Automated quantitative inspection of nuclear assets and canisters

Student thesis: Doctoral Thesis


Structural integrity of nuclear assets and their associated infrastructure is key to national security, energy generation, and efficient deployment of capital. Furthermore, building nuclear infrastructure is inherently complex and it is common to seek extensions in asset and infrastructure lifetime as they age. Therefore, this thesis explores various Non-Destructive Evaluation (NDE) techniques to advance the inspection of nuclear canisters and their welds for three scenarios: 1) In-situ screening of legacy canisters sealed by Resistance Seam Welds (RSWs) while on a racking system with considerable accessibility constraints; 2) Ex-situ characterisation of legacy canisters sealed by RSWs; & 3) Inspection of fusion welds performed via Gas Tungsten Arc Welding (GTAW) for an upgraded canister of the future design at the point of manufacture. For the in-situ scenario a unique ultrasonic guided wave, a Feature Guided Wave (FGW), was exploited to screen the legacy RSW. For the RSW, four weld-guided wave modes were identified, and the fundamental flexural (F0) weld guided mode was down selected. Several Finite Element (FE) models explored applicable transduction strategies and documented reflection coefficients from transverse cracks. Multiple experiments were conducted on pseudo RSW structures as well as flat plate and cylindrical RSWs. For the pseudo weld experiments, it was shown that FGWs, like that of the analytical free plate counterparts, could easily be excited in flat plate-like features with traditional transduction techniques. For the flat plate RSW experiments, it was shown that reflections from 6.5 mm wide through wall and 1.00 mm deep defects could be observed with Signal to Noise Ratios (SNRs) of 16.33 dB and 8.21 dB respectively. Lastly, it was shown that for cylindrical canister like RSWs, reflections could also be observed from 1.00 mm deep defects with a SNR of 11.85 dB being reported. This work clearly showed the benefit of deploying such a system on-site at Sellafield as full circumferential screening of the RSW could be performed giving greater insight to the structural health of the canisters in-situ. To address the ex-situ inspection of the legacy canister, an automated robotic eddy current system was developed. The data rich platform allowed for a complete digital record to be established of the impedance data gathered and is well suited for further advancements in eddy current inversion to leverage in the future. The robotic deployment of the eddy current array was combined with force torque feedback and enabled major sources of noise, resulting from lift-off and wobble, to be reduced. Two different datasets were reported on. The first being eddy current scans of canister bodies with known stress corrosion cracks. All of the stress corrosion cracks were detected, and the resulting SNR of images generated from the impedance data was increased through post processing of the eddy current data. The second dataset is concerned with eddy current scans of RSWs. These scans make use of the aforementioned FGW, to localise defects and perform targeted raster scans in the area of concern. Basic inversion on the EDM notch width was shown to give results with 96.4% accuracy, and it was shown that time savings of up to ~95% could be realised by performing targeted eddy current raster scans. This work clearly shows the benefit of performing an ex-situ inspection in this manner due to the minimal levels of operator handling and time savings that can be realised on large production volumes like that at Sellafield. Lastly, inspection of the envisaged canister of the future design at the point of manufacture was explored as it would enable operational efficiencies and the ability to be able to create a Non-Destructive Evaluation (NDE) digital twin to monitor the structural health of the component over its lifetime. Technically, inspecting welded components at the point of manufacture is challenging due to the elevated temperature and resulting thermal gradients in the component introducing beam bending effects due to refraction and positional inaccuracies in the ultrasonic data. A novel thermal compensation strategy was developed that leveraged thermal weldment simulations to correct for positional inaccuracies. Initially, the thermal compensation strategy was trialled on simulated data and the positional accuracy was shown to increase by at least 85%. Experimental results also showed a similar trend with a 63.6% improvement in reflector positional accuracy. The results show how high-quality ultrasonic images can be generated at the point of manufacture and how a similar strategy could be deployed to establish an inspection record from manufacture until the asset is retired.
Date of Award14 Feb 2023
Original languageEnglish
Awarding Institution
  • University Of Strathclyde
SponsorsEPSRC (Engineering and Physical Sciences Research Council)
SupervisorCharles Norman MacLeod (Supervisor) & Anthony Gachagan (Supervisor)

Cite this