Improving Thermophysical Property Models with Volume Translation

Process simulation is a powerful tool for the design, construction, and maintenance of reaction and separation systems. However, the accuracy of property models has not kept place with the capacity for simulations. Systems that include light gaseous components in highly non-ideal or polar solutions are common and will become even more prevalent as industry looks to develop more non-petroleum fuel sources. Mixtures involving hydrogen, methane, carbon dioxide, ethanol, and water along with the light alkanes will form the bulk of fuel processing for the foreseeable future.

Current state-of-the-art property models rely on volumetric relations like the Peng-Robinson or Redlich-Kwong-Soave equations, which have remained largely unchanged for the last thirty years. While this persistence is a testament to the robustness of the methods, there is still a pressing need for improvement. Errors of more than five percent in the enthalpy and more than twenty percent in the liquid volume are common, which is unacceptable when a margin of only a few percent can dictate the feasibility of a process. Most changes to property models have been incremental and the result of regression of large sets of data. Regression is very accurate for systems where data is available, but inevitably is inadequate because data over the full range of systems of interest will never be realized.

This project aims to characterize the models that best represent PρTx relations and the ability to successfully predict phase coexistence using a minimum amount of experimental data. A complete understanding of the relationship between model parameters and data representation will help guide experimental design so measurements will have the greatest utility for property representation.