SOFT COMPUTING ALGORITHMS ACCELERATE AND IMPROVE THE HISTORY MATCHING PROCESS: ELK HILLS, 29R RESERVOIR

FIRINCIOGLU, Tuba¹, OZGEN, Cetin², O'BRIEN, William Jay¹, ALBERTONI, Alejandro¹, GIRBACEA, Radu Axente¹, and SAVAGE, Bill¹, (1) , [email protected], (2) CO

This paper presents the application of soft computing techniques to history matching problems. The objective was to provide the engineer with the necessary information to speed up and to improve the quality of their history match results. In our approach, the data required are the observed data, the simulation results, and the history match parameter set that describes each run. The observed data (production and pressure) are compared to the simulation results to calculate the mismatch associated with each run for pressure and each phase. Next, soft computing (learning) algorithms are used to establish the relationships among the history match parameters and the mismatches associated with the wells. Through this approach, the engineer can analyze the effects of different parameters on simulation results for a range of values without making additional sensitivity runs. Upon establishing the relationships among the history match parameters and the mismatches, genetic algorithms are used to find the global minimum, hence yielding the set of parameters for the next simulation run. Given an acceptable mismatch criteria, sets of parameters can be generated for multiple solutions to the same problem, and the associated statistics can be displayed. The combinations of history matching parameters that minimize the mismatch are proposed as different simulation run alternatives. These alternatives provide the best solutions with the given knowledge and significantly accelerate completion of the history matching task by eliminating the historically accepted approach of making a run, analyzing the results, and adjusting parameters until the history match is completed. We present two application examples. One is the successful application of this technology to a complex fractured (dual porosity) porcelanite oil reservoir with an aquifer (ElkHills 29R, Bakersfield). This field has 28 years of history with 37 production wells. Another is a well test in a faulted reservoir with a strong aquifer.