Obtaining unsaturated soil properties for high volume change oil ...

Unsaturated Soils: Theory and Practice 2011Jotisankasa, Sawangsuriya, Soralump and Mairaing (Editors)Kasetsart University, Thailand, ISBN 978-616-7522-77-7Obtaining Unsaturated Soil Properties for High Volume Change OilSands MaterialsObtaining unsaturated soil properties for high volume change oil sandsDelwyn materials G. Fredlund, Jeff Stone and Jason StiansonGolder Associates Ltd., 1721 – 8th Street East, Saskatoon, SK., Canada, S7H 0T4,Del_Fredlund@golder.com, Jeff_Stone@golder.com, Jason_Stianson@golder.comAndrea D.G. Fredlund, Sedgwick J. Stone & J. StiansonTotal Golder E&P Associates Canada, Ltd., 2900, 1721 240-4th – 8th Street Avenue, East, SW Saskatoon, Calgary, SK., Alberta Canada, T2P 4H4, S7H 0T4, Andrea.Sedgwick@total.comDel_Fredlund@golder.com, Jeff_Stone@golder.com, Jason_Stianson@golder.comA. SedgwickTotal E&P Canada, 2900, 240-4th Avenue, SW Calgary, Alberta T2P 4H4, Andrea.Sedgwick@total.comABSTRACT: This paper describes the manner in which conventional soil-water characteristic curve testsconducted on high volume change material can be used in conjunction with independently measuredshrinkage curves to provide the unsaturated soil property functions required for saturated-unsaturatednumerical modeling of the drying process. The laboratory test procedure used to measure the shrinkage curveis described along with the interpretation of the shrinkage curve results. A regression curve-fitting procedureis used to obtain a closed-form equation for the shrinkage curve. These results are combined with an equationfor the soil-water characteristic curve to identify the correct air entry value and residual conditions for thematerial. The paper also describes the manner in which each of the unsaturated soil properties can be used forthe numerical simulation of the drying process.KEYWORDS: soil-water characteristic curve, volume change, tailings1 INTRODUCTIONMany of the estimation procedures used tocharacterize unsaturated soil property functions arebased on the assumption that the soil will undergo asmall amount of volume change as soil suction isincreased. While this assumption may be acceptablefor sands and coarse-grained materials, it is notacceptable for fine-grained silts and clays,particularly those that are deposited as slurry andthen left to dry and increase in strength. Projectsassociated with the mining of Oil Sands in Alberta,Canada have tailings that are initially in a slurryform at high water content. The engineeringchallenge involves converting the tailings into amaterial with sufficient strength for trafficability.Thin lift deposition is a potential solution, andinvolves tailings deposition in thin lifts that aresubsequently allowed to dry and gain strength. Thefine-grained tailings undergo considerable volumechange as soil suction is increased during the dryingprocess.Geotechnical engineers are requested to undertakenumerical model simulations of the drying ofthe initially wet tailings material. The tailings aregenerally exposed to random weather conditions;however, there will be a drying of the tailings withtime since the climate is semi-arid in the FortMcMurray region. Questions related to the rate ofdrying and the thickness at which the tailings can bedeposited can be addressed as a transient saturatedunsaturatedseepage problem (SoilVision, 2010).The required saturated and unsaturated soilproperties take on the form of nonlinear functionswhich can be estimated from measured watercontent versus soil suction relationships (i.e., soilwatercharacteristic curves, SWCC). It is importantthat the soil property functions be properlyquantified for usage as input to numerical modelsimulations.The objective of this paper is to describe themanner in which conventional soil-watercharacteristic curve tests can be conducted on highvolume change material and used in conjunctionwith independently measured shrinkage curves toprovide the unsaturated soil property functionsneeded for numerical simulations of the dryingprocess. The laboratory test procedure used tomeasure, interpret and apply the shrinkage curve isdescribed in this paper.A regression curve-fitting procedure is used toobtain a closed-form equation for the shrinkagecurve. The shrinkage curve results are thencombined with a conventional soil-watercharacteristic curve to identify the correct air entryvalue and residual conditions for the material. TheSWCC information is then used to calculate theunsaturated soil property functions for the tailingsmaterial.415

2 ROLE OF THE SOIL-WATERCHARACTERISTIC CURVE, SWCCThe SWCC shows the relationship between theamount of water in a soil and various applied soilsuctions (Fredlund and Rahardjo, 1993). There aretwo primary reference points on the SWCC; namely,the air-entry value and the residual suction. Theapparent air-entry value for a soil is dependent onhow the amount of water in the soil is quantified.When a soil undergoes volume change as soilsuction is increased, the air-entry value appears tooccur at different values. The differing air-entryvalues are related to how the amount of water in thesoil is defined. Each variable to designate theamount of water in the soil has significance but it isimportant that the correct interpretations be appliedat various stages of the analysis (Fredlund, 2002).The variables used to quantify the amount ofwater in the soil are:i) gravimetric water content, w;ii) volumetric water content, θ; with theinstantaneous total volume used in calculatingvolumetric water content;iii) volumetric water content, θ, with the volume ofwater, V w , referenced to original total volume ofthe specimen, V o , (i.e., V w / V o ); andiv) degree of saturation, S.Each of the above designation for the amount ofwater in the soil can be used to plot a soil-watercharacteristic curve. Each of the SWCCs wouldprovide similar information to the geotechnicalengineer provided the soil does not undergo volumechange as soil suction is increased. However, whenthe soil undergoes volume change, as is the caseunder consideration, then the geotechnical engineermust be able to plot each of the SWCCs and use theappropriate curves in the proper manner forevaluating unsaturated soil property functions to beused in performing numerical simulations ofphysical processes.Gravimetric water content is the most basicmeasurement of the amount of water in a soilbecause it requires only the measurement of mass.Volumetric water content, θ, has commonly beenused in agriculture-related disciplines and is definedas the amount of water in the soil referenced to theinstantaneous total volume of the soil specimen. Thetotal volume of the soil specimen must be knownwhen computing the volumetric water content in thismanner. Therefore, it is necessary to know the totalvolume of the soil specimen corresponding toequilibrium soil suction conditions. When modelingwater seepage through a soil, it is the change involumetric water content that defines the change inwater storage in the soil as suction changes.However, there are two possible ways to computevolumetric water content. If the overall volume ofthe soil changes by a small amount during theincrease in soil suction, then either designation ofvolumetric water content can be used. However, ifthe volume changes are substantial, then theinstantaneous total volume should be used for thecalculation of the water storage function.The degree of saturation, S, references thevolume of water in the soil to the instantaneousvolume of voids and therefore also requires the totalvolume of the specimen. When a soil undergoesvolume change as soil suction increases, then the airentryvalue and residual conditions need to beevaluated from the plot of degree of saturationversus soil suction. In addition, it is the degree ofsaturation versus soil suction plot that must be usedduring the integration process to calculate thepermeability function. The permeability functionmay also need to be computed in a different mannerto accommodate changes in void ratio prior toreaching the air-entry value of the soil.It is important to note that the unsaturated soilproperty functions need to be computed in distinctlydifferent manners when the soil undergoessubstantial volume change as soil suction increases.It is the instantaneous volumetric water contentdesignation that must be used to compute waterstorage and the degree of saturation designation thatmust be used to define the air-entry value of the soil.The degree of saturation designation plays the majorrole involved in estimating the unsaturated hydraulicconductivity function of the soil.The soils of primary concern in this paper are thehigh water content tailings that are a by-productfrom Oil Sands extraction. These materials may startwith a natural water content that is well above theliquid limit of the material and undergo a largevolume change upon drying.The difficulties associated with the testing ofsoils that undergo large volume changes withincreasing soil suctions are not a recent discovery.Fredlund (1964) showed that shrinkage curvemeasurements needed to be used in conjunction witha conventional measurement of the SWCC in orderto properly interpret the unsaturated soil behaviourof high volume change soils. Fredlund (loc. cit.)performed a series of tests on highly plastic Reginaclay that had a liquid limit of 75, a plastic limit of25, and had 50% clay size particles.3 SHRINKAGE LIMIT AND THE SHRINKAGECURVEThe shrinkage limit of a soil has been one of theclassification properties since the inception of soilmechanics (ASTM D427). Mercury immersion wasoriginally used for the measurement of the volume416

of the soil specimen. This technique is no longerconsidered acceptable in most countries.The shrinkage limit is defined as the watercontent corresponding to the minimum volume that asoil can attain upon drying to zero water content. Itis a fictitious water content that generally fallsslightly below the plastic limit of the soil.It is the entire shrinkage curve from an initiallysaturated soil condition to completely dry conditionsthat is called the shrinkage curve and is significantfor the interpretation of SWCC data. Figure 1 showsthe drying curve for an initially saturated soil. As aclay soil dries, a point is reached where the soilstarts to desaturate. This point is generally quiteclose to the plastic limit of the soil. Consequently,there is an approximate correlation between theplastic limit of a soil and its air-entry value. Uponfurther drying, another point is reached where thesoil dries without any further change in overallvolume. This can be referred to as the true“shrinkage limit” of the soil and the gravimetricwater content appears to approximately correlatewith the residual soil conditions.Void Ratio3.02.52.01.51.00.50.0ResidualWater ContentPlasticLimitShrinkage LimitLiquid LimitSaturation Line0% 20% 40% 60% 80% 100%Gravimetric Water ContentFigure 1. Shrinkage curve with its relationship to the AtterbergLimit classification properties.3.1 Measurement of the shrinkage curveAn experimental procedure was developed for themeasurement of the entire shrinkage curve frominitial high water content conditions to completelydry conditions. A digital micrometer was used forthe measurement of the volume at various stages ofdrying as shown in Figure 2. Brass rings weremachined to contain the soil specimens (i.e., therings have no bottom). The rings with the soil wereplaced onto wax paper and drying was allowed. Thedimensions of the soil specimens were selected suchthat cracking of the soil was unlikely to occur duringthe drying process. The initial dimensions selectedfor the shrinkage curve specimens were a diameterof 3.7 cm and a thickness of 1.2 cm.Figure 2. Digital micrometer for the measurement of thediameter and thickness of shrinkage specimens.The mass and volume of each soil specimen weremeasured on a daily basis. Four to six measurementsof the diameter and thickness of the specimen weremade at differing locations on the specimens. Figure3 shows typical measurements of water content andvoid ratio as the soil dried. It was observed that asthe specimen diameter began to decrease, with thespecimen pulling away from the brass ring, the rateof evaporation increased significantly (i.e., abouttwice as fast).The increase in the evaporation rate was relatedto the increased surface area from which evaporationwas occurring. Consequently, it is recommendedthat the measurements of mass and volume shouldbe increased to once every two to three hours oncethe soil shows signs of pulling away from the sidesof the ring.Void Ratio2.52.01.51.00.50.0Plastic LimitLiquid Limit0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%Gravimetric Water ContentBox #11 - S1Box #11 - S2Box #6 - S1Box #6 - S2Aged TT - S1Aged TT - S2Aged TT - S3100% SaturationBest FitFigure 3. Typical shrinkage curve for an oil sand clay soil.3.2 Equation for the Shrinkage CurveThe shrinkage curve has the form of a hyperboliccurve. Fredlund et al. (1997, 2002) proposed anequation to best-fit data for the shrinkage curve. Theequation has parameters with physical meaning andis of the following form:417

⎛ 1 ⎞⎜ ⎟⎝ csh⎠c⎡ shw ⎤e( w)= ash⎢ + 1c ⎥(1)sh⎣bsh⎦where: a sh = the minimum void ratio (e min ), b sh =slope of the line of tangency, (e.g., drying fromsaturated conditions), c sh = curvature of the shrinkagecurve, and w = gravimetric water content. Theratio,ash Gs= (2)b Sshis a constant for a specific soil; G s is the specificgravity and S is the degree of saturation.Once the minimum void ratio of the soil isknown, it is possible to estimate the remainingparameters required for the designation of theshrinkage curve. The minimum void ratio the soilcan attain is defined by the variable, a sh . The c shparameter provides the remaining shape of theshrinkage curve. The curvature of the shrinkagecurve is controlled by varying the c sh parameter.3.3 Integration of the Shrinkage Curve and thelaboratory measured SWCCA laboratory measured SWCC describes therelationship between gravimetric water content andsoil suction. The Fredlund and Xing (1994) equation(Equation 3) can be used to best-fit the SWCCwhich can then be combined with the shrinkagecurve.⎡⎤⎡ ⎛ ψ ⎞ ⎤ ⎢ ⎥⎢ ln ⎜1+⎟ ⎥ ⎢⎥hr⎢ 1 ⎥w( ψ ) = w ⎢s1− ⎝ ⎠ ⎥6m⎢f⎛ 10 ⎞ ⎥ ⎢ ⎥n fln 1⎡ ⎡ψ⎤⎤⎢ ⎜ + ⎟ ⎥ ⎢⎛ ⎞ ⎥⎢ h ⎢ln ⎢exp(1)⎥⎥⎣ ⎝+r ⎠ ⎥⎦ ⎢⎜ ⎥a ⎟⎢ ⎢ ⎢⎥f ⎥⎣ ⎣ ⎝ ⎠ ⎦⎦⎥⎣⎦(3)where: w(ψ) = gravimetric water content at anyspecified suction, ψ; w s = saturated gravimetricwater content; h r = residual soil suction; a f , n f , and m f= the fitting parameters for the Fredlund and Xing(1994) SWCC equation. Equation 3 is written for thegravimetric water content designation; however, itshould be noted that it can be best-fit to any of thedesignations of water content versus soil suctionbecause of the flexibility of the equation with threefitting soil parameters.It is possible to compute the degree of saturationversus soil suction as well as any other designationfor the amount of water in the soil by combiningEquations 1 and 3.4 MEASUREMENT AND INTERPRETATIONOF OIL SANDS TAILINGS RESULTSSoil-water characteristic curves were measured onsamples of Oil Sands tailings that were thickenedusing a pilot deep cone thickener. The tailings werecomprised of fines (< 44 microns) and 10% and 45%greater than 44 microns sand (i.e., 0.1 and 0.8 sandto fines ratios, SFR, respectively). The tailings werefrom development studies on Oil Sands tailings fromnorthern Alberta, Canada. The shrinkage curveresults were presented in Figure 3. The 0.1 SFRtailings have plastic and liquid limits of 30 and 55,respectively. The 0.8 SFR tailings have plastic andliquid limits of 15 and 38, respectively.Approximately 60% of the material classifies as claysize particles. The slurry material has a gravimetricwater content of about 100%. The intent is to depositthe thickened tailings material in lifts of varyingthicknesses that are allowed to dry.As water is removed from the thickened finetailings, the volume of the material decreases andthere is a slow increase in shear strength. As thematerial begins to desaturate near the plastic limitthere is a substantial increase in shear strength.Figure 4 shows the gravimetric water content, w,plotted versus soil suction for two samples testedwith each of the sand to fines ratios (SFR). Box #11had a SFR of 0.8 and a starting gravimetric watercontent of about 70%. Box #6 was the samematerial; however, it was dried to about 25% beforethe SWCC test was performed. Box #2 had a SFR of0.1 and the initial gravimetric water content wasnear 70%. Box #5 was the same material; however,it was dried to about 47% before the SWCC wasperformed.The initially wet samples show a break in thecurvature in the region around 1 kPa. The driedsamples show a break in the curvature around10 kPa. However, the curvature is not distinct and itshould not be taken to represent an air-entry valuefor the material.It is necessary to use the shrinkage curve resultswhen interpreting the meaning of the SWCC results.A best-fit shrinkage curve equation can be combinedwith the Fredlund and Xing (1994) equation for theSWCC. It is then possible to plot the degree ofsaturation, S, versus soil suction as shown in Figure5.The results show that there is a distinct air-entryvalue for the 0.8 SFR material at about 100 kPa. Theair-entry value for the 0.1 SFR material is about1,000 kPa. These are the true air-entry values forthese two materials and the degree of saturationSWCCs are the curves that must be used for theestimation of the unsaturated hydraulic conductivityfunction. The residual condition can also be clearlyidentified as having a residual suction of about3,000 kPa and a residual degree of saturation of 20%for the 0.8 SFR tailings. The 0.1 SFR tailings have aresidual suction of about 15,000 kPa and a residualdegree of saturation of about 20%.418

Water Content100%90%80%70%60%50%40%30%20%10%0%GravimetricVolumetric (instantaneous)Volumetric (based on Vo)Degree of SaturationFitted curves0.01 0.1 1 10 100 1000 10000 100000 1000000Suction (kPa)Figure 8. Gravimetric water content, volumetric water content(based on both instantaneous and initial volumes) and degree ofsaturation versus soil suction for Box #5 tailings.100%90%80%70%60%50%40%30%20%10%0%Degree of Saturation5 CONCLUSIONSThe results clearly show that it is imperative tomeasure the shrinkage curve when testing highvolume change soils for their unsaturated soilproperty functions. It is also necessary to combinethe shrinkage curve and the gravimetric watercontent SWCCs in order to calculate all of thevolume-mass properties versus soil suction. The(instantaneous) volumetric water content SWCC isnecessary for computing the water storagecharacteristics of the thickened Oil Sands tailings.The degree of saturation SWCC is necessary forobtaining the correct physical features of the SWCC(i.e., air-entry value and residual conditions), and thesubsequent unsaturated permeability function for thethickened Oil Sands tailings.Water Content100%90%80%70%60%50%40%30%20%10%0%0.01 0.1 1 10 100 1000 10000 100000 1000000Suction (kPa)GravimetricVolumetric (instantaneous)Volumetric (based on Vo)Degree of SaturationFitted curvesFigure 9. Gravimetric water content, volumetric water content(based on both instantaneous and initial volumes) and degree ofsaturation versus soil suction for Box #11 tailings.Void Ratio2.01.81.61.41.21.00.80.60.40.20.0100%0.01 0.1 1 10 100 1000 10000 100000 1000000Suction (kPa)Box #11Box #6Box #2Box #5Box #11 fitBox #6 fitBox #2 fitBox #5 fit90%80%70%60%50%40%30%20%10%0%Degree of SaturationREFERENCESASTM – D427, 1988 Annual Book of ASTM Standards, Vol.4.08, Soil and Rock, Building Stones; Geotextiles.Fredlund, D. G., 1964. Comparison of soil suction and onedimensionalconsolidation characteristics of a highly plasticclay. National Research Council of Canada, Division ofBuilding Research, Technical Report No. 245, Ottawa,Ontario, Canada.Fredlund, D.G. 2002. Use of soil-water characteristic curve inthe implementation of unsaturated soil mechanics. UNSAT2002, Proceedings of the Third International Conferenceon Unsaturated Soils, Recife, Brazil, March 10-13, pp. 887-904.Fredlund, M.D., Wilson, G.W., and Fredlund, D.G. 2002.Representation and estimation of the shrinkage curve.UNSAT 2002, Proceedings of the Third InternationalConference on Unsaturated Soils, Recife, Brazil, March 10-13, pp. 145-149.Fredlund, D. G., and Rahardjo, H. 1993. Soil mechanics forunsaturated soils. New York: John Wiley & Sons.Fredlund, M. D., Fredlund, D. G. and Wilson, G. W., 1997.Prediction of the soil-water characteristic curve from grainsize distribution and volume-mass properties. InProceedings of the Third Brazilian Symposium onUnsaturated Soils, NONSAT ’97, Rio de Janeiro, Brazil,April 22-25, 1: 13-23.Fredlund, D. G. and Xing, A. 1994. Equations for the soilwatercharacteristic curve. Canadian Geotechnical Journal,31(3): 521-532.Lambe, T. W. 1951. Soil Testing for Engineers. New York:John Wiley and Sons.SoilVision Systems (2010), User’s Manual for SVFlux,Saturated-Unsaturated Numerical Modeling. Saskatoon,Canada.Figure 10. Void ratio versus soil suction plot for the Oil Sandstailings.420