Summary of test results of MQXFS1 – the first short model 150 mm aperture Nb 3 Sn quadrupole for the High-Luminosity LHC upgrade Stoyan Stoynev (on behalf of HL-LHC AUP collaboration) MT25 Conference, Amsterdam, Nederlands 30 August 2017
Introduction • MQXFS1 is the first 150 mm diameter, 1.5 m long Nb 3 Sn quadrupole from LARP/HiLumi – Coils fabricated by CERN (#103 and #104) and LARP (#3 and #5) • Three test cycles performed – MQXFS1a in Feb/May 2016 – MQXFS1b in Sep/Dec 2016 • Increased azimuthal pre-stress (35%) – MQXFS1c in May/July 2017 • Increased axial pre-stress (65%) • Main goals – Training and performance characteristics – Magnetic measurements (see Suzana’s talk) MQXFS1 and VMTF at FNAL – Magnet protection studies (see Emmanuele’s talk) 2 30-Aug-17
Cable properties and SSL estimates • Coils made of Nb 3 Sn strand – RRP 108/127 (#3 and #5) – RRP 132/169 (#103 and #104) Short- sample Nom. Nom. Current Field Gradient Current Field Gradient Current Gradient Nom. Field % Iss 4.3 K 4.3 K 4.3 K 1.9 K 1.9 K 1.9 K 1.9 K 1.9 K 1.9 K 1.9 K kA T T/m kA T T/m kA T/m T % Coil 103 19.550 13.383 155.164 21.50 14.599 169.083 16.47 132.667 11.425 77 Coil 104 19.775 13.525 156.783 21.78 14.769 171.026 16.47 132.667 11.425 76 Coil 3 20.118 13.740 159.245 22.28 15.080 174.573 16.47 132.667 11.425 74 Coil 5 19.725 13.493 156.424 21.85 14.813 171.526 16.47 132.667 11.425 75 Magnet 19.550 13.383 155.164 21.50 14.599 169.083 16.47 132.667 11.425 77 SSL at 1.9 K: 21.5 kA (coil #103) 3 30-Aug-17
Instrumentation Voltage taps schematics • Voltage taps on the IL and OL • CERN and LARP strain gauge systems with gauges – On coils, shell, rods – Azimuthal/longitudinal sensitivity QA picture and initial positions • Adjustable quench antenna (QA) • Magnet is protected with IL/OL heaters 4 30-Aug-17
Strain gauges monitoring • ”CERN” and “LARP” SG systems used in parallel – Different (AC/DC) sources – Different configurations – Different DAQ and sampling frequency Coil 5, azimuthal strain ~17 kA • Results consistent and also in agreement with Finite Element Analysis (FEA) calculations • Coil pre-stress levels verified ~15 kA with the SG and coil-unloading “shifted” to higher current after the first test cycle More on mechanical analysis in Giorgio’s talk 5 30-Aug-17
Quench training • Fairly fast training in the first test cycle • Slower in the next cycles with multiple detraining quenches – All in coil 3, lead end – Between wedge and end-part (see dedicated poster presentation) Training in the three test cycles In all tests the ultimate current (8% field above LHC operational field) reached Reached 95.8% of SSL@4.9K and 88.4% of SSL@1.9K Good training memory; partial loss of memory after axial pre-stress increase 6 30-Aug-17
Quench training in coils • Reordering the training sequences in terms of coils • “CERN” coils and the mirror -magnet coil show similar training • “LARP” coils training fast initially but curve flattening fast – Training obstructed by a “weak spot” • Likely same location in coils 3 and 5 • See analysis in Thomas’ poster presentation Quench current normalized to SSL per coil 4.5 K 7 30-Aug-17
Ramp rate dependence and temperature margin • Very weak quench current dependence on the ramp rate • High ramp-rate dependence consistent in all test cycles • Cable degradation of ~5-6% – At 4.5K - consistent between test cycles within 1 % • Behavior consistent with a magnet still training – ~ same absolute quench current at 1.9 K and 4.5 K 8 30-Aug-17
Quench locations • Only in the first test cycle training was rich in locations • Later many detraining quenches in coil 3 • Quenches up-to 300 A/s similar to training quenches • At higher ramp rate all quenches in coil 5 – outer layer, mid-plane region • At higher temperature quenches in coil 5 – one exception in the first test cycle – all in inner layer, likely around first wedge like in coil 3 (@1.9K) White marks: positions of QA sensors that helped in the location analysis In yellow: Likely region of most quenches in coil 3 (and 5) See the dedicated poster presentation 9 30-Aug-17
RRR and training First quench current vs RRR (per coil) • RRR measured per segment and averaged per coil • Negative correlation observed vs the first quench current • Underlying reasons not “(m)” – mirror magnet test understood (investigation on going) RRR in coil segments 10 30-Aug-17
Splice resistance measurements • Splice resistance measurements in all test cycles – Consistent results • Different soldiering techniques used in “LARP” and “CERN” splices – coils (03,05) vs (103,104) • CERN splices 2-3 times less resistive (as expected) • All splices good : R < 1 n Ω (acceptable level) coil Splice R [n Ω ] 03 A 0.34 A = IL 03 B 0.30 05 B 0.29 B = OL 103 A 0.15 104 A 0.10 Uncertainty: 0.06 n Ω Instrumental offsets not relevant in the context 11 30-Aug-17
Summary • The first 150-mm dimeter Nb 3 Sn magnet demonstrated strong performance • It reached 95.8% of SSL@4.5 K and 88.4% of SSL@1.9 K – 15% above LHC operational current at 1.9 K – target currents reached in all test cycles • Behavior at high ramp rate and temperature did not change with higher pre-stress levels or over time with testing • Multiple detraining quenches were observed after pre-stress increase – mostly same coil, same location – no real training plateau reached 12 30-Aug-17
Summary • Very good training memory – no sizable loss after azimuthal pre-stress increase – confirmed observations from MQXFS1 with thermal cycle test • Axial pre-stress (after azimuthal pre-stress) increase lead to partial loss of training memory – Possibly related to “un -loading ” of the coil during the procedure • RRR coil values were observed to be negatively correlated with the first quench currents in the coils • Splice resistances were measured to be very small 13 30-Aug-17
Back up • Holding current tests performed successfully – 8 hours at operational current and later at 17760 A – 2 hours at ultimate current (17890 A) MQXFS1b Cool-down Warm-up Start: End: 17 September 2016 13 December 2016 14 30-Aug-17
Back up 15 30-Aug-17
Recommend
More recommend
