You are on page 1of 111

DISTURBED STATE MODELLING FOR CREEP

RESPONSE OF SUBGLACIAL TILL

by

AMIT PANDEY

___________________________

A Thesis Submitted to the Faculty of the


DEPARTMENT OF CIVIL ENGINEERING AND
ENGINEERING MECHANICS
In Partial Fulfillment of the Requirement for the Degree of
MASTER OF SCIENCES IN CIVIL ENGINEERING
In the Graduate College
THE UNIVERSITY OF ARIZONA

2005
Dedicated to

Uncle Sushil K. Puri


and
cousin brother Mr. Sunil Punetha

May God bless their Soul..


4

ACKNOWLEDGEMENTS

I wish to express my sincere gratitude to my advisor, Professor C.S.Desai for providing

guidance and valuable suggestions and advice during my Masters program at The

University of Arizona.

Special thanks to Dr. D.N. Contractor, Dr S. Merry for their suggestions and for

serving on my committee.

The love, patience, encouragement, guidance and moral support given to me by

my parents, sister and relatives are deeply appreciated. Special thanks to my friend Mr.

Shantanu sane for his encouragement, technical support and help during preparation of

this thesis.

I am grateful to my friends and roommates Dhananjay, Rajarashi, Vineet,

Navneeth, Rohit , Wen, Bridgette and Elizabeth for their kind words and encouragement.

My special thanks to Hemant and Chandana for their love during my stay at Tucson.

I am honored to have a friend like Saurabh, without him it would not be possible

for me to come here for my studies. I would like to thank my B-Tech advisor Prof

T.N.Singh, Associate Professor, IIT-Bombay for his encouragement and support.

The research herein was supported through research grant No. EAR-0229513 by

the National Science Foundation, Washington, DC. This support is sincerely appreciated.
5

TABLE OF CONTENTS

INTRODUCTION .............................................................................................................. 8
TU UT T

1.1
U U Necessity of Present Research ............................................................................ 8
U U

1.2
U U Objective and Scope of Research ....................................................................... 9
U U

1.3
U U Organization of Text ........................................................................................... 9
U U

DEFORMATION OF GLACIAL TILL ........................................................................... 11


U U

1.1
U U Introduction ....................................................................................................... 11
U U

1.2
U U Properties of Till ............................................................................................... 12
U U

1.3
U U Flow Relations for Till ...................................................................................... 14
U U

1.4
U U Flow Relation .................................................................................................... 21
U U

1.5
U U Yield Stress and effective velocity ................................................................... 21
U U

RHEOLOGICAL MATERIAL MODELS ................................................................... 25


U U

1.1
U U Introduction ....................................................................................................... 25
U U

1.2
U U Viscoelastic (ve) Overlay Model ...................................................................... 26
U U

1.3
U U Elastoviscoplastic (evp) Perzynas Model (1966) ......................................... 28
U U

1.4
U U Viscoelasticviscoplastic (vevp) Model ............................................................. 29
U U

THE DISTURBED STATE CONCEPT ........................................................................... 31


U U

1.1
U U Introduction ....................................................................................................... 31
U U

1.2
U U Multicomponent DSC and Overlay Model ....................................................... 33
U U

LABORATORY CREEP EXPERIMENTS, PARAMETER DETERMINATION AND


U

BACK PREDICTION ....................................................................................................... 37


U

1.1
U U Introduction ....................................................................................................... 37
U U

1.2
U U Mechanics of Viscoplastic Solution.................................................................. 39
U U

1.3
U U Results and Discussions .................................................................................... 41
U U

1.4
U U Determination of Viscous Parameters .............................................................. 63
U U

1.5
U U General Procedure............................................................................................. 65
U U

1.6
U U Material Parameters for Perzynas (evp) Model ............................................... 67
U U

1.7
U U Back-prediction of Creep Tests ........................................................................ 75
U U
6

1.8
U U Material Parameters in Overlay Model ............................................................. 85
U U

1.8.1
U U Viscoelastic (ve) Overlay Model .................................................................. 85
U U

1.8.2
U U Viscoelasticviscoplastic (vevp) Overlay Model ........................................... 86
U U

COMPUTER IMPLEMENTATION OF THE DSC AND FINITE ELEMENT


U

ANALYSIS ....................................................................................................................... 87
U

1.1
U U Introduction ....................................................................................................... 87
U U

1.2
U U Results and Discussions .................................................................................... 87
U U

SUMMARY AND CONCLUSIONS ............................................................................. 106


U U

1.1
U U New Contribution............................................................................................ 107
U U

REFERENCES ............................................................................................................... 108


U U
7

ABSTRACT

The existing relations for constitutive behavior of glacial tills are based essentially

on empirical considerations. Due to the lack of well established relations, most analyses

have been based on one of the two simplified assumptions, that the subglacial till is either

a Newtonian viscous material or a perfectly plastic material. The simplified assumption

for till behavior, presented a major insight into actual movement of ice sheets. However,

there is a need for constitutive modeling based on fundamental considerations. Such a

model can play an important role in analysis and prediction of ice sheet movement.

A general concept, called the Disturbed State Concept (DSC) is applied to analyze

the creep behavior of the subglacial till. The DSC is a general theory for material

modelling that has been successfully applied previously to a wide range of practical

problems.

The creep model parameters for the subglacial till are evaluated using data from

laboratory tests on the Material Testing System (MTS) for the stress- strain (CTC)

behavior and triaxial creep device for creep behavior.

The creep model parameters were determined from the creep test data. The back

prediction was done by the one-dimensional procedure. The DSC model with specific

parameters is implemented in a non-linear static finite element program (DSC-SST2D). It

is used to solve the elastoviscoplastic (evp) and viscoelasticviscoplastic (vevp) behavior

for the subglacial till.

Based on the results of this research, it can be stated that DSC with vevp Overlay

model can best describe the constitutive behavior of the subglacial till, thus improving the

understanding of subglacial processes.


8

CHAPTER 1

INTRODUCTION

Improving our understanding of subglacial processes is important, because they

may exert important influences on global climate changes. Research done on the behavior

of subglacial tills has shown that simple rheological models have been used in the past to

describe the deformation mechanism of ice sheets, whereas stress system in the glacier is

very complex. The flow relation has to be generalized to cover such cases. In an elastic

material deformation and stress are proportional whereas in a Newtonian viscous

material, the rate of deformation is proportional to the stress. In a perfectly plastic

material a gradual increasing stress produces no deformation until a critical value, the

yield stress is reached, after which rapid deformation begins.

1.1 Necessity of Present Research

The northern hemisphere ice sheets have played an important role in global

climate through their influence on solar albedo, atmospheric circulation and continental

freshwater runoff. Understanding climate dynamics and sea level history thus requires

the consideration of the factors that control the extent and thickness of ice sheets. Of

these, the mechanics of ice movement is most probably the most fundamental. This is a

very complex problem, conventional and advance computer methods are needed to gain

insight. The computer methods will rely on constitutive models that have been calibrated

using appropriate testing.


9

1.2 Objective and Scope of Research

The objective of this research is to gain further understanding of the creep behavior of

glacial tills. To accomplish this, the DSC model will be used. This model will be

calibrated from the laboratory tests performed on a till from Manitoba, Canada (Fig.1.1).

The scope of the research is as follows.

1. Conduct appropriate laboratory creep experiments on the subglacial till;

2. Determination of creep parameters;

3. Back prediction using the one dimensional creep procedure and comparing the

results with laboratory observations;

4. Apply the DSC-creep model in appropriate manner to describe the creep behavior

of subglacial till;

5. Verify the DSC-creep model by comparing the model with the results of

independent tests; and

6. Further prediction of creep results for unloading and interpretation of results.

1.3 Organization of Text

A brief review of the studies done to understand the deformation of subglacial till

and the existing flow relation for till is presented in Chapter 2

In Chapter 3, the review of various rheological models is described in brief.

In Chapter 4, the DSC, Multicomponent DSC and the Overlay model is described

in brief.

Chapter 5 discusses the results from triaxial creep experiments, the procedure for

one dimensional creep program and the procedure for evaluating the creep parameters for

viscoelastic (ve) model, Perzynas elastoviscoplastic model (evp) and


10

viscoelasticviscoplastic (vevp) model. Chapter 5 also discusses the results from the one

dimensional procedure and compares them with the laboratory observations.

Chapter 6 briefly describes the implementation of DSC to finite element

procedure. Chapter 6 also discusses the results from nonlinear finite element program

DSC-SST2D by Desai, 1999 and compares these results with the laboratory observations.

FIGURE 1.1-The site from where the till was excavated for laboratory testing (Roy
1998).
11

CHAPTER 2

DEFORMATION OF GLACIAL TILL

1.1 Introduction

In theoretical analysis of glacier sliding, the ice is assumed to move over a bed

that is rigid and impermeable. In fact, glacial retreat often reveals not bedrock but glacial

deposits (till). As a large-scale example, the southwestern portion of the Laurentide ice

sheet (now Great Lake basin) was underlain by glacial sediments, as was the central part

over the present Hudson Bay, Northern Manitoba, Canada. Geologists have long known

that till can, under certain conditions, deform in shear (MacClintock and Dreimanis,

1964) but they have, until recently, ignored this fact.

Observations in Iceland and Antarctica show a change in view, Boulton and Jones

(1979) showed that, near the margin of the Breidamerkurjkull glacier, shear in a layer of

water- saturated till accounted for 90% of the velocity measured at the glaciers surface.

The yield stress in shear of this till is only 3 to 8kPa (Boulton and Dent, 1974); such a

shear stress produces significant deformation in ice. Seismic reflection surveys on Ice

Stream B in West Antarctica showed that the ice is underlain by a layer of till several

meters thick that is highly porous and saturated with water. Deformation of subglacial till

could explain why the ice is moving several hundred meters per year although the driving

stress is only about 20kPa (Blankenship and others, 1986, 1987). The driving stress is

the driving force distributed over the entire frictional surface that resist over the ice base.

The driving force is the component of the glaciers weight that is operating parallel to the

frictional surface i.e., in the direction of ice movement.


12

Deformation of subglacial till could explain the low surface slopes, which

corresponds to basal stresses of only 7 to 22kPa , of some of the southern lobes of the

Laurentide ice Sheet, as inferred from the geological evidence (Mathews, 1974).

Conditions (hydraulic conductivity, diffusivity) in subglacial till may also control the

behavior of some surging glaciers (Clarke and others, 1984).

1.2 Properties of Till

The term till, as used by glaciologist, refers to deformable subglacial sediments

irrespective of whether they are originally deposited by a glacier or by water. To a

geologist, till is a poorly graded sediment with little structure deposited by the glacier.

The particles in till range from clay and silt (fines) to pebbles, cobbles and boulders

(coarse and clast). Water, contained in pores within a matrix of the fine- grained material,

is an essential component of a deformable till.

Porosity is the fraction of the total volume that consists of pores. Typical values

are 0.2 to 0.4 (Boulton and Dent, 1974). The upper values appear to be typical of a

saturated till that is deforming subglacially.

Steady state flow of a fluid through a porous medium is assumed to be governed

by Darcys Law. It states that the flux through the porous medium varies directly with the

pressure gradient and inversely with the viscosity of the fluid. For a parallel sided till

layer lying on an impermeable surface of slope , the flux q in m / s is given by

k p
q= g sin (2.1)
s
13

Here, is the density of the fluid, is viscosity Pa s , g the acceleration due to

gravity, p the pressure of the fluid in the pores, and s a distance coordinate measured in

the direction of flow. The quantity k is the hydraulic intrinsic permeability. It depends on

the porosity, and when the till is compressed its porosity and permeability decrease

(consolidation). Because the amount of compression depends on the hydrostatic

pressure of the overlying material and on the pressure of the water in the pores, the

permeability depends on these quantities as well. Measured values of k for deformable

tills are in the range 2 10 13 to 4 10 15 m 2 .

gk
Hydraulic conductivity is expressed as = (m / s ) . For water at 0 0 C ,

= 1.787 103 Pa s and = 5.5 10 6 k .

Change in water pressure at the surface are propagated into a till layer according

to the diffusion equation. The equation for one dimension is given by

2 p p
D 2 = (2.2)
z t

Here p is the water pressure, z depth, t time, and D is the hydraulic diffusivity.

Diffusivity depends on the porosity and permeability of the till and depends on the

viscosity and compressibility of water. Values of D are in the range

4 10 4 to 6 10 7 m 2 / s (Boulton and Dent, 1974).

Stresses applied to the till are distributed between the solid matrix and the water

pores. The deformable properties of the granular materials depend upon what is called the

effective stress obtained by subtracting the pore-water pressure p , taken as positive, from

the normal components x , y , z of the total stress. As regard the deformation in shear,
14

one possible approximation is to treat till as a perfectly plastic material i.e., it does not

deform if the applied stress is less than the yield stress or shear stress 0 i.e., rigid plastic.

The Yield stress may be obtained from Mohr- Coulomb criterion as

0 = c 0 + N tan (2.3)

where, 0 is shear stress, N is the normal stress, c0 and is the cohesion and friction

angle of the material respectively.

Dilatancy, a property granular material, is an added complication in till

deformation. A dilatant material expands on shearing because the grains rearrange

themselves in a way that increases the pore space; this weakens the material. A

consolidated till can recover porosity and permeability in this way. On the other hand, a

dilated till collapses or stiffens when deformation stops. Subglacial till deforms

appreciably only if it is water-saturated and pore pressures are high. Boulton and other

(1974) demonstrated dilatant behavior in laboratory tests on an Icelandic till.

1.3 Flow Relations for Till

Till is a complex material and, its deformation rate depends not only on the

applied shear but also on effective stress, porosity, volume fraction of fines, and strain. It

is inhomogeneous and may be anisotropic if deformation has produced a preferred

orientation in the grains. In studies done before, till is assumed to be water-saturated, ice

free, and isotropic. Only steady state deformation in simple shear is considered and the

strain rate is assumed to depend only on shear stress and effective pressure. The steady

state assumption implies that the till has been deformed to large strains so that change in

strength due to dilatant expansion is completed. The objective is to obtain a flow relation
15

that can be used in modelling the flow of glaciers with deformable beds, especially the

ice streams in West Antarctica. The extensive studies of the behavior of clay rich soil

have never produced a flow relation of type explained above. In soil mechanics, the usual

interest is in condition of failure; in glaciology it is in the relation between strain rate and

residual strength that is, the shear stress needed to produce continuing deformation after

failure.

Mechanical tests on samples of till in the laboratory can suffer from some

drawbacks. As in ice testing, it is difficult to carry on a test long enough to reach a steady

state. Deformation rates are higher than those in the field and so the measured strength

may be different. Large clasts have to be removed from the sample before testing.

Because interactions between clasts are part of the deformation process, their removal

changes the mechanical properties. Measurements of the shear strain rate and pore- water

pressure in subglacial till are difficult, only few have been made. The reasons for it are as

follows-

1. The base of glacier is inaccessible, except near the margin of a glacier. Also, the

stress conditions at the margin are not same as in the interior,

2. Glaciers are only found in remote and inhospitable places, e.g. Antarctica or

Alaska/ Yukon, and

3. The installation of borehole could alter the in-situ stress and water pressure.

Existing flow relations for till are empirical relations between such measurements

of strain rate, shear stress, and effective pressure. Other factors that affect the mechanical

properties, the size distribution of the particles are ignored. The factor could be till
16

composition, clay mineralogy etc. The review of data from different locations and the

existing flow relations for till are as follows.

Data: Breidamerkurjkull

Breidamerkurjkull is a temperate glacier in Iceland (Boulton and Hindmarsh,

1987). The data consist of seven triplets of values of shear strain rate & , shear stress ,

and effective normal pressure N measured in a 0.5m thick layer of till deforming under

about 10m of ice near the glacier margin. Strain rates and water pressures are averages

over unspecified periods. Moreover, longitudinal stresses, which may predominate over

shears near the terminus, were neglected. The data fitted a relation of the from

& = B1 a N b (2.4)

where, a = 1.33 , b = 1.80 , B1 = 34.8(kPa) 0.47 a 1 .

Equation 4 implies that & 0 as N ; in other words, the till grains can move freely

past each other. This is unrealistic because till cannot dilate indefinitely. Fowler and

Walder (1993) therefore rewrote the equation as

& = B2 a ( N + N 0 )b (2.5)

The constants obtained from the same data are, a = 1.47 , b = 2.34 , B2 = 172(kPa) 0.87 a 1 ,

N = 4.5kPa .

Boulton and Hindmarsh obtained a second relation by assuming that strain depended not

on the total stress but on the amount by which it exceeded a yield stress. The relation is

& = B3 ( 0 )a N b (2.6)
17

with a = 0.065 , b = 1.25 , B3 = 121(kPa) 0.625 N 1 . The yield stress was obtained form Eq.

3.1.3 with c0 = 3.75kPa and tan = 0.625 . These parameters appear to have been

obtained independently by laboratory tests on the till.

Although Boulton and Hindmarsh measured velocity-depth profiles in the till,

they used only strain rates averaged and, with the simplifying assumption that a and b are

integers, concluded that the shape of the profiles was consistent with a relation of the

from

& = B4 ( 0 ) (2.7)
N2

However, the shape of the velocity depth profiles is determined to some extent not by the

behavior of the fine-grained matrix but by the position of boulders.

Data: Trapridge Glacier

Trapridge Glacier is a small glacier in the Yukon, Canada. The ice is less than

100m thick. Continuously recording tilt and pressure sensors showed that till deformation

beneath Trapridge Glacier is more complicated than that in an idealized sheet of till

shearing uniformly in one direction (Blake, 1992). In four days average values in 1988

were & = 36a 1 , N = 292kPa ; the six days average values in 1988 were & = 4.1a 1 ,

N = 78kPa . The shear stress , calculated in the usual way, was 77 kPa in both cases.

Data: Ice Stream B

Ice Stream B is one of the five ice sheet that drain ice from the interior of West

Antarctica to the Rose Ice Shelf. It is about 300km long and 30 to 80km wide. An

analytic, steady state solution for ice stream B, shows that till viscosity decreases slowly
18

downstream, probably in response to decreasing effective pressure downstream (Kamb,

1991). Surface velocities are in the range 50 to 825m / a compare with less than

10m / a for the ice on either side. An extensive seismic reflection survey revealed a layer,

6m thick on average, at the base of the ice. The velocities of both compressional ( P) and

shear ( S ) waves in this layer were abnormally low. Their values implied that the material

in the layer was highly porous (porosity about 0.4) and saturated with water at high pore

pressure that supported nearly all of the weight of the 1050 m of ice. The estimated

effective pressure was 50 40kPa , compared with the overburden pressure of 9.2 MPa .

The surface velocity at UpB (Upstream portion of Ice Stream B) is

440m / a although the basal shear stress is only 20kPa . The surface velocity of

440m / a results entirely form deformation of 6m of till, the average strain rate is 73a 1 .

Eq.2.4 and Eq.2.5, with the minimum effective pressure of 30kPa , predict a strain rate of

only 4a 1 . Eq.2.6, with N = 30kPa and 0 = 2kPa , predict & = 10a 1 . The yield stress

calculated from Eq.2.3, with N = 30kPa and Boulton and Hindmarshs (1987) value of

c 0 and , is greater than the basal shear stress. The data from Ice Stream B and Trapridge

Glacier therefore do not support the Boulton-Hindmarsh (1987) relation.

A simple one dimensional coupled ice- till flow model for ice stream flow on

deforming till was proposed by (Alley and others. 1987). The model is based on data of

ice- stream B. In this model it is assumed that ice stream behavior will be quantitatively

and qualitatively different on deforming till as opposed to water lubricating rock. The

physical concept of till conservation, ice conservation plus a flow law for till is used.

From these a coupled equation is derived by Alley and others (1987) for the time rate of
19

change of ice thickness and till thickness, the accumulated rate of ice and till, and till

properties.

For one-dimensional flow of ice of constant density, continuity of ice can be

stated as

( )
hu
= b&
h
(2.8)
x t

where x is the horizontal coordinates in the direction of ice flow with the origin taken in

the bed under the head of the ice- stream, h is the ice thickness, u is the ice velocity in

the x direction averaged over thickness, b& is the net accumulation rate of ice(net ice flow

minus basal melting) in ice thickness equivalents. Similarly, for till of constant density

flowing in the x direction, continuity yields

(
hb u b ) h
= b&b b (2.9)
x t

where, h is the till thickness, u b is the till velocity in the x direction, u b , averaged over

thickness, and b& b is the till generation rate.

It is assumed that till exhibits linear viscous rheology. For one-dimensional flow

in simple shear the flow law reduces to

u b
b = (2.10)
z

where b is the basal shear stress, is the Newtonian viscosity, and z is the vertical

coordinate, positive upward. Eq.2.8 to Eq.2.10 incorporates the physics of one-

dimensional model.
20

Because the till is thin compared to overlaying ice, the shear stress acting in the

till is nearly constant with depth. For linear viscous till with no slip boundaries the

equations given by Alley and others, 1987 are expressed as

h & g h
2
2h h h h hh
=b h hhb 2 + 2hb + h b b (2.11)
t x x x x x x

and

hb g h hb 2h h hb hhb h
2
h
= Kh hhb 2 + hb + 2h (2.12)
t x 2 x x x x x x

The Eq.2.11 and Eq.2.12 describe time-dependent ice stream flow on a deformable bed.

For materials that deviate from linear-viscous rheology, the viscosity listed in the Table

2.1 is the effective viscosity at the existing state of stress.

Table 2.1-Viscosity of till and other materials


Material Viscosity (Pa. s) References
Water 1.8 10 3 Weast, 1973
Mudflow 4.5 10 2 Joshnson, 1970
Till, UpB 8.0 10 9 Alley and others, 1987
Till, Breidamerkurjkull 3.2 1018 Boulton and others, 1987
Till, Blue Glacier 1.8 1011 Engelhardt et. al, 1978
Ice, at bed, UpB 3.5 1014 Paterson, 1994
Mantle rocks 10 21 Turcotte and Schubert,
1982
21

1.4 Flow Relation

The above described flow relations for the steady-state shear deformation of water

saturated till can be written in the from

& = B( 0 )a N b 0 (2.13)

& = 0 <0 (2.14)

later a term N 0 is added to N to keep & finite as N 0 . The simplest case is to assume

that till behaves as a Newtonian viscous material so that 0 = 0 , a = 1 , b = 0 and 1 / B is

the viscosity. This relation has been used in two analyses of flow in ice streams by

MacAyeal (1989) and Alley and others (1987b)

1.5 Yield Stress and effective velocity

For lack of a well-established flow relation, most analyses of bed deformation

have been based on one of two simplifying assumptions, namely that subglacial till

deformation as either a Newtonian viscous material (a = 1) or a perfectly plastic

one (a ) . Table 3.2 list some measured values of yield stress 0 (kPa ) and its

component, cohesion c0 (kPa ) and friction angle (deg ) .

The values were measured on samples in the laboratory. The yield stresses were

measured in the field either on the fresh samples or, at Columbia and Trapridge glaciers,

by a rod driven into the till from the bottom of a borehole. Ice movement drags the rod

through the till, the mechanical properties of which can be inferred the way rod bent.

Bending is measure when rod is extracted or remotely by strain gauges bounded to it (a

ploughmeter). At Breidamerkurjkull, the upper layer had a high proportion of slit and
22

clay and was deforming subglacially, while the lower layer, with much slit and clay, was

not found. The data from this glacier show that laboratory tests give higher yield stresses

than those obtained by field measurements; a yield stress measured in the field is even

less than the cohesion measured in the laboratory. All the pair of values of c 0 and in

the table would give a yield stress for Ice Stream B that exceeds the basal shear stress

there. The field measurements of the yield stress are probably more useful, although they

suffer from the drawback that the effective pressure is not known.

TABLE 2.2- Measurement of yield stress, cohesion and friction angle.

0 (kPa ) c0 (kPa ) (deg ) Glacier Reference

- 15 30 Typical till Clarke, 1987b


- 0-40 31-39 New England till Lingle and
others, 1974
7.5 28 27 Bredamerkur Boulton and
upper layer others, 1974
32.5 21 28 Bredamerkur Boulton and
lower layer others, 1974
3.75 32 Bredamerkur Boulton and
Hindmarsh, 1987
60 5 31 Stor R.L.Hooke, pers.
comm..
2 - - Ice Stream B Kamb, 1991
5.5-13 - - Columbia Humphrey and
others. 1993
48-57 - - Trapridge Fisher and
Clarke, 1994
23

Till will deform only if its yield stress is less than the basal stress of the overlying

ice, which is often about 100kPa . With the values of c 0 and for typical till (Table 3.2),

this implies that the effective stress must be less than 145kPa . Because the overburden

pressure of 250m of ice is 2.2 MPa , till will deform only if its pore water pressure is at

least 90 to 95% of the overburden pressure.

TABLE 2.3 Estimates of effective viscosity of till.

(Pa s ) Glacier Method Reference:

5 1010 5 1011 Breidamerkur data*P P Boulton and


Hindmarsh, 1987
3 1010 1.5 1011 Trapridage data*P P Blake, 1992

3 10 9 3 1010 Trapridage ploughmeter Fischer and Clarke,


1994
2 10 8 5 10 8 Columbia Bent drill rod Humphrey and
others, 1993
*
8 10 9 Ice Stream B data P P Alley and others,
1986
2 10 9 1.4 1010 Ice Stream B fits to model Alley and others,
1987b
10 8 1.5 10 9 Ice Stream B fits to model MacAyeal, 1989
*
P Measurement of basal stress and either strain rate or velocity and till thickness.
P

Table 2.3 lists the estimates of what the viscosity of till would be if it were to

deform as a Newtonian viscous material; there is no evidence that it does. The

ploughmeter and drill rod values are based on the same measurement as before, but
24

analyzed under the assumption that till is Newtonian viscous not perfectly plastic. The

last two ranges of values are based on modeling studies in which viscosity was adjusted

to give the best fit to surface velocities along the ice stream.

The spread of the estimates by the factor 5000 and by 50 on the same glaciers,

strongly suggest that till is not a Newtonian viscous material; in other words, its viscosity

is not constant but depends on the stress.

The reviews done in this chapter are the part of research published by references

listed in this chapter and mainly from the book Physics of Glaciers by Patersons, 1994.

The purpose of this review is to present the results from field investigation and laboratory

results performed on subglacial tills (Table 2.1, 2.2 and 2.3) and later compare these

results with the result calculated in the subsequent chapters.


25

CHAPTER 3

RHEOLOGICAL MATERIAL MODELS

1.1 Introduction

The theory of rheological models used as a combination of several mechanical

elements (springs, dashpots, frictional resistance to movement) connected in series or

parallel or both. These systems represent only the mechanical behavior of the material

under condition of testing, but have nothing in common with the real material. The

following are the three basic elements forming mechanical models.

1. A perfectly elastic spring representing truly elastic deformation.

2. A dashpot representing viscous deformation (truly Newtonian material).

3. A mass resting on a plane with a frictional force equal to the yield limit which

prevents the movement of the block under the action of the forces below the yield

limit. This represents truly elastic deformation.

The details of various rheological models are present in Handbook on Mechanical

Properties of Rocks, Vol-3, by Lama and Vutukuri, 1978. The Viscoelastic (ve ) ,

Elastoviscoplastic (evp ) and Viscoelasticviscoplastic (vevp ) are discussed briefly in this

chapter. The creep analysis for subglacial till is done for evp, Perzynas model and vevp

model. The details of these models are presented in Mechanics of Material and

Interface-The Disturbed State Concept, Desai, 2001. The following descriptions are

adopted from this book.


26

1.2 Viscoelastic (ve) Overlay Model

Consider the model in Fig.3.1 (Desai, 2001) It is required to determine four elastic

(E1 , 1; E2 , 2 ) and two viscous (1 , N 1 ) parameters. Consider the uniaxial or triaxial

loading response. Let 0 be the applied stress, which is held constant, and let t1 and t 2 be

the thickness of the unit 1 and unit 2 respectively. Then the equilibrium of forces gives

1t1 + 2 t 2 = 0 (t1 + t 2 ) (3.1)

here, t1 + t 2 = 1 . Therefore,

d 2 t d 1
= 1 (3.2)
dt t 2 dt

At, t = 0 , the dashpot is not operational; hence, the strain 0 , is given by

0 E1 0
0 = and (0 ) = (3.3)
E1t1 + E 2 t 2 E1t1 + E 2 t 2

Now with respect to unit 1

= e +d (3.4)

where e and d are the strains in the spring and the dashpot respectively, Hence

d 1 d 1
= + 1 (3.5)
dt E1 dt

where, 1 is the stress in unit 1.

With respect to unit 2,

2
= (3.6)
E2

Hence, using Eq.3.25


27

d 1 t 1 d 1
= (3.7)
dt E 2 t 2 dt

Combining Eq.3.28 and Eq.3.30, we have

1 1 t 1 d 1
+ + 1 = 0 (3.8)
E1 E 2 t 2 dt

the solution of which can be obtained as

t a

1 = 1 (0 ).e b
(3.9)

where, a = E1 E 2 t 2 , b = E1t1 + E 2 t 2 . Therefore,

E1 , 1

, N E 2 , 2
0
(a) (b) t

FIGURE- 3.1 (a) Viscoelastic Model (b) Viscoelastic Response. (Desai, 2001)

Therefore,

ta
d 1 t a
= 1 (0 ). e b
(3.10)
dt b

and from Eq.3.25

t a
d 2 E12 E 2 t1 0
= .e b
(3.11)
dt b2

Substitution of Eq.3.33 to Eq.3.30 gives


28

ta
d E12 E 2 t1 0
= .e b
(3.12)
dt b2

Integration of Eq.3.35 leads to the solution for the overlay strain, , as

ta
d E12 E 2 t1 0 0
= .e b
+ (3.13)
dt b2 E2t 2

1.3 Elastoviscoplastic (evp) Perzynas Model (1966)

Here we consider one unit, Fig.3.2, and one overlay with thickness t = 1 . The

elastic (E , ) , viscous (, N ) and y (classical plasticity) or F (HISS model) are assigned

to the spring, dashpot, and the slider, respectively.

At time t = 0 , the dashpot will not be operational; hence, the instantaneous elastic

0
strain, e = , will occur. Then, for time t > 0 , the dashpot will be operational, and
E

( )
viscoplastic strains vp , (Fig.3.12), due to the slider ( y orF ) , will remain. The final

viscoplastic strain will be equal to that from inviscid plasticity governed by the yield

function, F . The evp-type model gives the timewise variation of the plastic strains, but

the final magnitude of the viscoplastic will be equal to the plastic strain from inviscid

plasticity.
29

E ,

Unit 1
Unit 2
e

vp
, N
e

t
(a) (b)

FIGURE 3.2- (a) Perzynas Elastoviscoplastic Model (b) Elastoviscoplastic Response

(Desai, 2001)

1.4 Viscoelasticviscoplastic (vevp) Model

Consider the two units in the model, (Fig.3.3), and assume two overlays, each with

thickness = 0.5 . The parameters are assigned as shown in Table 3.1. The resulting model

is shown in Fig.3.13

At time t = 0 , only the springs will deforms the dashpots are not operational. The

instantaneous elastic strain e , will be given by


e = (3.14)
E1t1 + E 2 t 2

Then, for time t > 0 , the dashpot in unit 1 will operate first, as the slider in unit 2 has

nonzero stress. This will lead to viscoelastic deformations. Now for later time t > 0 ,
30


E1 , 1
E2 , 2 e

1 , N1
ve
F ( y )
2 , N 2
e vp

t
Unit 1 Unit 2 (a) (b)

FIGURE 3.3- (a) Viscoelasticviscoplastic Model (b) Viscoelasticviscoplastic

Response. (Desai, 2001)

When the yield condition in slider 2 is reached, the dashpot in unit 2 will also

become operational.. If continuous yielding (e.g. 0 model) is considered for the slider,

both viscoelastic and viscoplastic deformation can occur simultaneously during A B .

Upon unloading, e will be recovered first, then the viscoelastic strain, ve , will be

recovered, and finally the viscoplastic strain will remain Fig.3.3(b).


31

CHAPTER 4

THE DISTURBED STATE CONCEPT

1.1 Introduction

The initial idea of this theory was introduced by Desai (2001) to characterize the

softening response of an overconsolidated soil by expressing the observed response in

terms of its response in its normally consolidated state as the reference state. This idea

was later formalized as the Disturbed State Concept (DSC) to model the behavior of

wide range of materials.

In the DSC, it is assumed that applied force causes disturbance or change in the

materials micro structure. As a result, an initially relative intact (RI) material modifies

continuously, through a process of natural self-adjustment, and a part of it approaches the

fully adjusted (FA) state at randomly distributed locations in the material. Thus, the

observed or average response can be represented in terms of the responses of the

materials in RI and FA state, which are called reference states (Fig.4.1). In this concept,

the relative intact state (RI) can be characterized as elastic, elastic- perfectly plastic, or

following any constitutive model like the Hierarchical single-surface plasticity (HISS)

model. The fully adjusted state (FA), can be assumed to imply the sate in which the

material can continue to carry the shear stress level reached up to that state under given

initial hydrostatic stress and can continue to deform in shear with constant volume; which

is called critical state.


32

The DSC, can yield to a unified material model which can be specialized to

include a wide range of material behaviors. It has been compared to nonlocalized

continuum damage model, which assumes that the observed behavior of a material at a

point is influenced by the behavior of its surroundings. If a deforming material is

composed of both continuous and discontinuous parts, then the overall behavior at a point

within the material is influenced by the continuous and discontinuous parts surrounding

these models, the behavior of a material element is a smear or weighted average of both a

continuous and discontinuous response (Desai, 2001).

RI

a Df Du
Dc

Observed

FA

D=0

FIGURE 4.1 Schematic of Stress-Strain behavior with DSC.(Desai, 2001).


33

Primary Secondary
Creep Creep


Tertiary
Creep

Permanent
Creep

FIGURE 4.2 - Schematic of Creep Behavior.

1.2 Multicomponent DSC and Overlay Model

The evp model based on Perzynas theory simulates the viscoplastic response,

particularly the time-dependent plastic response during the secondary creep regime (Fig.

4.2). However, in general, a material may experience both viscoelastic and viscoplastic

creep responses, which include creep deformation during both the primary and secondary

regimes. Hence, it is appropriate to develop and use a model that can simulate the

combination response as closely as possible. The following description is adopted from

Desai ( 2001).
34

Unit 1 Unit n
Unit 2

E1B

B
F1 , ( y1 ) B

FIGURE 4.3- Overlay model (Desai, 2001)

TABLE 4.1 Specializations of Multi-component DSC Model (Desai, 2001)

Specialization Plastic No. of Thickness Parameters


Model Overlays
Elastic (e) Von Mises 1 1 .0 E , , , N , and very high y

Maxwell Von Mises 2 0.5,0.5 E1 ,1 , 1 , N 1 , y1 = 0

E 2 , 2 , 2 , N 2 , y 2 = 0

Viscoelastic (ve) Von Mises 2 0.5,0.5 E1 ,1 , 1 , N 1 , y1 = 0

E 2 , 2 , 2 , N 2 , y 2 = very high

Elastoviscoplastic (evp) Any 1 1 .0 E , , , N ,


(Perzynas type)
Viscoelasticviscoplastic Von Mises 1 0.5 E1 ,1 , 1 , N 1 , y1 = 0
(vevp) 1 0.5
E 2 , 2 , 2 , N 2 , y 2 or F
35

Models for
component Fa

1 2 n

FIGURE 4.4 - Multi- component DSC (Desai, 2001)

In the basic DSC, an element of same material, which is composed of two

material parts in the RI and FA reference states. During deformation, the extents of the RI

and FA parts changes, which is defined through the disturbance , D . DSC for a material

element composed of more than one or two, or a higher number of, different materials. In

that case, the behavior of the component material provides reference state responses.

Then the observed stress, a , can be expressed as

A1 A A
a =1 + 2 2 + ........................... + n n (4.1)
A A A

where A1 , A2 ,........... An are the areas of the components (Desai, 2001), with total area

A = Ai (i = 1,2....n ) and 1 , 2 ,...... n are the corresponding stresses. For convenience,

we assume that the width of the material element is constant and equals b .

Eq.4.1 can be written as

a = 1 d1 + 2 d 2 + ............ n d n (4.2)
36

Ai
where d i = is the ratio of the area of component i to the total area A ; here,
A

d i = 1 . The three dimensional incremental form of Eq.4.2 can be written as

~
a
(
d = d t1 + d t 2 + ................ + d t n
~
1
~
2
~
n
) (4.3)

Eq.4.3 can now be written as


d = t1 C 1 d 1 + t 2 C 2 d 2 + t 3 C 3 d 3 + ................ + t n C n d n
a
(4.4)
~ ~ ~ ~ ~ ~ ~ ~ ~

where C m (m = 1,2,....n ) are the corresponding increment strain vectors. If it is assumed


~

that the strains in all components are equal (= d ) , Eq.4.4 becomes

~
(
d = t1 C 1 + t 2 C 2 + t 3 C 3 + ................ + t n C n d
a
~ ~ ~ ~
) ~
(4.5)

or,

d = C eq d
a
(4.6)
~ ~
~

where, Ceq is the equivalent constitutive matrix for the material element. In the
~

symbolic representation of the multicomponent DSC model from Eq.4.6. Each

component unit (1,2,3,.............n ) can be characterized using elastic, viscoelastic,

Elastoviscoplastic, or any appropriate model. Thus, it can provide a hierarchical frame

work from which elastic (e ) , viscoelastic (ve ) , elastoviscoplastic (evp ) , and

viscoelasticviscoplastic (vevp ) .
37

CHAPTER 5

LABORATORY CREEP EXPERIMENTS, PARAMETER

DETERMINATION AND BACK PREDICTION

1.1 Introduction

A general yield function in the hierarchical approach for constitutive modelling of

materials is used to characterize the creep behavior of Tiskilwa glacial till. A numbers of

creep tests were performed in the laboratory at different confining pressures and constant

loading conditions. The two creep parameters (fluidity parameter) and N in the model

were determined by using results available from the laboratory creep tests.

The laboratory creep tests (Fig.5.1) were performed at the constant stress

conditions. The test were performed on the Tiskilwa till specimens of height = 5.6 inch

h
and diameter = 2.8 inch ( = 2 : 1 , ASTM Standards). The creep tests were performed at
d

drained conditions, to keep the test conditions very similar to in-situ till environment.

Table 5.1 shows the creep test performed under different stress conditions. The constant

load for the creep tests were determined on the basis of the conventional triaxial

compression (CTC) tests on the till samples. The creep tests were performed at various

percentages of the peak stress of triaxial test results. The samples were extruded from the

shelby tube and then were installed in a triaxial device. The confining stress was applied

till the sample consolidates, which took about five days. Then the axial stress was

increased so that the deviatoric stress ( 1 2 ) reached the desire percentage of the peak

stress ( 1 2 ) in the triaxial test for the applied confining pressure. The axial strain was
38

recorded with the time. The sample was kept in the same stress environment till it shows

no or very little variation in the axial stain (which is normally 7- 8 days, as observed).

FIGURE 5.1 Triaxial Creep Testing Setup

TABLE 5.1 List of creep experiment performed in the laboratory at various deviatoric
stress ( 1 2 )kPa for constant confining stresses 3 (kPa ) .
3 (kPa ) 20% 40% 60% 80%
( 1 2 )kPa ( 1 2 )kPa ( 1 2 )kPa ( 1 2 )kPa
100 kPa - 1 2 -

200kPa 1 1 2 1

400kPa 1 1 1 -
39

1.2 Mechanics of Viscoplastic Solution

The mechanics of viscoplastic solution can be best explained by considering a

creep test. Consider that an undrained creep test is to be performed on a preconsolidated

sample by imposing a stress, ij , at time t = 0 + . In the creep test, the imposed stress ij is

kept constant (Fig.5.2b) for the entire duration of the test while the (creep) strains are

measured with time. The following description is adopted from Samtani and Desai (1999)

and Desai (2001).

Let FA represents the yield surface passing through point A (Fig.5.2c-1) at ijo . Let

point B (Fig.5.2c-1) represents the imposed state of stress ij which remains constant

with time and under which creep strains will be measured with time t (Fig.5.2 a). A state

of stress (such as represented by B) outside a yield surface (e.g. FA ) will generate

viscoplastic strains due to stress imbalance (in this case ij ij0 ). As time passes, the

strains (elastic+ viscoplastic) accumulate (Fig.5.2d). The accumulating viscoplastic strain

trajectories (Fig.5.4e) modifies the value of the hardening function ( ) (Fig.5.2f), and

consequently change the position of the yield surface (Fig.5.3c-2, 5.2c-3). When the load

(
increment is applied at t = 0 + the value of F is greater than zero Ft =0+ > 0 . At )
( )
sometime t t > 0 + , the material will harden and hence i < 0+ and 0 < Fi < F0+ .

Similarly, at time t = t + 1 we will have i +1 < i and 0 < Fi +1 < Fi . After a long duration

of time, F will become (almost) equal to zero (Fig.5.2g), which means that the yield

surface will pass through stress point B (Fig.5.2c-4). At this time steady state conditions

are deemed to be achieved and the displacement ceases.


40

t=0 t=
(a)

0
~
~

(b) Time

*B *B
*B *B
J 2D J 2D J 2D
J 2D
F0

(c-1) (c-2) (c-3) (c-4) J1


J1 J1 J1
d vp = d p


~ (d)

vp (e)

A
(f)
B
t

F (g)
F =0
t

FIGURE 5.2- Mechanics of Viscoplastic Solution (Samtani and Desai, 1991, Desai,
2001)
41

At time t = the solution given by the viscoplastic model tends to the solution

given by the inviscid model, e.g. the accumulated viscoplastic strains tends to the value

of the accumulated plastic strains given by the inviscid model (Desai and Zang 1987).

1.3 Results and Discussions

The triaxial laboratory creep experiments (Fig 5.3) were performed at the constant

stress conditions. The static load for the creep tests were determined from the

Conventional triaxial compression (CTC) - Undrained test performed on the till samples

previously. Creep tests were than performed at various percentages of the peak loads ( p )

calculated from the CTC tests.

Figure 5.3 Triaxial Creep Experiment


42

The Creep tests performed for 3 = 100kPa and at constant ( 1 3 )kPa as shown in

Fig.5.4 to Fig.5.7

0.008

Axial Strain .
0.0015
0.006
Axial Strain

0.001
0.004 0.0005
3 =100.0 kPa
B B

0
1- 3 = 42.4 kPa
0.002 0 0.05 0.1 0.15
B B B B

Time (min)
0
0 2000 4000 6000
Time (min)
FIGURE 5.4- Axial stress-strain results for constant 3 = 100kPa and
( 1 3 ) = 0.4 p kPa ( p = 106kPa )

0.01

0.008 0.002
Axial Strain .
Axial Strain

0.0015
0.006
3 =100.0 kPa
B B
0.001
1- 3 =63.6 kPa 0.0005
0.004
B B B B

0
0.002 0 0.05 0.1 0.15
Time (min)
0
0 2000 4000 6000 8000
Time (min)
Figure 5.5- Axial stress-strain results for constant 3 = 100kPa and
( 1 3 ) = 0.6 p kPa ( p = 106kPa )
43

Figure 5.6- Creep specimen after the test was performed for 3 = 100kPa and constant
( 1 3 ) = 0.6 p kPa ( p = 106kPa )

0.01
.

0.008
.

0.002
Axial Strain

0.006 0.0015
Axial Strain

3 =100.0 kPa 0.001


0.004
B B

1- 3=63.6 kPa
B B B B
0.0005
0
0.002 0 0.05 0.1 0.15
Time (min)

0
0 2000 4000 6000 8000
Time (min)
Figure 5.7- Axial stress-strain results for constant 3 = 100kPa and
( 1 3 ) = 0.6 p kPa ( p = 106kPa ) (Independent test)
44

The axial strain recorded for the test done at ( 1 3 ) = 42.2kPa (Fig.5.4) at 1

sec, 5 sec and 10 sec are 0.0002, 0.00035 and 0.0004 respectively. There is no change in

axial strain rate after 4000 min.

The value of axial strain rate at t = 0 was 0.01561.The axial strain recorded for

the test done at ( 1 3 ) = 63.6kPa (Fig.5.5) at 5 sec, 10 sec and 15 sec were 0.0016,

0.00148 and 0.00130 respectively. The strain rate (Fig.5.8) decreases sharply to

0.000733, 0.000234 and 2.44E-05 in 10 sec, 15 min and 30 min. The stain rate decreases

gradually and the values recorded at 2000 min was 2.14E10-7 . No appreciable change in

stain rate was recorded after 4000 min. One Independent test (Fig.5.7) is also performed

for validation purpose.

0.0008
Strain Rate (/min)

0.0006
3 =100.0 kPa
B B

1- 3= 63.6 kPa
0.0004
B B B B

0.0002

0
1 10 100 1000
ln(t) (min)

Figure 5.8- Time vs. Strain rate for constant 3 = 100kPa and
( 1 3 ) = 0.6 p kPa ( p = 106kPa )
45

The Creep tests performed for 3 = 200kPa and at constant ( 1 3 )kPa as shown in

Fig.5.9 to Fig.5.17

0.0025
.

0.002

.
0.00003
0.000024
0.0015
Axial Strain

0.000018

Axial Strain
0.000012
0.001 3 =200.0 kPa
0.000006
0
B B

1- 3=37.0 kPa
B B B B

0.0005 0 0.05 0.1 0.15


Time (min)

0
0 4000 8000 12000
Time (min)
Figure 5.9- Axial stress-strain results for constant 3 = 200kPa and
( 1 3 ) = 0.2 p kPa ( p = 185kPa )

Figure 5.10- Creep specimen after the test was performed for constant 3 = 200kPa and
( 1 3 ) = 0.2 p kPa ( p = 185kPa )
46

0.016
.

.
0.005
0.012 0.004
0.003

Axial Strain
Axial Strain

0.002
0.008
3 =200.0 kPa
B B
0.001
1- 3=74.0 kPa
B B B B
0
0.004 0 0.05 0.1 0.15
Time (min)

0
0 2000 4000 6000
Time (min)
Figure 5.11- Axial stress-strain results for constant 3 = 200kPa and
( 1 3 ) = 0.4 p kPa ( p = 185kPa )

Figure 5.12- Creep specimen after the test was performed for constant 3 = 200kPa and
( 1 3 ) = 0.4 p kPa ( p = 185kPa )
47

0.035
. 0.030

.
0.025 0.015

Axial Strain
0.020 0.01
Axial Strain

3 =200.0 kPa 0.005


0.015
B B

1- 3 =111.0 kPa
0
B B B B

0.010 0 0.05 0.1 0.15


0.005 Time (min)
0.000
0 1000 2000 3000
Time (min)

Figure 5.13 Axial stress-strain results for constant 3 = 200kPa and


( 1 3 ) = 0.6 p kPa ( p = 185kPa )

Figure 5.14- Creep specimen after the test was performed for constant 3 = 200kPa and
( 1 3 ) = 0.6 p kPa ( p = 185kPa )
48

. 0.032
0.028
0.024
3 =200.0 kPa
0.02
B B

1 =111.0 kPa
B B
Axial Strain

0.016
0.012
0.008
0.004
0
0 1000 2000 3000
Time (min)
Figure 5.15- Axial stress-strain results for constant 3 = 200kPa
and ( 1 3 ) = 0.6 p kPa ( p = 185kPa ) (Independent test)

0.08
.

0.02
Axial Strain .

0.06
0.015
0.01
0.04
Axial Strain

3 =200.0 kPa
B B

0.005
1- 3 =148.0 kPa
0
B B B B

0.02 0 0.05 0.1 0.15


Time (min)
0
0 1000 2000 3000
Time (min)
Figure 5.16- Axial stress-strain results for constant 3 = 200kPa and
( 1 3 ) = 0.8 p kPa ( p = 185kPa )
49

Figure 5.17- Creep specimen after the test was performed for constant 3 = 200kPa and
( 1 3 ) = 0.8 p kPa ( p = 185kPa )

3.0E-05
Strain Rate (/min) .

2.5E-05
2.0E-05
3 =200.0 kPa
1.5E-05 B B

1- 3=37.0 kPa
B B B B

1.0E-05
5.0E-06
0.0E+00
1 10 100 1000 10000
ln (t) (min)
Figure-5.18 Time vs. Strain rate for constant 3 = 200kPa and
( 1 3 ) = 0.2 p kPa ( p = 185kPa )
50

The value of axial strain rate at t = 0 was 0.000219.The axial strain recorded for

the test done at ( 1 3 ) = 0.2kPa (Fig.5.9) at 5 sec, 30 sec and 1 min were 1.8254E-05,

3.6507E-05 and 5.4761E-05 respectively. The stain rate (Fig.5.18) decreases gradually

and the values recorded at 450 min, 2500 min and 12000 min were 5.18E-07, 2.66E-07

and 2.70E-08 respectively. There was very less change in strain rate after 13000 min

The axial strain rate at t = 0 was 0.1947. The axial strain recorded for the

test done at ( 1 3 ) = 0.4kPa (Fig.5.11) at 1sec, 2 sec, 5 sec and 10 sec were

0.003245, 0.003446, 0.003920 and 0.004121 respectively. The strain rate (Fig.5.17)

decreases gradually to 0.009481, 0.001750 and 0.000656, in 5 sec, 30 sec and 1 min. The

stain rate decreases gradually and the values recorded at 60 min, 500 min and 2000 min

were 5, 1.77E-05 and 1.21E-06 and 1.78754E-07 respectively. There was very little

change in strain rate after 4000 min.

0.001
Strain Rate (/min)

0.0008
3 =200.0 kPa
0.0006
B B

1- 3 =74.0 kPa
B B B B

0.0004
0.0002

0
1 10 100 1000 10000
t (min)
Figure-5.19 Time vs. Strain rate for constant 3 = 200kPa and
( 1 3 ) = 0.4 p kPa ( p = 185kPa )
51

The axial strain recorded for the test done at ( 1 3 ) = 0.6 p kPa

(Fig.5.13, Fig.5.14) at 1 sec, 2 sec, 5 sec and 10sec 0.00823, 0.01057, 0.01148 and

0.01285 respectively. The value of axial strain rate at t = 0 was 0.4936. The strain rate

(Fig.5.20) decreases sharply to 0.01173, 0.000124, and 3.41E-05 in 10 sec, 10 min and 60

min. The stain rate decreases gradually and the values recorded at 100 min, 250 min and

600 min were 1.24E-05, 4.46E-06 and 2.34E-07 respectively. There was very little

change in strain rate after 2000 min.

0.0015
Strain Rate (/min) .

0.0012
3 =200.0 kPa
0.0009
B B

1- 3 =111.0 kPa
B B B B

0.0006

0.0003

0
1 10 100 1000 10000
ln (t) (min)
Figure 5.20-Time vs. Strain rate for constant 3 = 200kPa and
( 1 3 ) = 0.6 p kPa ( p = 185kPa )
52

0.01
Strain Rate (/min) .
0.008
3 =200.0 kPa
0.006 B

B
B

1- 3 =148.0 kPa
B B B

0.004

0.002

0
1 10 100 1000
ln (t) (min)

FIGURE 5.21- Time vs. Strain rate for constant 3 = 200kPa and
( 1 3 ) = 0.8 p kPa ( p = 185kPa )

The axial strain recorded for the test done at ( 1 3 ) = 0.8kPa (Fig.5.16 ) at 1

sec, 2 sec, 5 sec and 10sec 0.00426, 0.00834, 0.01418 and 0.01964 respectively. The

axial strain rate at t = 0 was 0.2559. The strain rate (Fig.5.21) decreases sharply to

0.0437, 0.0107 and 0.000572 in 10 sec, 1min and 10 min. respectively. The strain rate in

2000min was 1.63E-07. There was very less change in strain rate after 1000 min.
53

0.0025

0.002
.

.
0.0002

Axial Strain
0.00015
0.0015
Axial Strain

3 =400.0 kPa 0.0001


0.00005
B B

1- 3 =50.0 kPa
0.001
B B B B

0.0005 0 0 0 0
Time (min)
0
0 2000 4000 6000 8000 10000
Time (min)
Figure 5.22- Axial stress-strain results for constant 3 = 400kPa and
( 1 3 ) = 0.2 p kPa ( p = 250kPa )

0.008
.

0.006 0.0015
Axial Strain

0.001
Axial Strain

0.004 3 =400.0 kPa


B B
0.0005
1- 3=500.0 kPa
0
B B B B

0.002 0 0.05 0.1 0.15


Time (min)
0
0 2000 4000 6000 8000
Time (min)
Figure 5.23- Axial stress-strain results for constant 3 = 400kPa and
( 1 3 ) = 0.4 p kPa ( p = 250kPa )
54

Figure 5.22- Creep specimen after the test was performed for constant 3 = 400kPa and
( 1 3 ) = 0.4 p kPa ( p = 250kPa )

0.008
.

0.0008
0.006
0.0006
Axial Strain
Axial Strain

0.0004
0.004
0.0002
3 =400.0 kPa
0
B B

1- 3=550.0 kPa
0.002 B B B B

0 0.05 0.1 0.15


Time (min)
0
0 2000 4000 6000
Time (min)

Figure 5.25- Axial stress-strain results for constant 3 = 400kPa and


( 1 3 ) = 0.6 p kPa ( p = 250kPa )
55

Figure 5.26- Creep specimen after the test was performed for constant 3 = 400kPa and
( 1 3 ) = 0.6 p kPa ( p = 250kPa )

2.0E-04
.
Strain Rate (/min)

1.5E-04 3 =400.0 kPa


B B

1- 3=50.0 kPa
B B B B

1.0E-04

5.0E-05

0.0E+00
1 10 100 1000 10000
ln (t) (min)
FIGURE 5.27- Time vs. Strain rate for constant 3 = 400kPa and
( 1 3 ) = 0.2 p kPa ( p = 250kPa )
56

The axial strain recorded for the test done at ( 1 3 ) = 0.2 p kPa (Fig.5.22) at 1

sec, 5 sec and 15 sec 0.000146, 0.000182 and 0.0002 respectively. The value of axial

strain rate (Fig.5.27) at t = 0 was 0.00875. The strain rate decreases gradually to 7.96E-

05, 6.08E-06 and 1.82E-07 in 10 min, 30 min and 1500 min. There was very less change

in strain rate after 7000 min.

The value of axial strain rate at t = 0 was 0.03506.The axial strain recorded for

the test done at ( 1 3 ) = 0.4 p kPa (Fig.5.23 ) at 1 sec, 2 sec, 5 sec and 10 sec 0.0032,

0.0058, 0.0066 and 0.0069 respectively. The strain rate (Fig.5.28) decreases sharply to

0.0001643, 1.52E-05 and 2.19E-07 in 1 min, 60 min and 4000 min. There was negligible

change in strain rate after 5000 min.

0.0002
.
Strain Rate (/min)

0.00015
3 =400.0 kPa
0.0001
B B

1- 3=100.0 kPa
B B B B

0.00005

0
1 10 100 1000 10000
ln(t) (min)
FIGURE 5.28- Time vs. Strain rate for constant 3 = 400kPa and
( 1 3 ) = 0.4 p kPa ( p = 250kPa )
57

0.0003

Strain Rate (/min) .


0.00025
0.0002
0.00015 3 =400.0 kPa
B B

1- 3=150.0 kPa
0.0001 B B B B

0.00005
0
1 10 100 1000 10000
ln (t) (min)
FIGURE 5.29-Time vs. Strain rate for constant 3 = 400kPa and
( 1 3 ) = 0.6 p kPa ( p = 250kPa )

The axial strain recorded for the test done at ( 1 3 ) = 0.6 p kPa (Fig.5.25) at

sec, 2 sec, 5 sec and 10 sec 0.0004, 0.000637 and 0.000692 respectively. The value of

axial strain rate at t = 0 was 0.01202. The strain rate (Fig.5.29) decreases sharply to

0.0001639, 2.41E-05 and 6.07E-07 in 1 min, 40 min and 1100 min. There was negligible

change in strain rate after 4000 min.

Fig.5.30 shows the plot of time vs. axial strain for the tests performed

for 3 = 100kPa for 0.4. p kPa and 0.6. p kPa . The axial strain for 0.6. p kPa is higher

than 0.4. p kPa for same time interval. The value of elastic modulus (E ) calculated from

CTC test for 3 = 100kPa is 33373kPa . The comparison between the calculated and

observed value of strains are shown in Table 5.2. It can be inferred that the total strain

recorded up to 10 sec is elastic strain. The plasticity growth in the samples had started

after 10 sec.
58

0.01
. 0.008
0.006
Axial Strain

0.004 60% 40%


0.002
0
0 2000 4000 6000
Time (min)

Figure 5.30- Axial stress-strain results for constant 3 = 100kPa and


( 1 3 ) = x. p kPa ( p = 106kPa ) (x= 0.4 and 0.6)

Table 5.2- The observed and calculated values of strains from creep test and CTC for
3 = 100kPa

( 1 3 )kPa i (Observed) at t = 10 sec e (Calculated from CTC)


0.4 p = 42.4 0.0013 0.00125
0.6 p = 63.6 0.0018 0.0018

Fig.5.31 shows the plot of axial strain and time for the tests performed

at 3 = 200kPa for 0.2. p kPa , 0.4. p kPa , 0.6. p kPa and 0.8. p kPa . The comparison

between the calculated and observed value of strains are shown in Table 5.3. The value of

elastic modulus (E ) calculated from CTC test for 3 = 200kPa is 44044kPa . The

comparison between the calculated and observed value of strain are shown in Table 5.3.

It can be inferred that the plasticity has started before 10 sec during the creep test.
59

0.08
.
0.06
20% 40% 60% 80%
Axial Strain

0.04

0.02

0
0 1000 2000 3000
Time (min)
Figure 5.31- Axial stress-strain results for constant 3 = 200kPa and
( 1 3 ) = x. p kPa ( p = 185kPa ) (x=0.2, 0.4, 0.6 and 0.8)

Table 5.3- The observed and calculated values of strains from creep test and CTC for
3 = 200kPa

( 1 3 )kPa i (Observed) at t = 10 sec e (Calculated from CTC)


0.2 p = 37 2.8254E-05 0.00084
0.4 p = 74 0.004121 0.00168
0.6 p = 111 0.01285 0.00252
0.8 p = 148 0.01964 0.00336

Fig.5.32 shows the plot of axial strain and time for the tests performed

at 3 = 400kPa for 0.2. p kPa , 0.4. p kPa and 0.6. p kPa . The comparison between the

calculated and observed value of strains are shown in Table 5.4 The value of elastic

modulus (E ) calculated from CTC test for 3 = 400kPa is 80000kPa . The comparison
60

between the calculated and observed value of strain are as shown in the Table 5.4, it can

be inferred that the plasticity has started before 10 sec during the creep test.

0.008
.

0.006

40% 60% 20%


Axial Strain

0.004

0.002

0
0 2000 4000 6000
Time (min)
Figure 5.32- Axial stress-strain results for constant 3 = 400kPa and
( 1 3 ) = x. p kPa ( p = 250kPa ) (x=0.2, 0.4 and 0.6)

Table 5.4- The observed and calculated values of strains from creep test and CTC for
3 = 400kPa

( 1 3 )kPa t (Observed) at t = 10 sec e (Calculated from CTC)

0.2 p = 50 0.000192 0.000625

0.4 p = 100 0.001259 0.00125

0.6 p = 150 0.000692 0.00187

The instantaneous strain t is calculated by assuming that it has occurred approximately

after t= 10 sec. This was adopted from the creep test data as shown in the Table 5.2, 5.3

and 5.4. Elastic strain was computed approximately for t= 10 sec by dividing
61

( 1 3 )kPa applied in the creep test by the Elastic Modulus (E). The Elastic Modulus

(E) is calculated from the CTC tests. The values of elastic strain can be constant. The two

values are included in 5.2, 5.3 and 5.4 for 100 kPa, 200 kPa and 400 kPa confining

pressure respectively.

It is difficult to measure the instantaneous strain i . Hence, it is approximately

adopted after 10 sec. However, for some tests e and i at 10 sec do not correlate well.

Then it may be appropriate to consider i for less than 10 sec.

0.015

0.012
Axial Strain

0.009
400kPa 200kPa
0.006

0.003

0
0 1000 2000 3000 4000 5000 6000 7000
Time (min)

Figure 5.33- Axial stress-strain results for constant 3 and ( 1 3 ) = 0.2. p kPa
62

. 0.015
0.012
0.009
Axial Strain

0.006
0.003 100kPa 200kPa 400kPa
0
0 2000 4000 6000
Time (min)
Figure 5.34- Axial stress-strain results for constant 3 and ( 1 3 ) = 0.4. p kPa

0.03
0.025
Axial Strain .

0.02
100kPa 200kPa 400kPa
0.015
0.01
0.005
0
0 1000 2000 3000
Time (min)
Figure 5.35- Axial stress-strain results for constant 3 and ( 1 3 ) = 0.6. p kPa
63

The values of elastic strain ( e ) calculated from CTC for 3 = 100kPa, 200kPa and

400kPa are 0.00127, 0.00168 and 0.00125 respectively. The calculated values correlated

quite well with the plots in Fig.5.33, Fig.5.34 and Fig.5.35.

1.4 Determination of Viscous Parameters

The following description is adopted from Samtani and Desai (1999) and Desai (2001).

In the Perzynas theory (Perzyna, 1966), the total strain tensor, ij , is decomposed into an

elastic part, ije , and a viscoplastic part , ijvp , as follows:

ij = ijvp + ije (5.1)

where,
1
ije = C ijkl
e
ilvp (5.2)

d ijvp F Q
= < > (5.3)
dt F0 ij

1 d ijvp
e
C ijkl , is the elastic tensor, is the inelastic strain rate tensor, is a material
dt

parameter (fluidity parameter) and is a function of F. Here F is the yield function and

Q is the potential function; for associative plasticity Q F . The angle bracket <> has the

meaning of switch-on off operator as follows.

F F F
< >=< >, if > 0 (5.4)
F
F0 F0 0

F F
and, < >= 0 , if 0 (5.5)
F0 F0
64

Squaring both sides we get,

2 2
d ijvp
= [ < >]2 Q (5.6)
dt
ij

Eq.5.6 can be rewritten as

2 2
1 d ij
= [ < >]2 1 Q

vp
(5.7)
2 dt
2 ij

The LHS is the second invariant of the inelastic strain- rate tensor, we have

2
dI 2vp 1 d ij
vp

= (5.8)
dt 2 dt

Thus, we can rewrite Eq.5.7 as

2
dI 2vp 2 1
Q
= [ < >] (5.9)
dt 2 ij

or,

dI 2vp
< >= dt (5.10)
1 Q Q
2 ij ij

Eq.5.10 is the general equation for creep experiments using Perzynas theory.

In general, Q = Q(J1 , J 2, D , J 3 D ) which, means that

Q Q Q Q 2
= ij + Sij + Sik S kj J 2 D ij (5.11)
ij J1 J 2 D J 3 D 3

Depending upon the condition of the test Eq.5.11 can either be utilized in the full

or only the dominant term can be considered.


65

dI 2vp
Let, a= dt (5.12)
1 Q Q
2 ij ij

Thus, we can write Eq.5.10 as,

< >= a (5.13)

The value of a is known from the experiments. In the creep test with constant

dI 2vp Q
stress, is found by measuring the visco-plastic strain (Eq.5.8). The gradient
dt ij

Q
can be calculated using Eq.5.11. Although ij is constant, the value will change as
ij

Q changes with increasing viscoplastic strain (hardening).

The choice of flow function < > , depends upon the experimental data. Since

F F
< > involves some function of , a plot of vs. a is required. Fig. shows a
F0 F0

schematic of such a plot. An appropriate function can be found to fit the curve by plotting

the experimental data in different spaces. Two commonly used functions are discussed

next.

1.5 General Procedure

The procedure for finding the viscous parameters and N was described by Samtani,

(1991).

1. Calculate J 1 , J 2 D before the application of the creep stress, ij . Thus, these J 1 ,

J 2 D would correspond to point A at t =0. (Fig.5.2a, c-1). Since, point A

represents steady state equilibrium condition under the present (consolidation)


66

stress, the yield surface passing through point A will be represented by F = 0 .

Find the value of ( =A) (Fig.5.2f) corresponding to this state of stress.


B B

2. Calculate J 1 , J 2 D after the application of the creep stress, ij at t =0+. These J 1 ,


P P

J 2 D would correspond to point B (Fig.5.2c-1) since the creep stress will remain

unchanged. At t = point B (Fig.5.2c-4) will be captured and the value of the

yield function will be F=0. Find the value of (=B) (Fig.5.2f) corresponding to
B B

this state of stress.

3. Calculate ( vp )A and ( vp )B using the hardening law determined from inviscid

plasticity analysis (e.g. the hyperbolic law, =


a1
1
) with vp = (d )(d ) .
vp vp

vp
ij ji

The trajectory of the viscoplastic strains is a function of time. Thus at time t= 0,

vp = ( vp )A and at time t= , vp = ( vp )B .

dI 2vp
4. Build a table containing vp , (t ) , F (t ) and = dt . Note that at
1 Q Q
2 ij ij

t= , vp = ( vp )B , = B and F = 0 .

F F F
5. Plot and on various different spaces. e.g. ln versus ln( ) .
F0 F0 F0

versus which would respectively yield the power law and the exponential law

for flow function.

6. Perform the regression analysis on the various spaces and select the space which

gives the best regression analysis, select the law corresponding to the selected

space and determine the fluidity parameter and the material parameter N . Note
67

that the material parameter N is dimensionless while the fluidity parameter has

the dimension of stress 1time 1 .

1.6 Material Parameters for Perzynas (evp) Model

There are two parameters, and N or N , that need to be determined from

appropriate laboratory tests as shown in Fig. . According to Perzynas theory, the total

strain, & , can be decomposed as

& = & e + & vp (5.14)


~ ~ ~

where,

& = C& & e (5.15)


~ ~ ~

or

& = C (e )1 & (5.15)


~ ~ ~

dQ
& vp = < > (5.16)
~ d
~

where, is the stress vector, C is the elastic constitutive matrix for isotropic
e
~ ~

matrix for material, Q is the plastic potential function; Q F for associative plasticity.

Squaring both sides of Eq.5.16 and multiplying them by 1 leads to


2

T
F
2 F

2
( )
1 vp T vp
& &
1
= ( < > )
2






(5.17)
~ ~

or
68

T
F
1 2 F
I&2vp = ( < > ) (5.18)
2
~ ~

where I&2vp is the second invariant of the viscoplastic strain rate tensor, & . Therefore,
vp
~

2 I&2vp
< >= T
=a (5.19)
F F


~ ~

The flow function by the power law is given by

N
F
= (5.20)
F0

Now Eq.5.20 is written as

N
F
=a (5.21)
F0

Therefore,

F
ln ( ) + N ln = ln (a ) (5.22)
F0

F
Plot of a vs. is shown schematically in Fig.5.36. The slope of the average
F0

F
straight line gives N , and the intercept when = 1 gives . Other forms of flow
F0

function are possible, e.g., the exponential form is given by

N
F
= exp 1.0 (5.23)
F0

Then the Eq.5.22 leads to


69

N
F
ln ( ) + N ln = ln (a ) (5.24)
F0

ln (a )
N

ln

ln F
F0

FIGURE 5.36 Determination of Creep Parameter (Desai, 2001)

ln(F/pa)
0

-2 0.0 0.5 1.0 1.5 2.0 2.5

3 =100.0 kPa
-4 B

B
B

1- 3= 63.6 kPa
B B B

-6

-8
-10

-12
-14
Figure 5.37-Determination of creep parameters from stress-strain results for constant
3 = 100kPa and ( 1 3 ) = 0.6 p kPa ( p = 106kPa )
70

The values of N and are calculated using the general procedure described by

N
F
Samtani, (1991). The Fig.5.37 to Fig.5.39 shows the plot of ln ( ) vs. ln . The
F0

slope of the curve gives the value of N and the y intercept gives the value of ln ( ) .

ln(F/pa)
0
0 1 2 3 4 5
-2
3 =200.0 kPa
B B

-4 1- 3 = 70.72 kPa
B B B B

-6

-8

-10

-12

Figure 5.38- Determination of creep parameters from stress-strain results for constant
3 = 200kPa and ( 1 3 ) = 0.4 p kPa ( p = 185kPa )
71

ln(F/pa)
0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
-2
3 =200.0 kPa
-4 B

B
B

1- 3 =104.58 kPa
B B B

-6

-8

-10

-12

-14
Figure 5.39- Determination of creep parameters from stress-strain results for constant
3 = 200kPa and ( 1 3 ) = 0.6 p kPa ( p = 185kPa )

TABLE 5.5- Values of creep parameters at constant axial stress 1 (kPa ) for various
confining stresses 3 (kPa ) .

3 (kPa ) 20% 1 (kPa ) 40% 1 (kPa ) 60% 1 (kPa ) 80% 1 (kPa )

(
kPa 1 min 1 ) N (
kPa 1 min 1 ) N (
kPa 1 min 1 ) N (
kPa 1 min 1 ) N

100kPa - - 6.23E-08 1.34 7.6028E-06 2.16 - -

200kPa 2.356E-06 1.203 9.782E-06 1.506 3.421E-05 2.068 5.843E-05 3.208

400kPa 8.453E-05 1.221 6.450E-05 1.260 4.620E-04 1.540 - -


72

TABLE 5.6- Average values of creep parameters at constant axial stress 1 (kPa ) for
various confining stresses 3 (kPa ) .

3 (kPa ) (
Average kPa 1 min 1 ) (kPa min ) (Pa s ) NAverage
B B

100kPa 3.83E-06 2.61E+05 1.57E+10 1.75

200kPa 2.62E-05 3.82E+04 2.29E+09 1.996

400kPa 2.04E-04 4.90E+04 2.94E+08 1.340

( )
The fluidity parameter kPa 1 min 1 is the inverse of viscosity (Pa s ) . The value of

viscosity from Table 2.1 and Table 2.3 shows good correlation with the observed values

(Table 5.6). The slight difference in the value of the viscosity may be due to the change

in till composition and hence change in physico- mechanical behavior at various

locations.

4
0.0094x
y = 0.8214e
3 2
R = 0.9769
2
N

0
0 40 80 120 160
Deviatoric stress (kPa)
FIGURE 5.40.- Variation of N with ( 1 3 ) kPa for 3 = 200kPa .
73

8.0E-05 2
y = 4E-09x - 7E-08x - 5E-07
Fluidity Parameter . 2
R = 0.9912
6.0E-05

4.0E-05

2.0E-05

0.0E+00
0 40 80 120 160
Deviatoric stress (kPa)
FIGURE 5.41- Variation of with ( 1 3 ) kPa for 3 = 200kPa .

The values of N and increase for 3 = 200kPa with the increase in the value

of deviatoric stress ( 1 3 ) (Fig.5.40, Fig.5.41).The minimum value of N calculated

from the trend is 1.08 at ( 1 3 ) 29kPa . Similarly, minimum value of is negative

at ( 1 3 ) 8.75kPa .

The values of N and increases for 3 = 400kPa with the increase in the value

of deviatoric stress ( 1 3 ) (Fig.5.42 and Fig.5.43). The minimum value

of N calculated from the trend is 1.01 at ( 1 3 ) 65kPa . Similarly, minimum value of

is 5.0 E 04 at ( 1 3 ) 62.5kPa .
74

2.0

1.5

1.0
N

2
0.5 y = 5E-05x - 0.0065x + 1.423

0.0
0 40 80 120 160
Deviatoric stress (kPa)

FIGURE 5.42- Variation of N with ( 1 3 ) kPa at 3 = 400kPa .

6.0E-04 2
y = 8E-08x - 1E-05x + 0.0005
Fluidity Parameter .

4.0E-04

2.0E-04

0.0E+00
0 40 80 120 160
Deviatoric stress (kPa)

FIGURE 5.43- Variation of with ( 1 3 ) kPa at 3 = 400kPa .


75

1.7 Back-prediction of Creep Tests

Due to specific forms of the stress and strain tensor representing the tests it is

possible to analytically back predict the creep tests. The procedure for back prediction of

creep test was described step by step for a particular time interval by Samtani, (1991).

1. Choose the same time interval as selected during the creep test. Knowing the

F
state of stress, find for the particular time interval, t.
F0

2. Knowing , N and from the flow function, determine the value of a. For the

Tiskilwa till we have,


N
F
a = (5.25)
F0

3. We can write the axial strain as

1 Q Q
& = 2a (5.26)
2 ij ij

4. Determine the magnitude of strain as follows,

d
i +1 = (t i +1 t i ) + i (5.27)
dt i +1

d
Where, is known from Eq.5.27, (ti +1 ti ) is the time step t is known from
dt i +1

step 1 and i is known from previous time step.

5. By starting from the known condition of i =0 when t=0, and repeating from step

1 to step 4, the creep test can be back-predicted.

Fig. 5.40 to Fig. 5.54 shows the back-prediction of the creep tests using the HISS

parameter obtained from the CTC tests. The HISS and DSC parameters were calculated
76

by CTC undrained test for 3 = 100 kPa, 200 kPa and 400 kPa (Shantanu, 2005). The

average value of creep parameters were used for all three confining pressures. The

parameter were used in one dimensional program for the prediction and validation of

laboratory experiments and later (Chapter-4) used in DSC- SST2D program for the

similar purpose. The average values of creep parameters (Table. 5.6) are used for back

prediction.

Table 5.7- HISS and DSC parameters for 3 = 100kPa (Sane, 2005)

Elastic HISS Hardening


Parameters Parameters Parameters

E 33733 kPa 0.01648 a1 1.29E-07


0.4 0.3344 1 1.4

Disturbance Parameters R 35.342 kPa


Du 1.0 m -0.5
A 15.89 n 3.6
Z 2.08

Note- p a = 101.235 kPa, 0 =0 and S r =1


77

0.01

0.008
.

3 =100.0 kPa
0.006
B B
Axial Strain

1- 3= 42.4 kPa
B B B B

0.004
Back prediction Experiment
0.002

0
0 1000 2000 3000 4000 5000 6000
Time (min)
Figure 5.44- Back prediction of stress-strain results for constant 3 = 100kPa and
( 1 3 ) = 0.4 p kPa ( p = 106kPa )

0.01

0.008
Axial Strain

3 =100.0 kPa
0.006
B B

1- 3 = 63.6 kPa
B B B B

0.004
Experimental Backprediction
0.002

0
0 1000 2000 3000 4000
Time (min)
Figure 5.45- Back prediction of stress-strain results for constant 3 = 100kPa and
( 1 3 ) = 0.6 p kPa ( p = 106kPa )
78

The back predictions by the one -dimensional procedure for 3 = 100kPa are

shown in Fig.5.44 to Fig.5.46. The back predictions for both cases show satisfactory

results.

The back prediction value is higher than the experimental for

( 1 3 ) = 0.4 p kPa (Fig.5.44).

The back prediction for ( 1 3 ) = 0.6 p kPa (Fig.5.45) shows increase in strain

rate after 4000 min. The elastic strain value from the back prediction is higher as

compared to experimental value. This may be due to neglecting the effect of stain value

in the other two directions for one-dimensional procedure.

0.01
.

0.008

0.006
Axial Strain

3 =100.0 kPa
B B

0.004 1- 3 = 63.6 kPa


B B B B

Back prediction Experiment


0.002
0
0 1000 2000 3000 4000
Time (min)

Figure 5.46- Back prediction of stress-strain results for constant 3 = 100kPa and
( 1 3 ) = 0.6 p kPa ( p = 106kPa ) (Independent test)
79

The validation by the one dimensional procedure for 3 = 200kPa is shown in

Fig.5.47- Fig.5.51.The one dimensional procedure provides good results as compared to

the experimental result.

Table-5.8- HISS and DSC parameters for 3 = 200kPa (Sane, 2005)

Elastic HISS Hardening


Parameters Parameters Parameters
E 44044 kPa 0.01648 a1 1.15E-08
0.4 0.5691 1 1.4
Disturbance Parameters R 35.342 kPa
Du 1.0 m -0.5
A 15.89 n 3.6
Z 2.08

Note- p a = 101.235 kPa, 0 =0 and S r =1

0.0025

0.002
.

0.0015 3 =200.0 kPa


B B

1- 3 =34.85 kPa
Axial Strain

B B B B

0.001

Back prediction Experiment


0.0005

0
0 2000 4000 6000 8000 10000 12000
Time (min)

Figure 5.47- Back prediction of stress-strain results for constant 3 = 200kPa and
( 1 3 ) = 0.2 p kPa ( p = 185kPa )
80

. 0.018
0.015
0.012
3 =200.0 kPa
Axial Strain

B B

0.009 1- 3 =72.72 kPa


B B B B

0.006
Back prediction Experiment
0.003
0
0 500 1000 1500 2000 2500 3000 3500 4000
Time (min)
Figure 5.48- Back prediction of stress-strain results for constant 3 = 200kPa and
( 1 3 ) = 0.4 p kPa ( p = 185kPa )

0.04
.

0.03
Axial Strain

0.02
3 =200.0 kPa
B B

1- 3 =104.58 kPa
B B B B

0.01
Back prediction Experiment

0
0 500 1000 1500 2000 2500 3000
Time (min)
Figure 5.49- Back prediction of stress-strain results for constant 3 = 200kPa and
( 1 3 ) = 0.6 p kPa ( p = 185kPa )
81

0.04
.

0.03
Axial Strain

3 =200.0 kPa
B B

0.02 1- 3 =104.58 kPa


B B B B

0.01 Back prediction Experiment

0
0 500 1000 1500 2000 2500 3000
Time (min)

Figure 5.50- Back prediction of stress-strain results for constant 3 = 200kPa and
( 1 3 ) = 0.6 p kPa ( p = 185kPa ) (Independent test)

0.1

0.08
.

0.06
Axial Strain

3 =200.0 kPa
B B

1- 3 = 139.40 kPa
0.04
B B B B

Back prediction Experiment


0.02

0
0 500 1000 1500 2000 2500 3000
Time (min)
Figure 5.51- Back prediction of stress-strain results for constant 3 = 200kPa and
( 1 3 ) = 0.8 p kPa ( p = 185kPa )
82

The back prediction for ( 1 3 ) = 0.2 p kPa (Fig.5.47) shows satisfactory results

in the elastic range. The back prediction values of strain are higher after 500 min to 4000

min. The values of strains are similar after 4500 min. The back prediction gives higher

values of strain for other three cases (Fig.5.48, Fig.5.49 and Fig.5.51). The strain rate is

also higher for the given value of time interval. This is due to neglecting the strains in

other two directions. The strain rate increases for long time interval.

The validation by the one dimensional procedure for 3 = 200kPa is shown in

Fig.5.50. The one dimensional procedure provides good results as compared to the

experimental result expect for the time t > 1500 min the values predicted by the one

dimensional procedure gives high value as compared to Independent creep test.

Table 5.9- HISS and DSC parameters for 3 = 400kPa (Sane, 2005)

Elastic HISS Hardening


Parameters Parameters Parameters
E 80000 kPa 0.01648 a1 2.10E-09
0.4 0.5691 1 1.2
Disturbance Parameters R 35.342 kPa
Du 1.0 m -0.5
A 15.89 n 3.6
Z 2.08

Note- p a = 101.235 kPa, 0 =0 and S r =1


83

0.0028
. 0.0024
0.002
0.0016 3 =400.0 kPa
Axial Strain

B B

1- 3 =50.0 kPa
0.0012 B B B B

0.0008
0.0004 Back prediction Experiment

0
0 1000 2000 3000 4000 5000 6000
Time (min)
Figure 5.52- Back prediction of stress-strain results for 3 = 400kPa and constant
( 1 3 ) = 0.2 p kPa ( p = 250kPa )

0.008
.

0.006
3 =400.0 kPa
Axial Strain

B B

1- 3 =100.0 kPa
0.004
B B B B

0.002 Back prediction Experiment

0
0 2000 4000 6000 8000
Time (min)
Figure 5.53- Back prediction of stress-strain results for 3 = 400kPa and constant
( 1 3 ) = 0.4 p kPa ( p = 250kPa )
84

0.0088
.
0.0066
3 =400.0 kPa
Axial Strain

B B

1- 3 =150.0 kPa
0.0044 B B B B

0.0022 Back prediction Experiment

0
0 1000 2000 3000 4000 5000 6000
Time (min)

Figure 5.54- Back prediction of stress-strain results for 3 = 400kPa and constant
( 1 3 ) = 0.6 p kPa ( p = 250kPa )

The back predictions by the one -dimensional procedure for 3 = 400kPa are

shown in Fig.5.52 to Fig.5.54. The back predictions for all cases show satisfactory

results.

The back prediction for ( 1 3 ) = 0.2 p kPa (Fig.5.52) shows high value of

strain in the elastic range. The back prediction values of strain are lower after 400 min to

5000 min. The back prediction gave higher value of strain after 5000 min.

The back prediction curves give similar trend for other two cases (Fig.5.53 and

Fig.5.54). The values of strain are higher after 4500 min. The reason for it may be

neglecting the strains in other two directions. The strain rate also increases for long time

interval.
85

1.8 Material Parameters in Overlay Model

The parameters in the Overlay models are determined from the triaxial creep test

performed in the laboratory. The following procedure to determine the creep parameters

for Viscoelastic model and Viscoelasticviscoplastic model is adopted from Desai (2001).

1.8.1 Viscoelastic (ve) Overlay Model

Consider the model in Fig. . It is required to determine four elastic (E1 , 1; E2 , 2 )

and two viscous (1 , N 1 ) parameters. They can be obtained from the following simplified

procedure.

For t = 0 . Eq. gives,

0
0 = (5.28)
E1t1 + E 2 t 2

As, t , Eq. gives the ultimate strain, u , as

0
u = (5.29)
E2t 2

The values of 0 and u can be obtained from the measured response. The later can be

adopted as an asymptotic value. Then the solution of Eq. provides the values of E1 and

d
E 2 .Now the gradient of (0) at t = 0 can be measured approximately from stress-
dt

strain curve as

d
(0) t +1 t (5.30)
dt t 0

where, t +1 and t are the values of strains near t = 0 , and t is a small time step.

Substitution of Eq. in Eq. gives


86

d E12 t1 0
(0) = (5.31)
dt (E1t1 + E 2 t 2 )2
Substituting of the E1 and E 2 computed above in Eq. gives the value of .

The values of 1 and 2 can be found by the relation (Desai, 2001)

1 E1 1 E
1 = 1 and 2 = 1 2 (5.32)
2 3K 2 3K

where,

1
K= (5.33)
3 1

1.8.2 Viscoelasticviscoplastic (vevp) Overlay Model

The model parameters are as shown in Fig. Assume that the significant

viscoplastic strains will not occur during earlier times. This will approximately be the

situation be the situation if the irreversible strains do not occur until the yield stress, y ,

in the slider is exceeded. In that case, parameter (E1 , 1 , 1 , 1 , E 2 , 2 ) can be adopted to

be same as those of the viscoelastic model. The parameter 2 , 2 and y (F ) can be

adopted to be same as those for the elastoviscoplastic (evp) Perzynas model.


87

CHAPTER 6

COMPUTER IMPLEMENTATION OF THE DSC AND FINITE

ELEMENT ANALYSIS

1.1 Introduction

The DSC has been implemented into two- dimensional Finite Element computer

program DSC-SST2D (Desai, 2001 & Desai, 1999). A brief formulation of the Finite

Element equation and a description is given below.

1.2 Results and Discussions

DSC_SST2D program is developed by Desai and his co-workers for static

problems. The values of the parameter for evp and vevp model were used to make the

data file for DSC_SST2D finite element program (Desai, 1999). This program is used to

study and analyze the effectiveness of the constitutive model. The results from the

program are compared with the experimental observations. Fig.6.4 to Fig. 6.27 shows the

plot of DSC- vevp Overlay model and compares them with the experimental results.

Fig.6.4 to Fig.6.27 shows the results from the DSC-vevp Overlay models using the

HISS, disturbance and creep parameters obtained from the CTC tests. The HISS and DSC

parameters were calculated by CTC- Undrained test and creep parameters were obtained

from triaxial drained creep tests.


88

0.01

0.008
.

0.006 3 =100.0 kPa


Axial Strain

B B

1- 3 = 42.4 kPa
B B B B

0.004

0.002 FEM_vevp Experiment FEM_evp

0
0 1000 2000 3000 4000 5000 6000
Time (min)
Figure 6.4 - Finite element prediction of stress-strain results for constant 3 = 100kPa
and ( 1 3 ) = 0.4 p kPa ( p = 106kPa )

0.01

0.008
.

0.006
Axial Strain

3 =100.0 kPa
B B

0.004 1- 3 = 63.6 kPa


B B B B

0.002 Experiment FEM_vevp FEM_evp

0
0 1000 2000 3000 4000 5000 6000 7000
Time (min)
Figure 6.5 - Finite element prediction of stress-strain results for constant 3 = 100kPa
and ( 1 3 ) = 0.6 p kPa ( p = 106kPa )
89

Fig.6.4 and Fig.6.5 shows the predictions by the DSC- vevp overlay model and

DSC- evp overlay model for 3 = 100kPa . The value for strain predicted is little less at

the initial stage but matches well with the experimental results.

The plot for ( 1 3 ) = 0.4 p kPa (Fig.6.4) shows a good correlation between

the experimental and vevp model results. The evp model predicts lower values initially

and higher values of strain after t > 3000 min . The strain rate by vevp model is almost

equal to experimental results for similar time interval.

The plot for ( 1 3 ) = 0.6 p kPa (Fig.6.5) shows a good comparison between

the vevp overlay model and experimental results. The evp model underestimates the

values of strain but shows increasing trend. The values of strain predicted by the vevp

model are little bit higher for long time range than for short time.

0.0030

0.0025
.

0.0020
Axial Strain

0.0015 3 =200.00 kPa


B B

1- 3 =34.85 kPa
B B B B

0.0010

0.0005 FEM_vevp Experiment FEM_evp

0.0000
0 3000 6000 9000 12000 15000 18000
Time (min)
Figure 6.6- Finite element prediction of stress-strain results for constant 3 = 200kPa
and ( 1 3 ) = 0.2 p kPa ( p = 185kPa )
90

Fig.6.6 to Fig.6.9 shows the predictions by the DSC-creep vevp overlay model

for 3 = 200kPa . The prediction results by the model are very close to the experimental

values.

The plot for ( 1 3 ) = 0.2 p kPa (Fig.6.6) shows a good correlation between

the experimental and vevp model prediction. The values of strain predicted are almost

similar to the experimental results. The strain rate is almost equal to experimental results

for all time intervals and shows the similar trend. The evp model predicts fewer values of

strains initially but the values are higher for t > 12000 .

0.018

0.015
Axial Strain .

3 =200.00 kPa
0.012 B

B
B

1- 3 =70.72 kPa
B B B

0.009

0.006

0.003 Experimental FEM_vevp FEM_evp

0.000
0 2000 4000 6000 8000
Time (min)

Fig 6.7- Finite element prediction of stress-strain results for constant 3 = 200kPa and
( 1 3 ) = 0.4 p kPa ( p = 185kPa )
91

0.035
0.03
.

0.025
Axial Strain

0.02 3 =200.00 kPa


B B

1- 3 =104.58 kPa
B B B B

0.015
0.01
0.005 Experiment FEM_vevp FEM_evp
0
0 500 1000 1500 2000 2500 3000
Time (min)

Figure 6.8- Back prediction of stress-strain results for constant 3 = 200kPa and
( 1 3 ) = 0.6 p kPa ( p = 185kPa )

0.08

0.06
Axial Strain .

3 =200.00 kPa
0.04 B

B
B

1- 3 =139.40 kPa
B B B

0.02
Experiment FEM_vevp FEM_evp

0
0 500 1000 1500 2000 2500 3000
Time (min)
Figure 6.9- Back prediction of stress-strain results for constant 3 = 200kPa and
( 1 3 ) = 0.8 p kPa ( p = 185kPa )
92

The plot for ( 1 3 ) = 0.4 p kPa (Fig.6.7) shows a good correlation between the

experimental and vevp model prediction. The values of strain predicted by the model are

little less initially up to 2000 min. The model predicts the results quite well later. The

strain rate shows similar trend for all time interval. The strain values by evp model

shows lower values up to 2000 min and later the values are higher than experimental

results.

The plot for ( 1 3 ) = 0.6 p kPa (Fig.6.8) compares good with the experimental

results. Very less or no error is observed at any time. The strain rate also shows good

correlation with the experimental values. The evp model predicts lower values of strains

up to 1500 min and later shows increase in the values of strains.

The plot for ( 1 3 ) = 0.8 p kPa (Fig.6.9) compares good with the experimental

results. The values of strain are almost similar initially up to 300 min. The strain values

are little higher than the experimental results. The strain rate also shows good correlation

with the experimental values at any time interval.


93

. 0.0025

0.002

0.0015 3 =400.00 kPa


Axial Strain

B B

1- 3 =50.00 kPa
B B B B

0.001

0.0005 Experiment FEM_vevp

0
0 2000 4000 6000
Time (min)
Figure 6.10- Finite element prediction of stress-strain results for constant 3 = 400kPa
and ( 1 3 ) = 0.2 p kPa ( p = 250kPa )

0.008

0.006
Axial Strain .

3 =400.00 kPa
0.004
B B

1- 3 =100.00 kPa
B B B B

0.002
Experiment FEM_vevp
0.000
0 2000 4000 6000 8000
Time (min)
Figure 6.11- Finite element prediction of stress-strain results for constant 3 = 400kPa
and ( 1 3 ) = 0.4 p kPa ( p = 250kPa )
94

0.008

0.006
Axial Strain .

3 =400.00 kPa
0.004
B B

1- 3 =150.00 kPa
B B B B

0.002 Experiment FEM_vevp

0
0 2000 4000 6000
Time (min)
Figure 6.12- Finite element prediction of stress-strain results for constant 3 = 400kPa
and ( 1 3 ) = 0.6 p kPa ( p = 250kPa )

Fig.6.10 to Fig.6.12 shows the predictions by the DSC-creep vevp overlay model

for 3 = 400kPa . The prediction results by the model are very close to the experimental

values.

The plot for ( 1 3 ) = 0.2 p kPa (Fig.6.10) compares results quite well with the

experimental results. The values of strain predicted by the model are little high up to

4000 min. The model predicts the results quite well later. The strain rate shows similar

trend for all time interval.

The plot for ( 1 3 ) = 0.4 p kPa (Fig.6.11) and ( 1 3 ) = 0.4 p kPa (Fig.6.12)

compares good with the experimental results. Very less or no error is observed at any

time. The strain rate also shows good correlation with the experimental values.
95

The results in the plots Fig.6.4 to Fig.6.11 compares well with the experimental

results for 3 = 100kPa, 200kPa and 400kPa. The comparison of the results proves that

DSC- vevp Overlay model closely describes the behavior of subglacial till.

The DSC-vevp model is then used to study the unloading process for all the three

cases i.e., 3 = 100kPa, 200kPa and 400kPa. The results from the unloading curves are as

shown in the Fig.6.12 to Fig.6.23. The plot from the unloading curves were used to

calculate the viscoelastic (ve) and viscoplastic (vp) part for all cases as shown in the

Table 6.2 to Table 6.4

0.008

0.006
Axial Strain .

0.004 3 =100.00 kPa


B B

1- 3= 42.00 kPa
B B B B

0.002

0.000
0 4000 8000 12000 16000
Time (min)

Figure 6.12-Stress-strain results for constant 3 = 100kPa and


( 1 3 ) = 0.4 p kPa ( p = 106kPa ) from FE-vevp Overlay model
96

0.01

0.008
Axial Strain .

0.006
3 =100.00 kPa
B B

1- 3 = 63.60 kPa
0.004
B B B B

0.002

0
0 3000 6000 9000 12000
Time (min)

Figure 6.13-Stress-strain results for constant 3 = 100kPa and


( 1 3 ) = 0.6 p kPa ( p = 106kPa ) from FE- vevp Overlay model

Table 6.1- Values of e , vp and ve for 3 = 100kPa .

( 1 3 )kPa i (t = 6500 min ) e ( CTC) vp ve


0.4 p = 42.4 0.00782 0.00125 0.0061 0.00041
0.6 p = 63.6 0.00933 0.0018 0.007 0.00055

The results shows increase in the value of viscoplastic ( vp ) and viscoelastic

strain ( ve ) with increase in deviatoric stress ( 1 3 )kPa for 3 = 100kPa as shown in

the Table.6.1. The value of ve is less as compared to vp for same vale of ( 1 3 )kPa .
97

0.0028
0.0024
.

0.0020
0.0016
Axial Strain

0.0012
3 =200.00 kPa
B B

0.0008 1- 3 = 34.85 kPa


B B B B

0.0004
0.0000
0 5000 10000 15000 20000 25000 30000
Time (min)
Figure 6.14 -Stress-strain results for constant 3 = 200kPa and
( 1 3 ) = 0.2 p kPa ( p = 185kPa ) from FE- vevp Overlay model

0.015

0.012
.
Axial Strain

0.009

0.006 3 =200.00 kPa


B B

1- 3 = 70.72 kPa
B B B B

0.003

0.000
0 3000 6000 9000 12000 15000 18000
Time (min)
Figure 6.15 -Stress-strain results for constant 3 = 200kPa and
( 1 3 ) = 0.4 p kPa ( p = 185kPa ) from FE- vevp Overlay model
98

0.032
0.028
0.024
.

0.02
Axial Strain

0.016
3 =200.00 kPa
0.012
B B

1- 3 =104.58 kPa
B B B B

0.008
0.004
0
0 2000 4000 6000
Time (min)
FIGURE-6.16 Stress-strain results for constant 3 = 200kPa and
( 1 3 ) = 0.6 p kPa ( p = 185kPa ) from FE- vevp Overlay model

0.080
.

0.060
Axial Strain

0.040
3 =200.00 kPa
B B

1- 3 =139.40 kPa
B B B B

0.020

0.000
0 2000 4000 6000
Time (min)
FIGURE- 6.17 Stress-strain results for constant 3 = 200kPa and
( 1 3 ) = 0.8 p kPa ( p = 185kPa ) from FE-vevp Overlay model
99

Table 6.2-Values of e , vp and ve for 3 = 200kPa .

( 1 3 )kPa i e (CTC) vp ve
0.2 p = 37 0.002609 (t = 18300 min) 0.00084 0.00175 9.0E-06
0.4 p = 74 0.0145 (t = 8335 min) 0.00168 0.011 0.00182
0.6 p = 111 0.02958 (t = 3335 min) 0.00252 0.02 0.00706
0.8 p = 148 0.0765 (t = 3335 min) 0.00336 0.06 0.0134

0.08 Total strain

Elastic strain
0.06
Viscoplastic
Strain

0.04 strain
Viscoelastic
0.02 strain

0
0 50 100 150 200
Deviatoric Stress (kPa)

Fig.6.18- Plot of ( 1 3 )kPa vs. Strain for 3 = 200kPa .

The results (Fig.6.18) shows increase in the value of Viscoelastic strain ( ve ) and

Viscoplastic strain ( vp ) with increase in deviatoric stress ( 1 3 )kPa for 3 = 200kPa

as shown in the Table.6.2. Also, the value of ve is less as compared to vp for a given

value of deviatoric stress ( 1 3 )kPa . The plot (Fig.6.18) shows high rate of increasing

in the value of vp and very less rate of increase in ve . Thus, it can be concluded that
100

viscoelastic strain in very less significant for higher values of deviatoric stress

for 3 = 200kPa .

0.0025
.

0.002

0.0015
Axial Strain

0.001 3 =400.00 kPa


B B

1- 3 =50.00 kPa
B B B B

0.0005

0
0 3000 6000 9000 12000 15000
Time (min)
FIGURE-6.19 Stress-strain results for constant 3 = 400kPa and
( 1 3 ) = 0.2 p kPa ( p = 250kPa ) from FE- vevp Overlay model

0.008

0.006
Axial Strain .

0.004
3 =400.00 kPa
0.002
B B

1 =500.00 kPa
B B

0.000
0 3000 6000 9000 12000
Time (min)
FIGURE-6.20 Stress-strain results for constant 3 = 400kPa and
( 1 3 ) = 0.4 p kPa ( p = 250kPa ) from FE- vevp Overlay model
101

0.008
.
0.006
Axial Strain

0.004
3 =400.00 kPa
B B

1- 3=150.00 kPa
B B B B

0.002

0
0 2000 4000 6000 8000 10000
Time (min)
FIGURE 6.21- Stress-strain results for constant 3 = 400kPa and
( 1 3 ) = 0.6 p kPa ( p = 250kPa ) from FE- vevp Overlay model

Total strain Elastic strain


Viscoplastic strain Viscoelastic strain
0.0084

0.0063
Strain

0.0042

0.0021

0
0 50 100 150 200
Deviatoric Stress (kPa)

Fig 6.22- Plot of ( 1 3 )kPa vs. Strain for 3 = 400kPa .


102

Table 6.3- Values of e , vp and ve for 3 = 400kPa .

( 1 3 )kPa i e (CTC) vp ve
0.2 p = 50 0.00243 (t = 6335 min) 0.000625 0.0018 5.0E-06
0.4 p = 100 0.00701 (t = 7500 min) 0.00125 0.0049 0.00086
0.6 p = 150 0.00754 (t = 5000 min) 0.00187 0.0051 0.00057

The results (Fig.6.22) shows increase in the value of Viscoelastic strain ( vp ) and

Viscoplastic strain ( ve ) with increase in deviatoric stress ( 1 3 )kPa for 3 = 400kPa

as shown in the Table.6.3. The plot (Fig.6.22) shows high rate of increasing in the

viscoplastic strain and almost no increase in the values of viscoelastic strain . Thus, it can

be concluded that viscoelastic strain in very less significant for higher values of

deviatoric stress for 3 = 400kPa .

Total strain Elastic strian


Viscoplastic strain Viscoelastic strain
0.016

0.012
Strain

0.008

0.004

0
0 100 200 300 400 500
Sigma 3 (kPa)
Fig- 6.23 Plot of 3kPa vs. Strain for ( 1 3 ) = 0.4 p kPa
103

Fig.6.23 shows the plot of 3kPa vs. ( 1 3 ) = 0.4 p kPa for different 3kPa .

The curve shows high values of viscoplastic strain as compared to viscoelastic strain for

given values of 3kPa . The Viscoplastic strain contributes more towards total strain than

viscoelastic strain. The value of viscoplastic and viscoelastic strain shows similar trend.

Fig.6.24 shows the quadratic trendline plot for the values of viscoelastic and

viscoplastic strain. The maximum value of ve is 0.00145 for 3 250kPa . Similarly, the

maximum value of vp is 0.0098 for 3 166.67 kPa .

Viscoplastic strain Viscoelastic strain


Poly. (Viscoplastic strain) Poly. (Viscoelastic strain)
0.016
2
y = -3E-07x + 0.0001x - 0.0041
0.012
Strain

0.008
2
y = -6E-08x + 3E-05x - 0.0023
0.004

0
0 100 200 300 400 500
Sigma 3 (kPa)

Fig-6.24 Trend line plot of 3kPa vs. Strain for ( 1 3 ) = 0.4 p kPa
104

Total Strain Elastic strain


Viscoplastic stain Viscoelatic strain
0.03
0.025
0.02
Strain

0.015
0.01
0.005
0
0 100 200 300 400 500
Sigma 3 (kPa)
Fig.6.25- Plot of 3kPa vs. Strain for ( 1 3 ) = 0.6 p kPa

Viscoplastic stain Viscoelatic strain


Poly. (Viscoplastic stain) Poly. (Viscoelatic strain
0.025 2
y = -7E-07x + 0.0003x - 0.0196
0.02
0.015
Strain

2
y = -3E-07x + 0.0002x - 0.0125
0.01
0.005
0
0 100 200 300 400 500
Sigma 3 (kPa)

Fig.6.26- Trend line plot of 3kPa vs. Strain for ( 1 3 ) = 0.4 p kPa
105

Fig.6.25 shows the plot of 3kPa vs. ( 1 3 ) = 0.6 p kPa for different 3kPa .

The curve shows very high values of viscoplastic strain as compared to viscoelastic strain

for given values of 3kPa . The viscoplastic strain contributes more towards total strain

than viscoelastic strain. There is not significant change in the value of viscoelastic strain

as compared to viscoplastic strain.

Fig.6.26 shows the quadratic trendline plot for the values of viscoelastic and

viscoplastic strain. The maximum value of ve is 0.0125 for 3 333.33kPa . Similarly,

the maximum value of vp is 0.021 for 3 214.28kPa .


106

CHAPTER 7

SUMMARY AND CONCLUSIONS

A constitutive model based on Multicomponent DSC (MDSC) to predict the creep

behavior of subglacial till has been used (Desai, 2001). The model takes into account the

factor such as induced non-associative ness, anisotropy, softening due to creep loading,

inelastic response during loading for large time interval. The model also takes into

account the elastic (e ) , viscoelastic (ve ) , viscoplastic (vp ) and inviscid plasticity during

the creep loading.

The model considers that the actual response can be related to two reference states

called the intact state and the fully disturbed state through a disturbance function D. The

intact state is modeled by the hierarchical associative plasticity model hardening

isotropically, 0 and the fully disturbed state is modelled by the critical state.

The inelastic loading is modeled by using the interpolation functions

which are made function of the disturbance. The disturbance increases during loading and

increases with time.

Laboratory creep experiments were done on simple triaxial creep device. The

experiments were performed in drained conditions on Tiskilwa subglacial till.

Verification of the model is done with respect to the laboratory creep results and

shows that the model can predict well, the drained creep behavior of subglacial till. The

predictions of drained creep tests shows that reasonably good predictions of these tests

can be obtained by using the parameters found only from undrained CTC test and drained

creep tests.
107

1.1 New Contribution

Significant new contributions are achieved in this research. They are

1. A constitutive model based on Disturbed State Concept (DSC) is applied to

predict the drained creep behavior of normally consolidated fully saturated

subglacial till

2. The creep parameters in the MDSC model have been evaluated from the creep

tests conducted by using the triaxial creep device.

3. The model with the average parameters is used to predict the test data from which

the parameters were determined and independent tests not used to find the

parameter.
108

REFERENCES

1. Alley, R.B., Blankenship, D.D., Bentley, C.R., and Rooney, S.T., Deformation

of till beneath ice stream B, West Antarctica, Nature, Vol. 322, 1986, pp 57-59.

2. Alley, R.B., Blankenship, D.D., Rooney, S.T., and Bentley, C.R., Till Beneath

Ice Stream B4- A Coupled Ice-till Flow Model, Journal of Geophysical

Research, v. 92, 1987b, pp 8931-8940.

3. Annual Report National Science Foundation-Sponsored Research Project EAR-

0229513, Collaborative Research: Testing and Constitutive Modeling of Fine-

Grained Glacial Tills Deposited by the Laurentide Ice Sheet, 2004- 2005.

4. Blake, E.W., The Deformation Bed beneath a Surge- Type Glacier:

Measurement of Mechanical and Electrical Properties. PhD thesis, Univ. of

British Columbia, Vancouver, 1992.

5. Blankenship, D.D., Benteley C.R., Rooney S.T. and Alley R.B. Till beneath Ice

Stream B: 1. Properties Derived from Seismic Travel Times. J.Geophy. Res.,92,

1987, pp 8903-8911.

6. Boulton G.S. and Dent D.L.,The nature and rates of post- depositional changes

in recently deposited till from south-east Iceland. Geogr. Ann., 56-A, 1974, pp

121-134.

7. Boulton G.S., Dent D.L. and Morris E.M. Subglacial Shearing and Crushing,

and the Role of Water Pressure in Tills from South- East Iceland. Geogr. Ann.,

56A, 1974, pp 135- 145.


109

8. Boulton, G.S. and Hindmarsh R.C.A., Sediment deformation beneath glaciers:

rheology and geological consequences. Jou. Geophy. Res., 92, 1987, pp-9059-

9082.

9. Boulton, G.S. and Jones A.S.,Stability of Temperate Ice caps and Ice sheets

Resting on Beds of Deformable Sediment, Jou. Glaciol., 24, 1979, pp 29- 42.

10. Clarke, G.K.C., Sub-glacial Till: A Physical Framework for its Properties and

Processes. Jou. Geophy. Res., 92, 1987(b), 9023- 9036.

11. Clarke, G.K.C., Collins S.G. and Thompson D.E., Flow, Thermal Structure, and

Subglacial Conditions of the Surge-type Glacier, Can. J. Earth Sci., 21, 1984, pp

232- 240.

12. Desai C.S., Mechanics of Material and Interface- The Disturbed State Concept,

CRC Press, USA, 2001.

13. Desai C.S., User Manual- DSC-SST2D, Computer code for Static, Dynamic,

Creep and Thermal Analysis: Solid, Structure, and Soil- Structure Problems,

Part- 1, 2 and 3, Version 2, Tucson, Arizona, USA, 1999.

14. Desai C.S., and Zhang. D., Viscoplastic Model for Geological Materials and

Generalized Flow Rule, Int. J. Num. Analyt. Meth. in Geomech., Vol.11, 1987

pp 603- 620.

15. Engelhardt, H.F., Harrison W.D., and Kamb B., Basal Sliding and Conditions at

the Glacier Bed as Revealed by Bore-hole Photography, Jou. Glaciol., 20, 1978

pp 469- 508.

16. Fischer, U.H. and G.K.C. Clarke, Ploughing of Subglacial Sediment. J.Glaciol,

40, 1994.
110

17. Humphrey, N., Kamb, B., Fahnestock, M., and Engelhardt, H., Characteristics of

the bed of the lower Columbia Glacier, Alaska: Journal of Geophysical Research,

v. 98, 1993, pp 837-846.

18. Johnson, A.M.,Physical Processes in Geology, Freeman, Cooper, and Co., San

Fransico, Calif., 1970.

19. Kamb, B.Rheological Non-linearity and Flow Instability in the Deforming Bed

Mechanism of Ice Stream Motion. Jou. Geophys. Res., 96, pp 16585- 16595,

1991.

20. Lama R.D. and Vutukuri V.S.,Handbook on Mechanical Properties of Rocks,

Vol-3, Trans Tech Pub., 1978.

21. Lingle, C.S., and Brown, T.J.,A Subglacial Aquifer Bed Model and Water

Pressure Dependent Basal Sliding relationship for a West Antarctic Ice- stream,

in van der Veen, C.J., and Oerlemans, J., eds., Dynamics of the West Antarctic

Ice Sheet: D. Reidel Publishing Company, 1987, pp. 249-285.

22. MacAyeal, D.R.,Large- scale Ice Flow over Viscous Basal Sediment: Theory

and application to Ice Stream B Antartica Jou. geophys. Res., 94, 1989, pp

4071- 4087.

23. MacClintock, P and Dreimanis A.,Reorientation of Till Fabric by Overriding

Glacier in the St. Lawrence Valley. Am. J. Sci., 262, 1964, pp 133-142.

24. Mathews, W.H.,Surface profile of the Laurentide Ice Sheet in its Marginal

Areas. J. Glaciol., 13, 1974, pp 37-43.

25. Patersons, W.S.B., The Physics of Glaciers, 3rd ed., Pergamon Press, New York,
P P

1994
111

26. Perzyna. P., Fundamental Problems in Viscoplasticity, Adv. in Appl. Mech., 9,

1966, pp 243- 277.

27. Roy, M., 1998, Pleistocene stratigraphy of the Lower Nelson River Area:

Implications for the evolution of the Hudson Bay Lowland of Manitoba, Canada.

Unpublished MS Thesis, Universite du Quebec a Montreal, Montreal, Quebec,

Canada.

28. Samtani, N.C. and Desai C.S., Constitutive Modelling and Finite Element

Analysis of Slowly moving landslide using Hierarchical Viscoplastic Material

Model. Report to NSF, Dept. of Civil Engineering, The Univ. of Arizona,

Tucson, USA, 1991.

29. Sane S., Constitutive Modelling and Finite Element Analysis of Glacial Till.

PhD thesis, Tucson, USA., 2005.(under preparation)

30. Turcotte, D.L., and Schubert G., Geodynamics, John Wiley, New York, 1982.

31. Weast, R.C. (Ed.),Handbook of Chemistry and Physics, 54th ed., CRC Press,
P P

Cleveland, Ohio, 1973.

You might also like