A Systematic Approach to Local Grid Refinement in Geothermal Reservoir Simulation

Abstract

Thermodynamic conditions of geothermal reservoir fluids, such as pressures. Temperatures, phase compositions, and concentrations of dissolved solids and non-condensible gases, may show strong spatial variability in the vicinity of production and injection wells. Large spatial variations also can occur near reservoir heterogeneities, such as faults and lithologic contacts. In numerical simulation of flow and transport, fine gridding is required to accurately represent steep changes in fluid conditions. This can necessitate prohibitively large numbers of grid blocks in conventional finite difference discretization, where finer gridding is applied globally with reference to a global system of coordinates.

This paper presents a more economical alternative, in which fine gridding is applied only to the regions in which large gradients in thermodynamic conditions need to be resolves. We have implemented a scheme for local grid refinement into the general-purpose geothermal reservoir simulator TOUGH2. We discuss the concept and algorithms used for local grid refinement, as well as applica5ions to production from and injection into fractured two-phase reservoirs. The accuracy and economy of local grid refinement are demonstrated by comparison with globally fine grid simulations.