Enkf Reservoir Assisted History Matching Method Combining Single Well Sensitivity Localization

Abstract

The invention belongs to the technical field of oil reservoirs, and discloses an EnKF oil reservoir auxiliary history fitting method fusing single well sensitivity localization, which comprises the following steps of quickly calculating and tracking the propagation time of pressure waves from a well point to a stratum grid by utilizing geological model static parameter field information on the basis of a process function equation; determining the dynamic maximum sensitivity area of each single well according to the propagation time, constructing a localization matrix based on the single well sensitivity areas, correcting the gradient of a data assimilation method by combining an EnKF method, eliminating pseudo-correlation, gradually fitting production and dynamically updating an oil reservoir model, and finally giving the optimal estimation of the model. Concept and actual example show that compared with a standard and an EnKF method based on distance truncation, the established FMM-EnKF model set oil reservoir production dynamic fitting effect is better, and the oil reservoir model geological characteristics after inversion are more accurate.

Description

technical field [0001] The invention belongs to the technical field of oil reservoirs, and in particular relates to an EnKF oil reservoir auxiliary history matching method integrated with single well sensitivity localization. Background technique [0002] At present, the existing technologies commonly used in the industry are as follows: [0003] History fitting is a key link that takes the longest time in reservoir numerical simulation. Automatic history fitting based on optimization theory with the help of reservoir numerical simulation technology has become a current research hotspot. Random Perturbation Approximation (SPSA), Ensemble Kalman Filter (EnKF) and their improved methods are currently widely used fitting algorithms, which mainly use the average sensitivity gradient estimated by a model set to approximate the real gradient, obtain the optimization direction and realize Optimization fixes for mesh parameters. In practical applications, considering the calculati...

AI Technical Summary

Problems solved by technology

[0006] (1) Most of the current mainstream automatic reservoir history matching algorithms have problems such as inaccurate calculation of the localized ensemble Kalman filter history gradient of single well dynamic sensitivity and false correlation, which lead to the inversion of reservoir model parameters ( Such as porosity, permeability, net-to-gross ratio, relative permeability and other formation static parameters) correction errors, reservoir model inversion distortion, deviation from the actual reservoir;

[0007] (2) Existing technical methods cannot accurately obtain dynamic sensitive areas of single wells

[0008] (3) The traditional distance truncation method performs localization processing, and the results obtained by eliminating gradient pseudo-correlation are difficult to match the actual formation conditions

Images