Beyond Simulate-Then-Optimize: Geothermal AI for Geothermal Dynamics Prediction, Design, and Discovery

Geothermal modeling commonly follows a simulate-then-optimize workflow in which a calibrated deterministic simulator predicts future system behavior and supports operational decisions. This paper argues that next-generation enhanced geothermal systems require a broader framework because subsurface conditions are partially observed, heterogeneous, uncertain, and affected by interventions. It reframes geothermal computation as a coupled cycle of inference, intervention, and discovery, outlining how AI can support uncertainty-aware prediction, adaptive decision-making, model calibration, and the discovery of site-specific physical mechanisms. The resulting research agenda positions AI not only as an accelerator for existing simulation modules but also as a foundation for adaptive geothermal knowledge systems.

📄 In ACM SIGKDD Explorations Newsletter, Volume 28, Issue 1, pp. 102–114.
🔗 Paper Link (ACM Digital Library)