Quench statistics Outline: Reminder of the facts about natural quenches in the main dipole circuit in sectors 4-5 and 5-6 observed quench characteristics and propagation of quenches quench behaviour in the tunnel vs. SM18 quench data symmetric quench propagation phenomenon What can be done to speed up the quench training in the machine A. Siemko and E. Todesco
Quench statistics . Outline: Reminder of the facts about natural quenches in the main dipole circuit in sectors 4-5 and 5-6 observed quench characteristics and propagation of quenches quench behaviour in the tunnel vs . SM18 quench data symmetric quench propagation phenomenon - PowerPoint PPT Presentation
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Quench statistics
Outline:
Reminder of the facts about natural quenches in the main dipole circuit in sectors 4-5 and 5-6
observed quench characteristics and propagation of quenches
quench behaviour in the tunnel vs. SM18 quench data
symmetric quench propagation phenomenon
What can be done to speed up the quench training in the machine
A. Siemko and E. Todesco
MAC 23 Meeting 13/o6/2008 A. Siemko
Quench Training, Retraining, Memory Effect,…
Reminder of the “jargon” used
Extract of Natural Training Quenches at 1.8K to Reach Ultimate Field of 9 Tesla
8.0
8.5
9.0
9.5
Quench Number
Mag
netic
Fie
ld a
t Que
nch
B [T
esla
]
QuenchRetraining
MemoryEffect
MAC 23 Meeting 13/o6/2008 A. Siemko
Quench Training in Sectors 4-5 and 5-6
0 5 10 15 20 25 309000
9500
10000
10500
11000
11500
12000
5.340
5.540
5.740
5.940
6.140
6.340
6.540
6.740
6.940S45S56
Quench number
Que
nch
curr
ent
[A]
Corr
espo
ndin
g en
ergy
[Te
V]
MAC 23 Meeting 13/o6/2008 A. Siemko
Training quench characteristics
Typical current decay curve
dI/dt+10A/s
-100 A/s
MAC 23 Meeting 13/o6/2008 A. Siemko
HYDRAULIC BEHAVIOUR DURING A QUENCH
18 bar
2 min.
3 h
Pressure build-up
Pressure discharge
MAC 23 Meeting 13/o6/2008 A. Siemko
Example of a training quench and quench propagation
Natural quench in A22R4 at 9859 A (magnet name 3176) 4 magnets quenched (3 after quench propagation) Sequence of events:
Magnet Cryogenic cell Local time t quench [s] I quench [kA] E [MJ]A22R4 21R4 16:50:34.947 0 9.859 4.957B22R4 21R4 16:51:24.679 49.732 6.011 1.843C22R4 21R4 16:52:07.532 92.589 3.829 0.748C21R4 19R4 16:52:41.798 126.855 2.644 0.357
Total 7.905
MAC 23 Meeting 13/o6/2008 A. Siemko
What can be expected from series tests in SM18 Powering to Nominal Field of 8.33 T
0
50
100
150
200
250
300
350
400
450
500
550
0 1 2 3 4 5 6 7 notreachedNumber of quenches to reach 8.33T
Num
ber o
f Cry
o-di
pole
s 01 02 03
Histogram of the number of quenches to reach 8.33 Tesla for 1252 LHC dipoles
during 1st powering 38 % of MB magnets has reached the nominal field without quench
42 % of MB magnets has required 1 quench to reached the nominal field
MAC 23 Meeting 13/o6/2008 A. Siemko
What can be expected from series tests in SM18
A. Siemko and P. Pugnat
Thermal Cycle performed on ~10 % of MB magnets
The number of quenches that can be expected in the tunnel can be estimated although the sample is not “entirely” random
MAC 23 Meeting 13/o6/2008 A. Siemko
What can be expected from series tests in SM18
Retraining data for magnets submitted to a TC in Sector 4-5
• From thermally cycled magnet sample {TC} the reduction of the average number of quenches to reach I nominal (11850A) and I commissioning (12000A) is equal to ~ 82%
• Assuming the same reduction for other magnets {NoTC} Average number of Training Quenches
per sector can be calculated
Magnets tested virginMagnets tested after
thermal cycle
82% reduction of number of
quenches to go to nominal
1232
115 115
?
MAC 23 Meeting 13/o6/2008 A. Siemko
Extrapolation method
Additional hypotheses: No “problematic” MBs in the machine after series tests in SM18 No long time relaxation of the trained magnets No retraining for magnets submitted in SM18 to a thermal cycle No important detraining ? Estimation for detraining:
Aim: critically review the quench data taken at SM18 to see if they partially justify the 5-6 results
Montecarlo analysis with pessimistic hypotheses, more than what used the previous method
A method based on extrapolation:
We take the first virgin quench of all magnets of 5-6 (available for all magnets)
We sum the correlation between 1st quench after thermal cycle and 1st virgin quench measured in 136 magnets, split per firm We take the correlations that are available, i.e. that ones of poorly
quenching magnets (pessimistic hypothesis) This correlation is affected by a random part that must be taken into
account one needs a MonteCarlo
The method gives only the first quench for each magnet Up to now, all 5-6 quenches were in different magnets
MAC 23 Meeting 13/o6/2008 A. Siemko
MONTECARLO ANALYSIS
MonteCarlo results versus sector 5-6 HC data Qualitatively is fine: a lot of Noell, a few Ansaldo, no
Alstom Now, after 27 Noell quenches, we are below of around
400-600 A
8
9
10
11
12
0 20 40 60 80First quench number
Cur
rent
(kA
)
AlstomAnsaldoNoellAnsaldo HCNoell HC
5-6: Montecarlo vs hardware commissioning
MAC 23 Meeting 13/o6/2008 A. Siemko
MOTECARLO ANALYSIS
Noell magnets lost some memory, but not completely
8
9
10
11
12
0 20 40 60 80First quench number
Cur
rent
(kA
)
Noell MonteCarloNoell HCNoell virgin
5-6
MAC 23 Meeting 13/o6/2008 A. Siemko
TRAINING EFFECT MEASURED IN SM18
In general magnets gain current from 1st virgin quench to 1st quench after thermal cycle The gain the larger when the virgin quench is lower
Noell shows some anomalous behavior It is the only manufacturer that has some magnets
with a detraining loss
-2
-1
0
1
2
3
4
8 9 10 11 12 13 141st virgin quench (kA)
1st q
uenc
h at
c - 1
st v
irgin
que
nch
(kA
)
AlstomAnsaldoNoell
MAC 23 Meeting 13/o6/2008 A. Siemko
TRAINING IN SM18 VS. TRAINING IN SECTOR 5-6
… the detraining loss looks worse in the sector 5-6 data
-2
-1
0
1
2
3
4
8 9 10 11 12 13 141st virgin quench (kA)
1st q
uenc
h at
c - 1
st v
irgin
que
nch
(kA
)
NoellNoell HC
MAC 23 Meeting 13/o6/2008 A. Siemko
Phenomenon of Symmetric Quenches
In sector 5-6 five symmetric quenches were observed after quench propagation caused by a thermo-hydraulic wave One quench (in B16.R5 at ~7.4 kA) has developed the high “MIITs” and
resulting high hot spot temperature
There is a weakness in the magnet protection.
MAC 23 Meeting 13/o6/2008 A. Siemko
ANALYSIS AND FUTURE STRATEGY
Development of “tools” to speed-up training is undergoing
Overshooting with current to have more training quenches at the same time Requires development of dedicated electronics to prevent diodes
overheating and to control maximum number of quenching magnets Forced quench training
Tried on 4th June (3 quenches at the same time) and on 5th June (4 quenches) but not conclusion yet
MAC 23 Meeting 13/o6/2008 A. Siemko
Conclusions
During the high current quenches in MB magnets of RB45 and RB56 circuits:
all individual systems (PC, PIC, QPS, EE and CRYO) performed as designed
typical quench propagation time from magnet to magnet was observed to be in the rage of 40-60 seconds – slightly slower as compare to the initial estimates (better).
Statistical analysis of the quench data taken at SM18 do not explain the quench behavoiur of dipoles in sectors 4-5 and 5-6
Much more pronounced than expected detraining effect was observed in a lot of Noell magnets.
Critical review the production data of the magnets can help to understand the quench behaviour observed in sector 5-6
MAC 23 Meeting 13/o6/2008 A. Siemko
Conclusions
There is a weakness in the magnet protection in case of symmetric quenches
Remedy is under study
Much more pronounced than expected detraining effect was observed in a lot of Noell magnets.
Critical review the production data of the magnets can help to understand the quench behaviour observed in sector 5-6
“Tools” to speed-up training of magnets in the machine are under development
MAC 23 Meeting 13/o6/2008 A. Siemko
Additional information (quench data)Sector # Date Time Magnet Cold mass Magnet ID