Cross-Verification of Adjoint and MCMC Workflows for Estimation of Prediction Uncertainties Including Historical Data

Reservoir simulation workflows leading from history to prediction are built on a number of alternative optimization and sampling techniques with different characteristics. Adjoint techniques derive analytical sensitivities directly from the flow equations of the simulator. In a model update step those sensitivities are used for property modifications on grid cell level. Derivative-free techniques like Markov Chain Monte Carlo are used for optimization and uncertainty quantification including history data. Both techniques are different in nature and support alternative modelling strategies like local vs. global, deterministic vs. stochastic. In this work we apply both techniques in alternative workflow designs to a recent “hierarchical benchmark case study for history matching, uncertainty quantification and reservoir characterisation”. An implementation of an adjoint technique is applied for analytical sensitivity calculations and local adjustments of rock properties in a history matching workflow. Markov Chain Monte Carlo is used for global optimization and uncertainty quantification. Well production data is used in the model calibration phase and history matched simulation models are carried forward to prediction. Both methods are used for cross-verification of prediction results.