Thermodynamic Models from Fluctuation Solution Theory Analysis of Molecular Simulations
Advancing the chemical engineering fundamentals
Thermodyanmics: Molecular Simulation & Related Approaches (T2-1d)
Keywords: Fluctucations, Liquids, Simulation, Derivative, Properties
Most engineering approaches to property prediction (group contribution methods and equations of state) are driven and developed by parameterization revisions based on experimental data. Current global research trends indicate that atomic scale modeling R&D expenditures grow relatively faster than expenditures for experimental research. Also the price ratio for property determination by experiment versus atomic scale simulation clearly develops in favor of the latter. This suggests that ’experimental data’-driven approaches will be more and more difficult to develop and to maintain in future, unless we develop less data-demanding approaches or other less expensive means of providing equally reliable data.
From isobaric-isothermal Molecular Dynamic (NPT-MD) simulations we calculate radial distribution functions (RDFs) for molecular pairs using for example the CHARMM force field. We integrate simulation results to obtain total correlation function integrals (TCFIs). Small errors in RDFs can lead to large errors in TCFIs. This problem was overcome by introducing a simple expression to reproduce the indirect interactions of the RDFs thereby ensuring finite TCFIs. The simple expression also allows extrapolation of RDFs, which allows integration to infinity. One output of this analysis is activity coefficient derivatives which can be used together with experimental data in the development of liquid phase Gibbs free energy models. Earlier works have included test systems of slightly and moderately non-ideal binary mixtures for which measured data exist, such as benzene/ethanol (at 298.15 K) and benzene/methyl acetate (at 303.15 K). When comparing the results of our computations with reliable data on both systems the results agreed to within experimental uncertainty. Recently, we have studied both near-critical systems and systems with no experimental data where group contribution methods (mixture and pure components) are difficult to develop. The results are compared with experimental data to evaluate the reliability of the method.
Presented Tuesday 18, 11:40 to 12:00, in session Thermodyanmics: Molecular Simulation & Related Approaches (T2-1d).