SIMULATION OF PHASE BEHAVIOR USING DIFFERENT EQUATIONS OF STATE (EOS) IN PVTp SOFTWARE

Abstract

Equation of State (EOS) is crucial for both production forecasts and reservoir simulation, but tailored models typically don't match experimental findings, which undermines field development decisions. The Peng-Robinson (PR) and Soave-Redlich-Kwong (SRK) EOS were used to test the performance of Peng-Robinson and Soave-Redlich-Kwong in simulating crude oil phase behavior. A systematic tuning methodology was built that entailed sequential regression of the CO 2/C7+ binary interaction coefficients, C7 + critical properties, and acentric factor. The tuned model was checked in constant composition expansion and differential liberation experiments. The untuned PR EOS had better accuracy in terms of bubble point pressure with an Absolute Relative Error (ARE) of 0.015% than the SRK EOS of Absolute Relative Error (ARE) 8.22%. After tuning, PR EOS obtained a precise bubble point with an oil formation volume factor error of less than 2%. The CO 2 /C7+ binary reaction coefficient was the most sensitive to saturation pressure. But there was a trade-off when GOR prediction error rose to 10.13%, which showed that single-objective regression is not capable of producing an optimal combination of all the PVT properties. The C7+ acentric factor had to be adjusted at 21.22%, indicating heavy deviation of the fraction from the ideal behavior. The tuned EOS is a reliable input to reservoir simulation, and the uncertainty of the input is measured. The trade-off that was noted in GOR shows the necessity of multi-objective optimization in future EOS calibration processes. The presented systematic methodology is flexible to the various types of crude oil to enhance compositional modelling.