oa Assisted History Matching with Application of Adjoint Method Sensitivity Computation: Case Study North German Basin Oilfield

History matching, which is an inverse problem, is traditionally performed by a trial and error approach to minimize the mismatch between observed and simulated data. Modification of parameters on sequential simulation usually leads to rock properties which are way far from geological interpretation. This renders the predictive power of the simulation model doubtful. In the presented approach, the adjoint method is used to capture the derivatives of the mismatch (sensitivities) with respect to each parameter at the grid level. Adjoint methods derive the analytical sensitivities based on prior knowledge of fluid flow equations implemented in a dynamic simulator. During the modification step, the sensitivity and rock property updates are iteratively calculated and implemented grid cell by grid cell until convergence is reached. The workflow is applied to history matching of an abandoned North German oilfield model with long production life. The outcome suggested that, with the use of this technique, improvements can be achieved beyond the scope of manual approach using a small number of simulation runs. Both the history match quality and the predictive capability of the dynamic simulation model are improved.