Asia, Australasia

Nuclear Energy: The perfect alignment

Issue 4 and Volume 15.

Korea Hydro & Nuclear Power has employed a high precision photogrammetry technique to successfully ensure the correct alignment of nuclear fuel rod assemblies within a nuclear reactor vessel following a scheduled refueling outage.

Kenneth Edmundson, Geodetic Systems, Inc., USA, Giuseppe Ganci, Gancell Pty. Ltd, Australia & Park, Chan-Hong, Visiontech, South Korea

Measuring nuclear fuel rod assemblies is a difficult task because the environment is hazardous. The assemblies lie 12 m underwater and are radioactive. Faced with this, Korea Hydro & Nuclear Power (KHNP) approached Visiontech of Seoul, South Korea for assistance with the measurement of fuel rod assemblies in a nuclear reactor vessel.

The Yongwangg facility is a Korean Standard Nuclear Power Plant (KSNP). Refueling outages typically take place at intervals of 12-24 months. Refueling involves removing a number of spent fuel assemblies and loading an equal number of fresh assemblies. It is critical that the fuel assemblies are in their proper location when the top of the reactor is lowered into place. Any that fall outside of the allowable tolerance may result in interference and potential damage to the assembly, the top of the reactor vessel, or both.

This situation could lead to an unplanned, and very costly extended outage. Moreover, after refueling, the nuclear and thermal characteristics and the safety and operating parameters of the core will be different from those prior to refueling. For these reasons, it is necessary to establish the relative positions of the fuel assemblies. While basic methods have been used to accomplish this task, the need for measurement advancements are greatly needed to establish an accurate, efficient and automated process.

Photogrammetry, which is based on digital camera technology and computer vision techniques, was the chosen tool to solve this problem. Visiontech worked together with technology partners Geodetic Systems, Inc. (GSI) of the USA and Gancell of Australia to develop the necessary software, hardware, and procedures for this undertaking. GSI developed the V-STARS photogrammetric line of 3D coordinate measurement systems.

Metrology is the science of measurement. Engineers, manufacturers, scientists, and quality control technicians utilize various measurement technologies for their own unique disciplines. Photogrammetry is a three-dimensional coordinate measuring technique that uses photographs as the fundamental medium for metrology. The V-STARS system is comprised of a high resolution digital INCA3 camera, a laptop or desktop computer, and software for measuring and processing images. This technology is well known for its precision and flexibility for usage in extreme conditions and environments, compared to other metrology devices that require pristine laboratory conditions.


The INCA3 camera is lowered into the nuclear reactor vessel
Click here to enlarge image

The extreme conditions present both physical and mathematical challenges to the metrologist. To work underwater, the camera must be housed in a watertight canister. To work online, power and communications must be provided from the surface to the camera through the canister. Imaging underwater complicates the traditional collinearity model of photogrammetric reconstruction as light must travel through water, air and last but not least, the canister window.

Typically in industrial photogrammetry, artificial targets of either white or retro-reflective material are used to signalize points of interest on the object to be measured. In the KSNP case, targets could not be placed on the fuel assemblies so it was necessary to detect and measure the natural features of the assembly itself.

KSNP reactor core

The KSNP utilizes a pressurized water reactor. The reactor core consists of 177 fuel assemblies, which are positioned to approximate a right circular cylinder with a diameter of 3m. Each individual fuel assembly is 4.5m long, and has 236 fuel rod positions in a 16 by 16 array. Each fuel assembly has five guide tubes welded to spacer grids and is enclosed by end fittings at the top and bottom. The top surfaces (or hold-down plates) of the upper end fittings (UEFs) were the designated features to be measured. They were assumed to lie in a plane approximately 12m below the surface of the water in the reactor vessel.


Section of photographic image showing measured ellipses.
Click here to enlarge image

The UEF plate is a single-piece construction with five connected rings for the guide tubes. The central guide ring is attached to four surrounding rings positioned at 45°, 135°, 225° and 315°. The central and surrounding rings have outer diameters of 50 mm and 62 mm, respectively. The centre-to-centre distance between adjacent outer rings is 102.87 mm.

It would be necessary to provide measurements for the center points of the five rings of each UEF plate in the XY plane only. With 177 fuel assemblies, 885 points needed to be measured. The required point-to-point accuracy was stated at 0.76mm (one-sigma) in X and Y. A mock test measurement was completed in which six fuel assembly UEF plates were measured in air. The camera derived proper illumination from temporary underwater lights, positioned around the upper interior of the reactor vessel. Although rudimentary, the test results indicated the measurement requirements outlined by KSNP could be met. In February 2005, an expert team commenced work on a feasible solution.

Sizing up the problem

As data acquisition would be performed underwater, the photogrammetric camera would have to be remotely controlled. The INCA3 camera receives power and communications via a combined power/ethernet cable. Functions such as strobe power, shutter speed and camera triggering are controlled through the V-STARS software installed on the host computer. When an image is captured, it is compressed and transferred by Ethernet to the computer where it can be examined, measured or re-acquired if necessary. The camera can also be powered up or down remotely.

The development of an underwater canister suitable to house the INCA3 camera was a major priority. The main canister body was constructed of a solid piece of 16 mm aluminum rolled and welded to a prescribed diameter based on the size of the INCA3 camera. The window was made from a piece of non-browning fused silica selected for its radiation resistance. The canister’s rear plate included a mounting bracket with a Manfrotto quick release mechanism to hold the camera in place. Both end-plates featured double rubber O-rings for redundancy to ensure a watertight canister. As an added precaution, a leak detector alarm was installed in the unlikely event that the housing seal was breached. Power and communications were provided to the camera through the canister via a standard 30 m cable, which was wrapped in a polyurethane jacket to make it suitable for use in an underwater nuclear environment.

The next challenge was moving the camera underwater. The optimal network design consisted of a minimum of nine camera stations arranged in a regular grid three m above the plane of the UEF plates. Two images would be acquired at each camera station with a 90° roll between them about the camera axis. This setup would also achieve camera self-calibration. To position the camera at the correct height above the UEF plates, the canister was attached at the end of a series of interlocking poles and then lowered into the reactor vessel. A specialized underwater pan/tilt device with the necessary weight rating was selected, which would allow the camera to be rotated about its axis, as well as tilted up to nearly 90° from the vertical.

With the underwater canister completed just in time for the next refueling outage, an initial test date was set in 2005. The primary objective was to collect sufficient test imagery to facilitate software development for the measurement of the rings on each UEF. Using the camera and canister systems, test images were collected to establish the exposure and shutter speed needed to image the natural features of the UEF plates. Other factors including camera stability, camera temperature, and image contrast were also noted during the session. The development of a special software programme for natural feature detection and measurement was shaping up.

Getting down to business

Armed with new imaging technology, the metrology team executed the main data acquisition project in March 2006. A total of 52 images were collected, and 46 were of sufficient quality for both measurement and analysis.

For the purpose of experimentation, images were collected at more camera stations than were called for in the optimal network design. In this session, all images were collected prior to measurement. Measurement and analysis were made outside of the nuclear containment area.

As the procedure is refined, images will be measured immediately upon acquisition, with the final bundle adjustment performed as soon as the last image is captured. The nominal coordinates of all ring centres were utilized to automate the measurement.

In each image a minimum of four rings were measured manually and the remainder by resection driveback. Less than ten minutes were required to measure all images. The adjusted coordinates were compared to the design coordinates of the rings. As expected, the vectors from the cluster of five points on each UEF have consistent magnitude and direction.

The accuracy of a photogrammetric network can be independently assessed via the measurement of some form of external standard. It was decided to utilize the accurately machined UEF hold-down plates themselves as a measure of accuracy. The four computed distances between the adjusted centers of adjacent outer rings were compared to the nominal distance of 102.87 mm. This comparison was made for each of the 177 UEF plates for a total of 708 distances. The results are shown in the table on the left.

Click here to enlarge image

The results described clearly demonstrate the suitability of computer vision techniques such as edge detection for the measurement of natural features in close-range, industrial photogrammetry. This approach shows significant potential for the measurement of not only circular objects and holes, but for slots and square shaped cutouts, and features of irregular shape. This measurement capability is clearly valuable in many areas of manufacturing, engineering, and quality control where the measurement of such features is critical, such as the nuclear and automotive industries.