A broadly applicable methodology is presented to reliably predict crude oil liquid viscosity from only a gas chromatographic assay composition (C30+ is recommended). The viscosity model employs a Walther-type correlation of double log viscosity with log temperature to predict the viscosity of dead and live crude oils and mixtures. The model has three parameters: the slope and intercept of the Walther plot and a viscosibility factor to account for pressure effects. Simple mass based mixing rules are applied on these three parameters to obtain mixture viscosity. The three parameters were correlated to component molecular weight and therefore a gas chromatographic assay is the only required input apart from the temperature and pressure.
The methodology was developed from a Western Canadian dataset of two bitumens, one heavy oil, and one condensate, and then tested on an independent dataset of 10 conventional and heavy crude oils from the Gulf of Mexico, the Middle East, Asia, and Europe. The model provides un-tuned viscosity predictions within a factor of two of the measured values for dead and live crude oils ranging in viscosity from 0.5 to 500,000 mPa.s. A single multiplier is used to tune the model. Models tuned to dead oil data predict live oil viscosities and mixtures of oils with solvents to within 30% of the measured values. Models tuned to the viscosity at the saturation pressure predict the effect of temperature and pressure to within 20% of the measured values.
The method retains its accuracy when components are lumped into a few pseudo-components and is ideally suited for use in simulators for accurate liquid phase viscosity predictions over a wide range of compositions, pressures, and temperatures. It would be necessary to include the proposed mixing rules in numerical simulators. An additional advantage of the method is the reduction in viscosity measurements needed to construct an accurate viscosity model.