Combining Capacitance Resistance Model with Geological Data for Large Reservoirs
Abstract Capacitance resistance models (CRM) constitute reduced physics-based models that provide quantitative first order estimate of inter-well connectivity using production and injection data (without geology). However, application of these models to mature fields with large number of wells along with historical data and varying operating conditions is challenging and computationally expensive. In this study we present a novel hybrid approach that combines CRM with geological data for application to large reservoirs. CRM models become more challenging, when applied in larger domains, because the number of unknowns they assume for every injector and producer pair are inherently connected. To overcome this complication, we obtain radius of investigation (ROI) for every injector using Fast Marching Model (FMM) that characterizes the field based on the geological data. Specifically, FMM is used to determine unrelated producer-injector pairs and ROI for each injector that reduce the number of unknowns and hence computational complexity. This hybrid model approach is validated by comparing results with a standalone CRM model for both synthetic and field data. We validate the proposed method results using synthetic data for a producing field with 5 producers and 4 injectors that is generated by numerical simulation. The FMM accurately identifies unrelated injection-production pairs to reduce the number of unknowns in CRM model by 35% (20 to 13). After conceptual validation, we apply this hybrid approach to a mature field with 29 injectors and 46 producers where the number of unknowns are reduced by 57% after the non-related pairs are determined with FMM. This results in significant speedup of computations as compared to standalone CRM approach. In totality, we have developed and validated the proposed hybrid model of geology and CRM using synthetic and field data. On applying our learnings successfully, we propose a hybrid model that combines two reduced-physics models (CRM and FMM) to decrease computational speed and reduce uncertainty in the field. With increasing focus on digitization, these workflows can help organizations reallocate water injection without CAPEX to deliver return on investment for digitization projects.