Article Soil–Water Retention Curves Derived as a Function of Soil Dry Density
Yulong Chen
Department of Hydraulic Engineering, Tsinghua University, Beijing 100084, China; [email protected]
Received: 18 June 2018; Accepted: 23 August 2018; Published: 27 August 2018
Abstract: The soil–water retention curves (SWRC) of soil plays a key role in unsaturated soil mechanics, which is a relatively new field of study having wide applications particularly in geotechnical and geo-environmental engineering. SWRCs were used to evaluate the ability of unsaturated soils to attract water with various water contents and matric suctions. Drying and wetting SWRCs for a sandy soil with different dry densities were studied in a laboratory. Proton nuclear magnetic resonance, image processing technology, and mercury intrusion porosimetry were used to characterize the microscopic mechanisms of pore size distribution in the soil. Soil–water retention in the soil samples was strongly dependent on the dry density. With zero matric suction, soil samples with a higher dry density had a lower initial volumetric water content. Volumetric water content changed at a slower rate when values of matric suction increased in soils with a higher dry density. Soil samples had residual matric suction and a larger air-entry value with a smaller slope of the SWRC when they had a higher density. Dry density change is mainly responsible for the large pores. The number of large pores decreased as dry density increased. As the dry density increased, the area of macropores occupying the largest portion decreased, while the area of mesopores and micropores increased. Minipores accounted for the smallest proportion of total area and they were nearly constant. The proportion of large diameter pores decreased relative to pores with small diameters in the tested soils. The total pore volume was lower for soil specimens that had larger dry densities, as compared to relatively loose specimens. There was hysteresis between the drying and wetting curves for all soil samples. Hysteresis decreased as the dry density of the soil increased. The different liquid–solid contact angle was the main factor causing hysteresis of SWRC.
Keywords: soil–water retention curve; dry density; unsaturated soil; nuclear magnetic resonance; image processing technique; mercury intrusion porosimetry
1. Introduction The measurement of soil parameters for unsaturated soils requires complicated laboratory tests and it is always time consuming. To avoid cumbersome laboratory work, the soil-water retention curve (SWRC) is often used to estimate various unsaturated soil property functions, such as shear strength function, permeability function, and thermal property function. At present, many methods or soil property indexes have been adopted to estimate SWRC. Geotechnical engineering problems associated with unsaturated soils can be classified into three general factors: stress, deformation, and flow. Most practical engineering solutions involve all three factors simultaneously. SWRCs involve flow phenomena, which mainly deal with capillary flow. A SWRC contains the basic information required to describe the mechanical behavior of unsaturated soil [1]. Therefore, SWRCs can play a significant role in determining the behavior of unsaturated soils. SWRCs have also been used to predict the hydraulic conductivity, shear strength, and volume changes of unsaturated soils [1–6].
GeoHazards 2020, 1, 3–19; doi:10.3390/geohazards1010002 www.mdpi.com/journal/geohazards GeoHazards 2020, 1 4
A SWRC is typically a sigmoid and it illustrates the relationship between suction and the water content of a soil. The drying and wetting SWRCs of soils are a function of stress [7,8], temperatureGeoHazards 2018[9–11, 1, ]x, FOR grain-size PEER REVIEW distribution [12], and dry density [13,14]. It is well accepted2 of 17 that the initial dry density of soil has a significant influence on the SWRC. Rajkai et al. [15] studied the A SWRC is typically a sigmoid and it illustrates the relationship between suction and the water estimationcontent of of water a soil. retention The drying characteristic and wetting fromSWRCs bulk of densitysoils are anda function particle of sizestress distribution [7,8], temperature parameters. Tarantino[9–11], [16 grain-size] proposed distribution an SWRC [12], equation and dry for density deformable [13,14]. soils It is based well accepted on an empirical that the powerinitial dry function of thedensity water of ratio.soil has Gallage a significant and influence Uchimura on [the17] SW investigatedRC. Rajkai et the al. effect[15] studied of dry the density estimation and of grain size distributionwater retention on characteristic SWRC’s parameters from bulk density of sandy and particle soils. Shengsize distribution and Zhou parameters. [18] studied Tarantino the effect of deformation[16] proposed on an the SWRC soil-water equation retention for deformable properties soils and based found on an the empirical influence power ofinitial function void of the ratio on the air-entrywater ratio. value Gallage and the and slope Uchimura of SWRC. [17] investigated the effect of dry density and grain size distributionA number ofon theoretical SWRC’s parameters or empirical of sandy models soils. have Sheng been and developed Zhou [18] to describestudied the a SWRC, effect butof the relateddeformation microscopic on the mechanisms soil-water retention underlying properties variations and found that the result influence from dryof initial density void changesratio on the remain air-entry value and the slope of SWRC. largely unknown. It is therefore useful to study ways to enhance the predictive methods of SWRCs. A number of theoretical or empirical models have been developed to describe a SWRC, but the Due torelated efficiency microscopic and non-destructiveness, mechanisms underlying the variatio nuclearns magneticthat result resonancefrom dry density (NMR), changes image remain processing techniquelargely and unknown. mercury It is intrusion therefore porosimetry useful to study have ways been to enhance used in the the predictive field of geotechnical methods of SWRCs. engineering to characterizeDue to efficiency the structure and non-destructiveness, of geomaterial the matrices. nuclear magnetic In this study, resonance drying (NMR), and image wetting processing SWRCs for sandytechnique soils were and determined mercury intrusion in the porosimetry laboratory usinghave b aeen pressure used in platethe field apparatus. of geotechnical In addition, engineering the NMR techniqueto characterize was used the to structure characterize of geomaterial water migration matrices. and In this the study, redistribution drying and process wetting at SWRCs various for matric suctionssandy during soils were drying determined and wetting. in the Imagelaboratory processing using a pressure was applied plate apparatus. to obtain theIn addition, area of pores the NMR between particlestechnique by imaging was used the to topcharacterize surface water of specimens migration with and the cutting redistribution ring. The process pore sizeat various distribution matric and porositysuctions of samples during weredrying investigated and wetting. with Image a mercury processing porosimetry was applied test. to Weobtain discuss the area the of results pores in the between particles by imaging the top surface of specimens with cutting ring. The pore size context of the influence of the dry density of soils on drying and wetting hysteresis. distribution and porosity of samples were investigated with a mercury porosimetry test. We discuss the results in the context of the influence of the dry density of soils on drying and wetting hysteresis. 2. Experimental Methodology 2.A Experimental soil sample was Methodology collected from a natural slope in Japan. The physical properties and grain size distributionA soil curves sample for was the collected test samples from a are natural shown slope in Tablein Japan.1 and The Figure physical1, respectively. properties and The grain soils size were classifieddistribution using thecurves Unified for the Soil test Classification samples are shown System. in Table The soil 1 and type Figure was a1, siltyrespectively. sand. The soils wereA pressure classified plate using apparatus the Unified was Soil employed Classification to establishSystem. The the soil water type was content a silty under sand. certain matric suction levelsA pressure [19]. The plate pressure apparatus plate was was employed thoroughly to establish saturated the beforewater content use and under was soakedcertain matric in distilled watersuction overnight levelsprior [19]. toThe testing. pressure A plate tube was from thoroughly the plate saturated outlet extended before use through and was the soaked water in to the atmospheredistilled towater enable overnight air to prior escape to fromtesting. the A cavitytube from between the plate the outlet ceramic extended and rubberthrough backing the water as to it was the atmosphere to enable air to escape from the cavity between the ceramic and rubber backing as it being wet. was being wet.
100
80
60
40
Percent smaller(%) 20
0 1E-3 0.01 0.1 1 10 Grain size (mm)
FigureFigure 1. 1.Grain Grain sizesize distribution curve curve for for test test soil. soil.
GeoHazards 2018, 1, x FOR PEER REVIEW 3 of 17 GeoHazards 2020, 1 5 Table 1. Physico-mechanical properties of soil.
Table 1. Physico-mechanicalProperties properties of soil.Value Specific gravity, Gs 2.71 Properties Value Effective grain size, D10 (mm) 0.0035 MeanSpecific grain gravity, size,G sD50 (mm) 2.71 0.14 Effective grain size, D10 (mm) 0.0035 Coefficient of uniformity, Cu 54.40 Mean grain size, D50 (mm) 0.14 Coefficient of gradation, Cc 1.95 Coefficient of uniformity, Cu 54.40 CoefficientMaximum of gradation, void ratio,Cc emax 1.951.59 MaximumMinimum void void ratio, ratio,emax emin 1.591.01 Minimum void ratio, e 1.01 Optimum water content,min wopt (%) 16.01 Optimum water content, w (%) 16.01 opt 3 Maximum dry density, ρdmax (g/cm3 ) 1.720 Maximum dry density, ρdmax (g/cm ) 1.720
The pressure plate was then placed in its apparatus. The plate outlet was connected with a piece of tubingThe pressure to the pressure plate was plate then placedapparatus in its outlet apparatus. which The extended plate outlet to the was burette. connected The with connecting a piece ofsurfaces tubing of to thethe pressurepressure plate plate apparatus apparatus outlet and whichits lid extendedwere wiped to the with burette. methylated The connecting spirit to surfacesremove ofmoisture the pressure and dust plate and apparatus the lid was and secured its lid were in posi wipedtion withon the methylated pressure plate spirit apparatus to remove bowl. moisture The andrelease dust tap and was the closed lid was and secured the regulator in position turned on on the gradually pressure until plate the apparatus pressure bowl. reached The levels release lower tap wasthan closed the desired and the pressure. regulator We turned allowed on a gradually few hours until for the the pressure pressure to reached equilibrate levels and lower adjusted than thethe desiredregulator pressure. as necessary. We allowed We did a few not hours allow for for the the pressure pressure to equilibrateto exceed andthe adjusteddesired pressure the regulator as this as necessary.introduced We problems did notallow with forhysteresis. the pressure A check to exceed was themade desired for leaking pressure pressure as this introducedplates by observing problems withregular hysteresis. bubbles A rising check in was burette. made forLeaking leaking plates pressure may plates result by from observing plates regularthat were bubbles inadequately rising in burette.saturated, Leaking developed plates amay leak, result or from from gas plates that thatwas were leaking inadequately into the tube saturated, connecting developed the plate a leak, to the or frompressure gas thatplate was apparatus leaking outlet. into the tube connecting the plate to the pressure plate apparatus outlet. PressurePressure waswas sustainedsustained inin thethe pressurepressure placeplace devicedevice untiluntil thethe waterwater levellevel inin thethe buretteburette becomingbecoming static.static. ThisThis processprocess maymay requirerequire 77 daysdays althoughalthough moremore timetime waswas neededneeded atat lowlow pressures.pressures. DuringDuring thisthis time,time, wewe ensured ensured that that the the water water level level in thein the burette burette was was maintained maintained at the at same the same level aslevel the as plate the inside plate theinside apparatus the apparatus to prevent to prevent adding anadding extra componentan extra component of water potentialof water frompotential outside from the outside apparatus. the Weapparatus. allowed We a few allowed days ofa few equilibrium days of eq beforeuilibrium taking before measurements. taking measurements. TheThe experiment involved involved sample sample preparation, preparation, ce ceramicramic disk disk saturation, saturation, and and recording recording drying drying and andwetting wetting SWRCs. SWRCs. The Thetest was test wasinitiated initiated by saturating by saturating the ceramic the ceramic disk diskby immersion by immersion in water in water and andsubjecting subjecting it to negative it to negative pressure pressure (−101.78 (− 101.78kPa). After kPa). 24 Afterh, the 24disk h, was the diskremoved was from removed the saturation from the saturationtank, installed tank, ininstalled the pressure in the plat pressuree apparatus. plate Weight apparatus. of oven-dried Weight of sand, oven-dried required sand, to achieve required target to 3 achievedensity targetwas computed. density was Soil computed. samples with Soil samplesdry densities with dryof 1.30, densities 1.40, of1.50 1.30, and 1.40, 1.60 1.50 g/cm and3 were 1.60 used g/cm in werethe experiment. used in the Soils experiment. were mixed Soils with were water mixed to withachieve water a gravimetric to achieve water a gravimetric content waterof 10% content for all the of 10%experiments. for all the To experiments. ensure that Toall ensurethe soil thatsamples all the ha soild the samples same structure had the samefor all structure tests, a predetermined for all tests, a 2 predeterminedsample mass was sample pressed mass into was a pressed2 cm-high into cutting a 2 cm-high ring with cutting the ring 30 cm with2-area, the 30i.e., cm inner-area, diameter i.e., inner of diameter61.8 mm of(Figure 61.8 mm 2). (FigureAfter a 2soil). After sample a soil was sample comp wasacted, compacted, it was then it was satu thenrated saturated using the using vacuum the vacuumsaturation saturation method. method. The sample The samplewas assembled was assembled on the onsaturated the saturated ceramic ceramic disk in disk the inpressure the pressure plate plateapparatus. apparatus.
(a) ρ = 1.30 g/cm3 (b) ρ = 1.40 g/cm3 (c) ρ = 1.50 g/cm3 (d) ρ = 1.60 g/cm3
Figure 2. Soil samples with different dry densities. Figure 2. Soil samples with different dry densities.
GeoHazards 2020, 1 6
The saturated soil sample underwent different amounts of matric suction in order to establish the SWRC in drying and wetting, and the correlating alteration in the sample moisture was noted. Afterwards, the sample was dried via strengthening matric suction to 0.5 kPa, 1.0 kPa, 2.0 kPa, 4.0 kPa, 7.0 kPa, 10.0 kPa, 15.0 kPa, 30.0 kPa, 50 kPa, 100 kPa, and 200 kPa. For matric suction, pore air pressure was increased. The magnitude of applied matric suction was thus equivalent to the applied pore air pressure. On each increment of matric suction, water flowed out from the specimen until equilibrium was reached. Equilibrium was generally attained in 24–48 h. To determine the amount of water that was lost from the specimen, the soils were removed from the cell and weighed, which provided the water content. The specimen was then moved into the sample tube of the NMR apparatus and an NMR measurement was made. NMR tests were conducted on the treated specimens by the AniMR-150 NMR imaging system, which is manufactured by Niumag Electric Technology Company, Ltd. in the city of Suzhou. Acquisition parameters and sampling parameters were set before starting the NMR experiment. This process was repeated until all the suction steps were applied. Maximum matric suction that can be applied to the specimen is limited by the air entry value of the ceramic disk, which was 300 kPa for the present apparatus. The maximum matric suction applied to soil specimens was 200 kPa. To simulate the wetting path, pore air pressure was reduced from 200 kPa, while still keeping the water pressure at the constant value of zero. After decreasing the pore air pressure, water flowed into the soil specimen through the ceramic disk until equilibrium was reached. Soils were removed from the cell and weighed again to determine the water content sucked by the specimen. The same procedure was repeated for lower matric suction values. Upon the specimen achieving zero matric suction in the wetting process (i.e., pore air vented to the atmosphere and the water level kept at the center of the specimen), the soil sample was taken out and the correlating moisture content was established by oven-drying the specimen. The actual water content of the soil samples was calculated using the final moisture content of the various suction levels of soil samples. The pore size distribution was measured with mercury intrusion porosimetry (MIP) in an AutoPore IV apparatus (Pmax = 400 MPa). This allowed study of pore radii ranging from 3.7 nm to 1000 µm. By measuring the volume of mercury that intruded into the sample with each pressure change, the volume of pores in the corresponding size class was calculated. The samples for the MIP test were carefully trimmed into pieces that had an approximately cubic shape and a volume of approximately 1 cm3. They were then vacuum dried with lyophilization using a Freeze Dryer apparatus (ALPHA 1–2 Ld Plus—Martin Christ Gefriertrocknungsanlagen GmbH, Osterode am Harz, Germany) to remove the pore water without damaging the original texture of the soil sample. MIP testing was conducted on the soil after SWRC test.
3. Experimental Results
3.1. SWRC Test The experimental data of soils with different densities are presented in Figure3. There were three stages in the process of desaturation of a soil. These are the transition stage, the boundary effect stage, and the residual stage of unsaturation. They are described as follows: Boundary effect stage: At this stage, the pore water is in a state of tension. However, due to capillary force, the soil was still mostly saturated, and this phase continues to the entrance value of the soil. Transition stage: The phase starts at the end of the inlet and residual water content. The soil begins to saturate with the air inlet value, and the water in the pores becomes greater with more airflow. When the suction increases, the flow of water is liquid. As the suction increases, the soil dries quickly. The connectedness of water in the pore or the void continues to decrease as the suction increases. GeoHazards 20202018, 1, x FOR PEER REVIEW 5 of 177
Residual stage of saturation: Big increases in suction result in very little change in water content Residual stage of saturation: Big increases in suction result in very little change in water content (or degree of saturation). Water and soil are closely combined, and the water movement mainly (or degree of saturation). Water and soil are closely combined, and the water movement mainly occurs occurs as vapor flow. There is a small amount of hydraulic flow through the pore; but, there is some as vapor flow. There is a small amount of hydraulic flow through the pore; but, there is some water water movement as film flow. movement as film flow. The air-entry value and the residual state (residual water content and residual suction) are The air-entry value and the residual state (residual water content and residual suction) are normally determined by following this procedure: draw the tangent line through the inflection point, normally determined by following this procedure: draw the tangent line through the inflection point, followed by a horizontal line through the initial point and another tangent line through the point followed by a horizontal line through the initial point and another tangent line through the point where the curve starts to drop linearly in the high suction range. Finally, the intersections of these where the curve starts to drop linearly in the high suction range. Finally, the intersections of these tangent lines indicate air-entry value (AEV) and the residual state. tangent lines indicate air-entry value (AEV) and the residual state. Because of the impact of dry density, the dry SWRCs in the laboratory revealed the variations in Because of the impact of dry density, the dry SWRCs in the laboratory revealed the variations in SWRCs, which were established by the differences in SWRC parameters, for example, the air-entry SWRCs, which were established by the differences in SWRC parameters, for example, the air-entry value (ψa), residual matric suction (ψr), and slope of the SWRC (s). To examine the impact of value (ψ ), residual matric suction (ψ ), and slope of the SWRC (s). To examine the impact of preliminary preliminarya density on SWRC parameters,r SWRCs of unsaturated soil with various drying densities density on SWRC parameters, SWRCs of unsaturated soil with various drying densities were reviewed. were reviewed. The air-entry value, AEV or ψa, is characterized as the matric suction in which the air begins by The air-entry value, AEV or ψa, is characterized as the matric suction in which the air begins by going into the biggest pores in the soil throughout a drying process [20]. Thus, the air-entry value is going into the biggest pores in the soil throughout a drying process [20]. Thus, the air-entry value is the substrate suction force of the surface tension coming out of the surface tension of the water surface the substrate suction force of the surface tension coming out of the surface tension of the water surface in the largest pore. in the largest pore. The volumetric water content of the soil was nearly constant as the matric suction increased from The volumetric water content of the soil was nearly constant as the matric suction increased from zero (corresponding to saturation state) to the AEV of the soil. Then the water content stably decreased zero (corresponding to saturation state) to the AEV of the soil. Then the water content stably to the residual water content, θr, as the matric suction increased beyond the AEV. The θr is the water decreased to the residual water content, θr, as the matric suction increased beyond the AEV. The θr content in a residual state. A large matric suction change is required to remove additional water from is the water content in a residual state. A large matric suction change is required to remove additional the soil and the water phase is discontinuous in this state. The soil matric suction corresponding to water from the soil and the water phase is discontinuous in this state. The soil matric suction the residual water content is called the residual soil matric suction, ψr. The definitions of air-entry corresponding to the residual water content is called the residual soil matric suction, ψr. The value (ψa), residual matric suction (ψr), and the slope of SWRC (s) are given by Vanapalli et al. [21]. definitions of air-entry value (ψa), residual matric suction (ψr), and the slope of SWRC (s) are given The slope of the SWRC can be measured as [(ψa − ψr)/(logψr − logψa)]. by Vanapalli et al. [21]. The slope of the SWRC can be measured as [(ψa − ψr)/(logψr − logψa)].
0.5 0.5
Drying 0.4 0.4
0.3 0.3
0.2 Wetting 0.2
0.1 0.1 Volumetric water content water Volumetric Volumetric water content water Volumetric
0.0 0.0 0.1 1 10 100 1000 0.1 1 10 100 1000 Suction (kPa) Suction(kPa) (a) ρ = 1.30 g/cm3 (b) ρ = 1.40 g/cm3 0.5 0.5
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1 Volumetric water content Volumetric water content water Volumetric
0.0 0.0 0.1 1 10 100 1000 0.1 1 10 100 1000 Suction (kPa) Suction (kPa) (c) ρ = 1.50 g/cm3 (d) ρ = 1.60 g/cm3
Figure 3. Influence of dry density on the SWRC for unsaturated soil. Figure 3. Influence of dry density on the SWRC for unsaturated soil.
GeoHazards 2020, 1 8 GeoHazards 2018,, 1,, xx FORFOR PEERPEER REVIEWREVIEW 6 of 17
The initial dry density of soil samplessamples hashas aa significantsignificant effecteffect onon SWRCsSWRCs (Figure(Figure4 ).4). AtAt thethe beginning of of zero zero suction, suction, the the initial initial volume volume of ofthe the soil soil samples samples with with high high dry drydensity density was low. was The low. decreaseThe decrease in the in initial the initial volumetric volumetric water water content content may maybe due be to due the to reduction the reduction of thethe of volumevolume the volume ofof thethe of voidvoid the invoid increased in increased initial initial dry density. dry density. The volumetric The volumetric water water content content of all of of all the of soils the soils decreased, decreased, as the as suctionthe suction increased, increased, but butthey they decreased decreased at atdifferent different rates. rates. The The increase increase of of dry dry density density of of the the sample produced a lower volumetric water contentcontent reductionreduction rate.rate.
0.5 1.3g/cm1.3g/cm3 1.4g/cm1.4g/cm3 0.4 1.5g/cm1.5g/cm3 1.6g/cm1.6g/cm3 0.3
0.2
0.1 Volumetric water content Volumetric water content
0.0 0.1 1 10 100 1000 Suction(kPa) Figure 4. InfluenceInfluence ofof drydry densitydensity ofof soilssoils onon thethe SWRCSWRC duringduring drying.drying. Figure 4. Influence of dry density of soils on the SWRC during drying.
Figure 5 shows the propensity of soil samples with largerlarger drydry densitydensity toto havehave greatergreater airair entryentry values.Figure This5 may shows be thedue propensity to the existence of soil of samples the sma withll mean larger pore dry size density in soil to samples have greater caused air by entry high dryvalues. density. This mayAccording be due to to Vanapalli the existence et al. of the[21], small the air-entry mean pore value size infrom soil each samples sample caused was by compared high dry withdensity. the Accordingdry density, to Vanapalliand the results et al. [21showed], the air-entrythat thethe valueentryentry fromvaluevalue each increasedincreased sample linearlylinearly was compared withwith thethe with drydry density.the dry density, and the results showed that the entry value increased linearly with the dry density.
5 Test data 4 Fitting curve
3
2 2 y=8.860x-9.950 AEV (kPa) AEV (kPa) R2=0.9533 1
0 1.3 1.4 1.5 1.6 3 Dry density (g/cm )
FigureFigure 5. 5. VariationVariationVariation ofof air-entryair-entry vava valueslueslues againstagainst drydry density. density.
Compared with the sample with a low density, the sample with a high density loses water at a faster rate (Figure 4). Figure 6 reveals the slope of the SWRC with dry density. The greater the dry
GeoHazards 2020, 1 9
Compared with the sample with a low density, the sample with a high density loses water at a GeoHazards 2018, 1, x FOR PEER REVIEW 7 of 17 faster rate (Figure4). Figure6 reveals the slope of the SWRC with dry density. The greater the dry GeoHazards 2018, 1, x FOR PEER REVIEW 7 of 17 density,density, the lower the slope of the SWRC. Thus, the water content of of the high-density soil soil is is greater greater thandensity,than the the low-densitythe low-density lower the soil soil slope at at a aof matric matric the SWRC. suction suction Thus, beyond the their water air-entry content values of the high-density (see Figure4 4).). soil is greater than the low-density soil at a matric suction beyond their air-entry values (see Figure 4). 0.6 0.6 Test data FittingTest data curve Fitting curve 0.4 0.4 y = -0.663x + 1.3906 Ry 2= = -0.663x 0.9989 + 1.3906 R2 = 0.9989 0.2 Slope of Slope SWRC 0.2 Slope of Slope SWRC
0.0 1.3 1.4 1.5 1.6 0.0 3 1.3 1.4Dry density (g/cm 1.5) 1.6 Dry density (g/cm3) Figure 6. Variation of the slope of drying SWRC with the dry density. Figure 6. Variation of the slope of drying SWRC with the dry density. Figure 6. Variation of the slope of drying SWRC with the dry density. Figure 7 reveals the difference of residual matric suction in conjunction with the dry density of Figure7 reveals the difference of residual matric suction in conjunction with the dry density of the soilFigure specimen. 7 reveals An the increase difference in preliminary of residual dry matric density suction increases in conjunction the residual with matric the dry suction density (ψ rof). the soil specimen. An increase in preliminary dry density increases the residual matric suction (ψ ). Thesethe soil findings specimen. agree An with increase the resu in preliminarylts of Croney dry and density Coleman increases [22]. the residual matric suction (ψrr). These findingsfindings agree with the resultsresults ofof CroneyCroney andand ColemanColeman [[22].22]. 25 25 Test data FittingTest data curve 20 Fitting curve 20
15 15
10 10 y =85.25x2 - 197.5x + 118.8 Residual suction (kPa) 5 Ry 2=85.25x2 = 0.9962 - 197.5x + 118.8 Residual suction (kPa) 5 R2 = 0.9962 0 1.31.41.51.6 0 3 1.31.41.51.6Dry density (g/cm ) Dry density (g/cm3) Figure 7. Variation of the residual matric suction with the dry density. Figure 7. Variation of the residual matric suction with the dry density. Figure 7. Variation of the residual matric suction with the dry density. The water content of soil at a matric suction is not unique. The drying curve is always located aboveThe the water wetting content curve. of Thesoil atsoil a matrichas different suction su isction not unique.during Thethe dryingdrying andcurve wetting is always processes, located whichabove isthe called wetting hysteresis. curve. TheIn this soil study, has different the hysteresis suction between during the the drying drying and and wetting wetting SWRCs processes, was quantifiedwhich is called (Figure hysteresis. 8). The area In this between study, the the drying hysteresis and betweenwetting SWRCs the drying was and computed wetting on SWRCs a log scale was overquantified the matric (Figure suction 8). The range area from between 0.1 kPa the todrying 200 kPa. and Thewetting hysteresis SWRCs from was the computed high-density on a log sample scale wasover less the thanmatric that suction from the range low-density from 0.1 samplekPa to 200 (Figure kPa. 8).The The hysteresis possible fromreasons the for high-density this behavior sample may bewas less less pore-volume than that from and the greater low-density capillary sample potential (Figure created 8). The by possiblethe denser reasons soil sample. for this behavior may be less pore-volume and greater capillary potential created by the denser soil sample.
GeoHazards 2020, 1 10
The water content of soil at a matric suction is not unique. The drying curve is always located above the wetting curve. The soil has different suction during the drying and wetting processes, which is called hysteresis. In this study, the hysteresis between the drying and wetting SWRCs was quantified (Figure8). The area between the drying and wetting SWRCs was computed on a log scale over the matric suction range from 0.1 kPa to 200 kPa. The hysteresis from the high-density sample was less than that from the low-density sample (Figure8). The possible reasons for this behavior may be less
GeoHazardspore-volume 2018, and1, x FOR greater PEER capillary REVIEW potential created by the denser soil sample. 8 of 17
1200 Test data Fitting curve
800
y = -2720x + 4569 400 R2 = 0.9909 Hysteresis (kPa) Hysteresis
0 1.3 1.4 1.5 1.6 3 Dry density (g/cm )
FigureFigure 8. 8. VariationVariation of of the the hysteres hysteresisis with the dry density.
3.2. NMR Test 3.2. NMR Test Transverse relaxation corresponds to transverse magnetization decay. Transverse magnetization Transverse relaxation corresponds to transverse magnetization decay. Transverse magnetization decay is described by an exponential curve, characterized by the time constant T2. After time T2, decay is described by an exponential curve, characterized by the time constant T2. After time T2, transverse magnetization has lost 63% of its original value. The transverse relaxation time (T2) of pore transverse magnetization has lost 63% of its original value. The transverse relaxation time (T2) of fluid in the soil sample is proportional to the pore diameter [23]. Therefore, the T2 distribution reflects pore fluid in the soil sample is proportional to the pore diameter [23]. Therefore, the T2 distribution the characteristics of the soil pore structure. The T2 distribution curve of various dry densities is reflects the characteristics of the soil pore structure. The T2 distribution curve of various dry densities shown in Figure 9. The maximum T2 is the value of the largest NMR signal, indicating the optimal is shown in Figure9. The maximum T is the value of the largest NMR signal, indicating the optimal radius of the pore filled with water at the2 specified matric suction value. radius of the pore filled with water at the specified matric suction value. The T2 distribution curve area and the maximum T2 decrease with the increase of the matric The T distribution curve area and the maximum T decrease with the increase of the matric suction. This2 indicates that the pore water discharge occurs2 from the larger pores to the smaller pores. suction. This indicates that the pore water discharge occurs from the larger pores to the smaller pores. Also, we observed the same change trend for the T2 distribution curve area and the maximum T2 with Also, we observed the same change trend for the T distribution curve area and the maximum T with dry density, indicating that the change of dry density2 is mainly responsible for the large pores.2 As dry density, indicating that the change of dry density is mainly responsible for the large pores. As dry dry density increases, the number of large pores decreases. density increases, the number of large pores decreases. 80 80 1.3g/cm3 3 1.4g/cm3 1.3g/cm 3 1.5g/cm3 1.4g/cm 60 60 3 1.6g/cm3 1.5g/cm 1.6g/cm3
40 40 NMR signal NMR signal
20 20
0 0 0.01 0.1 1 10 100 1000 0.01 0.1 1 10 100 1000 T2 (ms) T2 (ms) (a) ψ = 0 kPa (b) ψ= 50 kPa
GeoHazards 2018, 1, x FOR PEER REVIEW 8 of 17
1200 Test data Fitting curve
800
y = -2720x + 4569 400 R2 = 0.9909 Hysteresis (kPa) Hysteresis
0 1.3 1.4 1.5 1.6 3 Dry density (g/cm ) Figure 8. Variation of the hysteresis with the dry density.
3.2. NMR Test Transverse relaxation corresponds to transverse magnetization decay. Transverse magnetization decay is described by an exponential curve, characterized by the time constant T2. After time T2, transverse magnetization has lost 63% of its original value. The transverse relaxation time (T2) of pore fluid in the soil sample is proportional to the pore diameter [23]. Therefore, the T2 distribution reflects the characteristics of the soil pore structure. The T2 distribution curve of various dry densities is shown in Figure 9. The maximum T2 is the value of the largest NMR signal, indicating the optimal radius of the pore filled with water at the specified matric suction value. The T2 distribution curve area and the maximum T2 decrease with the increase of the matric suction. This indicates that the pore water discharge occurs from the larger pores to the smaller pores. Also, we observed the same change trend for the T2 distribution curve area and the maximum T2 with GeoHazardsdry density,2020 ,indicating1 that the change of dry density is mainly responsible for the large pores. As11 dry density increases, the number of large pores decreases.
80 80 1.3g/cm3 3 1.4g/cm3 1.3g/cm 3 1.5g/cm3 1.4g/cm 60 60 3 1.6g/cm3 1.5g/cm 1.6g/cm3
40 40 NMR signal NMR signal
20 20
0 0 0.01 0.1 1 10 100 1000 0.01 0.1 1 10 100 1000 T2 (ms) T2 (ms) GeoHazards 2018, 1, x FOR( aPEER) ψ =REVIEW 0 kPa (b) ψ= 50 kPa 9 of 17 GeoHazards 2018, 1, x FOR PEER REVIEW 9 of 17 80 80 80 80 1.3g/cm3 1.3g/cm3 3 3 1.3g/cm 3 1.4g/cm 1.3g/cm1.4g/cm3 3 3 60 1.4g/cm 3 3 1.5g/cm 60 1.4g/cm1.5g/cm 60 3 60 3 1.5g/cm 3 3 1.6g/cm 1.5g/cm1.6g/cm 1.6g/cm3 1.6g/cm3 40 40 40 40 NMR signal NMR signal NMR signal NMR signal 20 20 20 20
0 0 00.01 0.1 1 10 100 1000 00.01 0.1 1 10 100 1000 0.01 0.1 1 10 100 1000 0.01 0.1 1T2 (ms) 10 100 1000 T2 (ms) T2 (ms) T2 (ms) (c) ψ= 100 kPa (d) ψ= 200 kPa (c) ψ= 100 kPa (d) ψ= 200 kPa Figure 9. T2 distribution curves for various dry densities at different matric suctions.
Figure 9. T22distributiondistribution curves curves for for various various dry dry de densitiesnsities at at different different matric matric suctions. suctions. 4. Discussion 4. Discussion Discussion 4.1. Mechanism of Density Effect 4.1. Mechanism Mechanism of of Density Density Effect The presence of water creates surface tension in the unsaturated soil, and it cements the soil The presence of water creates surface tension in the unsaturated soil, and it cements the soil particles. The cement is defined as the matrix suction, and indicates that the water has a tension particles. The The cement cement is is defined defined as as the the matrix matrix suction, suction, and and indicates indicates that that the the water water has a tension phenomenon shown by the meniscus between the two soil particles (see Figure 10). phenomenon shown by the meniscus between the two soil particles (see Figure 1010).).
Figure 10. Condition of soil structure. Figure 10. Condition of soil structure.
Aitchison [24] determined that the matric suction (ua − uw) obeys the capillary model as follows: Aitchison [24] determined that the matric suction (ua − uw) obeys the capillary model as follows: −=2cosγ (1) uuaw−=2cosTRs γ s uuawTRs s (1) where Ts is the surface tension force of water, in units of N/m; Rs is the radius of curvature of the where Ts is the surface tension force of water, in units of N/m; Rs is the radius of curvature of the meniscus, in units of m; γ is the contact angle, in units of degree. From Equation (1), soil pores having meniscus, in units of m; γ is the contact angle, in units of degree. From Equation (1), soil pores having small radii will have high suction because it corresponds to the rise of water level in capillary tubes small radii will have high suction because it corresponds to the rise of water level in capillary tubes with smaller radii. Figure 11 shows different air–water interfaces at different radii curvatures. with smaller radii. Figure 11 shows different air–water interfaces at different radii curvatures.
GeoHazards 2020, 1 12
Aitchison [24] determined that the matric suction (ua − uw) obeys the capillary model as follows:
ua − uw = 2Ts cos γ /Rs (1) where Ts is the surface tension force of water, in units of N/m; Rs is the radius of curvature of the meniscus, in units of m; γ is the contact angle, in units of degree. From Equation (1), soil pores having small radii will have high suction because it corresponds to the rise of water level in capillary tubes with smaller radii. Figure 11 shows different air–water interfaces at different radii curvatures. GeoHazards 2018, 1, x FOR PEER REVIEW 10 of 17
FigureFigure 11. 11. CapillaryCapillary tubes tubes indicating indicating the the air–water air–water interfaces interfaces at at different different radii curvatures [25]. [25].
SoilsSoils are are subjected to to drying and wetting. In In the the drying drying stage, stage, when when air air displaces displaces water water in in the the porepore or when pore pore water water pressure pressure de decreases,creases, it will increase the strength of the soil as as matric matric suction suction increases.increases. When When soil soil is is in in a awetting wetting stage, stage, as aswate waterr infiltrates infiltrates the thesoil soil structur structuree and and displaces displaces pore pore air, poreair, pore water water pressure pressure will will gradually gradually increase. increase. This This diminishes diminishes soil soil matric matric suction suction and and forces forces soil soil particlesparticles apart. In In unsaturated unsaturated soil soil this condition is defined defined as failure state. UponUpon dry density increase ofof thethe soilsoil specimen,specimen, thethe size size and and number number of of pores pores in in the the soil soil matrix matrix is islowered. lowered. Thus, Thus, the the radius radius of of the the curvature curvature of of the the meniscus meniscus is is lowered lowered and and the the correlating correlating suction suction is isgreater. greater. As As a a result, result, the the matric matric suction suction (AEV) (AEV) requiredrequired forfor airair toto enter into the soil soil matrix matrix increases. increases. SoilsSoils with with small small pore pore size size drain drain slowly. slowly. Therefore, Therefore, the the water water content content in in the the soil soil with with small small pores pores is is greatergreater than than that that in in soil soil with with large large pores pores at at the the matr matricic suction suction that that exceeds exceeds the the air-entry air-entry value. For For the the samesame volumetric volumetric water water content, content, the the soil soil with with higher higher dry dry density density has has a a greater greater matric matric suction suction value. value. ThisThis is is because the the increase increase in in the the dry dry density density of of a a soil soil sample sample results results in in decreased decreased soil soil permeability permeability leadingleading to to a a slower slower de-saturation de-saturation process process and and also also to to an an increase increase in in the the AEV AEV and and the the residual residual suction. suction. TheThe reduction ofof hysteresishysteresis with with increasing increasing dry dry density density may may result result from from the the reduced reduced pore-volume pore-volume and andgreater greater capillary capillary potential potential caused caused by the by higherthe higher dry densitydry density of the of soilthe sample.soil sample. ToTo further document the mechanismmechanism of density effect, images with resolutionresolution of 38763876 ×× 25842584 were made made of of samples samples with with different different dry dry densities densities prior prior to SWRC to SWRC tests. The tests. area The of the area pores of the between pores particlesbetween is particles the crucial is the issue. crucial The issue. process The is processas follows: is as (1) follows: convert (1) the convert image to the a imagegrayscale to a image; grayscale (2) determineimage; (2) determinethe differentiation the differentiation between particles between and particles pores; and(3) establish pores; (3) the establish area of thethe areapore of(see the Table pore 2). The important part of this process is to determine the exact threshold binarization value (eigenvalue) to distinguish particles and pores, and to evaluate the pore area. Using the intermediate color of the image, the threshold is applied to the concentration between the filter particle and the pore. The threshold can significantly influence the particle shape. The image processing program, Image J, was utilized for analysis of the images. This program can capture a binary image, transform a color image into a grayscale image, and compute the area of pores between particles based on Otsu algorithm. From the binary images, we observed that an increase in dry density caused a larger black portion area and a smaller white portion area.
GeoHazards 2020, 1 13
(see Table2). The important part of this process is to determine the exact threshold binarization value (eigenvalue) to distinguish particles and pores, and to evaluate the pore area. Using the intermediate color of the image, the threshold is applied to the concentration between the filter particle and the pore. The threshold can significantly influence the particle shape. The image processing program, Image J, was utilized for analysis of the images. This program can capture a binary image, transform a color image into a grayscale image, and compute the area of pores between particles based on Otsu algorithm. From the binary images, we observed that an increase in dry density caused a larger black portion area and a smaller white portion area. GeoHazardsGeoHazardsGeoHazardsGeoHazardsGeoHazardsGeoHazardsGeoHazards 20182018 20182018 2018,2018, 1 1,2018,,2018 , 1 x1x,,, , FOR x1xFORx1 , , ,FOR FOR FOR 1xx1x, , FOR FOR FORx xPEERPEER FOR FOR PEERPEERPEER PEERPEERPEER REVIEWPEERREVIEWPEER REVIEWREVIEWREVIEW REVIEWREVIEWREVIEW REVIEWREVIEW 11111111 ofof 11 11of of 171711 11 of17of17 of of17 17 17 17 Table 2. Calculation steps for determining the area of pores between particles. TableTableTableTableTableTableTable 2.2. 2. 2. CalculationCalculation 2.Calculation2.CalculationCalculation 2.CalculationCalculationCalculation CalculationCalculation stepssteps stepsstepssteps stepsstepssteps forstepsstepsfor forforfor determining determiningforforfor determining determining determiningforfor determiningdeterminingdetermining determiningdetermining thethe thethethe areathetheareathe area areaareathethe areaareaarea ofof area area ofof of porespores of ofofporesporespores of of poresporespores pores poresbetweenbetween betweenbetweenbetween betweenbetweenbetween betweenbetween particles.particles. particles.particles.particles. particles.particles.particles. particles.particles. 3 Dry Density/(g/cm3 ) 3 Step DryDryDryDryDryDry DensityDensityDry DensityDensity DensityDensity Density//(g/cm(g/cm//(g/cm(g/cm//(g/cm(g/cm//(g/cm(g/cm33)) 3 )) 33)) 3)) StepStepStepStepStepStep Step 1.301.301.301.301.301.30 1.30 1.401.401.401.40 1.401.401.40 1.40 1.50 1.501.501.501.501.50 1.50 1.60 1.601.601.601.601.60 1.60
OriginalOriginalOriginalOriginalOriginalOriginalOriginalOriginal
GrayscaleGrayscaleGrayscaleGrayscaleGrayscaleGrayscaleGrayscale
BinaryBinaryBinaryBinaryBinaryBinaryBinaryBinary
TheTheTheTheTheThe thresholdthresholdThe thresholdthreshold thresholdthreshold threshold valuesvalues valuesvalues valuesvalues values werewere werewere werewere were determineddetermined determineddetermined determineddetermined determined artificialartificial artificialartificial artificialartificial artificially.ly.ly.ly. Thely.Thely. TheThely.ly. TheThe darkdarkThe Thedarkdark darkdark darkdarkpartpart partpart partpart inpartinpart inin the thein inthe the in in thetheimageimage thetheimageimage imageimage imageimage illustratesillustrates illustratesillustrates illustratesillustrates illustratesillustrates thethe thethe thethe thethe The threshold values were determined artificially. The dark part in the image illustrates the particlesparticlesparticlesparticlesparticlesparticlesparticles whilewhile whilewhile whilewhile while thethe thethe thethelightlight thelightlight lightlight lightportionportion portionportion portionportion portion illustratesillustrates illustratesillustrates illustratesillustrates illustrates thethe thethe thethepopo thepopores. res.popores.res. pores. res. TheThe res. res.TheThe TheThe area areaThe The areaarea areaarea ofareaofarea ofof the theof of the the of of thetheporespores thetheporespores porespores porespores waswas waswas waswas computedcomputedwaswas computedcomputed computedcomputed computedcomputed byby byby thebytheby thethebyby thethe thethe particles while the light portion illustrates the pores. The area of the pores was computed by the programprogramprogramprogramprogramprogramprogram andand andand andand was wasand waswas waswas subtracted subtractedwas subtractedsubtracted subtractedsubtracted subtracted byby byby the thebyby the theby thearthear artheareaea areaarea of ofareaea ofof eathethe of of thethe of thecuttingthecutting cuttingthecutting cuttingcutting cutting ring.ring. ring.ring. ring.ring. Usuallyring.Usually UsuallyUsually UsuallyUsually Usually thethe thethe theporesthepores poresthepores porespores pores inin inin soilssoils in in soilssoils in soilssoils aresoilsare areare aresortedaresorted sortedaresorted sortedsorted sorted intointo intointo intointo into program and was subtracted by the area of the cutting ring. Usually the pores in soils are sorted microporesmicroporesmicroporesmicroporesmicroporesmicroporesmicropores ((d d (( d
GeoHazards 2020, 1 14 GeoHazards 2018, 1, x FOR PEER REVIEW 12 of 17 GeoHazards 2018, 1, x FOR PEER REVIEW 12 of 17 100 100
80 80
60 Macropore 60 MesoporeMacropore MiniporeMesopore 40 MicroporeMinipore Area ratio (%) 40 Micropore Area ratio (%) 20 20
0 1.3 1.4 1.5 1.6 0 3 1.3Dry 1.4 density (g/cm 1.5) 1.6 Dry density (g/cm3) Figure 12. Relationship between the area ratio of different types of pores and dry density. Figure 12. Relationship between the area ratio of different types of pores and dry density. Figure 12. Relationship between the area ratio of different types of pores and dry density. The primary variations of SWRCs verified by the mercury porosimetry test are induced by the pore Thesize primarydistribution variations of the soil.of SWRCs Figure verified verified13 demons by tratesth thee mercury the pore porosimetry size distribution test are of inducednumerous by dry the poredensity size soils distribution after SWRC of theexamination. soil. Figure When 13 demonstratesdemons contrasttratesing the the outcomes pore size distributionof soil materials of numerous with various dry drydensity densities, soils after the primary SWRC examination. discovery is When that the contrast contrasting pore ingsize the distribution outcomes altersof soil with materials the dry with density various in thedry soil. densities, The ratio the primaryof bigger discovery pores to smalleris that the pore pores lowers. size distribution Compared alters with thewith relatively the dry densityloose soil, in the total soil. Thepore The ratio ratiovolume of bigger of the pores soil with to smallerhigh dry pore pores denss lowers.ity is lower. Compared Compared These resultswith with the the indicate relatively relatively that loose loosethere soil, soil,is a correlationthe total pore between volume pore of thevolume soil with caused high by dry dry densitydens densityity is and lower. SWRCs. These results indicate that there is a correlation between pore volumevolume causedcaused byby drydry densitydensity andand SWRCs.SWRCs. 0.04
0.04 3 1.6g/cm 1.5g/cm1.6g/cm3 0.03 1.4g/cm1.5g/cm3 0.03 1.3g/cm1.4g/cm3 1.3g/cm3 0.02 0.02
0.01 Incremental volume (ml/g) 0.01 Incremental volume (ml/g)
0.00 0.001E-4 1E-3 0.01 0.1 1 10 100 1000 10000 μ 1E-4 1E-3 0.01Pore 0.1 diameter( 1 10m) 100 1000 10000 Pore diameter(μm) Figure 13. Effect of the dry density on the pore size distribution of soils. Figure 13. Effect of the dry density on the pore size distribution of soils. Figure 13. Effect of the dry density on the pore size distribution of soils. 4.2. Water–Air–Soil Particle System 4.2. Water–Air–Soil Particle System 4.2. Water–Air–SoilThe water–air–soil Particle particle System system in a soil sample changes with the degree of saturation throughoutThe water–air–soil a SWRC. Many particle studies syst haveem beenin a conducsoil sampleted on changes the water–air–soil with the degreeparticle ofrelationships saturation The water–air–soil particle system in a soil sample changes with the degree of saturation ofthroughout unsaturated a SWRC. soils [27–29]. Many studies Yu and have Chen been [28] conduc describedted onthree the basic water–air–soil water–air–soil particle particle relationships systems throughout a SWRC. Many studies have been conducted on the water–air–soil particle relationships inof unsaturated soilssoils: [27–29]. closed-air, Yu andbi-opened, Chen [28] and described closed-water three (Figure basic water–air–soil 14). Three partially particle saturated systems of unsaturated soils [27–29]. Yu and Chen [28] described three basic water–air–soil particle systems conditionsin unsaturated were soils: described closed-air, by Kohgo bi-opened, et al. [29],and closed-wateras insular air (Figure saturation, 14). pendularThree partially saturation, saturated and fuzzyconditions saturation. were described In the closed-air by Kohgo system et al. [29],(Fig ureas insular14a), which air saturation, corresponds pendular to a highsaturation, degree and of saturation,fuzzy saturation. pore air In is theclosely closed-air contained system by pore (Fig wateure r14a), and existswhich in corresponds the form of anto aira highbubble. degree The bi-of saturation, pore air is closely contained by pore water and exists in the form of an air bubble. The bi-
GeoHazards 2020, 1 15 in unsaturated soils: closed-air, bi-opened, and closed-water (Figure 14). Three partially saturated conditions were described by Kohgo et al. [29], as insular air saturation, pendular saturation, and fuzzy saturation. In the closed-air system (Figure 14a), which corresponds to a high degree of saturation, pore GeoHazards 2018, 1, x FOR PEER REVIEW 13 of 17 air is closely contained by pore water and exists in the form of an air bubble. The bi-opened system (Figure 14b) is the transition between a closed-air system and a closed-water system. In a closed-water opened system (Figure 14b) is the transition between a closed-air system and a closed-water system. system (Figure 14c), the pore water is discontinuous and separated by pore air and soils. These three In a closed-water system (Figure 14c), the pore water is discontinuous and separated by pore air and systems in unsaturated soil can be distinguished by the AEV and residual matric suction value in soils. These three systems in unsaturated soil can be distinguished by the AEV and residual matric the SWRC [30]. Kohgo et al. [31] summarized three influences of matric suction on the mechanical suction value in the SWRC [30]. Kohgo et al. [31] summarized three influences of matric suction on properties of unsaturated soils: (1) In the insular air saturation (closed-air) system, pore air exits in the the mechanical properties of unsaturated soils: (1) In the insular air saturation (closed-air) system, form of air bubbles surrounded by water. An increase in matric suction therefore increases effective pore air exits in the form of air bubbles surrounded by water. An increase in matric suction therefore stresses; (2) In the pendular saturation (closed-water) system, matric suction can create capillary forces. increases effective stresses; (2) In the pendular saturation (closed-water) system, matric suction can This means an increase in matric suction enhances yield stresses and influences resistance to plastic create capillary forces. This means an increase in matric suction enhances yield stresses and deformations; (3) Both of the suction effects mentioned in (1) and (2) should be considered in the fuzzy influences resistance to plastic deformations; (3) Both of the suction effects mentioned in (1) and (2) saturation (bi-opened) system. should be considered in the fuzzy saturation (bi-opened) system.
(a) Closed-air system (b) Bi-opened system (c) Closed-water system
Figure 14. Three basic possible saturation conditions in unsaturated soils. Figure 14. Three basic possible saturation conditions in unsaturated soils. Oh and Vanapalli [32] divided typical SWRC into three zones: (i) saturation zone, (ii) transition zone,Oh and and (iii) Vanapalli residual [zone.32] divided In the typicalsaturation SWRC zone, into all three the voids zones: are (i) saturationcompletely zone, filled (ii) with transition water, thuszone, the and matric (iii) residual suction zone.is extremely In the saturation low. In the zone, transition all the zone, voids inter-particle are completely voids filled are with filled water, partially thus withthe matric water suctionand partially is extremely with air, low. and In thechanges transition in soil zone, moisture inter-particle are sharply voids reflected are filled by partially variation with of matricwater andsuction. partially In the with residual air, and zone, changes most in of soil the moisture voids are are filled sharply with reflected air; water by is variation only coating of matric soil particlessuction. Inin thethe residualform of zone,adsorbed most water of the film. voids are filled with air; water is only coating soil particles in the form of adsorbed water film. 4.3. Hysteresis 4.3. Hysteresis A SWRC depends on the pore-size distribution of the soil. This, in turn, has relations to the A SWRC depends on the pore-size distribution of the soil. This, in turn, has relations to the porosity and grain-size distribution. Many reports have shown the significance of hysteresis effects whenporosity analyzing and grain-size problems distribution. involving Many solute reports transp haveort, shownwater thetransfer, significance multiphase of hysteresis flow, and/or effects microbialwhen analyzing activities problems [33–36]. involving Neglecting solute hysteresis transport, effects water transfer, in calculations multiphase can flow, result and/or in significant microbial discrepanciesactivities [33– 36between]. Neglecting measured hysteresis and predicted effects inresults calculations [37–39]. can result in significant discrepancies betweenThere measured may be andseveral predicted reasons results for the [37 hysteresis–39]. effect. These include: (1) the ink bottle effect causedThere by the may inhomogeneity be several reasons of the for thedifferent hysteresis shapes effect. and These sizesinclude: of the interconnected (1) the ink bottle pores. effect Drying caused progressby the inhomogeneity is controlled by of thethe differentsmaller pore shapes radius, and sizesr, butof wetting the interconnected relies on the pores.larger Dryingpore radius progress R (see is r R Figurecontrolled 15a); by (2) the various smaller liquid–solid pore radius, contact, but wetting angles reliesfor advancing on the larger (wetting) pore radius and receding(see Figure (drying) 15a); water(2) various meniscus liquid–solid (Figure contact15b); (3) angles air trapped for advancing in the newly (wetting) wet soil and (e.g., receding pore doublet (drying) [40]); water and meniscus (4) the shrinkage(Figure 15 b);and (3) expansion air trapped of soil in the under newly wetting wet soil and (e.g., drying pore causes doublet various [40]); alterationsand (4) the inshrinkage the soil structure.and expansion By rearranging of soil under the wetting aggregates and dryingand granules, causes various volume alterations change alters in the the soil pore structure. size distribution.By rearranging the aggregates and granules, volume change alters the pore size distribution.
GeoHazards 20202018,, 1, x FOR PEER REVIEW 14 of 1617
(a) ink bottle effect (b) contact angle effect
Figure 15. Potential mechanisms for hysteresis [41]. Figure 15. Potential mechanisms for hysteresis [41]. In the wetting process, the ink-bottle effect leads to a delay that occurs at the residual state. ConsideringIn the wetting that the process, median thecurve ink-bottle part is the effect mo leadsst significant to a delay part that of SWRC, occurs and at the because residual the state. ink- Consideringbottle effect occurs that the mostly median at curve high partmatric is thesuction, most significantthis factor partis often of SWRC, considered and because negligible. the ink-bottleSwelling effectand shrinking occurs mostly effects at should high matricbe worth suction, emphasizing this factor for isclayey often soils, considered but can negligible. be neglected Swelling for typical and shrinkinggranular soils effects with should low plasticity. be worth emphasizingThe air entrapped for clayey effect soils, in the but wetting can be neglectedprocess can for also typical be neglected granular soils[42]. The with effect low plasticity.of the contact The angle air entrapped is the most effect significant in the wettingfactor of process hysteresis can for also many be neglected types of soils, [42]. Theparticularly effect of for the soils contact with angle a relatively is the most coarse significant texture. factor This ofassumption hysteresis foris in many keeping types with of soils, the particularlydiscoveries of for Wang soils with et al. a [43], relatively Likos coarse and Lu texture. [44], Moseley This assumption and Dhir is [45], in keeping and Ishakoglu with the and discoveries Baytas of[46]. Wang et al. [43], Likos and Lu [44], Moseley and Dhir [45], and Ishakoglu and Baytas [46]. Some ofof thethe hysteresishysteresis phenomenonphenomenon may may be be due due to to measurement measurement artifacts. artifacts. For For example, example, itmay it may be duebe due to differencesto differences between between tension-induced tension-induced and and pressure-induced pressure-induced desaturation desaturation [[47].47]. AA potentially important aspect of desorption methods under tension is liquid displacement (drainage) even in the absence of a continuous gaseous phase due to cavitationcavitation initiated by entrapped gas bubbles or the vapor pressure of thethe liquidliquid [[48]48].. Impurities and surface heterogeneity in in soil and rock water are conducive to lowering the cavitation tension threshold.threshold. 4.4. Interpretation for Rainfall Induced Slope Failures 4.4. Interpretation for Rainfall Induced Slope Failures Most slope failures are caused by rainfall infiltration [49–52]. Before the rain, the soil near the Most slope failures are caused by rainfall infiltration [49–52]. Before the rain, the soil near the slope is unsaturated. The matrix suction significantly contributes to the shear strength of the soil, slope is unsaturated. The matrix suction significantly contributes to the shear strength of the soil, thereby contributing to the stability of the slope. However, rainwater permeates the soil during the thereby contributing to the stability of the slope. However, rainwater permeates the soil during the rainfall process, thus reducing the matrix suction. In these conditions, the shear strength of the soil rainfall process, thus reducing the matrix suction. In these conditions, the shear strength of the soil may decrease, and the substrate suction may decrease. Therefore, the slope may lose its initial balance. may decrease, and the substrate suction may decrease. Therefore, the slope may lose its initial To examine the impacts of the preliminary relative density (D ), Orense and Farooq [53,54] balance. r employed three materials (silt, gravel, and sand) in the advanced triaxial drainage device. Constant To examine the impacts of the preliminary relative density (Dr), Orense and Farooq [53,54] shear stress is applied to the sample, which indicates the slope condition and the water is injected employed three materials (silt, gravel, and sand) in the advanced triaxial drainage device. Constant to represent the precipitation. The pore water pressure is measured using only a small sensor that shear stress is applied to the sample, which indicates the slope condition and the water is injected to responds to positive pressure. Figure 16 shows how the sand responds to various relative densities. represent the precipitation. The pore water pressure is measured using only a small sensor that With the development of pore water pressure, the specimen did not succeed. After the failure, the responds to positive pressure. Figure 16 shows how the sand responds to various relative densities. loose sample has a faster strain rate, while the dense sample has a low strain rate. The low initial With the development of pore water pressure, the specimen did not succeed. After the failure, the matrix suction of the loose sample can dissipate rapidly by water injection. As a result, loose samples loose sample has a faster strain rate, while the dense sample has a low strain rate. The low initial tend to quickly lower the intensity and start to break. matrix suction of the loose sample can dissipate rapidly by water injection. As a result, loose samples tend to quickly lower the intensity and start to break.
GeoHazards 2020, 1 17 GeoHazards 2018, 1, x FOR PEER REVIEW 15 of 17
Figure 16. Effect of specimen relative density (Drr) during failure under water infiltrationinfiltration [[53].53].
5. Conclusions (1)(1) TheThe soilsoil specimensspecimens having having a drya dry density density revealed revealed decreased decreased preliminary preliminary volumetric volumetric water content. water content.A high density A high typicallydensity typically altered the altered volumetric the volumetric water content water atcontent a protracted at a protracted rate. rate. (2)(2) HigherHigher density density soil soil had had a greater air-entry value and residual matric suction, with a decreased SWRCSWRC slope. slope. (3)(3) TheThe change change of dry density was mainly responsible for large pores. The number of large pores decreaseddecreased as as a consequence of dr dryy density increment. As the dry density increased, the area of macroporesmacropores occupying occupying the the largest portion decrea decreased,sed, while that of mesopores and micropores increases.increases. Areas Areas of of minipores minipores always always accounted accounted for for the the least least portion portion and and were were nearly constant. constant. TheThe proportion proportion of large large diameter diameter pores decrease decreasedd compared with pores that had small diameters inin the the tested soil. The The calculated calculated total total pore volume was lower for soil specimensspecimens possessing a higherhigher dry dry density density in comparison to the relatively loose soils. (4)(4) ThereThere was an observable hysteresis between the dr dryingying and wetting curves for all of the tested soilsoil samples. samples. Hysteresis Hysteresis decreased decreased while while the soil’s dry density increased.
Funding: This research was funded by the China Postdoctoral Science Foundation (No. 2018T110103). Funding: This research was funded by the China Postdoctoral Science Foundation (No. 2018T110103). ConflictsConflicts ofof Interest:Interest: TheThe authorauthor declares declares no no conflicts conflicts of interest. of interest.
References
1. Fredlund, D.G.;D.G.; Rahardjo,Rahardjo, H. H.Soil Soil Mechanics Mechanics for for Unsaturated Unsaturated Soils Soils; John; John Wiley Wiley & Sons: & Sons: Hoboken, Hoboken, NJ, USA, NJ, USA, 1993. 2. Vanapalli,1993. S.K.; Fredlund, D.G.; Pufahl, D.E.; Clifton, A.W. Model for the prediction of shear strength with 2. respectVanapalli, to soilS.K.; suction. Fredlund,Can. D.G.; Geotech. Pufahl, J. 1996 D.E.;, 33 Clifton,, 379–392. A.W. [CrossRef Model for] the prediction of shear strength with 3. Mualem,respect to Y.soil A suction. new model Can. Geotech. for predicting J. 1996, 33 the, 379–392. hydraulic conductivity of unsaturated porous media. 3. WaterMualem, Resour. Y. A Res. new1976 model, 12, for 513–522. predicting [CrossRef the hydrau] lic conductivity of unsaturated porous media. Water 4. Khalili,Resour. Res. N.; 1976 Khabbaz,, 12, 513–522. M.H. A unique relationship of chi for the determination of the shear strength of 4. unsaturatedKhalili, N.; Khabbaz, soils. Geotechnique M.H. A unique1998, 48 relationship, 681–687. [CrossRef of chi for] the determination of the shear strength of 5. Fredlund,unsaturated D.G.; soils. Xing, Geotechnique A.; Fredlund, 1998 M.D.;, 48, 681–687. Barbour, S.L. The relationship of the unsaturated soil shear to the 5. soil-waterFredlund, characteristicD.G.; Xing, A.; curve. Fredlund,Can. Geotech.M.D.; Barbour, J. 1996, 33S.L., 440–448. The relationship [CrossRef of] the unsaturated soil shear to 6. Ajdari,the soil-water M.; Habibagahi, characteristic G.;Ghahramani, curve. Can. Geotech. A. Predicting J. 1996, effective 33, 440–448. stress parameter of unsaturated soils using 6. neuralAjdari, networks.M.; Habibagahi,Comput. G.; Geotech. Ghahramani,2012, 40 A., 89–96.Predicting [CrossRef effective] stress parameter of unsaturated soils using 7. Tavakolineural networks. Dastjerdi, Comput. M.H.; Habibagahi, Geotech. 2012 G.;, 40 Nikooee,, 89–96. E. Effect of confining stress on soil water retention curve 7. andTavakoli its impact Dastjerdi, on the M.H.; shear Habibagahi, strength of G.; unsaturated Nikooee, E. soils. EffectVadose of confining Zone J. 2014 stress, 13 on.[ soilCrossRef water] retention curve and its impact on the shear strength of unsaturated soils. Vadose Zone J. 2014, 13, doi:10.2136/vzj2013.05.0094.
GeoHazards 2020, 1 18
8. Ng, C.W.; Pang, Y.W. Influence of stress state on soil-water characteristics and slope stability. J. Geotech. Geoenviron. Eng. 2000, 126, 157–166. [CrossRef] 9. Wilkinson, G.E.; Klute, A. The temperature effect on the equilibrium energy status of water held by porous media. Soil Sci. Soc. Am. J. 1962, 26, 326–329. [CrossRef] 10. Peck, A.J. Change of moisture tension with temperature and air pressure: Theoretical. Soil Sci. 1960, 89, 303–310. [CrossRef] 11. Chahal, R.S. Effect to temperature and trapped air on matric suction. Soil Sci. 1965, 100, 262–266. [CrossRef] 12. Yang, H.; Rahardjo, H.; Leong, E.C.; Fredlund, D.G. Factors affecting drying and wetting soil-water characteristic curves of sandy soils. Can. Geotech. J. 2004, 41, 908–920. [CrossRef] 13. Zhou, W.H.; Yuen, K.V.; Tan, F. Estimation of soil–water characteristic curve and relative permeability for granular soils with different initial dry densities. Eng. Geol. 2014, 179, 1–9. [CrossRef] 14. Li, B.; Chen, Y.L. Influence of dry density on soil-water retention curve of unsaturated soils and its mechanism based on mercury intrusion porosimetry. Trans. Tianjin Univ. 2016, 22, 268–272. [CrossRef] 15. Rajkai, K.; Kabos, S.; Genuchten, V.; Jansson, P.E. Estimation of water-retention characteristics from the bulk density and particle-size distribution of Swedish soils. Soil Sci. 1996, 161, 832–845. [CrossRef] 16. Tarantino, A. A water retention model for deformable soils. Geotechnique 2009, 59, 751–762. [CrossRef] 17. Gallage, C.P.K.; Uchimura, T. Effects of dry density and grain size distribution on soil–water characteristic curves of sandy soils. Soil Found. 2010, 50, 161–172. [CrossRef] 18. Sheng, D.; Zhou, A.N. Coupling hydraulic with mechanical models for unsaturated soils. Can. Geotech. J. 2011, 48, 826–840. [CrossRef] 19. Ford, E.J. Soil Water Retention Determination Using the ‘Wetlab’ Facility at CSIRO, Davies Laboratory; CSIRO Land and Water: Acton, Australia, 1997. 20. Brooks, R.H.; Corey, A.T. Properties of porous media affecting fluid flow. J. Irrig. Drain. Div. 1966, 92, 61–90. 21. Vanapalli, S.K.; Fredlund, D.G.; Pufahl, D.E. The influence of soil structure and stress history on the soil–water characteristics of a compacted till. Géotechnique 1999, 49, 143–159. [CrossRef] 22. Croney, D.; Coleman, J.D. Soil structure in relation to soil suction (pF). J. Soil Sci. 1954, 5, 75–84. [CrossRef] 23. Mohnke, O. Jointly deriving NMR surface relaxivity and pore size distributions by NMR relaxation experiments on partially desaturated rocks. Water Resour. Res. 2014, 50, 5309–5321. [CrossRef] 24. Aitchison, G.D. Relationship of Moisture and Effective Stress Functions in Unsaturated Soils. Gold. Jubil. Int. Soc. Soil Mech. Found. Eng. 1961, 20, 47–52. 25. Janssen, D.J.; Dempsey, B.J. Soil-moisture properties of subgrade soils. Transp. Res. Rec. 1981, 790, 61–66. 26. Zhao, T.; Wang, J. Soil-water characteristic curve for unsaturated loess soil considering density and wetting-drying cycle effects. J. Central South Univ. Sci. Technol. 2012, 43, 2445–2453. 27. Wu, S.; Gray, D.; Richart, F.E. Capillary effects on dynamic modulus of sands and silts. J. Geotech. Eng. 1984, 110, 1188–1203. [CrossRef] 28. Yu, P.J.; Chen, Y.I. The pore air-water configuration and their effects on the mechanical properties of partially saturated soils. J. Hydraul. Eng. 1965, 1, 16–24. 29. Kohgo, Y.; Asano, I.; Hayashida, Y. An elastoplastic model for unsaturated rockfills and its simulations of laboratory tests. Soils Found. 2007, 47, 919–929. [CrossRef] 30. Zhang, Y.; Ishikawa, T.; Tokoro, T.; Nishimura, T. Influences of degree of saturation and strain rate on strength characteristics of unsaturated granular subbase course material. Transp. Geotech. 2014, 1, 74–89. [CrossRef] 31. Kohgo, Y.; Asano, I.; Hayashida, Y. Mechanical properties of unsaturated low quality rockfills. Soils Found. 2007, 47, 947–959. [CrossRef] 32. Oh, W.; Vanapalli, S. Relationship between Poisson’s Ratio and Soil Suction for Unsaturated Soils. In Proceedings of the 5th Asia-Pacific Conference on Unsaturated Soils, Pattaya, Thailand, 14–16 November 2011. 33. Kaluarachchi, J.J.; Parker, J.C. An efficient finite element method for modeling multiphase flow. Water Resour. Res. 1989, 25, 43–54. [CrossRef] 34. Whitmore, A.P.; Heinen, M. The effect of hysteresis on microbial activity in computer simulation models. Soil Sci. Soc. Am. J. 1999, 63, 1101–1105. [CrossRef] 35. Si, B.C.; Kachanoski, R.G. Unified Solution for Infiltration and Drainage with Hysteresis Theory and Field Test. Soil Sci. Soc. Am. J. 2000, 64, 30–36. [CrossRef] GeoHazards 2020, 1 19
36. Zhang, Q.; Werner, A.D.; Aviyanto, R.F.; Hutson, J.L. Influence of soil moisture hysteresis on the functioning of capillary barriers. Hydrol. Process. 2009, 23, 1369–1375. [CrossRef] 37. Gillham, R.W.; Klute, A.; Heermann, D.F. Hydraulic properties of a porous medium: Measurement and empirical representation. Soil Sci. Soc. Am. J. 1976, 40, 203–207. [CrossRef] 38. Kool, J.B.; Parker, J.C. Development and evaluation of closed-form expressions for hysteretic soil hydraulic properties. Water Resour. Res. 1987, 23, 105–114. [CrossRef] 39. Mitchell, R.J.; Mayer, A.S. The significance of hysteresis in modeling solute transport in unsaturated porous media. Soil Sci. Soc. Am. J. 1998, 62, 1506–1512. [CrossRef] 40. Dullien FA, L. Porous Media: Fluid Transport and Pore Structure; Academic Press: Cambridge, MA, USA, 2012. 41. Tuller, M.; Or, D. Retention of water in soil and the soil water characteristic curve. Encycl. Soils Environ. 2004, 4, 278–289. 42. Maqsoud, A.; Bussière, B.; Aubertin, M.; Mbonimpa, M. Predicting hysteresis of the water retention curve from basic properties of granular soils. Geotech. Geol. Eng. 2012, 30, 1147–1159. [CrossRef] 43. Wang, X.D.; Peng, X.F.; Lu, J.F.; Liu, T.; Wang, B.X. Contact angle hysteresis on rough solid surfaces. Heat Transf. Asian Res. 2004, 33, 201–210. [CrossRef] 44. Likos, W.J.; Lu, N. Hysteresis of capillary cohesion in unsaturated soils. In Proceedings of the 15th ASCE Engineering Mechanics Conference, New York, NY, USA, 2–5 June 2002; Volume 150, pp. 2–5. 45. Moseley, W.A.; Dhir, V.K. Capillary pressure-saturation relations in porous media including the effect of wettability. J. Hydrol. 1996, 178, 33–53. [CrossRef] 46. Ishakoglu, A.; Baytas, A.F. The influence of contact angle on capillary pressure–saturation relations in a porous medium including various liquids. Int. J. Eng. Sci. 2005, 43, 744–755. [CrossRef] 47. Chahal, R.S.; Yong, R.N. Validity of the soil water characteristics determined with the pressurized apparatus. Soil Sci. 1965, 99, 98–103. [CrossRef] 48. Tuller, M.; Or, D. Hydraulic conductivity of variably saturated porous media: Film and corner flow in angular pore space. Water Resour. Res. 2001, 37, 1257–1276. [CrossRef] 49. Lim, T.T.; Rahardjo, H.; Chang, M.F.; Fredlund, D.G. Effect of rainfall on matric suctions in a residual soil slope. Can. Geotech. J. 1996, 33, 618–628. [CrossRef] 50. Ng, C.W.; Zhan, L.T.; Bao, C.G.; Fredlund, D.G.; Gong, B.W. Performance of an unsaturated expansive soil slope subjected to artificial rainfall infiltration. Geotechnique 2003, 53, 143–157. [CrossRef] 51. Huang, M.; Lee Barbour, S.; Elshorbagy, A.; Zettl, D.J.; Si, B.C. Infiltration and drainage processes in multi-layered coarse soils. Can. J. Soil Sci. 2011, 91, 169–183. [CrossRef] 52. Oh, W.T.; Vanapalli, S.K. Influence of rain infiltration on the stability of compacted soil slopes. Comput. Geotech. 2010, 37, 649–657. [CrossRef] 53. Farooq, K.; Orense, R.; Towhata, I. Response of unsaturated sandy soils under constant shear stress drained condition. Soils Found. 2004, 44, 1–13. [CrossRef] 54. Orense, R.; Farooq, K.; Towhata, I. Deformation behavior of sandy slopes during rainwater infiltration. Soils Found. 2004, 44, 15–30. [CrossRef]
© 2018 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).