Development of a New Artificial-Neural-Network Model for Predicting Minimum Miscibility Pressure in Hydrocarbon Gas Injection

Abstract

Abstract Miscible gas injection is one of the most effective enhanced oil recovery techniques and minimum miscibility pressure (MMP) is an important parameter in miscible gas injection processes. Exact determination of this parameter is critical to the designing of the injection equipments and the project investment prospect. The purpose of this paper is to develop a new artificial neural network (ANN) model to predict the minimum miscibility pressure of hydrocarbon gas flooding. Different MMP correlations and models regarding the kind of injection gas and the mechanism of miscibility have been proposed, which are respectively based on mathematical and thermodynamic calculations. Almost all the correlations proposed in the literature are representative of either condensing or vaporizing mechanism or give reasonable results only in the limited range of data they are based on. In this article, by taking into consideration both condensing and vaporizing mechanisms and by using a wider range of data, the new model is introduced. Experimental data from different crude oil reservoirs carried out by slim tube test have been used to obtain a new model in which both mechanisms are included. An Iranian oil reservoir is a candidate for miscible gas recycling with API=23 and initial pressure of 5500 psi. Sampling and recombination is done on the reservoir fluids. Flash and differential expansions are performed for fluid characterization and then miscible experiments are carried out for MMP determination with slim-tube apparatus. The significance of this correlation is that MMP can be determined for any composition of oil and gas, no matter which mechanism is dominant in achieving miscibility. The sensitivity analysis is done and consequently the percentage of error for the model is compared with the literature data. It is shown that the results obtained from the new MMP model are more accurate when compared with other most common correlations reported.