Summary
This paper presents a statistical upscaling workflow for warm solvent injection (WSI) processes, a more environmentally friendly alternative to traditional thermal-based heavy oil extraction methods. The complexity of the heat and mass mechanisms involved in WSI makes flow simulation and optimization challenging. A two-step flow-based upscaling workflow is presented for handling static (facies proportions, porosity, and permeability) and dynamic properties (longitudinal and transverse dispersivity). The first step involves quantifying the effect of numerical dispersivity for a homogeneous model, while the second step incorporates the scaleup of uncertainty in heterogeneity. The method is flexible for handling anisotropic dispersivity upscaling for 3D models. Several novel aspects include (1) considering facies distributions (e.g., sand vs. shale layers), (2) extending the method to 3D, and (3) implementing a cloud transform to sample from the conditional probability distributions of longitudinal and transverse dispersivity considering porosity and net-to-gross (NTG) ratio. An ensemble of coarse-scale models is simulated, demonstrating the proposed workflow’s effectiveness in capturing spatial heterogeneity and improving WSI simulation accuracy in heterogeneous reservoirs.