Time to scale the harvesting of geothermal heat for conversion to grid electricity. And not just because geothermal is clean. No, we choose geothermal because it is firm, inexhaustible, safe, and competitively harvestable almost anywhere (cheap). How best to scale geothermal? We urge the development of... geothermal arrays.
— Bob Metcalfe
Wind and solar are abundant but intermittent. Geothermal energy is different: it draws on the continuous heat of the Earth, delivering capacity factors above 90% — far exceeding the 15–45% typical of wind and solar — without fuel, combustion, or weather dependence.
Geothermal energy is also safe, produces no emissions during operation, and the subsurface heat resource is practically inexhaustible on human timescales. The remaining challenge is cost and scale.
The Internet scaled by networking small, standard, competitively sourced computers — not by building ever-larger mainframes. We apply the same logic to geothermal energy. Instead of pursuing ever-deeper, ever-more-expensive single wells, we propose geothermal arrays: networks of smaller, standardized, closed-loop deep borehole heat exchangers (DBHEs) linked by a micro-grid and operated autonomously.
Each well in the array is a self-contained unit with its own circulation pump, heat exchanger, organic Rankine cycle (ORC) converter, and micro-grid connection. Standard mechanical and electrical interfaces allow competing suppliers to manufacture subsystems and robot vans to install and maintain them autonomously — driving down cost through learning curves, competition, and shared infrastructure.
We target a levelized cost of electricity (LCOE) below 1¢/kWh.
Our group sits at the intersection of geothermal engineering and high-performance scientific computing. We develop GPU-accelerated, full three-dimensional models of deep borehole heat exchanger arrays, enabling the simulation of multi-well interactions at resolutions previously impractical.
These tools — built in the Julia programming language and deployed on modern GPU hardware — allow us to explore array configurations, thermal drawdown dynamics, drilling strategies, and economic scenarios at the speed required for engineering iteration and policy analysis.
Moreover, we study the use of generative and agentic AI to assist geothermal engineering design, using LLM-generated hypotheses together with simulation-based validation.