Performance of the 1-m Model of the 6 kA Superconducting Quadrupole for the LHC Insertions

Abstract

The LHC dispersion suppressors and matching sections will be equipped with individually powered superconducting quadrupoles with an aperture of 56 mm. In order to minimise the cost of the powering circuits, the quadrupole has been designed on the basis of an 8 mm wide NbTi Rutherford-type cable for a nominal current of 5300 A, corresponding to a gradient of 200 T/m at 1.9 K. In order to validate the design options a model magnet program has been launched. In this report we describe the construction features of the first 1-m long magnet, and present its training performance and the results of protection studies.

Full text

Page 1

PERFORMANCE OF THE 1-M MODEL OF THE 6 kA
SUPERCONDUCTING QUADRUPOLE FOR THE LHC INSERTIONS
J. Lucas, L. Bottura, H. Dariol, R. Ostojic, S. Sanfilippo, A. Siemko,
F. Sonnemann, D. Tommasini, I. Vanenkov, CERN, Geneva, Switzerland
Abstract
The LHC dispersion suppressors and matching sections
will be equipped with individually powered superconduct-
ing quadrupoles with an aperture of 56 mm. In order to
minimise the cost of the powering circuits, the quadrupole
has been designed on the basis of an 8 mm wide NbTi
Rutherford-typecablefora nominalcurrentof5300A,cor-
responding to a gradient of 200 T/m at 1.9 K. In order to
validate the design options a model magnet program has
been launched. In this report we describe the construction
features of the first 1-m long magnet, and present its train-
ing performance and the results of protection studies.
1 INTRODUCTION
In order to increase the flexibility and performance of
the collider and to decrease the cost of the powering in-
frastructure, the LHC dispersion suppressors and match-
ing sections will be equipped with individually powered
6 kA superconducting quadrupoles.
ture quadrupolewas designedusinga previouslydeveloped
8.2mmwideRutherford-typecable, andhasa nominalgra-
dient of 200 T/m at 1.9 K and 160 T/m at 4.5 K [1]. In
order to validate the magnet design a model program com-
prising two 1 m long single aperture magnets and a twin
aperture magnet was recently launched. In this report, we
describe the construction of the first single aperture mag-
net, and present its training performance, as well as the re-
sults of magnet protection studies.
The 56 mm aper-
2 MAGNET CONSTRUCTION
The 1 m longsingle aperturequadrupoleconsists of a stand
alone collared coil, assembled in a vertically split iron
yoke, Fig. 1. The coils were wound using an 8.2 mm
Rutherford-type cable with 34 strands (strand diameter
0.48 mm, filament diameter 10 µm), a mid thickness of
0.84 mm and a keystone angle of 1.05 degree. The cable
has a copper-to-superconductor ratio of 1.3, and a critical
current density of 3100 A/mm2at 4.2 K and 5 T. The main
parameters of the magnet are given in Table 1.
One trial and four production coils were wound in two
layers from a single cable with the layer jump in the lead
side. Ineachlayertherearetwoconductorblocksseparated
by a copper wedge. The coils were moulded at 185oC and
high pressure. After moulding the coils were measured un-
der a pressure of 50 MPa, and their size compared to the
nominal coil size. In Fig. 2, the size difference of all five
coils is shown in eight different positions along the coil.
Figure 1: Cross section of the magnet.
Table 1: Properties of the magnet at 1.9 K and nominal
current
Nominal gradient
Nominal current
Peak field in coil
Current density in copper
Current density in superconductor
Number of turns in inner/outer layer
Inductance
Fyper octant
Fxper octant
200 T/m
5165 A
6.31 T
1455 A/mm2
1891 A/mm2
15/21
4.55 mH/m
-427 kN/m
306 kN/m
The coils are on the average oversized by 17 µm, with a
dispersion of 31 µm. The winding and curing procedures
are therefore repetitive and well under control.
The magnet is protected by two sets of strip heaters. The
inner heaters are placed in between the two layers and in-
duce a quench in 28 turns of a coil at a time. The outer
heaters are placed on top of the coil and cover17 turns. Al-
though there was considerable concern about the integrity
of the inner heaters during winding and curing of the coils,
all inner heaters showeda highlevel of dielectric insulation
in all phases of coil production. Two of the fourproduction
coils were instrumented with two spot heaters, located in
the high and low field regions, and voltage taps for deter-
mining the hot spot temperatures. Additional voltage taps
were added to the instrumented coils to localise the origin
of the quenches.
2145Proceedings of EPAC 2000, Vienna, Austria

Page 2

-60
-40
-20
0
20
40
60
80
100
0
123456789
Average error
Error in size [µm]
Coil position
Figure 2: Coil size difference along the coil. Positions 1-4
and 5-8 correspond to two opposite sides of the coil with
1 and 5 at the lead end. Error bars correspond to min-max
values for all measurements.
The coils were assembled in a self supporting collared
structure. The austenitic steel collars are held together with
full length keys welded to both end-plates in order to par-
tially react the longitudinal magnetic forces. The keying
was performed in several passes using a collaring press
with four collaring rams and four keying rams displaced
by 45 degrees. As shown in Fig.3, the collar assembly was
compressed at the four poles over 200 mm to allow key
bending during their gradual insertion. In trial collaring
tests nodamageofthekeysorthecollarslotswas observed.
Figure 3: Collaring of the magnet. Full length keys are
pushed in over 100 mm with four 20 tonne rams while the
four60tonnecollaringramscompresstheassemblyateach
pole over 200 mm.
The coils were instrumented with capacitive gauges for
measurement of the azimuthal pre-stress. The measure-
ments were done at collaring and at cold, and showed good
uniformity between the four poles. After collaring, the av-
erage value of the pre-stress in the inner and outer layers
was 110 MPa and 80 MPa and dropped due to creep after
one week to 104 MPa and 75 MPa, respectively. These
values are compatible with the average oversize of the
coils (Fig. 2).
After completion of the pole connections, the self-
supporting collared coil was finally assembled in the ver-
tical split iron yoke (Fig. 1). This arrangement is an adap-
tation of the structure normally used for testing the 1 m
long single aperture models of the LHC main dipole [2]. In
order to mount the collared coil, a ferromagneticinsert was
provided which centred the coil within the outer yoke and
its clamped shell.
3TEST RESULTS
3.1Magnet training
The magnet was tested twice with a thermal cycle in be-
tween. The tests started by magnet training at 1.9 K and
4.4 K, Fig.4.The first quench occurred at 5285 A
(203 T/m), slightly above the nominal current in the LHC.
After the second quench the current went above 6 kA, and
reached the conductor limit of 6900 A (270 T/m) in 10
quenches. At 4.4 K no training was observed, the quench
current being stable above 5300 A (205 T/m). After a ther-
mal cycle, the first quench at 1.9 K occurred at 5466 A
(210 T/m), showing a weak memory effect. However the
training proceeded much faster and the magnet reached its
conductor limit after only four quenches. The seven initial
training quenches in the second test were performed with a
full energy dump in the magnet. Small detraining was ob-
served near the conductor limit related to the peak temper-
ature above 240 K. A number of quenches were performed
with the bath temperature slightly above 4.4 K in order to
confirm the conductor limit.
3000
3500
4000
4500
5000
5500
6000
6500
7000
7500
510 15 2025 30 3540
120
140
160
180
200
220
240
260
280
Current [A]
Field gradient [T/m]
Quench Number
1.9 K first cycle
4.4 K first cycle
1.9 K second cycle
4.40 K to 4.54 K sweep
Nominal current 1.9K
Conductor limit 1.9 K
Nominal current 4.4 K
Conductor limit 4.4 K
Figure 4: Training of the magnet at 1.9 K and 4.4 K.
Most training quenches at 1.9 K were located in the re-
gion of the layer jump. This is a mechanically weak point,
which in addition corresponds to the location of the peak
field in the magnet. In the initial test, the first training
quenches occurred in coil number 2, while in the second
one they moved to coil 3. Measurements of the pre-stress
in the coils showed that the straight section is still com-
pressed, pointing to a weakness of the layer jump of struc-
tural origin. The quenches at the conductor limit were lo-
Proceedings of EPAC 2000, Vienna, Austria 2146

Page 3

cated in the straight section and in the coil ends, indicating
that the design of the end spacers was well optimised.
During the excitation, the coils lose pre-stress in both
layers with identical rate of 0.95 MPa/kA2. Nevertheless,
at nominal current the compression of the coil poles is still
more than 10 MPa in the inner and 25 MPa in the outer
layer.
3.2Protection studies
The efficiency of the magnet protection critically depends
on the heater delay times, which were independently mea-
sured for the inner and outer heaters. The measurement
results, Fig. 5, show that the heater delay decreases with
current from about 80 ms at low currents to about 20 ms at
the nominal current of 5100 A, similar to what was mea-
sured previously on this cable type [3]. Surprisingly, al-
though the inner heaters act on turns in higher magnetic
field, the heater delay times for the inner and outer heaters
are almost identical. This could partly be related to the fact
that the polyimideinsulation of the inner heaters is 100 µm
on each side of the stainless steel strip, while it is 75 µm
for the outer heaters.
0
20
40
60
80
100
0
100020003000 400050006000
Delay [ms]
Current [A]
Outer layer QH
Inner layer QH
Figure 5: Heater delay as a functionof current. The heaters
were fired with an initial power of 21 W/cm2and a time
constant of 112 ms.
The hot spot temperature in the magnet during a quench
was derivedonthebasis ofthe MIITSmeasuredaftera spot
heater provoked quench. The test was performed for cases
when the magnet was protected by prompt firing of all in-
ner or outer heaters, or of half of the inner or outer heaters.
The hot spot temperature for these cases is shown in Fig.6
as a function of magnet current. As expected, the protec-
tion scheme with only half of the outer heaters results in
the highest temperature, which nevertheless remains well
below 300 K. The hot spot temperature is reduced from
234 K to 200 K if all outer heaters are used, and by an ad-
ditional 20 K if the inner heaters are fired. The maximum
voltage to ground measured in these cases was 27 V and
20 V for protection with outer and inner heaters, respec-
tively. As the outer heaters protect the magnet sufficiently,
the useofinnerheatersseems unnecessary,especiallysince
they create considerable difficulties in coil winding.
40
60
80
100
120
140
160
180
200
220
240
2000 2500 30003500 4000 4500500055006000
Hot spot temperature [K]
Current [A]
All inner layer
All outer layer
Half inner layer
Half outer layer
Figure 6: Hot spot temperature as a function of current.
In order to study the effect of the additional heater delay
related to quench detection time for magnets installed in
the LHC tunnel, the firing of the outer heaters was gradu-
ally delayed in steps of 5 ms. The hot spot temperaturewas
foundto beproportionalto thedelay,with theproportional-
ity factor of 5 K/ms. A quench detection and filtering time
of 20 ms would therefore imply an additional temperature
increase in the hot spot of 93 K. A temperature of 293 K
was measured in the worst case.
4 CONCLUSION
A 1 m long single aperture model of the 6 kA supercon-
ducting quadrupole for the LHC insertions was built and
tested in CERN. The magnet features a two layer coil
wound using an 8.2 mm wide Rutherford-type cable with
34strandsanda copper-to-superconductorratioof1.3. The
two layers were wound with a single length of cable and
cured at the same time with the inner heaters in between.
The magnet was tested in two occasions separated by a
thermal cycle, and performed very well. All initial train-
ing quenches at 1.9 K were above the nominal gradient of
200 T/m in the LHC, and the magnet reached its conduc-
tor limit of 270 T/m in 10 and 4 quenches in the first and
second training tests, respectively. There were no training
quenches at 4.3 K. Protection studies showed that the mag-
net is sufficientlywell protectedwith theouterheatersonly,
so that coil winding can be significantly simplified.
REFERENCES
[1] J. Lucas et al., “Design and Construction of a 1-m Model of
the Low Current Superconducting Quadrupole for the LHC
Insertions”, IEEE Trans. Appl. Superconductivity, Vol. 10,
No. 1, pp. 150-153.
[2] N.Andreev et al., “The 1 m Long Single Aperture Coil test
Program for LHC” , Proc. EPAC-96, Vol. 3, pp. 2258-2260.
[3] M. Lamm et al., “Tests of the 70 mm Quadrupole for the LHC
low-β Insertions”, ASC’98, IEEE Trans. Appl. Superconduc-
tivity, Vol. 9, No. 2, pp. 455-458.
2147Proceedings of EPAC 2000, Vienna, Austria