History matching and future production optimisation in reservoir modelling is being done more and more with the help of automated or assisted computing techniques, such as design of experiments, Ensemble Kalman Filtering or adjoint- and other gradient-based techniques. With the advent of these techniques, reservoir engineers also get less trivial applications of computer assisted optimisation at their disposal. In this paper we will present two such non-trivial applications.
In the first example we demonstrate how an adjoint-based history match, updating gridblock by gridblock permeability, can flag the problematic areas of a model. The results show that good data matches, requiring unrealistic values for updated parameters are not necessarily useless, but can actually provide a lot of information about unmodelled aspects of reservoir behaviour. Some of the learnings from such an exercise are of geological, others of operational nature. In the second example we use, again adjoint-based, gradients of net present value (NPV) with respect to production and injection rates to optimise well work-over timing. In the field under investigation, each injector well starts its life as a producer, for both geomechanical and operational reasons. Making use of the gradients of NPV with respect to the modelled production rate before the conversion and the modelled injection rate after conversion, the process of shifting the conversion moment forward or backward in time has been automated.
The first example allows for the creation of multiple history matched models, by updating static and dynamic models in parallel, while ensuring the geological concepts are maintained. The second example, well conversion moment optimisation, has been used to validate schedules for operational hoist allocation.
Both applications have had a significant impact on the view of the asset team on the practical usage of modelling: in the first one, a very useful multi-disciplinary platform to help steer thoughts on model improvement was found, while the second helped assess the overall impact of operational decisions on field development.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.