Effect of suction change on water content and total volume of an expansive clay

Zhan et al. 1J Zhejiang Univ SciA 2007 8(5):699-706699Journal of Zhejiang Unlversity SCIENCE AISSN 1673-565X (Prit; ISSN 1862- 1775 (Online)www.zju. educnjzus; www. springerink.comE-mail: jzus@zju.edu.cngZUSEffect of suction change on water content and total volumeof an expansive clay'ZHAN Liang-tongt", CHEN Ping', NG C.w.W.3('Department of Civil Engineering, Zhejiang University, Hangzhou 310027, China)(Department of Civil Engineering, Zhejiang Sci-Tech University, Hangzhou 310018, China)(Department ofCivil Engineering. Hong Kong University of Science and Technology, Hong Kong, China)E-mail: zhanlt@zju.edu.cn; chen_ 1230@yahoo.com.cnReceived Oct. 10, 2006; revision acepted Feb. 20, 2007Abstract: A laboratory study was carried out on both natural and compacted specimens to investigate the complex soil-waterinteraction in an unsaturated expansive clay. The laboratory study includes the measurement of soil-water characteristic curves, 1Dfree swelling tests, measurement of swelling pressure and shrinkage tests. The test results revealed that the air-entry value of thenatural specimen was quite low due to cracks and fissures present. The hydraulic hysteresis of the natural specimen was relativelyinsignificant as compared with the compacted specimen. Within a suction range 0 to 500 kPa, a bilinear relationship between freeswelling strain (or swelling pressure) and initial soil suction was observed for both the natural and compacted specimens. As aresult of over-consolidation and secondary structures such as cementation and cracks, the natural specimens exhibited significantlower swelling (or swelling pressure) than the compacted specimen. The change of matric suction exerts a more significant effecton the water phase than on the soil skeleton for this expansive clay.Key words: Expansive soil, Water content, Suction, Swelling, Shrinkagedoi:10.163 1/jzus.2007.A0699Document code: ACLC number: TU4INTRODUCTIONstrength and soil swelling (or a development ofswelling pressure under a confined condition). FieldExpansive clay is widely distributed in the world,studies show that the soil-water interaction inducedand often causes damages to light buildings, pave-by wetting-drying cycles is very complex, and in-ments, and slopes, which has been reported in manyvolves the coupled effects among the changes in wa-countries around the world (Nelson and Miller, 1992;ter content, suction, stress, deformation and shearLiu, 1997). The damage of expansive soils is closelystrength (Ng et al, 2003; Zhan, 2003). For example,related to the strong soil-water interaction in thethe wetting- induced swelling of an expansive clay is ashallow soil layers subject to seasonal wetting-drying function of initial suction (or initial water content),cycles. During dry seasons, evapotranspiration causes initial dry density and confining stress, with the as-loss of water content in the shallow soil layer, andsociated swelling pressure also depending on thehence the expansive soil shrinks and cracks. Duringstress-path employed in the tests (Brackley, 1975).wet seasons, rainfall infiltration results in an increaseA laboratory study was carried out on bothin water content and a decrease in soil suction in thenatural and compacted specimens to improve ourshallow soil layer, which leads to a reduction of shearunde中国煤化工-water interaction inanC N M H Ghe laboratory studyincluaes tne Tmeasurement 0I soll-water characteristic↓Corresponding author"Project No. 50408023) supported by National Natural Sciencecurves, free swelling tests, measurement of swellingFoundation of Chinapressure, and shrinkage tests. On the basis of the test00Zhan et al. 1J Zhejing Univ SciA 2007 8(5):699-706results, the relationships among matric suction, wateras 18.5%, with the maximum compaction pressurecontent and void ratio for the expansive soil werebeing 800 kPa. The dry density of the obtainedobtained, and the difference in the soil-water charac-specimens was equal to the average value of theteristic between natural and compacted specimensnatural specimens (i.e, 1.56 Mg/m ). The compactionwas identified.water content (i.e, 18.5%) was on the dry side of theoptimum water content. The optimum water contentas determined by the proctor compaction test wasMATERIALS AND SPECIMEN PREPARATION20.5%, corresponding to a maximum dry density ofMETHOD1.66 Mg/m'. After completion of preparation, theinitial suction of the natural specimens was measuredThe soil used in this laboratory study was aby a tensiometer, with the values ranging from 20 tobrownish-yellow expansive clay taken from the re-30 kPa. The initial suction of the compacted specimensearch slope in Zaoyang, Hubei of China (Ng et al,was measured by a high suction probe, with the2003). The basic physical properties of the soil aremeasured value being about 540 kPa (Zhan, 2003). Itpresented in Table 1. In accordance with the USCS was obvious that the difference in the initial suctionclassification system, the soil is clasified as a siltybetween the natural and compacted specimens is at-clay. With respect to expansion potential, the soil cantributable to the difference in their initial water con-be classified as a medium expansive soil in accor-tent.dance with the criterion proposed by Sridharan andFig.1 shows a comparison of the appearance ofPrakash (2000).natural and compacted specimens. The two specimensThe procedure for preparing natural specimenswere both in an air-drying state. Open cracks andincludes cutting appropriate size of soil blocks from fissures were well developed in the natural specimen,the block samples and then trimming them into di-whereas no obvious cracks appeared on the com-mensions ftting an oedometer ring (i.e, 70 mm in pacted specimen. In addition, there appeared somediameter and 19 mm in height). The compactedblack mottles on the natural specimen, which werespecimens were prepared with static compactionidentified as iron and manganese oxides (approxi-method. The compaction water content was adoptedmately 2.5%). It is generally believed that the blackTable 1 Soil properties obtained from the soil samplesSoil propertiesValuesRemarksClay mineralPercentage of ilit, montmorillonite, kaolinite16%, 21%, 4%ClassificationPercentage of sand, sit, clay3%, 58%, 39%USCS casificationSity clayConsistencyLiquid limit (LL)50.5limitPlasticity index (PI)31Linear shrinkage limit Ws12%DensitySpecific gravity G,2.67Dry density (Mg/m)1.56Flow propertiesSaturated permeability ks (m/s)2.7x10-10Oedometer testAir-entry valuc (kPa)25Copresibility'Compressibility index C。0.18Swelling index C,0.038Coefficient of consolidation C, (m^/s)1.6x10^From 100 to 200 kPaSwelling andOne-dimension fee swelling中国煤化工-air-dry to saturatedshrinkageSwelling pressure (kPa)air-dry to saturatedVolumetric strain due to shrinkageYHC N M H Gaturated to oven-dryShear strengthEffective cohesion c' (kPa)11.7parametersEffective angle of friction φ'24°denotes measured from compacted specimensZhan et al. 1J Zheiang Univ SciA 2007 8(5)699-706701oxides have a cementation effect between soil parti-both ends to prevent soil swelling upon wetting so ascles. However, the secondary structure was com- to maintain the desired dry density. After saturationpletely lost on the compacted specimen as a result ofthe constraint at both ends was removed and thegrinding and remoulding process.specimens together with the steel rings were quicklymoved to pressure plate apparatus for SWCC tests.After the removal of constraint, an obvious swellingwas observed in the compacted specimen, but theswelling of the natural specimen was negligible. Thesignificant swelling of the compacted specimen wasmainly attributed to its larger initial suction as com-pared with the natural specimen. The increase inheight due to the swelling was approximately 0.8 mma)(b)for the compacted specimen with a height of 19 mm,and it is equivalent to a volumetric strain of about 4%.Fig.1 Comparison of appearance between (a) naturalTherefore, the dry density of the compacted specimenand (b) compacted specimens at air drying conditionafter saturation was less than that of the natural(Zhan and Ng, 2006)specimen.TESTING PROGRAM， APPARATUS ANDPressure plate testsPROCEDURETwo conventional apparatuses, i.e, 2-bar volu-metric pressure plate extractor and 5-bar pressureThe laboratory investigation was carried out onplate extractor, were used to measure soil-wateroth the natural and compacted specimens. The in-characteristic curves (SWCCs) for both natural andvestigation on the compacted specimen was to pro-compacted specimens. The former apparatus wasvide a reference because it basically represents theused to measure SWCC along both desorption andparticle-level properties of the soil material. The fol-adsorption paths up to a suction of 200 kPa, and thelowing testing procedures were conducted to inves- ltter one was used for measuring desorption curve uptigate the soil-water characteristics of the expansiveto a higher suction (i.e., 500 kPa). For the test con-soil: Both the natural and compacted specimens wereducted with 5-bar pressure plate extractor, 8 duplicatefirst saturated at constant volume condition. Then, allspecimens were used to obtain a complete curve. Thethe specimens were moved into pressure plate ex-test procedures suggested by ASTM (2000) weretractors to measure soil-water characteristic curvesfollowed.(SWCCs). When each of the applied suction wasequalized, one or two specimens were taken out fromFree swelling tests and measurement of swellingthe extractor. The specimens with known values ofpressuresuction were then used to conduct ID free swellingThe specimens taken out of the pressure plateests, at the end of which each specimen wasextractor after the SWCC tests and four air-dryingone-dimensionally compressed to its initial heightspecimens, which possessed different values of initial(i.e., the height before free swelling tests), and hencesuction, were used for the free swelling tests. The testthe swelling pressures for different specimens corre-procedure included: (1) the unsaturated specimensponding to different suctions were measured. Inhoused in a steel ring was firstly wetted under a smalladdition, shrinkage tests were carried out on both thevertical loading (e.g., 2 kPa), and the vertical swellingcompacted and natural specimen to investigate thewas monitored until it approached a steady value,volume change behavior of the expansive clay uponwhich was recorded as the final swelling for thedrying.spec中国煤化工men was then loadedby (YHC N MH Gent rate (i.e., 0.01Saturation processmm/min) unt1 the iniual neignt of the specimen priorDuring the saturation process, the soil specimensto wetting was obtained. The pressure required tohoused in steel rings were completely constrained on recover the initial height was recorded as swelling702Zhan et al.1IJ Zhejiang Univ SciA 2007 8(5):699-706pressure for the specimen. It should be noted that theThe solid lines in Fig.2 show the SWCCs meas-swelling pressure measured by the above method is ured for the compacted specimens. It is noticed fromusually larger than that from constant-volume method both the two desorption curves that water content(Gens and Alonso, 1992).does not decrease as usual but increases slightly at thebeginning. It may be atributed to the gradual swellingShrinkage upon dryingof the compacted specimens as a result of the removalShrinkage tests on both the compacted andof constraint at both ends. The slight difference be-natural specimen were also carried out to investigatetween the two desorption curves measured by the twothe shrinkage characteristic of the expansive clayapparatuses could be due to the slight difference inupon drying. The shape and dimension of the speci- their initial dry density. The slope of the desorptionmens are identical to the specimen for an oedometer curves starts to increase significantly at a suction intest. The compacted and natural specimens were excess of 25 kPa, which can be regarded as theprepared in the same way as that for the measurement air-entry value for the compacted specimens. Theof SWCC. After saturation, each specimen was sub- hysteresis between the drying and wetting curvesjected to drying in a temperature and humidity con-appears to be more significant as compared with thetrolled room. When the mass of each specimen kept natural specimen. As before, the final water content atunchanged in the temperature and humidity controlled the end of adsorption is slightly larger than the initialroom, further drying was made by placing the water content.specimens at a dryer air condition and finally in anoven. The dimension (height and diameter) and mass昌3of each specimen were measured periodically (e.g..3190every 1% change in water content). The height anddiameter were measured by a caliper and a PI tape,respectively. The relationships between void ratio and=三2:water content can be deduced from the measurement.EXPERIMENTAL RESULTS1000Matric suction (kPa)Soil-water characteristic curvesFig.2 shows the SWCCs in term of gravimetricFig.2 Comparison of swCC between natural andcompacted specimenswater content measured for the natural and compactedspecimens. The dash lines show SWCCs for theA comparison of SWCC between the natural andnatural specimen. With an increase in suction, thcompacted specimens indicates that there is a distinctwater content in the natural specimens starts to de- difference between the two types of specimen forcrease at a quite low suction (less than 1 kPa). It seems suctions less than 200 kPa, whereas the two desorp-that the air-entry value of the natural specimen is quite tion curves tend to merge together when suction ex-low due to cracks and fissures present. The hysteresisceeds 200 kPa. Just prior to the desorption tests, thebetween the adsorption and desorption curves is rela- initial saturated water content of the compactedtively insignificant. It should be noted that the final specimens is significantly larger than that of naturalwater content at the end of the adsorption curve isspecimens, which implies the initial dry density of theunusually larger than the start point of the desorption compacted specimens is less than that of naturalcurve (ie, at saturated condition). This finding isspecimens. This is consistent with the significantconsistent with the results obtained by Wang (2000) swelli中国煤化工mens after the re-for a similar expansive soil. Wang (2000) demon-movale in the initial drystrated the abnormal result was due to the structure densitMYHc N M H Gnce in water reten-change that occurred in the specimens during dry- tion characteristic between the two types of speci-ing/wetting cycle on the basis of microscopic analysis. mens, particularly at low suction range (Vanapalli etZhan et al. 1J Zhjiang Univ SciA 2007 8(5):699-706703al, 1999; Ng and Pang, 2000). The compacted Relationship between free swelling strain and ini-specimens retain more water at a given suction but tial suctionexhibits a significantly greater desorption rate onceFig.4 shows the variations of ID free swellingmatric suction exceeds its air-entry value. Therefore, with the initial suction obtained for the natural andit is postulated that the loose compacted specimens compacted specimens. In Fig.4a, the data points cor-posses a relatively uniform fabric and much more responding to a suction of 100000 kPa were obtainedpores of intermediate sizes (corresponding to the from air-drying specimens. The suction value wassuction range from 10 to 200 kPa) than the naturalestimated from the water content of the air-dryingspecimens. The postulation can also help to explain specimens and the extended SwCC (Zhan, 2003). Aswhy the compacted specimens exhibit a more sig- expected, the measured swelling strain increases withnificant hysteresis than the natural specimens. The the value of initial suction for both natural and com-cracks and fissures present in the natural specimenspacted specimens. For a given initial suction, thare likely to explain why the desorption rate of the swelling of the natural specimen is always less thannatural specimen at suctions less than 10 kPa is that of the compacted specimen, regardless of largerslightly larger than that of the compacted specimen.initial dry density of the natural specimen. TheFig.3 shows two SWCCs in terms of degree of maximum values of ID free swelling measured for thesaturation for the compacted and natural specimens, natural and compacted specimens are approximatelywhich correspond to the two desorption curves 8% and 16%， respectively, corresponding to anmeasured by 5-bar pressure plate extractor shown inair-drying initial state. The less swelling of the naturalFig.2. The degree of saturation was deduced from the specimens is likely attributable to its highervolume measurement for each specimen taken out over-consolidation ratio (OCR) than the compactedfrom the 5-bar pressure plate extractor after equilib- specimens. Oedometer tests indicated that therium at each suction. It should be noted here that 8 pre-consolidation pressure of the natural specimensduplicated specimens were used for the measurement (230 kPa) was significantly greater than that of theof each of the SWCCs. It can be seen that the com- compacted specimens (140 kPa). Another reason maypacted specimens kept saturated until the air-entrybe the cementation effect of the iron and manganesevalue (25 kPa) was reached, and then de-saturated at aoxides present in the natural specimens. The cemen-rapid rate.The natural specimens started ttation tends to bond clay particles, reduce particlesde-saturate at quite low suction, but the de-saturation surface for accessing water, and hence reduce therate was significantly lower than the compacted swelling tendency (Hillel, 1998). It should be notedspecimens. Therefore, the two SWCCs in terms of that the cementation may contribute to thedegree of saturation intersect at a suction of about 60 over-consolidation nature of the natural specimens.kPa, and the difference between the two types ofTo enlarge the left-lower part of Fig.4a, the dataspecimen becomes significant even at a suction in points corresponding to initial suctions not more thanexcess of 200 kPa.500 kPa were re plotted in Fig.4b together with thevalue of initial void ratio for each specimen. It can be100seen that the inconsistency between the two specimenswith the same initial suction is likely attributable to thedifference in the initial void ratio. Within the suctionrange considered, the relationship between swelling.strain and initial suction appears to be bilinear on a台8(linear scale. The inflection points for the compacted75 H + HNaturaland natural specimens correspond to an initial suctionof 200 and 100 LPa. recnntivalvPrior to the inflec-7001000tion P中国煤化工ye), the variation ofMatric suction (kPa)swellYHCN M H Gction is more sig-Fig.3 SWCCs in terms of degree of saturation fornificant than that at relatively high suction range. Thenatural and compacted specimensobserved behavior is consistent with the Suc-704Zhan et al. 1J Zhejiang Univ SciA 2007 8(5):699-706tion-Decrease yielding locus in the elasto-plasticand fissures). The open cracks and fissures provided amodel proposed by Gens and Alonso (1992) for un-certain space allowing for the expansion of the soilsaturated expansive soils.matrix, resulting in a less inherent constraint. As be-fore, the data points for initial suctions not more than18Number denots void ratio8:35年500 kPa were re-plotted in Fig.5b. It can be seen thatthe bilinear characteristic discussed above is kept forE 12the relationship between swelling pressure and initial0.5704suction. It is noticed that the values of swelling0.s40↑pressure for the two natural specimens with an initial6suction less than 200 kPa appear to be larger than the■Compactedcorresponding values for the compacted specimens....▲Nrtu.....The larger swelling pressure of natural specimen010100100010000100000may be attributed to their significantly lower initialInitial matric suction (kPa)void ratios as compared with the correspondingcompacted specimens.Number denotes void ratio0.6870000.6820.52510.709中0.697 .800 F0.71940.628600↑0.6687450.6510.5400.6350.636名400F” 0.570+日0.767 0.59. Compacted0.630▲Natural司200-恋100 200300400 500 60000(b)(a)Fig.4 Variation of 1D free swelling with initial suctionfor natural and compacted specimens. (a) Logarithmic250 r40.687scale; (b) Linear scale200 F0.697Relationship between swelling pressure and initial50 F0.709平40.719suction40.635After full swelling, swelling pressure for each100 F).630 70.7450.659tspecimen was measured by loading the specimen to50 Hits initial height. Fig.5a shows the variations of10.767swelling pressure with the initial suction for both the010200300400500natural and compacted specimens. The swellingcharacteristic reflected by the relationships betweenswelling pressure and initial suction in Fig.5a is ba-Fig.5 Variation of swelling pressure with initial suctionsically consistent with that reflected by the relation-ships between swelling strain and initial suctionshown in Fig.4a. The maximum values of swellingpressure measured for the compacted and natural Shrinkage upon dryingspecimens are about 800 kPa and 400 kPa respec-Pig 6 shows the shrinkaoe curves for the com-tively, corresponding to an air-dry initial state. Apartpactg中国煤化工:ach two). The twofrom the two reasons previously explained for the less shrin;YHC N M H Glicated compactedswelling of the natural specimens, the less swellingspecimens (A and B) are reasonably consistent withpressure observed for the natural specimen was alsoeach other. However, an obvious divergency waslikely related to its secondary structure (i.e, cracksobserved for the other two curves for the naturalZhan et al. 1J Zhejiang Univ SciA 2007 8(5)699-706specimens (C and D) at water content less than 20%.compacted and natural specimens, respectively. TheThe observed divergency is attributed to the differ- data indicate that the magnitude of shrinkage upon theence in the quantity and opening of cracks developingdrying from a saturated state to an air-dry state isin the two natural specimens during drying. It shouldclose to the magnitude of 1D free swelling upon thebe noted here that the void ratio was deduced from thereverse wetting path for both the compacted andmeasurement of external dimension (height and di-natural specimens. However, it should be noted thatameter), which included the spaces of the cracks. Itthe stress state during the wetting path (Ko condition)was found that there are more and wider open cracksis different from that during the drying path (no stressdeveloped in specimen C as compared with specimenapplied).D. Thus, the more cracks developed in the specimen,the larger the deduced void ratio even if the shrinkageRoles of suction on the soil skeleton and the waterof intact soil mass is identical. Shrinkage limit (w)phasecan be identified from the shrinkage curves accordingThe relationship between void ratio and suctionto its definition (ie, the water content at which dry-can be approximately deduced from the combinationing-shrinkage ceases). The values of shrinkage limitof the shrinkage curve and SWCC measured. Theidentified for the compacted and natural specimensobtained relationships for the compacted and naturalare 10.8% and 13.0%, respectively.specimens are shown in Fig.7, together with SwCC interm of“water ratio" (volume of water to volume of0.85psolids, wG) (Romero and Vaunat, 2000). It should benoted here that a change in void ratio (e) representsw, (compacte)-10.8%he change in soil skeleton, and that a change irw, (atural)=13.0%“water ratio" (wG) reflects the change in water phase.0.65A comparison between the e~s and wGg~s curves. Recompicted (A)indicates the water phase is more susceptible to a0.55ORecompacted (B)士Natural (C)0.9女Natual D0.4550152025303:Gravimetric water content w (%)0.8Fig.6 Shrinkage curves for natural and compacted speci-mensE 0.2As compared with the natural specimen, th甘e-s (compacted)compacted specimens possess a lower initial dry0.6↑母wG,s (compacted)士e-s (natura)density and hence a low resistance to the dry-古wG-s (natural)ness-induced shrinkage. Hence, the gradient of the0-5.1)01000Matric suction (kPa)shrinkage curve for the compacted specimen is gen-erally larger than that for the natural specimen. ItFig.7 Comparison between the effects of suctionshould be noted that the fabric of compacted speci-change on soil skeleton and water phasemens is relatively uniform so no obvious crackingwas observed throughout the drying process (seeFig.1). The final void ratio of natural specimen Dunexpectedly coincides with that of the compactedspecimens. The total volumetric strain due to the芭0shrinkage from an initial saturated state to a com-pletely dry state can be calculated from the shrinkage.6-一curves, with the values for the compacted and natural中国煤化工specimens are 14.8%~15.1% and 8.6%~11.4%, re-YHCNMHG 100spectively. As discussed before, the values of the IDsucton s ur cliccuve suress p' (kPa)free swelling strain measured from air drying speci-Fig.8 Comparison between the relationships of voidmens are 16.1%~16.7% and 7.8%~8.9% for theratio to changes in suction and effective stress06Zhan et al.1J Zhejiang Univ SciA 2007 8(5):699-706change of matric suction than the soil skeleton for(4) The change of matric suction exerts a moreboth the compacted and natural specimens. In other significant effect on the water phase than on the soilwords, the change of matric suction exerts a moreskeleton for this expansive soil. The soil skeleton issignificant effect on the water phase than on the soilmore susceptible to an increase in extermal stressesskeleton for this expansive soil. In order to assess thethan an increase in matric suctions.effect of suction change on the soil skeleton, the re~lationship between void ratio (e) and suction for theReferencescompacted specimen was plotted in Fig.8 togetherASTM, 2000. Annual Book of ASTM Standards. Americanwith the isotropic compression curve for the saturatedSociety for Testing and Materials. West Conshohocken,PA.compacted specimen. It can be seen that the responseBrackley, 1.J.A., 1975. Swell Under Load. Proceeding of 6thof soil skeleton to a change in suction tends to beRegion Conference for Africa SMFE, 1:65-70.similar to the pre-yield response of soil to a change inGens, A., Alonso, E.E, 1992. A framework for the behavior ofcompression stress (i.e., elastic response). However,unsaturated expansive clays. Canadian Geotechricalonce yielding is reached, the soil skeleton is muchJournal, 29:1013-1032.more susceptible to an increase in external stress thanHillel, D., 1998. Environmental Soil Physics. Academic Press,San Diego, CA, USA.an increase in matric suction. In other words, theLiu, T.H, 1997. Problems of Expansive Soils in Engineeringstiffness of soil skeleton with respect to a change inConstruction. Architecture and Building Press of China,external stress is generally lower than that with re-Beijing (in Chinese).spect to change in matric suction.Nelson, J.D., Miller, D.J.1992. Expansive Soils- _Problemsand Practice in Foundation and Pavement Engineering.John Wiley & Sons, Inc.Ng, C.W.W, Pang, Y.W, 2000. Influence of stress states onCONCLUSIONsoil-water characteristics and slope stability. Journal ofGeotechnical and Geoenvironmental Engineering, ASCE,On the base of the laboratory test results, the126(2):157-166. [doi:10.1061/(ASCE)1090-0241(2000)following conclusions can be drawn. These experi-126:2(157)]mental findings will helpful to interpret the complexNg, C.W.W, Zhan, LT, Bao, C.G, Fredlund, D.G, Gong,B.W, 2003. Performance of an unsaturated expansivesoil-water interaction observed in unsaturated expan-soil slope subjected to artificial rainfall infiltration.sive soil foundation and slope, and improve our un-Geotechnique, 53(2):143-157.derstanding on the associated deformation and failureRomero, E., Vaunat, J, 2000. Retention Curves of Deformablemechanism.Clays. Experimental Evidence and Theoretical Ap-(1) The air-entry value of natural expansive clayproaches in Unsaturated Soils, Balkema, Rotterdam,is quite low due to cracks and fissures present, asp.91-108.compared with that of the compacted specimens (.e, Sridharan, A., Prakash, K, 2000 Casfcation pocdures forexpansive soils. Proc. Instn. Civ. Engrs. Geotech. En-25 kPa). The hydraulic hysteresis of the natural ex-geng, 143:235-240.pansive clay is relatively insignificant.(2) The 1D free swelling strain (or swellingVanapalli, S.K, Fredlund, D.G, Pufahl, D.E, 1999. The in-fluence of soil structure and stress history on thepressure) increased with the value of initial suctionsoil-water characteristic of a compacted tll. Geotech-for both natural and compacted specimens. For aWang, B.. 2000. Stress Effets of Soil Water Characteristicsnique, 49(2):143-159.given initial suction, the swelling of the naturalCurve on Slope Stability in Expansive Soils. Mphilspecimen was always less than that of the compactedThesis, the Hong Kong University of Science . andspecimen. The less swelling potential of the naturalTechnology, Hong Kong, China.specimens may be attributable to both itsZhan, LT, 2003. Field and Laboratory Study of an Unsatu-over-consolidation nature and secondary structuresrated Expansive Soil Associated with Rain-induced(cementation and cracks).Slope Instability. Ph.D Thesis. the Hong Kong Univer-中国煤化工ng Kong, China.(3) Within a suction range from 0 to 500 kPa, aZhan,I-rength characteristicsbilinear relationship between free swelling strain (orYHCN M H Ganadion Gotechiswelling pressure) and initial soil suction was ob-cal Journal, 43(7):751-763. [doi:10. 1139/T06-036]served for both the natural and compacted specimens.