Viscosity and thermal conductivity are related properties and models for both are required for reservoir and process simulation. In most heavy oil processes, the viscosity must be reduced by heating and/or dilution with solvents. To design and optimize these processes, accurate viscosity models are required for both reservoir and process simulation. Current models are challenging to apply to heavy oils. Thermal conductivity is required for the simulation of heat exchange operations in refineries. Current models are either intended for liquid phases or computationally intensive.

This thesis presents the development of predictive viscosity and thermal conductivity models for reservoir and process simulation. The models were developed based on an experimental dataset collected in this thesis that includes the viscosity and thermal conductivity of whole and diluted heavy oils, partially deasphalted oils, asphaltenes, distillation cuts and pure hydrocarbons. The Expanded Fluid (EF) and the Generalized Walther (GW) viscosity models were updated to predict the viscosity of whole and diluted crude oils and their fractions (such as deasphalted oils). The EF model is suitable for process simulation and is applicable across the whole phase diagram. The required inputs are a distillation assay, the oil specific gravity, experimental or predicted fluid density at the process conditions, and pressure. The GW model is suitable for reservoir simulation and is only applicable to liquids well below their critical point. The inputs are a distillation assay, the oil specific gravity, temperature, and pressure. The EF concept was also used to develop a thermal conductivity model suitable for process simulation using the same inputs as the EF viscosity model.

The updated EF and GW viscosity models and the EF thermal conductvity model are applicable to crude oils over a wide range of API gravities, temperatures and pressures. They have fewer parameters than other models, the parameters have physical significance, and they are easily correlated to fluid properties. The predicted viscosities and thermal conductivities are within 50% and 3% of the experimental values, respectively. The deviations are less than obtained with other available methods. A straightforward tuning procedure allows the models to fit data to within the experimental error.