Beschreibung
For processes and products of biomass-based chemistry, a vast number of chemical species are available as possible design choices. An experimental evaluation of properties of all possible chemical species poses a prohibitive effort. Using property prediction methods like predictive perturbed-chain polar statistical associating fluid theory (predictice PCP-SAFT) and the conductor-like screening model for real solvents (COSMO-RS) experiments can be avoided altogether, but accuracy is still not sufficient for certain tasks. This thesis is concerned with increasing accuracy of property predictions. Using rigorous statistical methods, information from predictive PCP-SAFT and from experiments can be combined to yield a good trade-off between accuracy and experimental effort. USing the same statistical methods to combine information from predicitive PCP-SAFT and from COSMO-RS does not require any experimental information while yielding a higher accuracy than each individual property prediction methos. Finally, predictive PCP-SAFT is further developed into a new framework: segment-based equation of state parameter prediction (SEPP). SEPP yields a higher accuracy than predictive PCP-SAFT while using a lower number of parameters. The range of applicability is extended to chemical species containing carbon, hydrogen, oxygen, and nitrogen. Furthermore, chemical species that exhibit association can be predicted with low additional computational effort. SEPP is a broadly applicable framework for predicting thermophysical equilibrium data and thus is a valuable tool for chemical engineering.
