Cold mass integration of the ATLAS barrel toroid magnets at CERN
The ATLAS Barrel Toroid, part of the ATLAS Detector built at CERN, is comprised of 8 coils symmetrically placed around the LHC beam axis. The coil dimensions are 25 m length, 5 m width and 0.4 m thickness. Each coil cold mass consists of 2 double pancakes of aluminum stabilized NbTi conductor held in an aluminum alloy casing. Because the magnet is conduction cooled a good bonding between the superconducting winding and the coil casing is a basic requirement. Due to the high load level induced by the Lorentz forces on the double pancakes, a pre-stressing technique has been developed for the assembling of the double pancake windings in the coil casing. This prestressing technique uses inflatable bladders made of extruded aluminum tubes filled with glass microballs and epoxy resin then cured under pressure. The paper describes the design of the system as well as the problems occurred during the assembling of the 8 superconducting ATLAS coils and the ATLAS B0 prototype coil, and the behavior of the Barrel Toroid coils with respect to this prestress during the cold tests
IEEE TRANSACTIONS ON APPLIED SUPERCONDUCTIVITY, VOL. 16, NO. 2, JUNE 2006 553
Cold Mass Integration of the ATLAS Barrel Toroid
Magnets at CERN
Jean-Michel Rey, Michel Arnaud, Christophe Berriaud, Romain Berthier, Sandrine Cazaux, Alexey Dudarev,
Michel Humeau, René Leboeuf, Jean-Paul Gourdin, Christophe Mayri, Chhon Pes, Herman ten Kate, and
Abstract—The ATLAS Barrel Toroid, part of the ATLAS De-
tector built at CERN, is comprised of 8 coils symmetrically placed
around the LHC beam axis. The coil dimensions are 25 m length,
5 m width and 0.4 m thickness. Each coil cold mass consists of 2
double pancakes of aluminum stabilized NbTi conductor held in an
aluminum alloy casing. Because the magnet is conduction cooled a
good bonding between the superconducting winding and the coil
casing is a basic requirement. Due to the high load level induced
by the Lorentz forces on the double pancakes, a pre-stressing tech-
nique has been developed for the assembling of the double pancake
windings in the coil casing. This prestressing technique uses inﬂat-
able bladders made of extruded aluminum tubes ﬁlled with glass
microballs and epoxy resin then cured under pressure.
The paper describes the design of the system as well as the prob-
lems occurred during the assembling of the 8 superconducting
ATLAS coils and the ATLAS B0 prototype coil, and the behavior
of the Barrel Toroid coils with respect to this prestress during the
Index Terms—Magnet, pressure effects, superconducting coils.
HE BARREL Toroid Magnet is part of the Magnet System
of the ATLAS Detector for the LHC , . It consists of
eight ﬂat superconducting coils extending over a length of 26
meters with an inner bore of 9 meters and an outer diameter of
20 meters and producing a toroidal magnetic ﬁeld. The B0 coil
,  which is a shortened racetrack coil, having a length of
only 9 meters is a technological prototype using all the previous
R&D efforts to manufacture it. In order to ensure the reliability
of the magnet an innovative process for the pre-stressing of the
windings in the coil casing has been developed at DAPNIA/
STCM Saclay . This process is described as well as the result
of its use during the manufacturing of the B0 prototype and the
8 barrel toroid coils.
At an early stage of the magnet design mechanical calcu-
lations showed that the shear stress appearing between the
impregnated superconducting winding and its aluminum sup-
porting structure (named “coil casing”) were far above the shear
limit of the epoxy resin. This stress conﬁguration has been con-
sidered as very unsafe for a conduction cooled magnet storing
Manuscript received September 19, 2005. This work is part of the ATLAS
Detector Program on the LHC and was supported by the ATLAS Collaboration.
J.-M. Rey, M. Arnaud, C. Berriaud, R. Berthier, S. Cazaux, M. Humeau,
R. Leboeuf, C. Mayri, C. Pes, and P. Védrine are with the CEA Saclay—DSM/
DAPNIA, 91191 Gif sur Yvette, France (e-mail: firstname.lastname@example.org).
A. Dudarev and H. ten Kate are with ATLAS Magnet Team at CERN, CH
1211, Geneva 23, Switzerland.
Digital Object Identiﬁer 10.1109/TASC.2006.870838
high energy. It led to the development of the pre-stressing
technique using high pressure epoxy ﬁlled inﬂatable aluminum
tubes named “bladders”.
Although the process consists mainly in putting strain in one
part of the structure, the expression “pre-stressing”, being more
usual, will be used in the article.
Each cold mass coil consists of two double pancakes of in-
sulated conductors in a coil casing. The conductor is a pure
aluminum stabilized cable obtained by a co extrusion process.
After winding the double pancakes are insulated with epoxy
resin using a vacuum impregnation technique. As the conductor
carries a 20 500 A current, each double pancake takes 2.46
amp-turns in a magnetic ﬁeld gradient of 4.85 T. The Lorentz
forces appearing on the winding compress it as the magnet is en-
ergized. The indirect cooling technique requires a good bonding
of the double pancake and the coil casing.
Avoiding shear stress at the double pancake/coil casing in-
terface is possible if the central part of the coil casing is put in
tension during assembling. Then as the magnet is energized, the
tension in the coil casing is released by the magnetic compres-
sion of the double pancake, but shear stress does not appear at
The technique used to create the pre-stress is to use inﬂatable
bladders on each side of the superconducting winding. Some of
them are bent to be placed in the corners of the pancake wind-
ings. The bladders are ﬁlled with micro balls of glass and im-
pregnated with liquid epoxy resin, then pressurized and cured
under pressure. Once ﬁlled with liquid resin and pressurized the
bladders act as hydraulic jacks, putting the double pancake coils
in compression and the coil casing in tension. After curing the
epoxy resin under pressure the bladders work like compressed
shims due to the tensile strain stored in the coil casing. Fig. 1
shows a section of the cold mass with the magnetic ﬁelds and
compressive stress values appearing on the winding due to the
ﬁeld and to the bladders.
Major parts have been subcontracted in the European industry
and shipped to CERN for assembling .
One of the critical problems in the assembly scenario was the
handling of the 25 m
5 m double pancake winding, each of
them weighting 7500 kg with a bending displacement lower than
2 mm/m over the coil length. To solve this problem the handling
operations have been made using the coil casing as part of the
1051-8223/$20.00 © 2006 IEEE
554 IEEE TRANSACTIONS ON APPLIED SUPERCONDUCTIVITY, VOL. 16, NO. 2, JUNE 2006
Fig. 1. Upper part: magnetic ﬁeld and its induced strain and bladder induced
strain as a function of the conductor location along the coil casing width (along
X axis). Lower part: cross section of the ATLAS race-track coil. The bladders
are located on each side of the conductor windings. The bladders of the lower
winding are not represented.
The assembling scenario follows this sequence:
— The double pancake windings are covered with glass ﬁber
cloth pre-impregnated with slow reactivity epoxy resin
— The coil casing is put in place over one double pancake,
and both parts are provisionally clamped together.
— The provisional assembling is turned over.
— The bladders and shims are inserted between the coil
casing and the double pancakes.
— A complementary pre-preg layer is put in place on the
double pancake winding.
— The casing covers are put in place over the double pancake.
— The operations are repeated for the second double pancake.
The pre-stressing operations can start at this point.
Fig. 2 illustrates the turn over operations.
Several years of R&D activity ,  have been necessary
to optimize the process and qualify the ﬁlling materials for the
bladders. The efﬁciency of the bladder has been deﬁned as the
ratio of the measured force applied by the bladder to the calcu-
lated force applied by an hydraulic jack having the same ex-
ternal section as the bladder. This has been measured and is
83%. A 500 mm long full section mock up has been realized to
control the elongation of the coil casing during the integration
process. It allowed a test of the complete pre-stressing sequence,
Fig. 2. A cold mass comprising one coil casing and two double pancake wind-
ings during a turn over operation.
Fig. 3. View of a bladders on top of a double pancake inserted in the coil casing.
The cut bladder has been inﬂated and is already 3 times thicker than before
inﬂation. One bladder is already inserted in the gap between the coil casing and
the double pancake winding.
the qualiﬁcation of the hydraulic jacks for the resin injection,
and a measurement of the coil casing elongation.
The pre-stress sequence is the following:
— The bladders are inﬂated with gaseous nitrogen to 20 bars
(2 MPa) to expand and ﬁll the gap between the double
pancake and the coil casing.
— The gaseous nitrogen is released and the bladders are ﬁlled
with glass micro balls having a diameter between 0.45 mm
and 0.85 mm.
— Once ﬁlled the bladders are provisionally sealed.
The assembly process continues until the previous operations
are completed on both double pancakes for one coil casing.
REY et al.: COLD MASS INTEGRATION 555
— The complete injection tubing is assembled to link the 106
bladders for one cold mass (altogether 846 high pressure
unions per coil).
— The thermal protection is put in place to allow the curing
of the epoxy resin.
— The cover pressing system is installed. It comprises load
plates and rubber bladders linked together and inﬂated with
Nitrogen to 5 bars (0.5 MPa).
The pre-stressing sequence can then start:
— The remaining gas is evacuated from the bladder at a pres-
sure of 5 mbars.
— The bladder is impregnated with epoxy resin. The pressure
applied on the resin is increased to 2 MPa for two hours
in order to impregnate completely the micro balls in the
— The pressure is then increased to 12 MPa by steps of
2 MPa. The tubing and the connections are checked at
each step to prevent leakages.
— The system is cured under pressure.
Pressurizing of the liquid resin has been achieved using a
high pressure oil pump to pressurize twinned resin jacks. The
high pressure oil is injected in one jack that pushes on a second
coaxial jack ﬁlled with epoxy resin. In order to cure the magnet
once under pressure the winding is powered to generate Joule
heating. The current, up to 430 A, goes in the high purity alu-
minum stabilizer, allowing a maximum temperature of 125
for the full hardening of the epoxy resin.
HE B0 EXPERIENCE
Despite the efforts made to achieve a safe process for the B0
prototype, several leaks occurred from the bladders during the
integration. There is no way back to dismantle the magnet once
the epoxy resin is injected in the bladders and the magnet was
cured. The bladder manufacturing technique was reanalyzed af-
terwards to improve the operational safety. The changes are
summarized in Table I. Nevertheless the mean elongation mea-
sured on the B0 coil casing at the places where the bladders had
not been leaking was greater than the speciﬁed value of 0.2 mm.
Table II gives the elongation of the coil casing as measured in
the location with good and leaking bladders. The conclusions
reached at that time were that even with leaking bladders an
elongation is induced in the structure. It is shown in Table II that
0.13 mm is the mean value of elongation at the location of the 3
leaking bladders. This fact has been attributed to the compaction
of the glass micro balls under the effect of gravity that does not
allow the bladder to ﬂatten again once put under pressure.
OIL CASING ELONGATION MEASURED IN B0 AND BT
Fig. 4. A completely assembled cold mass during the curing under pre-stress
The tests of the B0 magnet were done during the winter 2000
and proved to be very successful , the 40 ton magnet suffered
altogether around 40 deliberately induced quenches without sig-
niﬁcant degradation proving the efﬁciency of its design.
HE BT EXPERIENCE
The integration of the 8 cold masses for the ATLAS Barrel
Toroid took place during the year 2003. The pre-stressing op-
erations went ﬁne, only one among the total of 848 bladders
leaked at 50 bars.
The measured values of the coil casing elongation are given
in Table II. In the case of BT the mean elongation represents the
mean value of 21 points of measure. The pre-stressing of the
8 coils winding is homogeneous although the measurement of
the elongation shows a fairly large scatter between the different
coils. As at least 3 weeks took place between the initial (before
pre-stressing) and ﬁnal (after curing of the coil) width measure-
ments the temperature change is the most probable explanation
of this scatter. A speciﬁc tool to perform the measurement has
been developed but it has been available only for coil 4 and later.
Nevertheless the precision is better, the standard deviation being
less than 30% of the mean elongation measured. The elongation
of coil 6 was not determined because the measurement of the
coil width before pre-stressing had been forgotten.
556 IEEE TRANSACTIONS ON APPLIED SUPERCONDUCTIVITY, VOL. 16, NO. 2, JUNE 2006
On B0 the coil casing was a thick monolithic aluminum piece
made of plates welded by electron beam welding. This was a
critical process and has been replaced on BT by a bolted as-
sembling of thinner plates assembled using TIG welding. The
change of manufacturing technique can also explain the lower
elongation of the BT coil casing compared to B0.
A result of the technique developed was that the bladders
being inﬂated in place were able to withstand fairly large
dimensional changes from one impregnated double pancake
to the other. No other mechanical device would have allowed
matching the size and shape dispersion of the impregnated
winding as easily as the inﬂatable bladder technique.
The tests of the 8 magnets conﬁrmed the efﬁciency of the
process, no one of them showed training –.
An innovative and original process has been used for the as-
sembling of the cold masses of the ATLAS Barrel Toroid. It uses
inﬂatable aluminum bladders ﬁlled with glass micro balls and
epoxy resin pressurized at 120 bars during curing. The prob-
lems arising during the integration of the B0 prototype have
been solved for the Barrel Toroid coils.
The developed technique allows matching the dimensional
changes of the impregnated windings and therefore does not
need precise machining on large pieces. Even in the case of leaks
appearing in one of the bladders a pre-stress was generated in the
structure. With only one bladder leaking over 848 pressurized
during the integration process of the 8 ATLAS Barrel Toroid
coils the technique developed is considered safe efﬁcient and
The authors express their warm thanks to all the staff mem-
bers of JINR, Technicatome, CERN and DAPNIA having cured
the 8 coils during long overnight shifts.
 A. Daël et al., “Progress in the design of the barrel toroid magnet for
the ATLAS experiment and associated R&D at CEA-Saclay and INFN-
Milano,” Proc. 15th Int. Conf. Magnet Technology Science press, pp.
 H. H. J. ten Kate, “ATLAS superconducting toroids and solenoid,”
IEEE Trans. Appl. Supercond., vol. 15, no. 2, pp. 1267–1270, 2005.
 A. Dael et al., “Construction of the ATLAS B0 model coil,” IEEE
Trans. Appl. Supercond., vol. 11, no. 1, pp. 1597–1600, Mar. 2001.
 ——, “Synthesis of the technological developments for the B0 model
coil and the ATLAS barrel toroid coils,” IEEE Trans. Appl. Supercond.,
vol. 10, no. 1, pp. 361–364, Mar. 2000.
 J. M. Rey et al., “Prestressing concepts and related materials qualiﬁca-
tions for the ATLAS barrel toroid coil,” Adv. Cryogenic Eng. Mater. ,
vol. 46, pp. 65–72, 2000.
 P. Vedrine et al., “Manufacturing and integration progress of the
ATLAS barrel toroid air core magnet at CERN,” IEEE Trans. Appl.
Supercond., vol. 14, no. 2, pp. 491–494, Jun. 2004.
 J. M. Rey et al., “Epoxy resin developments for the ATLAS and CMS
superconducting magnets impregnation,” Adv. Cryogenic Eng. Mater.,
vol. 42, pp. 37–42, 1996.
 N. Dolgetta et al., “Review of the ATLAS B0 model coil test program,”
IEEE Trans. Appl. Supercond., vol. 14, no. 2, pp. 495–504, 2004.
 A. Dudarev et al., “First full size ATLAS barrel toroid coil successfully
tested,” presented at the Applied Superconductivity Conf., Jacksonville,
USA, Oct. 2004.
 C. Berriaud et al., “On-surface test of the ATLAS barrel toroid coils:
acceptance criteria and results,” in this conference.
 A. Dudarev et al., “On-surface test of the ATLAS barrel toroid coils:
overview,” in this conference.