There are major computational challenges and substantial uncertainties associated with model inversion and performance optimization in oil field applications. As production and (in some cases) 4D seismic and electromagnetic data are collected, there are inevitably discrepancies between the predicted and actual reservoir responses. This drives the need for data assimilation, usually referred to as history matching in petroleum engineering. History matching involves the solution of a computationally demanding, highly ill-conditioned, inverse problem. Key complications that arise are the uncertain nature of the geology (and thus the need to “properly” sample the posterior distribution), combined with the need to retain geological realism in history-matched reservoir descriptions. Additional complexity enters as a result of the unknown rock-physics quantities needed to integrate multiphysics data sets. Computational optimization is also of great importance for oil and gas production, given the complexity of the reservoir flow response and the very high costs associated with large-scale field development. An emerging area is the joint optimization of field development and operation, in which decision variables could include the sequence and type (producer or injector, vertical or horizontal) of wells to be drilled, well locations, and time-varying controls. This problem involves real, integer and categorical variables and is thus a MINLP problem. Additional issues that arise are the need to perform these optimizations under uncertainty (i.e., robustly), and the need to treat multiple, possibly conflicting, objective functions.
This workshop is expected to include discussion of computational procedures addressing the following topics, among others:
This workshop will include a poster session; a request for poster titles will be sent to registered participants in advance of the workshop.
Lou Durlofsky, Chair
(Stanford University)
Eldad Haber, Co-chair
(University of British Columbia)
Jan-Dirk Jansen, Co-chair
(Delft University of Technology)
Albert Reynolds
(University of Tulsa)
Xiao-Hui Wu
(ExxonMobil)