Molecular Dynamics Simulation of Supercritical Spray Phenomena

Results showed that the way the fluid-wall interaction is modeled in molecular dynamics simulations has a strong effect on the resultant simulation of liquid injection into a gas. It was found that modeling the wall as individual atoms (atomistic model) interacting with the fluid resulted in the fluid remaining in an injection tube absent any pressure or body force to force it into a gaseous region while a stochastically modeled diffusely reflecting wall with no attraction between the fluid and the wall resulted in the fluid being injected into the gas even in the absence of pressure or body forces. Since a non-atomistically modeled wall has reduced computational requirements compared to an atomistic wall, efforts centered on an examination of possible nonatomistic continuous fluid-wall interaction models that would not only correctly reproduce fluid thermodynamic properties such as pressure but would also reproduce the same fluid injection behavior as an atomistic wall model. Several non-atomistic fluid-wall models were examined in terms of their ability to correctly predict the fluid pressure in the injection tube over a range of fluid densities as well as their ability to reproduce atomistic wall injection behavior, but none were found able to reproduce atomistic wall behavior. Accomplishments/New Findings In molecular dynamics (MD) simulations, the solid boundary conditions are modeled in two ways: Atomistic wall model: the wall has a molecular structure included in the modeled system. Continuous solid boundary model: the wall structure is modeled through a mathematical or a functional model. In the mathematical model, the wall model acts as an impulsive force acting on the fluid particle only when it crosses the imposed boundary, an example being a diffusely reflecting wall as is used in rarefied gas dynamics. The functional wall model corresponds to a force that acts on the fluid particles but within a range of interaction and not only at the boundary. Atomistic wall models are usually used if the main physical property studied is the fluid/solid interactions. The clear disadvantage of the atomistic model is its computational cost. The attraction of the functional model versus the mathematical model is the 'longer' range of interaction that might limit the fluid slip values at the boundary. However, no attempt to compare these models is available in the literature. No work to assess the consequences these models would have on especially dense phases and to determine which criteria has a predominant effect in order to choose one or the other model has been published. Pressure calculations are used to initially assess wall models differences since it is closely related to the fluid structure observed in a bounded medium at the nanoscale. Pressure is computed using the virial formula P.^r.l/jj?^).^ (i) V B 6V\r^. drtj "I where • is the interatomic potential between two atoms i and j and r is their separation distance. Pressure was first calculated for liquid argon confined by an atomistic wall, which is supposed to be the closest wall model to reality, especially at the nanoscale. Figure 1 shows that the calculated pressure matches the experimentally measured pressure over a wide range of liquid densities. The atomistic wall also gives a fluid density that is slightly lower in the center of the bounded region than the mean density due to the attractive (wetting) forces between the wall and fluid atoms.