Geothermal 2.0

NREL launched a 2-year project that integrates physics-based modeling and machine learning to improve geothermal reservoir management. NREL researchers are improving subsurface characterization and modeling of fluid movement through geologic lithologies, fractures, and faults using our Eagle supercomputer. This project focuses on two difficult reservoir management decisions: where to drill and complete makeup wells and how to allocate injection into and production from new and existing wells.

As our PFA projects varied in spatial extent, geologic setting, technical approach, data products, and reporting, NREL was tasked with determining the impact of the projects and comparing their outcomes. Our scientists are evaluating metrics for measuring and valuing exploration knowledge, assessing remaining data gaps in publicly available national exploration knowledge, defining the typical characteristics of geothermal plays, and developing a best practices report for public use throughout the United States and abroad. This project will enhance geothermal exploration practices and the rate of discovery going forward.

NREL partnered with the New Zealand company Upflow on the Geothermal Operational Optimization with Machine Learning (GOOML) project. The project team is applying machine learning to decades of data from hundreds of wells and multiple power generation facilities—including steam, flash, and binary generators—from New Zealand’s massive Wairakei geothermal field.

The GOOML project is creating “digital twin” system models and applying machine learning analysis of these large operational data sets. Findings from the models will improve operations through optimization of daily processes, scheduling of critical asset maintenance, and prediction and detection of potential trouble events. Geothermal operators can use the technology to achieve a step change in the availability and efficiency of geothermal operations without drilling new wells or constructing new plants. This can reduce the levelized cost of electricity for existing and future geothermal installations and increase the economic competitiveness of geothermal.

Preliminary results demonstrate increases in total power generation ranging from of 2% to 11%.

Techno-economic analysis of geothermal systems typically requires assumptions for reservoir and wellbore parameters such as producer-injector well ratio, production temperature drawdown, and production-injection temperature, pressure, and flow rate. To decrease uncertainty of these parameters, NREL analyzed field data from monthly production reports at 19 power plants in California and Nevada covering 196 production wells and 175 injection wells. Read the full report.

NREL is partnering with international research institutions on the Geothermica-funded DEEPEN project to de-risk geothermal exploration in deep, supercritical magmatic plays. The project is developing a play fairway methodology specific to magmatic plays that includes the roots zone of the geothermal system (supercritical fluids, magma bodies, the brittle-ductile transition zone, and other elements). The project is also focused on developing new tools to help with subsurface imaging of deep and hot bodies.