We are delighted to announce that the work led by DGYM PhD student Ziyou Liu, ‘Efficient optimization of coupled geothermal reservoir modeling and power plant off-design based on deep learning’, is published in Advances in Geo-Energy Research and promoted by SPE Data Science & Engineering Analytics Technical Section (DSEATS) in the December Newsletter. 

Link: https://connect.spe.org/dsea/discussion/newsletter-dec-2025

Ziyou’s work was carried out under the supervision of Prof. Bicheng Yan and collaboration with Dr. Manojkumar Gudala.

Brief description of the work:

A CNN–Transformer hybrid reservoir surrogate model and a power plant surrogate model based on fully connected neural network is constructed to flexible predict the geothermal production temperature and electricity output of the power plant, respectively. Based on the heat source conditions between the geothermal reservoir and the power plant, the surrogate models of the reservoir and the power plant are coupled into one integrated forward model. Finally, a multi-objective optimization workflow considering the reservoir uncertainty is proposed to optimize the coupled reservoir–power plant system for the best thermodynamic or economic outcomes. This work significantly accelerates the optimization procedures of both geothermal reservoir and power plant.