History matching processes for complex and large reservoirs have always posed difficulties to reservoir engineers. To help reservoir engineers during history matching, various assisted history matching (AHM) algorithms have been developed. While AHM can help automate various aspects of history matching, oftentimes, algorithms suffer from slow convergence. This work proposes an ensemble based markov-chain Monte Carlo (MCMC) based algorithm with efficient sampling from the given distribution of properties. For efficient sampling properties during AHM, streamline trajectories are used to find the connection between source(s) and producer well. Streamline tracking based on output of the full-physics simulator is used as a guideline to capture the fluid flow patterns, and only properties of grid cells along the streamline trajectories are considered prime candidates for history matching. The proposed algorithm was applied to a sector model of a reservoir as a test case study to history match water cut on a well-by-well basis.

You can access this article if you purchase or spend a download.