Comparison of Laboratory Methods for Measuring Thermal Conductivity of Unsaturated Soils

Abstract

Experiments were conducted to explore three different laboratory approaches for determining the relationship between thermal conductivity (λ) and saturation (S) ("thermal dryout curves") for unsaturated coarse-grained porous media. These include: (i) a single-sample approach involving evaporation from a sample with an embedded thermal conductivity probe; (ii) a multiple-sample approach involving subsamples compacted to various saturations; and (iii) an instrumented Tempe cell approach affording concurrent measurement of λ(S) and the soil-water characteristic curve (SWCC). Dryout curves were obtained using each approach for F-75 Ottawa sand, a poorly-graded river sand, and a mixture of spherical glass beads. Conductivity sharply decreased at a critical saturation between 0.05 and 0.15 for all three materials. The single-sample approach required the longest time to produce a dryout curve (~500 hours) and resulted in λ values ~10% to 20% higher than the other approaches, an observation attributed to the presence of a sharp drying front associated with the evaporation testing procedures. The multiple-sample approach required the least amount of time (~15 hours) but produced a relatively sparse data set and is limited by potential errors associated with variability in sample preparation. The instrumented Tempe cell approach required a moderate amount of time (~123 hours) but resulted in the most robust λ(S) function and clearly defined thermal regimes. A clear advantage of the Tempe cell approach is that the SWCC may also be obtained, as often required for modeling coupled heat and moisture transport in many geotechnical applications.