On the estimation of bulk viscosity of dilute nitrogen gas using equilibrium molecular dynamics approach

In this work, we present a study for the estimation of bulk viscosity using the equilibrium molecular dynamics-based Green-Kubo method. We have performed a parametric study to find optimal hyper-parameters to estimate bulk viscosity using the Green-Kubo method. Although similar studies exist for shear viscosity, none has been reported so far specifically for bulk viscosity. The expected uncertainty in bulk viscosity for a given length and number of molecular dynamics trajectories used in statistical averaging is determined. The effect of system size, temperature, and pressure on bulk viscosity has also been studied. The study reveals that the decay of autocorrelation function for bulk viscosity is slower than that for shear viscosity and hence requires a longer correlation length. A novel observation has been made that the autocorrelation length required for convergence in the Green-Kubo method for both shear and bulk viscosity of dilute nitrogen gas is of the same mean collision time length units irrespective of simulation pressure. However, when the temperature is varied, the required autocorrelation length remains unaffected for shear viscosity but increases slightly with temperature for bulk viscosity. The results obtained from the Green-Kubo method are compared with experimental and numerical results from the literature with special emphasis on their comparison with the results from the nonequilibrium molecular dynamics-based continuous expansion/compression method. Although the primary focus and novelty of this work are the discussion on bulk viscosity, a similar discussion on shear viscosity has also been added. Published under an exclusive license by AIP Publishing.

Automated summary

In this work, a study is presented for estimating bulk viscosity using the Green-Kubo method based on equilibrium molecular dynamics. The study explores optimal hyper-parameters for estimating bulk viscosity and investigates the effect of system size, temperature, and pressure. The study reveals that the autocorrelation function for bulk viscosity decays slower than that for shear viscosity, and the required autocorrelation length slightly increases with temperature for bulk viscosity.