A geothermal energy project led by engineers at the University of California, Irvine, has been selected by the U.S. Department of Energy Office of Science to advance to a second phase. This three-year phase will support the multi-agent AI expert for subsurface reasoning and optimization initiative, known as MAESTRO. The project aims to deploy advanced artificial intelligence tools to access subterranean heat and convert it into electricity.
Mohammad Javad Abdolhosseini Qomi, a UC Irvine professor of civil and environmental engineering and the lead principal investigator for MAESTRO, stated that while significant geothermal energy exists underground, extracting it safely and economically remains a challenge. He indicated the MAESTRO project is designed to address gaps in understanding complex underground environments through an integrated AI framework.
MAESTRO is one of several projects to receive a DOE Genesis Phase 2 award. These awards are intended to leverage AI to achieve specific goals, in this instance, sustainable energy generation that can bolster national security and economic stability. According to Qomi, receiving the Phase 2 grant confirms that his team met Phase 1 requirements by making "transformative contributions" to the field of geothermal energy.
The project involves a broad collaboration, bringing together researchers from five University of California campuses, including Irvine, Berkeley, Riverside, San Diego, and Santa Cruz. Additionally, four national laboratories—Los Alamos, Lawrence Livermore, Lawrence Berkeley, and Pacific Northwest—are participating, alongside three private-industry partners. Qomi noted that this team combines expertise in geophysics, geomechanics, geochemistry, AI and machine learning, and the engineering required for developing and operating sustainable energy resources.
Enhanced geothermal systems (EGS) aim to harness energy by circulating fluid through low-permeability rock, located as deep as 2.5 miles underground, to extract heat and power turbines. However, the optimal conditions for EGS are not universally present. Geologists and engineers must identify suitable sites where fractures can be created and controlled, and factors such as stress state, rock properties, water availability, drilling accessibility, existing cracks, and seismology characteristics are understood.
Many of these subsurface properties cannot be directly observed. MAESTRO scientists will infer them using AI tools to run complex simulations. These simulations will integrate substantial real-time geophysical observational data and prior learning. The objective is to develop models that can predict how cracks will form during fracturing operations, assisting operators in assessing the probability of EGS success—a capability not available with current tools.
Kevin Rosso, a Battelle Fellow and geochemist at Pacific Northwest National Laboratory, highlighted the project's strength in creating unique partnerships across geophysics, geochemistry, applied math, and computation. Maziar Raissi, a UC Riverside associate professor of mathematics, noted that the project will advance how AI systems learn and reason about physical problems, describing it as "AI that learns by doing: simulation in the loop."
MAESTRO’s AI will operate using a self-improving feedback loop. It will identify areas of predictive uncertainty, run targeted simulations to fill knowledge gaps, and update its models. A safety mechanism is integrated to ensure that AI recommendations are physically consistent and trustworthy.
Another significant outcome of the MAESTRO project is a greater understanding and control of induced seismicity—earthquakes that can result from hydraulic fracturing. The technologies and AI tools developed for this purpose may also contribute to early warning systems for naturally occurring earthquakes. Emily Brodsky, a UC Santa Cruz Distinguished Professor of Earth and planetary sciences, expressed excitement about addressing key hurdles to scaling up geothermal power, such as induced seismicity.
Working alongside Qomi is co-principal investigator Russ Detwiler, a UC Irvine professor of civil and environmental engineering. Detwiler stated that professionals in the sustainable energy development industry will gain access to real-time guidance, which can reduce financial risks and time costs for geothermal projects. For local communities, he noted that "continuous underground monitoring will greatly enhance safety in earthquake-prone regions."
Detwiler also confirmed that MAESTRO will be an open-source project, making its digital resources available to academic researchers, government policymakers, and the public. These tools will be packaged into a free, browser-based application, allowing anyone to explore geothermal potential, assess risk, and review data across the country.
U.S. Rep. Dave Min, whose district includes UC Irvine, acknowledged the university's leadership in this national effort. He stated that harnessing artificial intelligence to unlock geothermal resources can strengthen energy security, lower costs, and build a more resilient energy future. The investment, he added, is a testament to the researchers at UC Irvine and across the University of California system. The second phase of the MAESTRO project is expected to span three years, continuing its research and development toward these goals.
