Superconducting Magnetsfor the LHC
Lucio Rossi
CERN – TE department
CERN Accelerator School
26 February 2009 - Divonne
Lucio Rossi - Superc. Mag. 2
A few references
Basic Superconductivity:M. Tinkham, Superconductivity, Gordon & Breach Publisher
A.C. Rose-Innes, E.H. Rhoderick, Introduction to Superconductivity, Pergamon Press
W. Buckel, Superconductivity, Fundamental and Applications, VCH Publisher
J. Evetts (editor), Coincise Encyclopedia of Magnetic and Superconducting Materials, Pergamon Press
H.W. Weber High Tc Superconductivity, Plenum Press
Applied Superconductivity M.N. Wilson, Superconducting Magnets, Clarendon Press Oxford
H.A. Brechna, Superconducting Magnet Systems, Springer Verlag
K.-H. Mess, P. Schmüser, S. Wolff, Superconducting Accelerator Magnets, World Scientific
B. Seeber (editor), Handbook of Applied Superconductivity, IoP Publishing
L. Dresner, Stability of Superconductors, Plenum Publ. Corp.
Y. Iwasa, Case Studies in Superconducting Magnets, Plenum Publ. Corp.
Previous CERN Academic Training• 1999 : Ph. Lebrun - Superfluid Helium
• 2000 : L. Rossi - Superconducting Magnets
• 2002 : D. Larbalestier - Superconducting Materials
CAS School on SC and cryogenics for accelerator and detectors, Erice 2002,
Report CERN 2003-
Lucio Rossi - Superc. Mag. 3
Content
• Accelerators and magnet shape• Accelerator Magnets : basic design, magnet types• The heart: superconductors (and superconductivity)• Mechanical structure• Making it suitable for LHC: Field Quality• Alignment issues• Snapshot at construction in Industry• Quench and stability
Lucio Rossi - Superc. Mag. 4
Circular accelerator: magnet festival
E 0.3 B R
Tev, T, km
LHC:
18 km MB
2.5 km MQ
2 km of
other
Quads
> 8000 Sc
Magnets!!!
Lucio Rossi - Superc. Mag. 5
Rationale for using SC
• The LHC has a circumference of 26.7 km, out of whichsome 20 km of main superconducting magnets operating at 8.3 T. Cryogenics will consume about 40 MWelectrical power from the grid.
If the LHC were not superconducting:
• If it used resistive magnets operating at 1.8 T(limited by iron saturation), the circumferencewould have to be about 100 km, and the electrical consumption 900 MW (a good-size nuclear power plant), leading to prohibitive capital and operation costs.
Lucio Rossi - Superc. Mag. 6
Accelerators and Magnet Shape
• The cost and the difficulties scales as the amount of the stored energy: Vol B2/2 0
• Rush for high fieldsmall volume of field
• Thin field tubes that follow the particle trajectories
• Dipole to bend• Quadrupoles to focus• Sextupoles and Octupoles• Correctors (from Dip to Decap)• Correctors are mostly local
Detector Magnets
Lucio Rossi - Superc. Mag. 7
• Magnetic fields needed for - electric charge identification- momentum spectrometry- p = mv = q B; = q/p B L
BL is often the comparison parameter
• If momentum analysis is done by tracking inside the field volume:
- p/p 1/BL2 large volume better than high field - Field homogeneity appreciated but NOT critical (field knowledge of 0.1% usually suffices)
Lucio Rossi - Superc. Mag. 8
Accelerator Magnets Basic Design - I
In practice the above current distributions are approximate, so the field contains also higher order harmonics (see later).It can be shown that if the cos(n ) is approximate by step function, there is a “magic” angle that makes nil the first higher order harmonics.
Intersecting ellipses generate
uniform field.
Two intersecting ellipses, rotated
of 90 , generate a perfect
quadrupole fields:.
ba
JbdB y
0
xba
baJB
yba
baJB
y
x
)(
)(
0
0
All these configurations follow:
Js = J cos( ) , Js = J d cos(2 ), …
d = coil thickness
a, b, ellipses parameter
Lucio Rossi - Superc. Mag. 9
Uniform shell current density with
cut to nul the first higher order
harmonic.
Approximation of cos with coil blocks (left) and multiple shells (centre) and of intersecting ellipses (from Wilson book).
DEVIATION FROM PERFECT FIELD: 10-4 (0.01%) that is taken as the unit.
Accelerator MagnetsBasic Design - II
Lucio Rossi - Superc. Mag. 10
Accelerator MagnetsBasic Design - IV
Reality
e.m. forces
NOT SELF-
SUPPORTING
How to contain
them
More difficult in
twin magnets!
Joverall 500 A/mm2 ! e.m. forces are not kept by conductors but tend to torn apart the winding.
Principle
coil-collars-cylinder structure
Lucio Rossi - Superc. Mag. 11
Outer Yoke
Force
Inclined
Force
Collar/Yoke
Forces
Shrinking
Cylinder
Iron Yoke
Packing rods
Austenitic Steel
Collars
Collaring rods
Coils
Ft=340 tonnes
• The approach chosen by LHC is to have a design where most of the force can be taken by collars
• Then to have a vertical gap almost closed at RT and the outer skin welded such as to have a strong prestress
• Part of this is lost during cool-down but there is enough to assure azimuthal prestress of the coil up to 9 T and to have at least radial contact radial between collar and yoke
• The choice of stainless steel has released the tolerances on the skin stress: ideal 150 15 MPaWe can live even if all tolerance goes in the too low or too high direction.
Lucio Rossi - Superc. Mag. 14
• Cu-NbTi cable @ 1.9 K. Gop = 223 T/m• e.m. forces kept by collars only (iron is a mere flux return yoke)
• Two-in-one concept (two apertures are de-coupled both magnetically and mechanically).
• 3.5 m long, coil ap.56 mm, straight, alignment given by inertia tube• Correctors are fixed on the inertia tube
MQ
Main Quad
CERN- CEA
collaboration
MQ X-sect
Lucio Rossi - Superc. Mag. 15
MQY wide aperture quadrupole
• Four-layer, graded shell coil.
• Free standing collars, fully
supporting the forces.
• Two-in-one iron yoke.
70 mm ID coil
G = 160 T/m at 4.5 K
I = 3620 A
E = 141 kJ/m/aperture
Lmag = 3.4
Lucio Rossi - Superc. Mag. 16
CERN-Oxford Inst. (MQY)
Fermilab (MQXB)
KEK (MQXA)
Three 70 mm quadrupoles developed for the LHC IRs
Lucio Rossi - Superc. Mag. 17
The zoo of the 6000 SC corrector magnets
MCS Sextupole Magnets
MOOctupoleMagnets MQSXA
QuadrupoleOctupoleSextupoleMagnets
MCDODecapoleOctupoleMagnets
Superconductivity
• Zero resistance !!
• But at low temperature
• And within certain limit of field and current
• Concept of critical surface
Lucio Rossi - Superc. Mag. 18
f3(J,B,T0) f2(J,T,B0)
f1(B,T,J=0
)
J
B
T
Lucio Rossi - Superc. Mag. 19
Who provide the current: type II superconductors
Flux penetration in the material is in quanta:
= h/2e 2 10-15 Wb
B
J
F
Lorentz force : Fp = -Jc B : to avoid
movements and heating it is needed a pinning
given by defects.
NbTi: Fp max 15 GN/m3 (or 15 N/mm3 !!) Jc
3 GA/m2 (3000 A/mm2 ) at 5 T
Lucio Rossi - Superc. Mag. 20
Practical Materials
Long journey from material discovery to
magnet application
Criterion Number
Superconducting 10,000
Tc 10 K .and. Bc2 10 T 100
Jc 1 GA/m2 @ B > 5 T 10
Magnet-grade superconductor 1
From science to technology
Lucio Rossi - Superc. Mag. 23
Niobium-Titanium
Critical surface of NbTi (from Wilson textbook)
Critical current of best Cu/NbTi with
typical 3 T field shift at superfluid
helium (INFN-LASA lab, february 2000)
Critical current density vs field measured on
NbTi multiflamentray wire at 4.22 and 2.17 K
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
11000
0 1 2 3 4 5 6 7 8 9 10 11 12
Btot (T)
Jc(A
/mm
2)
LHe HeII
10
100
1000
10000
0 5 10 15 20 25 30 35 40 45
Applied Field (T)
JE
(A/m
m²)
YBCO Insert Tape (B|| Tape Plane)
YBCO Insert Tape (B Tape Plane)
MgB2 19Fil 24% Fill (HyperTech)
2212 OI-ST 28% Ceramic Filaments
NbTi LHC Production 38%SC (4.2 K)
Nb3Sn RRP Internal Sn (OI-ST)
Nb3Sn High Sn Bronze Cu:Non-Cu 0.3
YBCO B|| Tape Plane
YBCO B Tape Plane
2212
RRP Nb3Sn
Bronze
Nb3Sn
MgB2
Nb-TiSuperPower tape used
in record breaking
NHMFL insert coil 2007
18+1 MgB2/Nb/Cu/Monel
Courtesy M. Tomsic,
2007
427 filament strand
with Ag alloy outer
sheath tested at
NHMFL
Maximal JE for
entire LHC NbTi
strand production
(CERN-T. Boutboul
'07)
Complied from
ASC'02 and
ICMC'03 papers
(J. Parrell OI-
ST)
4543 filament High Sn
Bronze-16wt.%Sn-
0.3wt%Ti (Miyazaki-MT18-
IEEE’04)
Domain of iron
dominated
magnets
Je vs field for all practical superconductor
Lucio Rossi - Superc. Mag. 25
E-J curve
Transition at fixed temperature: V = k In , so we have to adopt a criterion to define Ic.
Electric field. Ic is the current generating an electric field Ec= 10-5 V/m E = Ec (J/Jc)n
Resistivity. Ic is the current showing an apparent resistivity of c = 10-14 m.
The exponent n, called also n-value or n-index, is related to the homogeneity of the
material or of the superconducting properties. For good superconductors n 30 – 60 or
more. Near critical surface, B > 0.9 Bc2 the n-values drops down to 20 or below.
8T
0
5
10
15
20
0 200 400 600 800 1000 1200
Jsc (A/mm2)
E(µ
V/m
)
c =10-14
m
Ec=10 µV/m
5T
1
10
100
100 1000 10000
log Jsc
log
Ec =10
-14
Ec=10 µV/m
Lucio Rossi - Superc. Mag. 26
Superconductors are not stable!
I
T TLHe
Iop
Superconductors are NOT stable against perturbation albeit very small.
E of J are enough to drive superconductor normal!
Heat capacity drops at low temperature (T<< TDebey) :
C T3 T = E/ C. So even small E generates sensible T
operating point of the magnet beyond critical surface QUENCH
TEMPERATURE
CURRENT
Ic curve
Imagnet
All current in sc
Current
in Cu
Current
in sc
TC
S
TC
All current in
Cu
TEMPERATURETCTCS
JOULE
POWER
Electrodynamic stability: intimate
contact between the superconductor and a
good conductivity material.
Adiabatic (or intrinsic)stability: to cure
the flux rearrangement that generates heat
Direct cooling : LHe and more HEII are
very good coolant, capable to remove
heating in milliseconds! Latent heat 10-
1000 times that of solid specific heat!
Lucio Rossi - Superc. Mag. 28
-250
-200
-150
-100
-50
0
50
100
150
200
250
-1 -0.5 0 0.5 1
B (T)
M (
mT
)
shielding current layer in a superconducting
filament (persistent currents)
M 1/Bn
n =
0.5...1.5
M JcDeff
Courtesy of L. Bottura
Magnetization and fine filaments
Lucio Rossi - Superc. Mag. 29
+ I - I
B
m
Multipoles generated areproportional to 1/B
n
If M is symmetric only allowedmultipoles are present
Courtesy of L. Bottura
Effect of magnetization
Lucio Rossi - Superc. Mag. 30
Magnetization for LHC NbTi
Strands:Width of the magnetization loop of all manufacturers (all 01, 02 billets)
15
20
25
30
35
40
45
50
1 51 101 151 201 251 301 351 401 451 501 551 601 651 701 751 801 851 901 951100110511101115112011251130113511401145115011551Strand identifier (ordered sequentially)
Wid
th o
f th
e m
agnetization loop (
mT
)
01B
01E
02B02C
02G 02K
02D
Cable Magnetization (CERN)Control limits are TDB strand limits
20
21
22
23
24
25
26
27
28
29
30
B10
001
B10
009
B10
017
B10
025
B10
033
B10
041
B10
049
B10
057
B10
065
B10
073
B10
081
B10
089
B10
097
B10
105
B10
113
B10
121
B10
129
B10
137
B10
145
B10
153
B10
161
B10
169
B10
177
B10
185
B10
193
B10
201
B10
209
B10
217
B10
225
B10
233
B10
241
B10
249
B10
257
B10
265
Map Id
Wid
th o
f th
e m
ag
neti
zati
on
lo
op
(m
T)
at
T=
1.9
K, B
=0.5
T
01B cable
Due to field
imperfection
generated by M:
Rejection of
conductor
Limit in the dynamic
range of the magnets
In LHC Dfil= 6-7 m
Courtesy of S. LeNaour
Lucio Rossi - Superc. Mag. 31
dB/dt
resistive
contact Rc at
cross-over
point
induced eddy currents
in the loop I -dB/dt
and I 1/Rc
superconducti
ng path in the
strands
Needs for 10-20 kA cable for
protection
Needs very high packing factor: 90% !!
Needs a system simple that keep
strands The strand are fully
transposed
BUT field changes
over a period !
Ends problems
Junctions
BICC
Rutherford cable
Lucio Rossi - Superc. Mag. 32
Controlling the contact resistance
CERN has developed
the controlled
oxidation method
Coating wire with 0.3-
0.5 of SnAg then H.T.
cable in air
What are the acceptable
limits ?
Too low (< 15-20) gives
field errors (ad He
consumption
Too high (>100-200) may
raise instability or current
distribution
Rc measured by CERN on the cables for the inner dipole layer
0
20
40
60
80
100
120
140
160
180
200
B10001
B10034
B10067
B10100
B10133
B10166
B10199
B10232
B10265
B10298
B10331
B10364
B10397
B10430
B10463
B10496
B10529
B10562
E00005
E00039
E00072
E00105
E00139
Rc [
µO
hm
]
average Rc = 42
µohm for the
production of
ALSTOM cables
average Rc =
33 µohm for
Lucio Rossi - Superc. Mag. 33
Snap-back phenomena-1
.2-1
-0.8
-0.6
-0.4
-0.2
0
0 500 1000 1500 2000time (s)
b3 (
units)
0.05 A/s
0.5 A/s0.25 A/s
0.1 A/s
•during the injection platform of an LHC dipole, the
sextupole error term integrated over the magnet
length decays with a time constant of ~ 1000sec.
•when the ramp is started, the error term 'snaps back'
to its earlier value
•it is difficult to find a time constant of 1000 seconds
in any of the usual coupling modes
-20
-10
010
20
30
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Field (T)
Magnetization (m
T)
Injection + 20 mT
- 20 mT
•BICCs produce a field component which alternates
20 mT along the magnet
•imagine the hysteresis curve of NbTi filaments
subjected to this oscillation
•a 20 mT increase produces very little change in
filament magnetization
•a 20 mT decrease produces a large change in filament
magnetization
•thus the hysteresis curve acts as a 'rectifier' enabling
the oscillating BICCs to produce a dc level
After Rob Wolf
Lucio Rossi - Superc. Mag. 34
LHC MB X-sect: conductor (Rutherford cable)
Conductor
position
optimization:
Control of harmonics
Balance of margin
among blocks
Stable against
inevitable errors
Minimum shear among
conductors
Balance between T
margin of inner/outer
Lucio Rossi - Superc. Mag. 35
Cu Spacer
Precise at 20 m
Used to steer
production
Change of Cu wedge
0.2-0.5 mm of inner
wedges in July 2001
(3 CM, 15 CC).
Effect in 2002.
~35 CM old Xsect.
LHC MB X-sect: copper wedges
Profile II
-80
-60
-40
-20
0
20
40
Batc
h P
RO
T
Batc
h A
Batc
h 1
Batc
h 2
Batc
h 3
Batc
h 4
Batc
h 5
Batc
h 6
Batc
h 7
Batc
h 8
Batc
h 9
Batc
h 1
0
Batc
h 1
1
Batc
h 1
2
Batc
h 1
3
Batc
h 1
4
Batc
h 1
5
Batc
h 1
6
Batc
h 1
7
Batc
h 1
8
Batc
h 1
9
Batc
h 2
0
Batc
h 2
1
Batc
h 2
2
Batc
h 2
3
Batc
h 2
4
Batc
h 2
5
Batc
h 2
6
Batc
h 2
7
Batc
h 2
8
Batc
h 2
9
Batc
h 3
0
Batc
h 3
1
Batc
h 3
2
Batc
h 3
3
Batc
h 3
4
Batc
h 3
5
Batc
h 3
6
Batc
h 3
7
m P1out
P2out
Lucio Rossi - Superc. Mag. 36
Polyimmideinsulation
Around cable and
around coils
Important
elements are
dimensions, 3%
of thickness, and
creep (Apical
creeps less than
kapton)
Rutherford Cables InsulationGround isolation
Four layers 125AH
LHC MB X-sect: conductor and ground Insulation, Interlayer
Inter
layer
To allow
HEII to
flow
-2 layers of Apical 200 AV insulation
-1 layer Pixeo to glue cables together at 185°C (-0,+5 critical)
Lucio Rossi - Superc. Mag. 37
LHC MB X-sect: insulated CBT
CBT
Cold bore tube
StSt tubes
Insulation done at
CERN
Special Insulation
technique > 20 kV
Clearance between
coils and insulated
CBT is about 0.5 mm
over the 15 m length
Lucio Rossi - Superc. Mag. 38
LHC MB X-sect: Quench Heater
Strips of stainless
steels partially
coated with copper
to adjust resistance
Encapsulated like a
sandwich in two
foils of 75 m of
polyimide
Fired by current
pulse, heat must
diffuse from strip to
coils in 20-50 ms !!
Lucio Rossi - Superc. Mag. 39
LHC MB X-sect: Collars
Collars and collaring are the main controllers of the final coil shape
Lucio Rossi - Superc. Mag. 40
LHC MB X-sect: magnetic insert
Introduced to
ease the
mechanical
assembly
It serves for FQ
By tapering we
cured unwanted
quadrupole and
octupole
components
The iron InsertPunched
together with
the yoke
lamination
Lucio Rossi - Superc. Mag. 41
The iron yoke:
Stray field
15% field
increase (but
big gain in
protection)
If saturates
affect FQ
(sextupole)
Trim of
magnetic
length
LHC MB X-sect: yoke laminations
One supplier for
the steel 45,000
tons
Precise vertical gap
Temperature probe
Regular
Nested
Lucio Rossi - Superc. Mag. 42
LHC MB X-sect: Bus Bars & fillers
Bus Bars
Quad BB
Focusing
Quad BB
Defocusing
Corrector
circuit bus
bars on top
of Quad BB
Dipole
circuit BB
160 km of
main
BusBars!!
We provide:
-technology
-SC 02 cables
-Polyimmide
foils and tapes
Lucio Rossi - Superc. Mag. 43
LHC MB X-sect: Shrinking cylinder and support
Two half
shells,
welded on
the magnet
Many
difficulties
Curvature
released from
1 to 2.5
mm: still not
OK.
To c ure this
we went to
sorting
Precise support
Lucio Rossi - Superc. Mag. 44
LHC MB X-sect: beam screen and HXT
Copper Heat Exchange Tubes
HEII satur.
Beam
Screen
Inserted at
CERN just
before
insertion in
the tunnel
Lucio Rossi - Superc. Mag. 45
3D
Inner layer lyre side
End Spacers: critical for Quench
Two slightly different design (in cable
profile)
Only pre-series (4-5 suppliers can do)
Lucio Rossi - Superc. Mag. 52
LHC Main Dipole -end partCorrector Magnets (spool pieces)
Assembly
in CMAs
is purely
mechanic
al
(tolerance
s of B axis
wrt mech.
frame
given by
supplier
Lucio Rossi - Superc. Mag. 56
Interconnection between two superconducting magnets
6 superconducting bus
bars 13 kA for B, QD, QF
quadrupole
20 superconducting bus
bars 600 A for corrector
magnets (minimise
dipole field harmonics)
42 sc bus bars 600 A for corrector
magnets (chromaticity, tune, etc….)
+ 12 sc bus bars for 6 kA (special
quadrupoles)
13 kA Protection
diodeTo be connected:
• Beam tubes
• Pipes for helium
• Cryostat
• Thermal shields
• Vacuum vessel
• Superconducting
cables
Lucio Rossi - Superc. Mag. 57
Critical ProcessWinding-Curing-Coil formation
• Coils are cured under press
• Then measured all along 15 m
• Then collared with shims
• Shims influence also prestress and then coil movements (quench)
• Shift of radius of tens of micron as well deformation can easily drive harmonics out of tolerance
Lucio Rossi - Superc. Mag. 58
Steering production (and check assembly) Field Measurements - CERN supply
It has also helped to detect a number of defects.
It has also been used to detect subtle electrical shorts
Introduced first to steer the FQ toward beam dynamics targets.Note the uniformity among different manufacturers
10.07
10.09
10.11
10.13
10.15
0 200 400 600 800 1000Magnet progressive number
Int
transf
func (
Tm
/kA
)
-40
-20
0
20
40
Units
Firm 1
Firm 2
Firm 3
AT-MAS
Cold mass
upper limit for single magnet (3 sigma)
lower limit for single magnet (3 sigma)
0
5
10
15
20
0 200 400 600 800 1000
Magnet progressive number
Tota
l num
ber
of fo
und d
efe
cts
Lucio Rossi - Superc. Mag. 59
Stering production with actions
-10
-5
0
5
10
0 100 200 300 400 500 600 700 800 900 1000 1100
Magnet progressive number
b3 inte
gra
l (u
nits)
all
Firm 1
Firm 2
Firm 3
Cold mass
upper limit for systematic
lower limit for systematic
AT-MAS & MTM
Cro
ss
-se
cti
on
2
Cross-section 3
Lucio Rossi - Superc. Mag. 60
W W
V2
V1
V2
V1
97.26
R 281236
LINE V1
R 281236
97.26
R 281236
LINE V2V2
2
V1
1
14343 LINE V1
14343 LINE V2
9.14
2
V1/V2
V1/V2
Ø 0.6
V1/V2
Ø 2
3 D: curvature measurementsEnd tolerances
Lucio Rossi - Superc. Mag. 61
LTD 500Mechanical mole
Target
Accuracy:10ppm at 2on static target
BUT 0.2 mm when changing position
Measuring Instruments : Laser tracker (Leica) and moles
Lucio Rossi - Superc. Mag. 63
The longitudinal welding
• Pre-developed at CERN
• Installed directly CMAs
• Two weldings synchronized
• Root welding STT: high quality very sophisticated control, a world PRIMA for this conditions and austenitic steel
• Problem on the press, now almost over : still quality of welding
• Each CM leak tested 26 bar !!!
Lucio Rossi - Superc. Mag. 64
Magnet performance and Training Curve
Second Generation Dipoles MBL1N1, MBL1AJ2, MBL1AJ1, MBL1N2, and MBP1A1(all training quenches at superfluid helium)
6.00
6.50
7.00
7.50
8.00
8.50
9.00
9.50
10.00
Ma
gn
etic
Fie
ld a
t Q
uen
ch B
[T]
MBL1N1 MBL1AJ2 MBL1N2
Th
. C
ycle
Fa
st
Th
. C
ycle
Th
. C
ycle
Th
. C
ycle
MBP1A1MBL1AJ1
Th
. C
ycle
Lucio Rossi - Superc. Mag. 65
The spectrum of disturbances
Continuous Distributed Perturbances:
AC losses (hysteretic and coupling losses, eddy currents)
Intrinsic dissipation due to smooth transition (Iop too near or above Ic!)
Thermal load (vacuum degradation,…). This could be a serious effect in
cryocooled system.
These perturbations are usually predictable and estimate must be done at design level
coupling losses can depends on interstrand resistance, i.e. on manufacture
technique and on prestress and e.m. forces more difficult to evaluate
Continuous Point Perturbances:
Joints inside coils
Release of mechanical energy (hysteresis of the stress-strain relation)
Localised heat input (suspension rods with bad thermal anchoring)
These effects are well understood and predictable (it does not mean easy to cure !)
Lucio Rossi - Superc. Mag. 66
The spectrum of disturbances - II
Transient Distributed/Point Perturbances
Flux jumps. This effect is cured almost definitely for NbTi. Effects
could be seen on NbSn with very high current density and very large effective
filament diameter. This effect can be detected at low field, during current ramp.
Mechanical origin: movements, friction, sudden release of elastic
energy…
crack in the resin
Basically these last two mechanism are now understood, in principle, and
acoustic emission experiments did prove it almost visually.
Still they are less predictable and more difficult to avoid. They depend on
magnet geometry, material properties, local conditions and on many details.
They can depend on magnet history (previous quench, overheating, thermal
induced stress, etc.)
Lucio Rossi - Superc. Mag. 67
Temperature and enthalpy margins
The main action to take is to have a reasonable energy margin, larger than the
expected energy release, to make unlikely to pass the critical surface: but the
specific heat of solids are pretty low near LHe and we can rely only on T 1-2 K
I
T TLHe
Iop
Volumetric Specific Heat
0.E+00
2.E+04
4.E+04
6.E+04
8.E+04
1.E+05
1.E+05
1.E+05
0 5 10 15 20T [K]
H [
J/(
m3K
)]
Al
NbTi
Cu
G10
Lucio Rossi - Superc. Mag. 68
Point Disturbances : MPZEnergy density is not the only criterion, since most of the perturbations are localized.
l
T
T0
Tcs
Conductor length
MPZ : the Minimum Propagating Zone
with a simple balance between power
dissipated in the normal zone and heat
conducted along the cable we found:
2
)(2
c
opcs
J
TTkl
If there is no stabiliser, only NbTi, we see that l 1 m H 1 nJ only !!
Lucio Rossi - Superc. Mag. 69
Stability of LHC
1
10
100
1000
10000
0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1
4.3K 6T H16
4.3K 6T H8
4.3K 6T H1
1.9K 9T H16
1.9K 9T H8
1.9K 9T H1
If there is no stabiliser, only NbTi, we see that l 1 m H 1 nJ only !!
If each strand hehaved as single wire, H 1-10 J
Lucio Rossi - Superc. Mag. 70
PROTECTION
A superconducting magnet, whatever the stability margin, it will quench. And magnet
integrity has to be preserved.
When working at current density like in the LHC dipoles, where dissipation per unit
volume following a quench is JCu2 6 10-10 m 1018 A/m2 = 600 MW/m3
Excessive voltage rise insulation breakdown.
Temperature growth melting or serious trouble to insulators and conductor
Temperature gradients excessive stress with subsequent de-training.
Impressive damage caused by a short circuit developed during a quench in a LHC dipole protype
Lucio Rossi - Superc. Mag. 71
Hot Spot Temperature
Let’s suppose that heat is coming only by Joule effect and
conduction is not significant
)( )(
)()( )()()( 2
0
0
22
md
T
T
TUTJdTT
TCdttJdTTCdtTtJ
m
op
The function U(T) is a computable a
priori, based only on material properties.
If the magnet is discharged on an external
–dumping- resistors, RD, Td=0.5 Lmag/RD.
U vs T
0
1
2
3
4
5
6
7
8
9
10
0 50 100 150 200 250 300
T [K]
U [
10
16A
s /m
4]
LHC dipole
ATLAS
MIITs
The goal is to speed up the quench
propagation by any means, to
avoid too high hot spots:
1) Heater : activated in 20 ms !!
2) Benefit of quench-back
This goes against having LHe
inside the coils (i.e. is against
stability)!
Lucio Rossi - Superc. Mag. 72
Protection scheme for a dipole string
LHC protection scheme (courtesy of F. Rodriguez Mateos , CERN)
Lucio Rossi - Superc. Mag. 73
Quench statistics (single magnet test)
Only 2.5% rejected
Half for quench half for electrical
NCs (QH)
Except 1, all repaired successfully