In this study, minimum miscibility pressure (MMP) which is a key parameter in design of an efficient miscible gas injection project is aimed to be determined by means of adaptive neuro-fuzzy inference system (ANFIS). 27 features including concentrations of different components of reservoir oil and injected gas, molecular weight and specific gravity of C
in reservoir oil and finally reservoir temperature were taken as inputs to the ANFIS. Principal component analysis (PCA) algorithm was used to reduce the dimensionality of the data. Using the back propagation gradient descent method in combination with the least squares method, ANFIS model was trained. The model’s predictions were compared with experimental results and also the results obtained from the commonly used MMP correlations in the literature. Based on these comparisons, it was found that the proposed ANFIS model has potential in predicting MMP values and also the effect of each individual parameter on these values.
Keywords: miscible gas injection, adaptive neuro-fuzzy inference system, ANFIS, minimum miscibility pressure, MMP, principal component analysis, PCA