Optimal Experimental Design for the Characterization of LiquidLiquid Equilibria

Abstract

Measurements of liquid–liquid equilibria (LLE) are characterized by high experimental effort, long equilibration times, high sample consumption, and hence high costs. To reduce these efforts, we analyze LLE measurements using optimal experimental design (OED). Using OED, we identify experiments with the highest information content for LLE systems of both type I and type II. The NRTL and UNIQUAC gE models are employed. The optimal designs are compared to conventional designs. The optimal designs employ very few (3–5) distinct experiments located in regions where the curvature of the binodal curve is high. Thereby, at least half of the experimental effort can be saved without losing either prediction or parameter accuracy.