Thermo-Hydro-Mechanical Behaviour of Composite Geosynthetic Lining Systems Under High Temperature and Low Pressure

Faculty of Engineering and IT
School of Civil Engineering
Thermo-Hydro-Mechanical Behaviour of Composite Geosynthetic

Lining Systems under High Temperature and Low Pressure
by Ali Ghavam-Nasiri
A thesis submitted in fulfilment of the requirements for the degree of Doctor of Philosophy
January 2018
Abstract
Aim. Composite geosynthetic clay lining systems (CGCLS), a combination of high density polyethylene
geomembranes and geosynthetic clay liners (GCLs), have been used in brine ponds to manage the large
amount of waste water produced by coal seam gas extraction. Exposure to sunlight combined with the
thermal properties of brine water can lead to temperatures of up to 85-95°C at the liner. Recent studies
indicate that the risk of desiccation of the GCL component of a CGCLS is higher under brine pond
conditions, compared to landfill conditions. The goal of the thesis is to investigate whether GCLs may
desiccate in engineering applications with high temperatures and low overburden loads, such as those
encountered on brine ponds.
Methodology. A multi-phase, thermo-hydro-elastic theory of soil behaviour is first adopted to the study of
desiccation of GCLs. The theory has been implemented in a computer simulation environment
(CODE_BRIGHT). Next, the dependence of the GCL’s soil-water chararcteristic curve (SWCC) on
temperature and overburden load are characterised experimentally and new forms of the SWCC equations
are developed and implemented in CODE_BRIGHT. A set of column studies are conducted, replicating
hydraulic, mechanical and thermal conditions experienced by CGCLS in brine pond applications on coal-
seam gas sites. The CGCLS is placed on a sandy subsoil. The aims of the column studies are to assess
whether desiccation of bentonite occurs and to generate experimental data that can be used to validate
the above-mentioned theory. Once validation is completed, the simulation tool is used to explore the risk
of desiccation over a range of applied temperatures and overburden loads and to investigate the ability of
new liner configurations to reduce or remove that risk.
Findings. The new developed SWCCs perform well in predicting the effects of void ratio on SWCC based on
the available data. It is found that air-entry value increases as the net vertical stress increases for the
experiments under the same temperature. However, the proposed SWCCs underestimate the effects of
elevation of temperature, especially for the wet side of the curves. A non-linear thermo-hydro-elastic
theory is found to be capable of replicating the observed change in time of temperature and water
contents in the subsoil with reasonable accuracy, even with minimal back-fitting of data and independent
determination of material properties of both GCLs and subsoil. In addition, the theory qualitatively predicts
the desiccation of bentonite. On the other hand, the predictions of deformation of GCL during the
hydration and dehydration stages are poor. The limitations of the theory in predicting deformation of GCL
are most likely due to assumptions of elasticity, as well as difficulties the adopted theory has in
-i-
approximating void ratio-suction relationships that exhibit high levels of non-linearity and hysteresis. The
thesis has established experimentally, for the first time, that GCLs in composite liners, subject to surface
temperatures of 78°C and overburden loads of 20kPa, may experience high levels of desiccation cracking.
This observation has been confirmed numerically. On the other hand, permeability tests with distilled
water conducted on saturated specimens have shown no decline in hydraulic conductivity in desiccated
GCLs relative to virgin ones. The findings of this thesis call for caution in using the type of GCL studied here
in applications in which they are exposed to temperatures higher than 60°C. Increasing the overburden
load does not seem to be an effective way of removing the risk of desiccation, at least within a range of 20
to 50kPa. Designs that reduce the temperature on top of the GCL (such as adding one or more geonet),
are found to be more promising.
Keywords: void ratio and temperature-dependent SWCC, GCL, composite lining systems, desiccation,
column studies, brine pond, nonlinear thermo-hydro-elastic.
-ii-
Acknowledgements
This thesis was made possible by contributions, advices, and technical supports from my supervisor,
Professor Abbas El-Zein. I am sincerely thankful to him for his outstanding supervision through entire my
candidature. I am also grateful to Professor David Airey for his exceptional supervision and guidance
through the experimental works.
I deeply appreciate assistance that I have received from a large number of scholars, colleagues, and
technical staff in this project. I am grateful to Professors David Toll, Antonio Gens, Hywel Thomas, Enrique
Romero, Kerry Rowe, and Malek Bouazza, Dr Alfonso Rodríguez Dono, Dr Majid Sedighi, Dr. Benjy Marks,
and Dr. Francois Guillard, Mr Garry Towell, Mr Paul Geddes and Mr Sergio de Carvalho for their generous
advices and help.
This project was funded by Australian Research Council (ARC) Discovery grant DP130100203 and the
research was undertaken on the NCI National Facility in Canberra, Australia, which is supported by the
Australian Commonwealth Government. I specially thank Dr. Rika Kobayashi from Australian National
University Supercomputer Facility and Dr. Joachim Mai, HPC Support Specialist at Intersect.
Finally, I would like to thank my family for their endless support that allowed me to pursue the challenging
work of my interest.
-iii-
List of Publications
In addition to the manuscript that was written for this thesis, one journal and five conference papers were
published.
Journal Papers:
Ghavam-Nasiri, A., El-Zein, A. (2016). Effects of defects in geomembranes on reducing desiccation potential
of geosynthetics clay liners. Japanese Geotechnical Society Special Publication, 2(70), pp.2418–2422.
Conference Papers:
Ghavam-Nasiri, A., El-Zein, A., Airey, D., Rowe, R. K., (2018). Soil-water characteristics curve of geosynthetic
clay liners: Effects of temperature and void ratio. 11th International Conference on Geosynthetics
(11ICG), Seoul, Korea.
Ghavam-Nasiri, A., El-Zein, A., Airey, D., Rowe, R. K., (2017). Hydration and desiccation of geosynthetic clay
liners in composite lining systems under brine pond conditions: A laboratory investigation. In
Proceedings of the 2nd Symposium on Coupled Phenomena in Environmental Geotechnics (CPEG2),
Leeds, UK.
Ghavam-Nasiri, A., El-Zein, A. (2017). Desiccation of GCLs under high temperature and low pressure: An
experimentally validated model. In Proceedings of the 2nd Symposium on Coupled Phenomena in
Environmental Geotechnics (CPEG2), Leeds, UK.
El Zein, A., Ghavam-Nasiri, A., Rowe, R. (2015). Brine ponds on coal-seam gas extraction sites: are
geosynthetic clay liners at risk of desiccation? SARDINA 2015 15th International Waste Management
and Landfill Symposium, Padova, Italy: CISA, Environmental Sanitary Engineering Centre.
El-Zein, A., Ghavam-Nasiri, A., Bouazza, A., and Rowe, R. K. (2014). Performance of GCLs in Brine Ponds for
Coal-Seam Gas Extraction Sites: An Investigation. In 7th International Congress on Environmental
Geotechnics, (7ICEG2014). Melbourne, Australia, pp. 1209-1216.
-iv-
Table of Contents
1 Background and Aims ............................................................................................................................ 1
1.1 General .......................................................................................................................................... 1
1.2 Objectives ...................................................................................................................................... 3
1.3 Structure of the Thesis .................................................................................................................. 4
2 Literature Review .................................................................................................................................. 6
2.1 Introduction................................................................................................................................... 6
2.2 Coal Seam Gas and Brine Ponds .................................................................................................... 6
2.3 Geomembranes and Geosynthetic Clay Liners ............................................................................. 9
2.4 Hydration of Geosynthetic Clay Liners ........................................................................................ 11
2.5 Desiccation of Geosynthetic Clay Liners...................................................................................... 16
2.6 Shear Strength of GCL ................................................................................................................. 20
2.7 Soil-Water Characteristics Curves ............................................................................................... 21
2.8 Volume Change and Temperature Effects in Water Retention of GCLs ..................................... 26
2.9 Defects in Geomembrane ........................................................................................................... 28
2.10 Summary and Gap Identification................................................................................................. 29
3 Research Design .................................................................................................................................. 30
3.1 Experimental Determination of Soil-Water Characteristic Curves of GCL and Subsoil Material
Properties ................................................................................................................................................ 31
3.1.1 Aims ..................................................................................................................................... 31
3.1.2 Step 1 ................................................................................................................................... 31
3.1.3 Step 2 ................................................................................................................................... 33
3.2 Column Studies ............................................................................................................................ 33
3.3 Development and Validation of a Theoretical Thermo-Hydro-Mechanical Model..................... 34
3.4 Engineering Applications ............................................................................................................. 34
3.5 Summary...................................................................................................................................... 35
-v-
4 A Non-Linear Thermo-Hydro-Elastic Theory for Desiccation of GCL ................................................... 36
4.1 Introduction ................................................................................................................................. 36
4.2 Governing Equations ................................................................................................................... 37
4.3 Soil-Water Characteristic Curves ................................................................................................. 40
4.3.1 Approach based on van Genuchten (1980) (VG version) .................................................... 41
4.3.2 Selection of 𝑒𝑟𝑒𝑓 ................................................................................................................. 45
4.3.3 Extension to Fredlund and Xing (1994) (FX version) ........................................................... 45
4.4 Intrinsic Permeability for Liquid Phase ........................................................................................ 46
4.5 Relative Hydraulic Conductivity for Liquid and Gas .................................................................... 47
4.6 Flux of Vapour ............................................................................................................................. 47
4.7 Flux of Heat.................................................................................................................................. 48
4.8 Non-linear Elasticity .................................................................................................................... 48
4.9 Numerical Implementation ......................................................................................................... 49
4.10 Summary...................................................................................................................................... 51
5 Experimental Determination of Soil-Water Characteristic Curves of GCL .......................................... 52
5.1 Introduction ................................................................................................................................. 52
5.2 Material ....................................................................................................................................... 52
5.3 Experimental Program................................................................................................................. 53
5.4 Sample Preparation ..................................................................................................................... 54
5.5 Axis-translation: Pneumatic Suction-controlled Oedometer (Barcelona Cell) ............................ 56
5.6 Dew Point Method: WP4C and VSA ............................................................................................ 58
5.7 Mass-volume Relations for GCL .................................................................................................. 60
5.8 Results ......................................................................................................................................... 62
5.8.1 Axis-translation .................................................................................................................... 62
5.8.2 Dewpoint method ............................................................................................................... 70
5.8.3 Vapour equilibrium technique (VET) experiments (Monash University) ............................ 77
5.9 Summary of SWCC Experiments .................................................................................................. 79
-vi-
5.10 Effect of Void Ratio and Temperature on SWCC ......................................................................... 82
5.10.1 Approach based on van Genuchten (1980) (VG version) .................................................... 82
5.10.2 Approach based on Fredlund and Xing (1994) (FX version) ................................................ 86
5.11 Comparison to Existing GCL SWCCs............................................................................................. 89
5.12 Application to Existing GCL SWCC Measurements ...................................................................... 91
5.13 Summary and Limitations of Predictions .................................................................................... 95
6 Column Studies of Desiccation of GCL ................................................................................................ 97
6.1 Introduction ................................................................................................................................. 97
6.2 Materials...................................................................................................................................... 97
6.3 Experimental Program................................................................................................................. 98
6.4 Testing Apparatus........................................................................................................................ 99
6.5 Selection of Subsoil ................................................................................................................... 102
6.6 Permeability of GCL and Subsoil................................................................................................ 103
6.7 SWCC and Unsaturated Hydraulic Conductivity of Sand ........................................................... 106
6.8 Thermal Conductivity of GMB, GCL, and Subsoil ...................................................................... 108
6.9 Mechanical Properties of GCL ................................................................................................... 109
6.10 Results ....................................................................................................................................... 109
6.10.1 Permeability ...................................................................................................................... 109
6.10.2 HYPROP.............................................................................................................................. 111
6.10.3 Thermal conductivity ......................................................................................................... 113
6.10.4 Mechanical Properties of GCL ........................................................................................... 115
6.10.5 Summary of material properties ....................................................................................... 117
6.10.6 Column studies .................................................................................................................. 118
6.11 Summary.................................................................................................................................... 124
7 Validation of the Proposed Non-Linear Thermo-Hydro-Elastic Theory of GCL Desiccation ............. 125
7.1 Introduction ............................................................................................................................... 125
7.2 Experimental Programs ............................................................................................................. 126
-vii-
7.3 Material Properties ................................................................................................................... 128
7.3.1 Mechanical properties: 𝜅 and 𝜅𝑠 ...................................................................................... 128
7.3.2 Soil-water characteristic curves ........................................................................................ 132
7.3.3 Intrinsic permeability......................................................................................................... 133
7.3.4 Relative hydraulic conductivity ......................................................................................... 134
7.3.5 Diffusive flux of vapour ..................................................................................................... 134
7.3.6 Conductive flux of heat ..................................................................................................... 134
7.4 Initial and Boundary Conditions ................................................................................................ 134
7.5 Results ....................................................................................................................................... 135
7.5.1 Validation for model #1THM ............................................................................................. 135
7.5.2 Validation for models #2THM and #3THM (Southen and Rowe, 2005a) .......................... 143
7.5.3 Validation for model #4TH (Bouazza et al, 2014) .............................................................. 146
7.6 Summary.................................................................................................................................... 147
8 Engineering Applications ................................................................................................................... 148
8.1 Introduction ............................................................................................................................... 148
8.2 Numerical Modelling ................................................................................................................. 148
8.3 Material Properties ................................................................................................................... 151
8.4 Initial and Boundary Conditions ................................................................................................ 152
8.5 Sensitivity Analyses ................................................................................................................... 153
8.6 Results ....................................................................................................................................... 154
8.6.1 Effect of temperature: Case #1 ......................................................................................... 154
8.6.2 Effect of overburden stress: Case #2 ................................................................................. 157
8.6.3 Performance of Designs D1 and D2 ................................................................................... 158
8.6.4 Sensitivity analyses ............................................................................................................ 161
8.7 Summary.................................................................................................................................... 162
9 Conclusions and Recommendations for Further Studies .................................................................. 164
9.1 Summary of Research................................................................................................................ 164
-viii-
9.2 Main Research Findings and Significance.................................................................................. 165
9.2.1 SWCC Dependence on Void Ratio and Temperature ........................................................ 165
9.2.2 Non-Linear Thermo-Hydro-Elasticity under High Temperature Low Overburden Load ... 165
9.2.3 Desiccation of GCLs in Brine Ponds ................................................................................... 166
9.3 Research Limitations ................................................................................................................. 166
9.3.1 Theoretical Limitations ...................................................................................................... 166
9.3.2 Experimental Limitations................................................................................................... 167
9.4 Research Suggestions and Future Works .................................................................................. 168
References ................................................................................................................................................. 171
Appendix A: Derivatives of the Proposed SWCCs ..................................................................................... 183
Appendix B: Unsaturated Mass-Volume Relations for GCL ...................................................................... 186
Appendix C: Sydney Water Quality ........................................................................................................... 188
Appendix D: Calibration of TDRs and Thermocouples .............................................................................. 189
Appendix E: Thermo-Hydro-Mechanical Interactions ............................................................................... 191
-ix-
Table of Figures
Figure 2-1. Schematic of coal seam gas extraction (credit: Australian Science Media Centre, accessed on
July 2016) http://theconversation.com/three-myths-the-coal-seam-gas-industry-wants-you-to-believe-
24422 ............................................................................................................................................................. 7
Figure 2-2. A typical brine pond on a coal seam gas site (www.aqob.com.au, accessed on May 2014) ...... 8
Figure 2-3. Swelling pressure of two commercial bentonites with distilled water: (left) MX80 Na-bentonite;
(right) Calcigel Ca-bentonite (after Schanz et al., 2013) ............................................................................. 14
Figure 2-4. Swelling pressure of MX80: (left) distilled water; (right) different NaCl concentrations (after
Wang and Liu, 2012) .................................................................................................................................... 14
Figure 2-5. Effect of polymer treatment on swelling pressure of GCL: (left) without treatment; (right) with
treatment (after Janssen et al., 2015) ......................................................................................................... 14
Figure 2-6. Shear strength of GCL (Fox and Stark, 2015): a) Typical shear stress-displacement behaviour of
hydrated GCLs; b) Peak and residual shear strength envelopes for a hydrated W/NW NP GCL ................ 20
Figure 2-7. Errors in total suction measurements (adopted from Agus and Schanz, 2007): a) Device error;
b) Estimate of error because of temperature fluctuations ......................................................................... 22
Figure 2-8. Evolution of SWCC with deformation of soil: a) Experiments under wetting and drying paths
(Tarantino, 2009); b) Idealized model (Gallipoli, 2012) .............................................................................. 26
Figure 2-9. Effect of temperature and density on SWCC of Boom clay (Romero et al., 2001) ................... 27
Figure 3-1. Research design......................................................................................................................... 31
Figure 3-2. Overview of suction range covered in this thesis including all methods and conditions. ........ 32
Figure 4-1. Schematic of unsaturated porous media composed of solid, liquid, and gas phases containing
(modified from Olivella et al, 1996) ............................................................................................................ 38
Figure 4-2. Surface tension vs. temperature for pure water ...................................................................... 43
Figure 4-3. Correlation between 𝑃/𝑃0 and void ratio, e. ........................................................................... 44
Figure 4-4. Correlation between 𝜅𝑙/𝜅0 and porosity based on the Kozeny’s model ................................. 47
Figure 5-1. Hydration setup......................................................................................................................... 55
Figure 5-2. Barcelona Cell (modified from a drawing by Tomás Pérez Revilla, UPC) .................................. 56
-x-
Figure 5-3. Water baths for SWCC tests under 40°C ................................................................................... 57
Figure 5-4. Schematic of dew point method devices: a) WP4C (Fredlund et al., 2012); b) VSA (VSA manual)
..................................................................................................................................................................... 59
Figure 5-5. Effect of load on SWCC at 20°C: Degree of saturation vs suction; b) Void ratio vs suction...... 65
Figure 5-8. Effect of temperature on SWCC under net axial stress of 20kPa: Degree of saturation vs suction;
b) Gravimetric water content vs suction ..................................................................................................... 66
Figure 5-9. Effect of temperature on SWCC under net axial stress of 50kPa: Degree of saturation vs suction;
b) Gravimetric water content vs suction ..................................................................................................... 66
Figure 5-10. Summary of all axis-translation tests: a) Degree of saturation; b) Gravimetric water content;
c) Volumetric water content; d) Void ratio ................................................................................................ 67
Figure 5-11. GCL height measurement setup: the setup is a modification of ISO 9863-1 (2016)............... 70
Figure 5-12. Comparison of VSA and WP4C: specimen #2D37 at 20°C: a) degree of saturation vs. suction;
b) gravimetric water content vs. suction; c) volumetric water content vs. suction; d) void ratio vs. suction
..................................................................................................................................................................... 71
Figure 5-13. Comparison of VSA and WP4C: specimen #15D37 at 35°C: a) degree of saturation vs. suction;
b) gravimetric water content vs. suction; c) volumetric water content vs. suction; d) void ratio vs. suction
..................................................................................................................................................................... 72
Figure 5-14. VSA measurements for drying and wetting at 60°C on specimen #16D37: a) degree of
saturation vs. suction; b) gravimetric water content vs. suction; c) volumetric water content vs. suction; d)
void ratio vs. suction ................................................................................................................................... 73
Figure 5-15. Summary of dew point method experiments on the drying path: a) Degree of saturation vs.
suction; b) void ratio vs. suction.................................................................................................................. 74
Figure 5-16. Summary of dew point method experiments on the wetting path: a) Degree of saturation vs.
suction; b) Gravimetric water content vs. suction. ..................................................................................... 74
Figure 5-17. Void ratio vs suction for all. w: wetting path; d: drying path.................................................. 74
Figure 5-18. VSA measurements: a) Gravimetric water content vs. time; b) Total suction vs. time .......... 76
Figure 5-19. SWCC of GCL X2000 using VET performed at Monash University: a) Gravimetric water content
vs. relative humidity under 20°C (modified from Rouf et al. 2016); b) Gravimetric water content vs. total
-xi-
suction. Total suction are calculated using Kelvin equation (Eq. 5-2); c) Degree of saturation vs. total
suction; d) Volumetric water content vs. total suction ............................................................................... 78
Figure 5-20. Summary of all SWCC experiments on X2000: a) Gravimetric water content vs suction; b)
Degree of saturation vs. suction. ................................................................................................................ 81
Figure 5-21. Void ratio vs suction for all experiments on X2000 ................................................................ 82
Figure 5-22. van Genuchten (1980) SWCC fit for reference curve on the drying path ............................... 83
Figure 5-23. Determination of 𝑒𝑟𝑒𝑓 ............................................................................................................ 84
Figure 5-24. Calibration of 𝛽𝑒 and 𝜆 for all experiments under 20°C (𝑅2 = 0.9982) ................................ 85
Figure 5-25. Predictions based on VG version for experiments under elevated temperature conditions
(𝑅2 = 0.9227) ............................................................................................................................................. 86
Figure 5-26. Fredlund and Xing (1994) SWCC fit for reference curve on the drying path .......................... 87
Figure 5-27. Calibration of 𝛽𝑒, 𝑛𝑓, and 𝑚𝑓 for all experiments under 20°C (𝑅2 = 0.9978) ..................... 87
Figure 5-28. Predictions based on FX version for experiments under elevated temperature conductions
(𝑅2 = 0.9346) ............................................................................................................................................. 88
Figure 5-29. SWCC tests on X2000 and comparison with SWCC of GCL from literature. a) Degree of
saturation vs. suction; b) Void ratio vs. suction. SR07: Southen and Rowe (2007); AB10: Abuel-Naga and
Bouazza (2010) ............................................................................................................................................ 90
Figure 5-30. Determination of the VG reference curves for G1 and G2 (data digitized from Southen and
Rowe, 2007)................................................................................................................................................. 92
Figure 5-31. Void ratio at air entry for G1 and G2 (data digitized from Southen and Rowe, 2007) ........... 93
Figure 5-32. VG version of the SWCC for G1 (R2=0.8624) and G2 (R2=0.8009) ........................................... 93
Figure 5-33. Determination of the FX reference curves for G1 and G2 (data digitized from Southen and
Rowe, 2007)................................................................................................................................................. 94
Figure 5-34. FX version of the SWCC for G1 (R2=0.8565) and G2 (R2=0.8039) ............................................ 94
Figure 5-35. X-ray image of GCL: a) specimen #42D50 at nearly fully saturation; b) specimen #12D50 at
matric suction of 170kPa; c) specimen #11D50 at matric suction of 350kPa ............................................. 96
Figure 6-1. Particle size distribution and compaction curves for SW and SP .............................................. 98
Figure 6-2. Schematics of the soil column test apparatus: a) Front view; b) Side view; c) Cross section. 100
Figure 6-3. TDRs and data logger: a) MP306; b) CS655; c) CR800. Figures are not drawn to scale. ......... 101
-xii-
Figure 6-4. Hydration of GCL from S1 and S2 (𝑤0 is initial gravimetric water content of subsoil)........... 102
Figure 6-5. Resulting void ratio vs gravimetric water content vs degree of saturation: results from the
compaction tests on SW using a standard effort ...................................................................................... 102
Figure 6-6. HYPROP ................................................................................................................................... 107
Figure 6-7. Thermal conductivity setups: a) Hot Disk AB device; b) thermocell (from Ali et al., 2016) .... 108
Figure 6-8. Reference compression curves for saturated soils (Fredlund et al., 2012): a) Specific volume vs.
natural logarithm of effective isotropic stress; b) Void ratio vs. base-10 logarithm of effective vertical stress
................................................................................................................................................................... 109
Figure 6-9. Saturated hydraulic conductivity of an intact GCL using a flexible-wall permeameter: a) 𝑘𝑠 vs
time; b) ratio of outflow to inflow vs time. Specimens were saturated with tap water TDS=100 – 136 mg/l
................................................................................................................................................................... 110
Figure 6-10. Saturated hydraulic conductivity of a desiccated GCL using a flexible-wall permeameter: a) 𝑘𝑠
vs time; b) ratio of outflow to inflow vs time ............................................................................................ 110
Figure 6-11. Effect of void ratio on saturated hydraulic conductivity of subsoil (SW) using a rigid-wall
permeameter. ........................................................................................................................................... 111
Figure 6-12. Hyprop measurements with time for SW: a) Suction vs time at z1 and z2 ; b) Gravimetric water
content vs. time......................................................................................................................................... 111
Figure 6-13. SWCC for SW. VG: van Genuchten (1980). FX: Fredlund and Xing (1994) ............................ 112
Figure 6-14. Relative hydraulic conductivity for SW. VG: van Genuchten. FX+VG: effective degree of
saturation based on the Fredlund and Xing (1994) combined with the relative conductivity function from
van Genuchten (1980) ............................................................................................................................... 112
Figure 6-15. Thermal conductivity of subsoil (SW) ................................................................................... 114
Figure 6-16. Thermal conductivity vs. gravimetric water content for GCL under 25kPa overburden stress
(modified from Ali et al., 2016) ................................................................................................................. 115
Figure 6-17. Net axial stress vs void ratio for saturated GCL at 35°C ........................................................ 116
Figure 6-18. Suction and void ratio relationship for GCL: a) Void ratio vs. suction; b) Void ratio vs.
(suction+100)/100. .................................................................................................................................... 116
Figure 6-19. Temperature vs Time: a) Column A; b) Column B ................................................................. 119
Figure 6-20. Volumetric water content vs Time: a) Column A; b) Column B ............................................ 120
-xiii-
Figure 6-21. Movement of surface of the lining system: a) Hydration stage; b) Heating stage (GCL was
placed on day 14, prior to which the soil was allowed to reach hydraulic equilibrium in the column) ... 121
Figure 6-22. GCLs after hydration stage: a) Specimen #5D190 used in column A; b) Specimen #6D190 used
GCL in column B......................................................................................................................................... 122
Figure 6-23. GCLs after heating stage: a) Specimen #5D190 used in column A; b) Specimen #6D190 used
GCL in column B......................................................................................................................................... 122
Figure 6-24. X-ray image of specimens at the end of heating stage: a) specimen #5D190 was used in column
A; b) specimen #6D190 was used in column B. ......................................................................................... 123
Figure 6-25. X-ray image of specimen #4D190 after hydration from subsoil for 72 days. Gravimetric water
content of the GCL reached 98%. Specimen was used in a trial test identical to the column studies ..... 123
Figure 7-1. Schematic of the Validation experiments: a) #1THM (this thesis); b) #2THM and #3THM
(Southen and Rowe, 2005a); c) #4TH (Bouazza et al., 2014) .................................................................... 126
Figure 7-2. Determination of 𝜅 for GCL2 and GCL3 (digitized data from Southen and Rowe, 2011) ....... 131
Figure 7-3. Determination of 𝜅𝑠 for GCL2 and GCL3 (digitized data from Southen and Rowe, 2007) ..... 131
Figure 7-4. Determination of 𝜅 for SC2 (digitized data from Southen and Rowe, 2011) .......................... 132
Figure 7-5. SWCCs for SC2 and SC3 (experimental data digitized from Southen and Rowe, 2011) ......... 133
Figure 7-6. Change in volumetric water content with time at different elevations: a) Column A; b) Column
B. VG version of the SWCC was used for the simulation........................................................................... 136
Figure 7-7. Change in volumetric water content with time at different elevations: a) Column A; b) Column
B. FX version of the SWCC was used for the simulation ........................................................................... 137
Figure 7-8. Temperature vs. time: a) Column A; b) Column B. VG version of the SWCC was used for the
simulation .................................................................................................................................................. 138
Figure 7-9. Deformation of surface of lining system with time: a) Hydration stage; b) Heating stage. ... 139
Figure 7-10. Prediction of development of net stress-xx vs. time (compression positive). VG: the VG version
of the SWCC was used in the simulation; FX: the FX version of the SWCC was used in the simulation. .. 141
Figure 7-11. Predictions of change in degree of saturation and gas pressure at the middle of GCL during
hydration and heating stages: a) degree of saturation; b) gas pressure .................................................. 142
Figure 7-12. Predictions for profile of degree of saturation for #1THM. The VG version of the SWCC was
used for simulation.................................................................................................................................... 142
-xiv-
Figure 7-13. Predictions for profile of degree of saturation for #1THM. The FX version of the SWCC was
used for simulation.................................................................................................................................... 143
Figure 7-14. Comparison of predictions and experimental data for profile of volumetric water content for
#2THM and #3THM at the end of heating stage ....................................................................................... 144
Figure 7-15. Comparison of predictions and experimental data for profile of temperature for #2THM and
#3THM at the end of heating stage........................................................................................................... 145
Figure 7-16. Predictions of net horizontal stress vs time at the middle of GCL for #2THM and #3THM .. 145
Figure 7-17. Comparison of observed desiccation cracking of #2THM (i.e. G1-S1-L5) and #3THM (i.e. G2-
S2-L3) (adopted from Southen and Rowe, 2005a) .................................................................................... 146
Figure 7-18. Comparison of temperature distribution for #4TH: a) Profile of temperature at the end of
heating stage; b) Temperature vs. time for top reservoir, 20mm below reservoir, and 150mm below
reservoir. ................................................................................................................................................... 147
Figure 8-1. Schematic of the numerical models: a) Base case (i.e. #1THM); b) Design D1 (one layer of
geonet); c) Design D2 (two layers of geonet). Thickness of the geonet was selected based on thickness of
Trinet, a commercial geonet by GEOFABRICS Australia. ........................................................................... 150
Figure 8-2. Trinet, a geocomposite drainage system produced by GEOFABRICS Australia: triple layer net +
single sided geotextile ............................................................................................................................... 150
Figure 8-3. Prediction of profile of temperature for Case #1 under temperatures of 40, 55, and 78°C and
an overburden stress of 20kPa at the end of the heating stage (day 97) ................................................. 155
Figure 8-4. Prediction of profile of degree of saturation for Case #1 under temperatures of 40, 55, and 78°C
and an overburden stress of 20kPa at the end of the heating stage (day 97) .......................................... 156
Figure 8-5. Effect of temperature on development of net horizontal stresses at the middle of GCL for Case
#1 under temperatures of 40, 55, and 78°C and an overburden stress of 20kPa..................................... 156
Figure 8-6. Effect of overburden stress on development of net horizontal stresses at the middle of GCL for
Case #2 under a temperature of 78°C and net vertical stresses of 20, 35, and 50kPa ............................. 158
Figure 8-7. Effect of overburden stress on: a) change in void ratio vs. time. Graphs are plotted for middle
of GCL; a) change in degree of saturation vs time .................................................................................... 158
Figure 8-8. Profile of temperature at the end of heating stage (i.e. on day 200 of simulation and after 142
days of heating): a) Design D1 with one layer of geonet; b) Design D2 with two layers of geonet.......... 159
-xv-
Figure 8-9. Comparison of performance of designs D1 and D2: Net horizontal stress vs. time under
temperatures of 70, 80, and 90°C ............................................................................................................. 160
Figure 8-10. Effect of 𝜆𝑘 on profile of degree of saturation at the end of heating stage......................... 161
Figure 8-11. Sensitivity of developed net horizontal stresses vs. time at the middle of GCL to: a) unsaturated
hydraulic conductivity parameter of SW1, 𝜆𝑘; b) Poisson’s ratio of GCL1, 𝜈; c) Thermal expansion
coefficient of GCL1, 𝛼𝑇.............................................................................................................................. 162
-xvi-
Table of Tables
Table 2-1. Composition and potential energy content (Hamawand et al., 2013)......................................... 7
Table 2-2. Average concentration of total dissolved solids and some inorganic components in brine and
seawater (mg/L) ............................................................................................................................................ 9
Table 2-3. Summary of recent studies on hydration of GCL from subsoil .................................................. 15
Table 2-4. Summary of selected laboratory experiments on effect of thermal gradients on composite lining
systems ........................................................................................................................................................ 18
Table 2-5. Summary of selected publications on numerical simulation of desiccation of mineral liners... 19
Table 2-6. Failure envelopes for hydrated needle-punched GCLs .............................................................. 21
Table 2-7. A summary of recent studies conducted on SWCC of GCLs. ...................................................... 24
Table 5-1. GCL X2000 properties reported by GEOFABRICS Australia. Typical values are presented. ....... 53
Table 5-2. Experimental program................................................................................................................ 54
Table 5-3. Summary of GCL specimens ....................................................................................................... 55
Table 5-4. Height of solids for X2000 based on the typical values reported by the manufacturer† ........... 62
Table 5-5. Preparation and SWCC tests using axis-translation technique .................................................. 63
Table 5-6. Specimen #11D50 SWCC data under target vertical stress of 20kPa and temperature of 20°C 68
Table 5-10. Specimen #37D50 SWCC data under target vertical stress of 20kPa and temperature of 40°C
..................................................................................................................................................................... 69
..................................................................................................................................................................... 69
Table 5-12. GCL specimens mass per area and testing condition ............................................................... 71
Table 5-13. Comparison of VSA and WP4C: specimen #2D37 at 20°C ........................................................ 75
Table 5-14. Comparison of VSA and WP4C: specimen #15D37 at 35°C ...................................................... 75
-xvii-
Table 5-15. VSA measurements for drying and wetting at 60°C on specimen #16D37 .............................. 75
Table 5-16. RH for solutions at 20°C............................................................................................................ 77
Table 5-17. VET measurements at 20°C and under 0kPa ............................................................................ 78
Table 5-18. VET measurements at 20°C and under 20kPa .......................................................................... 79
Table 5-19. VET measurements at 70°C under <1kPa load ......................................................................... 79
Table 5-20. Summary of all tests performed on GCL X2000 ....................................................................... 80
Table 5-21. van Genuchten (1980) SWCC (Eq. 4-7) fitting parameter for the reference curve .................. 85
Table 5-22. Fitting parameters for the proposed VG version of the SWCC (Eq. 4-16) ................................ 85
Table 5-23. Fredlund and Xing (1994) SWCC (Eq. 4-19) fitting parameter for the reference curve ........... 89
Table 5-24. Fitting parameters for the FX version of the proposed SWCC Eq. 4-20 ................................... 89
Table 5-25. Comparison of GCL properties ................................................................................................. 89
Table 5-26. van Genuchten (1980) SWCC (Eq. 4-7) fitting parameter for the reference curves ................ 92
Table 5-27. Fitting parameters for VG version of the proposed SWCC (Eq. 4-16) ...................................... 93
Table 5-28. Fredlund and Xing (1994) SWCC (Eq. 4-19) fitting parameter for the reference curves.......... 95
Table 5-29. Fitting parameters for the FX version of the proposed SWCC Eq. 4-20 ................................... 95
Table 6-1. Description of experiment stages............................................................................................... 99
Table 6-2. Hydration of GCL from different subsoils ................................................................................. 103
Table 6-3. Composition of the synthetic brine .......................................................................................... 104
Table 6-4. Thermal conductivity of subsoil (SW) ....................................................................................... 114
Table 6-5. Thermal conductivity of GCL under 25kPa overburden stress (digitized data from Ali et al., 2016)
................................................................................................................................................................... 115
Table 6-6. Net stress vs void ratio for GCL ................................................................................................ 116
Table 6-7. Summary of GCL properties: Hydraulic, thermal, and mechanical .......................................... 117
Table 6-8. Summary of the subsoil properties .......................................................................................... 118
Table 7-1. Main Features of the Validation Experiments .......................................................................... 127
Table 7-2. Properties of the GCLs for the validation experiments ............................................................ 129
Table 7-3. Properties of the subsoils and air gap for the validation experiments .................................... 130
-xviii-
Table 7-4. Initial conditions and porosities for simulation of the four sets of experiments (compression
positive) ..................................................................................................................................................... 135
Table 7-5. Boundary conditions for simulation of the four sets of experiments ...................................... 135
Table 7-6. Comparison of experimental data and numerical estimates for Gravimetric water content of
GCL. GCLs #5D190 and #6D190 were used in column experiments A and B, respectively. VG: the VG version
of the SWCC was used. FX: the FX version of the SWCC was used. .......................................................... 140
Table 8-1. Material Properties .................................................................................................................. 151
Table 8-2. Initial conditions and porosities (compression positive) .......................................................... 153
Table 8-3. Boundary conditions for simulations of effect of temperature (case #1) and overburden stress
(case #2) and performance of designs D1 and D2. For all cases, the equilibrium and hydration stages (i.e.
day 0 to 58) were identical to the base case. ............................................................................................ 153
Table 8-4. Details of sensitivity analyses. All models were built based on the base case (i.e. #1THM). .. 154
-xix-
Notations
𝑎 cross-sectional area of the reservoir containing the influent permeant
𝑎𝑓 a parameter to be determined from air-entry value
𝑎3 a coupling parameter
𝐴 cross-sectional area of specimen
𝐴𝑠ℎ𝑎𝑓𝑡 cross section area of the piston shaft, m2
𝐴𝑝𝑖𝑠𝑡𝑜𝑛 cross section area of piston, m2
𝑏 vector of body force
𝐷 a model parameter, m2s-1K-nPa
𝑒 void ratio
𝑒𝑟𝑒𝑓 reference void ratio
𝑒𝑏𝑢𝑙𝑘 bulk void ratio
𝐸𝑠 internal energy of solid phase
𝐸𝑙 internal energy of liquid phase
𝐸𝑎 internal energy of gas phase
𝑓𝑄 external supply of heat
𝑓𝑎 external supply of air
𝑓𝑤 external supply of water
𝑔 the gravitational acceleration, 9.81m/s2
𝐺𝑠 specific gravity
ℎ pressure head, (m)
𝑖
ℎ time- and space- average suction
ℎ𝑖 average suction at 𝑡 𝑖
ℎ𝑗𝑖 suction at 𝑧𝑗 and 𝑡 𝑖
𝐻𝐺𝐶𝐿 height of GCL, m
-xx-
𝐻𝑠 height of solids, m
∆ℎ1 head loss across the specimen at 𝑡1
∆ℎ2 head loss across the specimen at 𝑡2
𝑖𝑐 conductive flux of heat
𝑖𝑣𝑎𝑝𝑜𝑢𝑟 diffusive flux of vapour
𝐈 identity matrix
𝑗𝐸𝑎 advective flux of heat by air mass motion
𝑗𝐸𝑙 advective flux of heat by liquid mass motion
𝑗𝐸𝑠 advective flux of heat by solid mass motion
𝑗𝑔𝑎 total flux of air in gas phase
𝑗𝑔𝑤 total flux of water in gas phase
𝑗𝑙𝑎 total flux of air in liquid phase
𝑗𝑙𝑤 total flux of water in liquid phase
𝑘𝑟𝑔 gas relative hydraulic conductivity
𝑘𝑠 saturated hydraulic conductivity
𝑘𝑠 saturated hydraulic conductivity, m/s
𝐿 length of the specimen
𝑚𝑓 a parameter to be determined by residual water content
𝑚𝑝𝑖𝑠𝑡𝑜𝑛 mass of piston, kg
𝑛 model parameter that controls slope of curve
𝑀𝑏 mass per area of bentonite, kg.m-2
𝑀𝑔𝑒𝑜 are mass per area of geotextiles, kg.m-2
𝑛𝐷 a model parameter in vapour diffusion
𝑛𝑓 a parameter to be determined by rate of water extraction from soil once air-entry

value is exceeded
𝑝𝑣 partial pressure of pore-water vapour, kPa
-xxi-
𝑝𝑣0 saturation pressure of water vapour over a flat surface of pure water at the same
temperature, kPa
𝑃 model parameter that is representative of air-entry value, kPa
𝑃0 model parameter at reference 𝜎0 and 𝑒𝑟𝑒𝑓 , kPa
𝑃𝑔 gas pressure (compression is positive), kPa
𝑃𝑙 liquid pressure (compression is positive), kPa
𝑃𝑙𝑜𝑎𝑑 gas pressure applied to top of loading piston, kPa
∆𝑄 quantity of flow for a given time interval ∆𝑡, taken as average of inflow and out
flow
𝑅 universal (molar) gas constant, i.e. 8.31432 J/(mol K)
𝑠 suction, kPa
𝑠𝑇𝑜𝑡𝑎𝑙 total suction, kPa
𝑆𝑒 effective degree of saturation of liquid phase
𝑆𝑟 degree of saturation of liquid phase
𝑆𝑟𝑔 degree of saturation of gas phase
𝑆𝑟𝑠 residual degree of saturation of liquid phase
𝑆𝑠 maximum degree of saturation of liquid phase
𝑡1 time at start of permeation trial, date
𝑡2 time at end of permeation trial, date
∆𝑡 interval of time over which the flow ∆𝑄 occurs (i.e. 𝑡2 − 𝑡1 )
𝑇 temperature, °C
𝑇𝑐 critical temperature, i.e. 373.95°C
𝛼 model parameter reperasentative of air-entry value, (1/m)
𝛼𝑇 linear thermal expansion coefficient, 1/°C
𝛼𝑒 model parameter, kPa
𝛽𝑒 model parameter
𝜃 volumetric water content
-xxii-
𝜃𝑟 residual volumetric water content
𝜃𝑠 saturated volumetric water content
𝜅 slope of unload/reload curve in the 𝑒 − ln(−𝜎𝑛𝑒𝑡 ) diagram
𝜅0 the intrinsic permeability of the matrix at reference porosity, 𝜙0 , m2
𝜅𝑙 intrinsic permeability of the liquid phase
𝑠+0.1
𝜅𝑠 slope of unload/reload curve in the 𝑒 − ln ( 0.1
) diagram
𝜆 model parameter in van Genuchten SWCC
𝜆𝑇 thermal conductivity of unsaturated porous media, Wm-1K-1
𝜆𝑑𝑟𝑦 thermal conductivity of dry porous media, Wm-1K-1
𝜆𝑘 a fitting parameter for relative hydraulic conductivity
𝜆𝑠𝑎𝑡 thermal conductivity of water saturated porous media, Wm-1K-1
𝜇𝑤 the dynamic viscosity of water, it is 0.001 𝑃𝑎. 𝑠 at 20°C
𝜌𝑏 density of bentonite, kg.m-3
𝜌𝑔 density of gas
𝜌𝑔𝑒𝑜 density of geotextile, kg.m-3
𝜌𝑙 density of liquid
𝜌𝑤 density of water, 998.2 kg/m3 at 20°C
𝜎 surface tension at current temperature, N/m
𝜎0 surface tension at reference temperature, it is 0.072 N/m at 20°C
𝜎𝑇𝑜𝑡𝑎𝑙 tensor of total stress
𝜎𝑚𝑒𝑎𝑛 the mean total stress, kPa
𝜎𝑛𝑒𝑡 𝑎𝑥𝑖𝑎𝑙 the axial net stress, kPa
𝜎𝑛𝑒𝑡 the net stress, kPa
𝜓𝑟 soil suction corresponding to the residual water content
𝜏 tortuosity
𝜔𝑔𝑎 mass fraction of air in gas phase
-xxiii-
𝜔𝑔𝑤 mass fraction of water vapour in gas phase
𝜔𝑙𝑎 mass fraction of air in liquid phase
𝜔𝑙𝑤 mass fraction of water in liquid phase
𝜔𝑣 molecular mass of water vapour, i.e. 18.016 kg/kmol
𝜙 porosity
𝜙0 the reference porosity where the reference intrinsic permeability, 𝜅0 , is measured
-xxiv-
Abbreviations
CGCLS ` Composite geosynthetic clay lining system
CSG Coal seam gas
CBM Coalbed methane
CLS Composite lining system
FX Fredlund and Xing
GCL Geosynthetic clay liner
GMB Geomembrane
HDPE High density polyethylene
LVDT liner variable differential transducer
NP Needle-punched
NPTT Needle-punched thermally treated
SWCC Soil-water characteristics curve
THM Thermo-hydro-mechanical
VET Vapour equilibrium technique
VG van Genuchten
VSA Vapour Sorption Analyser
-xxv-
1 Background and Aims
Chapter 1
Background and Aims
1.1 General
Worldwide demand for energy has led to the development and discovery of new sources of energy such
as coal seam gas (Hamawand et al., 2013). Coal seam gas (CSG) or coalbed methane (CBM) refers to a
natural gas extracted from coalbeds. Extraction of CSG first commenced in the United States around 30
years ago (Hamawand et al., 2013). Today, it is a major source of energy in the United States, Canada,
Australia, China and Europe. Extraction of CSG is associated with the production of a large amount of brine
with high concentration of salt, organic and inorganic components, and heavy metals (Hamawand et al.,
2013) and regulation in most countries require safe and reliable disposal systems for this waste water.
Brine ponds, also called evaporation ponds, are relatively simple and efficient solutions for the
management of the large amount of the waste water, and composite geosynthetic clay lining systems
(CGCLS) have been used on coal seam gas sites as the main basal lining system for brine ponds to prevent
groundwater contamination (Bouazza et al., 2013).
Composite lining systems (CLSs) usually consist a primary liner such as a geomembrane (GMB) and a
secondary liner made of clay or geosynthetic clay liners (GCLs). In addition, double or multiple-layer
composite lining systems may be constructed by including more layers of GMB and clay and/or GCL. A
composite geosynthetic clay lining system (CGCLS) refers to a specific type of CLS that is made of a GCL
underlying a GMB. The GMB plays an important role in controlling the transport of contaminants through
the barrier because of its extremely low permeability and high resistance to aggressive chemicals usually
present in leachates. The GCL generally further decreases diffusive and advective contaminant transport
-1
on account of the significant expansion of the GCL after hydration, which creates close contact between
the GCL and the geomembrane and reduces the hydraulic transmissivity of the interface.
The performance of CLSs (including both clay and GCL as the primary liner) has been extensively
studied for landfill applications (e.g. Ruhl and Daniel, 1997; Southen and Rowe, 2005a; Southen and Rowe,
2005b; Southen and Rowe, 2011; Azad et al., 2011; Azad et al., 2012; Hoor and Rowe, 2013) and most of
our knowledge of CLS design and application comes from landfill studies. On the other hand, CGCLSs are
exposed to considerably different conditions in brine ponds, namely higher temperatures and lower
overburden stress. Recent studies indicate that the risk of desiccation of the GCL component of a CGCLS is
higher under brine pond conditions, compared to landfill conditions (Hoor and Rowe, 2013; Rowe and
Verge, 2013). The performance of a CGCLS deteriorates significantly if the GCL component is desiccated.
This may lead to a serious risk of contamination to the environment. Therefore, it is important to
understand the behaviour of CGCLSs and assess the safety of these lining systems under brine pond
conditions. While brine ponds and solar ponds are especially pertinent here because of the combination
of high-temperature and low-pressure to which they subject the GCL, other instances in which thermal
loads expose the liner to higher-than-usual temperatures include low level nuclear waste repositories (e.g.
Bonaparte et al., 2008) and heap leach pads (e.g. Abdelaal and Rowe, 2015). Hence, the studies reported
in this thesis are relevant to all of the above-mentioned engineering problems.
Generally, GCLs operate under unsaturated conditions at all stages as a component of a CGCLS. They
are delivered onsite at a very low gravimetric water content (typically less than 10%) and, even when
properly hydrated by the subsoil, rarely reach full saturation. Furthermore, as mentioned above, if
subjected to thermal loads, partial dehydration occurs. Hence, a thermo-hydro-mechanical (THM) theory
for unsaturated porous media is required to describe the complex behaviour of these lining systems. While
theories of soil behaviour under unsaturated conditions have been developed for multi-phase soils (e.g.
Olivella et al., 1994; Thomas and He, 1997), their ability to predict desiccation of GCL systems, especially
at temperatures higher than those encountered in landfills (55oC), has remained an open question.
In addition, a key constitutive equation in unsaturated soil mechanics is the relationship between
water content and suction, known as soil-water characteristic curve (SWCC). Despite some work in this
area, this relationship is not yet well characterised for GCLs. For example, surface tension of water is one
of the main factors that influence the SWCC and temperature can considerably change surface tension of
water (Grant and Salehzadeh, 1996) and, consequently, the SWCC. Nevertheless, our knowledge of the
effects of temperatures on the SWCC of GCLs remains poor. Furthermore, the SWCC of a wide range of
soils depends on void ratio (e.g., Russell, 2014). However, the effects of different applied loads – hence
different void ratios – on the SWCCs of GCLs are not well characterised. In fact, SWCC measurements for
GCLs have not covered conditions of high temperature and low pressure encountered in brine ponds,
-2
despite the fact that even a small change in this constitutive equation may lead to very different soil
behaviour and levels of desiccation risk.
1.2 Objectives
The goal of this thesis is to identify conditions under which CGCLSs subjected to high temperatures and
low pressures are prone to desiccation and the consequent loss of performance. It reports an experimental
and computational investigation that has the following objectives:
1. Perform SWCC measurements for GCLs with different void ratios and controlled temperatures in
order to develop SWCC equations that better reflect conditions of high-temperature and low
overburden stress encountered in brine ponds and other comparable engineering applications.
2. Perform a set of column studies of GMB-GCL-Subsoil systems subjected to high thermal gradients
and low overburden stress in order to generate data that can validate a predictive THM theory of
behaviour under unsaturated conditions.
3. Assess the ability of the THM theory by Olivella et al. (1994), based on non-linear elasticity and
implemented in the THM software CODE_BRIGHT (Olivella et al., 1996), to explain the results
generated in 2, hence developing a model that can predict the dehydration and possible
desiccation caused by high thermal gradients under low overburden pressure.
4. Use the model developed in 3 in order to identify conditions under which the risk of desiccation is
likely to be high. This is done by assessing the effects of different parameters on the development
of tensile stresses in the GCL and hence the risk of desiccation.
The material used in this thesis is ELCOSEAL geosynthetic clay liner grade X2000, manufactured by
GEOFABRICS Australia. The GCL is made of a powder Na-bentonite and polypropylene geotextiles. The
carrier is made of a layer of non-woven geotextile on top of a woven geotextile, while the cover is made
of a non-woven geotextile. In addition, the GCL is fibre-reinforced by needle-punching and the fibres are
thermally locked. This product was selected for the thesis for two reasons. First, ELCOSEAL X2000 is widely
used in Australia. Second, the product has been subject to past studies at Monash University in which its
saturated permeability, thermal conductivity at variable saturation conditions, and water retention
properties under controlled temperature conditions and high suctions (i.e. higher than 1.5MPa) along the
wetting and drying path have been well characterised. The two important advantages of the availability of
this data is that first, the data was cross checked for those experiment that were performed in thesis in
parallel with Monash University and second, only a small number of parameters was back-fitted during
the process of model calibration.
-3
A precise assessment of desiccation risk should ideally take into account the possibility that the brine
solution might escape through leaks in the geomembrane into the transmissive layer between the GCL and
the Geomembrane, and subsequently into the GCL and the subsoil. This can have at least three important
effects on the GCL. First, it is possible that the leaked brine might partially counter-act the thermally
dehydration process by providing some moisture from above. Second, the presence of salt in the brine
introduces osmotic suction into the GCL which can modify significantly its water retention properties.
Third, it has been shown that the hydraulic conductivity of a GCL with Na-bentonite can increase by up to
one or two orders of magnitude because of cation exchange with calcium-rich leachates under low
overburden stress (Bouazza, 2002).
It is important for theoretical and practical reasons to assess the desiccation risk, prior to any leakage
through defects in the geomembrane, especially since even when defects are present, most of the surface
of the geomembrane remains intact. Therefore, given the scope of the investigation which encompasses
both experimental and numerical investigations, no contact between the brine and the GCL is assumed to
take place in this thesis.
The thesis makes the following original contributions to the literature on the unsaturated
behaviour of GCLs:
1. New SWCCs for GCL which take into account the effects of void ratio and temperature;
2. New experimental data on the performance of CGCLSs under temperature of up to 80oC and
pressure below 50kPa;
3. A validated non-linear thermo-hydro-elastic theory of CGCLSs that applies to the conditions
described in 2;
4. An assessment of the safety and performance of CGCLSs under these conditions.
1.3 Structure of the Thesis
The thesis is structured as follows. Chapter 2 presents a review of literature on thermally-induced

desiccation of GCLs, performance of composite lining systems under thermal gradients, and the effects of
temperature and deformation on the SWCCs of GCLs. The research methodology and the relationship
between the different components of the investigations are described in Chapter 3. Chapter 4 discusses
the theory of desiccation of GCLs. Experimental measurements of the SWCCs of GCLs under different
temperature and loading conditions are reported in Chapter 5. Column studies representing a CGCLS under
brine conditions are described in Chapter 6. In Chapter 7, experimental data generated in Chapters 5 and
6 are used to validate a numerical model of desiccation, based on the theory presented in Chapter 4. The
validated model is used in Chapter 8 to conduct parametric analyses of CGCLSs under brine pond
-4
conditions in order to assess desiccation risk. Lastly, conclusions and implications for practice and research
are elicited in Chapter 9.
-5
2 Literature Review
Chapter 2
Literature Review
2.1 Introduction
Over the last decade or so, coal-seam gas extracted in the US, Australia, China, and parts of Europe has
become a significant source of energy on global markets (Hamawand et al., 2013). This development has
been partly responsible for a long-term decline in the price of oil, with far-reaching consequences for the
global economy. Composite geosynthetic clay liner systems have been widely used for protection of
groundwater from contamination on coal seam gas sites. Although, risk of desiccation of GCLs is high under
brine pond conditions which may impose serious risk of contamination to the environment, especially
groundwater.
In this chapter, review of literature on performance of composite lining systems, hydration and
desiccation of GCLs, and shear strength of GCLs is presented. In addition, a key relationship in hydraulic
performance and desiccation of GCL is water retention of GCL and effect of load and temperature on the
relationship. Therefore, a detailed review of literature on SWCC of GCL is performed. Moreover, a brief
review is conducted on water retention models for porous media considering effect of temperature and
deformation. Finally, effect of defects in GMB and desiccation of GCL under leaks is reviewed.
2.2 Coal Seam Gas and Brine Ponds
Research on coal seam gas started in the US as a potential new source of energy in response to the 1970s
energy crisis. Table 2-1 compares the composition and energy content of natural gas, coal seam gas, and
biogas. Generally, the potential energy content of coal seam gas is lower than natural gas and higher than
biogas, which corresponds roughly to the methane content of each type of gas.
-6
Table 2-1. Composition and potential energy content (Hamawand et al., 2013)
Other Energy
CH4 CO2
Components (kWh/m3)
Natural gas 97.0% 0 3.0% 11.30
Coal seam gas 95.0% 3.0% 2.0% 9.96
Biogas 68.0% 26.0% 6.0% 7.50
The fast spread of coal-seam mining has generated heated debates about its actual and potential
environmental impacts. One of the more intractable environmental problems associated with coal-seam
gas is the brine water often generated in the process of extraction. Hydraulic fracturing is the most
common technique used for stimulation of coalbeds to enhance flow of gas and Figure 2-1 illustrates the
various steps involved in the extraction. Fracturing liquids and additives are injected into the coalbeds
under high pressure to introduce a network of small fractures within the gas-bearing geological formations
(Australian Government, 2014). Following this process, a large amount of water is usually extracted from
fractured formations. The volume of the water depends on the specific geological formations. The
extracted water is mixed with gas and generally the ratio of gas to water production increases with time.
The high salinity of extracted water precludes its re-injection into aquifer zones where it would pollute
groundwater reserves (Hamawand et al., 2013; Australian Government, 2014). Table 2-2 shows a
comparison of average concentration of total dissolved solids and some inorganic components found in
brine from different basins and seawater.
Figure 2-1. Schematic of coal seam gas extraction (credit: Australian Science Media Centre, accessed on July 2016)
http://theconversation.com/three-myths-the-coal-seam-gas-industry-wants-you-to-believe-24422
-7
A review of coal seam gas extraction by Moore (2012) shows that disposal of waste water in unlined
ponds or discharge to surface water in the US (Powder River basin) leads to significant contamination of
soil, surface water, and groundwater which indicates the importance of using a well-engineered lining
systems in these ponds. One potentially effective response to this problem is to divert the water into
evaporation ponds. Brine ponds have been used as a traditional method for removing water from saline
solutions for centuries, in view of their simplicity and low cost of construction and maintenance.
Modern versions of brine ponds are usually lined with a composite liner made of a high-density
polyethylene geomembrane (HDPE-GMB) and a geosynthetic clay liner (GCL), over an attenuation layer
made of natural soil, separating the liner from the aquifer (Figure 2-2). Although, recent studies show that
GCLs in composite lining systems may desiccate under temperature gradients (Southen and Rowe, 2005a)
and risk of desiccation increases with lower overburden stress (Rowe and Verge, 2013; Hoor and Rowe,
2013).
GCL desiccation significantly reduces the performance and life span of the composite lining system.
Most of investigations on GCL desiccation have been conducted for landfill conditions where lower
temperature and higher overburden stresses prevail. Hence, an important question mark remains about
the susceptibility to thermal dehydration and desiccation of GMB-GCL composite systems in brine ponds,
as well as other applications of lining systems in which surface temperatures exceed 60oC and effective
stresses in the GCL are lower than 50kPa.
Figure 2-2. A typical brine pond on a coal seam gas site (www.aqob.com.au, accessed on May 2014)
-8
Table 2-2. Average concentration of total dissolved solids and some inorganic components in brine and seawater (mg/L)
Greater Powder Bowen Basin in Bowen Basin in

Raton San Juan
Green River River Upper Seam Lower Seam Seawater
(USA)[1] (USA)[1]
(USA)[1] (USA)[1] (Australia)[2] (Australia)[2]
Total dissolved
2148 997 2512 4693 5149 5342 35000
solids
Total alkalinity
1488 1384 1107 3181 699 1022 -
as CaCO3
Arsenic (As) 0.027 0.001 0.01 0.001 - - -
Chlorine (Cl) 336 21 787 624 2805 2947 19700
Calcium (Ca) 13 32 14 53 43 34 410
Lead (Pb) 0.003 ND 0.023 0.023 - - -
Sodium (Na) 824 356 989 1610 2077 2259 1900
[1]: Dahm et al. (2011); [2]: Kinnon et al. (2010); ND: Not Detectable
2.3 Geomembranes and Geosynthetic Clay Liners
As mentioned earlier, HDPE-GMB and GCLs are two main components of lining systems. Different
configurations of GMBs and GCLs have been used in practice such as single or multiple layers of GMB and
GCL for both basal and capping lining systems. Existing studies of the thermal dehydration of basal lining
systems are based on a single layer of GCL covered by a GMB on the top, facing the source of contamination
(e.g., Southen and Rowe, 2005a; Bouazza et al., 2014). Thermal dehydration and desiccation of double-
layer composite lining systems have also been investigated by Azad et al. (2011) and Azad et al. (2012).
An HDPE-GMB is usually the main protective layer used in CGCLSs because of its extremely low
permeability to water, chemical resistance to aggressive chemicals and efficiency in controlling diffusive
transport of inorganic contaminants (Rowe and Sangam, 2002). The earliest processes of GMB
manufacturing date back to the 1930s. Today, geomembranes for geotechnical and civil engineering
applications are manufactured under a low-pressure polymerisation method (Muller, 2007). According to
ASTM D883-12e “Standard Terminology Relating to Plastics”, a polyethylene resin with a density of 0.941
g/cm3 and above is classified as HDPE. The durability and performance of HDPE geomembranes have been
widely studied for environmental applications (e.g., Rowe and Sangam, 2002; Muller, 2007; Rowe et al.,
2009; Touze-Foltz et al., 2016).
In practice, HDPE geomembranes are susceptible to defects and punchers (Giroud and Bonaparte,
1989a; Giroud and Bonaparte, 1989b; Giroud and Daniel, 2004; Touze-Foltz et al., 2008). Rowe and
Abdelatty (2013) presented a thorough summary of the quantity and probability of defects in GMBs and
the associated leakage through the defects for a wide range of industrial applications. One reason for
defects occurring in GMBs is the formation of wrinkles after installation of the GMB (due to diurnal heating
-9
and cooling cycles) and the tendency of these materials to crack (Muller, 2007). Another type of defects
happens because of presence of sharp objects above or underneath the GMB (Saidi et al. 2008). Therefore,
secondary liners are sometimes used to provide an additional protection layer, countering any leakage
through the primary liner (Giroud and Bonaparte, 1989b).
GCLs have gained extensive popularity, compared to the more conventional CCLs because of a
number of advantages such as low permeability to water, additional space available for waste as a result
of the smaller thickness, good capability to handle differential settlements, and easy and fast installation
(Bouazza and Gates, 2014). GCLs are usually made of a thin layer of sodium bentonite encapsulated within
layers of geosynthetics. The sandwich structure is needle-punched, stitch-bonded, or bonded with an
adhesive. The geosynthetics are either geotextiles or geomembranes.
Montmorillonite is the main mineral component of bentonite. It is composed of alternating sheets of

aluminium silicates. Owing to the unique formation of bentonite, most of the minerals carry a net negative
charge on their surfaces which lead to the creation of a layer of water, close to the clay surface, with
reduced mobility and helps explain bentonite’s high plasticity and low hydraulic conductivity (Shackelford
et al., 2000). The latter property is crucial for barrier systems. In addition, the swelling characteristics of
hydrated bentonite allows the GCL to expand and come into close contact with the overlying GMB (Touze-
Foltz et al., 2016). This leads to a reduced hydraulic transmissivity of the interface between the GCL and
GMB and significantly decreases leakage through defects in the GMB – which is important in maintaining
good performance since GMB defects are difficult to prevent in practice as described earlier.
It has been shown that the type and composition of bentonite and GCL manufacturing process can
affect some GCL properties. Needle-punched (NP) GCLs have gained more attention than other types of
GCLs because of their coherent shear responses. Thermal treatment of the fibres can further enhance
shear strength under higher normal stresses (Zornberg et al., 2005) and may reduce fibre pull out under
shearing (Fox and Stark, 2015). Sodium bentonite is commonly used for GCL manufacturing since
permeability of sodium bentonite to water is about one order of magnitude lower than that of calcium
bentonite (Bouazza and Gates, 2014).
Though, a high increase (one or two order of magnitude) in hydraulic conductivity of GCL with sodium
bentonite has been observed as a result of cation exchange when in contact with calcium-rich soils or
calcium-rich leachates under low overburden stress (Bouazza, 2002). In addition, Lee and Shackelford
(2005) showed that the greater the cation exchange capacity and sodium montmorillonite content of the
bentonite in a GCL, the more susceptible it is to chemical attack.
In response to the issues related to cation exchange, an innovative type of GCL, known as dense
factory-prehydrated (DPH) GCL has recently been introduced that benefits from the well-known effects of
-10
prehydration and densification of bentonite on hydraulic conductivity of Na-bentonite. Furthermore, DPH
GCLs commonly contain additives such as polymers that can improve workability, sealing capacity, and
resistance of GCL to aggressive permeants (Mazzieri et al., 2013; Mazzieri and Emidio, 2015). Nevertheless,
these new GCLs have not yet become widely used and their performance remain the subject of research.
There is now a rich literature on the hydraulic performance and chemical compatibility of GCLs (e.g.,
Benson et al., 2010; Rosin-Paumier and Touze-Foltz, 2012; Xue et al., 2012; Liu et al., 2013; Mazzieri et al.,
2013; Bouazza and Gates, 2014; Athanassopoulos and College, 2015; Bradshaw et al., 2016; Tian et al.,
2016). Other topics related to GCL performance have also received considerable attention such as
contaminant transport (Rowe et al., 2016; Shackelford et al., 2016; Touze-Foltz et al., 2016), gas
permeability and gas diffusion (Bouazza, 2002; Rouf et al., 2016a), transmissivity of GMB-GCL interface
and flow through GMB defects (El-Zein, 2008; El-Zein et al., 2012; Touze-Foltz et al., 2016), thermal
conductivity (Singh and Bouazza, 2013; Ali et al., 2016), effects of freeze and thaw cycles (Podgorney and
Bennett, 2006; Makusa et al., 2014), and shear strength (Zornberg et al., 2005; Fox and Stark, 2015; Risken
et al., 2016). A full review of this literature is beyond the scope of this thesis and the reader is referred to
review papers by Bouazza et al. (2013) and Touze-Foltz et al. (2016).
It is evident from the literature that, for an Na-bentonite GCL to deliver the required low hydraulic
conductivity, it must be hydrated with a suitable permeant (i.e. fresh and soft water with low
concentration of cations such as calcium or magnesium) prior to exposure to contaminants. In addition, it
must remain hydrated during its operating life. Although, dehydration of GCL may occur due to thermal
gradients and may ultimately lead to the desiccation of GCL because of extreme moisture loss. Both
hydration and dehydration of GCLs has been found to be highly sensitive to the water retention properties
of both the GCL and underlying soil providing moisture for hydration or rehydration. Hence, a detailed
review of the literature on hydration of GCLs from subsoil, thermally-induced desiccation of GCLs, and
SWCCs of GCLs, is presented in the following sections.
2.4 Hydration of Geosynthetic Clay Liners
Hydration of the GCL component of CGCLSs has been extensively studied because the degree of saturation
of a GCL is one of the main factors that directly affects the performance of CGCLSs (Touze-Foltz et al.,
2016). In practice, GCL must hydrate from the subsoil in CGCLS applications for basal barriers because the
GCL is covered with the GMB which prevents hydration from the top. Moreover, the composition of
permeant from the contained hazardous material is generally aggressive and unsuitable for GCL hydration.
On the other hand, pre-hydrating the GCL before placement of the GMB and the waste is not advisable
-11
because, subsequently, it would make it extremely difficult to walk on the site and lay the GMB on the
hydrated GCL.
It has been shown that the rate of hydration of the GCL from the subsoil, as well as the equilibrium
water content in the GCL, depend on a number of factors such as the type of GCL and its method of
manufacture, the nature of the subsoil and its initial water content, the applied overburden stress and the
CGCLS temperature variations. Table 2-3 shows a summary of recent studies on hydration of GCL
component of CGCLSs from the subsoil.
Upward flow of water and diffusion of vapour from the subsoil are the two main mechanisms involved
in the hydration of GCL. Hydraulic and transport properties of GCL and subsoil may change with changes
in void ratio (Beddoe et al., 2011; Bannour et al., 2014; Seiphoori et al., 2016), chemical attack (Shackelford
et al. 2016), and temperature (Chevrier et al. 2012; Barclay and Rayhani, 2013). Therefore, it is important
that relevant coupling between thermal, hydraulic, mechanical and chemical variables be taken into
account in any predictive models of hydration of GCLs. Generally, suction in the GCL before placement on
the subsoil is high (≅ 100𝑀𝑃𝑎) because of the low water content, and at equilibrium with humidity of air
which depends on the local climate. On the other hand, suction in the subsoil depends on the type and
moisture content of the subsoil which is controlled by the location of the water table and the patterns of
evaporation and infiltration of rainwater (Zhu and Mohanty 2002).
Movement of moisture from the subsoil to the GCL occurs after the installation of the GCL because
of higher suction potential and lower vapour concentration in GCL relative to those in the subsoil. The rate
of hydration increases as the moisture content of the subsoil increases (Rayhani et al., 2011; Anderson et
al., 2012; Sarabian and Rayhani, 2013). Chevrier et al. (2012) have observed an increase in both rate of
hydration and final water content of GCL as initial water content of subsoil increases. They report the same
trend for a sand (SP) and a clay (CL) subsoil under different temperatures. On the other hand, Sarabian
and Rayhani (2013) compare the hydration of a GCL in contact with four types of soil (SP, SM, SC, and CL)
with the same initial gravimetric water content of 10% and conclude that the rate of hydration and final
water content of GCL is lower for a finer subsoil. This is because, at the same initial water contents, the
difference in suction potential between GCL and subsoil is lower in a finer soil compared to a coarser soil.
In contrast, Bouazza et al. (2016) reported that a soil with higher clay content (CL) was able to hydrate a
GCL to a higher water content than a soil with lower clay content (CH). They found that the latter soil had
a higher smectite content which was able to increase the water retention and suction potential of the
subsoil.
Sarabadani and Rayhani (2014) report that overburden stress on a CGCLS increases the hydration rate
because of better contact between subsoil and GCL. They also report that higher loads result in less
-12
swelling and lower final water content of GCL at equilibrium. In addition, experiments under simulated
field conditions show that daily and seasonal thermal cycles can significantly reduce GCL hydration from
subsoil (Rowe et al. 2011; Sarabian & Rayhani 2013). However, Chevrier et al. (2012) report that under
isothermal conditions, hydration of GCL increases as ambient temperature increases and conclude that
hydration of GCL is achieved primarily by vapour transfer.
Almost all studies on hydration of GCL are experimental investigations with numerical works of
Siemens et al. (2012) and Siemens et al. (2013) being the exception. Despite the complexity of governing
mechanisms described above, hydration simulations have been performed based on water flow only and
have not included the effects of vapour transfer. Siemens et al. (2013) consider the effects of overburden
stress in the context of 2D flow simulations. They compare results from transient analyses of GCL hydration
under loads of 2 and 100kPa considering change in hydraulic conductivity of GCL based on experiments by
Rowe and Hosney (2013) and assuming that the air-entry of the GCL under 100kPa is two orders of
magnitude higher than the GCL under 2kPa. They have found that a higher normal stress leads to a lower
GCL swelling, lower equilibrium water content and a shorter time taken to reach equilibrium.
No experimental or numerical data exist that directly reports the developed horizontal net stresses
during hydration of a GCL from a subsoil. Numerical simulations have usually ignored this process,
including work by Southen and Rowe (2011), Rowe and Verge (2013), Hoor and Rowe (2013) and Siemens
et al. (2012, 2013).
A number of swelling pressure experiments, under constant volume, have been conducted on
expansive soils (Schanz and Tripathy, 2009; Wang and Liu, 2012; Schanz et al., 2013). Mineralogy and
density of the expansive material and chemistry of the permeant are the three main factors that control
swelling pressure of an expansive soil. Experiments on two commercial bentonites by Schanz et al. (2009)
show that swelling pressures of hydrated specimens by distilled water exceed 14 and 10MPa for dense
sodium (MX80) and calcium (Calcigel) bentonites, respectively. Conversely, the reported swelling
pressures for low densities are relatively small (Figure 2-3). Figure 2-4 shows that decline in swelling
pressure because of increase of concentration of salt (NaCl) is more pronounced for Na-bentonite (MX80)
specimens with low densities. Swelling pressure experiments have also been performed on different GCLs
(Junfeng et al., 2006; Di Emidio et al., 2008; Janssen et al., 2015). Janssen et al. (2015) reported a range of
swelling pressures from nearly 30 kPa to 130 kPa for GCLs with different types of bentonite. Polymer
treatment of bentonite used in the GCLs increased swelling pressure of these GCLs from 40 kPa to around
170kPa (Figure 2-5).
-13
Figure 2-3. Swelling pressure of two commercial bentonites with distilled water: (left) MX80 Na-bentonite; (right) Calcigel Ca-
bentonite (after Schanz et al., 2013)
Figure 2-4. Swelling pressure of MX80: (left) distilled water; (right) different NaCl concentrations (after Wang and Liu, 2012)
Figure 2-5. Effect of polymer treatment on swelling pressure of GCL: (left) without treatment; (right) with treatment (after
Janssen et al., 2015)
-14
Swelling pressure of a GCL can reach 130kPa and polymer treatment of bentonite can increase the
swelling pressure to 170kPa. This is consistent with swelling pressure of bentonite with similar density to
GCL. Typically, a swelling pressure test is performed under constant volume with isotropic stress condition;
although, hydration of GCL in field applications is associated with expansion and development of deviatoric
stresses which must not exceed the shear strength of GCL.
Table 2-3. Summary of recent studies on hydration of GCL from subsoil
Temperature (°C)
Bentonite type
Subsoil type
Load (kPa)
GCL type
References Comments
Silty sand
Rayhani et al. (2011) NP, NPTT GB (SM), Sand 2 22 -
(SP)
Silty sand
Cycles of 20 Same GCLs and subsoils as Rayhani
Rowe et al. (2011) NP, NPTT GB (SM), Sand 2
to 60 et al. (2011) were used.
(SP)
4, 7, 9.88,
Silica Sand,
Chevrier et al. (2012) NP GB 14.1, 21.16, 5, 20, 45 -
Clay
28.2
Silty sand 2D simulations (SEEP/W) on
Siemens et al. (2012) NP, NPTT GB (SM), Sand N/A N/A Rayhani et al. (2011) and sensitivity
(SP) analyses were performed.
Clayey sand 22, Cycles of Same GCLs as Rayhani et al. (2011)
Anderson et al. (2012) NP, NPTT GB 1
(SC) 20 to 60 were used.
Sarabian and Rayhani 23, Cycles of Same GCLs as Rayhani et al. (2011)
NP, NPTT GB Clay (CL) 1
(2013) 23 to 60 were used.
Barclay and Rayhani Sand (SP), Clay 22, 35, 45, Same GCLs as Rayhani et al. (2011)
NP, NPTT GB 2
(2013) (CL) 55 were used.
Effect of overburden stress was
considered in context of 2D flow
Siemens et al. (2013) NPTT GB Silty sand 2, 100 N/A simulations (SEEP/W) on (Rayhani
et al. 2011) and sensitivity analyses
were performed.
Same GCLs as (Rayhani et al. 2011)
Sarabadani and Sand (SP), Clay 0, 0.5, 1, 2,
NP, NPTT GB 22 and subsoils as (Barclay & Rayhani
Rayhani (2014) (CL) 5, 8
2013) were used.
Clay content of CL was higher than
Sily Sand (SM),
CH; however, CL was able to
Clayey Sand
Bouazza et al. (2016) NPTT PB 2.5 21±1 hydrate a GCL to higher water
(SC), Clay (CL,
content than CH because smectite
CH)
content of CH was higher than CL.
GB: Granular Bentonite; PB: Powder Bentonite; NP: Needle-Punched; NPTT: Needle-Punched Thermally-Treated
-15
2.5 Desiccation of Geosynthetic Clay Liners
Desiccation of GCLs due to thermal gradients has been first examined and reported for cover of landfills
(Melchior, 2002). Desiccation of cohesive soils occurs as a result of dehydration and moisture loss. The
physical theory of moisture movement in porous media under temperature gradients was first described
by Philip and de Vries (1957) considering conservation of heat and mass of liquid. The theory was later
improved to consider conservation of momentum and mass of gas, as well as the effect of the presence of
chemical species such as salt or contaminants (e.g. Olivella et al., 1994 and Thomas and He, 1997).
Four main mechanisms are involved in the desiccation of the GCL component in a CGCLS under
thermal gradients: 1) downward heat transfer over the entire CGCLS and subsoil along the temperature
gradient created by high surface temperatures and relatively constant, lower temperatures in the aquifer;
2) downward vapour transfer from GCL and top layers of subsoil to cooler areas near the aquifer as a result
of the increase in concentration of vapour; 3) upward flow of water from subsoil to GCL because of the
increase in suction potential in the GCL, hence partially countering progressive dehydration; and 4)
shrinkage of the GCL and development of tensile stresses. Desiccation of GCL occurs if the developed
stresses exceed the tensile strength of the GCL. An important factor, accompanying these processes, is the
decline in the hydraulic conductivity of the bentonite in the GCL as a result of the drop in moisture content,
which slows down the ability of the subsoil to rehydrate the GCL (point 3 above).
Theories of GCL desiccation due to thermal gradient have been developed as an extension to
desiccation of compacted clay liners in composite lining systems. One of the pioneer studies on simulation
of CCL desiccation was conducted by Döll (1997) who developed a computer code, called SUMMIT, based
on two mass balance equations coupling flow of liquid and transport of liquid vapour and heat following
the mechanistic approach developed by Philip and de Vries (1957) and Milly (1984). Although the model
developed by Döll (1997) is a significant advance, it is based on a rigid body assumption (no swelling or
shrinkage of clay). This is a significant limitation, given the high void ratio and compressibility of GCL which
may experience large deformations under thermal gradients (Southen and Rowe, 2011).
A model later proposed by Zhou et al. (1998), called ZR3, includes the effects of thermo-hydro-
mechanical coupling in unsaturated media and expresses the governing equations using four conservation
equations and four independent primary variables: displacement, capillary pressure, air pressure and
temperature. A number of researchers have implemented and extended ZR3 to simulate desiccation of
composite mineral liner systems including CCLs (Zhou and Rowe, 2003; Zhou and Rowe, 2005) and GCLs
(Southen and Rowe, 2011; Azad et al., 2012; Hoor and Rowe, 2013; Rowe and Verge, 2013) in composite
lining systems.
-16
The ZR3 model has made a significant contribution to the understanding of composite lining systems;
nevertheless, most of the studies based on this model have a number of limitations: 1) the model is
implemented in a 1D solver context which limits the model’s ability to estimate tensile stresses in GCLs, 2)
mostly Lloret and Alonso (1985) SWCCs is used which has a limited range of application especially near the
dry end of the curve.
A number of other coupled models are available that can potentially simulate the process of
desiccation such as those developed by Olivella et al. (1994) and Thomas and He (1997). Both of these
models have been developed for the investigation of the behaviour of high-level nuclear wastes
repositories in deep salt and rock formations. One of the advantages of the model of Olivella et al. (1994)
is the possibility of considering the osmotic effects on the process of desiccation – a potentially important
factor in considering the behaviour of GCLs in brine ponds, subject to leakage from the overlying
geomembranes.
In contrast to landfills, the use of GMB-GCL systems in brine ponds on coal-seam gas sites is relatively
recent. More importantly, the thermo-hydro-mechanical conditions prevalent in landfill liners differ to
those found in brine ponds in three crucial ways, each of which may increase the risk of drying and
desiccation.
First, water in brine ponds is exposed to solar radiation and temperatures on top of the liner can
exceed those found in landfills by far, reaching 70°C in brine ponds (Bouazza et al., 2014) and up to 90°C
in solar ponds with controlled-salt concentration (Alcaraz et al., 2016). By contrast, maximum
temperatures encountered in landfills are typically 50-60°C (Southen and Rowe, 2005a; Yeşiller et al.,
2005), though exceptions involving much higher temperatures can occur due to chemical reactions (e.g.,
Jafari et al., 2014). In addition, while high temperatures in landfills tend to decline as biodegradation slows
down, those in brine ponds may last much longer, hence subjecting the liner to a more persistent thermal
load (Bouazza et al., 2016). Second, the waste buried in landfills applies significant overburden pressures
(typically 200kPa or more) on the underlying liner which increases compressive effective stresses in the
bentonite and, in theory, reduces the likelihood of desiccation. In brine ponds, on the other hand, the
effective stresses experienced by the GCL are much lower and usually less than 50kPa. Third, if the brine
water comes into contact with the GCL’s bentonite, its salinity may induce osmotic suction in the clay
which can change the pace of drying.
Hence, each one of these three factors may increase the risk of desiccation and fissures may provide
preferential pathways of contaminant transport through defects in the geomembrane. Such defects can
be kept to a minimum through good site construction and management practices but cannot be entirely
prevented (e.g. Giroud and Bonaparte, 1989a; Giroud, 1997; Rowe, 1998; El-Zein, 2008; Rowe, 2012).
-17
Despite this, very little research has been conducted on the behaviour of GMB-GCL systems under
conditions similar to those encountered in brine ponds, i.e. high temperatures and low overburden stress
(Tables 2-4 and 2-5). Bouazza et al. (2014) studied the behaviour of GCLs under 70°C applied temperature,
with no overburden stress. Though, no desiccation was observed and this may be attributed to two factors.
First, the introduction of an air gap between the GMB and GCL reduced temperature significantly at the
top of the GCL. Second, a clay subsoil with high moisture content and low hydraulic conductivity under the
GCL might have slowed down the thermal dehydration of the GCL. In addition, the simulations presented
by Bouazza et al. (2014) considered heat and moisture but not deformations. Rowe and Verge (2013)
studied the effects of low overburden stress by conducting thermo-hydro-mechanical simulations of GCL
desiccation, albeit with applied temperatures of up to 44°C which is considerably lower than those that
may occur in brine ponds. The authors reported that risk of desiccation increases with larger temperature
gradients and lower overburden stress.
Table 2-4. Summary of selected laboratory experiments on effect of thermal gradients on composite lining systems
Height of Soil Column (m)
Bottom Temperature (°C)
Surface Temperature (°C)
Thermal Gradient (°C/m)
Subsoil Type
Load (kPa)
References Main Findings, Developments, and Limitations
25 Initial water content of subsoil and temperature

Southen and Rowe Silty
80.0 ? ? 0.25 to gradient were found to be two important
(2004) Sand
40 factors in desiccation of GCL.
Two large-scale laboratory experiments on two
15.0 20.0 24.8 different subsoils were performed. It was found
Southen and Rowe Silty
to 55.0 to 1.0 to that thermally driven redistribution of moisture
(2005a) Sand
95.0 29 29 content can cause desiccation of GCL
component in CGCLSs.
23.0 22.0 23
Azad et al. (2011) 100.0 to to N/A to N/A
42.5 23.0 66.5 The performance of double CGCLSs under
20 thermal gradient was investigated.
Silty
Azad et al. (2011) 100.0 30.5 45.0 0.258 to
Sand
61
22.0 0 Sand Effect of temperature gradients on hydration of

Barclay and Rayhani GCL from subsoil was investigated. It was found
2.0 to 22.0 0.257 to and
(2013) that surface temperatures can significantly
55.0 100 Clay affect hydration of GCL from subsoil.
Effect of air gaps in CGCLSs under high
temperature was investigated. It was found
Bouazza et al. (2014) 0.0 70.0 21.0 0.321 152.6 Clay that the presence of an air gap can significantly
reduce temperature exposed the GCL
component of a CGCLS.
-18
Table 2-5. Summary of selected publications on numerical simulation of desiccation of mineral liners
Thermal Gradient (°C/m)

Depth of Groundwater
Bottom Temperature
Surface Temperature
Analysis dimensions
Brine Pond
Load (kPa)
GMB/GCL
GMB/CCL
Landfill
Main Findings, Developments, and
(°C)
(°C)
(m)
References
Limitations
25 Findings: SWCC and permeability of CCL

5.5 to 2.1 to and subsoil material play important role in
Döll (1997)     N/A to 10 1D
14.5 2.7 desiccation of CCL. Limitations: Rigid body
40 assumption.
Developments: A fully coupled thermo-
Zhou and Rowe hydro-mechanical model proposed.
    N/A 35 28 1.1 6.4 1D
(2003) Limitations: The theory only examined
under rigid body assumption.
Findings: Performance of CCLs and GCLs are
examined under a wide range of landfill
20
Zhou and Rowe 4.01 1.8 to applications.
    N/A to 10 1D
(2005) to 5.5 7.5 Limitations: No-displacement (fixed)
50 boundary condition is used for surface of
lining system.
Southen and 15 to
    55 29 N/A 26 1D Limitations: Rigid body assumption.
Rowe (2005b) 70
Findings: Good agreement between
experimental results and predictions
Southen and 50 to
    55 29 N/A 26 1D achieved. Limitations: SWCC model by
Rowe (2011) 80 Lloret and Alonso (1985) was used which
has a limited range of application.
32 22 Developments: Performance of double
Azad et al. 24 to
    100 to to N/A 1D layer composite lining systems are
(2012) 66.5 examined.
45 24
Findings: The placement of GCLs directly
35 over coarse-grained soils or drainage layers
Hoor and Rowe 50 to 6.7 to
    to 10 3.75 1D should be evaluated carefully. Limitations:
(2013) 100 12.0 Development of compressive stresses
55
during hydration stage are not considered.
Findings: SWCC model by Lloret and Alonso
(1985) can lead to poor desiccation
estimates. Brine pond conditions has the
38 3.75
Rowe and 10 to 3.1 to potential for GCL desiccation in composite
    to 10 to 1D
Verge (2013) 100 11.2 lining systems. Limitations: Development of
52 9.00 compressive stresses during hydration
stage are not considered. Tensile strength
of GCL is not considered.
Findings: No desiccation observed for the
GCL installed on a clay subsoil with 52%
saturation degree under the high thermal
Bouazza et al.
    N/A 70 21 N/A 152.6 2D gradient applied; however, introduction an
(2014) air gap on top of GCL significantly limited
the temperature rise at GCL. Limitations:
Rigid body assumption.
Bouazza et al. (2017) experimentally compared different liner setups, through column studies, in
terms of the extent to which surface temperatures, under brine pond conditions, are transmitted to the
liner and found that installing a primary geomembrane with an air gap, over a secondary geomembrane
with a GCL underneath it, may significantly reduce the thermal gradient applied to the secondary liner.
-19
However, the thermal protection was found to disappear if high-temperature water leaks through a hole
in the geomembrane. In addition, the authors only measured temperature distribution and water content
in the GCL before and after heat is applied. Hence, their results cannot be used for validating numerical
models.
2.6 Shear Strength of GCL
A typical relationship between shear stress and displacement of a hydrated GCL is presented in Figure
2-6-a. The shear stress reaches a peak strength (𝜏𝑝 ) with a relatively small displacement (𝛿𝑝 ) of less than
50mm. Shear strength reduces as displacements exceed 𝛿𝑝 and ultimately becomes constant at a
minimum shear strength (𝜏𝑟 ). The corresponding displacement to the minimum shear strength may be as
large as 0.1 to 0.5 m or more which depends on the normal stress and the material. Figure 2-6-b shows
peak and residual shear strength envelopes for a hydrated needle-punched (NP) GCL with a woven (W)
cover and nonwoven (NW) carrier. The shear strength parameters for the peak envelope are 𝑐𝑝 = 98.2
kPa and 𝜙𝑝 = 32.6° (Fox and Stark, 2015).
Shear strength envelopes of GCLs depend on a number of factors such as normal stress during
hydration (𝜎ℎ ), GCL configuration, and shear displacement rate. Table 2-6 presents shear strength
parameters for needle-punched GCLs. The shear strengths are based on linear failure envelopes. As a
result, the peak cohesion for the NPTT and NP GCLs varies from 17.2 to 23.3 kPa and from 42.3 to 98.2
kPa, respectively. Although, Chiu and Fox (2004) reported a zero peak cohesion based on a nonlinear
regression envelope for a large data base of internal shear strength for NP GCLs.
(a)
(b)
Figure 2-6. Shear strength of GCL (Fox and Stark, 2015): a) Typical shear stress-displacement behaviour of hydrated GCLs; b)
Peak and residual shear strength envelopes for a hydrated W/NW NP GCL
-20
Table 2-6. Failure envelopes for hydrated needle-punched GCLs
𝜎ℎ 𝑐𝑝 𝜙𝑝 𝑐𝑟 𝜙𝑟
Reference GCL type Comments
(kPa) (kPa) (Degrees) (kPa) (Degrees)
Zorenburg et al. (2005) NPTT-W/NW 𝜎𝑛 17.2 28.3 N/A N/A 7.2 < 𝜎𝑛 < 103
Zorenburg et al. (2005) NPTT-W/NW 𝜎𝑛 23.3 23.8 12.3 9.8 7.2 < 𝜎𝑛 < 575
Fox and Stark (2015) NP-W/NW 𝜎𝑛 42.3 41.9 1 4.7 𝐹𝑝 = 850 𝑁/𝑚
Fox and Stark (2015) NP-W/NW 𝜎𝑛 98.2 32.6 1 4.7 𝐹𝑝 = 1600 𝑁/𝑚
NPTT: Needle punched thermally treated; W: woven; NW: non-woven; 𝜎ℎ = 𝜎𝑛 means that the normal stress during
hydration and shearing are the same; 𝜎𝑛 : normal stress; N/A: not applicable; 𝐹𝑝 : the peel strength.
2.7 Soil-Water Characteristics Curves
Soil water characteristic curves (SWCCs), also referred to as soil water retention curves or soil water
release curves, are defined by ASTM D6836 (2008) as a relationship between, on the one hand, soil suction
and, on the other hand, volumetric water content (VWC), gravimetric water content (GWC), or degree of
saturation. Suction potential is given as a sum of four components of matric, osmotic, gas, and gravitational
potentials (Gens, 2010). Matric suction is defined as the difference between gas and liquid pressures and
reflects both capillary rise and physico-chemical forces binding water to soil particles. Osmotic suction
occurs because of difference of dissolved salt concentrations over a semi-permeable membrane. The gas
potential relates to the applied gas pressure and elevation differences defines gravitational potentials. It
has been shown that only the internal components of suction potential, matric and osmotic suction, have
important effects on the mechanical responses of soil (Gens, 2010). Hence, total soil suction is usually
expressed as the sum of matric suction and osmotic suction, only.
The SWCC of a GCL is an important function that partly determines its hydraulic behavior, volume
change, and strength. The range of suctions in a GCL is wide and a large number of different methods have
been used to measure or estimate the SWCCs of GCLs. Experiments on different soils show that soil suction
can vary from zero at a fully saturated state to hundreds of MPa at an oven-dry state (105°C) (Wang et al.
2016). In a recent study, Schneider and Goss (2012) suggested an average suction of nearly 620MPa
(negative pressure of 6.3×106cm height of water) for oven dry soils. Although, earlier research considered
a suction slightly below 1000 MPa at zero relative humidity (Fredlund & Xing, 1994). Beddoe et al. (2011)
measured total suctions as high as 700MPa for a dry GCL.
A number of SWCC measurement methods have been used for different types of soils and porous
construction materials; however, the proper application of each method is limited to a specific range of
suctions. In addition, different methods have different capabilities in terms of the ability to apply and
control a specific condition (e.g., applying normal stress or confinement stress) and/or measuring or
controlling additional parameters other than water content and suction (e.g., volume change).
-21
The commonly-used suction measurement techniques are broadly divided into two main categories:
a) direct measurement of matric suction and b) indirect measurement of matric, osmotic, or total suction.
In the direct methods, some devices such as suction plates, pressure plates, membranes and tensiometers
have been widely used to directly measure the liquid pore pressure. Usually a ceramic disk is required to
separate liquid and gas phases for direct methods and the maximum value of suction is limited to the air-
entry of the ceramic disk used. An example of these methods is the axis-translation technique (Agus and
Schanz, 2005; Mancuso et al., 2012). On the other hand, in the indirect methods, soil suction is determined
based a correlation between relative humidity or water content and properties of a sensor or soil such as
thermal conductivity and capacitance (psychrometer) (Cardoso et al., 2007; Bulut and Leong, 2009),
temperature of condensation (dew point potentiameter) (Campbell et al., 2007), mass (filter paper),
permittivity and dielectric constant (time domain reflectometry) (Papa and Nicotera, 2012; Carravetta et
al., 2012).
Many sources of errors have been reported for measurements of total suction by means of the
indirect methods. Agus and Schanz (2007) discussed inaccuracy of several commonly used sensors and
found that at low suctions, errors are high because of the nature of the sensors and their sensitivity to
temperature fluctuations (Figure 2-7). Therefore, the direct methods are recommended for suctions lower
than 1000kPa. Schelle et al. (2013) compared accuracy and capability of four different methods
comprising: 1) hanging column (the suction plate); 2) axis-translation (the pressure plate); 3) HYPROP
(b)
(b)
(a)
Figure 2-7. Errors in total suction measurements (adopted from Agus and Schanz, 2007): a) Device error; b) Estimate of error
because of temperature fluctuations
-22
simple evaporation method (tensiometer); and 4) WP4C Dewpoint PotentiaMeter for moderate to dry soil
moisture range. They concluded that the combination of HYPROP and WP4C Dewpoint PotentiaMeter
provides acceptably accurate and relatively fast results for the entire range of soil suction. Though, in these
methods, it is not possible to apply load to the specimen nor measure the volume change of a specimen.
SWCC of GCL was studied by Barroso and Touze-Foltz (2006) using the contact filter paper (CFP)
method. Both CFP and non-contact filter paper (NCFP) methods were later successfully applied to study a
wide range of suctions from 10kPa to 100MPa (Acikel et al. ,2011; Hanson et al., 2013; Acikel et al., 2015;
Risken et al., 2016). In the CFP method, it was found that capillary break was more likely to occur while
filter paper is in contact with a non-woven geotextile than a woven geotextile (Acikel et al., 2015).
Seiphoori et al. (2016) performed experiments on reconstituted GCLs using NCFP and a dew point method
to investigate a range of suctions from 20kPa to 110MPa (under no load) and concluded that the SWCC of
GCL was entirely controlled by the bentonite at suctions higher than 2MPa.
Measuring SWCCs over the entire range of suctions in a GCL is a challenging task. Table 2-7 presents
a summary of recent studies on SWCC of GCLs. The composite structure of the GCL and the possibility of
capillary breakage because of the presence of cover and carrier geotextiles considerably reduces the range
of applicability of the direct methods to measure suction. For example, Hanson et al. (2013) limited
measurements by the axis-translation method to suctions from 0 to 300kPa. In addition, Southen and
Rowe (2007) reported scatter measurements of matric suction in one set of their experiments because of
poor contact and capillary breaks for high suctions using the axis-translation method.
On the other hand, a number of experiments have been successfully conducted by measuring relative
humidity using relative humidity sensors (Beddoe et al., 2010; Abuel-Naga and Bouazza, 2010; Beddoe et
al., 2011; Hanson et al., 2013) or controlling relative humidity using vapor equilibrium techniques (VET)
(Rouf et al., 2014; Rouf et al., 2016b). Rouf et al. (2014) reported that the time to reach equilibrium
depends on a number of factors and it may exceed 1 month for VET. In addition, the methods based on
relative humidity are reliable only for the dry end of the SWCC curve. Practically, the lower limit for relative
humidity sensors is 5MPa and VET has been used for suctions of 3MPa and higher (Rouf et al., 2016b).
Other methods such as the filter paper method or the dew point method have been used to cover this
gap. Nevertheless, these methods are subject to some limitations such as incapability to apply load and
measurement of volume change.
-23
Table 2-7. A summary of recent studies conducted on SWCC of GCLs.
Temperature (°C)
Range of suction
Bentonite type
Measurement
Load (kPa)
Permeant
GCL type
method
Path
References Main Findings
Barroso and The filter paper method was

NP, 10kPa to
Touze-Foltz GB, PB 0 ? W CFP WP used for measurement of a
AB 5.5MPa
(2006) wide range of suction.
The air-entry of GCLs found to
Southen and 0, 0.5, 10kPa to
NP GB, PB ? DW AT DP be between 50 and 200kPa
Rowe (2007) 3, 100 9MPa
depending on the applied load.
0 to
1.5MPa, New techniques developed
Beddoe et al. NP, HCT,
GB 2 ? W DP based on HCTs and CRHs to
(2010) NPTT RHS 10MPa to
evaluate SWCC of GCL.
350MPa
Abuel-Naga
300kPa to A new moisture content control
and Bouazza NPTT PB 50a 25 W TP, RHS WP
80MPa method developed.
(2010)
Acikel et al. CFP, Applicability of filter paper
NP PB 0 20 W ? WP
(2011) NCFP method was evaluated.
Wetting and drying curves of
four different types of GCLs
0 to
were established.
Beddoe et al. NP, HCT, 500kPa, WP,
GB 2 22 W Manufacturing method and
(2011) NPTT RHS 5MPa to DP
configuration of geotextiles
700MPa found to have significant effect
on SWCC of GCLs.
30kPa to
Effect of cycles of wetting and
300kPa,
drying was investigated on
Hanson et al. AT, CFP, 10kPa to WP,
NP GB ? ? DI SWCC of three different GCLs
(2013) RHS 100MPa, DP
on both wetting and drying
5MPa to
curves.
400MPa
Vapour equilibrium technique
was used to establish SWCC of
GCL under two loads of 0 and
20kPa. Void ratios were the
same for suctions higher than
Rouf et al. 3MPa to WP, 110MPa; however, specimens
NP GB 0, 20 20 WV VET
(2014) 357MPa DP with 0 load showed higher void
ratios under suctions lower
than 110MPa. The time to reach
equilibrium depends on a
number of factors and it
generally exceeds 1 month.
Bannour, 10, 50, 0.1, 0.5, No change in air-entry and
VET,
Stoltz, et al. NP GB 100, 21 WV, W 2.8, 4.2, WP slope of SWCC observed under
OST
(2014) 200 8.5MPa different applied loads.
Insignificant change in
Bannour, GB, 0, 50, 4.2, hydration of two types of GCLs
Barral, et al. NP, ST 20 WV VET WP
PB-Ca 100 8.5MPa was observed under high
(2014)
suctions.
-24
Table 2-7 (continued). A summary of recent studies conducted on SWCC of GCLs.
Temperature (°C)
Range of suction
Bentonite type
Measurement
Load (kPa)
Permeant
GCL type
method
Path
References Main Findings
Upper theoretical suction limits

of 66 and 146kPa were
established for initially dry and
initially wet CFP method,
Acikel et al. CFP, 10kPa to respectively, based on pore size
NP GB 0 22 SDW WP
(2015) NCFP 100MPa distribution of Whatman no. 42
filter paper. It was found that
capillary break was more likely
to occur while in contact with a
non-woven geotextile.
A comparison of SWCC of
reconstituted GCL and SWCC of
a GCL from literature was
Seiphoori et al. DPC, 20 kPa to
RCON GB 0 ? WV WP performed. It was found that
(2016) NCFP 110MPa
SWCC of GCL was entirely
controlled by bentonite at
suctions higher than 2MPa.
30kPa to
300kPa, Effects of different permeant on
DI,
Risken et al. AT, CFP, 10kPa to WP, SWCC, permeability, and shear
NP GB 2 ? TAP,
(2016) RHS 100MPa, DP strength of GCL were
CaCl2
5MPa to investigated.
400MPa
For RH≤97.7%, the effect of GCL
manufacturing method on
Rouf et al. NP, 3MPa to WP, SWCC was insignificant.
GB, PB 0, 20 20 WV VET
(2016b) NPTT 357MPa DP However, bentonite form and
mineralogy were found to be
important.
a: Confining pressure, GB: Granular Na-bentonite; PB: Powder Na-bentonite; PB-Ca: Powder Ca-bentonite; NP: Needle-punched; NPTT:
Needle-punched thermally-treated; ST: Stitch-bonded; AB: Adhesive-bonded; RCON: Reconstituted GCL; DW: Distilled water; DI:
Deionized water; W: water, WV: Water vapour; SDW: Sterile distilled water; TAP: Tap water; CaCl2: 0.1M CaCl2; CFP: Contact filter paper;
NCFP: Non-contact filter paper; WP: Wetting path; DP: Drying path; AT: Axis-translation; HCT: High-capacity tensiometer; RHS: Relative
humidity sensor; VET: Vapour equilibrium technique; TP: Thermocouple psychrometer; OST: Osmotic technique; DPC: Dew point chilled-
mirror potentiometer
SWCCs often exhibit hysteretic behaviour, with a non-unique relationship between soil suction and
degree of saturation under cycles of wetting and drying. Hysteresis in SWCC can be explained, at a pore
scale, by the change in the curvature of the meniscus as one moves between hydration or dehydration
paths at different points of the SWCC curves (Huang et al. 2016). Beddoe et al. (2011) and Hanson et al.
(2013) studied SWCC on wetting and drying paths for different GCL types. It was found that manufacturing
method and configuration of geotextiles can significantly affect SWCC of GCLs and the magnitude of
hysteresis between wetting and drying curves (Beddoe et al., 2011).
-25
2.8 Volume Change and Temperature Effects in Water Retention of GCLs
Most of the popular SWCC models such as van Genuchten (1980) and Fredlund & Xing (1994) are based
on the assumption that temperature and applied stress are constant. It has been well-established that the
SWCC of a soil alters as a result of change in volume (e.g. Khalili et al., 2008; Tarantino, 2009; Gallipoli,
2012; Russell, 2014; Huang et al., 2016) and change in temperature (e.g. Romero et al., 2001; Ye et al.,
2009; Zhou et al., 2014; Gao and Shao, 2015; Roshani and Sedano, 2016). Therefore, it is very important
to quantify the change in the SWCC of a GCL as a result of a change in void ratio and temperature, because
composite lining systems are expected to be exposed to variable loads and temperatures in practice.
The effect of applied stress and void ratio have been investigated for a wide range of soils and a
number of approaches have been developed to include these effects in SWCCs. For example, Tarantino
(2009) reported a significant change in the air-entry suction from about 30 to 150kPa for a silt under zero
vertical stress and normally consolidated to 100, 300, 500kPa along the drying path (Figure 2-8-a). Gallipoli
(2012) proposed a modified version of van Genuchten (1980) SWCC and introduced a power function for
the evolution of the air-entry suction due to void ratio change. The power nature of the relationship
between air-entry suction and void ratio was later confirmed mathematically and validated experimentally
for fractal soils by Russell (2014). Zhou and Ng (2014) extended the model by Gallipoli (2012) to consider
the effects of pore structure using a micro-structural state variable.
Currently, the effect of volume change of a GCL on the evolution of its SWCC is not well understood,
with only two studies conducted so far, one by Southen and Rowe (2007) along a drying path with suctions
between 10kPa to 9MPa and another one by Bannour et al. (2014) along the wetting path with suctions
between 100kPa and 8.5MPa. Bannour et al. (2014) investigated the effects of loads of 0, 50, and 100kPa
on the SWCC of a GCL using a combination of osmotic techniques and VET. In their experiments, no change
(a) (b)
Figure 2-8. Evolution of SWCC with deformation of soil: a) Experiments under wetting and drying paths (Tarantino, 2009); b)
Idealized model (Gallipoli, 2012)
-26
in the air-expulsion or slope of the SWCC was observed under different applied loads. On the other hand,
Southen and Rowe (2007) used the axis-translation technique under loads of 0, 0.5, 3.0, and 100kPa and
found that the air-entry suction of GCLs varied between 50 and 200kPa depending on the applied load.
The reason for the difference between the findings of Southen and Rowe (2007) and Bannour et al.
(2014) may be related to the difference in paths investigated. In other words, it may be that, as shown in
these studies, air-entry suction does depend on applied load along the drying path, but air-expulsion
suction does not, along the wetting path. It is to be noted that, as mentioned earlier, experiments by
Tarantino (2009) on both wetting and drying paths show that both air-entry and air-expulsion values
change with consolidation stress (Figure 2-8-a).
Grant and Salehzadeh (1996) developed a chemical-thermodynamic explanation for the effects of
temperature on SWCC for non-deformable media – based on the model of Philip and de Vries (1957) for
liquid-gas interfacial tension – and implemented it in the van Genuchten (1980) SWCC. In this model, the
effects of temperature are assumed to be only due to a change in the surface tension of pure water.
However, experiments indicate that other processes may be involved such as change in inter-particle bond
strength, particle volume and pore water volume, soil fabric, and intra-aggregate fluid chemistry (Romero
et al., 2001; Gao and Shao, 2015). Figure 2-9 shows the effect of temperature and initial density on SWCC
of Boom clay (Romero et al., 2001). The model by Grant and Salehzadeh (1996) has been implemented in
more comprehensive THM theories considering effects of both deformation and temperature on SWCC
such as the non-isothermal model by Zhou et al. (2014), known as non-isothermal SFG model, based on
the original framework proposed by Sheng et al. (2008), known as SFG model.
Figure 2-9. Effect of temperature and density on SWCC of Boom clay (Romero et al., 2001)
-27
Despite the large number of experiments on SWCC of GCLs, the only systematic investigation of the
effects of temperature appears to have been conducted by Risken (2014) using a filter paper method under
2kPa load and temperatures of 2, 20, 40°C. It was observed that temperature had a small effect on air-
entry value of GCL along the wetting path, with water retention of GCL decreasing with increasing
temperature. However, no effect was observed on the drying path. Risken (2014) concluded that the
decrease in the retention capacity of GCL may be attributed to a reduction of water surface tension with
increasing temperature. In addition, it was suggested that complications associated with filter paper
calibration under conditions other than 20°C and 0kPa may be responsible for masking the effect of
temperature on SWCC of GCL.
2.9 Defects in Geomembrane
Defects in GMB are commonly reported in industrial applications (Touze-Foltz et al., 2008; Touze-Foltz et
al., 2016) and leakage and contaminant transport through the defects have been extensively studied (e.g.
El-Zein, 2008; Rowe et al., 2009; El-Zein et al., 2012; Rowe and Abdelatty, 2013; Rowe and Ewais, 2014).
The number of defects per installation area, known as density of defects, depends on the quality of
installation, surface preparation, materials, quality assurance – quality control (QA/QC). Based on a
number of studies on landfills, reservoirs, and surface impoundments, a frequency of 3-5 seam defects per
hectare is expected with a high QC and it is reported that the size of defects can vary from an equivalent
of 1-3mm diameter circle for seam defects to larger 10mm diameter holes for accidental punchers (Giroud
and Bonaparte, 1989a). Nevertheless, no data is available for density and size of defects in brine ponds on
coal seam gas sites.
Leakage rate through defects has been studied extensively (e.g. Touze-Foltz et al., 2008). In practice,
the regulatory limit on leakage rates depends on regional standards and guidelines and the limit can
depend on a specific-site condition. For example, US EPA (1992) proposed a leakage rate of 50 to 200
litre/day/hectare (lpdh), known as action leak rate, to trigger an action response to prevent contamination
of environment because of exceedance of the leakage limit.
Conversely, experiments and simulations of thermally-driven desiccation of GCL, found in the

literature, almost always assume intact GMBs above the GCL, idealised as perfect hydraulic barriers and
free of defects. Hence, as temperature gradients trigger a process of dehydration in the GCL, the only
source of rehydration is assumed to be the underlying subsoil. Though, in reality, the presence of defects
has three important effects. First, it provides another possible source of rehydration for the GCL. Second,
if the brine solution comes into contact with the GCL, chemical modification of the bentonite may occur
which could lead to significant changes in the hydraulic properties of the GCL (Yesiller et al., 2014). This
-28
may include a change in the SWCC of the GCL. Third, presence of salt in brine may impose osmotic suctions
on GCL and change the rate of flow of liquid. The first effect can be investigated by considering the effect
of water leakage from one or more defects in the geomembrane on the water content of a GCL undergoing
thermal dehydration. The second and third effects require thorough investigations of the effect of brine
solutions on GCLs which are beyond the scope of this thesis. Nonetheless, the experimental, theoretical
and numerical investigations reported here – based on an assumption of no contact between brine and
GCL – are necessary prerequisites for any research taking such contact into account.
2.10 Summary and Gap Identification
A review of engineering applications of the composite geosynthetic clay lining systems was presented in
this chapter. The large number of studies that have been conducted on the performance of composite
lining systems show that exposure of lining systems to thermal gradients results in desiccation of GCL and
significant loss of performance. In addition, it has been shown that the higher the temperature, the greater
the risk of desiccation. On the other hand, it was found that temperature applied on top of the GCL was
limited to less than 60°C in the existing experimental and numerical studies. Therefore, there is a
knowledge gap in relation to the performance of these lining systems under temperatures higher to 60°C,
and up to 90°C encountered in industrial applications such as brine ponds on coal seam gas sites.
Furthermore, processes and mechanisms of hydration and desiccation of GCL were described in this
chapter. The SWCC of GCL was identified as an important constitutive equation. Despite the significant
number of investigations of SWCC of GCLs, there has been a limited number of systematic studies
conducted on the effects of temperature and void ratio on SWCC of GCL.
-29
3 Research Design
Chapter 3
Research Design
The research design is based on four main tasks:
1. experimental determination of SWCC of GCL, and other material properties;

2. column studies simulating a GMB-GCL system over a subsoil, subjected to high surface
temperature and low overburden load;
3. development and validation of a THM theory for hydration and desiccation of GCL component in
CGCLSs, and
4. parametric analyses using the developed THM theory to assess risk of desiccation of GCL.
In addition, the project conducted in this thesis is part of a larger project carried out jointly with
Monash University. Results from experiments conducted at Monash University on the same GCL product
(GEOFABRICS ELCOSEAL X2000) will hence be used (determination of material properties of GCL including
SWCC and column studies of GCL subject to temperature gradients). These experiments complement, but
do not duplicate, the experiments reported in this thesis. Finally, in Task 3, in addition to experimental
datasets developed as part of this thesis and those developed at Monash University, datasets found in the
literature, representing landfill conditions, are used as supplementary resources in the validation exercise.
Figure 3-1 illustrates the research design and connections between the research tasks.
-30
#1. Experimental:
SWCC and other #2. Experimental:
Material Properties Column Studies
for GCL and Subsoil
Column Studies and

Material Properties
#3. Theoretical: Performed at
SWCC of GCL
Develop and Validate a Thermo-hydro- Monash University
Performed at
mechanical Theory for Behaviour of
Monash University
CGCLSs under Brine Pond Conditions
Existing Column
Study Datasets
Found in the
#4. Numerical: Literature
Engineering Applications
Figure 3-1. Research design
3.1 Experimental Determination of Soil-Water Characteristic Curves of GCL and Subsoil

Material Properties
3.1.1 Aims
In Task #1, SWCCs of GCL under different loads and temperatures are measured. New versions of the
existing popular SWCC models of van Genuchten (1980) and Fredlund and Xing (1994) are proposed for
curve fitting. The SWCC investigation is conducted in two steps: 1) establish base SWCC for wetting and
drying paths under 20kPa overburden load and ambient temperature and 2) establish the effect of changes
in load and temperature.
3.1.2 Step 1
To cover the wide suction range of GCL, the SWCCs of GCL are measured using axis-translation technique,
dew point method, and vapour equilibrium technique. All the data based on vapour equilirium technique
are obtained from Monash University as discussed above.
In practice, GCLs experience hydration during installation and may dehydrate during operation
because of exposure to thermal gradients on coal seam gas sites. The wetting path of SWCCs governs
-31
hydraulic performance of GCLs during hydration; however, it is the drying path of the SWCCs that matters
while the GCLs dehydrate.
Given the possible hysteresis of SWCCs, different drying curves might apply for different initial water
contents, at the end of the hydration phase. Given the time-consuming nature of SWCC measurement,
establishing the full pattern of hysteresis would require a separate research project, possibly a PhD thesis
in its own right, and is beyond the scope of this project. Instead, the aim is to cover, as much as possible,
the full range of suctions for the drying curve, and establish at least part of the wetting curves starting
from as-received water content (𝑤 ≅ 9%), within the experimental and time constraints available. In
addition, measurements conducted as part of the project, are supplemented by others found in the
literature. Figure 3-2 shows an overview of the range of suctions covered in this thesis including all
temperature and load testing conditions. The different techniques used are the axis translation method,
the dew point method (using a Vapour Sorption Analyser, VSA), and the vapour equilibrium technique.
The axis-translation technique is used to derive the drying path of the SWCC for matric suctions
between 0 and 700 kPa, using two Barcelona cells. The upper limit of 700kPa is due to limitations in the
ability of the instrument to measure suction of a GCL (see Section 2.7 for more details). The dew point
method is used to measure total suctions in a range of 10 to 250MPa on both wetting and drying paths,
using two instruments, a W4PC and a Vapour Sorption Analyser (VSA), both manufactured by Decagon.
The upper limit of 10MPa is selected for this method for two reasons. First, measurements for suctions
Drying path measured in this thesis

Wetting path measured in this thesis
Drying path NOT measured
Wetting path NOT measured
Sr
1 10 100 1000 10000 100000 1000000

Suction (kPa)
Figure 3-2. Overview of suction range covered in this thesis including all methods and conditions.
-32
lower than 10MPa seem to carry considerable error (see Figure 2-7 in Section 2.7). Second, the required
time to reach equilibrium significantly increases for lower suctions. The upper limit of 250MPa is near the
capacity of the VSA intrument. The wetting path is measured with the VSA starting from the as-received
water content (𝑤 ≅ 9%) with relatively high total suction (≅ 100𝑀𝑃𝑎) and reaching 10MPa, under
controlled relative humidity conditons. Then the drying path is followed from 10MPa to 250MPa.
The measurements at Monash University were performed using the vapour equilibrium technique
(VET) to determine the wetting path for total suctions from as-received water content (with suction of ≅
100𝑀𝑃𝑎) to 1.5MPa and the drying path from as-received water content (≅ 100𝑀𝑃𝑎) to 415 MPa. In
other words, VET data is only available on the drying and wetting paths for suction higher and lower than
≅ 100𝑀𝑃𝑎, respectively.
3.1.3 Step 2
The effects of load and temperature are investigated by conducting further SWCC measurements under
higher temperatures and overburden loads, using the three methods. The Barcelona cells have been
designed to allow control of load and temperature up to about 40oC. On the other hand, no load can be
applied when using the WP4C and VSA instruments. Hence, loads of 10, 20 and 50 kPa and temperatures
of 20 and 40°C are applied to the specimens in further measurements using the axis-translation technique.
The dew-point method is used to measure SWCCs under temperatures of 0, 35 and 60°C and no applied
load. Finally, the VET is performed under loads of 0 and 20kPa and temperatures of 20 and 70°C. A detailed
description of the experimental program is presented in Section 5.3 and Table 5-20.
Moreover, in Task #1, other required material properties for a THM simulation are established such
as thermal conductivity, mechanical properties, and hydraulic conductivity. SWCC of subsoil is measured
using a simplified evaporation method (using the HYPROP device) only for the drying path and under zero
load and room temperature.
3.2 Column Studies
In Task #2, column studies are performed under conditions representing those found in brine ponds on
coal seam gas sites. The columns are instumented with TDRs, thermocouples, and linear variable
displacement transducers (LVDTs). The instumentation enables continuous monitoring of heat and
moisture flow inside the column and tracking settlements of surface of CGCLS. A well-graded sand is
selected for the subsoil. An initial subsoil water content that is able to hydrate the GCL to substantial
degree of saturation is identified. This is done through a number of pilot tests, in which the final water
contents of GCLs are measured, after placing them in contact with the subsoil at different initial water
contents.
-33
The column studies are conducted in three stages:
1) equilibriumn of subsoil with gravity (14 days);
2) hydration of GCL from subsoil (44 days); and
3) application of heat to the surface of the CGCLS (39 days).
In stage 1, the column is filled with subsoil with the target moisture content and compacted to the target
void ratio in layers of 50 mm soil. The column is sealed at the top and bottom and kept at 20°C for 2 weeks
to reach equilibrium with the gravitational field. Following this stage, the CGCLS is installed on top of the
column and a load of 20kPa is applied to the CGCLS under isothermal condition at 20°C. After 44 days of
hydration, heat is applied to the top and load is kept constant (20kPa) for an additional 39 days. Hence the
total legnth of the experiement is 97 days. The experiment is duplicated to test repeatability. A more
detailed description of the column studies is presented in Chapter 6.
3.3 Development and Validation of a Theoretical Thermo-Hydro-Mechanical Model
The theoretical component of the research, Task #3, comprises two main stages:
1) a thermo-hydro-mechanical (THM) theory for behaviour of CGCLSs under evaporation pond

conditions is formulated based on the work of Olivella et al. (1994) and
2) the theory is validated using various available experimental sets.
In the first stage, the data from Task #1 is used and new versions of SWCC of van Genuchten (1980)
and Fredlund and Xing (1994) are developed to consider the effects of temperature and void ratio based
on extensions of the models of Grant and Salehzadeh (1996) and Gallipoli (2012), respectively. Then the
THM theory of Olivella et al. (1994) is extended to include the temperature and void-ratio dependent
SWCCs. The details of theories and implementations in CODE_BRIGHT are discussed in Chapter 4.
Next, the validation of the theory is performed against the column studies results for brine pond
conditions obtained from Task #2, as well as a dataset developed at Monash University (Bouazza et al.,
2014) and two sets of existing column studies (Southen and Rowe, 2005a). The different datasets cover
different thermal and mechanical loadings, representing conditions encountered in both landfills (low
temperature and high overburden load) and brine ponds (high temperature and low overburden load)
3.4 Engineering Applications
In Task#4, the theory, implemented in CODE_BRIGHT, is used to simulate the hydration, dehydration and
desiccation of a composite liner over a subsoil, under brine-pond conditions, while assessing the effects of
-34
various environmental and design factors and material properties on the pace and extent of dehydration.
The aim of the simulations is to identify conditions under which desiccation of the GCL’s bentonite may
occur and, based on this, to assess performance of two alternatvie composite lining system designs and
guidelines for the safe operation of these liners.
3.5 Summary
The overall research design was presented in this chapter and its individual components described. To
assess the safety of composite geosynthetic clay lining systems for a wide range of applications, an
experimentally validated non-linear thermo-hydro-mechanical theory for desiccation of GCLs is developed.
The experimental part comprises two sets of column studies and a number of SWCC tests under controlled
temperature and overburden stress conditions. In addition, the THM theory by Olivella et al. (1994) is
extended to include two new void ratio and temperature-dependent SWCCs based on the van Genuchten
and Fredlund and Xing models.
-35
4 A Non-Linear Thermo-Hydro-Elastic Theory for Desiccation of GCL
Chapter 4
A Non-Linear Thermo-Hydro-Elastic Theory for
Desiccation of GCL
4.1 Introduction
Desiccation of cohesive soils subjected to a thermal load at the surface is believed to occur in two main
stages. First, progressive dehydration is driven by evaporation as a result of thermodynamic imbalance
between the clay and the surrounding environment. This leads to an increase in water vapour
concentration in the clay pores and causes downward flow of water vapour as a result of concentration
gradients – since upwards flow is prevented by the presence of the GMB on top of the GCL. As moisture is
lost negative water pressure develop and the resulting hydraulic gradients lead to a process of partial
rehydration with the water flowing upwards from the underlying natural soil back towards the GCL.
Although, this counteracting mechanism is attenuated by a reduction in the hydraulic conductivity of the
GCL and top layers of subsoil that accompanies the moisture loss and increase in suction. This makes it
increasingly harder for the underlying soil to rehydrate the GCL. On the other hand, rehydration by the
brine pond from the top is not desirable and essentially precluded by the presence of the GMB (except,
possibly through relatively infrequent and widely spaced defects in the GMB).
Second, as moisture is progressively lost from the GCL, bentonite has a tendency to shrink. While
shrinkage can occur freely in the vertical direction, it is significantly resisted in both horizontal directions,
as a result of the friction between the bentonite and adjacent layers (geotextiles, GMBs and subsoil).
Consequently, tensile stresses can develop in the bentonite. If dehydration continues, these stresses may
eventually exceed the tensile strength of the bentonite, leading to the formation of fissures and fractures.
This creates preferential paths for water and contaminants and increases the hydraulic conductivity of the
-36
GCL, hence significantly compromising its performance. While bentonite has a good capacity for self-
healing if rehydrated, such a process of self-repair depends on the availability of a suitable rehydration
liquid (i.e. water with low concentration of minerals such as calcium and magnesium) and on the extent of
fracturing, and may fail to bring back the liner to its required insulation capacity.
It is clear from this description that predicting desiccation of composite liners has to take into account
various couplings between thermal transfer, liquid water flow, gas and water vapour flow, as well as the
mechanical response of the soil. In addition, the applied thermal gradients and overburden load, as well
as the thermal, hydraulic and mechanical properties of both the GCL and the underlying soil, (including key
constitutive equations such as water retention and stress-strain curves) will be critical in determining the
risk of desiccation. In what follows, a multi-phase thermo-hydro-mechanical theory of unsaturated soils is
presented as a basis for predicting the likelihood of desiccation of composite liners subjected to large
thermal gradients. The theory is essentially a version of multi-phase theory developed in the 1990s by
Alonso et al. (1990), Oliveilla et al. (1994), and Thomas and He (1997) and later modified by Zhou and Rowe
(2003) to model desiccation of GCLs.
4.2 Governing Equations
Principally, the formulation of governing constitutive equations of non-isothermal consolidation of

unsaturated media is derived by considering the thermodynamic conservation equations such as mass,
heat, and momentum conservation (balance) equations over a representative volume element (RVE) (also
called representative elementary volume (REV) or the unit cell). Originally, Oliveilla et al. (1994) formulated
the problem in multi-phase and multi-species of three phases of solid phase, liquid phase, and gas phase
and three species of salt, water, and air. However, in this study, as Figure 4-1 illustrates, the species are
reduced to two species of water and air. Therefore, the composition of the three phases is:
1) solid phase: comprises soil grains

2) liquid phase: comprises liquid water and dissolved air
3) gas phase: comprises dry air and water vapour
The processes involved in desiccation can be simulated by four coupled, non-linear equations
obtained from statements of conservation of water, air, momentum and internal energy for the medium.
CODE_BRIGHT – a fully coupled multi-phase finite element code within a Cartesian coordinate system that
offers both 2D and 3D elements – is used as the simulation environment in this study. All models built in
this thesis were 2D axisymmetric. CODE_BRIGHT is based on the THM theory developed by Olivella et al
(1994) and Olivella et al (1996). The balance equations are described briefly here. Variable t is time in all
formulations shown in this chapter.
-37
Balance of momentum:
∇. 𝜎𝑇𝑜𝑡𝑎𝑙 + 𝑏 = 0 4-1
Balance of mass of water:
𝜕
(𝜔𝑤 𝜌 𝑆 𝜙 + 𝜔𝑔𝑤 𝜌𝑔 𝑆𝑟𝑔 𝜙) + ∇. (𝑗𝑙𝑤 + 𝑗𝑔𝑤 ) = 𝑓 𝑤 4-2
𝜕𝑡 𝑙 𝑙 𝑟
Balance of mass of air:
𝜕
(𝜔𝑎 𝜌 𝑆 𝜙 + 𝜔𝑔𝑎 𝜌𝑔 𝑆𝑟𝑔 𝜙) + ∇. (𝑗𝑙𝑎 + 𝑗𝑔𝑎 ) = 𝑓 𝑎 4-3
𝜕𝑡 𝑙 𝑙 𝑟
Balance of internal energy:
𝜕
(𝐸 𝜌 (1 − 𝜙) + 𝐸𝑙 𝜌𝑙 𝑆𝑟 𝜙 + 𝐸𝑔 𝜌𝑔 𝑆𝑟𝑔 𝜙) + ∇. (𝑖𝑐 + 𝑗𝐸𝑠 + 𝑗𝐸𝑙 + 𝑗𝐸𝑔 ) = 𝑓 𝑄 4-4
𝜕𝑡 𝑠 𝑠
where:
𝑏 = vector of body force
𝐸𝑎 = internal energy of gas phase
𝐸𝑙 = internal energy of liquid phase
𝐸𝑠 = internal energy of solid phase
𝑓 𝑄 = external supply of heat
𝑓 𝑎 = external supply of air
𝑓 𝑤 = external supply of water
Figure 4-1. Schematic of unsaturated porous media composed of solid, liquid, and gas phases containing (modified from
Olivella et al, 1996)
-38
𝑖𝑐 = conductive flux of heat
𝑗𝐸𝑎 = advective flux of heat by air mass motion
𝑗𝐸𝑙 = advective flux of heat by liquid mass motion
𝑗𝐸𝑠 = advective flux of heat by solid mass motion
𝑗𝑔𝑎 = total flux of air in gas phase
𝑗𝑔𝑤 = total flux of water in gas phase
𝑗𝑙𝑎 = total flux of air in liquid phase
𝑗𝑙𝑤 = total flux of water in liquid phase
𝑆𝑟 = degree of saturation of liquid phase
𝑆𝑟𝑔 = degree of saturation of gas phase
𝜌𝑔 = density of gas
𝜌𝑙 = density of liquid
𝜎𝑇𝑜𝑡𝑎𝑙 = tensor of total stress
𝜔𝑔𝑎 = mass fraction of air in gas phase
𝜔𝑔𝑤 = mass fraction of water in gas phase
𝜔𝑙𝑎 = mass fraction of air in liquid phase
𝜔𝑙𝑤 = mass fraction of water in liquid phase
𝜙 = porosity
For a detailed description of the balance equations, treatment of fluxes, and implementation of
boundary conditions, the reader is referred to Olivella et al. (1994), Olivella (1995), and Olivella et al.
(1996). Key assumptions of the theory are:
1. Dry air is considered a single species and it is the main component of the gaseous phase. Henry’s
law governs phase equilibrium in dissolved air.
2. Vapour concentration is in equilibrium with the liquid phase, in accordance with the psychrometric
law.
3. The state variables are solid displacements, liquid pressure, gas pressure, and temperature.
4. Balance of momentum for the medium as a whole is reduced to the equation of stress equilibrium
together with a mechanical constitutive model to relate stresses with strains.
-39
5. Small strains and small strain rates are assumed for solid deformation; advective terms due to
solid displacement are neglected.
6. Balance of momentum equations for dissolved species and for fluid phases are reduced to
constitutive equations, Fick's law and Darcy's law, respectively.
7. Physical parameters in constitutive laws (e.g., concentration of vapour, surface tension, dynamic
viscosity) are pressure and temperature dependent.
Key constitutive relationships are described in the next sections of this chapter and a table of
interactions of thermo-hydro-mechanical processes for an unsaturated media is presented in Appendix E.
4.3 Soil-Water Characteristic Curves
Soil-water characteristic curves show the relationship between soil suction and moisture content. The
earlier empirical equations such as van Genuchten (1980) and Fredlund and Xing (1994) were proposed
for the conventional conditions of SWCC measurements under constant vertical stress and constant
temperature. However, it has been observed that changes in void ratio and temperature alters the suction
and moisture content relationship. Therefore, it is important to consider these effects on the SWCC of GCL
because GCLs experience relatively large void ratio change as a result of change in suction and load. In
addition, CGCLS used for basal protection layer in brine ponds are exposed to significant temperature
elevation.
Gallipoli (2012) presented a model based on van Genuchten SWCC to consider the effect of change in
void ratio on SWCC of deformable media. The model by Gallipoli (2012) is established and evaluated for
SWCC experiments under constant volume (i.e. constant void ratio). In the model, effect of void ratio
change is expressed as a power function for a parameter that is a representative of air-entry value in van
Genuchten’s SWCC. This method has been used in other SWCC equations such as Russell (2014), where
the SWCC is explicitly expressed in terms of the air-entry value. In addition, Grant and Salehzadeh (1996)
implemented a chemical-thermodynamic model in van Genuchten (1980) to consider the effect of
temperature under non-deformable condition based on the theory of Philip and de Vries (1957). A detailed
review of literature on this topic was presented in Section 2.8.
In this section, implementaion of combined theories of Grant and Salehzadeh (1996) and Gallipoli
(2012) in van Genuchten (1980) SWCC is described. In addition, application of the same approach to the
Fredlund and Xing (1994) SWCC is presented.
-40
4.3.1 Approach based on van Genuchten (1980) (VG version)
The relationship of soil suction and volumetric water content is expressed by van Genuchten (1980)
as:
𝜃𝑠 − 𝜃𝑟
𝜃 = 𝜃𝑟 + 4-5
(1 + |𝛼ℎ|𝑛 )𝑚
where:
𝜃 = volumetric water content
𝜃𝑠 = saturated volumetric water content
𝜃𝑟 = residual volumetric water content
ℎ = pressure head, (m)
𝑛 = model parameter that controls slope of curve
𝛼 = model parameter representative of air-entry value, (1/m)
and 𝑚 is a parameter that controls slope of the curve. It is defined as:
1
𝑚 =1− 4-6
𝑛
van Genuchten SWCC can be expressed in terms of effective degree of saturation and matric suction
by the following equation:
1 −𝜆
𝑆𝑟 − 𝑆𝑟𝑠 𝑠 1−𝜆
𝑆𝑒 = = (1 + ( ) ) 4-7
𝑆𝑠 − 𝑆𝑟𝑠 𝑃
where:
𝑆𝑒 = effective degree of saturation of liquid phase
𝑆𝑟 = degree of saturation of liquid phase
𝑆𝑟𝑠 = residual degree of saturation of liquid phase
𝑆𝑠 = maximum degree of saturation of liquid phase. In this study, it is always assumed that 𝑆𝑠 = 1.0.
𝑃 = model parameter that is representative of air-entry value, kPa
𝑠 is matric suction defined as:
-41
𝑠 = 𝑃𝑔 − 𝑃𝑙 4-8
where:
𝑃𝑔 = gas pressure (compression is positive), kPa
𝑃𝑙 = liquid pressure (compression is positive), kPa
and 𝜆 is a parameter that controls the shape of the SWCC in 𝑆𝑒 - ln 𝑠 space and is defined as:
1
𝜆 =𝑚=1− 4-9
𝑛
therefore, the relationship between 𝑃 and 𝛼 is:
𝜌𝑤 𝑔 × 10−3
𝑃= 4-10
𝛼
where:
𝜌𝑤 = density of water at reference temperature, it is 998 kg/m3 at 20°C
𝑔 = 9.81 m/s2
Following the method proposed by Grant and Salehzadeh (1996), it is possible to modify equation 4-7
to include effect of temperature as a result of change in surface tension:
−𝜆
1
1−𝜆
𝑠
𝑆𝑒 = 1+( 4-11
𝜎 )
𝑃0 𝜎
0
( )
where:
𝜎 = surface tension at current temperature, N/m
𝜎0 = surface tension at reference temperature, it is 0.072 N/m at 20°C
𝑃0 = model parameter measured at a reference temperature, kPa
In addition, Grant and Salehzadeh (1996) calibrated a surface tension equation based on a linear
approximation proposed by Philip and de Vries (1957) with a coeffcient of determination (R2) of 0.99899
for temperatures between -10 and 50°C for pure water. Although, the error increases for higher
temperatures (Figure 4-2). The equation can be expressed by:
-42
𝜎 = 0.11766 − 0.0001535 (𝑇 + 273.15) 4-12
where:
𝜎 = surface tension, N/m
𝑇 = temperature, °C
In this study, a more accurate formulation proposed by the International Association for the
Properties of Water and Steam (IAPWS) was used as follows (IAPWS, 2014):
𝑇 + 273.15 1.256 𝑇 + 273.15

𝜎 = 0.2358 (1 − ) (1 − 0.625 (1 − )) 4-13
𝑇𝑐 + 273.15 𝑇𝑐 + 273.15
where:
𝑇𝑐 = 373.95°𝐶 , critical temperature
Figure 4-2 shows predictions of Grant and Salehzadeh (1996) and IAPWS (2014) against experimental
data for surface tension of pure water at different temperatures.
Gallipoli (2012) suggested a power relationship for 𝑃 based on an extension on the model proposed
by Gallipoli et al (2003) as follows:
𝑃 = 𝑃0 𝑒 −𝛽𝑒 4-14
where:
𝑒 = void ratio
0.080
0.075
0.070
𝜎(𝑁/𝑚)
0.065
0.060
Grant and Salehzadeh (1996)
0.055 IAPWS (2014)
Experimental (IAPWS, 2014)
0.050
0 20 40 60 80 100
T (°C)
Figure 4-2. Surface tension vs. temperature for pure water
-43
𝑃0 = model parameter measured at a reference void ratio of 𝑒0 , kPa
𝛽𝑒 = model parameter
In a very similar approach, Russell (2014) proposed a model for the air-entry value of fractal soils and
verified the model for a wide range of different soils based on constant volume SWCC tests. A governing
equation is presented for the evolution of air-entry value, 𝑠𝑎𝑒 , in the model by Russell (2014):
𝑠𝑎𝑒 = 𝛼𝑎𝑒 𝑒 −𝛽𝑒 4-15
where:
𝛼𝑎𝑒 = model parameter, kPa
𝑒 = void ratio
𝛽𝑒 = model parameter
Figure 4-3 shows the correlation between 𝑃/𝑃0 and void ratio for different values of 𝛽𝑒 . Gallipoli et
al. (2003) used a value of 𝛽𝑒 = 8.433 for a compacted kaolinite with void ratios from 0.6 to 1.3. In addition,
Russell (2014) suggested a range of 2.45 to 2.95 for 𝛽𝑒 for fractal soils with void ratios smaller than 1.5.
Considering equations 4-11 and 4-14, effects of change in temperature and void ratio on van
Genuchten SWCC can be expressed as:
1 −𝜆
1−𝜆
𝑠 4-16
𝑆𝑟 = 𝑆𝑟𝑠 + (𝑆𝑠 − 𝑆𝑟𝑠 ) 1 +
𝜎 𝑒 −𝛽𝑒
𝑃0 ( )
𝜎0 𝑒𝑟𝑒𝑓
( ( ) )
10
1
𝑃/𝑃0 = 𝑒 −𝛽𝑎𝑒
0.1
0.01
0 1 2 3 4 5 6
e
Figure 4-3. Correlation between 𝑃/𝑃0 and void ratio, e.
-44
where:
𝑒𝑟𝑒𝑓 = reference void ratio
4.3.2 Selection of 𝑒𝑟𝑒𝑓
Ideally, the SWCC measurements for the proposed SWCC should be performed at constant void ratio.
Nevertheless, it is not possible to conduct such tests on GCLs or clays for full range of suctions because of
a number of limitations such as small thickness of GCL and large volume change of bentonite as a result of
change in suction. In this thesis, the void ratio at the air-entry value is selected for the reference void ratio.
In general, selection of the reference void ratio depends on the calibration procedure. For example,
considering 𝑒𝑟𝑒𝑓 = 1.0 would change equation 4-16 to a simpler form. For GCLs, such a void ratio at air
entry suction requires a very large applied load and tends to yield a very large 𝑃0 which may not be easily
comparable with the 𝑃0 values available in this study or in the literature.
4.3.3 Extension to Fredlund and Xing (1994) (FX version)
The developed approach presented in the previous section can be applied to Fredlund and Xing’s SWCC.
The model by Fredlund and Xing (1994) in terms of volumetric water content is expressed as follows:
𝜃𝑠
𝜃 = 𝐶(𝑠) 𝑚𝑓
𝑠 𝑛𝑓 4-17
{ln [exp(1) + (𝑎 ) ]}
𝑓
where the correction factor 𝐶(𝑠) is:
ln(1 + 𝑠/𝜓𝑟 )
𝐶(𝑠) = 1 − 4-18
ln[1 + (106 /𝜓𝑟 )]
where:
𝑠 = suction, kPa
𝑎𝑓 = a parameter to be determined from air-entry value, kPa
𝑛𝑓 = a parameter to be determined by rate of water extraction from soil once air-entry value is
exceeded
𝑚𝑓 = a parameter to be determined by residual water content
𝜓𝑟 = soil suction corresponding to the residual water content, kPa.
-45
The SWCC is forced to be zero at a suction of 106 kPa by the correction factor, 𝐶(𝑠). As mentioned
earlier, assuming saturated degree of saturation of soil is 1.0 then equation 4-17 can be re-written in terms
of degree of saturation as:
𝑛𝑓 −𝑚𝑓
𝑠
𝑆𝑟 = 𝐶(𝑠) {ln [exp(1) + ( ) ]} 4-19
𝑎𝑓
Effect of temperature and void ratio can be incorporated in Eq. 4-19 following same approach that
described for van Genuchten’s SWCC in Section 4.3.1 by:
𝑠
𝑆𝑟 = 𝐶(𝑠) ln exp(1) + 4-20
𝜎 𝑒 −𝛽𝑒
𝑎𝑓0 𝜎 (𝑒 )
{ [ ( 0 𝑟𝑒𝑓 ) ]}
where:
𝛼𝑓0 = model parameter at reference 𝜎0 and 𝑒0 , kPa
4.4 Intrinsic Permeability for Liquid Phase
The relationship between the intrinsic permeability of the liquid phase, 𝜅𝑙 , and porosity is assumed to be
given by Kozeny’s model as:
𝜙 3 (1 − 𝜙0 )2
𝜅𝑙 = 𝜅0 4-21
(1 − 𝜙)2 𝜙03
where
𝜙 = the porosity
𝜙0 = the reference porosity where the reference intrinsic permeability, 𝜅0 , is measured
𝜅0 = the intrinsic permeability of the matrix at reference porosity, 𝜙0 , m2
where the intrinsic permeability (m2) is defined as:
𝜇𝑤
𝜅0 = 𝑘𝑠 4-22
𝜌𝑤 𝑔
where
𝑘𝑠 = the saturated hydraulic conductivity, m/s
-46
𝜇𝑤 = the dynamic viscosity of water, 0.001 𝑃𝑎. 𝑠 at 20°C
𝜌𝑤 = the density of water, 998.2 kg/m3 at 20°C
𝑔 = the gravitational acceleration, 9.81m/s2.
Figure 4-4 shows relationship between porosity and κl /κ0 for different reference porosities based on
the Kozeny’s model.
100
10
𝜅𝑙 /𝜅0
0.1
0.01
0.25 0.35 0.45 0.55 0.65
𝜙
Figure 4-4. Correlation between 𝜅𝑙 /𝜅0 and porosity based on the Kozeny’s model
4.5 Relative Hydraulic Conductivity for Liquid and Gas
The van Genuchten’s relative hydraulic conductivity model for the liquid phase can be defined as:
𝜆𝑘 2
𝑘𝑟𝑙 = √𝑆𝑒 (1 − (1 − 𝑆𝑒 1/𝜆𝑘 ) ) 4-23
where:
𝜆𝑘 = a fitting parameter for relative hydraulic conductivity.
The gas relative hydraulic conductivity is given by:
𝑘𝑟𝑔 = 1 − 𝑘𝑟𝑙 4-24
4.6 Flux of Vapour
Both advective and diffusive processes are considered in describing vapour transport. Fick’s law governs
diffusive flux of vapour:
𝑣𝑎𝑝𝑜𝑢𝑟
𝑖𝑣𝑎𝑝𝑜𝑢𝑟 = −(𝜏𝜙𝜌𝑤 𝑆𝑟 𝐷𝑚 𝐈)∇𝜔𝑔𝑤 4-25
-47
where:
𝑖𝑣𝑎𝑝𝑜𝑢𝑟 = diffusive flux of vapour
𝜏 = tortuosity
𝐈 = identity matrix
𝜔𝑔𝑤 = mass fraction of water vapour in gas phase
and molecular diffusion of vapour in the gas phase can be calculated from:
𝑣𝑎𝑝𝑜𝑟 (273.15 + 𝑇)𝑛𝐷

𝐷𝑚 = 𝐷( ) 4-26
𝑃𝑔
where:
𝑇 = temperature in degree Celsius,
𝑛𝐷 = a model parameter,
𝐷 = a model parameter, m2s-1K-nPa
Dispersive flux of vapour is not considered here because of lack of data.
4.7 Flux of Heat
Conductive heat flux driven by thermal gradients (Fourier’s law), as well as convective heat flux driven by
liquid, gas, and vapour flow are all incorporated in the model. The dependence of thermal conductivity on
the degree of saturation maybe given by (Olivella, 1995):
𝜆 𝑇 = 𝜆𝑑𝑟𝑦 (1 − 𝑆𝑟 ) + 𝜆𝑠𝑎𝑡 𝑆𝑟 4-27
where:
𝜆𝑑𝑟𝑦 = thermal conductivity of dry porous media, Wm-1K-1
𝜆𝑠𝑎𝑡 = thermal conductivity of water saturated porous media, Wm-1K-1
4.8 Non-linear Elasticity
The bentonite component in a GCL can experience very large strains due to suction changes during
hydration or dehydration. We selected a non-linear elastic model as the mechanical constitutive model.
This model is based on the elastic part of Barcelona Basic Model (BBM) (Alonso et al., 1990) and is capable
of representing volume changes due to suction and stress. The elastic volume change can be defined as:
-48
−𝜅 𝑑𝜎𝑛𝑒𝑡 −𝜅𝑠 𝑑𝑠 𝑑𝑠 𝑠 + 𝑃𝑎 𝑑𝜎𝑛𝑒𝑡
𝑑𝜀𝑣𝑒 = + + 𝑎3 [ln(−𝜎𝑛𝑒𝑡 ) + ln ( ) ] + 𝛼 𝑇 ∆𝑇 4-28
1 + 𝑒 𝜎𝑛𝑒𝑡 1 + 𝑒 𝑠 + 𝑃𝑎 𝑠 𝑃𝑎 𝜎𝑛𝑒𝑡
where:
𝜅 = slope of unload/reload curve in the 𝑒 − ln(𝜎𝑛𝑒𝑡 ) diagram
𝑠+𝑃𝑎
𝜅𝑠 = slope of unload/reload curve in the 𝑒 − ln ( ) diagram
𝑃𝑎
𝑎3 = a coupling parameter
𝛼 𝑇 = linear thermal expansion coefficient, 1/°C
𝑃𝑎 = atmospheric pressure, i.e. 100kPa
and
𝜎𝑛𝑒𝑡 = 𝜎𝑚𝑒𝑎𝑛 − max(𝑃𝑔 , 𝑃𝑙 ) 4-29
where:
𝜎𝑚𝑒𝑎𝑛 = the mean total stress
Note that, following geotechnical convention, compressive stresses and compressive liquid and gas
pressures are positive.
4.9 Numerical Implementation
Solution of the resulting system of nonlinear partial differential equations (PDEs), i.e. Eq. 4-1 to 4-4,
requires a numerical approach, which can be formulated in two parts of spatial and temporal
discretization. In this section, a summary of the numerical scheme used in CODE_BRIGHT is described and
details of implementation of the new SWCCs (Eq. 4-16 and 4-20) in CODE_BRIGHT are presented.
The standard Galerkin and finite difference methods are used for spatial and temporal discretization,
respectively. The implicit scheme is implemented in CODE_BRIGHT. The scheme is linear and uses two
intermediate points of 𝑡 𝑘+𝜀 and 𝑡 𝑘+𝜃 between each time step, 𝑘, with initial time, 𝑡 𝑘 , and final time, 𝑡 𝑘+1 .
The state variables in this scheme are:
𝑢𝑥 = displacement in x direction
𝑢𝑦 = displacement in y direction
𝑢𝑧 = displacement in z direction
𝑃𝑙 = liquid pressure
-49
𝑃𝑔 = gas pressure
𝑇 = temperature
Olivella et al. (1996) discussed the performance of two different approaches to numerical
implementation specifically in relation to the treatment of storage terms: the conservative and capacitive
approaches. The conservative approach directly discretises the storage terms; while the chain rule is used
in the capacitive approach to derive time derivatives in terms of state variables. Olivella et al. (1996) argue
that the difference in accuracy of the methods are not very significant; however, the conservative
approach is computationally more efficient than the capacitive approach if the Newton-Raphson is
adopted as the iterative scheme for solution of the system of equations because in the capacitive
approach, second order derivatives must be computed. The reader is referred to Milly (1985), Allen and
Murphy (1986), and Celia et al. (1990) for comparison of performance of these methods.
It is assumed that variation of porosity occurs at slow rate; therefore, computation of porosity at the
intermediate points is not necessary and the values of porosity at 𝑡 𝑘 are used for integrations. Spatial
discretization of the PDEs can be derived by applying the weighted residual method. It can be presented
for one finite element as:
𝑟(𝑋) = 𝑑̇ + 𝑎(𝑋) + 𝑏(𝑋) = 0 4-30
where:
𝑟 = the residuals,
𝑑̇ = the storage terms, they are calculated by means of material properties such as degree of
saturation, density, porosity, and specific energy,
𝑎 = the flux terms. Both advective and non-advective terms are calculated explicitly in terms of
gradients,
𝑏 = the sink/source terms and boundary conditions,
𝑋 = the vector of state variables defined as:
𝑢𝑥
𝑢𝑦
𝑢𝑧
𝑋= 𝑃 4-31
𝑙
𝑃𝑔
{𝑇}
The overdot indicates time derivatives. Time discretization using the finite difference method leads
to:
-50
𝑑𝑘+1 − 𝑑𝑘
𝑟(𝑋 𝑘+1 ) = + 𝐴(𝑋 𝑘+𝜀 )𝑋 𝑘+𝜃 + 𝑏(𝑋 𝑘+𝜀 , 𝑋 𝑘+𝜃 ) = 0 4-32
∆𝑡𝑘
where:
𝐴 = the explicit expression of matrix of state variable gradients based on Darcy’s and Fick’s laws.
Calculation of the matrix on the intermediate points of 𝑡 𝑘+𝜀 and 𝑡 𝑘+𝜃 depends on the solution strategy.
Applying the Newton-Raphson method leads to the following form:
𝜕𝑟(𝑋 𝑘+1 ) 𝑘+1,𝑙+1

(𝑋 − 𝑋𝑘+1,𝑙 ) = −𝑟(𝑋𝑘+1,𝑙 ) 4-33
𝜕𝑋𝑘+1
where:
𝑙 = iteration indicator
Detailed descriptions of the numerical implementation have been presented by Olivella et al. (1994),
Olivella (1995), and Olivella et al. (1996). In equation 4-33, the first derivatives of degree of saturation with
respect to the state variables must be computed for implementation of the new SWCCs (Eq. 4-16 and 4-20)
in CODE_BRIGHT. Details of deriving the derivatives are included in Appendix A.
4.10 Summary
In this chapter, the governing equations and constitutive models for the non-linear thermo-hydro-
mechanical theory were presented. The theories of Grant and Salehzadeh (1996) and Gallipoli (2012) were
combined to develop a new void ratio and temperature-dependent SWCC based on the van Genuchten
(1980) SWCC. A similar approach was used to establish the Fredlund and Xing version of the SWCC.
Furthermore, numerical implementation of the SWCCs in CODE_BRIGHT was briefly discussed.
-51
5 Experimental Determination of Soil-Water Characteristic Curves of GCL
Chapter 5
Experimental Determination of
Soil-Water Characteristic Curves of GCL
5.1 Introduction
SWCCs define the relationship between suction and water content in a porous media. SWCC of GCLs are
important in understanding the performance of CGCLSs. It has been shown that the SWCC relationship
may change under different temperatures and/or void ratios (e.g. Romero et al., 2001; Khalili et al., 2008).
Despite the relatively large number of experiments on SWCCs of GCL, our understanding of the effects of
change in temperature and void ratio on the water retention capacity of GCLs remains incomplete. This
has been discussed in Section 2.8 where a detailed review of the relevant literature was presented. In this
chapter, GCL SWCC experiments under different temperatures and loads are described. Results are
presented and empirical equations are fitted.
5.2 Material
The material used in this study is ELCOSEAL geosynthetic clay liner grade X2000 manufactured by
GEOFABRICS Australia. The GCL is made of a powdered Na-bentonite and polypropylene geotextiles with
non-woven and woven carrier and non-woven cover configuration. In addition, it is fibre-reinforced by
needle-punching and the fibres are thermally locked. Table 5-1 shows properties of the GCL provided by
the manufacturer. The as-received gravimetric water content of the GCL was between 8.5 and 9.5% which
depends on the manufacturing process and type of GCL.
-52
Table 5-1. GCL X2000 properties reported by GEOFABRICS Australia. Typical values are presented.
ELCOSEAL Grade
Property Test method Unit
X2000
Configuration (Carrier/Cover) - W+NW/NW
𝑘𝑠 ASTM D5887 m/s 1.6 × 10-11
𝑀𝐺𝐶𝐿 @ 𝑤=0% ASTM D5993 g/m2 4960
𝑀𝐵𝑒𝑛𝑡𝑜𝑛𝑖𝑡𝑒 @ 𝑤=0% ASTM D5993 g/m2 4250
𝑀cover AS 3706.1 g/m2 300
𝑀carrier AS 3706.1 g/m2 410
Hydrated peak internal shear

strength under 10 kPa normal ASTM D6243 kPa 35
stress
Hydrated peak internal shear
strength under 30 kPa normal ASTM D6243 kPa 60
stress
Bentonite particle size AS 1286-3.6.2 passing 75𝜇𝑚 ≥ 75%
passing 0.5𝜇𝑚 ≥ 55%
Bentonite swell index ASTM D5890 mL/2g ≥ 24
Montmorillionite content XRD QM % of bulk specimen ≥ 70
% of particle passing 0.5𝜇𝑚 ≥ 90
Calcium carbonate content
XRD QM % of bulk specimen ≥2
(CaCO3)
NH4 DM
Cation exchange capacity cmol/kg 70-110
BSM
W: Woven; NW: Non-Woven; XRD: X-ray diffraction; QM: Quantitative mineralogy analysis; DM: displacement method; BSM:
Barium saturation method; The typical values are the arithmetic mean of the measured values.
5.3 Experimental Program
Two different methods were used to cover a wide range of suction. SWCC tests based on axis-translation
method were performed for matric suctions from 0 to 700 kPa on the drying path, using pneumatic
suction-controlled odometers (Barcelona Cells). In addition, the dew point method was used for
measurement of total suctions from 100 to 10 MPa and 30 to 250MPa on the wetting and drying path,
respectively, using WP4C and vapour sorption analyzer (VSA). It has been shown that WP4C measurements
are accurate for suctions higher than 1000 and the accuracy improves as suction increases (e.g. see
Campbell et al., 2007). On the other hand, VSA is able to control humidity of a specimen in the chamber
and provide a path for drying and wetting SWCCs with high confidence. In addition, the time required to
reach equilibrium using VSA is relatively short. Therefore, WP4C was used to verify suction measurements
by VSA. A brief review of the dew point method, also known as chilled-mirror method has been discussed
in Section 2.7.
Moreover, SWCC measurements at Monash University were conducted on the same type of GCL using
the vapour equilibrium technique (VET) for a range of total suctions between 3 and 410 MPa on both
-53
Table 5-2. Experimental program
Net vertical stress Temperature Matric/Total

Test method Suction Path
(kPa) (°C) suction
Axis-translation 10, 20, 50 20, 35, 40 0 to 700kPa Matric Drying
WP4C 0 20, 35 100 to 10MPa Total Wetting
WP4C 0 20,35 30 to 250 MPa Total Drying
VSA 0 20, 35, 60 100 to 10MPa Total Wetting
VSA 0 20, 35, 60 30 to 250 MPa Total Drying
VETc 0, 20 20 3 to 70MPa Total Wetting
VETc 0, 20 20 110 to 360MPa Total Drying
VETc <1 70 1.5 to 80MPa Total Wetting
VETc <1 70 230 and 410MPa Total Drying
a: Matric suction; b: Total suction; c: Data from Monash University
drying and wetting paths. The data from Monash University is included in the results presented in this
chapter. The wetting path was investigated for suctions lower than ≅ 100𝑀𝑃𝑎 (i.e. the suction
corresponding to the as-received GCL moisture content) and the drying path was followed for suctions
higher than ≅ 100𝑀𝑃𝑎.
All experiments were under temperature-controlled conditions. The axis-translation, WP4C, and VSA
experiments were conducted under 20 and 35°C. Additional experiments at 40 and 60°C were also
performed using axis-translation and VSA, respectively. The VET tests were obtained for 20 and 70°C.
Moreover, constant net vertical stresses were applied to the specimens in the axis-translation (10, 20, and
50 kPa) and VET (<1 and 20kPa) experiments. The rationale for the specific sets of measurements has been
presented in Section 3.1, where Figure 3-2 summarises the ranges of the suctions covered. A summary of
the experiments performed is presented in Table 5-2.
5.4 Sample Preparation
Cutting dies made of steel were used to minimize bentonite loss when cutting GCL. Specimens from the
GCL sheets were cut at the as-received moisture content. The mass of each specimen was measured and
then the specimen was sealed in double plastic zip bags and stored in an air-tight container. The average
mass and mass per area of the specimens are presented in Table 5-3. It is evident that the smaller size
specimens, specimens with 37 and 50 mm diameters, have noticeably lower average mass per area
compared to the larger specimens, i.e. 160 and 190mm diameters, which is likely due to bentonite loss
during the cutting process. It should be mentioned that mass per area of bentonite can significantly affect
hydraulic properties of a GCL.
-54
Table 5-3. Summary of GCL specimens
Average mass of
Number of Average mass per area at
Diameter specimens at as-
specimens as-received moisture Comments
(mm) received moisture
prepared content (g/m2)
content (g)
37 24 5.34 4968 For VSA and WP4C
50 48 9.67 4924 For Barcelona cells
160 4 106.67 5305 For PVC columns
190 6 154.83 5461 For Perspex columns
For SWCC tests using the axis-translation method, a hydration setup was used to hydrate GCL
specimens under specific loads. The setup allows measuring volume change during hydration without
disturbing the specimen. It includes a hydration ring with a diameter of 50mm and height of 20mm made
of stainless steel, two porous stones for top and bottom of specimen, a hydration mould, a dial gauge, and
a load hanger (see Figure 5-1). Hydration of a specimen inside a stainless steel ring is very important
because the hydration ring permits only one dimensional deformation of the GCL during hydration. In
addition, the ring prevents bentonite loss and keeps the specimen shape as a perfect circle during
hydration.
Two porous stones were placed on top and bottom of the specimen inside the hydration ring and
then the ring was submerged in tap water for more than a month to hydrate under a specific load applied
by the hanger. Swelling of the specimen was measured using the dial gauge and it was assumed that the
specimen had reached maximum hydration once swelling stopped. The normal stress during the hydration
Figure 5-1. Hydration setup
-55
stage was 2 kPa for the specimens that were tested under 10 kPa in the Barcelona cell. On the other hand,
for SWCC tests where specimens were subjected to net normal stresses of 20 and 50 kPa, the load during
hydration was 20 kPa (Table 5-5). The specimens were hydrated at room temperature varying between 22
and 24°C. It is generally not possible to fully saturate the GCL specimens without applying a back pressure.
The specimens prepared for SWCC testing in this study reached a very high degree of saturation; although,
full saturation was never achieved.
5.5 Axis-translation: Pneumatic Suction-controlled Oedometer (Barcelona Cell)
Two pneumatic suction-controlled oedometers were acquired for this study. The oedometers were
designed and manufactured at the Department of Geotechnical Engineering at Polytechnic University of
Catalonia (UPC). Figure 5-2 shows a diagram of the device. A fully automated control and monitoring
system was setup for the oedometers to perform SWCC testing in a temperature-controlled environment.
The setup included one liquid pressure and volume (GDS) controller, one liner variable differential
transducer (LVDT), one water bath, two gas pressure transducers and two manual high precision air
regulators for each oedometer.
Axis-translation is based on the principle that a desired matric suction can be applied to a soil by
controlling pore-gas and pore-liquid pressure. The liquid pressure is kept constant using a liquid volume-
pressure controller (GDS) and gas pressure is adjusted to reach the target matric suction. A porous ceramic
disk (or a membrane) is required in axis-translation to separate the gas and liquid phases and the maximum
possible suction is hence limited to the air-entry value of the porous ceramic disk. In the pneumatic
LVDT
Loading piston
Diaphragm Load pressure
Soil sample
Porous ceramic disk Pore air pressure
Flushing diffused air Pore water pressure

(outlet) (inlet/outlet)
Figure 5-2. Barcelona Cell (modified from a drawing by Tomás Pérez Revilla, UPC)
-56
oedometers, temperature was controlled by an air-conditioning system for experiments at 20°C and water
baths were used for experiments at 35°C (Figure 5-3). Load was applied by compressed air over a
diaphragm separating pore gas pressure and load pressure. The net stress is the differential pressure over
the diaphragm plus the pressure from the weight of the piston as follows:
𝑚𝑝𝑖𝑠𝑡𝑜𝑛 . 𝑔 𝐴𝑠ℎ𝑎𝑓𝑡
𝜎𝑛𝑒𝑡 𝑎𝑥𝑖𝑎𝑙 = 𝑃𝑙𝑜𝑎𝑑 − 𝑃𝑔 + − 𝑃𝑙𝑜𝑎𝑑 5-1
𝐴𝑝𝑖𝑠𝑡𝑜𝑛 𝐴𝑝𝑖𝑠𝑡𝑜𝑛
where
𝑃𝑙𝑜𝑎𝑑 = gas pressure applied to top of loading piston, kPa
𝑃𝑔 = gas pressure, kPa
𝐴𝑠ℎ𝑎𝑓𝑡 = cross section area of the piston shaft, m2
𝑚𝑝𝑖𝑠𝑡𝑜𝑛 = mass of piston, kg
𝐴𝑝𝑖𝑠𝑡𝑜𝑛 = cross section area of piston, m2.
The guidelines from ASTM D6836 (2008) and ASTM C1699 (2009) were followed in this study. It is
very important to saturate the porous ceramic disk before each test. For saturation of the ceramic disk, it
was submerged in de-ionised and de-aired water for at least 24 hours and de-ionised and de-aired water
was flushed with 700 kPa pressure through the ceramic disk for a short period of less than 1 hour and
repeated again with freshly de-aired water until no air was observed to flow out from diffused air flushing
system. A kaolin paste was prepared according to ASTM C1699 (2009) by mixing 125g of kaolin powder
and 150g distilled water. The paste with a thickness of ≈1mm was directly applied onto the saturated
ceramic disk. The kaolin paste ensures good hydraulic contact between GCL and the ceramic disk. It should
be noted that capillary breakage has been frequently reported in GCL SWCC measurements (e.g. Southen
and Rowe, 2007; Abuel-Naga and Bouazza, 2010; Beddoe et al., 2010). In addition, a 45° rotation of the
Figure 5-3. Water baths for SWCC tests under 40°C
-57
specimen after installation is recommended for a better hydraulic contact. A number of preliminary tests
were conducted and it was observed that using the above recommendations considerably reduced the
required time to reach equilibrium under higher suctions (i.e. 350 and 700kPa).
After saturation, the specimen was placed in the oedometer and the desired suctions were applied.
To be able to determine air-entry value of the specimen, it is recommended that the first applied suction
be below one-half the anticipated air-entry value (ASTM D6836, 2008). The oedometers were placed in
the temperature controlled environment with the target temperature at least 24 hours before the SWCC
measurements start, in order to make sure the cells are in thermal equilibrium with their environment.
Suctions from 0 to 700 kPa were imposed in successive stages to determine the soil water
characteristic curves. At each applied suction, hydraulic equilibrium must be reached before a reading of
water content is made. According to ASTM C1699 (2009), equilibrium is reached when water outflow is
less than 50mm3 in 48 hours. Achieving equilibrium may take several days, weeks, even months for some
materials depending on the nature of the material and applied suction. Furthermore, the equilibrium state
for a GCL can be confirmed based on the rate of axial strain similar to ASTM D2435/D2435M (2011) under
each matric suction step.
Flushing out any diffused air through the ceramic disk is essential in axis-translation technique
particularly at high suctions. During SWCC testing, air dissolves in water under high air pressures in the
oedometer and diffuses through the saturated ceramic disk. Then air bubbles appear after passing the disk
under lower liquid pressure at the bottom of the disk. For flushing air bubbles, water is pumped through
a spiral groove under the ceramic disk and air bubbles are forced to flow out to an air trap. The level of
water in the air trap should be adjusted back to a fixed level before readings on the liquid pressure and
volume controller are restarted. If flushing is not performed, errors occur in water volume change readings
equal to volume of air bubbles accumulated under the disk and possibly significantly altered SWCCs would
be observed (Roshani and Sedano, 2016).
5.6 Dew Point Method: WP4C and VSA
“Dewpoint PotentiaMeter WP4C” and VSA devices are based on an accurate measurement of partial
pressure of vapour in a small sealed chamber above a specimen using the dew point of vapour under
equilibrium condition with the specimen (Campbell et al. 2007). At equilibrium, the partial pressure of
vapour in the chamber and the partial pressure of pore-water vapour of the specimen are equal. A small
fan in the chamber reduces the time it takes to reach equilibrium. Figure 5-4 illustrates schematics of WP4C
and VSA. The air space acts as a perfect semi-permeable membrane; therefore, total suction is measured
-58
using the dew point method. Total suction can be computed from relative humidity using the Kelvin
equation (Fredlund et al. 2012):
−𝑅(𝑇 + 273.16)𝜌𝑤 𝑝𝑣
𝑠𝑇𝑜𝑡𝑎𝑙 = ln ( ) 5-2
𝜔𝑣 𝑝𝑣0
where:
𝑠𝑇𝑜𝑡𝑎𝑙 = total suction, kPa
𝑅 = universal (molar) gas constant, i.e. 8.31432 J/(mol K)]
𝑇 = temperature, °C
𝜌𝑤 = density of water, kg/m3
𝜔𝑣 = molecular mass of water vapour, i.e. 18.016 kg/kmol
𝑝𝑣 = partial pressure of pore-water vapour, kPa
𝑝𝑣0 = saturation pressure of water vapour over a flat surface of pure water at the same temperature,
kPa.
𝑝𝑣
The term is called relative humidity, RH, or water activity, 𝑎𝑤 .
𝑝𝑣0
The partial pressure of vapour, 𝑝𝑣 , is evaluated by measuring dew point temperature of vapour above
the specimen. This is done by cooling a mirror and carefully monitoring reflections from the mirror using
an optical sensor. The mirror temperature is controlled to obtain a uniform thickness of dew over the
(a)
(b)
Figure 5-4. Schematic of dew point method devices: a) WP4C (Fredlund et al., 2012); b) VSA (VSA manual)
-59
mirror. The saturated vapour pressure at the dew point is equal to the partial vapour pressure at specimen
temperature. The temperature of the mirror (i.e. dew point temperature), 𝑇𝑑 , is measured with a
thermocouple and the infrared thermometry method is used to monitor temperature of the specimen, 𝑇𝑠 .
The saturated vapour pressure only depends on temperature for pure water (Campbell et al., 2007):
17.502𝑇
𝑝𝑣0 = 0.611 exp ( ) 5-3
240.97 + 𝑇
Substituting 𝑇𝑑 and 𝑇𝑠 in equation 5-3 gives the partial pressure of pore-water vapour and saturation
pressure of water vapour at specimen temperature, respectively. Therefore, total suction of specimen can
be calculated based on equation 5-2 as follows:
−𝑅(𝑇 + 273.16)𝜌𝑤 𝑇𝑑 𝑇𝑠
𝑠𝑡𝑜𝑡𝑎𝑙 = ln ( − ) 5-4
𝜔𝑣 𝑇𝑑 + 240.97 𝑇𝑠 + 240.97
VSA provides two different methods to perform SWCC tests: 1) the dynamic vapour sorption method
(DVS); and 2) the dynamic dew point isotherm method (DDI). In the DVS method, a specimen is exposed
to a controlled humidity and change in mass of specimen is tracked until the specimen reaches equilibrium
and rate of change of mass of specimen reduces to a limit (say 0.02% per hour). On the other hand, the
DDI method dries or wets the specimen under a constant flow of air; however, neither change in mass of
specimen nor humidity is controlled; as a result, the specimen may not be at equilibrium at the time of
suction measurements. Therefore, the DVS method is preferred and it is used to perform tests in this
thesis.
5.7 Mass-volume Relations for GCL
Mass-volume relations for GCL are easy to derive and available in the literature. These relationships are
briefly presented here for a clearer presentation of the method used in this thesis to calculate void ratio
and degree of saturation.
SWCCs are usually expressed as a function of volumetric water content or degree of saturation.
Degree of saturation depends on void ratio of the specimen. However, GCL is a composite material and
void ratio of the components can be different. Therefore, bulk void ratio of GCL was introduced by Petrov
et al. (1997) as follows:
𝐻𝐺𝐶𝐿 − 𝐻𝑠
𝑒𝑏𝑢𝑙𝑘 = 5-5
𝐻𝑠
where:
-60
𝐻𝐺𝐶𝐿 = height of GCL, m
𝐻𝑠 = height of solids, m
Height solids for GCL is defined as:
𝑀𝑏 𝑀𝑔𝑒𝑜
𝐻𝑠 = + 5-6
𝜌𝑏 (1 + 𝑤) 𝜌𝑔𝑒𝑜
where:
𝑀𝑏 = mass per area of bentonite, kg.m-2
𝑀𝑔𝑒𝑜 = are mass per area of geotextiles, kg.m-2
𝜌𝑏 = density of bentonite solids, kg.m-3
𝜌𝑔𝑒𝑜 = are density of geotextile polymer solids, kg.m-3
𝑤 = gravimetric water content of GCL.
Equation 5-6 in terms of specific gravity values of bentonite, cover geotextile, and carrier geotextile
is as follows:
1 𝑀 𝑀 𝑀
𝐻𝑠 = (( ) +( ) +( ) ) 5-7
𝜌𝑤 𝐺𝑠 (1 + 𝑤) 𝑏𝑒𝑛𝑡𝑜𝑛𝑖𝑡𝑒 𝐺𝑠 𝑐𝑜𝑣𝑒𝑟 𝐺𝑠 𝑐𝑎𝑟𝑟𝑖𝑒𝑟
where:
𝐺𝑠 = specific gravity of each component
Then degree of saturation is:
𝜌𝑑 (1 + 𝑒𝑏𝑢𝑙𝑘 )
𝑆𝑟 = 𝑤 5-8
𝜌𝑤 . 𝑒𝑏𝑢𝑙𝑘
where:
𝜌𝑑 = dry density of GCL, kg/m3
𝜌𝑤 = density of water, it is 998 kg/m3 at 20°C
and degree of saturation in terms of volumetric water content is:
(1 + 𝑒𝑏𝑢𝑙𝑘 )
𝑆𝑟 = 𝜃 5-9
𝑒𝑏𝑢𝑙𝑘
-61
Table 5-4. Height of solids for X2000 based on the typical values reported by the manufacturer†
𝑀𝐺𝐶𝐿 𝑀𝑐𝑜𝑣𝑒𝑟 𝑀𝑐𝑎𝑟𝑟𝑖𝑒𝑟 𝑀𝑏 𝑤 𝐺𝑠 𝑏𝑒𝑛𝑡𝑜𝑛𝑖𝑡𝑒 𝐺𝑠 𝑐𝑜𝑣𝑒𝑟 𝐺𝑠 𝑐𝑎𝑟𝑟𝑖𝑒𝑟 𝜌𝑤 𝐻𝑠

(g/m2) (g/m2) (g/m2) (g/m2) (%) (kg/m3) (mm)
4960 300 410 4250 0 2.65 0.91 0.91 998 2.389
†: The typical values are the arithmetic mean of the measured values by the manufacturer.
More details of the derivation of these mass-volume relations for GCL under unsaturated condition
are presented in Appendix B.
Typical height of solids can be computed as 𝐻𝑠 = 2.389𝑚𝑚 for the GCL used in this study based on
the typical values of mass per area of GCL, bentonite, and geotextile (Table 5-4 shows the parameters used
for calculation of height of solids). On the other hand, for each specimen used in this thesis, the actual
height of solids based on actual mass per area of the specimen was calculated and it is assumed that the
typical values of mass per area of geotextiles (reported by the manufacturer) are the actual values for all
specimens and any difference in the mass of GCL for the specimen is because of difference in mass per
area of bentonite. This assumption complies with guidelines for measurement of mass per unit area of
GCLs provided by ASTM D5993 (2009) “Standard Test Method for Measuring Mass Per Unit of Geosynthetic
Clay Liners”. In other words, for all specimens in this thesis, mass per area of cover and carrier are
considered to be 300 and 410 g/m3, respectively.
5.8 Results
5.8.1 Axis-translation
To study the effects of overburden stress and temperature on the low-suction range of SWCC, experiments
using axis-translation technique were conducted under loads of 10, 20, and 50 kPa and temperatures of
20, 35, and 40°C. Table 5-5 shows the details of preparation and test conditions performed using axis-
translation.
Figures 5-5 to 5-7 show the effect of load on SWCC and void ratio under different test conditions of
20, 35, and 40°C, respectively. Moreover, Figures 5-8 and 5-9 present effect of temperature on retention
capacity of GCL in terms of degree of saturation and gravimetric water content under net axial stresses of
20 and 50kPa, respectively. A summary of all axis translation experiments is presented in Figure 5-10. In
addition, details of all experiments are tabulated in Tables 5-6 to 5-11.
-62
Table 5-5. Preparation and SWCC tests using axis-translation technique
Hydration Test
Dry mass of Mass per area of Test net
Specimen Diameter overburden
specimen specimen @ 0% axial stress Temperature
ID (mm) Stress
(g) (g/m2) (kPa) (°C)
(kPa)
#11D50 50 9.680 4930 20 20 20±1.0a
#17D50 50 9.086 4627 2 10 20±1.0a
#23D50 50 10.340 5266 20 20 35±0.1b
#37D50 50 9.548 4863 20 20 40±0.1b
#39D50 50 9.280 4726 20 50 35±0.1b
#42D50 50 9.047 4608 20 50 40±0.1b
a: air-conditioning was used; b: water bath was used
For all tests, an increase in the degree of saturation was observed after applying the target vertical
stress and start of the first suction step (Tables 5-6 to 5-11). As discussed before, it was not possible to
reach full saturation during the hydration stage despite the long period in which the specimens were
submerged. The increase in degree of saturation when load is applied is due to the expulsion of the air
bubbles trapped in the GCL pores. It is also associated with a small amount of water flow, which depends
on the hydration stress. The reduction in the volume of the specimen is larger than the change in the
volume of water and the degree of saturation of the specimen rises as a result.
At 20°C, an increase in net axial stress from 10kPa to 20kPa decreased void ratio of the specimen for
the same suctions (Figure 5-5-b) and increased air-entry suction and slope of the SWCC (Figure 5-5-a). This
is consistent with reported effects of void ratio change on SWCC of porous media (e.g. see Romero et al.,
2001; Gallipoli et al., 2003; Tarantino, 2009).
The significant difference between void ratios of tests under 10kPa (i.e. specimen #17D50) and 20kPa
(i.e. specimen #11D50) may be due to the different hydration conditions of the specimens. As described
before, the specimens for tests under 10kPa and 20kPa were hydrated under 2kPa and 20kPa overburden
stresses, respectively (Table 5-5). Therefore, fibres have very likely been pulled out of the specimen that
was hydrated under 2kPa load as a result of swelling pressure of bentonite which leads to higher void
ratios and higher gravimetric water contents, at saturation. On the other hand, the stresses acting on the
specimen that was hydrated under 20kPa include both internal forces from fibres and the applied 20kPa
overburden stress during the hydration stage which significantly limits swelling of the specimen.
The observed trend for effect of the load for tests under 35°C is consistent with those conducted
under 20°C to some extent (Figure 5-6). A higher net axial stress caused lower void ratio and slightly
increased air-entry value, while it seems that the slope of the SWCC for the experiment under net stress
-63
of 50kPa is the same or less than the slope of experiment under 20kPa. Though, more data points are
required to be able to determine any difference in slope of the SWCC between the two experiments.
The measured SWCCs under 40°C seem to be on a scanning curve at the beginning of the experiments
instead of following the drying path (Figure 5-7-a). This could be due to insufficient saturation of specimens
#37D50 and #42D50 prior to start of the tests. Moreover, the effect of load on void ratios under 40°C
condition is inconsistent with the other two sets of experiments under temperatures of 20 and 35°C (Figure
5-7-b). The test with the higher load of 50kPa has a higher initial void ratio; however, at the end of the
test, the higher load results in a smaller void ratio compared to the lower load of 20kPa.
There are a number of difficulties associated with determination of degree of saturation at the
beginning of the test. Tests under elevated temperatures are even more difficult. First, the dry mass of the
specimen at the beginning of the test is unknown because oven drying a specimen prior to SWCC testing
may alter the mineralogy of the bentonite; therefore, the specimen must be hydrated from as-received
moisture content. In addition, some bentonite loss may occur, while the specimen is transferred from the
saturation set-up to the oedometer. Therefore, the actual dry mass of specimen can be only reliably
determined at the end of the test.
Second, determination of the exact height of the specimen at the beginning of the test is challenging.
In general, the standard height of soil specimens to be tested in the Barcelona cells is 20mm. However,
GCLs are manufactured and the actual height of the specimen depends on a number of factors such as
degree of saturation and mass per area of the GCL components (for a given type of GCL). The specimens
used in this thesis were saturated under a constant overburden stress of 20kPa, except for one specimen
that was hydrated under 2kPa as discussed above. Unloading the specimens after hydration, results in an
elastic swelling followed by a gradual plastic swelling because of fibre pullout. Therefore, the specimen
must be quickly transferred to the oedometer and a net stress of at least 20kPa must be applied to avoid
fibre pullout. It is clear that only a rough estimate of height of specimen at the beginning of the test is
possible and the exact height of specimen must be measured at the end of the test.
Third, the readings from LVDTs are sensitive to temperature change. The sensors were calibrated for
each temperature; however, it is difficult to measure height of the specimen accurately while the
oedometer is not at equilibrium and distribution of temperature within the device and at the specimen
are unknown. As described in Section 5.5, the cell (including LVDT) was left at the target test temperature
for at least 24 hours to get to thermal equilibrium in the thermal bath. Transferring the specimen to the
-64
oedometer was performed quickly; however, the measurements of height of specimen at the beginning
of the test are not accurate, which add to the difficulties of establishing an accurate measurement of the
degree of saturation at the beginning of the test.
100 5.0
4.5
80
4.0
60
Sr (%)
e (-)
3.5
40
10kPa, 20°C
3.0
20 20kPa, 20°C
2.5
0 2.0
0.1 1 10 100 1000 0.1 1 10 100 1000
(a) Suction (kPa) (b) Suction (kPa)
Figure 5-5. Effect of load on SWCC at 20°C: Degree of saturation vs suction; b) Void ratio vs suction
100 5.0
4.5
80
4.0
60
Sr (%)
e (-)
3.5
40
20kPa, 35°C
3.0
20 50kPa, 35°C
2.5
0 2.0
1 10 100 1000 1 10 100 1000
100 5.0
4.5
80
4.0
60
Sr (%)
e (-)
3.5
40
20kPa, 40°C
3.0 20kPa, 40°C
20 50kPa, 40°C
2.5 50kPa, 40°C
0 2.0
1 10 100 1000 1 10 100 1000
-65
To evaluate the effect of elevation of temperature on SWCC along the drying path, two main factors
can be considered among others (Zhou et al., 2014): 1) reduction of surface tension of water; 2) softening
of the soil. The former factor results in a lower retention capacity in terms of both degrees of saturation
and gravimetric water content because a lower surface tension of water reduces capillary pressure and
the capacity of the soil to hold water (Grant and Salehzadeh, 1996). Conversely, softening can result in a
lower void ratio (Campanella and Mitchell, 1968) which can lead to two counter-acting processes: 1)
increasing capillary pressure because of smaller pore sizes (Khalili et al., 2008; Gallipoli, 2012); 2) squeezing
water out and reducing the total amount of water. In addition, other factors such as change in pore water
chemistry and the different thermal expansion of different phases can be involved (Romero et al., 2001).
For the tests under 20kPa, the retention capacity of GCL in terms of gravimetric water content appears
to decline as temperature increases (Figure 5-8), which is consistent with our understandings of effect of
temperature on retention capacity of porous media. However, for the tests under 50kPa, higher
100 250
80 200
60 150
Sr (%)
w (%)
40 20kPa, 20°C 100

20kPa, 35°C
20 20kPa, 40°C 50
0 0
0.1 1 10 100 1000 0.1 1 10 100 1000
Figure 5-8. Effect of temperature on SWCC under net axial stress of 20kPa: Degree of saturation vs suction; b) Gravimetric water
content vs suction
100 250
80 200
60 150
Sr (%)
w (%)
40 100 50kPa, 35°C

50kPa, 35°C
20 50kPa, 40°C 50 50kPa, 40°C
0 0
1 10 100 1000 1 10 100 1000
Figure 5-9. Effect of temperature on SWCC under net axial stress of 50kPa: Degree of saturation vs suction; b) Gravimetric water
content vs suction
-66
temperature shows a higher retention capacity (Figure 5-9). Given that specimen #45D50 had a lower mass
of GCL per area than #39D50 and specimen #45D50 was not probably fully hydrated prior to test as
discussed before, the observed trend in retention capacity seems to be inconsistent with other sets of
experiments and the literature.
In terms of degrees of saturation, for all tests, a higher temperature shows lower retention capacity
under the same net vertical stress, except for the dry end of the test under 20kPa net vertical stress and
40°C where degrees of saturation are higher (Figure 5-10). The available data is very limited and more
experiments are required to build an understanding of behaviour. A major difficulty in the axis-translation
method experiments is that a relatively long time is required to reach equilibrium for each point at higher
suctions. The length of experiments to obtain each curve was more than 2 months in this study.
The results from the axis-translation method will be used later in this chapter to derive the wet end
of a new SWCC equation for GCLs.
100 250
80 200
10kPa, 20°C 10kPa, 20°C
60 20kPa, 20°C 150 20kPa, 20°C
Sr (%)
w (%)
20kPa, 35°C 20kPa, 35°C

40 100
20kPa, 40°C 20kPa, 40°C
50kPa, 35°C 50kPa, 35°C
20 50
50kPa, 40°C 50kPa, 40°C
0 0
0.1 1 10 100 1000 0.1 1 10 100 1000
1 5.0
4.5
0.8
10kPa, 20°C 10kPa, 20°C
4.0
0.6 20kPa, 20°C 20kPa, 20°C
3.5
θ
20kPa, 35°C 20kPa, 35°C

0.4
20kPa, 40°C 3.0 20kPa, 40°C
50kPa, 35°C 50kPa, 35°C
0.2
2.5
50kPa, 40°C 50kPa, 40°C
0 2.0
0.1 1 10 100 1000 0.1 1 10 100 1000
(c) Suction (kPa) (d) Suction (kPa)
Figure 5-10. Summary of all axis-translation tests: a) Degree of saturation; b) Gravimetric water content; c) Volumetric water
content; d) Void ratio
-67
Net Axial Suction Axial

Stage w (%) θ Sr (%) e Equ. Time
Stress (kPa) (kPa) Strain (%)
Seating 2.42 - 186 0.779 97.49 3.97 - -
1 22.11 0.73 186 0.783 98.22 3.93 0.66 2 hours
2 21.67 11.21 183 0.794 100.00 3.80 3.30 12 hours
3 22.10 20.64 172 0.782 99.99 3.58 7.76 2 days
4 21.90 39.61 161 0.752 97.03 3.44 10.61 1 week
5 22.03 79.87 151 0.741 97.01 3.24 14.61 5 days
6 21.03 170.33 133 0.681 90.26 3.07 18.10 19 days
7 20.00 350.21 102 0.562 76.49 2.77 24.09 39 days
Net Axial Axial

Suction
Stage w (%) θ Sr (%) e Strain Equ. Time
Stress (kPa) (kPa)
(%)
Seating 2.04 - 246 0.834 99.96 5.04 - -
1 9.26 1.00 240 0.832 100.00 4.91 2.17 24 hours
2 11.67 40.01 215 0.800 97.75 4.51 8.67 3 days
3 11.34 80.00 186 0.709 87.11 4.38 10.83 4 days
4 10.94 169.83 155 0.623 77.45 4.10 15.48 14 days
5 9.69 349.48 126 0.537 67.74 3.81 20.30 24 days
6 7.69 699.59 103 0.459 58.67 3.60 23.78 29 days

Seating 2.80 - 165 0.683 84.94 4.11 - -

1 19.32 2.70 174 0.736 92.03 3.99 2.27 1.00
2 21.22 10.14 174 0.741 92.75 3.97 2.70 1.00
3 20.29 20.01 165 0.708 88.88 3.93 3.55 5.00
4 21.42 38.59 157 0.691 87.27 3.81 5.88 6.00
5 20.32 78.93 141 0.649 82.99 3.60 10.06 11
6 20.60 170.03 122 0.591 76.75 3.36 14.71 15
7 21.53 347.95 100 0.515 68.04 3.12 19.32 15
8 20.84 698.38 82 0.441 59.11 2.93 23.04 14
-68

Seating 2.80 - 173 0.779 99.06 3.69 - -

1 50.39 7.39 173 0.779 99.06 3.69 0.00 7 days
2 50.79 21.30 161 0.735 93.90 3.61 1.56 5 days
3 49.87 39.99 151 0.708 91.11 3.49 4.14 7 days
4 50.23 79.32 127 0.644 84.80 3.16 11.29 11 days
5 52.69 166.16 107 0.587 79.36 2.83 18.21 11 days
6 50.60 348.90 83 0.510 71.98 2.43 26.91 14 days
7 50.31 698.10 66 0.444 65.03 2.15 32.84 15 days

Seating 2.42 - 172 0.719 89.86 4.00 - -

1 21.49 2.97 169 0.731 92.18 3.82 3.46 15 h
2 21.16 10.69 156 0.682 86.39 3.76 4.78 8 days
3 20.14 19.77 154 0.679 86.03 3.73 5.28 2 days
4 19.94 40.52 147 0.664 84.77 3.62 7.56 2 days
5 19.67 80.35 130 0.619 80.28 3.37 12.49 9 days
6 20.75 169.82 114 0.581 76.92 3.10 18.00 35 days
7 19.87 349.17 98 0.530 71.69 2.84 23.23 19 days
8 19.78 700.12 81 0.458 62.93 2.68 26.31 20 days
Net Axial
Suction Axial
Stage Stress w (%) θ Sr (%) e Equ. Time
(kPa) Strain (%)
(kPa)
Seating 2.42 - 206 0.719 86.60 4.90 - -
1 49.97 10.30 179 0.709 87.76 4.20 11.89 16 days
2 51.06 19.93 172 0.697 86.85 4.08 13.91 7 days
3 50.34 39.79 162 0.682 85.73 3.89 17.02 7 days
4 51.13 78.25 145 0.662 85.17 3.50 23.69 14 days
5 51.44 168.70 119 0.604 80.13 3.06 31.18 15 days
6 50.68 348.22 93 0.529 73.25 2.60 38.90 12 days
7 50.82 699.35 70 0.423 59.66 2.43 41.79 16 days
-69
5.8.2 Dewpoint method
VSA and WP4C were used to measure total suctions for the dry ends of the curves. In practice, composite
geosynthetic clay liners are expected to experience temperatures as high as 80°C on coal seam gas sites;
however, the experiments in this study were performed under temperatures of 20, 35, and 60°C using VSA
and WP4C was only used for 20 and 35°C because of the limitation of the devices. Table 5-12 shows mass
per area of GCLs and SWCC testing conditions.
For the experiments under 20 and 35°C, the starting point was the as-received water content and
suction of the specimen was measured at the starting point using WP4C (the blue triangles on Figures 5-12
and 5-13). Then VSA was used to impose controlled relative humidity and bring the suction ultimately to
≈10MPa. Wetting was performed in two steps then the drying path was followed in three steps to a
suction of nearly 250MPa. The upper and lower limits for suctions were selected based on the accuracy
and capacity of the VSA. In each step of SWCC testing, the suction measurements of VSA were cross
checked with WP4C. For the experiments under 60°C, only VSA was used, following the protocol described
above. The only difference was that the first reading was performed on the wetting path instead of the as-
received water content, since temperature-controlled suction measurement with the VSA requires
following a certain path (either wetting or drying).
The height of the specimen for all experiments was measured under a small load of 2kPa after each
step was completed. The load was applied by a steel block, while the specimen was confined in its cup.
However, it is not possible to apply a load to the specimen in the dewpoint method during suction
measurements. Initially, the height of the specimens was measured by a calliper; however, the readings
showed significant scatter. Therefore, the calculated void ratios were not accurate. To improve accuracy
of void ratio measurements, a setup similar to the apparatus described by ISO 9863-1 (2016) was used
Figure 5-11. GCL height measurement setup: the setup is a modification of ISO 9863-1 (2016)
-70
Table 5-12. GCL specimens mass per area and testing condition
Dry mass of Mass per area of

Diameter Test Stress Test Temperature
Specimen ID specimen GCL @ 0%
(mm) (kPa) (°C)
(g) (g/m2)
#2D37 37 5.080 4724 0 20±0.1

#15D37 37 5.364 4989 0 35±0.1
#16D37 37 5.035 4683 0 60±0.1
#23D37 37 5.510 5124 0 20±0.1
which significantly enhanced the accuracy of measurements (Figure 5-11). The main modification to the
procedure is that ISO 9863-1 (2016) recommends unconfined measurements of thickness under the
applied load; however, the thickness of a specimen in this study was measured under confined condition
in its holding cup. This modification is necessary for GCL specimens to avoid bentonite loss during testing.
VSA measures mass of specimen during SWCC tests to establish equilibrium; in addition, the change
in mass of specimen was independently tracked using a high-precision electronic balance with a
14 20
(a) 18 (b)
12
16
10 14
20°C
8 12
Start point (WP4C)
Sr (%)
w(%)
10
6 VSA (wetting)
8
WP4C (wetting) 6
4
VSA (drying) 4
2
WP4C (drying) 2
0 0
1000 10000 100000 1000000 1000 10000 100000 1000000
Suction (kPa) Suction (kPa)
0.1 3.2
0.09 (c) (d)
3
0.08
0.07
2.8
0.06
0.05 2.6
θ
0.04
2.4
0.03
0.02
2.2
0.01
0 2
1000 10000 100000 1000000 1000 10000 100000 1000000
Figure 5-12. Comparison of VSA and WP4C: specimen #2D37 at 20°C: a) degree of saturation vs. suction; b) gravimetric water
content vs. suction; c) volumetric water content vs. suction; d) void ratio vs. suction
-71
measurement accuracy of ±0.01mg after reaching equilibrium. VSA measures tare weight of cup at the
start of each step; however, an identical cup to the specimen cup was used for tare weight measurement
to prevent disturbing specimen and bentonite loss. The weight change based on the tare weight is accurate
enough to establish equilibrium; however, the actual water content was measured based on the
independent weight measurements using the high-precision electronic balance. It is important to conduct
such independent confirmation of mass of the specimen after each suction measurement step for both
VSA and WP4C, because of the possibility that a small amount of water exchange might occur during the
measurements. A consistent water content ensures that WP4C and VSA measurements have been
performed on the same path.
Comparison of VSA and WP4C measurements under temperatures of 20 and 35°C are presented in
Figures 5-12 and 5-13, respectively, and show excellent agreement. In addition, the data for the two dew
point methods are presented in Tables 5-13 and 5-14 for experiments under 20 and 35°C, respectively. The
results for 60°C are presented in Figure 5-14 and Table 5-15. The total suctions for VSA were calculated
14 20
18
12 (a) (b)
16
10 14
35°C
8 12
Start point (WP4C)
Sr (%)
w(%)
10
6 VSA (wetting)
8
WP4C (wetting) 6
4
VSA (drying) 4
2
WP4C (drying) 2
0 0
1000 10000 100000 1000000 1000 10000 100000 1000000
0.12 3
0.1 (c) 2.9
0.08 2.8
0.06 2.7
θ
0.04 2.6
0.02 2.5 (d)
0 2.4
1000 10000 100000 1000000 1000 10000 100000 1000000
Figure 5-13. Comparison of VSA and WP4C: specimen #15D37 at 35°C: a) degree of saturation vs. suction; b) gravimetric water
content vs. suction; c) volumetric water content vs. suction; d) void ratio vs. suction
-72
based on the measured relative humidity and temperature at equilibrium by the Kelvin’s equation (5-2),
while WP4C directly reports total suction. Degree of saturation and volumetric water content of specimens
were calculated from gravimetric water content and bulk void ratio based on the equations described in
Section 5.7.
Hysteresis was observed for all experiments and the drying path showed slightly higher degree of
saturation for all tests (Figures 5-12 to 5-14). Moreover, void ratios show a relatively consistent trend,
increasing as suction decreases and vice versa. However, a small hysteresis is observed in void ratio
changes and void ratios on the drying path are higher than those on the wetting path, except for one
reading for test under 35°C, which may be a measurement error.
Figure 5-15 shows the effects of temperature on SWCC of GCL for the drying path. The differences in
degrees of saturation are not significant nor following a consistent trend; therefore, it can be concluded
that temperature has no significant effect on SWCC of GCL on the dry end of the curves. On the wetting
path, water retention capacities in terms of gravimetric water content are almost identical (Figure 5-16-
14 20
18
12
(a) 16 (b)
10 14
8 12
Sr (%)
w(%)
10
6 8
60°C
4 6
Wetting 4
2 Drying 2
0 0
1000 10000 100000 1000000 1000 10000 100000 1000000
0.12 3.00
0.10 (c) (d)

2.80
0.08
2.60
0.06
θ
2.40
0.04
2.20
0.02
0.00 2.00
1000 10000 100000 1000000 1000 10000 100000 1000000
Figure 5-14. VSA measurements for drying and wetting at 60°C on specimen #16D37: a) degree of saturation vs. suction; b)
gravimetric water content vs. suction; c) volumetric water content vs. suction; d) void ratio vs. suction
-73
b). However, the experiment under higher temperature shows higher degrees of saturation (Figure 5-16-
a). This is likely because of lower void ratios of specimen #16D37 (Figure 5-17) which had the lowest mass
per area of GCL (Table 5-12).
12 16
14
10
12
8
10
Sr (%)
w (%)
6 8
0kPa,20°C 6 0kPa,20°C
4
0kPa,35°C 4 0kPa,35°C
2
0kPa,60°C 2 0kPa,60°C
0 0
1000 10000 100000 1000000 1000 10000 100000 1000000
Figure 5-15. Summary of dew point method experiments on the drying path: a) Degree of saturation vs. suction; b) void ratio
vs. suction.
14 20
18
12
16
10 14
8 12
Sr (%)
w (%)
10
6 8
0kPa,20°C 0kPa,20°C
4 6
0kPa,35°C 0kPa,35°C
4
2 0kPa,60°C
0kPa,60°C 2
0 0
1000 10000 100000 1000000 1000 10000 100000 1000000
Figure 5-16. Summary of dew point method experiments on the wetting path: a) Degree of saturation vs. suction; b) Gravimetric
water content vs. suction.
3.00
2.90
0kPa,20°C,w
2.80
0kPa,20°C,d
e
0kPa,35°C,w
2.70
0kPa,35°C,d
2.60 0kPa,60°C,w
0kPa,60°C,d
2.50
1000 10000 100000 1000000
Suction (kPa)
Figure 5-17. Void ratio vs suction for all. w: wetting path; d: drying path
-74
Table 5-13. Comparison of VSA and WP4C: specimen #2D37 at 20°C
VSA VSA WP4C

VSA VSA VSA VSA VSA WP4C WP4C WP4C WP4C WP4C
Path suction Sr suction
RH T (°C) w (%) θ e T(°C) w (%) θ e Sr (%)
(kPa) (%) (kPa)
Start point - - - - - - - 84070 20.00 7.34 0.04 2.82 5.37
Wetting 0.800 20.01 30188 12.47 0.066 2.90 8.85 30860 20.00 12.46 0.07 2.90 8.84
Wetting 0.929 20.01 9917 17.78 0.093 2.95 12.40 10610 19.90 17.78 0.09 2.95 12.39
Drying 0.802 20.01 29884 13.96 0.073 2.93 9.80 30440 19.90 13.95 0.07 2.93 9.80
Drying 0.550 20.00 80690 9.28 0.049 2.92 6.55 79170 20.50 9.28 0.05 2.92 6.56
Drying 0.165 19.97 242988 4.11 0.022 2.90 2.92 241870 20.00 4.13 0.02 2.90 2.93
Table 5-14. Comparison of VSA and WP4C: specimen #15D37 at 35°C
VSA VSA WP4C

VSA VSA VSA VSA VSA WP4C WP4C WP4C WP4C WP4C
Path suction Sr suction
RH T (°C) w (%) θ e T(°C) w (%) θ e Sr (%)
(kPa) (%) (kPa)
Start point - - - - - - - 94880 34.90 7.50 0.04 2.81 5.69
Wetting 0.797 35.03 32044 12.62 0.069 2.89 9.31 33920 34.90 12.54 0.07 2.89 9.25
Wetting 0.928 35.04 10580 17.76 0.096 2.95 12.83 12300 35.00 17.63 0.10 2.95 12.74
Drying 0.800 35.06 31533 14.02 0.077 2.86 10.44 33620 35.00 13.96 0.08 2.86 10.39
Drying 0.549 35.03 84908 8.94 0.050 2.84 6.70 88020 34.90 8.88 0.05 2.84 6.66
Drying 0.159 34.86 260279 3.99 0.022 2.89 2.94 259390 34.50 4.03 0.02 2.89 2.97
Table 5-15. VSA measurements for drying and wetting at 60°C on specimen #16D37
T Suction w θ e Sr
Path RH
(°C) (kPa) (%) (%)
Wetting 0.60 59.70 78035 7.99 0.047 2.65 6.42
Wetting 0.790 59.97 35631 12.33 0.070 2.73 9.60
Wetting 0.932 59.86 10657 17.38 0.097 2.79 13.24
Drying 0.798 60.09 34215 13.44 0.076 2.75 10.41
Drying 0.599 60.01 77602 9.40 0.054 2.69 7.44
Drying 0.168 59.99 269376 3.10 0.018 2.65 2.49
Figures 5-15-a and 5-16-a show that at high suction range, the differences in degrees of saturation
are not significant nor following a consistent trend for experiments performed at temperatures between
20 and 60°C for both the wetting and drying paths. In addition, Figures 5-15-b and 5-16-b indicate that
temperature has no effect on retention capacity of GCL in terms of gravimetric water content. Therefore,
it can be concluded that temperature has no significant effect on SWCC of GCL on the dry end of the curve.
Void ratios show a relatively consistent trend, increasing as suction decreases and vice versa (Figure 5-17).
However, a small hysteresis is observed in void ratio changes and void ratios on the drying path are higher
than those on the wetting path, except for one reading for test under 35°C at suction of 260MPa, which
-75
may be a measurement error. For all experiments, linear correlations were observed between moisture
content or void ratio and logarithm of suction with an exception for void ration of the test under 35°C as
described above.
The time required to reach equilibrium is relatively short for VSA and depends on the applied suction,
the amount of change in suction from previous step, and temperature. VSA recorded around 10 to 12
readings per hour. Figure 5-18 shows VSA measurements during testing for 20, 35, and 60°C. Clearly, the
20
18 20°C (#2D37)
16 35°C (#15D37)
60°C (#16D37)
14
12
w (%)
10
0
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 18.00
Time (day)
1000000
100000
Total suction (kPa)
10000
20°C (#2D37)
35°C (#15D37)
60°C (#16D37)
1000
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 18.00
Time (day)
Figure 5-18. VSA measurements: a) Gravimetric water content vs. time; b) Total suction vs. time
-76
rate of change in water content and suction depends on the temperature. In this study, the time to reach
equilibrium for each point was less than a week.
5.8.3 Vapour equilibrium technique (VET) experiments (Monash University)
The VET was used for SWCC measurement at Monash University for the same GCL type that was used in
this thesis (i.e. ELCOSEAL GCL grade X2000 manufactured by Geofabrics Australia). The experiments were
performed under loads of 0 and 20kPa and temperatures of 20 (published by Rouf et al., 2016) and 70°C
(unpublished data from Rouf, 2017). In these experiments, specimens were exposed to a controlled-
relative humidity environment at temperature-controlled condition. A number of different super
saturated salt solutions were used to control RH (Table 5-16). The as-received moisture content of the
GCLs was between 7 and 10% which gives a total suction of slightly less than 100MPa (i.e. 𝑅𝐻 ≅ 48%) for
GCL specimens (Rouf et al., 2016). All VET experiments followed the wetting path for total suctions lower
than 100MPa and the drying path for total suctions higher than 100MPa (Figure 5-19). For more details on
these experiments, the reader is referred to Rouf et al. (2014, 2016).
Kelvin’s equation (Eq. 5-2) was used to convert relative humidity measurements under 20°C (Figure
5-19-a) to total suctions (Figure 5-19-b). The digitized data from Rouf et al. (2016) and conversion of RH to
total suction are presented in Tables 5-17 to 5-19. In addition, the total suction measurements under 70°C
are included in Figure 5-19-b and Table 5-19 presents the data. The void ratio of the GCLs was not
measured for this experiment. Therefore, it’s not possible to calculate degree of saturation and volumetric
water content.
The results from experiments under different loads and temperatures indicate that both effects of
load and temperature become more significant for lower range of suctions. The GCL under the lowest load
and temperature showed the highest water retention capacity in terms of gravimetric water content and
vice versa.
Table 5-16. RH for solutions at 20°C
Super saturated Rouf et al. (2016) Super saturated Rouf et al. (2016)
ASTM E104 ASTM E104
salt solution (WP4C) salt solution (WP4C)
LiBr 7.09 ± 0.04 6.4 ± 0.6 KI 70.60 ± 0.12 68.9 ± 0.3
LiCl 11.05 ± 0.30 11.3 ± 0.3 NaCl 75.45 ± 0.15 75.3 ± 0.2
CH3CO2K 22.75 ± 0.03 22.5 ± 0.4 KCl 84.95 ± 0.16 84.2 ± 0.3
K2CO3 44.50 ± 0.29 43.2 ± 0.4 KNO3 93.70 ± 0.22 -

NaBr 59.30 ± 0.50 57.6 ± 0.4 K2SO4 97.70 ± 0.04 97.3 ± 0.5
-77
40
20°C #1M(0kPa,20°C)
35
#2M(20kPa,20°C)
30
#3M(0kPa,70°C)
25
w (%)
20
15
drying
10
5 (b)
(a)
0
1000 10000 100000 1000000
Total suction (kPa)
30 0.25
25
0.2
20
0.15
Sr (%)
θ (-)
15
0.1
10
drying drying
0.05
5 (c) wetting (d) wetting
0 0
1000 10000 100000 1000000 1000 10000 100000 1000000
Total suction (kPa) Total suction (kPa)
Figure 5-19. SWCC of GCL X2000 using VET performed at Monash University: a) Gravimetric water content vs. relative humidity
under 20°C (modified from Rouf et al. 2016); b) Gravimetric water content vs. total suction. Total suction are calculated using
Kelvin equation (Eq. 5-2); c) Degree of saturation vs. total suction; d) Volumetric water content vs. total suction
Table 5-17. VET measurements at 20°C and under 0kPa
Total suction
Path RH a w (%)a θ ea Sr (%)
(kPa)
Drying 0.071 357225 2.57 0.014 2.87 1.86
Drying 0.110 298099 3.94 0.022 2.78 2.95
Drying 0.228 199663 6.02 0.033 2.82 4.45
Drying 0.445 109350 7.44 0.046 2.40 6.45
Wetting 0.593 70573 8.91 0.050 2.68 6.91
Wetting 0.706 47017 12.47 0.072 2.61 9.94
Wetting 0.755 37955 13.56 0.077 2.68 10.52
Wetting 0.850 21949 16.63 0.087 2.98 11.62
Wetting 0.937 8788 20.40 0.127 2.35 18.06
Wetting 0.977 3142 32.81 0.194 2.52 27.07
a: Digitized data from Rouf et al. (2016)
-78
Table 5-18. VET measurements at 20°C and under 20kPa
Total suction w
Path RH a θ ea Sr (%)
(kPa) (%)a
Drying 0.071 357225 2.30 0.016 1.91 2.51
Drying 0.110 298099 3.45 0.022 2.32 3.09
Drying 0.228 199663 5.63 0.033 2.53 4.63
Drying 0.445 109350 6.84 0.043 2.32 6.14
Wetting 0.593 70573 7.93 0.055 1.99 8.27
Wetting 0.706 47017 11.38 0.075 2.16 10.96
Wetting 0.755 37955 12.20 0.079 2.22 11.45
Wetting 0.850 21949 14.82 0.105 1.95 15.81
Wetting 0.937 8788 19.91 0.150 1.76 23.59
Wetting 0.977 3142 27.51 0.177 2.24 25.56
a: Digitized data from Rouf et al. (2016)
Table 5-19. VET measurements at 70°C under <1kPa load
Total suction
Path w (%)a
(kPa)a
Wetting 1500 35.00
Wetting 3601 20.47
Wetting 25241 11.49
Wetting 80867 6.88
Drying 229127 4.56
Drying 409550 1.49
a: Data from Monash University (unpublished)
5.9 Summary of SWCC Experiments
A summary of all SWCC experiments performed on X2000 is presented in Table 5-20 and Figure 5-20. The
summary includes all the data from axis-translation, dew point method, and VET tests. As discussed in
Section 2.7, suction measurement for a full range of GCL SWCC is a challenging task. In this study, as
described before, matric suctions were measured for low suctions and total suctions for high suctions.
Total suction is the sum of matric suction and osmotic suction. The magnitude of osmotic suction depends
on concentration of dissolved salts. Therefore, at a specific water content, total suction is higher than
matric suction and the difference between matric and total suction increases as water content decreases
(e.g. see Leong et al., 2003). Beddoe et al. (2011) argued that, for GCLs, the difference between matric and
total suction at high range of suction is relatively negligible compared to the magnitude of suction. As a
result, no correction was performed to adjust the values of measured suctions. The magnitude of osmotic
suction for experiments in this study is not clear and more experiments are required to quantify osmotic
suctions.
-79
Table 5-20. Summary of all tests performed on GCL X2000
Target Target Total/ Starting Starting End End

Specimen
Set Load Temperature Method Device Path Matric Suction w Suction w
ID
(kPa) (°C) Suction? (kPa) (%) (kPa) (%)
1 0 20 #23D37 DPM WP4C Wetting Total 87830 7.81 11990 17.88
2 0 20 #23D37 DPM WP4C Drying Total 29580 14.75 241060 4.77
3 0 20 #23D37 DPM VSA Wetting Total 27657 13.96 10208 17.90
4 0 20 #23D37 DPM VSA Drying Total 30148 14.76 246990 4.74
5 0 20 #M1-1 VET DCH Wetting Total 70573 8.91 3142 32.81
6 0 20 #M1-2 VET DCH Drying Total 109350 7.44 357225 2.57
7 0 35 #15D37 DPM WP4C Wetting Total 94880 7.50 12300 17.63
8 0 35 #15D37 DPM WP4C Drying Total 33620 13.96 259390 4.03
13 <1 60 #M3-1 VET DCH Wetting Total 80867 6.88 1500 35.00
14 <1 60 #M3-2 VET DCH Drying Total 229127 4.56 409550 1.49
15 10 20 #17D50 ATM BC Drying Matric 0 245.71 700 103.11
17 20 20 #M2-1 VET DCH Wetting Total 70573 7.93 3142 27.51
18 20 20 #M2-2 VET DCH Drying Total 109350 6.84 357225 2.30
a: Test was performed at Monash University; DPM: Dew point method; ATM: Axis-translation method; DCH: Desiccator
chamber; BC: Barcelona Cell
Another difficulty is that different paths of drying or wetting were followed for different parts of the
curves. As discussed in Section 3.1, in the context of this thesis, the wetting path is important for the
hydration stage and the drying path for dehydration. However, on the wet side of the SWCC curve, only
the drying path was measured because the tests were time consuming and a full characterisation of both
drying and wetting paths would have required more than the time available for the thesis. Hence, priority
was given to the dehydration part of the SWCC because of the project’s focus on desiccation.
In Figure 5-20, the following labelling has been used. Data with label “0kPa,20°C” includes dew point
method experiment on specimen #23D37 and the VET experiment #1M. Labels “0kPa,35°C” and
“0kPa,60°C” refer to dew point method experiments on specimens #15D37 and #16D37, respectively.
These experiments with 0kPa load were performed on both wetting and drying paths and total suctions
were measured. Labels “10kPa, 20°C”, “20kPa, 40°C”, and “50kPa, 40°C” refer to matric suction
measurements on the drying path using axis-translation performed on specimens #17D50, #37D50, and
-80
#42D50, respectively. Label “20kPa,20°C” includes matric suctions on the drying path for specimen #11D50
and total suctions on both drying and wetting paths for experiment #2M. A detailed description of the test
conditions are presented in Table 5-20.
Figure 5-21 shows relationship between void ratio and suction. The graph presents data for all
experiments performed on X2000. The slope of change in void ratio with change in suction depends on
250
0kPa, 20°C
0kPa, 35°C
200
0kPa, 60°C
0kPa, 70°C
150 10kPa, 20°C

20kPa, 20°C
20kPa, 35°C
w (%)
100
20kPa, 40°C
50kPa, 35°C
50kPa, 40°C
50
0
0.1 1 10 100 1000 10000 100000 1000000
Suction (kPa)
(a)
100
90
80
0kPa, 20°C
70
0kPa, 35°C
60 0kPa, 60°C
Sr (%)
50 10kPa, 20°C
40 20kPa, 20°C
20kPa, 35°C
30
20kPa, 40°C
20
50kPa, 35°C
10 50kPa, 40°C
0
0.1 1 10 100 1000 10000 100000 1000000
(b) Suction (kPa)
Figure 5-20. Summary of all SWCC experiments on X2000: a) Gravimetric water content vs suction; b) Degree of saturation vs.
suction.
-81
5.0
4.5
4.0
3.5 0kPa, 20°C

e
0kPa, 35°C
0kPa, 60°C
3.0
10kPa, 20°C
20kPa, 20°C
2.5
20kPa, 35°C
20kPa, 40°C
2.0 50kPa, 35°C
50kPa, 40°C
1.5
0.1 1 10 100 1000 10000 100000 1000000
Suction (kPa)
Figure 5-21. Void ratio vs suction for all experiments on X2000
suction and three different zones can be identified. The relationship for the drying path is discussed and
the slopes of the curves are estimated in the following sections.
5.10 Effect of Void Ratio and Temperature on SWCC

5.10.1 Approach based on van Genuchten (1980) (VG version)
Ideally, the effect of void ratio in the proposed SWCC equation (Eq. 4-16) should be quantified by
conducting SWCC measurements with void ratios held constant during each experiment, and repeated at
different void ratios with same temperature. Conversely, the swelling/shrinkage of the GCL’s bentonite
that accompanies hydration/dehydration during SWCC measurements does not allow void ratio to be kept
constant. Therefore, the calibration of the equation to experimental data will be conducted based on
available experimental points in the Sr-s-e-T space.
The parameters of equation 4-16 were determined through calibration to experimental

measurements, in four steps. The procedure is based on selection of one set of data under a specific load,
temperature, and path as the reference curve. Although, in general, it would be possible to establish an
imaginary reference curve based on minimum error for a reference temperature. The four steps are:
-82
1. Establish the reference curve: Eq. 4-7 (van Genuchten, 1980) to be used to determine 𝑃0 , 𝜆, 𝑆𝑠 ,
and 𝑆𝑟𝑠 for a set of data that will be considered as the reference set of data. In this study, it is
always assumed that 𝑆𝑠 = 1.0.
2. Determination of the air-entry value of the reference curve defined in step 1.
3. Determination of 𝑒𝑟𝑒𝑓 : the void ratio at the air-entry suction defined in step 2.
4. Calibration of 𝛽𝑒 and 𝜆 for Eq. 4-16: To be done by minimizing prediction error of Eq. 4-16 for all
availabe data sets at the same temperature as the reference curve using same 𝑃0 , 𝑆𝑠 , and 𝑆𝑟𝑠
defined in step 1.
Only drying curves were used in the fitting exercises presented in this section because the behaviour
of the GCL under low suction range following the wetting path is not clear. Moreover, no calibration has
been proposed for prediction of effect of temperature for two reasons. First, in this thesis, the theory by
Grant and Salehzadeh (1996) was followed, which does not consider any calibration. Second, as mentioned
earlier, the available data for the effect of temperature on the SWCC of GCL is limited and further
investigation is required to be able to improve the theory and predictions.
Following the method described above, it is required to establish a reference curve. In theory, there
is no constraint on the selection of the reference curve or even using an imaginary representative
reference curve since the solution for this problem is not unique. Clearly, using different reference curves
results in different 𝑃0 and 𝑒𝑟𝑒𝑓 which may lead to different solutions for 𝛽𝑒 and 𝜆. In this section, the sets
100
90
80
70
𝑠𝑎𝑒 = 67𝑘𝑃𝑎
60
Sr (%)
50
40
Reference curve:
30 𝑃 = 168𝑘𝑃𝑎
0kPa, 20°C 𝜆 = 0.296, 𝑆𝑟𝑠 = 0
20 𝑅 2 = 0.9907
10kPa, 20°C
10 Reference curve (VG) Eq. 4-7
0
1 10 100 1000 10000 100000 1000000
Suction (kPa)
Figure 5-22. van Genuchten (1980) SWCC fit for reference curve on the drying path
-83
of data corresponding to experiments under temperature of 20°C and loads of 0 and 10kPa were selected
for determination of the reference curve. This is because the number of data points is large and almost a
complete range of drying curve has been measured under this condition (i.e., tests performed on
specimens #17D50, #2D37 and #1M on drying path). Figure 5-22 shows the van Genuchten SWCC (Eq. 4-7)
fit for the reference curve. A good fit (𝑅 2 = 0.9907), based on the least-squares method, was obtained
for values of 𝑃 = 168𝑘𝑃𝑎, 𝜆 = 0.296, and 𝑆𝑟𝑠 = 0 (Table 5-21).
In step 2, the air-entry value of the reference curve was determined. An air-entry value of 𝑠𝑒𝑎 =
67𝑘𝑃𝑎 can be estimated for the reference curve by drawing a tangent line to the curve passing through
the inflection point (Figure 5-22). In step 3, void ratio at the air-entry value was estimated. Figure 5-23
shows the relationship between void ratio and suction. Three distinct slopes can be identified for the full
range of suction. This is consistent with behaviour of soils such as white clay (Khalili et al., 2008).
In step 4, the least-squares method was used to calibrate the values of 𝛽𝑒 and 𝜆 for the proposed
SWCC equation (4-16) based on all experiments under 20°C.  was allowed to change in order to provide
more flexibility in the fitting exercise. Table 5-22 presents the fitting parameters for the proposed SWCC
and Figure 5-24 shows prediction for the experiments under 20°C on the drying path. A good fit was
achieved with a high coefficient of determination of 0.9982 for these experiments. Calibration of the
model based on the available data suggests 𝜆 = 0.333 and 𝛽𝑒 = 1.46. The other parameters were
obtained from the reference curve. Therefore, 𝑃0 = 168 𝑘𝑃𝑎, 𝑒𝑟𝑒𝑓 = 4.46, 𝜎0 = 0.0727 𝑁/𝑚, 𝑆𝑠 = 1.0,
5.00
4.50
y = -0.106ln(x) + 4.9051 10kPa, 20°C
R² = 1
4.00
0kPa, 20°C
𝑒𝑟𝑒𝑓 = 4.46
3.50
e
y = -0.365ln(x) + 5.9741
3.00 R² = 0.997
2.50 y = -0.025ln(x) + 3.1344

R² = 0.0695
2.00
1 10 100 1000 10000 100000 1000000
Suction (kPa)
Figure 5-23. Determination of 𝑒𝑟𝑒𝑓
-84
and 𝑆𝑟𝑠 = 0.0. Finally, the calibrated model under 20°C condition was used to predict degrees of saturation
for the higher temperature experiments (𝑅 2 = 0.9227). Note that 𝜆 obtained for the proposed SWCC is
about 12% higher than that obtained for the conventional VG equation. This is due to the fact that more
points were fitted to the proposed equation and some dependence of lambda on the range of void ratios
considered would be expected.
Figure 5-25 shows that the predictions for the dry side of the curves is good. On the wet side,
predictions are better for 35°C than those of 40°C. The reason for the poor estimates on 40°C is that the
GCLs were not fully hydrated and as discussed before, probably scanning curves were followed instead of
the drying path.
100
90
80
70
60
Sr (%)
50 0kPa, 20°C
10kPa, 20°C
40
20kPa, 20°C
30
Predictions for 0kPa, 20°C
20 Predictions for 10kPa, 20°C
10
0
0.1 1 10 100 1000 10000 100000 1000000
Suction (kPa)
Figure 5-24. Calibration of 𝛽𝑒 and 𝜆 for all experiments under 20°C (𝑅2 = 0.9982)
Table 5-21. van Genuchten (1980) SWCC (Eq. 4-7) fitting parameter for the reference curve
𝑃
Curve 𝜆 𝑆𝑠 𝑆𝑟𝑠 Based on experiments
(kPa)
Reference curve (VG) Eq. 4-7 168 0.296 1.0 0.0 0kPa, 20°C+10kPa, 20°C
Table 5-22. Fitting parameters for the proposed VG version of the SWCC (Eq. 4-16)
𝑃0 𝜎0
Curve 𝜆 𝛽𝑒 𝑒𝑟𝑒𝑓 𝑆𝑠 𝑆𝑟𝑠
(kPa) (N/m)
Proposed SWCC (Eq. 4-16) 168 0.333 1.46 4.46 0.0727 1.0 0.0
-85
14 14
12 12
10 10
8 8
Sr (%)
Sr (%)
6 6
4 4
0kPa, 35°C 0kPa, 60°C
2 2 Predictions for 0kPa, 60°C
0 0
10000 100000 1000000 10000 100000 1000000
100 100
80 80
60 60
Sr (%)
Sr (%)
40 40
20 20kPa, 35°C 20 20kPa, 40°C

Predictions for 20kPa, 35°C Predictions for 20kPa, 40°C
0 0
1 10 100 1000 1 10 100 1000
100 100
80 80
60 60
Sr (%)
Sr (%)
40 40
20 50kPa, 35°C 20 50kPa, 40°C

0 0
1 10 100 1000 1 10 100 1000
Figure 5-25. Predictions based on VG version for experiments under elevated temperature conditions (𝑅2 = 0.9227)
5.10.2 Approach based on Fredlund and Xing (1994) (FX version)
Similar steps to the four steps described in Section 5.10.1 were followed for determination of the
parameters. In step 1, the reference curve was established. The reference data sets used were the same.
Clearly, the Fredlund and Xing (1994) SWCC provides better estimates than the van Genuchten (1980)
model for the dry end of the curve (Figure 5-26).
-86
Fitting parameters for the reference curve are presented in Table 5-23. In step 2, the air-entry was
estimated as 𝑠𝑎𝑒 = 62𝑘𝑃𝑎 and in step 3, a reference void ratio of 4.46 was determined. In step 4, the
least-squares method was used to calibrate 𝛽𝑒 , 𝑛𝑓 , and 𝑚𝑓 for all experiments under 20°C (Figure 5-27).
Table 5-22 presents the fitting parameter for the proposed SWCC (Eq. 4-20) based on Fredlund and Xing
(1994) model (𝑅 2 = 0.9978). Predictions of the model for the experiments under elevated temperature
100
90
80
70
60
Sr (%)
50
Reference curve:
𝑎𝑓 = 277𝑘𝑃𝑎
40
𝑛𝑓 =0.90, 𝑚𝑓 = 1.10
𝜓𝑟 = 2100𝑘𝑃𝑎
30 𝑅 = 0.9951
20 0kPa, 20°C
10kPa, 20°C
10
Reference curve (FX) Eq. 5-14
0
1 10 100 1000 10000 100000 1000000
Suction (kPa)
Figure 5-26. Fredlund and Xing (1994) SWCC fit for reference curve on the drying path
100
90
80
70
60
Sr (%)
50 0kPa, 20°C
10kPa, 20°C
40
20kPa, 20°C
30
20 Predictions for 10kPa, 20°C
10
0
0.1 1 10 100 1000 10000 100000 1000000
Suction (kPa)
Figure 5-27. Calibration of 𝛽𝑒 , 𝑛𝑓 , and 𝑚𝑓 for all experiments under 20°C (𝑅2 = 0.9978)
-87
conditions are presented in Figure 5-28. The predictions using the FX version (𝑅 2 = 0.9346) are better
than those of the VG version (𝑅 2 = 0.9227). However, predictions for experiments under 40°C remains
poor because of insufficient hydration of GCL prior to testing.
14 14
12 12
10 10
8 8
Sr (%)
Sr (%)
6 6
4 4
0kPa, 35°C 0kPa, 60°C
2 2
0 0
10000 100000 1000000 10000 100000 1000000
100 100
90 90
80 80
70 70
60 60
Sr (%)
Sr (%)
50 50
40 40
30 30
20 20kPa, 35°C 20 20kPa, 40°C
10 Predictions for 20kPa, 35°C 10 Predictions for 20kPa, 40°C
0 0
1 10 100 1000 1 10 100 1000
100 100
90 90
80 80
70 70
60 60
Sr (%)
Sr (%)
50 50
40 40
30 30
20 50kPa, 35°C 20 50kPa, 40°C
10 Predictions for 50kPa, 35°C 10 Predictions for 50kPa, 40°C
0 0
1 10 100 1000 1 10 100 1000
Figure 5-28. Predictions based on FX version for experiments under elevated temperature conductions (𝑅2 = 0.9346)
-88
Table 5-23. Fredlund and Xing (1994) SWCC (Eq. 4-19) fitting parameter for the reference curve
𝑎𝑓 𝑛𝑓 𝑚𝑓 𝜓𝑟
Curve Based on experiments
(kPa) (kPa)
Reference curve (FX) Eq. 4-19 277 0.90 1.10 2100 0kPa, 20°C+10kPa, 20°C
Table 5-24. Fitting parameters for the FX version of the proposed SWCC Eq. 4-20
𝑎𝑓0 𝑛𝑓 𝑚𝑓 𝛽𝑒 𝑒𝑟𝑒𝑓 𝜎0 𝜓𝑟
Curve
(kPa) (N/m) (kPa)
Proposed FX SWCC Eq. 4-20 277 1.06 1.20 1.60 4.46 0.0727 2100
5.11 Comparison to Existing GCL SWCCs
This Section presents a comparison of SWCCs of X2000 and SWCCs from the literature including the
experiments by a) Southen and Rowe (2007) on two types of GCLs, one granular (SR07G1) and one
powdered Na-bentonite(SR07G2), under different loads following the drying path, and b) Abuel-Naga and
Bouazza (2010) on a powdered Na-bentonite (AB10) under 50kPa at 25°C following the wetting path
(Figure 5-29-a). A summary of the specifications of the GCLs used in these experiments is presented in
Table 5-25.
The selected experiments from the literature have some similarities with X2000 and provide a wider
view of the behaviour of GCLs. The GCLs used in the experiments by Southen and Rowe (2007), i.e. SR07G1
and SR07G2, and Abuel-Naga and Bouazza (2010), i.e. AB10, have very similar configuration to X2000.
Those GCLs are needle-punched and thermally treated and their covers are nonwoven geotextiles and
their carriers are nonwoven geotextiles and reinforced with a slit-film or a woven geotextile (Table 5-25).
Table 5-25. Comparison of GCL properties
Experiment This study SR07G1 SR07G2 AB10

ELCOSEAL
GCL Bentofix NS Bentofix BFG 5000 ?
X2000
Bentonite type PB-Na GB-Na PB-Na PB-Na
Bonding NPTT NPTT NPTT NPTT
Cover geotextile NW NW NW NW
Carrier geotextile NW+W W+slit film W+slit film W+slit film
MGCL(g/m2) 4960 4650 5500+800 5500
Mbentonite (g/m2) 4250 4340 5000+800 5000
Mcover 300 200 300 ?
Mcarrier 410 105 200 ?
SR07G1: Southen and Rowe (2007) on G1; SR07G2: Southen and Rowe (2007) on G2; AB10: Abuel-Naga and Bouazza (2010);
PB-Na: Powder sodium bentonite; GB-Na: Granular sodium bentonite; NPTT: Needle punched thermally treated; NP: Needle
punched; NW: Non-woven; W: Woven
-89
Furthermore, the mass per area of bentonite of SR07G1 (4340 g/m2) is close to X2000 (4250 g/m2).
However, the masses of bentonite per area of SR07G2 and AB10 are considerably higher than X2000.
As described in Section 5.7, degree of saturation is calculated based on the void ratio of GCL (bulk
void ratio). Figure 5-29-b shows the change in void ratio of GCL as a result of change in suction. It shows
that void ratio of GCL is controlled by the applied vertical stress and a higher load can result in a lower void
100
80
0kPa, 20°C 0kPa, 35°C

60
0kPa, 60°C 10kPa, 20°C
Sr (%)
20kPa, 20°C 20kPa, 35°C
40 20kPa, 40°C 50kPa, 35°C
50kPa, 40°C SR07G1 (0kPa)
SR07G1 (3kPa) SR07G1 (100kPa)

20
SR07G2 (0kPa) SR07G2 (0.5kPa)
SR07G2 (100kPa) AB10 (50kPa,25°C)

0
0.1 1 10 100 1000 10000 100000 1000000
Suction (kPa)
(a)
6.0
5.0
4.0
e
3.0
2.0
1.0
0.0
0.1 1 10 100 1000 10000 100000 1000000
(b) Suction (kPa)
Figure 5-29. SWCC tests on X2000 and comparison with SWCC of GCL from literature. a) Degree of saturation vs. suction; b)
Void ratio vs. suction. SR07: Southen and Rowe (2007); AB10: Abuel-Naga and Bouazza (2010)
-90
ratio. Other factors such as mass per area of bentonite can affect the trend. The data on the effect of
temperature on void ratio is very limited and does not show a significant effect. However, for all specimens
(except SR2007 3kPa), a similar trend for change in void ratio with change in suction was observed and all
specimens showed relatively large void ratio change for a range of suctions between 10 and 10000kPa,
while change in void ratio for higher suctions became insignificant.
Overall, comparison of the results with existing SWCC of GCLs show general agreement, especially for
the dry end of the retention curves (Figure 5-29-a). The relatively large difference between air-entry value
of G2 and other GCLs can be due to higher mass per area of bentonite. Void ratios vary considerably on
both wet and dry end of curves (Figure 5-29-b). The reason may be the different methods used in
calculation of the bulk void ratios of GCL or inaccurate measurements of volume change.
5.12 Application to Existing GCL SWCC Measurements
A large number of experiments have been conducted on the water retention capacity of GCLs. Conversely,
a limited number of investigations exist that have been conducted systematically to determine the effect
of change in volume and temperature on SWCC of GCL (as discussed in Section 2.8). In this section, the
SWCCs developed in Section 5.10 of this thesis are applied to experiments by Southen and Rowe (2007).
These experiments will be used in Chapter 7 for validation of the theory by modelling two sets of column
studies reported by Southen and Rowe (2005a).
Two types of GCLs were examined by Southen and Rowe (2007). Specification of the GCLs are
presented in Table 5-25. All tests were conducted following the drying path at room temperature. SWCC
tests were performed on SR07G1 under 0kPa net vertical stresses using pressure plate extractor from
nearly saturated condition (~10kPa) to a suction of around 600kPa. Tests under loads of 3 and 100kPa
were conducted using pressure membrane extractors for suctions of approximately 1500, 2000, 4500, and
9000kPa. In addition, under 100kPa load, SWCC of GCL was determined for a low suction around 10kPa.
The experimental program for SR07G2 was very similar to SR07G1. The suction range in pressure plate
extractor tests (under 0kPa load) were extended to 1500kPa suction and in the pressure membrane
extractor tests a net stress of 0.5kPa was applied to GCLs instead of 3kPa.
For determination of parameters of the VG version of the developed SWCC (Eq. 4-16), an alternative
procedure is considered in this section. First, an imaginary reference curve was fitted to all available data
using the least-squares method (Figure 5-30 and Table 5-26). The reason for fitting to all available data is
that, first, the number of data points under each condition is not large enough to perform a reliable
parameter fitting and second, the data points are scattered for both G1 and G2 experiments. The
coefficients of determination for G1 and G2 are 0.7906 and 0.8009, respectively. The reference curve
-91
100
90
G2:
80
G1:
70 𝑠𝑎𝑒 = 145𝑘𝑃𝑎
60 G2 reference curve:
𝑃 = 1414𝑘𝑃𝑎
Sr (%)
50 𝜆=0.309, 𝑆𝑟𝑠 = 0
𝑅 2 = 0.8009
40
Experiments-G1
30
Experiments-G2
VG Ref. Curve-G1 G1 reference curve:
20
𝑃 = 363𝑘𝑃𝑎
VG Ref. Curve-G2 𝜆=0.285, 𝑆𝑟𝑠 = 0
10 𝑅 2 = 0.7906
0
1 10 100 1000 10000 100000 1000000
Suction (kPa)
Figure 5-30. Determination of the VG reference curves for G1 and G2 (data digitized from Southen and Rowe, 2007)
provides 𝑃0 and 𝜆. This curve can be used as representative of GCL behaviour based on the best fit values
if the original van Genuchten (1980) SWCC is to be used. Next, air-entry value of the reference curves and
their corresponding void ratios at air-entry suction, 𝑒𝑟𝑒𝑓 , are estimated. The coefficients of determination
of the fitted curves are low, especially for void ratios (Figure 5-31); therefore, the estimates can be
inaccurate. Finally, the least-squares method is used to calibrate 𝛽𝑒 and 𝜆 based on the established values
of 𝑃0 and 𝑒𝑟𝑒𝑓 in the previous steps. Figure 5-32 shows predictions of the calibrated SWCCs.
Following the procedure described above, values of 𝑃0 , 𝜆, 𝑒𝑟𝑒𝑓 , and 𝛽𝑒 are 363kPa, 0.324, 3.20, and
1.002 for SR07G1 and the parameters for SR07G2 are 1414, 0.309, 2.58, and 0, respectively (Table 5-27).
The value of 𝛽𝑒 for SR07G2 is 0. This indicates that no correlation is found between deformation and
change in SWCC. On the other hand, volume change shows a significant effect on SWCC of SR07G1 and the
coefficient of determination for G1 improved from 0.7906 to 0.8624. The tests were performed at room
temperature; however, the exact temperature is unknown. Therefore, a reference tension surface of
0.0727 N.m-1 corresponding to 20°C was adopted.
Table 5-26. van Genuchten (1980) SWCC (Eq. 4-7) fitting parameter for the reference curves
𝑃
Reference Curve 𝜆 𝑆𝑠 𝑆𝑟𝑠 Based on experiments
(kPa)
G1-Reference curve (VG) Eq. 4-7 363 0.285 1.0 0.0 All G1 experiments
G2-Reference curve (VG) Eq. 4-7 1414 0.309 1.0 0.0 All G2 experiments
-92
Table 5-27. Fitting parameters for VG version of the proposed SWCC (Eq. 4-16)
𝑃0 𝜎0
Curve 𝜆 𝛽𝑒 𝑒𝑟𝑒𝑓 𝑆𝑠 𝑆𝑟𝑠
(kPa) (N/m)
G1- VG SWCC (Eq. 4-16) 363 0.324 1.0 3.20 0.0727 1.0 0.0
G2- VG SWCC (Eq. 4-16) 1414 0.309 0.0 2.58 0.0727 1.0 0.0
4.5
4 y = -0.272ln(x) + 4.5651
R² = 0.3945
3.5
2.5
e
G1:
2
𝑒𝑎𝑒 = 3.20
1.5 Experiments-G1 G2:
Experiments-G2 𝑒𝑎𝑒 = 2.58
1
y = -0.266ln(x) + 4.3457
Log. (Experiments-G1)
R² = 0.3764
0.5
Log. (Experiments-G2)
0
1 10 100 1000 10000 100000 1000000
Suction (kPa)
Figure 5-31. Void ratio at air entry for G1 and G2 (data digitized from Southen and Rowe, 2007)
100
90
80
70
60
Sr (%)
50
40
Experiments-G1
30 VG Predictions-G1
VG Ref. Curve-G1
20 Experiments-G2
VG Predictions-G2
10 VG Ref. Curve-G2
0
1 10 100 1000 10000 100000 1000000
Suction (kPa)
Figure 5-32. VG version of the SWCC for G1 (R2=0.8624) and G2 (R2=0.8009)
-93
For the FX version of SWCC, using the least-squares method for determination of 𝑎𝑓0 leads to an
unacceptable fitting with extremely sharp slopes because the data is highly scattered. Therefore, a trial
and error approach was used and values of 𝑎𝑓0 for G1 and G2 were assumed to be 360 and 1600kPa,
respectively. Moreover, a value of 𝜓𝑟 = 2100𝑘𝑃𝑎 was selected for the same reason as above. Fitting
parameter for the reference curves based on Fredlund and Xing (1994) model are presented in Table 5-28
100
90
G2:
80
G1:
70 𝑠𝑎𝑒 = 160𝑘𝑃𝑎
G2 Reference curve:
60 𝑎𝑓 = 1600𝑘𝑃𝑎
𝑛𝑓 = 1.23, 𝑚𝑓 = 0.59
Sr (%)
50 𝜓𝑟 = 2100𝑘𝑃𝑎
𝑅 = 0.8039
40
30 Experiments-G1
Experiments-G2 G1 Reference curve:
20 𝑎𝑓 = 360𝑘𝑃𝑎
FX Ref. Curve-G1 𝑛𝑓 = 2.32, 𝑚𝑓 = 0.45
10 FX Ref. Curve-G2 𝜓𝑟 = 2100𝑘𝑃𝑎
𝑅 = 0.8290
0
1 10 100 1000 10000 100000 1000000
Suction (kPa)
Figure 5-33. Determination of the FX reference curves for G1 and G2 (data digitized from Southen and Rowe, 2007)
100
90
80
70
60
Sr (%)
50
40
Experiments-G1
30 FX Predictions-G1
FX Ref. Curve-G1
20 Experiments-G2
FX Predictions-G2
10 FX Ref. Curve-G2
0
1 10 100 1000 10000 100000 1000000
Suction (kPa)
Figure 5-34. FX version of the SWCC for G1 (R2=0.8565) and G2 (R2=0.8039)
-94
and the parameters for the FX version of the proposed SWCC are tabulated in Table 5-29. It seems that
both reference curves provide better estimates for the dry end of curves than those based on van
Genuchten (1980) model (Figure 5-33). No correlation was found between degree of saturation and void
ratio for G2 since 𝛽𝑒 is 0 (Figure 5-34), which is the same conclusion drawn earlier from the fitting exercise
for G2 using VG version of the proposed SWCC.
Table 5-28. Fredlund and Xing (1994) SWCC (Eq. 4-19) fitting parameter for the reference curves
𝑎𝑓 𝜓𝑟
Curve 𝑛𝑓 𝑚𝑓 Based on experiments
(kPa) (kPa)
G1-Reference curve (FX) Eq. 4-19 360 2.32 0.45 2100 All G1 experiments
G2-Reference curve (FX) Eq. 4-19 1600 1.23 0.59 2100 All G2 experiments
Table 5-29. Fitting parameters for the FX version of the proposed SWCC Eq. 4-20
𝑎𝑓0 𝜎0 𝜓𝑟
Curve 𝑛𝑓 𝑚𝑓 𝛽𝑒 𝑒𝑟𝑒𝑓
(kPa) (N/m) (kPa)
G1 - FX SWCC Eq. 4-20 360 1.65 0.65 1.17 3.18 0.0727 2100
G2 - FX SWCC Eq. 4-20 1600 1.23 0.59 0 2.58 0.0727 2100
5.13 Summary and Limitations of Predictions
Different methods were used for SWCC measurements of GCL to cover the full range of suctions because
the composite nature of GCL significantly limits the applicability of the available methods. One important
issue was the possibility of capillary break at the interface between the geotextile and the ceramic disk in
the axis-translation method. In addition, SWCCs were measured mostly for the drying path because of
difficulties associated with SWCC measurements under the wetting path, especially for the wet side of the
curve with suctions lower than 3MPa.
The proposed equations performed well in predicting the effects of void ratio on SWCC based on
available data. It was found that the air-entry value increases as the net vertical stress increases under the
same temperature. Incorporating the effect of void ratio led to an improvement in fitting SWCC data to
analytical curves for the GCL used in this study. In addition, a similar improvement was observed in
analysing SWCC data for one GCL (G1) tested by Southen and Rowe (2007) but not for another one (G2).
The SWCCs underestimated the effects of elevation of temperature, especially for the wet side of the
curves. Therefore, it seems that the model based on change of surface tension with temperature is not
sufficient for predicting the effect of temperature on SWCC and other mechanisms may be involved, such
as changes in pore fluid chemistry as a result of increases in temperature. On the other hand, modified
-95
versions of van Genuchten (1980) and Fredlund and Xing (1996) appear to have equal capabilities in
representing experimentally determined SWCCs and their changes with void ratio and temperature.
The dependence of soil water characteristic curves of GCL on temperature and void ratio is difficult
to establish because of the highly time-consuming nature of the measurements, the complexity of the
experimental procedure and the presence of hysteresis in the curves in addition to the fundamental
difficulty of establishing void ratio of GCL accurately. The aim of the curve-fitting exercise in the previous
sections was to provide the best possible equation representing the SWCC, based on available
experimental data, while accounting for the effects of void ratio and temperature. The van Genuchten
version of the equation will be used in the thermo-hydro-mechanical modelling conducted in later
chapters to predict the risk of desiccation of GCLs subjected to thermal gradients.
Three limitations of the SWCC curves obtained here must be kept in mind. First, a full characterisation
of the dependence of degree of saturation on suction, void ratio and temperature requires a larger number
of experimental points than those developed here. Second, hysteresis has been neglected here, and
although measurements along the wetting and drying paths have been made separately, there were not
enough points to develop two separate drying and wetting equations, even less to follow the hysteresis
along cycles of wetting and drying. Third, the calculated degrees of saturation in this study were based on
the bulk void ratio of GCL (Figure 5-20-b). However, X-ray images taken from specimens under different
suctions revealed considerable cracking may occur (Figure 5-35). Therefore, it is important to further
investigate the effect of cracks on the behaviour of GCL during SWCC tests and provide a better estimate
of void ratios.
(a) (b) (c)
Figure 5-35. X-ray image of GCL: a) specimen #42D50 at nearly fully saturation; b) specimen #12D50 at matric suction of 170kPa;
c) specimen #11D50 at matric suction of 350kPa
-96
6 Column Studies of Desiccation of GCL
Chapter 6
Column Studies of Desiccation of GCL
6.1 Introduction
In this chapter, laboratory investigations of the desiccation of composite geosynthetic clay lining systems
under brine pond conditions on coal seam gas sites are presented. The investigations were made of two
parts which, together with the SWCC experiments reported in Chapter 5, generated a data set that will be
used for the validation of the non-linear thermo-hydro-elastic theory described in Chapter 4.
The first part of the investigation consists of a set of experiments to determine key material
properties, namely mechanical properties of GCL, hydraulic and thermal conductivities of GCL and subsoil,
and SWCC of subsoil. These are important for validation of the THM theory because they allow the
reduction of the need for parameter fitting and back calculations to a minimum, and therefore enhance
the quality of the validation exercise.
The second part consists of a soil column study that replicates, to the extent possible in the laboratory,
the conditions encountered on brine ponds. The column tests were conducted in two columns under
identical conditions in order to confirm the repeatability of all findings.
6.2 Materials
An HDPE-GMB was used as the primary barrier with a thickness of 2mm. The GCL used in the column
studies was ELCOSEAL geosynthetic clay liner grade X2000 manufactured by GEOFABRICS Australia. The
GCL acts as the secondary barrier. Specifications and manufacturing properties and SWCC of the GCL have
been described in Chapter 5. Two different sandy soils, a well graded sand (SW) and a poorly graded sand
-97
100% 2200
90%
2100 SW
80% SP
70% ρ Max
2000
ρ dry (kg/m3)
Percent Finer
60%
50% 1900
40%
30% 1800
SW
20%
1700
10% SP
0% 1600
0.01 0.1 1 10 6 8 10 12 14 16 18
Equivalent Particle Size (mm) w (%)
Figure 6-1. Particle size distribution and compaction curves for SW and SP
(SP), were initially tested as candidate subsoils to be used in the column. Figure 6-1 illustrates particle size
distribution (ASTM D2487, 2011) and compaction curves (ASTM 698, 2012) for the SW and SP soils. The
SW was selected after performing a number of trial hydration tests, so as to maximize the extent of
hydration of the GCL prior to heating.
6.3 Experimental Program
Two identical column tests were conducted and a set of data was generated for desiccation of GCL in
composite geosynthetic clay lining systems. Additional experiments were conducted to determine
material properties: permeability, thermal conductivity and mechanical properties. The column studies
followed three main stages: 1) filling the column with subsoil material and reaching equilibrium under
gravity under overburden stress of 5kPa (14 days), 2) installation of a composite geosynthetic clay lining
system on top of the subsoil and hydration of GCL from subsoil (44 days), and 3) heating (39 days);
therefore, the total length of the experiment was 97 days (Table 6-1). The following factors were
considered in designing the column study to represent brine pond conditions:
1. A well-graded sand was selected for subsoil with an initial gravimetric water content of 11% which
was able to hydrate the GCL to a gravimetric water content of nearly 100% within 45 days of
hydration under isothermal conditions.
2. Moisture content of the given subsoil height was allowed to reach equilibrium under gravity before
installation of the composite lining system. The top and bottom of the column were sealed during
the experiment to prevent moisture loss. O-rings were used for sealing purposes.
3. An overburden stress of 20kPa was selected that represents a 2m height of fluid in a brine pond.
This load was kept constant during hydration and heating stages for the sake of simplicity.
-98
Table 6-1. Description of experiment stages
Duration Load Top temperature Bottom temperature

Stage Start day End day
(days) (kPa) (°C) (°C)
Equilibrium 0 14 14 5 20±1°C 20±1°C
Hydration 14 58 44 20 20±1°C 20±1°C
Heating 58 97 39 20 78±1°Ca 23±1°Cb
a: the target temperature was 80°C; however, the actual temperature in piston reached 78±1°C because of heat loss in the
circulation loop; b: temperature was increased to a target temperature of 23°C two hours after the top temperature was
increased to target temperature of 80°C.
4. A high target temperature of 80°C was selected for the surface temperature during heating to
represent a condition which is expected to occur in practice. The actual temperature in the heating
piston reached 78±1°C because of heat loss.
Equilibrium and hydration stages were under isothermal conditions with temperature at the top and
bottom of the column at constant 20±1°C. During the heating period, a constant temperature of 78±1°C
was maintained at the top of the lining system and temperature at the bottom was increased to a constant
temperature of 23±1°C two hours after heating of the top of soil column had started. This was done to
simulate the transient increase of the bottom temperature because a constant temperature of 20°C was
assumed for an aquifer at 60 mm below bottom of the column. The depth of the aquifer was calculated
based on water content profile prior to heating. In other words, at equilibrium under gravity, an initial
gravimetric water content of 11% changes to a profile of water content that is equivalent of a condition
that the aquifer is at a depth of 660 mm from surface for the subsoil used.
The height of brine in a pond can vary in practice from zero in an empty pond to roughly 5m.
Therefore, the lining system is expected to experience very low pressures around 2 to 5 kPa during the
hydration stage from the facilities installed above the lining system. The pressure can ultimately increase
to nearly 50 kPa during operation with 5m of brine.
6.4 Testing Apparatus
The columns were made of Perspex tubes with diameter of 190mm. The height of the soil column was
600mm and a composite lining system consisting of the HDPE-GMB on top of the GCL was installed on top
of each 600mm soil column (Figure 6-2). The GMB was cut in a diameter of 190mm using a steel cutting
die and was able to move freely upward and downward in the column. A loading frame and an aluminium
piston were used to apply load to the top of the composite lining system (Figure 6-2). The temperature
was controlled on the top and bottom of the column using two cooling/heating units (Julabo F12-ED
Refrigerated/Heating Circulator) by circulating temperature-controlled liquid through the temperature
control compartments at the top and bottom of the soil columns during the three stages described above.
-99
Oil (Julabo Thermal G oil) was used in the heating system to control temperature at the top and tap water
was used for bottom of the column. Heating and cooling capacities of the units are 2 and 0.16 kW,
respectively. Figure 6-2 shows a diagram of the loading frame and column.
Four TDRs were installed horizontally into each column with 150mm spacing between the TDRs and
positioned 75mm from the bottom and top of the column. The top TDR only measures water content
(MP306-ICT International). The other three TDRs measure both water content and temperature of the soil
(CS655-Campbell Scientific). This is because MP306 is able to handle temperatures as high as 90°C; while,
maximum temperature for CS655 is limited to 70°C. Using an MP306 at the top of column hence allowed
a high target temperature of 80°C for the heating piston. Calibrations of the TDRs and thermocouples are
presented in Appendix D.
The elevations of the TDRs are illustrated in Figure 6-2. The top of the base plate is considered as
reference elevation. The three CS655 were installed at elevations of 75, 225, and 375mm and the MP306
was installed at 525mm. In addition, two linear variable displacement transducers (LVDT) measured the
displacement of the top of the loading piston and base of each column and one thermocouple measured
the temperature of liquid delivered into each aluminium loading piston.
(c)
Inner diameter = 190mm
(a) (b)
Figure 6-2. Schematics of the soil column test apparatus: a) Front view; b) Side view; c) Cross section.
-100
Data from the TDRs and thermocouples were collected at 1 minute intervals using a CR800 data logger
(Campbell Scientific) and the average over 15min was used to reduce the volume of data. Figure 6-3
illustrates the TDRs and the data logger. A separate data logger, a USB-1608G (Measurement Computing),
was used to collect data from the LVDTs with a frequency of one reading per second and the average over
(a) (b)
MP306 CS655
(c)
CR800
Figure 6-3. TDRs and data logger: a) MP306; b) CS655; c) CR800. Figures are not drawn to scale.
-101
15 minutes was recorded. A higher frequency of reading was selected for LVDTs compared to the other
sensors because a higher noise was observed in the LVDT readings.
6.5 Selection of Subsoil
After performing a number of hydration tests, the SW was selected as subsoil to be used in the desiccation
experiments. The reason is that the SW with a significantly lower initial water content than the SP can
hydrate a GCL to a higher water content during 4 weeks (Figure 6-4). It was observed that the SW with an
initial gravimetric water content of 11% was able to hydrate a GCL to 81 and 100% gravimetric water
contents during 28 and 45 days of hydration, respectively, whereas the SP with 18% water content was
able to hydrate a GCL to 71 and 78% during 28 and 35 days of hydration, respectively (Table 6-2).
Therefore, the SW and the initial gravimetric water content of 11% was selected for subsoil material.
120
Gravimetric Water Content of GCL (%)
100
80
60
SW (w0=11%)
40
SP (w0=16%)
20
SP (w0=18%)
0
0 20 40 60 80
Time (day)
Figure 6-4. Hydration of GCL from S1 and S2 (𝑤0 is initial gravimetric water content of subsoil).
0.54 100
e Sr 90
0.52
80
0.50 70
0.48 60
Sr (%)
50
e
0.46 40
0.44 30
20
0.42
10
0.40 0
0 2 4 6 8 10 12 14 16
w (%)
Figure 6-5. Resulting void ratio vs gravimetric water content vs degree of saturation: results from the compaction tests on SW using
a standard effort
-102
Table 6-2. Hydration of GCL from different subsoils
Initial Initial 𝑀𝐺𝐶𝐿 GCL gravimetric water content (%)c after

Subsoil subsoil Stress
GCL ID subsoil dry
type (kPa) 7 14 21 28 35 45
𝑤0 (%) Sr (%) (g/m2)
days days days days days days
#1D190a SW 11 57 5 5097 - - 58 - - -
#2D190a SW 11 57 5 4977 - - - 81 - 100
#2D160b SW 11 56 20 4837 - 42 - - - -
#1D160b SP 18 72 20 4760 49 61 71 - - -
#3D160b SP 18 72 20 5250 - 54 64 71 78 -
#4D190a SP 16 63 20 5120 35 43 52 61 - -
a: height of column = 600mm; b: height of column = 260mm; c: initial water content of GCL ≅9%.
Figure 6-5 shows the resulting void ratios in the standard compaction tests for different target water
contents for SW. It illustrates that a void ratio of approximately 0.45 can be achieved for 11% gravimetric
water content if the subsoil (SW) is compacted using a standard compaction effort (in a compaction
mould). Whereas, for compaction of the subsoil in the column studies in this thesis, the standard
compaction hammer was not used because of the likelihood of damaging the TDRs and the Perspex
columns. Instead, the subsoil was compacted to a target void ratio of 0.50 in layers of 50 mm soil with a
target gravimetric water content of 11% using a wooden hammer. Therefore, the moisture content in
terms of volumetric water content and degree of saturation were, 19% and 58%, respectively. Water
content specimens were taken for each layer from the mixing batch according to ASTM D2216, 2010.
6.6 Permeability of GCL and Subsoil
Saturated hydraulic conductivities of the GCL and subsoil were measured following the guidelines from
ASTM D5084 (2010) and ASTM D5856 (2015), respectively. The constant head method was conducted for
the GCL using a flexible-wall permeameter and the falling head and constant head tailwater method was
performed for the sandy subsoil using a rigid-wall permeameter. Tap water with a total dissolved solids
(TDS) of 100-136 mg/l was used for saturation of sand and GCL specimens (Sydney water quality is
presented in Appendix C). While, distilled water and tap water were used as permeants during testing for
GCL and sand, respectively.
In addition, the hydraulic conductivity of the desiccated GCL at the end of the heating stage was
measured while in contact with a synthetic brine using a flexible-wall permeameter. The constant head
method was performed and two conditions were considered: 1) the desiccated GCL is first hydrated with
tap water then comes in contact with brine 2) only brine is used as permeant. A procedure similar to
guidelines from ASTM D6766 (2012) was used.
-103
Table 6-3. Composition of the synthetic brine
Ions from Ions mole Concentraion Total

Chemical Molar Concentration
Formula each fraction of ions Ions concentraion
compound Mass (g/L) compound (%) (g/L) of ions (g/L)
Sodium
NaCl 58.44 100.00 Na+ 39.34 39.34 Na+ 74.71
chloride
Cl- 60.66 60.66 Cl- 60.66
Sodium
NaHCO3 84.01 50.00 Na+ 27.37 13.68 HCO3- 36.32
bicarbonate
HCO3- 72.63 36.32 CO32- 28.31
Sodium
Na2CO3 105.99 50.00 Na+ 43.38 21.69
carbonate
CO32- 56.62 28.31
Total 200.00 200.00 200.00
Atomic weight of chemical elements: H = 1.01; C = 12.01; O = 16.00; Na = 22.99; Cl = 35.45
Composition of the synthetic brine is given in Table 6-3. The synthetic brine had three chemical
compounds of sodium chloride, sodium bicarbonate, and sodium carbonate with concentrations of 100,
50, and 50 g/L, respectively. Therefore, the total concentration of dissolved solids was 200 g/L (similar to
tests reported by Rowe and Shoaib, 2014 and 2017). The dissolved ions in solution were Na+, Cl-, HCO3-,
and CO3-2 with a total concentration of ions of 74.71, 60.66, 36.32, and 28.31 g/L, respectively. The
concentration of salts is very high compared to those of found for a fresh brine on coal seam gas sites.
Though, this amount of TDS is appropriate because the purpose of using brine ponds is to evaporate water
and concentrate the salts. The brine will ultimately reach high concentrations of salt or become super
saturated. Therefore, because of leaks through defects in GMB there is a risk of exposure of GCL to such
high concentration of salts.
Saturated hydraulic conductivity is calculated for the constant head method as:
∆𝑄. 𝐿
𝑘𝑠 = 6-1
𝐴. ∆ℎ. ∆𝑡
where:
𝑘𝑠 = saturated hydraulic conductivity, m/s,
∆𝑄 = quantity of flow for a given time interval ∆𝑡, taken as average of inflow and out flow, m3,
𝐿 = length of the specimen, m,
𝐴 = cross-sectional area of specimen, m2,
∆𝑡 = interval of time over which the flow ∆𝑄 occurs (i.e. 𝑡2 − 𝑡1 ), s,
-104
𝑡1 = time at start of permeation trial
𝑡2 = time at end of permeation trial
and:
∆ℎ1 + ∆ℎ2
∆ℎ = 6-2
2
where:
∆ℎ1 = head loss across the specimen at 𝑡1 , m, and
∆ℎ2 = head loss across the specimen at 𝑡2 , m.
In the constant head method, flow must reach steady state and a plot of hydraulic conductivity for
consecutive measurements must not show any significant upward or downward trend. In addition, based
on ASTM D5084 (2010), the ratio of outflow to inflow rate must remain between 0.75 and 1.25 for four
consecutive hydraulic conductivity determinations over an interval of time to consider the flow steady.
For the falling head and constant tailwater head method, saturated hydraulic conductivity is
calculated as:
𝑎. 𝐿 ∆ℎ1
𝑘𝑠 = ln ( ) 6-3
𝐴. ∆𝑡 ∆ℎ2
where:
𝑎 = cross-sectional area of the reservoir containing the influent permeant, m2.
Guidelines from ASTM (D5084 and D5856) limits the maximum hydraulic gradient, 𝑖, depending on
the range of hydraulic conductivity. The limit is higher for a soil with lower hydraulic conductivity. The
maximum limit is 30 for a very low hydraulic conductivity range of less than 1×10-9m/s; however, higher
gradients are possible, provided the hydraulic conductivity remains constant. The hydraulic gradient is
defined as:
∆ℎ
𝑖= 6-4
𝐿
The saturated hydraulic conductivity test is valid only if specimen has reached 100% saturation. Full
saturation can be achieved by applying back pressure for GCLs. Submerged GCLs in water usually do not
reach full saturation condition. Saturation of the specimen can be verified by measuring the Skempton B
coefficient prior to application of hydraulic gradient. The test specimen is considered saturated if 𝐵 ≥
0.95. For an accurate B-value determination, all pore-water pressure must be dissipated and completion
-105
of the primary consolidation is necessary. It is possible to measure B-value after completion of a hydraulic
conductivity test if no hydraulic gradient is acting on the specimen.
A deviation from the guidelines was that height of the specimens was less than the standard
recommended minimum of 25mm. The saturated GCLs used in this experiment had thicknesses less than
15mm. Another deviation from the guidlines was that higher gradients than the recommended values
were applied. The maximum recommended gradient in ASTM D5084 (2010) is 30; however, gradients
higher than 30 were applied to the specimens in this thesis. This is because of the small height of the
specimen and the limited accuracy of pressure measurments of the GDSs used in the experiments.
Moreover, no suitable differential pressure transducer was available for the experiments to control and
measure the gradient.
6.7 SWCC and Unsaturated Hydraulic Conductivity of Sand
The SWCC and unsaturated hydraulic conductivity of the subsoil were measured by HYPROP (Figure 6-6).
HYPROP uses the simplified evaporation method to calculate suction and hydraulic conductivity
relationships with water content (Schindler, 1980; Schelle et al., 2013). The test starts with a saturated
specimen and measures the soil suction using two tensiometers at different elevations of 𝑧1 = 0.25𝐿 and
𝑧2 = 0.75𝐿 where L is the height of the specimen. Weight change of the specimen with time is monitored
as water evaporates from the top of the specimen. This process usually takes less than a week for a sand
specimen and depends on the rate of evaporation. The test stops when the tensiometers become dry. In
this thesis, tap water was used to saturate the sand specimen.
Both suction and unsaturated hydraulic conductivity are calculated based on linear discretization of
the data in time and space. In the following formulation, a superscript and a subscript refer to the time
and space discretization, respectively. Over a time interval from 𝑡 𝑖−1 to 𝑡 𝑖 soil suction is determined as the
arithmetic mean of measured suctions of tensiometers:
ℎ1𝑖 + ℎ2𝑖
ℎ𝑖 = 6-5
2
where:
ℎ𝑖 = average suction at 𝑡 𝑖
ℎ𝑗𝑖 = suction at 𝑧𝑗 and 𝑡 𝑖
-106
Water content, 𝜃 𝑖 , is calculated based on weight change of the specimen at time 𝑡 𝑖 and final oven
dried weight of specimen. Therefore, the relationship between water content and suction can be defined
as 𝜃 𝑖 (ℎ𝑖 ) at each time 𝑡 𝑖 . HYPROP calculates the unsaturated hydraulic conductivities based on the
assumption that the water content profile is linear with depth and the Darcy’s law is valid for each time
interval. Therefore, the unsaturated hydraulic conductivity can be calculated for each time interval as:
𝑖 𝑞𝑖
𝐾 𝑖 (ℎ ) = − 6-6
∆ℎ𝑖
∆𝑧 + 1
where:
𝑖
ℎ = time- and space- average suction. It is defined as:
𝑖 ℎ1𝑖−1 + ℎ2𝑖−1 + ℎ1𝑖 + ℎ2𝑖

ℎ = 6-7
4
and
∆𝑉 𝑖
𝑞𝑖 = 6-8
2∆𝑡 𝑖 𝐴
where:
Figure 6-6. HYPROP
-107
∆𝑡 𝑖 = the time interval, i.e. 𝑡 𝑖 − 𝑡 𝑖−1
∆𝑉 𝑖 = water volume loss between 𝑡 𝑖−1 and 𝑡 𝑖
𝐴 = cross sectional area of the specimen
∆ℎ𝑖 = is change in suction between 𝑡 𝑖−1 and 𝑡 𝑖
∆𝑧 = 𝑧2 − 𝑧1
Volume change measurements were not possible in this method and the main assumption was that
volume change as a result of change in suction was negligible during the test for evaluation of the SWCC
in terms of degree of saturation or volumetric water content. This assumption is practically valid for sand.
6.8 Thermal Conductivity of GMB, GCL, and Subsoil
Thermal conductivity of the subsoil was measured using the Hot Disk AB device. A simple experimental
setup was used for measurement of thermal conductivity of subsoil under an unsaturated condition
(Figure 6-8-a). Hot Disk AB has been developed based on extension of the research by Gustafsson (1991)
on measurement of thermal conductivity of materials using transient responses to a heat pulse. In
addition, thermal conductivity of the GCL (Ali et al., 2016) and a HDPE-GMB (Singh and Bouazza, 2013)
were measured at Monash University using a thermocell (Figure 6-8-b). The experiments on the GCL were
performed under a steady-state heat flow under different overburden stresses. The reader is referred to
the papers for more details of test methods.
Loading
blocks
Hot Disk AB
sensor
Sample
holding rings
(a) (b)
Figure 6-7. Thermal conductivity setups: a) Hot Disk AB device; b) thermocell (from Ali et al., 2016)
-108
6.9 Mechanical Properties of GCL
A non-linear elastic model has been considered in this thesis for simulation of mechanical responses of
materials. The constitutive equation (Eq. 4-28) is presented in Section 4.8. In this equation, 𝜅 and 𝜅𝑠 are
𝑠+100
the slope of unload/reload curve in the 𝑒 − ln(𝜎𝑛𝑒𝑡 ) and 𝑒 − ln ( 100
) diagrams, respectively. 𝜅𝑠 can be
established from the SWCC experiments. To determine 𝜅, consolidation tests similar to ASTM
D2435/D2435M (2011) were performed using the Barcelona Cells.
The non-linear elastic model adopted in this thesis only considers elastic responses of the porous
media. However, in general, a yield stress can be established on volume change vs suction or net stress
curves. Figure 6-8 shows relationship between the slope of virgin compression line, 𝜆, and slope of the
unload/reload curve, 𝜅, in the 𝑣 − ln(𝜎𝑛𝑒𝑡 ) space for saturated soils.
In the experiments performed on the GCLs in this thesis, two different slopes of change in void ratio
with net stress were established under nearly saturated condition (Figure 6-17). Moreover, three distinct
zones with different slopes of volume change as a result of change in suction were observed over the full
range of suction (Figure 5-21).
6.10 Results
6.10.1 Permeability
Figure 6-9 shows results of hydraulic conductivty tests on intact GCL specimens under an effective stress
of 20kPa. The hydraulic conductivities of intact specimens fell within 5.5 × 10−11 and 6.5 × 10−11 𝑚/𝑠.
Tap water was used for saturation of the specimens. Gradients of 63, 65, and 147 were applied to the
intact specimens and the back pressures corresponding to gradients were 405, 10 and 410kPa,
(a) (b)
Figure 6-8. Reference compression curves for saturated soils (Fredlund et al., 2012): a) Specific volume vs. natural logarithm of
effective isotropic stress; b) Void ratio vs. base-10 logarithm of effective vertical stress
-109
respectively. In addition, permeability tests were conducted on desiccated specimens obtained from the
column studies. The specimens were hydrated with distilled water and allowed to reach full saturation
under a back pressure of 500kPa for two weeks. Figure 6-10 shows the permeability tests on the desiccated
specimens. Hydraulic conductivity of desiccated specimens fell within 2.2 × 10−11 and 3.2 × 10−11 𝑚/𝑠.
The desiccated specimens were exposed to the synthetic brine with a total dissolved solids of 200 g/L (see
Table 6-3) after saturation with distilled water; however, no change in permeability was observed over a
period of one week despite the high concentration of salts. Moreover, a dry desiccated specimen was
exposed to the synthetic brine; however, the experiment had not reached equilibrium within one week
because of limitations of the experimental setup to apply high back pressures.
Saturated hydraulic conductivity of the subsoil (SW) was measured using a rigid-wall permeameter
for 3 different void ratios. The specimen was permeated with tap water at room temperature. Figure 6-11
shows that saturated hydraulic conductivity increases as void ratio increases. The target void ratio of the
subsoil for the column studies was 0.5 and a hydraulic conductivity of 3.0 × 10−4 𝑚/𝑠 is estimated.
8.0E-11 2.5
i=63
7.5E-11
i=65
2.0
7.0E-11 i=147
6.5E-11 Limits
1.5
Qinf/Qeff
ks (m/s)
6.0E-11
5.5E-11 1.0
5.0E-11
0.5
4.5E-11
4.0E-11 0.0
0.0 1.0 2.0 3.0 4.0 5.0 0.0 1.0 2.0 3.0 4.0 5.0
(a) Time (days) (b) Time (days)
Figure 6-9. Saturated hydraulic conductivity of an intact GCL using a flexible-wall permeameter: a) 𝑘𝑠 vs time; b) ratio of outflow
to inflow vs time. Specimens were saturated with tap water TDS=100 – 136 mg/l
3.2E-11 1.4
3.0E-11 1.2
1
2.8E-11
Q inf/ Qeff
0.8
ks (m/s)
2.6E-11
0.6 i=135
2.4E-11 i=153
0.4
i=176
2.2E-11 0.2
Limits
2.0E-11 0
0.00 2.00 4.00 6.00 8.00 0.00 2.00 4.00 6.00 8.00
(a) Time (day) (b) Time (day)
Figure 6-10. Saturated hydraulic conductivity of a desiccated GCL using a flexible-wall permeameter: a) 𝑘𝑠 vs time; b) ratio of
outflow to inflow vs time
-110
3.8E-04
Permeant: tap water
3.6E-04 (TDS = 100 – 136 mg/l)
3.4E-04
𝑘𝑠 (𝑚/𝑠)
3.2E-04
3.0E-04
2.8E-04
2.6E-04
0.540 0.560 0.580 0.600 0.620
e
Figure 6-11. Effect of void ratio on saturated hydraulic conductivity of subsoil (SW) using a rigid-wall permeameter.
6.10.2 HYPROP
Figure 6-12 shows suction and water content measurements with time for the SW soil. The test was
performed at 21.5±1°C and void ratio of the specimen was 0.44. The test was started at high degree of
saturation (99.20%) and stopped when the top tensiometer was dried at the beginning of day 4 of the
experiment. Suction in the top and bottom tensiometers reached 200 and 84kPa, respectively. The
generated data was used to calculate the relationship between suction and water content and the
unsaturated hydraulic conductivity following the procedure described in Section 6.7.
SWCC models by van Genuchten (1980) (Eq. 4-7) and Fredlund and Xing (1994) (Eq. 4-19) were
combined with the relative hydraulic conductivity model by van Genuchten (1980) (Eq. 4-23) to fit the
experimental data. The relative hydraulic conductivity model described by Equation 4-23 is written in
terms of effective degree of saturation and a fitting parameter, 𝜆𝑘 . This form of the equation allows either
250 18
Suction bottom (z1) 16
200
Suction top (z2) 14
12
Stop
Suction (kPa)
150
Stop
10
w (%)
100 8
6
50 4
(a) 2 (b)
0 0
0 1 2 3 4 5 0 1 2 3 4 5
Time (day) Time (day)
Figure 6-12. Hyprop measurements with time for SW: a) Suction vs time at z1 and z2 ; b) Gravimetric water content vs. time
-111
of the two SWCC models (van Genuchten or Fredlund and Xing) to be used for prediction of the effective
degree of saturation. Figure 6-13 shows SWCC estimates based on these models. In Figure 6-16, the label
“VG” refers to Van Genuchten based SWCC (with van Genuchten relative hydraulic conductivity model).
Label “FX+VG” refers to the combination of SWCC predictions by Fredlund and Xing (1994) with the relative
hydraulic conductivity model of van Genuchten.
100
90
Experimental
80 VG (Eq. 4-7)
70 FX (Eq. 4-19)
60
Sr (%)
50 FX:
VG: 𝑎𝑓 = 2.06𝑘𝑃𝑎
40 𝑃 = 2.33𝑘𝑃𝑎 𝑛𝑓 = 2.57
𝜆 = 0.582 𝑚𝑓 = 1.07
30 𝑆𝑟𝑠 = 8% 𝜓𝑟 = 100𝑘𝑃𝑎
𝑅 2 = 0.9970 𝑅 2 = 0.9979
20
10
0
0.1 1 10 100 1000 10000 100000 1000000
Suction (kPa)
Figure 6-13. SWCC for SW. VG: van Genuchten (1980). FX: Fredlund and Xing (1994)
1.0E+00
1.0E-02 Experimental
VG (Eq. 4-23)
1.0E-04
FX+VG (Eq. 4-23)
1.0E-06
𝑘𝑟𝑙 (−)
1.0E-08
VG: FX+VG:
1.0E-10 𝜆𝑘 𝑉𝐺 = 0.298 𝜆𝑘 𝐹𝑋+𝑉𝐺 = 0.272
𝑅 2 = 0.7794 𝑅 2 = 0.9998
1.0E-12
1.0E-14
1.0E-16
1.0E-18
0.01 0.1 1 10 100 1000 10000 100000 1000000
Suctio(kPa)
Figure 6-14. Relative hydraulic conductivity for SW. VG: van Genuchten. FX+VG: effective degree of saturation based on the
Fredlund and Xing (1994) combined with the relative conductivity function from van Genuchten (1980)
-112
For each combination, the least-squares method was used to simultaneously fit the SWCC and the
unsaturated hydraulic conductivity to the experimental data. For the VG model, values of 2.33kPa and
0.582 are estimated for P and 𝜆, respectively (𝑅 2 = 0.9970). In addition, a residual degree of saturation,
𝑆𝑟𝑠 , with a value of 8% was selected for the subsoil to reduce the fitting error (Figure 6-13). A relative
hydraulic conductivity function (van Genuchten, 1980) with a shape parameter, 𝜆𝑘 , of 0.298 gave a
reasonably good fit to experiments for low suctions; however, the estimates for the dry end carries errors
of several orders of magnitude (𝑅 2 = 0.7794).
For the FX model, values of 2.06kPa, 2.571, and 1.066 were calculated for 𝑎𝑓 , 𝑛𝑓 , and 𝑚𝑓 , respectively.
In addition, a recommended value of 𝜓𝑟 = 100kPa was assumed (Fredlund el al. 2012). The predicted
relative hydraulic conductivities show a significant improvement for the combination of FX+VG compared
to VG (Figure 6-14). The value of 𝜆𝑘 is estimated to be 0.272. The coefficient of determination for SWCC
and relative hydraulic conductivity estimates were 0.9979 and 0.9998, respectively.
Suction measurements at early stages of drying show low accuracy (Figure 6-13). Moreover, the
precision of suction measurements is limited to ±0.001 pF, i.e. ≈±0.100kPa. Therefore, HYPROP filters out
unsaturated hydraulic conductivity estimates in the early stages of drying (Figure 6-14). However, the
accuracy of measurements improves later as the specimen dries. The available data for unsaturated
hydraulic conductivity of the subsoil is very limited and only a narrow range has been measured. This is
because of difficulties in performing unsaturated hydraulic conductivity tests such as those described
above. Therefore, our understanding of the unsaturated hydraulic conductivity of the subsoil remains
limited.
6.10.3 Thermal conductivity
The experiments were performed for a range of void ratios and degrees of saturation of subsoil using the
Hot Disk AB device (see Figure 6-7-a). It was observed that the thermal conductivity of the subsoil was a
linear function of the degree of saturation within the range of examined void ratios (Figure 6-15 and Table
6-4). Thermal conductivities of 𝜆𝑑𝑟𝑦 = 0.367 and 𝜆𝑠𝑎𝑡 = 2.987 W/(m.K), corresponding to a void ratio of
0.5, were adopted for the subsoil since the target void ratio of the subsoil is 0.50 (see Section 6.5) and it is
expected to remain approximately unchanged during testing.
The thermal conductivity of the same GCL has been reported for different water contents and
overburden stresses by Ali et al. (2016) using the thermocell. Figure 6-16 and Table 6-5 shows thermal
conductivities measured under 25kPa load. Considering a linear relationship between gravimetric water
content and thermal conductivity implies a dry thermal conductivity, 𝜆𝑑𝑟𝑦 , of 0.151W/(m.K). In addition,
based on the reported maximum gravimetric water content (called reference water content, 𝑤𝑟𝑒𝑓 ) under
25kPa, a saturated thermal conductivity of 0.826 W/(m.K) is estimated.
-113
Table 6-4. Thermal conductivity of subsoil (SW)
𝑤 𝑒 𝑆𝑟 𝜆
Test ID
(%) (%) (W/mK)
1 16.511 0.51 85.22 2.3070

2 0.100 0.57 0.46 0.4067
3 0.100 0.57 0.46 0.3976
4 0.100 0.57 0.46 0.4002
5 0.000 0.57 0.00 0.3542
6 0.000 0.58 0.00 0.3838
7 16.586 0.59 75.12 2.3810
8 14.108 0.59 63.09 2.1810
9 0.000 0.61 0.00 0.3670
10 7.779 0.63 32.64 0.9885
11 20.220 0.67 80.17 2.6790
12 0.000 0.68 0.00 0.4265
13 12.292 0.68 47.63 1.6560
14 0.000 0.70 0.00 0.3181
15 0.000 0.70 0.00 0.2960
16 4.692 0.76 16.39 0.8665
As discussed in Section 4.7, change in thermal conductivity is expressed as a function of degree of

saturation; however, for the GCL used in this study, the relationship based on degree of saturation is
unknown. Moreover, the assumed linear relationship based on gravimetric water content is weak
(R2=0.7023); therefore, the estimated dry and saturated thermal conductivities may carry some error.
Despite this, the estimated values fall within the reported range of thermal conductivities of GCLs in the
literature. A range of 𝜆𝑑𝑟𝑦 = 0.10 to 𝜆𝑠𝑎𝑡 = 0.80 for both powder and granular sodium bentonite GCLs and
3.5
Experimental
3.0
Linear (Experimental)
2.5
λ (W/(m.K))
2.0
y = 0.0262x + 0.3667
1.5 R² = 0.9803
1.0
0.5
0.0
0 20 40 60 80 100
Sr (%)
Figure 6-15. Thermal conductivity of subsoil (SW)
-114
Table 6-5. Thermal conductivity of GCL under 25kPa overburden stress (digitized data from Ali et al., 2016)
𝑤a 𝜆 𝑤a 𝜆
Test ID Test ID
(%) (W/mK) (%) (W/mK)
1 80 0.246 7 135 0.592
2 88 0.423 8 150 0.656
3 102 0.575 9 153 0.647
4 113 0.640 10 175 0.689
5 125 0.484 11 180 0.700
6 136 0.536 12 199 0.729
a: water content is calculated based on 𝑤𝑟𝑒𝑓 = 218% (see Ali et al., 2016)
𝜆𝑑𝑟𝑦 ≅ 0.05 to 𝜆𝑠𝑎𝑡 ≅ 0.60 W/(m.K) for a powder sodium bentonite GCL were reported by Southen and
Rowe (2011) and Bouazza et al. (2014), respectively. Moreover, a linear relationship between thermal
conductivity of GCL and degree of saturation were observed in these studies. However, more
investigations are required to develop better understanding of thermal behaviour of GCLs.
Singh and Bouazza (2013) reported a thermal conductivity of 0.34 W/(m.K) for a 1.5mm-thickness
HDPE-GMB. They found it to be consistent with the reported thermal conductivity of HDPE polymers (𝜆 =
0.40 − 0.45 W/(m.K)).
0.900
0.800
0.700
0.600
λ (W/(m.K))
0.500
0.400
y = 0.0031x + 0.1505
0.300 R² = 0.7023
0.200
0.100
0.000
0 50 100 150 200
w (%)
Figure 6-16. Thermal conductivity vs. gravimetric water content for GCL under 25kPa overburden stress (modified from Ali et
al., 2016)
6.10.4 Mechanical Properties of GCL
Table 6-6 and Figure 6-17 present the relationship between void ratio and net axial stress for a highly
saturated GCL. The test was conducted under 35°C on specimen #39D50. It is possible to estimate 𝜅 and
𝜆(0), i.e. slope of virgin compression line at saturated condition, from 𝑒 − ln(𝜎𝑛𝑒𝑡 𝑎𝑥𝑖𝑎𝑙 ) curve (Figure
6-17); therefore, values of 𝜅 = −0.068 and 𝜆(0) = −0.277 are adopted for GCL.
-115
Table 6-6. Net stress vs void ratio for GCL
𝜎𝑛𝑒𝑡 𝑎𝑥𝑖𝑎𝑙
Stage e
(kPa)
Seating 2.80 4.07

1 18.67 3.94
2 22.63 3.92
3 24.62 3.91
4 30.58 3.87
5 37.04 3.82
6 44.47 3.76
7 50.45 3.71
4.00
y = -0.068ln(x) + 4.0172
3.95
R² = 0.9979
3.90 2.303
3.85 𝜅
Void ratio
3.80
3.75 2.303
3.70 𝜆 0
y = -0.277ln(x) + 4.6954
3.65 R² = 0.9746
3.60
3.55
1 10 100
Net axial stress (kPa)
Figure 6-17. Net axial stress vs void ratio for saturated GCL at 35°C
5.00 5.00
4.50 4.50
y = -0.15ln(x) + 3.6259
4.00 4.00 R² = 0.5125
3.50 3.50
e
2.303
𝜅𝑠
3.00 3.00
2.50 2.50
y = -0.127ln(x) + 4.0527
2.00 2.00
R² = 0.5914
1.50 1.50
0.1 10 1000 100000 1 10 100 1000 10000
(a) Suction (kPa) (b) (s+100)/100 (kPa/kPa)
Figure 6-18. Suction and void ratio relationship for GCL: a) Void ratio vs. suction; b) Void ratio vs. (suction+100)/100.
-116
Figure 6-18 shows the relationship between suction and volume change. The figure includes data from
𝑠+100
SWCC tests performed on all X2000 GCL specimens. The slope of 𝑒 − ln ( 100
) diagrams indicates a value
of 𝜅𝑠 = −0.150 (Figure 6-18-b).
Clearly, different slopes control volume changes because of change in suction or net stress. Therefore,
coefficient of determination for the fitted line is low. To improve the fit, it is required to use more advanced
models which are capable of considering such behaviour. The nonlinear elastic constitutive model used
for the mechanical constitutive model is limited to only one slope for suction change and one for net stress
change. The dependence of these parameters on suction and temperature can be further investigated and
considered in more advanced models which may yield different swelling and shrinking behaviour of GCL
(Alonso et al. 1990).
6.10.5 Summary of material properties
Table 6-8 summarizes properties of subsoil including soil classification, specific gravity, permeability,
SWCC, and thermal conductivity. Hydraulic, thermal, and mechanical properties of the GCL are tabulated
in Table 6-7. Other properties of the GCL are presented in Chapter 5. The data will be used in Chapter 7
for validation of the developed THM theory.
Table 6-7. Summary of GCL properties: Hydraulic, thermal, and mechanical
Test/Property Method Unit Result

Permeability ASTM D5084 (2010)
Tap water with an average
𝑘𝑠 m/s 5.5 × 10−11 to 6.5 × 10−11
TDS of 100-136 mg/L
Distilled water m/s 2.2 × 10−11 to 3.2 × 10−11
Permeability with Incompatible
ASTM D6766 (2012)
Aqueous Solutions
A salt solution with TDS of
𝑘𝑠 m/s 2.2 × 10−11 to 3.2 × 10−11
200 g/L
Thermal Conductivity -
𝜆𝑑𝑟𝑦 W/(m.K) 0.151
𝜆𝑠𝑎𝑡 W/(m.K) 0.826
Mechanial Properties -
𝜅 - -0.068
𝜆(0) - -0.277
𝜅𝑠 - -0.150
-117
Table 6-8. Summary of the subsoil properties
Test/Property Method Unit Result

Soil Classification ASTM D2487 (2011) -
Soil Type SW
Specific Gravity ASTM D854 (2014)
𝐺𝑠 - 2.65
Compaction test ASTM D698 (2012)
Optimum water content N/A
Maximum dry density N/A
Permeability ASTM D5856 (2015)
𝑘𝑠 m/s 3.0 × 10−4 to 3.6 × 10−4
SWCC: VG SEM1
𝑃 kPa 2.33
𝜆 - 0.582
𝑆𝑟𝑠 % 8.0
SWCC: FX SEM1
𝑎𝑓 kPa 2.06
𝑛𝑓 - 2.571
𝑚𝑓 - 1.066
𝜓𝑟 kPa 100
Relative hydraulic
SEM1
conductivity
𝜆𝑘 𝑉𝐺 - 0.298
𝜆𝑘 𝑉𝐺+𝐹𝑋 - 0.272
Thermal Conductivity -
𝜆𝑑𝑟𝑦 W/(m.K) 0.367
𝜆𝑠𝑎𝑡 W/(m.K) 2.987
1. SEM: simplified evaporation method.
6.10.6 Column studies

Figure 6-19 shows temperature change with time for the two columns. The temperature curves are almost
identical for the columns showing good repeatability. The heating was started at day 58 of the experiments
and it was continued for 39 days. A heating power outage occurred on day 73 and 18 hours later heating
was resumed. Temperature reached equilibrium shortly after the start of heating and remained relatively
unchanged during the heating stage, except for the power outage period.
In Figure 6-20, taking one elevation at a time, the TDRs from the two columns show essentially
identical readings of water content, with one exception at elevation 225mm. The TDRs at elevation 225mm
appear to be an anomaly. In the hydration stage, TDR-225mm shows higher water contents in column A
than column B and, around day 33, an unexplained drop occurred in the latter. On day 58, when the GCL
is removed, weighed and replaced, then heating is applied, a sudden increase in water content is recorded
-118
by the TDR-225mm in both columns. The increase is larger than that recorded by the TDRs closer to the
surface and is hence unexpected. After heating, the TDR-225mm in the two columns showed similar
trends. A number of explanations for TDR-225mm are possible. The volumetric water content readings
from TDRs are very sensitive to the quality of contact between the rods and the soil.
90
80
70
60 Top piston (Column A)

Temperature (°C)
Y=375 mm (Column A)
50
Y=225 mm (Column A)
40 Y=75 mm (Column A)
30
20
GCL placed Heating power
Column Heating outage
10 packed on the top
started
0
0 7 14 21 28 35 42 49 56 63 70 77 84 91 98
(a) Time (day)
90
80
70
Top piston (Column B)

60
Temperature (°C)
Y=375 mm (Column B)
50
Y=225 mm (Column B)
40 Y=75 mm (Column B)
30
20
GCL placed Heating power
Column Heating outage
10 packed on the top started
0
0 7 14 21 28 35 42 49 56 63 70 77 84 91 98
Time (day)
(b)
Figure 6-19. Temperature vs Time: a) Column A; b) Column B
-119
A small displacement and separation of the TDR rods from the soil can change the readings
significantly. This may have happened on the day the GCL was removed for weighing and the heating was
applied. After the column experiments were completed, the hydration stage was repeated under identical
conditions to check again the behaviour of the TDRs. The new curves obtained for all elevations were very
close to those shown in Figure 6-20, and were therefore not reproduced here. However, the curves for the
TDR-225mm are shown in Figure 6-20 and are much more consistent with the other measurements.
Therefore, the anomaly is assumed to be a measurement error.
0.35
(a)
0.3
0.25
Y=75 mm - Col A
θ (-)
Y=225 mm - Col A
Y=225 mm - (repeat)
0.2
Y=375 mm - Col A
Y=525 mm - Col A
0.15
0.1
0 7 14 21 28 35 42 49 56 63 70 77 84 91 98
Time (days)
Column GCL placed Heating Heating power
packed on the top started outage
0.35
(b)
0.3
Y=75 mm - Col B
0.25
Y=225 mm - Col B
θ (-)
Y=225 mm - (repeat)
0.2 Y=375 mm - Col B
Y=525 mm - Col B
0.15
0.1
0 7 14 21 28 35 42 49 56 63 70 77 84 91 98
Time (days)
Column GCL placed Heating Heating power
packed on the top started outage
Figure 6-20. Volumetric water content vs Time: a) Column A; b) Column B
-120
At the end of the equilibrium period, the top two TDRs at elevations of 375mm and 525mm were
close to steady-state in column B; however, a small rate of change in water content was still observable in
all other TDRs. After the installation of the GCL, a decrease in water content was detected by the three top
TDRs and an increase in the lowest one, signalling loss of moisture from the subsoil hydrating the GCL and
redistribution of moisture within the subsoil. At the beginning of the heating stage, moisture content
readings of the topmost TDR (lowest curve in the Figure 6-20) showed a quick and significant rise followed
by a gradual decrease. The rise in TDR readings is most probably due to moisture travelling downwards
from the GCL into the subsoil through water vapour gradients, followed by condensation. The decrease
reflects the slower process of partial dehydration of the GCL by the top layers of the soil along the suction
gradients.
Moreover, the moisture content readings at elevation of 375mm show a small rise then remain
relatively constant during heating stage. The lower most TDR readings remained mostly unaffected by
heating. A small alteration was observed that can be a consequence of the measurement of mass of the
GCL.
Movement of the surface of the lining systems are presented in Figure 6-21. The two columns showed
similar behaviour. Figure 6-21-a shows the hydration stage. After applying the load of 20kPa, an instant
settlement was observed and the LVDT readings were set to 0 (only the effect of hydration was recorded).
Some consolidation (i.e. time-dependent settlement with no change in net vertical stress) occurred shortly
after applying the load, followed by no volume change for nearly one week. Then both columns started
swelling. The rate of swelling slowed down about four weeks after GCL installation. Maximum swelling at
the end of the hydration stage was around 1.40mm. At the end of the hydration stage, the thickness of
GCLs (under no load) increased from nearly 9mm to 10.50mm and 11.00mm for column A and B (Figure
Figure 6-21. Movement of surface of the lining system: a) Hydration stage; b) Heating stage (GCL was placed on day 14, prior
to which the soil was allowed to reach hydraulic equilibrium in the column)
-121
6-22), respectively. In addition, water content of the specimens increased from an initial gravimetric water
content of nearly 9.00% to 105% and 126% for columns A and B, respectively.
As mentioned before, at the end of the hydration stage, the experiment was stopped briefly in order
to measure the thickness and weight of GCLs. After re-installation of the GCLs, 20kPa load was re-applied
and target temperature of the loading piston was set to 80°C. Then LVDT reading was re-zeroed (only the
effect of heating was recorded). Relatively sharp settlements after the start of heating were observed in
both columns (Figure 6-21-b). The rate of settlement slowed down after 3 days and the final settlement
reached 3.0 to 3.5mm. At the end of the heating stage, thickness and water content of specimens was
7.30mm and 8.5% and 7.49mm and 7.8% for columns A and B, respectively.
The GCLs used in columns A and B are shown in Figures 6-22 and 6-23 after hydration (44 days) and
heating (39 days) stages, respectively. X-ray images of the GCLs after the heating stage are presented in
Figure 6-24, showing that desiccation cracks have developed as a result of thermal dehydration. This is
confirmed by comparison to Figure 6-25 which shows an X-ray image of a hydrated GCL, specimen #4D190,
at 98% gravimetric water content. The GCL of Figure 6-25 was allowed to hydrate from the subsoil for 72
(a) (b)
Figure 6-22. GCLs after hydration stage: a) Specimen #5D190 used in column A; b) Specimen #6D190 used GCL in column B
(a) (b)
Figure 6-23. GCLs after heating stage: a) Specimen #5D190 used in column A; b) Specimen #6D190 used GCL in column B
-122
days and shows no cracking. The X-ray images of Figure 6-24 show that cracking patterns from the two
columns are very similar. The fissures follow a dense pattern and the network of cracks is relatively
uniform.
(a) (b)
Figure 6-24. X-ray image of specimens at the end of heating stage: a) specimen #5D190 was used in column A; b) specimen
#6D190 was used in column B.
Figure 6-25. X-ray image of specimen #4D190 after hydration from subsoil for 72 days. Gravimetric water content of the GCL
reached 98%. Specimen was used in a trial test identical to the column studies
-123
6.11 Summary
All in all, these findings appear consistent with our understanding of moisture redistribution through
liquid water and vapour water redistribution in the subsoil triggered by heating. The sudden rise in water
content in the highest parts of the subsoil seems to indicate that downwards vapour movement and
condensation occur at a much faster rate than the rehydration of the GCL by the subsoil. Experiments
conducted by Southen and Rowe (2005a) (with maximum temperatures of around 55°C) did not record a
sudden increase in water content in the subsoil upon heating. It is possible that the higher temperature
used in our experiments (78°C) has produced more and/or faster water vapour transfer. On the other
hand, the relative importance of liquid and vapour transfer depends on the subsoil composition (e.g.,
smectite content) as shown by Bouazza et al. (2016). Furthermore, while the GCLs in the present study
were allowed to hydrate from the subsoil, those in Southen and Rowe (2005a) were hydrated by
submersion which led to higher post-hydration water contents.
The brine pond conditions represented by the two columns (80°C surface temperature, 20kPa
overburden) have led to desiccation cracking of the GCL and therefore high risk of loss of performance in
contact with a brine. The challenge, however, is to identify the range of conditions under which cracking
occur (e.g. temperature, overburden stress, initial water content, subsoil type) and possible engineering
design solutions that can prevent desiccation. This will be achieved by validating the THM theory described
in Chapter 4, against the datasets developed in Chapters 5 and 6, and then using the theory to predict the
behaviour of the liner under various conditions, not covered in the column studies conducted here.
-124
7 Validation of the Proposed Non-Linear Thermo-Hydro-Elastic Theory of GCL Desiccation
Chapter 7
Validation of the Proposed Non-Linear Thermo-
Hydro-Elastic Theory of GCL Desiccation
7.1 Introduction
The nonlinear thermo-hydro-elastic theory for desiccation of GCL has been described in Chapter 4. In the
present chapter, the theory is validated against four sets of experiments. All experiments subjected a GMB-
GCL liner, resting on a subsoil, to a thermal load applied to the top surface, hence inducing dehydration of
GCL. The first set was the experiment performed in this thesis and presented in Chapter 6. The model built
to simulate this set of experimental data is referred to as #1THM. Two models #2THM and #3THM were
developed to simulate two sets of experiments from the research conducted by Southen and Rowe
(2005a). Models #2THM and #3THM were used to simulate the experiments referred to as G1-S1-L5 and
G2-S2-L3 in their paper, respectively. The subsoil for both experiments were silty sand subsoils. Southen
and Rowe (2011) provided a comprehensive description of material properties of GCLs and subsoil used in
these experiments and compared numerical simulations performed in 1D to experimental results. In
addition, Southen and Rowe (2007) performed a number of SWCC experiments on the GCLs used in #2THM
and #3THM. Application of the proposed SWCCs to these experiments was described in Section 5.12.
Finally, one set of experiments was selected from Bouazza et al. (2014), called #4TH, with the GCL on top
of a clay subsoil.
The four sets of experiments were modelled using the modified version of CODE_BRIGHT, as a multi-
layered problem, under axisymmetric conditions. Both VG and FX versions of the void ratio and
temperature-dependent SWCCs were included in the code. GCLs were modelled as uniform units, and
mesh densities for models were selected based on a number of sensitivity analyses to element size. In
-125
addition, an adaptive time stepping strategy was used for time marching with an initial time step of 10 -5
seconds and a maximum growth rate of 1.4 which was adjusted in each time step according to the relative
errors of the primary variables.
The new SWCCs were used in simulations #1THM, #2THM, and #3THM and both effects of
temperature and void ratio change on the SWCC of materials were considered for these models. For #4TH,
only the effect of temperature was considered on the SWCC of materials since data for calibration of the
proposed SWCC was not available. Moreover, only a hydro-thermal validation was performed on #4TH
because mechanical properties of the materials were unknown and modelling the air-gap in a THM context
requires a number of additional assumptions which are beyond the scope of this thesis.
7.2 Experimental Programs
#1THM was performed in three stages of hydrological equilibrium (prior to placement of GCL), hydration
of GCL from subsoil, and thermal dehydration. The three stages reflect field conditions for brine ponds. A
detailed description of the experiment is presented in Chapter 6 and GCL SWCC measurements are
described in Chapter 5. However, in order to speed up the testing procedure for #2THM, #3THM, and
#4TH, the sets of experiments hydrated the GCL, prior to thermal loading, by immersing it in water rather
than allowing the subsoil to hydrate it (see Southen and Rowe, 2005a and Bouazza et al, 2014). Immersing
a GCL in water can lead to a much higher degree of saturation in the GCL compared to that achieved in the
#2THM
#1THM #3THM #4TH
(a) (b) (c)
Figure 7-1. Schematic of the Validation experiments: a) #1THM (this thesis); b) #2THM and #3THM (Southen and Rowe,
2005a); c) #4TH (Bouazza et al., 2014)
-126
Table 7-1. Main Features of the Validation Experiments
Description
Units #1THM[1] #2THM[2] #3THM[2] #4TH[3]
/Parameter
Column Diameter mm 190 600 600 100
Column Height mm 600 1000 1000 300
Top Temperature °C 78 55 55 70
Load kPa 20 50 70 0
Testing duration: day 97 282 282 75
Equilibrium day 14 50 49 10
Hydration day 44 N/A N/A N/A
Heating day 39 232 233 65
GCL: GCL2 GCL3
GCL1 GCL4
(SR07G1) (SR07G2)
GCL type - ELCOSEAL X2000 Bentofix NS Bentofix BFG 5000 ELCOSEAL X1000
Cover+Carrier - NW+W/NPTT NW+SW/NPTT NW+SW/NPTT NW+W/NPTT
Bentonite type Powder Granular Powder Powder
𝑀𝐺𝐶𝐿 g.m-2 4960Typical 4545 5500+800 4380MARV
𝑀𝐵𝑒𝑛𝑡𝑜𝑛𝑖𝑡𝑒 g.m-2 4250Typical 4240 5000+800 4000MARV
𝑀𝑔𝑒𝑜𝑡𝑒𝑥𝑡𝑖𝑙𝑒 g.m-2 710Typical 305 500 380
𝐺𝑠 𝐵𝑢𝑙𝑘 - 2.08 2.35 2.3 2.27
𝐻𝑠 mm 2.39 1.94 2.74 1.93
ℎ0 mm 9.0 8.0 8.0 14.0
𝑒0 - 2.77 2.62 2.88 6.85
𝑤0 % 9.0 105.0 81.0 ~300
𝜙0 - 0.735 0.724 0.742 0.873
𝜌𝑑0 kg.m-3 551 568 788 313
𝑆𝑟0 % 6.77 98.10 61.97 ~100.00
Subsoil: SC2 SC3
SW1 CL4
(SR11S1) (SR11S2)
Type - Sand (SW) Silty sand Silty sand Clay
𝐺𝑠 - 2.65 2.71 2.74 2.66
𝜌𝑑𝑟𝑦 kg/m3 1763 1750 1790 1380
𝑤0 % 11 4.20 6.30 18
𝑆𝑟0 % 58.30 21.03 32.72 52.00
𝜙0 - 0.333 0.35 0.345 0.48
𝑒0 - 0.500 0.545 0.528 0.92
[1]: This thesis; [2]: Southen and Rowe (2005a); [3]: Bouazza et al. (2014); SR07G1 and SR07G2: GCLs were described in
Section 5.12; SR11S1 and SR11S2: subsoils S1 and S2 were described in Southen and Rowe (2011).
field, as well as different stress conditions, prior to application of the thermal load, both of which can have
a significant impact on the likelihood of desiccation.
The GCLs used in the experiments #1THM to #4TH are labelled as GCL1 to GCL4 and subsoils are
named as SW1, SC2, SC3, and CL4, respectively. Table 7-1 describes the main features of the four sets of
experiments and Figure 7-1 illustrates schematics of the column studies. It is important to note that
-127
temperatures applied to the top surface in the experiments by Southen and Rowe (2005a) and Bouazza et
al (2014) are closer to those found in landfills, rather than brine ponds.
The Southen and Rowe (2005a) experiments applied a temperature of 55°C. Bouazza et al. (2014)
applied a high temperature of 70°C at the top. However, the authors included an air gap between the GCL
and the GMB (Figure 7-1-c), which significantly reduced thermal transfer to the GCL and likely subjected it
to temperatures of less than 55oC.
7.3 Material Properties
Properties of GCLs and subsoils are presented in Tables 7-2 and 7-3. A detailed description of the selection
of material properties is presented in the following sections. No backfitting was performed for parameters
of the GCLs. In addition, no backfitting was performed for SW1 parameters based on VG model. However,
for the model based on FX version of the SWCC, the relative hydraulic conductivity, 𝜆𝑘 𝑉𝐺+𝐹𝑋 , of SW1 for
combined VG+FX model was backfitted to 0.250. As discussed in Section 7.3.4, a value of 0.272 was
obtained for 𝜆𝑘 𝑉𝐺+𝐹𝑋 from the best fit to the unsaturated hydraulic conductivity experimental data using
the least-squares method. For model #2THM, the slope of SWCC for SC2 was set to 0.714 for a better
estimate of the air-entry and slope of experimental SWCC, 𝜆, (see Section 7.3.2) which significantly
improves quality of simulation predictions. The same slope was used for SC3 in model #3THM. It should
be noted that this parameter also was used as 𝜆𝑘 for relative hydraulic conductivity of SC2 and SC3 (i.e.
𝜆𝑘 = 𝜆). No backfitting performed for #4TH.
7.3.1 Mechanical properties: 𝜅 and 𝜅𝑠
The developed net horizontal stresses during hydration and heating depend on the mechanical properties
of GCL. 𝜅, 𝜆(0), and 𝜅𝑠 for GCL1 (X2000) were established in Section 0. Slopes of the virgin compression
line, 𝜆(0), and unloading and reloading line, 𝜅, were measured under 35°C for GCL. In the simulation, 𝜆(0)
and 𝜅 were used for hydration and heating stages, respectively. In addition, values of 𝜅 and 𝜅𝑠 for GCL2
and GCL3 were estimated from experiments reported by Southen and Rowe (2011).
In Figure 7-2, constant stress swell tests (CSST) and constant volume stress tests (CVST) are presented
(Southen and Rowe, 2011). Based on the data, values of 𝜅 for GCL2 and GCL3 of -0.649 and -1.152,
respectively, were estimated. Experiments of SWCC measurements of a GCL by Southen and Rowe (2007)
illustrate relationships between void ratio (bulk void ratio) of GCL and suction under different total stresses
𝑠+0.1
(Figure 7-3). The data is scattered; however, the slope of the fitted lines on the 𝑒 − ln ( 0.1
) diagrams for
all loading conditions was used as representative of 𝜅𝑠 (Figure 7-3). The values of 𝜅𝑠 are estimated as
−0.445 and −0.380 for GCL2 and GCL3, respectively.
-128
Table 7-2. Properties of the GCLs for the validation experiments
#1THM #2THM #3THM #4TH

Parameter Unit
GCL1[1] GCL2[2] GCL3[2] GCL4[3]
SWCC VG (Eq. 4-16):
𝑃0 KPa 168 363 1414 16
𝜆 - 0.333 0.324 0.309 0.231
𝑆𝑠 - 1 1 1 1
𝑆𝑟𝑠 - 0 0 0 0
𝑒𝑟𝑒𝑓 - 4.46 3.20 2.58 0
𝛽𝑒 - 1.46 1.00 0 0
𝜎0 N.m-1 0.0727 0.0727 0.0727 0.0727
SWCC FX (Eq. 4-20):
𝑎𝑓0 kPa 277 - - -
𝑛𝑓 - 1.06 - - -
𝑚𝑓 - 1.20 - - -
𝜓𝑟 kPa 2100 - - -
𝑒𝑟𝑒𝑓 - 4.46 - - -
𝛽𝑒 - 1.60 - - -
𝜎0 N.m-1 0.0727 - - -
Intrinsic Permeability (Eq. 4-21):
𝑘𝑠 m s-1 3.00×10-11 3.00×10-11 3.00×10-11 1.00×10-11
𝜙0 - 0.811 0.724 0.804 0.873
Relative hydraulic conductivity (Eq. 4-23):
𝜆𝑘 - 0.333 0.324 0.309 0.231
Diffusive flux of vapour (Eq. 4-25):
𝐷 m2s-1K-nPa 5.90×10-6 5.90×10-6 5.90×10-6 5.90×10-6
𝑛𝐷 - 2.3 2.3 2.3 2.3
𝜏 - 0.67 0.67 0.67 0.67
Conductive flux of heat (Eq. 4-27):
𝜆𝑑𝑟𝑦 Wm-1K-1 0.151 0.1 0.1 0.1
𝜆𝑠𝑎𝑡 Wm-1K-1 0.826 0.8 0.8 0.5
𝑐𝑠 J kg-1K-1 1000 1000 1000 800
𝜌𝑠 kg m-3 2076 2344 2296 2268
Non-linear elasticity (Eq. 4-28):
𝜅 - -0.069 -0.649 -1.152 -
𝜆(0) - -0.277 - - -
𝜅𝑠 - -0.150 -0.445 -0.380 -
𝑎3 - 0 0 0 -
𝜐 - 0.3 0.3 0.3 -
𝑇𝑠 MPa 0.1 0.1 0.1 -
𝐾𝑚𝑖𝑛 MPa 0.1 0.1 0.1 -
𝛼𝑇 °C-1 1.0×10-5 1.0×10-5 1.0×10-5 -
[1]: This thesis; [2]: Southen and Rowe (2011); [3]: Bouazza et al. (2014)
-129
Table 7-3. Properties of the subsoils and air gap for the validation experiments
#1THM #2THM #3THM #4TH #4TH

Parameter Unit
SW1 SC2 SC3 CL4 Air gap
SWCC VG (Eq. 4-16):
𝑃0 kPa 2.33 10 8 1088 -
𝜆 - 0.582 0.714[2] 0.714[3] 0.487 -
𝑆𝑠 - 1 1 1 1 -
𝑆𝑟𝑠 - 0.08 0.015 0.015 0 -
𝑒𝑟𝑒𝑓 - 0 0 0 0 -
𝛽𝑒 - 0 0 0 0 -
𝜎0 N.m-1 0.0727 0.0727 0.0727 0.0727 -
SWCC FX (Eq. 4-20):
𝑎𝑓0 kPa 2.06 - - - -
𝑛𝑓 - 2.57 - - - -
𝑚𝑓 - 1.07 - - - -
𝜓𝑟 kPa 100 - - - -
𝑒𝑟𝑒𝑓 - 0 - - - -
𝛽𝑒 - 0 - - - -
𝜎0 N.m-1 0.0727 - - - -
𝑘𝑠 m s-1 3.00×10-4 3.63×10-7 4.50×10-7 1.00×10-8 1.00×10-15
𝜙0 - 0.333 0.35 0.345 0.48 0.99
𝜆𝑘 𝑉𝐺 - 0.298 0.714 0.714 0.487 -
𝜆𝑘 𝑉𝐺+𝐹𝑋 - 0.250[1] - - - -
𝐷 m2s-1K-nPa 5.90×10-6 5.90×10-6 5.90×10-6 5.90×10-6 5.90×10-6
𝑛𝐷 - 2.3 2.3 2.3 2.3 2.3
𝜏 - 0.67 0.67 0.67 0.67 1.0
𝜆𝑑𝑟𝑦 Wm-1K-1 0.367 0.5 0.5 0.7 0.03
𝜆𝑠𝑎𝑡 Wm-1K-1 2.987 3.5 3.5 1.7 0.03
𝑐𝑠 J kg-1K-1 800 800 800 800 1000
𝜌𝑠 kg m-3 2645 2705 2735 2655 1.024
𝜅 - -0.01 -0.009 -0.009 - -
𝜅𝑠 - -0.0001 -0.0001 -0.0001 - -
𝑎3 - 0 0 0 - -
𝜐 - 0.3 0.3 0.3 - -
𝑇𝑠 MPa 0.1 0.1 0.1 - -
𝐾𝑚𝑖𝑛 MPa 0.1 0.1 0.1 - -
𝛼𝑇 °C-1 1.0×10-5 1.0×10-5 1.0×10-5 1.0×10-5 -
[1]: 𝜆𝑘 𝑉𝐺+𝐹𝑋 was adjucted from 0.272 (obtained from the best fit to experimental data, see Section 6.10.2) to 0.250 to improve
perdictions for water content of subsoil in simulation of column study #1THM. [2]: The value of 0.714 gives a better fit to SWCC
of SC2 for both slop and air-entry value than a value of 0.542 obtained from the lease-square method. [3]: Same value of 0.714
was selected for SC3.
-130
6.00
GCL2
GCL2 CSST
5.00
GCL2 CVST
GCL3
GCL3 CSST
4.00 y = -1.152ln(x) + 7.9705
R² = 0.9623 GCL3 CVST
Log. (GCL2)
Log. (GCL3)
3.00
e
y = -0.649ln(x) + 5.1279
2.00 R² = 0.9599
1.00
0.00
10 100 1000
Net stress (kPa)
Figure 7-2. Determination of 𝜅 for GCL2 and GCL3 (digitized data from Southen and Rowe, 2011)
5.00
GCL2-Exp.
4.50
y = -0.445ln(x) + 3.8738 GCL3-Exp.
4.00 R² = 0.471 Log. (GCL2-Exp.)
3.50 Log. (GCL3-Exp.)
3.00
2.50
e
2.00 y = -0.38ln(x) + 3.584

R² = 0.3406
1.50
1.00
0.50
0.00
1.00 10.00 100.00
(s+100)/100 (kPa/kPa)
Figure 7-3. Determination of 𝜅𝑠 for GCL2 and GCL3 (digitized data from Southen and Rowe, 2007)
𝜅 for SC2 (i.e. subsoil for model #2THM) was established based on experiments reported by Southen
and Rowe (2011) (Figure 7-4). A value of 𝜅 = −0.009 was estimated for SC2. The mechanical properties
of other subsoils were not available. Same value of 𝜅 was adopted for SC3 and a value of 𝜅 = −0.01 was
-131
0.56
SC2
0.55
Log. (SC2)
0.54
0.53
e
y = -0.009ln(x) + 0.5907
0.52
R² = 0.9197
0.51
0.50
10 100 1000 10000
Net stress (kPa)
Figure 7-4. Determination of 𝜅 for SC2 (digitized data from Southen and Rowe, 2011)
adopted for SW1. Moreover, a small value for 𝜅𝑠 = −0.0001 was assumed for all subsoils. These
assumptions were made because volume change of sandy and silty subsoils as a result of change in suction
were expected to be small.
Poisson’s ratio and thermal expansion coefficient of the materials were unknown and values similar
to those used in the literature were selected (such as Hoor and Rowe, 2013 and Rowe and Verge, 2013).
The non-linear elastic mechanical constitutive equation used (Eq. 4-28) can result in a 0 or negative value
of bulk modulus under some combination of parameters and state variables; however, in this study, it was
assumed that bulk modulus of the materials remain positive. Therefore, a minimum value of bulk modulus
of 100kPa was selected. In addition, a traction value (𝑇𝑠 ) of 100kPa was selected that allows the nonlinear
elastic model to remain stable as long as tensile stresses are less than 100kPa. This assumption was sound
because the purpose of the model is to detect the onset of desiccation but not to predict post-desiccation
behaviour.
Experimental studies on unsaturated soils show that tensile strength and apparent cohesion increase
with suction. The relationship between the apparent cohesion and suction is non-linear and reaches an
asymptotic value at large matric suctions (Hoyos et al., 2012). However, there is very little experimental
evidence concerning the tensile strength of GCLs. A simple linear correlation between tensile strength and
suction was suggested by Morris et al. (1992).
7.3.2 Soil-water characteristic curves
SWCCs were determined for GCL1, 2, and 3 in Chapter 5 based on the experimental data and the retention
curves for GCL4 and CL4 were reported by Bouazza et al. (2014). SWCC of subsoil SW1 was measured using
HYPROP and the curve were established in Chapter 6. For subsoils SC2 and SC3, experimental data were
reported by Southen and Rowe (2011). Figure 7-5 shows SWCC curves for the van Genuchten (1980)
-132
100
90 CS2-Exp
CS3-Exp
80 CS2-VG:
𝑃 = 10𝑘𝑃𝑎 CS2-VG
70 𝜆=0.714
CS3-VG
𝑆𝑟𝑠 = 1.5%
60 𝑅 2 = 0.6491
Sr (%)
50 CS3-VG:
𝑃 = 8𝑘𝑃𝑎
40 𝜆=0.714
𝑆𝑟𝑠 = 1.5%
30 𝑅 2 = 0.4795
20
10
0
1 10 100 1000 10000
Suction (kPa)
Figure 7-5. SWCCs for SC2 and SC3 (experimental data digitized from Southen and Rowe, 2011)
model. Estimate of SWCC parameters of subsoil based on the least-squares method leads to a slightly
lower slope; however, a sharper slope of 0.714 gives a better fit to the experimental measurement of
SWCC at the beginning of the transition zone on the drying path (Figure 7-5). Same slope of 0.714 was used
for the subsoil (SC3) in simulation #3THM.
7.3.3 Intrinsic permeability
Saturated hydraulic conductivities of the materials were measured and used for calculation of intrinsic
permeability of materials in the simulations. Hydraulic conductivity tests for GCL1 and SW1 are described
in Sections 6.6 and 6.10.1 of this thesis. Hydraulic conductivity of GCL2 and 3 were reported by Southen
and Rowe (2005a) to be ~3.0×10-11 and ~4.0×10-12 m.s-1 for specimens exhumed from column studies at
the end of heating stage, respectively. However, values of 5.0×10-11 and ~3.0×10-11 m.s-1 were used in
simulations by Southen and Rowe (2005b) and Southen and Rowe (2011), respectively, for both GCLs.
Therefore, a value of 3.0×10-11 m.s-1 was adopted for the GCL2 and 3. In addition, Southen and Rowe
(2005a) reported ranges of 2.5×10-8 to 7.0×10-7 m.s-1 and 1.5×10-7 to 7.5×10-7 m.s-1 for SC2 and SC3 using
the falling head method, respectively. Average values of experimental measurements of 3.63×10-7 and
4.50×10-7 m.s-1 were selected for SC2 and SC3, respectively. Saturated hydraulic conductivities of GCL4 and
CL4 reported by Bouazza et al. (2014) were used in #4THM.
-133
7.3.4 Relative hydraulic conductivity
As described in Sections 6.7 and 6.10.2, HYPROP was used to estimate unsaturated hydraulic conductivity
of SW1. The value of 𝜆𝑘 𝑉𝐺 was estimated based on the experimental data; however, given the
uncertainties in measurements of unsaturated hydraulic conductivity for full range of suction, the value of
𝜆𝑘 𝐹𝑋+𝑉𝐺 was adjusted from 0.272 (obtained from the best fit to the experimental data for measurement
of unsaturated hydraulic conductivity, see Section 6.10.2) to 0.250 to improve predictions of moisture
content of the column studies using the FX version of the proposed SWCC.
7.3.5 Diffusive flux of vapour
Parameters defining diffusive flux of vapour were selected based on similar simulations performed by Hoor
and Rowe (2013) on the desiccation of GCL due to thermal gradients except that a tortuosity of 1.0 was
selected for the air gap.
7.3.6 Conductive flux of heat
Thermal conductivities of GCL1 and SW1 were measured experimentally for a wide range of degrees of
saturation and details of the experiments were presented in Sections 6.8 and 6.10.3. For simulation
#1THM, values of specific heat capacity of the GCL1 and subsoil SW1 identical to those used by Southen
and Rowe (2011) were selected.
In addition, thermal conductivities and specific heat capacities were provided for simulations #2 and
3THM by Southen and Rowe (2011) and Bouazza et al. (2014) reported the required parameters for model
#4TH.
7.4 Initial and Boundary Conditions
Boundary conditions aim to mimic, as faithfully as possible, the actual loading and heating sequences for
each experiment. The geomembrane, assumed to be impermeable to water and air, was hence
represented as a boundary condition applied on the top of the GCL (no flux barrier). Bottom boundary in
#1THM and #4TH experiments were no flux of gas and liquid (i.e. sealed). In modelling #4TH, the 5mm air
gap was represented explicitly in the modelling. Bottom of columns in experiments #2THM and #3THM
are reported to be exposed to air; therefore, a constant atmospheric air pressure and no flux of liquid were
selected for the simulation (Figure 7-1).
Table 7-4 summarizes the initial conditions and porosities used in the simulations, while Table 7-5
shows the boundary conditions, loading, and heating sequences. The initial liquid pressures were
-134
Table 7-4. Initial conditions and porosities for simulation of the four sets of experiments (compression positive)
𝑃𝑙 𝑃𝑙
𝑃𝑔 𝑇 𝜎𝑖𝑖 𝜙
Model Location (MPa) (MPa)
(MPa) (°C) (MPa)
for VG model for FX model
#1THM GCL1 -73.90 -58.60 0.100 20 0.100 0.735
Subsoil (SW1) 0.0970 0.0971 0.100 20 0.100 0.333
#2THM GCL2 0.0328 - 0.100 20 0.100 0.724
Subsoil (SC2) 0.0815 - 0.100 20 0.100 0.350
#3THM GCL3 -3.3946 - 0.100 20 0.100 0.742
Subsoil (SC3) 0.0881 - 0.100 20 0.100 0.345
#4TH GCL4 0.100 - 0.100 20 - 0.873
Subsoil (CL4) -1.7515 - 0.100 20 - 0.480
Table 7-5. Boundary conditions for simulation of the four sets of experiments
Time interval 𝑇𝑡𝑜𝑝 𝑇𝑏𝑜𝑡𝑡𝑜𝑚 𝑃𝑙 𝑡𝑜𝑝 𝑃𝑙 𝑏𝑜𝑡𝑡𝑜𝑚 𝑃𝑔 𝑡𝑜𝑝 𝑃𝑔 𝑏𝑜𝑡𝑡𝑜𝑚 𝐿𝑜𝑎𝑑

Experiment
(days) (°C) (°C) (MPa) (MPa) (MPa) (MPa) (MPa)
0 to 14 (Equilibrium) 20 20 0.005
#1THM 14 to 58 (Hydration) 20 20 No-flux No-flux No-flux No-flux 0.020
58 to 97 (Heating) 78 23 0.020
0 to 50 (Equilibrium) 20 20 0.0
#2THM No-flux No-flux No-flux 0.100
51 to 282 (Heating) 55 29 0.050
0 to 49 (Equilibrium) 20 20 0.0
#3THM No-flux No-flux No-flux 0.100
51 to 282 (Heating) 55 24 0.070
0 to 10 (Equilibrium) 20 20
10 to 11 (Heating) 30 21
#4TH 11 to 16 (Heating) 50 21 No-flux No-flux No-flux No-flux 0.0
16 to 25 (Heating) 65 21
25 to 75 (Heating) 70 21
calculated from the SWCCs, based on the initial degree of saturation of the GCLs and the subsoils (given in
Tables 7-2 and 7-3).
7.5 Results
7.5.1 Validation for model #1THM
Comparison of predictions and experimental measurements of volumetric water content are presented in
Figures 7-6 and 7-7 for simulations using VG and FX versions of the proposed SWCC, respectively. The
predictions of volumetric water contents using the two different SWCCs were similar. The models provided
relatively accurate predictions for elevations of 75mm and 375mm and the predictions for elevation
525mm were close to TDR readings from column B. As discussed in Section 6.10.6, readings of TDRs at
-135
elevation of 225mm of both columns were inconsistent; however, sources of possible errors in the TDRs
readings are unknown. Predictions by #1THM for this elevation only matched with reading from column B
for a short period of time before applying heat.
0.35
Column A
0.30
0.25
θ (-)
0.20
0.15
0.10
0 7 14 21 28 35 42 49 56 63 70 77 84 91 98
(a) Hydration Heating
started Time (day) started
0.35
Column B
0.30
0.25
θ (-)
0.20
0.15
0.10
0 7 14 21 28 35 42 49 56 63 70 77 84 91 98
(b) Hydration Heating
Figure 7-6. Change in volumetric water content with time at different elevations: a) Column A; b) Column B. VG version of the
SWCC was used for the simulation
-136
Figure 7-8 shows change in temperature with time for the simulation based on the VG version. Results
for the FX version were similar and hence they are not shown here. The predictions were higher for all
readings in the soil body. There are two possible causes for this discrepancy. First, the assumption made
0.35
Column A
0.30
0.25
θ (-)
0.20
0.15
0.10
0 7 14 21 28 35 42 49 56 63 70 77 84 91 98
(a) Hydration Heating

0.35
Column B
0.30
0.25
θ (-)
0.20
0.15
0.10
0 7 14 21 28 35 42 49 56 63 70 77 84 91 98
(b) Hydration Heating

Figure 7-7. Change in volumetric water content with time at different elevations: a) Column A; b) Column B. FX version of the
SWCC was used for the simulation
-137
in the modelling of no-flow through the sides of the model is likely to be valid for flow of gas and liquid
but less so for heat. Better isolation is required to completely prevent heat losses.
Another possible problem is related to the installation of the TDRs. The CS655 TDRs are able to
measure both water content and temperature. Only the TDR bars were inserted into the soil body, which
is enough for water content measurements; however, to obtain accurate temperature measurements, the
whole TDR, including the white epoxy-protected part, need to be buried inside the column since a part of
90
Column A
80
70
Temperature (°C)
60
50
40
30
20
56 63 70 77 84 91 98
(a) Time (day)
90
Column B
80
70
Temperature (°C)
60
50
40
30
20
56 63 70 77 84 91 98
(b) Time (day)
Figure 7-8. Temperature vs. time: a) Column A; b) Column B. VG version of the SWCC was used for the simulation
-138
the thermocouple is attached to one of the TDR bars and the rest is sealed in the epoxy protected part. A
larger column (∼250mm dimeter) would be required to accommodate the whole TDR; however, diameter
of the columns in this study was limited to 190mm.
Increasing the diameter beyond this size would have introduced a range of difficulties related to
preparation, packing and heating of the soil column. After the start of the heating stage, all temperature
predictions for the subsoil (i.e. elevations of 75mm, 225mm, and 375mm) showed a peak followed by a
1.6
1.4
1.2
1.0
0.8
u-yy-Num. VG
u-yy (mm)
0.6 GCL
u-yy-Num. FX
installed
0.4 Column A-Exp.
0.2 Column B-Exp.
0.0
-0.2
-0.4
-0.6
0 7 14 21 28 35 42 49 56 63
(a)
Time (days)
2.0
1.0
0.0
u-yy (mm)
-1.0
Heating u-yy-Num. VG
started
u-yy-Num. FX
-2.0 Column A-Exp.
Column B-Exp.
-3.0
-4.0
56 63 70 77 84 91 98
(b)
Time (days)
Figure 7-9. Deformation of surface of lining system with time: a) Hydration stage; b) Heating stage.
-139
Table 7-6. Comparison of experimental data and numerical estimates for Gravimetric water content of GCL. GCLs #5D190 and
#6D190 were used in column experiments A and B, respectively. VG: the VG version of the SWCC was used. FX: the FX version of
the SWCC was used.
End of hydration End of heating

Day 58 Day 97
Column A (GCL ID: #5D190) 105% 8.4%
Column B (GCL ID: #6D190) 126% 7.8%
Numerical prediction (VG) 89% 7.5%
Numerical prediction (FX) 133% 17.3%
gradual decrease in temperature. This is due to the fact that as the top part of the soil loses moisture, its
thermal conductivity declines; however, such behaviour was not observed in any of the TDR
measurements.
Predictions of the movement of the top surface of the lining system versus time is compared with
experimental data during the hydration and heating stages in Figure 7-9. Large differences were observed
in both trends and values between numerical predictions and experimental data for both stages.
During the hydration stage, the FX version provided slightly better estimates than those of VG version
for the final swelling of GCL (Figure 7-9-a). However, neither models predicted observed deformations
during the early stages of hydration. This is due to the fact that the non-linear elasticity model does not
capture the consolidation of GCL after loading. Furthermore, the constitutive model only incorporates one
slope, 𝜅𝑠 , for the relationship between e and ln(s), when in fact Figure 6-18 shows three distinct zones with
three different slopes. The average 𝜅𝑠 used in the simulation seems to yield reasonable predictions of final
deformations, but not of the transient response.
The mismatch between predicted and measured deformation during the heating stage (Figure 7-9-b)
was investigated further. The effects of changes in the thermal expansion coefficient (𝛼 𝑇 ) on simulated
deformations was found insignificant (namely, using 𝛼 𝑇 =0, instead of 𝛼 𝑇 =10-5 °C-1, made no difference to
predicted deformations). Hence, the most likely reason for the mismatch is the increase in gas pressure in
the GCL and top subsoil, as a result of heating, which led to a decrease in the simulated net stress since
the total stress had been kept constant at 20kPa. Consequently, the non-linear elastic constitutive model
yielded expansion because volume change was only allowed in the y direction. This was followed by
shrinkage, as a result of loss of moisture and suction increase in the GCL and dissipation of gas pressure
after about 3 weeks. Another contributing factor to the gap between predicted and simulated
deformations, may be plastic and thermo-plastic effects, not taken into account in the constitutive model.
-140
Table 7-6 presents a comparison between experimental observations and the numerical simulations
for gravimetric water content of GCL at the end of the hydration and heating stages. Gravimetric water
content of the GCLs was initially ∼9%. It reached 105 and 126% at the end of the hydration stage and
decreased to 8.4 and 7.6% at the end of the heating stage for columns A and B, respectively. The model
using the VG version of SWCC underestimated water contents. Gravimetric water content was estimated
to be 89% and 7.5% at the end of hydration and heating stages, respectively. On the other hand, the FX
model overestimated water content for both end of hydration (133%) and heating stages (17.3%).
The simulations based on both the VG and FX versions of the SWCC show that at the end of heating
stage tensile stresses develop in the GCL (Figure 7-10). This is consistent with experimental observations,
since the X-ray images of GCLs showed significant cracking as a result of GCL dehydration (Figure 6-24).
Both models showed a peak in the developed tensile stresses followed by a reduction to nearly 0 (Figure
7-10). This may be due to the development of gas pressures in the GCL and the dissipation of those
pressures (Figure 7-11-b).
Profiles of the degree of saturation for both models are very similar for the subsoils (Figures 7-12 and
7-13). Simulations indicate that during the hydration stage, the GCL adsorbs moisture mostly from the top
100 mm of subsoil and the rest of profile at the end of the hydration stage shows only a small decline in
degree of saturation compared to the start of the hydration stage (i.e. day 14). At the end of the heating
stage, the GCL and the top layers of the subsoil significantly dehydrate and moisture travels to the bottom
50
Mid-GCL Num. VG
40 Mid-GCL Num. FX
30
Net stress-xx (kPa)
20
Heating
10 started
-10
14 21 28 35 42 49 56 63 70 77 84 91 98
Time (day)
Figure 7-10. Prediction of development of net stress-xx vs. time (compression positive). VG: the VG version of the SWCC was
used in the simulation; FX: the FX version of the SWCC was used in the simulation.
-141
layers and condenses since the bottom of the column is sealed. Therefore, a significant increase occurs in
water content of bottom layers. This is contradictory to real field conditions, where levels of groundwater
are expected to remain unaffected by changes from the top and such a sealed barrier does not exist.
1.0 0.18
0.9 Heating Mid-GCL VG 0.17
0.8 started Heating
Mid-GCL FX 0.16
0.7 started
0.6 0.15
Pg (MPa)
Sr (-)
0.5 0.14
0.4 0.13
0.3 Mid-GCL VG
0.12
0.2 Mid-GCL FX
0.1 0.11
0.0 0.1
14 21 28 35 42 49 56 63 70 77 84 91 98 14 21 28 35 42 49 56 63 70 77 84 91 98
(a) Time (day) (b) Time (day)
Figure 7-11. Predictions of change in degree of saturation and gas pressure at the middle of GCL during hydration and heating
stages: a) degree of saturation; b) gas pressure
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
0.609
Y (m) GCL
0.606
0.603
0.6
0.6
Day 14 (GCL installation) VG

0.5 Day 58 (End of hydration stage) VG
Day 97 (End of heating stage) VG

0.4
Y(m) Subsoil
0.3
0.2
0.1
0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Sr (-)
Figure 7-12. Predictions for profile of degree of saturation for #1THM. The VG version of the SWCC was used for simulation.
-142
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
0.609
Y (m) GCL
0.606
0.603
0.6
0.6
Day 14 (GCL installation) FX

0.5 Day 58 (End of hydration stage) FX
Day 97 (End of heating stage) FX

0.4
Y(m) Subsoil
0.3
0.2
0.1
0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Sr (-)
Figure 7-13. Predictions for profile of degree of saturation for #1THM. The FX version of the SWCC was used for simulation.
Overall, the developed models based on the nonlinear thermo-hydro-elastic theory were able to
provide reliable predictions for moisture content of subsoil. In addition, predictions of water content of
GCLs were reasonable. However, the nonlinear elastic constitutive model has a limited range of valid
applications and the predicted stresses capture desiccation only qualitatively. More complex mechanical
constitutive models are required in order to capture other important processes such as elasto-plasticity
and the effects of temperature and suction on stiffness parameters.
7.5.2 Validation for models #2THM and #3THM (Southen and Rowe, 2005a)
The VG version of the proposed SWCC was used to simulate experimental studies performed by Southen
and Rowe (2005a). The model #2THM was calibrated to the experiment by backfitting a single parameter.
Then, a completely blind simulation was performed on #3THM based on the calibrated data from #2THM.
The only parameter that was calibrated for #2THM was the slope of the SWCC for the subsoil (SC2). As
discussed before, using the least-squares method for fitting purpose leads to slightly lower air-entry value
and slope of curves. Water content of the subsoil at the end of the heating stage was found to be highly
sensitive to the slope of the SWCCs for the subsoil.
-143
Simulations on #2THM and #3THM gave excellent predictions for volumetric water content of subsoils
at the end of the heating stage (Figure 7-14). Moreover, #3THM predictions of water content of GCL were
almost identical to corresponding experimental results at different stages. However, the simulation on
#2THM slightly overestimated GCL water content. It can also be seen from Figure 7-14 that the numerical
simulations conducted here seem to offer an improvement on predictions made by Southen and Rowe
(2011), most likely as a result of a better approximation of SWCCs of subsoils using the VG version of the
SWCC (only effect of temperature was considered because void ratio change of the subsoils were small).
The models by Southen and Rowe (2011) were based on the Lloret and Alonso (1985) SWCC, which gives
poor estimates for the dry end of curves.
Estimates of temperature profiles show considerable deviation from observations at the middle part
of the columns for both experiments (Figure 7-15). However, the predictions are very close to those of
conducted by Southen and Rowe (2011).
Figure 7-16 shows the predictions of net horizontal stresses during the simulations at the middle of
GCLs. GCL3 hydrated from the subsoil quickly after installation of the GCL because of relatively high initial
suction of the GCL (2745kPa) compared to the subsoil (12kPa) and compressive stresses developed in the
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

1.008
1.006
Y(m) GCL
1.004
1.002
1
1.00
#2THM-Exp. Final
0.90
#2THM-Initial
0.80 #2THM-Num. VG Final (This study)
#2THM-Num. Final (Southen and Rowe, 2011)
0.70
#3THM-Exp. Final
0.60 #3THM-Initial
Y(m) Subsoil
#3THM-Num. VG Final (This study)

0.50 #3THM-Num. Final (Southen and Rowe, 2011)
0.40
0.30
0.20
0.10
0.00
0 0.05 0.1 0.15 0.2 0.25 0.3
θ (-)
Figure 7-14. Comparison of predictions and experimental data for profile of volumetric water content for #2THM and #3THM
at the end of heating stage
-144
GCL. On the other hand, water content of GCL2 remained nearly the same during the equilibrium period
because suction potentials were close.
After the start of the heating and loading, compressive net stresses increased because the GCL was
confined in the horizontal direction. The stresses gradually declined because of moisture loss and
1.0
0.8
0.6
Y(m)
0.4 #2THM-Exp.
0.2 #3THM-Exp.
0.0
20 25 30 35 40 45 50 55 60
T (°C)
Figure 7-15. Comparison of predictions and experimental data for profile of temperature for #2THM and #3THM at the end of
heating stage
0.08
#2THM VG-mid GCL

0.06
#3THM VG-mid GCL
0.04
Net horizontal stress (MPa)
0.02
0.00
-0.02
-0.04
-0.06
-0.08
0 50 100 150 200 250 300
Time (day)
Figure 7-16. Predictions of net horizontal stress vs time at the middle of GCL for #2THM and #3THM
-145
shrinkage of the GCLs. Tensile net stresses in the horizontal direction developed in GCL2 earlier than in
GCL3 and at the end of heating stage, predicted tensile stress in GCL2 was higher than that of GCL3. This
is consistent with experimental observations of Southen and Rowe (2005a) who reported that desiccation
cracking of the GCLs was discovered at the end of experiment (Figure 7-17) and the cracks in GCL2 were
wider than GCL3. As discussed earlier, the non-linear thermo-hydro-elastic model used in this thesis can
only qualitatively show risk of desiccation. Prediction of width of cracks and exact values of stresses are
beyond the capability of the model.
Figure 7-17. Comparison of observed desiccation cracking of #2THM (i.e. G1-S1-L5) and #3THM (i.e. G2-S2-L3) (adopted from
Southen and Rowe, 2005a)
7.5.3 Validation for model #4TH (Bouazza et al, 2014)
The available experimental data is limited to temperature for this set. This is due to inaccuracies associated
with the ability of relative humidity measurements provided by Bouazza et al. (2014) to represent actual
degree of saturation in the subsoil which make calibration difficult. Therefore, no parameter fitting or back
calculation was performed for the simulation.
Figure 7-18 shows a comparison of predictions of temperature distributions for model #4TH. A large
deviation can be seen between predictions and experimental data for the temperature profile at the end
of the heating stage for the middle upper part of the column (Figure 7-18-a). In general, the trend of change
in temperature shows good agreement with experimental data (Figure 7-18-b). The presence of the air
gap between the GMB and GCL reduced the actual imposed temperature applied to the top of GCL that
may significantly reduce risk of desiccation of the GCL. This is likely why no desiccation cracking was
detected in this experiment.
-146
Figure 7-18. Comparison of temperature distribution for #4TH: a) Profile of temperature at the end of heating stage; b)
Temperature vs. time for top reservoir, 20mm below reservoir, and 150mm below reservoir.
7.6 Summary
The thermo-hydro-mechanical theory presented in Chapters 4 and 5 and implemented in CODE_BRIGHT

has been tested against four different sets of experimental data. The experimental data were from three
different sources reflecting different conditions of heating and overburden load. The theory has been
shown to be capable of providing reasonable predictions of changes in temperature and water content in
the subsoil, and of changes in water content in the GCL. The theory is also capable of yielding qualitative
predictions of the onset of desiccation in the GCL. However, the assumption of non-linear elasticity, the
paucity of knowledge on tensile behaviour of GCLs, as well as lack of consideration of various plasticity and
hysteresis effects (e.g., fibre pull out in hydration, thermo-plasticity, SWCC hysteresis) presents sever
limitations on the ability of the theory to predict deformation and evolution of stresses in the GCL in the
tensile region.
In the following chapter, the model will be used to identify conditions under which desiccation is likely
to occur under brine-pond conditions.
-147
8 Engineering Applications
Chapter 8
Engineering Applications
8.1 Introduction
A non-linear thermo-hydro-elastic theory considering the effects of temperature and void ratio on the
SWCC was introduced in Chapters 4 and 5. The theory was validated in Chapter 7 for a range of applied
surface temperatures and overburden loads and was found to capture changes in moisture content in the
GCL and the underlying subsoil, and to provide a qualitative indication of the onset of desiccation, although
it did not reproduce observed deformations of GCL. In this chapter, the theory and associated numerical
model are used to a) examine the effect of temperature and load on desiccation of GCL and b) simulate
two alternative composite liner designs with the aim of minimizing the likelihood of desiccation in brine
ponds under high temperature exposure and low overburden loads. In addition, the model is used to
assess the sensitivity of dehydration and desiccation to three material properties that are difficult to
measure, namely coefficient of unsaturated hydraulic conductivity 𝜆𝑘 of the subsoil, and the GCL’s
Poisson’s ratio and coefficient of thermal expansion.
8.2 Numerical Modelling
Figure 8-1 shows schematics of the numerical models. The base case for the numerical models was #1THM
which was developed and validated in Chapter 7 under brine pond conditions. It was shown in Chapters 6
and 7 that the GCL desiccates under the conditions examined. Three sets of analyses were conducted.
First, the numerical model was used to explore a broader range of applied temperatures and loads and
their effects on the onset of desiccation. This was done by modifying top temperature and load boundary
conditions of the base case. The material properties and duration of stages in this set were identical to
-148
#1THM (i.e. 14, 44, and 39 days for equilibrium, hydration, and heating stages, respectively, with a total
duration of 97 days).
Next, two alternative designs were used to reduce risk of desiccation of GCLs using combinations of
geonets and geomembranes (Figures 8-1-b and 8-1-c). Material properties of the GCL and the subsoil were,
once again, identical to #1THM. However, the length of the heating stage was increased from 39 to 142
days in order to come closer to steady-state; therefore, the total length of simulations was 200 days. In
the designs, the additional protection layers consisted of one layer of geomembrane on top of a layer of
geonet for design D1 (Figure 8-1-b). In design D2 (Figure 8-1-c), two layers of geonet were used. The
thickness and mass per unit area of the geonets were selected based on measured values for Trinet geonet
(triple layer net + single sided geotextile), a commercial geocomposite drainage geonet produced by
GEOFABRICS Australia (Figure 8-2).
The designs were developed following the experimental and numerical studies performed by Bouazza
et al. (2014, 2017). It was observed that a layer of air gap between GMB and GCL significantly decreased
the imposed heat on top of the GCL (Bouazza et al., 2014) and hence lowered the risk of GCL desiccation
(Rowe and Verge, 2013; Hoor and Rowe, 2013). However, separation of GCL and GMB significantly reduces
the efficiency of the design because of large increase in the transmissivity of the GCL/GMB interface (El-
Zein, 2008; El-Zein et al., 2012; Touze-Foltz et al., 2016). Therefore, in the designs analysed here, additional
layers of geonet and geomembrane were considered on top of the base case to avoid separation of the
GCL and GMB. This is similar to the experiments conducted by Bouazza et al. (2017).
In the simulations of designs D1 and D2, the thermal and hydraulic properties of the geonet were
identical to those of air gap defined for model #4TH (Bouazza et al., 2014) and already discussed in Chapter
7. In addition, the geomembrane and geonet were explicitly modelled in these simulations (unlike the base
case in which no-flux boundary conditions for liquid and gas were applied on top of the GCL and the
geomembrane therefore was not included in the model).
Finally, sensitivity analyses were performed with identical conditions and duration of stages to #1THM
(i.e. total simulation length of 97 days) to study effects of three parameters of 𝜆𝑘 , 𝜈, and 𝛼 𝑇 on behaviour
of the system because of uncertainty in the measurements of these parameters.
Similar constitutive equations to those used in Chapter 7 were selected for all simulations in this
chapter. Only the VG version of the developed void ratio and temperature-dependent SWCC was
considered in the simulations. In addition, linear elastic constitutive equations were assumed for the
geonet and the geomembrane.
-149
Base Case
(#1THM) D1 D2
(a) (b) (c)
Figure 8-1. Schematic of the numerical models: a) Base case (i.e. #1THM); b) Design D1 (one layer of geonet); c) Design D2 (two
layers of geonet). Thickness of the geonet was selected based on thickness of Trinet, a commercial geonet by GEOFABRICS Australia.
Figure 8-2. Trinet, a geocomposite drainage system produced by GEOFABRICS Australia: triple layer net + single sided geotextile
-150
8.3 Material Properties
Properties of the materials are tabulated in Table 8-1. Parameters for GCL1 and SW1 were identical to
those used for simulation of #1THM in Chapter 7. These parameters are included in the table for clarity.
As described before, for the geonet, hydraulic and thermal conductivities and a specific heat capacity
identical to those of air gap defined for simulation of #4TH (Bouazza et al., 2014) were used and a density
of 250 kg.m-3 was adopted.
Table 8-1. Material Properties
Parameter Unit GCL1 SW1 Geonet Geomembrane

SWCC VG (Eq. 4-16):
𝑃0 KPa 168 2.33 - -
𝜆 - 0.333 0.582 - -
𝑆𝑠 - 1 1 - -
𝑆𝑟𝑠 - 0 0.08 - -
𝑒𝑟𝑒𝑓 - 4.46 0 - -
𝛽𝑒 - 1.46 0 - -
𝜎0 N.m-1 0.0727 0.0727 - -
𝑘𝑠 m s-1 3.00×10-11 3.00×10-4 1.00×10-15 1.00×10-15
𝜙0 - 0.811 0.333 - -
𝜆𝑘 - 0.333 0.298 - -
𝐷 m2s-1K-nPa 5.90×10-6 5.90×10-6 - -
𝑛𝐷 - 2.3 2.3 - -
𝜏 - 0.67 0.67 - -
𝜆𝑑𝑟𝑦 Wm-1K-1 0.151 0.367 0.03 0.40
𝜆𝑠𝑎𝑡 Wm-1K-1 0.826 2.987 0.03 0.40
𝑐𝑠 J kg-1K-1 1000 800 1000 800
𝜌𝑠 kg m-3 2076 2645 250 945
𝜅 - -0.069 -0.01 - -
𝜆(0) - -0.277 - - -
𝜅𝑠 - -0.150 -0.0001 - -
𝑎3 - 0 0 - -
𝜐 - 0.3 0.3 - -
𝑇𝑠 MPa 0.1 0.1 - -
𝐾𝑚𝑖𝑛 MPa 0.1 0.1 - -
𝛼𝑇 °C-1 1.0×10-5 1.0×10-5 1.0×10-5 1.0×10-5
Linear elasticity:
𝐸 MPa - - 10 600
𝜈 - - - 0.3 0.49
-151
For the geomembrane, a low hydraulic conductivity of 1.00×10-15 m.s-1 and a specific heat capacity of
800 J.kg-1.K-1 were used based on the numerical simulation performed by Bouazza et al. (2014) and a typical
density of 945 kg.m-3 was considered (Muller, 2007). Thermal conductivity of the geomembrane was taken
from experimental data by Singh and Bouazza (2013).
HDPE-GMBs are almost incompressible materials; therefore, a Poisson’s ratio of 0.49 was used
(Muller, 2007). In addition, a typical value of 600MPa was assumed for Young modulus of the
geomembrane (Muller, 2007). Mechanical properties of the geonet are unknown; however, values of 0.30
and 10MPa were chosen for Poisson’s ratio and Young modulus of geonet, respectively.
8.4 Initial and Boundary Conditions
Initial and boundary conditions are presented in Tables 8-2 and 8-3, respectively. Initial conditions identical
to those used in simulation #1THM were used for GCL1 and SW1; they are shown in Table 8-2 for clarity.
In addition, for the base case, boundary conditions for all time intervals identical to those of simulation
#1THM were used, as shown in Table 8-3.
For both the geonet and geomembrane, an atmospheric gas pressure was assumed. In addition, it
was assumed that these layers were initially in thermal equilibrium with other components; therefore, an
initial temperature of 20°C was considered. A porosity of 0.90 was used for the geonet which is consistent
with the nature of the material. CODE_BRIGHT requires non-zero porosity of all materials. A high porosity
was selected for the geomembrane to avoid convergence difficulties. Convergence of models with lower
porosities of geomembrane was difficult and small time steps were required because of high errors
induced in conservation of mass of liquid. However, porosity of the geomembrane has insignificant effects
on the results since no flow of fluids occurs through this layer due to its extremely low permeability.
As described before, cases #1 and #2 are designed to illustrate effects of temperature and overburden
stress on desiccation of GCL, respectively, and provide a comparison of risk of desiccation under landfill
vs. brine pond conditions (Table 8-3). Therefore, the analyses were performed considering three
temperatures of 40, 55, and 78°C for case #1 and three overburden stresses of 20, 35, and 50kPa for case
#2. In addition, safety of the two alternative designs, D1 and D2, were examined under brine pond
conditions for three temperatures of 70, 80, and 90°C applied to the surface of the composite lining
systems for a total duration of 200 days (i.e. 142 days of heating). The longer heating period was selected
for the simulations of designs D1 and D2 in order to investigate trend of change in developed stresses in
GCL.
Finally, it was assumed that the boundary conditions for the equilibrium and hydration stages are
identical to those of the base case for all simulations.
-152
Table 8-2. Initial conditions and porosities (compression positive)
𝑃𝑙 𝑃𝑔 𝑇 𝜎𝑖𝑖 𝜙
Location
(MPa) (MPa) (°C) (MPa)
GCL1 -73.90 0.100 20 0.100 0.735
SW1 0.0970 0.100 20 0.100 0.333
Geonet 0 0.100 20 0.100 0.90
Geomembrane 0 0.100 20 0.100 0.90
Table 8-3. Boundary conditions for simulations of effect of temperature (case #1) and overburden stress (case #2) and performance
of designs D1 and D2. For all cases, the equilibrium and hydration stages (i.e. day 0 to 58) were identical to the base case.
Time interval 𝑇𝑡𝑜𝑝 𝑇𝑏𝑜𝑡𝑡𝑜𝑚 𝑃𝑙 𝑡𝑜𝑝 𝑃𝑙 𝑏𝑜𝑡𝑡𝑜𝑚 𝑃𝑔 𝑡𝑜𝑝 𝑃𝑔 𝑏𝑜𝑡𝑡𝑜𝑚 𝐿𝑜𝑎𝑑

Case/Design
(days) (°C) (°C) (MPa) (MPa) (MPa) (MPa) (MPa)
0 to 14 (Equilibrium) 20 20 0.005
Base case
14 to 58 (Hydration) 20 20 No-flux No-flux No-flux No-flux 0.020
(#1THM)
58 to 97 (Heating) 78 23 0.020
0 to 58 Same as Base case
#1
58 to 97 (Heating) Varies 23 No-flux No-flux No-flux No-flux 0.020
#2
58 to 97 (Heating) 78 23 No-flux No-flux No-flux No-flux Varies
D1 and D2
58 to 200 (Heating) Vaires 23 No-flux No-flux No-flux No-flux 0.020
8.5 Sensitivity Analyses
Nine sets of sensitivity analyses were performed for three key material properties (Table 8-4): 1)
unsaturated hydraulic conductivity of the subsoil; 2) Poisson’s ratio of the GCL; and 3) thermal expansion
coefficient of the GCL. The rationale for choosing these three parameters is described next.
Past work has shown that the behaviour of the composite lining systems was highly sensitive to the
unsaturated hydraulic conductivity of the subsoil (Ghavam-Nasiri and El-Zein, 2016). In addition, in this
thesis, this property of the subsoil (SW1) was measured over a narrow range of suction (see Section
6.10.2). Poisson’s ratio and thermal expansion coefficient of the GCL are unknown and are expected to
have a significant effect on estimates of developed stresses in GCL (Rowe and Verge, 2013).
Given the uncertainty associated with 𝜆𝑘 (a parameter determining relative unsaturated hydraulic
conductivity), values of ±10% of the best fit to experimental data were examined (i.e. 𝜆𝑘 =0.268, 0.328 in
addition to the best fit value of 𝜆𝑘 =0.298). For Poisson’s ratio, three values of 0.25, 0.30, and 0.35 were
considered and thermal expansion coefficients of 0, 1.0×10-6, 1.0×10-5 1/°C were studied. Similar values of
-153
Poisson’s ratio and thermal expansion coefficients have been considered in the literature for simulation of
GCL desiccation (Southen and Rowe, 2011; Rowe and Verge, 2013).
Table 8-4. Details of sensitivity analyses. All models were built based on the base case (i.e. #1THM).
Parameter Unit Value

𝜆𝑘 for SW1 - 0.298-10%, 0.298, 0.298+10%
𝜈 for GCL1 - 0.25, 0.30, 0.35
𝛼 𝑇 for GCL1 - 0, 1.0×10-6, 1.0×10-5
8.6 Results
8.6.1 Effect of temperature: Case #1
Figure 8-3 presents effects of applied temperature to the top of the composite lining system on the profile
of temperature at the end of the heating stage (day 97). Two distinct breaking points can be identified on
the curve for surface temperature of 78°C which are separating, from the top, two straight lines followed
by a curve. In addition, one breaking point for 55°C was observed. However, for 40°C no obvious change
in slope occurred. This is because under 78°C condition, the thermal conductivities of the GCL and the top
layers of the subsoil significantly decrease as a result of drying. It can be seen from Figure 8-4 that water
content in GCL and nearly 50 mm top layer of subsoil are almost uniform in depth which helps explain the
predicted straight lines for curve under 78°C condition in Figure 8-3. However, for lower layers of the
subsoil, the profile of the temperature is curved because of the change in degree of saturation and thermal
conductivity with depth. This is consistent with results from validation of the models #2THM and #3THM
(Section 7.5.2), #4TH (Section 7.5.3), and existing studies on thermal desiccation of GCLs (such as Southen
and Rowe, 2011 and Bouazza et al., 2014).
Under 40°C condition, the profile of degree of saturation remained nearly identical to the profile at
the end of hydration stage. On the other hand, it was predicted that water content of the GCL continued
to increase (Figure 8-4). This implies that suction potential of GCL was greater than the developed potential
in vapour transport as a result of thermal gradients; the former and latter processes are responsible for
upward flow of water and downward flow of vapour, respectively. The behaviour of the system under 55°C
was similar to that observed under 78°C; however, changes in degrees of saturation of subsoil were
insignificant for 55°C compared to 78°C. In addition, as expected, moisture loss in GCL under 55°C was
smaller than under 78°C. This is consistent with our understanding of the effects of temperature on
transport of moisture in porous media (Philip and de Vries, 1957).
-154
Figure 8-5 shows that the predicted net horizontal stresses at the middle of GCL decrease (i.e.,
become less compressive) as applied temperature increases. This is consistent with our understanding of
the dehydration process in which higher thermal gradients, leads to faster dehydration and smaller
compressive stresses. This is also consistent with existing studies on the effects of thermal gradients on
risk of desiccation of GCL (e.g. Azad et al., 2011; Azad et al., 2012; Rowe and Verge, 2012; Hoor and Rowe,
2013). Note that only under the 78°C scenario, are tensile stresses predicted to develop in the GCL. Keeping
temperatures on top of the GCL below 55°C may hence be a sound approach to reducing the risk
desiccation.
Figure 8-3. Prediction of profile of temperature for Case #1 under temperatures of 40, 55, and 78°C and an overburden stress of
20kPa at the end of the heating stage (day 97)
-155
Figure 8-4. Prediction of profile of degree of saturation for Case #1 under temperatures of 40, 55, and 78°C and an overburden
stress of 20kPa at the end of the heating stage (day 97)
Figure 8-5. Effect of temperature on development of net horizontal stresses at the middle of GCL for Case #1 under temperatures
of 40, 55, and 78°C and an overburden stress of 20kPa
-156
8.6.2 Effect of overburden stress: Case #2
Effects of overburden load on redistribution of moisture contents and temperatures have been found to
be insignificant for the subsoil, hence they are not presented here. On the other hand, overburden stress
directly alters the net horizontal stresses in GCL because the liner is confined horizontally (Figure 8-6). In
the simulations, the overburden stress was 20kPa during the hydration stage in all cases and, at the
beginning of the heating stage, either kept at 20kPa or increased to 35 or 50kPa, while the other boundary
conditions remained the same.
Figure 8-6 shows effect of overburden stress during the heating stage on net horizontal stresses at
the middle of the GCL. The higher overburden stress is associated with a lower risk of desiccation because
of the additional developed compressive stresses. This is consistent with existing numerical simulations of
the effect of overburden stress on desiccation of GCL (Rowe and Verge, 2013; and Hoor and Rowe, 2013).
Hence, under 78oC and 20kPa or 30kPa, net horizontal stresses are predicted to become tensile.
Furthermore, under 78oC and 50kPa, though net horizontal stress does not become tensile, it reaches
values below 5kPa, indicating high risk of desiccation. Therefore, it appears that, under the conditions
studied, reducing temperature from 80oC to around 60oC may be a more effective way of reducing the risk
of desiccation than increasing overburden stress to 50kPa.
Figure 8-7 illustrates effect of overburden stress on change in void ratio and degree of saturation with
time at the middle of the GCL. During the hydration period, both void ratio and degree of saturation
gradually increase. After applying load and heat, void ratio shows considerable change and as expected,
higher loads show lower void ratios. However, change in degree of saturation is insignificant and at the
end of heating stage all curves converge to an identical value of degree of saturation. The small change in
degree of saturation of GCL has negligible effect on redistribution of temperature in GCL; therefore, it is
not presented here.
-157
Figure 8-6. Effect of overburden stress on development of net horizontal stresses at the middle of GCL for Case #2 under a
temperature of 78°C and net vertical stresses of 20, 35, and 50kPa
Figure 8-7. Effect of overburden stress on: a) change in void ratio vs. time. Graphs are plotted for middle of GCL; a) change in
degree of saturation vs time
8.6.3 Performance of Designs D1 and D2
As mentioned earlier, a heating period of 142 days was selected for the simulation of designs D1 and D2;
therefore, the total length of each simulation was 200 days. Profiles of temperature are presented in Figure
8-8 for designs D1 and D2 at the end of the heating stage. Clearly, the layers of geonet have significantly
reduced the temperature applied to the top of the GCL. The drop in temperature depends on the number
of layers of geonet. It also depends on the temperature gradient applied to the systems.
-158
(a)
(b)
Figure 8-8. Profile of temperature at the end of heating stage (i.e. on day 200 of simulation and after 142 days of heating): a)
Design D1 with one layer of geonet; b) Design D2 with two layers of geonet
-159
For design D1 with one layer of geonet, temperature at the top of the GCL was reduced by 14.5, 17.2,
and 19.9°C, under applied temperatures of 70, 80 and 90°C, respectively. The corresponding temperature
reductions for design D2 comprising two layers of geonet, temperature reduction were 22.5, 26.8, and
30.9°C.
In all cases studied under designs D1 and D2, the developed net horizontal stresses remained
compressive (Figure 8-9). In addition, stresses appeared to reach a steady-state condition under which
they stayed compressive. As expected, the simulations showed that design D2 was safer than D1 and
higher thermal gradients imposed higher risk of GCL desiccation. However, as described in Chapter 7, the
predictions are only qualitatively able to determine risk of desiccation. For example, it is very likely that
the GCL in design D1 under temperature of 90°C condition might desiccate since the predicted stresses are
close to 0.
All in all, findings from the simulations indicate that the alternative designs D1 and D2 may be viable
solutions to the problem of desiccation of GCLs under high temperature and low overburden load. It was
found that maintaining temperature at top of the GCL under 60°C is a more effective solution for reducing
risk of desiccation than increasing overburden stress to 50kPa. These findings require experimental
confirmation which can be achieved by conducting the columns studies described in Chapter 6 for the
analysed designs.
Figure 8-9. Comparison of performance of designs D1 and D2: Net horizontal stress vs. time under temperatures of 70, 80, and
90°C
-160
8.6.4 Sensitivity analyses
Figure 8-10 shows the effects of unsaturated hydraulic conductivity parameter, 𝜆𝑘 , on the profile of degree
of saturation. Depth of the dried soil at the top of lining systems was controlled by 𝜆𝑘 and a lower value
caused a deeper thickness of dried soil. In addition, the developed net horizontal stresses during hydration
showed significant sensitivity to this parameter. However, the stresses at the end of the heating stage
were all tensile (Figure 8-11-a).
As expected, higher values of Poisson’s ratio showed higher developed net horizontal stresses after
GCL installation and loading (Figure 8-11-b). However, the trend is reversed during hydration. This is at
variance with Southen and Rowe (2011) and Rowe and Verge (2013) who found that a higher Poisson’s
ratio always led to a higher compressive stress. The different findings here may be due to different
methods of determination of net stresses. In the existing studies, 1D models were used to predict effect
of Poisson’s ratio on net horizontal stresses; however, 2D axisymmetric models were used in this study.
Note that, at the end of the heating stage, the net horizontal stress was similar for all three values of
Poisson’s ratio.
Figure 8-10. Effect of 𝜆𝑘 on profile of degree of saturation at the end of heating stage
-161
(a) (b)
(c)
Figure 8-11. Sensitivity of developed net horizontal stresses vs. time at the middle of GCL to: a) unsaturated hydraulic conductivity
parameter of SW1, 𝜆𝑘 ; b) Poisson’s ratio of GCL1, 𝜈; c) Thermal expansion coefficient of GCL1, 𝛼 𝑇
The curves for different thermal expansion coefficients were almost identical. This is consistent with
existing studies that showed this parameter has little effect on prediction of net horizontal stresses in a 1D
context (Rowe and Verge, 2013).
8.7 Summary
In this chapter, the risk of desiccation of GCLs used in composite lining systems under conditions
encountered on brine ponds was found to be high. A number of scenarios were considered with different
thermal gradients and overburden stresses and it was shown that both increase in temperature and
decline in overburden stress increase risk of desiccation of GCL. Furthermore, under the conditions
considered here, reducing temperatures on top of the GCL to around 55oC appears to keep net horizontal
stresses safely compressive and therefore significantly reduce risk of desiccation. However, maintaining
overburden load at 50kPa, up from 20 kPa, still leads to stresses that are almost tensile.
-162
Furthermore, the design D2, involving an additional geomembrane and two Trinet geonets of 5mm
thickness each, succeeded in reducing temperatures on top of the GCL to less than 60 oC, even when the
applied temperature on top of the liner was 90oC.
In addition, sensitivity of net horizontal stresses to the three parameters of unsaturated hydraulic
conductivity parameter for subsoil, Poisson’s ratio of GCL, and thermal expansion coefficient of GCL was
assessed. These parameters were chosen because they are difficult to measure and little is known about
their actual values. It was found that 𝜆𝑘 of subsoil had a significant effect on the distribution of moisture
in both GCL and subsoil and consequently, affected the evolution of stresses in GCL. Poisson’s ratio also
had a considerable effect on the evolution of stresses in GCL. The effect of the thermal expansion
coefficient, on the other hand, was insignificant. Nevertheless, at the end of heating stage, none of the
three parameters was found to have a strong effect on the final net horizontal stresses.
-163
9 Conclusions and Recommendations for Further Studies
Chapter 9
Conclusions and Recommendations for Further
Studies
9.1 Summary of Research
Geosynthetic clay liners are powerful engineering materials, versatile and easy to use, which makes them
cost-effective in providing hydraulic and chemical insulation in a range of engineering applications. They
have been used in brine ponds on coal-seam gas sites where temperature on the surface of the GCL can
reach 90oC and overburden loads are usually less than 50kPa which increases the risk of desiccation of
bentonite. Given the relatively short period over which experience with GCLs has accrued (just over two
decades), their adequacy in applications under high temperature and low overburden load remains an
open question. The aim of this thesis has been to address this question through a research project that
was carried out based on five main parts:
1. Adopt a non-linear thermo-hydro-elastic theory of soil behaviour based on the THM theory of
Olivella et al. (1994, 1996) to the study of desiccation of GCLs and use the CODE_BRIGHT code to
implement it (Chapter 4).
2. Adopt a specific GCL (X2000) and a well graded sandy subsoil (SW) and experimentally measure
their material properties as defined by the constitutive equations associated with the non-linear
thermo-hydro-elastic theory. Most importantly, characterise the dependence of the GCL’s SWCC
on temperature and overburden load and develop new forms of the SWCC to account for this
dependence. Two new SWCCs were developed based on classical van Genuchten (1980) and
Fredlund and Xing (1996), by combining theories of Grant and Salehzadeh (1996) and Gallipoli
(2012) (Chapters 4 and 5).
-164
3. Conduct a set of column studies in the laboratory, replicating hydraulic and thermal conditions in
coal-seam gas sites (to the extent possible), using a liner made of a geomembrane, the GCL X2000
and the sandy subsoil, above a simulated aquifer layer; the aim of the column studies was to assess
whether desiccation of bentonite occurs and to generate experimental data for the validation of
the non-linear thermo-hydro-elastic theory. The experiments were conducted in two identical
columns to test for repeatability (Chapter 6).
4. Validate the non-linear thermo-hydro-elastic theory by comparing its predictions to experimental
results derived in part 3, using material properties measured in part 2, as well as experimental
datasets taken from the literature (Chapter 7).
5. Use the validated non-linear thermo-hydro-elastic theory to explore the risk of desiccation over a
range of applied temperatures and overburden loads and investigate the ability of new liner
configurations to reduce or remove that risk (Chapter 8).
9.2 Main Research Findings and Significance

9.2.1 SWCC Dependence on Void Ratio and Temperature
The proposed SWCCs performed well in predicting the effects of void ratio on SWCC based on the available
data. It was found that air-entry value increases as the net vertical stress increases for the experiments
under the same temperature. Incorporating the effect of void ratio led to an improvement in fitting SWCC
data to analytical curves for the GCL used in this thesis. In addition, a similar improvement was observed
in analysing SWCC data for one GCL (G1) tested by Southen and Rowe (2007) but not for another one (G2).
The proposed SWCCs underestimated the effects of elevation of temperature, especially for the wet
side of the curves. Therefore, it seems that the model based on change of surface tension with
temperature is not enough for the prediction of the effect of temperature on SWCC and other mechanisms
may be involved, such as changes in pore fluid chemistry as a result of increase in temperature. On the
other hand, modified versions of van Genuchten (1980) and Fredlund and Xing (1996) appear to be equally
capable to representing experimentally determined SWCCs and their changes with void ratio and
temperature.
9.2.2 Non-Linear Thermo-Hydro-Elasticity under High Temperature Low Overburden Load
A non-linear thermo-hydro-elastic theory was established based on the thermo-hydro-mechanical theory

of Olivella et al. (1994, 1997). The two new void ratio and temperature-dependent SWCCs were added to
the theory. The theory was found to be capable of replicating the observed change in time of temperature
and water contents in the subsoil with reasonable accuracy, even with minimal back-fitting of data. In
addition, the theory qualitatively predicted the desiccation of bentonite. Previously, the theory had only
-165
been applied to GCLs under the less extreme conditions found in landfills. On the other hand, the
predictions of deformation of GCL during the hydration and dehydration stages was poor. Nevertheless,
the void ratios at the end of the heating stage predicted by the theory were close to observed values. The
limitations of the theory in predicting deformation of GCL are most likely due assumptions of elasticity, as
well as difficulties the adopted theory has in approximating void ratio-suction relationships that exhibit
high levels of non-linearity and hysteresis. In addition, at placement of GCL considerable settlement of the
GCL was observed which was not predicted by the theory.
9.2.3 Desiccation of GCLs in Brine Ponds
The thesis has established experimentally, for the first time, that GCLs in composite liners, subject to
surface temperatures of 78oC and overburden loads of 20kPa, may experience high levels of desiccation
cracking. This observation was established experimentally and later confirmed numerically. On the other
hand, permeability tests with distilled water conducted on saturated specimens showed no decline in
hydraulic conductivity in desiccated GCLs relative to virgin ones. Similar tests with a brine solution also
showed no change, although these were conducted for only one week which limits their significance.
The findings of this thesis call for caution in using the similar types of GCLs studied here (needle-
punched X2000) in applications in which they are exposed to temperatures higher than 60oC. Increasing
the overburden load does not seem to be an effective way of removing the risk of desiccation, at least
within a range of 20 to 50kPa. Designs that reduce the temperature on top of the GCL (such as adding a
geonet), are more promising. For example, numerical simulation of a new configuration with two geonets
of 10mm total thickness was found to significantly lower thermal loads on the GCL and prevent desiccation.
However, these conclusions must be seen as preliminary, and subject to a set of research limitations, most
important of which is that the project conducted here concerned one particular GCL (X2000) and that
other GCLs, by the same manufacturer or by different ones, may perform differently (e.g., GCL X3000 by
Geofabrics has higher bentonite density). The limitations of the thesis are discussed next.
9.3 Research Limitations
The findings of the thesis are inevitably constrained by the limitations of the theoretical and experimental
approaches adopted here, including a number of assumptions made in this research.
9.3.1 Theoretical Limitations
A relatively complex theory was developed; however, it is limited in a number of aspects. One main
limitation of the study is that the presence of salt and the related interactions were not considered.
Moreover, it was assumed that the GMB remains intact. Therefore, the developed theory is limited in
-166
predictions of effect of leaks from defects and performance of the lining system exposed to a real brine.
In addition, the safety of composite lining systems was examined based on likelihood of desiccation of GCL
component of the lining systems; however, in a broader context, safety of the lining systems must be
assessed based on risk of contamination of aquifer because of failure of the lining systems. Therefore,
interactions of other processes must be considered in assessment of failure of the lining systems that are
likely to occur in practice such as self-healing of GCL and the effect of brine on the hydraulic performance
of a desiccated GCL. In addition, although alternative designs analysed in this thesis have yielded promising
results the evaluation of their performance and safety has been conducted through numerical simulations
only and it is crucial to confirm these findings experimentally as well.
Validation of the developed non-linear thermo-hydro-elastic theory was performed for one specific
brine pond condition. In addition, the non-linear mechanical constitutive equation used has limited range
of valid application. Risk of desiccation was assessed by simply estimating net horizontal stress at GCL.
However, a number of other important processes are involved in the development of stresses such as
plasticity during both the hydration and dehydration stages and thermo-plasticity during the dehydration
stage. Moreover, crack initiation and cracking mechanism of GCL are not well understood because of the
complex nature of the composite material and uncertainties in important properties of GCL such as tensile
strength and Poisson’s ratio.
Another limitation of the theory is that hysteresis of the SWCC was not considered. The hydraulic
behaviour of GCL on the wetting and drying paths are different. Therefore, the evolution of the SWCC and
the transition between these two main curves can have important effect on the development of stresses
and desiccation of GCL.
9.3.2 Experimental Limitations
Different methods were used for SWCC measurements of GCL to cover the full range of suctions because
the composite nature of GCL significantly limits the applicability of the available methods. One important
issue was the possibility of capillary break at the interface between the geotextile and the ceramic disk in
the axis-translation method. In addition, SWCCs were measured mostly for the drying path because of
difficulties associated with SWCC measurements under the wetting path, especially for the wet side of the
curve with suctions lower than 3MPa.
The SWCC experiments were performed under a limited range of temperature-controlled conditions.
Composite lining systems are expected to be exposed to temperatures as high as 80°C; however, in this
thesis, maximum temperature was limited to 40 and 60°C for the wet (suctions lower than 700kPa) and
dry (suctions higher than 10MPa) sides of the curves, respectively. Fourth, the effects of osmotic suction
on the total suction potential and water content relationship have not been studied (matric and total
-167
suctions were measured for the wet and dry sides of the curves, respectively). Fifth, the number of data
points available for the fitting exercise are limited especially for the wet side of the SWCCs under elevated
temperatures, because of the length of time it takes to measure each point. More data is required to be
able to establish a more accurate SWCC model.
Considering the column studies, very limited experimental data is available for the water content of
the GCL and no data exists for the development of the stresses in the GCL. In addition, mechanisms of
crack formation and propagation in the GCL are unknown.
9.4 Research Suggestions and Future Works
Given the above limitations, five research directions that can be pursued in combination or separately are
suggested.
First, for better assessment of the safety of composite lining systems, it is necessary to investigate,
experimentally, numerically and theoretically, the effects of brine on hydraulic performance of GCL before
and after desiccation. It is possible, for example, to modify the column studies apparatus developed in this
thesis to allow for direct contact between brine and GCL which can illustrate effects of defects in the GMB.
This would require a better characterisation of the effects of salt on the SWCC of GCL along the wetting
and drying paths, including scanning curves.
Second, despite efforts in this thesis and work by other authors, our understanding of the effects of
temperature on SWCC of GCLs remains limited. Performing additional temperature-controlled
experiments, especially on the wet side of the curves, is required to improve the developed SWCC models
and the accuracy of fitting exercise.
Third, conducting similar experiments to those conducted here on different types of GCLs is necessary
to characterize the effects of GCL type and manufacturing process on its performance and safety in
applications with high temperature gradients. This is important because different GCL types and
manufacturing processes have significant effect on hydration of GCL from subsoil and water retention
capacity of GCL which are important factors in the response of a composite lining system to applied
thermal gradients.
Fourth, the limitations of the thermo-hydro-elastic theory adopted in this thesis are evident and
applying more sophisticated mechanical constitutive models, incorporating elasto-plasticity, thermo-
plasticity and better characterisation of the tensile strength of bentonite, is important. Promising work by
Ghorbani, El-Zein and Airey (2017) has shown that elasto-plasticity applied to the study of GCLs can lead
to better predictive capabilities.
-168
Finally, exploration of new liner configurations and new materials (e.g., polymer bentonite) seem to
be promising approaches to deal with desiccation of GCLs under high thermal gradients. The low thermal
conductivity of air is a suitable property and a layer of geonet can act as a good thermal barrier on top of
the composite lining system because of high porosity of the material. In addition, using new materials and
improving endurance of bentonite in GCLs can reduce the risk of development of cracks in bentonite during
dehydration.
-169
(Abdelaal & Rowe 2015) (El-Zein et al. 2012) (Romero et al. 2001)
(EN ISO 9863-1 2016) (Roshani & Sedano 2016)
(Abuel-Naga & Bouazza 2010)
(EN ISO 9863-2 1996) (Rosin-Paumier & Touze-Foltz 2012)
(Acikel et al. 2011)
(Fox & Stark 2015) (Rouf et al. 2014)
(Acikel et al. 2015)
(Fox & Stark 2015) (Rouf et al. 2016a)
(Agus & Schanz 2005)
(Fredlund & Xing 1994) (Rouf et al. 2016b)
(Agus & Schanz 2007)
(Fredlund et al. 2012) (Rowe & Abdelatty 2013)
(Alcaraz et al. 2016)
(Gallipoli 2012) (Rowe & Ewais 2014)
(Ali et al. 2016)
(Gallipoli et al. 2003) (Rowe & Hosney 2013)
(Allen & Murphy 1986)
(Gao & Shao 2015) (Rowe & Sangam 2002)
(Alonso et al. 1990)
(Gens 2010) (Rowe & Verge 2013)
(Anderson et al. 2012)
(Ghavam-Nasiri & El-Zein 2016) (Rowe et al. 2009)
(ASTM C1699 2009)
(Ghorbani et al. n.d.) (Rowe et al. 2011)
(ASTM D2216 – 10 2010)
(Giroud & Bonaparte 1989a) (Rowe et al. 2016)
(ASTM D2487-11 2011)
(Giroud & Bonaparte 1989b) (Russell 2014)
(ASTM D5084 - 16 2010)
(Giroud & Daniel 2004) (Saidi et al. 2008)
(ASTM D5856-15 2015)
(Grant & Salehzadeh 1996) (Sarabadani & Rayhani 2014)
(ASTM D5993 2009)
(Gustafsson 1991) (Sarabian & Rayhani 2013)
(ASTM D6766-12 2012)
(Hamawand et al. 2013) (Schanz & Tripathy 2009)
(ASTM D6836 - 02 2008)
(Hanson et al. 2013) (Schanz et al. 2013)
(ASTM D698 − 12 2012)
(Hoor & Rowe 2013) (Schelle et al. 2013)
(ASTM D854-14 2014)
(Hoyos & Pérez-Ruiz 2012) (Schindler 1980)
(ASTM D883− 12 2016)
(Huang et al. 2016) (Schneider & Goss 2012)
(ASTMD2435/D2435M 2011)
(Jafari et al. 2014) (Seiphoori et al. 2016)
(Athanassopoulos & College 2015)
(Janssen et al. 2015) (Shackelford et al. 2000)
(Australian Government 2014)
(Junfeng et al. 2006) (Sheng et al. 2008)
(Azad et al. 2011)
(Khalili et al. 2008) (Siemens et al. 2012)
(Azad et al. 2012)
(Kinnon et al. 2010) (Siemens et al. 2013)
(Babu et al. 2002)
(Lee & Shackelford 2005) (Singh & Bouazza 2013)
(Bannour, Barral, et al. 2014)
(Leong et al. 2003) (Southen & Rowe 2004)
(Bannour, Stoltz, et al. 2014)
(Liu et al. 2013) (Southen & Rowe 2005a)
(Barclay & Rayhani 2013)
(Lloret & Alonso 1985) (Southen & Rowe 2005b)
(Barroso & Touze-Foltz 2006)
(Lu et al. 2017) (Southen & Rowe 2007)
(Beddoe et al. 2010)
(Makusa et al. 2014) (Southen & Rowe 2011)
(Beddoe et al. 2011)
(Mancuso et al. 2012) (Tarantino 2009)
(Benson et al. 2010)
(Mazzieri & Emidio 2015) (Thomas & He 1997)
(Bonaparte et al. 2008)
(Mazzieri et al. 2013) (Touze-Foltz et al. 2016)
(Bouazza & Gates 2014)
(Melchior 2002) (van Genuchten 1980)
(Bouazza 2002)
(Milly 1984) (Wang & Liu 2012)
(Bouazza et al. 2013)
(Milly 1985) (Wang et al. 2016)
(Moore 2012) (Ye et al. 2009)
(Morris et al. 1992) (Yeşiller et al. 2005)
(Muller 2007) (Yesiller et al. 2014)
(Bradshaw et al. 2016)
(Olivella 1995) (Zhou & Ng 2014)
(Bulut & Leong 2009)
(Olivella et al. 1996) (Zhou & Rowe 2003)
(Campanella & Mitchell 1968)
(Olivella, Carrera, A. Gens, et al. 1994) (Zhou et al. 1998)
(Campbell et al. 2007)
(Papa & Nicotera 2012) (Zhou et al. 2014)
(Cardoso et al. 2007)
(Petrov & Rowe 1997) (Zhu & Mohanty 2002)
(Carravetta et al. 2012)
(Philip & de Vries 1957) (Zornberg et al. 2005)
(Celia et al. 1990)
(Podgorney & Bennett 2006)
(Chevrier et al. 2012)
(Rayhani et al. 2011)
(Dahm et al. 2011)
(Richards & Neal 1937)
(Di Emidio, Gemmina; Mazzieri 2008)
(Risken 2014)
(Döll 1997)
(Risken et al. 2016)
(El-Zein 2008)
-170
References
Abdelaal, F.B. & Rowe, R.K., 2015. Durability of Three HDPE Geomembranes Immersed in Different Fluids
at 85°C. Journal of Geotechnical and Geoenvironmental Engineering, 141(2).
Abuel-Naga, H. & Bouazza, A., 2010. A novel laboratory technique to determine the water retention curve
of geosynthetic clay liners. Geosynthetics International, 17(5), pp.313–322.
Acikel, A.S. et al., 2015. Applicability and accuracy of the initially dry and initially wet contact filter paper
tests for matric suction measurement of geosynthetic clay liners. Géotechnique, 65(9), pp.780–787.
Acikel, A.S. et al., 2011. Water retention behaviour of unsaturated geosynthetic clay liners. In 13th
International Conference of the International Association for Computer Methods and Advances in
Geomechanics, IACMAG 2011. pp. 626–630.
Agus, S.S. & Schanz, T., 2005. Comparison of Four Methods for Measuring Total Suction. Vadose Zone
Journal, 4(4), p.1087.
Agus, S.S. & Schanz, T., 2007. Errors in Total Suction Measurements. In T. Schanz, ed. Experimental
Unsaturated Soil Mechanics. Springer Proceedings in Physics. Berlin, Heidelberg: Springer Berlin
Heidelberg, pp. 59–70.
Alcaraz, A. et al., 2016. Enhancing the efficiency of solar pond heat extraction by using both lateral and
bottom heat exchangers. Solar Energy, 134, pp.82–94.
Ali, M.A. et al., 2016. Thermal Conductivity of Geosynthetic Clay Liners. Canadian Geotechnical Journal,
53(June), pp.1–12.
Allen, M.B. & Murphy, C.L., 1986. A Finite-Element Collocation Method for Variably Saturated Flow in Two
Space Dimensions. Water Resources Research, 22(11), pp.1537–1542.
Alonso, E.E., Gens, A. & Josa, A., 1990. A constitutive model for partially saturated soils. Géotechnique,
40(3), pp.405–430.
Anderson, R., Rayhani, M.T. & Rowe, R.K., 2012. Laboratory investigation of GCL hydration from clayey
sand subsoil. Geotextiles and Geomembranes, 31, pp.31–38.
ASTM C1699, 2009. Standard Test Method for Moisture Retention Curves of Porous Building Materials
Using Pressure Plates. , p.4.
ASTM D2216 – 10, 2010. Standard Test Methods for Laboratory Determination of Water (Moisture)
Content of Soil and Rock by Mass. Annual Book of ASTM Standards, pp.1–7.
ASTM D2435/D2435M -11, 2011. Standard Test Methods for One-Dimensional Consolidation Properties of
-171
Soils Using Incremental Loading. ASTM Standard Guide, p.15.
ASTM D2487-11, 2011. Standard Practice for Classification of Soils for Engineering Purposes (Unified Soil
Classification System)1. ASTM Standard Guide, pp.1–12.
ASTM D5084 - 16, 2010. Standard Test Methods for Measurement of Hydraulic Conductivity of Saturated
Porous Materials Using a Flexible Wall Permeameter. ASTM International, West Conshohocken, pp.1–
24.
ASTM D5856-15, 2015. Standard Test Method for Measurement of Hydraulic Conductivity of Porous
Material Using a Rigid-Wall, Compaction-Mold Permeameter. Annual Book of ASTM Standards.
ASTM D5993, 2009. Standard Test Method for Measuring Mass Per Unit of Geosynthetic Clay Liners.
Annual Book of ASTM Standards, 99(Reapproved 2009), pp.1–4.
ASTM D6766-12, 2012. Standard Test Method for Evaluation of Hydraulic Properties of Geosynthetic Clay
Liners Permeated with Potentially Incompatible Aqueous Solutions. ASTM International, West
Conshohocken, PA, USA, pp.1–9.
ASTM D6836 - 02, 2008. Standard Test Methods for Determination of the Soil Water Characteristic Curve
for Desorption Using a Hanging Column, Pressure Extractor , Chilled Mirror Hygrometer , and / or
Centrifuge. ASTM Standard Guide, p.20.
ASTM D698 − 12, 2012. Standard Test Methods for Laboratory Compaction Characteristics of Soil Using
Standard Effort. , pp.1–13.
ASTM D854-14, 2014. Standard Test Methods for Specific Gravity of Soil Solids by Water Pycnometer. ,
(May), pp.1–8.
ASTM D883− 12, 2016. Standard Terminology Relating to Plastics. ASTM Standard Guide, pp.1–15.
Athanassopoulos, C. & College, H., 2015. Hydraulic Conductivity of a Polymer - Modified GCL Permeated
with High - pH Solutions. Geosynthetics 2015, (1), pp.181–186.
Australian Government, 2014. Hydraulic fracturing (“fraccing”) techniques, including reporting

requirements and governance arrangements,
Azad, F.M. et al., 2011. Laboratory investigation of thermally induced desiccation of GCLs in double
composite liner systems. Geotextiles and Geomembranes, 29(6), pp.534–543.
Azad, F.M. et al., 2012. Modelling of thermally induced desiccation of geosynthetic clay liners in double
composite liner systems. Geotextiles and Geomembranes, 34, pp.28–38.
Babu, G.S. et al., 2002. Desiccation behaviour of selected geosynthetic clay liners. In H. Zanzinger, R. M.
-172
Koerner, & E. Gartung, eds. Clay Geosynthetic Barriers. Nurmberg, Germany, pp. 295–302.
Bannour, H., Stoltz, G., et al., 2014. Effect of stress on water retention of needlepunched geosynthetic clay
liners. Geotextiles and Geomembranes, 42(6), pp.629–640.
Bannour, H., Barral, C. & Touze-Foltz, N., 2014. Water retention curves of GCLs at early stage of hydration
using over saturated salt solutions. In 10th International Conference on Geosynthetics. Berlin,
Germany.
Barclay, A. & Rayhani, M.T., 2013. Effect of temperature on hydration of geosynthetic clay liners in landfills.
Waste Management & Research, 31(3), pp.265–272.
Barroso, M.P. & Touze-Foltz, N., 2006. Validation of the use of filter paper suction measurements for the
determination of GCLs water retention curves. In J. Kuwano & J. Koseki, eds. 8th International
Conference on Geosynthetics (8ICG). Yokohama, JAPAN, pp. 171–174.
Beddoe, R.A., Rowe, R.K. & Take, W.A., 2010. Development of suction measurement techniques to quantify
the water retention behaviour of GCLs. Geosynthetics International, 17(5), pp.301–312.
Beddoe, R. a., Take, W.A. & Rowe, R.K., 2011. Water-Retention Behavior of Geosynthetic Clay Liners.
Journal of Geotechnical and Geoenvironmental Engineering, 137(November), pp.1028–1038.
Benson, C.H., Oren, A.H. & Gates, W.P., 2010. Hydraulic conductivity of two geosynthetic clay liners
permeated with a hyperalkaline solution. Geotextiles and Geomembranes, 28(2), pp.206–218.
Bonaparte, R. et al., 2008. Design , Construction , and Performance of Low- Level Radioactive Waste
Disposal Facility. In Sixth International Conference on Case Histories in Geotechnical Engineering. pp.
1–21.
Bouazza, A., 2002. Geosynthetic Clay Liners (GCLs). In 4th International Pipeline Conference, Parts A and B.
ASME, pp. 211–216.
Bouazza, A. et al., 2014. Heat and moisture migration in a geomembrane–GCL composite liner subjected
to high temperatures and low vertical stresses. Geotextiles and Geomembranes, 42(5), pp.555–563.
Bouazza, A. et al., 2017. Heat mitigation in geosynthetic composite liners exposed to elevated
temperatures. Geotextiles and Geomembranes, 45(5), pp.406–417.
Bouazza, A. et al., 2016. New insight on geosynthetic clay liner hydration: the key role of subsoils
mineralogy. Geosynthetics International, pp.1–12.
Bouazza, A. et al., 2013. Unsaturated geotechnics applied to geoenvironmental engineering problems

involving geosynthetics. Engineering Geology, 165, pp.143–153.
-173
Bouazza, A. & Gates, W.P., 2014. Overview of performance compatibility issues of GCLs with respect to
leachates of extreme chemistry. Geosynthetics International, 21(2), pp.151–167.
Bradshaw, S.L., Benson, C.H. & Rauen, T.L., 2016. Hydraulic Conductivity of Geosynthetic Clay Liners to
Recirculated Municipal Solid Waste Leachates. Journal of Geotechnical and Geoenvironmental
Engineering, 142(2), p.4015074.
Bulut, R. & Leong, E.C., 2009. Indirect Measurement of Suction. In Laboratory and Field Testing of
Unsaturated Soils. Dordrecht: Springer Netherlands, pp. 21–32.
Campanella, R.G. & Mitchell, J.K., 1968. Influence of temperature variations on soil behavior. Journal of
Soil Mechanics & Foundations Div.
Campbell, G.S., Smith, D.M. & Teare, B.L., 2007. Application of a Dew Point Method to Obtain the Soil
Water Characteristic. In Experimental Unsaturated Soil Mechanics. Berlin, Heidelberg: Springer Berlin
Cardoso, R. et al., 2007. A Comparative Study of Soil Suction Measurement Using Two Different High-Range
Psychrometers. In Experimental Unsaturated Soil Mechanics. Berlin, Heidelberg: Springer Berlin
Carravetta, A., De Paolis, N. & Martino, R., 2012. Field Scale Water Content Measurement by TDR
Technique. In Unsaturated Soils: Research and Applications. Berlin, Heidelberg: Springer Berlin
Celia, M.A., Bouloutas, E.T. & Zarba, R.L., 1990. A general mass-conservative numerical solution for the
unsaturated flow equation. Water Resources Research, 26(7), pp.1483–1496.
Chevrier, B. et al., 2012. Influence of subgrade, temperature and confining pressure on GCL hydration.
Geotextiles and Geomembranes, 33, pp.1–6.
Dahm, K.G. et al., 2011. Composite Geochemical Database for Coalbed Methane Produced Water Quality
in the Rocky Mountain Region. Environmental Science & Technology, 45(18), pp.7655–7663.
Döll, P., 1997. Desiccation of Mineral Liners Below Landfills with Heat Generation. Journal of Geotechnical
and Geoenvironmental Engineering, 123(11), pp.1001–1009.
El-Zein, A., 2008. A general approach to the modelling of contaminant transport through composite landfill
liners with intact or leaking geomembranes. International Journal for Numerical and Analytical
Methods in Geomechanics, 32(3), pp.265–287.
El-Zein, A., McCarroll, I. & Touze-Foltz, N., 2012. Three-dimensional finite-element analyses of seepage and
contaminant transport through composite geosynthetics clay liners with multiple defects. Geotextiles
-174
and Geomembranes, 33, pp.34–42.
Di Emidio, Gemmina; Mazzieri, F.V. impe W.F., 2008. Hydraulic Conductivity of a Dense Prehydrated Gcl:
Impact of Free Swell and Swelling Pressure. EuroGeo4, (320), pp.2–8.
EN ISO 9863-1, 2016. Geosynthetics — Determination of thickness at specified pressures BSI Standards
Publication.
EN ISO 9863-2, 1996. Geotextiles and geotextile-related products — Determination of thickness at

specified pressures - Part 2: Procedure for determination of thickness of single layers of multilayer
products.
Fox, P.J. & Stark, T.D., 2015. State-of-the-art report: GCL shear strength and its measurement – ten-year
update. Geosynthetics International, 22(1), pp.3–47.
Fredlund, D.G., Rahardjo, H. & Fredlund, M.D., 2012. Soil-Water Characteristic Curves for Unsaturated
Soils. In Unsaturated Soil Mechanics in Engineering Practice. Hoboken, NJ, USA: John Wiley & Sons,
Inc., pp. 184–272.
Fredlund, D.G. & Xing, A., 1994. Equations for the soil-water characteristic curve. Canadian Geotechnical
Journal, 31(4), pp.521–532.
Gallipoli, D., 2012. A hysteretic soil-water retention model accounting for cyclic variations of suction and
void ratio. Géotechnique, 62(7), pp.605–616.
Gallipoli, D., Wheeler, S.J. & Karstunen, M., 2003. Modelling the variation of degree of saturation in a
deformable unsaturated soil. Géotechnique, 53(1), pp.105–112.
Gao, H. & Shao, M., 2015. Effects of temperature changes on soil hydraulic properties. Soil and Tillage
Research, 153, pp.145–154.
Gens, A., 2010. Soil–environment interactions in geotechnical engineering. Géotechnique, 60(1), pp.3–74.
van Genuchten, M.T., 1980. A Closed-Form Equation for Predicting the Hydraulic Conductivity of
Unsaturated Soils. Soil Science Society of America Journal, 44(5), p.892.
Ghavam-Nasiri, A. & El-Zein, A., 2016. Effects of defects in geomembranes on reducing desiccation
potential of geosynthetics clay liners. Japanese Geotechnical Society Special Publication, 2(70),
pp.2418–2422.
Ghorbani, J., El-Zein, A. & Airey, D.W., Thermo-elasto-plastic analysis of geosynthetic clay liners subjected
to thermal dehydration. To be published.
Giroud, J.P. & Bonaparte, R., 1989a. Leakage through liners constructed with geomembranes—part I.
-175
Geomembrane liners. Geotextiles and Geomembranes, 8(1), pp.27–67.
Giroud, J.P. & Bonaparte, R., 1989b. Leakage through liners constructed with geomembranes—Part II.
Composite liners. Geotextiles and Geomembranes, 8(2), pp.71–111.
Giroud, J.P. & Daniel, D.E., 2004. Liquid migration in an encapsulated bentonite layer due to geomembrane
defects. Gesosynthetics International, 309(4), pp.311–329.
Grant, S.A. & Salehzadeh, A., 1996. Calculation of Temperature Effects on Wetting Coefficients of Porous
Solids and Their Capillary Pressure Functions. Water Resources Research, 32(2), pp.261–270.
Gustafsson, S.E., 1991. Transient plane source techniques for thermal conductivity and thermal diffusivity
measurements of solid materials. Review of Scientific Instruments, 62(3), pp.797–804.
Hamawand, I., Yusaf, T. & Hamawand, S.G., 2013. Coal seam gas and associated water: A review paper.
Renewable and Sustainable Energy Reviews, 22, pp.550–560.
Hanson, J.L., Risken, J.L. & Yesiller, N., 2013. Moisture-Suction Relationships for Geosynthetic Clay Liners.
In Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering.
Paris, France, pp. 3025–3028.
Hoor, A. & Rowe, R.K., 2013. Potential for Desiccation of Geosynthetic Clay Liners Used in Barrier Systems.
Journal of Geotechnical and Geoenvironmental Engineering, 139(10), pp.1648–1664.
Hoyos, L. & Pérez-Ruiz, D., 2012. A Refined Approach to Barcelona Basic Model Using the Apparent Tensile
Strength Concept. In C. Mancuso, C. Jommi, & F. D’Onza, eds. Unsaturated Soils: Research and
Applications. Berlin Heidelberg: Springer, pp. 103–110.
Huang, Q., Zhao, C. & Cai, G., 2016. The formulation of the water retention curve with deformation and
hysteresis. In Z. Chen et al., eds. Unsaturated Soil Mechanics - from Theory to Practice. Proceedings
of the 6th Asia Pacific Conference on Unsaturated Soils. Guilin, China: CRC Press, pp. 525–530.
Jafari, N.H., Stark, T.D. & Rowe, R.K., 2014. Service Life of HDPE Geomembranes Subjected to Elevated
Temperatures. Journal of Hazardous, Toxic, and Radioactive Waste, 18(1), pp.16–26.
Janssen, J. et al., 2015. Hydraulic conductivity and swelling pressure of GCLs using polymer treated clays
to high concentration CaCl2 solutions. In Proceedings of the XVI ECSMGE Geotechnical Engineering
for Infrastructure and Development. pp. 2687–2692.
Junfeng, S. et al., 2006. Study on Swell Pressure Stress of Bentonite in Geosynthetic Clay Liners. Acta
Geologica Sinica - English Edition, 80(5), pp.763–769.
Khalili, N., Habte, M.A. & Zargarbashi, S., 2008. A fully coupled flow deformation model for cyclic analysis
of unsaturated soils including hydraulic and mechanical hystereses. Computers and Geotechnics,
-176
35(6), pp.872–889.
Kinnon, E.C.P. et al., 2010. Stable isotope and water quality analysis of coal bed methane production
waters and gases from the Bowen Basin, Australia. International Journal of Coal Geology, 82(3–4),
pp.219–231.
Lee, J.-M. & Shackelford, C.D., 2005. Impact of Bentonite Quality on Hydraulic Conductivity of Geosynthetic
Clay Liners. Journal of Geotechnical and Geoenvironmental Engineering, 131(1), pp.64–77.
Leong, E.C., Tripathy, S. & Rahardjo, H., 2003. Total suction measurement of unsaturated soils with a device
using the chilled-mirror dew-point technique. Géotechnique, 53(2), pp.173–182.
Liu, Y., Gates, W.P. & Bouazza, A., 2013. Acid induced degradation of the bentonite component used in
geosynthetic clay liners. Geotextiles and Geomembranes, 36, pp.71–80.
Lloret, A. & Alonso, E.E., 1985. State surfaces for partially saturated soils. In Proceedings of the 11th
International Conference on Soil Mechanics and Foundation Engineering. San Francisco, Calif, pp.
557–562.
Lu, Y., Abuel-Naga, H. & Bouazza, A., 2017. Water retention curve of GCLs using a modified specimen
holder in a chilled-mirror dew-point device. Geotextiles and Geomembranes, 45(1), pp.23–28.
Makusa, G.P. et al., 2014. Freeze–thaw cycling concurrent with cation exchange and the hydraulic
conductivity of geosynthetic clay liners. Canadian Geotechnical Journal, 51(6), pp.591–598.
Mancuso, C., Nicotera, M. V. & Papa, R., 2012. Performances of Two High Capacity Tensiometers. In C.
Mancuso, C. Jommi, & F. D’Onza, eds. Unsaturated Soils: Research and Applications. Berlin,
Heidelberg: Springer Berlin Heidelberg, pp. 11–17.
Mazzieri, F. et al., 2013. Permeation of two GCLs with an acidic metal-rich synthetic leachate. Geotextiles
and Geomembranes, 40, pp.1–11.
Mazzieri, F. & Emidio, G. Di, 2015. Hydraulic conductivity of a dense prehydrated geosynthetic clay liner.
Geosynthetics International, 22(1), pp.138–148.
Melchior, S., 2002. Field studies and excavations of geosynthetic clay barriers in landfill covers. In H.
Zanzinger, R. M. Koerner, & E. Gartung, eds. Clay geosynthetic barriers. Proceedings of the
International Symposium. Nuremberg, Germany: AA Balkema Publishers, pp. 321–330.
Milly, P.C.D., 1985. A mass-conservative procedure for time-stepping in models of unsaturated flow.
Advances in Water Resources, 8(1), pp.32–36.
Milly, P.C.D., 1984. A Simulation Analysis of Thermal Effects on Evaporation From Soil. Water Resources
Research, 20(8), p.1087.
-177
Moore, T.A., 2012. Coalbed methane: A review. International Journal of Coal Geology, 101, pp.36–81.
Morris, P.H., Graham, J. & Williams, D.J., 1992. Cracking in drying soils. Canadian Geotechnical Journal,
29(2), pp.263–277.
Muller, W., 2007. HDPE Materials and Geomembrane Manufacture. In HDPE Geomembranes in
Geotechnics. Berlin, Germany: Springer Berlin Heidelberg, pp. 11–33.
Olivella, S., Carrera, J., Gens, a., et al., 1994. Nonisothermal multiphase flow of brine and gas through
saline media. Transport in Porous Media, 15(3), pp.271–293.
Olivella, S., Carrera, J., Gens, A., et al., 1994. Nonisothermal multiphase flow of brine and gas through
saline media. Transport in Porous Media, 15(3), pp.271–293.
Olivella, S., 1995. Nonisothermal multiphase flow of brine and gas through saline media (Part 2). PhD
thesis, Departamento de Ingeniería del Terreno, UPC, Barcelona.
Olivella, S. et al., 1996. Numerical formulation for a simulator (CODE_BRIGHT) for the coupled analysis of
saline media. Engineering Computations, 13(7), pp.87–112.
Papa, R. & Nicotera, M. V, 2012. Use of TDR Probes to Measure Water Content in Pumiceous Soils. In
Unsaturated Soils: Research and Applications. Berlin, Heidelberg: Springer Berlin Heidelberg, pp.
107–112.
Petrov, R.J. & Rowe, R.K., 1997. Geosynthetic clay liner (GCL) - chemical compatibility by hydraulic
conductivity testing and factors impacting its performance. Canadian Geotechnical Journal, 34(6),
pp.863–885.
Philip, J.R. & de Vries, D. a., 1957. Moisture movements in porous materials under temperature gradients.
Transactions, American Geophysical Union, 38(2), pp.222–232.
Podgorney, R.K. & Bennett, J.E., 2006. Evaluating the Long-Term Performance of Geosynthetic Clay Liners
Exposed to Freeze-Thaw. Journal of Geotechnical and Geoenvironmental Engineering, 132(2),
pp.265–268.
Rayhani, M.T. et al., 2011. Factors affecting GCL hydration under isothermal conditions. Geotextiles and
Geomembranes, 29(6), pp.525–533.
Richards, L.A. & Neal, O.R., 1937. Some Field Observations with Tensiometers (1). Soil Science Society of
America Journal, 1(C), p.71.
Risken, J.L., 2014. Development and Use of Moisture-Suction Relationships for Geosynthetic Clay Liners.
California Polytechnic State University, San Luis Obispo.
-178
Risken, J.L., Hanson, J.L. & Yesiller, N., 2016. Estimation of engineering behavior of geosynthetic clay liners
using moisture-suction relationships. Japanese Geotechnical Society Special Publication, 2(69),
pp.2373–2378.
Romero, E., Gens, A. & Lloret, A., 2001. Temperature effects on the hydraulic behaviour of an unsaturated
clay. Geotechnical and Geological Engineering, 19(3/4), pp.311–332.
Roshani, P. & Sedano, J.Á.I., 2016. Incorporating Temperature Effects in Soil-Water Characteristic Curves.
Indian Geotechnical Journal, 46(3), pp.309–318.
Rosin-Paumier, S. & Touze-Foltz, N., 2012. Hydraulic and chemical evolution of GCLs during filter press and
oedopermeametric tests performed with real leachate. Geotextiles and Geomembranes, 33, pp.15–
24.
Rouf, M. et al., 2014. Evaluation of a geosynthetic clay liner water retention curve using vapour equilibrium
technique. In Unsaturated Soils: Research & Applications. CRC Press, pp. 1003–1009.
Rouf, M.A. et al., 2016a. Gas flow unified measurement system for sequential measurement of gas
diffusion and gas permeability of partially hydrated geosynthetic clay liners. Canadian Geotechnical
Journal, 53(6), pp.1000–1012.
Rouf, M.A. et al., 2016b. Water vapour adsorption and desorption in GCLs. Geosynthetics International,
23(2), pp.86–99.
Rowe, R.K. et al., 2011. GCL hydration under simulated daily thermal cycles. Geosynthetics International,
18(4), pp.196–205.
Rowe, R.K. & Abdelatty, K., 2013. Leakage and Contaminant Transport through a Single Hole in the
Geomembrane Component of a Composite Liner. Journal of Geotechnical and Geoenvironmental
Engineering, 139(3), pp.357–366.
Rowe, R.K. & Ewais, A.M.R., 2014. Ageing of exposed geomembranes at locations with different
climatological conditions. Canadian Geotechnical Journal, 52(3), pp.326–343.
Rowe, R.K. & Hosney, M.S., 2013. Laboratory investigation of GCL performance for covering arsenic
contaminated mine wastes. Geotextiles and Geomembranes, 39, pp.63–77.
Rowe, R.K., Jones, D.D. & Rutter, A., 2016. Polychlorinated biphenyl diffusion through HDPE
geomembrane. Geosynthetics International.
Rowe, R.K., Rimal, S. & Sangam, H., 2009. Ageing of HDPE geomembrane exposed to air, water and leachate
at different temperatures. Geotextiles and Geomembranes, 27(2), pp.137–151.
Rowe, R.K. & Sangam, H.P., 2002. Durability of HDPE geomembranes. Geotextiles and Geomembranes,
-179
20(2), pp.77–95.
Rowe, R.K. & Verge, A., 2013. Prediction of geosynthetic clay liner desiccation in low stress applications.
Geosynthetics International, 20(5), pp.301–315.
Ruhl, J.L. & Daniel, D.E., 1997. Geosynthetic Clay Liners Permeated with Chemical Solutions and Leachates.
Journal of Geotechnical and Geoenvironmental Engineering, 123(4), pp.369–381.
Russell, A.R., 2014. How water retention in fractal soils depends on particle and pore sizes, shapes,
volumes and surface areas. Géotechnique, 64(5), pp.379–390.
Saidi, F., Touze-Foltz, N. & Goblet, P., 2008. Numerical modelling of advective flow through composite
liners in case of two interacting adjacent square defects in the geomembrane. Geotextiles and
Geomembranes, 26(2), pp.196–204.
Sarabadani, H. & Rayhani, M.T., 2014. Influence of Normal Stress on Hydration of GCLS from Subsoil. The
Journal of Solid Waste Technology and Management, 39(4), pp.292–303.
Sarabian, T. & Rayhani, M.T., 2013. Hydration of geosynthetic clay liners from clay subsoil under simulated
field conditions. Waste Management, 33(1), pp.67–73.
Schanz, T., Khan, M.I. & Al-Badran, Y., 2013. An alternative approach for the use of DDL theory to estimate
the swelling pressure of bentonites. Applied Clay Science, 83–84, pp.383–390.
Schanz, T. & Tripathy, S., 2009. Swelling pressure of a divalent-rich bentonite: Diffuse double-layer theory
revisited. Water Resources Research, 45(5), p.n/a-n/a.
Schelle, H. et al., 2013. Water retention characteristics of soils over the whole moisture range: a
comparison of laboratory methods. European Journal of Soil Science, 64(6), pp.814–821.
Schindler, U., 1980. Ein Schnellverfahren zur Messung der Wasserleitfahigkeit im teilgesattigten Boden
and Stechzylinderproben. Archiv fur Acker-und Pflanzenbau und Bodenkunde.
Schneider, M. & Goss, K.-U., 2012. Prediction of the water sorption isotherm in air dry soils. Geoderma,
170, pp.64–69.
Seiphoori, A. et al., 2016. Water retention and swelling behaviour of granular bentonites for application in
Geosynthetic Clay Liner (GCL) systems. Soils and Foundations, 56(3), pp.449–459.
Shackelford, C.D. et al., 2000. Evaluating the hydraulic conductivity of GCLs permeated with non-standard
liquids. Geotextiles and Geomembranes, 18(2–4), pp.133–161.
Sheng, D., Fredlund, D.G. & Gens, A., 2008. A new modelling approach for unsaturated soils using
independent stress variables. Canadian Geotechnical Journal, 45(4), pp.511–534.
-180
Siemens, G. et al., 2012. Numerical investigation of transient hydration of unsaturated geosynthetic clay
liners. Geosynthetics International, 19(3), pp.232–251.
Siemens, G.A. et al., 2013. Effect of confining stress on the transient hydration of unsaturated GCLs. In
Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering.
pp. 1187–1190.
Singh, R.M. & Bouazza, A., 2013. Thermal conductivity of geosynthetics. Geotextiles and Geomembranes,
39, pp.1–8.
Southen, J.M. & Rowe, R.K., 2007. Evaluation of the water retention curve for geosynthetic clay liners.
Geotextiles and Geomembranes, 25(1), pp.2–9.
Southen, J.M. & Rowe, R.K., 2004. Investigation of the Behavior of Geosynthetic Clay Liners Subjected to
Thermal Gradients in Basal Liner Applications. Journal of ASTM International, 1(2), pp.1–13.
Southen, J.M. & Rowe, R.K., 2005a. Laboratory Investigation of Geosynthetic Clay Liner Desiccation in a
Composite Liner Subjected to Thermal Gradients. Journal of Geotechnical and Geoenvironmental
Engineering, 131(7), pp.925–935.
Southen, J.M. & Rowe, R.K., 2005b. Modelling of thermally induced desiccation of geosynthetic clay liners.
Southen, J.M. & Rowe, R.K., 2011. Numerical modelling of thermally induced desiccation of geosynthetic
clay liners observed in laboratory experiments. Geosynthetics International, 23(5), pp.289–303.
Tarantino, A., 2009. A water retention model for deformable soils. Géotechnique, 59(9), pp.751–762.
Thomas, H.R. & He, Y., 1997. A coupled heat-moisture transfer theory for deformable unsaturated soil and
its algorithmic implementation. International Journal for Numerical Methods in Engineering, 40(18),
pp.3421–3441.
Touze-Foltz, N. et al., 2016. A review of the performance of geosynthetics for environmental protection.
Wang, Y., Ma, J. & Guan, H., 2016. A mathematically continuous model for describing the hydraulic
properties of unsaturated porous media over the entire range of matric suctions. Journal of
Hydrology.
Wang, Z. & Liu, L., 2012. The Swelling Pressure of Na–Bentonite: Study with a Density Functional Approach.
Chemistry Letters, 41(10), pp.1346–1348.
Ye, W., Chen, B. & Chen, Y., 2009. Effect of temperature on soil-water characteristics and hysteresis of
compacted Gaomiaozi bentonite. Journal of Central South University of Technology, 16, pp.821–826.
-181
Yesiller, N. et al., 2014. Effects of hydration fluid on moisture-suction relationships for geosynthetic clay
liners. In Unsaturated Soils: Research & Applications. CRC Press, pp. 1023–1029.
Yeşiller, N., Hanson, J.L. & Liu, W.-L., 2005. Heat Generation in Municipal Solid Waste Landfills. Journal of
Geotechnical and Geoenvironmental Engineering, 131(11), pp.1330–1344.
Zhou, A.-N.N., Sheng, D. & Li, J., 2014. Modelling water retention and volume change behaviours of
unsaturated soils in non-isothermal conditions. Computers and Geotechnics, 55, pp.1–13.
Zhou, C. & Ng, C.W.W., 2014. A new and simple stress-dependent water retention model for unsaturated
soil. Computers and Geotechnics, 62, pp.216–222.
Zhou, Y., Rajapakse, R.K.N.D. & Graham, J., 1998. Coupled Heat-Moisture-Air Transfer in Deformable
Unsaturated Media. Journal of Engineering Mechanics, 124(10), pp.1090–1099.
Zhou, Y. & Rowe, R.K., 2003. Development of a technique for modelling clay liner desiccation. International
Journal for Numerical and Analytical Methods in Geomechanics, 27(6), pp.473–493.
Zhu, J. & Mohanty, B.P., 2002. Upscaling of soil hydraulic properties for steady state evaporation and
infiltration. Water Resources Research, 38(9), pp.17-1-17–13.
Zornberg, J.G., McCartney, J.S. & Swan, R.H., 2005. Analysis of a Large Database of GCL Internal Shear
Strength Results. Journal of Geotechnical and Geoenvironmental Engineering, 131(3), pp.367–380.
-182
Appendix A: Derivatives of the Proposed SWCCs
The first derivatives of degree of saturation with respect to the state variables are required to be computed
for implementation of the new SWCCs in CODE_BRIGHT since the Newton-Raphson method is used.
VG version of SWCC:
The effective degree of saturation is a function of gas pressure, liquid pressure, void ratio, and
temperature in this equation:
1 −𝜆
𝑃𝑔 − 𝑃𝑙 1−𝜆
𝑆𝑒 = (1 + ( ) ) A1
𝑃
and
1 −𝜆
𝑃𝑔 − 𝑃𝑙 1−𝜆
𝑆𝑟 = 𝑆𝑟𝑠 + (𝑆𝑠 − 𝑆𝑟𝑠 ) (1 + ( ) ) A2
𝑃
where a more general form of 𝑃 can be expressed as:
𝜎 𝑒 −𝛽
𝑃 = 𝑃0 ( ) 𝑒 A3
where:
𝑒𝑟𝑒𝑓 = the reference void ratio
The derivative with respect to displacement is easy and is defined with the 𝐵 matrix. In addition, the
partial derivative of void ratio with respect to the state variables is zero since it is assumed that the
variation of void ratio occurs at a slow rate. The derivatives with respect to 𝑃𝑔 , 𝑃𝑙 , 𝑇 are:
𝜆
𝑃 − 𝑃𝑙 1−𝜆 1+𝜆
(𝑆𝑠 − 𝑆𝑟𝑠 )𝜆 ( 𝑔
𝜕𝑆𝑟 𝑃 ) 𝑆𝑒 𝜆 A4
=−
𝜕𝑃𝑔 (1 − 𝜆)𝑃
𝜆
𝑃𝑔 − 𝑃𝑙 1−𝜆 1+𝜆
𝜕𝑆𝑟 (𝑆𝑠 − 𝑆𝑟𝑠 )𝜆 ( ) 𝑆𝑒 𝜆 A5
𝑃
=
𝜕𝑃𝑙 (1 − 𝜆)𝑃
-183
𝜕𝑆𝑟 𝜕𝑆𝑟 𝜕𝜎
= A6
𝜕𝑇 𝜕𝜎 𝜕𝑇
where:
1
𝑃 − 𝑃𝑙 1−𝜆 1+𝜆
(𝑆𝑠 − 𝑆𝑟𝑠 )𝜆 ( 𝑔
𝜕𝑆𝑟 𝑃 ) 𝑆𝑒 𝜆 A7
=
𝜕𝜎 (1 − 𝜆)𝜎
and considering the surface tension equation by IAPWS (2014), i.e. Eq. 4-13, gives:
𝑇 0.256 𝑇 1.256
𝜕𝜎 (−0.111062𝑇𝑐 − 0.185103𝑇) (1 − ) + 0.147375𝑇𝑐 (1 − )
𝑇𝑐 𝑇𝑐 A8
=
𝜕𝑇 𝑇𝑐2
where 𝑇 and 𝑇𝑐 are absolute temperatures (K).
FX version of SWCC:
Following the same approach, degree of saturation for the FX version of SWCC can be re-written in a more
general form as:
𝑠
𝑆𝑟 = 𝐶(𝑠) {ln [exp(1) + ( ) ]} A9
𝑎𝑓
where 𝑎𝑓 is:
𝜎 𝑒 −𝛽
𝑎𝑓 = 𝑎𝑓0 ( ) 𝑒 A10
The derivatives with respect to 𝑃𝑔 , 𝑃𝑙 , 𝑇 are:
𝑠 𝑛𝑓
𝑓(𝑠)−1−𝑚𝑓 (𝑠 exp(𝑓(𝑠))𝑓(𝑠) + 𝑚𝑓 𝑛𝑓 (𝑎 ) (𝑠 + 𝜓𝑠 )𝐶(𝑠) ln[1 + (106 /𝜓𝑟 )])
𝜕𝑆𝑟 𝑓 A11
=−
𝜕𝑃𝑔 𝑠 exp(𝑓(𝑠)) (𝑠 + 𝜓𝑟 ) ln[1 + (106 /𝜓𝑟 )]
𝑠 𝑛𝑓
𝑓(𝑠)−1−𝑚𝑓 (𝑠 exp(𝑓(𝑠))𝑓(𝑠) + 𝑚𝑓 𝑛𝑓 ( ) (𝑠 + 𝜓𝑠 )𝐶(𝑠) ln[1 + (106 /𝜓𝑟 )])
𝜕𝑆𝑟 𝑎𝑓 A12
=
𝜕𝑃𝑙 𝑠 exp(𝑓(𝑠)) (𝑠 + 𝜓𝑟 ) ln[1 + (106 /𝜓𝑟 )]
-184
𝜕𝑆𝑟 𝜕𝑆𝑟 𝜕𝜎
= A13
𝜕𝑇 𝜕𝜎 𝜕𝑇
where:
𝜕𝑆𝑟 𝑚𝑓 𝑛𝑓 (exp(𝑓(𝑠)) − exp(1))𝑓(𝑠)−1−𝑚𝑓 𝐶(𝑠)

= A14
𝜕𝜎 exp(𝑓(𝑠)) 𝜎
where:
1
𝐶(𝑠) 𝑚𝑓 A15
𝑓(𝑠) = ( )
𝑆𝑟
and the derivative of surface tension with respect to temperature is defined by equation A8.
-185
Appendix B: Unsaturated Mass-Volume Relations for GCL
Volumetric water content, 𝜃, can be directly calculated as follows if gravimetric water content, w, is
known:
𝜌𝑑
𝜃=𝑤 B1
𝜌𝑤
where:
𝜌𝑑 = dry density of specimen, kg/m3
𝜌𝑤 = density of water, it is 998 kg/m3 at 20°C
or if volume of the water is known:
𝑉𝑤
𝜃= B2
𝑉
where:
𝑉𝑤 = volume of water, m3
𝑉 = volume of specimen, m3.
Bulk void ratio of GCL was introduced by Petrov and Rowe (1997) as follows:
𝐻𝐺𝐶𝐿 − 𝐻𝑠
𝑒𝑏𝑢𝑙𝑘 = B3
𝐻𝑠
where:
𝐻𝐺𝐶𝐿 = height of GCL, m
𝐻𝑠 = height of solids, m.
and the height solids for GCL is defined as follows (Petrov and Rowe, 1997):
𝑀𝑏 𝑀𝑔𝑒𝑜
𝐻𝑠 = + B4
𝜌𝑏 (1 + 𝑤0 ) 𝜌𝑔𝑒𝑜
where:
𝑀𝑏 = mass per area of bentonite, kg/m2
𝑀𝑔𝑒𝑜 = mass per area of geotextiles, kg/m2
𝜌𝑏 = density of bentonite solids, kg/m3
-186
𝜌𝑔𝑒𝑜 = density of geotextile solids, kg/m3
𝑤0 = initial gravimetric water content of GCL.
Equation B4 in terms of specific gravity values of bentonite, cover geotextile, and carrier geotextile is
as follow:
1 𝑀 𝑀 𝑀
𝐻𝑠 = (( ) +( ) +( ) ) B5
𝜌𝑤 𝐺𝑠 (1 + 𝑤0 ) 𝑏𝑒𝑛𝑡𝑜𝑛𝑖𝑡𝑒 𝐺𝑠 𝑐𝑜𝑣𝑒𝑟 𝐺𝑠 𝑐𝑎𝑟𝑟𝑖𝑒𝑟
where
𝐺𝑠 = specific gravity of each component
Degree of saturation is:
𝜃
𝑆𝑟 = B6
𝜙
where:
𝜙 = porosity
and porosity can be calculated as follows:
𝜙= B7
1 + 𝑒𝑏𝑢𝑙𝑘
Substituting equation B7 in B6 gives degree of saturation as:
(1 + 𝑒𝑏𝑢𝑙𝑘 )
𝑆𝑟 = 𝜃 B8
and degree of saturation in terms of gravimetric can be derived by substituting equation B1 in B8 as

follows:
𝜌𝑑 (1 + 𝑒𝑏𝑢𝑙𝑘 )
𝑆𝑟 = 𝑤 B9
𝜌𝑤 . 𝑒𝑏𝑢𝑙𝑘
In addition, Bulk specific gravity of GCL can be defined as:
𝜌𝑠 𝑏𝑢𝑙𝑘 𝑀𝐺𝐶𝐿
𝐺𝑠 𝑏𝑢𝑙𝑘 = = B10
𝜌𝑤 𝜌𝑤 𝐻𝑠 (1 + 𝑤0 )
-187
Appendix C: Sydney Water Quality
Table C1. Sydney water quality modified from:
http://www.sydneywater.com.au/web/groups/publicwebcontent/documents/document/zgrf/mdq0/~edisp/dd_044727.pdf
ADWG ADWG
Property/Parameter Unit Limit Limit Value
(Health) (Aesthetic)
True colour TCU or HU na 15 <2 – 4
Turbidity NTU na 5 0.1 – 0.2
Total dissolved solids mg/L na 600 100 – 136
pH pH units na 6.5 - 8.5 7.8 – 8.0
Conductivity mS/m na na 18 – 21
Total hardness mg CaCO3 /L na 200 46 - 61
Calcium hardness mg CaCO3 /L na na 30 – 40
Magnesium hardness mg CaCO3 /L na na 19 – 23
Alkalinity mg CaCO3 /L na na 31 – 42
Temperature degrees C na na 15 – 23
Dissolved oxygen % saturation na >85% 103 – 124
Free chlorine mg/L 5 0.6 <0.04 – 0.04
Monochloramine mg/L 3 0.5 0.60 – 1.40
Trihalomethanes mg/L 0.25 na 0.010 – 0.124
Aluminium mg/L na 0.2 0.010 – 0.022
Ammonia (as NH3) mg/L na 0.5 0.11 – 0.40
Arsenic mg/L 0.01 na <0.001
Cadmium mg/L 0.002 na <0.001
Calcium mg/L na na 12.1 – 16.5
Chloride mg/L na 250 25.5 – 30.3
Chromium mg/L 0.05 na <0.0004
Copper mg/L 2 1 0.008 – 0.050
Cyanide mg/L 0.08 na <0.005
Fluoride mg/L 1.5 na 0.96 – 1.10
Iron mg/L na 0.3 0.010 – 0.030
Lead mg/L 0.01 na <0.001
Nickel mg/L 0.02 na <0.001
Magnesium mg/L na na 4.1 – 5.3
Manganese mg/L 0.5 0.1 <0.001 – 0.002
Mercury mg/L 0.001 na <0.0001
Nitrate (as NO3) mg/L 50 na 0.21 – 1.11
Nitrite (as NO2) mg/L 3 na 0.003 – 0.100
Phosphorous mg/L na na 0.007 – 0.009
Potassium mg/L na na 1.6 – 2.3
Reactive silica (as SiO2) mg/L na <80 mg/L 2.5 – 4.9
Selenium mg/L 0.01 na <0.003
Silver mg/L 0.1 na <0.003
na = no published health or aesthetic guideline value, ADWG = Australian Drinking Water Guidelines 2011
-188
Appendix D: Calibration of TDRs and Thermocouples
Volumetric water content measurements using TDRs for subsoil SW are presented in Figure D1 for both
CS655 and MP306 sensors. A third degree polynomial was fitted to CS655 and a linear regression was used
for MP306 sensors. Calibration was performed on the data from all available six CS655 and two MP306
TDRs. This is because the output from sensors for each type were essentially identical.
0.35 0.2
CS655
0.30 MP306
Poly. (CS655)
0.15 Linear (MP306)
0.25
0.20
0.1
θ
θ
0.15
0.10
0.05
y = 0.0057x - 0.0764
0.05 y= 11.333x3 - 5.6988x2 + 1.8893x + 0.0009 R² = 0.9755
R² = 0.9762
0.00 0
0.000 0.100 0.200 0.300 10 20 30 40
Sensor output (a) (b) Sensor output
Figure 0D1. Volumetric water content calibration of TDRs for subsoil (SW): a) CS655; b) MP306
Volumetric water content calibration equation for CS655 is:
𝜃𝐶𝑆655 = 11.33𝑥 3 − 5.6988𝑥 2 + 1.8893𝑥 + 0.0009 D1
and for MP306:
𝜃𝑀𝑃306 = 0.0057𝑥 − 0.0764 D2
where:
𝑥 = the sensor (volumetric water content) output
The CS655 sensors are able to measure both water content and temperature at the same time. These
sensors are used to track temperature change of the soil inside columns. Moreover, thermocouples are
used for measurement of temperature of liquid inside the top loading piston. Calibration check for all
temperature sensors (including CS655 and thermocouples) are presented in Figure D2. A thermal bath was
used for controlling temperature of water and outputs from the sensors were compared with a mercury
thermometer with a precision of ±0.1°C. Only one calibration equation was derived for CS655 temperature
sensors since the standard deviation of sensors readings were small and the sensor outputs were almost
-189
identical. However, separate calibration check was performed on each thermocouple, i.e. thermo#1 used
in column A and thermo#2 used in column B.
100 100
90 90
y = 1.0541x - 2.7599 y = 1.0559x - 2.8317
80 80
R² = 0.9998 R² = 0.9998
70 70
60 60
T(°C)
T(°C)
50 50
40 40
30 30
20 Thermo#1 (Col A) 20 Thermo#2 (Col B)
10 Linear (Thermo#1 (Col A)) 10 Linear (Thermo#2 (Col B))
0 0
0 20 40 60 80 100 0 20 40 60 80 100
(a) Sensor output (°C) (b) Sensor output (°C)
80
70
y = 0.9832x + 0.7912
60 R² = 0.9998
50
T(°C)
40
30
CS655
20
Linear (CS655)
10
0
0 20 40 60 80
(c) Sensor output (°C)
Figure D2. Calibration of thermocouples: a) Thermo#1; b) Thermo#2; c) CS655
Temperature calibration equation for CS655 is:
𝑇𝐶𝑆655 = 0.9832𝑥 + 0.7912 D3
and for thermo#1:
𝑇𝑡ℎ𝑒𝑟𝑚𝑜#1 = 1.0541𝑥 − 2.7599 D4
and for thermo#2:
𝑇𝑡ℎ𝑒𝑟𝑚𝑜#2 = 1.0559𝑥 − 2.8317 D5
where:
𝑥 = the sensor (temperature) output, °C
-190
Appendix E: Thermo-Hydro-Mechanical Interactions
The interactions between thermal, hydraulic, and mechanical processes for an unsaturated media are presented below:
Effects: form thermal phenomena form hydraulic phenomena form mechanical phenomena
on thermal phenomena:
 Heat storage Heat storage is proportional to temperature 1. Flow of fluids modifies the amount of water and Change in porosity changes the amount of
air stored. space available for fluids
2. Phase change modifies heat storage through the
latent heat of vapour.
 Heat conduction Fourier’s law governs Flow of fluids modifies thermal conductivity Change in porosity changes thermal
conductivity
 Heat advection - Heat transport by flow of fluids and vapour -
 Phase change Vapour pressure is a function of temperature Vapour pressre is a fuction of suction -
on hydraulic phenomena:
 Storage of fluids and vapour 1. Densities of fluid and vapour are funcions of 1. Density of liquid is a funcion of liquid pressure. Change in porosity changes the amount of
temperature. 2. Vapour density is a funciton of suction and gas space available for fluids
2. Phase change alters amount of water in liquid and pressure.
gas phase.
3. Mass of dissovled air is a funciotn of temperature.
 Flow of fluids 1. Fluids viscosities are functions of temperature. 1. Darcy’s law governs Permeability of liquid is a function of
2. Degree of saturation changes with temperature. 2. Permeability to liquid and gas are functions of porosity.
degree of saturation to liquid and gas, respctively.
 Vapour transport Vapour pressure is a function of temperature. 1. Fick’s law governs Diffusion coefficient is a function of porosity.
2. Vapour diffusion depends on degree of
saturation.
 Dissolved air trasport Diffusion coefficient is a function of temperature Fick’s law governs Diffusion coefficient is a function of porosity.
on mechanical phenomena:
 Stress/strain field Thermal expansion of materials Deformation depends on suction Deforamation depends on stress
-191
-192

Thermo-Hydro-Mechanical Behaviour of Composite Geosynthetic Lining Systems Under High Temperature and Low Pressure

Enviado por

Dados do documento

Título original

Direitos autorais

Formatos disponíveis

Compartilhar este documento

Compartilhar ou incorporar documento

Opções de compartilhamento

Você considera este documento útil?

Este conteúdo é inapropriado?

Direitos autorais:

Formatos disponíveis

Thermo-Hydro-Mechanical Behaviour of Composite Geosynthetic Lining Systems Under High Temperature and Low Pressure

Enviado por

Direitos autorais:

Formatos disponíveis

Faculty of Engineering and IT

School of Civil Engineering

Thermo-Hydro-Mechanical Behaviour of Composite Geosynthetic

1 Background and Aims ............................................................................................................................ 1

1.1 General .......................................................................................................................................... 1

1.2 Objectives ...................................................................................................................................... 3

1.3 Structure of the Thesis .................................................................................................................. 4

2 Literature Review .................................................................................................................................. 6

2.2 Coal Seam Gas and Brine Ponds .................................................................................................... 6

2.3 Geomembranes and Geosynthetic Clay Liners ............................................................................. 9

2.4 Hydration of Geosynthetic Clay Liners ........................................................................................ 11

2.5 Desiccation of Geosynthetic Clay Liners...................................................................................... 16

2.6 Shear Strength of GCL ................................................................................................................. 20

2.7 Soil-Water Characteristics Curves ............................................................................................... 21

2.9 Defects in Geomembrane ........................................................................................................... 28

2.10 Summary and Gap Identification................................................................................................. 29

3 Research Design .................................................................................................................................. 30

3.1.1 Aims ..................................................................................................................................... 31

3.1.2 Step 1 ................................................................................................................................... 31

3.1.3 Step 2 ................................................................................................................................... 33

3.2 Column Studies ............................................................................................................................ 33

3.3 Development and Validation of a Theoretical Thermo-Hydro-Mechanical Model..................... 34

3.4 Engineering Applications ............................................................................................................. 34

4.1 Introduction ................................................................................................................................. 36

4.2 Governing Equations ................................................................................................................... 37

4.3 Soil-Water Characteristic Curves ................................................................................................. 40

4.3.1 Approach based on van Genuchten (1980) (VG version) .................................................... 41

4.3.2 Selection of 𝑒𝑟𝑒𝑓 ................................................................................................................. 45

4.3.3 Extension to Fredlund and Xing (1994) (FX version) ........................................................... 45

4.4 Intrinsic Permeability for Liquid Phase ........................................................................................ 46

4.5 Relative Hydraulic Conductivity for Liquid and Gas .................................................................... 47

4.6 Flux of Vapour ............................................................................................................................. 47

4.7 Flux of Heat.................................................................................................................................. 48

4.8 Non-linear Elasticity .................................................................................................................... 48

4.9 Numerical Implementation ......................................................................................................... 49

5 Experimental Determination of Soil-Water Characteristic Curves of GCL .......................................... 52

5.1 Introduction ................................................................................................................................. 52

5.2 Material ....................................................................................................................................... 52

5.3 Experimental Program................................................................................................................. 53

5.4 Sample Preparation ..................................................................................................................... 54

5.5 Axis-translation: Pneumatic Suction-controlled Oedometer (Barcelona Cell) ............................ 56

5.6 Dew Point Method: WP4C and VSA ............................................................................................ 58

5.7 Mass-volume Relations for GCL .................................................................................................. 60

5.8 Results ......................................................................................................................................... 62

5.8.1 Axis-translation .................................................................................................................... 62

5.8.2 Dewpoint method ............................................................................................................... 70

5.8.3 Vapour equilibrium technique (VET) experiments (Monash University) ............................ 77

5.9 Summary of SWCC Experiments .................................................................................................. 79

5.10.1 Approach based on van Genuchten (1980) (VG version) .................................................... 82

5.11 Comparison to Existing GCL SWCCs............................................................................................. 89

5.12 Application to Existing GCL SWCC Measurements ...................................................................... 91

5.13 Summary and Limitations of Predictions .................................................................................... 95

6 Column Studies of Desiccation of GCL ................................................................................................ 97

6.1 Introduction ................................................................................................................................. 97

6.3 Experimental Program................................................................................................................. 98

6.4 Testing Apparatus........................................................................................................................ 99

6.5 Selection of Subsoil ................................................................................................................... 102

6.6 Permeability of GCL and Subsoil................................................................................................ 103

6.7 SWCC and Unsaturated Hydraulic Conductivity of Sand ........................................................... 106

6.8 Thermal Conductivity of GMB, GCL, and Subsoil ...................................................................... 108

6.9 Mechanical Properties of GCL ................................................................................................... 109

6.10 Results ....................................................................................................................................... 109

6.10.1 Permeability ...................................................................................................................... 109