US 11,927,717 B2
Optimized methodology for automatic history matching of a petroleum reservoir model with Ensemble Kalman Filter (EnKF)
Yevgeniy Zagayevskiy, Houston, TX (US); Hanzi Mao, College Station, TX (US); Harsh Biren Vora, Houston, TX (US); Hui Dong, Austin, TX (US); Terry Wong, Spring, TX (US); Dominic Camilleri, Katy, TX (US); Charles Hai Wang, Houston, TX (US); and Courtney Leeann Beck, Denver, CO (US)
Assigned to Landmark Graphics Corporation, Houston, TX (US)
Appl. No. 17/047,152
Filed by Landmark Graphics Corporation, Houston, TX (US)
PCT Filed May 9, 2018, PCT No. PCT/US2018/031794
§ 371(c)(1), (2) Date Oct. 13, 2020,
PCT Pub. No. WO2019/216892, PCT Pub. Date Nov. 14, 2019.
Prior Publication US 2021/0149077 A1, May 20, 2021
Int. Cl. G01V 99/00 (2009.01); E21B 49/00 (2006.01); G06N 20/20 (2019.01); G06Q 10/06 (2023.01); G06Q 50/02 (2012.01); G06F 17/00 (2019.01); G06Q 10/04 (2023.01)
CPC G01V 99/005 (2013.01) [E21B 49/00 (2013.01); G06N 20/20 (2019.01); G06Q 10/06 (2013.01); G06Q 50/02 (2013.01); E21B 2200/20 (2020.05); G01V 2210/663 (2013.01); G01V 2210/665 (2013.01); G06F 17/00 (2013.01); G06Q 10/04 (2013.01)] 13 Claims
OG exemplary drawing
 
1. A method for history matching a model of a reservoir based on actual production data from the reservoir over time, the method comprising steps of:
generating an ensemble of reservoir models using geological data representing petrophysical properties of the reservoir;
locating one or more logging tools in the reservoir;
acquiring production data corresponding to a first time instance from the one or more logging tools located in the reservoir;
performing normal score transformation on the ensemble of reservoir models to transform an original distribution of the ensemble of reservoir models into a normal distribution of the ensemble of reservoir models and performing normal score transformation on the acquired production data to transform an original distribution of the acquired production data into a normal distribution of the acquired production data;
generating a weighting function comprising a plurality of weights using a distance function, wherein the distance function determines a distance from each of a plurality of points in at least one grid block to a well within the reservoir;
updating the generated ensemble based on the transformed acquired production data using an ensemble Kalman filter (EnKF) and the weighting function;
transforming the updated generated ensemble back to the original distribution of the ensemble of reservoir models and transforming the acquired production data back to the original distribution of the acquired production data; and
performing a flow simulation for at least one future time period using the updated ensemble to obtain one or more predictions of future reservoir behavior comprising bottomhole pressures, fluid production rates, or formation permeability for the reservoir based on the updated ensemble.