One of the widely known use cases of 3D simulation model is the field development optimization. This task usually requires multiple numerical experiments (simulation runs) under different scenario conditions. Analysis of the huge space of input parameters and multiple runs is time consuming. Also, there is a lack of logic in such process driven by common management requests "run the case". In the scope of the paper, the system approach for analysis of broad parameters space is presented combined with robust techniques for analysis of multiple numerical experiments. The method is demonstrated on the example of solving 2 tasks: determination of the optimal gas usage distribution and selection of the best water injection strategy.
2 separated 3D reservoir simulation models were used in the study. Models have been constructed in commercially available reservoir simulator and another tool was used to substitute variable model parameters by modifiers. Gas injection rates, maximum reservoir production rate, water injection rates, positions of water injection wells, and target reservoir pressure were investigated to find the optimal (in terms of cumulative oil production) reservoir development strategy. On the final step, optimal reservoir development strategy was utilized in the integrated reservoir simulation model.
Based on the analysis of multiple runs with different combinations of reservoir management parameters, the optimal development strategy was formulated for 5 reservoirs. Implementation of optimal positions of water injectors is adding 2% to recovery factor. Usage of proper injection gas allocation strategy is allowing keeping average reservoir pressure above minimum miscibility pressure during the whole reservoir development time and substantially increases cumulative oil production. Summarizing, implementation of the optimal reservoir development strategy (maximum reservoir production rates, gas and water injection rates, and injection wells position) is absolutely crucial for success of reservoir development. Advanced optimization approach using multiple realizations of input parameters was used. More than 100 runs have been simulated and analyzed with prior and posterior distributions of parameters.