The Influence of the Pore Fluid on Desiccation of a Deformable Porous MaterialH. Péron !cl@@auth, L.B. Hu !cl@@auth, T. Hueckel !cl@@auth, L. LalouiThe Influence of the Pore Fluid on Desiccationof a Deformable Porous MaterialH. P´eron1, L.B. Hu2, T. Hueckel2, and L. Laloui11Soil Mechanics L aboratory, Ecole Polytechnique F´ed´erale de Lausanne – EPFL,Station 18, CH-1015 Lausanne, [email protected], [email protected] of Civil and Environmental Engineering, Duke University, Durham,N.C. 27706, U.S.A. [email protected], [email protected]. This paper aims at elucidating the influence of pore fluid propertieson the shrinkage strains of deformable materials subjected to drying. Results ofisothermal drying tests of two kinds of silts saturated with three different pore fluidsare presented. Results show different strain amounts during drying and the final voidratios, depending on pore fluid. The rates of drying are consistent with saturatedvap or pressures, while their amount with the surface tension. The shrinkage limitapp e ars also to be controlled by a limitation in skeleton compressibility.IntroductionThe study of deformation due to drying is fundamental in the understanding ofthe patterns of stress generated by the constrained shrinkage and cracking insoils. When drying shrinkage is constrained, reaction forces arising located atthe constraints generate stresses, which depend on the amount of shrinkage.Our previous papers (P´eron et al. 2006, Hu e t al. 2006) revealed how thedevelopment of such stress in the elastic domain leads to soil cracking.Notably, seve ral previous studies established that most of the desiccationshrinkage develops when soil is saturated (Croney and Coleman 1954, Fleureauet al. 1993). The state of full saturation when cracking initiates was hypoth-esized by several authors (Konrad and Ayad 1997, Abu-Hejleh and Znidarcic1995, Lloret et al. 1998). Recently, P´eron et al. (2006) showed that cracksoccur at the end of saturation range or at the early stage of desaturation.The goal in this work is to investigate a possible relationship betweenthe total amount of shrinkage, its rate and the mechanisms of liquid removaland deformation. In particular, the role of the evaporation flux versus surfacetension, in addition to soil compressibility and transport properties are of in-terest. Slow, constant humidity environment, (isothermal) air-drying of thinand long rectangular soil slabs was carried out. Two different soil types werePage: 1 job: Peron macro: svmult.cls date/time: 26-Sep-2006/9:002 H. P´eron, L.B. Hu, T. Hueckel, and L. Lalouiused: clayey silt, and a powder of rock. The soils were wetted with three dif-ferent pore fluids, characterized by distinctly different surface tension, latentheat, saturation vapor pressure and viscosity.Desiccation TestsDesiccation Tests ProcedureThe silts employed were: a clayey silt (referred to as silt A), and a powderof granite (silt B). They were saturated with three fluids: tap water, andtwo alcohols: ethanol/water mixture (with a volume ratio of 50%/50%), andethylene-glycol/water mixture (65%/35%) with the properties shown in Ta-ble 1. The two alcohols were used in mixtures with water to avoid flocculation,which is especially acute in pure ethanol. Saturation vapor pressure and latentheat were calculated as for pure mixtures.Table 1. Properties of fluids used for material preparationPure waterEthylene-glycolsolutionEthanolsolutionSpecific Latent Heat [MJ/m3] 2448 1479 1591Saturation Vapor Pressure [Pa] 2337 1639 3253Surface Tension [mN/m] 72.6 56.4a30aViscosity [cm2/s] 0.01 0.0423b0.029bDensity [g/cm3] 1 1.071 0.92aVasquez et al. (1995) J Chem Eng Data 40;Won et al. (1981) J Chem Eng Data 26bFernandez & Quigley (1988) Can Geotech J 25;Hay duk & Malik (1971) J Chem Eng Data 16(2)Two silts were used: a clayey silt (referred as silt A) and granite powder(silt B). Particles greater than 90 µm (sandy fraction) have been removed.In silt A, the clay fraction is 25%; clay minerals are illite (10% of the totalamount of mineral species), smectite (10%) and chlorite (5%). The liquid limitis wL= 31.8%, plastic limit wP= 16.9%, and the unit weight of solid particlesis γs= 27.1 kN/m3. Silt B has low clay content (less than 7%, mainly illiteand chlorite). The liquid limit is wL= 29.5%, the plastic limit wP= 20.1%,and the unit weight of solid particles is γs= 27.7 kN/m3.The tests were performed in a climate chamber with a controlled relativeair humidity and temperature, of 40% and 18◦C, respectively. Strains in threedirections were measured with calipers, which were fixed on the support. Av-erage fluid content of the whole cake was recorded versus time by continuallyPage: 2 job: Peron macro: svmult.cls date/time: 26-Sep-2006/9:00The Influence of the Pore Fluid on Desiccation . . . 3weighing an identical cake , dried in the same time in identical conditions as theone used for strain measurement. Details of the tests are reported elsewhere(P´eron et al. 2006, Hu et al. 2007).Density changes (due to the fact that the more volatile fluid of the mixtureevaporates faster) were tested for the calculation of void ratio evolution duringdrying, as well as for the fluid properties dete rmination.Desiccation Test ResultsThe main result is the relationship between the void ratio evolution and thevolumetric fluid content for both materials plotted in Fig. 1A and B. In thesegraphs the thin dotted line denotes the theoretical void ratio evolution underthe assumption that the material remains wate r saturated at constant volumeof solids. It is noted that at the beginning of drying, all the curves do notmuch deviate from this line. However, well before all fluid is evaporated, theshrinkage stopped. The volumetric water content value at that point identifiedas the shrinkage limit tends to depend on the pore fluid. Void ratio stabilizesfinally, reaching a “limiting void ratio.” The curves obtained could be concep-tualized as piecewise bilinear, with the first part merging with the saturationline, and a second part with void ratio equal to a constant “limiting voidratio.”For either silt, water involves the highest amount shrinkage and ethanolthe lowest. Ethylene-glycol induces intermediate values, closer to water in thecase of silt B, and closer to ethanol in the case of silt A. Shrinkage in the caseof silt A was much more significant for all three liquids, than in the case ofsilt B. The values of the limiting