The task of appropriately forecasting production and estimating reserves, particularly from liquid rich shale reservoirs is quite difficult seemingly in part due to multiphase flow effects. In the quest to find dependable ways of forecasting production from shale reservoirs, several methods have been used such as reservoir simulation, empirical and analytical methods. The objective of this paper is to present the results of a study to determine how to reliably forecast production from shale volatile oil reservoirs using compositional reservoir simulation, hybrid (combination) decline curve analysis (DCA) methods and a new data-driven semi-analytical and statistical technique called the Principal Components Methodology (PCM).
To start, we simulated production with four different reservoir fluids (volatile oils) using a commercial compositional reservoir simulator. Then, we tested a variety of traditional and hybrid DCA models (i.e., a model such as the SEPD model for transient flow combined with a different model like the Arps hyperbolic model with a fitting value of the parameter ‘b’) on simulated data for each fluid sample. Further, we also used PCM to forecast production and finally, we compared all the results.
Due to the complicated physics of flow in shale volatile oil reservoirs, we propose that the most reasonable reservoir simulation method should be compositional. While compositional simulation can be thorough, taking into account complex PVT and reservoir features, time required and lack of data can be limitations. Alternatively, empirical methods can be used to estimate production. However, existing empirical techniques have not been entirely satisfactory for production forecasting in liquid rich shale reservoirs. Also, presumably due to multiphase flow effects, lengthy transition periods between transient linear flow and boundary-dominated flow were observed on the diagnostic plots. Therefore, we concluded that hybrid DCA models are more appropriate for multiphase flow analysis. Rigorous production data analysis and forecasts using analytical methods can also be considered. Present analytical techniques assume single-phase flow, thus making their application for multiphase flow analysis somewhat unsuitable. Here, we have developed a semi-analytical/statistical approach for forecasting production that bypasses existing complications associated with empirical forecasting methods and predicts production with reasonable certainty.