More and more hard-to-recover deposits involved into the development, and most of the fields under development are in mature production stage. Therefore, reservoir engineers are faced with task of quantifying the predicted indicators uncertainty.
There are several approaches to achieving this goal, ranging from expert assessment and sensitivity analysis to methods based on proxy modeling, response surfaces, Kalman filter, and ensemble smoothing. In the opinion of the authors Multi Data Assimilation modification of the ensemble smoothing is the most technologically feasible way of solving the problem creating and then analyzing a representative ensemble of model implementations.
This method, in contrast to popular methods based on the response surface, allows, without loss of accuracy, to run parallel calculations throughout the ensemble of implementations, which saves considerable time for making calculations. ES-MDA method application to create a representative ensemble of realizations of the real-field filtration model, which has a long development history is considerate in this paper.
This paper describes the experience of one of the first ES-MDA applications in Russia to such a complex facility as a three-phase multilayer sector filtration model with more than 60 years of development, with more than 150 wells, most of which with casing leakage.