Abstract: To address the shortcomings of common history matching methods such as large amounts of computation, abnormal update of reservoir parameters, and distortion of reservoir model corrections, ensemble smoother algorithm was adopted to repeatedly assimilate the same data by multiple iterations in the ensemble Kalman filter (EnKF) algorithm to derive the core formula of the ES-MDA algorithm (ensemble smoother with multiple data assimilations) and write automatic reservoir history matching software. In a case study of marine sandstone reservoir of Brent Oilfield in the North Sea, the automatic history matching program of the reservoir based on the ES-MDA was applied to the reservoir to conduct history matching for water injection, oil production and development of the oilfield. The results showed that the predicted data obtained from the numerical modeling of the reservoir matched the actual measured data by more than 90%, and characterized the porosity distribution of the real reservoir more accurately. The ES-MDA algorithm is advantaged by stable algorithm, high operation efficiency and accurate model updating. The research results are of great significance for realizing computer-based automatic reservoir history matching and real-time optimization of reservoir production.

Key words: history matching, mathematical model, simulation algorithm, data assimilation, ensemble smoother, ensemble Kalman filter