Testing during manufacture Inspired by XFEL cryomodule assembly

Testing during manufacture Inspired by XFEL cryomodule assembly at CEA/Saclay XFEL Kapton foil discovered on XM 7 warm coupler at AMTF O. Napoly, CEA-Saclay, Irfu/SACM i

Preliminary Thoughts/Principles i • A good operator is worth ~100 QC tests • invest in the selection of competent (or even qualified), rigorous and motivated, in one word reliable technicians. • The scope of testing is inherently limited by the ambition of the assembly schedules: the faster the schedule, the fewer tests ! • In XFEL, all reception/assembly/test/transfer/documentation work for one cryomodule needs to be concentrated in 40 hours (1 week), actually ~160 man. hour labor. • In ESS elliptical medium beta: 9 cryomodules in 18 months • Testing can be detrimental: invasive to the accelerator components and disruptive to the assembly • The leading principle for testing is the balance between the rejection rate and the extra-work/extra-risk of the testing procedure • 0% rejection likelihood after 100 modules is a waste of effort • 10 % rejection likelihood is globally ineffective if it is accompanied by 10% failures driven by the testing alone. 16 October 2014 ESS Workshop on QC 2

Preliminary Thoughts/Principles i • Corrective actions and decision making for negative tests: are they always specified ? Exemption is key. • Documentation for acceptance test needs to include corrective actions, of possible • A global system test if the best way to identify work faults: perform early on site in-house RF test is the ultimate. • Work faults is related to bad procedures or to bad operators 16 October 2014 ESS Workshop on QC 3

Concrete Examples i • Leak test of cavities at reception, requested for DESY-Saclay safe transport contract: abandoned after XM 4 because: 1. no cavity was found leaky 2. leak test is performed before coupler assembly 3. angle valve mis-manipulation • RGA rejection rate: about 10 RGA tests are performed per module: 1. not a single rejection up to XM 20 2. mass spectrometers are very unreliable and introduce delays on cold coupler assembly 3. Corrective action in case of negative RGA on string or module ? ? 4. Replace mass spectrometer by leak detector ? 16 October 2014 ESS Workshop on QC 4

Concrete Examples i • Beware of elaborate but not built in-house control devices and software: • the good example is the RF-tuning and control (and TDR) crates built at CEA in 4 units. • the example of improvement is the commandcontrol system of the 8 DESY and 3 CEA pumping units many uncontrolled venting event during training and ramp-up phase. • Not so successful: • ‘automatic’ alignment software, too rigid • tuner motor and piezo-tuner electric tests, not robust to operator (mis-)use. • Failure handling is not properly treated. 16 October 2014 ESS Workshop on QC 5

Auto-quality check Matrix i Non-conform product Detection of Non. Conformance 16 October 2014 Origin of Non-Conform Assembly ESS Workshop on QC 6

Discussion Themes 1. 2. 3. 4. 5. 6. 7. i Invasiveness of leak checks Decision taking on RGA Definition of hold points with CEA during assembly Definition of Testing vs. Measurement QC on 1. 1 MW ESS coupler assembly vs. 250 k. W XFEL coupler QC on fastening hardware torque and on torque specification On site cryomodule testing 16 October 2014 ESS Workshop on QC 7