Learning Reservoir Dynamics Metrics of Improved Oil Recovery Projects by Harnessing Real-Time Analysis and Data Analytics

摘要

Real-time analysis and data analytics have become cornerstones in reservoir management of waterflood operations or conformance program. Connectivity between injector-producer pairs and premature breakthrough of injected water or gas are perennial issues that can make or break the economics of secondary and tertiary recovery projects. In this study, we aim to harness the advances in modern data analytics and real-time analysis to systematically evaluate a suite of standard diagnostics tools and propose novel ones for improved recovery projects. Although the scope of these reservoir dynamics evaluation tools can be extensive, our current investigation utilizes data from the Permian Basin. A suite of reservoir models under varying conditions involving water injection helped understand and evaluate a number of diagnostics tools and devise new characteristics plots. We performed over 8,000 model runs and used data analytics to assess these tools. These tools include water/oil ratio (WOR) vs. time plot, Chan diagnostics, reciprocal-productivity index (RPI) plot, gas/oil ratio (GOR) vs. time plot, among others. We investigated the well-spacing effect ranging from 20 to 320 acres, grid effects, and heterogeneity effects in evaluating these tools. We also explored heterogeneity measures, such as the Dykstra-Parsons method and an index based on final hydrocarbon pore-volume injection (HCPVI), and ultimate recovery. Both cluster analysis and K-means statistics aided this screening process. This study illuminates initial-RPI (IRPI) that has a linear relationship with the ultimate recovery. Cluster analysis of the spread and uncertainty in final recovery vs. IRPI reveals scale invariance. In other words, this diagnostic plot can correctly identify and cluster the spread in final recovery under various well and reservoir quality scenarios, irrespective of the well spacing. Comparison of the final RPI and that at breakthrough with those at initial conditions suggests that IRPI can be a relevant reservoir performance indicator. This study shows critical parameters for oil recovery under waterflooding are reservoir flow paths and connectivity between layers, reservoir storativity, fluid properties, the thickness of the oil/water transition zone, and fluid mobilities. We also observed that the water breakthrough time does not show a clear relationship with IRPI. Nonetheless, the HCPVI at breakthrough time exhibited a linear correlation with the ultimate oil recovery. In the absence of water production or the presence of water channeling a linear trend emerges for the final HCPVI plot. Cluster analysis and real-time production data analysis have demonstrated the strength of a new reservoir dynamics indicator plot of ultimate hydrocarbon recovery vs. initial reciprocal productivity index. Combination of this indicator and traditional diagnostics and heterogeneity index can quantify the spread of final recovery efficiently.