Improved Steamflood Analytical Model

Summary The Jones (1981) steamflood model incorporates oil displacement by steam as described by Myhill and Stegemeier (1978), and a three-component capture factor based on empirical correlations. The main drawback of the model, however, is the unsatisfactory prediction of the oil production peak: It is usually significantly lower than the observed value. Our study focuses on improving this aspect of the Jones model. In our study, we simulated the production performance of a five-spot-steamflood-pattern unit and compared the results against those based on the Jones model (1981). To obtain a satisfactory match between simulation and Jones-analytical-model results, at the start and height of the production peak, the following refinements to the Jones model were necessary. First, the dimensionless steam-zone size AcD was modified to account for the decrease in oil viscosity during steamflood and its dependence on the steam injection rate. Second, the dimensionless volume of displaced oil produced VoD was modified from its square-root format to an exponential form. The modified model gave very satisfactory results for production performance for up to 20 years of simulated steamflood, compared to the original Jones model. Engineers will find the modified model an improved and useful tool for the prediction of steamflood-production performance. Introduction Steamflooding is a major enhanced-oil recovery (EOR) process applied to heavy oil reservoirs. A steamflood typically proceeds through four development phases: reservoir screening, pilot tests, fieldwide implementation, and reservoir management (Hong 1994). Steamflood-performance prediction is essential to provide information for the proper execution of each development phase. Three mathematical models (statistical, numerical, and analytical models) are often used to predict steamflood performance. Statistical models are based on the historical data of steamflood performance from other reservoirs which have similar oil and rock properties. A statistical model, however, does not include all the flow parameters, and thus may be inaccurate for a particular reservoir. Numerical models usually require a large amount of data input with lengthy calculations using computers; and they are usually CPU-, manpower- and time-consuming and also expensive. They may be extremely comprehensive and better serve as tools for research or advanced reservoir analysis. Meanwhile, analytical models are more economical, but at the expense of accuracy and flexibility. They serve as tools for engineering screening of possible reservoir candidates for field testing (Hong 1994). For many years, attempts have been made to provide analytical models for steamflood-production-performance prediction (Marx and Langenheim 1959; Boberg 1966; Mandl and Volek 1969; Neuman 1975; Myhill and Stegemeier 1978; Gomaa 1980; Jones 1981; van Lookeren 1977; Farouq Ali 1970; Miller and Leung 1985; Rhee et al. 1978; Aydelotte et al. 1982). None of these analytical models gives a comparison with simulation results. Miller and Leung (1985) presented comparison between their analytical model and simulation results for cumulative production vs time, but the comparison for production rate vs time is not available.