Thermal runaways in LHC main circuit interconnections: Experiments Gerard Willering

With contributions of:TE/MPE: Arjan VerweijTE/MSC-CI: N. BourceyTE/MSC-TF: M. Bajko, G. Deferne, G. Dib, M. CharrondiereTE/MSC-SCD: L. Bottura, D. Richter, G. Peiro, C. Scheuerlein, S. HeckTE/MSC-LMF: P. Fessia, K. Chaouki, R. Principe, S. TriquetEN/MME: T. Regnalia, P. PerretTE/EPC: G. Hudson, M. CerqueiraEN/ICE: A. Rijllart, D. KudryavtsevTE/CRG: V. Bendaand many more...

Thermal runaways in LHC main circuit interconnections: Experiments

Gerard Willering

1

Technology Department

Gerard Willering – Splice review – 18 October 2010 - CERN

Experiments to the effects of defected and consolidated LHC main bus splices are conducted in three phases

1. Thermal runaways in interconnections with defectsAugust 2009 – Februari 20105 quadrupole busbar samples in test station FRESCAGoal: Validation of model → safe operating current before consolidation

2. Proof of principle of the consolidation with shuntsMarch 2010 – June 20104 quadrupole busbar samples in test station FRESCAGoal: Validation of model → Proof of principle of the consolidation proposal

3. Final validation of the consolidation with shunts in a realistic test setupMarch 2010 – October 20102 dipole busbar samples inbetween two special SSS magnetsGoal: Validation of model and final validation of the shunts

2

Technology Department

Gerard Willering – Splice review – 18 October 2010 - CERN

Contents

3

Technology Department

Gerard Willering – Splice review – 18 October 2010 - CERN

Sample Defects LNSC

(MM)RADD (T = 300 K)(ΜΩ)

RADD (T = 10 K)(ΜΩ)

RRRCable

RRR Busbar

1 Single-sided 47 63 0.37 170 ~3002A Double-sided 35+27 43+32 0.43+0.24 100-130 ~2702B Single-sided 35 42 0.26 ~170 ~2903A Double-sided 39 + 30 51+39 0.31+0.28 140-170 ~1903B Single-sided 21 27 0.22 120 ~160

Definition of a defect:1. Stabilizer discontinuity2. Non-stabilized cable with a specified length

1

1

Gamma-ray image of sample 1, indicating the single-sided defect.

2

Important parametersRcable = 1.3 µΩ/mmRquad-busbar = 0.1 µΩ/mmRadd = Rmeasured - R8cm

RRRbus

RRRcable

30 mm non-stabilized cable

Guaranteed by Kapton tape

Preparation of the defect.

Defect preparation

4

Technology Department

Gerard Willering – Splice review – 18 October 2010 - CERN

Sample 2ASingle-sided defect

Sample 2BDouble-sided defect

Test layout with normal LHC pieces and geometry and with lots of instrumentation (RQ circuit)

DiscontinuityDiscontinuity

Heater

Sample preparation

5

Technology Department

Gerard Willering – Splice review – 18 October 2010 - CERN

Test station FRESCA

- In the FRESCA teststation the sample length is limited to 1.7 m, which gives 0.8 meter of busbar on each side of the interconnection. -24 Voltage taps- 10 Thermocouples- 5 heaters- The ends of the busbars are thermalized (a lot of copper in direct contact with helium).- Measurements are performed with constant current.

- Due to limitations of the test station (Helium volume, length of sample, vincinity of the current leads) the quadrupole interconnections are chosen to test.

6

Technology Department

Gerard Willering – Splice review – 18 October 2010 - CERN

Voltage innon-stabilized cable

Temperature innon-stabilized cable

Temperature in busbar

Typical measurement data

Thermal Runaway7 kA, 43+32 µΩ defect

Fingerprint of a local thermal runaway:- Relatively low busbar temperature.- Accelerated voltage increase in the non-stabilized cable.Main characteristic: Thermal runaway time trun

7

Technology Department

Gerard Willering – Splice review – 18 October 2010 - CERN

Thermal runaway time

- Except for sample 3B, all samples would melt within 1 second with a current of 12 kA.- The MIITs (kA^2/s) for an exponentially decaying current with timeconstant τ is reached by a constant current in t = 0.5*τ. - For the quadrupole circuit with τ = 20 s, we can correlate the safe currents for the sample conditions with a cross-section at trun = 10 s.

- Although there is a correlation, safe currents can not be drawn from the measurements.

Sample Defects LNSC

(MM)1 Single-sided 472A Double-sided 35+272B Single-sided 353A Double-sided 39 + 303B Single-sided 21

8

Technology Department

Gerard Willering – Splice review – 18 October 2010 - CERN

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

0.0 2.0 4.0 6.0 8.0 10.0 12.0

Ra

dd

bigg

est d

efec

t at

10 K

(µ

Ω)

I at trun=10s (kA)

The current at trun versus the additional resistance R add shows a good correlation.The allowed power at 10 K is between 16 and 27 W.

Since we varied the applied field on the sample, the effective Radd varied giving us a wider range in measurements. Therefore more than 5 points (number of samples) are shown.

16 W

27 W

Measurement characterization

9

Technology Department

Gerard Willering – Splice review – 18 October 2010 - CERN

Melt-down of a non-stabilized cable

To perform multiple thermal runaway measurements, the current is cut-off when the maximum temperature reaches in between 100 and 300 K. Out of 175 run-aways we did, we choose the smallest defect of 20 mm at 9 kA to demonstrate that the incident can be reproduced. In fact, each of the 175 measurements would lead to a melt-down.

10

Technology Department

Gerard Willering – Splice review – 18 October 2010 - CERN

Melt-down of a non-stabilized cable

With an increased protection cut-off voltage the thermal runaway was conducted until the cable melted over the full width over a length of 1.5 to 3 mm.- The temperature was at least 1360 K to melt the copper in the cable.- Remarkably, at the moment of melt-down, the thermocouple in the busbar 15 mm from the hotspot only measured 50 K.

Sample 3BLNSBC = 21 mmRadd = 27 μΩI = 9 kAtrun = 13 s

Experiments to the effects of defected and consolidated LHC main bus splices are conducted in three phases

1. Thermal runaways in interconnections with defectsAugust 2009 – Februari 20105 quadrupole busbar samples in test station FRESCAGoal: Validation of model → safe operating current before consolidation

2. Proof of principle of consolidation with shuntsMarch 2010 – June 20104 quadrupole busbar samples in test station FRESCAGoal: Validation of model → Proof of principle of the consolidation proposal

3. Final validation of the consolidation with shunts in a realistic test setupMarch 2010 – October 20102 dipole busbar samples inbetween two special SSS magnetsGoal: Validation of model → safe operating current before consolidation

11

Technology Department

Gerard Willering – Splice review – 18 October 2010 - CERN

Content

12

Technology Department

Gerard Willering – Splice review – 18 October 2010 - CERN

First try: Discontinuity of the copper was not guaranteed due to solder creep in the voids.

Second try: Discontinuity guaranteed by cutting away part of the stabilizer

Important parameters for shunts:- Thickness of the shunt- Non-soldered shunt length (see with white arrows).

Shunt preparation

13

Technology Department

Gerard Willering – Splice review – 18 October 2010 - CERN

Sample 4 3 mm thick shunts

11 m

m5

mm

7 m

m0

mm

Sample LNSS (MM)

SHUNT THICKNESS(MM)

Radd before shunt

(T=300 K)(ΜΩ)

RADD SHUNTED

(T=300 K)(ΜΩ)

4A-3mm 7 & 5 3 51 & 39 3.9 & 3.14A-1.5mm 7 & 5 1.5 51 & 39 6.4 & 5.44B-3mm 11 & 0 3 27 & 0 2.8 & -0.54B-1.5mm 11 & 0 1.5 27 & 0 5.1 & -0.1

Shunts reduced to 1.5 mm thickness

Sample 4 without shunt

Sample 4 with shunt

Shunt preparation

14

Technology Department

Gerard Willering – Splice review – 18 October 2010 - CERN

0

10

20

30

40

50

6 8 10 12 14 16 18 20 22

Ru

naw

ay ti

me

(s)

Current (kA)

1 3A

2A

3B2B

100

1000

10000

6 8 10 12 14 16 18 20

MII

Ts

unt

il ru

naw

ay (1

06

A2 s

)

Current (kA)

1

3A

2A

3B

2B

-Runaway time for the shunted samples much higher than for non-shunted samples.

- All the shunted samples can carry 13 kA for more than 24 seconds.

- The same data, but the MIITs are calculated (kA2*s) - The shunted samples with 1.5 and 3 mm thick shunts can handle the MIITs of 15.5 kA with τ = 20 s.- These samples do not have the worst case parameters and not the worst case conditions. Therefore no direct conclusions for LHC conditions.

Result on measurements with shunts

Experiments to the effects of defected and consolidated LHC main bus splices are conducted in three phases

1. Thermal runaways in interconnections with defectsAugust 2009 – Februari 20105 quadrupole busbar samples in test station FRESCAGoal: Validation of model → safe operating current before consolidation

2. Proof of principle of consolidation with shuntsMarch 2010 – June 20104 quadrupole busbar samples in test station FRESCAGoal: Validation of model → Proof of principle of the consolidation proposal

3. Final validation of the consolidation with shunts in a realistic test setupMarch 2010 – October 20102 dipole busbar samples inbetween two special SSS magnetsGoal: Validation of model → safe operating current before consolidation

15

Technology Department

Gerard Willering – Splice review – 18 October 2010 - CERN

Content

Goal: Test in realistic conditions of a worst case scenario, with a non-soldered shunt length of 8 mm and low RRR values.- 2 Special SSS spare magnets are connected to the testbench in SM18.- In total 35 meter of RQ busbar and 35 meter of RB busbar. - Two instrumented RB (M3) interconnections.- No magnets in the test-circuit

16

Technology Department

Gerard Willering – Splice review – 18 October 2010 - CERN

Quadrupole lines

Dipole lines

Que

nch

stop

per

Quadrupole lines

Dipole lines2 interconnections

2 interconnections

Preparation of final validation test

Test conditions are rather special:

- First time 2 magnets in serie on the test-bench -> Test bench elongation- The quench needs to be stopped between M3 and M1 line.

- Additional copper strips (Lyra) for cooling- Large, 30 liter reservoir for helium

- Instrumentation wire feed-through-box - Heating power of more than 300 W for long time with significant loads on cryogenic system

17

Technology Department

Gerard Willering – Splice review – 18 October 2010 - CERN

Preparation of final validation test

U-profile/wedge

High precision measurements on resistance are important for the validation of shunt and model.- In the test the U-profile/wedge have a low RRR- In the tests the shunts have a much lower RRR than foreseen for the LHC conditions since they are not annealed

18

Technology Department

type RRR Typical LHC valueU-profile RQ 174, 176, -, - > 200U-profile RB 182, - > 200Shunt RB 156, 156, 160, 160 > 300Busbar RQ 252, 264, - , - > 200 (lab tests)Busbar RB 258, 303, - > 200 (lab tests)

shunt shunt

Gerard Willering – Splice review – 18 October 2010 - CERN

RRR measurements

Thermal runaway measurements on the interconnection with the largest non-soldered length (8 mm).

19

Technology Department

10000

11000

12000

13000

14000

15000

16000

17000

18000

0 2 4 6 8 10 12

Safe

curr

ent [

A]

lwc = Distance between contact [mm]

Safe operating current for a shunted RB joint, assuming an infinitely long non-stabilized cable on both sides of the joint; tau=100 s, RRR_shunt=100

16 mm2, RRR_bus=100

16 mm2, RRR_bus=160

32 mm2, RRR_bus=100

32 mm2, RRR_bus=160

Arjan Verweij, 18/1/2010

type Calculations chamonix SM-18 sampleRRR shunt 100 160RRR busbar 160 > 250Shunt cross-section 32 mm2 45 mm2

Cooling Adiabatic Superfluid helium

What to expect in the test conditions?I > 16 kA @ τ = 100 s

Gerard Willering – Splice review – 18 October 2010 - CERN

Expected results

Figure from A.Verweij (chamonix 2010 workshop and first splice review)

Current cycles for thermal runaway measurements at 1.9 K.

20

Technology Department

0.0

2.0

4.0

6.0

8.0

10.0

12.0

14.0

16.0

0 50 100 150 200 250 300

Curr

ent (

kA)

time (s)

LHC 7TeV

LHC design

SM-18 test

SM-18 limit

LHC cycle: 13 kA, tau = 100 s Very quick recovery of the normal zone

Gerard Willering – Splice review – 18 October 2010 - CERN

Current cycles for test

Test cycle: 14 kA, τ = 100 sTest cycle: 14 kA, τ = 140 sTest cycle: 14 kA for 22 s, then τ = 140 s. Still no signs of thermal runaway in the most critical shunt!!

Therefore we went to constant currents of 13 and 14 kA (power supply limit).

21

Technology Department

Gerard Willering – Splice review – 18 October 2010 - CERN

Q9 Q8

Quench behavior at a constant current of 13 kA

- No significant heating of the interconnection in 180 s.- No significant heating in the busbar Q9-1 in 180 s.- Normal zone does not enter the Q8 busbars.

- Very stable conditions at 13 kA in busbar and interconnection!!!

Q9-busbar have an RRR ≈ 250 → R9.4meter ≈ 2.3 µΩ at 10 KQ8-busbars have an RRR ≈ 300 → R16.5meter ≈ 3.5 µΩ at 10 K

22

Technology Department

Gerard Willering – Splice review – 18 October 2010 - CERN

Q9 Q8

Quench behavior at a constant current of 14 kA

- Small temperature increase in the interconnection in 85 s.- The full 35 meter of busbar between the quench-stoppers become normal- Accelerated heating effect in busbars Q8 and Q9-2.

- Limitation factor is not the shunted interconnection, but the busbar.

Q9-busbar have an RRR ≈ 250 → R9.4meter ≈ 2.3 µΩ at 10 KQ8-busbars have an RRR ≈ 300 → R16.5meter ≈ 3.5 µΩ at 10 K

- In the straight section the busbars are encapsulated in a G10 casing and close to each other. - In the region closer to the interconnection superfluid helium is available for cooling.

23

Technology Department

Gerard Willering – Splice review – 18 October 2010 - CERN

Temperature profiles in the shunts at 13 and 14 kA

- Additional proof of thermally stable conditions with measurements by two thermocouples in the shunts of Interconnection 1.

Interconnection 1

Interconnection 2

T3 T4

I = 13 kA I = 14 kA

Quench propagation velocity – dipole busbar

24

Technology Department

- No propagation below 12 kA (with the busbar cooled by superfluid helium).

- Arjan’s calculation (RRR = 200) is a bit more optimistic than the measurements (RRR 250 - 300).

0

0.2

0.4

0.6

0.8

1

1.2

1.4

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Prop

agati

on s

peed

(m/s

)

I (kA)

Dipole busbar propagation velocity

dipole RRR 100 (calc)

dipole RRR 200 (calc)

V15-V18 (dipole)

V25-V27 (dipole, lyra)

Gerard Willering – Splice review – 18 October 2010 - CERN

Quench propagation velocity

Quench propagation velocity – quadrupole busbar

25

Technology Department

- No propagation below about 9 kA (with the busbar cooled by superfluid helium)

- At higher currents the measured velocity might be overestimated since the temperature and therefore the resistance can be increased.

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 160

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2 Quadrupole busbar propagation velocity

quad RRR 100 (calc)quad RRR 200 (calc)V32-V31 (quad)V5-V29 (quad, lyra)

I (kA)

Prop

agati

on sp

eed

(m/s

)

Gerard Willering – Splice review – 18 October 2010 - CERN

Quench propagation velocity

26

Technology Department

MIITs >> 30000 kA2s at 13 kA (no sign of thermal runaway) MIITs > 18000 kA2s at 14 kA (Start of thermal runaway in busbars)(LHC 13 kA, 100 s – MIITs = 8500 kA2sLHC 11.8 kA, 100 s – MIITs = 6800 kA2s)

Gerard Willering – Splice review – 18 October 2010 - CERN

Simple and short conclusion: The proposed shunts work!

27

Technology Department

Gerard Willering – Splice review – 18 October 2010 - CERN

Summary and Conclusion

Thermal runaways in interconnections with defects- Clear proof of the damage a defect can have with the melted sample.- Measurements provided largely sufficient experimental data for model validation (by A. Verweij).- Conclusions on safe current/energy cannot be drawn directly from this measurements, since test conditions are different from machine conditions.

Proof of principle of consolidation with shunts- Clear improvement of the thermo-electric stability by applying shunts on the samples with defects.- Boundary conditions of the test-station prohibit direct conclusions on the stability of the consolidated interconnection, but indicates that the principle good.- Sufficient experimental data for model validation (by A. Verweij).

Final validation of the consolidation with shunts in a realistic test setup- A consolidated interconnection with a copper shunt having a cross-section of 45 mm^2, a double defect in the interconnection, a non-soldered lenght of 8 mm and an RRR of 160 is more stable than the busbar itself in the straight section. - In the condition a quench starts in the interconnection itself a continuous current of 13 kA does not show any sign of a thermal runaway in the first 180 seconds.- At a continuous current of 14 kA provokes an excellerated temperature increase in the encapsulated part of the busbars, with a temperature of about 40 K after 85 s.- In terms of thermo-electrical stability the shunt is overdesigned.