Study of the variation of thermal conductivity with water saturation using nuclear magnetic resonance

In this paper, we present a method that allows one to correct thermal conductivity measurements for the effect of water loss when extrapolating laboratory data to in situ conditions. The water loss in shales and unconsolidated rocks is a serious problem that can introduce errors in the characterization of reservoirs. For this study, we measure the thermal conductivity of four sandstones with and without clay minerals according to different water saturation levels using an optical scanner. Thermal conductivity does not decrease linearly with water saturation. At high saturation and very low saturation, thermal conductivity decreases more quickly because of spontaneous liquid displacement and capillarity effects. Apart from these two effects, thermal conductivity decreases quasi-linearly. We also notice that the samples containing clay minerals are not completely drained, and thermal conductivity reaches a minimum value. In order to fit the variation of thermal conductivity with the water saturation as a whole, we used modified models commonly presented in thermal conductivity studies: harmonic and arithmetic mean and geometric models. These models take into account different types of porosity, especially those attributable to the abundance of clay, using measurements obtained from nuclear magnetic resonance (NMR). For argillaceous sandstones, a modified arithmetic-harmonic model fits the data best. For clean quartz sandstones under low water saturation, the closest fit to the data is obtained with the modified arithmetic-harmonic model, while for high water saturation, a modified geometric mean model proves to be the best.