Artificial intelligence empowering geothermal energy development: A full-lifecycle review from exploration to operation

Abstract

Geothermal energy (GE), driven by heat from the Earth's interior, is a renewable energy source characterized by abundant reserves, wide geographic distribution, and stable supply. It can continuously provide baseload electricity and supports both power generation and direct heating. Compared with intermittent renewables such as wind and solar, GE's inherent stability gives it strategic value in grid regulation and energy security. However, the global development of GE remains in its early stages, accounting for less than 0.5 % of total renewable energy capacity. Key limitations include high uncertainty in resource exploration, costly and technically demanding drilling, strong reservoir heterogeneity, long project cycles, and restricted economic feasibility.

Artificial intelligence (AI)—particularly machine learning (ML) and deep learning (DL)—offers new opportunities to overcome these challenges. AI excels at processing large datasets, identifying complex patterns, and forecasting system behavior, making it applicable across the GE lifecycle—from resource assessment and drilling optimization to reservoir modeling, production forecasting, and plant operation. In particular, ML/DL shows significant potential in enhanced geothermal systems (EGS), where reservoir complexity and sparse data pose major barriers. In recent years, rapid advancements in algorithms and insights from other industries have accelerated the adoption of ML/DL in GE development. Nevertheless, issues such as limited model generalizability, insufficient data quality, and lack of interpretability remain persistent obstacles.

This review systematically examines representative applications of AI across the GE lifecycle, identifies key breakthroughs and current bottlenecks, and outlines future research directions. It aims to provide both theoretical guidance and practical insights for intelligent geothermal development.