IEEE/CSC & ESAS EUROPEAN SUPERCONDUCTIVITY NEWS FORUM, No. 19, January 2012
Submitted to ESNF Nov. 16, 2011; accepted Nov. 30, 2012. Reference No. ST286, Category 6
The published version of this manuscript appeared in IEEE Trans. Appl. Supercond. 22, (4200809) (2012)
1
2019808
The ITER Magnets: Design and Construction
Status
N. Mitchell, A. Devred, P. Libeyre, B. Lim, F. Savary and the ITER Magnet Division,
ITER Organization, Route de Vinon, 13115 St. Paul lez Durance, France

Abstract— The ITER magnet procurement is now well
underway. The magnet systems consist of 4 superconducting coil
sets (toroidal field (TF), poloidal field (PF), central solenoid (CS)
and correction coils (CC)) which use both NbTi and Nb3Sn-based
conductors. The magnets sit at the core of the ITER machine and
are tightly integrated with each other and the main vacuum
vessel. The total weight of the system is about 10000 t, of which
about 500 t are Nb3Sn strands and 250 t, NbTi. The reaction of
the magnetic forces is a delicate balance that requires tight
control of tolerances and the use of high-strength, fatigueresistance steel forgings. Integration and support of the coils and
their supplies, while maintaining the necessary tolerances and
clearance gaps, have been completed in steps, the last being the
inclusion of the feeder systems. Twenty-one procurement
agreements have now been signed with 6 of the ITER Domestic
Agencies for all of the magnets together with the supporting
feeder subsystems. All of them except one (for the CS coils) are
so-called Build to Print agreements where the IO provides the
detailed design including full three-dimensional CAD models.
The production of the first components is underway (about 175 t
of Nb3Sn strand was finished by July 2011) and manufacturing
prototypes of TF coil components are being completed. The
paper will present a design overview and the manufacturing
status.
Index Terms—Magnetic
magnets, Fusion reactors
confinement,
Superconducting
I. MAGNET REQUIREMENTS
T
A. Operational Performance
HE function of the ITER magnets, Fig 1, i s to form,
control and drive the tokamak plasma [1]. For this
purpose, the Toroidal Field (TF) coils will provide a s teady
field of 5.30T at a radius of 6.20m. The Central Solenoid (CS)
and Poloidal Field (PF) coils will be capable of 30000
inductively driven 15MA plasma pulses with a burn duratio n
of about 400s. Hybrid and non-inductive plasmas will also be
within the coil capabilities (where the plasma current duration
is extended by non-inductive current drive methods), with
burn durations up to 12000s. An extensive database of plasma
scenarios (coil currents and plasma details) has been simulated
and is used to confirm the coil capabilities [2]. The CS and PF
coils are desi gned also for feedback control of plasma
Manuscript received 12 September 2011.
disturbances at frequencies up to the 100Hz range.
In terms of the number of operational c ycles for other
conditions, the coil system is designed for 100 cooldown and
warm-up cycles, 1,000 TF charging cycles, 50 TF fast
discharges and 10 TF quenches. A fast discharge of the CS or
PF coils is les s onerous th an the ref erence inductive plasma
cycle and so no extra limits are imposed.
Achieving reliable quench detection for the coi ls [3] is
critical for successful operation, not only because o f the
obvious need to protect against coil damage but because false
positives will i nterrupt machine availability, especially of the
TF coils. The magnet quench detection system will be able t o
distinguish between superconductor quench and the
electromagnetic perturbations caused by a plas ma disruption
with a reliability of >90% for the CS, Correction Coils (CC)
and PF co ils and >99.9% for the TF coils. The quench
detection system of the coils (particularly the TF) must be
designed to allow possible tuning during commissioning as
experience with plasma behaviour is obt ained. The magnets
need not be ready f or further operation (i.e. completely
recooled) until 96hrs after a f ast discharge of the TF system.
In the event of a fast discharge of only CS , PF or CC, the
magnets will be re-cooled ready for further operation within
2hrs. In the event of a fast discharge of any or all of the
magnets, the Helium expelled from the system will be
collected and stored. There is negligible expulsion from CS,
PF or CC dur ing a fast discharge but, due to eddy current
heating of the radial plates, almost all the TF helium is
expelled as the temperature rises to about 50-60K.
The magnet system cooldown will not occur at an y time
faster than 1K/h and the maximum temperature difference
between inlet and outl et supplies will be 50 K . It will be
demonstrated by analysis that the maximum temperature
differences inside the modules and structure do not exceed 50
K. Cooldown from RT to 4K will be achieved within 30 days.
B. Tolerances
Tolerances are essentially in any real engineering
component and affect the ITER coils in three ways. Firstly, the
departure from ideal shapes and locations of the c oils
generates so called ‘error field’. Approximately, the allowable
departure from axisymmetry of either toroidal or poloidal field
is about 3x10-4 T [4]. The correction coils are placed and sized
so that they can correct the most significant error modes in the
poloidal field (up to mode 3 in either poloidal or toroidal
1 of 9
IEEE/CSC & ESAS EUROPEAN SUPERCONDUCTIVITY NEWS FORUM, No. 19, January 2012
2
2019808
directions). Thee two main sources of thee error field are the
unnpaired lengthhs of busbars that are req uired to conneect the
cooil terminals too the main cryyostat feeders,, though it is second
more significannt) arises f rom
m coil
orrder. The secoond source (m
poositioning errrors. Seconddly, the pe nalty for alllowing
increased toleraances is generrally that highh strength struuctural
a
A marrgin of
material is lost, replaced by weak filler material.
100% is allowedd in the desiggn between maximum alloowable
strresses and thoose calculatedd by stre ss annalysis for thee ideal
metric imperfeections. Thirdlly, the
geeometry, to alllow for geom
cooils and their connections must all fit togeether when installed
in the cry ostat, maintain minimum clearrances (generrally a
minimum gap oof 50mm is reequired betweeen componentts with
ween themselvves and
independent flexxing) during operation betw
alsso to surroundding componennts such as thee Vacuum Vesssel.
Detailed and self-consistennt tolerances have been deveeloped
foor all the magnet compoonents, considering these three
requirements annd the variouss manufacturinng operations. Costs
arre reduced by avoiding tighht tolerances and a balance has to
bee maintained between perfo
formance and latitude allow
wed to
suuppliers.
Fig. 1. Overview of IT
TER and the Magnet System
12-13T (for the TF and CS coils) and NbTi operation up to 6T
for the PF.
For any set of coils, the heat exchangger bath mayy be
reduuced from 4.22K to 3.8K too allow the caapability to c orrect
posssible faults or design/annalysis errorrs discoveredd in
operation (for example, an underestimatee of the nucclear
heatting or a drop in conductor current sharinng temperaturee).
E. Safety
Because of thee severe effectt of magnet faaults on the ovverall
macchine availabiility, and the difficulty off repair, theree are
ms built intoo the
mulltiple monitorring and prottection system
desiign. These incclude inherentt features, dettection/monitooring
systtems (that operate continnuously whille the coils are
charrged), and testing systemss (that are a pplied periodically
wheen plasma pullsing is interruupted or w henn the magnetss are
disccharged). Parrticular attenttion is paid to reducing the
probbability of potential casscade sequennces, where the
exisstence of an innitial fault inccreases the proobability of otthers
(forr example, heaat from a shoort degrades a protection baarrier
or increases locaal voltages), annd common mode faults where
seveeral componnents (due for example to a com
mmon
mannufacturing errror) have the same initial faault.
F
From the safetty viewpoint the magnets contain high foorces
andd high stored and mobile ennergy. So theree exists theoreetical
ms which incclude
poteential to im pact neighbouring system
tritiuum confinement barriers, either by hea ting or by imp
mpact.
Studdies have shoown that al l magnet faults are confined with
the magnet sysstem with thhe possible exception off an
unddetected coil quench (when >10kg of liqquid metal maay be
spraayed on to t he vacuum veessel). A sepparate safety class
quennch detectionn system has been includedd for the TF coils,
locaated outside the cryostatt in the Helium lines, fully
indeependent (pressure based) and separatedd from the noormal
quennch detection system, to preevent such an initiating evennt.
F. Operational Margins and Performannce Verificatioon
C. Nuclearr heating
The design deecision was made during the ITER Enginneering
Deesign Activityy (1993-2001) to limit the nuclear heatingg from
ma to 14kW, a level compparable
the reference 500MW plasm
wiith that of thhe AC losse s, in the ra nge 10-20kW when
avveraged over a 1800s plasm
ma cycle. Thiss is compatible with
the cable-in-connduit conducttor design witth a central cooling
chhannel, prevennts the nuclear heating domiinating the cryyoplant
sizze and assists in smoothing the cryoplant loads (as AC losses
cooncentrate in thhe ramp up annd ramp downn times when thhere is
noo nuclear heaating). At thee same time, some optionns are
avvailable to operate with exxtra nuclear heeat loads if nuclear
heeating predictiions turn out to be inaccuraate. This is achhieved
byy adjusting thhe cryoplant operating connditions and/or the
plasma pulsing rate (i.e. tradinng of nuclear vs AC losses)).
m cooling
D. Helium
The coils aree cooled withh supercritical Helium at ann inlet
mperature of 4.5K (correspponding to a heat exchangeer bath
tem
at 4.2K). This temperature prrovides good heat capacity of the
heelium flow annd matches booth Nb3Sn opeeration in the range
The coils have been designedd with supercoonducting marrgins
baseed on the resuults from the ITER model coil program inn the
A phase of the project, supplemented and updated by tests
EDA
on short conductoor samples. Thhe margins, inn terms of alloowed
degradation of thhe superconduuctor that cann occur beforee the
macchine operatioon is comproomised, are coonservative. They
for a 30% l oss of superconnductor area while
typiically allow fo
maiintaining the full operationn temperaturee margin of 0.7K.
Sincce superconduuctor degradattion is also geenerally associated
withh a more graddual supercondducting transittion (the so-c alled
n value), the fulll operation can be maintaained beyond this
leveel of degradatiion if the coil cooling is incrreased.
Margins also exist to c over uncertainnties in the coil
operating condittions, particullarly for nucclear heating and
plassma disruptionns.
G. Repair
Some in-situ repair possibillities are bu illt into the maagnet
ws:
systtem, as follow
Inn situ repair (in all case s hands-on acccess is requuired,
posssibly with locaal removal of the thermal shhield):
2 of 9
IEEE/CSC & ESAS EUROPEAN SUPERCONDUCTIVITY NEWS FORUM, No. 19, January 2012
3
2019808
Retightening of the CS precompression structure. A
loss of 5% of the initial prec ompression is required to
be within the design capability. Above this the situation
must be reviewed and operating window restrictions
may be imposed until retightening can be scheduled.
 Retightening of the TF Precompression rings. A loss of
5% of the initial pretension is r equired to be within the
design capability. A review is required if operation is
continued with a higher loss.
 Retightening of bolts on TF gravity support plates, CC
support plates and PF coil clamps
 Bridging of a failed PF double pancake
 Repair of He pipes and He inlets in the CS bore and at
TF and PF coil terminals
Removal (medium difficulty):
 Extraction of CS stack (no removal of feeders/coils)
 Removal and replacement of TF Precompression rings
(this requires removal of the CS)
 Removal of PF1
 Replacement of inner poloidal keys (partial removal of
CS and PF1/6 coils)
Removal (high difficulty):
 Removal of a TF coil
 Removal of PF2
To ease re moval of the CS (which may be required fo r
replacing the Precompression Rings) the upper connections of
the CS to the feeders are aligned to allow the CS s tack to be
lifted vertically once the joints are opened, without removal of
the feeders.
Because of the long timescale and hig h cost of
manufacturing replacement coils once manufacturing is
complete and tooling is dismantled, one spare CS module and
one spare TF coi l are included in the procurement. It is as a
result a requirement that all CS modules are interchangeable.
Certain large components can only be p laced in the lower
cryostat during assembly. Therefore, 3 spare precompression
rings and jumpers for the PF coil double pancake bridging are
stored below the tokamak assembly in case of need.

II. SYSTEM OVERVIEW AND INTEGRATION
The overall operational requirements of the magnets define
the basic functionality but obviously leave open many design
choices, many of which affect also the design of the rest of the
machine. The following section gives a brief summary of the
final design choices. More details are given in [5].
The magnet system for ITER consists of 18 TF coils, a CS
coil stack with 6 coil modules, six PF coils and 18 Correction
Coils (CC), Fig 1 and Table 1. The 18 TF coils determine the
basic toroidal segmentation of the machine and were chosen to
meet the requirements of access ports (both number and size)
and the ripple at the plasma edge.
Both CS and TF coils operate at high field and use Nb3Sntype superconductor. The PF co ils and CCs use NbTi
superconductor. All coils are cooled with supercritical Helium
with a co il inlet temperature of 4.5 K. The conductor is a
cable-in-conduit conductor with a c ircular multistage cable
consisting of about 1000 strands cabled around a small central
cooling spiral tube. The cable is contained in a circular jacket
for the TF coils or a jacket with an outer square section for the
other coils. The op erating currents are 40-45 kA for the CS,
45-55kA for the PF coils and 68 kA for the TF coils. The CCs
use a reduced size cond uctor, 16kA, with about 300 strands
and without the central cooling channel.
The coil electrical insulation system is composed of
multiple layers of polyimide film-glass impregnated with
epoxy resin (a radiation hard cyanate ester based r esin in the
case of the TF coils). Epoxy-glass and f illed epoxy resin is
used extensively to fill tolerance gaps.
The TF coil case encloses the winding pack and is the main
structural component of the magnet system [6]. The TF coil
inboard legs are wedged all along their side walls in operation,
with friction playing an important role in supporting the outof-plane magnetic forces. In the curved regions above and
below the inboard leg, the coils are struc turally linked by
means of three upper and t hree lower precompression rings
formed from unidirectional bonded glass fibre and by four
upper and four lower sets of poloidal shear keys arranged
normal to the coil centreline. In the outboard region, the outof-plane support is provid ed by f our sets o f Outer Intercoil
Structures (OIS) integrated with the TF coil cases and
positioned around the perimeter within the constraints
provided by the access ducts to t he vacuum vessel. The OIS
form four toroidal rings and act as shear panels in combination
with the TF coil cases. Th e TF conductor is co ntained in
grooves on each side of flat D shaped steel radial plates. Seven
DPs are assembled together to form the winding pack of one
TF coil.
The CS assembly [7] consists of a vertical stack of six
independent winding pack modules, which is su pported from
the bottom of the TF coils through its pre-load structure. The
pre-load structure consists of a set of toroidally segmented tieplates located at the inner and outer diameters of the coil
stack. It provides axial pressure on the stack throug h key
blocks . The CS modules use a square section jacket which is
self-supporting against the radial and vertical magnetic forces.
The six PF coils (PF1 to PF6) [8, 9], are attached to the TF
coil cases through flexible plates or sliding supports allowing
radial displacements. The square conductor provides the coil
structural support.
Outside the TF coils, and mounted on t heir cases, are
located three independent sets of CC [ 10], each consisting of
six coils arranged around the toroidal circumference above, at
and below the equ ator. Within each set, pai rs of coils on
opposite sides of the machine are connected in series.
The coil supplies (helium, current and instrumentation [11])
are grouped into feeder ducts [12, 13] that run from outside
the cryostat, initially within cryostat extensions and then
inside the main cryostat, largely above and below the coils.
The valves, current leads and instru mentation cubicles are
placed in galleries just outside the bioshield [14]. The current
leads make use of HTS technology [15].
3 of 9
IEEE/CSC & ESAS EUROPEAN SUPERCONDUCTIVITY NEWS FORUM, No. 19, January 2012
4
2019808
whiich are custom
m machined too achieve a tigght fit of the shear
keys in those lineers.
The assembledd 1/9 machine segments aree transferred too the
pit in the tokamaak hall, Fig 2. The vault sttructure of thee TF
coill inboard legs is built up cirrcumferentiallly, with correcction
ms every 3 pa irs if
of the accumulatted assembly error by shim
neceessary (althouugh the intentioon is that the coils are machhined
mming at asseembly).
accuurately enoughh to avoid shim
The upper andd lower OIS structures agaiin act as referrence
surffaces for the assembly operaation. Once thhe vault structuure is
mplete, the rem
maining OIS keys are installled. After the OIS
com
panel structure liinking the ouuter legs is c omplete, the upper
andd lower preccompression rings are innstalled and pretenssioned. The ring pretensioning can be adjustedd to
mpensate for any settlinng effects. Retensioningg, if
com
neceessary, can bee performed with the CS in-ssitu.
All Correctionns Coils (low
wer, side andd upper) are then
attached to the TF coil cases. The PF coi l installation starts
withh a survey of the support poositions on the TF coil casees. If
requuired, approprriate shims arre prepared annd installed beefore
the PF coils are atttached to the TF coils.
Coil current ceentre alignmennt could be doone on the bassis of
the geometry of the winding paack, measuredd at the final sttages
of manufacturingg. However, direct measureement of the locus
of the current ceentre line is exxpected to bee possible at room
mperature with a small AC current in the coil to an accuuracy
tem
bettter than 1mm. A prototype system is beingg developed [66].
The feeder insstallation (apaart from the CC header rinng) is
the next step inn the magneet assembly. Those underr the
macchine are ei theer lowered innto position thhrough the borre of
the TF coils or brrought in throuugh the side poorts.
Finally, the CS is assemblled outside, complete withh its
prelload structure and all terminnals and coolinng pipes. It is then
low
wered into the inner bore of the TF coils and attached too the
F coil inboardd legs.
suppports at the boottom of the TF
ABLE 1 COIL MAIN PARAM
METERS
TA
Paraameters
TF
Number of coils
Number of turns per coil or WP
Stored magnetic energy GJ
Maximum operatiing current kA
Nominal peak fielld T
Electrical dischargge time constant**s
CS
PF
18
6
6
134
549
1155-459
41GJ
7
4
68
45
45
11.8
13.0
6.0
15
11.5
18
Conductor total leength m
82260
36114
633142
Coil total weight (with case and
structures) t
298 (1
coil)
954 (all)
2163
(all)
S/C strand total weight t
410
104
240
Winding Pack weight t
110
688
1361
Coil Case weight t
188t
NA
NA
403MN
NA
NA
Vertical force per coil, max MN
NA
327
160
In plane bursting force per half coill
MN (max for CS and for each PF)
205
220
43,13,30,
19,64,54
Length of the coil centerline m
34.1
10.7
255-75
4570m
6019
60009144067
760/ 415
903/601
3877-879
C
Cross section of thhe steel at TF
innboard mid planee (jacket + RP,
ccase) or in coil secction (CS, PF) m2
0.562
0.86
0.444,0.2
0.322,0.29
0.377,0.82
He mass flow ratee in conductor g/s
8
8
8-14
He mass flow ratee of structures kg//s
2.5


Centring force perr coil
WP Conductor lenngth
Conductor Unit Length Range m
*Note that thiss is the overall theermal equivalent exponential valuee as
teemperature depenndent resistors aree used and there iss a detection delaay time
III. ASSEMBLY
This section is a brief summ
mary of the main assembly steps.
The lower PF
F coils (P5 annd P6) and thhe lower Corrrection
Cooils (CCs) are the first cooils to b e placed in positioon, on
mporary suppports at the bottom of thee cryostat. Various
tem
eleements of the support structture are also put in position at this
tim
me (such as thhe lower precompression riings and the gravity
wer correctionn coils
suupports). The feeders for thhe side and low
(thhe lower header rings) are also positionedd.
The TF coils are assembleed in pairs in the assembly hall at
the ITER site as part of a 1/9 vacuum vesssel sector com
mplete
wiith its thermall shield, Fig 2. The tightly toleranced intterface
regions betweeen the two coils will h ave been preecision
h factory but additional shimming can be
machined in the
wer OIS acts as the
prrovided as re quired. The upper and low
reference surfacce and joint too link the coills for this operation.
Eaach coil has a thin 'sandwich' of innsulation, conntained
beetween two steel sheets, that is fixed to one of thhe coil
weedging surfacces (welded orr bonded) durring manufactturing,
annd shims can be attached by welds to thhe outer sheet. Slots
foor the shear keeys are oversizzed to allow the use of slott liners
4 of 9
Figg. 2. 1/9 VV segm
ment and 2 TF coils, with temporaryy supports, insidee the
cryostat
IEEE/CSC & ESAS EUROPEAN SUPERCONDUCTIVITY NEWS FORUM, No. 19, January 2012
5
2019808
NG
IV. MANUFACTURIN
In this secttion we givve a brief description of the
mponents. Thiss is the
manufacturing ooperations for the main com
me originally foreseen forr the coils annd for
reference schem
whhich the desiggn is optimisedd. Individual Domestic Ageencies,
responsible forr the procureement, may adjust this in the
interests of costt reduction.
A. Conducctor
The componeents of the coonductor are the supercondducting
strrands, copperr wires, an oppen spiral thaat forms the central
coooling channeel, steel wraps for AC losss barriers to cable
suubstages and for protection of the final cable, and exxtruded
seections of the jacket (eitheer circular forr the TF coils or a
sqquare outer seection with a circular hole for the CS a nd PF.
Thhe manufacturring is illustratted in Fig 3 [116]. The strands and
coopper wires aree cabled to a multistage circcular cable, thhe final
m jacket sectioons, 1suubstage of whiich has 6 ‘petaals’. The 7-11m
2m
mm oversize on the cable, are butt welded
d together to form a
fuull unit length (this process allows easy weld inspectioon and
tessting) before the cable is pulled into it.. The jacket is then
coompacted by rolling, onto the cable. Thhe process reequires
caareful control of dimensionaal tolerances so that the cabble can
fitt the jacket wiith a 1-2mm clearance and a pulling forcce <5t,
mental to the jacket
wiithout excessiive compactionn that is detrim
mechanical prooperties (it haas to wi thstannd the Nb3Snn heat
treeatment withoout grain bounddary sensitisattion).
Fig. 3. Steps in Conduuctor Manufacturiing Process
B. TF Coill Winding Pacck
The conductoor is wound innto a mould holding the shhape of
the conductor during the reeaction heat treatment [177, 18].
Duuring the mannufacture of the TF Modell Coil, it was found
that the conducctor may expperience a peermanent elonngation
affter the heat treatment up to 0.05%. During heat treaatment,
the conductor is held flat byy clamps but is allowed to move
raadially. The innitial winding dimensions must take accoount of
this possible chhange in lengtth after heat trreatment so t hat the
reacted conducttor can fit in thhe RP groovess.
Once the DP is wound but before heat treatmennt, the
coonductor jackeet at the i nnerr transition reegion is ope need and
the helium inlett formed. Thee conductor teerminations arre also
foormed after winding but before heatt treatment of the
coonductor. Theyy consist of thhe terminals (2 per WP) annd the
DP to DP joints (6 per WP).
The conductorrs are then heat treated (about 650°C
C for
2000hrs) in a furnace with a conntrolled inert atmosphere. After
heatt treatment, thhe top and botttom single paancakes are moved
ween
verttically up andd down respecctively to cre ate a gap b etw
m and allow the removal of the mouldd. The RP is then
them
inseerted between the two separrated pancakees as shown inn Fig
4. The removal of the mould and the in serttion of the RP
P are
mplex operatioons requiringg a purpose-ddesigned handdling
com
toolling and a special set of insertion movements. The
alloowable maxim
mum strain on the conductor (±0.1%) limitts the
pancake separatioon.
Fig. 4. Transffer of Conductor DP from Mould to Radial Plate
The conductor is then locallly lifted to alloow the applicaation
ween
of the turn insulation. Thee correct aliignment betw
conductor and RP groove hass to be mainttained during this
mber of outer glass
proccess, using maarks and adjuusting the num
wraaps, to preventt a progressivee buildup of misalignment. The
widdth of the RP grooves iss adapted to allow a s mooth
embbedding of the reacted conductor and accommoodate
winnding tolerancees and possiblle permanent elongations of the
reaccted conductoor. The clearrance betweeen conductor and
grooove is 3mm inn the upper annd lower regioon and 2mm inn the
outbboard region. The resultingg gaps will bee filled with glass
clotth shims durinng application of the turn inssulation.
The RPs are thhick plates maade of austeniitic stainless steel,
typee 316LN. Thee manufacturinng process forr the RPs reliees on
macchining of pllates (produceed by hot roolling, forgingg, or
hippping [17, 18]) to form sub-ssections whichh are assembleed by
welding. The butt welding is simplifiedd by maintaiining
recttangular butt sections and completing the radial groooves
afteer welding, byy local machinning. A final machining off the
plattes is possiblee in case that the groove needs to be adjuusted
fit the final shaape of the condductor.
to fi
Each jacketed conductor iss locked in thhe RP groovee by
meaans of a coveer plate (CP). The CPs haave holes at 20cm
w during the vacuum presssure
spaccing to enable resin flow
VPI) process. The CPs a re fixed by laser
imppregnation (V
welding with a continuous seaam of at least 1mm depth inn the
outbboard leg and at least 2mm depth in the inn-board.
The whole DP assembly, with grounnd and conduuctor
insuulation, is thenn vacuum-presssure impregnaated in a moulld.
5 of 9
IEEE/CSC & ESAS EUROPEAN SUPERCONDUCTIVITY NEWS FORUM, No. 19, January 2012
6
2019808
C. TF Struuctures
Each leg of the TF Coil Case (TFCC) is formed by a subasssembly of 4 components, with an inner U-section and a
cloosure plate subb-assembly [119].
BU BP
AU
AP
Fig. 6. Keyy Interfacing Planees and Surfaces on Sector AU
Fig. 5. The 4 Sub-Asseemblies of the TF
F Coil Case
mblies have numerous inntegral
Both U-secttion sub-assem
strructural attachhments such as the Intercoill structures, thhe precoompression flaanges, the TF coil support leg, and attachhments
foor the CS, PF
F, CC, vacuum
m vessel t herm
mal shield (V
VVTS)
suupports and maachine assembbly tooling.
The casing opperation compprises three maain phases [200]:
 Inseertion of the WP into the casse;
 Clossing by weldinng the TFCC elements;
 Finaal impregnatioon of the TFCC
C-WP gaps.
The final toolerances acceeptable on inndividual coiils are
deetermined by the coil fit during assembbly in the cryyostat,
together with the allowablee magnetic fiield errors annd the
maintenance off operational clearances. The tolerances on case
seections, windinng pack and the casing opeeration are ballanced
to give maxim
mum manufaccturing latitudde. The indiividual
suubassemblies are built upp by a s ucccession of foorging,
machining, welding
l
and maachining steps.
s The key to cost
reduction in t hiis process is a predictable and localizedd weld
distortion so thhat the nu mbber of machinning steps, a nd the
mount of overrmetal, are minimized andd eliminated on the
am
finnal closure weelds.
The AU s ub-assembly is the key to maintaininng the
tolerances. The flat inner surfface of the AU
U sub-assemblly will
prrovide precise geometric loccation and unniform force trransfer
interface for thhe winding paack. The base line design reequires
mized shim over the com
mplete
the installationn of a custom
interface with a precise flatness annd perpendicuularity
U sub-assembly are
tolerance. The angular facees of the AU
deefined with a wedged 20° angle. These faces will defiine the
plane for refereence to the currrent centre liine plane of thhe WP
duuring insertionn and they theen are used as references for the
location of the TF coil in the ITER machinee, Fig 6.
The case su b-assemblies are designedd with assem
mbly
cleaarance gaps too the WP. Thee design clearance will be 7mm
P angled facees and W P-AU
U) or 10 mm (WP vs BU––AP–
(WP
BP)). The go al of these c leaarance gaps is to re movee the
requuirement for tight tolerancees on all the WP-TFCC maating
ment of the WP relative too the
surffaces and to allow alignm
CC at insertioon. The finall geometric loocation and force
TFC
trannsfer between the WP and Case will be achieved by fiilling
the inter-space wiith a charged epoxy.
The WP will be lowered intto the AU elem
ment, aligningg the
P by means of an overhead bridge crane annd three alignm
ment
WP
guiddes. The a lignnment guides will align t he WP while it is
slightly lifted oveer its final possition so its weight is still taken
m the crane and the g uidess do not havee to apply a large
from
forcce, Fig 7 . Pre-machined G110 shims willl be placed onn the
AU surface to suppport it.
WP mid-plane
mid-plane
A mid-plane
AU
mid-plane
G
=
A*
=
J**
B*
A
B
J
=
=
Figg. 7. WP and Case Mid-Planes andd Reference Points
t for WP Insertion in
AU Sub--Assembly
The BU sub-aassembly willl be lifted using tooling and
braccing to minim
mize distortioon. The sub- assembly willl be
wered over thhe WP and thhe interfaces to the AU sublow
asseembly broughht into locatioon. The matcching of the weld
faces will bee checked annd the geom
metric
prepparation surfa
locaation of the BU reference marks surveyeed with respeect to
the AU reference marks and thee WP fiducials.
There is a t woo stage weld sequence. Ann initial laser root
mm made witth the TF cooil in the verrtical
weld of 10–20m
(inssertion) orienttation. A finaal structural weld made byy the
W) with the TF coil rotated too the
Subbmerged Arc Method (SAW
horiizontal positioon. This weld is to be made after filled resin
imppregnating the WP to TFCC gap is compleetely cured.
After welding of the AU-BU
U interface it will be necesssary
P relative to the TFCC struucture in ordeer to
to block the WP
6 of 9
IEEE/CSC & ESAS EUROPEAN SUPERCONDUCTIVITY NEWS FORUM, No. 19, January 2012
7
2019808
remove the WP support jigs and braces; this is necessary to
insert the final case sub-assemblies AP and BP. At this stage
the CCL fiducial marks of the WP will be transferred to the
TFCC prior to closing it with the sub -assemblies AP and BP.
When fiducial marking transfer is complete, cover plat es AP
and BP will b e installed. The first step will be the laser root
welding of the closure plates. After the clos ing of all welds
with laser techniques and the W P-TFCC gap filling by
vacuum pressure impregnation with a resin/filler [6], the TFC
will be rotated to the horizontal position for completion of the
case welding by Submerged Arc Weld (SAW).
E. PF Coils
A simplified flow diagram of the PF co il manufacturing
process is shown in Fig. 9. Starting from the conductor spools,
two in hand Double Pancakes (DP) are wound, insulated,
vacuum-impregnated with resin, then stacked to form coils,
and bonded with a further vacuum impregnation step.
D. Gravity Supports
The TFC assembly rests, via 18 Gravity Supports (GS), on a
tokamak pedestal ring s upported by 18 columns off the
concrete floor slab of the tokamak hall. The GS consists of a
flexible plate assembly (FPA), clamping bars, studs and bolts,
as shown in Fig. 8. The FPA is pre-assembled in the factory
from 21 flexible plates (with cooling pipes and thermal shields
sections attached), separated by 20 short spacer plates at t op
and bottom. The FPA components are pressed together at top
and bottom between pairs of horizontal pre -stressing bars by
tie-rods. The FPA is designed with sufficient pre-compression
and friction to prevent both sliding and separation between its
components.
The FPA of each GS flexes within allowable stresses to
accommodate the radial thermal contraction of the TFC
assembly. But the GS is r igid versus c yclic out-of-plane
bending caused by TFC torsion during plasma operation, and
versus vertical and horizontal forces parallel to the flexible
plates due to deadweight, asymmetric plasma VDE and
earthquakes. The 1 8 GS transmit these loads to the pedestal
ring, which is an integral part of the cryostat structure.
Fig. 9. Simplified flow diagram of the PF coil winding, impregnation, and
assembly
Tolerance control is required throughout the manufacturing
to ensure that subc omponents interface correctly. The
manufacturing scheme has therefore been defined with an
allocation of overall tolerances in the coil current centerline
position at the different manufacturing stages, and the
possibility to allow reductions in position errors during the
winding – DP stacking. Assembly errors and manufacturing
errors for the PF coil s are defined in Fig. 1. The targ et
tolerances for the PF coil winding pack are defined in Table 2.
Clamping
bars
Thermal anchor and
radiation shield
Fig. 10. PF Coil Tolerances
TABLE 2 TARGET TOLERANCES FOR PF COILS IN MM
Manufacturing
Assembly
Pre-stressing
bars
Fig. 8. Details of a Gravity Support
PF1, 6
x
±3
y
±3
z
±3
x
±2
y
±2
z
±1
PF2, 5
±3
±3
±3
±2
±2
±1
Coil
Flexible plates
±4
±4
±3
±2
±2
±1
PF3, 4
The shaping of the termination is performed after
completion of winding. This operation is needed to allow the
7 of 9
IEEE/CSC & ESAS EUROPEAN SUPERCONDUCTIVITY NEWS FORUM, No. 19, January 2012
8
2019808
prroper fitting of the final filllers into the winding pack. To do
this, a p ortion of conductor (about 3 m) is opened, thhe turn
moved and thee joggling is performed by shaped
insulation is rem
h close the winding table.
moulds, brought
The windingg DP is then placed in a mould and vaacuum
mpregnated with resin. Durring this p roccess, the connductor
im
ennds, which caarry the terminnations to be used for the joints
beetween doublee pancakes, must be keppt unbonded to the
m. This is to provide
wiinding pack over a length of typically 5m
flexibility wheen the jointss are made. These
the required fl
coonductor ends are, thereforee, provided with an anti-addhesive
wrrap before impregnation. This wrap is removed after
im
mpregnation.
The DP is heated-up to aboout 80 degreess, by passing current
through the connductor. At thee same time thhe resin is putt under
vaacuum, degasssed and heatedd up to 80 degrrees. Once thee DP is
w starts at the lowest
hoot it is put undder vacuum. The resin flow
pooint of the DP, in sin glee location, annd gets ou t in the
diametrically oppposite positioon, at the highhest point of thhe DP.
Thhe resin is forcced to face thee bottom surfaace of the DP and to
floow through thhe winding to reach to the suurface of it. The DP
siddes have no resin passagges. The whoole impregnattion is
wly, to allow the resin to wet the insulatiion, to
peerformed slow
takke advantage of the capilllary effect and to avoid crreating
traapped bubbless in the windinng.
After the imppregnation is complete, thee DP is put first at
mospheric pressure
r
and thhen brought to an overpreessure.
atm
Thhis helps to reduce and diffuse eventtual trapped microbuubbles of inertt atmosphere into the resin bath. Then, with the
P under pressuure, the Curinng Cycle takess place, for about 24
DP
hoours at around 130 degrees.
Eight DPs aree assembled toogether to forrm the windingg pack
off PF 1,3,4,5 cooil and six andd nine DPs aree assembled toogether
foor PF 2 an d 6 winding pack respectivelyy. For handlinng, the
P is clamped and lifted at a large number of locationns by a
DP
suuitable Jig to avoid local turn deformatioon and relaxat ion. A
layyer of dry gllass is placedd between thee individual DPs to
alllow resin peenetration and bonding during the vaacuum
mpregnation off the ground insulation and to absorb toleerance
im
vaariations in thee gap betweenn the DPs.
Following VPI of the winding pack, cooling pipes for
heelium inlets annd outlets, andd manifolds aree pre-assembleed and
coonnected to thee helium inlets and outlets. An insulationn break
is attached on each inlet annd outlet. Alll cooling pipes are
wrrapped with inndividual grouund insulation up to the insuulation
brreaks.
F. CS Coills
Conductor winding is shoown in Fig. 11. Each connductor
lenngth is receivved on a sp ool and is useed to wind eiither a
HP) or a qua dpancake (QP
P). The co nducctor is
heexapancake (H
unnwound from the spool to pass through a straighteninng unit
beefore reachingg the windingg table. The conductor is wound
wiithout any insulation, which impliess that spacerrs are
regularly inserteed between tuurns and panccakes to acco unt for
the further insttallation of the insulation. Multiple paancake
wiinding with a single leength meanss that windiing is
alternatively perfformed inwards
r and outwaards in successsive
plannes, starting and ending on the outsidee. The windinng is
circcular, which means that thee radius of currvature is connstant
alonng a turn, withh the exceptioon of the jogggle area wheree the
conductor is mooving from one turn to the next. This is
achiieved throughh a short straigght section folllowed by a cuurved
secttion. Much atttention has too be paid to thhis area sinc e any
erroor in th e startting point couuld impact thee geometry off the
next turn. The coorrect bending radius of the conductor in each
secttion is achieveed using a set of bending rollers located at the
entrrance of the winding table. Most of the winding is plaanar,
withh the exceptiion of the pancake to paancake transiition,
perfformed alternaately at the innner and at the outer radius.
Fig. 11. CS module winding
The location of the He inleets and outletts is only deffined
afteer winding thee HP or QP once the futuure location off the
inneer and outer tiie plates is esttimated. The conductor jackket is
thenn locally machhined in order to insert the prefabricated cover
inclluding the He inlet or outlet.
Once the uppper terminal is manufactuured, a cl am
mping
systtem is installeed around thee hexapancake or quadpanncake
andd it is transferreed to the heat treatment oveen to form Nb3Sn.
The whole sttack of wounnd conductor is clamped and
trannsferred to thee insulation staation. In orderr to allow spacce to
insttall the wrappping tool, thhe conductor is progressiively
unspprung by remooving the spaccers and lowerring down thee turn
to be insulated annd putting it inn its final posittion. A continnuous
control of the possition of the innsulated turnss is performedd and
inteerturn and inteerpancake glasss cloth layers inserted to filll the
spacce. The grounnd insulation is installed around the whole
moddule at the insuulation stationn.
The impregnattion mould iss installed aroound the coil. He
inleets and outlets are plugged to avoid any penetration of resin
duriing impregnattion. Resin i s injected insidde the mould from
botttom to top so as to avoid bubbble formationn.
18 outer tie-plates and 9 innner tie-plates are manufactuured.
Theese tie-plates are either cuut from cold worked and shot
peenned laminatedd plates o r machined out of forged pieeces.
Theey can be spllit into three parts (two ennds and a ceentral
secttion) welded together, sincce the thickneess of the endds is
largger than that off the central part. The only machined surffaces
are the flat surfacces which trannsmit the loadd to the keybloocks,
andd the holes for bolting the tiee-plates to the keyblocks.
8 of 9
IEEE/CSC & ESAS EUROPEAN SUPERCONDUCTIVITY NEWS FORUM, No. 19, January 2012
9
2019808
ACKNOWLLEDGEMENT
The invaluablee assistance off members off the ITER maagnet
mestic
diviision and the contributions of the membeers of the dom
wledged.
agenncies are grateefully acknow
The views and opinnions expressed herein do not neceessarily reflect those of
the ITER Organizatioon.
RENCES
REFER
[1]
Fig. 12. Installlation of the condductor insulation during CS windinng
The stacking is performed in the assembbly hall and reequires
the use of a special support abble to withstannd the weight of the
S plus temporrary structuress. The supportt must be veryy rigid
CS
m. The
ass the key bloccks must remaain in a planee within 1 mm
deeformation shoould be such that the stackking of the moodules
remains perpenddicular to the reference plann determined by the
keeyblocks. Slotts are providedd in the suppoort to allow vertical
t modules with their buusbar extensioons. A
movement of the
mplate is firstt installed on the stacking support to sim
mulate
tem
the 9 supports located at the bottom of thee TF coils. Thhen the
lower keyblockks are installled and bolteed to the st acking
suupport. The G10 buffer plates are in stalled after having
deetermined the correct thicknnesses of the loower and uppeer ones
takking into accoount the real height of eachh module. Theen they
arre machined inn order to havve the CS equaatorial plane loocated
wiith the correctt offset to the machine equaatorial plan att room
mperature. Thhe tolerances for the CS modules are givven in
tem
Taable 3 (see Figg 10 for notatioon).
ARGET TOLEERANCES FOR THE CS COILS MM
TABLE 3: TA
x
y
z
+/-2
+/-2
+/- 1
x
y
z
Stack assemblyy *
+/- 0.5 ( CS3L) lineearly
+/- 3
+/- 3
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
Each CS module
Manufacturingg
[12]
to +/-2 (CS3U))
[13]
Installation ***
+/-2
+/-2
+/-0
* Change in x and y betw
ween adjacentt modules < 2 mm
** extra xx, y identical for all modulees at installatioon
[14]
The precomppression system is then i nsstalled at the top of
the CS assemblly. A set of suupernut bolts is used to appply the
multaneously on all the tie plates.
prrecompressionn by pulling sim
Thhis is achieved by tighteninng progressiveely the bolts so that
the tensile load is equal in alll tie-plates. When enough gap is
prrovided, shimss are inserted between tie-pllates and keybblocks.
Affter insertion of the shims and tighteninng of the ho riizontal
sccrews, the bollts are unscreewed to releaase the load on the
threads.
V. CONCLUSIONS
The main reequirements, design and manufacturing routes
foor the ITER cooil system have been describbed. Constructtion of
way and thhe many tecchnical
the componennts is underw
chhallenges are being successffully solved.
[15]
[16]
[17]
[18]
[19]
[20]
9 of 9
ITER Project Reequirements, ITER IDM reference 27ZRW8, versioon 4.6,
May 2010
ITER Magnet Design Descripption Document Part 1, I TER IDM
KM, version 1.8, Sept 2009
reference 2NPLK
M. Coatanea, J.L. Duchateau, B. Lacroix et al, “Investigationns on
Quench Detectioon for the ITER TF Magnet Systeem”, paper presennted at
MT22 and publiished in IEEE Traans Applied Superrconductivity, 20112
J. Knaster et all, “ITER non-axiisymmetric errorr fields induced by its
magnet system”, Fusion Engineering and Design, in press Aug 20111
N. Mitchell et al, “Status of thee ITER Magnets””, Fusion Engineeering
and Design, vol 84, issues 2-6, p1113-121, June 2009
F. Savary, R. Gallix, J. Knaster et al, “The Toroiidal Field Coils for the
ITER Project”, paper presented at MT22 and pubblished in IEEE Trans
Applied Supercoonductivity, 2012
P. Libeyre, D. Bessette, M. Je well et al, “Adddressing the Te chhnical
Challenges for the Construction of the ITER Ceentral Solenoid”, paper
presented at MT22 and p published in IE
EEE Trans Appplied
Superconductiviity, 2012
F. Simon, C. B oyer et al, “De sign of the ITER PF coils joints”, paper
presented at MT22 and p published in IE
EEE Trans Appplied
Superconductiviity, 2012
B. Lim, F. Sim
mon et al, “Devellopment of the IT
TER PF C oils”, paper
presented at MT22 and p published in IE
EEE Trans Appplied
Superconductiviity, 2012
A. Foussat, W. Wu, H. Li, “Quaalification phase of key technologiies for
ITER Correctioon Coils”, paper presented at MT22 and publishhed in
IEEE Trans Appplied Superconducctivity, 2012
F. Rodriguez-Mateos, A. F oussat et al, “Thermo-Mechanical
Instrumentation for the ITE R Magnets’ Structurees”, paper presennted at
MT22 and publiished in IEEE Traans Applied Superrconductivity, 20112
P. Bauer, Y. Chhen, N. Dolgettaa et al, “Ke y Com
mponents of the ITER
Magnet Feeders”, paper presentedd at MT22 and puublished in IEEE Trans
Applied Supercoonductivity, 2012
C. Gung, Y. Ilinn, N. Dolgetta ett al, “Progress in Design, Analysis, and
Manufacturing Studies of the ITER Feeders”, papper presented at MT22
c
, 2012
and published inn IEEE Trans Appplied Superconductivity
Y. Song, P. Baauer, Y. Bi, “Staatus of R&D for ITER Feeder S ystem
Procurement in China”, paper preesented at MT22 and published in IEEE
Trans Applied Superconductivity, 2012
A. Ballarino, P. Bauer et al, “Thee ITER Current Leads”, paper pressented
at MT22 and puublished in IEEE Trans Applied Supperconductivity, 2012
A. Devred, D. Bessette et al, “S
Status of ITER Conductor Producction”,
paper presentedd at MT22 an d published in IEEE Trans Appplied
Superconductiviity, 2012
N. Koizumi, K. Matsui, T. Hemmi et al, “D
Development of ITER
Toroidal Field Coil in Japan”, paaper presented at MT22 and publishhed in
IEEE Trans Appplied Superconducctivity, 2012
A. Bonito-Olivaa et al, “Status off the F4E Pr ocureement of the EU ITER
TF Coils, papeer presented at MT22 and publlished in IEEE Trans
Applied Supercoonductivity, 2012
M. Iguchi, Y. Chida, K. Takanoo, “Development of structures for ITER
Toroidal Field Coil in Japan”, paaper presented at MT22 and publishhed in
IEEE Trans Appplied Superconducctivity, 2012
E. Baynham, R. Gallix, J. Knnaster, N. Mitchhell, F. Savary, “The
TER”,
Insertion of the WP in the St ructural Casing of thhe TF Coils of IT
Transactions onn Applied Superconductivity, vol. 20, issue 3, pp. 389 –
393, 2010