The goal of this work was to apply an ensemble-based technique for data assimilation (Ensemble Smoother with Multiple Data Assimilation, ES-MDA) to link probabilistic history match and uncertainties assessment of a petroleum benchmark field. The first step was the definition of the model uncertainties and its parameterization to stablish the prior ensemble composed by 500 models. After this stage, I applied standard ES-MDA followed by the localization technique which was considered taking into account the area of influence of each well. Then, I conducted production forecast to assess field behavior until the end of the field productive life and compared with the reference response. To evaluate the results, I used two mais indicators: the Normalized Quadratic Deviation with Signal (NQDS) and the Sum of Normalized Variance (SNV). I used NQDS to verify which models were within the specified tolerance while SNV was responsible for evaluating the loss of variability of the ensemble. Through the standard ES-MDA, it was possible to improve knowledge about the reservoir behavior and obtain good data matches with low computation effort. However, I observed certain loss of variability and spurious correlation in the models. Afterwards, I applied ES-MDA with localization to improve results and this approach helped to increase variability of the ensemble and generate smoother images. It was crucial to compare the production forecast with the reference response to validate the methodology. The main advantage of this work is the ability of analyzing the reduction of uncertainties aligned with the benchmark case with a known response, assessing how strong the loss of variability of the ensemble was and comparing the convergence of the forecast with the reference response. Furthermore, this technique required a low computation effort when compared with other similar methods.

You can access this article if you purchase or spend a download.