Joint History Matching of Production and 4D-Seismic Related Data for a North Sea Field Case

Abstract

Abstract This paper shows the results of an assisted history-matching process applied to a North Sea field case operated by Statoil. The dynamic data considered, that is the data varying with fluid flow, are both production data and 4D-seismic related data. The history-matching process entails a geology to fluid-flow simulation workflow, including the population of the geological model by facies and petrophysical properties using geostatistical algorithms, the simulation of fluid saturation and pressure changes in the reservoir and the computation of time-lapse seismic attributes. This workflow also addresses scaling issues: upscaling of petrophysical properties from the fine geological grid to the reservoir coarse grid; pressure and saturation downscaling from the reservoir grid to the geological grid where a petro-elastic model is applied to compute the seismic related attributes. A least-square objective function is defined to measure the mismatch between the measured dynamic data (production and 4D seismic-related data) and the corresponding simulated responses. Then, a gradient-based optimization algorithm is run to minimize this function, perturbing parameters identified as uncertain from a preliminary sensitivity study based upon Morris's method. To ensure the overall consistency of the workflow, the perturbations applied to the selected parameters are propagated all along the workflow. This methodology was applied to a North Sea field case. A good production data match was obtained perturbing the spatial distribution of some facies and petrophysical properties. The implemented perturbation process was driven by the gradual deformation method: many likely distributions were combined all together leading to a more suitable distribution, which still honored the statistical properties of the model. The match of the available 4D seismic-related data was also locally improved. In addition, the use of 4D seismic data on top of production data made it possible to reduce the data misfit at some wells. This study shows the ability of the assisted history-matching process to handle both production and 4D seismic data on a real field case.