advanced-manufacturing-techniques
The Role of Advanced Robotics in Candu Reactor Inspection and Maintenance
Table of Contents
The Indispensable Role of Inspection and Maintenance in CANDU Reactors
CANDU reactors, a Canadian-designed pressurized heavy water reactor (PHWR) technology, form a cornerstone of electricity generation in several countries. In Canada alone, these units supply roughly 15% of the nation's electricity, with additional operating fleets in Romania, South Korea, Argentina, and China. The horizontal pressure tube design, use of natural uranium fuel, heavy water moderator, and on-line refueling capability provide distinct operational and economic advantages over light water reactor designs. However, these very features also introduce uniquely complex inspection and maintenance challenges that push the boundaries of conventional human-based approaches.
A typical CANDU unit contains between 380 and 480 horizontal fuel channels, each comprising a zirconium-niobium alloy pressure tube surrounded by a Zircaloy-2 calandria tube. These channels operate under demanding conditions: internal pressures near 11 MPa, coolant temperatures of approximately 310°C, and intense fast neutron flux. Over decades of service, pressure tubes can experience deformation such as diametral creep and sag, as well as degradation mechanisms including delayed hydride cracking, crevice corrosion at rolled-joint locations, and fretting wear from fuel bundle vibration. Manual access to inspect these critical components is severely constrained by several factors: high radiation fields that can exceed 1 Sv/h in the reactor vault, limited physical space within the fuel channel annulus, the sheer number of components requiring examination, and the need to minimize outage duration for economic reasons.
Advanced robotics have therefore transitioned from being a convenient alternative to an absolute necessity for lifecycle management. These robotic systems enable precise, repeatable, and comprehensive inspection and maintenance interventions while dramatically reducing personnel radiation dose and shortening planned outage periods. The economic case is compelling — every day of unplanned outage can cost a large CANDU station several million dollars in replacement power costs, making the speed and reliability that robotics bring a critical business imperative. With aging infrastructure across the global CANDU fleet, the role of robotics will only grow as operators seek to extend operating licenses beyond the original 30-year design life.
Why Robotics Are Critical for Safe and Cost-Effective Operations
Human entry into the reactor vault region or the fuel channel annulus is often completely impractical or unsafe. The shutdown radiation environment within the vault can reach 0.5 to 1 Sv/h in localized areas, which would limit a worker wearing full protective equipment to a few minutes per year under ALARA dose limits. Even with the use of temporary shielding tents and specialized protective gear, the As Low As Reasonably Achievable (ALARA) principle demands that utilities pursue every feasible dose reduction measure. Robotics satisfy this requirement comprehensively by performing visual surveillance, non-destructive testing, and light repair tasks from a safe distance, with the operator typically located in a low-dose area such as a control trailer outside the reactor building.
Beyond safety, robotics deliver a level of consistency and data quality that human technicians cannot match. A robotic crawler equipped with a calibrated ultrasonic transducer, operating under computer control, will gather flaw detection data with sub-millimeter repeatability across hundreds of fuel channels. This consistency is vital for establishing reliable degradation trends, which inform probabilistic safety assessments and long-term fitness-for-service evaluations. When a human technician works in a high-dose environment wearing multiple layers of protective clothing, with restricted mobility and a respirator limiting stamina, the quality and speed of work inevitably decline. Robots do not fatigue, their attention does not wander, and their sensors maintain precise calibration over extended shifts. Furthermore, the data collected by robotic systems is inherently digital and traceable, enabling comprehensive quality assurance records that satisfy regulatory scrutiny from bodies such as the Canadian Nuclear Safety Commission (CNSC).
Another critical advantage is the ability to operate in environments where human access is physically impossible. The annular gap between a pressure tube and its surrounding calandria tube is typically only a few millimeters wide — far too narrow for any human hand or conventional tool. Yet robotic inspection probes routinely navigate these confined spaces, carrying sensors that measure wall thickness, detect cracks, and assess surface condition with remarkable precision. This capability alone has transformed the approach to pressure tube lifecycle management, allowing operators to detect degradation at the earliest possible stage and plan remedial actions well before any safety margin is compromised.
Core Functions Robots Perform in Reactor Maintenance
The robotics fleet deployed across the worldwide CANDU fleet handles a diverse spectrum of specialized tasks that collectively form the backbone of condition-based maintenance programs. These functions extend well beyond straightforward visual checks, and they continue to evolve with each inspection and refurbishment campaign. The following list details the primary roles these systems fulfill, each supported by decades of operational experience and continuous technological refinement.
- Volumetric non-destructive examination: Robots carry phased-array ultrasonic probes and eddy current sensor arrays to inspect pressure tube walls. They detect and characterize thickness loss from corrosion, crack-like flaws from delayed hydride cracking, and hydride blistering at contact points with the calandria tube. These examinations must resolve flaws smaller than 0.5 mm in length while operating in a confined, curved geometry. Advanced signal processing algorithms now allow automated classification of flaw types in real time, reducing the need for offline expert analysis.
- Dimensional metrology: Laser profiling tools and optical triangulation sensors attached to robotic delivery systems measure pressure tube sag, ovality, diametral creep, and gap changes between the pressure tube and surrounding calandria tube. These measurements track time-dependent deformation that directly feeds engineering fitness-for-service assessments and determines the safe operating life of each channel. The precision required is extreme: dimensional changes as small as 0.1 mm can significantly affect the structural integrity assessment of a pressure tube.
- Remote visual testing: High-resolution cameras with programmable lighting and pan-tilt-zoom capabilities document surface conditions, fretting wear marks at spacer bundle locations, corrosion deposits on end fittings, and general cleanliness of the fuel channel inner surface. Modern camera systems capture 4K video and high-resolution still images that are automatically stitched into panoramic views of each channel, allowing engineers to inspect every square millimeter without entering the radiation zone.
- Material sampling and debris retrieval: Specialized end-effectors can capture small metallic debris, scrape samples from suspect locations, or collect filter deposits for off-site laboratory analysis. This enables root cause investigations without spreading contamination or requiring complex manual handling. Recent advancements include grippers with integrated force sensors that can distinguish between loose debris and tightly adhered deposits, ensuring that sampling does not damage the underlying pressure tube surface.
- Reactor face cleaning and decontamination: Over decades of operation, activated corrosion products and particulate matter, known as crud, accumulate on fuel channel end fittings and the reactor face. Robotic cleaning tools use pressurized water jets, mechanical brushing, and vacuum recovery to remove these deposits, restoring cooling flow paths and reducing radiation fields that affect subsequent maintenance work. Effective decontamination can reduce general area dose rates by 30 to 50 percent, significantly improving working conditions for all subsequent outage activities.
- Fuel handling and refurbishment support: During major refurbishment projects, robotic arms assist with guide tube alignment, flange bolt torqueing, and the careful handling of highly irradiated components during removal and replacement. These tasks demand positional accuracy measured in tenths of millimeters. The latest generation of refurbishment robots incorporates vision-guided alignment systems that automatically locate bolt holes and adjust tool positioning without requiring manual jigging.
- Weld inspection and repair: Robotic systems equipped with specialized ultrasonic and radiographic sensors inspect critical welds on feeder pipes, headers, and steam generator components. When defects are identified, the same robot can deploy orbital welding heads to perform precision repairs, all while maintaining continuous monitoring of weld parameters to ensure code compliance.
- Calandria tube inspection: The gap between pressure tubes and calandria tubes must be maintained to ensure adequate cooling and to prevent contact that could lead to hydride blistering. Robotic probes specifically designed for this annular space measure gap dimensions and inspect the outer surface of the pressure tube and inner surface of the calandria tube for signs of wear, corrosion, or fretting damage.
Types of Robotic Units Operating Inside CANDU Stations
The robotic fleet inside a modern CANDU station is not a single machine but a carefully integrated suite of purpose-built tools, each with a distinct locomotion method, sensor payload, and control interface tailored to the unique geometry and environmental challenges of the reactor core. Understanding the capabilities of each type is essential for appreciating how comprehensive the robotic inspection and maintenance ecosystem has become.
Fuel Channel Inspection Crawlers
These are the workhorses of CANDU lifecycle management, and they have seen continuous refinement over four decades. Crawlers travel inside an empty pressure tube after the fuel bundles have been removed, or along the outer surface of the fuel channel between the calandria tubesheet and the channel closure. They carry modular inspection heads that can rotate 360 degrees to scan the full circumference of the pressure tube wall. The drive mechanism typically uses magnetic wheels to provide adhesion to the zirconium alloy surface, or spring-loaded friction pads that press against the tube walls for traction. Advanced units, such as the AURORA system used extensively at Ontario Power Generation's Darlington and Bruce Power's Bruce complex, combine ultrasonic, eddy current, and visual sensors on a single deployable platform that can be swapped between channels in under an hour. Operators control the crawler from a low-dose control station while it streams live telemetry and inspection data for real-time engineering evaluation. A typical inspection campaign requires the crawler to operate continuously for 10 to 12 hours per channel, surviving cumulative gamma doses that can degrade conventional electronics and polymer components. Radiation-hardened electronics, tungsten shielding, and optical signal transmission are standard design features. The latest crawlers also incorporate onboard data storage and processing capabilities, allowing them to operate even during brief communication interruptions.
Articulated Manipulator Arms
Mounted on mobile bases or fixed to the reactor face scaffolding, these six-axis industrial manipulators perform small-scale mechanical work with high precision. They are deployed for tasks such as removing bolted closures on fueling machine ports, cutting and capping instrument tubes, positioning repair sleeves over defective pressure tube sections, and replacing gaskets on heat exchanger bonnets. The arms are typically teleoperated with force feedback technology that allows the technician to sense the torque applied to a fastener, even though the manipulator is several meters away behind a heavy shielding wall. IAEA technical reports on robotics in nuclear installations confirm that modern master-slave manipulators achieve positional accuracy under 0.3 mm, which is essential when working near brittle, irradiated components that could fail catastrophically if mishandled. These arms also incorporate collision avoidance systems and redundant position encoders to prevent inadvertent contact with adjacent structures. Recent developments include force-torque sensors at the wrist that provide haptic feedback to the operator, enabling delicate tasks such as threading a repair sleeve onto a pressure tube end fitting without damaging the sealing surfaces.
Miniature Submersible ROVs for Moderator System Work
The heavy water moderator within the calandria vessel is itself a complex thermal-hydraulic system that occasionally requires internal inspection following upset events or during aging management programs. Small remotely operated vehicles (ROVs) with neutrally buoyant propulsion systems can navigate the annulus region between the calandria wall and the fuel channel assemblies while completely submerged in heavy water. These submersibles are equipped with radiation-tolerant cameras, LED lighting arrays, and ultrasonic thickness gauges. They map corrosion patterns on the calandria wall, inspect the integrity of horizontal flux detector housings, verify the position and operability of liquid injection shutdown system nozzles, and examine gasket surfaces on the calandria closures. The ROVs must be extremely compact — some are less than 100 mm in diameter — to enter through the available access ports that are only slightly larger than a human fist, without requiring a large-scale moderator drain operation that would extend the outage by days. The propulsion system uses small ducted thrusters that generate minimal turbulence to avoid disturbing settled particulate matter that could obscure the operator's view. Some advanced ROVs are also equipped with sonar imaging systems that can provide situational awareness even in murky water conditions.
Drone-Style Aerial and Hybrid Robots for Vault Surveys
Small multi-rotor drones equipped with protective cages have found a valuable niche in general area surveys during maintenance outages. They can rapidly scan concrete vault walls for spalling, corrosion, and cracking; inspect overhead crane rails and bridge structures for wear and alignment; and document the condition of pipe hangers, cable trays, and ventilation ductwork in areas that are difficult or dangerous to access. The absence of physical tracks or rail systems allows them to reach elevated points without the need for scaffold construction, which saves days of outage schedule and reduces worker dose. Collision-tolerant carbon fiber cages protect the rotors from occasional contact with structural elements, and the drones stream 4K video to engineers who can review footage live and instantly flag anomalies. Their use at several CANDU stations has significantly cut the time required for vault walk-downs, which historically required a team of workers in full protective suits with supplied air, working in rotation to manage dose. The drones are also being equipped with thermal imaging cameras to detect hot spots in electrical equipment and insulation deficiencies on steam piping, adding an extra dimension to condition monitoring.
Pipe Crawling and Conduit Inspection Robots
Beyond the reactor core itself, CANDU stations contain hundreds of kilometers of pipes, conduits, and ducts that require periodic inspection. Specialized pipe-crawling robots, often with articulated sections that can navigate 90-degree bends and vertical rises, inspect feeder pipes, cooling water lines, and ventilation ducts for wall thinning, blockage, and structural damage. These robots use magnetic or pneumatic traction systems and carry cameras, ultrasonic sensors, and laser profilometers. In feeder pipes, which are particularly susceptible to flow-accelerated corrosion, these crawlers have proven invaluable for detecting wall thinning long before it reaches a critical level. The ability to inspect pipes from the inside without cutting them open or removing insulation has saved many CANDU stations significant maintenance costs and reduced outage durations.
Real-World Deployments and Industry Collaboration
Canada's nuclear operators and research organizations have actively driven robotics innovation through long-standing partnerships with universities, federal laboratories, and specialized engineering firms. At the Chalk River Laboratories in Ontario, Canadian Nuclear Laboratories (CNL) has developed a series of prototype systems including the ARTEMIS platform, a modular crawler whose drive wheels and sensor sleds can be reconfigured for different fuel channel diameters and inspection requirements. CNL's reactor safety test facilities include full-scale mock-ups of the fuel channel assembly where robots are tested under simulated radiation, temperature, and hydraulic conditions before deployment in live reactors. These facilities allow for rigorous validation and operator training without the constraints of an operating plant schedule. The test beds are instrumented with hundreds of sensors that provide detailed performance data, enabling engineers to fine-tune control algorithms and identify potential failure modes before field deployment.
Bruce Power operates one of the largest CANDU fleets in the world, with eight units at its site on Lake Huron. The company has been a leader in adopting advanced robotics for its ongoing refurbishment program, which will extend the operating life of its units well beyond 2060. During the Bruce Unit 6 refurbishment, robotic systems were used to inspect over 480 fuel channels, perform dimensional measurements on all calandria tubes, and assist with the removal and replacement of major components. The lessons learned from this project are now being applied to subsequent unit refurbishments, with continuous improvements in robot reliability, inspection speed, and data analysis automation. Bruce Power has also invested in developing a dedicated robotics training center where operators can practice on full-scale mock-ups under realistic conditions, ensuring that the transition from training to field deployment is seamless.
The CANDU Owners Group (COG) funds joint research projects that pool resources and share results among utilities in Canada, Korea, Romania, and Argentina. One recent COG-led initiative delivered a next-generation pressure tube inspection head that incorporates real-time artificial intelligence capable of flagging potential crack-like anomalies without requiring an off-site expert to review every ultrasonic scan individually. This machine learning system, trained on thousands of labeled flaw signatures from previous inspections, reduced the per-channel inspection time by 40% during a trial at the Point Lepreau Generating Station in New Brunswick. Such collaborations demonstrate the power of shared investment in solving common technical challenges across the international CANDU community. Another COG project focused on developing a common data format for inspection results, enabling direct comparison of data from different robots and sensor types without requiring custom data conversion for each campaign. This standardization effort has significantly improved the efficiency of data analysis and trend tracking across the fleet.
Overcoming Radiation and Reliability Hurdles
Designing robots that can survive and operate reliably inside a CANDU reactor presents formidable materials science and electronics engineering challenges. The total ionizing dose can exceed 1 MGy over the course of a single multi-channel inspection campaign. This level of radiation exposure causes embrittlement and molecular scission in polymers, darkening and loss of transmission in optical fibers, parametric drift and eventual failure in semiconductor devices, and single-event upsets in digital logic circuits. Engineers combat these effects through a combination of strategies: selecting radiation-hardened electronic components where available, encapsulating sensitive electronics in high-density tungsten or lead shielding, using long optical fiber bundles to relay sensor data to control equipment located safely outside the biological shield, and designing modular payloads that can be quickly swapped if a sensor degrades mid-campaign.
Redundancy is built into every critical subsystem. Inspection crawlers typically feature dual drive motors, parallel communication channels using both wired and wireless links, and backup battery power feeds that allow safe retrieval if the primary power source fails. Even with these precautions, field experience shows that the most common failure modes in CANDU inspection robotics are not radiation effects but rather mechanical issues: connector corrosion from prolonged exposure to humid, warm environments, cable fatigue from repeated bending around small radius rollers, and seal failures that allow moisture ingress into electronic compartments. Lessons learned from each deployment campaign are systematically documented and fed back into a continuous design improvement cycle managed by organizations such as the Electric Power Research Institute (EPRI) and the CANDU Owners Group. EPRI has published comprehensive guidelines on radiation-hardened electronics and robotic reliability that are widely referenced in the industry. Regular preventive maintenance and pre-deployment testing protocols have been developed to catch potential failures before robots are inserted into the reactor, reducing the risk of in-service failures that could delay an outage.
Another critical hurdle is the thermal environment. While the reactor is shut down for inspection, residual heat keeps the fuel channel region at temperatures that can exceed 40°C, with localized hot spots near end fittings and closure plugs. Robotics must be designed to operate reliably at these temperatures, which can affect lubricants, battery performance, and electronic cooling. Active thermal management systems using heat pipes, liquid cooling loops, and phase-change materials are incorporated into some high-duty-cycle systems to maintain component temperatures within safe operating ranges.
Data Integration and the Digital Twin Paradigm
Modern CANDU inspection robotics generate terabytes of raw sensor data per year across a multi-unit station. The real value of this data lies not in isolated examination reports but in its integration into a living digital twin of the reactor core. When an inspection crawler completes a scan of pressure tube D12, the resulting wall-thickness map is automatically registered with the as-built three-dimensional geometric model of that channel, and historical degradation trends are plotted against inspection data from the previous outage — whether that was five or ten years earlier. Maintenance planners can then run predictive models that forecast the specific operating year and outage window when a particular tube will approach its fitness-for-service limit for parameters such as diametral creep or gap closure with the calandria tube.
The digital twin paradigm extends beyond individual components to encompass the entire reactor system. Thermal-hydraulic models, neutron flux distributions, and materials degradation models are integrated with the inspection data to create a holistic representation of reactor state. This allows engineers to ask "what-if" questions: What would happen if a particular pressure tube experienced accelerated creep due to a manufacturing anomaly? How would that affect the neighboring channels? What is the optimal timing for a channel replacement to minimize outage impact while maintaining safety margins? The answers to these questions inform not only maintenance planning but also long-term investment decisions and regulatory submissions.
EPRI has published comprehensive implementation guidelines for digital twin technology in the nuclear industry, which many CANDU operators have adopted and adapted. These frameworks integrate robotic data streams with enterprise asset management systems, scheduling software, and probabilistic safety assessment tools. The result is a transformation of inspection from a costly, periodic compliance activity mandated by regulatory requirements into a continuous, predictive engineering function that directly informs capital investment decisions and life optimization strategies. Operators can prioritize maintenance on the channels that most need it, defer work on those that have decades of remaining safe life, and plan refurbishment outages with confidence backed by high-resolution data. The economic benefits are substantial: optimized maintenance scheduling alone has been estimated to save multi-unit CANDU stations tens of millions of dollars per year by avoiding unnecessary work and extending the intervals between major refurbishments.
The Path Ahead: Autonomy, AI, and Telerobotics
The next frontier for CANDU robotics is the progression from teleoperation to greater autonomous capability. Current systems, while highly capable, remain largely under direct human control, requiring a skilled technician to direct every movement and interpret every signal. Research programs underway at Canadian universities and industrial laboratories are working to equip robots with simultaneous localization and mapping (SLAM) algorithms specifically adapted for the highly metallic, reflective, and geometrically complex environment inside a calandria. Deep learning models trained on large libraries of labeled ultrasonic and eddy current scans are being taught to detect and classify flaw types in real time, distinguishing between benign manufacturing features, harmless fretting wear, and active crack growth that requires immediate engineering evaluation. These AI systems are also being trained to recognize when sensor data quality is degraded — for example, due to coupling loss or electrical noise — and to alert the operator or automatically adjust inspection parameters to compensate.
The long-term goal is a robotic inspector that can be given a high-level mission — such as "Scan all fuel channels at elevation 0 to 8, flag any wall loss exceeding 15% of nominal thickness, and report any gaps below 1.5 mm" — and then execute that mission independently, navigating around obstacles, adjusting inspection parameters based on real-time findings, and calling for human intervention only when it encounters an unexpected or ambiguous condition. Coupled with low-latency, high-bandwidth communication networks such as the 5G systems being installed at refurbished CANDU stations, this could eventually allow remote operations from a centralized expert hub serving multiple plants, reducing the need for highly specialized inspection crews to travel to each site individually. This concept of "robotics-as-a-service" is already being explored by several nuclear service providers, who envision a future where a single team of experts can oversee inspection campaigns at multiple stations simultaneously, with local support staff handling robot deployment and retrieval.
Another area of active development is life extension robotics designed for large-scale construction tasks during major refurbishment projects. As many CANDU units approach or execute mid-life refurbishment, robots are being designed for heavy work: removing and replacing entire fuel channel assemblies, precision cutting of pressure tubes that are under residual tension, automated orbital welding of new calandria tube inserts, and handling of massive shield plugs and channel closures. These heavy robots, some weighing several tonnes, will operate from the reactor face for months at a time, dramatically shortening the critical path schedule and significantly lowering total project costs by reducing the need for manual work in high-dose areas. Prototype systems have already demonstrated the ability to remove and replace a fuel channel closure plug in under four hours, compared to the 12 to 16 hours typically required for manual operations with extensive shielding and dose controls.
Collaborative robots, or cobots, designed to work alongside human technicians in low-dose areas are also gaining traction. These lighter, more flexible robots assist with tasks such as tool delivery, parts handling, and repetitive measurements, freeing human workers to focus on higher-value activities. Cobots are equipped with sophisticated safety features, including force-limited joints and vision-based collision detection, that allow them to operate safely in close proximity to personnel without the need for physical barriers. Their adoption is accelerating as plant operators recognize the productivity gains and dose reduction benefits they offer even in areas where manual work is still feasible.
The integration of augmented reality (AR) with robotic control systems is another promising trend. Using AR headsets, operators can see virtual overlays of inspection data, robot status, and plant diagrams directly in their field of view, improving situational awareness and reducing the cognitive load associated with interpreting complex telemetry displays. Some CANDU operators are now testing AR systems that allow a remote expert to "see" what the robot's camera sees and provide guidance to the on-site operator by drawing annotations directly onto the live video feed. This capability is especially valuable during complex interventions where specialized knowledge is required but the expert cannot be physically present due to dose constraints or travel limitations.
Conclusion
Advanced robotics are no longer a niche experiment or a technology demonstration within the CANDU industry. They have become a foundational element of safe, reliable, and economically sustainable reactor operation. These systems shield workers from radiation exposure that would otherwise accumulate quickly, deliver engineering data of unmatched quality and consistency, and enable proactive, condition-based maintenance strategies that keep plants operating safely decades beyond their original design life. With each new inspection campaign, the continued integration of smarter sensors, advanced artificial intelligence, and more capable autonomous navigation brings the industry closer to a vision of truly predictive nuclear asset management. The sustained collaboration among utilities, research institutions, and specialized suppliers worldwide ensures that the robotic fleet serving CANDU technology will continue to evolve, meeting the growing demands of aging infrastructure with ingenuity, precision, and unwavering attention to safety. As the global CANDU fleet ages and operators seek to maximize the return on their nuclear investments, the role of advanced robotics will only expand, driving further innovation in sensor technology, data analytics, and autonomous operations. The future of CANDU reactor inspection and maintenance is robotic, and that future is already unfolding with each outage, each inspection campaign, and each new generation of purpose-built machines designed to keep these vital power plants running safely and efficiently for decades to come. The Canadian Nuclear Association continues to highlight the importance of robotic innovation in maintaining the competitiveness and safety of the country's nuclear fleet.