Loading...

Intelligent and Sequential Reservoir Model Updating and Uncertainty Assessment during EOR Process

Jahanbakhshi, Saman | 2017

937 Viewed
  1. Type of Document: Ph.D. Dissertation
  2. Language: Farsi
  3. Document No: 50322 (06)
  4. University: Sharif University of Technology
  5. Department: Chemical and Petroleum Engineering
  6. Advisor(s): Pishvaie, Mahmoud Reza; Bozorgmehry Boozarjomehry, Ramin
  7. Abstract:
  8. Hydrocarbon reservoir management and development as well as planning of enhanced oil recovery (EOR) processes are based on the reservoir dynamic model. Thus, successful implementation of EOR scenarios greatly depends on the quality of the dynamic model and accuracy of the associated parameters in order to correctly describe fluid flow through porous media. First, a dynamic model is constructed based on the prior knowledge. However, because of the various types of error during model building, the prior model is not so accurate and perfect. Accordingly, new observation data, such as production and 4D seismic data, are utilized to calibrate the prior model and characterize the reservoir under a Bayesian framework, which is known as history matching in the petroleum literature. Moreover, optimum reservoir management and risk analysis require quantification of uncertainty in the reservoir characterization as well as in the reservoir future performance prediction. Among all of the available sequential methods for reservoir model updating, ensemble-based Kalman filters have exhibited promising results and also benefit from several technical advantages, such as computational efficiency, ease of implementation and assessment of uncertainty along with history matching.In this dissertation, ensemble-based Kalman filters are first applied for characterization of PUNQ real-field model. Then, joint estimation of relative permeability and capillary pressure curves as well as joint estimation of absolute and relative permeabilities are considered. In either cases, accurate estimation of the model parameters is obtained using measurement data and prior knowledge. Furthermore, a hybrid assimilation scheme is proposed to better estimate model parameters for the cases in which there exist both local and non-local unknown parameters. In the hybrid approach, estimation process of the local and non-local parameters is separated. The accuracy of the estimated model parameters as well as quality of history-match are greatly improved in the hybrid method. Thereafter, ensemble-based Kalman filters are employed to characterize channelized reservoirs. However, because of inherent assumption of Gaussianity, these methods are not directly applicable to channelized reservoirs wherein the distribution of the petrophysical properties is non-Gaussian and multimodal. Therefore, ensemble-based Kalman filters are combined with level set parametrization in order to reshape the multimodal histogram into a Gaussian one. Thus, categorical variables (facies type in each block) together with petrophysical properties of each facies are simultaneously estimated. In addition, impact of initial ensembles on posterior distribution of ensemble-based assimilation methods is analyzed. Also, sampling performance as well as uncertainty quantification of these methods are compared in terms of their ability to accurately and consistently evaluate unknown reservoir model parameters and reservoir future performance. Finally, polynomial chaos expansion (PCE) is used to investigate evolution and propagation of uncertainty from uncertain model parameters to the model outputs. Along with, the constructed PCE is used to analyze the sensitivity of the model outputs to the uncertain model parameters
  9. Keywords:
  10. Automatic History Matching ; Data Assimilation ; Channelized Reservoirs ; Ensemble Kalman Filter ; Enhanced Oil Recovery ; Sequential Reservoir Model Updating ; Uncertainty Assessment

 Digital Object List

 Bookmark

No TOC