Optimal physical property data for process simulations by optimal experimental design

Abstract

Abstract Simulations of chemical processes require accurate physical properties, which are usually estimated from experiments. Experimental effort can be minimised by optimal experimental design (OED). OED tailors experimental measurements to minimise the expected uncertainty of the fitted parameters. However, in process design and optimisation, the primary purpose of an experiment is usually not to determine property parameters as accurately as possible but to enable most accurate process simulations. Therefore, in this work, we present OED of physical property measurements leading to optimal predictions of process simulations by the so-called c -optimal experimental design. A c -optimal design minimises the uncertainty of the simulation results by weighting parameters by their influence on the process model. The c -optimal OED is employed to design liquid-liquid equilibrium measurements as a basis for simulations of an extraction and a hybrid extraction-distillation process. For the same simulation accuracy, the c -optimal design can almost half the number of experiments compared to state-of-the-art OED that neglects the process. Our work shows that c -optimal design can reduce experimental effort in chemical engineering successfully by tailoring experiments to their process application.