Application of soft computing approaches for modeling saturation pressure of reservoir oils

Accurate determination of bubble pressure of reservoir fluid at reservoir conditions is one of the important parameter which is necessary for various calculations in petroleum engineering. This study presents two improved algorithms based on machine learning approaches for efficient estimation of saturation pressure of reservoir oil. To achieve the research purpose, a large data set, comprising of more than 750 crude oil samples with different composition and geographical origins, was collected from the literature for development of the models. The efficiency of the proposed models was tested against sixteen well-known empirical correlations. The proposed models show good performance in terms of accuracy with the lowest error percentage and highest R2 values.