Skip to main content
U.S. Department of Energy
Office of Scientific and Technical Information

GOOML - Real World Applications of Machine Learning in Geothermal Operations

Conference ·
OSTI ID:1983899
GOOML (Geothermal Operational Optimization with Machine Learning) is a machine-learning based framework that enables geothermal power plant operators to explore optimization opportunities for their assets in an efficient and robust digital environment. Backed by real-world data sources, thermodynamic constraints and steamfield intelligence, the GOOML environment provides new tools to explore how to best operate steamfields as well as test new scenarios and configurations prior to implementation in the field. To prove the effectiveness of GOOML, we have undertaken optimization experiments using reinforcement learning (RL) to generate operational suggestions using a balance of mass-take targets, sustainability considerations and net generation. Our experiments use the GOOML construct to explore different field parameters and perform multiple reinforcement learning experiments. Like a comprehensive laboratory workbench, we can change out components of a steamfield to perform testing under a variety of conditions (restrict mass, increase pressure, reroute steam, etc.). This flexibility allows us to explore conditions that would require significant infrastructure changes in a real-world setting at a fraction of the cost and time in a digital environment. The results highlight the benefits of using digital twins and advanced data analytics for the geothermal industry.
Research Organization:
National Renewable Energy Laboratory (NREL), Golden, CO (United States)
Sponsoring Organization:
USDOE Office of Energy Efficiency and Renewable Energy (EERE), Renewable Power Office. Geothermal Technologies Office
DOE Contract Number:
AC36-08GO28308
OSTI ID:
1983899
Report Number(s):
NREL/CP-6A20-86437; MainId:87210; UUID:3f72f976-36ec-43fe-8e5d-f2375ec21d25; MainAdminID:69670
Country of Publication:
United States
Language:
English