Loading...

Closed Loop Mangement of Naturally Fractured Reservoir Using Data Assimilation Methods

Bagherinezhad, Abolfazl | 2017

554 Viewed
  1. Type of Document: Ph.D. Dissertation
  2. Language: Farsi
  3. Document No: 49598 (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. In this research, the aim is to investigate the use of data assimilation method for better reservoir model updating in the reservoir management. In addition, multi-objective optimization concept is studied for the production optimization. The application of these methods is applied for reservoir management in the naturally fractured reservoirs. To update the reservoir, ensemble based methods, especially ensemble Kalman filter and ensemble smoother, are used. To tackle the challenges encountered with these methods, modifications are proposed to obtain a better history matching and more accurate reservoir characterization. The proposed framework for history matching is implemented for the history matching of a real field (Brugge field). Furthermore, a new method named “Parameterization with Prior Statistics Correction” is proposed to improve the performance of Kalman based algorithms in the non-Gaussian prior problems. This can be used for parameter estimation for the case of high uncertainty in the prior information, in addition of non-Gaussian parameters. The proposed model updating framework is investigated for the fractured reservoir characterization considering different fracture modeling approaches. Finally, an approach for fractured network characterization is proposed. In this approach, discrete parameters of fracture (i.e. length, angle, etc.) are estimated using single porosity simulation employing upscaling step. Implementing ensemble based method shows that there are two different steps in the characterization of fracture network. In the first step called “global characterization” the main conduits of flow (major fractures) are estimated. Then, in second step named “local calibration”, the fracture parameters are tuned to match the measured data in the wells. Effects of different data on these steps are investigated carefully. It is shown that different data (i.e. production history, hard data, seismic data) can be considered as complementary for the fracture characterization. To optimize the production in the fractured reservoirs, the multi-objective genetic algorithm is used. First, a framework for optimization is proposed. It is implemented for the bi-objective problem, including life-cycle performance and short-term profit in the water-flooding process. In this optimization problem, there is redundant degree of freedom. The standard implementation of genetic algorithm cannot construct the whole Pareto front (set of optimal solutions). To have a more diverse and smooth Pareto front, some modification in the crowding comparison operator is proposed. For production optimization in the naturally fractured reservoirs, special concentrations are conducted on the physics. In this type of reservoir, the fracture and its distribution can affect the production profile considerably. So, we select the cumulative oil production and water front velocity as two objectives in the production optimization. In addition, the uncertainty in the fracture parameters is quantified using “robust optimization." By some experiments, it is shown that in this optimization, the physics of the reservoir should be considered carefully. Finally, use of uncertainty in the economic condition is investigated to select the trade-off solution among the Pareto solutions
  9. Keywords:
  10. Genetic Algorithm ; Data Assimilation ; Fractured Reservoirs ; Multiobjective Optimization ; Uncertainty ; Ensemble Kalman Filter ; Model Updating ; Reservoir Management

 Digital Object List

 Bookmark

No TOC