Abstract A model was developed to represent the physical displacement mechanism of tertiary oil recovery in an aqueous-phase surfactant flood. The chemical aspects were not modeled. In particular, the residual oil saturation in the presence of surfactant must be specified to use the model. This model was used to investigate the relationship between the system parameters (mobility ratio, partition coefficient, parameters (mobility ratio, partition coefficient, adsorption) and the performance variables (oil cut, chemical breakthrough, recovery efficiency at breakthrough). The model is an extension of Buckley-Leverett analysis and applies to the flow of two fluids in a system in which composition and saturation are variables. This model assumes a homogeneous one-dimensional system, the absence of dispersion, equilibrium mass transfer, and constant composition injection (infinite slug). Analysis applies to systems of two mobile phases (oil and water) and one immobile phase (reservoir rock) where three components (oil, water, and chemical) transfer between mobile phases. and chemical transfers to the rock. The model predicts that oil recovery and surfactant breakthrough may be retarded in low-tension surfactant floods where the surfactant partitions preferentially into the oil phase. This partitions preferentially into the oil phase. This prediction is confirmed by experimental core-flood prediction is confirmed by experimental core-flood results. Introduction In designing and optimizing a surfactant-flooding process, one is confronted with many mechanisms process, one is confronted with many mechanisms and corresponding physicochemical properties of the rock and fluid interactions that affect performance of a surfactant flood. Important properties are relative permeabilities, viscosities, interfacial tensions, dispersion coefficients, and adsorption isotherms. Laboratory investigation of these mechanisms is hampered by the high degree of coupling among mechanisms, which makes it difficult to analyze process sensitivity to each property. property. Therefore, a simple mathematical model was developed to interpret results of core-flooding experiments and to apply in cases in which surfactant is injected continuously (infinite slug). A sensitivity study of the effect of 14 model parameters on oil recovery revealed that recovery parameters on oil recovery revealed that recovery is affected strongly by pore-to-pore displacement efficiency (governed by interfacial tension) and fluid mobilities, and by interphase mass transfer of chemical (surfactant), oil, and water. The effect of this transfer phenomenon on surfactant-flood oil recovery previously has received little attention. ASSUMPTIONS OF THE MODEL SYSTEM The system is one-dimensional with uniform properties. properties. Two mobile fluid phases, an aqueous displacing phase, and an oleic displaced phase are considered, as well as one immobile phase (rock). Three mobile components (oil, water, and chemical) are considered. Each is assumed to behave as if it was a pure component. Mass transfer of chemical, water, and oil between the mobile phases and transfer of chemical to the immobile phase is allowed. The mass transfer rates are assumed sufficiently fast compared with fluid flow that chemical equilibrium exists across phase boundaries everywhere in the reservoir. phase boundaries everywhere in the reservoir. Initially, the reservoir contains pure oil at its waterflood residual oleic-phase saturation (Sorw), and the rest of the pore space contains pure water. The injected fluid is a single aqueous phase at constant composition (infinite slug injection). This injected composition lies on the two-phase envelope of a ternary diagram (that is, no phase extraction). The system is self-sharpening in that only the injected and the initial compositions exist in the composition profile. (Sufficient valid conditions for this assumption are derived in Appendix B.) There are no chemical reactions. In each phase, each component occupies the volume it would have in its pure state (VE = O, the excess volume of mixing is zero). SPEJ p. 42