A hybrid neural–genetic algorithm for predicting pure and impure CO2 minimum miscibility pressure

"> Accurate prediction of the minimum miscibility pressure (MMP) in a gas injection process is crucial to optimizing the management of gas injection projects. Because the experimental determination of MMP is very expensive and time-consuming, searching for a fast and robust mathematical determination of CO2-oil MMP is usually requested. This paper presents a new model based on a hybrid neural-genetic algorithm for predicting pure and impure CO2-oil MMP. The CO2-oil MMP of a reservoir fluid was correlated with the reservoir temperature, the composition of the oil, and that of the solution gas. The developed model is able to reflect the impacts on the CO2–oil MMP of the molecular weight of the C5+ fraction, reservoir temperature, and solution gas in the oil. The validity of this new model was successfully approved by comparing the model results to the calculated results for the common pure and impure CO2-oil MMP correlations. The new model yielded the accurate prediction of the experimental slim-tube CO2-oil MMP with the lowest mean absolute percentage error (MAPE), the standard deviation of error (SD), the root mean square error (RMSE), and the highest correlation coefficient among tested impure and pure CO2-oil MMP correlations. The results demonstrate that the hybrid neural-genetic model can be applied successfully and provide high accuracy and reliability for MMP forecasting.