Just as a clogged kitchen sink can bring household routines to a halt, a blockage in a nuclear reactor's heat exchanger can slow operations and require urgent attention. Researchers at the U.S. Department of Energy's (DOE) Argonne National Laboratory are now using artificial intelligence (AI) to spot these hidden clogs early. This will help molten-salt-cooled reactors (MSCR) run smoothly and reduce expensive interruptions. The study is published in Nature Scientific Reports.
MSCRs are a promising type of next-generation nuclear reactor that can help meet the country's energy needs. The design of such reactors, including the design of key components such as heat exchangers, is still under consideration by the research community.
The research team proposed a first-of-its-kind heat exchanger. This heat exchanger - a type known as a compact matrix-type - features an embedded hardware-software system that allows for efficient monitoring.
"If one of those channels becomes plugged, either partially or completely, we want to be able to detect it, locate it and determine the severity. It's like getting a high-resolution look into where the fault is occurring in the system." - Alex Heifetz, Argonne principal electrical engineer
Nuclear reactors use heat exchangers to move heat away from the reactor core, where the fission chain reactions happen. Fission occurs when the center of an atom splits into smaller parts and releases energy. In the heat exchanger, the hot salt from the reactor core transfers thermal energy to another fluid, which eventually makes steam to rotate a turbine to make electricity.
Keeping Reactor Systems Flowing Smoothly
In the proposed design, each matrix-type heat exchanger typically has 2,000 to 4,000 small pipes, called channels. Larger exchangers have even more. If the salt coolant gets close to its freezing point, it can start to solidify and block these channels. This slows the flow and makes the reactor less efficient. Finding these blockages early is important.
Most heat exchangers today check the temperature and flow rate only at the input and output of the heat exchanger. They do not check inside each channel, meaning problems in the early stages can go unnoticed.
Detecting Problems Before They Escalate
The U.S. Nuclear Regulatory Commission requires nuclear power plants to have an inspection plan for safety-related components. With support from DOE's Advanced Research Projects Agency-Energy, the team proposed a new way to monitor heat exchangers with embedded sensors and AI algorithms. Their system would use distributed fiber-optic temperature sensing and AI to find unusual temperature patterns or anomalies. A temperature anomaly may suggest the channel has become blocked by solidified coolant.
"If one of those channels becomes plugged, either partially or completely, we want to be able to detect it, locate it and determine the severity," said Alex Heifetz, an Argonne principal electrical engineer and co-author of the paper. "It's like getting a high-resolution look into where the fault is occurring in the system."
The AI system also explains to the operator why it raised an alarm. It gives details about the problem, why it matters and how the AI made its decision. This AI-generated explanation, called explainable AI, provides supporting information and helps operators understand and trust the system.
Real-Time Monitoring in Action
The system would use fiber-optic sensors to collect a large volume of temperature data from inside the heat exchanger in real time. The AI checks for channels with unusual temperatures and decides if a blockage is likely.
The sensors would be installed on the mechanical support structures in the heat exchanger, not inside the channels. This means that the sensors do not require through-wall penetration and thus do not affect safety or the heat exchanger's operation. This heat exchanger design with embedded instrumentation is a good candidate for regulatory approval.
"One of the biggest challenges in these systems is detecting faults when the data looks almost identical to normal operation," said Konstantinos Prantikos, a Purdue graduate research assistant, visiting scholar at Argonne and paper co-author. "Our approach can detect barely distinguishable patterns in large volumes of data that would be challenging for a human to inspect. Unlike humans, AI doesn't get tired of looking at thousands of numbers. It also explains its decisions in a way that helps us to understand and trust the systems."
Ensuring Reliable and Cost-Effective Operations
MSCRs run at much higher temperatures than traditional pressurized water-cooled reactors. Their salt coolant freezes at approximately 500 degrees Celsius (932 degrees Fahrenheit). These tough conditions make early fault detection and strong monitoring tools important. They address problems that need to be solved before these reactors can be commercialized.
The new monitoring system addresses these challenges by providing operators with real-time updates on each channel's health. By finding problems early, the system helps develop a remediation plan that would prevent a total heat exchanger failure.
"Repairing or replacing a failed heat exchanger is an expensive procedure that will most likely force an unplanned reactor shutdown," Heifetz said.
Minimizing unplanned shutdowns is critical, as MSCRs are typically designed to operate for up to 10 years before needing new fuel. Reliable monitoring keeps reactors running and means maintenance repairs are done only when needed.
"AI is always on guard. It doesn't sleep," said Prantikos, underscoring the system's continuous vigilance.
Paving the Way for Next-Generation Nuclear Reactors
By combining advanced sensors with explainable AI, Argonne is making reactor structural monitoring more accurate and responsive. This new way to find and diagnose problems will help operators keep reactors working well and reduce risks. As MSCRs move closer to real-world deployment, these tools will support the next generation of nuclear technology.