Upscaling of CO2-Brine Flow with Capillary Heterogeneity Effects

Tuesday, 16 December 2014
Avinoam Rabinovich, Stanford University, Stanford, CA, United States, Kasama Itthisawatpan, Stanford University, Department of Energy Resources Engineering, Stanford, CA, United States and Louis J Durlofsky, Stanford University, Los Altos Hills, CA, United States
Large-scale simulation of CO2 storage operations can be expensive computationally, particularly when the effects of fine-scale capillary pressure heterogeneity are included. These effects, which are often ignored in traditional reservoir simulation, can significantly impact the movement of CO2 and should therefore be included in carbon storage simulations. The application of upscaling techniques that handle capillary heterogeneity could lead to reductions in computational cost, though general and robust procedures for this problem have yet to be presented. In this work, we develop and apply a new upscaling technique for two-phase flow in heterogeneous porous media with capillary heterogeneity effects. Fine-scale capillary pressure is modeled using the Leverett J-function and varies with permeability. We show that accurate upscaled models can be constructed by first upscaling capillary pressure under the assumption of capillary equilibrium, and then computing coarse-scale relative permeability functions using a global dynamic upscaling procedure. Some of the key numerical treatments associated with the overall procedure, such as near-well upscaling and the use of global coarse-scale iteration to improve model accuracy, will be described.

The new upscaling approach is applied to synthetic heterogeneous two-dimensional aquifer models with flow driven by CO2 injection. Injection via both vertical and horizontal wells is considered. Fine-scale reference results are compared with coarse-scale results generated using both the new upscaling approach and alternative (simpler) methods. In general, of the techniques considered, our new approach is shown to provide the best overall accuracy. Different CO2 injection rates and well locations are considered, which allows us to explore the robustness of the upscaled model. Various strategies for enhancing model robustness will also be applied and assessed.