Distributed Pressure Sensing for Production Data Analysis

Abstract

Well testing is an essential tool to estimate reserves and forecast production. The assessment depends on the analytical solution of the continuity and diffusivity equations which results in average reservoir properties. The key challenge is to acquire real-time data of pressure pulse signatures as they propagate and reach the boundaries. A potential solution is to use a permanent downhole pressure gauge or an array of distributed pressure sensors (DPS) placed at each hydraulic fracture cluster to characterize the flow. This work presents the elements of an innovative analytical model that uses this data to derive formation properties, visualize averaged flow dynamics, and evaluate the Stimulated Reservoir Volume (SRV). The real-time distributed pressure data, together with flow rate history, provides information that can be used to characterize the flow and estimate the boundary effect of hydraulic fractures and fissures. First, the numerically generated synthetic data is analyzed at each cluster to eliminate pressure drops due to friction. Next, analytical solutions of the continuity equation as well as trilinear models are used to invert reservoir properties to verify the proposed model. Based on the results, an advance statistical analysis is used to characterize the contribution of each variable to the flow rate. The numerical results suggest that there are key variables to identify different flow regimes. Numerical simulations are used to gauge the accuracy of the analytical model at predicting reservoir properties and flow patterns. Statistical analysis evinces that there are key parameters of the formation, fractures, and fissures that control the well productivity. The numerical analysis showed that for every reservoir type there are different combination of fracture parameters that can optimize the flow. Moreover, the results describe a method to obtain hydraulic fracture properties around each pressure sensor (DPS) and forecast their productivity. Finally, statistical learning was investigated as a potential solution to derive reservoir properties, including hydraulic and natural fractures, using the pressure pulse signature data without the need of inversion. The results show that there are key parameters that determine flow patterns. The importance of the accurate recognition and analysis of the multiple linear flow regimes at each cluster is in the potential to estimate the size of the SRV around hydraulic fractures during the transient life of the well. Moreover, this paper explains the procedure used for analyzing the change in the flow rate to obtain reservoir properties.