Abstract A new method is developed to represent the phase behavior of multicomponent systems. This method uses fewer pseudocomponents than true components, but unlike conventional methods in which pseudocomponents are often chosen arbitrarily, the pseudocomponents are often chosen arbitrarily, the method uses regression analysis to find a "best" set of pseudocomponents.The method is applied to two surfactant systems of the type used for tertiary oil recovery. One system contains a crude oil (from the North Burbank Unit, Osage County, OK) and the other contains a pure hydrocarbon, 1-phenyltetradecane. For both systems the representation in terms of the pseudocomponents chosen by the regression analysis is significantly more faithful than that obtained by conventional methods.Since the considerations discussed here are general, they should be applicable to a wide range of phase studies in multicomponent systems. For example, they should be illuminating when applied to oil recovery by gas injection (carbon dioxide, natural gas, etc.), and to extraction processes, as well as to surfactant systems. Introduction Systems containing surface active agents have attracted a great deal of attention in connection with tertiary oil recovery. In many of these systems optimum oil recovery has been found to be strongly correlated with the phase behavior of these systems. To understand completely the basis for these correlations, one must be able to represent the phase behavior of systems containing surfactants adequately.Surfactant systems for tertiary oil recovery usually contain at least five components: oil, water, surfactant. cosurfactant, and electrolyte. The isothermal, isobaric phase diagram of these systems can be represented in a phase diagram of these systems can be represented in a four-dimensional space. Because physical representation (in three dimensions) of such a diagram is impossible. various techniques have been developed to attempt to represent the phase behavior in lower dimensional spaces. All these techniques correspond to projections of the original diagram onto lower dimensional spaces. Although almost unlimited methods of projection exist, only a small fraction convey useful information.Two projection schemes having some similarities but different intents and consequences are straight mathematical projection and "pseudocomponent" projection. Straight mathematical projection refers to the process of directly projecting the four-dimensional data for the entire phase diagram along some specified direction onto a three-dimensional space. All information parallel to the direction of the projection is lost. For example, if the rays of projections are parallel to the oil/water edge of the phase diagram, the resulting representation contains no information about the relative amounts of oil and water in the phases. In principle, this problem can be circumvented by generating two representations corresponding to projections of the same data along two different directions. For example. a second representation could be produced corresponding to a projection parallel to the water/alcohol edge of the phase diagram. parallel to the water/alcohol edge of the phase diagram. Although neither representation contains complete information, the pair of representations does contain all information about the system. The problem with this method is that the information is not perceived easily and, since the intent of using phase diagrams is usually to make visible a summary of the phase trends, such mathematical representations are not very useful.The second and generally more useful projection scheme uses pseudocomponents. A pseudocomponent is some mixture of pure components treated as a single component. SPEJ P. 289
Read full abstract