Modeling Viscosity in Wide Ranges of Conditions: A Molecular Simulation Study

VICTOR R VASQUEZ; EUGENIA A MACEDO; MARCELO S ZABALOY

Austin Convention Center, Austin, TX

Congreso; AIChE 2004 Annual Meeting, November 7-12,; 2004

American Institute of Chemical Engineers

The Lennard-Jones (LJ) fluid is a realistic model because it reproduces the basic trends found for real fluids, in wide ranges of conditions, for physical properties such as viscosity. On the other hand, the LJ fluid is such that its dimensionless properties, e.g., pressure and viscosity, are universally related to its dimensionless density and temperature. Therefore, the LJ fluid can be used as a corresponding states fluid, where the dimensionless variables are expressed in terms of the intermolecular potential parameters, i.e., in terms of parameters with some physical meaning at molecular level. A corresponding states nature, realism and reference to molecular level parameters are features that make the LJ fluid appealing as a basis to build a model for the viscosity of real pure substances and their mixtures. The molecular simulation field is rich in works where complex molecules are modeled using segment-segment LJ interactions as a term within the intermolecular potential function. However, such results cannot be used in a straightforward manner to build a general-purpose analytical viscosity model, valid in wide ranges of conditions. Also, some papers are found where the viscosity of mixtures of simple LJ fluids is computed, but typically at a limited number of conditions. A more fundamental study, consisting of computing the viscosity of mixtures of simple LJ fluids in wide enough ranges of temperature, pressure, and composition, is lacking for mixtures of LJ fluids with large differences in their intermolecular potential parameters. One of the aims of this work is to fill that gap.In this work, we have computed the viscosity of binary LJ mixtures using a molecular dynamics (MD) periodic perturbation method at a large number of temperature, density, and composition conditions. The method is fast and basically consists of imposing a periodic external field on the system and of computing the viscosity from the developed periodic velocity profile. We used the MD results to optimize the analytical form of the composition dependence of an LJ-based one-fluid analytical model for the viscosity of mixtures of real fluids. We evaluated these MD-based mixing rules by comparing predicted and experimental viscosities for real mixtures.