Radiation monitoring/ALARA | Robotics

As Low As a Robot’s Arm

20 June 2011



A snake-arm robot has been used to perform outage inspections in a Candu reactor. The robot offers new inspection tools and reduces worker dose. By Rob Buckingham


There are 438 civil nuclear reactors in the world across 35 countries. Of these reactors 123 units are over 30 years old (measured from first criticality) and are located in 20 countries. A further 12 units in UK, USA, India and Switzerland are more than 40 years old [1]. The climate for extending the life of these high-value assets has changed considerably in the past decade due to concerns over the security of supply, emissions and meeting expected demand. Of the 104 US power plants, 59 have been granted extended licenses to operate for 60 years. A further 19 are applying for license extensions [2].

Safe, profitable operation of these plants is driving the replacement of critical components, often as part of power upgrades or refurbishment of other systems. It is reasonable to expect ageing plant to require increasing levels of inspection and maintenance, particularly for safety-critical systems. If such systems were not in a nuclear environment, many of the tasks would be straightforward to perform. High levels of radiation can convert even the simplest task into a complex, expensive and time consuming problem.

Remote handling has always been part of the nuclear toolkit, but historically, the main driver for remote handling has been to conduct tasks in areas where dose levels prohibit manual entry. New requirements to establish the current state of plant and then monitor the same plant for a further 20, 30 or 40 years beyond the original design life will generate requirements for remotely-operated solutions. This may be driven by unacceptable dose levels and the drive for ALARA, or may simply be due to the challenges posed by the restrictive geometry of the plant.

The assignment of a dollar value to operator received dose changes the economics of nuclear inspection and maintenance. Arguably, the creation of a virtual market for received dose, with the price effectively set and paid for by wider society, better reflects the real cost of nuclear power generation. If this figure is set appropriately it can be used to reduce nuclear worker dose and stimulate innovation by enabling the investment to be justified to stakeholders.

In May 2010 a new type of robot was deployed at Pickering NPP, Ontario. The main commercial driver was dose reduction. The potential for dose saving using robotics is easy to calculate. Set-up and tear-down took less than 90 minutes. In this time the operators spent less than 30 minutes in the raised-dose area ensuring the equipment was operational. If the personnel dose received is 1 unit per hour and the robot requires two operators for set up and tear down, then the received dose is 1 unit. This means dose will be reduced by 9 units if (a) the robot does work that would take 10 man-hours or 19 units if (b) the robot does 20 man-hours of work. If the dose has a value of $X/unit and there are three robot deployments per year for five years then the ‘value’ of using the robot is: 9 * x * 3 * 5 = 135X for (a) and 185X for (b).

Assigning a dollar value to X is not easy. X could include the cost of recruiting and training of new staff each year, each of whom might only work for a few days before being burnt out. X could also include the value of an owner being able to declare in the annual report that nuclear worker dose has been reduced by investing in robotics. X could include increased profit due to decreased outage times, if the robot gets the job done faster than the manual equivalent.

There are other components of X that are more difficult to quantify. It is difficult to work effectively wearing an air suit in a confined space at elevated temperatures in a radiation environment. Well-trained, experienced operators perform remarkably well. Humans are incredibly resourceful and adaptable and can make complex tasks look simple. But what if they become burnt out before they become experts? It is not easy to quantify this cost but if the task is to gather inspection data that could prevent an incident then the quality of the data collected is important.

The robot

In this particular case, the robot in question was a snake-arm model called Safire [3], developed by OC Robotics and Ontario Power Generation. It was deployed at one of the six 540MWe CANDU reactors at the Ontario Power Generation Pickering site in Ontario, Canada. CANDU reactors have horizontal fuel channels. The primary coolant circuit uses D2O to remove heat from the fuel bundles through feeder pipes that connect to either end of the fuel channels. The pressurized heavy water enters at the core at ~250°C and exits at ~290°C. The outflow feeder pipes converge on header tanks within the Upper Feeder Cabinets (UFC), Figure 1, before rising into the steam generators.

The integrity of the thick-walled stainless steel feeder pipes is essential for safe operation. One requirement of any inspection is to ensure that no leaks occur by ensuring that pipe wall thickness remains above predefined safe limits. If a pipe is found that is near the limit then it has to be replaced (at considerable cost). Thinning has been observed due to flow-assisted corrosion at pipe bends and also adjacent to welds. Pipe thinning has also been observed externally, caused by fretting between pipes and also between pipes and pipe supports (hangers). One preventative measure is to place sacrificial protective chafe shields in areas of concern. These then need to be inspected to make sure they have not moved or become damaged themselves.

The potential for fretting necessitates direct examination to establish the presence and extent of any damage. Inspections rely on making visual and other non-destructive assessments. These are conducted during outages and require people, wearing appropriate protective clothing, to enter the vault with measurement equipment and conduct inspections as fast as possible in order to minimize the dose received.

SAFIRE was designed for the UFCs in CANDU reactors. The UFC is a challenging work environment combining radiation, elevated temperatures, and confined spaces. Access to the UFC is via a narrow staircase adjacent to the reactor face inside the reactor vault. Inside the UFC there are three suspended catwalks that run between the 440 feeders. To reach the hangers an inspector must lie on the catwalk and reach down under the catwalk with a camera on a stick and take pictures. There are some areas that cannot be viewed and it is also important to be able to revisit specific locations in order to generate trend data.

SAFIRE, (Figure 2), is a remotely controlled robot that is equipped with a 2.2m long, 12.5mm wide, 18 degrees-of-freedom arm that literally snakes under the catwalk and between the hangers (Figure 3), carrying cameras to take images of the pipe work. The arm is mounted on a vehicle which can be driven along the walkways allowing SAFIRE to view the complete cabinet without manual intervention. The arm can also reach above the walkway to inspect the header tanks and other systems within the UFC (Figure 4).

SAFIRE is controlled from within a trailer parked up to 500m from the UFC. The operators, who are highly experienced permanent inspection staff, are able to sit in the comfort of a trailer and drive SAFIRE remotely (Figure 5). In the trailer, they are able to think and talk without incurring a dose.

Early in the development of snake-arm robots a critical decision was made to use wire ropes (high-tech metal constructions used routinely in robotic surgery) to deliver mechanical power into the arm to control arm shape and stiffness. The consequences of this engineering decision are profound and worthy of some explanation. SAFIRE has three main elements: a tracked vehicle to drive along the catwalks; a motor pack that contains the motors and electronics that power the arm; and the arm itself. More than 95% of the weight of the system is in the vehicle and motor pack; this makes the system very stable. The arm itself is light and moves slowly, so it cannot damage the environment. Because the arm is light it can also be longer. Wire ropes help the arm to be well-damped, which means that the motion of the camera located at the tip of the arm is smooth. This is important for the operator driving the system over prolonged periods. Wire ropes also make the arm slightly compliant. This is used to advantage by allowing the arm to brush against components. Getting an arm stuck in the environment would also be somewhat embarrassing. Training can be used to limit driver error, but a further feature of the wire rope design is that by removing rope tension (completely or partially) the arm will go floppy and then it can be pulled out, like a standard video probe. In combination these features that have been purposefully designed into the system make both the risk and consequences of failure low.

The job

SAFIRE completed factory acceptance tests in December 2009. Site acceptance tests followed in February 2010. There then followed an intensive period of rehearsals using both proprietary simulation software (Figure 6), and real hardware within UFC mock-ups (Figures 3 and 4). Training and procedure development is a major component of outage preparation. This was enhanced by providing a virtual test environment using the actual control software but without hardware connected. This proprietary software was also be used as part of the design process to ensure that the snake-arm kinematics was compatible with a CAD model of the environment.

SAFIRE was deployed for the first time at Pickering during May 2010. Set-up time was 70 minutes from the time the operators carried the equipment into the vault until the system began taking live images under control of the same operators in the remote trailer. The inspections were conducted over a 12-hour period using two operators who received no further dose and enjoyed a relaxed lunch. Tear-down was completed in 30 minutes from entry into the vault to the equipment being packaged outside the vault for transfer to a contamination inspection area. The robot was designed with smooth flat surfaces and sealed units where possible for cleaning. Other elements that are likely to be contaminated, such as the vehicle tracks, were designed to be replaceable.

The operation of the system is intuitive: the basics can be learned in minutes. To increase data capture rates, two operators are used to fly the arm. One is responsible for motion and the other is responsible for gathering data. SAFIRE uses proprietary nose-following software which allows the operator to control the 18 degree-of-freedom arm with ease by just focusing on where the camera is moving.

SAFIRE has a number of cameras and lighting systems. Two identical bespoke compact PTZ cameras with matched lights are mounted on the vehicle and a fixed base station. These provide scene views and can also be used for inspections. The main inspection tool is a tip-mounted camera tool which contains three fixed focal length cameras – two pointing forward and one facing sideways. High power LED lights are used and matched to the focal length of the cameras (Figure 7). The tool is the size of a large matchbox. It includes a motor to rotate the tool around its longitudinal axis to enable different views to be obtained from the side camera. Video from the cameras is transferred over 500m of fibre to the operator console, which allows the video streams to be recorded, time stamped and annotated. Other variants of SAFIRE could include different tools, including UT probes.

Since the arm is stable, the system could review images and decide to re-take pictures or change the angle of view. Inspectors recorded images of parts of the reactor system that have not been seen since the reactor was built.

SAFIRE is designed to be a dose-tolerant workhorse, gathering data during every outage in order to build up a picture of the state of the feeders and track any changes. SAFIRE has demonstrated the expected benefits of replacing manual inspection with automated inspection. These include: rapid set-up and tear down and remote operation that enabled considerable operator dose reduction; high quality images acquired from new vantage points and video captured from a stable platform. Less tangible results included the ability of the most experienced inspectors to view and re-view plant condition in real time. In addition, there are opportunities to improve procedures and performance; in addition, SAFIRE’s value could be extended by adding more tools.

Extending nuclear plant life is now a key issue for nuclear operators. Inspection of plant condition will be an increasingly important element of assessing plant integrity as a part of license renewal. Snake-arm robots create the opportunity to deploy new and existing inspection and repair technologies in dose-efficient ways.

Snake-arm robot technology will enable plant life to be extended both by conducting necessary repairs and by providing high quality data about the physical condition of the plant. Introducing new technology is a challenging business which requires perseverance and investment. SAFIRE demonstrates that these goals are reasonably achievable.


Author Info:

Rob Buckingham, co-founder, OC Robotics

Figure 1 - ALARA Figure 1 - ALARA
Fig. 2: SAFIRE tracked vehicle and motor pack Fig. 2: SAFIRE tracked vehicle and motor pack
Fig. 3: SAFIRE with arm deployed in a mockup environment Fig. 3: SAFIRE with arm deployed in a mockup environment
Fig. 4: SAFIRE deploying above the mock-up walkway Fig. 4: SAFIRE deploying above the mock-up walkway
Fig. 5: Operator console Fig. 5: Operator console
Fig. 6: Sample user display images of simulation used for rehearsals Fig. 6: Sample user display images of simulation used for rehearsals


Privacy Policy
We have updated our privacy policy. In the latest update it explains what cookies are and how we use them on our site. To learn more about cookies and their benefits, please view our privacy policy. Please be aware that parts of this site will not function correctly if you disable cookies. By continuing to use this site, you consent to our use of cookies in accordance with our privacy policy unless you have disabled them.