Evaluating Cubic Equations of State with Various α Functions for Viscosity Predictions of 124 Industrial Important Fluids Based on Residual Entropy Scaling

Le résumé fourni par la source

Abstract Accurate prediction of viscosity remains a challenge in industry due to the lack of reliable simple universal models. This work investigates the accuracy of four cubic equations of state (EOS) with different α functions for pure fluid viscosity predictions based on the residual entropy scaling (RES). The cubic EOS are Peng–Robinson (PR), Soave–Redlich–Kwong (SRK), Patel-Teja-Valderrama (PTV), and Yang-Frotscher-Richter (YFR), and the α functions are those proposed by Twu et al., Coquelet et al., Mahmoodi and Sedigh, and Heyen et al. The investigation utilizes the approximately 54,000 experimental viscosity data of 124 pure fluids at pressures below 60 MPa, mainly obtained from the NIST TDE (ThermoData Engine) database. Compared to a previous study (ACS Omega2025, 10, 6124), which focused on the Soave α function, the adoption of modified α functions led to improved viscosity predictions for 39 out of the 124 studied substances. Notable enhancements are observed for R1234ze­(E) and SO2 with PTV-Heyen and for CO2 with PR-Twu. The absolute average deviation (AAD) between experimental values and model predictions is 3.1% (PR, SRK, and YFR), 3.0% (PTV), 3.4% (PR-Twu and SRK-Coquelet), 3.2% (PTV-Heyen and SRK-MS), 3.5% (PR-Coquelet), and 3.3% (PR-MS). As a reference, the AAD of the various reference models implemented in REFPROP 10.0 is 2.7%. This work demonstrates the potential of integrating optimized α functions to improve the predictive capabilities for viscosity within the cubic EOS + RES framework for certain specific pure fluids. Additionally, a recommended combination of cubic EOS and α function is provided for each fluid studied.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.