AI can significantly improve grid management reports DOE

Artificial intelligence (AI) has the potential to significantly improve key areas of grid management, a new US Department of Energy report finds.

The report, focussed on potential opportunities for AI in the grid, finds that AI is able to improve grid planning, permitting, operations and reliability and resilience as well to advance the broader transition to a clean energy economy.

However, it is crucial that new AI use cases do not introduce new risks to the grid, the report states. Thus AI models for grid applications should be rigorously validated as well as scalable in performance and adherent to power grid governance standards.

The report AI for Energy was prepared as a requirement of an October 2023 Executive Order on the safe development and use of AI.

Have you read?
How predictive AI is helping smooth the path to net zero
Collaboration key to advancing AI in energy sector say WEC panelists

It identifies priority use cases in the four broad areas of grid management where AI can be immediately deployed to improve the grid.

Among these are AI-accelerated power grid models for capacity and transmission studies, large language models to assist compliance and review with federal permitting, advanced AI to forecast renewable energy production for grid operators and smart grid applications of AI to enhance resilience.

Beyond the grid, AI is seen to support a range of applications to help advance an equitable clean energy economy.

Examples include optimising planning for electric vehicle (EV) charging networks, enabling virtual power plants, generating design of structural materials for manufacturing and discovering alternatives for critical materials.

Employing a portfolio of these AI-enabled solutions, while mitigating any potential risks, can support transformations needed across the economy to tackle the climate crisis, reduce costs and improve lives, states the report.

The report also comments that while AI applications for energy hold the promise of great opportunities, there are also potential risks and widespread deployment requires thoughtful consideration of the societal impact.

Careful consideration of how AI deployment affects different stakeholders and industries can mitigate downstream risks or unforeseen hazards – for example, AI itself may lead to significant load growth that adds burden to the grid.

Additionally, AI should be both equally accessible by all – including equal access to workforce opportunities in this growing industry – and designed so it doesn’t cause disparate harms.

“Artificial intelligence can help crack the code on our toughest challenges from combating the climate crisis to uncovering cures for cancer,” said US Secretary of Energy Jennifer M. Granholm.

“DOE is accelerating its AI work on multiple fronts to not only keep the US globally competitive but also to manage AI’s increasing energy demand so we can maintain our goal of a reliable, affordable and clean energy future.”

AI actions

Alongside the report on AI opportunities, the DOE published a report on the risks, which identifies four categories of potential risk, i.e. unintentional failure modes of AI, adversarial attacks against AI, hostile applications of AI and compromise of the AI software supply chain.

It is proposed over the coming months to convene energy stakeholders and technical experts to collaboratively assess the potential risks of these to the grid, as well as ways in which AI could potentially strengthen grid resilience and the ability to respond to threats.

Other initiatives highlighted include the new ‘VoltAIc’ initiative aimed to use AI to help streamline siting and permitting at the Federal, state and local levels. As part of this DOE has partnered with Pacific Northwest National Laboratory (PNNL) to develop PolicyAI, a policy-specific large language model test bed to develop software to augment the National Environmental Policy Act and related reviews.

A new Working Group on powering AI and data centre infrastructure has also been established, which is due to make recommendations on meeting the energy demand by June.

AI grid priority use cases

Grid planning

• Completing, correcting and harmonising sparse data on grid infrastructure to inform predictive asset replacement
• Assessing dynamic system conditions to inform upgrades, maintenance and new resource needs, as well as dynamic assessments of available grid capacity
• Preventing avoidable losses through predictive maintenance
• Detecting faults in solar panels, dams, wind turbine blades, generators, etc.
• Processing aerial images for remote job-site inspections
• Informing adoption of grid-enhancing technologies and accelerating interconnection queues to get projects connected to the grid
• Enabling modelling for distributed energy resource adoption to anticipate distribution system upgrades and implications for load and load shape.

Grid permitting and siting

• Organising, extracting, consolidating information across Federal, state, and/or local regulations to improve the efficiency of administrative processes
• Accelerating environmental review process, e.g. for comment processing, information extraction, drafting documents, automating compliance checks, etc.
• Optimising placement of renewable energy and transmission projects to facilitate effective and efficient siting and permitting
• Generating size/location data for rooftop solar panels, optimal placement of wind turbines, etc.
• Identifying and managing sites for geothermal energy, using satellite imagery and seismic data
• Placing hydropower dams in a way that satisfies energy and ecological objectives.

Grid operations and reliability

• Improving variable renewable energy forecasting
• Improving demand forecasting using AI trained on historical data, including weather, climate, economic and load
• Improving power system optimisation, reducing the computational intensity of modelling
• Setting real-time pricing to optimise the operation and/or economics of distributed energy resources, storage, etc.
• Anticipating system anomalies to avoid disruption.

Grid resilience

• Enabling proactive monitoring to make critical infrastructure more resilient to severe weather
• Monitoring, detecting and diagnosing anomalous events, e.g. extreme weather events, cyber¬attack
• Improving coordination with other interdependent systems such as natural gas and water to regain operation after disruption
• Enhancing situational awareness across the system with coupled AI and digital twins
• Improving the accuracy and interpretability of landslide predictions, sea level rise, storm surge, etc.
• Simulating disruption/disaster scenarios to inform resilience strategies
• Enhancing system efficiency and coordination to restart the grid during full or partial blackouts
• Optimising the deployments of repair crews to accelerate response
• Identifying the fastest path to system restoration.

AI clean energy priority use cases

Transportation

• Optimising EV charger planning, permitting, and siting
• Optimising EV charger usage and pricing for a variety of customers to balance user charging preferences and grid load
• Enabling vehicle to grid operations and providing grid services through EV or EV supply equipment assets
• Enabling EV fleet coordination through vehicle to vehicle and advanced charging.

Buildings

• Unlocking virtual power plant adoption through improved customer segmentation and incentive allocation
• Drive materials innovation in building materials, e.g. low carbon cement
• Optimising energy use in buildings
• Modelling buildings to predict energy, load shape, appliance disaggregation, future consumption and coordination with power system
• Coordinating demand response programmes, the internet of things, smart appliances, distributed energy resources, etc.
• Optimising HVAC performance and operation to energy efficiency and/or demand response priorities
• Estimating marginal emissions factors, providing customers feedback about energy & emissions, suggesting behavioural interventions
• Leveraging AI and digital twins to optimise operations and resilience across the built environment.

Industrials and manufacturing

• Improving manufacturing quality control and better sort feedstocks for recycling streams
• Reducing carbon footprint of industry, data centres by optimising energy consumption, cooling, etc.
• Revolutionising component design through generative inverse design, particularly when paired with advanced manufacturing techniques
• Optimising predictive maintenance and operations optimisation to improve manufacturing efficiency and performance.

Agriculture

• Optimising colocation of renewable energy with agriculture for synergistic benefits, e.g. agrivoltaics
• Supporting bioeconomy and biomanufacturing R&D
• Using deployable field sensors and satellite imagery to better map and predict agricultural yields
• Optimising precision agriculture and delivery of fertilizer and water to crops.