Semiannual Stakeholder Webinar Highlights Roles of Artificial Intelligence and Machine Learning in Water Power

A stakeholder webinar on August 22, 2024, hosted by the U.S. Department of Energy’s (DOE’s) Water Power Technologies Office (WPTO), focused on artificial intelligence (AI) and machine learning (ML) and their potential use across the water power sector.   

Before the discussion turned to AI and ML, Acting Director Matthew Grosso provided updates from WPTO. He discussed the recently released Hydropower Supply Chain Gap Analysis, which identifies five major gaps in the U.S. hydropower supply chain, along with recommendations to address those gaps. A robust domestic supply chain is critical to support new hydropower development and upgrades and refurbishments at existing facilities so this resource can continue to help ensure the electricity grid remains reliable and stable.

He then highlighted recent projects at the DOE national laboratories and prize winners. WPTO announced $1.7 million in “Seedlings” projects, which are funded through the office’s Seedling and Sapling program to encourage and support new and innovative research ideas at DOE national laboratories. In July, 15 teams were awarded $1.2 million in cash prizes in Phase II of the Innovating Distributed Embedded Energy Prize, which challenges competitors to develop novel technologies to harness and convert the power of ocean waves into usable types of energy.

Grosso also discussed technical assistance opportunities for marine energy and hydropower developers to advance their technologies. The Testing Expertise and Access for Marine Energy Research (TEAMER) program helps technology developers and researchers advance their devices while also building knowledge, fostering innovation, and driving commercialization of marine energy technologies. Applications for the latest TEAMER request for technical support are due Oct. 4, 2024. Meanwhile, the Hydropower Testing Network connects hydropower technology developers to testing capability providers. (On Sept. 18, 2024, WPTO opened applications for technology developers interested in receiving testing services! Applications are due Oct. 30, 2024.)

Grosso concluded by highlighting a now open funding opportunity for up to $112.5 million to accelerate the design, fabrication, and testing of multiple wave energy converters, which harness power from ocean waves. “This is our largest ever single funding opportunity,” said Grosso. “We hope it will be a game changer for the marine energy industry in the years to come.”

New DOE Office Coordinating AI 

Two guest speakers from DOE’s new Office of Critical and Emerging Technologies (CET) joined the webinar to talk about their office’s role working with DOE and its national laboratories, which involves identifying potential uses for AI in the energy sector and ensuring those uses are developed safely. Michael Mazur, a CET policy advisor, said the office was established to help implement the 2023 Executive Order on the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence. 

“In parallel with promoting the benefits of AI, we’re really focused on managing the risks,” said Mazur. “We’re deploying AI testbeds, or places where we can try it out—whether it’s new hardware, new software, or other new concepts. And we’re asking ourselves the hard questions: When an AI model is out there, what is it capable of doing? So, we’re doing all those things in protective environments so we can quantify those risks, and then use that to inform the future broader regulatory space.”

Charles Yang, another CET policy advisor, provided more specifics on how the office is assessing ways that AI can support a modern power grid—from grid planning and operations to reliability and resilience concerns. He also shared a recent DOE report  that discusses how AI can help ensure clean energy can meet the challenge of powering AI itself, a growing concern in the sector. 

Yang concluded by talking about DOE’s recent announcement of Frontiers in AI for Science, Security, and Technology, or FASST. “The goal of FASST is to tackle a lot of the issues we see in this space,” said Yang. “So, this is both advancing scientific discovery and national security, but also addressing the energy challenge through continued innovations and energy efficiency, building out that AI workforce, and finally, developing the technical expertise that we need for AI governance.”

Artificial Intelligence, Machine Learning, and Water Power

Swara Salih, a marine energy data analyst at WPTO, spoke next about the basics of AI and ML to set the stage for how it can be used in water power research.  Machine learning “learns” in much the same way as humans: perform an action, make a mistake, learn from the mistake, and adjust the next action accordingly. Artificial Intelligence is the culmination of many machine learning algorithms, making decisions after multiple iterations, and like ML, is dependent on human coding. “Don’t be intimidated by it,” said Salih. “Understand that these are computer programs that simply are running highly iterative processes that expedite and make efficient our own processes.”

WPTO is using these types of learning in operations and maintenance, controls, and designs. Salih and WPTO Hydropower Engineer Kenneth Chandonait provided more specifics on a few WPTO-funded marine energy and hydropower projects that have used different forms of AI.

• Design and Control Case Study: Desalination Research and Development – Membrane Test Stand: This study used ML to predict how easily water and salt could move through the membrane of a wave-powered desalination unit. The results showed the ML algorithm made accurate predictions, regardless of wave characteristics such as shape and speed. 
• Operations and Maintenance Case Study: Embedded Instrumentation for Detection of Flow Separation: This study used ML to develop a novel pressure sensing system on marine energy turbines to optimize turbine performance and predict maintenance needs.
• Environmental Monitoring Case Study: Deep Learning for Automated Identification of Eels in Sonar Data: This study used deep learning to automatically detect adult American eels from sonar data around hydropower facilities. 
• Operations and Maintenance Case Study: Domain-Informed Models for Degradation and Prognostics in Hydro Components: This study used an AI-enabled framework to predict degradation in hydropower bearings, which support and guide rotating parts of turbines. AI could improve maintenance operations by anticipating potential equipment failures so operators can schedule maintenance and plan outages accordingly. 
• Design and Control Case Study: Cybersecurity Situational Awareness Tool: This study involved an intrusion detection system that used rules- and ML-based methods to identify possible cyberattacks at a hydropower facility. 
• Water Temperature Case Study: DOE-Tennessee Valley Authority Climate Research and Development Collaboration: In partnership with the Tennessee Valley Authority, researchers built a foundational dataset for training ML models on daily stream temperature. This data can be valuable for maintaining water temperatures downstream from hydropower facilities. 

Considerations for the Future 

Since AI and ML models need to be trained using large amounts of data, Chandonait says, “our primary areas of emphasis are looking at standardizing that data.” He emphasizes the importance of WPTO’s partnerships through projects like Hydropower Fleet Intelligence, with groups like the Hydropower Research Institute, and with operators, owners, utilities, and other interested parties, for advancing that data to a point where larger insights could be made across the fleet. Once the data has been standardized and aggregated, WPTO’s next step is to work on facilitating and validating explainable algorithms, methodologies, and models for integration into industry digital systems.

In terms of where ML and AI might take the industry, Salih says it is important to consider how this overall tool can “make our jobs easier, help us make predictions on where best to make investments…to just really enhance what we’re able to do.”

Open Opportunities at a Glance

• TEAMER: Applications for the program’s 14^th request for technical support are due Oct. 4, 2024.
• Oceans of Opportunity funding opportunity: Concept papers are due Oct. 25, 2024.
• Hydropower Testing Network: Applications for technology developers to receive testing services are due Oct. 30, 2024. 

Webinar Timestamps

• 0:00—Introduction, Maxine Hillman, Moderator
• 5:15—Matthew Grosso, WPTO Acting Director 
• 15:55—Michael Mazur, Policy Advisor, Office of Critical and Emerging Technologies
• 23:01— Charles Yang, Policy Advisor, Office of Critical and Emerging Technologies
• 29:42—Swara Salih, Marine Energy Data Analyst 
• 41:40—Kenneth Chandonait, Hydropower Engineer
• 57:12—Q&A
• 1:18:15—Opportunities and Upcoming Events, Maxine Hillman

Stay in the know with WPTO! In addition to attending these webinars, stakeholders can learn more about WPTO’s Hydropower and Marine Energy programs. Subscribe to the bimonthly Hydro Headlines and Water Column newsletters and the comprehensive, monthly Water Wire newsletter to get information on funding opportunities, events, and other news.