Viscosity of ‘live’ water-in-crude-oil emulsions: experimental work and validation of correlations

Abstract

A method for measuring emulsion viscosity under pressurized (1–100 bar) and temperature-controlled flow conditions is described. It makes use of a flow simulator shaped as a hollow wheel. As the wheel is rotating, the liquid inside will have a relative motion in the opposite direction of the rotation. The torque acting on the wheel shaft is measured and transformed, via a calibration model, into the viscosity of the fluid. The method has been applied to water-in-oil emulsions with several live North Sea crude oils with saturation pressures up to 100 bar and water cuts in the range 0–90%. The method is shown to be a useful way of obtaining estimates of emulsion viscosity for live oil systems in a relatively small scale apparatus. Even though the crude oils studied were rather different and had very different viscosities, the relative viscosities as a function of water cut up to at least 60% showed small variation and might be represented by a general ‘master curve’. The relative viscosities obtained experimentally have been compared to a temperature-dependent Richardson-type correlation [Rønningsen, H.P., 1995. Correlations for predicting viscosity of W/O-emulsions based on North Sea crude oils. Proc. SPE Int. Symp. Oil Field Chem., Houston, TX, USA, SPE 28968], and three other correlations proposed by Mooney [J. Colloid Sci. 6 (1951) 162], Pal and Rhodes [J. Rheol. 337 (1989) 1021] (P&R) and Pal [J. Colloid Interface Sci. 231 (2000) 168], respectively. Overall, when being tuned to measured data, the P&R correlation gives the best match, closely followed by the Mooney correlation, but for low and medium water cuts, the Rønningsen correlation is comparable. This correlation does not require any measurements for tuning. Although it may be slightly conservative at low to medium water cuts, and somewhat optimistic at very high water cuts, for practical purposes in design of flow lines, it provides a reasonably accurate first estimate of the effective liquid phase viscosity. The recent correlation of Pal [J. Colloid Interface Sci. 231 (2000) 168] was found to be extremely sensitive to the choice of tuning point and, thus, to be rather inaccurate either at low water cuts or high water cuts.