Real-Time Canister Welding Health Monitoring and Prediction System for Spent Fuel Dry Storage
- Intelligent Automation Inc.
- Oak Ridge National Laboratory
- Orano Federal Services, LLC
The possibility of chloride-induced stress corrosion cracking (CISCC) in welded stainless-steel dry storage canisters (DSC) for spent nuclear fuel (SNF) has been identified as a potential concern regarding long-term performance of the canister’s containment boundary. During canister fabrication, the welding procedure introduces high tensile residual stress and sensitization in the heat-affected zone (HAZ), which might render the welds susceptible to the incubation of pitting and transition to crack initiation and growth when exposed to an aggressive chemical environment. To reduce the potential for a through-wall CISCC and breach of the canister’s containment boundary, Intelligent Automations, Inc. (IAI), along with Oak Ridge National Laboratory (ORNL) and Orano TN, aims to develop and test an innovative multi-sensor network with Machine Learning (ML) system called CanisterMonitor™ for in-situ DSC chloride induced stress corrosion crack (CISCC) diagnosis and prognosis. The goal is to design and manufacture a suite of sensing system to real-time monitor the canister health, and use a machine-learning tool for predictive detection and interpretation of pits and cracks on canister. Toward this goal, the team has been mainly focused on implementing CanisterMonitor™ prototype with Acoustic Emission (AE) sensing capability in phase I, and evaluate the AE sensor signal correlation with CISCC. Our Phase I efforts are summarized as follows: Investigated CISCC on stainless steel via literature surveys and simulated crack growth rate on stainless steel. The team performed the literature review to understand the CISCC and acoustic emission on the canister structure. Simple crack propagation problem was simulated in stainless steel. We have also set up the welded stainless-steel model to simulate the canister welding. This guided the plan for the corrosion test. Designed CanisterMonitor™ architecture. The CanisterMonitor™ includes several sensing mechanisms, data acquisition hardware, and interface hardware. We focused on the AE sensing in phase I. We adapted from the previously developed bridge structural health monitoring system and customized it for the AE sensor node and wireless gateway for CanisterMonitor™. Performed corrosion test on U-bent and welded stainless steel 304L samples and collected a large amount of AE signatures due to SCC. ORNL provided us a representative canister welded specimen with submerged arc welds (SAW). A cut out of the sample was used for submersion corrosion tests performed at IAI. The AE sensors was mounted on the specimens and AE events were monitored. There are totally more than 6000 AE events recorded for a 35-day period. We have also performed accelerated environmental corrosion test with welded specimens placed in a salt fog cabinet for corrosion test at ORNL. Both specimen with and without artificial crack to simulate the SCC were used in the corrosion test to monitor the crack growth. However, no crack growth was noticed from visual inspection after 4-week test. AE signals were recorded and analyzed. There is no indication of crack growth from AE signals. Designed and developed Support Vector Machines (SVN) networks for AE signal classification and similarity detection. The team performed the signal processing, feature extraction, and machine learning on collected AE signals for CISCC detection and quantification. The experimental data was preprocessed before we applied machine learning algorithms for signal classification. We used a well-known SVM-based machine learning tool, called LibSVM, to classify the AE signals into two classes. One class of signals contains low-frequency noise and other class does not have much of such noise. Although the detection accuracies for both classes are not very high. However, the classification model, which was blindly trained with all the original imbalanced training data (<1 % noise signal), we will further investigate the way we currently extract features and collected more data in phase II. Collaborated with nuclear canister OEM for transition. The team worked closely to develop the commercial plans for successfully transitioning the technology to Orano. We have shared our research progress with Orano, who widely supplies the TN spent nuclear fuel (SNF) storage canister for nuclear plants. We identified the potential application with Orano, and engaged them for further discussions and determine the best way to involve them in Phase II for suitable transition plans. Overall, Orano sees definitive potential benefits to implementing the CanisterMonitoring™ system for monitoring of DSC health under extended storage conditions. In brief, our Phase I efforts have demonstrated the feasibility of the proposed CanisterMonitor™ solution to monitor the CISCC to reduce the potential for a through-wall CISCC and breach of the canister’s boundary. CanisterMonitor™ can also be applied to other structural health monitoring applications. CanisterMonitor™ in general can provide a solution for non-destructive evaluation of a structure to better assist the end users for system diagnosis.
- Research Organization:
- U.S. Department of Energy
- Sponsoring Organization:
- USDOE Office of Science (SC)
- DOE Contract Number:
- SC0019887
- OSTI ID:
- 1638370
- Type / Phase:
- SBIR (Phase I)
- Previous Contract Number:
- DE-SC0019887
- Report Number(s):
- FinalReport-IAI-2471
- Country of Publication:
- United States
- Language:
- English
Similar Records
Novel Plating Process for Mitigating Stress Corrosion Cracking in Dry Storage Canister Welds
Crack Growth Rate and Large Plate Demonstration of Chloride-induced Stress Corrosion Cracking in Spent Nuclear Fuel Storage Canisters