Forecasting Production of Shale Volatile Oil Reservoirs Using Simple Models
Conference Paper
Overview
Identity
Additional Document Info
Other
View All
Overview
abstract
AbstractDue to the advent of shale reservoirs as important sources of oil and gas production, there has been a critical need for reliable techniques for forecasting production and estimating reserves from these plays. This work presents the results of a study to determine whether simple decline models can be used to forecast production and estimate reserves of both oil and gas in shale volatile oil reservoirs with reasonable accuracy. We found that hybrid models (i.e., one model such as the Duong model for transient flow coupled with a different model, such as the Arps hyperbolic model with an appropriate value of the parameter "b") are appropriate for forecasting oil production and that it is possible to forecast solution gas production with availability of adequate data.To establish the "truth case" for comparison, we simulated production with four different reservoir fluids using a commercial compositional reservoir simulator. We tested a variety of hybrid and simple DCA models on simulated data (and later on field data) with a work flow that included identifying flow regimes with diagnostic plots for each fluid sample. We analyzed 0.5 to 3 years of production history to estimate model parameters for forecasting future production. Also, we forecasted gas production using a technique similar to one recently presented in the literature.Lengthy transition periods between transient linear flow and boundary dominated flow (BDF) were observed on the diagnostic plots. This is presumably due to multi-phase flow effects, as this transition period is typically much shorter in single-phase shale reservoirs. As in single-phase flow, the Arps decline model proved to be good for forecasting in the BDF regime. For transient flow, we examined the Duong model (which includes transient linear flow as a special case) and YM-SEPD (modified form of Stretched Exponential Production Decline model). In some cases, we applied the Arps model for both the transition period and BDF regimes (with different b values). We concluded that hybrid DCA models are more appropriate for multi-phase flow analysis than simple DCA models. In most of the cases, the hybrid YM-SEPD and Arps models led to more accurate oil production forecasts than other alternatives. Our analyses of solution gas forecasts indicated that solution gas production forecasting is possible but there is still need for more research in this area.
