Ocean Framework (Third Party Simulation), C#, .NET, C++
Petrobras operates with a diverse portfolio of reservoir simulators, including the CMG family (IMEX, GEM, STARS), widely used in enhanced recovery studies, heavy oil modeling, and compositional simulation. Although Petrel offers native support for some commercial simulators, integrating the CMG family within the Petrel environment required a dedicated coupling layer capable of translating Petrel's reservoir model into CMG input formats and bringing simulation results back into the project environment.
Running external simulators from Petrel without native integration creates significant operational friction: manual model exports, deck editing outside the environment, loss of traceability between the geological model and simulation results, and difficulty automating workflows that involve multiple runs, such as history matching and uncertainty analysis. In the context of BR-Kalman, the absence of this integration would make parallel execution of ensembles with CMG simulators unfeasible.
DeepSoft developed CMG Coupling, a BR-Kalman module responsible for integrating the CMG family of simulators into Petrel. The solution is built on top of the Ocean Framework's Third Party Simulation architecture, SLB's official mechanism for incorporating external simulators as first-class citizens within Petrel.
Through this architecture, CMG Coupling handles the conversion between Petrel's internal data structures (grids, properties, wells, history, events) and the input decks expected by CMG simulators, runs the simulator in a controlled manner, and reimports the results (saturations, pressures, rates, computed histories) into the project environment. The outcome is a user experience identical to that of a native simulator: the reservoir engineer configures, runs, and analyzes CMG simulations without leaving Petrel.
As a BR-Kalman module, CMG Coupling also enables parallel execution of hundreds of realizations on high-performance computing (HPC) clusters, an essential element for ensemble-based history matching.
IMEX, GEM, and STARS simulators now operate inside Petrel as first-class simulators, eliminating manual exports and out-of-environment deck editing.
The module enables CMG simulators in BR-Kalman's data assimilation workflows, allowing parallel execution of ensembles on HPC clusters.
Coherence between the geological model, the simulation deck, and the results is maintained by Petrel itself, reducing operational errors and preserving the project's technical history.