Analysis and comparison of in-situ combustion chemical reaction models

Abstract

In-situ combustion (ISC) is a promising enhanced oil recovery method for heavy oil, late stage, or very deep reservoirs. One persistent challenge in modeling ISC is determining the correct chemical reaction model for heavy oil combustion. Reaction models containing more reactions or hydrocarbon pseudocomponent species are thought to model better the combustion of heavy oil, but smaller models are preferable for upscaling. Furthermore, most reaction models in prior work have been chosen on an ad hoc basis, and no work has performed a comparative analysis between oil samples and reaction models. Here we apply a reaction model optimization workflow to calibrate seven different ISC reaction models to experimental data for combustion of two heavy oil samples. We show that both the stoichiometry and kinetic parameters for a given model can vary substantially between oil samples. Results also show that a greater number of parameters, reactions, or pseudocomponent species do not necessarily improve the performance of the reaction model. On the other hand, a greater number of pseudocomponent intermediate coking stages appears to improve the accuracy. Overall, our findings suggest that reaction models with fewer reactions or parameters describe heavy oil combustion as well or better than comparatively larger models and that the number of stages of pseudocomponent decomposition is the most important aspect of a reaction model. These results allow for better selection from existing ISC reaction models as well as provide heuristics for future design of improved models for heavy oil combustion.