Since their first successful application in the field of oil recovery optimization about 10 years ago, methods for optimization based on approximate gradients have gained consistently growing popularity. This is largely due to the flexibility these methods offer for use with commercial simulators, which are treated as a black box, as well as to the relative efficiency when used for robust optimization. In this presentation I will summarize the main theory underlying these methods, discuss their connections, and review published applications to real fields and remaining challenges. Furthermore, I will present recent developments on e.g. applications to field development optimization problems, and the latest results from ongoing research on improved sampling strategies.
Copyright. All Rights Reserved.