Digital Twin for Model Based Operational Support: Innovative Applications

Abstract Within BP, network simulation and optimization are a well-established methodology for increasing and sustaining production capacity. To improve on existing workflows, BP has enhanced its previous petroleum/process engineering-focused toolkit and has globally deployed a production system digital twin end-to-end; Reservoir, Wells, Plant and Export, for model-based surveillance and optimization. The production system digital twin is a cloud-based system that connects sensor data from each asset's data historian to an equipment data model and first principle steady state simulation tools to create a reliable status of the well network and processing facilities. This facilitates multi-disciplinary collaboration and is remotely accessible by global teams. The integrated digital twin has three modes of operation: monitoring, simulation, and optimization. In monitoring mode, the models are automatically updated hourly with real-time data and key simulation results collected and saved. These monitoring simulations provide virtual sensor output via bespoke algorithms, revealing information that either real sensors cannot measure or are not installed to measure. Engineers utilize what-if simulation or optimization mode to test scenarios and explore capacity increase options or analyze optimization potential. This paper will feature three practical applications to demonstrate realized values. It will describe elements of the integrated digital twin models deployed in BP's Gulf of Mexico assets, and outline the challenges and lessons in maintaining and auto-calibrating the digital twin.