experimental investigations of the soil-water characteristics of a volcanic soil

13
Experimental investigations of the soil-water characteristics of a volcanic soil Charles W.W. Ng and Y. W. Pang Abstract: Rain-induced landslides are common around the world. To analyse transient seepage and to predict pore- water pressure distribution in unsaturated slopes subjected to rainfall infiltration, it is essential to study soil-water char- acteristics and water permeability functions. The soil-water characteristic curve (SWCC) is a relationship between suc- tion and water content or degree of saturation. Conventionally, only the drying soil-water characteristic curve of soil specimens is determined in a pressure-plate extractor without the application of any external stress. In this paper, the influences of initial dry density and initial water content, history of drying and wetting, soil structure, and the stress state upon the desorption and adsorption soil-water characteristics of a completely decomposed volcanic soil in Hong Kong are examined and discussed. The experimental results presented are obtained by using a conventional volumetric pressure-plate extractor and a newly modified one-dimensional stress-controllable pressure-plate extractor with deforma- tion measurements. The SWCC of a recompacted specimen is very different from that of a natural specimen with the same initial soil density and initial water content. The SWCC of the recompacted specimen is highly dependent on the history of drying and wetting. The rates of desorption and adsorption are substantially higher at the first drying and wetting cycle than at the second drying and wetting cycle. The size of the hysteresis loop of the recompacted specimen is considerably larger than that of the natural specimens. The SWCC of soil is stress-state dependent. For recompacted specimens subjected to different stress states, the higher the applied stresses, the lower the rate of desorption and the smaller the size of the hysteresis loops. However, for natural specimens, the size of the hysteresis loops seems to be independent of the stress state. Under a higher applied stress, natural specimens exhibit lower rates of desorption and adsorption. Key words: volcanic soil, SWCC, drying and wetting, stress-state dependent. Résumé : Les glissements de terrain induits par la pluie sont fréquents dans le monde. Pour analyser l’infiltration transitoire et prédire la distribution de la pression interstitielle dans les talus non saturés soumis à une infiltration de pluie, il est essentiel d’étudier les caractéristiques sol-eau et les fonctions de perméabilité à l’eau. La courbe caractéristique sol-eau (SWCC) est une relation entre la succion et la teneur en eau ou le degré de saturation. Conventionnellement, seulement la courbe caractéristique sol-eau de l’assèchement de spécimens de sol est déterminée dans un extracteur à plaque de pression sans application de quelque contrainte extérieure. Dans cet article, on examine et discute l’influence de la densité initiale et de la teneur en eau initiale, de l’histoire du séchage et mouillage, de la structure du sol, et de l’état des contraintes sur les caractéristiques de la désorption et adsorption sol-eau d’un sol volcanique complètement décomposé de Hong Kong. Les résultats expérimentaux présentés sont obtenus au moyen d’un extracteur à plaque conventionnel avec pression volumétrique et un extracteur à plaque nouvellement modifié avec pression unidimensionnelle contrôlable et mesure de déformation. Les courbes SWCC de spécimens recompactés sont très différentes de celle d’un spécimen naturel avec les mêmes densités et teneurs en eau initiales. Les courbes SWCC d’un spécimen recompacté dépend énormément de l’histoire du mouillage et séchage. Tant les taux de désorption que d’adsorption sont appréciablement plus élevés au premier qu’au deuxième cycle de séchage-mouillage. La dimension de la boucle d’hystérèse des spécimens recompactés est considérablement plus grande que celle des spécimens naturels. La courbe SWCC du sol dépend de l’état des contraintes. Pour les spécimens recompactés soumis à différents états de contraintes, plus les contraintes appliquées sont élevées, plus le taux de désorption est faible et plus la dimension de la boucle d’hystérèse est petite. Cependant, pour les spécimens naturels, la dimension des boucles d’hystérèse semble être indépendante de l’état des contraintes. Sous des contraintes appliquées plus élevées, les spécimens naturels montrent de plus faibles taux de désorption et d’adsorption. Mots clés : sol volcanique, SWCC, séchage mouillage, dépendance de l’état des contraintes. [Traduit par la Rédaction] Ng and Pang 1264 Introduction Slope instability in various unsaturated soils is attracting increasing attention in many countries around the world, es- pecially in the subtropical and tropical zones such as Hong Kong and Singapore. The causes of landslides in these slopes are attributed to a number of factors. Rain-induced Can. Geotech. J. 37: 1252–1264 (2000) © 2000 NRC Canada 1252 Received June 4, 1999. Accepted May 10, 2000. Published on the NRC Research Press website on December 11, 2000. C.W.W. Ng and Y.W. Pang. Department of Civil Engineering, Hong Kong University of Science and Technology, Clearwater Bay, Hong Kong.

Upload: y-w

Post on 14-Mar-2017

212 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Experimental investigations of the soil-water characteristics of a volcanic soil

Experimental investigations of the soil-watercharacteristics of a volcanic soil

Charles W.W. Ng and Y. W. Pang

Abstract: Rain-induced landslides are common around the world. To analyse transient seepage and to predict pore-water pressure distribution in unsaturated slopes subjected to rainfall infiltration, it is essential to study soil-water char-acteristics and water permeability functions. The soil-water characteristic curve (SWCC) is a relationship between suc-tion and water content or degree of saturation. Conventionally, only the drying soil-water characteristic curve of soilspecimens is determined in a pressure-plate extractor without the application of any external stress. In this paper, theinfluences of initial dry density and initial water content, history of drying and wetting, soil structure, and the stressstate upon the desorption and adsorption soil-water characteristics of a completely decomposed volcanic soil in HongKong are examined and discussed. The experimental results presented are obtained by using a conventional volumetricpressure-plate extractor and a newly modified one-dimensional stress-controllable pressure-plate extractor with deforma-tion measurements. The SWCC of a recompacted specimen is very different from that of a natural specimen with thesame initial soil density and initial water content. The SWCC of the recompacted specimen is highly dependent on thehistory of drying and wetting. The rates of desorption and adsorption are substantially higher at the first drying andwetting cycle than at the second drying and wetting cycle. The size of the hysteresis loop of the recompacted specimenis considerably larger than that of the natural specimens. The SWCC of soil is stress-state dependent. For recompactedspecimens subjected to different stress states, the higher the applied stresses, the lower the rate of desorption and thesmaller the size of the hysteresis loops. However, for natural specimens, the size of the hysteresis loops seems to beindependent of the stress state. Under a higher applied stress, natural specimens exhibit lower rates of desorption andadsorption.

Key words: volcanic soil, SWCC, drying and wetting, stress-state dependent.

Résumé: Les glissements de terrain induits par la pluie sont fréquents dans le monde. Pour analyser l’infiltrationtransitoire et prédire la distribution de la pression interstitielle dans les talus non saturés soumis à une infiltration depluie, il est essentiel d’étudier les caractéristiques sol-eau et les fonctions de perméabilité à l’eau. La courbecaractéristique sol-eau (SWCC) est une relation entre la succion et la teneur en eau ou le degré de saturation.Conventionnellement, seulement la courbe caractéristique sol-eau de l’assèchement de spécimens de sol est déterminéedans un extracteur à plaque de pression sans application de quelque contrainte extérieure. Dans cet article, on examineet discute l’influence de la densité initiale et de la teneur en eau initiale, de l’histoire du séchage et mouillage, de lastructure du sol, et de l’état des contraintes sur les caractéristiques de la désorption et adsorption sol-eau d’un solvolcanique complètement décomposé de Hong Kong. Les résultats expérimentaux présentés sont obtenus au moyend’un extracteur à plaque conventionnel avec pression volumétrique et un extracteur à plaque nouvellement modifié avecpression unidimensionnelle contrôlable et mesure de déformation. Les courbes SWCC de spécimens recompactés sonttrès différentes de celle d’un spécimen naturel avec les mêmes densités et teneurs en eau initiales. Les courbes SWCCd’un spécimen recompacté dépend énormément de l’histoire du mouillage et séchage. Tant les taux de désorption qued’adsorption sont appréciablement plus élevés au premier qu’au deuxième cycle de séchage-mouillage. La dimension dela boucle d’hystérèse des spécimens recompactés est considérablement plus grande que celle des spécimens naturels. Lacourbe SWCC du sol dépend de l’état des contraintes. Pour les spécimens recompactés soumis à différents états decontraintes, plus les contraintes appliquées sont élevées, plus le taux de désorption est faible et plus la dimension de laboucle d’hystérèse est petite. Cependant, pour les spécimens naturels, la dimension des boucles d’hystérèse semble êtreindépendante de l’état des contraintes. Sous des contraintes appliquées plus élevées, les spécimens naturels montrent deplus faibles taux de désorption et d’adsorption.

Mots clés: sol volcanique, SWCC, séchage mouillage, dépendance de l’état des contraintes.

[Traduit par la Rédaction] Ng and Pang 1264

Introduction

Slope instability in various unsaturated soils is attractingincreasing attention in many countries around the world, es-pecially in the subtropical and tropical zones such as HongKong and Singapore. The causes of landslides in theseslopes are attributed to a number of factors. Rain-induced

Can. Geotech. J.37: 1252–1264 (2000) © 2000 NRC Canada

1252

Received June 4, 1999. Accepted May 10, 2000.Published on the NRC Research Press website onDecember 11, 2000.

C.W.W. Ng and Y.W. Pang. Department of CivilEngineering, Hong Kong University of Science andTechnology, Clearwater Bay, Hong Kong.

I:\cgj\Cgj37\Cgj06\T00-056.vpWednesday, December 06, 2000 8:39:50 AM

Color profile: Generic CMYK printer profileComposite Default screen

Page 2: Experimental investigations of the soil-water characteristics of a volcanic soil

failures are the most common ones (Lumb 1962; Brand1984; Fukuoka 1980; Wolle and Hachich 1989; Fourie 1996;Lim et al. 1996; Ng and Shi 1998; Geotechnical EngineeringOffice 2000). In Hong Kong, most slopes have soils that aredesiccated from the ground surface to a considerable depth,and pore-water pressure at shallow depths in these slopes isgenerally negative with respect to atmospheric pressure. Thepresence and magnitude of matric suction have been foundto be crucial to the stability of some unsaturated soil slopes.

Infiltration of rainwater or ingress of a wetting front(Lumb 1975) lead to the development of a perched water ta-ble, a rise in the level of the main groundwater table, andwashouts (soil erosion due to concentrated water flow), re-sulting in an increase in pore-water pressure or a reductionin soil matric suction. This, in turn, results in a decrease inshear strength on the potential failure surface to a pointwhere equilibrium can no longer be sustained in the slopeand then failures occur. To understand and analyse transientseepage and pore-water pressure distribution in unsaturatedslopes subjected to rainfall infiltration, it is essential to studysoil-water characteristics and water permeability functionsof soils. The soil-water characteristic curve (SWCC) is a re-lationship between suction and water content or degree ofsaturation. Studies of SWCCs can assist in the understandingof pore-water pressure distributions in the slope and causesof landslide failures during rainstorms. Subsequently, pre-ventative measures can then be developed. Recently, labora-tory measurements of SWCCs have been used to study thegroundwater flow mechanism and to investigate the causesof major landslides in a decomposed volcanic soil in HongKong (Sun and Campbell 1998; Tung et al. 1998).

In this paper, the influences of several factors, includinginitial void ratio and water content, soil structure, history ofdrying and wetting, and stress state, upon the soil-watercharacteristics of a completely decomposed volcanic (CDV)soil in Hong Kong are investigated in the laboratory. Ofthese factors, the influence of stress state cannot be readilystudied using conventional methods because the conven-tional pressure-plate extractor is not equipped for applying avertical or confining stress to the specimen. A modifiedstress-controllable pressure-plate extractor (Ng and Pang2000) is used to measure the soil-water characteristics undervarious stress states. The total net normal stress can be con-trolled one-dimensionally and axial deformation can be mea-sured using this extractor. In addition, emphasis is placednot only on the desorption characteristics, but also on the ad-sorption characteristics of the volcanic soil. The traditionalcapillary model is applied to assist in understanding someobserved adsorption characteristic of the volcanic soil.

Background and brief literature review

The SWCC is a useful parameter for analysing transientwater flows in unsaturated soils and plays a similar role as aconsolidation curve describing the relationship between voidratio (or volume) and effective stress in saturated soils(Rahardjo and Leong 1997; Barbour 1998). Engineeringproperties of an unsaturated soil can be predicted using theSWCC and saturated soil properties (Brooks and Corey1966; van Genuchten 1980; Fredlund et al. 1994; Vanapalliet al. 1996; Fredlund 1998; Barbour 1998).

Soil slopes generally experience variations in environmen-tal and climate conditions throughout the year. Therefore,soil elements in an unsaturated slope are subjected to re-peated cycles of drying and wetting, which result in the hys-teresis of SWCC and hydraulic conductivity of the soils(Topp and Miller 1966; Maulem 1986; Hillel 1998; Lin andBenson 2000). These repeated drying and wetting cycles canlead to changes in soil structure through shrinking andswelling (Hillel 1982, 1998; Maulem 1986). Bell (1992)suggested that repeated drying and wetting could cause ag-gregation of soil particles in residual soils. Hence, the soil-water characteristic and hydraulic conductivity of most soilsare likely to be dependent on the history of drying and wet-ting.

Traditionally, only the drying soil-water characteristiccurves of soil specimens are measured under zero net normalstress conditions. However, in the actual field situations, theinfluence of stress state may not be negligible. Even thoughthe total net normal stress on the soil elements in an unsatu-rated slope is seldom altered, the stress state at each elementat different depths is different, which may affect the soil-water characteristic of these elements. To more realisticallypredict pore-water pressure distributions and hence the slopestability of an unsaturated slope, it is thus essential to inves-tigate the influence of stress state on soil-water characteris-tics. Although it is theoretically recognized that the stressstate of a soil has some influence on soil-water characteris-tics (Fredlund and Rahardjo 1993), few experimental resultscan be found in the literature, with the exception of those byVanapalli et al. (1998, 1999), who conducted a series of ex-periments to investigate the influence of soil structure andstress history on the soil-water characteristics of arecompacted clayey till. Recompacted fine-grained soil spec-imens were first loaded in an oedometer to different void ra-tios before their soil-water characteristics were measured ina conventional pressure-plate extractor. The results ofVanapalli et al. suggest that the rate of change of water con-tent with respect to matric suction of the specimensrecompacted at dry of optimum water content is higher thanthat of wet of optimum specimens due to their different soilstructures. Moreover, the stress history appears to have a sig-nificant effect on the soil-water characteristics of the speci-mens recompacted at dry of optimum, but not at wet ofoptimum. In their experiments, Vanapalli et al. essentiallyconsidered the overconsolidation effects but not, strictlyspeaking, the actual stress effects on SWCCs.

Descriptions of soil specimen

The volcanic soil used in this paper is a completely de-composed ash tuff and can be classified as a firm, moist,orangish brown, slightly sandy silt–clay with high plasticityaccording to the Geotechnical Control Office (1988). Thesoil specimens were obtained from an “undisturbed”200 mm × 200 mm × 200 mm block sample excavated froma hillside slope in Shatin, Hong Kong, for failure investiga-tions after a major landslide in July 1997 (Sun and Campbell1998). The grain-size distribution, as determined by sieveand hydrometer analyses (British Standards Institution1990), is shown in Fig. 1. In addition, some index propertiesof the soil are summarized in Table 1. The gravel, sand, silt,

© 2000 NRC Canada

Ng and Pang 1253

I:\cgj\Cgj37\Cgj06\T00-056.vpWednesday, December 06, 2000 8:39:51 AM

Color profile: Generic CMYK printer profileComposite Default screen

Page 3: Experimental investigations of the soil-water characteristics of a volcanic soil

and clay contents determined in the tuff are 4.9, 20.1, 36.6,and 37.1%, respectively, and it can be classified as a well-graded soil.

Experimental program

An experimental program was undertaken to investigatethe soil-water characteristics of a volcanic soil. In additionto three natural (or undisturbed) soil specimens, sevenrecompacted soil specimens with different initial conditionswere used for testing. Details of the testing program aresummarized in Table 2.

EquipmentThree different types of extractor were used for measuring

the SWCC. A conventional volumetric pressure-plate extrac-tor was used to measure the soil-water characteristics of un-saturated soils for suction ranges of 0–200 kPa. In thecontext of this paper, “conventional volumetric pressure-plate extractor” refers to a commercially available pressureextractor in which the principle of axis translation is appliedto control matric suction in a soil specimen and no externalload is imposed on the specimen (Fredlund and Rahardjo1993). This extractor was also adopted to study the hyster-esis of the soil-water characteristics associated with the dry-ing and wetting of the soil. For measuring the soil-watercharacteristic for suction ranges greater than 200 kPa (onlyone test), a 500 kPa pressure-plate extractor was adopted.

The conventional volumetric pressure-plate extractor wasnot equipped for applying any vertical or confining stress tothe specimen. Hence, an extractor for measuring the stress-dependent soil-water characteristic curve (SDSWCC) of un-saturated soils under one-dimensional (K0) stress conditionswas designed (Ng and Pang 2000) to investigate the effectsof stress on the soil-water characteristic of the volcanic soil.A schematic diagram of the modified pressure-plate extrac-tor is shown in Fig. 2. An oedometer ring equipped with a

high air entry ceramic plate at its base is located inside anair-tight chamber. Vertical load via a loading piston is ap-plied to a soil specimen inside the oedometer ring through aloading frame. Possible pulling away of soil specimen fromthe sides of the oedometer ring was minimized by applying avertical load and by testing the samples under small suctionranges. The airtightness of the chamber is maintained usingrubber O-rings at the openings. Along the piston, a load cellis attached inside the airtight chamber for determining theactual vertical net normal stress applied to the soil specimen.A dial gauge is attached for measuring the vertical displace-ment of the soil specimen. The size of specimen that can beaccommodated in this extractor is 70 mm in diameter and20 mm in height.

Similar to the conventional volumetric pressure-plate ex-tractor, the pore-air pressure,ua, is controlled through acoarse porous stone and a thin woven geotextile located atthe top of the specimen. The compression of the geotextilewas accounted for by calibration using a steel dummy speci-men in the extractor. The applied air is saturated using anair-saturator. The pore-water pressure,uw, is controlled at the

© 2000 NRC Canada

1254 Can. Geotech. J. Vol. 37, 2000

Fig. 1. Particle-size distribution of completely decomposed volcanic (CDV) soil.

Specific gravity (Mg/m3) 2.62Maximum dry density (kg/m3) 1603Optimum moisture content (%) 22Initial moisture content (%) 30Gravel content (≤62 mm, %) 4.9Sand content (≤2 mm, %) 20.1Silt content (≤63 µm, %) 36.6Clay content (≤2 µm, %) 37.1Liquid limit (%) 55.4Plastic limit (%) 33.4Plasticity index (%) 22

Table 1. Index properties of a completely de-composed volcanic soil (sandy silt–clay).

I:\cgj\Cgj37\Cgj06\T00-056.vpWednesday, December 06, 2000 8:39:53 AM

Color profile: Generic CMYK printer profileComposite Default screen

Page 4: Experimental investigations of the soil-water characteristics of a volcanic soil

atmospheric pressure through the high air entry ceramicplate mounted at the base of the specimen. The high air en-try ceramic plate will remain saturated if the matric suctiondoes not exceed the air-entry value of the plate (200 kPa).The matric suction,ua – uw, imposed on the soil specimenwill be the difference between the applied air and pore-waterpressures. The net normal stress can be controlled one-dimensionally and axial deformation can be measured usingthis extractor. Other details of this modified extractor areavailable in Ng and Pang (2000).

Preparation of soil specimensThe soil was oven-dried at 45°C for 48 h prior to the prep-

aration of a “recompacted” soil specimen. The soil was thenpulverized using a rubber hammer. Water was added to ob-tain the desired water contents (see Table 2) and the soil wasthoroughly mixed. After keeping the mixed soil in a plastic

bag for moisture equalization for about 24 h in atemperature- and moisture-controlled room, the soil wasthen statically recompacted to the desired density in anoedometer ring 70 mm in diameter and 20 mm high. A “nat-ural” soil specimen, on the other hand, was directly cut fromthe block sample into an oedometer ring with a cutting edge.Both natural and recompacted soil specimens were then sub-merged in deaired water inside a desiccator subjected to asmall vacuum for about 24 h to ensure almost full saturation.The 24 h duration was sufficient to ensure almost full satu-ration, as the soil has a high saturated hydraulic conductivityof 2.88 × 10–6 m/s, as measured by Ng and Pang (1998).

Testing procedures for a conventional volumetricpressure-plate extractor

A complete drying and wetting cycle was imposed oneach specimen during the tests. Initially the specimen was

© 2000 NRC Canada

Ng and Pang 1255

Specimenidentity*

Initial drydensity(Mg/m3)

Initialvoidratio

Relativecompaction(%)

Initial watercontent(%)

Appliedstress(kPa)

No. ofwetting–dryingcycles

Max. suctionapplied(kPa)

Measuredfinal voidratio

CDV-R1(C) 1.47 0.782 92 30.3 0 3 400 0.758CDV-R2(C) 1.50 0.747 94 30.3 0 1 200 0.736CDV-R3(C) 1.53 0.712 95 30.3 0 1 200 0.703CDV-R4(C) 1.47 0.782 92 22.0 0 1 200 0.767CDV-R5(C) 1.47 0.782 92 15.0 0 1 200 0.761CDV-R2(M) 1.47 0.782 92 30.3 40 1 200 0.724CDV-R3(M) 1.47 0.782 92 30.3 80 1 200 0.701CDV-N1(C) 1.47 0.782 92 30.3 0 1 100 0.771CDV-N2(M) 1.47 0.782 92 30.3 40 1 200 0.743CDV-N3(M) 1.47 0.782 92 30.3 80 1 200 0.695

* N and R denote natural and recompacted specimens, respectively; C and M in parentheses denote conventional and modified one-dimensionalvolumetric pressure-plate extractor, respectively.

Table 2. Summary of the testing program for determining soil-water characteristics.

Fig. 2. Schematic diagram of the modified one-dimensional volumetric pressure-plate extractor (Ng and Pang 2000) (1 bar = 100 kPa).

I:\cgj\Cgj37\Cgj06\T00-056.vpWednesday, December 06, 2000 8:39:58 AM

Color profile: Generic CMYK printer profileComposite Default screen

Page 5: Experimental investigations of the soil-water characteristics of a volcanic soil

subjected to an increasing matric suction in a series of stepsto measure the soil-water characteristic along the dryingpath. As the matric suction increased, water was expelledfrom the soil specimen into the ballast tube (see Fig. 2). Themovement of water in the ballast tube was continuouslymonitored using a ruler fixed along the ballast tube until theequilibrium condition was reached. Typically, 2–7 dayswere required to achieve equilibrium at a given suction. Anytrapped air was removed via the air trap removal attachment.After reaching equilibrium, the water levels in both the airtrap and the ballast tube were adjusted to their respectivelevels. Change of volume of water in the specimen betweentwo successive suctions was then precisely measured usingthe marked burette, which has a resolution of 0.1 mL. Infact, changes could be estimated with confidence to0.05 mL. For the specimens tested, the estimated accuracyof the measured volumetric water content was better than0.2%. After reaching the required maximum suction, a wet-ting process was then initiated by decreasing the value ofmatric suction in a series of steps. Water in the ballast tubewas absorbed by the soil specimen. The change in the vol-ume of water during wetting was determined as previouslydescribed. At the end of the test, each soil specimen wasoven-dried at 45°C for 24 h to determine its water content.The degrees of saturation at various matric suctions werethen determined from the change of volume of water and thefinal water content (i.e., backward calculations). Volumechange of the soil specimen was assumed to be zero for de-termining the volumetric water content and the degree ofsaturation using the conventional volumetric plate extractor.

Similar procedures cannot be used for soil specimenCDV-R1(C) because the soil-water characteristic for suctionranges greater than 200 kPa cannot be obtained using thevolumetric pressure-plate extractor. Hence, after the suctionreached 200 kPa, the soil specimen was removed from thevolumetric pressure-plate extractor and weighed immedi-ately. It was then placed in a 500 kPa pressure-plate extrac-tor and subjected to a suction value of 400 kPa. Uponreaching the equilibrium condition, the soil specimen wasweighed and put back into the conventional volumetric pres-sure-plate extractor to continue the wetting path. In total,three repeated cycles of drying and wetting were performedon this specimen.

Testing procedures for the modified volumetricpressure-plate extractor

The saturated soil specimens (i.e., CDV-R2(M), CDV-R3(M), CDV-N2(M), and CDV-N3(M)), were loaded to therequired net normal stresses (i.e., 40 and 80 kPa) in anoedometer for measuring the stress-dependent soil-watercharacteristics. These specimens were allowed to have freedrainage at the top and bottom for 24 h and hence they werepreconsolidated under their required corresponding net nor-mal stresses. The purpose of this preconsolidation was toeliminate the effects of consolidation on volume changesduring the subsequent measurements of soil-water character-istics at various suctions. After consolidation, the soil speci-mens were then removed from the oedometer and subjectedto the same applied net normal stresses in the modified volu-metric pressure-plate extractor. Consolidation carried out in

the oedometer, which is equipped with a low air entry valueporous disc, was to reduce the testing time.

Vertical deformation of the specimen was continuouslymonitored in the oedometer to ensure the end of primaryconsolidation was reached by studying the relationship be-tween deformation and root time. It was found that 24 h wassufficient for the completion of primary consolidation. Afterplacing the consolidated specimen in the modified volumet-ric pressure-plate extractor, good contacts between the speci-men and the ceramic plate were ensured by the applied netnormal stress.

For each drying and wetting path, the required net normalstress was maintained constant throughout the test and thevertical load and axial displacement of the soil specimenwere measured (see Fig. 2). To account for deformation ofthe loading piston subjected to an applied load, calibrationwas carried out. Thus, the actual vertical deformation andhence the actual volume change of the soil specimen weremeasured.

Soil-water characteristics of recompactedsoils

Influence of initial dry density on soil-watercharacteristics

The soil-water characteristics of specimens recompactedat the same initial water content but different dry densities(i.e., CDV-R1(C), CDV-R2(C), and CDV-R3(C)) are com-pared in Fig. 3. The structures of these three specimens arenot identical, as different compaction efforts were used dur-ing the preparation of specimens (Lambe 1958). However,this series of tests is appropriate for investigating the soil-water characteristics of a fill slope or embankment formedby various degrees of compaction at different depths.

The air-entry values for all three specimens fall within asmall range of matric suction values (see Fig. 3). The air-entry value for each specimen can be estimated by using a“judging-by-eye” method to extend a line from the constant-slope portion of the first drying soil-water characteristiccurve to intersect the suction axis at 100% saturation. Theestimated values from all three first drying curves vary be-tween 4 and 5 kPa (refer to Table 3). Only a very small in-crease in the air-entry value is observed between CDV-R1(dry density = 1.47 Mg/m3) and CDV-R3 (dry density = 1.53Mg/m3). This is likely due to the small increase in dry den-sity (i.e., only 4% increase in dry density between the twospecimens). By studying the experimental results presentedin Table 4 of Tinjum et al. (1997), two clay specimens com-pacted at 19.7% water content (dry unit weight = 17.1 kN/m3)and 20.0% water content (dry unit weight = 19.8 kN/m3) ofsoil M are suitable for comparisons. An increase in dry unitweight of 15.7% results in an increase in the air-entry valuefrom 2.5 kPa to 7.6 kPa. Croney and Coleman (1954) alsoreported that a specimen of silty sand with a high initialcompacted density had a higher air-entry value than that of aspecimen with a low initial compacted density. In addition,the higher the initial dry density, the slower the rate ofdesorption. This is consistent with the current test results.

As shown in Fig. 3, there is a marked hysteresis betweenthe drying and wetting curves for all soil specimens. This islikely attributed to the geometric nonuniformity of the

© 2000 NRC Canada

1256 Can. Geotech. J. Vol. 37, 2000

I:\cgj\Cgj37\Cgj06\T00-056.vpWednesday, December 06, 2000 8:39:58 AM

Color profile: Generic CMYK printer profileComposite Default screen

Page 6: Experimental investigations of the soil-water characteristics of a volcanic soil

individual pores, resulting in the so-called “ink-bottle” effect(Hillel 1982, 1998). The difference in the contact angles atthe receding soil–water interface during drying and at theadvancing soil–water interface during wetting may contrib-ute to the observed hysteresis. Moreover, any trapped air inthe “blind” or “dead-end” pores inside the soil specimen orin the testing system may account for the observed hyster-esis. There is a general trend that the size of the hysteresisloop decreases as the initial void ratio decreases. Similar testresults were also reported by Croney and Coleman (1954).The ink-bottle effect may be more pronounced in soils witha large pore-size distribution than that with a small pore-sizedistribution. Inaddition, an average large pore-sizedistributionin a loose specimen may lead to a larger difference in the re-ceding and advancing contact angles than that in a densespecimen.

At the end of the wetting path (i.e., 0.1 kPa suction), noneof the wetting curves reaches full saturation, and the looser the

sample, the lower the degree of saturation. The “non-return”of the wetting paths may be attributed to air trapped in thespecimens. For the observed difference in the degree of satu-ration achieved between a loose and a dense specimen, it ismore difficult for capillary force to displace air trapped inthe large pores than that in the small pores, and this may ac-count for the observed difference. Full saturation is very dif-ficult to achieve in a loose specimen through capillary actionalone. Moreover, the ink-bottle effect is likely to be morepronounced in a loose specimen than in a dense specimen.

Influence of initial water content on soil-watercharacteristics

Soil specimens must be recompacted to the same initialwater content and dry density to be qualified as “identical”(Lambe 1958). Specimens recompacted to the same dry den-sity at different initial water contents or using different

© 2000 NRC Canada

Ng and Pang 1257

Fig. 3. Influence of initial density on soil-water characteristics.

Specimenidentity Path

Estimated air-entryvalue (kPa)

CDV-R1(C) 1st drying 4CDV-R1(C) 2nd drying 2CDV-R1(C) 3rd drying 2CDV-R2(C) 1st drying 4CDV-R3(C) 1st drying 5CDV-R4(C) 1st drying 2CDV-R5(C) 1st drying 1.5CDV-R2(M) 1st drying 4CDV-R3(M) 1st drying 5CDV-N1(C) 1st drying 1.5CDV-N2(M) 1st drying 3CDV-N3(M) 1st drying 5

Table 3. Estimated air-entry values for a completely decomposedvolcanic soil.

Fig. 4. Schematic compaction curves at various dry densities andwater contents.

I:\cgj\Cgj37\Cgj06\T00-056.vpWednesday, December 06, 2000 8:40:05 AM

Color profile: Generic CMYK printer profileComposite Default screen

Page 7: Experimental investigations of the soil-water characteristics of a volcanic soil

compaction efforts are different soils due to different inher-ent structures. Thus, samples A, B, and C shown in Fig. 4are not identical, even though they have the same dry den-sity. The influence of soil structure, which is defined as thearrangement of soil particles, fabric, and bonding, on soil-water characteristics may be investigated by comparing thesoil-water characteristics of the specimens that have thesame dry density but were recompacted at different initialwater contents or with different compaction efforts.

Figure 5 shows the variation in the degree of saturationwith respect to matric suction for specimens CDV-R1(C),CDV-R4(C), and CDV-R5(C) recompacted to the same drydensity at wet of optimum, at optimum, and dry of optimuminitial water content, respectively. According to Benson andDaniel (1990) and Vanapalli et al. (1999), specimen CDV-R5(C), which is recompacted at dry of optimum initial watercontent, should have relatively large pores between the clodsof soil. On the contrary, specimen CDV-R1(C), which isrecompacted at wet of optimum initial water content, shouldhave no visible, relatively large interclod pores, and thus thissoil would be relatively homogenous. As all three specimenshave the same initial dry density, it is perhaps reasonable toassume that the total amount of voids would also be thesame. However, their pore-size distributions are likely to be

different. A capillary-tube analogy (assuming thatr1 ≥ r2 ≥r3 , wherer1, r2, and r3 are the radii of the tubes) shown inFigs. 6a and 6b can be used to illustrate the difference inpore-size distributions of CDV-R5(C) and CDV-R1(C), re-spectively. It is likely that the pore-size distribution of CDV-R4(C) lies somewhere in between.

Figure 5 and Table 3 show that the air-entry value ofCDV-R5(C) is lower than that of CDV-R1(C). A decreasingair-entry value is consistent with a decreasing compactionwater content. This is because a smaller air pressure is re-quired to enter the relatively large interclod pores (i.e.,r1 inFig. 6a) in specimen CDV-R5(C). The current experimentaldata are consistent with results published by Tinjum et al.(1997) and Vanapalli et al. (1999). Regarding the rate ofdesorption, there is no significant difference between thethree specimens, however.

During the wetting process, the degree of saturation in-creases as the matric suction decreases. The end points ofthe wetting paths at low suction (0.1 kPa) are different fromthose of the drying paths because of air trapped in the soil.This difference is higher for specimens recompacted with alarger initial water content. The overall ink-bottle effect wouldbe more pronounced in the uniform pore-size distribution(Fig. 6b) than in the nonuniform pore-size distribution

© 2000 NRC Canada

1258 Can. Geotech. J. Vol. 37, 2000

Fig. 5. Influence of initial water content (w) on soil-water characteristics.

Fig. 6. Capillary-tube analogy to illustrate the influence of pore-size distribution on saturation characteristics.

I:\cgj\Cgj37\Cgj06\T00-056.vpWednesday, December 06, 2000 8:40:10 AM

Color profile: Generic CMYK printer profileComposite Default screen

Page 8: Experimental investigations of the soil-water characteristics of a volcanic soil

(Fig. 6a), assuming that the two samples have the sameamount of total voids. This leads to a large hysteresis loopfor the specimen recompacted at wet of optimum comparedwith that for the specimen recompacted at dry of optimum.

Influence of the drying and wetting history on soil-water characteristics

The soil-water characteristics of soil specimen CDV-R1(C) subjected to three repeated drying and wetting cyclesare shown in Fig. 7. In all three cycles, marked hysteresisloops between the drying and wetting paths can be seen. Thesize of the hysteresis loop is the largest in the first cycle butseems to become independent of the drying and wetting his-tory after the completion of the first cycle. As the matricsuction decreases during wetting, none of the three wettingcurves returns to its original degree of saturation. As dis-cussed earlier, it is difficult to attain full saturation condi-tions mainly because of trapped air. The degree of saturationat low suction (i.e., 0.1 kPa) decreases as the number of dry-ing and wetting cycles increases, but at a reduced rate.

The desorption characteristic with respect to matric suc-tion is dependent on the drying and wetting history. The rateof desorption is relatively high during the first drying cyclecompared with that during the second and third drying cy-cles. This may be due to the presence of relatively largevoids initially. During the first wetting, significant volumechange is likely to take place and this results in smallervoids in the sample due to collapse of soil structures of avirgin soil (Chiu et al. 1998; Ng et al. 1998). Thus, a smallerrate of desorption for the second and third drying and wet-ting cycles is expected. Moreover, the estimated air-entryvalue reduces during the second drying but remains almostunchanged in the subsequent drying cycles (see Table 3).

The adsorption characteristic of the first wetting processalso seems to be different from that of the subsequent wet-ting processes. The value of matric suction at which the soilstarts to absorb water significantly is higher during the firstwetting cycle (i.e., about 50 kPa) than during the subsequentcycles (i.e., about 10 kPa). The rates of adsorption are sub-stantially different for the first and subsequent wetting cyclesat suctions ranging from 50 to 10 kPa. This might be causedby some soil structure changes after the first drying cycle.

According to Bell (1992), drying initiates cementation byaggregation formation, leading to some relatively largeinterpores formed between the aggregated soil lumps. Theselarge interpores reduce the specimen’s rate of absorptionalong a certain range of the wetting path.

Influence of stress state on recompactedand natural soils

Verification of constant-volume assumptionSoil-water characteristics are conventionally determined

using a pressure-plate extractor with the assumption that novolume change takes place throughout the test. This assump-tion is studied and verified using the newly modified extrac-tor.

The test results from the recompacted (i.e., CDV-R2(M)and CDV-N3(M)) and natural (i.e., CDV-N2(M) and CDV-N3(M)) soil specimens are shown in Figs. 8 and 9, respec-tively. The curves labeled “constant volume assumed” areobtained by ignoring volume change taking place throughoutthe tests. Figures 8 and 9 show that there is no significantdifference between the drying paths with and without vol-ume-change corrections until the matric suction reaches100 kPa, at which measurable shrinkage of the soil starts tooccur. The traditional method of interpretation by neglectingany volume reduction clearly underpredicts the volumetricwater content and degree of saturation of the soil specimenand leads to a higher rate of desorption than that from a testconsidering volume changes.

During the wetting process, the difference between thetwo wetting paths does not stay constant, indicating that dif-ferent degrees of swelling take place throughout. The influ-ence of volume change for the natural soil specimens seemsto be consistent with that for the recompacted soil speci-mens. In general, the higher the normal stress applied to asoil specimen, the smaller the average pore size of the speci-men, and the stiffer the specimen. In turn, it would have ahigher resistance to volume change due to drying. From nowon, volume-change corrections are applied for the test resultsobtained from the modified pressure volumetric plate extrac-tor in this paper.

© 2000 NRC Canada

Ng and Pang 1259

Fig. 7. Influence of drying and wetting cycles on soil-water characteristics of specimen CDV-R1(C).

I:\cgj\Cgj37\Cgj06\T00-056.vpWednesday, December 06, 2000 8:40:13 AM

Color profile: Generic CMYK printer profileComposite Default screen

Page 9: Experimental investigations of the soil-water characteristics of a volcanic soil

Influence of net normal stressBy using both the conventional and the modified volumet-

ric pressure extractor, soil-water characteristic and stress-dependent soil-water characteristic curves (SDSWCC) aremeasured and compared. The soil-water characteristics ofthe recompacted soil specimens vertically loaded with 0, 40,and 80 kPa net normal stresses underK0 conditions (CDV-R1(C), CDV-R2(M), and CDV-R3(M)) are shown in Fig. 10.There is a general tendency for the soil specimen subjectedto a higher stress to possess a slightly larger air-entry value(Table 3). As the matric suction keeps increasing, all speci-mens start to desaturate but at different rates. The higher theapplied load on the specimen, the lower the rate ofdesorption. This is likely to be caused by the presence of asmaller average pore-size distribution in the soil specimenunder higher applied load.

Upon completion of the drying phase, the tests proceed tothe wetting phase. There is a marked hysteresis between the

drying and wetting curves for all soil specimens, asexpected. The size of the hysteresis loops seems to have re-duced with the increase in applied stress, for the range of netnormal stresses considered. This is consistent with the re-sults shown in Fig. 3 that the denser the specimen, thesmaller the size of the hysteresis loop. At the end of thetests, the end points of all three wetting curves are lowerthan their corresponding starting points, which is likelycaused by trapped air. Similar to the results shown in Fig. 3,the denser the specimen or the higher the applied load, thecloser the end points to the starting points.

Figure 11 shows the influence of stress state on the soil-water characteristics for the natural soil specimens (CDV-N1(C), CDV-N2(M), and CDV-N3(M)). The size of thehysteresis loops does not seem to be governed by the appliedstress level. Similar to the recompacted soil specimens, anatural soil specimen, which is subjected to higher appliedstresses, possesses a slight higher air-entry value (see Table 3),

© 2000 NRC Canada

1260 Can. Geotech. J. Vol. 37, 2000

Fig. 9. Comparison of measured soil-water characteristics with and without volume change considered for the natural soil specimens.

Fig. 8. Comparison of measured soil-water characteristics with and without volume change considered for the recompacted soil specimens.

I:\cgj\Cgj37\Cgj06\T00-056.vpWednesday, December 06, 2000 8:40:23 AM

Color profile: Generic CMYK printer profileComposite Default screen

Page 10: Experimental investigations of the soil-water characteristics of a volcanic soil

and lower rates of desorption and adsorption, as a result ofsmaller pore-size distribution. However, the influence ofstress state on the soil-water characteristics of the naturalspecimens appears to be more significant than that on thesoil-water characteristics of the recompacted specimens.This may be attributed to the fact that the recompacted spec-imens, which were recompacted at the wet side of the opti-mum (30.3%), would have a relatively uniform pore-sizedistribution, whereas the structure of the natural residual soilgenerally involves a wide range of pore sizes (Bell 1992).The natural specimens, therefore, can be reasonably postu-lated to have relatively nonuniform pore-size distributions.As a load is applied, relatively large pores in the naturalspecimens may be reduced in size significantly and the spec-

imen would probably become more homogenous. On thecontrary, the pore-size distributions of the recompactedspecimens may not be affected substantially compared withthose in the natural specimens. Hence, the influence of stressstate on soil-water characteristics appears to be more signifi-cant in natural soil specimens than in recompacted soil spec-imens.

Comparison between natural and recompacted soilsThe soil-water characteristics of a recompacted (CDV-

R2(M)) and a natural (CDV-N2(M)) specimen are comparedin Fig. 12. Specimen CDV-R2(M) was recompacted to thesame density at the same initial moisture contents as thenatural specimen. The hysteresis loop in the recompacted

© 2000 NRC Canada

Ng and Pang 1261

Fig. 11. Influence of stress state on soil-water characteristics of natural CDV specimens.

Fig. 10. Influence of stress state on soil-water characteristics of recompacted CDV specimens.

I:\cgj\Cgj37\Cgj06\T00-056.vpWednesday, December 06, 2000 8:40:31 AM

Color profile: Generic CMYK printer profileComposite Default screen

Page 11: Experimental investigations of the soil-water characteristics of a volcanic soil

soil is considerably larger than that in the natural specimen.As discussed earlier, soil recompacted at wet of optimum(i.e., CDV-R2(M)) is generally believed to be more homoge-nous, whereas the natural soil specimen (CDV-N2(M)) hasrelatively nonuniform pore-size distributions due to variousgeological processes such as leaching in the field. As thetwo specimens have the same density, it is reasonable topostulate that CDV-N2(M) would have larger pores thanthose in CDV-R2(M), statistically. Thus, CDV-N2(M) has aslightly lower air-entry value (see Table 3) and a higher rateof desorption than CDV-R2(M)) for suctions up to 50 kPa.The rates of desorption of the two soil specimens appear tobe the same for high suctions.

On the other hand, the rates of adsorption for the twospecimens are considerably different. The rate of wettingcurve obtained from the natural soil specimen is substan-tially higher than that of the recompacted specimen. This ob-served behaviour may be explained by the capillary-tubeanalogy discussed previously (Fig. 6). The difference inpore-size distributions between CDV-N2(M) and CDV-R2(M)is illustrated in Fig. 6. As the natural soil specimen has anonuniform pore-size distribution in which some relativelylarge pores (r1) are together with some relatively small pores(r3), the presence of these small pores would facilitate theingress of water to the specimen as the soil suction reduces.On the contrary, the lack of small pores and the presence ofthe relatively uniform, medium-sized pores (r2) in therecompacted specimen would slow the rate of water enteringthe soil specimen. The rates of adsorption of the two soilspecimens appear to be the same for small suctions (lessthan 5 kPa).

Summary and conclusions

To gain a fundamental understanding of the desorptionand adsorption characteristics of a volcanic soil in HongKong, an experimental program was carried out to test bothrecompacted and natural soil specimens. Various factors in-

fluencing soil-water characteristics were considered in theexperimental program, including initial dry density, initialwater content, soil structure, history of drying and wetting,and stress state. Based on the experimental investigationsundertaken, the following conclusions are drawn:

(1) There is a marked hysteresis between the drying andwetting curves for all soil specimens recompacted to threedifferent densities at the same water content. The size of thehysteresis loops decreases as the initial density of the speci-men increases. This is probably due to the difference in theinterconnection voids among the soil specimens. Althoughno apparent difference in the air-entry value of the speci-mens is observed (due to the small increase in density), therate of desorption appears to be governed by the soil density.The soil specimens with a lower density exhibit a higher rateof desorption.

(2) For soil specimens recompacted to the same densityat optimum, dry of optimum, and wet of optimum watercontents, the soil specimen recompacted at wet of opti-mum, which has a relatively uniform pore-size distributionresulting in more air entrapment, has the largest hysteresisloop among the three samples. The soil specimenrecompacted at dry of optimum, which has a relatively non-uniform pore-size distribution, has the smallest hysteresisloop. The hysteresis loop for the soil specimenrecompacted at optimum lies between those for the othertwo specimens. Although the air-entry value is somewhat afunction of the initial moisture content, the rates ofdesorption and adsorption do not seem to be significantlyaffected by the initial water contents for the ranges of suc-tion and water content considered.

(3) Soil-water characteristics of the recompacted soil de-pend on the history of drying and wetting. There is a distinctdifference in both desorption and adsorption characteristicsbetween the first and the second drying and wetting cycles.The rates of desorption and adsorption are substantiallyhigher at the first than at the second drying and wetting

© 2000 NRC Canada

1262 Can. Geotech. J. Vol. 37, 2000

Fig. 12. Influence of soil structure on soil-water characteristics.

I:\cgj\Cgj37\Cgj06\T00-056.vpWednesday, December 06, 2000 8:40:35 AM

Color profile: Generic CMYK printer profileComposite Default screen

Page 12: Experimental investigations of the soil-water characteristics of a volcanic soil

cycle. This may be caused by changes in the soil structureduring the first drying and wetting cycle. However, the soil-water characteristics of the subsequent drying and wettingcycles appear to remain approximately the same.

(4) The soil-water characteristics of a natural specimenare very different from those of the same soil recompactedto the same density at the same water content. The hysteresisloop for the natural specimen is considerably smaller thanthat for the recompacted soil. Also, the natural specimen hasa slightly lower air-entry value and a higher rate ofdesorption than the recompacted specimen for suctions up to50 kPa. The rates of desorption of the two soil specimensappear to be the same for high suctions. On the other hand,the wetting curve for the natural soil specimen is substan-tially higher than that for the recompacted specimen. Thisobserved behavior can be explained by the capillary-tubeanalogy.

(5) For recompacted specimens subjected to differentstress states under various applied loads, the higher the ap-plied load on the specimen, the lower the rate of desorption.This is likely caused by the presence of an average smallerpore-size distribution in the soil specimen under the higherapplied load. As expected, there is marked hysteresis be-tween the drying and wetting curves for all soil specimensunder various applied loads. The hysteresis loops seem tohave reduced in size with an increase in the applied load.

(6) The size of the hysteresis loops does not seem to beaffected by the net normal stresses applied to natural speci-mens. Similar to the recompacted soil specimens, the naturalsoil specimen subjected to a higher applied load possesses aslight larger air-entry value and lower rates of desorptionand adsorption because of the presence of an averagesmaller pore-size distribution.

(7) Traditionally, the soil-water characteristic curve hasbeen interpreted by neglecting any volume change through-out the test. By using the newly modified volumetric pres-sure-plate extractor, the traditional method slightlyunderpredicts the volumetric water content or degree of satu-ration in soil specimens for suctions higher than 100 kPa.

Acknowledgements

The authors gratefully acknowledge financial supportfrom the two research grants HKUST6053/97E andHKUST6046/98E provided by the Research Grant Councilof the Hong Kong Government of the Special AdministrativeRegion (HKSAR).

References

Barbour, S.L. 1998. Nineteenth Canadian Geotechnical Collo-quium: The soil-water characteristic curve: a historical perspec-tive. Canadian Geotechnical Journal,35: 873–894.

Bell, F.G. 1992. Engineering properties of soils and rocks.Butterworth-Heinemann Ltd., Woburn, Ma.

Benson, C.H., and Daniel, D.E. 1990. Influence of clods on hy-draulic conductivity of compacted clay. Journal of GeotechnicalEngineering, ASCE,116(8): 1231–1248.

Brand, E.W. 1984. Landslides in south Asia: a state-of-art report.In Proceedings of the 4th International Symposium on Land-slides, Toronto, Vol. 1, pp. 17–59.

Brooks, R.H., and Corey, A.T. 1966. Properties of porous media af-fecting fluid flow. Journal of the Irrigation and Drainage Divi-sion, ASCE,92(IR2): 61–88.

British Standards Institution. 1990. British standard methods oftest for soils for civil engineering purposes. Part 2. British Stan-dards Institution, London.

Chiu, C.F., Ng, C.W.W., and Shen, C.K. 1998. Collapse behaviourof a loosely compacted soil in Hong Kong.In Proceedings ofthe 2nd International Conference on Unsaturated Soils, Beijing,Vol. 1, pp. 25–30.

Croney, D., and Coleman, J.D. 1954. Soil structure in relation tosoil suction (pF). Journal Soil Science,5(1): 75–84.

Fourie, A.B. 1996. Predicting rainfall-induced slope instability.Proceedings of the Institution of Civil Engineers, GeotechnicalEngineering,119: 211–218.

Fredlund, D.G. 1998. Bringing unsaturated soil mechanics into en-gineering practice.In Proceedings of the 2nd International Con-ference on Unsaturated Soils, Beijing, Vol. 2, pp. 1–36.

Fredlund, D.G., and Rahardjo, H. 1993. Soil mechanics for unsatu-rated soils. John Wiley & Sons, Inc., New York.

Fredlund, D.G., Xing, A., and Huang, S. 1994. Predicting the per-meability function for unsaturated soils using the soil-watercharacteristic curve. Canadian Geotechnical Journal,31: 533–546.

Fukuoka, M. 1980 Landslides associated with rainfall.Geotechnical Engineering, Journal of Southeast Asia Society ofSoil Engineering,11: 1–29.

Geotechnical Control Office. 1988. Guide to rock and soil descrip-tions (GEOGUIDE 3). Geotechnical Control Office, PublicWorks Department of Hong Kong, Hong Kong.

Geotechnical Engineering Office. 2000. Report on the Shek KipMei Landslide of 25 August 1999. Geotechnical EngineeringOffice, Civil Engineering Department of the Hong Kong SpecialAdministrative Region, Hong Kong.

Hillel, D. 1982. Introduction to soil physics. Academic Press, NewYork.

Hillel, D. 1998. Environmental soil physics. Academic Press, SanDiego, Calif.

Lambe, T.W. 1958. The structure of compacted clay. Journal of theSoil Mechanics and Foundation Engineering Division, ASCE,84(2): 1–34.

Lim, T.T., Rahardjo, H., Chang, M.F., and Fredlund, D.G. 1996.Effect of rainfall on matrix suctions in a residual soil slope. Ca-nadian Geotechnical Journal,33: 618–628.

Lin, L., and Benson, C.H. 2000. Effect of wet–dry cycling onswelling and hydraulic conductivity of GCLs. Journal ofGeotechnical and Geoenvironmental Engineering, ASCE,126(1): 40–49.

Lumb, P.B. 1962. Effects of rain storms on slope stability.In Pro-ceedings of a Symposium on Hong Kong Soils, Hong Kong,pp. 73–87.

Lumb, P.B. 1975. Slope failures in Hong Kong. Quarterly Journalof Engineering Geology,8: 31–65.

Maulem, Y. 1986. Hydraulic conductivity of unsaturated soils: pre-diction and formulas.In Method of soils analysis. Part 1. Physi-cal and mineralogical methods.Edited byA. Klute. AmericanSociety of Agronomy, Madison, Wis., pp. 799–823.

Ng, C.W.W., and Pang, Y.W. 1998. Ling Ping Road landslide in-vestigation — specialist testing of unsaturated soils. Report forthe Geotechnical Engineering Office of the Hong Kong SpecialAdministrative Region, Hong Kong.

Ng, C.W.W., and Pang, Y.W. 2000. Influence of stress state on soil-water characteristics and slope stability. Journal of Geotechnicaland Geoenvironmental Engineering, ASCE,126(2): 157–166.

© 2000 NRC Canada

Ng and Pang 1263

I:\cgj\Cgj37\Cgj06\T00-056.vpWednesday, December 06, 2000 8:40:35 AM

Color profile: Generic CMYK printer profileComposite Default screen

Page 13: Experimental investigations of the soil-water characteristics of a volcanic soil

© 2000 NRC Canada

1264 Can. Geotech. J. Vol. 37, 2000

Ng, C.W.W., and Shi, Q. 1998. A numerical investigation of thestability of unsaturated soil slopes subjected to transient seep-age. Computers and Geotechnics,22: 1–28.

Ng, C.W.W., Chui, C.F., and Shen, C.K. 1998. Effects of wettinghistory on the volumetric deformations of an unsaturated loosefill. In Proceedings of the 13th Southeast Asian GeotechnicalConference, Taipei, Taiwan, Vol. 1, pp. 141–146.

Rahardjo, H., and Leong, E.C. 1997. Soil-water characteristiccurves and flux boundary problems.In Unsaturated soil engi-neering practice.Edited by S.L. Houston and D.G. Fredlund.American Society of Civil Engineers, New York, pp. 88–112.

Sun, H.W., and Campbell, S.D.G. 1998. The Lai Ping Road land-slide of 2 July 1997. Landslide Study Report (LSR 27/98),Geotechnical Engineering Office, Civil Engineering Departmentof the Hong Kong Special Administrative Region, Hong Kong.

Tinjum, J.M., Benson, C.H., and Blotz, L.R. 1997. Soil-water char-acteristic curves for compacted clays. Journal of Geotechnicaland Geoenvironmental Engineering, ASCE,123(11): 1060–1069.

Topp, G.C., and Miller, E.E. 1966. Hysteresis moisture characteris-tics and hydraulic conductivities for glass-bead media. Soil Sci-ence Society of America, Proceedings,30: 156–162.

Tung, Y.K., Ng, C.W.W., and Liu, J.K. 1998. Lai Ping Road land-slide investigation — three dimensional groundwater flow com-putation. Report for the Geotechnical Engineering Office, CivilEngineering Department of the Hong Kong Special Administra-tive Region, Hong Kong.

Vanapalli, S.K., Fredlund, D.G., Pufahl, D.E., and Clifton, A.W.1996. Model for the prediction of shear strength with respect tosoil suction. Canadian Geotechnical Journal,33: 379–392.

Vanapalli, S.K., Pufahl, D.E., and Fredlund, D.G. 1998. The effectof stress state on the soil-water characteristic behavior of a com-pacted sandy-clay till.In Proceedings of the 51st CanadianGeotechnical Conference, Edmonton, pp. 81–86.

Vanapalli, S.K., Pufahl, D.E., and Fredlund, D.G. 1999. The effectof soil structure and stress history on the soil-water characteris-tics of a compacted till. Géotechnique,49(2): 143–159.

van Genuchten, M.T. 1980. A closed-form equation for predictingthe hydraulic conductivity of unsaturated soils. Soil Science So-ciety of America Journal,44: 892–898.

Wolle, C.M., and Hachich, W. 1989. Rain-induced landslides insouth-eastern Brazil.In Proceedings of the 12th InternationalConference on Soil Mechanics and Foundation Engineering, Riode Janeiro, Vol. 3, pp. 1639–1644.

I:\cgj\Cgj37\Cgj06\T00-056.vpWednesday, December 06, 2000 8:40:36 AM

Color profile: Generic CMYK printer profileComposite Default screen