Comparison of Various Deterministic Forecasting Techniques in Shale Gas Reservoirs

Abstract

Abstract There is a huge demand in the industry to forecast production in shale gas reservoirs accurately. There are many production estimation methods including several variations of decline curve analysis (DCA), analytical simulation, and numerical simulation. Each one of these methods has its advantages and disadvantages, but only the DCA techniques can use readily available production data to forecast rapidly and, to some extent, accurately. Traditional DCA methods in use in the industry, particularly Arps’ decline model, were originally been developed for wells in boundary Dominated Flow (BDF). By contrast, in shale reservoirs, the dominant flow regime is long-duration transient flow. Therefore, the petroleum industry needed to develop newer models to match data in transient flow regimes and then for forecast production using these transient flow models, followed, in necessary, by BDF models. The Stretched Exponential model, the Duong model and the Arps model with a minimum terminal decline rate all have the ability to match and forecast wells with transient flow followed by BDF. In this paper we propose revisions to the Duong model, to provide better fits to data in BDF regimes. A thorough analysis of actual well production and analytical simulation results were performed on selected wells in gas shales to compare the various DCA models to ascertain the model that provides the lowest discrepancy in estimates of remaining reserves. Individual well and grouped well analyses were performed to check the efficacy of the various models. We concluded that, in most cases, the newer decline models, such as the Duong method and its modifications, provide more accurate estimates of reserves for individual and grouped data sets than the Arps decline methods. The outcome of this research should assist the industry to forecast gas production rapidly and more accurately in shale reservoirs. The grouped data methodology will enable us to forecast production in shale reservoirs even more rapidly. The grouped data approach will prove to be especially valuable when only limited data are available from wells with less than a few months of production history.