Opportunities for Using AI in the Nuclear Energy Industry

There is no shortage of ideas for using artificial intelligence to address the complex problems facing developers of new nuclear reactors as well as operators of existing nuclear power plants.

For any nuclear energy project that involves the development of artificial intelligence applications, the focus must be on delivering results for the organization that relate to the key performance measures (KPIs) / project management milestones in its strategic development and operational management plans.

The success of every project, whether it involves artificial intelligence, or any other software as a service, is composed of three parts. Did the use of the computer tool result in the project being on time, within budget, and were the facilities, components, or services desired by the customer delivered as specified?

Fundamentally, the goal of deploying artificial intelligence applications is to reduce the amount of friction in moving information (data) encountered within standard engineering and/or business processes or used for decision making, resulting in an increase the efficiency and effectiveness overall of the organization. The key question is whether the investment of resources in AI produce measurable results?

A key caveat for any of these ideas is that major challenges in data validation and verification of the results of the use of AI for managing nuclear energy await any developer entering this field.

What Kinds of Nuclear Energy Questions Can AI Address?

In this blog post three questions will be explored as examples. This doesn’t mean they are the best questions, but they are useful to illustrate the issues.

The ideas presented here are not unique, and if you put a half dozen Chief Nuclear Officers in a room, over lunch they would come up with a much more definitive list with better detail.

There are many opportunities and challenges for the uses of AI in the nuclear energy industry. These examples are intended to illustrate some of the thinking going on at this time.

Nuclear Plant Management: How can artificial Intelligence (AI) play a significant role in the management of nuclear energy across various aspects including safety, efficiency, and maintenance.

Development of Advanced Reactors: How can artificial Intelligence (AI) revolutionize the design of advanced nuclear reactors by offering numerous applications that enhance safety, efficiency, and innovation in nuclear energy reactor design, fuels, and production of electricity, process, heat, hydrogen, desalination, etc.?

Data Centers Use of SMRs for Reliable Power: How can AI address the challenges and opportunities facing data centers that are considering the use of reliable power from Small Modular Reactors (SMRs)?

Nuclear Plant Management for Existing Reactors

Safety Monitoring and Control: AI systems can be used to continuously monitor and analyze vast amounts of data from sensors in nuclear power plants. These systems can be designed to quickly detect anomalies, predict potential issues, and initiate corrective actions to ensure the safety and stability of nuclear reactors. AI-based control algorithms could be develop to dynamically adjust reactor parameters to optimize performance and respond to changing operational conditions, ensuring stable and efficient operation.

Predictive Maintenance: AI algorithms can analyze historical operational data to predict when critical components of nuclear power plants might fail or require maintenance. This proactive approach can prevent costly downtime and ensures the reliability of nuclear energy production.

Radiation Monitoring and Management: AI-powered systems could be employed for real-time monitoring of radiation levels within nuclear facilities and their surroundings. These systems can identify patterns and trends in radiation data, helping operators to mitigate potential risks and ensure regulatory compliance. They also can monitor radiation exposure for workers which means not waiting for the end of the month to pull badges.

Fuel Management: AI algorithms can be developed to optimize the utilization of nuclear fuel by predicting demand, managing inventory, and optimizing outage / refueling schedules. This helps maximize the efficiency of nuclear reactors and reduce operational costs. Also, better scheduling will reduce the loss of revenue from each outage.

Simulation and Training: AI-driven simulations could be used in simulators and digital twins to train nuclear plant operators to run a new reactor or to train new operators for an existing plant. Also, AI can be used in various scenarios, including emergency response procedures. These simulations provide a safe and realistic environment for operators to enhance their skills and decision-making abilities.

Human-Machine Interaction: AI interfaces could provide intuitive ways for operators to interact with complex reactor systems, facilitating decision-making and enhancing overall operational efficiency.

Security and Threat Detection: AI-based surveillance systems could be to enhance the security of nuclear facilities by using sensors to detect and identify potential security threats, such as unauthorized access, mis-use of facilities or cybersecurity intrusion attempts.

Regulatory Compliance: AI systems can assist in ensuring compliance with regulatory requirements by automating the analysis of vast amounts of data and documents related to safety protocols, environmental regulations, and operational procedures.

Development of Advanced Reactors

In addition to all the uses for existing reactors mentioned so far, three areas seem ripe for the use of AI in development of advanced reactors.

Design Optimization: AI algorithms could be used to explore vast design spaces more efficiently than traditional methods, identifying optimal reactor configurations that maximize safety, performance, and cost-effectiveness.

Materials & Fuels Sciences: AI-driven material modeling could accelerate the discovery and development of novel materials and fuels with improved properties for performance, reliability in areas such as radiation resistance and thermal conductivity.

Supply Chain Management: The development of a master equipment list could drive production of procurement specifications, performance characteristics, etc., the help keep track of the schedule for delivery of key components tied to the project management system.

Using AI to Address the Use SMRs to Power Data Centers

There has been a lot of interest in the use of small modular reactors to power large (hyper) data centers being built by major platforms like Amazon, Facebook, Goggle, Microsoft, and others.  While AI may not in the near term be definitive in leveraging the use of SMRs to power data centers, some of the challenges and opportunities of using AI can create cost effective solutions to current challenges and enhancements to ROI for these kinds of deals.

Currently, only one US firm has a licensed design of an SMR. With an agreement in principle to build 24 SMRs for two data centers, 12 at each site, if the firm broke ground tomorrow, it would be the at least three to five years before the first SMRs were completed.

For SMR designs that have not yet submitted their reactor designs for NRC review, any SMRs they might build, e.g., at TVA or OPG, etc., are at least seven years away – four years for the NRC to complete the safety reviews and another three to build the first of a kind unit.

Even with these timelines in mind, it is still useful to consider the needs of data centers considering the use of power from Small Modular Reactors (SMRs) face both challenges and opportunities.

How could AI speed up the regulatory review process and construction of the first SMRs saving customers and reactor developers alike tens if not hundreds of millions of dollars? How can AI produce better ROI for the nuclear utility building an SMR and a more competitive power purchase deal for a data center to attract new customers? Here is a brief review of some ideas about the challenges and opportunities to achieve these kinds of results.

Challenges

Regulatory Hurdles: The regulatory landscape surrounding SMRs pose challenges for SMR developers seeking data centers as customers, as licensing and permitting processes can be lengthy and complex. Better understanding of regulatory requirements, and the data needed to comply with them, can be addressed by AI in terms of accessing NRC documents and the managing the complex housekeeping associated with document management, revisions, and submissions (topical reports) leading up to a license application. Assembly of the license application itself lends it self to the US of AI for the purposes of continuing, complete compliance with license applications requirements, indexing, change control and other complex document management functions

Initial Capital Investment: While SMRs offer potential cost savings over traditional nuclear reactors due to their smaller size and modular design, the initial capital investment required to deploy SMRs is still be significant. At even the very competitive rate of $4,000/KW, a 300 MW SMR will cost $1.2 billion.  Data centers will never be developers of SMRs, but they can be significant customers through power purchase agreements.

Even so data centers must carefully evaluate the financial feasibility of integrating SMR-powered solutions into their infrastructure. Data validation, verification of analytic methods of economic / financial feasibility, tracking formulas and data cell references in complex spreadsheets are all ripe for the use of AI.

Public Perception and Acceptance: Public perception of nuclear power, particularly SMRs, can influence stakeholder / investor / customer attitudes. Data centers may encounter resistance from local communities or environmental groups opposed to nuclear energy, requiring effective communication and outreach efforts to address concerns and build trust.

Active monitoring of the news media, surveys of public opinion, and testimony by citizens and experts (pro and con) are local, state, and federal / congressional hearings can produce better understanding of key issues considered important by various stakeholder groups and initiation of effective responses to their concerns.

Safety and Security: Data centers must have a means to work with the reactor vendors and EPCs to assess the robustness of SMR designs and implement measures to mitigate risks and safeguard critical infrastructure and personnel especially if the data center is located adjacent to the SMR. AI modeling of accident event management could address these issues.

Opportunities

Scalability and Flexibility: SMRs offer scalability and flexibility that align with the evolving: energy needs of data centers. Modular design allows for incremental capacity expansion, enabling data centers to scale their power generation capacity according to demand growth while minimizing upfront investment and construction lead times.

For existing nuclear power plants with power purchase agreements with data centers, a new SMR may make more sense than a power uprate to meet evolving increases in demand for electricity. Further, it may make sense for existing nuclear utilities to build SMRs to meet the needs of data centers.

A new SMR will have a service life of at least 60 years. With many US nuclear plants being 40 years old or more, the SMRs will outlast the current plant yet continue to be able to take advantage of the grid connections, labor force, and especially management expertise of the nuclear utility.  TVA and OPG probably have this kinds of factors built in to their plants to build SMRs at Clinch River, TN, and Darlington, ON. New SMRs can be added to the site if the business case supports this level of investment.

Reliable Baseload Power: SMRs can provide a reliable baseload power source with high availability and low operating costs, ensuring uninterrupted electricity supply to data centers. This reliability is crucial for maintaining continuous operations and meeting stringent uptime requirements.

AI can be used for load following and grid stability management given multiple sources of power generation, e.g., nuclear, gas, and renewables on the grid.

By deploying SMRs, data centers can enhance energy independence and resilience by diversifying their energy sources and reducing reliance on grid-supplied electricity. This strategic diversification minimizes exposure to grid disruptions, price volatility, and geopolitical risks associated with fossil fuel dependence.

Low-Carbon Energy: SMRs produce electricity with minimal greenhouse gas emissions, offering data centers an opportunity to reduce their carbon footprint and demonstrate environmental responsibility. Integration of SMR-powered solutions aligns with sustainability goals and enhances corporate social responsibility initiatives.

AI can be used to manage the excessive heat generated by the facility by intelligently shedding computational loads to other data centers when ambient heat outside the plant affects cooling water supplies.

Long-Term Cost Savings: Despite the initial capital investment, SMRs can offer long-term data centers cost savings over the operational lifespan, thanks to lower fuel costs, reduced maintenance requirements, and predictable operating expenses. Data centers can benefit from stable energy costs and improved financial predictability, enhancing competitiveness in the market.

Use of AI in financial analyses can help developers and data center owner/operators to quickly arrive at feasible business cases for inking power purchase agreements with both SMRs and existing nuclear utilities.

Overall, while challenges exist, the opportunities presented by leveraging AI for deployment of SMRs for data centers include scalability, reliability, sustainability, and cost-effectiveness, positioning SMR-powered solutions as a promising option for meeting the growing energy demands of the digital economy.

# # #