Stress dependent water retention curve and volumetric behaviour

of intact and re-compacted loess

By
HOSSEN, SK Belal

A Thesis Submitted to

The Hong Kong University of Science and Technology

in Partial Fulfilment of the Requirements for

the Degree of Master of Philosophy

in Civil Engineering

December 2015, Hong Kong

HKUST Library
Reproduction is prohibited without the author’s prior written consent
 ABSTRACT

Aeolian loess is widely spread all over the world and covers approximately 10% of the earth
surface. Because of the aeolian origin, loess has a metastable and high porosity structure, which is
sensitive to hydraulic drying/wetting and loading. This sensitivity may induce significant
volumetric deformation and collapse, which influences water retention characteristics. Natural
loess deposited as deep as 10 m or more are subjected to seasonal variations of water content
(suction). Given that soil is a stress-dependent material, it is, therefore, necessary to understand
stress effects on water retention characteristics of loess with volumetric measurements. However,
very limited studies have been reported on water retention curve (WRC) of loess. Volumetric
changes due to drying/wetting and effects of stress state on water retention characteristics have
been rarely reported in the literature. To better understand intact loess behaviour, re-compacted
loess is used as a reference material. The principal objectives of this study are, therefore, to
measure the WRC of loess over a wide range of suctions and to investigate the effects of net
stress on WRC and wetting/drying induced volumetric behaviour of intact and re-compacted
loess.

Three series of tests were carried out on intact and re-compacted specimens to achieve the above
objectives. In test series-I, WRC tests of intact and re-compacted loess under zero stress were
carried out for a wide range of suctions (zero to 290 MPa). Two different hydraulic paths WD
(wetting followed by drying) and DW (drying followed by wetting) were followed to cover
suction range, 0-0.4 MPa and 0.88-290 MPa, respectively and volumetric measurements were
made as well along the drying and wetting paths. Suction was imposed by axis translation,
osmotic and vapour equilibrium techniques to cover different ranges of suction, 0-0.4 MPa, 0.88-
4 MPa and 7-290 MPa, respectively. On the other hand, stress effects on WRC and
wetting/drying induced volumetric behaviour of intact and re-compacted loess were investigated
in test series II and III, respectively in which tests were conducted at different net stresses, i.e.,
zero, 50 and 110 kPa, respectively, at low suction ranges (0-0.4 MPa). In that case, net stress was
applied by using a modified one-dimensional stress-controllable volumetric pressure plate while
suction was controlled by the axis translation technique.

Experimental results revealed that hysteretic behaviour is different for intact and re-compacted
loess for the studied suction range (i.e., zero-290 MPa). Intact loess shows more pronounced
hysteresis around suction value of 10 kPa compared to re-compacted loess due to larger
proportion of constricted pores (macro-pores larger than 100 m) of intact loess. Stress has a
significant effect on water retention curve (WRC) of intact and re-compacted loess. Only wetting
WRC of intact loess is influenced by an increase in net vertical stress while drying WRC is
insensitive to stress effects. On the contrary, stress effect is more pronounced on drying WRC of
re-compacted loess. At high suction range (i.e., 0.88-290 MPa), re-compacted loess shows 50%
higher drying induced shrinkage (or volumetric strain) than that of intact loess. This difference in
volumetric behaviour of intact and re-compacted loess is because the former soil has a higher
yield stress and a more stabilized structure than those of the latter. Under a constant net stress, re-
compacted loess shows significantly higher wetting induced collapse compared to that of intact
loess.

ii
 ACKNOWLEDGEMENTS

I would like to acknowledge and extend my heartfelt gratitude to my supervisor, Chair

Professor Charles W. W. Ng, for his constructive guidance and suggestions throughout my
study. This research would not have completed without his kind help and encouragement.

I would also like to express my sincere appreciation to academic staff Prof. Jidong Zhao,
Prof. Y. H. Wang for their teaching on advanced theories of soil mechanics and geotechnical
engineering. I am grateful to Prof. Limin Zhang and Prof. Chao Zhou for serving as the
member of the thesis supervision committee.

Many thanks go to the technicians of the geotechnical laboratory, Mr. Kenny C. K. Ma and
Mr. Michael M. H. Chung, for their assistance in laboratory tests and instruments of this
research.

I am deeply grateful to Prof. C.F. Chiu, Prof Chao Zhou and PhD candidate Hamed Sadeghi
for their valuable help and discussion on my research in various ways. Thanks are also
extended to my loess soil research colleagues and other fellow mates for theirs helpful
comments. Last but not least, I would like to express my deepest gratitude to my family and
friends for their continuous support and encouragement.

v
 TABLE OF CONTENTS

Title page……………………………………………………………………………………..i
Abstract……………………………………………………………………………………...ii
Authorization page………………………………………………………………………...iii
Signature page……………………………………………………………………………...iv
Acknowledgements………………………………………………………………………….v
Table of contents…………………………………………………………………………...vi
Abbreviations and Nomenclatures………………………………………………………...x
List of Tables……………………………………………………………………………….xi
List of Figures……………………………………………………………………………...xii
List of Plates………………………………………………………………………………xvi

CHAPTER 1 INTRODUCTION................................................... 1
1.1 BACKGROUND OF THE RESEARCH .................................................................................. 1
1.2 OBJECTIVES OF THE RESEARCH...................................................................................... 2
1.3 LAYOUT OF THE THESIS ................................................................................................. 2

CHAPTER 2 LITERATURE REVIEW ........................................ 4

2.1 COLLAPSIBLE SOIL ........................................................................................................ 4
2.1.1 Definition.............................................................................................................. 4
2.1.2 Classification ........................................................................................................ 4
2.1.3 Factors contribute to collapse ............................................................................... 5
2.2 LOESS .......................................................................................................................... 5
2.2.1 Introduction .......................................................................................................... 5
2.2.2 Geographic distribution ........................................................................................ 6
2.2.3 Geological formation ............................................................................................ 6
2.2.4 Mineralogy ........................................................................................................... 7
2.2.5 Structure ............................................................................................................... 7
2.2.6 Types of bonding................................................................................................... 8
2.2.7 Collapse mechanism ............................................................................................. 8
2.3 WATER RETENTION CURVE AND VOLUMETRIC BEHAVIOUR OF LOESS .............................. 9

vi
 2.4 SUCTION CONTROLLING TECHNIQUES FOR MEASUREMENT OF WATER RETENTION CURVE
(WRC) ............................................................................................................................ 12
2.4.1 Axis-translation technique (ATT) ........................................................................ 13
2.4.2 Osmotic technique (OT) ...................................................................................... 13
2.4.3 Vapour equilibrium technique (VET) .................................................................. 15

CHAPTER 3 TEST MATERIAL, TEST PROGRAM AND

TESTING DEVICES ...................................................................... 29
3.1 TEST MATERIAL .......................................................................................................... 29
3.2 TEST PROGRAM .......................................................................................................... 30
3.2.1 Test series I: Measurement of WRC and volume change along drying and wetting
processes for a wide suction range (zero to 290 MPa) under zero stress ...................... 30
3.2.2 Test series II: Measurement of WRC and wetting/drying induced volume change
for intact loess at different net vertical stresses ............................................................ 31
3.2.3 Test series III: Measurement of WRC and wetting/drying induced volume change
for re-compacted loess at different net vertical stresses ............................................... 32
3.3 TESTING DEVICES ....................................................................................................... 32
3.3.1 Modified one-dimensional stress controllable volumetric pressure plate apparatus
.................................................................................................................................... 32
3.3.2 Osmotic device.................................................................................................... 33
3.3.3 VET device.......................................................................................................... 34
3.3.3.1 Calibration of humidity probe ...................................................................... 34

CHAPTER 4 SPECIMEN PREPARATION AND TEST

PROCEDURES ............................................................................... 52
4.1 SPECIMEN PREPARATION ............................................................................................. 52
4.1.1 Preparation of intact specimen ........................................................................... 52
4.1.2 Preparation of re-compacted specimen ............................................................... 53
4.2 TEST PROCEDURES ...................................................................................................... 54
4.2.1 WRC and volumetric behaviour upon drying and wetting of intact and re-
compacted loess for a wide suction range (zero-290 MPa) under zero stress ............... 54
4.2.1.1 WRC tests using pressure plate apparatus by means of ATT ........................ 54
4.2.1.2 WRC tests using osmotic device by means of OT......................................... 55

vii
 4.2.1.3 WRC tests using VET device by means of VET ........................................... 57
4.2.2 WRC and wetting/drying induced volumetric behaviour of intact and re-compacted
loess under different net vertical stresses at low suction range (0.1-400 kPa) by means
of ATT ......................................................................................................................... 59

CHAPTER 5 WATER RETENTION PROPERTIES OF

INTACT AND RE-COMPACTED LOESS .................................. 70
5.1 WATER RETENTION CURVES OF INTACT AND RE-COMPACTED LOESS FOR A WIDE SUCTION
RANGE (ZERO-290 MPA) UNDER ZERO STRESS ................................................................... 70

5.2 STRESS EFFECTS ON WRC OF LOESS AT LOW SUCTION RANGE (0.1-400 KPA) ................ 72
5.2.1 WRC of intact loess at different net vertical stresses ........................................... 72
5.2.2 WRC of re-compacted loess at different net vertical stresses ............................... 73
5.3 COMPARISON OF WRC OF INTACT AND RE-COMPACTED LOESS UNDER DIFFERENT NET
VERTICAL STRESSES ......................................................................................................... 74
5.4 SUMMARY .................................................................................................................. 74

CHAPTER 6 VOLUMETRIC BEHAVIOUR OF INTACT AND

RE-COMPACTED LOESS ............................................................ 81
6.1 VOLUMETRIC BEHAVIOUR OF INTACT AND RE-COMPACTED LOESS UPON DRYING AND
WETTING FOR A WIDE SUCTION RANGE (ZERO-290 MPA) UNDER ZERO STRESS .................... 81

6.2 STRESS EFFECTS ON WETTING AND DRYING INDUCED VOLUMETRIC BEHAVIOUR OF LOESS
AT LOW SUCTION RANGE (0.1-400 KPA) ............................................................................ 82

6.2.1 Collapse and shrinkage of intact loess along wetting and drying processes under
different net vertical stresses ....................................................................................... 82
6.2.1.1 Test results of intact loess ............................................................................ 82
6.2.1.2 Discussion on volume change of intact loess ................................................ 83
6.2.2 Collapse and shrinkage of re-compacted loess along wetting and drying processes
under different net vertical stresses.............................................................................. 85
6.3 COMPARISON OF WETTING/DRYING INDUCED VOLUMETRIC BEHAVIOUR OF INTACT AND
RE-COMPACTED LOESS UNDER DIFFERENT NET VERTICAL STRESSES.................................... 86

6.4 SUMMARY .................................................................................................................. 87

viii
CHAPTER 7 CONCLUSIONS AND FURTHER WORK ......... 94
7.1 MAJOR CONCLUSIONS ................................................................................................. 94
7.1.1 Water retention properties of intact and re-compacted loess ............................... 94
7.1.1.1 WRC of intact and re-compacted loess for a wide suction range (zero-290
MPa) ....................................................................................................................... 94
7.1.1.2 WRC of intact and re-compacted loess at different net vertical stresses ........ 95
7.1.2 Volumetric behaviour of intact and re-compacted loess ...................................... 95
7.1.2.1 Wetting and drying induced volumetric behaviour of intact and re-compacted
loess for wide suction range (zero-290 MPa) ........................................................... 95
7.1.2.2 Collapse and shrinkage of intact and re-compacted loess along wetting and
drying processes under different net vertical stresses ............................................... 95
7.1.2.3 Comparison of stress dependent wetting and drying induced volumetric
behaviour of intact and re-compacted loess .............................................................. 96
7.2 RECOMMENDATIONS FOR FURTHER RESEARCH............................................................. 96

REFERENCES ............................................................................... 98

ix
 ABBREVIATIONS AND NOMENCLATURES
WRC water retention curve
PTF pedo-transfer function
ATT axis-translation technique
OT osmotic technique
VET vapour equilibrium technique
(σ-ua) net stress
(ua- uw) matric suction
PEG polyethylene glycol
RH relative humidity
GWC gravimetric water content
MWCO molecular weight cut-off value
s suction
VWC volumetric water content
mw adsorption rate
md desorption rate
AEV air entry value
Sr degree of saturation

x
 LIST OF TABLES
Table 3.1 Index properties of loess ...................................................................................... 36
Table 3.2 Summary of test program for test series I ............................................................. 37
Table 3.3 Summary of specimen details for WRC tests using ATT under series I ................ 37
Table 3.4 Detail test program of WRC test using OT under series I ..................................... 38
Table 3.5 Summary of specimen details for WRC tests using OT under series I .................. 39
Table 3.6 Detail test program of WRC test using VET under series I ................................... 40
Table 3.7 Summary of specimen details for WRC tests using VET under series I ................ 41
Table 3.8 Summary of test program and specimen details for test series II. .......................... 42
Table 3.9 Summary of test program and specimen details for test series III. ........................ 42
Table 4.1 Salt solutions with RH and imposed suction for WRC tests at high suction range 61
Table 5.1 Parameters of water retention curves for intact loess ............................................ 77
Table 5.2 Parameters of water retention curves for re-compacted loess ................................ 77

xi
 LIST OF FIGURES

Figure 2.1 A classification of collapsible soil (Rogers, 1995) .............................................. 17

Figure 2.2 Global distribution of loess deposits (Pecsi, 1990; adopted from Delage et al.,
2005) .......................................................................................................................... 18
Figure 2.3 A schematic map of the distribution of the Chinese loess (Jin and Liu, 2011) ..... 19
Figure 2.4 Deposition of aeolian loess in the proximity of large ice-sheets (Andersen and
Borns, 1997, adopted from Delage et al., 2005) .......................................................... 20
Figure 2.5 (a) A basic structural unit of loess (Tan, 1988), and (b) A schematic of loess
structure from mineralogical standpoint (Osipov and Sokolov, 1994) ......................... 20
Figure 2.6 Types of bonding in collapsible soils (Clemence and Finbarr, 1981) ................... 21
Figure 2.7 Water retention curve of the intact loess from the different approaches used
(Muñoz-Castelblanco et al., 2012) .............................................................................. 22
Figure 2.8 Water retention curve of the undisturbed collapsible loess soil, determined using
filter paper and pressure plate techniques (Haeri et al., 2014)...................................... 23
Figure 2.9 (a) Water characteristic curves at various net mean stresses (i.e., w – log s), and (b)
degree of saturation versus matric suction curves at various net mean stresses (Chen et
al., 1999). ................................................................................................................... 24
Figure 2.10 Results from wetting-induced collapse tests conducted at different mean net
stress values. (a) volumetric strain as a function of the applied matric suction, and (b)
degree of saturation as a function of the applied matric suction (Haeri et al., 2014)..... 25
Figure 2.11 Results of the UCL tests. (a) volumetric strains with decreasing matric suction at
constant axial load, and (b) water retention curves of loess, deduced from the wetting
paths of the UCL tests (Zhou et al., 2014) ................................................................... 26
Figure 2.12 Collapse and compression tests performed on natural unsaturated loess from
Northern France (Munõz-Castelblanco et al., 2011). ................................................... 27
Figure 2.13 Principles of different suction control techniques. (a) axis-translation technique,
(b) osmotic technique, and (c) vapour equilibrium technique (Ng and Menzies, 2007) 28
Figure 3.1 Particle size distribution of loess ........................................................................ 43
Figure 3.2 Compaction curve of loess.................................................................................. 43
Figure 3.3 Hydraulic path of WRC tests of loess for test series I.......................................... 44
Figure 3.4 Stress path for test series II and III ...................................................................... 45

xii
Figure 3.5 Schematic diagram (a) Test setup for determination of initial suction of the
specimen, and (b) Modified one-dimensional stress-controllable volumetric pressure
plate devices ............................................................................................................... 46
Figure 3.6 RH and Temperature observed by humidity probe in desiccator contained
saturated salt solution of known RH. (a) RH vs. Time, and (b) Temperature vs. Time. 47
Figure 3.7 Calibration result of humidity probe against known values of RH of saturated salt
solution ...................................................................................................................... 48
Figure 4.1 Measurement of Brix value (%) with elapsed time during drying and wetting
processes .................................................................................................................... 62
Figure 4.2 Weight equalization of intact specimens under different suctions inside desiccator
(a) drying, and (b) wetting .......................................................................................... 63
Figure 4.3 Weight equalization of re-compacted specimens under different suctions inside
desiccator (a) drying, and (b) wetting.......................................................................... 64
Figure 4.4 Measured RH and temperature of different desiccators containing salt solution. (a)
RH and (b) Temperature variations with time inside the desiccators ........................... 65
Figure 5.1 Comparison of water retention curve of intact and re-compacted loess for wide
suction range (zero-290 MPa) ..................................................................................... 78
Figure 5.2 WRC of intact and re-compacted loess specimens in (w-s) plane ........................ 78
Figure 5.3 Wetting and drying WRCs of intact loess under different net vertical stresses .... 79
Figure 5.4 Wetting and drying WRCs of re-compacted loess under different net vertical
stresses ....................................................................................................................... 79
Figure 5.5 Comparison of stress dependent WRC of intact and re-compacted loess ............. 80
Figure 6.1 Comparison of changes in void ratio with suction along drying and wetting paths
between intact and re-compacted loess under zero stress (a) at low suction range (0.1-
400 kPa), and (b) at high suction range (880-290 000 kPa). ........................................ 89
Figure 6.2 Oedometer test results of saturated intact and re-compacted specimens .............. 90
Figure 6.3 Wetting and drying induced volumetric behaviour of intact loess under different
net vertical stresses ..................................................................................................... 90
Figure 6.4 (a) Stress paths followed in WRC tests of intact loess and the evolution of LC
yield surface during the test I-110, and (b) postulation of SI yield curves .................... 91
Figure 6.5 Wetting and drying induced volumetric behaviour of re-compacted loess under
different net vertical stresses ....................................................................................... 92
Figure 6.6 Comparison of stress dependent wetting and drying induced volumetric behaviour
of intact and re-compacted loess ................................................................................. 93
xiii
 LIST OF PLATES

Plate 3.1 Osmotic device for determination of WRC at medium suction range (0.88-4.0 MPa)
under zero stress ......................................................................................................... 49
Plate 3.2 Desiccator assembly to impose suction on specimen using saturated salt solution to
measure WRC at high suction range (7-290 MPa) under zero stress............................ 50
Plate 3.3 Calibration technique of humidity probe used to measure RH and temperature
during WRC tests (a) humidity probe, and (b) calibration of humidity probe using
saturated salt solution ................................................................................................. 51
Plate 4.1 Preparation of intact specimen (49 mm diameter and 10 mm height)..................... 66
Plate 4.2 The WF compression apparatus for preparing re-compacted specimen (70 mm
diameter and 19 mm height) using static compaction method ..................................... 67
Plate 4.3 Re-compacted specimen (70 mm diameter and 19 mm height) .............................. 68
Plate 4.4 Preparation of re-compacted specimen (49 mm diameter and 10 mm height) ........ 68
Plate 4.5 WRC test using pressure plate apparatus. Test setup of (a) determination of initial
suction, and (b) wetting and drying WRC test under zero stress .................................. 69

xiv
CHAPTER 1
INTRODUCTION

1.1 Background of the research

Loess is an accumulation of wind-blown silt and mostly known as collapsible soil. It covers
approximately 6.6% (631,000 km2) of the total area of China and 10% of the land area of the
earth surface (Rogers et al., 1994; Peng et al., 2015). Having a metastable state provided by
open structure and partial saturation, this special type of soil is highly sensitive to wetting and
drying processes as well as mechanical loading (Muñoz-Castelblanco et al., 2011; Jiang et al.,
2014). These events could then influence volumetric behavior (Haeri et al., 2014; Garakani et
al., 2015) and water retention curves (Muñoz-Castelblanco et al., 2012). The former which
emerges as collapse or shrinkage would create serviceability problems to geo-structures. As a
result, the most important geotechnical consideration about this type of soil is collapsibility
(Delage et al., 2005; Yuan and Wang, 2009). The latter, on the other hand, plays a key role in
stability of natural unsaturated loess deposits, for which failure could be a catastrophic geo-
environmental hazard. Therefore, it is essential to understand the water retention and
volumetric behaviour of loess soil. Although numerous studies on water retention
characteristics and volume change behaviour of different types of soils have been conducted
but up to now, experimental results for loess are limited.

Natural loess deposits are subjected to seasonal variations of water content from nearly
saturated to very dry states depending on the season and environmental conditions. This
implies that wide range of suction can be induced especially in the humid region during dry
and wet seasons. Hence, it is necessary to measure the water retention curves (WRCs) for a
wide range of suction considering volumetric measurements due to drying/wetting. Most of
the studies, however, determined WRC of intact loess without considering volumetric
changes along the hydraulic paths and suction was not measured or controlled directly at
higher range (beyond the practical working range of axis translation technique i.e. ≥ 1500
kPa) (Huang et al., 2010; Muñoz-Castelblanco et al., 2012; Haeri et al., 2014). Moreover, in
order to have a better understanding of behaviour of loess, it is worth considering re-
compacted loess as a reference material to the intact loess that behaves differently because of
the existence of microstructure. Several studies have been carried out on re-compacted loess
(Chen et al., 1999; Roohnavaz et al., 2011; Zhan et al., 2014), however, based on the

1
literature, water retention curves of intact and re-compacted loess have been investigated
independently. In other words, it can be mentioned that no study has been conducted so far to
compare water retention curves of intact and re-compacted loess for wider suction range (e.g.
zero-300 MPa).

Apart from hydraulic drying/wetting effects on volumetric behaviour, unsaturated loess also
tends to exhibit significant volumetric compression when subjected to loading. On the other
hand, during WRC tests, it is commonly assumed that soil is non-deformable and stress has
no significant influence on its water retention characteristics (ASTM, 2002). However,
previous studies (Ng and Pang, 2000; Chiu and Ng, 2012) attributed that WRC is stress
dependent due to modification of pore structure. Therefore, it is needed to consider stress
effect on water retention and wetting/drying induced volumetric behaviour of loess. Limited
studies (Haeri et al., 2014; Zhou et al., 2014) were carried out on intact loess to understand
the stress effects on volumetric behaviour based on wetting tests under constant load whereas,
Chen et al. (1999) investigated the effects of net mean stress on WRC of compacted loess. To
the best of the knowledge, stress dependent water retention curve and wetting/drying induced
volumetric behaviour under constant net vertical stress of both intact and re-compacted loess
and their comparison have been rarely reported to the literature.

1.2 Objectives of the research

The principal objectives of this research include:

(1) To measure the WRC of intact and re-compacted loess for a wide suction range (zero to
290 MPa) under zero stress

(2) To investigate the volumetric behavior of intact and re-compacted loess upon wetting and
drying for a wide suction range (zero to 290 MPa) under zero stress

(3) To investigate the effects of net vertical stress on WRC and wetting/drying induced
volumetric behavior of intact and re-compacted loess at low suction range (zero to 400 kPa)

1.3 Layout of the thesis

This thesis consists of 7 chapters including this introduction.

2
In Chapter 2, at first collapsible loess soil is introduced with detail information. Following
that, WRC and volumetric behaviour of intact and re-compacted loess are reviewed. Finally,
different suction controlling techniques for WRC tests are discussed.

In Chapter 3, test material is introduced first. Afterwards, test program of different series of
tests is presented and finally, different testing devices for this laboratory study are described
in details.

In Chapter 4, the preparation procedure of intact and re-compacted specimens for different
tests is presented first. Afterwards, the procedure of WRC tests of intact and re-compacted
loess for wide suction range (zero to 290 MPa) using three suction controlling techniques is
described followed by WRC tests of intact and re-compacted loess under different net vertical
stresses at low suction range (zero to 400 kPa)

In Chapter 5, WRC tests of intact and re-compacted loess for full suction range (zero to 290
MPa) is presented first. Afterwards, the effects of net vertical stress on drying and wetting
WRC of intact and re-compacted loess at low suction range (zero to 400 kPa) are discussed
independently and finally, stress dependent WRC of intact loess is compared with re-
compacted loess.

In Chapter 6, results of volume change behaviour of intact and re-compacted loess upon
drying/wetting for full suction range (zero to 290 MPa) are presented and discussed.
Following that, effects of net vertical stress on wetting/drying induced volumetric behaviour
of intact and re-compacted loess are explained and finally, wetting/drying induced volumetric
behaviour of intact loess is discussed comparing with re-compacted loess.

In Chapter 7, major conclusions of this study are summarized and recommendations for
further work are also suggested.

3
CHAPTER 2
LITERATURE REVIEW

2.1 Collapsible soil

2.1.1 Definition

Large areas of the Earth’s surface, particularly in the Midwest and Southwest United States,
Northwest of China, South America, and Southern Africa, are covered by collapsible soils.
This type of soils are generally associated with an open structure formed by sharp grains, low
initial density, low natural water content, low plasticity, relatively high stiffness and strength
in the dry state, and often by particle size in the silt to fine sand range (Mitchell and Soga,
2005). The main geotechnical problem associated with this special type of soil is the
significant loss of shear strength and large volumetric deformation upon wetting with or
without extra loading (including self-weight) due to additional water from rainfall, irrigation,
broken water or sewer lines and rise of ground water, etc.

Rogers (1995) summarized major characteristics of collapsible soils from the literature.

1) ‘‘……a soil that undergoes an appreciable amount of volume changes upon wetting, load
application, or a combination of both.’’ (Sultan, 1969)

2) ‘‘…….any unsaturated soil that goes through radical rearrangement of particles and great
loss of volume upon wetting with or without additional loading.’’ (Dudley, 1970)

3) ‘‘…..additional settlement…. due to the wetting of a partially saturated soil, normally

without any increase in applied pressure.’’(Jemings and Knight, 1975)

2.1.2 Classification

Numerous soil types are considered under general category of collapsible soils. It includes
compacted soils and natural soils such as aeolian deposits, alluvial deposits, colluvial
deposits, residual deposits, and volcanic tuff. Rogers (1995) provided a classification of
collapsible soils shown in Figure 2.1.

4
2.1.3 Factors contribute to collapse

Collapsible soils remain at a stable state in unsaturated conditions but highly sensitivity to
volume change induced by water infiltration alone or water infiltration in combination with
external loading. Several influential factors contribute to collapse include 1) an open,
partially unstable, unsaturated fabric which results in low bulk density, high void ratio and
high porosity, 2) geologically young deposit, 3) a high enough net total stress that will cause
the structure to be metastable, 4) a bonding or cementing agent that stabilizes the soil in the
unsaturated condition, and 5) the addition of water to the soil, which causes the bonding or
cementing agent to be reduced (Rogers, 1995; Pereira et al., 2000).

2.2 Loess

2.2.1 Introduction

Loess is an Aeolian sediment formed by the accumulation of wind-blown silt. It is one of

most well-known collapsible soil and widely available all over the world (Cui et al., 2004;
Delage et al., 2005). It covers approximately 10% of the land area of the earth surface (Heller
and Evans, 1995; Rogers et al., 1994). Moreover, enormous loess deposits are also found in
China which cover 6.6% (631,000 km2) of the total area of China and nearly half (about
317,000 km2) of it forms the Loess Plateau in northwestern China, the greatest bulk
accumulation of loess in the world. (Yuan and Wang, 2009; Xu et al., 2011; Jiang et al.,
2012; Zhan et al., 2014; Wen and Yan, 2014; Peng et al., 2015).

Loess is typically a silt-based soil which has a metastable structure featured by open soil
fabric and weak inter-particle bonding (Barden et al., 1973; Dijkstra et al., 1995; Assallay et
al., 1997; Munõz-Castelblanco et al., 2011; Jiang et al., 2014; Haeri et al., 2014). It generally
occurs in unsaturated state in the arid and semi-arid regions of the world. (Muñoz-
Castelblanco et al., 2012; Wen and Yan, 2014). It is commonly yellow to buff in colour and
particle size of loess falls in a wide range depending on its composition and the environment,
but typically loess has quite uniform particle size in the range of 0.01–0.005 mm. It is
characterised by a significant calcareous content, higher porosity, low plasticity and relatively
low density but appreciable strength and stiffness in the dry or partially saturated state
(Muñoz-Castelblanco et al., 2012).

5
Because of Aeolian origin with metastable structure, loess is highly sensitive to collapse due
to wetting or changes in water content with or without constant vertical stress. This collapse
exhibits a larger settlement or significant volumetric deformation, which is associated with
the collapse of the open structure of the soil. As a result, the most important geotechnical
consideration of this type of soil is collapsibility. (Gao, 1996; Delage et al., 2005; Yuan and
Wang, 2009; Munõz-Castelblanco et al., 2011; Jiang et al., 2012; Zhou et al., 2014; Garakani
et al., 2015)

2.2.2 Geographic distribution

Loess deposits are widespread all over the world. In particular, extensive loess deposits exist
in China, United States, Russia, Argentina, Ukraine, and Eastern Europe. Figure 2.2 shows
the global distribution of loess deposits. In the Figure, number 1 (black areas) represents loess
and number 2 (grey areas) represents loess like sediments. In addition, China has one of the
largest loess deposits on earth in the Northwestern China, called Loess Plateau (Gao, 1996;
Jiang et al., 2012). The geographical distribution of Chinese loess is shown in Figure 2.3, a
schematic map proposed by Jin and Liu (2011). In this schematic map, the area covered with
yellow colour represents the loess distribution in different parts of China.

2.2.3 Geological formation

loess is a wind-deposited material and is mainly derived from the major valley train of
glaciers. Delage et al. (2005) describes the mechanisms of formation of aeolian loess
deposits:

-Thin particles produced by glacial abrasion are washed out, transported by proglacial flows
and redeposited close to the moraine (outwash plains)

- Sand, silt and clay particles that submitted to freeze-thaw cycles are eroded and transported
by violent and permanent cold and dry winds generated by high pressures above ice-sheets

- Larger sand particles are first deposited as duns and as sand cover

-Silt and clay particles are transported towards low pressure zones and high atmosphere
before being deposited due to changes in wind regime: fall in wind speed, obstacle, and
capture by herbaceous vegetation or snow cover.

6
2.2.4 Mineralogy

Loess typically contains 40–60 minerals (Egri, 1972). The most important minerals in sand
and silt fractions are quartz, feldspar, carbonate, mica, and gypsum while hydromica,
momtmorillonite, mixed layered kaolinite, and finely-dispersed quartz and calcite dolomite
are the major minerals in clay fraction. These typical minerals can be divided into two
groups, based on the role they play in the process of collapse (Egri, 1972):

1) Active minerals, which are changed due to the process of collapse, and which include
carbonates (mainly calcite), sulphates (gypsum), and salt.

2) Passive minerals, which are not changed due to the process of collapse, and which include
quartz, feldspar, and mica.

Osipov and Sokolov (1994) also mentioned that most loess has carbonate content from 1% to
15–25%, medium soluble mineral (gypsum) content 4–10%, readily soluble salts (chloride)
content <2%, half soluble oxides and hydroxides <2.4% and humus content 1–2%. These
salts, carbonates, and hydroxides play an important role in the bonding system of loess and
result in different collapsibility.

2.2.5 Structure

Loess has an open metastable structure and mainly consists of fine-grained particles.
However, loess structure, and therefore the type of loess (Sandy loess, Silty loess and Clayey
loess) also depends on the percentage of sand, silt, and clay present. Klukanova and
Frankovska (1994) demonstrated that typical loess, also called silty loess is characterized as a
skeleton structure, which presents micro-granularity.

The basic unit of loess structure is shown in Figure 2.5(a). Osipov and Sokolov (1994)
explained the basic unit of loess structure from mineralogical viewpoint using the schematic
as shown in Figure 2.5(b). The basic unit of the loess structure is formed by a core of quartz,
carbonates, and more rarely feldspar. The core is coated by a clay film composed mainly of
illite, momtmorillonite, mixed structures, sporadic iron, manganese hydroxides, and finely
dispersed quartz and calcite particles. In this schematic Region 1 (inner part) is the quartz or
feldspar nucleus; Region 2 and Region 3 are amorphous SiO gel and a calcite envelopes,
respectively; Region 4 is the clay shirt, saturated by FeO. The grain surface is amorphous.

7
Trace calcite etches the surface and the grain is covered by a spotted envelope of calcite. This
calcite coat is overlapped by a poly-mineral ‘‘shirt’’ consisting of clay minerals and
cemented by fine-dispersed calcite quartz, oxides of iron and amorphous silica. Typically, the
calcite envelopes and clay shirt are water resistant.

2.2.6 Types of bonding

The metastable structure of loess is characterized by weak in particle bonding by capillary

forces or chemical cementing agents. Clemence and Finbarr (1981) stated different types of
bonding in loess as shown in Figure 2.6. It is seen that in unsaturated condition, the big grains
can be held together by capillary tension, and in the fine grains, capillary forces provide silt-
silt and silt-sand bonds (A and B). Since coarse (sand) grains are usually coated by clays, the
bonding can be provided by aggregated clay grains (C), by flocculated clay (clay buttress)
(D), or by the combination of silt and clay (E). Some loess does not have high clay content,
so the coarse grain and silt size particles are not fully surrounded by clay; however, particles
are still bonded by clay bridges (F). However, it is also reported that some chemical
cementing agents such as iron oxide, calcium carbonate, etc. played an important role for
bonding in particular collapsible loess soils. Calcium-carbonate cementing can be important
for loess bonding although clay provides the dominant bonding effect in most of the
collapsible loess soil. The physical bonding strength provided these agents can be reduced
significantly due to wetting or water infiltration.

2.2.7 Collapse mechanism

Collapsibility is the major concern of loess soil. This special feature introduced loess as a
problematic soil in geotechnical engineering. The failure of metastable structure of loss upon
wetting with or without load is the primary reason for collapsibility. This collapse
phenomenon induced significant volumetric deformation and also reduced effective stress
which in turn results in partial loss of shear strength. Barden et al. (1973) listed three
conditions to observe collapse in a soil: 1) An open potentially unstable partly saturated
structure, 2) A high enough value of applied stress component to develop a metastable
condition, and 3) A high enough value of suction (or other bonding or cementing agent) to
stabilise intergranular contacts, and whose reduction on wetting will lead to collapse.

8
Based on the loess microstructure and the types of structural bonding exist in loess,
Klukanova and Frankovska (1994) suggested that the collapsing mechanism can be divided
into three phases:

Phase A: 1) Initial stage of destruction of the original microstructure due to increasing

moisture and external load, 2) Clay films, clay bridges, and buttress start to break, 3)
Aggregates, microstructure disintegrate, and 4) Intensity of dissolution of carbonate and their
migration in the soil increase

Phase B: 1) disintegration of microstructure continues-clay particles are water transported in

the soil, 2) Content of carbonate decreases, 3) other fabric elements compress, and 4) Total
volume of soil decreases

Phase C: 1) New microstructure develops-after collapse, 2) Basic structural units disintegrate,

they are no more coated by clay films and have no mutual contacts with clay bridges, 3) Clay
films of silty grains were destroyed and removed completely, 4) Clay particles were
aggregated and a coherent matrix formed in some place, 5) the soil acquires a heterogeneous
structure in contrast with the original homogeneous structure, and 6) void ratio is reduced

2.3 Water retention curve and volumetric behaviour of loess

Muñoz-Castelblanco et al. (2012) investigated the water retention properties of natural loess
from Northern France. The WRC of intact loess along the wetting and drying paths was
determined by using filter paper method and the high-capacity tensiometer (HCT). The high-
capacity tensiometer (HCT) and filter paper technique measured the water retention
properties of intact loess between 0 to 1 MPa and 10 kPa to 10 MPa, respectively. This study
obtained WRC with no hysteresis around the natural water content (14.4%) and two
hysteresis loops on both the wet and dry sides of the curve as shown in Figure 2.7. The
insensitivity of WRC to hydraulic path in that region corresponds to seasonal variation of
water content. However, WRC was determined assuming no volume change due to
drying/wetting for both suction measurement methods. The considered assumption may not
be valid for metastable structured collapsible loess since it induces large volumetric
deformation upon wetting and drying. As a result, volumetric water content or degree of
saturation must be treated with caution. In addition, water retention curve of loess at high
suction range up to 10 MPa was measured using filter paper technique, which is an indirect
method of suction measurement. In the literature, no WRC data can be found especially

9
beyond the practical working range of axis translation technique (0-1500 kPa) by direct
measurement or control of suction. On the other hand, the considered maximum suction of 10
MPa for water retention curve might not be enough for providing complete scenario of water
retention characteristics of natural loess of arid and semi-arid regions where soil is subjected
to high suctions in extremely dry season.

Haeri et al. (2014) measured drying and wetting WRC of undisturbed loess (Figure 2.8)
specimen for wide suction range using the filter paper and pressure plate techniques. The
undisturbed sample was collected from Hezar-pich site, Gorgan, Iran. During experiment, no
volumetric measurements were considered along the drying and wetting processes. Since
natural loess is susceptible to volumetric deformation during drying/wetting processes,
therefore, volume changes need to be taken into account during WRC tests. Moreover,
suction was not controlled at high suction range, rather it was deduced indirectly based on
filter paper measurements.

Huang et al. (2010) measured the drying WRC over limited suction range (10-1500 kPa) of
undisturbed loess sample from the loess plateau of China using the pressure plate apparatus.
In this test, water content was measured at different matric suctions once equilibrium
achieved. The aim of the study mainly given to predict the WRC of loess from indirect pedo-
transfer functions, (PTFs) and comparing with measured WRC for evaluation of PTFs rather
than investigating the water retention properties of loess. This study ignored the volumetric
measurement of specimen along the drying path and wetting WRC along with wetting
induced volume change were not considered as well.

In determination of WRC, it is commonly assumed that soil is non-deformable and stress has
no influence on water retention characteristics (ASTM, 2002). The first assumption becomes
violated once expansive or collapsible soils are considered. The second assumption is also
invalid which evidenced by different researchers with a conclusion that net stress is one of
the two stress state variables governing the behavior of unsaturated soils and affecting the
WRC accordingly (Ng and Pang, 2000; Chiu and Ng, 2012; Ng and Leung, 2012; Tavakoli
Dastjerdi et al., 2014; Zhou and Ng, 2014).

Chen et al. (1999) performed three triaxial shrinkage tests on unsaturated compacted loess,
collected from Shanxi province, China using the modified triaxial apparatus. This study
investigated the effect of net mean stress on water retention behavior of compacted loess. The

10
suction versus water content and suction versus degree of saturation relationship from the
triaxial shrinkage tests are presented on a semi-logarithmic plot in Figure 2.9(a) and (b).
Experimental results as shown in Figure 2.9 (a) and (b) revealed that no unique relationship
exist between the matric suction and the water content or degree of saturation that implies net
mean stress influences the WRC of compacted loess. However, this study only focused on the
stress effects on drying WRC, but no tests were conducted on wetting WRC and wetting
induced volumetric behaviour considering stress effects.

In addition to WRC tests on loess mentioned earlier, Haeri et al. (2014) also conducted five
wetting-induced compression tests on undisturbed cylindrical triaxial specimens at different
constant mean net stresses to investigate the effect of mean net stress on wetting-induced
volume changes of from its initial suction. The wetting induced compression test results are
shown in Figure 2.10. Experimental results revealed that volumetric deformation or collapse
increased with higher values of mean net stress whereas, WRC showed less sensitivity to the
mean net stress during wetting. In contrast, this study ignored the stress effects on WRC and
volumetric behaviour along the drying path that can provide the complete picture of stress
dependent WRC and volumetric behaviour of natural loess.

Zhou et al. (2014) conducted two wetting tests (UCL 1, UCL 2) on unsaturated natural loess
specimen (76 mm in dia and 152 mm in height) under constant axial load (UCL) at net mean
stress of 30 and 50 kPa, respectively using triaxial apparatus. This study showed that
unsaturated natural loess deforms progressively with reduction of matric suction (wetting)
which is more significant at low suctions (below 50 kPa) and the deformation in terms of
volumetric strain increased with stress level (high stress level induced more structural
collapse in the wetting process) (Figure 2.11 (a)). Furthermore, based on the wetting tests
under constant load, wetting branch of WRC was deduced for small suction range and found
that water retention capacity decreased slightly with increasing stress level as shown in
Figure 2.11 (b). No volumetric measurements along the drying path were conducted
considering stress effects in order to investigate the stress effects on drying induced
volumetric behaviour.

Jiang et al. (2012) studied wetting induced collapsibility of artificially structural loess sample
(ASLS) using single-oedometer, double-oedometer, and triaxial wetting tests. This study
concluded that wetting induced deformation is independent of the sequence of wetting and

11
loading based on results of single and double oedometer tests. Both loading and wetting can
reduce larger pores.

Munõz-Castelblanco et al. (2011) investigated compression and collapse behaviour of natural

unsaturated aeolian loess from Northern France by performing constant water content
oedometer compression tests with suction measurement, including 4 standard 24 h step
loading (SL) and 4 constant rate of strain (CRS) experiments. The ‘saturated’ reference
compression curve is provided by compression of one soaked sample. The Barcelona basic
model was used as a theoretical framework in order to interpret the results qualitatively. It
was observed that responses in suction are reasonably constant in all tests (Figure 2.12 ). The
small variation of suction is attributed to the compression of larger and weaker inter-grain dry
pores with little effect on the smaller water-saturated pores governing suction changes. Less
observed yield values compared to the compacted samples is due to the aeolian origin of the
loess. In the other words, yield is not related to any previous applied stress but to the bonding
due to carbonate re-precipitation between the grains. The maximum collapse occurs at
vertical net stress of 30 kPa. Oedometer testing was selected for the simplicity purpose
whereas triaxial experiments are more preferable since more reliable volume changes can be
captured by triaxial tests. However, re-compacted specimens could be considered as an
alternative choice for laboratory experiments. The behavior of re-compacted specimens can
be treated as a reference line for interpreting intact soil behavior.

2.4 Suction controlling techniques for measurement of water retention curve (WRC)

To measure the WRC of unsaturated soil, different suction controlling/measuring techniques

are used. Out of these, the most widely used suction controlling techniques are: (i) Axis
translation technique (Hilf, 1956), for suction up to 1.5 MPa; (ii) Osmotic technique (Zur,
1966), for suction up to 10 MPa (Delage et al., 1998); (iii) Vapour equilibrium technique
(Esteban and Saez, 1988), for suction ranging from 3 to 1000 MPa (Romero, 1999; Delage
and Cui, 2000; Villar, 2000). The first two techniques (axis translation and osmotic
technique) control the matric suction whereas vapour equilibrium technique can be used to
control total suction in an experiment. These three techniques are briefly introduced in the
following sections.

12
2.4.1 Axis-translation technique (ATT)

The axis translation technique is the most commonly used suction controlling technique for
unsaturated soil testing. This technique is introduced by Hilf (1956) to measure matric
suctions up to 1.5 MPa (Fredlund et al., 2012). Matric suction (i.e., ua- uw) of an unsaturated
soil is defined as the difference between the pore air pressure (ua) and pore water pressure
(uw). Figure 2.13 (a) shows the principle of ATT. This technique elevates the pore air
pressure ua to increase pore water pressure to be positive, preventing cavitation in the water
drainage system. As shown in Figure 2.13 (a), in the field, the total stress, pore air pressure
and pore water pressure are σ1, , . Generally, is assumed to be the atmospheric
pressure ( ) while is negative gauge pressure with respect to atmospheric condition.
When ATT is applied, it translates the origin of standard atmospheric condition by the same
amount of an applied air pressure, Therefore, σ1, , are increased to σ2, , whereas
net stress, (σ-ua) and matric suction (ua- uw) remains constant.

During application of ATT, ua is gerenally controlled through a low air entry value porous
stone while uw is controlled or measured using a saturated high air-entry value ceramic disc,
which separates the air and water phase of soil specimen during experiment. The maximum
suction that can be controlled or measured is limited by the air-entry value of the ceramic
disc. Although ATT is widely used for unsaturated soil tests but, there are some limitations to
using the axis translation technique. Firstly, the application of an air pressure higher than the
atmospheric pressure does not correspond to the field conditions (Delage et al., 2008).
Secondly, the applicability of the axis translation technique appears to be questionable at a
high degree of saturation when air is discontinuous and water is continuous in the soil pore
network (Bocking and Fredlund, 1980). At this point, Vanapalli et al. (2008) also
demonstrated that ATT is only valid when the air-water inter phase is continuous, which is
typically the case for soils at a degree of saturation lower than 90%. Furthermore, the theory
associated with the axis translation technique is applicable for soils with totally
interconnected pore-air voids and for soil particles that are incompressible (Vanapalli et al.,
2008).

2.4.2 Osmotic technique (OT)

Osmotic method is an alternative to control matric suction in unsaturated soil testing

especially beyond 1500 kPa. This method was introduced by soil scientists in the 1960s and

13
adapted in geotechnical engineering by Kassiff and Ben Shalom (1971). The principle of
osmotic technique is related to osmosis phenomenon as shown in Figure 2.13 (b) (Ng and
Menzies, 2007). This phenomenon is observed when a solvent and solution are separated by a
semi-permeable membrane. The diffusion of large-sized molecules is limited by semi-
permeable membrane, while small-sized molecules are allowed to cross the membrane. In
geotechnical unsaturated soil testing, the specimen is placed in contact with a semi-permeable
membrane behind which an osmotic solution is circulated. The membrane is permeable to
water and ions in the soil but impermeable to large solute molecules and soil particles (Zur,
1966). This results in a difference in osmotic potential or osmotic suction between the
specimen and solution across the membrane. The osmotic potential of the solution causes the
drainage of the soil specimen and this potential is finally balanced by negative pore water
pressure of the soil specimen. Based on energy analysis on soil-water by Zur (1966), it can be
concluded that the matric suction in the soil specimen is equal to the difference of osmotic
suction (osmotic potential) on both sides of the membrane. The maximum suction that can be
controlled is up to 10 MPa, reported by Delage et al. (1998).

Polyethylene glycol (PEG) solution is commonly used as an osmotic solution in geotechnical

engineering. The PEG is a polymer made up of large sized molecules consisting in long
chains with the following formula: HO-[CH2-CH2-O]n-H. According to the value of n that
characterises the length of the chain, the molecular weight of the PEG changes.
Commercially available PEG have molecular weights included between 1000 and 50 000
whereas, PEG 6 000 and 20 000 are commonly used in geotechnical testing. The relationship
between osmotic pressure and PEG concentration is well known for two molecular weights
(PEG 6,000 and 20,000).

Osmotic technique has several advantages as compared to other suction controlling

techniques. Firstly, the osmotic technique allows water exchanges between the soil sample
and PEG solution in the liquid phase which reduces equilibrium time significantly (Blatz et
al., 2008; Delage and Cui, 2008). Secondly, comparing to axis translation technique, no
artificial air pressure is used to apply suction to the soil during tests. As a result, the pore
water pressure remains negative like in reality (Kassiff and Ben Shalom, 1971). However, the
major problem with this technique is that the weakness of the membrane and its sensitivity to
microbial attack and mechanical shearing. The degradation of the membrane may allow for
PEG crossing and penetrating to the sample (Blatz et al., 2008; Delage and Cui, 2008). To

14
minimize the bacterial attack, it is strongly recommended to add few drops of penicillin in the
solution before use (Kassiff and Ben Shalom, 1971). This can provide satisfactory
performance up to two weeks.

2.4.3 Vapour equilibrium technique (VET)

Relative humidity (RH) control using solutions to impose suction is commonly known as
vapour equilibrium technique (VET) in geotechnical engineering, was initially developed by
soil scientists and the first application to geotechnical testing was introduced by Esteban and
Saez (1988) (Delage et al., 1998). VET is used to impose total suction on an unsaturated soil
specimen by controlling the RH in a closed chamber under isothermal equilibrium condition.

The principle of the vapour equilibrium technique using aqueous solutions for controlling RH
and hence total suction in an isolated chamber is shown in Figure 2.13 (c) (Ng and Menzies,
2007). A soil specimen is placed in a closed container containing an aqueous solution of a
given chemical compound. RH is imposed within the sealed container based on the property
of the chemical compound in the aqueous solution. The equilibrium RH generated by
different saturated salt solutions can be found in Delage et al. (1998). Water exchanges occur
between the solution and the specimen by vapour transfer, and hence a given suction is
applied to the specimen when vapour equilibrium is achieved in the closed system. Generally
two types of solutions namely, saturated saline solutions and unsaturated acid solution with
different concentrations can be used to control RH in a closed chamber during test. The use
of saturated saline solutions is advantageous since the concentration of the solution does not
change as water exchange takes place between specimens and the vapour environment
produced by solution but, the control of suction is limited for each specific type of salt based
on the properties and purity of the specific chemical component. On the other hand,
unsaturated acid solutions can be used to impose higher suctions with continuous range of
total suction by careful control of the concentration of the solution. However, the
concentration of the unsaturated acid solution changes during the water exchange which in
turn alter the target suction value. This is highly significant when the suction value higher
than 150 MPa. Due to highly non-linear relationship between acid concentration and RH,
small changes in the concentration results dramatic change on the applied suction (Blatz et
al., 2008). It can be noted that saturated salt solutions are mostly used in the vapour
equilibrium technique for geotechnical tests (Delage et al., 1998). Furthermore, some

15
literature reported that RH can also be controlled by mixing vapour saturated gas with dry gas
using a feedback system.

The advantage of this technique is that it can be used to apply high suctions (3-1000 MPa)
comparing to other suction controlling techniques. As a result, the behaviour of unsaturated
soil at high suction range can be studied and investigated. But, the main drawbacks of this
suction controlling technique are that it needs long equilibrium time to achieve isothermal
equilibrium condition since water exchanges between soil specimen and aqueous solution are
quite slow due to the very low kinetics of vapour transfer. However, several studies attempted
to reduce the equilibrium time by circulating the humid air through the base of the soil
specimen. (Bernier et al., 1997; Cui et al., 2002; Oldecop and Alonso, 2004). Secondly, strict
temperature control of the closed system is required because small variation of temperature
significantly affects the imposed suction. Thirdly, it is very difficult to maintain a constant
RH especially at low suction range. Delage et al. (1998) stated that 1% uncertainty of RH
gives a 1.38MPa absolute uncertainty for total suction at a given temperature of 25ºC.
Therefore, this study limits the application of VET under 10MPa suction.

16
Figure 2.1 A classification of collapsible soil (Rogers, 1995)

17
Figure 2.2 Global distribution of loess deposits (Pecsi, 1990; adopted from Delage et al.,
2005)

18
Figure 2.3 A schematic map of the distribution of the Chinese loess (Jin and Liu, 2011)

19
Figure 2.4 Deposition of aeolian loess in the proximity of large ice-sheets (Andersen and
Borns, 1997, adopted from Delage et al., 2005)

(a) (b)

Figure 2.5 (a) A basic structural unit of loess (Tan, 1988), and (b) A schematic of loess
structure from mineralogical standpoint (Osipov and Sokolov, 1994)

20
Figure 2.6 Types of bonding in collapsible soils (Clemence and Finbarr, 1981)

21
Figure 2.7 Water retention curve of the intact loess from the different approaches used
(Muñoz-Castelblanco et al., 2012)

22
Figure 2.8 Water retention curve of the undisturbed collapsible loess soil, determined using
filter paper and pressure plate techniques (Haeri et al., 2014)

23
Figure 2.9 (a) Water characteristic curves at various net mean stresses (i.e., w – log s), and (b)
degree of saturation versus matric suction curves at various net mean stresses (Chen et al.,
1999).

24
Figure 2.10 Results from wetting-induced collapse tests conducted at different mean net
stress values. (a) volumetric strain as a function of the applied matric suction, and (b) degree
of saturation as a function of the applied matric suction (Haeri et al., 2014)

25
 (a)

(b)

Figure 2.11 Results of the UCL tests. (a) volumetric strains with decreasing matric suction at
constant axial load, and (b) water retention curves of loess, deduced from the wetting paths of
the UCL tests (Zhou et al., 2014)

26
Figure 2.12 Collapse and compression tests performed on natural unsaturated loess from
Northern France (Munõz-Castelblanco et al., 2011).

27
 (a)

(b)

(c)

Figure 2.13 Principles of different suction control techniques. (a) axis-translation technique,
(b) osmotic technique, and (c) vapour equilibrium technique (Ng and Menzies, 2007)

28
CHAPTER 3
TEST MATERIAL, TEST PROGRAM AND TESTING
DEVICES

In this chapter, the studied test material is introduced first with its index properties. A
detailed test program is then presented in order to achieve the objectives of this laboratory
study and finally, different testing devices are discussed.

3.1 Test material

The loess soil used in this laboratory study was collected from Xi'an, Shaanxi province of
China. High quality intact loess block samples (0.25×0.25×0.25 m3) were retrieved at 1.5 m
depth from an excavated pit. The block samples were immediately wrapped with a thick
plastic membrane (non absorbent to water) followed by several layers of clothes and wax to
preserve the in-situ water content and soil structure. Finally, the intact soil block was placed
inside the tight wooden box. A series of laboratory tests were carried out to measure the
index properties of loess sample. This series of tests includes specific gravity, in situ water
content and dry density, Atterberg limits (liquid limit, plastic limit and plasticity index),
particle size distribution and standard proctor compaction test. The specific gravity, in situ
water content and dry density, Atterberg limit tests were conducted in accordance with the
procedures given in BS1377-2 (BSI, 1990). The particle size distribution of loess investigated
by sieve and hydrometer analyses (BSI, 1990) is shown in Figure 3.1. The sand, silt and clay
contents are 0.1%, 71.9%, and 28.0%, respectively. Based on the particle size distribution and
the Atterberg limit test results, the studied loess is classified as clay of low plasticity (CL)
according to the Unified Soil Classification System, USCS (ASTM, 2006). Furthermore, the
standard proctor test was conducted to determine the compaction curve of loess (ASTM,
2012). As shown in Figure 3.2, the maximum dry density of loess is 1680 kg/m3 (1.68
Mg/m3) and the corresponding optimum water content is 18.1%. Table 3.1 summarises the
measured index properties of loess.

29
3.2 Test program

The principal objectives of the experimental study are:

(1) To measure the WRC of intact and re-compacted loess for a wide suction range (zero to
290 MPa) under zero stress

(2) To investigate the volumetric behavior of intact and re-compacted loess upon wetting and
drying for a wide suction range (zero to 290 MPa) under zero stress

(3) To investigate the effects of net vertical stress on WRC and wetting/drying induced
volumetric behavior of intact and re-compacted loess at low suction range (0.1 to 400 kPa)

In order to achieve these objectives, three series of tests were conducted. Three series of tests
are individually introduced with detail information (objective, stress path, test program and
test condition).

3.2.1 Test series I: Measurement of WRC and volume change along drying and wetting
processes for a wide suction range (zero to 290 MPa) under zero stress

This test series was designed to measure the WRC and drying/wetting induced volumetric
behavior of intact and re-compacted loess (i.e. I-0 and R-0) for a wide suction range using
most commonly used suction controlling techniques such as axis-translation, osmotic and
vapour equilibrium technique. Based on the suction controlling technique and device used for
each technique, this series can be divided into three parts like WRC tests using ATT, OT and
VET, respectively. The program of test series I is summarized in Table 3.2. To cover wide
range of suction for measurement of WRC under zero stress, two different hydraulic paths
were followed in this study called WD (wetting followed by drying) and DW (drying
followed by wetting) as shown in Figure 3.3. According to Figure 3.3, at low suction range
(zero to 0.4 MPa), hydraulic path WD (A-B-C) was followed in which intact or re-compacted
specimen was subjected to wetting at different suction stages from its in-situ/as-compacted
state to 0.1 kPa following drying with same suction stages up to 0.4 MPa and suction was
controlled by ATT. On the other hand, from medium (0.88-4.0 MPa) to high (7-290 MPa)
suction range, specimens were subjected to hydraulic path DW (A-B'-C') and suction was
imposed by osmotic and vapour equilibrium technique, respectively. In order to follow the
hydraulic path DW, for drying WRC, different specimens were subjected to different target

30
suctions along the drying path (A-Bi) while wetting WRC tests was performed following path
(A-B-Ci) at the same suction stages after applying maximum suction of 290 MPa. (see Figure
3.3)

WRC tests using ATT were carried out on intact (I-ATT) and re-compacted specimen (R-
ATT) by volumetric pressure plate apparatus (see Figure 3.5 ) at low suction range (zero to
0.4 MPa) following hydraulic path WD. The suction was controlled by axis-translation
technique. The specimen details are summarized in Table 3.3. During drying and wetting
WRC tests in medium (0.88-4.0 MPa) and high (7-290 MPa) suction range under zero stress
following hydraulic path WD, suction was imposed by employing osmotic (see Plate 3.1 )
and VET device (see Plate 3.2), respectively. WRC tests using OT were conducted at three
suction stages using different concentrations of PEG solution to impose target suction. The
detail test program and specimen condition for WRC tests using OT are summarized in Table
3.4 and Table 3.5, respectively. On the other hand, WRC tests using VET were conducted at
five different suction stages by controlling RH produced by saturated salt solution in an
isolated desiccator. Table 3.6 and Table 3.7, respectively summarized the test program and
specimen details for WRC tests using VET.

Volumetric measurements were performed at different suction stages along the drying and
wetting paths. At low suction range (0.1-400 kPa), vertical deformation of specimen inside an
oedometer ring was observed under K0 condition in order to measure volume change at each
suction stage. On the other hand, at medium (0.88-4.0 MPa) and high (7-290 MPa) suction
range, volumetric measurements were conducted by wax method (ASTM D4943-08) once
specimen achieved equalization under target suction stage.

3.2.2 Test series II: Measurement of WRC and wetting/drying induced volume change for
intact loess at different net vertical stresses

To investigate the effect of net vertical stress on WRC and wetting/drying induced volumetric
behaviour of intact loess, three tests, namely I-0, I-50, I-110 were carried out under three
different net vertical stresses of zero, 50 and 110 kPa, respectively at low suction range (0.1-
400 kPa). It is noted that the value of net vertical stress 50 and 110 kPa corresponds to depth
of sampling in the field whereas zero was chosen as a reference. It is noted that, test I-0 was
considered from the WRC test of intact loess under zero stress at low suction range (0.1-400
kPa) that was performed under test series I. During each test, volumetric measurements were

31
taken along the wetting and drying paths of WRC test based on the vertical deformation of
the specimen inside the oedometer ring under K0 condition using modified one-dimensional
stress-controllable volumetric pressure plate devices (see Figure 3.5(b)). In test series II,
matric suction was applied and controlled by means of ATT.

Figure 3.4 shows the stress path of this test series to investigate the stress effect on WRC and
wetting/drying induced volumetric behaviour. According to stress path (A-B-C-D), at first,
the initial suction (A) was determined using test setup as shown in Figure 3.5 (a). Afterwards,
net vertical stress was applied to the specimen according to the test program under its
controlled initial suction (A→B). Then, the specimen was subjected to wetting by reducing
the suction to 0.1 kPa in steps (B→C) followed by drying up to 400 kPa in steps (C→D). The
detail information with specimen state is summarized in Table 3.8.

3.2.3 Test series III: Measurement of WRC and wetting/drying induced volume change for
re-compacted loess at different net vertical stresses

Three tests, namely R-0, R-50, R-110 were carried out under three different vertical stresses
of zero, 50 and 110 kPa, respectively at low suction range (0.1-400 kPa) to study the stress
effect on WRC and wetting/drying induced volumetric behaviour of re-compacted loess.
Stress dependent WRC tests of re-compacted loess were conducted according to stress path
as shown in Figure 3.4 and volume changes due to wetting and drying under constant net
vertical stress were also measured, similar to test series II. R-0 test was considered from test
series I same as I-0. A modified one-dimensional stress-controllable volumetric pressure plate
devices was employed and suction was controlled by ATT. The test program details are given
in Table 3.9.

3.3 Testing devices

3.3.1 Modified one-dimensional stress controllable volumetric pressure plate apparatus

A modified one-dimensional stress controllable volumetric pressure plate apparatus (Ng and
Pang, 2000) was used to measure the water retention curve (WRC) of loess specimen at low
suction range (0.1-400 kPa) under K0 condition with or without net vertical stress. The axis
translation technique (Hilf, 1956) was used to control and measure the matric suction of soil
specimens. The original setup was modified for measurement of initial suction. Modification
was done by incorporating a set of ancillary devices including a pressure transducer, a DC

32
power supply and a multimeter. The pressure transducer attached to the water compartment
was used for monitoring excess pore water pressure generated under null-type axis translation
conditions. The schematic diagram of modified test setup for measurement of initial suction
is shown in Figure 3.5 (a). After determination of initial suction, the pressure transducer is
replaced with a set of hysteresis attachments including an air trap, a ballast tube attached to
the ruler and a burette, to measure the water volume change along wetting and drying paths.
Moreover, a loading frame with a set of dead load was placed on the top of the loading piston
to apply the net vertical stress on the specimen inside the oedometer ring under K0 condition
during the test. The bottom of the piston holds a circular aluminium cap and coarse porous
stone with same diameter of the specimen which directly placed on top of the specimen with
ring. In addition, a dial gauge in contact with chamber top cap was fixed with the loading
piston to measure the vertical deformation of the specimen at different suction stages with a
resolution of 0.002 mm. Therefore, changes in void ratio due to mechanical (net vertical
stress) and hydraulic (suction) loading can be captured. Pore air pressure is controlled
through a pressure line connected to the airtight chamber while the pore water pressure is
controlled at the atmospheric pressure through a 5-bar high ceramic disk mounted at the
bottom of the specimen. Figure 3.5 (b) shows the schematic diagram of modified one-
dimensional stress controllable volumetric pressure plate apparatus used to investigate the
effect of net vertical stress on WRC.

3.3.2 Osmotic device

A simple test setup called osmotic device was employed for determination of WRC at
medium range of suction (0.88-4 MPa) as shown in Plate 3.1. This type of device for WRC
tests was proposed by Cui and Delage (1996). In this device, the soil specimen which in
contact with a semi permeable membrane was immersed in a glass container of PEG solution
and finally, the total assembly was placed on a magnetic stirrer. The suction value varied with
the concentration of PEG solution and higher suction value achieved with increase of PEG
concentration. In this study, PEG 20000 and semi-permeable membrane of molecular weight
cut-off value (MWCO) of 3500 were used for applying suction to the specimen by OT. A
good contact was ensured between soil specimen and semi-permeable membrane by placing
the cylindrical specimen inside the membrane tube of similar size and also fastening the both
ends of the membrane using a special Spectra/Por closure (75 mm gripping width). The top of

33
the glass container was isolated by using plastic film in order to avoid water evaporation from
the surrounding environment during tests.

3.3.3 VET device

To measure WRC of loess at high suction range (7-290 MPa), a test setup called VET device
was used in this study. It mainly consists of desiccator assembly contained saturated salt
solution. This simple setup can impose target suction as a result of constant RH environment
generated by saturated salt solution inside the desiccator. Different salt solutions are able to
produce different RH values in a closed environment and hence, wide range of suction can be
applied at a constant temperature. A typical setup of VET device is shown in Plate 3.2. A soil
specimen with a plastic base was placed on the perforated disk which provides support for
placement of specimens under constant RH environment to impose target suction. A humidity
probe was also used to monitor the RH generated by different salt solutions and the
temperature inside the isolated desiccator during tests. Therefore, a proper calibration of the
humidity probe is required beforehand, to calculate the target suction value at a constant
temperature. The details of the calibration are given in the following section.

3.3.3.1 Calibration of humidity probe

In this study, a humidity probe (EL-USB-2, LASCAR electronics) as shown in Plate 3.3 (a)
was used throughout the experiment to measure the RH generated by saturated salt solution
and the temperature inside an isolated desiccator. It can measure the RH between 0 and 100%
with internal resolution of 0.5% and accuracy ±3.0% and temperature between -35ºC and +
80ºC with internal resolution of 0.5 ºC and accuracy ± 0.5 ºC. The humidity probe was
calibrated against known values of RH produced by three different saturated salt solutions
such as LiCl, NaCl, KNO3. The calibration technique of the humidity probe is shown in Plate
3.3 (b). At first saturated salt solution was prepared and afterwards, the RH of the probe
observed in an isolated desiccator contained salt solution of known RH. The known RH
imposed by each solution inside an isolated desiccator was adopted from Delage et al. (1998).
Saturated salt solutions were prepared according to the procedure described in ASTM E 104
(2007). Based on solubility of the salt, a required amount of salt was placed in the glass
container for preparing target amount of salt solution. Distilled water was then added in about
2-ml increments and stirred well with a stirrer after each addition until significant amount of
salt dissolved in the solution. In order to confirm the saturation of the each salt solution, a

34
small amount of salt was added to the solution and stirred again. Saturation was ensured
when excess amount of salt remained undissolved at the bottom of the container. Since the
RH produced by each salt solution is sensitive to temperature, therefore it is necessary to
strictly control the temperature of the surrounding environment. As a result, salt solution
preparation was conducted in a temperature and humidity controlled room to minimize the
temperature effect. The salt solution was then transferred to the desiccator and kept it 1 hour
for temperature equalization. Afterwards, the humidity probe which functions as a data logger
was placed on the perforated disk inside the desiccator and recorded the humidity and
temperature at 30 minutes interval for first 6 hours followed by 6 hours interval afterwards.
The RH and temperature data with time of three saturated salt solutions logged by humidity
probe are shown in Figure 3.6. Based on the result shown in Figure 3.6 (a), it is seen that
measured RH of the probe for each salt solution reached an equilibrium value within 60-70
hours and showed good agreement with the known RH as shown in dotted line produced by
each salt solution. It is worth to mention that the known RH value of each salt solution at a
constant temperature of 24ºC was adopted from Delage et al. (1998). The temperature data
with time inside the isolated desiccator contained salt solution as shown in Figure 3.7 (b)
revealed that the temperature varies between 24ºC and 25ºC and it is assumed that the
average temperature inside the desiccator is 24.5±0.5 ºC, which is consistent with the
considered temperature as well. Finally, RH value logged by humidity probe at equilibrium
of each salt solution was plotted against the known values of RH and Figure 3.7 shows the
calibration result of the humidity probe.

35
Table 3.1 Index properties of loess

Index tests Value

Specific gravity 2.69

In situ water content (%) 10.9
In situ dry density (Mg/m3) 1.237
Atterberg limits
Liquid limit (%) 36
Plastic limit (%) 19
Plasticity index (%) 17
Particle size distribution
Gravel content (≥2 mm, %) 0
Sand content (2 mm-0.063 mm, %) 0.1
Silt content (0.063 mm-0.002 mm, %) 71.9
Clay content (≤0.002 mm, %) 28.0
Standard proctor compaction test

Optimum water content (%) 18.1

Maximum dry density (Mg/m3) 1.68
Unified Soil Classification System CL

36
Table 3.2 Summary of test program for test series I

Test ID* Suction controlling Suction range Hydraulic Apparatus

technique (MPa) path**

Axis Translation 0-0.4 WD Volumetric

Technique (ATT) pressure plate

I-0 Osmotic Technique 0.88-4 DW Osmotic device

R-0 (OT)

Vapour Equilibrium 7-290 DW VET device

Technique (VET)

* I: Intact, R: Re-compacted
** Refer to hydraulic paths in Figure 3.3

Table 3.3 Summary of specimen details for WRC tests using ATT under series I

Specimen Initial state of the specimen Measured Measured final

ID* initial void ratio
GWC (%) Dry density Void ratio
suction
(Mg/m3)
(kPa)
I-ATT 1.234 10.9 1.180 182 1.177
R-ATT 1.232 10.7 1.183 77 1.169

* I: Intact, R: Re-compacted, ATT: Axis-translation technique

37
Table 3.4 Detail test program of WRC test using OT under series I

Suction stage (MPa) 0.88 1.87 4.00

WRC tests PEG concentration 0.30 0.42 0.62
using OT (g PEG/ g Water)*

Target Brix value (%) ** 20.5 26.5 34.0

Drying path ID 0.88A ID 1.87A/ ID 4A
ID 1.87B
specimen ID
Intact (I)
Wetting path IW 0.88A/ IW 1.87A/ IW 4A/
IW 0.88B IW 1.87B IW 4B
specimen ID
Drying path RD 0.88A/ RD 1.87A RD 4A/
RD 0.88B RD 4B
Re-compacted specimen ID
(R) Wetting path RW 0.88A/ RW 1.87A/ RW 4A/
RW 0.88B RW 1.87B/ RW 4B
specimen ID

Notes: I: Intact, R: Re-compacted, D: Drying, W: Wetting, Number following ID/IW denotes

suction stages (values in MPa) in drying and wetting paths, A: Specimens used for
gravimetric water content measurement, B: Specimens used for volumetric measurement by
wax method

All tests were conducted using PEG 20 000 and MWCO 3500 cellulotic semi permeable
membrane

* From Tripathy and Rees (2013)

** From Tripathy and Rees (2013)

38
Table 3.5 Summary of specimen details for WRC tests using OT under series I

Final state of the

Initial state of the specimen
specimen
Specimen ID*
Dry density Void ratio GWC (%) Void ratio
GWC (%)
(Mg/m3)

ID 0.88A 1.230 1.186 10.97 -

ID 1.87A 1.235 1.179 9.40 -
ID 1.87B 1.234 1.180 - 1.100
13.0**
ID 4A 1.235 1.179 7.00 -
IW 0.88A 1.232 1.184 8.16 -
IW 0.88B 1.246 1.159 - 1.070
IW 1.87A 1.230 1.188 7.02 -
IW 1.87B 1.236 1.176 - 1.061
IW 4A 1.235 1.178 5.60 -
IW 4B 1.225 1.195 - 1.048

RD 0.88A 10.36 1.217 1.210 9.29 -

RD 0.88B 10.15 1.232 1.183 - 1.040
RD 1.87A 10.00 1.218 1.209 7.55 -
RD 4A 10.02 1.223 1.200 6.00 -
RD 4B 10.06 1.219 1.207 - 1.020
RW 0.88A 10.56 1.213 1.218 8.58 -
RW 0.88B 10.57 1.216 1.212 - 1.020
RW 1.87A 10.21 1.215 1.232 6.66 -
RW 1.87B 10.64 1.211 1.222 - 1.010
RW 4A 10.33 1.220 1.222 5.33 -
RW 4B 10.37 1.211 1.221 - 0.990

*I: Intact, R: Re-compacted, D: Drying, W: Wetting, Number following ID/IW denotes

suction stages (values in MPa) in drying and wetting paths, A: Specimens used for
gravimetric water content measurement, B: Specimens used for volumetric measurement by
wax method; **Average GWC of six intact samples

39
 Table 3.6 Detail test program of WRC test using VET under series I

Salt
KNO3 NaCl Mg(NO3)2.6H20 MgCl2.6H20 LiCl
solution

Measured
WRC tests 95 73 58 38 12
RH (%)*
using VET
Calculated
Suction 7 43 75 132 290
(MPa)**

Drying path ID 7A/ ID 290A

ID 43A ID 75A ID 132A
specimen ID ID 7B /ID290B

Intact (I) IW7A/

Wetting path
IW 43A IW 75A IW 132A -
specimen ID IW7B

RD 7A/ RD 43A/ RD290A/

Drying path
RD 7B/ RD 43B/ RD 75A RD 132A RD290B/
specimen ID
Re- RD 7C RD 43C RD290C
compacted
RW 7A/ RW 43A/
(R) Wetting path RW 7B/ RW 43B/ RW 75A RW 132A -
specimen ID RW 7C RW 43C

Notes: I: Intact, R: Re-compacted, D: Drying, W: Wetting, Number following ID/IW denotes

suction stages (values in MPa) in drying and wetting paths, A: Specimens used for
gravimetric water content measurement, B, C: Specimens used for volumetric measurement
by wax method

* RH measured by humidity probe during tests

** Suction value calculated based on equation (4-1)

40
Table 3.7 Summary of specimen details for WRC tests using VET under series I

Final state of the

Initial state of the specimen
specimen
Specimen ID*
Dry density Void ratio GWC (%) Void ratio
GWC (%)
(Mg/m3)

ID 7A 1.206 1.230 5.82 -

ID 7B 1.221 1.203 - 1.073
ID 43A 1.203 1.236 2.31 -
ID 75A 1.244 1.162 1.68 -
ID 132A 1.256 1.141 1.33 -
ID 290A 1.222 1.202 0.83 -
13.18**
ID 290B 1.141 1.200 - 1.010
IW 7A 1.251 1.151 4.14 -
IW 7B 1.267 1.123 - 0.986
IW 43A 1.276 1.108 1.83 -
IW 75A 1.206 1.231 1.39 -
IW 132A 1.226 1.194 0.98 -

RD 7A 10.23 1.220 1.224 4.73 -

RD 7B 10.54 1.227 1.200 - 1.003
RD 7C 10.49 1.237 1.185 - 1.019
RD 43A 10.48 1.223 1.200 1.94
RD 43B 10.47 1.227 1.192 - 0.955
RD 43C 10.49 1.242 1.200 - 0.935
RD 75A 10.67 1.230 1.202 1.50 -
RD 132A 10.49 1.216 1.212 1.27 -
RD 290A 10.98 1.221 1.202 0.76 -
RD290B 10.90 1.234 1.180 - 0.921
RD 290C 10.14 1.233 1.182 - 0.944
RW 7A 10.51 1.223 1.222 3.61
RW 7B 10.67 1.227 1.192 - 0.917
RW 7C 10.77 1.223 1.200 - 0.928
RW 43A 10.12 1.225 1.200 1.69 -
RW 43B 10.80 1.228 1.209 - 0.933
RW 43C 10.92 1.220 1.222 - 0.923
RW 75A 10.27 1.221 1.221 1.38 -
RW 132A 10.13 1.227 1.193 0.98 -
*I: Intact, R: Re-compacted, D: Drying, W: Wetting, Number following ID/IW denotes suction stages
(values in MPa) in drying and wetting paths, A: Specimens used for water content measurement, B,C:
Specimens used for volumetric measurement by wax method; **Average GWC of five intact samples

41
Table 3.8 Summary of test program and specimen details for test series II.

Test ID* Net vertical Initial state of the specimen Measured Measured
stress Dry GWC Void initial final void
(kPa) density (%) ratio suction ratio
(Mg/m3) (kPa)
I-0 0 1.234 10.9 1.180 182 1.177
I-50 50 1.272 11.9 1.115 175 1.097
I-110 110 1.112 11.2 1.419 230 0.975
*
I: Intact

Table 3.9 Summary of test program and specimen details for test series III.

Test ID* Net vertical Initial state of the specimen Measured Measured
stress Dry GWC Void initial final void
(kPa) density (%) ratio suction ratio
(Mg/m3) (kPa)
R-0 0 1.232 10.7 1.183 77 1.169
R-50 50 1.231 10.9 1.221 81 0.923
R-110 110 1.235 10.8 1.232 86 0.811
*
R: Re-compacted

42
 100

80
Percentage passing (%)

60

40

20

Clay Silt Sand

Fine Medium Coarse Fine
0
0.001 0.01 0.1
Particle size (mm)

Figure 3.1 Particle size distribution of loess

1900
Dry density (kg/m3 )

1800

1700

1600

1500

1400
0 5 10 15 20 25 30
Gravimetric moisture content (%)

Figure 3.2 Compaction curve of loess

43
 Note:
Hydraulic path WD (A-B-C)
A: Initial state, AB: Wetting of specimen in steps, BC: Drying
of the specimen in steps
Hydraulic path DW (A-B'-C')
A: Initial state, AB': Drying, B'C': Wetting
Drying WRC: A-Bi; A: Initial state, Bi: Drying of the specimen
at target suction
Wetting WRC: A-B-Ci; A: Initial state, B: Drying of the
specimen at highest suction(290 MPa), C i:Wetting of the
specimen at target suction

ATT (0-0.4 MPa) OT(0.88-4 MPa) VET(7-290 MPa)

C
B
WD B'
A
DW C'

WRC following path DW Suction, s (log scale)

B1 B2 B3 B4 B5
Drying WRC
s=0.1 kPa A
Wetting WRC B
C4 C3 C2 C1

Figure 3.3 Hydraulic path of WRC tests of loess for test series I

44
 D
A: Initial suction of the specimen

AB: Application of net vertical stress

under controlled initial suction
A B
Suction (kPa)

BC: Suction decrease to 0.1 kPa in

steps (Wetting path)

CD: Suction increase in steps up to

400 kPa (Drying path)

C
Net vertical stress (kPa)
Figure 3.4 Stress path for test series II and III

45
 (a) 1
1. Dead load
2. Dial gauge
2
3. Loading piston
4. Supplied air pressure
5. Airtight chamber
6. Aluminium cap
3 7. Coarse porous stone
4 5
8. Soil specimen
9. Ceramic disk (5 bar)
6
7 10. Water drainage tube
8
11. Pressure transducer
9
12. Multimeter
10 13. DC power supply
11

12

13

(b) 14. Loading frame

15. Water drainage tube
14 16. Dead load
17. Air trap 20
18. Ballast tube
19. Measuring ruler
20. Burette

18 19
17

15

16

Figure 3.5 Schematic diagram (a) Test setup for determination of initial suction of the
specimen, and (b) Modified one-dimensional stress-controllable volumetric pressure plate
devices

46
 (a)
100

80

60
RH (%)

LiCl NaCl KNO3

40

20

0
0 20 40 60 80 100
Time (hour)
(b)
26

25
Temperature (ºC)

24

KNO3 NaCl LiCl

23

22
0 20 40 60 80 100
Time (hour)

Figure 3.6 RH and Temperature observed by humidity probe in desiccator contained

saturated salt solution of known RH. (a) RH vs. Time, and (b) Temperature vs. Time

47
 100
KNO3

80 Temperature: 24.5 0.5ºC

NaCl
Known RH (%)

60
y = 0.9759x - 1.6535
40 R²= 1

20
LiCl
0
0 20 40 60 80 100
RH measured by humidity probe (%)
Figure 3.7 Calibration result of humidity probe against known values of RH of saturated salt
solution

48
 Plastic film

Glass container

Closure

Soil specimen

PEG 20000 solutions

Semi-permeable membrane

Magnetic stirrer

Plate 3.1 Osmotic device for determination of WRC at medium suction range (0.88-4.0 MPa)
under zero stress

49
 Desiccator

Soil specimen with

plastic base

Perforated disk

Saturated salt solution

Plate 3.2 Desiccator assembly to impose suction on specimen using saturated salt solution to
measure WRC at high suction range (7-290 MPa) under zero stress

50
 (a)

(b)

Desiccator

Humidity
probe

Perforated disk

Saturated
salt solution

Plate 3.3 Calibration technique of humidity probe used to measure RH and temperature
during WRC tests (a) humidity probe, and (b) calibration of humidity probe using saturated
salt solution

51
CHAPTER 4
SPECIMEN PREPARATION AND TEST PROCEDURES

This chapter mainly discussed about preparation of three different sizes of intact and re-
compacted specimens used for WRC tests. Afterwards, the test procedures of WRC test are
presented. At first, drying and wetting WRC test of intact and re-compacted loess for a wide
suction range (zero to 290 MPa) by using three different suction controlling techniques
(series I) is discussed and secondly, the stress effects on WRC of intact and re-compacted
loess at low suction range (zero to 400 kPa) ( series II and III) are presented.

4.1 Specimen preparation

In this study, three different sizes of cylindrical specimens were considered for preparing
intact and re-compacted specimens based on the requirement of the purposely designed tests
in series I, II, and III. In test series I, specimen size was 70 mm diameter and 19 mm height,
49 mm diameter and 10 mm height, 44 mm diameter and 10 mm height, that were used for
WRC tests by ATT, VET and OT, respectively. On the other hand, only oedometer
specimens with 70 mm in diameter and 19 mm in height were prepared for test series II and
III.

4.1.1 Preparation of intact specimen

Intact specimens were prepared by pushing the cutting edge of a stainless steel ring of desired
specimen size into intact block samples. Afterwards, a small part of soil with stainless steel
ring was separated from the block sample to avoid any type of disturbance of loess block.
Finally, the remaining soil in the top and bottom was trimmed using a wire saw to obtain the
intact specimen of desired size. The preparation of intact specimen was conducted in a
temperature (T = 22±0.5ºC) and humidity (RH = 60±2.5 %) controlled room to minimize the
alteration of in-situ water content or suction during preparation. Plate 4.1 demonstrates the
preparation of intact specimen of 49 mm diameter and 10 mm height. (A) and (B) show the
wooden box of intact loess and intact loess block sample, respectively collected from the
field. Afterwards, a stainless steel ring was inserted gently into the block sample to obtain an
intact specimen of 49 mm diameter and 10 mm height as shown (C) followed by trimming of
the specimen using a wire saw (D). Finally, the ring of the specimen was removed using a
cap-rod assembly (E). The in situ dry density, gravimetric water content (GWC) and initial

52
void ratio of intact specimens of test series I, and II are summarized in Table 3.5, Table 3.7
and Table 3.8.

4.1.2 Preparation of re-compacted specimen

Re-compacted specimens were prepared following the average GWC and dry density of
intact loess, which are 1.237 g/cm3 and 10.9%, respectively. At first, 500 g of loess soil was
oven-dried at 105º C for 24 hours and passed through the 2 mm-aperture sieve until
negligible amount of soil sample was retained on the sieve. Afterwards, required de-aired
water was sprayed to the dry soil using wash bottle to avoid any possible generation of soil
lump and mixed thoroughly in steps. According to the results of calibration tests, the amount
of water content was considered 1% in excess of the target GWC to compensate for loss of
water content caused by evaporation during preparation. The mixture was sieved through a 2
mm-aperture sieve to achieve a uniform aggregation and the retained soil was broken down
with rubber pestle and sieved again until negligible amount was retained on the sieve. The
prepared soil was transferred to the zip lock plastic bag and kept inside an ice chest for 24
hours in the temperature (T = 22±0.5ºC) controlled room to ensure the moisture equalization.
Finally, the soil sample was ready for preparing re-compacted soil specimen.

The re-compacted soil specimens were prepared by static compaction method using a WF
compression machine which consists of an alignment frame, proven ring and compaction
mould holds oedometer ring with extension collar (Plate 4.2). The soil specimen was
compacted inside an oedometer ring (70 mm in diameter and 19 mm in height) at a constant
axial displacement rate of 1.0 mm/min. Compaction was done in three layers (i.e. 6.33 mm
height of each layer), with scarification between each layer to ensure the homogeneity of the
specimen. In order to achieve the target dry density, static pressure was applied to each layer,
determined by the proven ring attached with the machine and the height of each layer was
also maintained carefully by attaching a measuring ruler with the compaction rod, commonly
known as displacement control technique. After finishing the compaction of the specimen,
the mould assembly with specimen was removed from the compression machine and
oedometer specimen (70 mm in diameter and 19 mm in height) with ring was taken out by
separating the extension collar of the mould assembly as shown in Plate 4.3. The final height
of the specimen was measured using a caliper readable to 2 digits after decimal.

53
To prepare small cylindrical re-compacted specimens (49 mm diameter and 10 mm height or
44 mm diameter and 10 mm height), re-compacted specimens of 70 mm in diameter and 19
mm in height were first prepared essentially following the same procedure as discussed
previously. Afterwards, specimen was then removed from the ring using cap-rod assembly as
shown in Plate 4.4 (A) and (B). The cutting edge of the small stainless steel ring of desired
specimen size was pushed into it (C) and the remaining soil sample was taken for water
content measurement (D). At the end, bottom and top of the specimen was trimmed using
wire saw (E) and specimen was removed gently from the ring (F and G). The initial dry
density, GWC and void ratio of re-compacted specimens of test series I and III are
summarized in Table 3.5, Table 3.7 and Table 3.9.

4.2 Test procedures

4.2.1 WRC and volumetric behaviour upon drying and wetting of intact and re-compacted
loess for a wide suction range (zero-290 MPa) under zero stress

4.2.1.1 WRC tests using pressure plate apparatus by means of ATT

The WRC of intact and re-compacted loess at low suction range (0.1-400 kPa) was measured
using the ATT following the hydraulic path WD (see Figure 3.3 ). Prior to conduct the tests,
the ceramic disk of each volumetric pressure plate apparatus was saturated and its air entry
value (5 bar) was checked accordingly. Then the specimen inside the oedometer ring was
placed on the ceramic disk. The initial suction of each specimen was determined according to
the test setup of null type pressure plate apparatus as shown in Plate 4.5 (a) and the values of
intact and re-compacted specimen (I-ATT, R-ATT) are reported in Table 3.3. During initial
suction measurement, air pressure inside the chamber was raised to 400 kPa and undrained
pore water pressure response of pressure transducer was monitored with time until
equilibrium was reached. The initial suction was the difference between air and water
pressure at equilibrium.

After determination of initial suction, the test setup (see Plate 4.5 (a)) was changed to the
configurations as shown in Plate 4.5 (b). According to the hydraulic path WD, wetting
process was started from its initial suction to 0.1 kPa followed by drying up to 400 kPa in
steps with similar wetting stages. The considered suction stages were: 0.1, 1, 2.5, 6, 15, 35,
85, 200, 400 kPa. The suction of 0.1, 1, 2.5 kPa were controlled by placing the ballast tube
below the soil specimen whereas, the suction ranging from 6 to 400 kPa were applied by

54
controlling the applied air pressure. On the other hand, the water pressure was essentially
kept at atmospheric level in wetting and drying paths. The water inflow/outflow rate and
vertical deformation of soil specimen were observed at 12 hours interval using the ballast
tube attached to the ruler and dial gauge fixed with loading piston, respectively at each
suction stage (see Plate 4.5 (b)). After that, flexible tubes connecting water compartment to
the hysteresis attachments were rolled by a solid metal cylinder to flush the diffused air
bubbles, if any. Suction was assumed to be equalized when water inflow/outflow rate was
less than or equal to 0.07cm3/day (Leung, 2011) and vertical deformation was less than 0.002
mm/day and typically, it required 5 to 15 days to reach equalization. Based on the total water
inflow/outflow and vertical deformation of the soil specimen, it is possible to calculate degree
of saturation, GWC and volume change as well void ratio at each suction stage. Consistent
results were obtained when water content was calculated from a back calculation analysis
based on final water content measurement at the end of wetting and drying cycle.

4.2.1.2 WRC tests using osmotic device by means of OT

WRC tests of intact and re-compacted loess at medium suction range (0.88 to 4 MPa) were
carried out according to the hydraulic path DW as shown in Figure 3.3. Suction was imposed
by OT using simple osmotic device (see Plate 3.1). Specimens were subjected to three suction
stages along the drying and wetting paths and the applied suctions were 0.88, 1.87, 4 MPa.
Different suction values can be applied by using different concentrations of PEG 20000
solution with Spectra/Por semi-permeable membrane of MWCO 3500. According to the
hydraulic path, during drying WRC tests, specimens were subjected to drying directly at
target suction from its in-situ/as-compacted state whereas, wetting WRC tests were
performed following the same suction stages after applying maximum suction of 290 MPa
from its initial state. Therefore, during drying WRC tests, specimen was immerse directly to
the glass container of different PEG concentrations (g of PEG/g of water) to impose target
suction of 0.88, 1.87, 4 MPa, respectively. On the other hand, during wetting WRC test, at
first, specimen was subjected to maximum suction of 290 MPa from its in-situ/as compacted
condition followed by wetting along the same suction stages of drying. Therefore, all intact
and re-compacted specimens were placed in an isolated desiccator containing saturated LiCl
salt solution that produced constant RH environment to apply maximum suction of 290 MPa
using VET. The details testing procedure using VET will be discussed in the following

55
Section 4.2.1.3. Once all specimens were achieved equilibrium under maximum suction of
290 MPa, they were used for wetting WRC tests using OT.

In order to perform the tests, at first, different PEG solutions with target concentration (see
Table 3.4) were prepared to impose the target suction to the specimen. Solutions were
prepared according to the PEG concentration (g of PEG/g of water) and suction (MPa)
relationship of PEG 20000 proposed by Tripathy and Rees (2013). Bulk solution was
prepared in a glass container by taking PEG of predetermined mass to obtain target
concentration and then, distilled water was added in steps with slow stirring using a stirrer for
dissolving the PEGs in water. The solution was prepared in a temperature (T = 22±0.5ºC) and
humidity (RH = 60±2.5 %) controlled room very quickly in order to minimize evaporation
during stirring and temperature effects. Afterwards, the solution container was sealed by
plastic wrapping and left overnight in the temperature and humidity controlled laboratory
environment. Before starting the tests, tube shaped (34 mm diameter or 54 mm flat width)
semi-permeable membrane was cut into proper length of 150 mm to accommodate the
prepared soil specimen conveniently and then, immersed in distilled water for approximately
30 minutes to remove the glycerine. It is noted that all laboratory works related to semi-
permeable membrane were performed by hands wearing rubber gloves to reduce the bacterial
attack. After that, the prepared intact or re-compacted specimen was placed inside the pre-
treated semi-permeable membrane using a plastic sleeve to place the specimen appropriately
into it with fewer disturbances. The plastic sleeve was removed gently from the membrane
and both ends were closed using Spectra/Por closure in such a way that a good contact was
maintained between specimen and membrane. Prior to immersing the specimen into PEG
solution container, it is recommended to add few drops of penicillin in the solution to resist
the bacterial attack (Kassiff and Ben Shalom, 1971). Before immerse the specimen, the
weight of the specimen with membrane and closure was taken using an electronic weight
balance with a sensitivity of 0.01 g. It is worth to mention that this balance was used for all
measurements of WRC tests by OT. At the same time, the Brix value of the PEG solution
was also measured using a well calibrated commercially available hand held refractometer
(ATAGO, MASTER-53α) as an initial value. It is mentioned that Brix value gives refractive
index of the solution that can be scaled to Brix scale (i.e. % Brix). Brix value is measured to
check the PEG concentration as well as to ensure that the applied suction remained constant
throughout testing period. Several specimens were immersed together in a glass container of
PEG solution for measurement of water content and volume change under target suction (see

56
Table 3.4 for detail test program). The solution container was isolated by using plastic film
and o-ring during experiment. Few specimens were selected that were withdrawn from the
solution container at 24 hours interval after immersing for weight measurement in order to
verify equalization under target suction and at the same time, Brix value was also measured
to monitor the PEG concentration for assuring that applied suction remained constant
throughout the tests. Equalization was assumed to be reached when the change of the
specimen weight was less than 0.01 gm/day and it took around 8 to 10 days to be equalized.

Figure 4.1 shows the measured Brix value with elapsed time at different suctions of 0.44,
1.87, 4 MPa along the drying (D) and wetting (W) paths. The known Brix value (Tripathy
and Rees, 2013) with respect to PEG concentration and corresponding suction stage is shown
with dotted line as a reference for measured Brix value. Results revealed that the measured
Brix values are found to agree very well with the known Brix (%) value and all the measured
values are varied within 1%. Based on the measured Brix value (%), suction can also be
determined as a check based on the Brix (%) and suction (MPa) relationship for PEG 20000
proposed by Tripathy and Rees (2013). After equalization at each suction stage, specimen
with membrane and closure were withdrawn from the PEG solution container and afterwards,
the closure and membrane were removed. Finally, the water content of the specimen at each
suction stage was measured by the oven drying method and volumetric measurement was
conducted using wax method (ASTM D 4943-08). Based on all these measurements, degree
of saturation, GWC and void ratio at each suction stage along the drying and wetting paths
can be calculated.

4.2.1.3 WRC tests using VET device by means of VET

WRC tests of intact and re-compacted loess at high suction range (7 to 290 MPa) were
carried out according to the hydraulic path DW as shown in Figure 3.3 . Suction was imposed
by VET using simple desiccator assembly as shown in Plate 3.2. Five suction stages were
considered for WRC tests using different salt solutions such as KNO3, NaCl,
Mg(NO3)2.6H2O, MgCl2.6H2O, LiCl that produced constant RH in a closed environment to
cover a wide range of suction (7-290 MPa). This test includes two stages: (i) drying of the
specimens directly at target suction from its in-situ/as-compacted state, and (ii) wetting of the
specimens at the same suction stages after being subjected to maximum suction of 290 MPa
from its initial state. Prior to the tests, five different salt solutions were prepared according to
the ASTM E 104 (2007) and the details of preparation of saturated salt solution are given in

57
Section 3.3.3.1. During drying WRC tests, after preparation, intact and re-compacted
specimens with plastic base were placed directly in the isolated desiccator contained salt
solution to impose target suction to the specimens. On the other hand, during wetting WRC
tests, at first, all intact and re-compacted specimens were placed directly in the desiccator
contained saturated LiCl salt solution (RH=12%) from its in-situ/ as compacted condition to
impose maximum target suction of 290 MPa. Afterwards, specimens were transferred to
different isolated desiccators to follow similar suction stages along the wetting path once
equalization achieved under maximum suction. At each suction stage of drying and wetting
paths, one specimen was selected for weight measurement with elapsed time to check the
equalization of suction and final water content measurement corresponding to that suction at
the end of the test. Furthermore, two specimens from all intact and re-compacted specimens
(one from each) that subjected to drying at maximum suction of 290 MPa for wetting WRC
tests were selected as representative specimen to check the equalization as well. Weight
measurements of specimen were taken at 30 minutes, 1 hour, 2 hour, 4 hour, 8 hour, 12 hour
and 24 hour from the start of the test followed by 1 day interval and continued until it reached
an equilibrium value to assure the suction equalization using digital balance with a sensitivity
of 0.01 g. During weight measurement, specimen was taken out of the desiccator very
carefully and weight taken within 10-15 seconds to minimize the disturbance of the constant
suction environment inside the desiccator. Figure 4.2 and Figure 4.3 show the variations of
water content with time of selected intact and re-compacted specimens, respectively at
different suction stages. Figure 4.3 (a) and (b) show the weight equalization time of intact
specimens under different suctions along the drying and wetting paths, respectively. Results
demonstrated that all intact specimens in drying path achieved equilibrium at target suction
within 2 weeks whereas 8-10 days were needed to be equalized along the wetting path. On
the other hand, according to Figure 4.3 (a) and (b), all re-compacted specimens along the
drying and wetting path achieved equilibrium within 2-3 weeks. Suction equalization was
deemed to be reached once the change of water content became less than 0.04%/day (weight
change less than 0.01 gm/day) and remained constant for at least 3 to 5 days to ensure the
equilibrium under constant RH condition inside the desiccator. Finally, a calibrated humidity
probe was placed inside the desiccator once each specimen reached equilibrium to measure
the RH produced by individual salt solution and temperature inside the desiccator at 1 hour
interval until equilibrium value achieved. Based on the measured RH and temperature at
equilibrium inside the desiccator, final suction value was calculated according to the Kelvin’s
law:
58
 (4-1)

where, is suction (kPa), R is universal gas constant [8.31432 (J/molK)], T is absolute

temperature [T=273.16+t°C(K)], is specific volume of water at t°C [(1/ρw) (m3/kg)],
is molecular mass of water vapour [18.016 g/mol], and RH is relative Humidity. Figure 4.4
shows variations of RH and temperature measurement using calibrated humidity probe inside
the desiccator contained salt solution. According to Figure 4.4 (a), it is seen that equilibrium
RH value is achieved within 12-18 hours for different salt solutions and equilibrium deemed
when the RH value change in the order of 0.5% over 6 hours time period. The measured RH
at equilibrium and reference RH of different salt solutions are shown by dotted line with
arrow and solid line in the vertical axis, respectively (see Figure 4.4 (a)) and the values are
listed in the Table 4.1. Measured and reference RH of different salt solutions showed
consistent result with minor variation for Mg(NO3)2.6H2O, MgCl2.6H2O. Furthermore, the
temperature variation with time inside the desiccator of different salt solutions is shown in
Figure 4.4 (b). Results confirmed that the temperature inside the desiccators varied between
23.5°C and 24.5°C and the average temperature is considered as 24±0.5ºC. Based on the
measured RH in the desiccator at a constant temperature of 24°C, suction value is calculated
according to equation (4-1) and listed in Table 4.1. After RH and temperature measurements,
specimens were taken out from the desiccator for final water content measurement by the
oven drying method and volume measurement using wax method (ASTM D 4943-08). Water
content measurement of specimens (defined with A) was conducted at each suction stage of
drying and wetting paths while volumetric measurement of specimens (defined with B) was
performed at selected suction stages. In addition, some volumetric measurements for re-
compacted specimens (defined with C) were repeated at different suction stages to check the
repeatability and consistent results were obtained (see Table 3.6 for detail test program). By
taking all these measurements, it is possible to calculate degree of saturation, GWC and void
ratio at each suction stage.

4.2.2 WRC and wetting/drying induced volumetric behaviour of intact and re-compacted
loess under different net vertical stresses at low suction range (0.1-400 kPa) by means of
ATT

To investigate the effects of net vertical stress on wetting and drying WRC and
wetting/drying induced volumetric behaviour, two series of tests called test series II and III

59
were carried out on intact and re-compacted loess, respectively. Tests (I-50, I-110, R-50, R-
110) were conducted according to the stress path as shown in Figure 3.4 by employing
modified one-dimensional stress-controllable volumetric pressure plate devices (see Figure
3.5 (b)) and suction was controlled by ATT. The detail information of test series II and III are
summarized in Table 3.8 and Table 3.9, respectively. According to the stress path followed in
series II and III, each test mainly consists of three stages: (i) measurement of initial suction,
(ii) application of net vertical stress under controlled initial suction according to test program,
and (iii) wetting of soil specimen from its initial suction to 0.1 kPa followed by drying up to
400 kPa at different suction stages.

Before conducting WRC test of intact and re-compacted specimen under different net vertical
stresses, the ceramic disk of volumetric pressure plate apparatus was saturated and its air
entry value was checked similar to WRC test under zero stress. The specimen inside the
oedometer ring was placed on the ceramic disk and initial suction value was determined
following the same procedure mentioned in section 4.2.1.1. The initial suction values of four
tested intact and re-compacted specimens are reported in Table 3.8 and Table 3.9,
respectively. After determination of initial suction, the test setup shown in Figure 3.5 (a) was
changed by replacing pressure transducer with hysteresis attachments shown in Figure 3.5
(b). Afterwards, net vertical stress, according to test program, was applied to the specimen
under controlled initial suction. According to the stress path (Figure 3.4), specimen was
subjected to wetting path followed by drying process in several steps following the same
testing procedure of WRC tests under zero stress at low suction range (0.1-400 kPa). Like
WRC test using ATT at low suction range, water inflow/outflow rate and vertical
deformation of soil specimens were recorded twice a day at each suction stage. Similar
suction equalization criterion was also followed and each suction stage took 5 to 15 days
along the wetting and drying paths. By taking all these measurements, it is possible to
calculate GWC, void ratio, and degree of saturation at each suction stage of wetting and
drying paths.

60
Table 4.1 Salt solutions with RH and imposed suction for WRC tests at high suction range

Salt solution Known RH(%) at Measured RH (%) at Calculated

24°C* 24°C** suction (MPa)***
KNO3 92 95 7
NaCl 75 73 43
Mg(NO3)2.6H20 53 58 75
MgCl2.6H20 33 38 132
LiCl 12 12 290

* From Delage et al. (1998)

** Humidity probe measurement inside the desiccator during tests
*** Based on measured RH using equation (4-1) at a constant temperature of 24°C

61
 40

s = 4 MPa, known Brix (%) = 34

30
Brix value (%)

s = 1.87 MPa, known Brix (%) = 26.5

20
s = 0.88 MPa, known Brix (%) = 20.5

10 s 0.88D s 0.88W s 1.87D

s 1.87W s 4D s 4W
0
0 2 4 6 8 10 12 14
Time (day)

Figure 4.1 Measurement of Brix value (%) with elapsed time during drying and wetting
processes

62
 (a)
14
14 Days 21 Days
12
ID 7A
Water content, w (%)

10 ID 43A
ID 75A
8
ID 132A
6 ID 290A
I 290 (IW 132A)
4

0
0 200 400 600 800
Time (hour)

(b)
5
7 Days 14 Days IW 132A
4 IW 75A
Water content, w (%)

IW 43A
3 IW 7A

0
0 200 400 600 800
Time (hour)

Figure 4.2 Weight equalization of intact specimens under different suctions inside desiccator
(a) drying, and (b) wetting

63
 (a)
12
14 Days 21 Days
RD 7A
10 RD 43A
Water content, w (%)

RD 75A
8
RD 132A
6 RD 290A

4 R290(RW 7A)

0
0 200 400 600 800
Time (hour)

(b)
5
14 Days 21 Days

RW 132A
4
RW 75A
Water content, w (%)

RW 43A
3 RW 7A

0
0 200 400 600 800
Time (hour)

Figure 4.3 Weight equalization of re-compacted specimens under different suctions inside
desiccator (a) drying, and (b) wetting

64
 (a)
KNO3 NaCl Mg(NO3)2 MgCl2 LiCl
100

80

60
RH (%)

40

20

0
0 6 12 18 24
Time (hour)

(b)
KNO3 NaCl Mg(NO3)2 MgCl2 LiCl
28

27
Temperature ( C)

26
Desiccator Temperature: 24 0.5ºC

25

24

23
0 6 12 18 24
Time (hour)

Figure 4.4 Measured RH and temperature of different desiccators containing salt solution. (a)
RH and (b) Temperature variations with time inside the desiccators

65
 Wooden box of loess sample Intact loess block sample
(A) (B)

Insertion of ring into block Trimming of the specimen

(C) (D)

Cap rod assembly with Intact specimen (49 mm

specimen diameter and 10 mm height)
(E) (F)

Plate 4.1 Preparation of intact specimen (49 mm diameter and 10 mm height)

66
 WF compression machine

Proven ring

Compaction rod
Alignment frame

Compaction mould with

oedometer ring

Plate 4.2 The WF compression apparatus for preparing re-compacted specimen (70 mm
diameter and 19 mm height) using static compaction method

67
 (A) (B)
70 mm diameter

19 mm height

Plate 4.3 Re-compacted specimen (70 mm diameter and 19 mm height)

(A) (B)

(C) (D)

(E) (F) (G)

Plate 4.4 Preparation of re-compacted specimen (49 mm diameter and 10 mm height)

68
(a)

Dead load

Dial gauge

DC power supply

Multimeter
Airtight
chamber

Pressure transducer

(b)

Burette

Air trap

Ballast tube attached to

ruler
Water drainage tube

Plate 4.5 WRC test using pressure plate apparatus. Test setup of (a) determination of initial
suction, and (b) wetting and drying WRC test under zero stress

69
CHAPTER 5
WATER RETENTION PROPERTIES OF INTACT AND RE-
COMPACTED LOESS

This chapter presents the WRC of intact and re-compacted loess for a wide suction range
(zero-290 MPa) following two different hydraulic paths WD (wetting followed by drying)
and DW (drying followed by wetting). A comparison is made between WRC of intact and re-
compacted loess for wide suction range in different aspects. In addition, the effects of net
vertical stress on WRC of intact and re-compacted loess at low suction range (0.1-400 kPa)
are discussed. Finally, stress dependent WRCs of intact and re-compacted loess are
compared.

5.1 Water retention curves of intact and re-compacted loess for a wide suction range
(zero-290 MPa) under zero stress

Figure 5.1 shows water retention curves of intact (I-0) and re-compacted loess (R-0) for a
wide suction range (zero-290 MPa) under zero stress. The results are presented in terms of
volumetric water content (VWC) versus suction. Volume changes during wetting and drying
processes are considered in calculation of VWC. The correction of VWC for volume changes
is considered for all figures reporting VWC as a variable in this study. Data points were
measured based on the axis translation, osmotic and vapor equilibrium techniques following
two different hydraulic paths WD and DW (see Figure 3.3) to cover wide suction range (zero-
290 MPa). According to the hydraulic path, at low suction range (zero-0.4 MPa), specimen
was subjected to wetting to 0.1 kPa from its in-situ/as-compacted states followed by drying in
steps whereas, at higher suction range (0.88-290 MPa), specimens were dried at different
suction stages to the maximum suction of 290 MPa from its in-situ/as-compacted (initial)
state followed by wetting with same suction stages. Hence, wetting of intact/re-compacted
loess from in-situ/as-compacted states delivers two primary wetting curves rather than the
main wetting curves. Mualem (1977) proved that the universal hysteresis relationship
between main boundary wetting and drying curves is still valid even if the wetting curve is
replaced by a wetting scanning curve. Further theoretical study of hysteresis by Mualem
(1984) clarified similarity between a primary hysteresis loop and the main hysteresis loop. In
other words, a primary wetting curve of a specific soil would be similar to the main wetting

70
curve. However, lower hysteresis could be expected if a primary wetting curve is given rather
than the main wetting curve since primary wetting curves should be higher than the main
wetting curve. Since it was important in this study to preserve the initial structure of loess,
therefore, drying and wetting WRC tests were started from in-situ/as-compacted states at low
and high range of suction. Furthermore, the WRCs of intact (I-0) and re-compacted loess (R-
0) are also presented in (w-s) plane as shown in Figure 5.2. During WRC tests at high suction
range, volumetric measurements were conducted at selective suction stage, whereas GWC
was measured at each suction stage. Therefore, Figure 5.2 provides more clear tread of water
retention characteristics of intact and re-compacted loess especially at high suction range.
However, the experimental results are discussed in the following parts based on Figure 5.1.
According to the results of Figure 5.1, there are some similarities and differences between
water retention characteristics of intact and re-compacted loess. In terms of similarities,
hydraulic hysteresis can be observed for both intact and re-compacted loess over the whole
studied range of suction. In order to quantify hysteresis, the concept of degree of hysteresis
proposed by Lu and Khorshidi (2015) is considered. Degree of hysteresis at a specific suction
is defined as the difference between wetting and drying water content over the mean water
content. Hysteresis is consistently higher in suction range of 10-100 kPa compared to suction
range above 1 MPa for both intact and re-compacted loess. The difference between hysteresis
in low and high range of suction can be explained by relative contribution of capillary and
adsorption phenomena. Capillary is the dominant mechanism to maintain water inside macro-
pores in low range of suction. On the other hand, water retention characteristics in micro-pore
region associated with high suction values is mainly governed by adsorption (Romero et al.,
2011). In the capillary dominant region, the combined effects of ink-bottle pore neck
mechanism, solid-liquid-air contact-angle mechanism, and entrapped air come to play a
significant role in retention behavior along wetting and drying paths, which eventually results
in a marked hydraulic hysteresis in the low suction range (Hillel, 1998). However, adsorption
dominates over capillary at high range of suction beyond residual state in which only
adsorption water can exist in soil due to surface hydration. Because the water retention in
this regime is different from capillary, the influences of the ink bottle and contact angle are
not applicable and hence less pronounced hysteresis compared to the low range of suction is
observed.

Regarding the differences, intact loess shows marginally higher water retention capability
along the drying and wetting paths than re-compacted loess between suction 1 and 290 MPa.

71
However, no clear trend is observed for WRC of intact and re-compacted loess at low suction
range (0.1-400 kPa). In addition, more gentle adsorption and desorption rates together with a
lower air entry value (3 kPa compared with 10 kPa of intact loess) can be seen for re-
compacted specimen. The most important difference is probably the hysteretic behaviour.
Intact loess indicates larger hysteresis than re-compacted loess for suction values less than 20
kPa especially around the air entry value (i.e. 10 kPa) because of the existence of the large
constricted pores (macro-pores larger than 100m can be constricted) which are only
accessible through smaller pores (Ng et al., 2016). On the other hand, re-compacted loess
shows more pronounced hysteresis in the suction range between 100 and 400 kPa compared
to intact loess due to different pore size distribution of two materials (Ng et al., 2016).
Muñoz-Castelblanco et al. (2012) reported noticeable hysteresis for an intact loess in both dry
and wet sides of WRC with respect to the in-situ suction. However, no hysteresis was
observed around the in-situ water content. It was postulated that part of retention curve with
no hysteresis corresponds to the natural variation of suction due to the seasonal changes.

5.2 Stress effects on WRC of loess at low suction range (0.1-400 kPa)

5.2.1 WRC of intact loess at different net vertical stresses

Figure 5.3 shows the WRCs of intact loess at different net vertical stresses of zero, 50, 110
kPa, respectively under K0 condition. It should be noted that all three tests were started from
its in-situ suction ranging from 175 to 230 kPa (shown by three arrows) and similar wetting
and drying cycle was applied for tests I-0, I-50 and I-110. Adsorption and desorption rates are
defined as the gradients of linear part of wetting (mw) and drying (md) branches of WRC,
respectively. Compared to the results of I-0, I-50 has a lower adsorption rate (0.662 vs.
1.049). However, the tendency of wetting branch of I-50 to shifting upwards could be due to
the lower initial void ratio of I-50 compared to that of I-0 rather than higher stress level. The
effect of net stress especially on wetting branch of WRC is more pronounced when results of
I-110 and I-0 are compared. It is seen that water retention capability of I-110 is increased
progressively with decreasing suction in a different fashion as compared to I-0, I-50. WRC of
I-110 has the least adsorption rate, and desorption rate amongst all three conducted tests. The
lower adsorption and desorption rates for an undisturbed sandy silt-clay under higher applied
net stress are reported by Ng and Pang (2000). Parameters related to the water retention
curves of intact loess are summarized in Table 5.1. Results indicate a consistent reduction in
adsorption rate, and desorption rate, with an increase of net stress. However, effect of net

72
stress is more pronounced on adsorption rate which is associated with wetting branch of
WRC. Results revealed that, the wetting branch shifts upwards with an increase of net stress
while the drying branch is not affected significantly. One possible reason is the existence of
large constricted pores in intact loess as evidenced by microstructural observations (Ng et al.,
2016). The large constricted pores control the water retention behaviour during wetting
processes. As net stress increases, the constricted pores collapse and less amount of water can
be adsorbed and adsorption rate decreases accordingly. On the other hand, drying is mainly
controlled by narrower channels surrounding large constricted pores. Since net stress cannot
significantly influence these channels, drying water retention curve is not affected much.
Since the wetting branch shifts upwards with an increase of stress and drying branch is not
affected much, hysteresis also reduces with an increase of stress. The WRC is a result of the
pore size distribution. Changes in pore size distribution are a consequence of stress changes
and suction changes and, also, on the particular constitutive behaviour of the studied soil.

5.2.2 WRC of re-compacted loess at different net vertical stresses

The wetting and drying WRCs of re-compacted loess (R-0, R-50, and R-110) at different net
vertical stresses of zero, 50, 110 kPa under K0 condition are shown in Figure 5.4. Similar to
intact series, all tests were started from its initial state and same wetting/drying cycle was
applied for each test. Compared to R-0, specimen under 50 kPa net vertical stress (R-50)
shows higher water retention capability and it increased progressively with reducing suction
along the wetting path. However, wetting WRC of R-110 shows almost similar results
compared to R-50. This implies that stress effect on wetting WRC of re-compacted loess
higher than 50 KPa net vertical stress is insignificant. Less sensitivity of stress effects on
wetting WRC of loess is reported by Haeri et al. (2014). On the other hand, Zhou et al. (2014)
showed that water retention capacity along the wetting path is decreased slightly with
increasing the stress level, which is opposite to the observed results of this study.

According to the results of stress dependent drying WRC of re-compacted loess, it is

evidenced that drying branch is influenced significantly and desorption rate is increasing with
the increase of stress level. The stress effects on both wetting and drying branch of WRC
curve are results of change of pore size distribution as well as stress effects on both macro
and micro-pore populations. The parameters related to WRC of re-compacted loess are
summarized in the Table 5.2. Results show that AEV, adsorption and desorption rate are
increased with stress level and effect of net vertical stress is more pronounced on desorption

73
rate compared to adsorption rate. The opposite trend of adsorption and desorption rate with
stress level for re-compacted sandy silt-clay is reported by Ng and Pang (2000). Since both
wetting and drying branch of WRC are influenced with net vertical stress and shift towards
each other, hydraulic hysteresis reduces with increasing stress level which is expected. This is
because of change of large pore size distribution to small pore size distribution after
application of stress. Hysteresis is more pronounced for large pore size distribution which
eventually caused by ink-bottle and contact angle effects. However, the effects are reduced
once stress applied as a result of more uniform small pore size distribution.

5.3 Comparison of WRC of intact and re-compacted loess under different net vertical
stresses

The comparison of stress dependent WRC of intact and re-compacted loess at different net
vertical stresses of zero, 50 and 110 kPa, respectively is shown in Figure 5.5. Results are
presented in terms of degree of saturation, Sr versus matric suction. Volume changes during
wetting and drying processes are considered in calculation of degree of saturation, Sr. The
correction of degree of saturation, Sr for volume changes is considered for all figures.
According to the results of I-0, R-0, intact and re-compacted loess show different water
retention characteristics in which intact loess shows higher adsorption and desorption rates
together with high air entry value. The trend of wetting and drying WRC of intact and re-
compacted loess under zero stress is different. However, stress dependent WRCs of intact and
re-compacted loess under 50 and 110 kPa net stress show that intact and re-compacted loess
have different water retention characteristics along the wetting and drying path. Results
revealed that intact loess shows marginally lower adsorption and desorption rate together
with lower AEV compared to re-compacted loess. In addition, significant hysteresis is
observed for both intact and re-compacted loess under the same constant net stress.

5.4 Summary

The WRC of intact and re-compacted loess for a wide suction range (zero-290 MPa) is
discussed and compared considering different aspects. Afterwards, the stress effects on WRC
of intact and re-compacted loess are also investigated. Stress dependent WRCs of intact and
re-compacted loess are measured at zero, 50 and 110 kPa net vertical stresses, respectively.
Finally, the stress dependent WRCs of intact and re-compacted loess are compared. Based on
the experimental results, following conclusions are drawn:

74
Hydraulic hysteresis is observed for both intact and re-compacted loess over the whole
studied range of suction. Hysteresis is consistently higher in suction range of 10-100 kPa
compared to suction range above 1 MPa for both intact and re-compacted loess. The
difference between hysteresis in low and high range of suction can be explained by relative
contribution of capillary and adsorption phenomena. In the capillary dominant region at low
suction range, the combined effects of ink-bottle pore neck mechanism, solid-liquid-air
contact-angle mechanism result a marked hydraulic hysteresis compared to high suction
range.

Intact loess shows higher adsorption and desorption rates together with a high air entry value
comparing with re-compacted loess.

The most important difference is probably the hysteretic behaviour. Intact loess indicates
larger hysteresis than re-compacted loess for suction values less than 20 kPa especially
around the air entry value (i.e. 10 kPa) because of the existence of the large constricted pores
which are only accessible through smaller pores. On the other hand, re-compacted loess
shows more pronounced hysteresis in the suction range between 100 and 400 kPa compared
to intact loess due to different pore size distribution of two materials.

Wetting WRC of intact loess is influenced by the increase of net vertical stress while drying
WRC is insensitive to stress effects. The is due to the collapse of the large macro-pores of
intact loess after application of stress that control the water retention behaviour during
wetting processes while stress effects on narrower channels (micro-pores) surrounding large
macro- pores is insignificant that influence drying branch of WRC. In addition, hysteresis is
reduced with increase of net vertical stress due to the modification of pore size distribution.

Drying WRC of re-compacted loess is influenced significantly with the increase of net
vertical stress while net stress higher than 50 kPa on wetting WRC is less pronounced. In
addition, AEV, adsorption and desorption rate of re-compacted loess are increased with stress
level and effect of net vertical stress is more pronounced on desorption rate compared to
adsorption rate.

Effects of net stress on water retention characteristics of intact and re-compacted loess is
different. With the increase of net stress, AEV, adsorption and desorption rate are decreased
for intact loess while opposite trend is observed for re-compacted loess.

75
Stress dependent WRC of intact and re-compacted loess indicates different water retention
characteristics in which intact loess shows marginally lower adsorption and desorption rate
together with lower AEV compared to re-compacted loess. However, significant hysteresis is
observed for both intact and re-compacted loess under constant net stress.

76
Table 5.1 Parameters of water retention curves for intact loess

Test ID Net vertical Adsorption Desorption rate Air entry value

stress rate (md) (AEV) (kPa)
(kPa) (mw)
I-0 0 1.049 0.468 9
I-50 50 0.662 0.426 6
I-110 110 0.424 0.360 4

Table 5.2 Parameters of water retention curves for re-compacted loess

Test ID Net vertical Adsorption Desorption rate Air entry value

stress rate (md) (AEV) (kPa)
(kPa) (mw)
R-0 0 0.420 0.295 3
R-50 50 0.450 0.430 8
R-110 110 0.525 0.654 10

77
 0.6
ATT OT VET
Volumetric water content, ϴw
0.5
Drying I-0
0.4
R-0
0.3

0.2
Wetting
0.1
Initial state
0
0.1 1 10 100 1000 10000 100000 1000000
Suction, s (kPa)

Figure 5.1 Comparison of water retention curve of intact and re-compacted loess for wide
suction range (zero-290 MPa)

50
Gravimetric water content, w (%)

Drying I R
40

ATT OT VET
30

20
Wetting
10

Initial state
0
0.1 1 10 100 1000 10000 100000 1000000
Suction, s (kPa)

Figure 5.2 WRC of intact and re-compacted loess specimens in (w-s) plane

78
 0.6
Volumetric water content, θw I-0
0.5 I-50
Drying

0.4 I-110

0.3

0.2 Wetting

0.1
Start of each test
0
0.1 1 10 100 1000
Matric suction (kPa)

Figure 5.3 Wetting and drying WRCs of intact loess under different net vertical stresses

0.6
Volumetric water content, θw

R-0
0.5 R-50
Drying

0.4 R-110

0.3

0.2 Wetting

0.1
Start of each test
0
0.1 1 10 100 1000
Matric suction (kPa)
Figure 5.4 Wetting and drying WRCs of re-compacted loess under different net vertical
stresses

79
 1
I-0
Degree of saturation, Sr
Drying
0.8
R-0
0.6

0.4
Wetting

0.2
Start of each test

0
0.1 1 10 100 1000
Matric suction (kPa)

1
I-50
Degree of saturation, Sr

Drying
0.8
R-50
0.6

0.4
Wetting

0.2 Start of each test

0
0.1 1 10 100 1000
Matric suction (kPa)

1
I-110
Degree of saturation, Sr

Drying
0.8
R-110
0.6

0.4
Wetting

0.2
Start of each test

0
0.1 1 10 100 1000
Matric suction (kPa)

Figure 5.5 Comparison of stress dependent WRC of intact and re-compacted loess

80
CHAPTER 6
VOLUMETRIC BEHAVIOUR OF INTACT AND RE-
COMPACTED LOESS

In this chapter, volume change behaviour of intact and re-compacted loess due to drying and
wetting for a wide suction range (zero-290 MPa) covered by following different two
hydraulic paths is presented first and comparison is made as well in different aspects.
Following that, the effects of net vertical stress on wetting/drying induced volumetric
behaviour of intact and re-compacted loess at low suction range (0.1-400 kPa) are discussed.
Finally, the stress dependent wetting/drying induced volumetric behaviour of intact and re-
compacted loess is compared and discussed.

6.1 Volumetric behaviour of intact and re-compacted loess upon drying and wetting for
a wide suction range (zero-290 MPa) under zero stress

Figure 6.1 depicts the volume changes of intact and re-compacted loess due to drying and
wetting and the results are presented in terms of void ratio versus suction. Figure 6.1 (a)
shows the wetting/drying induced volumetric behaviour at low suction range (0.1-400 kPa)
following hydraulic path WD (wetting followed by drying) from its in-situ/as-compacted
states whereas, volume changes at high suction range (880-290 000 kPa) following hydraulic
path DW (drying followed wetting) from its initial condition is plotted in Figure 6.1 (b).
There are some similarities and differences between two types of materials in terms of
volume change behaviour. According to the results, it is clearly seen that both intact and re-
compacted loess show negligible volume change due to drying and wetting at low suction
range while significant volume changes occur at high suction range. At high suction range,
both intact and re-compacted loess considerably shrink along the drying path and the
maximum drying induced shrinkage of intact and re-compacted loess is 8% and 12%,
respectively, in terms of volumetric strain. This high shrinkability is mainly attributed due to
the high initial porosity or loose structure of loess. On the other hand, along the wetting path,
re-compacted loess shrinks slightly between suction 290 and 7 MPa and afterwards,
substantial swelling is observed between 7 and 4 MPa followed by a gentle slope. The
maximum swelling of intact and re-compacted loess is about 4% and 5%, respectively, in
terms of volumetric strain. However, drying induced shrinkage is more pronounced than
wetting induced swelling at high suction range for both materials.

81
Difference in shrinkage behaviour of intact and re-compacted loess is significant. It is
evidenced that, re-compacted loess shrinks noticeably more and its shrinkage could reach
50% higher volumetric strain than that of intact loess for the considered suction range. This
observation is consistent with results of compression tests on saturated loess. Oedometer test
results on two saturated intact and re-compacted loess are presented in Figure 6.2 . Specimens
were prepared in the same way as described for WRC tests. Results of oedometer tests
indicated lower yield stress and higher compressibility for re-compacted loess than those of
intact loess. In fact, intact loess delivers a stiffer structure compared to the re-compacted
loess. Some research on compacted clay has clarified that drying induces shrinkage and the
induced shrinkage reduces to an asymptotic value as cycle number increases (Dif and
Bluemel, 1991; Alonso et al., 2005; Farulla et al., 2010). The behaviour is an outcome of a
continuous rearrangement of soil particles, which results in a less activate microstructure.
This trend is also consistent with the framework of BExM described by Alonso et al. (1999).
Based on this type of volumetric behaviour as a benchmark, the volumetric behaviour of
natural soil can be explained. In fact, intact loess has been subjected to countless cycles of
wetting and drying in the field. The application of wetting and drying cycles to the soil would
result in rearrangement of particles until a more stabilized structure be formed. On the other
hand, re-compacted specimen which has not experienced similar hydraulic events would
suffer more particle rearrangement leading to more volumetric compression during drying
process compared to the intact loess.

6.2 Stress effects on wetting and drying induced volumetric behaviour of loess at low
suction range (0.1-400 kPa)

6.2.1 Collapse and shrinkage of intact loess along wetting and drying processes under
different net vertical stresses

6.2.1.1 Test results of intact loess

Figure 6.3 shows the variations of void ratio with suction during wetting and drying
processes under different net vertical stresses of 0, 50, 100 kPa. According to test I-0,
insignificant volume changes occur in terms void ratio for a cycle of wetting from its initial
suction and drying up to 400 kPa. The difference between the maximum and minimum void
ratio is 1.3% during the test. Regarding the test I-50, small reduction in void ratio (0.02) is
observed due to the application of 50 kPa net vertical stress under constant initial suction.

82
According to the results of compression tests, Cc and Cr of intact loess are 0.383 and 0.016,
respectively. The same swelling index can be used for I-50 specimen if it is assumed that Cr
is independent of suction. Considering that, a change of 0.02 in void ratio is calculated due to
the application of 50 kPa net stress. This confirms that a vertical net stress of 50 kPa is less
than the yield stress of unsaturated intact loess. Similar to the test I-0, negligible deformation
is observed (1.4% change in void ratio) for I-50 during the following wetting and drying
sequences. Apart from the elastic compression caused by the application of 50 kPa net stress
to I-50 specimen, no collapse due to loading/wetting is observed for I-0 and I-50. The state of
these two specimens therefore remains within the elastic region. However, significant volume
changes are occurred for the test I-110 due to loading, wetting, and drying. It is also noted
that I-110 has a higher void ratio compared to the other two tests probably due to the natural
variability of intact loess. Noticeable compression is observed due to application of net
vertical stress of 110 kPa under its controlled initial suction and stress induced volume
change is around 8% in terms of void ratio. The comparison between results of I-50 and I-110
clarifies that the yield stress corresponding to the initial state should be between 50 and 110
kPa. Since there is no measurement to quantify this point, the exact value is unknown and
further studies are required. In addition, significant wetting induced collapse and considerable
drying induced shrinkage are observed for specimen under net vertical stress of 110 kPa.
Wetting induced collapse is around 10%, in terms of volumetric strain which is more
pronounced compared to drying induced shrinkage.

6.2.1.2 Discussion on volume change of intact loess

The observed trends for the evolution of void ratio with stress and suction are explained
within the framework of BBM (Alonso et al., 1990). Based on the concept of LC yield curve,
a higher yield stress would be expected for a higher suction value. In order to depict the LC
yield locus for the studied intact loess, yield points in the plane of suction-net stress are
needed. Pre-consolidation stress of saturated intact loess can be estimated from oedometer
tests, which is 19 kPa. The second point can be obtained from the results of I-50 presented in
Figure 6.3. Since no yielding is observed during wetting process for this test up to 0.1 kPa
suction, the yield suction can be assumed to be less than 0.1 kPa for 50 kPa net vertical stress.
Results of I-50 together with I-110 in Figure 6.3 are used to retrieve another yield point. It is
mentioned previously that, the initial state of I-50 specimen compressed to 50 kPa net vertical
stress is within the elastic region. On the other hand, the I-110 specimen having similar

83
suction experienced collapse due to the application of net stress. These observations imply
that the yield stress corresponding to the initial suction should be between 50 and 110 kPa.
Since no measurement was conducted to quantify this point, it is simply assumed that yield
stress associated with the initial suction is 90 kPa. Finally, the initial LC yield locus shown by
LC1 is plotted in Figure 6.4 (a) based on the information obtained from oedometer and WRC
tests.

The stress paths followed for the three conducted tests are shown in Figure 6.4 (a). All the
applied mechanical loading and hydraulic wetting/drying for I-0 and I-50 tests fall into the
elastic region bounded by the initial LC surface of intact loess, i.e. LC1. Apart from the
elastic compression caused by the application of 50 kPa net vertical stress to I-50 specimen,
no collapse due to loading/wetting is observed for these two specimens. In case of I-110,
upon the application of 110 kPa vertical stress, the LC1 is pushed towards right hand side
since the applied stress is larger than the yield stress corresponding to the initial suction. The
LC1 curve is then expanded to the new LC2 curve. Since the state of the soil is on the yield
surface, any disturbance caused by mechanical loading or hydraulic wetting would results in
repositioning of the yield surface. Therefore, progressive expansion of LC curves happened
during wetting process until the LC3 is reached by approaching saturation state.

Although the LC surface has been expanded from LC1 to LC3, such an expansion has not
occurred for the SI curve since I-110 shrinks more than I-0 and I-50 (for a suction range
between 0.1 and 400 kPa). Results of Figure 6.3 show that yield stress associated with
drying-induced shrinkage decreases and total shrinkage increases with increase of net stress.
This would suggest a lower yield stress associated with shrinkage for I-110 compared to the
other two specimens. This trend is against some studies showing reduction in shrinkability
with increase of net stress (e.g. Sheng et al., 2008; Chiu and Ng, 2012). However, the
observed trend is consistent with some previous studies (e.g. Dif and Bluemel, 1991; Alonso
et al., 2005; Farulla et al., 2010). One possible explanation is that inter-particle arrangement
of intact loess becomes destroyed once applied stress goes beyond yield stress. As a result,
intact loess having a loose structure suffers from severe particle rearrangement and hence
volume changes as net stress increases. As indicated in Figure 6.4 (b), it is postulated that
loading collapse yield curves and suction increase yield surfaces of intact loess are
uncoupled.

84
6.2.2 Collapse and shrinkage of re-compacted loess along wetting and drying processes under
different net vertical stresses

Figure 6.5 shows the volume change of re-compacted specimens along the wetting and drying
paths under different net vertical stresses of 0, 50 and110 kPa, respectively. The results are
presented in terms of void ratio versus suction. According to the tests results of R-0,
negligible volume changes occur due a cycle of wetting to 0.1 kPa from its as-compacted
state and drying up to 400 kPa in steps. The maximum volume change due to a cycle of
wetting and drying for re-compacted specimen under zero stress is around 1%, in terms of
void ratio. Regarding the tests R-50 and R-110, stress induced collapse is observed for re-
compacted specimen compressed to 50 and 110 kPa net vertical stress under its controlled
initial suction. The change of void ratio is 4% and 5%, respectively for R-50 and R-110. This
implies that the yield stress of unsaturated re-compacted loess is less than the vertical net
stress of 50 kPa. Apart from the loading induced compression caused by application of net
vertical stress 50 and 110 kPa, both specimens (R-50 and R-110) show significant wetting
induced collapse under constant net stress. Wetting induced collapse under constant net
vertical stress increases progressively with reducing suction and substantial collapse occurred
at nearly saturated condition (suction between 2 to 5 kPa) for both R-50 and R-110. However,
re-compacted specimen under 110 kPa net stress (R-110) shows higher wetting induced
collapse along the wetting path and finally, it deforms 50% more due to wetting in terms of
volumetric strain compared to R-50. This implies that wetting induced collapse or volumetric
deformation increases with increase of net stress. Similar stress dependent wetting induced
volumetric deformation of loess is also reported in the literature (Haeri et al., 2014; Zhou et
al., 2014). Significant volumetric deformation is a result of continuous rearrangement of
particles due to loading and wetting that increases with stress level and leading to more
volumetric deformation. This postulation is also consistent with the oedometer test results of
saturated re-compacted loess (Figure 6.2). According to test result, re-compacted loess has
lower yield stress and it also shows higher compressibility with increasing stress. On the
other hand, insignificant volume changes occur along the drying path for both R-50 and R-
110 for the studied suction range.

85
6.3 Comparison of wetting/drying induced volumetric behaviour of intact and re-
compacted loess under different net vertical stresses

Figure 6.6 depicts the comparison of wetting and drying induced volumetric behaviour of
intact and re-compacted loess at different net vertical stresses of zero, 50 and 110 kPa,
respectively. Results show the volume change in terms of void ratio with suction along the
wetting and drying paths. All tests were started from its initial suction and similar
wetting/drying cycle was applied under different net vertical stresses. Comparing results of I-
0 and R-0, negligible volume changes occur due to a cycle of wetting and drying. The
maximum volume change of intact and re-compacted loess under zero stress is around 1%, in
terms of void ratio. Intact and re-compacted specimen compressed to 50 kPa net vertical
stress (I-50, R-50) show different volumetric behaviour due to loading and wetting/drying.
Re-compacted specimen under 50 net stress shows considerable volume change (change of
void ratio is around 4%) after application of stress while small reduction in void ratio (0.02)
is observed for intact specimen under its controlled initial suction. This implies that yield
stress of unsaturated intact loess is higher than re-compacted loess under its initial suction.
On the other hand, along the wetting path, re-compacted loess shows progressive wetting
induced collapse with suction decrease while no volume change is observed for intact loess
due to wetting under the same net stress. However, insignificant volume change is observed
upon drying both intact and re-compacted loess under net stress 50 kPa for considered suction
range (0.1 to 400 kPa).

Comparing the results of I-110 and R-110, stress dependent wetting induced collapse and
drying induced shrinkage are observed for intact and re-compacted loess. It is noted that I-
110 has a higher void ratio compared to the R-110 probably due to the natural variability of
intact loess. However, both specimens (I-110 and R-110) show stress induced compression
under its initial suction in which intact shows higher volumetric compression compared to re-
compacted loess. This is because of high initial porosity of intact loess that collapse under net
stress of 110 kPa and higher volumetric compression is occurred. On the other hand,
significant wetting induced collapse is observed for both I-110 and R-110 and the maximum
wetting induced collapse is around 10% and 20%, respectively in terms of volumetric strain
considering its initial state. In addition, considerable drying induced shrinkage is observed for
I-110 compared to R-110. This experimental result evidenced that wetting induced collapse
and drying induced shrinkage are stress dependent for intact and re-compacted loess and

86
wetting induced collapse of re-compacted loess is more pronounced compared to intact loess
under same the net stress.

6.4 Summary

Volumetric behaviour of intact and re-compacted loess along the drying and wetting paths for
a wide suction range (zero-290 MPa) is discussed and compared. Afterwards, collapse and
shrinkage of intact and re-compacted loess along wetting and drying processes under
different net vertical stresses are investigated and analyzed. Finally, the wetting/drying
induced volumetric behaviour of intact and re-compacted loess at different net vertical
stresses is compared. Based on the all results and discussion, several conclusions can be
drawn:

Both intact and re-compacted loess considerably shrink and swell due to drying and wetting,
respectively at high suction range (0.88-290 MPa) while negligible volume change is
observed at low suction range (zero-0.4 MPa) for both wetting and drying paths.

Drying induced shrinkage is more pronounced than wetting induced swelling at high suction
range (0.88-290 MPa) for both intact and re-compacted loess. The maximum drying-induced
shrinkage of intact and re-compacted loess is 8% and 12%, respectively, whereas, maximum
swelling of intact and re-compacted loess is about 4% and 5%, respectively, in terms of
volumetric strain. High shrinkability is attributed due to the high initial porosity or loose
structure of loess.

Negligible volume change due to wetting and drying is observed for intact specimen under
zero and 50 kPa net vertical stress for considered suction range of 0.1-400 kPa. However,
significant wetting-induced collapse and drying-induced shrinkage are observed for the intact
specimen compressed to 110 kPa net stress in which wetting induced collapse is more
pronounced than drying induced shrinkage.

Insignificant volume change due to application of net vertical stress of 50 kPa is observed
while substantial volumetric deformation results after application of net stress 110 kPa under
its controlled initial suction for intact loess due to the collapse of large macro-pores. It also
implies that that the yield stress of intact loess corresponding to the initial state should be
between 50 and 110 kPa.

87
Stress and wetting induced collapse are observed for re-compacted specimen (R-50 and R-
110) compressed to 50 and 110 kPa net vertical stress. However, wetting induced collapse
under constant net stress increases progressively with reducing suction and substantial
collapse occurred at nearly saturated condition (suction between 2 to 5 kPa). In addition, R-
110 shows higher wetting induced collapse along the wetting path and finally, it deforms
50% more due to wetting in terms of volumetric strain compared to R-50.

Wetting and drying induced volumetric behavior of intact and re-compacted loess is stress
dependent for the considered suction range in which wetting induced collapse under constant
stress is more pronounced than drying induced shrinkage. Re-compacted loess shows
significantly higher wetting induced collapse compared to intact loess under the same net
stress.

88
 (a) 1.3
ATT
Wetting
1.2
Void ratio, e

1.1
Drying

1.0 Initial state

0.9 I-0 R-0

0.8
0.1 1 10 100 1000
Suction, s (kPa)

(b) 1.3
OT VET

1.2

Drying
Void ratio, e

1.1
Initial state
1.0

I-0 R-0
0.9
Wetting
0.8
10 100 1000 10000 100000 1000000
Suction, s (kPa)

Figure 6.1 Comparison of changes in void ratio with suction along drying and wetting paths
between intact and re-compacted loess under zero stress (a) at low suction range (0.1-400
kPa), and (b) at high suction range (880-290 000 kPa).

89
 1.5
Yield points

1.2
Start of the test Cc=0.383
Void ratio

0.9
Cr=0.016

0.6

0.3 Intact
3 kPa 19 kPa Re-compacted
0
1 10 100 1000
Vertical effective stress (kPa)

Figure 6.2 Oedometer test results of saturated intact and re-compacted specimens

1.5
1.4 I-0 I-50 I-110
1.3 Initial state
Wetting
1.2
Void ratio, e

1.1
1.0
0.9 Drying
After compression
0.8
0.7
0.1 1 10 100 1000
Matric suction (kPa)

Figure 6.3 Wetting and drying induced volumetric behaviour of intact loess under different
net vertical stresses

90
 (a) 400

300 Initial state

Suction (kPa)

200 I-0
I-50
I-110
100 LC 1
LC 2
LC 3
0
0 25 50 75 100 125 150 175 200
Net vertical stress (kPa)

(b) 3.5
3 LC 1

2.5 SI 1
Suction (MPa)

2 LC 3

1.5 SI 3

1
0.5
0
0 25 50 75 100 125 150 175
Net vertical stress (kPa)

Figure 6.4 (a) Stress paths followed in WRC tests of intact loess and the evolution of LC
yield surface during the test I-110, and (b) postulation of SI yield curves

91
 1.5
1.4 R-0 R-50 R-110
Wetting Initial state
1.3
1.2
Void ratio, e

1.1
1.0 After compression

0.9
0.8
Drying
0.7
0.1 1 10 100 1000
Matric suction (kPa)

Figure 6.5 Wetting and drying induced volumetric behaviour of re-compacted loess under
different net vertical stresses

92
 1.5
1.4 I-0 R-0
1.3 Wetting Initial state

1.2
Void ratio, e

1.1
Drying
1.0
0.9
0.8
0.7
0.1 1 10 100 1000
Matric suction (kPa)

1.5
1.4 I-50 R-50
Initial state
1.3
Wetting
1.2
Void ratio, e

1.1
1.0 After compression
0.9
Drying
0.8
0.7
0.1 1 10 100 1000
Matric suction (kPa)

1.5
1.4 I-110 R-110
Wetting
1.3
Initial state
1.2
Void ratio, e

1.1
After compression
1.0
0.9
0.8
Drying
0.7
0.1 1 10 100 1000
Matric suction (kPa)

Figure 6.6 Comparison of stress dependent wetting and drying induced volumetric behaviour
of intact and re-compacted loess

93
CHAPTER 7
CONCLUSIONS AND FURTHER WORK

The WRC and volumetric behaviour along the drying and wetting paths of intact and re-
compacted loess for a wide suction range (zero-290 MPa) are investigated in this study. In
addition, stress effects on WRC and wetting/drying induced volumetric behaviour of intact
and re-compacted loess are explored to understand the hydro-mechanical behaviour of loess.
To control wide range of suction for WRC measurement, three different suction controlling
techniques are used such as axis-translation, osmotic and vapour equilibrium technique.
Stress effects on WRC and wetting/drying volumetric behaviour are investigated by
employing a modified one-dimensional stress controllable volumetric pressure plate
apparatus.

In this chapter, major conclusions are summarized and some recommendations for further
work are given.

7.1 Major conclusions

7.1.1 Water retention properties of intact and re-compacted loess

7.1.1.1 WRC of intact and re-compacted loess for a wide suction range (zero-290 MPa)

Hydraulic hysteresis is consistently higher in suction range of 10-100 kPa compared to

suction range above 1 MPa for both intact and re-compacted loess. The difference between
hysteresis in low and high range of suction can be explained by relative contribution of
capillary and adsorption phenomena. At low suction range, the combined effects of ink-bottle
pore neck mechanism, solid-liquid-air contact-angle mechanism result a marked hydraulic
hysteresis compared to high suction range.

Hysteretic behaviour is different for intact and re-compacted loess for the studied suction
range (zero-290 MPa). A significant hysteresis is observed around the AEV (suction of 10
kPa) of intact loess compared to re-compacted loess. This is because of larger proportion of
constricted pores (macro-pores larger than 100 m) of intact loess that leading to a more
pronounced hysteresis. On the other hand, re-compacted loess shows higher hysteresis in the
suction range between 100 and 400 kPa than intact loess due to different pore size
distribution of two materials.

94
7.1.1.2 WRC of intact and re-compacted loess at different net vertical stresses

Wetting WRC of intact loess is influenced by the increase of net vertical stress while drying
WRC is insensitive to stress effects. The is due to the collapse of the large macro-pores of
intact loess after application of stress that control the water retention characteristics during
wetting processes while stress effects on narrower channels (micro-pores) surrounding large
macro- pores is insignificant that influence drying branch of WRC. In addition, hysteresis is
reduced with increase of net vertical stress due to the modification of pore size distribution.

Drying WRC of re-compacted loess is influenced significantly with the increase of net
vertical stress and effect of net stress is more pronounced on desorption rate compared to
adsorption rate.

7.1.2 Volumetric behaviour of intact and re-compacted loess

7.1.2.1 Wetting and drying induced volumetric behaviour of intact and re-compacted loess
for wide suction range (zero-290 MPa)

Both intact and re-compacted loess considerably shrink and swell due to drying and wetting,
respectively at high range of suction (0.88-290 MPa) in which drying induced shrinkage is
more pronounced than wetting induced swelling. High shrinkability is mainly attributed due
to the high initial porosity or loose structure of loess.

Noticeable shrinkage is observed for both intact and re-compacted loess at high suction range
(0.88-290 MPa). However, re-compacted loess shows 50% higher drying induced shrinkage
(or volumetric strain) than intact loess for the considered suction range. This difference in
volumetric behaviour of intact and re-compacted loess is because the former soil has a higher
yield stress and a more stabilized structure than those of the latter.

7.1.2.2 Collapse and shrinkage of intact and re-compacted loess along wetting and drying
processes under different net vertical stresses

Negligible volume change due to wetting and drying is observed for intact specimen under
zero and 50 kPa net vertical stress for the considered suction range of 0.1-400 kPa. However,
significant wetting-induced collapse and drying-induced shrinkage are observed for the intact
specimen compressed to 110 kPa net stress in which wetting induced collapse is more
pronounced than drying induced shrinkage.

95
It is observed that both wetting-induced collapse and drying-induced shrinkage are stress-
dependent for intact loess. It should be noted that the amount of drying induced shrinkage is
comparable with that of wetting induced collapse. Regarding shrinkage, the suction at
yielding decreases with increasing net stress. As inter-particle arrangement of intact loess
becomes destroyed once applied stress goes beyond yield stress, a lower suction is required to
yield the intact loess upon subsequent drying. The framework of loading-collapse (LC) and
suction-increase (SI) yield curves can be used to interpret the observed volumetric
phenomena of intact loess.

Re-compacted loess deforms progressively along the wetting path under constant net stress
while insignificant volume change is observed during drying process for the considered
suction range. In addition, wetting induced collapse is increased with stress level due to
continuous rearrangement of particles that leading to more volumetric deformation.

7.1.2.3 Comparison of stress dependent wetting and drying induced volumetric behaviour of
intact and re-compacted loess

Wetting induced collapse of intact and re-compacted loess is stress dependent. However, re-
compacted loess shows significantly higher wetting induced collapse compared to intact loess
under constant net stress.

7.2 Recommendations for further research

Experimental results of WRC tests for a wide suction range (zero to 290 MPa) showed that
hysteretic behaviour of intact and re-compacted loess is different especially at low suction
range (zero-0.4 MPa) compared to high suction range (0.88-290 MPa). One possible
explanation is given in this study in terms of different pore size distribution as well as
different macro (constricted pore) and micro-pore structures of intact and re-compacted loess.
This postulation needs to be supported more insight micro-structural investigation through
mercury intrusion porosimetry (MIP) and scanning electron microscopic (SEM) tests. These
tests can be done on intact/ re-compacted loess specimens at different suctions to interpret the
micro-structural effects on water retention properties of loess.

Intact and re-compacted loess showed different shrinkage and swelling behaviour upon
drying and wetting, respectively at high suction range (0.88-290 MPa). In other words, it
seems that effects of suction on wetting/drying induced volumetric behaviour are significant

96
and it also affects the micro-structure of loess. Therefore, it is necessary to investigate the
effects of microstructure on volumetric behaviour of intact and re-compacted loess at high
suctions.

In this study, WRC and wetting/drying induced volumetric behaviour of intact and re-
compacted loess were investigated for a wide suction range (zero to 290 MPa) under zero
stress using different suction controlling techniques such as axis translation, osmotic and
vapour equilibrium technique. It can be mentioned that, both intact and re-compacted loess
showed significant volume changes due to drying and wetting at high suctions. On the other
hand, stress effects on WRC and wetting/drying induced volumetric behaviour of intact and
loess were investigated using modified volumetric pressure plate apparatus for low suction
range (zero-0.4 MPa) in which suction was controlled by axis translation technique. Results
evidenced that stress significantly influences the volumetric behaviour of intact/ re-
compacted loess at low suction range (zero-0.4 MPa). Therefore, it is worth to consider the
stress effects on WRC along with wetting/drying induced volumetric behaviour at high
suction range to obtain complete picture of hydro-mechanical behaviour of intact and re-
compacted loess for wide suction range (zero to 290 MPa).

In order to achieve the abovementioned objective, it is necessary to develop a stress

controllable WRC device with volumetric measurements in which wide range of suction can
be imposed on a single specimen in order to avoid discrepancy among prepared specimens
and also reducing the testing time. This proposal will help to resolve the problem of
volumetric measurements for wide suction range following one method rather than using
different methods of volumetric measurements for different range of suction.

97
REFERENCES

Alonso, E.E., Gens, A., and Josa, A. 1990. A constitutive model for partially saturated soils.
Géotechnique, 40(3): 405-430.
Alonso, E.E., Romero, E., Hoffmann, C., and García-Escudero, E. 2005. Expansive
bentonite-sand mixtures in cyclic controlled-suction drying and wetting. Engineering
Geology, 81(3): 213-226.
Alonso, E.E., Vaunat, J., and Gens, A. 1999. Modelling the mechanical behaviour of
expansive clays. Engineering Geology, 54(1-2): 173-183.
Andersen, B.G., and Borns, H.W. 1997. The Ice Age of World. Scandinavian University
Press, Oslo.
Assallay, A.M., Rogers, C.D.F., and Smalley, I.J. 1997. Formation and collapse of metastable
particle packing and open structures in loess deposits. Engineering Geology, 48(1-2):
101-115.
ASTM. 2002. Standard test methods for determination of the soil water chararcteristic curve
for desorption using a hanging column, pressure extractor, chilled mirror hygrometer,
and/or centrifuge. ASTM standard D6836. American Society for Testing Materials,
West Conshohocken, Pa.
ASTM. 2006. Standard practice for classification of soils for engineering purposes (Unified
Soil Classification System). ASTM standard D2487. American Society for Testing
and Materials, West Conshohocken, Pa.
ASTM. 2007. Standard practice for maintaining relative humidity by means of aqueous
solutions. ASTM standard E104. American Society for Testing and Materials, West
Conshohocken, Pa.
ASTM. 2008. Standard test method for shrinkage factors of soils by the wax method. ASTM
standard D4943. American Society for Testing and Materials, West Conshohocken,
Pa.
ASTM. 2012. Standard test methods for laboratory compaction characteristics of soil using
standard effort. ASTM standard D698. American Society for Testing and Materials,
West Conshohocken, Pa.
Barden, L., McGown, A., and Collins, K. 1973. The collapse mechanism in partly saturated
soil. Engineering Geology, 7(1): 49-60.

98
Bernier, F., Volckaert, G., Alonso, E.E, and Villar, M.V. 1997. Suction-controlled
experiments on Boom clay. Engineering Geology, 47(4): 325-338.
Blatz, J., Cui, Y.J., and Oldecop, L. 2008. Vapour equilibrium and osmotic technique for
suction control. Geotechnical and Geological Engineering, 26(6): 661-673.
Bocking, K.A., and Fredlund, D.G. 1980. Limitations of the axis translation technique. In
Proceedings of the 4th International Conference on Expansive Soils, Denver,
Colo., 16-18 June 1980. American Society of Civil Engineering, New York.
pp. 117-135.
BSI. 1990. BS1377: methods of test for soils for civil engineering purposes. British Standards
Institution (BSI), London, UK.
Chen, Z., Fredlund, D.G., and Gan, J.K.-M. 1999. Overall volume change, water volume
change, and yield associated with an unsaturated compacted loess. Canadian
Geotechnical Journal, 36(2): 321-329.
Chiu, C.F., and Ng, C.W.W. 2012. Coupled water retention and shrinkage properties of a
compacted silt under isotropic and deviatoric stress paths. Canadian Geotechnical
Journal, 49(8): 928-938.
Clemence, S.P., and Finbarr, A.O. 1981. Design Considerations for collapsible soils. Journal
of the Geotechnical Engineering Division, 107(3): 305-317.
Cui, Y.J., Yahia-Aissa, M., and Delage, P. 2002. A model for the volume change behavior of
heavily compacted swelling clays. Engineering Geology, 64: 233-250.
Cui, Y.J., and Delage, P. 1996. Yielding and plastic behaviour of an unsaturated compacted
silt. Géotechnique, 46(2): 291-311.
Cui, Y.J., Marcial, M., Terpereau, J.M., Delage, P., Antoine, P., Marchadier, G., and Ye, W.
M. 2004. A geological and geotechnical characterisation of the loess of Northern
France. In Proceedings of the A.W. Skempton Memorial Conference on Advances in
Geotechnical Engineering, London, vol. 1, pp. 417-428.
Delage, P., and Cui, Y.J. 2000. L’eau dans les sols non saturés. Éditions Techniques de
l’ingénieur, Paris. Vol. C2, article C 301. Traité Construction.
Delage, P., and Cui, Y.J. 2008. An evaluation of the osmotic method of controlling suction.
Geomechanics and Geoengineering: An International Journal, 3(1): 1-11.
Delage, P., Cui, Y.J., and Antoine, P. 2005. Geotechnical problems related with loess
deposits in northern France. In Proceedings of the International Conference on
Problematic Soils, Famagusta, North Cyprus. pp. 517-540.

99
Delage, P., Howat, M.D., and Cui, Y.J. 1998. The relationship between suction and swelling
properties in a heavily compacted unsaturated clay. Engineering Geology, 50(1-2):
31-48.
Delage, P., Romero, E., and Tarantino, A. 2008. Recent developments in the techniques of
controlling and measuring suction. In Proceedings of the 1st European Conference on
Unsaturated soils: Advanced in Geo-Engineering, Durham, UK, 2-4 July 2008. Edited
by D.G. Toll, C.E. Augarde, D. Gallipoli, and S.J. Wheeler. Taylor & Francis Group,
London, UK. pp. 33-52.
Dif, A.E., and Bluemel, W.F. 1991. Expansive soils under cyclic drying and wetting.
Geotechnical Testing Journal, 14(1): 96-102.
Dijkstra, T.A., Smalley, I.J., and Rogers, C.D.F. 1995. Particle packing in loess deposit and
the problem of structure collapse and hydroconsolidation. Engineering Geology, 40
(1-2): 49-64.
Dudley, J.H. 1970. Review of collapsing soils. Journal of Soil Mechanics and Foundation
Division, 96: 925-947.
Egri, G. 1972. The physico-chemical properties and engineering problems of the loess soils.
Acta Gelo. Acad, Sci. Hung., 16: 337-345.
Esteban, V., and Saez, J. 1988. A device to measure the swelling characteristics of rock
samples with control of the suction up to very high values. In ISRM Symposium on
Rock Mechanics and Power Plants, Madrid, Spain, pp. 2.
Farulla, C.A., Ferrari, A., and Romero, E. 2010. Volume change behaviour of a compacted
scaly clay during cyclic suction changes. Canadian Geotechnical Journal, 47(6): 688-
703.
Fredlund, D.G., Rahardjo, H., and Fredland, M.D. 2012. Unsaturated soil mechanics in
engineering practice. John Wiley and Sons, Inc., Hoboken, New Jersey.
Gao, G. 1996. The distribution and geotechnical properties of loess soils, lateritic soils and
clayey soils in china. Engineering Geology, 42(1): 95-104.
Garakani, A.A., Haeri, S.M., Khosravi, A., and Habibagahi, G. 2015. Hydro-mechanical
behavior of undisturbed collapsible loessial soils under different stress state
conditions. Engineering Geology, 195, 28-41.
Haeri, S.M., Garakani, A.A., Khosravi, A., and Meehan, C.L. 2014. Assessing the hydro
mechanical behavior of collapsible soils using a modified triaxial test device.
Geotechnical Testing Journal, 37(2): 1-15.

100
Heller, F., and Evans, M.E., 1995. Loess magnetism. Reviews of Geophysics, 33(2): 211-
240.
Hilf, J.W. 1956. An investigation of pore-water pressure in compacted cohesive soils. Ph.D.
dissertation, US Dept. of the Interior, Bureau of Reclamation, Design and
Construction Division, Denver, Colo. Technical Memorandum No. 654.
Hillel, D. 1998. Environmental soil physics. Academic Press, San Diego, California.
Huang, M., Fredlund, D.G., and Fredlund, M.D. 2010. Comparison of measured and PTF
predictions of SWCCs for loess soils in china. Geotechnical and Geological
Engineering, 28(2): 105-117.
Jemings, J.E., and Knight, K. 1975. A guide to construction on or with materials exhibiting
additional settlement due to collapse of grain structure. In Proceedings of the Sixth
Regional Conference for Africa on Soils Mechanics and Foundation Engineering,
Druban, South Africa, 99-105.
Jiang, M., Hu, H., and Liu, F. 2012. Summary of collapsible behaviour of artificially
structured loess in oedometer and triaxial wetting tests. Canadian Geotechnical
Journal, 49(10): 1147-1157.
Jiang, M., Zhang, F., Hu, H., Cui, Y., and Peng, J. 2014. Structural characterization of natural
loess and remolded loess under triaxial tests. Engineering Geology, 181: 249-260.
Jin, C., and Liu, Q. 2011. Remagnetization mechanism and a new age model for L9 in
Chinese loess. Physics of the Earth and Planetary Interiors, 187(3-4): 261-275.
Kassif, G., and Ben Shalom, A. 1971. Experimental relationship between swell pressure and
suction. Géotechnique 21: 245-255.
Klukanova, A., and Frankovaska, J. 1994. The Slovak Carpathians loess sediments, their
fabric and properties. In Proceedings of the NATO Advanced Research Workshop on
Genesis and Properties of Collapsible Soils, Loughborough, UK, 413.
Leung, A. K. 2011. Field and laboratory studies of a saprolitic hillslope subjected to seasonal
variations. Ph. D thesis, The Hong Kong University of Science and Technology, HK.
Lu, N., and Khorshidi, M. 2015. Mechanisms for soil-water retention and hysteresis at high
suction range. Journal of Geotechnical and Geoenvironmental Engineering, 141(8):
04015032.
Mitchell, J.K., and Soga, K. 2005. Fundamentals of soil behaviour. John Wiley & Sons,
Hoboken, N.J.
Mualem, Y. 1977. Extension of the similarity hypothesis used for modelling the soil water
characteristics. Water Resources Research, 13(4): 773-780.
101
Mualem, Y. 1984. Prediction of the soil boundary wetting curve. Soil Science, 137(6): 379-
390.
Munõz-Castelblanco, J., Delage, P., Pereira, J.M., and Cui, Y.J. 2011. Some aspects of the
compression and collapse behaviour of an unsaturated natural loess. Géotechnique
Letters, 1(2): 17-22.
Muñoz-Castelblanco, J.A., Pereira, J.M., Delage, P., and Cui, Y.J. 2012. The water retention
properties of a natural unsaturated loess from northern France. Géotechnique, 62(2):
95-106.
Ng, C.W.W., and Leung, A.K. 2012. Measurements of drying and wetting permeability
functions using a new stress-controllable soil column. Journal of Geotechnical and
Geoenvironmental Engineering, 138(1): 58-68.
Ng, C.W.W., and Menzies, B. 2007. Advanced unsaturated soil mechanics and engineering.
Taylor & Francis, London and New York.
Ng, C.W.W., and Pang, Y.W. 2000. Experimental investigations of the soil-water
characteristics of a volcanic soil. Canadian Geotechnical Journal, 37(6): 1252-1264.
Ng, C.W.W., Sadeghi, H., Hossen, S.B., Chiu, C.F., Alonso, E.E., and Baghbanrezvan, S.
2016. Water retention and volumetric characteristics of intact and re-compacted loess.
Canadian Geotechnical Journal. [Accepted with revision]
Oldecop, L.A., and Alonso E.E. 2004. Testing rockfill under relative humidity control.
Geotechnical Testing Journal, ASCE, 27(3): 1-10.
Osipov, V.I., and Sokolov, V.N. 1994. Factors and mechanism of loess collapsibility. In
Proceedings of the NATO Advanced Research Workshop on Genesis and Properties
of Collapsible Soils, Loughborough, UK. pp. 49-63.
Pécsi, M. 1990. Loess is not just the accumulation of dust. Quaternary International 7/8: 1-
21.
Peng, J., Fan, Z., Wu, D., Zhuang, J., Dai, F., Chen, W., and Zhao, C. 2015. Heavy rainfall
triggered loess-mudstone landslide and subsequent debris flow in tianshui, china.
Engineering Geology, 186: 79-90.
Pereira, J.H. F., and Fredlund, D.G. 2000. Volume change behavior of collapsible compacted
gneiss soil. Journal of Geotechnical and Geoenvironmental Engineering, 126(10):
907-916.
Rogers, C.D.F. 1995. Types and distribution of collapsible soils. In Proceedings of the NATO
Advanced Research Workshop on Genesis and Properties of Collapsible Soils,
Loughborough, UK. pp. 1-17.
102
Rogers, C.D.F., Dijkstra, T.A., and Smalley, I. J. 1994. Hydroconsolidation and subsidence
of loess: Studies from China, Russia, North America and Europe in memory of Jan
Sajgalik. Engineering Geology, 37(2): 83-113.
Romero, E. 1999. Characterization and thermo-hydro-mechanical behaviour of unsaturated
Boom clay: an experimental study. Ph.D. thesis, Universitat Politècnica de Catalunya,
Barcelona, Spain.
Romero, E., Della Vecchia, G., and Jommi, C. 2011. An insight into the water retention
properties of compacted clayey soils. Géotechnique, 61(4): 313-328.
Roohnavaz, C., Russell, E.J.F., and Taylor, H.F. 2011. Unsaturated loessial soils: A
sustainable solution for earthworks. In Proceedings of the Institution of Civil
Engineers: Geotechnical Engineering, 164(4): 257-276.
Sheng, D., Fredlund, D.G., and Gens, A. 2008. A new modelling approach for unsaturated
soils using independent stress variables. Canadian Geotechnical Journal, 45(4): 511-
534.
Sultan, H.A. 1969. Collapsing soils. In Proceedings of the Seventh International Conference
on Soil Mechanics and Foundation Engineering, Seciality Session No. 5, Mexico
City.
Tan, T.K. 1988. Fundamental properties of loess from Northwestern China. Engineering
Geology, 25, 103-122.
Tavakoli Dastjerdi, M.H., Habibagahi, G., and Nikooee, E. 2014. Effect of confining stress
on soil water retention curve and its impact on the shear strength of unsaturated soils.
Vadose Zone Journal, 13(5).
Tripathy, S., and Rees, S.W. 2013. Suction of some polyethylene glycols commonly used for
unsaturated soil testing. Geotechnical Testing Journal, 36(5): 1-13.
Vanapalli, S.K., Nicotera, M.V., and Sharma, R.S. 2008. Axis translation and negative water
column techniques for suction control. Geotechnical and Geological Engineering,
26(6): 645-660.
Villar, M.V. 2000. Caracterización termo-hidro-mecánica de una bentonita de Cabo de Gata.
Ph.D. thesis, Universidad Complutense de Madrid, Madrid, Spain.
Wen, B.P., and Yan, Y.J. 2014. Influence of structure on shear characteristics of the
unsaturated loess in lanzhou, china. Engineering Geology, 168: 46-58.
Xu, L., Dai, F.C., Tham, L.G., Tu, X.B., Min, H., Zhou, Y.F., Wu, C.X., and Xu, K. 2011.
Field testing of irrigation effects on the stability of a cliff edge in loess, north-west
china. Engineering Geology, 120(1-4): 10-17.
103
Yuan, Z.X., and Wang, L.M. 2009. Collapsibility and seismic settlement of loess.
Engineering Geology, 105(1-2): 119-123.
Zhan, T.L.T., Yang, Y.B., Chen, R., Ng, C.W.W., and Chen, Y.M. 2014. Influence of clod
size and water content on gas permeability of a compacted loess. Canadian
Geotechnical Journal, 51(12): 1468-1474.
Zhou, Y.F., Tham, L.G., Yan, W.M., Dai, F.C., and Xu, L. 2014. Laboratory study on soil
behavior in loess slope subjected to infiltration. Engineering Geology, 183: 31-38.
Zur, B. 1966. Osmotic control of the matric soil water potential: I. Soil water system. Soil
Science, 102(6): 394-398.

104