| dc.description.abstract | A mechanistic reservoir simulation model is presented to study the time-rate relationships of flow in a multi-fractured horizontal well in unconventional reservoirs involving (a) a three-phase oil-gas-water system and (b) a two-phase gas-water system. In addition, non-mechanistic scenarios are also considered, including early-time high water production (i.e., flowback) and fracture permeability degradation. The results are compared to those from previously published models of decline curve analysis, and are used to develop a new time-rate model.
A "fit-for-purpose" numerical reservoir simulator is developed and, following validation against analytical solutions, is used to generate time-rate data. Several simulation cases were constructed from various reservoir and fluid properties that were gathered from published literature to model typical conditions of major US unconventional plays. The reservoir simulator models the pressure-dependent reservoir and fracture properties, the multiphase-multicomponent flow of black oil and gas-water systems, and accounts for non-laminar (Forchheimer/Klinkenberg) flow and gas adsorption in gas-water systems encountered in shale gas reservoirs. A new decline curve analysis model is proposed based on the data generated in this study, and is compared to the Modified-Hyperbolic and the Power-Law/Stretched-Exponential decline curve analysis (DCA) models.
Four flow regimes were successfully generated by the mechanistic model: an early-time fracture-dominated performance, a transient (linear) flow regime, a transitional flow behavior, and a boundary-dominated flow. The effects of spatial discretization on the accuracy of the numerical solution were investigated and proved to be significant, particularly for the wetting phase — leading to a conclusion that coarser grids tend to overpredict the wetting phase production.
The "non-mechanistic" scenarios were found to only affect the early-time production performance when compared to a given "mechanistic" scenario both in the three-phase black oil case and in the two-phase gas-water case. Production profiles generated in the "non-mechanistic" cases were shown to converge to the "mechanistic" ones no later than in 30 days. The additional computing time that was necessitated by the simulation of the "non-mechanistic" behavior, coupled with the lack of any significant impact on the production rates, suggests that the effects investigated should not be considered.
A "K1-Exponential (K1X)" DCA model is proposed based on the results generated in this work. The K1X model was shown to fit the flow regimes and its derivatives observed in both black oil and gas-water cases quite well. The features affecting the terminal decline phase of production were identified from the sensitivity analysis of production profiles to various simulation input parameters. Based on the sensitivity analysis results, the terminal decline parameter correlations to the reservoir, fracture and fluid properties were developed for the new K1X model, as well as for the standard DCA models. | en |
