Researchers develop AI to cut staff costs for advanced reactors
A new U.S.-funded research project aims to use artificial intelligence (AI), machine learning and expert safety analysis to reduce operational staff count and help secure the long-term competitiveness of nuclear power.
The U.S. Department of Energy (DOE) recently awarded $3.4 million to a research team led by North Carolina State University to perform preliminary development of a Nearly Autonomous Management and Control (NAMAC) system for advanced nuclear reactors.
Developers of next generation reactors believe design efficiencies and modular construction will cut costs compared with conventional large-scale plants. The NAMAC project was one of 10 projects awarded funding under the DOE's $24 million Modelling-Enhanced Innovations Trailblazing Nuclear Energy Reinvigoration (MEITNER) program. Led by the Advanced Research Projects Agency-Energy (ARPA-E), MEITNER aims to support the development of innovative technologies that improve the safety and reduce the cost of advanced reactors.
The NAMAC project aims to develop a "highly automated" management and control system, the research team said in a statement. The system will provide operational insights using artificial intelligence, continuous data monitoring and machine learning to predict future plant status.
"Ultimately, the team seeks to enable a significantly smaller operational staff to manage the plant, assisted by instrumentation, operator training, and smart procedures, reducing overall operational cost," the researchers said.
Falling wholesale prices have sliced nuclear margins, making operational efficiency critical. More than a quarter of U.S. operational reactors do not earn enough revenue to cover operating costs, according to a report published by Bloomberg New Energy Finance (BNEF) in May.
Forecast margins for conventional US reactors
(Click image to enlarge)
Source: Bloomberg New Energy Finance (BNEF), May 2018.
The target of reducing staff levels while maintaining high levels of safety was a key reason behind the NAMAC project's selection, Rachel Slaybaugh, ARPA-E Program Director, told Nuclear Energy Insider.
“If the next generation of reactors have features that enable them to run with less operating staff, they can be more cost competitive,” Slaybaugh said.
The NAMAC research team includes members from Ohio State University, New Mexico State University, Oak Ridge National Laboratory, Idaho National Laboratory, nuclear engineering firm Zachry Nuclear and advanced reactor developer TerraPower.
The project team will draw from a wide range of specialist expertise in areas such as modelling and simulation, artificial intelligence and machine learning, accident analysis and reactor codes and standards, Slaybaugh noted.
Analysis of safety-significant events and incorporation of existing safety guidelines will form a key part of the research.
"It is important that the new design appropriately integrates emerging sciences and technologies for all modes, including Emergency Operating Procedures (EOPs) and severe accident management,” the researchers said in a briefing document.
ARPA-E will work with the MEITNER research team to set project timetables and will review progress against milestones on a quarterly basis.
In the U.S. and Europe, conventional and advanced nuclear power operators are implementing analytics and increasing automation levels in a bid to cut costs.
SMR developer NuScale aims to use automation, component standardization and simpler design to reduce staff count to 0.7 per MW, lower than the industry average, Ross Snuggerud, NuScale’s Senior Operations Engineer, told Nuclear Energy Insider in April.
NuScale aims to be the first U.S. developer to build a commercial SMR plant, delivering a 12-module 600 MW plant to Utah Associated Municipal Power Systems (UAMPS) by the mid-2020s.
"We've put a lot of effort into developing high levels of automation and leveraging the simplicity of the design," Snuggerud said.
Since 2015, Exelon and GE Hitachi have been developing and implementing big data analytics at Exelon plants to enhance operational efficiency.
The analytics project has achieved powerful results, accurately predicting actionable fault events more than three months in advance, Mona Badie, Chief Digital Officer at GE Hitachi Nuclear Energy, told Nuclear Energy Insider in November.
Advances in wireless technology could be a "game changer" for conventional plant operators, Joe Donahue, Vice President of Nuclear Engineering at Duke Energy, said at the 2017 Nuclear Plant Digitalization Conference.
New wireless systems will help drive up operational efficiency by enabling operators to integrate advancing sensor technology into centralized analytics platforms, he said.
Nuclear digitalization projects are also tackling data quality and training challenges to pave the way for machine learning capabilities.
The NAMAC team hope their systems can be used to develop and validate EOPs and Severe Accident Management Guideline (SAMGs) for advanced reactor designs.
“A comprehensive, knowledge-based control system for credible, consistent management of plant operations will improve safety and optimize emergency management,” the research team said in their briefing note.
The algorithms produced by the research are expected to be incorporated into commercial software packages that will be used in future nuclear reactors, Slaybaugh said.
In addition, NAMAC research findings can be used to inform the design of instrumentation and control systems for advanced reactors, "which could also have commercial impacts,” she said.
By Neil Ford