Analysis and Comparison of Decline Models: A Field Case Study for the Intercampo Oil Field, Venezuela
Summary Intercampo oil field, which contains unconsolidated reservoirs driven by edge water and bottom water, is characterized by heavy oil with mid-high permeability and high oil saturation. The three classical models of the Arps model were applied in 13 horizontal and vertical wells in the oil field; also, the paper introduces two models that are not widely applied for decline analysis and forecasting in the wells. Decline features between vertical and horizontal wells were compared. The results accord well with the actual data from the oil field. The authors point out that these decline analysis models are applicable not only for vertical wells but also for horizontal wells. The authors would like to emphasize that four decline models discussed in the paper. In regard to screening and comparison of decline analysis models, this paper illustrates how to select and use a model, as well as the model's application conditions and their features. The screened models are recommended for production performance analysis of wells, reservoirs and oil fields. Introduction Existing decline curve analysis techniques, which include three Arps models (exponential, hyperbolic, and harmonic, 1945), and the Fetkovich model (1980), are derived empirically; the Arps models are still the preferred method for forecasting oil production and proven reserve. These methods have played a very important role in the exploration and development of oil fields worldwide (Arps 1945, Arps 1956, Fetkovich et al. 1980, Fetkovich et al. 1987, Fetkovich et al. 1996). Gentry and McCray (1978) presented a method to define decline curve. They claimed their equation might be superior to the Arps equations by defining certain decline curves. However, the model was derived from the hyperbolic model of the Arps model; their equation has a parameter qi of initial production rate computed by the Darcy Law. This means that the application of their method requires more parameters, such as relative permeability curve, radius of drainage, formation thickness, reservoir pressure at external drainage radius, and well bore terminal pressure. On this point, in their example the extrapolation with their model is not seen because the method is not a pure production-time relationship. Furthermore, use of this model to extrapolate future production is restricted by the data requirements. Li and Horne (2002, 2005) developed an analytical model, called the Li-Horne model, based on fluid flow mechanisms. The model was developed under the spontaneous water imbibition condition. Li and Horne also thought it difficult to predict which Arps equation a reservoir would follow. However, they made a conceptual error in their reasoning of the Arps models. In fact, we need to judge the decline type before using the Arps model to make production decline analysis. Li and Horne used only two special cases of decline exponent, n = 0 and 1, then compared the exponential model and harmonic model with any models. Hence, we think Li and Horne's comparison of several oil fields is not meaningful in cases where they did not get a concrete decline exponent n. When the Li-Horne model was applied to the actual oil fields, the values of a0 and b0 were regressed from the actual oilfield data, but not the calculation values from their equations. Because the models constants of the Arps and Li-Horne model regress from the actual oil fields, they include different reservoir type and fluid flow information (high permeability, low permeability, naturally fractured low permeability, complex, fault reservoir, etc.; single flow and multiphase flow, etc.). Therefore, the decline analysis models based on purely statistical models do not have any association with fluid flow mechanism, reservoir types, fluids characteristics, steady or unsteady flow, and single or multiphase flow. We are inclined to refer to this as an empirical rather than an analytical model. The other two decline analysis models introduced in this paper, the Orstrand-Weng model (Arps 1945, Weng 1992) and the T model, were both proposed for predicting oil field production in China during the 1980s. The main purpose of this paper is to compare application conditions and results among four models: Arps, Orstrand-Weng, T and the Li-Horne model.