Data-Driven Prediction of In-Situ Combustion Dynamics

Ogunbanwo, Olufolake (Stanford University) | Koh Yoo, Kuy Hun (Stanford University) | Gerritsen, Margot (Stanford University) | Kovscek, Anthony R. (Stanford University)



This paper presents a new workflow for the simulation of in-situ combustion (ISC) dynamics. In the proposed method, data from kinetic cell experiments, depicting the combustion chemistry, are tabulated and graphed based on the isoconversional principle. The tables hold the reaction rates used to predict the production and consumption of chemical species during in-situ combustion.

This new method of representing kinetics without the Arrhenius method is applied on one synthetic and two real kinetic cell experiments. In each case, the new method reasonably captures the reaction pathways taken by the reacting species as the combustive process occurs. A data-density sensitivity study on the tabulated rates for the real case shows that only four experiments are required to capture adequately the kinetics of the combustion process. The results are, however, found to be sensitive to the size of the time step taken. The method predicts critical changes in the reaction rates as the experiment is exposed to different temperature conditions, thereby capturing the speed of the combustion front, temperature profile, and fluid compositions of a simulated combustion tube experiment.

The direct use of the data ensures flexibility of the reaction rates with time and temperature. In addition, the non-Arrhenius kinetics technique eliminates the need for a descriptive reaction scheme that is typically computationally demanding, and instead focuses on the overall changes in the carbon oxides, oil, water and heat occurring at any time. Significantly, less tuning of parameters is required to match laboratory experiments because laboratory observations are easier to enforce.