The Earth’s subsurface holds a vast trove of renewable energy in the form of heat stored within rocks. If harnessed correctly, this geothermal energy could potentially power humanity for centuries. To extract this energy, engineers use a geothermal closed-loop system that requires advanced simulations and high-performance computing.
In the summer of 2022, Boise State Ph.D. in Computing student Brian Kyanjo worked on modeling geothermal systems at Idaho National Laboratory (INL). His team developed a thermal-hydraulic code in Python to simulate underground fluid temperatures and integrated this with a Multiphysics Object-Oriented Simulation Environment (MOOSE) based application to model heat flow through the rocks. To run these simulations, Kyanjo utilized INL’s powerful Falcon, Sawtooth, and Lemhi supercomputing resources.
Results showed that the configuration of the piping system, known as “spider legs,” strongly impacts performance. The fluid properties are also critical. Optimizing geothermal energy relies on factoring in these design considerations. However, running simulations over longer periods requires ample computational muscle. INL’s supercomputers enabled the team to model geothermal systems with unprecedented fidelity — shedding light on how we can best tap into the energy stored beneath our feet.