A systematic review of data analytics applications in above-ground geothermal energy operations

The advent of reliable and inexpensive sensors and advancements in general computing have made data-heavy algorithms feasible for operational, real-time decision-making applications in the geothermal energy industry. This systematic review aims to provide a starting point for researchers interested in developing data-driven systems, tools, and frameworks to enhance the performance and reliability of above-ground geothermal energy operations. The approach and results of the review are presented to answer the following research questions: how has data analytics been applied in above-ground geothermal operations, what data sets have been used in such studies, which types of machine learning or artificial intelligence algorithms have been used in geothermal studies, and at which stages of geothermal development have studies been applied. Published research articles were retrieved from four literature databases: the International Geothermal Association (IGA) online library, ScienceDirect, SpringerLink, and IEEE Xplore. A total of 830 publications were retrieved using the same search query across the selected databases, from which 63 research papers were selected based on a set of inclusion and exclusion criteria. A full-text evaluation of the selected research papers revealed that machine learning has been used in geothermal for design optimisation, performance monitoring, performance optimisation, fault detection, and other applications. Most of the trained models (95 %) were of the artificial neural network family with other model types generally used as performance benchmarks. The systematic review revealed significant potential for further research and applications in the areas of feature selection, systematic time-series feature engineering, and model evaluation.