Automated History Matching of Combustion Experiments Using the Ensemble Kalman Filter

In-situ combustion (ISC) and High Pressure Air Injection (HPAI) are enhanced oil recovery processes, in<br>which air or oxygen enriched air is injected into a reservoir. The oxygen present in the air reacts with the<br>crude oil in the reservoir. This results in a combustion front propagating through the reservoir, generating<br>heat and flue gases. Many chemical reactions take place over different zones and temperature ranges<br>during the process. Reaction schemes based on pseudo components are used in numerical simulations to<br>describe these reaction processes. The kinetics of these reactions are given by Arrhenius-type equations.<br>A still remaining challenge in combustion modelling is the estimation of the Arrhenius kinetic parameters.<br>Combustion tube experiments are therefore performed to obtain information about the burn characteristics<br>that depend on the crude and reservoir rock properties. The Arrhenius parameters can be estimated by<br>history matching these experiments. The obtained values have to be up-scaled before they can be used for<br>field-scale simulations. History matching the combustion tube experiments is an intensive and timeconsuming<br>process, because multiple reactions occur at the same time and because non-unique matches<br>are expected. An automated history-matching tool for combustion tube tests is desired to quantify<br>uncertainty in the obtained kinetic parameters and reduce time spend to obtain a good match. In this work<br>a method is described for the automated history matching of combustion tube experiments using the<br>Ensemble Kalman Filter (EnKF).<br>The Ensemble Kalman Filter is a sequential data assimilation technique that combines measurement series<br>with dynamic models. The EnKF uses a Monte Carlo approach in which model errors are represented by<br>an ensemble of realizations. The ensemble is integrated in time to make predictions on system parameters<br>and state variables and their uncertainties.<br>The method to history match combustion tube experiments using the Ensemble Kalman Filter is described<br>in this paper and applied to combustion tube experiments. The matching of the combustion tube test is<br>based on temperature profiles, oil and water production data and effluent composition. It is shown how the<br>initial uncertainty in the kinetic parameter ranges (activation energy and frequency factor) are reduced by<br>automated history matching of the experiments, using the EnKF method.