European Nb₃Sn and Nb–Ti strand verification for ITER: processing, measurements and statistical analysis

We present a large quantity ( ∼ 13000 ) of verification measurements (critical current ( I c ), n -value, hysteresis loss ( Q ), residual resistivity ratio (RRR), twist-pitch, diameter, plating thickness and copper-non-copper ratio) and statistical analysis of internal tin and bronze route (BR) Nb 3 Sn and Nb–Ti strands used in the ITER tokamak magnet system. The Durham laboratory partnered the relevant manufacturer and each independently processed-and-measured one of two short adjacent strands taken from the ends of thousands of piece-lengths, typically several kilometres long. Processing included heat-treatment of the Nb 3 Sn, which is an irreversible process (i.e. destructive of as-supplied strands). Here we show that when repeat processing-and-measurement of the same piece of Nb 3 Sn strand is not possible, although the similarity between adjacent strands is not known a priori , processing-and-measuring adjacent strands provides a proxy for single lab (with repeat) measurements or round-robin measurements (without processing). Processing-and-measuring adjacent strands provides limits for the errors introduced by laboratories and for the variability of the piece-lengths, rather than the specific values that round-robin measurements provide. This is because it is not possible to distinguish whether unidentified sources of error come from the differences between adjacent strands or from unidentified laboratory errors, such as from gas impurities during heat-treatment or strand handling. We categorise different types of processing-and-measurement: one type includes I c of BR Nb 3 Sn where the maximum lab errors are similar to the well-known estimated lab errors that are found using, for example, the commercially specified accuracy of the instruments. Furthermore, the possibility of unidentified sources of random lab error is excluded, and both piece-length variability and lab errors are specified—similar to round-robin measurements. In another type (e.g. RRR or Q of internal tin Nb 3 Sn), larger than expected variances are measured and unidentified sources of error operate. Categorising the different types of processing-and-measurement helps identify routes to improvement.