A complete general theory for non-equilibrium states is currently lacking. Non-equilibrium states are hard to reproduce experimentally, but creating computer simulations of relatively simple and non-equilibrium systems can act as a 'numerical laboratory', in which to study steady states far away from equilibrium.
The Joule-Thomson throttling experiment, being a system driven away from equilibrium during the throttling, was first performed by Lord Kelvin and Joule in 1852. They successfully cooled a gas in an adiabatic process. This study investigates the simulation of a Joule-Thomson throttling proposed by Hoover, Hoover and Travis (2014), who used a purely repulsive potential and successfully observed cooling. This was puzzling, as Van der Waals had noted that the Joule-Thomson experiment proved the presence of intermolecular attractive forces. It was found that the original simulation did not conserve enthalpy, which is a requirement of a Joule-Thomson throttling.
This study proposes the use of two families of pair potentials: the $mn$-family, first defined by Hoover and the $LJ/s$ first defined by Holian and Evans. These potentials offer an attractive component, while being well suited for molecular dynamics simulations, by being continuous in its derivatives and smooth without the need for further corrections.
The phase diagrams for these potentials are unknown, but are required to perform a successful throttling. This study develops two methods of predicting liquid-vapour coexistence and Joule-Thomson inversion curves without any a priori knowledge of the phase diagram: (i) Virial coefficient theory and (ii) a Barker-Henderson perturbation theory.
The theories successfully predicted liquid-vapour coexistence and Joule-Thomson inversion curves for a range of members of each family in two and three dimensions. One potential was then selected, and used to perform a two dimensional Joule-Thomson throttling, which displayed cooling of the gas while keeping the enthalpy constant.
