STATISTICAL REGRESSION OF BINARY VAPOR-LIQUID-EQUILIBRIUM DATA FOR TERNARY PHASE-EQUILIBRIUM PREDICTIONS

High-pressure vapor-liquid equilibrium data of more than 50 binary sys tems were correlated by a DDLC (density-dependent local-composition) m odel incorporated into the Soave-Redlich-Kwong equation of state. The Error-Propagation-Law Method based on the maximum likelihood principle and the simple least-squares method were applied to data reduction. F itting accuracies of the DDLC model by statistical regression were fou nd better than those obtained by the least-squares as well as those of the SRK equation of state by both methods. However, no improvements w ere obtained for the original SRK equation by the statistical method. Further, vapor-liquid equilibrium behaviors of eight ternary systems w ere predicted by utilizing the binary interaction parameters of both m odels obtained from experimental data of the constituent binaries by b oth statistical and conventional methods, respectively. Results showed that better prediction accuracies were obtained for the DDLC model by statistical regression. Similarly, no improvements were found for the SRK equation of state by statistical regression. In addition, the sup eriority of the statistical regression over the conventional method wa s demonstrated by various simulated data. (C) 1998 Elsevier Science B. V. All rights reserved.