An Optimized Equation of State Thermodynamic Model for Hydrocarbon Volumetric Calculations Based on Wide-Range Experimental Data

Abstract

Abstract Compositional modeling using equations of state is the standard thermodynamic modeling approach in the oil and gas industry. Despite its robustness, this approach yields non-unique solutions due to its dependence on the selection of heptane-plus characterization models, binary interaction sets, and volume shift schemes. This study investigates this dependence and optimizes the performance of equation of state models in predicting the volumetric properties of hydrocarbon mixtures based on it. The performances of the Peng-Robinson (1978) and the Soave-Redlich-Kwong (1971) equations of state were compared in a comprehensive study. Different combinations of published heptane-plus characterization models, binary interaction sets, and volume shift schemes were also compared. The investigated heptane-plus characterization models included 4 boiling point, 10 critical pressure, 11 critical temperature, and 4 acentric factor correlations. Five binary interaction sets were applied in addition to 5 volume shift schemes. 88,000 thermodynamic models were tested against multistage separation data for 71 hydrocarbon mixtures of crude oil and gas condensate samples. The performance of the models in predicting gas-oil ratio and liquid density was evaluated, and the influence of each model component was quantified through sensitivity analysis. Results show that the choice of the equation of state model, heptane-plus characterization models, binary interaction set, and volume shift scheme can significantly vary volumetric predictions. Based on the selected thermodynamic model, the average absolute percent error ranged from 1.5% to 33.5% for liquid density and 5.7% to 36.8% for gas-oil ratio. The sensitivity analysis revealed that the selection of the volume shift scheme, the equation of state model, and the heptane-plus critical pressure model is the most influential on volumetric predictions. The best-performing models were reported, 62% of which used the Peng-Robinson (1978) equation of state. In contrast, 82% of the worst-performing models used the Soave-Redlich-Kwong (1971) equations of state. This demonstrates that the Peng-Robinson (1978) equation outperforms the Soave-Redlich-Kwong (1971) equation in predicting the volumetric properties of hydrocarbons. The comprehensive comparative study provides great insight into the influence of different model components on volumetric predictions for hydrocarbon mixtures. Moreover, the recommended thermodynamic models provide an accurate predictive tool for hydrocarbon mixture volumetric properties that does not need prior tuning to experimental data.