Establishment and Application of the Multi-Peak Forecasting Model

Abstract After the development of the oil field, it is an important task to predict the production and the recoverable reserve opportunely by the production data. At present, the forecasting methods include reservoir numerical simulation, forecasting model, material balance and production decline, and so on. Of these methods, the forecasting model is a useful and effective one for entire forecasting. The famous Generalized Weng model, Weibull model, Rayleigh model, HCZ model, Hubbert model, Lognormal distribution model are major forecasting models. But these models are all single peak development model, they can't get the reliable results for the bi-peak and multi-peak development model. In this paper a generalized single peak forecasting model and its derived models were established. From the generalized single peak forecasting model, we can derive some famous forecasting models, such as Generalized Weng model, Weibull model, Rayleigh model, Arps Exponential Decline model, Modern model ?, Modern model?, and so on. Subsequently, a multiple peak forecasting model was proposed based on the generalized single peak forecasting model, by which the multi-peak model production, cumulative production and recoverable reserve can be obtained. Furthermore, the resolution method of multi-peak forecasting model parameters was also introduced by nonlinear automatic matching solving techniques. According to the practical application for some oil fields, both generalized single peak forecasting model and multiple peak forecasting model are all confirmed to be practical and effective. Introduction It is important to forecast the production and recoverable reserve during oil field development and management, and the forecasting model is one of the important methods to predict the future production. Forecasting models can be divided into Periodic model and Growth model according to their characteristics. The Periodic model includes generalized Weng model, Weibull model, Rayleigh model, Gamma model, generalized model, and so on. The Growth model includes HCZ model, Hubbert model (which is also called Logistic model)??logarithmic normal distribution model, normal distribution model, Hu-Chen model and other models1–12. The above mentioned models are all based on the single peak development model. On the other hand, bi-peak and multi-peak development models are more constantly displayed during the development of oil field because of putting into production of new block, large scale development adjustment, implement of tertiary oil recovery, and so on. When the oil field development models are bi-peak model, tri-peak model or multi-peak model, it is hard to get the reliable forecasted result only based on single peak model. Based on generalized single peak forecasting model, this paper established the multi-peak forecasting model and its nonlinear multi-parameter automatic matching method. In addition, real data from three oil fields have been adopted to forecast the production, and the results are satisfactory. So the multi-peak forecasting model was approved to be practical and effective.