Coarse-Grained Models for Crude Oils: A Direct Link Between Equations of State and Molecular Simulations

Abstract

A description of fluid systems with molecular-based algebraic equations of state (EoS) is common practice in the oil industry while direct molecular simulation is a mainstream tool in the physical sciences, but the two approaches are rarely closely coupled. We describe a framework whereby we link molecular simulations to continuum EoS, effectively bridging large size and time scales. The novelty in our approach is that the EoS is employed to develop a force-field model for the intermolecular interactions (i.e., a potential energy function that can be used directly in microscopic/molecular simulations of fluids). This multi-scale model has the advantage that molecular simulations performed using said force field may be employed to explore aspects of the fluid properties that are inaccessible to experiment (owing to limitations resulting from the resolution of the system size, the short timescale, or extremes in the experimental conditions) and/or to theory (e.g., an accurate description of the structural/microstructural or dynamical properties of dense fluids)The procedure is exemplified by describing a coarse grained model for the simulation of a light condensate and a heavier live crude oil. By fitting the SAFT EoS to experimental pure component properties (densities, vapor pressures), one is able to obtain parameters for all the constituents in the mixture including surrogate models for the heavy ends. We employ this SAFT model to predict the phase behavior of synthetic mixtures of crude oils composed of discrete components by means of direct multiphase molecular dynamics.As proof-of-concepts, the dew point of a synthetic seven-component mixture is predicted from molecular dynamics simulations without the need of any fitted parameter and is in excellent agreement with experimental data. Similarly, the largest reported simulation, equivalent to over one million atoms, of a 15-component crude oil mixture including a model resin and asphaltene, is described. The framework presented is completely generic in nature and a discussion on the way it can be adapted for the study of complex mixtures including surfactants, aqueous systems, systems containing precipitants such as asphaltenes or waxes in a relatively straightforward way is given.