On the Application of a Simplified Temperature-Dependent Friction-Theory Viscosity Model in Compositional and Thermal-Compositional Reservoir Simulations

Abstract

Since the introduction of Friction-Theory (FT) viscosity model, its performance has been investigated for a range of reservoir fluids, including heavy oils and mixtures. Modifications of FT viscosity model has shown versatility with capturing viscosity variation over a range of temperatures, pressure, and compositions. The main purpose of this paper is to evaluate and demonstrate the effect of viscosity variation on the reservoir simulation processing rates and demonstrate the application of FT viscosity model for usage in compositional and thermal-compositional reservoir simulations. A simplified temperature-dependent FT viscosity model with more flexibility for optimization is presented. Performance of FT viscosity model for extended and lumped compositional models (as low as two components) were compared for a suite of diverse fluids. Lumped EoS models were used in realistic sector models to compare performance of FT viscosity model with standard Lorenz-Bray-Clark (LBC), Corresponding State Principles (CSP), and tabular viscosity models. Processing (production/cumulative) rates for both miscible gas injection compositional and steam injection thermal-compositional simulation cases were investigated. The main observation was that processing rate variations have almost a 1:1 ratio with viscosity variations of homogenous simulation models and a ratio of 2:1 for thermal compositional and large heterogenous gas injection compositional simulation models. Comparisons to experimental data showed that FT model is more flexible to accurately represent reservoir oil viscosity data than LBC. For the cases studied, a higher error was generally seen for heavier oils viscosity data with LBC model (~13%). FT viscosity model also showed more flexibility compared to LBC and CSP models in predicting viscosities of oil and gas mixtures. For gas injection sector model study, more significant impact was observed on gas breakthrough time (1-2 years difference), merely due to more accurate representation of mixture viscosities of FT model. Todd-Longstaff modified black oil miscible formulation simplistic model to represent mixture viscosities need to be revisited. In thermal-compositional simulation cases studied, constant pressure tabular viscosity data underestimated the liquid viscosity in the undersaturated region, which led to an overestimation of cumulative oil production in situations where the reservoir pressure was high, and temperature was only moderately higher than the original reservoir temperature. At high temperatures however, the impact of pressure and temperature effects on the cumulative oil production has not been significant. Although FT viscosity model has shown great promise representing viscosity of different fluid types with variation in temperature, pressure, and composition, its implementation in a reservoir simulation software and the extent of the impact of more accurate representation of viscosity data on reservoir simulation processing rates has not been documented. Based on our studies, using the simplified temperature-dependent FT viscosity model will reduce the uncertainty in reservoir simulation processing rates.