Effects of density and viscosity in modeling heat as a groundwater tracer

Decoupled simulation of groundwater flow and heat transport assuming constant fluid density and viscosity is computationally efficient and simple. However, by neglecting the effects of variable density and viscosity, numerical solution of heat transport may be inaccurate. This study investigates the conditions under which the density and viscosity effects on heat transport modeling can be neglected without any significant loss of computational accuracy. A cross-section model of aquifer-river interactions at the Hanford 300 Area in Washington State was employed as the reference frame to quantify the role of fluid density and viscosity in heat transport modeling. This was achieved by comparing the differences in simulated temperature distributions with and without considering variable density and viscosity, respectively. The differences between the two sets of simulations were found to be minor under the complex field conditions at the Hanford 300A site. Based on the same model setup but under different prescribed temperature gradients across the simulation domain, a series of heat transport scenarios were further examined. When the maximum temperature difference across the simulation domain is within 15 degrees C, the mean discrepancy between the simulated temperature distributions with and without considering the effects of variable density and viscosity is approximately 2.5% with a correlation coefficient of above 0.8. Meanwhile, the speedup in runtime is roughly 225% when the maximum temperature difference is at 15 degrees C. This work provides some quantitative guidelines for when heat transport may be simulated by assuming constant density and viscosity as a reasonable compromise between accuracy and efficiency.