Pipeline Flow Of Heavy Crude Oil Emulsions

Abstract

Abstract Using a 53 mm test pipeline, emulsions containing up to 67% Lloydminster heavy crude oil in water hare been transported in laminar and turbulent flow. In combination with viscometry and velocity distribution measurements, the experiments showed that a homogeneous model was appropriate for laminar flows 0/ emulsions with median droplet sizes near 30 microns. Turbulent flow pressure drops could be interpreted with a conventional turbulent pipe flow model. Introduction Pipeline flow of oil in water emulsions occurs widely in field operations and has also been studied systematically in laboratory conditions. Zakin et. al(1) used oils with specific gravities between 0.78 and 0.923 in emulsions formed in a colloid mill. The emulsions were non-Newtonian at oil concentration of 50% by weight and above. Turbulent flow pressure drops were between 8 and 26% lower than values predicted from their laminar flow behaviour and the Dodge-Metzner correlation for turbulent non-Newtonian friction losses. Pal and Rhodes(2) used a mineral oil of comparatively low viscosity and inferred effective viscosities from the laminar and turbulent flow pressure gradients, assuming Newtonian fluid behaviour. With 45% and 55% oil in water emulsions, laminar flow effective viscosities were significantly lower than those in turbulent flow. The droplet size was unknown, however. With heavy crudes, inversion (3,4) of an oil in water emulsion to a water-in-oil emulsion can occur during pipeline flow with a catastrophic increase in pressure drop. On the other hand, Layrisse et. al. (5) found stable emulsions for a wide range of concentrations (to 65% oil) and shear rates with droplets whose mean diameters ranged between 6 and 86 µm. Good agreement was found between viscosities inferred from pipeline pressure drops and values measured with Couette viscometers. The rheology was strongly dependent on droplet size, however, with non Newtonian behaviour evident for smaller droplets, lower shear rates and the highest oil concentration. Turbulent flow predictions appeared to be within ± 20% of the expected values. In contrast with these results, Wyslouzil et. al. (6) found pipeline pressure drops to be significantly lower than values predicted using measured viscosities, in both the laminar and turbulent regions. The deviation was attributed to droplet migration away from the wall, producing a region of depleted concentration. With prolonged recirculation in a closed loop, an instabitity occurred which resembled the inversion reported by Gillies et. al.3. The initial droplet size of the Wystouzil experiments was in the range of 5–10 µm. Gillies and Shook(4) reported experiments conducted with high oil concentrations using laminar flows. Differences between viscometer and pipeline flow viscosities were also attributed to droplet migration and evidence for this was obtained in the form of velocity distributions measured with pitot tubes. In contrast with the droplets in the Wyslouzil et. al. experiments, the droplets were too large to allow their size to be determined photographically. A subsequent investigation, using the same oil, surfactant and preparation technique showed that the volume mean droplet size was approximately 120 microns in the Gillies and Shook (4) experiments.