Uncertainty of predicted process performance due to variations in thermodynamics model parameter estimation from different experimental data sets

Abstract

A method is presented using Monte Carlo Simulation coupled with the Latin Hypercube Sampling technique and parameter correlations to determine the effect of different sets of experimental data on the uncertainties in model parameters and, thus, the uncertainties in predicted process performance. Specifically, the estimation of binary interaction parameters for the NRTL equation was studied for three liquid-liquid tertiary systems: Diisopropyl-Ether + Acetic Acid + Water; 1,1,2-Trichloroethane + Acetone + Water; and Chloroform + Acetone + Water. From the experimental data available for these systems, regressions were done and the obtained parameters were used to perform simulations on a liquid-liquid extractor. The results show that the differences between the binary parameters from different sets of experimental data lead to significant differences in the uncertainty of predicted extractions in process units.

Type
Publication
Thesis, University of Nevada, Reno
Date