## Abstract

Electron beam welding and friction stir welding are the two processes now being considered for sealing copper canisters with Sweden's radioactive waste. This report outlines a strategy for verification and demonstration of the encapsulation process which here is considered to consist of the sealing of the canister by welding followed by quality control of the weld by non-destructive testing. Statistical methodology provides a firm basis for modern quality technology and design of experiments has been successful part of it. Factorial and fractional factorial designs can be used to evaluate main process factors and their interactions. Response surface methodology with multilevel designs enables further optimisation. Empirical polynomial models can through Taylor series expansions approximate the true underlying relationships sufficiently well. The fitting of response measurements is based on ordinary least squares regression or generalised linear methods. Unusual events, like failures in the lid welds, are best described with extreme value statistics and the extreme value paradigm give a rationale for extrapolation. Models based on block maxima (the generalised extreme value distribution) and peaks over threshold (the generalised Pareto distribution) are considered. Experiences from other fields of the materials sciences suggest that both of these approaches are useful. The initial verification experiments
More>>

Mueller, Christina;

^{[1] }Oeberg, Tomas^{[2] }- Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin (Germany)
- Tomas Oeberg Konsult AB, Lyckeby (Sweden)

## Citation Formats

Mueller, Christina, and Oeberg, Tomas.
Strategy for verification and demonstration of the sealing process for canisters for spent fuel.
Sweden: N. p.,
2004.
Web.

Mueller, Christina, & Oeberg, Tomas.
Strategy for verification and demonstration of the sealing process for canisters for spent fuel.
Sweden.

Mueller, Christina, and Oeberg, Tomas.
2004.
"Strategy for verification and demonstration of the sealing process for canisters for spent fuel."
Sweden.

@misc{etde_20564023,

title = {Strategy for verification and demonstration of the sealing process for canisters for spent fuel}

author = {Mueller, Christina, and Oeberg, Tomas}

abstractNote = {Electron beam welding and friction stir welding are the two processes now being considered for sealing copper canisters with Sweden's radioactive waste. This report outlines a strategy for verification and demonstration of the encapsulation process which here is considered to consist of the sealing of the canister by welding followed by quality control of the weld by non-destructive testing. Statistical methodology provides a firm basis for modern quality technology and design of experiments has been successful part of it. Factorial and fractional factorial designs can be used to evaluate main process factors and their interactions. Response surface methodology with multilevel designs enables further optimisation. Empirical polynomial models can through Taylor series expansions approximate the true underlying relationships sufficiently well. The fitting of response measurements is based on ordinary least squares regression or generalised linear methods. Unusual events, like failures in the lid welds, are best described with extreme value statistics and the extreme value paradigm give a rationale for extrapolation. Models based on block maxima (the generalised extreme value distribution) and peaks over threshold (the generalised Pareto distribution) are considered. Experiences from other fields of the materials sciences suggest that both of these approaches are useful. The initial verification experiments of the two welding technologies considered are suggested to proceed by experimental plans that can be accomplished with only four complete lid welds each. Similar experimental arrangements can be used to evaluate process 'robustness' and optimisation of the process window. Two series of twenty demonstration trials each, mimicking assembly-line production, are suggested as a final evaluation before the selection of welding technology. This demonstration is also expected to provide a data base suitable for a baseline estimate of future performance. This estimate can then be used for the safety assessment. In order to provide data for the analysis of welds it is required to use well defined methods for the investigation of the weld integrity. As the total length of the welds produced during the verification and demonstration exceeds 140 m and these needs to be examined with a reasonable resolution non destructive methods (NDT) with high data acquisition rates are required. Destructive methods like metallographic sectioning and examination or tensile testing can be used only to provide complementary information as the acquisition of these data are labour-intensive and time consuming. As the analysis is relying on the reliability of the NDT it is required to have a strategy for the determination of this. Three different ways to investigate the reliability of NDT signals are described. The performance demonstration, the parameter approach and the Integral Approach: ROC - Receiver Operating Characteristic and its relation to POD (probability of detection). The first step of a performance demonstration is to define the essential technical parameters of the system. The ROC and POD methods are appropriate tools to provide a clear measure of integral performance of the system though it has to be paid by high effort in test series with realistic test samples. With POD the user can learn about the detection capability whereas the ROC gives more information about the system's capability to distinguish between signal and noise. The modular approaches open the door to a promising technique - more efficient and with the capability also to optimize the system. The data evaluation methods chosen to be applied for this assessment is the quantitative POD (Probability of Detection) method according to MIL-STD 1823. The (1-POD) curve will provide the probability of missing a defect as function of defect size which can be used as input for probabilistic risk assessment. Finally, a hypothetical calculation example Appendix 2 is provided to illustrate the fitting of data and extrapolation using the extreme value models.}

place = {Sweden}

year = {2004}

month = {Aug}

}

title = {Strategy for verification and demonstration of the sealing process for canisters for spent fuel}

author = {Mueller, Christina, and Oeberg, Tomas}

abstractNote = {Electron beam welding and friction stir welding are the two processes now being considered for sealing copper canisters with Sweden's radioactive waste. This report outlines a strategy for verification and demonstration of the encapsulation process which here is considered to consist of the sealing of the canister by welding followed by quality control of the weld by non-destructive testing. Statistical methodology provides a firm basis for modern quality technology and design of experiments has been successful part of it. Factorial and fractional factorial designs can be used to evaluate main process factors and their interactions. Response surface methodology with multilevel designs enables further optimisation. Empirical polynomial models can through Taylor series expansions approximate the true underlying relationships sufficiently well. The fitting of response measurements is based on ordinary least squares regression or generalised linear methods. Unusual events, like failures in the lid welds, are best described with extreme value statistics and the extreme value paradigm give a rationale for extrapolation. Models based on block maxima (the generalised extreme value distribution) and peaks over threshold (the generalised Pareto distribution) are considered. Experiences from other fields of the materials sciences suggest that both of these approaches are useful. The initial verification experiments of the two welding technologies considered are suggested to proceed by experimental plans that can be accomplished with only four complete lid welds each. Similar experimental arrangements can be used to evaluate process 'robustness' and optimisation of the process window. Two series of twenty demonstration trials each, mimicking assembly-line production, are suggested as a final evaluation before the selection of welding technology. This demonstration is also expected to provide a data base suitable for a baseline estimate of future performance. This estimate can then be used for the safety assessment. In order to provide data for the analysis of welds it is required to use well defined methods for the investigation of the weld integrity. As the total length of the welds produced during the verification and demonstration exceeds 140 m and these needs to be examined with a reasonable resolution non destructive methods (NDT) with high data acquisition rates are required. Destructive methods like metallographic sectioning and examination or tensile testing can be used only to provide complementary information as the acquisition of these data are labour-intensive and time consuming. As the analysis is relying on the reliability of the NDT it is required to have a strategy for the determination of this. Three different ways to investigate the reliability of NDT signals are described. The performance demonstration, the parameter approach and the Integral Approach: ROC - Receiver Operating Characteristic and its relation to POD (probability of detection). The first step of a performance demonstration is to define the essential technical parameters of the system. The ROC and POD methods are appropriate tools to provide a clear measure of integral performance of the system though it has to be paid by high effort in test series with realistic test samples. With POD the user can learn about the detection capability whereas the ROC gives more information about the system's capability to distinguish between signal and noise. The modular approaches open the door to a promising technique - more efficient and with the capability also to optimize the system. The data evaluation methods chosen to be applied for this assessment is the quantitative POD (Probability of Detection) method according to MIL-STD 1823. The (1-POD) curve will provide the probability of missing a defect as function of defect size which can be used as input for probabilistic risk assessment. Finally, a hypothetical calculation example Appendix 2 is provided to illustrate the fitting of data and extrapolation using the extreme value models.}

place = {Sweden}

year = {2004}

month = {Aug}

}