Case studies of computer-aided design sensitivity to thermodynamic data and models

The design of chemical processes relies on simulation. However, the results are strongly dependent on thermodynamic models for the basic properties. W hile proper choice of the thermodynamic model is important, even with the b est available model, the uncertainties in the model parameters and the expe rimental data that are used to regress them can be significant. The uncerta inty induced from the different simulators is another factor to be consider ed in process design. Through a series of case studies on steady-state and dynamic simulation, we show how uncertainties of the thermodynamic data and models of the system can have a profound effect on the process design. In the meantime, the strategies for quantification of such thermodynamic-param eter-induced uncertainties via Monte Carlo simulation, with the Latin hyper cube sampling technique, equal probability sampling technique, and regressi on analyses, are described, The study indicates that the designs developed through use of these models are significantly sensitive to the parameter un certainties.