A significant body of work has demonstrated both the promise and difficulty of quantifying uncertainty in reservoir simulation forecasts. It is generally accepted that accurate and complete quantification of uncertainty should lead to better decision making and greater profitability. Many of the techniques presented in past work attempt to quantify uncertainty without sampling the full parameter space, saving on the number of simulation runs, but inherently limiting and biasing the uncertainty quantification in the resulting forecasts. In addition, past work generally has looked at uncertainty in synthetic models and does not address the practical issues of quantifying uncertainty in an actual field. Both of these issues must be addressed in order to rigorously quantify uncertainty in practice. In this study a new approach to reservoir simulation is taken whereby the traditional one-time simulation study is replaced with a new continuous process potentially spanning the life of the reservoir. In this process, reservoir models are generated and run 24 hours a day, seven days a week, allowing many more runs than previously possible and yielding a more thorough exploration of possible reservoir descriptions. In turn, more runs enabled better estimates of uncertainty in resulting forecasts. A new technology to allow this process to run continuously with little human interaction is real-time production and pressure data, which can be automatically integrated into runs. Two tests of this continuous simulation process were conducted. The first test was conducted on the Production with Uncertainty Quantification (PUNQ) synthetic reservoir. Comparison of our results with previous studies shows that the continuous approach gives consistent and reasonable estimates of uncertainty. The second study was conducted in real time on a live field. This study demonstrates the continuous simulation process and shows that it is feasible and practical for real world applications.

This dissertation describes research designed to enhance hydrometeorological forecasts. The objective of the research is to deliver an optimal methodology to produce reliable, skillful and economically valuable probabilistic temperature and precipitation forecasts. Weather plays a dominant role for energy companies relying on forecasts of watershed precipitation and temperature to drive reservoir models, and forecasts of temperatures to meet energy demand requirements. Extraordinary precipitation events and temperature extremes involve consequential water- and power-management decisions. This research compared weighted-average, recursive, and model output statistics bias-correction methods and determined optimal window-length to calibrate temperature and precipitation forecasts. The research evaluated seven different methods for daily maximum and minimum temperature forecasts, and three different methods for daily quantitative precipitation forecasts, within a region of complex terrain in southwestern British Columbia, Canada. This research then examined ensemble prediction system design by assessing a three-model suite of multi-resolution limited area mesoscale models. The research employed two different economic models to investigate the ensemble design that produced the highest-quality, most valuable forecasts. The best post-processing methods for temperature forecasts included moving-weighted average methods and a Kalman filter method. The optimal window-length proved to be 14 days. The best post-processing methods for achieving mass balance in quantitative precipitation forecasts were a moving-average method and the best easy systematic estimator method. The optimal window-length for moving-average quantitative precipitation forecasts was 40 days. The best ensemble configuration incorporated all resolution members from all three models. A cost/loss model adapted specifically for the hydro-electric energy sector indicated that operators managing rainfall-dominated, high-head reservoirs should lower their reservoir with relatively low probabilities of forecast precipitation. A reservoir-operation model based on decision theory and variable energy pricing showed that applying an ensemble-average or full-ensemble precipitation forecast provided a much greater profit than using only a single deterministic high-resolution forecast. Finally, a bias-corrected super-ensemble prediction system was designed to produce probabilistic temperature forecasts for ten cities in western North America. The system exhibited skill and value nine days into the future when using the ensemble average, and 12 days into the future when employing the full ensemble forecast.

This dissertation describes research designed to enhance hydrometeorological forecasts. The objective of the research is to deliver an optimal methodology to produce reliable, skillful and economically valuable probabilistic temperature and precipitation forecasts. Weather plays a dominant role for energy companies relying on forecasts of watershed precipitation and temperature to drive reservoir models, and forecasts of temperatures to meet energy demand requirements. Extraordinary precipitation events and temperature extremes involve consequential water- and power-management decisions. This research compared weighted-average, recursive, and model output statistics bias-correction methods and determined optimal window-length to calibrate temperature and precipitation forecasts. The research evaluated seven different methods for daily maximum and minimum temperature forecasts, and three different methods for daily quantitative precipitation forecasts, within a region of complex terrain in southwestern British Columbia, Canada. This research then examined ensemble prediction system design by assessing a three-model suite of multi-resolution limited area mesoscale models. The research employed two different economic models to investigate the ensemble design that produced the highest-quality, most valuable forecasts. The best post-processing methods for temperature forecasts included moving-weighted average methods and a Kalman filter method. The optimal window-length proved to be 14 days. The best post-processing methods for achieving mass balance in quantitative precipitation forecasts were a moving-average method and the best easy systematic estimator method. The optimal window-length for moving-average quantitative precipitation forecasts was 40 days. The best ensemble configuration incorporated all resolution members from all three models. A cost/loss model adapted specifically for the hydro-electric energy sector indicated that operators managing rainfall-dominated, high-head reservoirs should lower their reservoir with relatively low probabilities of forecast precipitation. A reservoir-operation model based on decision theory and variable energy pricing showed that applying an ensemble-average or full-ensemble precipitation forecast provided a much greater profit than using only a single deterministic high-resolution forecast. Finally, a bias-corrected super-ensemble prediction system was designed to produce probabilistic temperature forecasts for ten cities in western North America. The system exhibited skill and value nine days into the future when using the ensemble average, and 12 days into the future when employing the full ensemble forecast.

This dissertation describes research designed to enhance hydrometeorological forecasts. The objective of the research is to deliver an optimal methodology to produce reliable, skillful and economically valuable probabilistic temperature and precipitation forecasts. Weather plays a dominant role for energy companies relying on forecasts of watershed precipitation and temperature to drive reservoir models, and forecasts of temperatures to meet energy demand requirements. Extraordinary precipitation events and temperature extremes involve consequential water- and power-management decisions. This research compared weighted-average, recursive, and model output statistics bias-correction methods and determined optimal window-length to calibrate temperature and precipitation forecasts. The research evaluated seven different methods for daily maximum and minimum temperature forecasts, and three different methods for daily quantitative precipitation forecasts, within a region of complex terrain in southwestern British Columbia, Canada. This research then examined ensemble prediction system design by assessing a three-model suite of multi-resolution limited area mesoscale models. The research employed two different economic models to investigate the ensemble design that produced the highest-quality, most valuable forecasts. The best post-processing methods for temperature forecasts included moving-weighted average methods and a Kalman filter method. The optimal window-length proved to be 14 days. The best post-processing methods for achieving mass balance in quantitative precipitation forecasts were a moving-average method and the best easy systematic estimator method. The optimal window-length for moving-average quantitative precipitation forecasts was 40 days. The best ensemble configuration incorporated all resolution members from all three models. A cost/loss model adapted specifically for the hydro-electric energy sector indicated that operators managing rainfall-dominated, high-head reservoirs should lower their reservoir with relatively low probabilities of forecast precipitation. A reservoir-operation model based on decision theory and variable energy pricing showed that applying an ensemble-average or full-ensemble precipitation forecast provided a much greater profit than using only a single deterministic high-resolution forecast. Finally, a bias-corrected super-ensemble prediction system was designed to produce probabilistic temperature forecasts for ten cities in western North America. The system exhibited skill and value nine days into the future when using the ensemble average, and 12 days into the future when employing the full ensemble forecast. / Science, Faculty of / Earth, Ocean and Atmospheric Sciences, Department of / Graduate

The Capacitance Resistance Model (CRM) is a reservoir model based on a data-driven approach. It stems from the continuity equation and takes advantage of the usually abundant rate data to achieve a synergy of analytical model and data-driven approach. Minimal information (rates and bottom-hole pressure) is required to inexpensively characterize the reservoir. Important information, such as inter-well connectivity, reservoir compressibility effects, etc., can be easily and readily evaluated. The model also suggests optimal injection schemes in an effort to maximize ultimate oil recovery, and hence can assist real time reservoir analysis to make more informed management decisions. Nevertheless, an important limitation in the current CRM model is that it only treats the reservoir flow as single-phase flow, which does not favor capturing physics when the saturation change is large, such as for an immature water flood. To overcome this limitation, we develop a two-phase flow coupled CRM model that couples the pressure equation (fluid continuity equation) and the saturation equation (oil mass balance). Through this coupling, the model parameters such as the connectivity, the time constant, temporal oil saturation, etc., are estimated using nonlinear multivariate regression to history match historical production data. Incorporating the physics of two-phase displacement brings several advantages and benefits to the CRM model, such as the estimation of total mobility change, more accurate prediction of oil production, broader model application range, and better adaptability to complicated field scenarios. Also, the estimated saturation within the drainage volume of each producer can provide insights with respect to the field remaining oil saturation distribution. Synthetic field case studies are carried out to demonstrate the different capabilities of the coupled CRM model in homogeneous and heterogeneous reservoirs with different geological features. The physical meanings of model parameters are well explained and validated through case studies. The results validate the coupled CRM model and show improved accuracy in model parameters obtained through the history match. The prediction of oil production is also significantly improved compared to the current CRM model. A more reliable oil rate prediction enables further optimization to adjust injection strategies. The coupled CRM model has been shown to be fast and stable. Moreover, sensitivity analyses are conducted to study and understand the impact of the input information (e.g., relative permeability, viscosity) upon the output model parameters (e.g., connectivity, time constants). This analysis also proves that the model parameters from the two-phase coupled model can combine both reservoir compressibility and mobility effects. / text

The energy retrieval process in an Enhanced Geothermal System (EGS) depends on fracture networks to facilitate fluid movement, thereby enabling the extraction of heat from adjacent rocks matrix. Nonetheless, due to the inherent heterogeneity and intricate multi-physics characteristics of these systems, high-fidelity physics-based forward simulations ($f_h$) can be computationally demanding. This presents a considerable obstacle to the efficient management of these reservoirs. Therefore, creating an effective and robust optimization framework is essential, with the primary aim being to maximize the thermal extraction from Enhanced Geothermal Systems (EGS). A deep learning-assisted reservoir management framework incorporating a low-fidelity forward surrogate model ($f_l$) alongside gradient-based optimizers is developed to expedite reservoir management. A thermo-hydro-mechanical (THM) model for EGS is established by utilizing finite element-based reservoir simulation techniques. By parameterizing the fracture aperture and well controls, we carried out the THM simulation to produce 2500 datasets. Subsequently, we employed these datasets to train two distinct deep neural network (DNN) architectures to predict the variations in pressure and temperature distributions. Ultimately, these predictions from the forward model are used in calculating the total net energy. Instead of executing the optimization workflow with a large number of simulations from $f_h$, we directly optimize the well control parameters relative to the geological parameters using $f_l$. Since $f_l$ is efficient and fully differentiable, it could be combined with various gradient-based or gradient-free optimization algorithms to maximize the total net energy by determining the optimal decision parameters. Drawing from the simulation datasets, we analysed the effect of fracture aperture variation on temperature and pressure evolution. Our investigation revealed that the spatial distribution of the fracture aperture is a predominant factor in controlling the propagation of the thermal front. Variations of the fracture aperture exhibit a strong correlation with temperature fluctuations within the fracture, primarily due to thermal stress changes. When compared with a comprehensive physics simulator, our DNN-based forward surrogate model offers a significant computational acceleration, approximately 1500 times faster, without compromising predictive accuracy, achieving an $R^2$ value of 99%. The forward model $f_l$, when combined with gradient-based optimizers, enables optimization to proceed 10 to 68 times faster than when using derivative-free global optimizers. The proposed reservoir management framework exhibits both efficiency and scalability, facilitating the real-time execution of each optimization process.

No description available.

In this thesis, modeling and optimization in the field of storage management under stochastic condition will be investigated using two different methodologies: Simulation Optimization Techniques (SOT), which are usually categorized in the area of Reinforcement Learning (RL), and Nonlinear Modeling Techniques (NMT). For the first set of methods, simulation plays a fundamental role in evaluating the control policy: learning techniques are used to deliver sub-optimal policies at the end of a learning process. These iterative methods use the interaction of agents with the stochastic environment through taking actions and observing different states. To converge to the steady-state condition where policies and value functions do not change significantly with the continuation of the learning process, all or most important states must be visited sufficiently. This might be prohibitively time-consuming for large-scale problems. To make these techniques more efficient both in terms of computation time and robust optimal policies, the idea of Opposition-Based Learning (OBL-Type I and Type II) is employed to modify/extend popular RL techniques including Q-Learning, Q(λ), sarsa, and sarsa(λ). Several new algorithms are developed using this idea. It is also illustrated that, function approximation techniques such as neural networks can contribute to the process of learning. The state-of-the-art implementations usually consider the maximization of expected value of accumulated reward. Extending these techniques to consider risk and solving some well-known control problems are important contributions of this thesis. Furthermore, the new nonlinear modeling for reservoir management using indicator functions and randomized policy introduced by Fletcher and Ponnambalam, is extended to stochastic releases in multi-reservoir systems. In this extension, two different approaches for defining the release policies are proposed. In addition, the main restriction of considering the normal distribution for inflow is relaxed by using a beta-equivalent general distribution. A five-reservoir case study from India is used to demonstrate the benefits of these new developments. Using a warehouse management problem as an example, application of the proposed method to other storage management problems is outlined.

In this thesis, modeling and optimization in the field of storage management under stochastic condition will be investigated using two different methodologies: Simulation Optimization Techniques (SOT), which are usually categorized in the area of Reinforcement Learning (RL), and Nonlinear Modeling Techniques (NMT). For the first set of methods, simulation plays a fundamental role in evaluating the control policy: learning techniques are used to deliver sub-optimal policies at the end of a learning process. These iterative methods use the interaction of agents with the stochastic environment through taking actions and observing different states. To converge to the steady-state condition where policies and value functions do not change significantly with the continuation of the learning process, all or most important states must be visited sufficiently. This might be prohibitively time-consuming for large-scale problems. To make these techniques more efficient both in terms of computation time and robust optimal policies, the idea of Opposition-Based Learning (OBL-Type I and Type II) is employed to modify/extend popular RL techniques including Q-Learning, Q(λ), sarsa, and sarsa(λ). Several new algorithms are developed using this idea. It is also illustrated that, function approximation techniques such as neural networks can contribute to the process of learning. The state-of-the-art implementations usually consider the maximization of expected value of accumulated reward. Extending these techniques to consider risk and solving some well-known control problems are important contributions of this thesis. Furthermore, the new nonlinear modeling for reservoir management using indicator functions and randomized policy introduced by Fletcher and Ponnambalam, is extended to stochastic releases in multi-reservoir systems. In this extension, two different approaches for defining the release policies are proposed. In addition, the main restriction of considering the normal distribution for inflow is relaxed by using a beta-equivalent general distribution. A five-reservoir case study from India is used to demonstrate the benefits of these new developments. Using a warehouse management problem as an example, application of the proposed method to other storage management problems is outlined.

Reservoir management typically focuses on maximizing oil and gas recovery from a reservoir based on facts and information while minimizing capital and operating investments. Modern reservoir management uses history-matched simulation model to predict the range of recovery or to provide the economic assessment of different field development strategies. Geological models are becoming increasingly complex and more detailed with several hundred thousand to million cells, which include large sets of subsurface uncertainties. Current issues associated with history matching, therefore, involve extensive computation (flow simulations) time, preserving geologic realism, and non-uniqueness problem. Many of recent rate optimization methods utilize constrained optimization techniques, often making them inaccessible for field reservoir management. Field-scale rate optimization problems involve highly complex reservoir models, production and facilities constraints and a large number of unknowns. We present a hierarchical multiscale calibration approach using global and local updates in coarse and fine grid. We incorporate a multiscale framework into hierarchical updates: global and local updates. In global update we calibrate large-scale parameters to match global field-level energy (pressure), which is followed by local update where we match well-by-well performances by calibration of local cell properties. The inclusion of multiscale calibration, integrating production data in coarse grid and successively finer grids sequentially, is critical for history matching high-resolution geologic models through significant reduction in simulation time. For rate optimization, we develop a hierarchical analytical method using streamline-assisted flood efficiency maps. The proposed approach avoids use of complex optimization tools; rather we emphasize the visual and the intuitive appeal of streamline method and utilize analytic solutions derived from relationship between streamline time of flight and flow rates. The proposed approach is analytic, easy to implement and well-suited for large-scale field applications. Finally, we present a hierarchical Pareto-based approach to history matching under conflicting information. In this work we focus on multiobjective optimization problem, particularly conflicting multiple objectives during history matching of reservoir performances. We incorporate Pareto-based multiobjective evolutionary algorithm and Grid Connectivity-based Transformation (GCT) to account for history matching with conflicting information. The power and effectiveness of our approaches have been demonstrated using both synthetic and real field cases.