Shale Reserve Forecasting - Model Consistency and Uncertainty

Abstract

Three general categories of modeling are traditionally used to provide shale reserve forecasting -(1) decline curve analysis (DCA), (2) rate-time analysis (RTA), and (3) numerical model history matching (HM). The focus of this paper is aligning each of the three modeling approaches to ensure maximum consistency in terms of fundamental reservoir description, including (but not limited to) initial fluid in place, reservoir rock properties, PVT, well completion factors, fracture area and conductivity, well controls, and definition of infinite-acting and boundary-dominated flow regimes. The HM model approach, though more rigorous, is time consuming and cannot be used for the hundreds of wells in a typical shale field. We recommend, as have others, that history-matched numerical models be used to help calibrate RTA and DCA models in a consistent manner for all wells. Once a consistent model framework is achieved, reserve forecasting can be better understood by regulators, engineering and reserve teams within the operating company and their partners. Furthermore, a consistent modeling framework can provide more reliable uncertainty analysis to establish probabilistic reserves estimates in terms of P90-P50-P10 values (1P-2P-3P).Modeling methods used in forecasting shale reserves are based on production data that includes rates and pressures. DCA applies the boundary-dominated methods such as Arps, where multiple time regions are used to capture infinite-acting and boundary-dominated flow. RTA uses dimensionless pressure and rate solutions applicable to horizontal wells with multiple fractures, including superposition, pseudopressure and pseudotime. Numerical models solve the complex set of differential equations describing multiphase fluid flow using a properly-selected grid refinement (e.g. near fractures) and, in some cases, a dual porosity/dual permeability treatment of fracture-matrix interaction.