Experimental Determination and Theoretical Prediction of Effective Thermal Conductivity of Porous Media

The objectives of this study are two-fold: a) to develop an experimental method for the accurate determination of thermal conductivity of fluid saturated porous media and b) to predict the experimental results through microscopic scale modeling. A steady-state apparatus has been designed, tested, and implemented for obtaining thermal conductivity with improved accuracy by minimizing heat losses. The apparatus is capable of measuring thermal conductivity at various fluid saturation conditions. The steady-state apparatus has also been modified for transient measurements to obtain thermal diffusivity. The thermal conductivity of the consolidated rocks show strong sensitivity to the type of saturating fluid. An attempt has been taken to construct a 3-D numerical model of a porous medium based on available 2-D information such as a thin-section by using indicator simulation algorithm. Results from this study indicates that the consolidated porous media is too complex to be represented by a few statistics. A recursive model based on the concept of real space renormalization group has been developed to predict macroscopic properties by using the parameters characterizing the microscopic details. The macroscopic thermal conductivity was found to be relatively insensitive to the heterogeneity in the isotropically correlated systems even at large conductivity ratios of the individual components. The macroscopic thermal conductivity is a strong function of the aspect ratio describing heterogeneity in the anisotropically correlated systems even at small conductivity ratios of the components. A random walk model has been developed to calculate conductivity of extremely large systems with small correlation structures and with any number of components. A qualitative study on the effect of multiphase fluid saturation on the effective thermal conductivity has been conducted by using a thin section. Multiphase fluid saturation has been numerically generated by using Monte Carlo annealing algorithm. The effect of fluid saturation is not monotonic as has been commonly believed so far. Comparison between experimental results and the results from the numerical studies using spherepacks revealed that convective heat transfer may be a much stronger problem in experimental methods than ordinarily believed.