MuCool_01 TC3 Quench Performance Michael Tartaglia Fermilab TD Magnet Systems 3/21/14
TC3 Temperature Profile 5 months to review, then implement improvements 1) Thermal shield (He boiloff gas return pipe & rib cage) 2) Careful MLI installation Good steady vacuum, P=4 ∙10 -6 Torr NOTE: Cooling tube “bypass” (due to leak) is located at RTD2 location – worst cooling condition Temperatures are very stable, Eddy current heating is small: at 1.8 A/min dT=90mK at RTD2 (5.80K)
Coupling Coil Construction 12 layers per coil, stycast “wet layup” Cu/Nb-Ti Single Strand (L=600 H) Slip planes to reduce shear stress (?) There is no MICE Note or Publication on the actual coil fabrication LHe Cooling Tube (There are for prototype test coils) welded to outer Aluminum Ring Coil 1 Coil 8 Voltage Taps, protection diodes across each coil
150A Field Profile On Solenoid Coil 8 Forces on outer R Coils 765 N/m 450 N/m 630 N/m Axial Field at Coils 4,5 is near zero Coil 1 inner R
Quench Training Main concerns: frictional energy release caused by 1) stick-slip motion of coil wrt structure, 2) epoxy cracking from stress; 3) conductor motion
TC3 Q1 Iq=123.2A dI/dt=0.6A/min First “triggered” event is at about twice the current previously reached in Sept. 2013 (64A). The very first event seen in Sept. at 62A was a similar voltage spike. Diodes turn on in the range of 5 to 11 V (this is seen to increase somewhat with quench current ~ B Voltage Spike disturbance profiles are the same in all events; only the amplitude changes. Note coil 4 disturbance is always flat (suggesting radial coil motion?)
TC3 Q2 Iq=126.5A After this event, steps were taken to desensitize the quench protection half- coil difference signal. Quench Characterization system real time 8 th coil voltage traces clearly show a lot of voltage spike activity (not just in coils 1,2,3), especially at higher current. These are interpreted as the result of conductor motion in the solenoid field. Clearly, not all disturbances result in a quench.
TC3 Q3 Iq=127.7A Coil 1 All quench development plots shown on the same {V,t} scale 10ms Hcoil filter and raised threshold (4.5V) It is clear now that voltage spike disturbances can lead to a real quench at this low current.
TC3 Q4 Iq=129.7A 10 ms filter and 4.5 V threshold is still susceptible to tripping on spikes… From here on a 15ms validation delay for half coil above threshold is required.
TC3 Q5 Iq=140.2A Coil 1 Coil 5 is different here!
TC3 Q6 Iq=138.5A Coil 1 Coil 5 is noisy hereafter
TC3 Q7 Iq=143.1A Coil 1
TC3 Q8 Iq=144.7A Coil 1
TC3 Q9 Iq=145.3A Coil 1 dI/dt=0.9A/min from here on
TC3 Q10 Iq=133.8A Coil 1
TC3 Q11 Iq=147.3A Coil 1 A look at LTS Leads:
TC3 Q12 Iq=143.0A Coil 1
TC3 Q13 Iq=144.1A Coil 1
TC3 Q14 Iq=147.5A Coil 1
TC3 Q15 Iq=155.7A Coil 1
TC3 Q19 Iq=157.1A Coil 1
TC3 Q20 Iq=160.8A Coil 1
TC3 Q22 Iq=161.5A Coil 1
TC3 Q23 Iq=162.6A Coil 1
TC3 Q25 Iq=161.1A Coil 1
TC3 Q26 Iq=168.9A Coil 1 Hall Probe (current)
TC3 Q27 Iq=167.6A Coil 1 ∫ (I 2 dt)=0.115 MIITS (relationship to hot spot temperature is not known! H. Pan at LBNL is modeling this) Previous calcs predicted 130K at 210 A, but under different assumptions; here we are forcing all diodes to conduct.
TC3 Q28 Iq=171.9A Coil 1
TC3 Q29 Iq=177.3A Coil 1
TC3 Q30 Iq=178.2A Coil 1 Coil 2 also quenches
TC3 Q32 Iq=184.6A Coil 1 dI/dt=1.8A/min Quench Developments are all VERY SIMILAR Variable delay between time of disturbance and start of quench propagation depends upon: 1) Energy deposited >MQE 2) Distance from high field region Need T> critical surface (r,z) Expect Coil 1 diode to conduct Sometime soon! (14.6V)
TC3 Q16 Iq=152.6A Coil 8
TC3 Q17 Iq=150.8A Coil 8
TC3 Q18 Iq=151.9A Coil 8
TC3 Q24 Iq=166.1A Coil 8
TC3 Q31 Iq=179.3A Coil 8 dI/dt=1.8A/min from here on: 2 Pwr Supplies PS turning off ?
TC3 Q21 Iq=157.2A Coil 5,4 t( 5 ) = -.285 dV/dt(3V)=61 V/s t( 4 ) = -.235 dV/dt(3V)=65 t( 3 ) = -.130 dV/dt(3V)=93 t( 6 ) = -.130 dV/dt(3V)=35 Coils 3, 4, 5 motion at t=-.027 t( 2 ) = -.027 Quench starting in coils 5,4 must be at the coil end (axial field is small, radial field is maximal)
Quench Velocity at 3V Quench Characterization plots show that all quenches in a given coil, at very similar currents, develop in essentially the same way. Quench velocity appears to scale with the peak axial field on the quenching coil, as might be expected. Coil4/5 quench velocity is consistent with being at the coil end.
Peak Surface Temperature vs Iq Heng Pan at LBNL intends to model this.
CC - Comments • Add the strain gauge map around the coil • Strain was converted into stress • Stress is influenced by bending structure (short coil) • Ratcheting and irreversible behavior suggests motion and epoxy cracking as a cause of training • Voltage taps signals needs to be correlated with SG behavior on location • Gauges that are most active during Q1-5 are SG1,2,5,7,8 • Gauges that are most active during Q6-8 are SG2,4 • (the current has not been corrected for a small offset) Superconducting Magnet Group S. Caspi 3/21/2014 40
Deformed Shape warm/cold (No Lorentz Forces in this picture) Superconducting Magnet Group S. Caspi 3/21/2014 41
Change in Stress During Ramps 1-5 SG1 SG2 SG4 SG3 SG7 SG5 SG6 SG8 Superconducting Magnet Group S. Caspi 3/21/2014 42
Change in Stress During Ramps 6-8 SG1 SG2 SG4 SG3 SG7 SG5 SG6 SG8 Gauges show “racheting”, due to mechanical stress changes. Superconducting Magnet Group S. Caspi 3/21/2014 43
Strain Gauge Response (from Shlomo) … possibly unloading of coil pre -load with Lorentz force ~ I 2
Conclusion STAY TUNED… 1) Try to complete quench training to 210 A + 2% 2 quenches/day: <1.8 A/quench> about 1-2 more weeks Reached Minimum acceptable current for MICE (>180 A) 2) “Soak Test” (run 24 hours at 210 A) 3) Thermal Cycle (~2 weeks) 4) Test Quench Re-Training (<2 weeks) DONE by ~ MAY ASC’14 Test Results Abstract accepted for Oral Presentation
Recommend
More recommend
