(304f) A System for Thermophysical Data Analysis and Optimization

The one question that haunts every process engineer in today's world is, ?How good are my simulations?? Design decisions and customer guarantees, with important consequences, are made based on the results of process simulations. Unfortunately, there is no way to transcend the quality of the thermophysical methods and data that are employed in these simulations, necessitating special expertise. Competing with this requirement is the reality of budgets, and the need to ?do more with less.?

We introduce a system entitled Thermodynamic Data Manager (TDM), which is focused on empowering the process engineer to realize the best out of the simulation tools at his/her disposal. Even in the absence of thermodynamics experts, the process engineer ought to experience a substantial improvement in process design confidence. Although TDM is intended to work with the SimSci-Esscor process simulators, it can be used separately as a data analysis tool, the results of which can be transcribed to other downstream applications, even up to the plant level. TDM includes subsystems to enable the process engineer to organize and categorize thermophysical information, such as the creation of special purpose databanks. Customized visualization tools with overlay capability are included for easy comparison purposes.

TDM also allows the process engineer to superimpose data from multiple sources and tune appropriate thermodynamic model parameters. TDM assists the engineer in the identification of noise, and provides controls to minimize / eliminate its effect. Of special significance is the multi-property data optimization feature. Data of different kinds may be simultaneously regressed to optimize parameters for versatile thermodynamic models. A representative example would be the simultaneous correlation of relevant phase equilibrium, enthalpy and density data.

This paper first presents a brief description of the TDM system and its regression subsystem. Then it discusses the alpha regression problem in some detail and presents a solution procedure. Next, the alpha function developed in Twu et al (1991) becomes the object of an in-depth study of various factors that may affect the results of alpha regression. The regression results for Methane and n-Nonane using the Peng-Robinson equation of state are summarized and various observations and conclusions are drawn from them. The key conclusion is that changes in physical property values or in vapor pressure correlations may have a significant effect on the alpha regression results. Therefore, it is important for the regression practitioners and process engineers to be aware of changes taking place in the thermodynamic libraries and databanks being used for simulation studies and models. Such changed may necessitate a rerun of regression studies to fully assess their impact. TDM Regress provides a tool-of-choice to perform such studies.