Steady-state modeling of a project is typically performed by separate teams of engineers working various tasks such as the design of the subsystem, production, and processing facilities. This setup allows for each team to share design criteria, but a recent study from a pair of engineers at Chevron posited that these criteria were based on a limited set of predefined discrete assumptions for each subsystem, and that the setup does not enable teams to properly assess the changing conditions that take place during the full asset lifecycle. This gap, the study said, often results in less-than-optimal facilities design, which in turn leads to cost overruns and the potential for lost production.
At the 2018 SPE Annual Technical Conference and Exhibition (ATCE), Chevron facilities engineer Carlos Yengle, one of the authors of the study, spoke about the company’s efforts to build a fully integrated model that allows for the evaluation of diverse operational scenarios, with the goal of finding optimal settings for short- and long-term asset production needs. Yengle co-wrote the paper (SPE 191485) with Hemant Kumar, a petroleum engineer at Chevron, outlining the construction and testing of the model, which was built using commercial simulator packages and used as a support tool to help improve asset design.
Yengle said that the current methodology for steady-state modeling presumes that the process facility can process the fluid delivered by upstream systems at a predefined composition, pressure, and temperature. However, individual system complexity and the strong connectivity between systems make asset design and operation difficult, and changing one parameter within a system can affect the entire asset. Yengle and Kumar proposed a model that integrates the full asset, from the reservoir to the pipeline to the production facility.
This model, Yengle said, would still take into account the different modeling software programs that each team within a project may be using.
“All of these different models, they will have to be talking to each other in order to get a full asset integrated model,” Yengle said at ATCE. “At the same time, we need to respect the attention to detail for each discipline, the assurance for the models, the software they’re trusting for their models. We need to respect what they’re doing.”
The model included steady-state models for reservoirs, wells, production and injection networks, and processing facilities. Yengle said it provided realistic optimal operating conditions and long-term production forecasts of oil production, sales gas, and water injection requirements by incorporating the physics at appropriate levels to ensure higher accuracies than discrete models.
Chevron used asset data from an existing unnamed major capital project in the early development stages to pilot test the full integrated asset modeling approach, with the main objective being the validation of the asset’s design capacity. The processing facility comprised gas/liquid separators, followed by sales gas compression, fuel gas generation, and condensate production. It was modeled using AspenTech HYSYS, a commercial process simulator for facility design and operations, while phase equilibrium in the processing facility was evaluated using equations of state. RESOLVE, a global orchestration software platform, was used to integrate the coupled model with the processing facility, and as the platform to control, monitor, and manage the data transfer.
Yengle and Kumar wrote that a visual workflow allows the user to construct a logical flow of information from a set of easy-to-understand elements without writing code. By linking blocks of commands together into a single system, users can make the same decisions and take the same actions in a simulation as he or she would manually.
The pilot test showed a successful data transfer from the coupled model to the process facility model. The integration of processing facility steady-state models with the reservoir-network production simulation models is possible, but Yengle said it requires a high level of flexibility. He said that eliminating the artificial boundary conditions that come with independent production and processing facility design can help obtain a more realistic forecast for the asset’s production.
“I think the most important thing to take is that restraints from the different areas of the asset can be avoided, transferred, and utilized by any of the different integrated models,” Yengle said. “So all those different variables and all of those different systems can now operate by talking to each other. If there’s a constraint in the processing facility, let’s bring it back to the reservoir model and test that.”