Han, Jichao (2016-05). Multiobjective and Level Set Methods for Reservoir Characterization and Optimization. Doctoral Dissertation. Thesis uri icon

abstract

  • Proper management of oil and gas reservoirs as dynamic systems reduces operational expenditures, alleviates uncertainty, and increases hydrocarbon recovery. In this dissertation, we focus on two issues in reservoir management: multiobjective integration and channelized reservoir calibration. Multiple objectives, including bottom-hole pressure (BHP), water cut, and 4-D seismic data, are utilized in model ranking, history matching, and production optimization. These objectives may conflict, as they represent characteristics coming from different measurements and sources, and, significantly, of varying scales. A traditional weighted-sum method may reduce the solution space, often leading to loss of key information for each objective. Thus, how to integrate multiple objectives effectively becomes critical in reservoir management. This dissertation presents a Pareto-based approach to characterize multiobjective and potentially conflicting features and to capture geologic uncertainty, preserving the original objective space and avoiding weights determination as in the weight-sum method. For channelized reservoirs, identification of the channel geometry and facies boundaries, as well as characterization of channel petrophysical properties are critical for performance predictions. Traditional history matching methods, however, are unable to preserve the channel geometry. We propose a level set based method, integrated with seismic constraint and coupled with the Grid Connectivity Transform (GCT) for channelized reservoirs calibration. We first develop the Pareto-based model ranking (PBMR) to rank multiple realizations, taking into consideration seismic and production data. We demonstrate that this approach can be applied to select multiple competitive realizations compared with the weighted-sum method, and uncertainty range of each objective can be effectively addressed. Next, we extend the Pareto-based framework to full-field history matching and production optimization of the Norne Field in the North Sea. A hierarchical history matching workflow including global and local updates helps to capture the large- and fine-scale heterogeneity. A two-step polymer flood optimization consisting of the streamline-based rate optimization and the Pareto-based polymer optimization is shown to be beneficial for reducing the impact of heterogeneity and increasing production improvement as well as NPV. Finally, we propose a two-step history matching workflow for facies and property calibration of the channelized reservoirs, where the channel geometry is modeled using the level set method, and smaller scale heterogeneity is modeled using the GCT. Moreover, the seismic constraints incorporated into the level set improves facies model calibration.
  • Proper management of oil and gas reservoirs as dynamic systems reduces operational expenditures, alleviates uncertainty, and increases hydrocarbon recovery. In this dissertation, we focus on two issues in reservoir management: multiobjective integration and channelized reservoir calibration. Multiple objectives, including bottom-hole pressure (BHP), water cut, and 4-D seismic data, are utilized in model ranking, history matching, and production optimization. These objectives may conflict, as they represent characteristics coming from different measurements and sources, and, significantly, of varying scales. A traditional weighted-sum method may reduce the solution space, often leading to loss of key information for each objective. Thus, how to integrate multiple objectives effectively becomes critical in reservoir management. This dissertation presents a Pareto-based approach to characterize multiobjective and potentially conflicting features and to capture geologic uncertainty, preserving the original objective space and avoiding weights determination as in the weight-sum method. For channelized reservoirs, identification of the channel geometry and facies boundaries, as well as characterization of channel petrophysical properties are critical for performance predictions. Traditional history matching methods, however, are unable to preserve the channel geometry. We propose a level set based method, integrated with seismic constraint and coupled with the Grid Connectivity Transform (GCT) for channelized reservoirs calibration.

    We first develop the Pareto-based model ranking (PBMR) to rank multiple realizations, taking into consideration seismic and production data. We demonstrate that this approach can be applied to select multiple competitive realizations compared with the weighted-sum method, and uncertainty range of each objective can be effectively addressed.

    Next, we extend the Pareto-based framework to full-field history matching and production optimization of the Norne Field in the North Sea. A hierarchical history matching workflow including global and local updates helps to capture the large- and fine-scale heterogeneity. A two-step polymer flood optimization consisting of the streamline-based rate optimization and the Pareto-based polymer optimization is shown to be beneficial for reducing the impact of heterogeneity and increasing production improvement as well as NPV.

    Finally, we propose a two-step history matching workflow for facies and property calibration of the channelized reservoirs, where the channel geometry is modeled using the level set method, and smaller scale heterogeneity is modeled using the GCT. Moreover, the seismic constraints incorporated into the level set improves facies model calibration.

publication date

  • May 2016