Next-generation modelling workflows improve reservoir management decisions

Abstract

Sheldon Gorell, Tom Smart, Keshav Narayanan, Pom Sabharwal, and Sarah Bassett of Landmark Graphics discuss the development of better, more economic reservoir management through the use of enhanced workflows that can be carried out multiple times in practical timeframes. In today’s global business environment, exploration and production (E&P) asset teams are making increasingly challenging and costly development and production decisions earlier in the life of a reservoir. These decisions are being driven by a number of factors including more complex reservoirs, more complex and expensive wells, costly facilities, smaller targets requiring better strategies for optimal multi-reservoir development, high oil and gas prices, and the need to minimize risk. The ability to accurately model field development scenarios before making significant capital investments can optimize oil extraction and yield tremendous business savings. Historically, reservoir simulation and associated workflows have been time- and compute-intensive, often forcing asset teams to choose between a more thorough analysis and acceptable turnaround times. Typically, a majority of complex reservoir situations have been solved by methods of approximation, in which some degree of fidelity was sacrificed within one or more of the domains being modelled. Examples of such trade-offs range from simplified tank models for reservoirs with sophisticated surface network models to loosely coupling sophisticated high-resolution reservoir models with approximate or iterative methods for coupling them together with surface network modelling. Traditional reservoir engineering/management workflows have consisted of linear processes in which each discipline hands off its version of the model to the next specialist. Fine-scale earth models that capture all the complexities and heterogeneities of the subsurface can be constructed using seismic, well, and geologic data. Often, these earth models are up-scaled to a resolution that can be used effectively in conventional flow simulation where they become the basis for reservoir simulation. At this point, a number of ‘what if’ scenarios are run for different uncertainties and development options. Then there is yet another hand-off to the facilities engineering and economics domains. The key problem with this methodology is that each domain often produces its results in a ‘siloed’ manner, tossing them over the wall to the next person in the chain. This familiar process has been especially cumbersome due to the difficulties in transferring digital data and models. The tools and technology available within each of the disciplines enables a high degree of rigor but there has been limited support for integration between disciplines. The focus is often on building the best model possible in each of the silos instead of optimizing the asset to deliver on a core business objective. As a multi-step process (very much analogous to the old game of ‘telephone’ where a message gets passed from one person to the next), the original objective often gets lost, producing a suboptimal result. The linear process has also prevented asset teams from evaluating multiple development scenarios prior to capital investments, creating greater risk. Additional downsides from this process include the lack of a repeatable process, no audit trail, and unnecessary constraints. Inaccuracies often creep in because each step fails to fully capture the interdependencies in the reservoir development value chain.