Assisted history matching, reservoir model update and optimization

PhD Candidate Brage Strand Kristoffersen

Main Supervisor Carl Fredrik Berg

Sponsor: Equinor

Optimization of wells in a reservoir is a time consuming effort. Reservoir uncertainties and structural challenges can result in the procedure providing suboptimal locations. This project
investigates new types of parametrizations that, by replicating the decision-making process of geosteering, might enable a more efficient mapping of the search space. Features such as channels and faults can be dealt with and taken into account without increasing the complexity of the problem, yielding more realistic and productive well configurations.

Project result: Numerical geo-steering using neural networks on a reservoir model 
An automatic well planning procedure that is fast and finds trajectories with high economic value