A Viscosity Correlation for Mixtures of Heavy Oil, Bitumen, and Petroleum Fractions

Abstract

Abstract High viscosity is a major constraint in the recovery and transportation of heavy crudes and tar sand bitumens. Viscosity reduction may be achieved by mixing the crude with a light petroleum solvent. This paper presents a generalized correlation for calculating viscosities of such mixtures. A power-law mixing rule was generalized by use of the Einstein-type relationship for the viscosities of infinitely dilute solutions. Literature and in-house data were compiled to establish the con-elation. The final correlation requires only density and viscosity of the two fluids to predict blending viscosity at any mixture composition. The correlation is tested with heavy-oil/solvent blending data and gives an excellent prediction of the blending viscosities. Introduction To reduce viscosity, solvents frequently are used to dilute heavy crudes. This is one of the most efficient methods of pipeline transportation of heavy oils. Solvents also are injected into the reservoir for well cleaning, stimulation, fracturing and, less frequently, for miscible displacement. Engineering application of these processes often requires calculation of mixture processes often requires calculation of mixture viscosities. This paper documents the development of a simple but generalized correlation for predicting viscosities of binary mixtures of heavy oil, bitumen, and petroleum fractions, with particular emphasis on heavy-oil/solvent systems. Background Previous Work. Viscosity of liquid mixtures has been Previous Work. Viscosity of liquid mixtures has been studied extensively. Ref. 1 gives a brief review of the object. In general, the mixture viscosity as a function of composition is extremely complex. Theoretical considerations have offered little help in explaining these behaviors. Attempts such as McAllister's to derive a generalized expression for viscosities of all mixtures inevitably resulted in equations with many undetermined constants. There is no reliable method at present to allow an a priori prediction of these constants. These methods, therefore, can be classified only as descriptive. Literature reports few predictive methods, and those are mostly empirical and often specific to a particular group of mixtures. For mixtures of liquid hydrocarbons, including petroleum oils and fractions, the viscosity-composition petroleum oils and fractions, the viscosity-composition curve is generally a monotonic, concave-upward function, and rarely goes through a minimum. Regardless of the function's simplicity, a review by API showed that no single correlation would represent the viscosities of all hydrocarbon mixtures. Some of the reviewed correlations include Arrhenius (Eq. 1), Bingham (Eq. 2), and Kendal and Monroe (Eq. 3). ............(1) ............(2) ............(3) In these equations, VA and VB are volume fractions, MA and MB are mole fractions, and A, B, and are the viscosities of components A and B and their mixture, respectively. API recommended Eq. 3 for the blending of pure hydrocarbons and a graphical Wright method for mixtures of petroleum liquids. The latter calls for the use of the ASTM D341 viscosity-temperature charts. The procedure is to plot the viscosity-temperature lines of the oils and then to "blend" by linear proportioning along the log T axis. A hand-held calculator program, is now available to replace this tedious graphical manipulation. The viscosity ratios associated with the API data are mostly in the range of 1 to 100, where the ratio is calculated as the viscosity of the more viscous component divided by that of the less viscous one. In application to heavy-oil systems, we are interested in mixtures with viscosity ratio of 10(3) and higher. The only published method intended for blending heavy-oil systems was reported by Cragoe. Cragoe defined a function L such that .........(4) and proposed to calculate from the mixing rule ......................(5) SPEJ p. 277