Optimization and uncertainty quantification of in situ combustion chemical reaction models

Abstract

Heavy oil, deep, and waterflooded reservoirs contain significant petroleum resources but present many unique challenges for oil recovery. In situ combustion (ISC) is a thermal enhanced oil recovery method that has been proposed for such reservoirs. A persistent challenge in understanding ISC is modeling the underlying chemical reactions. While many different reaction models have been proposed, these models are typically developed on an ad-hoc basis for individual crude oil samples and manually calibrated to experimental data. While some work exists on applying optimization-based approaches to reaction calibration, existing approaches are not fully-generalized nor do they quantify uncertainty in the estimated parameters. Here we present such an approach for calibrating and quantifying uncertainty in ISC chemical reaction parameters. The workflow presented is generalized to any hydrocarbon pseudocomponent-based reaction model of ISC, provides a framework for quantification of uncertainty in estimated parameters, and is fully automated. Thus, no parameter initialization or manual calibration is needed. We apply this workflow to characterization of a heavy crude oil sample and demonstrate the capability of this workflow to predict ISC oxidation kinetics and upscale to a combustion tube simulation. Overall, this work provides an algorithmic approach to calibrating ISC kinetics that enables quantitative analysis and comparison of oil types, reaction models, and reservoir conditions for ISC.