Rafael Medeiros de Souza
Thesis: An Effective Optimization Method for Computational 4D Seismic History Matching.
This project aims to develop an innovative ensemble technique for computational 4D seismic history matching. The objective is to set up a computational optimization problem to derive a geologically reasonable reservoir model that matches both the fluid production data measured in boreholes (eg. oil/gas/water) and the 3D and 4D (time-lapse) seismic data.
This research addresses current challenges such as data integration, limited ensemble size, nonphysical updates and underestimation of uncertainties. The project goals include exploring the inversion statistics and the inter-relation between the ensemble inversions and uncertainty analysis to improve the inversion process and uncertainty analysis. This approach should provide better data integration, improved physical reasonableness of the updated models, increased reliability of their predictions and a quantitative uncertainty assessment. The method will be tested on a field data case study from the NW Shelf.
Why my research is important
This research project can improve reservoir management processes and decisions by developing an innovative modified Ensemble Kalman Filter method to help improve reservoir predictions and optimize resource recovery. 4D seismic data integration in computational production history matching poses a significant challenge. The benefits of developing a successful method, however, are very large. This approach may lead to faster reservoir model updating, optimised resource extraction, and better predictions of future reservoir performance. This can have a direct impact on field management, optimizing operational costs and improving production strategy. For example, improving the understanding of the reservoir flow regime in a typical giant gas field offshore WA could easily result in a 5-10% improvement in resource recovery, worth $1-10 Billion per field on average.