Characterization of a reservoir fluid requires representation of the fluid by a number of pseudocompounds and estimation of basic parameters such as the critical properties and acentric factor. Usually there are as much as a dozen characterization and thermodynamic models in a reservoir simulator. The impact of selected method on calculated properties such as interfacial tension, gas-to-oil-ratio (GOR) and pressure-temperature (PT) diagrams have been demonstrated. Recommendations for various systems are proposed.
Reservoir engineers usually calculate physical and thermodynamic properties needed for simulation of reservoir fluid flow and oil recovery through a reservoir simulator. Properties such as density, transport properties, vapor pressure and equilibrium ratios are normally calculated through generalized thermodynamic correlations and equations of states [1]. The input parameters for these equations are critical temperature (Tc), critical pressure (Pc), critical volume (Vc), acentric factor (?) and molecular weight (M). There are more than dozen methods for estimation of these properties for petroleum fractions. The most common methods used in various simulators are Lee-Kesler, American Petroleum Institute (API), Riazi-Daubert, Cavett and Twu which require boiling point (Tb) and specific gravity (SG) as input data. Among these methods the Lee-Kesler and Twu methods have been developed with critical data for heavy hydrocarbons calculated from vapor pressure data. The pseudocomponent method proposed for calculation of properties of heavy fractions requires a single characterization parameters (carbon number, molecular weight or boiling point) and the P/N/A composition of the fraction [2]. These methods are suitable for narrowboiling range fractions and petroleum cuts, howeverwhen they are directly applied to wider boiling range mixtures such as a heptane-plus (C7+) less accurate results are obtained. For this reason the C7+ fraction of heavy oils may be expressed in terms of several pseudocompounds through a distribution model. Riazi's distribution model has the following form [3]:
Equation(1) (Available in full paper)
where P*= (P - Po)/ Po. This model has three parameters A, B and Po. x is the cumulative mole, volume or weight fraction and P is a property such as Tb, M or SG. Po is the value of P at × =0, which corresponds to the initial value. Parameter B for specific gravity of most C7+ samples is fixed at 3 and for the boiling point is 1.5. When B=1 the model reduces to a simple exponential model. For light reservoir fluids such as gas condensates, value of B for molecular weight is one for most samples. However, the best optimum value of B varies from one mixture to another and from one property to another. Generally B for heavier fluids has a higher value. Use of this distribution model in representing a C7+ fraction by a number of pseudo-fractions is given in the previous publications [3]. In this work this model is used to represent a Kuwaiti reservoir fluid and calculation results are shown for GOR and the PT diagram with use of DB Robinson phase behavior simulator by several characterization methods [3]. However, for narrow fractions the interfacial tension is used to show impact of characterization methods selected from a simulator.
