Efficient Optimization of Energy Recovery From Geothermal Reservoirs With Recurrent Neural Network Predictive Models. Issue 3 (27th February 2023)

• Record Type: Journal Article
• Authors: Qin, Zhen; Jiang, Anyue; Faulder, Dave; Cladouhos, Trenton T.; Jafarpour, Behnam
• Is Part Of: Water resources research. Volume 59:Issue 3(2023)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2023-02-27
• Subjects: geothermal reservoirs -- recurrent neural network -- deep learning -- machine learning -- optimization -- energy recovery; Hydrology -- Periodicals; 333.91
• DOI: 10.1029/2022WR032653
• Languages: English
• ISSNs: 0043-1397

Abstract

Abstract: Improving the long‐term energy production performance of geothermal reservoirs can be accomplished by optimizing field development and management plans. Reliable prediction models, however, are needed to evaluate and optimize the performance of the underlying reservoirs under various operation and development strategies. In traditional frameworks, physics‐based simulation models are used to predict the energy production performance of geothermal reservoirs. However, detailed simulation models are not trivial to construct, require a reliable description of the reservoir conditions and properties, and entail high computational complexity. Data‐driven predictive models can offer an efficient alternative for use in optimization workflows. This paper presents an optimization framework for net power generation in geothermal reservoirs using a variant of the recurrent neural network (RNN) as a data‐driven predictive model. The RNN architecture is developed and trained to replace the simulation model for computationally efficient prediction of the objective function and its gradients with respect to the well control variables. The net power generation performance of the field is optimized by automatically adjusting the mass flow rate of production and injection wells over 12 years, using a gradient‐based local search algorithm. Two field‐scale examples are presented to investigate the performance of the developed data‐driven prediction and optimization framework. The prediction and optimization results from the RNN model are evaluated through comparison with the results obtained by using a numerical simulation model of a real geothermal reservoir. Plain Language Summary: The paper proposes the use of recurrent neural network (RNN) architectures for capturing the dynamics of historical well response data as a function of input control variables. A trained RNN is then used as an efficient input‐output dynamical model for optimization of energy recovery from geothermal reservoirs. Results from time‐consuming simulation‐based and fast RNN prediction models are presented and evaluated to compare the optimization strategies of the two approaches, indicating their consistency. The results suggest that RNN can be used as an efficient dynamic prediction tool for decision support and management of geothermal reservoir operations and development. Key Points: An efficient workflow is developed for long‐term optimization of energy recovery from geothermal reservoirs using deep learning‐based proxy models (a specific design of the recurrent neural network (RNN) is used in this paper) Using a limited set simulated data for training the RNN model improves its extrapolation power and leads to reliable long‐term predictions During training RNN captures the dynamics in well response data and uses it to predict long‐term net energy production under different well control settings (e.g., mass flow rates)