Improving your Separation Process Simulation with Accurate Binary Parameters

Process simulation software is extensively used for process development and for the optimization and improvement of existing plants, especially with the aim to reduce waste, by-products and increase energy efficiency and plant capacity. In both stages it quickly becomes apparent that a simulation model is only as good as the thermodynamic methods and parameters used. The correct modeling of separation processes requires exact models for the prediction of the real behavior of multicomponent systems using activity or fugacity coefficients in order to calculate separation factors between the phases in equilibrium.

The first, vital step is therefore to search for available experimental vapor-liquid equilibrium (VLE) data of the highest quality for the system being evaluated. It is often advisable to include further data types like azeotropic data, activity coefficients at infinite dilution, excess enthalpies for correct extrapolation to other temperatures, liquid-liquid equilibria and further more. Using of default parameters in your process simulator can be very risky when designing new processes. Investing some time on the front-end of your process design work to get the thermophysical properties right will always pay-off by insuring a successful start-up. It is important that the temperatures, pressure, and compositions are represented by the VLE data sets before fitting model parameters and a suitable equation of state (EOS) or gE model should be selected. If no VLE data of the subject system is available, the data can be measured or estimated via predictive methods like mod. UNIFAC or PSRK. In case of data estimation the performance of the models should be verified for similar mixtures where experimental data are available.

Significant improvements in accuracy can be achieved by making a conscience, up-front investment into obtaining and fitting high-quality VLE data to binary/model parameters for analysis of your separation process. During the presentation, this process will be exemplified using data and software tools from the DDB. Shortcomings of existing parameter sets in simulator software will be analyzed and explained in detail. This should help the engineer to identify crucial flaws in physical property predictions that are often not apparent without special training.