| Title | Bayesian Optimization of MMW Beam Ablation Under Material Uncertainties |
|---|---|
| Authors | Adriano CASABLANCA, Katerina ADAMOPOULOU, Franck MONMONT, Nikos NIKIFORAKIS |
| Year | 2025 |
| Conference | Stanford Geothermal Workshop |
| Keywords | MMW ablation, Bayesian optimization, optimization under uncertainty, computational physics |
| Abstract | Millimeter Wave (MMW) ablation is a novel technology designed to reach drilling depths of 10-20 km by direct energy irradiation, with significant applications for the development of Enhanced Geothermal Systems (EGS). Achieving these depths is critical to recover heat from high enthalpy geothermal reservoirs, which are otherwise inaccessible with conventional drilling technologies. Further development of MMW ablation requires a detailed understanding of beam-rock interactions to identify optimal beam parameters. In this study, we apply Bayesian optimization to a computational simulation of the ablation process, aiming to derive best practices for effective drilling when subject to uncertainties in material thermophysical properties. The simulation solves the multiphase conservation of energy in the solid, melt and vapor phase while modeling the irradiation from the MMW beam source. In this work, uncertainties arise from the lack of knowledge of the thermophysical properties of the rock and from the spatial randomness of the polycrystalline structure of the rock. The stochastic problem is formulated as a mean-variance optimization problem, where we seek to find the optimum incident beam variables satisfying the expected rate of penetration whilst minimizing its variance. Solutions are identified using Bayesian optimization, where sampling locations are determined by an acquisition function that combines marginalized Expected Improvement and objective variance. In general, the methodology yields optimal beam parameters as a function of stochastic material properties, providing valuable information for the future deployment of MMW drilling technology. |