RESERVOIR PERFORMANCE PREDICTION IN STEAM HUFF AND PUFF INJECTION USING PROXY MODELLING

The problems of cost and time inefficiency during reservoir simulation persists in designing a steam huff and puff injection scheme. Therefore, developing predictive proxy models using machine learning algorithms is a suitable solution to address this concern. This study employed polynomial regression (PR) and artificial neural network (ANN) models to develop predictive models of the steam huff and puff injection scheme. Using a one-well cylindrical synthetic reservoir model, a total of 6084 experimental observations were simulated, each with 28 different features as input parameters. For each experiment, cumulative oil production and maximum oil production rate were extracted as the output parameters for developing the models. Reservoir properties which could change after an injection cycle, were also modelled. The developed models were evaluated based on the fitting performance from adjusted R-square (????????????????2), the mean absolute error (MAE) and the root mean square error (RMSE).