A genetic algorithm-based model to predict CO₂-oil physical properties for dead and live oil

Abstract

A key parameter in a CO2 flooding process is the CO2 solubility as it contributes to oil viscosity reduction and oil swelling which together, in turn, enhance the oil mobility and oil relative permeability. Often CO2-Oil solubility parameters are established through time-consuming experimental means or using models or correlations available in the literature. However, one must recognize that such models or correlations to predict CO 2-Oil physical properties are valid usually for certain data ranges or site-specific conditions. Furthermore, it is to be noted that there is no reliable model available to predict CO2-live oil physical properties, as most of the available correlations are developed based on dead oil data. In this study, a genetic algorithm (GA)-based technique has been used to develop more reliable correlations to predict CO2 solubility, oil swelling factor, CO2-Oil density and CO2-Oil viscosity for both dead and live oils. These correlations recognize not only all major parameters that affect each physical property, but also take into account the effects of CO 2 liquefaction pressure and oil molecular weight (MW). These correlations have been successfully validated with published experimental data and compared against several widely used correlations. The GA-based correlations have yielded more accurate predictions with lower errors than other correlations tested. Furthermore, unlike these correlations, which are applicable to only limited data ranges and conditions, GA-based correlations can be applied over a wider range and conditions. Another important and useful aspect is that GA-based correlations can also be integrated into any reservoir simulator for CO2-flooding design and simulation.