Due to simplicity, consistency and flexibility, the Lohrenz-Bray-Clark (LBC) correlation is the most widely used viscosity model in reservoir engineering. Unfortunately the LBC viscosity model does not accurately predict liquid viscosity. Consequently it is necessary to tune the calculated viscosities. Tuning of the LBC viscosity model is normally performed by modifying the critical volumes of the C7+ components and/or the LBC coefficients. The tuning procedure is not straight forward. Special attention is required for three challenging fluid systems: viscosity of the condensed oil from gas condensates, viscosity changes in connection with gas injection and viscosity of heavy oils.
This paper describes guidelines for proper tuning and consistency checking of the LBC viscosity model. Instead of using various correlation-estimated critical volumes for the C7+ components as in most PVT software, the initial critical volumes of the C7+ components are calculated based on component viscosities estimated from a dead-oil empirical correlation. Global regression to the measured viscosity data is then made with modifiers to the initial critical volumes. This procedure guarantees monotonically increasing component viscosities for the C7+ components. In the case of any regression on LBC coefficients, it is very important to maintain a monotonic relation of viscosity vs. reduced density.
Viscosity data from three Norwegian offshore reservoirs, from gas condensate to heavy oil, are used as examples. The guidelines for tuning the LBC viscosity model presented in this paper provide practical insight and understanding of how to apply the LBC viscosity model to various fluid systems.
Viscosity is an important physical property for fluid flow calculations in reservoirs, tubing and pipelines. Empirical correlations and corresponding-states models have been developed for modeling viscosity under various pressure and temperature conditions.
The Lohrenz-Bray-Clark (LBC) correlation for dense gas mixtures was published in 1964 by Lohrenz et al.1 based on the original work by Jossi et al.2 for pure substances. The detailed formula is given in the next section.
The prediction capability of gas viscosity with the LBC correlation is reasonable, while the prediction of oil viscosity is usually poor. Other more recent corresponding-states viscosity models show better prediction capability for oil viscosity, for example, the Corresponding States Principle (CSP) method proposed by Pederson et al.3 Due to the simplicity and flexibility, the LBC correlation is the most widely used viscosity model, especially in most commercial compositional simulators.
The LBC correlation is very sensitive to mixture density and to the critical volumes of the heavy components. Adjustment of critical volumes of the heavy components and/or the LBC coefficients to match the experimental oil viscosity is usually necessary. However, the tuning procedure is not straight forward, especially for three challenging fluid systems: viscosity of the condensed oil from gas condensates, viscosity changes in connection with gas injection and viscosity for heavy oils.
Conventional tuning methods usually start with the initial critical volumes which are estimated based on various empirical correlations. The resulting component viscosity of the C7+ components is often non-monotonically increasing with molecular weight. This can potentially cause problems in compositionally sensitive processes, for example, viscosity of the condensed oil from gas condensates, and viscosity changes during gas injection.