Dear all,
GCMC is used to calculate the densities of ethane at the different chemical potentials when the temperature is 343.15 K. As shown in the table, the last 3 cases in which chemical potential is bigger than 8.3 Kcal/mol get unacceptable results.
Why do these cases get bad results? How can I make them get better?
Table The comparison of GCMC setting, result, and NIST Data
No. GCMC Set-Up GCMC Result NIST Data Relative Error，%
μ, Kcal/mol T, K T, K p, atm ρ,g/ml T, K p, atm ρ,g/ml T p ρ
1 -7.2 343.15 344.60 630.16 0.46 343.15 630.40 0.46 -0.42 0.04 -1.17
2 -7.4 343.15 344.98 507.76 0.44 343.15 507.20 0.44 -0.53 -0.11 -1.08
3 -7.6 343.15 345.34 382.72 0.42 343.15 382.85 0.42 -0.64 0.03 -0.60
4 -7.8 343.15 339.34 277.46 0.39 343.15 276.92 0.39 1.11 -0.20 -0.90
5 -8 343.15 324.93 192.39 0.34 343.15 192.87 0.35 5.31 0.25 3.44
6 -8.3 343.15 303.12 106.36 0.11 343.15 103.06 0.26 11.67 -3.21 58.40
7 -8.6 343.15 364.43 49.93 0.05 343.15 50.09 0.07 -6.20 0.31 38.02
8 -8.9 343.15 373.62 34.10 0.03 343.15 33.97 0.04 -8.88 -0.39 41.89
The TraPPE-UA force field is used to describe ethane. The input file is shown in the attachment. The version of LAMMPS run in this case is LAMMPS 64-bit 5Jun2019-MPI.
A bunch of thanks to you!
Sincerely,
Dongchen
in.gcmc_C2H6 (1.65 KB)
Dear all,
GCMC is used to calculate the densities of ethane at the different chemical potentials when the temperature is 343.15 K. As shown in the table, the last 3 cases in which chemical potential is bigger than 8.3 Kcal/mol get unacceptable results.
Why do these cases get bad results? How can I make them get better?
the quality of your results crucially depends on the degree of transferability of your force field parameterization and convergence of the GCMC runs.
all classical force field parameter sets have their limits and cannot be expected to represent all thermodynamic conditions equally well. how well for specific conditions depends on the specific choices made by the force field developers and what conditions those were tested with. so the best source of information are the relevant publications describing those parameterizations. as you go to lower densities, you have to keep in mind that the choice of interaction cutoff will have a larger impact on your results, and thus you will have to pay closer attention to the
as for the convergence, you can monitor this easily yourself and extend the run time of your simulation if necessary.
we have recently noticed some issues with fix rigid/nvt/small for recent LAMMPS versions when used in combination with fix deposit or fix gcmc, so you may want to try the latest patch version (currently 8 April 2021) to see if the corrections in that patch release have an impact on your results.
beyond this, you probably are best off discussing this with the authors of the force field parameterization to see if the force field is expected to reproduce the properties you are looking at to the level of accuracy that you expect or not.
axel.