Measurement And Prediction Of Hydraulic And Thermal Properties Of Frozen Soils:Thermo-Time Domain Reflectometry Technology

The hydraulic and thermal properties of frozen soils affect ground-coupled heat and moisture transfer process in cold regions and therefore are important for both engineering and environmental issues. For thermal property measurement, conventional steady state method is unsuitable for field application and heat pulse method performs poorly due to ice melting caused by the heating operation. Thermal property models are common in numerical modeling of frozen soils, but there exists a need for a model that predicts thermal conductivity accurately in both unfrozen and frozen soils. It has challenge to determine ice content in frozen soils because all existing approaches have intrinsic limitations. In addition, the measurement of soil freezing characteristic curve (SFC) is usually affected by supercooling phenomenon. The thermo-time domain reflectometry (Thermo-TDR) probe, which combines heat-pulse and TDR sensors into one unit, can be used to obtain SFC and has the potential to measure ice content in frozen soils. The first objective of this study was to identify an optimum heat application strategy for measuring soil thermal properties with Thermo-TDR probe in frozen soils while minimizing ice melting during the process. The optimized heating schemes were then applied for measuring thermal conductivity of frozen soils, monitoring soil ice content dynamics during freezing and thawing, and developing a simplified thermal conductivity model for frozen soils. Our second objective was to improve the measurement approach of SFC using Thermo-TDR technique and to evaluate the applicability of SFC for estimating other soil physical properties. The following lists the major findings of this study.First, we evaluated the performance of the Thermo-TDR method under various heating schemes consisting of three heat strengths and three heat-pulse durations. We demonstrated that when soil temperature was<-5℃, ice melting due to heat-pulse release could be minimized experimentally with a combination of 60-s heating duration and 450 J m-1 heating strength, or a 90-s heating duration with heating strength in the range of 450 to 600 J m-1.Second, we developed a simplified de Vries-based model for estimating thermal conductivity of both unfrozen and frozen soils. The simplified model follows the basic assumptions of the de Vries model, but simplifies or improves the estimations of model parameters. The model was validated using measured thermal conductivity results and data obtained from the literature. The results demonstrated that the simplified de Vries-based model provided accurate and consistent thermal conductivity predictions, and it performed better than other existing models in both unfrozen and frozen soils.Third, with the optimized heating scheme, we were able to estimate soil ice content changes during freezing and thawing based on Thermo-TDR measurements of volumetric heat capacity, liquid water content, and bulk density. Laboratory evaluations on three soils with different textures and initial water contents showed that the errors of Thermo-TDR measured ice content at temperatures<-5℃ were within ±0.05 m3 m-3 in sandy loam soils. The measurement errors became larger in soils with high clay contents (within ±0.1 m3 m-3).Fourth, the Thermo-TDR technique could be applied to determine ice contents in frozen soils by inverse calculation using the simplified de Vries-based model for thermal conductivity. Sensitivity analyses of liquid water content, bulk density, and thermal property results from the heat capacity-based method and the thermal conductivity-based method were performed using a hypothetical soil. The analysis indicated that the heat capacity-based method was more sensitive to the errors in liquid water content and other terms than thermal conductivity-based method. Thus the thermal conductivity-based method can provide more accurate ice content estimates. Experimental results of three soils agreed with the theory analysis.Fifth, we introduced a toothpick ice-seeding method to avoid supercooling during SFC measurement with the Thermo-TDR method, and compared SFC data measured in equilibrium systems versus those in dynamic systems at various freezing rates. Large errors in SFC measurement were observed in dynamic systems at fast freezing rate (10℃ h-1). A slow freezing rate (2.5℃ h-1) could provide accurate SFC data.Finally, we evaluated several empirical SFC models with measured SFC data on five soils and showed that the piecewise exponential model provided the best fits to measured values. Furthermore, the measured SFC data on five soils were applied to estimate soil moisture characteristic curves (SMC) and soil specific surface area. The dual porosity SMC model was fitted to sand box measurements and SFC data. The fitted curves agreed well with those obtained using sand box, pressure plate, and dew point method, with correlation coefficients ranged from 0.948 to 0.985, except for a sandy loam soil. The estimated soil specific surface areas using SFC data also agreed well with those obtained using dew point technique (with a correlation coefficient of 0.985).