Efficient reservoir characterization and production optimization using the augmented Lagrangian and level set methods

Updating of reservoir models with respect to dynamic data is a major challenge, which requires large effort from the field subsurface asset teams. Also, optimal use of available platform process capacities is a major challenge, especially for mature fiel ds with many wells, and for fields with new smart wells, where downhole valves can control flow from individual well segments. We want to improve today's technology for both these applications by applying modern technology for constrained optimization (b ased on the augmented Lagrangian method and level set methods) to the dynamic reservoir simulation models. In the augmented Lagrangian method the flow equations are considered as constraints to the minimization problem, and the solution of the flow equati ons and minimization problem is reduced to a series of much simpler minimization problems. Earlier results on similar problems show that this method can reduce the computational effort significantly compared to traditional methods. In this project these techniques will be further developed for more complicated model equations that are closer to equations used in real reservoir simulators. Also we will try to improve the methodology by combining with multiscale and multigrid type of methods. This will be the first attempt to apply such techniques to multiphase flow in reservoirs and will provide important new knowledge of this problem. Our aim is that the results of this project can be used in real reservoir simulators for the oil industry. Three Ph D students and one postdoc, as well as senior researchers at CIPR with experience from the oil industry, will work on the project. This will provide the students with a thorough theoretically education as well as ties to practical applications.