This question is about verifying a system simulated with MD. I am currently building a model of a two phase gas-liquid system, where I initially have $\ce{N2}$, $\ce{CO2}$, $\ce{Ar}$, $\ce{He}$ and $\ce{CH4}$ in the gaseous phase and $\ce{Na+}$ with $\ce{Cl-}$ in an aqueous phase, which is modeled with TIP3P water. The most abundant component of the gases is $\ce{N2}$, which comprises over 95% of the gas.

I have a certain, non-ambient $p/T$ setting with a relatively high pressure (30-40 kbar). My idea is to study partition coefficients using molecular dynamics in multiple different $p/T$ settings. I'm currently using OPLS/AA force field which models intramolecular potentials with harmonic potentials and intermolecular potentials through Ewald-summation and LJ-612 -potential.

Currently I have just equilibrated the system, so that $p,T,V$ and $E$ do not fluctuate much. Before moving onwards with this system, I would like to verify that it produces reasonable data which is on par with other classical computational simulations.

I have noticed that there are not many papers which have focused on a system similar to mine, and I'm probably not going to be able to find an exact match in which to compare to. So, while I've been looking through the literature, I've looked at the following parameters to get a feeling of how well the system is behaving:

  • Radial Distribution Functions (RDF).
  • Partition Coefficients of the gases.
  • Long term behaviour of the system (no odd behaviour such as crystal forming etc.).
  • Densities of phases ($\rho_{\text{gas}} < \rho_{\text{liquid}}$)

So far these have all been quite reasonable. Are there any other standard methods and/or values that I should be looking at? I am obviously constantly reading through literature to find other results for comparison.

  • $\begingroup$ You could try simulating a special case/known system that is close to your system using the same methodology and comparing the results with experiment. $\endgroup$ Feb 21, 2017 at 9:47
  • 2
    $\begingroup$ Thanks for the input Felipe! That's exactly what I've been doing. The problem (at the time I was writing the original question) was that I could easily find an abundance of information regarding to the behavior of single compounds (such as methane gas) in TIP3P water, but the amount of information about mixtures such as mine was sparse. I'm fully aware that I'm probably not going to find a 1:1 match from literature. I'm currently looking in to getting hydration free energies for the gaseous compounds and comparing those to experiment. $\endgroup$
    – Bdrs
    Feb 21, 2017 at 11:10
  • $\begingroup$ OPLA has been parametrized based on density and heat of vaporization of liquids. So, I'm afraid of using that for the gas phase. You'd better seek for another FF. Goodluck $\endgroup$ Mar 2, 2018 at 8:44
  • $\begingroup$ Also include Boltzmann distributions of bond lengths and angles? I know that much of this behaviour is captured in the RDF but I find that the distributions are easier to compare and also mean values are widespread in the literature $\endgroup$
    – obackhouse
    Mar 12, 2018 at 9:33

1 Answer 1


With the simple molecules in your system, using the Gibbs Ensemble Monte Carlo would be the way to go. Each phase gets its own simulation box, and particles are swapped back and forth until equilibrium is reached.

Cassandra, DL_MONTE and other programs can do this easily for you. Once equilibrium is reached you can sample each box in the NVT or NPT ensemble at that composition.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.