Dear RASPA-Community,
currently I am thinking to use RASPA in order to simulate some High-Pressure isotherms. Before starting with any kind of simulation in pores,
I thought it's a good idea to simulate both a rho-P (GCMC, supercritical) and a rho_T diagramm (Gibbs-Ensemble, subcritical).
To sum the results up: the subcritical results are good (as expected), but the high pressure results tend to have quite pronounced deviations (about 9% at 90 bar).
I was woundering about possible reasons for that, however I couldn't really come to a clear result. So my idea is that a: the mistake comes from the
use of the Peng-Robinson-EOS which is implemented as default EOS as far as I know b: at elevated pressure, it's not a valid assumption to neglect three-body
interactions anymore and/or the forcefield is not optimized for such high pressures. Is there anybody who conducted HP simulations and has an idea if
there is a possible solution or if a mistake of 9% is in spec at those elevated pressures?
Attached, you find my input file and a results plot.
Thanks for your input/help/ideas in advance :)
SimulationType MonteCarlo
NumberOfCycles 50000
NumberOfInitializationCycles 50000
PrintEvery 100
ContinueAfterCrash no
ChargeMethod Ewald
Forcefield TraPPE
RemoveAtomNumberCodeFromLabel yes
Cutoff 12.8
EwaldPrecision 1e-6
Box 0
BoxLengths 30 30 30
ExternalTemperature 298
ExternalPressure 9e6
Movies no
WriteMoviesEvery 10000
Component 0 MoleculeName methane
StartingBead 0
MolFraction 1
MoleculeDefinition TraPPE
IdealGasRosenbluthWeight 1
TranslationProbability 1
RotationProbability 1
ReinsertionProbability 1
SwapProbability 1
CreateNumberOfMolecules 0
WidomProbability 0
9% at those high pressure is still very, very good I would say. The validity of such a simple methane model is not "endless". Any classical model will show weaknesses at extreme conditions. Perhaps a full-atom model might do better, but these are harder to paremeterize.
Many thanks for the feedback, this really helped me to classify the results. Maybe I am going to try a full-atom model in future, just out of curiosity. In this case, I am going to share my results here for everybody else :)