Image analysis as a tool for improved use of the Digital Cherenkov Viewing Device for inspection of irradiated PWR fuel assemblies.

• The Digital Cherenkov Viewing Device (DCVD) is a tool used to measure the Cherenkov light emitted from irradiated nuclear fuel assemblies stored in water pools. It has been approved by the IAEA for attended gross defect verification, as well as for partial defect verification, where a fraction of the fuel material has been diverted. In this report, we have investigated the current procedures for recording images with the DCVD, and have looked into ways to improve these procedures. Using three different image sets of PWR fuel assemblies, we have analysed what information and results can be obtained using image analysis techniques. We have investigated several error sources that distort the images, and have shown how these errors affect the images. We have also described some of the errors mathematically, and have discussed how these error sources may be compensated for, if the character and magnitude of the errors are known. Resulting from our investigations are a few suggestions on how to improve the procedures and consequently the quality of the images recorded with the DCVD as well as suggestions on how to improve the analysis of collected images. Specifically, a few improvements that should be looked into in the short term are: • Images should be recorded with the fuel assembly perfectly centered in the image, and preferably without any tilt of the DCVD relative to the fuel in order to obtain accurate measurements of the light intensity. Image analysis procedures that may aid the alignment are presented. • To compensate for the distorting effect of the water surface and possible turbulence in the water, several images with short exposure time should be captured rather than one image with long exposure time. Using image analysis procedures, it is possible to sum the images resulting in a final image with less distortions and improved quality. • A reference image should be used to estimate device-related distortions, so that these distortions are compensated for. Ideally, this procedure can also be used to calibrate individual pixels. • The background should be carefully taken into account in order to separate the background level from diffuse signal components, allowing for the background to be subtracted. Accordingly, each measurement campaign should be accompanied by at least one background measurement, recorded from a section in the storage pool where no fuel assemblies are present. Furthermore, the background level should be determined from a larger region in the image and not from one individual pixel, as is currently done. • A database of measurements should be set up, containing DCVD images, information about the applied DCVD settings and the conditions that the DCVD was used in. Any partial defect verification procedure at any time could then be tested against as much data as possible. Accordingly, a database can aid in evaluating and improving partial defect verification methods using DCVD image analysis. Based on the findings and discussions in this report, some long-term improvements are also suggested.

Ämnesord

Natural Sciences  (hsv)

Physical Sciences  (hsv)

Naturvetenskap  (hsv)

Fysik  (hsv)

Nyckelord

DCVD

safeguards

spent fuel verification

Cherenkov light

image analysis