Flowsheeting packages: Reliable or fictitious process models?

Flowsheeting packages can be useful computational aids in process design and help in understanding the behaviour of processes. However, without quality input data and a good overall chemical engineering background, the simulation results can easily be misinterpreted. Kinetic data and thermodynamic property methods can be the most likely source of errors. A case study is presented, where the same process model of a typical vinyl chloride monomer plant can give widely differing simulation results, even when employing recommended thermodynamics property methods. In this work the PRSV thermodynamics method were used as a base case, and then in turn the extended NRTL method, the Antoine vapour pressure method and the Kabadi-Danner method. The Antoine method gave the worst predictions, with deviations as high a 293%, whilst the other two methods gave varying but still significant deviations, indicating differing sensitivity of the methods to different parts of the process. Recommendations were put forward to minimize misinterpretation of simulation results.