1. E. Todesco QUENCH BEHAVIOUR TEAM ACTIVITIES E. Todesco for the QBT CERN, Geneva Switzerland CERN, 7 th January 2016 MSC Technical Meeting

2. E. Todesco 2015 training analysis - 2 CONTENTS Tasks, composition, meetings HC data analysis 3000 vs 1000 and 2000 2008 vs 2015 Distribution shape and going towards 7 TeV Secondary quenches and quenches in both apertures Plateau quenches SM18 test Correlation to 3000 series production

3. E. Todesco 2015 training analysis - 3 TASKS A Quench Behaviour Team (QBT) has been formed in July 2015 www.cern.ch/qbt Common effort of MSC and MPE groups, and TE dept 7 meetings in July-December 2015 Average partecipation: 8, total number of colleagues: 17 Main scope Analysis of 2015 hardware commissioning data on training Slow trainer(s) to be replaced in LS2? Outlook for operation in next years Possibility of going to higher energy Correlation to production data Long-term goal: feedback to magnet design for future projects Margin: today 1200 LHC dipoles are operating at 80% of the loadline at 6.5 TeV with a training time at the edge of the shadow of HC

4. E. Todesco 2015 training analysis - 4 COMPOSITION B. Auchmann M. Bajko G. De Rijk P. Fessia P. Ferracin P. Hagen S. Le Naour J. C. Perez M. Modena F. Savary R. Schmidt J. Ph. Tock E. Todesco D. Tommasini A. Verweij G. Willering

5. E. Todesco 2015 training analysis - 5 CONTENTS Tasks, composition, meetings HC data analysis 3000 vs 1000 and 2000 2008 vs 2015 Distribution shape and going towards 7 TeV Secondary quenches and quenches in both apertures Plateau quenches SM18 test Correlation to 3000 series production

6. E. Todesco 2015 training analysis - 6 WHO IS QUENCHING Result I: 2015 confirmed that 3000 series magnets quench more than 2000/1000 series: 140 quenches of 3000 series, 27 of 2000 series and 5 of 1000 series Result II: In the 3000 series, the production is not homogeneous Differences are statistically significant Degradation at 3150-3200 (that went on 45) Number of quenches per magnet in 3000 series with statistical error

7. E. Todesco 2015 training analysis - 7 2008 VS 2015 What can we expect for the future? Two sectors can be compared: apparent paradox S56 powered at 11100 A 23 quenches in 2008, 16 quenches in 2015 S45 powered to 10300 A 3 quenches in 2008, 10 quenches in 2015 Adding statistical error with two sigma S56: 23±8 in 2008 versus 16±7 in 2015 S34: 3±6 in 2008 versus 10±6 in 2015 Result III: Data are compatible with a scenario where at each warm up we start in the same condition as at the beginning of the previous training

8. E. Todesco 2015 training analysis - 8 2015 VS VIRGIN TRAINING Did 3000 series totally lost memory? In virgin condition: 46% of 3000 series, i.e. 191±20 quenches In 2015: 34% of 3000 series, i.e. 142±19 quenches Result IV: data compatible with a partial but small preservation of memory HC «a bit better» than virgin training Training of virgin magnets during production

9. E. Todesco 2015 training analysis - 9 WHICH MAGNETS ARE BAD IN 56? Two extreme situations can be imagined: In 56 there are some bad magnets, always quenching, and other good magnets, never quenching In this case, all magnet quenching in 2008 will quench in 2015 In 56 all magnets are belonging to the same distribution, with a probability p of quenching In this case, the probability of having magnet quenching in 2008 and in 2015 will be p 2  (0.25) 2 =6% So it is enough to count how many magnet quenched both in 2008 and in 2015 3 magnets quenched in 2008 and 2015: 3358, 3330 and 3336 out of 84, ie 3% So magnets in 56 look equally bad Result V: S56 data compatible with magnets belonging to the same distribution (no bad or good magnets)

10. E. Todesco 2015 training analysis - 10 CONTENTS Tasks, composition, meetings HC data analysis 3000 vs 1000 and 2000 2008 vs 2015 Distribution shape and going towards 7 TeV Secondary quenches and quenches in both apertures Plateau quenches SM18 test Correlation to 3000 series production

11. E. Todesco 2015 training analysis - 11 SHAPE OF DISTRIBUTIONS Interesting result: HC data compatible with a gaussian Training follows the erf (errror function, integral of a gaussian) Statistical error (2  ) is associated to having 350 magnets Gaussian parameters:  =11.6 kA  =0.73 kA

12. E. Todesco 2015 training analysis - 12 SHAPE OF DISTRIBUTIONS Detail of previous plot

13. E. Todesco 2015 training analysis - 13 SHAPE OF DISTRIBUTIONS For 45, we quenched 75% of the magnets Erf fit does not look so nice …

14. E. Todesco 2015 training analysis - 14 SHAPE OF DISTRIBUTIONS For 45, we quenched 75% of the magnets Erf fit does not look so nice … But since the sample is 62 magnets only, statistical error is larger it fits within 2 sigma Gaussian parameters:  =10.75 kA  =0.45 kA

15. E. Todesco 2015 training analysis - 15 SHAPE OF DISTRIBUTIONS It does not look gaussian, the beginning is too steep First quenches are too low Tentative parameters:  =12.3 kA  =0.75 kA We now that sector 78 is much worse

16. E. Todesco 2015 training analysis - 16 SHAPE OF DISTRIBUTIONS We can separate the 78 data, having slower magnets But in this way data set is very small … but it fits Parameters:  =11.8 kA  =0.9 kA

17. E. Todesco 2015 training analysis - 17 SHAPE OF DISTRIBUTIONS The rest of 2000 This is a shaky fit, lower part still out Tentative parameters:  =12.9 kA  =1.0 kA

18. E. Todesco 2015 training analysis - 18 EXTRAPOLATION TO 7 TEV Result VI: HC data compatible with Gaussian distribution for the first quench Estimate for first quench based on gaussian fit: 450±100, plus the second quenches Assuming a priori a 5% quench probability for 1000 series It has been 1% for reaching 6.5 TeV

19. E. Todesco 2015 training analysis - 19 A FIRST SKETCH OF STRATEGY Main questions: Where is the second quench ? Close or beyond 7 TeV? A possible strategy Push 12 and 45 to 7 TeV before LS2 45 – to see the 3000 series second quench if any 12 – to see the 1000 and 2000 behaviour With  50 quenches more we could have more reliable guess for 7 TeV

20. E. Todesco 2015 training analysis - 20 PRODUCTION DATA To validate our analysis, we go back to production data All magnet made two quenches so we have access to full distributions 1000 case: first half gaussian, second half a bit slower Distribution falls faster on the short sample side as expected from the physics

21. E. Todesco 2015 training analysis - 21 PRODUCTION DATA And what about production data 1000 case: using two Gaussian distribution with the same average and two different sigma, very good agreeement Sigma on the upper part 30% smaller than in the lower part Gaussian parameters:  =12.0 kA   =0.70 kA   =0.45 kA

22. E. Todesco 2015 training analysis - 22 PRODUCTION DATA 2000 case Also in this case two Gaussian distribution with the same average and two different sigma give good agreeement Sigma on the upper part 30% smaller than in the lower part Gaussian parameters:  =11.35 kA   =0.90 kA   =0.60 kA

23. E. Todesco 2015 training analysis - 23 PRODUCTION DATA 3000 case Does not look Gaussian

24. E. Todesco 2015 training analysis - 24 PRODUCTION DATA Result VII: distribution of quench in the production is well fit by a asymmetric Gaussian Right side has a 30% lower sigma than left side Result VIII: 3000 series shows an anomalous distribution Another sign that something went wrong

25. E. Todesco 2015 training analysis - 25 CONTENTS Tasks, composition, meetings HC data analysis 3000 vs 1000 and 2000 2008 vs 2015 Distribution shape and going towards 7 TeV Secondary quenches and quenches in both apertures Plateau quenches SM18 test Correlation to 3000 series production

26. E. Todesco 2015 training analysis - 26 PLATEAU QUENCHES In 2015 we had 5 plateau quenches (until 3 November) All in 2000 series magnets Not an issue for operation, a mystery for magnet builders Initially all in sector 34, than also in other sectors Flattop quenches in 2015 [G. Willering]

27. E. Todesco 2015 training analysis - 27 TESTS IN SM18 Until now there is a gap between behaviour in SM18 and behaviour in the LHC tunnel We try to bridge this gap Three 3000 series spares available, 3096, 3100 and 3409 3409 only candidate I would dream to have a couple of 3150-3200 magnet to test (what went in 45)

28. E. Todesco 2015 training analysis - 28 TESTS IN SM18 Is this 3409 representative? This magnet was tested several times No sign of memory in the first two thermal cycle  0 quenches below 11100 A 3409 was already at SM18 on bench (no quench antenna) Additional quench performed, close to ultimate Another quench early in 2016, then discuss if/how to proceed Summary of test of 3409 [G. Willering]

29. E. Todesco 2015 training analysis - 29 CONTENTS Tasks, composition, meetings HC data analysis 3000 vs 1000 and 2000 2008 vs 2015 Distribution shape and going towards 7 TeV Secondary quenches and quenches in both apertures Plateau quenches SM18 test Correlation to 3000 series production

30. E. Todesco 2015 training analysis - 30 DIFFERENCES BETWEEN FIRMS Review of the main differences between firms [M. Modena, P. Fessia] Recall of the work done by M. Pojer and A. Devred in 2004 https://edms.cern.ch/document/501385/2 https://edms.cern.ch/document/486644/1 https://edms.cern.ch/document/486697/1 Among these differences: There was a non-negligible difference in the target for the azimuthal prestress in the ends between the three manufacturers. Firm3 had the lower target azimuthal prestress in coil ends The three manufacturers had a different procedure to put longitudinal stress on the coil ends. A difference in the resin used for impregnating the coil ends Since the beginning of 3000 series production shows good performance, it is clear that none of the above differences can explain (alone) the loss of performance experienced after the 100 th magnet.

31. E. Todesco 2015 training analysis - 31 3000 PRODUCTION During 3000 production there has been a double crisis (due to two different reasons), just before the worse batch that ended in 45 [P. Fessia] Crisis of the curing cycle, with bad curing in the range 3050-3150 Magnet 3030 to 3050 Magnet 3050 to 3160From 3160

32. E. Todesco 2015 training analysis - 32 3000 PRODUCTION During 3000 production there has been a double crisis (due to two different reasons), just before the worse batch that ended in 45 [P. Fessia] Crisis of the coil protection sheet, with partial decollaring in 3128 3130 3131 3132 3133 3134 3136 3137 3138 3139 3140 3141 3142 3143 3144 3145 3147 3149 3150 3151 3152 3153 3154 3156 In both cases, these crisis are just before the worse batch 3150-3200 that ended in 45 So the mystery is still there

33. E. Todesco 2015 training analysis - 33 FUTURE DIRECTIONS Continue the analysis of 3000 production and comparison to behaviour in the tunnel Clarify the role of secondary quenches (if any) Summarize what is the situation for the other magnets Manage to reproduce in SM18 the HC behaviour of 3000 series Agree with the other groups and with the management on the proposal for exploring the 7 TeV performance