ramess101 / IFPSC_10

Submission to Industrial Fluid Properties Simulation Challenge 10
1 stars 0 forks source link

Validation of TraPPE, Octane #10

Closed ramess101 closed 5 years ago

ramess101 commented 6 years ago

@mostafa-razavi

@Kaiveria and I have validated our TraPPE results for n-octane by comparing the viscosity-pressure dependence with that reported in the literature from two different simulation studies. The study from Kioupis (and Ed Maginn) is from 2000 while Nieto-Draghi et al. (Philippe Ungerer and Bernard Rousseau) is from 2006.

Notably, Kioupis and Maginn used a harmonic bond potential, while Nieto-Draghi et al. appear to use fixed bonds (although they never say this explicitly, they did not report using a harmonic potential). For this reason, we performed two sets of simulations, one with fixed bonds and one with the same harmonic potential used by Kioupis and Maginn.

image

There are a few reasons why our results demonstrate a small discrepancy with the literature:

  1. Nieto-Draghi reported their results at 347.52 K and purported that this was the same temperature as the Kioupis results. In fact, Kioupis and Maginn reported a value of 348.15 K. Furthermore, we accidentally performed our simulations at 347 K, which might explain why our pressures are slightly less than those reported by Nieto-Draghi, despite having the exact same density.

  2. Also, note that Kioupis and Maginn used nonequilibrium molecular dynamics (NEMD) while Nieto-Draghi et al. used EMD with the Einstein relationship at short times. The extrapolation of NEMD results to the zero shear limit using the Carreau model typically over predicts viscosity. The Einstein relation at short times is a cheap, but not as rigorous approach, for obtaining reliable viscosities.

For these reasons, we have deemed our results to be sufficiently consistent with those from the literature. However, we will rerun the simulations at 347.52 K to see if this resolves the small shift in pressure. Furthermore, we have some simulations running that start with the NPT ensemble to determine the average density. These new simulations use the Kioupis pressure to initialize the system whereas the old simulations (shown above) used the Nieto-Draghi density as the initial configuration.

These results will be included as supporting information for validation.