Re-Sampling the Ensemble Kalman Filter for Improved History Matching and Characterizations of Non-Gaussian and Non-Linear Reservoir Models

  • Author / Creator
    nejadi, siavash
  • Reservoir simulation models play an important role in the production forecasting and field development planning. To enhance their predictive capabilities and capture the uncertainties in model parameters, stochastic reservoir models should be calibrated to both geologic and flow observations. The relationship between production performance and model parameters is vastly non-linear, rendering history matching process a challenging task. The Ensemble Kalman Filter (EnKF) is a Monte-Carlo based technique for assisted history matching and real-time updating of reservoir models. EnKF works efficiently with Gaussian variables, but it often fails to honor the reference probability distribution of the model parameters where the distribution of model parameters are non-Gaussian and the system dynamics are strongly nonlinear. In this thesis, novel sampling procedures are proposed to honor geologic information in reservoirs with non-Gaussian model parameters. The methodologies include generating multiple geological models and updating the uncertain parameters using dynamic flow responses using iterative EnKF technique. Two new re-sampling steps are presented for characterization of multiple facies reservoirs. After certain number of assimilation steps, the updated ensemble is used to generate a new ensemble that is conditional to both the geological information and the early production data. Probability field simulation and a novel probability weighted re-sampling scheme are introduce to re-sample a new ensemble. After the re-sampling step, iterative EnKF is again applied on the ensemble members to assimilate the remaining production history. A new automated dynamic data integration workflow is implemented for characterization and uncertainty assessment of fracture reservoir models. This new methodology includes generating multiple discrete fracture network (DFN) models, upscaling the models for flow simulation, and updating the DFN model parameters using dynamic flow responses. The assisted history matching algorithm entails combining a probability weighted sampling with iterative EnKF. The performances of the introduced methodologies are evaluated by performing various simulation studies for different synthetic and field case studies. The qualities of the final matching results are assessed by examining the geological realism of the updated ensemble using the reference probability distribution of the model parameters and computing the predicted dynamic data mismatch.

  • Subjects / Keywords
  • Graduation date
    Spring 2015
  • Type of Item
  • Degree
    Doctor of Philosophy
  • DOI
  • License
    This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.