. In the past, it has been almost impossible to perform an exact

analysis for any realistic field -problem in soil engineering due to the lack of high-speed computers and powerful numerical techniques. As a

result, stress and deformation analyses in soil media have been restricted to ideal material properties, which are only a gross representation of actual material behavior. This idealization has not only been

restricted to material behavior but also to geometric and boundary conditions. The idealization of material properties and geometrical con-

ditions may simplify the mathematical complexity of.the problem, but generally it behavior. 2. In recent years, considerable advancement to a very mature leads to a divergence between observed and predicted soil

stage of development has been made in numerical stress analysis techniques such as the finite element and the finite difference methods. This advancement, plus larger and faster computers which can handle the

most complex computations in soil mechanics problems, has provided research and designing engineers with powerful tools. However, no stress

or deformation analysis, regardless of how intricate and theoretically exact it may be, can be useful unless a correct constitutive equation

which describes the actual behavior of the material has been used in the analysis. 3. The selection of a constitutive model is somewhat easier for In most

structural materials such as steel and concrete than for soils. cases,

steel and concrete are assumed to be linear elastic, and the ap-

plication of linear theories of elasticity for such materials yields reasonable answers for all practical purposes. However, the behavioral

characteristics of soils dictate that a linear elastic assumption will 1

only provide approximate solutions in representing actual material behavior. In cases in which solutions based upon theories of plasticity are used, the soil is usually assumed to be rigid plastic and its behavior is governed by a failure criterion such as Mohr's failure theory. Actually, the behavior Of soils is neither linear elastic nor compatible with rigid plastic or even elastic plastic classification. system: solids, liquids, and gases. The nonlinear properties of soils stem from the fact that soils are a three-phase Their kstress-strain behavior destress level, drainany correct Therefore, pends on many factors such as mineral composition, age condition, density, strain condition, etc.

modeling of soil by a constitutive relation should consider most if not all the variables which affect the stress and deformation behavior of soil.

4.

Modeling soil behavior in the form of a constitutive equation

has been the subject of many soil mechanics publications, and the concept of a constitutive relation in its broadest sense is still in a state of flux. To date, more than 20 different models have been proThese proposed models involve various forms of Review and discussion posed in soil mechanics literature for soils not subjected to creep or temperature effects. nonlinearity to be used for a particular problem.

of these models are two of the objectives of this report. Purpose of the Study 5. The initial motivation of this study was the necessity to Consequently, a com-

understand the nonlinear response of soils under different stress states when subjected to monotonically increasing loads. prehensive review of existing procedures published in soil mechanics literature was made with emphasis on the following three classes of constitutive models: a. b. c. Models derived from theories of elasticity. Models derived from theories of plasticity. Hybrid models.

The ability of each model to approximate the actual stress-strain

behavior of soil was considered; the theory and assumptions used in deriving the model were examined.; and the procedures used in obtaining soil parameters from experimental data needed to formulate the model were summarized. 6. The ultimate objective of this study was to develop a constitutive relation for granular soils which would be able to predict the nonlinear soil behavior under states of stress or deformation different from those conditions from which it was derived. The model was to be based on actual laboratory tests such as plane strain, triaxial compression, and other types of tests for evaluating the needed soil parameters. The constitutive model was to be general enough to be adopted in finite element and finite difference methods of stress or deformation analyses. Scope of the Study

1.

The aforementioned objectives were achieved by using laboraThese The

tory plane strain and triaxial compression test data for two granular materials, crushed Napa 'basalt and Painted Vock Dam material. soils were prepared at two relative densities: 70 and 100 percent.

description of these materials and other testing variables are presented in reports of previous studies.1,2 The testing equipment and procedures Sused also have been presented previously.3'34

PART II:

REVIEW OF ELASTIC CONSTITUTIVE MODELS

8.

Although the nonlinear behavior of soil has been recognized

since the birth of experimental soil mechanicS, the concept of linear elastic analysis has proved to be useful in the solution of many problems, particularly those involving very small-deformations. The adop-

tion of the theory of elasticity for solutions: of certain Classes of problems has been based primarily on practical experience in which reasonable agreement between theory and actual phenomena has been observed, 9. The simplest elastic model assumes that the soil~material is and isotropic, i.e., Hooke's law is valid. In this

linear, elastic, context,

the stress-strain relationship for soil may be expressed by

two elastic constants

ai =Ae k6j + 2Gc

C ekij+

,i
2G j-

j ,k =1 , 2
ekk'i)

(la3
(lb)

where ci = components of stress tensor A = Lame's constant

E

= dilatation = Kronecker's delta G = shear modulus = components of strain tensor

6 ij
Eij

i,j

,k = indices K = bulk modulus

Equation 1 can be written in other forms in which the elastic constants 4 and G are expressed by other known elastic constants such as the K and Poisson's ratio v

modulus of elasticity- E , the bulk modulus

The relationships between these different elastic constants for isotropic soils are shown in table 1.

: ~....... ;:i>i.. :(.. - . .... .

-----

Linear Models Linear elastic model

10. The linear elastic model (fig. 1) has been used in conjunction with the theory of elasticity for the solution of many practical problems such as beams on elastic foundations,

5 stresses

and deformations

E0

SE

STRAIN

Fig. 1.

The linear elastic model

beneath pavements,

and many other applications which can be found in Although the linear elastic

any standard soil mechanics textbook .

model usually has been applied to thick homogeneous layers of soil, it has also been used in stress and deformation analyses in which more than one homogeneous layer of soil is encountered. Burmister used the linear

elastic model to derive expressions for stres~ses and displacements in a two-layered airport pavement for which he obtained good agreement between actual and predicted pavement behavior under load.'

ll.

In general,

the elastic parameters required for formulating

a linear elastic model can be obtained from one or more of the following experimental tests: compression (i.e., triaxial compression, plane strain, one-dlmensional 7 Ko test), and sonic tests. Girijavallabban and Reese

used a linear elastic .model for the solution of stresses and deformations beneath a circular footing and obtained good agreement between the observed and predicted surface settlements. in a similar study by

Duncan et al., 8 the deformation beneath a uniformly loaded circular area placed on the surface of a homogeneous subgrade was analyzed using finite element techniques by :assigning constant each of three layers. E and v values to The resulting stresses and deformations compared

very well with those obtained from the elastic layer system developed by the California Research Corporation.9 A linear elastic model was also incorporated in the finite element program for plane strain problems 10 developed by Duncan and Dunlop to study the stability of slopes in homogeneous stiff fissured clay and 'shale. 12. In general, it appears that the linear elastic model is most useful in the analysis of stresses and deformations in homogeneous soils at low stress levels. However, for higher deviatoric stress levels in which the stress-strain curve deviates significantly from the linear form, the linear elastic model becomes of little or no value for analysis. Bilinear elastic isotropic model 13. The stress-strain behavior of soil in this model is assumed to be bilinear and can be defined by five soil parameters, as shown in fig. 2. These parameters are the initial elastic modulus (i.e., before
yield) EO
,

elastic modulus after yield

initial Poisson's

Ey

*YIELD I-.

E 0 . Pa0

STRAIN

Fig. 2.

The bilinear elastic model

ratio

,Poisson's.

ratio after yield

v 21

and the yield stress.

This model was used by D'Ap'polonia and Laxnbe in the Pinite element analysis of a footing resting :on Boston blue clay under an anis.0tropie state of stress, In their analysis, teitilela~stic modulus was E was assumed to taken as the average modulus for extension and compression at a shear stress equal to half the shear stress at failure; equal 0.001 E0 S and the yield stress was taken as 90 and 75 percent of the shear stress at failure for compression and extension, respectively. i14. Duncan
12

A bilinear constitutive model was also used by Dunlop and 6nd incorporated in their finite element program for plane They used a

strain problems to .stiidy the development of the failure zone aroun~d excavated slopes in homogeneous stiff fissured clay shales. normalized stress-strain curve for evaluating the elastic parameters in the same manner as that used by D'Appolonia and Lambe. Hiowever, they assumed E to be on the order of .0.0001 E 0 y 15. Because the conventional finite element program cannot be 'used with values of'Poisson's ratio ical difficulties,* v greater that 0.5 without numerv must be assigned a value less than 0.5. v to equal 0.1499 before yielding and

D'Appolonia and Lambe 11assumed ing from 0,4475 to 0-4i999.

0.14999995 after yielding, while Dunlop and Duncan 12assumed values rangIn general, the bilinear model has demonstrated reasonable agreement between predicted behavior obtained by finite element analysis :and field observations. Trilinear elastic isotropic model 16. in this model, the Actun stress-strain relationship for soil The first segis assumed to be approximated by three linear segments.

ment represents the initial part of the stress-strain curve; the second segment represents transient behavior between the initial and yield conditions; and the third segment represents the. stress-strain behavior
*The

numerical difficulties arise from the fact that the ma~trix 6f the elastic parameter which relates the stress to the strain matrix contains terms in the denominator equal to 1 - 2v Hence, a value of~ v t0.5 causes the denominator to equal zero.

This model was used in the finite element analysis by Ellison et al.13 to predict the load-deformation behavior of bored piles in London clay. The stress-strain curve was idealized by three segafter yield. ments, as shown in fig. 3, and the elastic modulus of each segment was related to the average undrained shear strength of soil E0 = SSu E1 = xIaS u Su as (2a) (2b)

E2 = A2 where E0 = initial elastic modulus E1 = intermediate elastic modulus E2 = elastic modulus after yield xI, The value of v

(2c)

x = constant parameters used to define the shape of the 2 stress-strain curve

-

in the analysis was assumed to be constant at all times and equal to 0.48. Ellison et al.13 showed that a trilineax model can

bE
U)

w W W (n

EO

EE=kx Eo SE2 Eo =
V1= 0.48 AXIAL STRAIN 61

Fig. 3. Trilinear elastic model (after Ellison et al.

13

accurately predict the load capacity aand load-deformation behavior of bored piles in clay.

Multilinear stress-strain model 17. While some interesting and useful. results have been obtained using the linear, bilinear, and trilineAr models, the multilinear model generally is the most useful since. it is more representative of the actual geometry of the stress-strain curve for soils. A number of procedures have been used to. incorporate this model in finite element

analyses to represent the nonlinear material behavior of soils. In modeling nonlinear soil behavior by multilinear or piecewise approximations, two procedures have been widely accepted: the iterative and incremental procedures. a. Iterative procedure. This procedure consists of first selecting a itil value of the elastic modulus E for each element in thefinitte element mesh. A certain change in the external:lOad. is applied and the resulting- stresses and strains are compared With the stress-strain relation
of the material -as shown in fig. 4a. If the calculated

stresses and strains are not compatible, then anotherz value of E is chosen forthe next analysis. The process is repeated until the difference in E calculated from one increment and E calculated from the previous increment is within an accepted tolerance. The iterative procedure is easy to program and use in the finite element
analysis and can be applied to both: loading, :nd unloading situations. However, the procedure cannot be used in problems with an initial stress of zero without some modification. b. Incremental procedure. The incremental procedure consists of subdividing the external load into many small and equal

increments which are applied incrementally. The stressstrain curve between each successive increment is assumed to be linear as shown in fig. 4b. The displacement increments wbich are related to strain are accumulated to give the total displacement at any stage of loading. At the beginning of this procedure, an initial value for E is assigned and the stresses and strains in each element are calculated. A new increment of load is added and another appropriate E value is selected based upon the stressstrain curve of the material at that particular increment. The process is repeeted and the nonlinear stress-strain behavior of the material is approximated by a series of straight lines. This procedure is" sligbtly more difficult
than they iterative procedure to program; however, it is

2I

b
Iii

STRAIN

a.

The Iterative procedure

b
Iii

STRAIN C

b, Fig. I.

The incremental procedure

Techniques for approximating nonlinearity of material

10

much more general since it provides a complete description of the stress-strain behavior of the material. It can be applied for. zero as well as nonzero initial stresses, but it cannot be .applied for materials exhibiting strain-

softetning behavior, iPe., a reduction in stress- with additional postpeak straining. In many cases, a mixed procedure which employs a. combination of incremental and iterative procedures is used. In such cases., the load is applied by increments; however, after each increment, an iterative procedure is performed to increase the accuracy of the nonlinear approximation. 18. To summarize, the nonlinear analysis in. both iterative and

incremental schemes consists of a sequence of linear approximations in which the modulus of elasticity E and Poisson's ratio v are the Thus,

only parameters needed to describe the behavior of soil material.

both procedures implicitly assumed that the material in each element is linear, elastic, isotropic, and independent of the stress level. ever, How-

if the effect of the stress level must be considered, 'then a fam-

ily of stress-strain curves under different confining pressures is needed to reflect the realistie material behavior.
not economical because it

Such a procedure is
is

requires a large computer space and it

tedious to obtain closure using iterative or incremental techniques. These disadvantages are responsible for the development of constitutive models in the form of analytical -functions.

Nonlinear Models Using Functional Forms

19. Because of the large computer space required to consider the a

effect of the stress level by the iterative Or incremental procedure, number of functional forms and curve-fitting techniques have been devised in an effort to approximate a family of Stress-strain curves by one general expression. Two functional forms have been widely used in

finite element analysis to achieve this purpose: and the spline function.

the hyperbolic function

11

Hyperbolic function 20. The hyperbolic approximation for idealizing the entire triaxial compression stress-strain curve was developed by Kondner15 and 16 They showed that the stress-strain relationship Kondner and Zelasko. for sand sheared under a constant mean normal stress can be approximated by a rectangular hyperbola whose shape is controlled by the initial slope and asymptotic value of the stress difference. and the axial strain e1 as The proposed hyd3 ) perbola was used to express the principal stress difference (o-

(1 " 3)
where

a + beI

major principal stress

03

minor principal stress

Cl = axial strain

a,b = parameters whose values depend on the sand tested and the octahedral normal stress applied The physical meaning of a and b can be. seefi in fig. 5a, in which Ei , and b a

is equal to the reciprocal of the initial tangent modulus (a1 - a3)ult ' the parameters Equation 3 can be simplified by expressing e b a and as depicted in fig. 5b.

is equal to the asymptotic value of the ultimate stress difference E1/(a1 - 03) as a linear function of , which would enable direct evaluation of The linear form of

equation 3 may be written as i

Sa + be

(4) Thus, by

where

is the intercept and a and b

is the slope of the line.

plotting the experimental data in the transformed form, the corresponding values of 21. can be easily obtained. More complicated forms of Kondner and Zelasko's hyperbolic 17 functions have been suggested by Hansen as

12

ASYMPTOTE

(i

ww
(L o./

II

(~~~0,l = '.o..1 IEt I =-T N6

0.

AXIAL STRAIN Cl

J <

IJ

=T.ANO k

-z -w a . (U) -w Lninear s

-I

a:
AXIAL STRAIN C

b.
Fig. 5.

Linear~

Representations of the byperbolic stress-strain function

13

(al

-Y3) =aa + e 1 be

(6) (6

Although in some cases Hansen's equation was found to give a slightly better fit to the experimental data than Kondner's equation, equation 3

has been favored by many researchers due to its simplicity. 22. Duncan and Chang expanded Kondne-r's hyperbolic stressvery conveniently in an incremental finite Et as a function

strain function and used it

element analysis by expressing the tangent modulus o

( 1

a),

the initial tangent modulus c and

E. as

, and the Mohr-Coulomb

soil shear strength parameters [i

Rf (l-_ sin 0)(01-e3]

2c cos 0 + 2a 3 sin0(7)

in which value of

Rf is

the ratio of (a
.

- a 3) at failure to the asymptotic

(aI - a3 )ult

The dependency of the initial tangent modulus

on the stress level was expressed by

Ea=
in which Ka and n

aPa (P3n a,
Pa

are experimentally determined parameters and Substituting equation 8 in equation 7

is the atmospheric pressure, yields

[
Et = The five parameters
-

Rt(l-si f

2c cos 0 + 2a3 sin

)(ali a3 1 s

2
Ka

a'3
F

0 ,

Rf *

Ka , and

may be determined con-

veniently from the results of a series of triaxial compression tests, 23. The Duncan and Chang model was used primarily to predict the and no serious attention was

stress-strain behavior of the material,

l14

given to predicting volume changes during shear (i.e., PoIsson's ratio

in this model was assumed to be constant). This model was used by

Chang and Duncan19 to predict soil movemenrt around a deep excavation, and close agreement was obtained between actual and predicted behavior. 24. A study made by Kulhawy et al. 20 on a number of soils shoved

that the variation of radial strain with respect to axial strain as obtained from triaxial compression tests is nonlinear and can be approxThe

imated by a hyperbolic function similar to that shown in fig. 5. slope at any point of this hyperbola was designated as the tangent Poisson's ratio v 'which is expressed as

dr
Vt d a where and vt
,

(=O)

ea

are the radial and axial strains, respectively.

The

value of

which reflects the nonlinear volume change characteris--

tics of the soil during primary loading, was found. to be dependent on the stress level in a manner similar to that of procedure adopted by Duncan and Chang et al.20 derived an expression for vt in Et
.

Using the same Et

,

deriving

Kuihavy

in terms of stress only as

G - F log

KP

1 [

"f"l

'3)(l 7 sin. 0)1

aKP(~)

2c: cos 0 + 20 3 sin 0

where

Ka , n

0 , and

Rf

are again experimentally determined G , F , and D may be obtained

and the three additional parameters

from volume change and axial strain measurements from the triaxial. com-

pression tests. 25.

The hyperbolic function model which incorporates the nonlinearity of the stress-strain curve as well as Poisson's ratio has also been used in a three-dimension finite element program by

15

Palmerton21 to study flexible pavement behavior.

He was able to obtain

close agreement between the predicted deformations and the field data. Because the parameters used in the hyperbolic function were obtained from triaxial compression tests,, their application to problems which are not axially symmetric generally is not Valid. For example in prob-

lems not categorized as axially symmetrical, the ratio of lateral strain to axial strain does not represent Poisson's ratio, nor does the slope

of stress difference versus axial strain represent the modulus of

elasticity of the material. Spline function 26. The mathematical expression used to span a given set of ex-

perimental points by several polynomials of different degrees in a manner similar to the one obtained by employing a mechanical spline or French curve is referred to as the spline function. In general, the spline function is not defined as a single expression over the entire range of data as is the hyperbolic function. Detailed derivations of spline functions and their mathematical properties have been presented by Ahlberg et al. and will not be discussed herein. Spline functions have been used as a valuable tool in curve fitting and have been applied to practical problems in science and engineering.23,24 A cubical spline was used by Desai25 to approximate the nonlinear stress-strain relationship of a cohesionless soil. He incorporated a spline function in a finite element analysis for predicting the load-deformation curve of footings. The application of spline functions discussed by :Cheek et al.
23

27.

for steady state seepage problems has 'een 28.

The advantage of the spline function is that the actual exAlso,

perimental data can be represented to any degree of accuracy.

the intermediate points and their derivatives at any instance can be readily obtained. However, spline functions require larger computer If the effect of confining pressure storage than hyperbolic functions.

must be accounted for, then splines are required for a number of stressstrain curves under various confining pressures, which requires considerably more computer storage.

16

Constitutive Models with Shear and Bulk Moduli

Nonlinear streSs-strain models with a variable modulus of E and constant Poisson's ratio V lead in most cases to

29. elasticity

erroneous estimates of volume changes that occur in Soils during shear. To accurately predict volume changes,
v

should also be varied in a

manner compatible with the variation of strains at any increment of stress. If a constitutive model is to be incorporated in a finite elethen any variation in the value of v should not exceed

ment program,

0.5; otherwise,

the mathematical formulations of the finite element This condition places a restriction on the

analysis become unstable.

nonlinear procedure used in accounting for the actual deformation of the material. Therefore., it appeaso that t and v are not necessarand another pair G and the bulk ily the most convenient material property parameters, of independent parameters such as :the shear nodulus modulus

K are more appropriate to use in many cases.

Constant bulk modulus model 30. As an alternative to using E and v for describing the

26 formulated a constinonlinear behavior of soil., Clough end Woodward

tutive matrix in terms of the shear modulus K , and used it G , and the bulk modulus

in finite element analyses to predict the deformation Their analysis was based on the assumption Based on these assumptions,

and stresses in embankments. that soil is

homogeneous and isotropic.

the stress-strain relationship for plane strain deformation may be expressed as

[']
m
where a

=v)(l-2v,") (i

01

(l - V)

(]

(12)

l-2

Ym

a, =major principal stress

= minor principal stress

17

maximum shear stress major principal strain

E,

=minor principal strain = maximum shear strain G and K in terms of E and v as

They also defined the values of

E 2(17211v+

(13)

K
a(l
+

')C

2') G and

(+)) K into equation 12,

By introducing the above definitions of

the stress-strain matrix may be written as

+ '(K

(K
4

G)

0'

:l

G) (K

G)

3(15)

S-

m0
31. The values of G and

K in the above analysis were obtained

from triaxial compression tests such that

G 0(16) and

"1

Clough and Woodward26 assumed

K to be constant and handled the stressG in equation 15.

strain nonlinearity by incrementing the shear modulus

The model was used to study the deformation of Otter Brook Dam during construction, and reasonable agreements were found between the predicted and actual deformations of the dam. Variable bulk and shear modulus model 32. Many investigators27'28,29 have shown that the bulk modulus 18

of soil is not constant but rather is confining pressure.

a function of the density and

Domaschuk and Wade

29

conducted two series of tests

on Chattahoochee Eiver sand over a wide range of relative densities. In the first test series, the sand was subjected to hydrostatic pressure

only, and they expressed the bulk modulus by

K where K.

=K.

= + o 1 , ma

(18)

and m are parameters whose values depend on the relative

and .cX is

density and the applied confining pressure, respectivelyy, the mean normal stress.

33.

29 The second test series conducted by Domaschuk and Wade

consisted of triaxial compression tests using a constant o in a manner similar to that in tests conducted by Kondner and Zelasko.16 They

showed that, when the stress-strain curves are plotted in terms of the octahedral shear Stress
*

cOct

and the octahedral shear strain

yoct

'

the general shape of each curve is expressed as

a rectangular hyperbola which can be

Toct

Yoct a + Byoct

(19)

where

is

the reciprocal of the initial shear modulus

Gi and , a

is

the reciprocal of the ultimal octahedral shear stress x rameters a and 0 are analogous but not identical with

The paand b

in equation 3. 34. An expression for the tangent shear modulus, t with respect to yoct Gt was derived

by taking the derivative of

to yield

Gt

-l Oct) .1
~) Gi

(20)

Equations 18 and 20 combined define the nonlinear behavior of soil under

primary loading. The validity of this model was checked by comparing the predicted and experimental stress-strain curves obtained from triaxial compression tests under a constant a
.

Good agreement was

19

obtained for each stress-strain curve.

35. creasing a

Domaschuk and Wade 2 9 indicated that *however, ;

G. 1

increased with in-

a mathematical form. express G

no attempt was made to relate the variables in m 30 Later, Al-Hussaini and Radhakrishnan0 expanded

Domaschuk and Wade's model and derived three empirical equations to

,

G t

:and

as
n

G,
Gt =

)n(21a) =
G. (21b)

df+ e

T(+St 2

and

V=

(21c)

where S
v
0

= ratio of the octahedral shear stress at failure

= initial Poisson's ratio

Tf

to

The constants

d,

e ,

Kb

KdI

and

can be determined

*from laboratory test results.

The equations were incorporated in a

finite element analysis to predict stresses and deformations in soil specimens sheared under plane strain conditions. Comparisons with the experimental data showed good agreement between the predicted and observed results. A variabl.e shear modulus model with constant v was used in conjunction with a finite element analysis by Clough and 32 Duncan31 and also by Girijavallabhan and Mehta.

Higher Order Elastic Material Models Although there is no complete theory of constitutive equations that encompasses all known physical phenomena, there have been a few successful attempts to develop theoretically sound constitutive models on the basis of their gross material behavior rather than their atomistic behavior. Such constitutive models are based on the assumption that matter can be replaced by a mathematical model whose kinematic 36.

20

.. . . ..............

.. .......

. ....... ..

..

or dynamic variables are piecevise continuous functions of the spatial coordinates. adequately, it For an ideal material to represent physical behavior should satisfy the principles of invariance,
33

determinism,

isotropy, and congsistenoy. elastic, and hypoelastic.

Higher order elastic material behavior Cauchy

has been divided into three major categoriesi. hyperelastic,

The major differences.:between these catego-

ries are that the behavior of :a hypoolastic material is path-independent while both hyperelastic and Cauchy.,elastic materials are path-dependent. Hyperelastic material model 37. Elastic bodies that possess an energy density functio.n are The constitutive equation for For referred to as hyperelastic materials.

hyperelastic materials is derived from 'laws of thermodynamics.33 adiabatic behavior, the conservation of energy requires that au DUij

j where a 38. iis the stress tensor, c J

(22)

U is the internal. energy density

function, and

is strain tensor.

For isotropic materials .hose strain energy density function

satisfies the invariance principles, three strain invariants as

U can be expressed in terms of

u= u(x1,I 2 ,I 3 )
where IiI, 12

(23)

, and

I3 are the first, second, and third strain in-

variants, respectively, which may be defined as

1 ie=k

(24a)
m n

(22fb)
(24c)

= immn~in

21

Employing equation 23 in a chain rule differentiation of equation 22

leads to the expression

BU
iJ

1I

a= 1 DE ij a I1+

BU 12 I12 ae ij T 1

U I3 K 3 ac .j3 ij(5

(25)

Employing equation 24, the variation of the strain invariants with respect to c.. may be obtained as

3,)
ij
=

(26a)

;e

j 8 ij a i rn m1 imjmJ Substituting equation 26 into

(26b)

(2 6 c)

where

6ij

is Kronecker's delta.

equation 25 yields
S*--

6i + U
23 )m

e iJ +

U ---

C .mcme( mj

(2T)7

iJ

a! 1

31 2

which can be simplified to

i
where a.(i = 1, 2, 3) is

li 6j

'2ij

3im`j

(28)

a response function that satisfies the

following condition:

3a. B.I= B,.

Buo Cauchy elastic material model 39. Cauchy material refers to elastic materials which do not For these materials it

(29)

possess elastic potential.

is not possible to

apply Green's theorem; therefore, an alternative method by Cauchy may be applied to obtain a constitutive equation.34 Cauchy's method is only a function of

based on the assumption that the state of stress is the current state of strain; thus,

22

ai
where fiJ

j--

(30)

is an unknown function which must be determined.

40.

The most direct approach for evaluating

is to expand

equation 30 as a polynomial such as

' a2'imE + a+ a3 6 imemne + aimemncnss +... (31)

ij

= a0 + a

where tensor

a0 , ei,

a1 ,

a 2 , ... a

are real coefficients.

Since the strain

is symmetric, then using the Cayley-Hamilton theorem: of

should. satisfy its matrix qnalysis, which implies that e /on acteristic equation, it is posSible to show that
ij + 0 2eij + 03 eimcm Oij = 0 l1

char-

(32

where

01
41.

02

end 03 are scaler polynomials that can be expressed a

I1
,

in terms of the strain invariants

12 , and

I3

Comparing equations 32 and 28, it

can be seen that the gen-

eral form of the Cauchy elastic model for infinitesimal deformation is similar to that of -a hyperelastic material even though the hyperelastic

is derived from thermodynamic consideration while Cauchy elastic material is based on matrix algebra. is considered more general. Hypoelastic material model 42. tensor path. ci, In: both hyperelastic and Cauchy :elastic material models, it di, is a function of the strain and that their relationship does not depend on the loading is assumed that the stress tensor Because of the thermodynamic restriction imposed on the hyperelastic material, the Cauchy elastic material

However, the stress-strain relationship for many engineering materials is a function of the stress path followed during shear. For such materials, Truesdel135 proposed a constitutive equation in which the rate of stress is expressed as a function of stress and the rate of deformation.

23

S:.. . :.

- ... . ........... ........-. ........ ....... ... ....... ......... . ....... .... .. ....

...

..

at = ij (m

(3 )

Equation 33 can be expanded into a general constitutive equation by 36 employing the Rivlin-Ericksen equation. Following procedures outlined by Rohani,37 a constitutive equation for rate-independent hypoelastic material can be obtained as aij =

nnaO ij

+ omndcmnBl6ij + amncnp, pma26ij + dEnn3 aij + amdemn 4aij + c=npdc pmB5aij + denn 6aimamj + arndEmn87aiscsJ + aimnnnp dEpm B8isasj
+ 5n (amdEmj + deimamj) + (ditimmj) + n3deij + n6(aimamnd nj

(34)

where

d is

an increment and the

B's

and

n's

are response functions

that can be expressed as polynomial functions of the stress invariants jl 1

J2
I

and

J3'

where J1 akk (35a)

Cr a=
2 3 20xsmin 3 im = in

(35b)
(m5h

(35c)

The solution of equation 34 for any stress path may be obtained by integration when the initial conditions are specified. 43. The degree of aij on the right-hand side of equation 34 For example, oij in hypovanish and dictates the grade of the hypoelastic material. consequently equation 34 reduces to

elastic materials of grade zero, all terms containing

24

dcr ij By substituting for

+ ;n 3ij n3 1 2G

(36) and after arranging

3K0 -2G and

terms, equation 36 may be reduced .futher to d nnj jijni =KdE nniJ + 21(deij - 3 de nnSi ) (37)

Equation

37 has the same form ofA incremental Hooke's law.

a.

If only terms

up to first power of

are retained, then equation 34 reduces to a

hypoelastic material of grade one and so on. 44.C The formulation of higher order elastic models which model experimental test results, presents a complicated problem, especially 'when dealing with Solls. Such difficulties are reflected by the limited Rohani
8

amount of research published to date on this subject.

derived

a nonlinear elastic constitutive equation for earth materials, and he obtained reasonable agreements with experimental results. Chang et al. 3 9 used a second order hyperelastic equation in an incremental form to develop a constitutive equation for Ottawa sand, which was: then incorporated in a finite element -program to predict sand behavior. Nelson and Baron teretic media. Their predicted results agreed only qualitatively with the experimental data. used an incremental constitutive equation of the Two separate models were used, in their study. In the hyperbolic type to investigate ground shock effect in nonlinear hysfirst model, both the bulk and the shear modulus were taken to be functions of the strain :invariants alone; in the second model,. they assumed that the bulk modulus was a function of combined invariance. However, they did not attempt to match results from the derived model with actual data. A first order hypoelastic constitutive equation was also used to interpret the behavior of granular material as by Coon and Evans tested under triexial compression. They obtained reasonable -greements between the predicted and actual stress-strain behavior. Concluding remarks 45. In the preceding discussion, it has been shown that the majority of elastic models which are based on curve-fitting techniques

25

? ::

II ..." .. :

.. -- -.: .... ..... . ... ......... ..... ....... ..-- --... .

(i.e., linear, bilinear, trilinear, and hyperbolic) provide good agreement between the observed and predicted soil behavior. qualified as constitutive models. However, these models lack sufficient experimental and theoretical verification to be A general constitutive model should None of the previously In addition, the be able to predict or define the behavior of soil media under any possible state of stress and deformation. discussed elastic models possess such qualities.

majority of the elastic models were derived from triaxial compression test data, a situation whnih implies axisymmetric stress and strain conditions, and, inappropriately., these models have been applied to design problems which might be better approximated as plane stress or plane strain problems. Such inconsistency between the developed model and actual field conditions may lead to erroneous estimates of soil behavior. A further restriction of incorporating elastic models in finite element programs is that Poissonts ratio must be kept below 0.5 because of instability problems. 46. This limitation places a restriction on accounting for actual soil behavior. 'Higher order elastic material models are very difficult to In some cases, these parameters are This drawback probably derive since experimental data under various stress ;tates are required to evaluate the needed parameters. extremely difficult if not impossible to obtain. investigated or applied. Nevertheless,

is the major reason why higher order elastic models have not been fully higher order elastic material models may prove to be useful in handling soil behavior asscciated with work softening and dilatancy and in predicting soil behavior under conditions different than those from which the parameters were derived. These higher order elastic material models may also provide finite element formulation free from the instability associated vhen v = 0.5 since the classical definition of Poissbn's ratio is no longer required.

26

PART III:

DEVELOPMENT OF THE CONSTITUTIVE MODEL

47.

In the previous part of this report it was concluded that a

constitutive equation based on higher order elastic continuum is probably the only way to generate a truly representative material model.
However, the procedure used to obtain the various parameters needed for not impossible. In this portion of such a model is very difficult if the report,

special forms of the general constitutive equation will be

used to generate a simple but practical constitutive model for granular materials. Procedures for obtaining material constants from various

tests are discussed and presented, and the proposed constitutive equation is evaluated by comparing the derived stress-strain relationship with observed material behavior.

Total Strain Deformation

48.,

The basic assumption of total deformation theory is that the a function of the current state of strain and is in-

state of stress is

dependent of the stress path.

The hyperelastic and Cauchy elastic maThe

terials, which were described in Part II, fall in this category. response coefficient

0i

in equation 32 (Cauchy elastic material),

which may take various forms for different materials, must be determined from experimental data. However, there is no reason (unless one is dic-

tated by experimental observation) for requiring all the response coefficients in the constitutive equation. For reasons of practicability 3 has been

and mathematical simplicity, the response coefficient

assumed to be zero in using equation 32 for describing the stress-strain behavior of soil. For this material, the tensorial dilatancy which conignored; how-

tributes to volume expansion of material under shear is functions of

ever, the scaler dilatancy may be accounted for by making I and I Thus, equation 32 becomes

01

and

02

1.

2
=j

euain2beos
0 +

O2 eij
replacing

(38)

The unknown

01

and

02

may be obtained by first

27

with

and reducing equation 38 to

J, = 301 + 021I By definition, the stress deviatoric tensor toric tensor Eij may be expressed as
Si J

(39) S and the strain devia-

0.j I

J1 3 --

(40a)

ii(ha (40b) i to

ij

=E

ij

3 ij

Using equation 40 in conjunction with equation 38 and equating j the following invariant equation can be obtained-

where

J' and I' are, respectively, the second invariants of the 2 . stress and strain deviatoric tensor and are defined as

2,= L

(4~2a)

12'

21

ij~ij

(42b) can be de-

From equations 39 and 41, the values of fined as

01 and 02

31 11

21

(43a)

2 2 0 121
Since

(43b)

01

and

02

are known in terms of the invariants, the stress-

strain relationship expressed in equation 38 may be written, after rearranging terms, as

28

3-

_-

Y27

ii-

+ Ii i a(44)

ij

In equation 44 it is only necessary to determine the functional forms of the Invariant2 JI and J. in terms of strain Invariants 11 and

The Constitutive Model S49. The invariants J and J1 in'equation 44 Can be expressed

by twocparameters J2 =P(11) 1 (45b)

where and

fI expresses the nonlinear pressure-volume change relationship f2 expresses the nonlinear shear stress-strain relationship. 50. The relationship bet"een J and II may be determined from

tests in which deviatoric stresses are not permitted, i.e., a spherical state of stress. A commn example of this is the case of isotropic consolidation. The relationship betwen J1 and. I can be ob ned fom tests in which only a deviatoric state of stress is applied however, such a condition is difficult to impose by .conventional means although an approximate relationship can be obtained using triaxial compression or plane strain shear devices. Isotropic compression test 151. This test is characterizqd by three. principal stresses and three principal strains such that 011. '22 33

oi

ii'i

(46a) (46b)
C22 C

1u =L where '11

22

33

=ij

i# j 11

022

and

033

are principal stresses, 29

and

are principal strains, the volumetric strain.

C33

is the hydrostatic pressure, and

is

52.

A typical stress-strain curve for granular material under Attempts have been made to

isotropic compression is shown in fig. 6.

W 9L

0
I

VOLUMETRIC STRAIN E

Fig. 6.

Typical hydrostatic compression test

relate the elastic behavior of individual particles to the overall elastic behavior of the mass of granular material using Hertz? t42 contact theory. Such an approach was used by Ko and Scott 27 vho showed that s for a simple cubical element of spheres can p as (47) the volumetric strain

be expressed in terms of hydrostatic pressure s where w = 3(1 - v E

=

3(16p)2/3

Actual data on sand showed that

c qoes not

vary with the two-thirds power of external pressure as predicted by Hertz' contact theory. Another empirical expression which is modified from Hertz' contact theory was used by El-Sobby28 as

SPiM 30

(48)

where data.

and

are constants that can be determined from experimental

Rohnni

53. 38

Other expressions used were strictly based on curve fittings. suggested the following: p =y 0e (49)

where

o
0

is the initial state of stress of the material that defines

the state of ease* and obtained experimentally.

8 is a parameter. as

Both

and

can be

A similar expression was suggested by

Domascxuk.and Vaaee

Sm
where X. I 54. is the initial

--- n(Ki +mp)]

m is a constant.

('50)

bulk modulus And

For reasons of simplicity,

an expression similar to equation

49 was adopted for describing the nonlinear stress-strain behavior under

isotropic compression as

0(e

1)

(51)

where

Jl

and

I1

are the first

stress and strain invariants,

respectively.

Triaxial compression test

55. This test is characterized by the symmetry of stresses and strains around one of the principal axes (i .e., major principal axis). Conditions under which the conventional triaxial test is be defined as performed may

11

202 2

33 e33

; ;

0
,

i#J i# j

(52a) (52b)

Il > e 22

'iJ =0

SIt

should be noted that a is not a material constant but rather a parameter which defines tfe initial state of stress of the soil tested.

31

in this test,

it

is customary to plot the stress difference

(o

a3

as a function of axial strain

(see fig. 5), and the resulting curve For granular material, the size and

can be aporoximated by a hyperbola.15'A,18

shape and size of such a hyperbola depend upon many factors such as relative density, drainage conditions, confining pressure, shape of particles, etc. A study by Domaschuk and Wade 2 9 showed that the hyperbolic shape of the stress-strain curve will be maintained in triaxial compression tests if the data are represented by the deviatoric stress S and the deviatoric strain
5

d where

Sd = 3 (011 C 033)(5) S
-

(53a)
(53b) J' and IV

11

3-3
e)

2
However, because respectively, S and e

are directly related to

2'

,J'

8d=.r-(oI

033)

(54a)

,fj
I

' d11

33
- 33) 5b

= 2V2 Ed =V3 (el

Therefore,

if

is to be plotted as a function of

for tri-

axial compression tests, the resulting stress-straixL curve should also be a hyperbola. Formulation of the hyperbolic function 56. For a given value of relative density and confining pressure the value of~ versus may be characterized by a rectangular hyperbolic function in a manner similar to that used by Kondner and Zelasko.16 The parameters are illustrated in fig. 7, and the resulting equation for the stress-strain curve can be expressed as

A17

3+

N -(55)

32

Fig. 7.
where a and y

Typical triaxial compression test

are parameters whose values depend on the material The phyoical meaning attached to y is equal to the inverse of r; and a a

properties and testing conditions. and y

can be seen in fig. 7, in which

the asymptotic value of V

(called the ultimate value of

is equal to the reciprocal of the initial slope of the stress-strain curve i Thus,

J1 1

(56a) (56b)

For linear elastic material, Yield criteria 57.

v =2G.

Mohr-Coulomb criteria have been generally accepted as useful

and practical failure criteria in theoretical and applied soil mechanics. In simplest form, these criteria may be stated as

33

c'

+ a'

tan 0'

(57)

where T '= shear stress on the failure plane cohesion

1' = normal stress on the failure plane

0'

angle of internal friction wide application and popularity, the Mohr-Coulomb theory

Despite its

has been the subject of controversy anong soils engineers regarding the experimental determination of c' and

0'

Probably the most severe

criticism is due to the fact that the Mohr-Coulomb theory does not account for the effect of the intermediate principal stress on material strength. Drucker and Prager 4 3 postulated a generalization of the

Mohr-Coulomb failure criteria which includes the effect of the intermediate principal stress on the behavior of soil. derived by them is can be expressed as The yield surface

conical in the principal stress space (fig. 8) and

-I l
C 11l 0"22 -=7c33

34

022

V/

Fig. 8.Three-dimensional represente.tion of Drucker-Prager 4 3 yield surface

'J1

fi where X' and K C

=K= + x

(58)

are physical constants as shown in fig. 9.

472

-r:~A N-'

Fig. 9. 58.

Drucker-Prager43 yield envelope X and K can be related to

.%I., K

Material constants

c' and c
,

0'
and

under special conditions.

The relationship between

0has been derived by Christian

a. Triaxial compression:

ior the following states of stress:

A/ 2 sin 0'( '3(3-sin 0'(59) 6c' cos01 - sin

(59b)
SN(3

b.

Rigid plastic unde.r plane strain conditions: ' = t .(60a)

49

12 tan2 0'

For cohesionless materials, equations 59b and 60b, 58 may be reduced to K

c'

is equal to zero.

Thus, according to

should also be equal to zero, and equation

35

It toI -

has been observed that the value of

J

(7

is not exactly similar

7 )
S-

and the ratio between the two quantities may be desigult

nated by the failure ratio

such that , (2(62)

Using equations 62, 61, and 56a and after simple substitution, rameter y may be expressed in terms of J and A, as

the pa-

=R

(63)

Knowing the value of

and

-p from equations 63 and 56b, respectively,

equation 55 may be expressed in the following form:

(64)

The value of

J1/3

can be eliminated from equation 64 by using its ap-

proximate value as expressed in equation 51; thus,

.I'jT

X'

(e'

1)(5

NV- 2-

(65)

The values of N
respectively, tutive equation
( $ij1
-

and

J1 /3

as expressed in equations

65 and 51,

can be used in equation 44 to obtain the desired consti-

XtOo(e

- 1)_

_ I,
-'

(6

It

should be noted that equation 66 can be expressed in terms of the

36

angle of internal friction by substituting Xt for 0 as indicated in equations 59a and 59b. Also, it can be expressed in terms of the shear
modulus by substituting p = 2G.

37

PART IV:

EDPERIMENTAL DETERMINATION OF MATERIAL CONSTATS

59,

Two types of test are necessary to evaluate the parameters isotropic. compression and triaxial Other tests such as uniaxial strain and plane strain

needed for the constitutive model: shear tests.

shear tests could also be used to develop the model. axial compression, model. Material Parameters a and and plane strain shear tests.

In this study, the material constants were obtained from isotropic compession, triHowever, results from uniaxial strain tests were used to verify the predictability of the

6W.

The material parameters

and

describe the behavior

of the granular material under a spherical state of stress (i.e., isotropic compression). The mathematical expression iivo2.vinv these conThe experimental daat.a and The mathematat ,.,r1,l are sho~nh u(0 eaJ d for the The Values of stants is given in equation 51. in plates 1 and 2, respecti.vely.

ical fit for crushed Napa basalt and Pailited Rock material tests are presented in table 2. Material Parameters and

61. The material constants u and y can be detertiinel from either triaxial compression or plane strain shear stie:oi--strain curves. To conveniently obtain these constants, equttion 53 should bE linearized in the form

+ +
where a 62. is the intercept on the

YA
N

1
2

(67)
axis and y is the slope

of the line. The transformed stress-strain curves for crushed Napa basalt and Painted Rock Dam material are shown in plates 3-6 for triaxial

38

compression tests and plates 7-10 for plane strain shear tests. of a and y obtained from straight lines that bett fit

Values

thi trans-

formed stress-strain curves are listed in table 3 and were used in for-

mulating the hyperbolic function described in equation 55. Comparisons of the calculated stress-strain curves and the experimental curves are
shown in plates llfor triaidal compression tests and in plates 15-18 These plots shov satisfactory agreement for plane strain shear tests.

between the experimental and calculated stress-strain curves, indicating that the proposed rectangular hyperbola reasonably predicts the stress-

strain behavior when expressed in terms of VfI

and -rTFj

63.

The ultimate value of the second invariant of the stress de-

viatoric tensor

(;T

is somewhat larger than the failure value

(If
ure ratio

This wouldL be expected since the hyperbola remains below the

asymptote for all finite values of

Rf
,

42. The
f

relationship between the

failure value and the asymptotic value of tionship between

is defined as the failPlate 19 shows the rela-

as indicated in equation 62.

()f

and (2)Ul

for the crushed Napa basalt

and Painted Rock Dam material under triaxial compression, and plate 20

shows the same relationship for plane strain shear tests.

These plots

indicate that Rf is 0.83 for crushed Napa basalt and 0489 for Painted Rock Dam material when tested in triaxial compression, and 0.59 for crushed Napa basalt and 0.69 for Painted Rock Dam material when tested under plane strain shear conditions.
Material Constant X'

64,

As indicated in equation 61, the material constant

X'

can

be obtained by measuring the slope of the generalized Mohr-Coulomb envelopes as suggested by Drucker and Prager. 4J3 These failure envelopes were constructed by plotting it, as a function of J /3 at failure for

various stress levels as shown in plates 21-24. It can be seen that the failure envelopes are not straight lines passing through the origin,
and the curvature is more pronounced for material tested in triaxial

39

F-...

,-- --- ...... ....... .... ,.. ....

....

compression (plates 21 and 22) than for that tested under plane strain conditions (plates 23 and 24), equal to However, the failure envelopes were X' for the materials tested approximated by straight lines passing through the origin with slopes V', A summary Of the values of is shown in table 4k

Material Parameter 65.

ti

With the exception of the unconsolidated undrained tests on

saturated soil, the initial slope of the stress-strain curve cannot be expected to remain constant -wnder different confining pressures. Such variation in the initial slope it is clearly shown in plates 11-18. The values of P obtained from the inverse of a for the crushed Napa basallt and the Painted Rock Dam material are presented in table 5. 29,30 Previous studies on granular' materials have Indicated that the initial shear modulus varies exponentially with the mean normal stress. Since ji is directly related to the shear mcduLus, ;j will vary exponentially with p and Jl/3 it can be expected

that the value of

J1 /3 for the Larushed Napa

66.

The relationships between

basalt and the Painted Rock Dam material are shown in plat'es 25-28. These plots indicated that the relationship between the two varieules may be approximated by a straight line, resulting in a conyenient expression for u ,

(-- )

(68)

where

and

a are constants whose values can be obtained from the u in equation 68 may be expressed in terms of I using equation 51 as O=C[ e (69)

experimental data (see table 6).

67.

The value of

the first strain invariant

-'T

4o

the stress tensor By substituting equation 68 in equation 66,

be expressed as

aij

may

06"-)'e

(70)

CRCi + X{ieI~11
'where c
,

and

V.' are soil parameters,

II

is

is the second invariant of the deviathe first strain invariant; V is Kyonecker's delta. toric strain; and 4 for evaluating the ma68. Fig. 30 summarizes the flow diagram terial constants used in equation 69.

EVALUATE CONSTARrs

EVALUATE (ONStAN7TS

EVALLUATE CONSTANT

EVALUATE CONSTANT

, sO'FI.kOPIC C OUPIEDISON'

TRIAXIAL OR PLANE STRAI'N SlISARITE.STS

TRIAXIAL IOM'RESSION COPRESSIOI ORPLANE STRAIN fE TTS

TRIAXIAL COMPRESSION ORPLANE STRAIN EATSS i

PLOTDATAIN T1 lISVASMIOB:

PLOTDATA IN THIFASHION,:

PLOTMOHR-COULOUB FAiL'!4 ENVELOPES AS:

PLOTDATA 1% THIS FASHION:

VOMLUEIRIC STRAIN

.I

IL

P. RESSUR.E7-f

L"G T

USINIT LEAST SQUARES METHODD, FIT PLOTTED DATA CURVE TO .p

USING, LEASTSQUARES METHOD, FIT STRAICIIT LINE

ASE FAILULR VELOP EN ASSTRAIGHT LRIE

ASSVVE STRAIGHIT LINE

A0 41 VALUES AND
VALUES, AD FROM CURVE FIT

-SLOPEOFENVELOPE 1
.... i
-

OBTAIN ..
VALUES A40 FRO, a EQUATION LINE OF 1Ij

EVALUIATE R

. I, 'T2el= MAWIAUM1

!=t),

EXPERIMENTAL VALUE

Fig. 10.

Flow diagram for evaluating material constants used in the constitutive equation

42

PART V: 69.

APPLICATION OF THE CONSTITUTIVE EQUATION

In previous Parts of this report, a constitutive equation was In the follow-

derived and the soil parameters associated with it were evaluated for both crushed Napa basalt and Painted Rock Dam material. ing paragraphs, the procedure of examr~ing the constitutive equation is illustrated by using examples of simple states of stresses and deformations. Hydrostatic State of Stress 70. The conditions of a hydrostatic state ot stress (i.e., spheri-

cal state of stress) are

al 1
aij = 0

0
all

0.
0 eiJ 0

0
ei1

0
0

10

'l0

Applying the above conditions in the expression , I.

for and for iJ j e ii
--

j -(1./3)6

yields

ij

I1 3

=-0 =0

iji

Thus, equation 69 for the hydrostatic stress state becomes

, ll oie=1 (e (Ti) (71)

The above equation exists only for the ease in which

1 /3 =0(e1 -

i = J or
The rela-

1) , which is the exact form of equation 51.

tionship between the observed and predicted stress-strain relationships

for the hydrostatic stress state is shown in plates I and 2. A good correlation between the observed and predicted values should be 43

anticipated since the constitutive model was derived from experimental data from hydrostatic compression tests.

Uniaxial State of Strain

71.

The uniaxial state of strain, commonly known as the el 0 0 0 0 E, 0 0 0 0

con-

dition, can be described as:

ij33
0

(3

ij

0 0

Also,

ii1 1

I 1

For

i # j

both

aij

and

iJ

vanish and equation 69 is

equal to

zero for all values of ai , which is in compliance with the conditions of the uniaxial.state of strain. However, for i .j the constitutive equation takes the form

aI

0 e

"-n

2V

0 e

11

(72)

and

033

.....
3C C11+

1) 3?.' IL(e~ell
1-

(73)

The predicted stress-strain relation was :compared with

K
0

test data on

the crushed Napa basalt end the Painted Rock Dam material as shown in plates 29 and 30. The results showed that, even though there is a the

difference in the values of the predicted and observed stresses, results are in agreement in at least a qualitative sense.

72.

The lack of quantitative agreement between the predicted and K test) may

the actual data for the uniaxial state of strain (i.e.,

44~

be due to many factors. normal stress

How~ever, the most, serious :One is the assumption I Iis only the result of the applied mean

that the volumetrit strain

J 1/3 i This assumption is a crude approximation :of the

actual behavior of granular material. Interpretation of Volumetric Deformation of Granular :Material T3. it has been observed that the volumetric deformation of granular material during drained shear ranges from dilatational to compressional, depending on msany factors such as density of material, stress level, itrain condition,.n sist of two components,.: the soil. shape and site of -particles. If secondary effects axe ignored, the total volumetric strain *canbe assumed to conone component is related to the. applied mean normal stress, and the other is due to shear deformations exhibited by

~I ld
vhere IIis the total volumetric strain, I 1Cis the component of I is the comid

volumetric strain due to compressional stresses, and ponent of volumetric strain due to shear deformation.
1 74. The value of

using equation 51. in the laboratory.

I1 can be obtained directly in terms of J /3 le1 can be obtained directly from a The value of I As an alternative, I ldcould be obtained by, shear~-

pure shear test; however, a pure: shear test, is very difficult to perform ing soil under a constant mean normal. stress, which is much ea~sier to perform than the pure sertest. Unfortunately, the experimental date, I are not available at the present time, but it is 1d hoped that such tests may be conducted Qfl crushed N~apa 'basalt and needed to obtain Painted Rock Dam material in the future, 75. tion. An early study by Stroganov 4 IId on sand under plane strain deformation shoved that interpreted by directly related to the shear deformais.

'For three-dimensional problems, Stroganov's hypothesis may be

lld2

S=

f(

)(75)

vhere

f 76.

is some unknown function of V

It

that can be determined from

f is constant,

appropriate experimental data. has been pointed out by Stroganov that as the friability coefficient.

and he referred to it

However, prelim-

inary examination made on crushed Napa basalt, by assigning different values for f as shown in plate 31, indicated that f is not constant

and can be positive or negative depending on the density of the material. Therefore, unless the actual value of f is obtained, it is not

possible to obtain quantitative agreement between the predicted and the .experimental data.

Cylindrical State of Strain

77.

The cylindrical state of strain, commonly known as triaxial

compression, can be defined as

a11

0E

oij

0 0

a33 0 0 33

sij

0 0

C33 0 0 33

Also.,

I,

c.ji +

78.

The same procedure can be used to Obtain the components of In the conventional triaxial test, a$3 is usually P ; the only stress component is

the stress tensor. 011

constant and can be designated as

According to equation 69, the major principal stress and devi-

atoric stress Can be expressed as

46

011~

-*I (*

3 1f+%3 ::

-t33

(7)

33

" ,

11*" 33

S:f:'1
.

r U

n2F 33 )

3"3)

(77) 77

13 < 3 gci 2C e<>L Pi c, 33 [e+ 0:

Predicted and Experimental Correlation for the

Cylindrical State of Strain

79. pressed: as

exFor the cylindrical state, the mean normal stress can be

31(

33) + '

(78)

Combining equations 78 and $1 yields

o'6(4

(1

03

033

(79)

By substituting equation 79 in equation T7, the resulting equation may

be expressed as (a01
(al
)L
.

33)
VT 33 -f 3)

33~
(,13

33)

(80)

33)

-1 33) +, ,331

( 11033 + a3

..

(a

- a 3)

It should be noted that the experimental data for the conventional triaxial tests (i.e.., cylindrical state of strain) were obtained under constant value of c33 * Thus, by incrementing all , the corresponding values of 80. 33) can be generated. (ell - C The correlation between the experimental stress-strain curves

47

and those predicted by equation 81 is shown in plates 32 and 33 for the Painted Rock Dam material and in plates 34 .and 35 for the crushed Napa basalt. Once again, the predicted and the experimental results are in reasonable agreement in a qualitative sense.

Plane strain State 81. For plane strain deformation, the stress and strain ten-

sors may be expressed by o 0 a1 0 a22 0 0 033

=

0 0 0 0 33

1 C 1 + C33

2; +(

+ C-ic3

The above equations can be substituted in equation 69 to obtain the governing equation for plane strain deformation.

48

PART VI:

CONCLUSIONS AND RECOMMENDATIONS

82.

The purpose of this study

was twofold:

first, to summarize to

the methodology and procedures used in modeling soils; secondly,

develop a mathematical model4.hich can describe and predict the nonlinear behavior of granular soils. 83. Based on the literature review it was found that the majority and hyperbolic) used

of soil models (i.e.,

linear, bilinear, trilinear,

in numerical techniques such as the finite ,element method are based on theories of elasticity and curve fittings.* While these models provide good agreement between the observed and predicted .soil behavior under restricted conditions, they cannot be used to predict stress-strain

behavior for other than those conditions from which they were derived. Consequently, they cannot be classified as constitutive models. Con-

stitutive models based on higher order elastic continuum are probably the only models which realistically represent material behavior. ever, the procedure used in developing such models is How-

difficult from

the analytical as well as the ,..,experimental-point of view.

84.

A nonlinear elastic constitutive model was derived for tw6 crushed Nap basalt and Painted Rock Dam material.

granular materials:

The derived model was based on the following experimental observations: a. The hydrostatic stress-strain curve for granular soil can be approximated by exponential function relating J1 /5 and Ii1 The magnitude of stress prior to failure for the soil tested is a function of the total strain. The stress-strain relationship as expressed in terms of J7 can be approximated by a rectangular Sversus hyperbola for both triaxial compression and plane strain shear. The failure points for 'both crushed Napa basalt and Painted Rock Dam material fell on the Drucker-Prager failure envelope. The portion of the stress-strain curve beyond the failure point could not be accounted for; thereforel, material which exhibits strain softening characteristics cannot be approximated by the proposed constitutive model.

b. c.

d.

e.

49

85.

The proposed model is the simplest type of nonlinear condoes not account for shear-dilatancy phenomena.

stitutive model, and it

Therefore, while there is a qualitative agreement between the predicted and actual material behavior, the quantitative agreement needs to be improved. Thus, a higher order constitutive equation, which includes It is also recommended that the shear-dilatancy phenomena, should be studied in order to significantly improve the accuracy of the model. analytical and experimental research be continued to include a plasticity model in an effort to improve the existing knowledge with regard to nonlinear behavior of soils.

50

LITERATURE CITED

1. Al-Hussaini, M. M., "Investigation of Plane Strain Shear Testing; Tests on Crushed Napa Basalt," Technical Report S-71-2, Report 2, Jun 1971, U. S. Army Engineer Waterways Experiment Station, CE, Vicksburg, Miss.
2. "Investigation of Plane Strain Shear Testing; Plane Strain and Triaxial.Compression Tests on Painted. Rock Dam Material," Technical Report S-71-2, Report 3, Sep 1972, U. S. Army Engineer Waterways Experiment Station, CE, Vicksburg, Miss.
___

3.

.......__

"Investigation of Plane Strain Shear Testing; WES High-

Capacity Plane Strain Shear Apparatus," Technical Report S-71-2, Report 1, Mar 1971, U. S. Army Engineer Waterways Experiment Station, CE, Vicksburg, Miss.
4.

'The Influence of End Restraint and the Method of Consolidation on the Drained Triaxial Compressive Strength of Crushed Napa Basalt," Miscellaneous Paper S-70-18, Jun 1970, U. S. Army Engineer Waterways Experiment Station, CE, Vicksburg, Miss.
_,

5.

Hetenyi, M. I., Beams on Elastic Foundation, 7th ed., University of Michigan Press, Ann Arbor, Mich., 1964. Yoder, E. J., Principles of Pavement Design, 2d ed., Wiley, New York, 1963.

6.

7. " Girijavallabhan, C. V. and Reese, L. S., "Finite-Element Method for Problems in Soil Mechanics," Journal, Soil Mechanics and Foundations Division, American Society of: Civil F~nsineers, ol 94, No. SM2, Mar 1968, pp '73-.496. 8. Duncan, J. Mi,, Monismith, C. L., and Wilson, E .L., "Finite Element Analyses of Pavements," Soil Stresses and Pavement Element Analyses.
Highway Research Record No. 228, pp 18-33, 1968, Highway Research

Board - National Research Council, Washington, D. C. 9. Michelo, J., "Analysis of Stresses and Displacements in an N-Layered Elastic System Under a Load Uniformly Distributed on a Circular Area," Internal Report, Sep 1963, California Research Corp. Duncan, J.. M. and Dunlop, P., "Slopes in Stitf-Fissured Clays and Shales," Journal,. Soil. Mechanics and Foundations Division. American Society of Civil Engineers, Vol 95, No. SM2, Mar 1969, pp 467-492. D'Appolonia, D. J. and Lambe, T. W., "Method for Predicting Initial Settlement," Journal, Soil Mechanics and Foundations Division. American Society of Civil Engineers, Vol 96, No. SM2, Mar 1970,

10.

11.

pp 523-544.
12. Dunlop, P. and Duncan, J. M., "Development of Failure Around Excavated Slopes," Journal, Soil Mechanics and Foundations Division, American Society of Civil Engineers, Vol 96, No. SM?, Mar 1970, pp 471-493. 51

. .. ...

13.

Ellison, R. D., D'Appolonia, E., and Thiers, G. R., No. SM4, Apr 1971, pp 661-678.

"Load-

Deformation Mechanism for Bored Piles," Journal, Soil Mechanics and Foundations Division. American Society of Civil Engineers, Vol 97, l1. Desai, C. S. and Abel, J. F., Introduction to the Finite Element Method, A Numerical Method for Engineering Analysis, Van Nostrand Reinhold Co., New York, 1972. 15. Kondner, R. L., "Hyperbolic Stress-Strain Response: Cohesive Soils," Journal, Soil Medhanics and Foundations Division, American

Society of Civil Engineers, Vol 89, No. SMI, Feb 1963, pp 115-143. 16. Kondner, R. L. and Zelasko, J. S., "Void Ratio Effects on the Hyperbolic Stress-Strain Response of a Sand," Proceedings, Second Pan-American Conference on Soil Mechanics and Foundation Engineer-

ing, Vol 1, 1963, pp 289-333.

17. Hansen, T. B., Discussion of "Hyperbolic Stress-Strain Response: Cohesive Soils," Journal, Soil Mechanics and Foundations Division. American Society of Civil Engineers, Vol 89, No. SMh, Jul 1963,

18.

pp 241-242. Duncan, J. M. and Chang, C. Y.,

"Nonlinear Analysis of Stress and

Strain in Soils," Journal, Soil Mechanics and Foundations Division,

American Society of Civil Engineers, Vol 96, No. SM5, Sep 1970, pp 1629-1653.
19. Chang, C. Y. and Duncan, J. M., "Analysis of Soil Movement Around a Deep Excavation," Journal, Soil Mechanics and Foundations Divi-

sion, American Society of Civil Engineers, Vol 96, No. SM5, Sep 1970,
pp 1655-1681. 20. Kulhawy, F. H., Duncan, J. M., and Seed, H. B., "Finite Element Analyses of Stresses and Movements in Embankments During Construction," Contract JReport S-69-8, Nov 1969, U. S. Army Engineer Waterways Experiment Station, CE, Vicksburg, Miss.; prepared under Contract No. DACW39-68-C-0078 by College of Engineering, Office of Research Services, University of California, Berkeley, Calif. Palmerton, J. B., "Application of Three-Dimensional Finite Element Analysis," Proceedings of the Symposium. Applicatibns of Finite Element Method in Geotechnical Engineering, Vicksburg, Miss., Sep 19T2, pp 155-2214. Ahlberg, J. H., Nilson, E. N., and Walsh, J. L., Theory of Splines and Their Applications, Academic Press, New York, 1967. Cheek, J. B., Jr., Radhaerishnan, N., and Tracy, F. T., "Application of Spline Interpolation Methods to Engineering Problems," Miscellaneous Paper 0-71-2, Jul 1971, U. S. Army Engineer Waterways Experiment Station,- CE, Vicksburg, Miss. Greville, T. N. E., "Data Fitting by Spline Functions," Technical Summary Report No. 893, Jun 1968, U. S. Army Mathematical Research Center, University of Wisconsin, Madison, Wis.

21.

22. 23.

24.

52

25.

Desai, C. S., "Nonlinear Analyses Using Spline Functions:," Journal, Soil Mechanics and Poundatiohs Division, American Society of Civil

26.

Engineers, Vol 97, No. SM10i:, Oct 1971, pp i4614480. Clough, R. W. and Woodward, R. J. IIi, "Analysis of Embankment Stresses and Deformations:," Journal, Soi.iMedhanics .and Foundations Division, American Society of:Civil Engineers, Vol 93, VNo. SM4, Jul 1967, pp 529-54:9. Ko, H. Y. and Scott, R. F., "Deformation of Sand in Hydrostatic
Compression," Journal, Soil Mechanics and Foundations Division American Society of Civil Engineers, Vol 93, No. SM3, May 1967,

27.

pp 137-156.
28. El-Sohby, M. A., "!Elastic:Behavior of Sand,:" Journal,. Soil MePp .1393-1409.
chanics and Foundations Division, American Society.of Civil

Engineers, Vol 95, No. :SM6, Nov 1969,

29.

Domaschuk, L. and Wade, N. ., "A Study of Bulk and Shear Moduli of a Sand," Journal, Soil Mechanics and Foundations: Division,

American Society of Civil Engineers, Vol 95, No. SM2, Mar 1969,
pp 561-581. 30. Al-Hussaini, M. M. and Radhakrishnan, N., "Analysis of Plane Strain Tests Using the Finite Element Method," Proceedings of the Symposium, Application of the Finite Element Method in Geotechnical Engineering, Vicksburg, Miss., Sep 1972., pp 215-257. Clough, G. W. and Duncan, J. M., "Finite Element Analyses of Port Allen and Old River Locks," Contract Report S-69-6, Sep 1969, U. S. Army Engineer Waterways Experiment Station, CE, Vicksburg, Miss.; prepared under Contract No. DACW39-68-C-0040 by College of Engineering, Office of Research Services, University of California, Berkeley, Calif. Girijavallabhan, C. V. and Mehta, K. C., "Stress-Strain Relationship from Compression Tests on Nonlinear Materials," Proceedings of the Symposium, Application of Finite Element Methods in Civil

31.

32.

Engineering,.Nov 1969, pp 457-480.

33. 34. 35. 36. Eringen, A. C., Nonlinear Theory of Continuous Media, McGraw-Hill, New York, 1962. Prager, W., Introduction to Mechanics of Continua, Ginn and Co.,

Boston, 1961.
Truesdell, C. A., "Hypoelasticity," Journal of Rational Mechanics and Analysis, Vol 4, No. 83, 1955, pp 83-133. Rivlin, R. S. and Ericksen, J., L., "Stress-Deformation Relation for Isotropic Material," Journal of Rational Mechanics and Analysis.

Vol 4, 1955, pp 323-425.

37.

Rohani, B., "Constitutive Equation for Continuous Media" (in preparation), U. S. Army Engineer Waterways Experiment Station, CE, Vicksburg, Miss.

53

38.

39.

Rohani, B., A Nonlinear Elastic-Viscoplastic Constitutive Relationship for Earth.Material, Ph. D. Dissertation, 1970, Texas A&M University, College Station, Tex. Chang, T. Y. et al., "An Integrated Approach to the Stress Analysis of Granular Materials," 1967, California Institute of Technology, Pasadena, Calif. Nelson, I. and Baron, M. L., "Investigation of Ground Shock Effects in Nonlinear Hysteretic Media; Development of Mathematical Material Model," Contract Report S-68-I, Report 1, Mar 1968, U. S. Army Engineer Waterways Experiment Station, CE, Vicksburg, Miss.; prepared under Contract No. DACA39-67-C-0O48 by Paul Weidlinger, Consulting Engineers, New York. Coon, M. D. and Evans, R. J., "Recoverable Deformation of Cohesionless Soils," Journal, Soil Mechanics and Foundations Division, American Society of Civil Engineers, Vol 97, No. SM2, Feb 1971, pp 375-391. Timoshenko, S. and Goodier, J. N., Theory of Elasticity, 2d ed., McGraw-Hill, New York, 1951, p 372. Drucker, D. C. and Prager, W., "Soil Mechanics and Plastic Analysis of Limit Design," quarterly of Applied Mathematics, Vol 10, No. 2, 1952, pp 157-175. Christian, J. T., "Two-Dimensional Analysis of Stress and Strain in Soils; Plane-Strain Deformation Analysis of Soil," Contract Report 3-129, Report 3, Dec 1966, U. S. Army Engineer Waterways Experiment Station, CE, Vicksburg, Miss.; prepared under Contract No. DA-22-079-eng-471 by Massachusetts Institute of Tecbnology, Cambridge, Mass. Stroganov, A. V., "Plane Plastic Deformation of Soil," Brussel Conference on Earth Pressure Problems, Vol 1, 1958, pp 94-103.

h0.

41.

42. h3.

hh.

h5.

54

CO
(Y)

0+ .0 CMj

+ TA

+ rCH

"-il
I

C)

P4 +

Cj

CCM

HCID 4-4
cMi

rA+

.. .. ... ..... . ...

.. .... .

..

...

....

Table 2 Summary of

and

for the Material Tests f

Material
Constant

Crushed NApaa 0 Basalt D=10 Dr Dr 100

percent percent

Painted Rock Dam Material D =TO1 100 D = 70 r r

percent percent

47 66.6 36.3

18 133

18 94

Q4

u'- N N-~? t- 0000 o 0"000 NA $4 0001 0

00 U'H 0 0

0 N~ 0

000 00 a0 0.

0 0 OxLA -= (Lx IN 00

o
Wo o oo

o.*
o

o
0.0 A" .

oI
0 0

4t

0 . 0

. ,0

00 0 0 0 0 0

.0 0 0

0 0 00 0

.. 0 0

( C4

000 0 0 0

00 t t

t . Io.I . .o o +) 0 00 o , 0 1 \'D C r- ,- 0\ o 0 oe. 1o0 t\ M I o 000 o , 1C 9 o 0 2 rj~,000 94 0o 9, 0o 0 00 0

4) P 00 0 0
s-I M 1-

o c
\0
r\, 0 0Ut\ H

N~ o 00 0 0 0

o0 8 H0 0 0 o3o 0 0
00

NC

JH

H4 UN0 N o 0 0 0 0 00 C 000

0c.4

H o

a) H

(d Po 0 tH 0

TI4) 4) U A.~ kHm $ 44 t d) m .4r P4 000 o ,-00o U 0 0 0 0 DC 0 00 00

.- 5 5u5

(nl m 0co0 0 1 0 .t H .- '.: It0 - (m 1 0 C000 0 0 0 0 000

\o CO 0O 0 0

0 0 0 0 000

&q to Ch "o 00 00 n 00 00

0 t N m~ OS' 0 j00 00 0 000 C

mn

mn

0 *10

LP'\ 0

Nl

n\

0 H UN U

\0

LU 0

N~ 0 H n

N -t

tf\

0 0

f'%

0 I'D

Ur\

1-

0 0 In

0 Ur\ 0 110 N~ 0 H4 m

50

E-4$4

-H

Table 4

Summary of

A'

Values
Painted Rock Dam Material

Crushed Napa Basalt

Dr =100" Type of Test percent

Dr =70 percent

Dr =100 percent

Dr

70

percent

Triaxial
compression

0.87
0.87

o.814
0.80

o.94 0.88

0.89 0.82

Plane
strain shear

Table 5

Summary of

Values
Paintea Rock Dam Material

Crushed XNapa. Basalt

D Type of Test Triaxi al compression a33 6o 125 300 425

500

100:

= 70

Dr =100 percent 28,570 32,260 55,550 66,66o

--

percent 10,750 19,610 23,800

--

percent 11,490 14,080 16,390

--

Dr = 70 percent 13,330 31,250 45,450

45,450
--

30,300

19,600 4,460 7,870

Plane strain shear

60 125

9,090 10,990

114,920 21,280

13,890 19,61o

300

T8,547 14,7o

25,000

23,26o

~41P
4 l 0 II

4]N
I

0 c
-

j ~ 0:~ 4)$~

0to

co

14 A4
C'1
f-4

k\

a-

V) CL

u._.1

"Z

wt oaz

0--

wJ tj C

I I

DJ

N " _""___(n

..

\LI

K~0.

I.-

I4Li if 38nss--d d ONINIAN6-3

PLATE 1

.~~

~~~~~~~~~~~~~ :'' 'T'7?

""""7

/- ---------

- --

>c
o ~

0 ~
z

, I.c

7
w1 a U

494J

.. N

PLT. ' 6.
,:, .4.

"

hI

I:'I.,A.

2.6.

I
CL /"0=.OO005I: 7.0,002746

Y=0.00i4t

"Y=O. 0044

/:0.00086

0l.4

0.0

0.04

0.04

0.0.

0.10

0.

0.14

0.14

T RANSFORMED STRESS- STRAIN CURVES FOR CRUSHED NAPA BASALT UNDER TRIAXIAL COMPRESSION Dr 100 PERCENT
PLATE 3

V I:O.OO t

I-- - -.--

o.o .. .. ......
V:.00/45

" I

Y= 0.0000061

:a !oooo
a:o.oooos,/

..

It

Is 2

..

..

V l...

..

1.0

0.6

00

f/ 0K .461%01
-00

TRANSFORMED STRESS-STRAIN CURVES FOR CRUSHED NAPA BASALT UNDER TRIA : IAL COMPRESSION
Dr= 7 0 PERCENT

PLATE 4

V:0-.0057

V0=.o0276

V:o.00136
/.

/ "o.oo~o..
W/O

IY=

1.4

0."

.0.4

_______

______

0,2

________

0 .05

0.10

0.15

0.20

0.25

TRANSFORMED STRESS- STRAIN CURVES FOR PAINTED ROCK DAM MATERIAL UNDER TRIAXIAL COMPRESSION
Dr=I O0 PERCENT

PLATE 5

.
a-00005
.42

0,0

- ro. 0.00002

/ )

c'-~o -o/o z
/ ,o.0002

.oo

0.8

0,05

0.0

0.15

0.25 0.20

TRANSFORMED STRESS-STRAIN CURVES FOR PAINTED ROCK DAM MATERIAL UNDER TRIAXIAL COMPRESSION
Dr70 PERCENT

PLATE 6

:~~~

~~ ............ . .. ...

. ..

.,

:i..

G3- 300

/ =0.0007
-.

.-.

000o068_

1/ j,,

,/-

-0 0032

S0.0

0.10

0.12

0.14

is

TRANSFORMED STRESS-STRAIN CURVES FOR CRUSHED NAPA BASALT UNDER PLANE STRAIN $HEAR Dr-- 0 0 PERCENT
PLATET

2.4

2.0

7:0r.004/9 a-=0O.000224
04c

3000/

0.02

0.04

0.0

0 8

016 0.10 o it

0.14F.I

TRANSFORMED STRESS-STRAIN CURVES FOR CRUSHED NAPA BASALT UNDER PLANE STRAIN SHEAR
Dr.7. PERCENT PLAT~E 8

l~cJ=60 IV:O.OOJa az Loo.oo oo67

.........

-!

7O.OOoo84S

0.4

4.....

0.0a

O.A

0.02

0.04

0.08

0.10

0. )z

0,14

i.s1

TRANSFORMED STRESS-STRAIN CURVES FOR PAINTED ROCK DAM MATERIAL UNDER PLANE STRAIN SHEAR
Dr= IO0 PERCENT PLATE 9

...... ..... .:

........ .... . ..... .... ..

2.4

ci Z60:
SVi

0.oooor,5

a-z0.00004J

1.8

1.4

0.02

0.04

0.04

0.04

0.10

011

0.1

014

TRANSFORMED STRESS- STRAIN CURVES FOR PAINTED ROCK DAM MATERIAL UNDER PLANE STRAIN SHEAR 0 r-7 0 PERCENT
PLATE 10

0304

I000

/'
8000

.oooooa~,OOO2SO.OO/O.0003'O0420001

100

0..

1j,=-. L

0.04

OJ .08

3 a.

0.20

0124

0. a

0.32

VARIATION OF

1TrJ' WITH 4I'T FOR CRUSHED NAPA BASALT IN THE TRIAXIAL COMPRESSION TEST
Dr-O0 PERCENT PLATE 11

300

477
i'03100, 0 3 /

tO0

..

...

.. .

.....

too-

Soo

' "

7007
,0 0
0 0 0 0 5 7

#"003swvi

OO
0.0,=

LEGENO
,08 0.12 0.' .= ---4* 4-0 0. =9,7, 0.: 0,5 PERiMEiNTAt DATA CALCULt AT D DMAl

VARIATION OF

4Jj WITH 4Tl

FOR CRUSHED NAPA

" BASALT IN THE

Dr= 7 0 PERCENT

TRIAXIAL COMPRESSION TEST

PLATE 12

0.O000f5

0,001031-

goo

oo , , !

i f ,

//2
,._ oo275-,,7 .- =0. 0000 31 0. /

=o0
oe

.0o0- s"0J57--

"

-s60

LEGEND
-0 0.0 0.10 i O..5 EXPERWlENTAL. DATA CALCULATED DATA 0,20 0.25

VARIATION OF

4Jz WITH WIj

FOR PAINTED ROCK DAM MATERIAL IN THE
TRIAXIAL COMPRESSION TEST Dr= f O0 PERCENT PLATE 13

Soot

bc:
aado

Al

300

a..oooo,,

, 056

---,---e--CALUL{ArTED

OATA

VARIATION OF

-,rJ

WITH -IIfj

FOR PAINTED ROCK DAM MATERIAL IN THE TRIAXIAL COMPRESSION TEST

Dr= 7 0 PERCENT

PLATE 1

8000

OOO00681O.OO0?77'.

.,

400

400

i0o

UK0RIMEML

DATA

C.ALCUL.ATOD ATA

00,.2_ _

VARIATION OF FOR CRUSHED:NAPA BASALT IN THE PLANE STRAIN SHEAR TEST

.Or=100PERCENT PLATE 15

................................ . ...

7OO

o.o~ot/#ogooO92V
...
-

0 '" "' :___

Soo 400O....

_______

. .0 0.....40,049..1

LEGEND
--.

0 &

#PERltM4tTM.

DATA CALCULATED DATA

0.05

9.1.0

0.15

0.2O

0.25

VARIATION OF
-O

WITH -11-

FOR CRUSHED NAPA BASALT IN THE PLANE STRAIN SHEAR TEST

7 Or-.= PERCENT 0

O000

0.0

900

zoo

:J-3

/..

o oooo4 oooola

-,It;

0~.00006710003Z-4

LEGEND
-r---CALCULATED DATA

0.,6

020 .

..

VARIATION OF
-41 WITH -Ar~ FOR PAINTED ROCK DAM MATERIAL IN THE PLANE STRAIN SHEAR TEST

Dr= IO0 PERCENT

PLATE 1T

O00

taoozjooooy
j..oo-'_

!
300

.b
/

0.000051 40.0025.97;

100
2C

6 //

0.000072 '00-6610 T-

S--

LEGEND

0__
0

. '
0.05

_
0,

--

I
0.1s

XPERI)ALNTAL DATA CALCULATEO OATA

0.20

0.20

"VARIATION OF
SWITH

4Tl

FOR PAINTED ROCK DAM

MATERIAL IN THE
D=70 PERCENT ""

PLANE STRAIN SHEAR TEST

PLATE 18

8000

/
Soof
600 -

400

0 D=IO0 PERCENT A Dr10 PERCENT

LEGEND

200

200

400

600

goo

1000

1200

1400

a. CRUSHED NAPA BASALT

800

to

,400

00

200

400

600

800

1000

0_

1400

b. PAINTED ROCK DAM MATERIAL

RELATIONSHIP BETWEEN

(452 )f. AND (457),It IN THE TRIAXIAL

COMPRESSION TEST PLATE 19

1000

8000

400___

A0l,

70r:AC0P4Tr

200

.400

f0c0

boo

1000

1.200

1400

a.CRUSHEF-D

NAPA BASALT

000

200-

200

400

c00

800

1000

1200

11400

b. PAINTED ROCK DAM MATERIAL

RELATIONSHIP BETWEEN

(-]J',)f AND
IN THE PLANE STRAIN SHEAR, TEST
PLATE 20

too

Dlmix 0 ' 0 PERCENT 7

200 =0

4000

200

400

600

800

1000

Iz00

VARIATION OF -,rJ,'WITH J
AT YfIELD FOR CRUSHED NAPA BASALT IN THE

TRIAXIAL COMPRESSION TEST

PLAM~ 21

2e000

-_

800

SD," PEO
600

"OO

PERNENT

7?PR.

/'
400
-_________J __-

)/
400 ro60 1000 1200

200

VARIATION OF

WITH r2 3 AT YIELDFOR PAINTED

ROCK DAM MATERIAL IN THE TRIAXIAL COMPRESSION TEST

PLATE 22

$200

4000

....

t--O

PERCENT ,

S./ ,
0o
400

Dr'70 PERCENT
. 0So100,0 Az0.80 . .

40

200

200

400

600

800

I000

1200

VARIATION OF J WITH 3 AT YIELD FOR CRUSHED NAPA BASALT IN THE PLANE STRAIN SHEAR TEST

PLATE 23

42 0 0

. ..

..... . .....

Soo

Or-zlO0 PERCENT

.:

Or= 70 PtERNT

"

=0, 82

2600

//

200

400

S00

B00

1000

1200

VARIATION OFr -...

WITH J-

AT YIELDFOR PAINTED ROCK DAM MATERIAL IN THE PLANE STRAIN SHEAR TEST

PLATE 24i.

100,0001 90,0001 ao,,ooo

70,00

71

'

---

60o,000500600-

"

40,000

20,000

2,000

go20

30

40

50

60 70 80 00100
Ji

200

300

400 500

VARIATION OF JJ.WITH FOR CRUSHED NAPA BASALT UNDER TRIAXIAL COMPRESSION

PLATE 25

7,OO

--

______

70,000Q60,000

300,000. 8.0,00

:3o000o

--

...

..
...

.
..

- .

..
-

.. . .
-

..'

'
"-::.. ..
-

---

20,000
20,600C

. .

8,$000 9,000 s,000

I
3,000

Z'000

1,000

1020

30

40

so

60 70$0 000100

200

300

400

Soo

VARIATION OF JL WITH J FOR CRUSHED NAPA BASALT UNDER~ PLANE STRAIN SHEAR
VT Amv~ 34

90o00o
60,000 90,000. .-.-

'.,--. ,*-

... -

-...-

--

$,_000

3-0,000

D....

.PoRCC...70

PERCENT

's,000

ii

S,6wo
: ;4,000.

"

Si~i~i

I0

2,0

30

40

50

60 10 60,90100

20O0

300

400

500

VARIATION OF JL WITH

FOR PAINTED ROCK DAM MATERIAL UNDER TRIAXIAL COMPRESSION PLATE 27

9o ~~o o . . ...... o

.. ........ ....... ..

iI

......

;00000-

--

80,000 70;000 ao, ooo

*

~
, ...
-

-I Im

... . ___

- .

S0,600~

.--

,4~o0

"1

...

r or --0'0PERC rN T

..

40,000
39,000-

dooo

8,000

10

040

60 70 8090100

200

300

400 SW

J1

3"

SII

VARIATION OF )I WITH

J'

FOR PAINTED ROCK DAM MATERIAL UNDER PLANE STRAIN SHEAR

1000

600

700

400

-J

2 o00 -

I
ID
=0,:70 PERCENT

LEGEND

too

r2
l '"

i
0

PAED;CTED CURVE
EXPERIM*ENTAL DATA

~~~

Oo

002

C.64

0.06

o. o5

0.1 ro

o ?Z

0 . .14

'O,16

MAJOR' PRINCIPAL STRAIN Ell

EXPERIMENTAL RESULTS

PREDICTED AND

S~STATE

FOR THE UNiAXIAL OF STRESS FOR THE CRUSHED NAPA BASALT PLATE 29'

/1.
900
__ _

Booo!.
?000

.
i-/,t.O

.
t4U

tl

...

100

'

LEEN

-100

PREDICTED CURVE EXPERIMENTAL DATA

0.01

0,

0.03

0.04

:0.05 ST"RAIN:, CII

6.66

0.0,7

:046B

MAJORt PRINCIPAL

PREDICTED AND EXPERIMENTAL RESULTS FOR THE UNIAXlAL STATE OF -STRESS FOR THE.PAINTED ROCK DAM M.ATE RIAL

SPLATE 30

... ....

o.o...... .........

.. . ................. . ...

..... .. .....

900

-f--O./

oo B700

fzo.

XP(P1MEVTAL CuRVC
(FROM NA T 29)
-

u
E. Soo ____ ___

4(

400
600

.....

..

400

I
,I. , , . ..

200

0.02

C 04

0,06

008

0,.10
STRAIN Eil,

0. 12

0.

14

O. 16

MAJOR PRINCIPAL

:EFFECT OF THE FlUNTION f .ON THE UNIAXIAL STRESS-STRAIN

REL ATI ONSHIP FOR

CRUSHED NAPA BASALT Dr- 100 PERCENT PLATE.3

1600

-.-

SCO 1,4-00

,r425

200

;' . ao

/ Io
/
,/

-J

-- O

U"

/
..
"_
-

S"I

Sgoo

I J
700---

600

S/
_ _ -

,_

800

400

30o"-

.1.

R0 -

A
----.___

LE GEND
PREOICTED CURVWE EXPERIMENTAL.ICATA 4

/
100
-

Is 20 25 30 PRINCIPAL STRAIN 0lfrERE.NCE

35 40 3PPERCENT

45

s0

PREDICTED AND OBSERVED STRESS-STRAIN CURVES FOR PAINTED ROCK DAM MATERIAL UNDER TRIAXIAL COMPRESSION
Dr= PLATE 32
10 0

PERCENT

17 00

--

1600-

1400

____

In

I
-

S..i/azoc
-So

-o-

~~~~t .... S~oo

'l1O

.9"

.. = 2

4-n

.. . . . . .

:.

.700

600

-0

zoo
to4---4

SI
0 5 t0 is 20 25

.LEGEND

PREDICTED CURVE EXPERIVENAL, DATA

I
AS so

3O ElfI 33

35

40

PRINC IPAL STRAIN DIFFERENCE

PERCENT

PREDICTED AND OBSERVED STRESS-STRAIN CURVES FOR PAINTED ROCK

DAM
MATERIAL UNDER

TRIAXIAL COMPRESSION
RDNF7, PERCENT

PLEATE33

i7000I600 -..
....

a..

I5

ooc

..-....

/I
13600 v
-o

-.

L.G.N.

700

. PRIoo . .
/

"

R"

......

. 0'

....

6 00:

//1
,..

300

"

00

"

Soo

300,.

NAPLALGEND
to EXEI.~oi DAA

1 Is

z 20

2O 5

30

36

.40

45

so

PRINC IPAL :STRAIN DIFFI'RENCE" (Ell E330 PE'RCENT

PREDICTED AND OBSERVED STRESS-STRAIN CURVES FOR CRUSHED NAPA BASALT UNDER

TRIAXIAL COMPRESSION
Dr=100 PERCENT

PLATaE 374

.170o

1600

4 1 00
1500

--

I
"'-

"~--

1300

:,-

$!Do-

--...

w 1200

z/

: .. oo
w

~~Go

le,

% Jo

S-

/.I400

4*...

I:,-

7300

:i

i=O

200

-'

LEGEND
'PREDICTED CURVE

".

--.

EXPEWI4ENTAL DATA

30 5 r I20 I PRINCIPAL STRAiN O;FFERENCE f.3,1,

35

40

45

0so

PERCENT

PREDICTED AND OBSERVED STRESS-STRAIN CURVES FOR CRUSHED NAPA BASALT UNDER TRIAXIAL COMPRESSION
Dr=
70

PERCENT PLATE 35

LctI I ;t C

eifd ion

DOCUMENT CONTROL OATA - R & D

IoNJAIRA TI,,5 AC rIVI C, (Cotpoet.t. Inh.ot) . . 2a. REPOR1T SECURIty CL:ssFrCe&hIo.

11. OSC.

iVmE

nOTeS

( ),p.

af OPAT.d

i,,eio. r.1

datlo)

5,

Au

ToORISI (first

earn.. RidS!e

inh*MC-(lea CorS)

A.RTROS

,T iY .i"
nJ.
CON TRAT ON SOAR T NO.

D ATE

l. tTAL.

No; OF

PAmGES

--

0P NO.

OF REV,

100
Ba. ORIGINS, TOWS REPORT I'I,MIJKERISI

, PR.oJE1"No. hA~tillOIA931

liscellsneous Paper S-Ti-1

..
0O.OI$RI'SU,TIONk STATAE[NT

Approved for public release; distr'Ibution unlimited.

A St 31M.W ELAMR CNO5TET12. SPOFOOITI MICITAMY ACTIVITy

Assistant Secretart Washington, D. C.

Ii. ABSTRACT

of the Army (R&D)

In the past or until recently the nsaority of stresss-de.formsation and stability snalyses

been restricted tO ideal material: behavior. Such idealizations in raateris properties and CIImeIricLl conditions may lead Adirvered feoormsron and pdedittedbehav~or. Renlisti stress and analyses of homogeneous earth masses or aoil-structu-e interaction problems using numerical techniqses such af the finite el(&eet ae finite dnilfe rncenmetho require the foreulalion of a constitutive model for the s0il and struc-

huve

tural materials. A literature reviev made in this study indicated that most procedures used in modeling soils are based on the theory of elasticity azd curve fitting. (This study is limited to constitutive models vhich are based on theory of elasticity. ) Linear, bilinear, trilinear, and hyperbolid nodels provide, under sj'ecial Conditions, good agreement between observed and prsdicted 'soil behavior. Unfortunately, these models lack sufficient erperimental and theoretical verification to be qualified as cons.itutive models. A general constitutive model should predict or reproduce soil behavior under any state of stress and not be restricted tothe state of stress from which it Is derived. Constitutive models based on higher order elastic continuus art probably the only hope for generating truly representative material models. However, the procedure used in obtaining the needed parameters for such models is very difficult if not impossible unless some simplified assumptions are made. A nonlinear elastic constitutive relationship was developed for two granular materials: crushed Napa basalt and Painted Rock Dammaterial. The behavior of the material was assumed to confors with Cauchy elastic material (i.e., the state of stress is only a function of the state of strain); also, the tensorial dilatoacy, which contributes to volume expansion of the material, was ignored. Previous laboratory? data obtained from hydrostatic compression, triaxial compression, and plane strain shear tests on both crushed Napa basalt swe Painted Rock Dammaterial vera used to obtain the needed parameters. The resulting constitutive model was used to predict the stress-strain relations for the uniaxial state of strain ,i.e., KO tests), and the predicted curves were compared with laboratory KC data. The results showed that there is a qualitative agreement between the data predicted by the model and those observed in the laboratory,. Fowever, the quantitative agreement between the predicted and observed data needs to be improved. T.e proposed constitutive relationshlp accounts for nonlinear pressure-volmsetric strain behavior, nonlinear shearing stress-strain behavior, an-d effect of superimposed hydrostatic presture on the behavior of the soils, The Consttlutive relationshIp, however, does not account for shear-dilatascy phenomena often observed during laboratory testing of soils. Therefore, this constitentive model should nc. be expected to predict

Mlfl

SConltinued),

FORM I473 RPPLACES O) fORM 1473. I JAN 64. W111CH IS ,INOV a6s 3 O0SOLETE FOR ARMY USEi.

Tlssil.'d

Security ClIsioa(ic.ton

SS r C

IClasTifiJEOio

Ilip

exncty b~ni~ivh

oi~ficthey rlpiral. to evaluiaitI ic

tmrcLchfl"y Uted eti,'t-kiv

oca Io noe whowver i

oretexmrijlenton-

wov;' i:. reirint

addit-~snal iintleure pal.owa~or;! 'hich

are oeeded to devolop the cnil~cv

IA

KE:Y

ONOS

~LIN. . ..
A wt

.,.
OLE

LIKB wt

LINK C BOLE VT

ROL

Carnt itutive equa.t0 ors Gramnudlr materials

Theory cl*elmstizitty

Security CISs~ificAtIOR

In accordance with ER 70-2-3, paragraph 6c(1')(b), dated 15 February 1973, a facsimile catalog card in Library of Congress format Is reproduced below.

Al-Eassain, Mosaid M A simple elastic constitutive equation for granular material, by M. M. Al-Bassaini. Vicksburg, U. S. Arsry Engineer Waterways Experiment Station, 1974. 1 Y. (various pagings) niUus, 2T em. (US.S. Waterways Experiment Station. Miscellaneous paper S-7-18) Sponsored by Aasitant tSecretary of the AxuV (S&n)J ProJect No. .4A61101A9lD. Includes bibliography. 1. Constitutive equations. 2. Granular mterials. 3. Soils4.4 7eory of elasticity. I. U. S. Office of the Chief of lILearch and Development. (SeriesU. S. WaterwayS Experiment Station, Vicksburg, Miss. Miscellsneous paper S-T4-18) TATW34m n. .S-74-.18