More from UWC 2020

The democratization of AI

Adam Dow, principal manager of enterprise risk management at Southern California Edison, organized a UWC session in the Technology and Innovation track entitled “The Democratization of AI—Using AI/ML to Assess Risk, Perform Inspections and Make Decisions.” The panel discussion was about how utilities are making use of artificial intelligence and machine learning on a broader scale than ever before, thanks to “low-code” or “no-code” development options and plug-and-play capabilities for use in the field. Joining Dow as panelists were Kumar Mankala, a data scientist and AI products lead at SCE; Ralph Granja, a principal consultant at Accenture; and Kim Gilbert, director of technical commercial engineering at Beyond Limits AI.

• Dow talked about the huge task of inspecting SCE’s electricity delivery infrastructure, which is spread across 50,000 square miles, and how AI enables predictive maintenance.
• For Dow, relying less on reactive maintenance and more on predictive maintenance means looking beyond the age of components and predicting failure based on factors including climate and weather, manufacturer, location, materials, and external stressors to create “an ensemble model for failure.” Compared to age-based models, that probability-based model predicts two to five times more potential failures. Good machine-learning predictive models come from a successful partnership between engineering resources and data science resources to “really doing the statistics and modeling that allow us to get to where we want to go,” Dow said.
• Mankala said the goal of the session was to “disprove the hypothesis that AI is difficult and only experienced data scientists can do it.” AI has progressed from the human-centric assumptions that initially limited its use, he said.
• Mankala proposed a “warp jump” to a new AI framework of the 2020s that doesn’t require users to have a specialized skillset—a “democratized” AI. Building that framework starts with focusing on a clear business need, then optimizing data collection and handling to be more user friendly. Mankala shared a rule of thumb: “If there is something that takes a human three seconds to assess, it can be automated by AI.”
• Granja spoke of his experience inspecting utility infrastructure for wildfire risks by using AI and machine learning to assess aerial and ground-based equipment images. The AI software reduced image analysis time by 98 percent, and Granja emphasized that each type of equipment required a separate learning curve.
• Gilbert gave attendees a look at what has been called cognitive AI, which she said is about “encoding human intelligence alongside machine learning.”
• Beyond Limits AI has licensed technology developed by NASA’s Jet Propulsion Laboratory. “What we do that’s a little different is we encode symbolic AI,” Gilbert said. Symbolic AI helps the software resolve conflicts and make recommendations in a human-like output.
• Gilbert emphasized that “it’s not all about the data, it’s about what the workforce knows.” A more experienced nuclear plant operator can often make quicker decisions. Cognitive AI captures operator knowledge through interviews and makes it part of the knowledge base, alongside numerical power plant data. The software can then provide recommendations, ”with the idea that the inexperienced operator can work at the same level as the more experienced operator.”

Approaches to decommissioning

The Decommissioning track of the UWC’s educational sessions began Tuesday with the session “Innovative and Historical Approaches to Decommissioning,” organized by Lynne Goodman of Marathon Consulting Group and Joe Carignan of Carignan and Associates.

The session addressed what is involved in decommissioning a nuclear power plant and approaches to decommissioning, including regulatory perspectives and new innovative decommissioning models. The featured presenters included Gerry van Noordennen, senior vice president of regulatory affairs for EnergySolutions, and Patricia Holahan from the Nuclear Regulatory Commission.

• D&D history: Carignan began the session by noting that nuclear power utilities face a steep learning curve when their plants close and move to decontamination and decommissioning, including adapting to changing safety and staffing issues. Fortunately, Carignan said, the industry has a solid history of decommissioning to draw upon. “We did this work in the past, we just need to do it faster, cheaper, and safer,” he said.
• Management approaches: Carignan also noted the various options available to utilities when it comes to managing the decommissioning of their facilities, from self-performing the work to hiring a general contractor to transferring the site license and assets to a third-party D&D company. Carignan said each option has its own set of advantages and challenges and that each utility has to select the best method for itself based on its individual circumstances. “Most of it comes down to risk,” he said. “How much risk is the owner willing to accept.”
• The five D&D models: Following Carignan, van Noordennen elaborated on the different management approaches to D&D, grouping the approaches into five different decommissioning models and giving real-life examples of each model. The five models and their examples include:

1. The utility self-performs the decommissioning (Humboldt Bay).
2. The utility hires a decommissioning contractor (San Onofre).
3. The utility integrates a decommissioning staff into its organization (Fort Calhoun).
4. The utility transfers its plant license to a decommissioning contractor (Zion and La Crosse).
5. The utility transfers its license and site assets to a decommissioning company (Vermont Yankee).

• Which model is best? “One model doesn’t fit all sites,” van Noordennen said, adding that the best model for a specific plant depends on a number of factors. These can include state regulations, the plant owner’s desire to retain staff and property, whether the plant is a merchant power plant or in a regulated market, the balance of the reactor’s trust fund, a utility’s desire to stay in the nuclear market, and the availability of nuclear disposal sites. Van Noordennen also said that new, future models that better serve specific sites are possible.
• Rulemaking: Holahan, who serves as the director of the NRC Division of Decommissioning, Uranium Recovery, and Waste Programs, provided an update on some of the rulemaking activities the agency is undertaking affecting D&D and waste disposal. These include a new decommissioning rule that is intended to streamline the decommissioning process by reducing the need for license exemptions from existing regulations.
• Holahan noted that the proposed rule was submitted to NRC commissioners by agency staff in May 2018, but the commissioners have yet to vote on it. The fact that a final rule has not yet been worked out, however, is not hindering the decommissioning process for closed reactors because the NRC has been able to process exemption requests in a timely manner, she explained.
• CABs: Holahan also highlighted the NRC’s report to Congress on best practices for establishing local community advisory boards (CAB) involved with the decommissioning of nuclear power plants. The July 1 report, which was required by the Nuclear Energy Innovation and Modernization Act, examined lessons learned from existing CABs, which are intended to foster communication between local communities and the owners/licensees of plants undergoing decommissioning.
• “We found that it is better to form CABs earlier in the decommissioning process, rather than later,” Holahan said of the NRC’s best practices. Other best practices include the development of a CAB charter or guiding document; consideration of local preferences for holding public meetings; having diversity in CAB membership; and establishing meeting frequency and topics for discussion based on site status, ongoing activities, and the level of stakeholder interest.

Subsequent license renewal

In the past year, the Nuclear Regulatory Commission marked a major achievement for a new regulatory process. The agency approved the first two subsequent license renewal (SLR) applications to extend the Turkey Point-3 and -4 and Peach Bottom-2 and -3 nuclear power plant lifetimes an additional 20 years, for a total operation of up to 80 years. Turkey Point received its SLR in December 2019, with Peach Bottom’s granted in March 2020.

The UWC session "Subsequent License Renewal: Is 80 the New 60?" included a summary of the efforts that led up to the submittal and approval of the first SLR applications. The session was sponsored by the Regulatory Relations track and was moderated by Greg Robison of Duke Power. Its speakers included Anna Bradford, director of the NRC's Division of New and Renewed Licenses in the Office of Nuclear Reactor Regulation; John Kotek, vice president of policy and public affairs for NEI; Keith Miller, a generation project manager for Dominion Energy; and Heather Galloway, SLR site liaison at Duke Energy’s three-unit Oconee plant.

The session’s speakers revealed lessons learned and planning insights for future SLR applicants.

• Bradford noted that the NRC has identified the top four technical issues that need review to assure safe operation of a nuclear plant to 80 years. “Those are neutron embrittlement of the reactor pressure vessel, stress corrosion cracking and other types of degradation of reactor pressure vessel internals, concrete and containment degradation, and electrical cable qualification and condition monitoring,” she said.
• Bradford noted that the NRC’s position on SLR is that it is the licensee’s responsibility to develop the technical basis for safe, long-term operation of the plant, and that absent a generic resolution or consensus on the four issues noted above, it is the licensee that needs to address each of the issues on a plant-specific basis.
• In addition to the SLR applications from Turkey Point and Peach Bottom, a third application was received by the NRC in 2018—for the Surry-1 and -2 units. Bradford said that the NRC completed its review of the Surry application and found that it met the agency’s regulatory requirements, but that the licensee was waiting for a coastal management certification from a separate authorizing agency. The certificate needs to be in place before the NRC can issue the SLR for Surry. “The actual issuance of that 80-year license is currently on hold,” she said.
• Bradford talked about the results of an industry survey of prospective SLR applicants conducted by the Nuclear Energy Institute in March of this year. The survey showed that 49 more units were forecasted to apply for SLRs.
• The NRC’s lessons learned so far, Bradford said, include that SLR reviews can be done in 18 months, preapplication submission meetings between the licensee and the NRC improve coordination, and the effective use of audits by the licensee reduce the requests from the NRC for additional information.
• John Kotek of NEI discussed the value of license renewal in a clean energy system. “More and more big companies and consumers are demanding clean electricity,” he said. “That’s causing states and utilities to respond with deep decarbonization goals.” He added that the further the states and utilities look into attaining the goals, the more they realize that maintaining the existing nuclear fleet is the least expensive way to avoid carbon emissions.
• Kotek said that state and federal policy frameworks will be needed that recognize the full value of nuclear and that are going to make it worthwhile for licensees to pursue SLRs.
• Keith Miller serves as Dominion Energy’s environmental lead for the SLR projects for the Surry and North Anna plants. “Dominion is in a particularly unique situation among utilities involved with SLR,” he said. “We were involved with the first license renewal of our stations in the early 2000s, and we’re one of the first few utilities to submit an initial license renewal application [for an additional 20 years].”
• Looking back at the process taken for Surry’s and North Anna’s first license renewals in the early 2000s, Miller noted that the units are very similar three-loop Westinghouse plants. “The similarities allowed for a single license renewal application to be submitted for both stations,” he said, “a first for the industry.” He added that one safety evaluation report was written for both plants.
• Miller explained that the plants’ design similarities that were leveraged for the first license renewal process were leveraged again during the recent SLR process. He added that about a year and a half was spent on developing the current SLR application, which was submitted to the NRC in October 2018. Other first license renewal and SLR applications were benchmarked for the Surry application, he noted.
• Heather Galloway said that Duke has announced plans to renew licenses to 80 years of operation for all 11 of its nuclear units. She noted that the Oconee plant is the first Duke plant to go through a second round of license renewal.
• Duke has formed a dedicated team to work on Oconee’s SLR application. Galloway said that the team is closely following the first plants (Turkey Point, Peach Bottom, and Surry) that worked on SLR application development. The team is also participating in license renewal working groups, and team members are currently chairs of the Electrical (S)LR and Mechanical (S)LR Working Groups.
• Oconee was the second plant to submit for its first license renewal, back in 1998, and it plans to submit an SLR application in late 2021.
• “SLR gives us a business opportunity to make changes that we may not otherwise make if we were going to be shutting down the plant after 60 years,” said Galloway, noting that Oconee was replacing three low-pressure turbines for each unit, at a cost of $58 million for each unit. Another job was a stator replacement at the two-unit Keowee hydroelectric station. “Oconee is the only nuclear plant in the country to have a hydro station as its backup power,” Galloway explained. Each unit at the Keowee station needed a stator at a cost of $50 million apiece. “This assures long-term operation of not just Oconee, but also the Keowee hydro station,” she said.

Xcel Energy/DOE/INL Transformational Project

Sponsored by the Executive/Leadership track, this late-morning UWC session served as a follow-up to the opening plenary by giving details about the Transformational Change Project implemented by Xcel Energy in cooperation with Idaho National Laboratory. The project’s goal was to improve Xcel’s business model, using lessons learned from the North Sea oil and gas industry.

■ Don Bostic, Xcel’s general manager of nuclear strategy and team leader for the new business model initiative, focused his presentation on one part of the initiative: integrated operations (IO) for the nuclear power industry. He defined integrated operations as the integration of people, processes, and technology to make and execute better decisions, faster. IO also includes the advancements in information technology that remove the physical boundaries between people, making real-time collaboration and remote operation possible for managing multiple distributed assets.

■ Bostic said the current business model for the nuclear industry is not competitive. He added that the Delivering the Nuclear Promise initiative had some effect, but it still wasn’t enough to keep the industry competitive for the future.

■ Technology-centered IO allows for common governance and decision-making in one central location for an organization, rather than the current system of operating fleets of autonomous sites with manual data handling and processes done remotely. With IO, data is processed through technology, analytics, and artificial intelligence, thus removing the possibility of human error in data collection.

■ “We are currently at an inflection point and we must adapt,” Bostic said. “Being carbon-free is not enough by itself for the future. But this is our decision at Xcel Energy. Everyone will have to look at their own circumstances and determine how you each want to respond to the challenges that face our industry.”

■ Hank Butterworth, Xcel’s general manager of nuclear transformation, presented information on the three steps the company took to implement IO. The first step was to do a top-down analysis. Xcel formed four teams to set the initial vision and structure using IO principles and challenging its performance targets. The second step was a capability analysis, which involved partnering with INL and Halden Labs to identify IO capabilities and to develop a “nuclear capability stack.” A capability stack is a structure of critical business capabilities needed to achieve a desired outcome. For Xcel, that meant assessing operational, supporting, and foundational capabilities.

■ Operational capabilities include operating, maintaining, and supporting the plant. Some supporting capabilities include adaptability, collaboration, innovation, and analytics. Finally, foundational capabilities include the physical plant, the plant information data stream, and plant by-products.

■ The final step in Xcel’s IO implementation was creating a transformational roadmap to serve as an integrated plan to institute the changes needed to the organization, processes, technology, and workforce. “Integrated operations is a way to get our costs down, to get to the point where we have less people to operate the facilities, and still operate very well,” Butterworth said.

■ Moderator Ken Thomas, senior consultant at INL, presented some ways in which the Department of Energy is making IO transformation information available to the public. He said the transformation is about more than the modernization of plant systems and work processes. It can also apply to addressing obsolescence of plants and reliability issues.

■ Thomas pointed out the special challenges that nuclear must consider in the IO transformation. For example, the nuclear industry cannot compromise on its safety culture, and any change cannot impact the revenue flow from power generation. Still, those challenges should not stand in the way of adopting and adapting to a new model. “The good news is that we all recognize that a new paradigm is needed,” Thomas said. “That’s what the theme of this year’s UWC meeting is all about. There’s a great deal of optimism that we can do this.