• Energy Research

To maximize the economic benefits of geothermal energy production, it is essential to optimize geothermal reservoir management strategies, in which geologic uncertainty should be considered. In this work, we propose a closed-loop optimization framework, based on deep learning surrogates, for the well control optimization of geothermal reservoirs. In this framework, we construct a hybrid convolution–recurrent neural network surrogate, which combines the convolution neural network (CNN) and long short-term memory (LSTM) recurrent network. The convolution structure can extract spatial information of reservoir property fields and the recurrent structure can approximate sequence-to-sequence mapping. The trained model can predict time-varying production responses (rate, temperature, etc.) for cases with different permeability fields and well control sequences. In this closed-loop optimization framework, production optimization, based on the differential evolution (DE) algorithm, and data assimilation, based on the iterative ensemble smoother (IES), are performed alternately to achieve a real-time well control optimization and to estimate reservoir properties (e.g. permeability) as the production proceeds. In addition, the averaged objective function over the ensemble of geologic parameter estimates is adopted to consider geologic uncertainty in the optimization process. Geothermal reservoir production cases are examined to evaluate the performance of the proposed closed-loop optimization framework. Our results show that the proposed framework can achieve efficient and effective real-time optimization and data assimilation in the geothermal reservoir production process. Renewable Energy, 211 ISSN:0960-1481 ISSN:1879-0682

To maximize the economic benefits of geothermal energy production, it is essential to optimize geothermal reservoir management strategies, in which geologic uncertainty should be considered. In this work, we propose a closed-loop optimization framework, based on deep learning surrogates, for the well control optimization of geothermal reservoirs. In this framework, we construct a hybrid convolution–recurrent neural network surrogate, which combines the convolution neural network (CNN) and long short-term memory (LSTM) recurrent network. The convolution structure can extract spatial information of reservoir property fields and the recurrent structure can approximate sequence-to-sequence mapping. The trained model can predict time-varying production responses (rate, temperature, etc.) for cases with different permeability fields and well control sequences. In this closed-loop optimization framework, production optimization, based on the differential evolution (DE) algorithm, and data assimilation, based on the iterative ensemble smoother (IES), are performed alternately to achieve a real-time well control optimization and to estimate reservoir properties (e.g. permeability) as the production proceeds. In addition, the averaged objective function over the ensemble of geologic parameter estimates is adopted to consider geologic uncertainty in the optimization process. Geothermal reservoir production cases are examined to evaluate the performance of the proposed closed-loop optimization framework. Our results show that the proposed framework can achieve efficient and effective real-time optimization and data assimilation in the geothermal reservoir production process. Renewable Energy, 211 ISSN:0960-1481 ISSN:1879-0682

Abstract Accurate and efficient simulation of fluid and heat flow in fractures has long been a topic of interest for fractured reservoirs, e.g., enhanced geothermal systems (EGS) and unconventional oil/gas formations. In this paper, we propose a flexible and effective modeling approach, the extended embedded discrete fracture model (XEDFM), to simulate fluid and heat flow in fractured reservoirs with 3-D non-planar fracture networks. Compared with the conventional embedded discrete fracture model (EDFM), the XEDFM possesses two major merits: (1) separation of fracture discretization and matrix gridding provides maximum flexibility in handling fractures with complex geometry/topology, regardless of the resolution of the matrix grid; and (2) the combination of connection-list strategy and the concept of non-neighboring connection facilitates the construction of fluid-heat flux between the fracture and the matrix/fracture. With systematically validated XEDFM, the impacts of fracture roughness and heat extraction strategy on hydrothermal behaviors and heat mining efficiency are investigated. Another example introduces a workflow for design and modeling of 3-D non-planar fracture networks, with which the performance of horizontal and vertical wells in tapping heat energy from EGS are explored. This work presents a flexible and effective approach for modeling fluid/heat transfer accurately in 3-D non-planar fractures, and provides a set of framework and efficient algorithms for non-planar fractures design, discretization and simulation, establishing the foundation to build and simulate models with complicated fractures.

Abstract Accurate and efficient simulation of fluid and heat flow in fractures has long been a topic of interest for fractured reservoirs, e.g., enhanced geothermal systems (EGS) and unconventional oil/gas formations. In this paper, we propose a flexible and effective modeling approach, the extended embedded discrete fracture model (XEDFM), to simulate fluid and heat flow in fractured reservoirs with 3-D non-planar fracture networks. Compared with the conventional embedded discrete fracture model (EDFM), the XEDFM possesses two major merits: (1) separation of fracture discretization and matrix gridding provides maximum flexibility in handling fractures with complex geometry/topology, regardless of the resolution of the matrix grid; and (2) the combination of connection-list strategy and the concept of non-neighboring connection facilitates the construction of fluid-heat flux between the fracture and the matrix/fracture. With systematically validated XEDFM, the impacts of fracture roughness and heat extraction strategy on hydrothermal behaviors and heat mining efficiency are investigated. Another example introduces a workflow for design and modeling of 3-D non-planar fracture networks, with which the performance of horizontal and vertical wells in tapping heat energy from EGS are explored. This work presents a flexible and effective approach for modeling fluid/heat transfer accurately in 3-D non-planar fractures, and provides a set of framework and efficient algorithms for non-planar fractures design, discretization and simulation, establishing the foundation to build and simulate models with complicated fractures.

Summary Enhanced coalbed-methane (ECBM) recovery by the injection of CO2 and/or N2 is an attractive method for recovering additional natural gas resources, while at the same time sequestering CO2 in the subsurface. For the naturally fractured coalbed-methane (CBM) reservoirs, the coupled fluid-flow and geomechanics effects involving both the effective-stress effect and the matrix shrinkage/swelling, are crucial to simulate the permeability change and; thus gas migration during primary or enhanced CBM recovery. In this work, a fully coupled multiphase multicomponent flow and geomechanics model is developed. The coupling effects are modeled by introducing a set of elaborate geomechanical equations, which can provide more fundamental understanding about the solid deformation and give a more accurate permeability/porosity prediction over the existing analytical models. In addition, the fluid-flow model in our study is fully compositional; considering both multicomponent gas dissolution and water volatility. To obtain accurate gas solubility in the aqueous phase, the Peng-Robinson equation of state (EOS) is modified according to the suggestions of Søreide and Whitson (1992). An extended Langmuir isotherm is used to describe the adsorption/desorption behavior of the multicomponent gas to/from the coal surface. With a fully implicit finite-difference method, we develop: a 3D, multiphase, multicomponent, dual-porosity CBM/ECBM research code that is fully compositional and has fully coupled fluid flow and geomechanics. It has been partially validated and verified by comparison against other simulators such as GEM, Eclipse, and Coalgas. We then perform a series of simulations/investigations with our research code. First, history matching of Alberta flue-gas-injection micropilot data is performed to test the permeability model. The commonly used uniaxial-strain and constant-overburden-stress assumptions for analytical permeability models are then assessed. Finally, the coupling effects of fluid flow and geomechanics are investigated, and the impact of different mixed CO2/N2 injection scenarios is explored for both methane (CH4) production and CO2 sequestration.