Using MD to Calculate the Diffusion Coefficients of Hydrocarbons

摘要

Developing more efficient engines depends greatly on the ability to model the combustion process computationally. Reliable computational simulations require knowledge of chemical kinetics as well as molecular transport properties. Gas-phase molecular transport properties such as viscosity, thermal conductivity, and diffusion play critical roles in combustion processes such as flame profile shapes, flame velocities, flame extinction, and pollutant product formation. Specifically, it has been shown that 10% inaccuracy in diffusion coefficients can lead to order of magnitude errors in macroscopic combustion models. Traditionally, diffusion coefficients were measured in experiments, but experiments are monetarily expensive, time consuming, and unreliable in extreme temperature and pressure conditions. Thus, gas kinetic theory (GKT) using classical mechanics with spherical intermolecular potentials is commonly used to derive the molecular diffusion coefficients necessary in combustion models. However, not only does GKT yield unreliable predictions in general, but it is inaccurate when used to calculate diffusion coefficients for molecules of significant non-spherical shape. We utilized molecular dynamics (MD) simulations in order to overcome the limitations of experiment and GKT. The advantage of using an MD simulation to calculate the diffusion coefficient of a given system is that it is less expensive than an experiment, less time-consuming, can simulate extreme conditions much more readily than an experiment, and it is more accurate than GKT for non-spherical molecules. We utilized molecular dynamics (MD) simulations, with a detailed atomistic model that includes intramolecular forces, Van der Waals interactions, and Columbic interactions, to predict the diffusion coefficients of various classes of hydrocarbons in nitrogen environments. We demonstrated the efficacy of our approach to calculate diffusion coefficients and the advantages it maintains over GKT by showing that the diffusion coefficients derived from MD simulations were more accurate than those from GKT when compared with experiments. Additionally, we note that the discrepancy between results obtained with MD and GKT increases with the eccentricity, size, and anisotropy of the molecule. The MD derived diffusion coefficients fit the expected temperature- and pressure-dependence trends. The ability, at a relatively low cost, to use MD to quickly and easily develop more accurate diffusion coefficients of hydrocarbons common in fuels will lead to more accurate macroscopic scale combustion models, which will lead to more efficient engines.