Application of ensemble Kalman filter in nonlinear reservoir problem

Key words:  oil reservoir numerical simulation  ensemble Kalman filter  automatic history matching  initial oil water interface  parameter estimation  nonlinear reservoir

Abstract:

The standard ensemble Kalman filter method is improved to history matching and inverse nonlinear reservoir problem. Reservoir model parameters can be estimated by automatic history matching with ensemble Kalman filter. Only reservoir model parameters are updated in this new method. The updated models are used to rerun the simulator from time zero to do the next prediction. The updated model is stochastically consistent with the updated dynamical variables by doing this. The improved method is used to estimate the depth of the initial oil water interface as well as gridblock rock property fields by matching production data. The results are compared with those obtained from standard ensemble Kalman filter. The estimation and prediction results are improved by using the new ensemble Kalman filter.