Primary Consolidation and Creep of Clays

Samson Abate Degago

CREBS IV, Delft

Norwegian Public Roads Administrations (SVV)

Formerly, Norwegian University of Science and Technology (NTNU)

January 08, 2014

Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

To begin with..
This study was motivated by the core theme of 1st CREBS workshop held in
Oslo in 2006.

In CREBS II (Pisa, 2007) a need for in-depth study, e.g. in form of a PhD
study, was stressed by Adjunct Professor Hans Petter Jostad.

This study was then initiated and conducted at Norwegian University of

Science and Technology (NTNU) (20072011) in collaboration with Norwegian
Geotechnical Institute (NGI) and Chalmers University of Technology.

Researchers who are directly involved in this work are acknowledged as

Hans Petter Jostad (NGI)
Gustav Grimstad (NTNU)
Steinar Nordal (NTNU)
Mats Olsson (Chalmers and NCC)
Peter Hedborg (Chalmers)

The work has also benefited from valuable feedbacks, discussions and review
critics by several other researchers.
Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

General v
vo

Measurements h
Laboratory
(Fast consolidation)

Significantly different consolidation times

v
vo

h H
Prediction In-situ
(Slow consolidation)
Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Two hypotheses on role of creep during primary consolidation

log 'v
Proposed by Ladd et al. (1977). Does creep act
as a separate phenomenon while excess pore
Laboratory
pressures dissipate during primary consolidation?

Vertical strain
log time
In-situ (field)
Thick Sample
(In-situ)
Hypothesis A
log 'v
Vertical strain

Thin Sample Hypothesis A

(Laboratory)
Laboratory
Hypothesis B

Vertical strain
EOP
In-situ (field)

Fig. Hypothesis A and B ( after Ladd et al., 1977)

Hypothesis B

Advocates of the two different creep hypotheses have independently presented

voluminous laboratory and field data to substantiate their opinions.
Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Experimental substantiation of the two hypotheses, e.g.

Fig.: EOP laboratory tests supporting hypothesis A Fig.: In-situ and EOP laboratory tests that support
(after Choi, 1982; Feng, 1991) hypothesis B (after Kabbaj et al., 1988)
Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Numerical substantiation of the two hypotheses, e.g.

Analysis of field cases using constitutive models based on the two hypotheses

Fig. : Measurement Vs. predictions at Sk-Edeby

test fill using hypothesis A model (after Fig. : Measurements Vs. predictions at Changi Airport
Mesri and Lo, 1989 ) using hypothesis B model (Cao et al., 2001)
Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

More on the two creep hypotheses

With an inclination to hypothesis A, Ladd et al. in 1977 concluded that

little definitive data exists to show which of the two hypotheses is more
nearly correct for the majority of cohesive soils.

Ever since, the topic became a topic of active debate and discussion and
remained to be an issue that needed to be resolved.

This discussion was re-started by NGI in 2006 at 1st CREBS workshop,

where advocates from both sides as well as others have attended.

In 2007, this study was initiated and carried out at NTNU, NGI and
Chalmers with additional funding from ICG (International Center for
Geohazards).
Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Main motivation and objectives CREBS I

How to extrapolate creep from short time observation
to long term predictions ?

The two conflicting hypotheses are well substantiated

with laboratory and field data. Why ?

Constitutive models based on the two hypotheses are

seen to produce acceptable field predictions. Study
and evaluate the models based on field cases.

To increase understanding on time- and stress-

compressibility of clays during primary consolidation.

To produce the most convincing creep hypothesis

and a numerical tool that can consistently explain
laboratory and field observations.
Fig. : Tentative list of problems as
presented in the 1st CREBS
workshop (Jostad, 2006)
Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Outline of the presentation

Laboratory studies
Part I: Specimens of varying thicknesses
Part II: Soil element compressibility (varying consolidation duration)

Field studies

Present the hypotheses for a specific case

A look at the relevant laboratory tests
Numerical studies
Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Laboratory studies I: Creep hypotheses for varying soil layer

thicknesses

EOP strain-effective stress relationships: the creep hypotheses

Hypothesis A Hypothesis B

Fig.: Principle sketch of the two creep hypotheses for varying soil layer thicknesses
Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

EOP strain-effective stress relationships:

laboratory tests

End effects (testing problem) ?

Strain rate effects ?

Evaluate the 508 mm thick specimen

(the action and the reaction)

Fig. : EOP V/Vo relationships for various

thicknesses (after Feng, 1991)
Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Hyphothesis A
Inconsistent EOP
criterion

Re-
interpretation

Hyphothesis B
Consistent EOP
criterion
Fig.: Original and re-interpreted volumetric straineffective stress relationships
Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Numerical study of raw experimental data with hypothesis B model

Similar load sequence and duration adopted from the actual test.
Identical set of soil parameters for the thin and thick specimen
Three load increments with respect to p c

c
open
508

3 c
open
127

4 c
0 mm
0 mm

Fig.: Axisymmetric FE-model of the triaxial specimens

Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Numerical study of raw experimental data with hypothesis B model

Fig.: Numerical simulation (smooth lines) vs. measurements (lines with symbols)
Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Strain-time relationships: the creep hypotheses

Hypothesis A

Fig. : Principle sketches of actionresponse relationships according to hypothesis A

Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Hypothesis B

Fig. : = 50 150 kPa, p c = 100 kPa

(allowed to creep at 150 kPa for
100 days)
Fig.: Effective stressStrain and Strain-Time relationships
according to hypothesis B
Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Hypothesis B
= 200 400 kPa

Fig.: Effective stressStrain and Strain-Time relationships

according to hypothesis B
Some typical experimental observations
Single load increment tests after exceeding initial p c

Aboshi (1973)

Konovalov & Bezvolev (2005 )

Imai & Tang (1992)

(Degago et al. (2011), Gotechnique 61(10))

Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Laboratory studies I: Creep hypotheses for varying soil layer

thicknesses

Final remarks

Laboratory tests on specimens of varying thicknesses imply hypothesis B.

EOP strain-effective stress relationship is not unique.
EOP strain increases with increasing consolidation duration

Numerical simulation results using hypothesis B model can explain experimental

measurements.
Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Laboratory studies II: Creep hypotheses for soil element

compressibility

The two hypotheses are best differentiated by consolidation duration of

soil layers than soil layer thickness

T B 4h
h 4h

Fig.: Interconnected tests

Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Creep hypotheses for soil element compressibility

EOP strain-effective stress relationships: the creep hypotheses

nh

Hypothesis A Hypothesis B

Fig.: Principle sketch of the two creep hypotheses for compressibility of soil elements
within a specimen
Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

EOP strain-effective stress relationships: laboratory test results

Fig.: EOP vertical straineffective stress of sub-specimens (interpreted from Feng, 1991)

Hypothesis B
Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Strain-time relationships: the creep hypotheses

Hypothesis A

At EOP, the strain-time relationships of all sub-specimens converge to

the same point

Fig.: Principle sketches of Strain-Time and Effective stressStrain

relationships according to hypothesis A
Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Hypothesis B

Fig.: Strain-Time and Effective stressStrain relationships

according to hypothesis B
Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Hypothesis B

Fig.: Strain-Time and Effective stressStrain relationships

according to hypothesis B
Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Hypothesis B

Figure : Strain-Time and Effective stressStrain

relationships according to hypothesis B
Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Strain-time relationships:
laboratory test results

Hypothesis B

Fig: Experimental results on

Batiscan and St. Hilaire clay
(Feng, 1991)
Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Strain-time relationships:
numerical study
Simulation using hypothesis B
(SSC) model

FE-code PLAXIS
c
open
508

4
0 mm

Fig.: Geometry adopted in FE simulation

Fig.: Experimental measurements (Feng, 1991) Vs Simulation results of Batiscan clay

Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Tests conducted during this study

(@Chalmers University of Technology)

Hypothesis A :- v

GWT
The sub-layer at the drainage face does not
experience any secondary consolidation until
EOP state of the bottom sub-layer
(Mesri & Vardhanabhuti, 2006).

Clay sub-layers H

Motivation

Will a soil element at the drainage face really wait

for the EOP state of the bottom sub-layer to start
its secondary consolidation?
(Jostad, 2006 @CREBS I)
Fig.: A soil layer consisting of
several soil sub-layers
Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

An idealized case

A clay layer placed on top of similar clay as compared to a clay layer

placed on top of a soil material with different coefficient of consolidation.

ho(t)

EOP
CC CB
(ClayClay) (ClayBentonite mix)

Fig.: Idealized cases

Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Expected strain-time relationship of the top clay: the creep hypotheses

/ho(t)

Hypothesis A Hypothesis B

Fig.: Predicted incremental nominal strain-time relationship of the top clay

Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Test set up and measurements

Conducted at Chalmers University of Technology

Incremental load sequence of 10, 20, 30 and 80 kPa ( EOP = 95 % EPP dissp.)
Two sets of tests

Deformation Pore pressure

(Bottom)
q q

2 cm Clay
Clay
Top open Bentonite
Clay &
Bottom closed
Clay mix

Pore pressure
(Middle) drainage tube Pore pressure
(Middle)

Measurements
Second
First setset
of experiment
of experiment

Fig.: Test set up and measurements

Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Fig.: Running the interconnected tests at Chalmers GeoEngineering laboratory

Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Experimental results

/ho(t)

Fig.: Test measurements

EOP is slightly more than expected for hypothesis B

EOP strain not unique !
Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Numerical study
Simulation using hypothesis B
(SSC) model

FE-code PLAXIS
4
4

/ho(t)

u (t)
0 cm
0 cm

C-C C-B
Fig.: Geometry adopted in FE simulation

Fig.: Measurements vs. simulation (30-80 kPa)

Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Laboratory studies II: Creep hypotheses for soil element

compressibility

Final remarks

Laboratory studies on soil element compressibility imply hypothesis B.

Local compressibility of a soil element is governed by its prevailing effective
stress-strain-strain rate on that particular soil element rather than what is
happening elsewhere in the soil layer.

This means that a soil element creeps during primary consolidation and
starts its secondary consolidation phase right after its primary consolidation
phase rather than wait until the completion of the primary consolidation of
all the other soil elements

Numerical simulation results using hypothesis B model can explain experimental

measurements.
Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Field studies
The two hypotheses could give significant practical differences when
predicting settlements of in-situ soil layers

However, on several occasions, the advocates of the two hypotheses have

independently presented acceptable predictions of in-situ settlements to
support the hypotheses.

In this study, the constitutive models for the two hypotheses are evaluated
based on the performance of a common and well-documented test fill.

This is mainly motivated by the analogy to the hypothetical case exercises

given to CREBS II participants in 2007 (Pisa) by Hans Petter Jostad.

Constitutive models for hypothesis A (ILLICON), hypothesis B (SSC) and

elasto-plastic model (SS) are considered.
Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Model comparisons Strain formulations

ILLICON strain decomposition log v

ep Cc* log v C * log t eo

v

( vj , e j , t j )
where C * merely decomposes the input and vo

out put e p into two arbitrary parts. vj 1

* Cc* log
ep Cc vj
Cc tj
SS is a rate-independent elasto-plastic model C * log 1

tj
SSC is a rate-dependent elasto-viscoplastic model vf
( vj 1
,ej 1,t j 1 ) ep
ILLICON is equivalent to SS model. EOP e log v
e

The SSC would give larger EOP strain than C * Cc* C Cc

both ILLICON and SS models.
Fig.: ILLICON strain formulations
(after Choi, 1982)
Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Model comparisons Excess pore pressure formulations

Continuity equation as used in ILLICON assumes that the excess pore pressure
dissipation is only affected by the so-called stress-compressibility.

(1 eo ) 2 kv u de de de det
w z 1 e z dt dt dt dt

In SSC and SS model the continuity equation is controlled by total strain rate.

ILLICON would give faster EPP dissipation than SS model.

SSC would give significantly slower EPP dissipation than both ILLICON and SS
model.
Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Comparison of the models based on analysis of Vsby test fill

ILLICON vs. SS
SSC vs. SS

ILLICON, SSC and SS models are indirectly compared based on

analysis of the test fills.

For a given set of soil data, the SS model is used in order to provide
reference predictions with respect to disregarding the effect of creep.
Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Analyses results ILLICON & SS Vsby test fill

ILLICON-Equivalent parameters were adopted for SS model.

Fig.: Settlement history predictions Fig.: Excess pore pressure profile

(ILLICON vs. SS) predictions (ILLICON vs. SS)
Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

ILLICON and SS model predictions vs. Measurements

While disregarding creep, both ILLICON and SS model gave an overall

acceptable predictions.

This should not imply that the soft clays considered do not undergo creep
deformation.

The acceptable predictions were mainly due to two factors, i.e. use of soil
data from disturbed samples and disregarding effect of large deformations.
Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

(1) Sample disturbance

Generally the OCR values used in ILLICON

and SS analysis were low and are believed to
be affected by sample disturbance.

For instance,
Vsby test fill, EOP OCR = 1.31 or 1.82 ?
(Leroueil and Kabbaj (1987))

In Sk-Edeby test fill, OCR = 1.0 ?

(field tests by SGI)
Fig. : Sample disturbance at Vsby test
fill (after Leroueil & Kabbaj, 1987)
Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

(2) Effect of large deformations (buoyancy)

ILLICON and SS model analyses disregarded

load reduction due to buoyancy forces.

Fig. : Applied load with and without consideration of

buoyancy effect (Vsby )

Fig.: Effect of buoyancy on predictions

Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Comparison of SSC vs. SS model

Use of OCR values from high quality sample data or clay age considerations
Effect of large deformation (buoyancy) taken into account

Fig.: Axisymmetric FE geometry adopted for Vsby test fill analysis

Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Analyses results SSC & SS Vsby test fill

Fig.: Settlement history predictions Fig.: Excess pore pressure profile

(SSC vs. SS) predictions (SSC vs. SS)
Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Field studies
Final remarks (based on Vsby, Sk-Edeby & Ellingsrud test fills)

When soil data are interpreted from tests on disturbed samples are used for
settlement analysis then some effect of creep is already incorporated.
A rate-independent elasto-plastic model, along with some simplifying
assumption, could give acceptable settlement and reasonable but
somehow low excess pore pressure responses.

An isotache model would significantly overestimate settlement and could

give unrealistically large excess pore pressure responses.

When soil data are interpreted from tests on high quality samples and used
for settlement analysis,
A rate-independent elasto-plastic model significantly underestimates
settlement and excess pore pressure responses
An isotache model would yield excellent prediction of settlements and
excess pore pressure.
Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Conclusions

In response to the important question raised by Ladd et al. in 1977,

this study has shown that there exist definitive data to demonstrate that
hypothesis B agrees very well with the measured behaviour of cohesive soils.

Several EOP laboratory tests considered in this study demonstrated the

validity of hypothesis B. In fact, this study disclosed that all the empirical data
that were previously used to support substantiate hypothesis A actually imply
hypothesis B.

The experienced p'c as well as EOP strain are rate dependent even for EOP
loading conditions and this fact has been experimentally supported by
several EOP tests and field observations.

The isotache theory (hypothesis B (SSC)) can explain and convincingly

capture important feature of various types of laboratory tests considered in
this study.
Introduction Laboratory studies I Laboratory studies II Field studies Conclusions

Conclusions

Great care needs to be exercised during interpretation and use of

preconsolidation stress (p c) in settlement analyses. With this aspect,
sample quality deserves extra attention.

Awareness regarding the significance of p c (OCR due to creep) on

settlement analysis needs to be stressed by the profession.

The isotache models are well suited to predict settlements of water

saturated soft clay deposits when the input data are deduced from
laboratory tests of good quality soil samples.

Future developments related to the compressibility of natural clays such

as anisotropy and destructuration should be focused on enhancing models
that are based on the isotache framework or similar.
Thank you for your attention !