Production Decline Analysis of Horizontal Shale Wells in Presence of Stress-Controlled Flow Rates

Abstract Decline curve analysis (DCA) methods curve-fit historical production data of a well and forecast the recoverable reserve. However, these methods do not consider the underlying physics of production and therefore they are not used as diagnostic tools. In this paper a new physics-based DCA method is introduced for reserve estimation and as a diagnostic tool for unconventional wells with long lateral length and with a great many hydraulic fractures. The method considers that the shale wells produce at flow rates that are influenced by in situ stresses. Namely, the reservoir matrix delivers the fluids to the fractures in an effective stress field, and the hydraulic fractures deliver the fluids to the well under fracture closure stress. The stresses change dynamically during the production due to withdrawal of the fluids and pressure depletion. We propose the production decline due to stresses can be captured by considering both the matrix and the fracture have stress-dependent permeability. We consider exponential function to capture this decline with a time-dependent exponent a representing the overall permeability loss ratio. The ratio changes over the production time as a function of the geo-mechanical properties and the changing stresses. It is the new DCA parameter to be determined by curve-fitting the production data. We used fourteen Marcellus shale gas wells, twenty shale oil wells (including wells from Permian, Eagle Ford, Anadarko basins) and utilized MS-Excel solver to demonstrate the application of the new stress-controlled DCA. On average a 30% loss in Marcellus gas reserve is estimated due to stress-controlled production rate decline. The shale oil wells analyzed also show strong dependence to stresses. The new approach could help operators identify the shale wells producing with flow rates under significant stress effects but also help improve the reserve estimation by adding new relevant physics and, hence, reducing uncertainty. In addition, it may be considered as a simple proxy model to use along with the complex reservoir flow simulation modeling approaches or as alternative to simulation in the absence of data needed for reservoir modeling.