Design Description Document: DDD 11 ITER_D_22HV5L v2.2. Magnet

Transcription

1 Design Description Document: DDD 11 ITER_D_22HV5L v2.2 Magnet Section 1. Engineering Description 14 Oct 2006

2 Table of Contents 1 Engineering Description System Description Design Philosophy Selection of Design Options Bucking and Wedging Superconductor Type and Operating Temperature Central Solenoid Jacket Material Use of Incoloy for the Extruded Jacket Use of Stainless Steel for the Extruded Jacket Conclusions Conductor Currents and Coil Voltage Levels TF Coil Centreline Shape TF Winding Pack: Plates and Square Conductors Insulation Conductor Jacket Material Cost and radial build Other Considerations Conclusions TF Coil Out-of-Plane Support Inner Intercoil Structure Optimisation of Poloidal Key Length Outer Intercoil Structure Nuclear Heating in the TF Coils PF and CC Coil Maintenance Strategy Conductor Joint Options Superconducting Cable Configuration Design Criteria Conductor Design Criteria Temperature Margin Copper and Helium Contributions to Stability Current Uniformity Hot Spot Electric and Magnetic Field Total Strand Strain in Cable and Transverse Load Degradation Hydraulic Friction Factors and Inlet Pressure Drop Cable AC Losses Structural Design Criteria Static Yield Criteria (Tresca stress limit) for Metallic Components Fracture Criteria for Metallic Components Fatigue Criteria for Metallic Components Yield Criteria for Coil and Structural Insulation Limiting Shear Stress for Coil Insulation Allowable Pre-tension in Precompression Rings Electrical Design Criteria Nuclear-Related Design Criteria Insulation Copper Operation Interfaces Nuclear Heat Loads on the Magnets PF Coil Current Scenarios Correction Coil Current Scenarios Plasma Control with the PF Coils Resistive Wall Mode Control with the Correction Coils...63 Design Description Document 11 Magnet Page i

3 Cryogenic Heat Loads Disruptions Internal Helium Pressure Physical Interfaces Power Supplies Cryoplant Coil Alignment Magnet Displacements Diagnostics Assembly Operation Sequences and Flexibility Cool Down and TF Coil Charging Vacuum Vessel Conditioning Hydrogen Plasma Pulsing DT Plasma Pulsing Scenario Flexibility Warm up and Cryostat Opening Faults and Safety Component Descriptions TF Coil TF Conductor TF Double Pancakes, Radial Plates and Joints Winding and Conductor Terminations Radial Plates Final Insulation TF Winding Pack Casing and finishing operations TF case design and manufacture Structures Inner Intercoil Structures Outer Intercoil Structures Gravity Supports PF Coil Supports Correction Coil Supports Central Solenoid CS Conductor CS Winding Pack Design and Manufacture CS Joints, Terminals and Helium Inlets/Outlets CS Preload Structure PF Coils PF Conductor PF Winding Pack Design and Manufacture PF Joints & Cooling Pipes Correction Coils CC Conductor CC Winding Pack CC Case Auxiliary Systems In-cryostat Feeders Cryostat Feedthrough Coil Terminal Box Current Leads Cooling Loops and Valves Structure Cooling Valve Boxes Electrical Insulation Breaks Grounding Scheme Instrumentation and Control Quench Detection Systems Coil Short Detection Systems Weights, Lengths and Volumes of the Main Magnet System Components Design Description Document 11 Magnet Page ii

4 1.3 Magnet Procurement Packages Introduction Overall Procurement Schedule Toroidal Field Coils Procurement Package Structures Procurement Package Poloidal Field Coils Procurement Package Central Solenoid Procurement Package Feeders Procurement Package Conductor Procurement Package Drawings TF WINDINGS TF STRUCTURES SPECIAL COMPONENTS SCHEMATICS POLOIDAL FIELD COILS INTERFACES FEEDERS AND CTBS CORRECTION COILS Annexes to Section 1.2 (Component Description) 1. Feeder Design (ITER_D_24N2XA) 2. Cooling Scheme (ITER_D_24N2YT) 3. Control and Instrumentation (ITER_D_24N32Y) 4. Electrical Insulation Design and Monitoring (ITER_D_24N34Z) 5. Grounding Scheme (ITER_D_24N35H and ITER_D_22GZA4) 6. Vacuum Design and Leak Detection (ITER_D_24N36W) 2.1 Conductor 2.2 Structural Analysis 2.3 Electromagnetic Analysis 2.4 Magnet Safety 2.5 Heat Load Tables Additional Sections of the DDD11 Technical Notes on Documentation The drawings associated with the document are provided as hyperlinks (CTRL+CLICK) within the text. For these links to function, the drawings must be placed in a subfolder with the name 'draw' in the same folder as the main document. The drawings are filed in IDM with a separate reference number ITER_D_24TYF4. All drawing references have the format 11AABBCCNNNNNN-LL.pdf, where 11 designates the magnets, AA, BB, CC are the two letter codes of the drawing breakdown structure (see PBS (ITER_D_23EKCU v1.3)), NNNNNN is the drawing sequential number and LL is the sheet number. Empty fields are filled by underscores. Design Description Document 11 Magnet Page iii

5 When using the PDF version of the main document, use ALT+back arrow to return to the main document after viewing a drawing. From the word version, the drawing opens in a separate window. All drawing references within the document are also bookmarked in Word (these bookmarks are not present in PDF format). The format of the bookmark is DrwAABBCCNNNNNN_LL. For multiple references to the same drawing, a 'b', 'c' and so on is added to the individual bookmarks. The drawing list in section contains cross-references to the bookmarks to each drawing in the text. These cross-references work also in the PDF document (CTRL+CLICK). The drawing list also gives a full description of all the drawings available within the magnet PBS files. Design Description Document 11 Magnet Page iv

6 Record of Changes Version Date Changes 0 July 2001 Issued in N 11 DDD R Rewritten and issued in N 11 DDD 178 R0.1. Corrections to tables of sizes of PF, CS winding packs, PF and CS conductor lengths Addition of Rogowski coil for plasma diagnostics Correction to long burn scenario (burn time) Update of drawings (side CC) All Annexes to Section 1.3 Procurement Packages are completely updated although they are still draft stage Corrections to piping in structural feeders (text and tables , 3 and addition of precompression structure cooling Corrections to description of precompression structure cooling section Additions to method of precompression section Corrections to Tables , , , , , , Section (Hydraulic Friction Factors) modified to include He inlet specification Section PF current scenarios modified to include part on PF control currents Figure on total heat loads added to section Section Disruptions added Section Component Weights updated Drawings updated Corrections to drawings and drawing numbers for CS quench detection pick up coils Corrections to Figures , 2 and Table (nuclear heat distribution in TF winding) Extra points in first 2 s of reference scenario, Table Vertical disruption loads on gravity supports defined in section Clarification on disruption conditions and requirements on quench detection systems section He cooling circuit filling conditions defined in section Lateral instabilty of CS, PF modules added in section with new Table and cross-references in section AC loss section rewritten to include latest models Approximate hybrid scenario (1000s burn) added to section Correction to CS voltages in Table Correction to cooldown description in section to Re-issued in IDM as ITER_D_22HV5L v2 Coil voltages updated Tables to 4 Amendment to voltage testing criteria section Changes to Nb3Sn and NbTi strand specifications Changes to cryoplant interface description section Changes to section , CS jacket material Design Description Document 11 Magnet Page v

7 Changes to section and 6, definitions of electric and magnetic field on the conductor and total cable strain Change to CS cable configuration (increased wrap coverage and changes to AC loss) in sections and Additions to PCR description section New pre-tension criteria for PCRs section Amendments to section Hydraulic Friction Factors and Inlet Pressure Drop Section (cryogenic heat loads) re-written and renumbered to New sections on plasma control (for shape and vertical stability with PF) and (for RWM with CC) Corrections to He mass flow rates and pumping power section Section 1.3 (procurement) updates, new scope, sharing details, new schedule, procurement packages removed as annexes to DDD Section 1.4 (drawings) added Drawing hyperlinks and bookmarks revised and updated Section (weights, lengths volumes) updated and restyled Nb3Sn conductor strain allocations changed, s/c design criteria adjusted to include latest theories of transverse load degradation section on He design pressures added Sections and substantially re-written for new casewinding insertion procedure, different material specifications Table corrected Section , CS preload structure, revised Section separator plates eliminated in p3, p4 Table on PF db/dt added Section 1.2.9(analysis results) eliminated, now in sections 2.1 and 2.2 Section (current leads) rewritten for HTS leads Design Description Document 11 Magnet Page vi

8 1 Engineering Description 1.1 System Description The magnet system for ITER consists of 18 Toroidal Field (TF) coils, a Central Solenoid (CS), six Poloidal Field (PF) coils and 18 Correction Coils (CCs), as shown on Drawing The TF coil case is shown on Drawing 1101CA_ The case encloses the winding pack and is the main structural component of the magnet system. The TF coil inboard legs are wedged all along their side walls in operation, with friction playing an important role in supporting the out-of-plane magnetic forces. In the curved regions above and below the inboard leg, the coils are structurally linked by means of two upper and two lower precompression rings formed from unidirectional bonded glass fibre and by four upper and four lower poloidal shear keys arranged normal to the coil centreline. In the outboard region, the out-of-plane support is provided by four sets of Outer Intercoil Structures (OIS) integrated with the TF coil cases and positioned around the perimeter within the constraints provided by the access ducts to the vacuum vessel. The OIS form four toroidal rings and act as shear panels in combination with the TF coil cases. There is low voltage electrical insulation toroidally between TF coils in the inboard leg wedged region, at the poloidal shear keys and between the OIS connecting elements. The CS assembly is shown on Drawing 11CS It consists of a vertical stack of six independent winding pack modules, which is hung from the top of the TF coils through its pre-load structure. At the bottom, there is a sliding connection to provide a locating mechanism and support against dynamic horizontal forces. This pre-load structure, which consists of a set of tie-plates located at the inner and outer diameters of the coil stack, provides axial pressure on the stack. The number of CS modules has been selected to satisfy the plasma shaping requirements. The CS stack is self supporting against the coil radial forces and most of the vertical forces, with the support to the TF coils reacting only the weight and net vertical components resulting from up-down asymmetry of the poloidal field configuration. The six PF coils (PF1 to PF6), Drawing are attached to the TF coil cases through flexible plates or sliding supports allowing radial displacements. The PF coils provide suitable magnetic fields for plasma shaping and position control and their position and size have been optimized accordingly, within the constraints imposed by the access and pumping ducts to the in-vessel components. Outside the TF coils are located three independent sets of CCs, each consisting of six coils arranged around the toroidal circumference above, at and below the equator, as shown on Drawing Within each set, pairs of coils on opposite sides of the machine are connected in series. These coils are used to correct error fields (particularly toroidal asymmetry) arising from positioning errors in the TF coils, CS and PF coils. They can also correct error fields from the NB systems and from blanket test modules where ferromagnetic material is used. The equatorial CCs are also used to stabilise plasma resistive wall modes (RWM). Both CS and TF coils operate at high field and use Nb 3 Sn-type superconductor. The PF coils and CCs use NbTi superconductor. All coils are cooled with supercritical Helium in the Design Description Document 11 Magnet Page 7

9 range K. The conductor type for the main coils is illustrated on Drawings 1101WP_ and 11CSU1_ It is a cable-in-conduit conductor with a circular multistage cable consisting of about 1000 strands cabled around a small central cooling spiral tube. The operating currents are 40-45kA for the CS and PF coils and 68kA for the TF coils. The upper and lower CCs use a reduced size conductor with about 300 strands and without the central cooling channel. The side (or equatorial) CCs use a reduced size conductor with a central cooling channel to improve the heat removal (resulting from RWM AC losses). The coil electrical insulation system is composed of multiple layers of polyimide film-glass impregnated with epoxy resin. Epoxy-glass is used extensively to fill tolerance gaps. The CS and PF coils are pancake wound with a conductor that has a square outer section. The TF coils use a conductor with a circular outer section that is contained in grooves in so-called radial plates. There is one radial plate for each double pancake and the conductor is contained in grooves on each side. The gravity supports are composed of pedestals (one under each TF coil), with flexible elements to allow radial displacements, Drawing 1101GS_ Each TF coil is electrically insulated from its own support (and grounded through the feeder ducts which are sufficiently resistive to suppress eddy currents). The magnets are supported fully independently of the vessel and associated components. All TF coils, the CS, the upper PF coils and CCs are designed to be removable from the machine in case of a major fault. Individual double pancakes of the PF coils may be disconnected and by-passed in-situ in case of fault, since the PF coils have accessible joints located on their external side, with jumpers already in position. In addition, the cryostat design allows the lower (trapped) PF coils to be rewound in situ under the machine. The main magnet parameters are listed in Tables to 4. In Table , the acronyms IM and EOB refer respectively to the plasma scenario times of Initial Magnetization and End of Burn. Detailed information on the weight of the various components of the magnet system can be found in section Information on coil operating voltages is given in section 2.4 Annex 6b (Assessment of the PF Coil Circuit Behaviour During Normal and Fault Conditions N41R W 0.1 ) Table Overall Magnet System Parameters Number of TF coils 18 Magnetic energy in TF coils (GJ) ~41 Maximum field in TF coils (T) 11.8 Centring force per TF coil (MN) 403 Vertical force per half TF coil (MN) 205 TF electrical discharge time constant (s) * 11 CS peak field (T) 13.0 Total weight of magnet system (t) ~10,135 *note that this is the equivalent value as temperature dependent resistors are used Design Description Document 11 Magnet Page 8

10 Table Parameters for TF Coils and CS TF Coil CS Modules CS 1, 2, 3 Maximum coil current (MA) at IM (13.0T) Number of turns per TF coil/cs module: Radial Toroidal/Vertical Total Conductor unit length (m) (double pancakes) (for 6 pancakes) 594 (for 4 pancakes) Turn voltage in normal operation (V) 26.5 (discharge) 20 (IM) Ground/Terminal voltage (kv) in normal 3.5/3.5 ** operation (including fast discharge) (2 coils in series) 19.5/19.5* Ground/Terminal voltage (kv) in faulted operation 15/ /19.5* Coil DC Ground Test Voltage (kv) 16 41* Number of current lead pairs 9 6 *** * These are the CS2U/L values which are higher than the other coils, however all coils are designed to the same conditions ** A voltage surge (of a few ms) caused by the jitter of the switches may reach about 5kV. *** The current leads for CS modules 1 (upper and lower) are connected in series outside the machine Table Parameters for PF Coils PF1 PF2 PF3 PF4 PF5 PF6 Max. coil current capacity (MA)* Number of turns per coil: Radial Vertical Effective Total (allowing for partial turns) Conductor unit length (m) (double pancake, two-in-hand) Maximum Turn voltage (V) in normal operation** Ground/Terminal voltage (kv)** in 12/12 17/17 17/17 17/17 17/17 12/12 normal operation Ground/Terminal voltage (kv)** in faulted operation 12/12 17/17 17/17 17/17 17/17 12/12 Coil DC Ground Test Voltage (kv) * Current capacity given by maximum conductor current (45kA) x number of turns, no field consideration ** Voltages are for a two in hand winding configuration, double pancakes, in normal operation. Design Description Document 11 Magnet Page 9

11 Table Parameters for Correction Coils Top Side Bottom Max. coil current capacity (ka)* Number of turns per coil: Number of single pancakes Conductor unit length (m) (double pancake) Maximum Turn voltage (V)** Ground/Terminal voltage (kv)** 0.05/ / /0.1 Ground/Terminal voltage (kv) 0.1/0.1 15/15 0.1/0.1 faulted operation Coil DC Ground Test Voltage (kv) * Current capacity given by maximum conductor current (10kA) x number of turns ** Voltages are for two coils in series, each with a one in hand winding configuration, in normal operation Design Philosophy The design philosophy refers to a set of basic principles that have been defined as guidelines for the designers. They are qualitative in nature and can be summarised as follows: 1) The magnets will form semi permanent components of the machine, whose removal would require a significant down-time (possibly of a period of years). They are therefore designed to be highly reliable and include, sometimes with some cost penalty, features to increase reliability and provide redundancy. 2) The magnets will use superconductor, as the power consumption of normal-conducting coils for long plasma burns is very high and unacceptable in a future reactor that is to generate net power. Although adding complexity in the form of a cryostat, the magnets can then operate at a constant temperature and take advantage of the high strength of cryogenic steels, allowing simplification and a size reduction of the support structure. 3) Critical elements of the magnets must have been developed and tested within the R&D programme in a form close to that used in the design. This implies, among others, that the superconductor and cable layouts should correspond in concept to those tested in the CS and TF model coils (minor geometric and parameter variations are generally acceptable). In addition to these guidelines, there are a set of external operational requirements on the magnets that have been reached after consideration of the overall machine design: 1) The magnets must be capable of providing a range of operating scenarios including variations in pulse length, plasma power and plasma shape to the extent that this flexibility can be achieved without significant cost increases on the basic design requirement (30,000 plasma pulses of 15MA current with a burn length of 400s, a repetition time of 1,800s and a total fusion power of 500MW, see also DRG1). In particular, the flexibility will extend to plasma currents up to 17MA with total fusion powers up to 700MW in pulsed operation mode and up to 10MA with total fusion power of 400MW in steady state (non-inductive drive) mode. Details of design scenarios are given in DRG1, Table and 6. 2) The magnets should be capable of continuing normal operation in the event of plasma disruptions (i.e. a disruption should not cause a magnet quench). Design Description Document 11 Magnet Page 10

12 3) The magnet system must be designed, built and operated so that credible magnet system failures, which could occur under normal or abnormal conditions (including the specified SL-2 safety earthquake), cannot cause damage to the nuclear systems confinement barriers that would result in release of radioactivity exceeding the specified dose limits. 4) There will be 18 TF coils (i.e. 18 equatorial access ducts and divertor ports). 5) The design will permit plasma operation with the toroidal field in either direction. 6) Requirements on the toroidal field level, the poloidal field coil current/field capacities, and the outward forces on the modules of the central solenoid are given in detail in the relevant sections of the component descriptions. As a result of initial magnet design assessments, the following requirements have been adopted within the magnet design as being reasonable targets for both design and operation: 1) The magnets will be designed for 100 cooldown and warm-up cycles, 1,000 TF charging cycles, 50 TF fast discharges and 10 quenches. 2) A fast discharge of the TF system will require a fast discharge of all coils (TF, CS, PF and CC). Fast discharge of a CS or PF coil will require an associated fast discharge only of the CS and PF system. A quench of the CC will require a fast discharge only of the CC system. 3) The magnets need not be ready for further operation (i.e. completely recooled) until 96hrs after a fast 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. 4) The toroidal field may be generated in either direction relative to the plasma current, so out of plane movements may be in either toroidal direction. 5) The toroidal field may in principal be operated below the maximum value. However, the out of plane support of the TF coils relies partially on friction between the coils in the inner leg. This support is generally reduced at lower toroidal fields and a reduction of the out of plane forces (i.e. a reduction of the plasma current) will be required. The flux function of the poloidal field, causing lower out of plane forces, may not need to be restricted. Calculations need to be made for specific cases, however (for example) a peak field of 2T and a plasma current of 5MA may be acceptable Selection of Design Options Bucking and Wedging There are two concepts available for supporting the centring forces of the TF coils. The first of these, known as bucking, reacts the centring force on each coil directly on a structure which, for ITER, has to be the CS itself in order to save space in the centre of the machine. The second, known as wedging, reacts the centring force through a central vault formed by the inboard legs of the TF coils themselves, leaving a free-standing CS in the bore. Each concept comes with advantages and constraints. The bucked system means that the CS has to be layer wound (to avoid joints in the high stress region where the TF coils make contact), but the CS hoop bursting forces are almost completely reacted by the TF centring forces, reducing the stresses in both the TF coils and the CS. The layer winding results in uniform current density over the whole height of the solenoid (restricting the plasma shaping capability). The breathing of the TF coils, as the CS cycles in the pulse, makes linking the TF coils to each other to support the magnetic out- Design Description Document 11 Magnet Page 11

13 of-plane forces from the poloidal field more difficult. The torsion of the CS by the TF coil out-of-plane tilting motion complicates its design. The wedged system means that the CS has to resist high cyclic tensile stresses due to the hoop bursting forces. The TF coils are subject to a large compressive hoop load at their inboard leg thus making the stress conditions in this region more critical. On the other hand, the wedged system allows a simpler TF coil structure that provides effective out-of-plane support of the TF coils. Analysis of bucked and wedged TF coil configurations has showed that similar out-of-plane support can be achieved with either configuration only if the bucked design uses structural reinforcement outside the CS (and inside the TF bore) in the form of a torsion cylinder. As a result, the bucked design is also partially wedged and the machine radial builds are similar with either configuration. In the ITER design, the high priority given to plasma shaping results in a need for a CS which can provide non-uniform current distributions along the vertical axis. Wedging of the TF coils is, therefore, the chosen option since it allows the use of a segmented CS which can provide the required non-uniform current distribution. Increased shaping also results in more severe out-of-plane load distributions at the ends of the inboard leg of the TF coil. The wedged design can provide an effective support against these out-of-plane loads. Wedging offers in addition the following advantages: i) the CS can be manufactured in repairable, or replaceable, sub-modules; ii) it simplifies component performance prediction in the TF coil inboard leg region by avoiding many of the sliding interfaces (and associated uncertainty of the friction behaviour) inherent in a bucked design. iii) it allows the use of keys in the TF coil inboard leg curved region, which provide effective out-of-plane support Superconductor Type and Operating Temperature The selection of the conductor operating temperature is linked both to the choice of operating fields, the choice of superconductor technology and the design of the cryoplant. The field level required from the TF coils (about 12T) restricts the choice of superconductor to the high field A15 types, of which Nb 3 Sn is by far the most developed. At somewhat lower fields (11T), NbTi cooled with 1.8K superfluid helium would be an option but the reduction in field would place unacceptable constraints on the plasma operation. Heat removal (especially of nuclear heat) is an additional constraint on 1.8K systems For the CS, the performance is substantially improved (in terms of available inductive flux) at high field operation as long as the superconductor can provide a high current density, and Nb 3 Sn at about T represents the optimum choice, compatible with the capabilities of the structural material. This is consistent with helium coolant supply temperatures in the range 4.4 to 4.7K and temperatures in the peak field region of up to about 5K. The PF coils may operate at substantially lower fields and it is possible to constrain the coil operating currents to allow the use of NbTi with supercritical helium (i.e. the helium inlet temperature is K). This operating temperature limits the maximum field on the PF coils to about 6-6.4T and imposes some minor penalties on both the coil size (the winding packs may not Design Description Document 11 Magnet Page 12

14 be too compact to avoid high fields) and the allowable coil operating scenarios, but these are felt to be acceptable in view of the considerable manufacturing simplification resulting from the use of NbTi with supercritical helium. The selection of Nb 3 Sn superconductor for TF and CS coils has important impacts on the configuration and manufacture of these coils. Nb 3 Sn superconductor is a brittle material that is formed from the ductile form by a heat treatment at about 650 C for about 200 hours. All conductor forming operations have to be completed before this heat treatment and subsequent conductor handling must be strictly controlled to avoid damage caused by excessive strain. The superconductor performance is affected by applied longitudinal (i.e. along the conductor) strain (generally, compression produces a degradation and limited tension an enhancement). As the thermal contraction coefficient is smaller than steel, the choice of structural material associated with the strands (usually in the form of a jacket) has an impact on the performance. The choice of the strand performance requirements is a balance between the best available technology, cost and individual coil requirements. Selection of over-ambitious parameters leads to excessive rejection rates in the manufacturing and ultimately extra cost/schedule delays. Both Nb 3 Sn and NbTi (particularly the former) have undergone substantial industrial development over the long duration of the ITER activities (some of it as a consequence of the ITER model coil activities) and it is appropriate to take account of this in the latest (2006) strand specifications. The requirements for the Nb 3 Sn TF and CS coils are different (the TF are substantially steady state, the CS pulsed). Obviously there is some advantage in using a lower loss strand in the CS coils if it is available. The strand values used for design are given in Tables and The coil cooling requirements have been calculated assuming strand with moderate to high hysteresis loss levels (hysteresis loss 1000mJ/cc in the TF and 600mJ/cc in the CS which is expected to be conservative as regards overall heat loads. Higher loss strands could be tolerated in the CS coils (see section 2.1 and 2.5) at the cost of a small extra heat load, if offering advantages in cost. The defining parameter for the acceptability of the Nb 3 Sn strand is the critical current density at the coil operating conditions. From this, the performance at the conventional measuring point of 12T and 4.2K can be derived using one of the Nb3Sn strand scaling formulae such as the Summer's scaling (this and others are defined in DRG1 Annex, Superconducting and Electrical Design Criteria). Application of a strand scaling formula requires data on the strand characteristic parameters which vary between formulae and depend on the strand internal layout /heat treatment history, and must be obtained by measurement of jc(t, B, ε). As an example the Summer's formula is applied in Table using two examples of possible Bc2om, and Tcom values (Co is a derived variable that is given by the critical current values). Strand jc values are conventionally given in the external-strain-free state whereas strands at the cable operating condition are degraded by thermal compression from the jacket as well as transverse load effects. The estimated total strand strain in the cable has been derived by interpretation from ITER model coil and short sample test results, using different scaling formulae (see sections , and ). The degradation in jc is represented entirely by an extra strain component in this model. This is not entirely satisfactory, particularly as regards the transverse load-rleated degradation, and is discussed further in section The values given are intended to include an allowance for cycling effects (large conductors can show some extra drop in performance over the first few hundred load cycles). Design Description Document 11 Magnet Page 13

15 Table Performance Assumptions for Nb 3 Sn Strand CS TF Jc (A/mm 2 ) requirement (at 10μV/m) Hysteresis loss (design basis/max acceptable) mj/cc of strand +/-3T cycle 260 at 12.7T effective field and 5.2K, -0.74%* 291 at 11.3T effective field and 5.7K, -0.77%* 600/ /1000 Cu:non-Cu RRR (after coating and heat >100 >100 treatment) Strand diameter 0.83mm 0.82mm n value 12T, 4.2K, from 10 to μv/m Cr plating thickness [μm] Reference heat treatment at which jc is achieved Possible scaling interpretations to derive the jc at 12T and 4.2K Jc (A/mm 2 ) 10μV/m 12T, 4.2K, unconstrained isolated strand with assumed strain -0.25% based on Summers scaling** 210 o C for 50hrs, 340 o C for 25hrs, 450 o C for 25hrs, 575 o C for 100hrs, 650 o C for 200hrs Transition rate 5 C/hr 210 o C for 50hrs, 340 o C for 25hrs, 450 o C for 25hrs, 575 o C for 100hrs, 650 o C for 200hrs Transition rate 5 C/hr Bc2om** (Summer's scaling) T Tcom** (Summer's scaling) K Co** (Summer's scaling) A/mm x x x x10 4 * these values are the assumed total strain at the design point ** values for use in the Summer's scaling law, see DRG1 Annex, Magnet Superconducting and Electrical Design Criteria and L.T. Summers, M.W. Guinan, J.R. Miller, and P.A. Hahn, A Model for the Prediction of Nb3Sn Critical Current as a Function of Field, Temperature, Strain, and Radiation Damage, IEEE Trans. on Mag., Vol. 27, No27, March 1991 Similarly for NbTi the strand acceptance conditions is its performance at the coil operating conditions, and the performance at the standard measuring conditions (5T and 4.2K) depends on the strand scaling formula parameters. The strand in the correction coils has the same requirements as P2/3/4 although the Cu:non Cu ratio is 1.4. Design Description Document 11 Magnet Page 14

16 Critical NbTi current density (A/ mm 2 coil operating point Table Performance Assumptions for NbTi Strand P1/6 P2/3/4 P5 Side CC T&B CC 994 A/mm A/mm A/mm 2 at 6.5K, 4T at 6.5K, 5T at 7.6K, peak field peak field 3.4T peak 196 A/mm 2 at 6.5K, 6T peak field 193 A/mm 2 at 7.3K, 4.2T peak field field Filament diameter (μm)** 7* 7* 7* 7* 7* Outer Ni coating thickness (μm) n 4.2K, between 10 and 100μV/m Strand diameter (mm) Cu : non-cu RRR >150 >150 >150 >150 >150 Possible scaling interpretations to derive the jc at 5T and 4.2K based on Bottura scaling *** jc at 5T and 4.2K, A/mm Scaling parameters Tco, Bc2o,Co A/mm 2 Scaling parameters n, alpha, beta, gamma 9.2K, 14.5T, 8.03x K, 14.5T, 8.03x K, 14.5T, 8.03x K, 14.5T, 8.03x K, 14.5T, 8.03x ,0.57, 1.0,0.57, 1.0,0.57, 1.0,0.57, 1.0,0.57, 0.9, , , , ,2.32 * these parameters are compatible with a strand without internal CuNi barriers. For some coils used with high frequency feedback control (side correction coils) it may be appropriate to specify a lower loss strand once the power supply controller design is finalised, section ** the filament diameter is set to avoid export licensing problems under the Wassenaar Agreement regarding sensitive technology and this is therefore the minimum value *** L. Bottura A practical fit for the critical surface of NbTi, IEEE Trans App Sup, Vol 10, No 1, March 2000 The preferred design configuration for the cryoplant is to supply liquid helium at close to atmospheric pressure (since this limits the impact of leaks). For initial design purposes, temperature differences within the heat exchanger of 0.1K and 0.1K along the cryogenic supply lines (0.1K) were assumed (more precise heat loads are used for simulations, see Table ). This implies that the base helium inlet temperature to coils and structures is chosen as 4.5K and that the maximum temperature of the heat exchanger bath is 4.3K (optimisation of the cold compressors and circulating pump work suggest the most efficient bath operating temperature is 4.15K for the reference pulsing scenario, see V. Kalinin et al, ITER Cryogenic System, paper presented at 15th International Toki Conference, Japan, Dec 2006)). To allow smoothing of the heat loads to the cryoplant (which is essential to avoid excessive cryoplant capacity), this coil inlet temperature may vary in the range K. The inlet cooling temperature for structural components may vary up to 4.9K. In the event Design Description Document 11 Magnet Page 15

17 that extra heat loads occur in operation (for example, due to extra nuclear heating, higher AC losses or more resistive conductor behaviour), the primary heat exchanger operating temperature can be reduced to 3.7K. Any drop in cooling power may in this case be compensated by adjustments to the machine plasma pulse rate Central Solenoid Jacket Material The main function driving the design of the CS is the generation of inductive flux to initiate, ramp up and maintain the plasma current. Flux generation in the solenoid is improved by: i) the choice of a maximum field compatible with the variation of the superconducting strand current density with field; ii) the use of the highest allowable tensile stresses in the structural support material. The structural support cannot be poloidally continuous because of eddy currents generated by coil pulsing and is most conveniently provided as an integral part of the superconducting cable (i.e. as a thick conductor jacket). In earlier designs of the CS, the option of applying this material after the heat treatment of the Nb 3 Sn cable was maintained, with spiral U channels being welded around the reacted cable contained in a thin Ti jacket for the heat treatment. Industrial assessments of the feasibility of this concept showed considerable problems in maintaining tolerances of the final conductor shape (due to the welding distortion) as well as the high cost and risk of cable damage. The design option finally selected is therefore a thick square shaped jacket with a circular cable inside, heat treated together. The jacket is made from extruded lengths (typically 5-10m long) which are assembled by butt welds. The flux generation capability of the solenoid at a given maximum field and outer radius is determined by the thickness of the winding, which is controlled by the performance of the superconducting cable (the higher the field, the lower the current density) and the allowable tensile stress in the jacket. The mechanical requirements for the CS conductor jacket material are primarily a high fatigue resistance to stress cycling, resulting from the pulsed operation (the CS swings from peak positive current to peak negative through a plasma pulse and hence experiences two stress cycles per plasma pulse). The fatigue resistance of candidate materials is substantially affected by the heat treatment and the frequent occurrence of butt welds, unavoidably in high stress regions. The jacket material can also affect the performance of Nb 3 Sn superconductor due to differential contraction with the strands from the heat treatment temperature. The result of the development programs associated with the ITER model coils has led to two options for the jacket material. 1) Incoloy 908, as developed and used in the CS Model Coil (CSMC), which is a nickel based superalloy. 2) A modified stainless steel capable of undergoing the heat treatment without embrittlement. Two variants exist, one as used in the TFMC jacket and one as developed from a high strength cryogenic steel known as JK2LB. The impact on the flux generation of the peak field is illustrated on Figure This figure shows a CS designed with a steel conductor (which has a high differential contraction with the superconductor) at a tensile stress of 400MPa. As the field increases, more structure and more superconductor is required and the winding becomes thicker. This reduces the flux. At about 13T, the flux reaches a maximum, and above 14T it drops. The weight of Design Description Document 11 Magnet Page 16

18 superconductor in the coil increases continuously. One the basis of such optimisation studies, the operating point is selected as 13T at initial magnetisation (EOB is slightly lower) Figure Optimisation of the CS Use of Incoloy for the Extruded Jacket There are two significant advantages of Incoloy. It is a precipitation hardened superalloy (hardening occurs during the early stages of the Nb 3 Sn heat treatment) with a very high fatigue resistance, and the thermal contraction between the reaction heat treatment temperature and 4K matches better than steel that of Nb 3 Sn. It was selected as the ITER CS reference on the basis that the process for heat treatment of Incoloy has been well established in the CSMC fabrication. However, the problems of Incoloy 908 are well known. It is highly sensitive to stress accelerated grain boundary oxidation (SAGBO) during the Nb 3 Sn heat treatment, which requires very strict control of the heat treatment atmosphere (O 2 <0.1ppm). In the CSMC, the weld filler was not optimised which led to hot cracking in many welds. This did not represent a problem for the CSMC which underwent limited cycling, but would not be acceptable for the CS itself. The issue is subsequently believed to have been solved with a new filler metal. There are still unresolved issues concerned with the possibility of cracking during annealing in fabrication, and the possibility of (undetectable) crack initiation even with low oxygen levels within the jacket in the presence of high tensile residual stresses. Design Description Document 11 Magnet Page 17

19 Use of Stainless Steel for the Extruded Jacket Stainless steel is a generic term and does not adequately define either the specific alloys required to withstand the Nb 3 Sn heat treatment without embrittlement due to carbide precipitation, or those that have good cryogenic properties. The two options for the CS are either a very low carbon 316LN (as used in the TFMC) or a specially developed cryogenic steel JK2LB. JK2LB is a high manganese steel with added Boron to reduce the tendancy to carbide precipitation during the heat treatment. The problem with a very low carbon 316LN variant is that the carbon levels need to be <0.018%, preferably <0.015%, and that below 0.02% is difficult to achieve reliably in arc furnaces due to carbon introduced from the electrodes. JK2LB allows a higher carbon content, in the range %. The sensitivity to carbon precipitation is also worsened by cold work of the material, so sample preparation must accurately reflect the conductor manufacturing. There is an advantage in a solution heat treatment of the jacket material before jacketing, although this reduces the final yield strength. Development of the final JK2LB specification (see section ) has required careful adjustment of the nitrogen content, since high nitrogen levels also enhance carbide precipitation but low nitrogen reduces the yield stress. Stainless steels that are co-reacted with the Nb 3 Sn strands also reduce the strand performance due to differential thermal contraction from the heat treatment temperature (650C) to 4K, resulting in a compression of the strands and lower cable space current density. The results from the TFMC suggest that this compression is less than would be expected from the contraction coefficients of the materials, probably due to some relaxation due to the 'springiness' of the large cable, but that values are significantly higher (i.e. the current is more degraded) than with Incoloy. Precise comparisons are difficult because of the additional performance degradation associated with transverse magnetic loads (see section and N. Mitchell, ITER_D_22FT5C v1.0 (Operating strain effects in Nb3Sn cable-in-conduit conductors)), which is different for Incoloy and steel. It appears that Incoloy Nb3Sn conductor offers an effective strain advantage of about 0.15% compared to steel at the operating conditions of ITER. The stack of 6 CS winding modules is held together by a vertical support structure. In operation, the outer modules are sometimes attracted towards the inner ones, sometimes repelled, causing either a vertical compression stress at the equator or gaps at the module interfaces. To suppress the gaps and restrain the outer modules, the vertical support structure provides a pre-compression. This precompression is achieved partly by wedges inserted at room temperature and partly by differential contraction between the CS stack and the vertical tie-plates of the precompression structure during cooldown. There is some advantage to a tie-plate material with a thermal contraction coefficient which is higher than that of the conductor jacket material since a lower pre-compression has to be applied at room temperature assembly. This difference is available with Incoloy jacket and steel structure. The two candidate steels have practically the same contraction from 650C to 4K. They differ significantly however in the thermal contraction from room temperature to 4K. JK2LB has a contraction of -0.2%, 316LN -0.29%. The combination of a JK2LB jacket and steel tie plates, the lower thermal contraction of JK2LB provides adequate extra precompression. Design Description Document 11 Magnet Page 18

20 Conclusions Due to the better thermal contraction match with Nb 3 Sn, an Incoloy jacket would offer a lower strand cost compared to a steel one, although this is partially offset by the higher cost of the Incoloy material and the extra precautions needed during winding. Also, assessments of the CS and TF model coil performance suggest both that the reduced compression of the strands with an Incoloy jacket is increasing the sensitivity to transverse load effects, and that the thermal compression of the strands with a steel jacket is significantly relaxed in large cables due to the 'twist' of the strands. Incoloy has well-known problems of cracking when heat treated (or welded) in the presence of even very low levels of oxygen. Although these problems can in principle be controlled, as with the CSMC manufacture, they complicate manufacture and represent a certain risk. The risks have been increased as the CS conductor now has more requirements for welding on the conductor jacket before heat treatment to form the helium inlet and joints (as it is a pancake, not a layer, winding) than the layer wound design in The ITER IT opinion is that Incoloy is not a suitably mature material for an industrial environment with these requirements, even though it has been successfully used in the CSMC development. Of the two options for steel, the JK2LB is selected as the reference since it is the only option for which a full qualification programme is being carried out. Although this is still ongoing, the present indications are that it satisfies fully the ITER mechanical property requirements (see section ) Conductor Currents and Coil Voltage Levels The coil voltage levels of the TF coils are dominated by the requirement for a fast discharge in the event of a quench. In the CS, PF and CC voltages in normal plasma operation are comparable to, or in excess of, those in a fast discharge. The current density in the conductor is increased as the time constant for the fast discharge is decreased, as less copper is required in the cable to provide thermal protection. For a fixed allowable maximum voltage, the time constant of the discharge is inversely proportional to the operating current, and there are therefore advantages in selecting the maximum possible conductor current. Alternatively, if the time constant is limited by external considerations (such as induced currents in other components), a higher conductor current allows a lower operating voltage level. The maximum conductor current is limited by two considerations: i) The capability of the switches in the power supply system. For the CS and PF coils, which require regular switching operations, this is 45kA. For the TF coils, where fast discharge is expected to be an infrequent event, this is 70kA. ii) The superconducting cable carries a local transverse load due to the magnetic forces that accumulates on one side of the cable, increasing in proportion to the square of the total current. The limit is well not defined, but permanent superconducting property changes have been observed in single strands and small cables under mechanical strain loading comparable to that expected in the ITER CS and TF. Assessments of the CSMC, CSI, TFI and TFMC also show a clear load related drop in superconductor performance compared to expectations based on strand data.. This performance drop has already been included in the CS and TF design margins. It is considered unacceptable to increase the TF conductor current above the present level of 68kA at 12T. Design Description Document 11 Magnet Page 19

21 The maximum coil terminal voltage or conductor to ground voltage that may be specified for any coil in normal operation has been defined as 20kV under normal conditions. This voltage level is well within the capability of a multilayer polyimide insulation wrap and can also be withstood by an epoxy-glass system (without voids). The choice of a combined polyimide film-glass-epoxy insulation system for all coils gives therefore a double level of protection up to 20kV against insulation breakdown. The corresponding test voltages on the coils may be up to 41kV (see section ). The actual coil design and test voltages may be less than these levels, depending on the application. The electrical design criteria (see ) require the insulation to be designed to tolerate also single shorts to ground or convertor faults such as failure of the thyristor triggers (as an anticipated upset condition). The first generally results in a doubling of the ground voltage level, and the second increases the terminal voltage (and of course the ground voltage) by a factor of about 1.4 (although only transiently). In some cases (for example in the PF coils) other anticipated upset conditions (in particular the incorrect simultaneous operation of the convertors and switching units) also result in significant overvoltages. Anticipated upset conditions can be the deciding factor for choosing the coil test voltage. In the TF coils, the discharge time is not the minimum compatible with the allowable coil terminal and ground voltages, due to the stresses caused by induced poloidal currents in the vacuum vessel during the discharge. The discharge system uses resistors with a positive temperature coefficient. As the current (and hence coil voltage) drops, the resistance increases and increases the voltage, giving a higher rate of energy extraction. The coil discharge is not exponential and values quoted for the discharge time constant are the 'equivalent' exponential value (which gives the same energy extraction to a constant resistance discharge circuit. The increase in resistance is over a factor of 2. For a given equivalent time constant this reduces the peak coil voltage and so gives lower induced currents in the vacuum vessel. More details are given in DRG1 section 1.9, table 1.9-1, PDD section 1.3.3, DDD1.5, sections 2.1.2,2.2.6 and , as well as the analysis of the vacuum vessel eddy current dependence on the discharge resistor characteristics given in G 40 RI W0.1, B. Bareyt, Update of the Assessment of the TF Coil Discharge Parameters and Passive Structure Electro-mechanical Stresses During a Fast Discharge, March The use of non-linear resistors reduces the vacuum vessel loads by over 25% compared to constant resistors and is critical to maintain acceptable loads with the TF time constant given in Table To take advantage of the allowable operating voltage levels, the TF coils are connected in series in pairs within the cryostat to a single discharge resistor, reducing the number of current leads, busbars and switches/discharge resistors by a factor of two. The TF coil voltages during a fast discharge are also sensitive to the different opening times of the 9 circuit breakers, and also to the complete failure of one or more breakers to open at all. The nominal (ideal) single coil terminal voltage is 3.55kV and the ground voltage +/-3.55kV. A failure of a circuit breaker to open, or switch jitter, could increase the ground voltage to 5-6.2kV. A single ground fault would increase the ground voltage to 7.1kV and to 12.5kV if it is combined with a single switch failure to open. A single short to ground (analysed in section 2.4, Appendix 6a, section 4.3) results in ground voltages elsewhere in the system increasing to about 20kV when combined with a failure of breakers to open. In the CS coils, the terminal and ground voltage limits are 6/12kV respectively, in the PF the limits are both in the range 12-16kV respectively (analysed in section 2.4, Appendix 6b). Design Description Document 11 Magnet Page 20

22 These have been selected after extensive simulations of plasma scenarios, particularly initiation and control. In normal operation, the coil voltage can be actively driven by the convertors (when energy may be directed both into and out of the coil) or passively by the resistors and switching units (when energy is directed out of the coil), or a combination of the two. To provide flexibility in plasma operation, a range of scenarios are possible which have different voltage requirements. Fast discharge of the coils can be included within the same voltage envelope. To allow full flexibility for choice of scenarios, the coils are designed to operate at the maximum terminal voltage that can be provided by the power supplies (although there may be power restrictions in using this). Some PF coils have multiple convertor units as well as common convertors for control purposes and this prevents the use of a symmetric virtual ground for each coil, which could limit the ground voltage to half the terminal voltage. Because of these multiple power supply arrangements (and also to permit future modifications of the control system), all the PF coil ground voltages are selected to be the same level as the terminal voltage. For the CS, a symmetric virtual ground system on the resistors (which dominate the applied voltage) allows a lower ground voltage. In the anticipated upset category, shorts to ground (analysed in section 2.4, Appendix 6b) result in minor increases in ground voltage levels. A more serious fault is the failure of the triggers to the convertor thyristors, which puts an AC voltage on the coil with an rms value at about the intended DC value (so the peak voltage level is about 1.4 times the intended DC value). This is a transient condition lasting only a few 10s of ms TF Coil Centreline Shape The shape of the TF coil perimeter has been reached by a process of iteration. The conventional lowest in-plane stress TF coil shape is the well-known bending moment free D. However, small departures from the ideal shape, as long as the resulting in-plane bending stresses are within acceptable limits, can give a better fit around the internal components and result in a reduction in coil material. The outer intercoil structures, required to support the out of plane magnetic loads, also provide an in-plane constraint and prevent the achievement of an ideal bending moment free condition. The initial TF coil shape was therefore chosen to fit closely around the in-vessel components on the inboard equatorial and top and bottom regions (i.e. allowing space for divertor, blanket, shielding and vacuum vessel). On the outside, the TF coil position is fixed by the NBI duct access constraint and the allowable toroidal field ripple level in the plasma. This initial shape is elongated vertically by a total (overall) of about 2m from the bending moment free shape for 18 coils. Various sensitivity studies were performed to assess if changes in coil shape could lead to an improvement in the in-plane stress system (although this was not unacceptably high in the initial configuration). i) Further increases in coil elongation rapidly increase the toroidal tensile forces in the OIS middle sections and/or the separation in the inner poloidal shear key region. There is also a cost increase as the perimeter/magnetic volume increases. This space is wasted as it is not used by the in-vessel components. ii) Radial outwards movement of the peak of the D (i.e. an increase in the curvature radius at the top and bottom of the inboard straight leg, effectively shortening the straight length of the inboard leg as the total coil height remains constant) result in an Design Description Document 11 Magnet Page 21

23 increase in in-plane stress (due to bending) in the inboard curved regions, as well as higher toroidal tensile forces in the OIS middle sections. Such a profile would in any case require some adjustment of the in-vessel components, particularly the configuration of the inner divertor plates, and may not be acceptable. It was concluded that the initial close-fit option was the best choice TF Winding Pack: Plates and Square Conductors Two options have been considered for the TF coil winding pack configuration: - the radial plate design where the TF conductors use a thin circular jacket and are placed in spiral grooves machined on steel radial plates; - the square conductor design where the conductor uses a thick-walled square jacket. In the following sub-sections, the two options are compared Insulation For both designs, radial plates and square conductors, the recommended procedure for the application of the conductor insulation is to use vacuum-pressure impregnation (VPI). After heat treatment, the conductor is wrapped with insulation tapes (a combination of glass and glass/polyimide) by carefully 'unspringing' it from its wound form to gain access without permanent plastic deformation or strand damage. It is then either transferred to the radial plate (if heat treated separately) or replaced in the groove after temporary local removal (if the radial plate is also used to hold the conductor during the heat treatment) and the cover plate is welded closed, or reformed into a double pancake. The use of polyimide film allows voltage testing of the conductor insulation before impregnation and gives a good guarantee that there is no pre-existing defect. An insulating layer is built up around each radial plate (or double pancake) and this assembly is filled with epoxy resin in a single impregnation step and cured. The plates (or double pancakes) are bonded together to form a winding pack with ground insulation in a final impregnation step. The radial plate and square conductor designs use therefore similar insulation manufacturing procedures. Even though the manufacturing procedures are similar, there are differences in the conductor geometry and operation conditions which are expected to give some advantages to the radial plate configuration in terms of the conductor insulation long term quality and reliability. i) The jacket with a circular outer cross section is the optimum shape to apply the glass and polyimide insulation tapes. The result is an insulation which is uniform in thickness and also uniform in the relative glass/polyimide/epoxy content. This insulation is robust since it can contain a high density of glass and polyimide film. ii) During the magnet operation, the Lorentz forces acting on each conductor are transferred to the plate, without accumulation of forces on the conductor and its insulation. As a result, almost no primary load is applied to the conductor insulation and there is no degradation leading to damage due to mechanical cycling. However the ground insulation experiences the same loads for each configuration. iii) With circular conductors in radial plates, the turn insulation is not subject to the stress concentration effects which are always present at corners of square conductors. iv) The turn insulation can be tested under DC conditions, also during the coil lifetime. Design Description Document 11 Magnet Page 22

24 v) The conductor and ground insulation are independent and physically separated by the radial plate. It is therefore impossible for a single insulation fault to affect both conductor and ground insulation. vi) A single conductor insulation fault can be detected by monitoring the resistance between conductor and radial plate. In the event of such a fault, action can be taken before a second fault induces severe damage to the coil system. However, there is also a disadvantage. The conductor insulation experiences a much higher voltage than with square conductors, due to the presence of the radial plate at an intermediate voltage level for a whole double pancake The square conductor insulation is subject to large primary stress due to the in-plane and out-of-plane (cyclic) loads and to stress concentrations at corners of conductors. In particular, the square conductors show local tension regions in the insulation, which would cause local debonding at the corners extending in the worst case to about 20% of the jacket surface. With the square conductor design, the conductor and ground insulation layers are not separated and there is no possibility to detect impending faults (although in most cases the voltage across the layer is low, <50V). The occurrence of an insulation fault leading to significant damage cannot, therefore, be excluded Conductor Jacket Material The choice of conductor jacket material is restricted in either option due to thermal expansion and quench pressure considerations. With a square winding pack, the jacket material must have the same thermal expansion coefficient as the case, otherwise unacceptable case stresses (as well as deformations) develop on cool down. With the radial plates, the plates themselves are sufficiently rigid to isolate the case from any effects of the differential thermal expansion of the conductor in the groove. However, the relative weakness of the groove cover plate, and its closure weld, means that differential expansion between conductor and plate creates deformation of the cover plate in the coil cross-section. This creates a wave shape to the surface of the plate with local regions of tension in the plate - plate insulation. For jackets made of Incoloy 908 (which has a high elastic modulus and a lower thermal contraction than steel) this tension would be unacceptable (reaching a peak of about 20 MPa insulation tension). The quench pressure represents a further design condition for the radial plates. With square conductors, the jacket is thick enough so that the jacket deformation (and associated insulation stress) is negligible. In the plates, the jacket wall is thinner and there is again a tendency to push out the cover plate. The insulation tensile stress can be limited to a small region and < 5 MPa (see section 2.2) with an adequate depth to the cover plate weld (> 3 mm). However, this load condition restricts the use of more compact plate configurations with, for example, two conductors contained in one groove. More detailed studies indicate that either Ti (Ti matches the thermal contraction of Nb 3 Sn and has been qualified for use in ITER by the TFI manufacture) or stainless steel would be compatible with maintaining acceptable insulation stresses, although it is possible that some local debonding between the radial plates could occur with Ti. Design Description Document 11 Magnet Page 23

25 Cost and radial build Although the radial plates allow the use of a highly reliable turn insulation, they cannot be used without a cost penalty due to the radial plate manufacture and additional coil manufacturing steps to transfer conductors onto the plates. The cost difference between the radial plate and square conductor design options has been estimated using the 1998 ITER design unit costs. It has been found that the total TF coil cost with radial plates is about 8% more expensive than with square conductors assuming an identical radial build. The cost of the radial plate option can be reduced if the plates themselves are used as a former to hold the conductor shape during the heat treatment. The conductor is held flat in the plane of the plate but allowed to move within the groove to accommodate small expansion differences. This option is not possible if the conductor jacket is Ti as the differential expansion at the reaction temperature between Ti and steel means that the insulation and tolerance gaps between conductor and sides of the groove are too small. With steel, the conductor is stretched relative to the steel by the cable during cooldown. Based on TFMC experience, if wound on the inside of the groove, the conductor will be in the centre after heat treatment. This procedure saves the cost of a set of formers, reduces the tolerance requirements on the plate grooves and reduces the complexity of the transfer from former to radial plate of a double pancake. The radial plate must be made of a suitable steel and must be completely annealed (to avoid distortion) before the heat treatment. The radial plate design implies, in addition, a radial build penalty. The stress analysis of the TF coil inboard leg indicates that at similar stress levels in the case and radial plate (or square conductor jacket), the radial plate design requires a radial build which is 30-50mm thicker than with square conductors Other Considerations Turn insulation voltage With radial plates, each plate is connected (through a resistor) to the conductor cross-over. The maximum turn insulation voltage is therefore 286V and the voltage between radial plates is 572V for a coil terminal voltage of 3.5kV. With square conductors, the turn to turn voltage is 26V and the voltage between pancakes is 572V. The higher turn insulation voltage, in the case of the radial plate design, is not seen as a disadvantage in view of the high insulation reliability of this design. Fast discharge and recool time: Radial plate design: in the event of a fast discharge of the TF system, eddy currents flow in the case and the radial plates. Heat conduction causes a quench of the superconductor after about 10s, based on the reference thermal conductivity of the insulation (see Annex 10 to DDD11 section 2.1 and 'Heat Transfer from Plates to Conductors: from Toroidal Field Model Coil Test Analysis to ITER Model, S. Nicollet et al, Cryogenics 43 (2003) ). Reassessment of the thermal conductivity of the insulation based on the TFMC results suggests a significantly lower value than the ITER reference (a factor of 6 smaller), which would increase the time until the superconductor quenches to 35s (and reduce the peak pressure). The radial plate temperature rises to about 60K and the conductor temperature to ~40-60K. During such an event, the helium in the TF coils is expelled and is collected in a cold (LN 2 temperature) pressure vessel (volume of about Design Description Document 11 Magnet Page 24

26 1,800m 3 and pressure of 1.8MPa). Recooling and recharge of the TF magnet is expected to take less than 4 days. Square conductor design: in this case, the TF coil case temperature rises but this is not expected to cause a conductor quench. Recooling of the case is expected to take about half a day. Fast discharges are expected to be rather infrequent Conclusions The evaluation of the two winding pack configurations requires a balanced judgement between considerations of totally different nature such as insulation quality, radial build, cost, etc. This judgement is therefore somewhat subjective since it critically depends on the weight and priority given certain aspects of the design. Considering that insulation faults are the most probable cause of magnet failure and considering the difficulties involved in the replacement of a TF coil in ITER, the considerations on insulation reliability during operation have been given a high, overriding, priority over other considerations. This is the basis for the use of radial plates in the ITER TF coils. For the conductor jacket, steel will be used. The evaluation of the TFI results (with a Ti jacket) suggest a larger load-related effect on the superconducting cable performance than in the TFMC (with a steel jacket) and because of this it is not possible at present to design a conductor that takes full advantage of the smaller differential contraction between Nb 3 Sn and Ti compared to Nb 3 Sn and steel. The larger strand requirement for the steel conductor will also be partially offset by easier and cheaper coil manufacture if the use of the radial plate as a heat treatment mold is confirmed. In operation, this choice reduces the radial plate-to- radial plate insulation stresses and reduces the risk of local debonded regions TF Coil Out-of-Plane Support Inner Intercoil Structure The out-of plane support for the TF coils consists of 3 main elements. Firstly, the inboard legs form a vertical torsion cylinder at the centre of the machine. Secondly, the coils are linked at each end of the inboard leg by an inner intercoil structure and thirdly, the OIS sections form four bands of shear panels at the outboard legs. Along the inboard legs, the torsional stiffness is provided by friction forces which result from the large toroidal compressive load on the wedged surfaces. Analysis shows that the minimum acceptable friction factor is about 0.2. This friction factor is not strictly required all along the inboard legs but just at the top and bottom in order to limit the extent of slip near the ends where the curved region starts. The main design driver for the inner intercoil structure (which is situated immediately above and below the inboard straight leg of the TF coils) has been the requirement to achieve acceptable tensile stresses in the curved part of the coil in these regions. The allowable stresses are driven by cyclic fatigue considerations and, depending on the case material, Design Description Document 11 Magnet Page 25

27 fabrication history and welding procedures, are expected to be in the range MPa. Although the outer intercoil structure (forming four toroidal bands around the outboard curved regions of the coil) has only a small influence on the stresses in the inboard curved region (due to the relatively high flexibility of the coils), the configuration of the inner intercoil structure can have a significant impact on the load conditions of the outer intercoil structures. Many configurations of inner intercoil structure have been analysed with a detailed finite element model but the only one that gives acceptable stresses in the coil case, combined with an acceptable stress distribution within the structure (and especially the keys/bolts associated with the structure), is a set of poloidal shear keys between the coils. The intercoil structure itself is absorbed into the coil case, so that (at least at assembly) the wedged region of the coils extends into the curved regions by using the full toroidal angular width available for each coil case. The shear keys run in between the coils in these curved regions, normal to the coil centreline, extending to the inner (plasma facing) surface. A set of three or four keys along the poloidal direction appears adequate. The keys provide full support between the coils and prevent the development of torsion of the case which can make a large contribution to the case tensile stresses. At the same time, the flexibility of the case in bending gives a uniform poloidal distribution of load on the 3 or 4 keys. Comparison of the effect of these keys is given in Table The values for 'no keys' are also without the precompression rings (discussed below), the values with poloidal keys include precompression rings and use four full length keys. It is clear that the keys substantially reduce both case cyclic tensile stresses and the loads on the outer intercoil structures. In the next section, the options to achieve this advantage, without generating higher stresses in the keyways in the case as a result of the shear loads, are examined. Table Effect of Inner Poloidal Keys Parameter No Poloidal Poloidal Keys Keys Peak stress/cyclic component in TF coil case (inner 520/ /265 curved region) MPa Peak out of plane displacement at EOB(mm) Upper/Lower OIS Shear Loads, Peak, MN 4.6/ /5.9 Upper/Lower Intermediate Shear Loads, Peak, MN 25.3/ /23.7 Total Poloidal Key Shear Loads (Upper/Lower) MN /51.2 In the inboard curved region, the radial expansion of the coils during energisation results in the opening of toroidal gaps between adjacent cases. Although small, the radial movement is sufficient to create a toroidal gap of about 0.35mm between the shear key and the key slot. During plasma operation, the shear loads acting on the keys tend to increase this gap to more than 1mm. Key shapes which can tolerate this sort of "breathing" without losing contact have to be square or rectangular and produce high stress concentrations in the key slots. Detailed evaluations of the key and key slot stresses are still underway but at present the preferred solution is to use circular cross-section keys. In order to suppress this undesirable breathing effect, and ensure that the keys do not become loose in their slots, the TF coils are put under a centripetal pre-load at assembly. This pre-load is provided by upper and lower pre-compression rings at the top and bottom of the TF coil inboard legs. The rings are tensioned at assembly and the load is transmitted to the TF coils by bolts oriented in the radial direction. The TF coils are therefore put into toroidal compression Design Description Document 11 Magnet Page 26

28 (effectively, the wedged region of the central vault is extended above and below the inner straight leg) and toroidal separation in the key region is much reduced. The rings are attached to each TF coil and therefore requires eddy current barriers. The impact of the rings on forces and stresses is summarised in Tables and 5. The required radial inward preload is 35MN at 4K applied at the top and at the bottom of the inboard leg of each TF coil. The rings also substantially reduce the toroidal loads in the intermediate OIS, see section Table Effect of Precompression Rings on Stresses and Fatigue Life in TF Coil Case Maximum Principal Stress (MPa) Maximum Stress Range (MPa) Allowed number of cycles In base metal In welds Bottom of straight leg With rings ,000 99,000 Without rings ,000 55,000 In front of first key With rings ,000 90,000 Without rings ,000 55,000 Table Effect of Precompression Rings on Maximum/Average Load on the Inner Intercoil Structure Shear Keys (4 Keys) in MN First Key Last Key (nearest inner leg) With ring 18.4/ /12.5 Without ring 20.0/ /11.6 To be effective, these precompression rings need to have a significantly lower elastic modulus than that of the case, so that the precompression is not sensitive to assembly tolerances and settling effects. A thermal contraction coefficient larger than that of the case is also advantageous. The space available for the precompression rings between the CS and the inner PF coils is limited and a material that can provide the necessary hoop force within that space is a unidirectional glass fibre-epoxy composite. Either S2 or A type would be suitable, with minor adjustments to area and pretensioning to suit the material strength and thermal contraction. A very high level of pretensioning is required at room temperature, up to about 30% of the ultimate strain (see section ). High strength steel (i.e., 40% cold work) may also be considered (the cross-section including insulation can be comparable to that of the glass fibre composite), although the radial pre-extension of the steel is only 9mm compared to about 25mm with fibreglass, making it more sensitive to settling effects in operation. Any solution based on other metallic materials would require a larger (about a factor of 1.5 larger) cross-sectional area than the glass fibre solution. The rings need to be placed close to the curved part of the coil, as extensive flange connections tend to rapidly reduce the effectiveness of the rings due to the flexibility of the flanges. The precompression is applied by stretching the rings using the 'Superbolts' in the flanges, Figure These are radial bolts between the rings and the back of the coil case in the upper and lower curved regions that have a built in pre-tensioning system. Some form of Design Description Document 11 Magnet Page 27

29 pre-tensioning system is required for these large bolts since the strength of a screw fastener increases with the square of its diameter whereas the torque required increases with the third power. Bolts with a diameter greater than about 2.5cm cannot be effectively tightened with wrenches. Standard hydraulic pre-tensioners occupy significant radial space (several tens of cms) and can only be applied when the CS is removed. Their use in this situation is also uncertain as many bolts must be tightened gradually to keep the PCR circular, and stud tensioners are not suitable for repeated removal and re-attachment. Superbolts can be activated with a very small radial space (few cms) and allow easy tightening in small steps. The stresses in the rings are dominated by their radial elongation at room temperature. The material is stronger at 4K and there are no significant extra stresses due to out-of-plane movement of the TF coils (this is very small as the rings are close to the poloidal shear keys that restrain any coil rotation). If settling in operation relaxes the pre-tension (especially if a steel ring is used)., it can be reapplied by accessing the 'Superbolt' tensioning nuts inside the cryostat, without removal of the CS. Figure Example of a superbolt system showing small jackbolts used for tensioning and hardened pressure washer Table summarises the main requirements for the rings for the glass fibre solution which has been selected as the reference (see also The Pre-compression System of the Toroidal Field Coils in ITER ITER_D_24MLHB). Table Parameters for the Glass Fibre Precompression Rings Material Glass fibre (unidirectional) Peak tensile stress at room temperature (MPa) 496 Ratio of peak to ultimate stress at RT (estimated) 4.6 Total cross-sectional area of three rings (m 2 ) (each at top and bottom) Radial displacement to apply precompression at room temperature (mm) 24 (0.9% strain) Optimisation of Poloidal Key Length There is room in the upper and lower curved regions for 4 keys between the coils. Various sensitivity studies have been carried out to determine the optimum position of the first key (generally, the key loads increase and the case stresses decrease as the innermost key moves Design Description Document 11 Magnet Page 28

30 closer to the straight inner leg), and the optimum distribution of loads between the individual keys. Alternative design options that have been considered are: * Elimination of first key (three key option). * First key (closest to the plasma) is made of 'soft' material. * Shorter first key (half of its original length), possibly with reduced length second key. The first two options were found unsatisfactory (and in any case no acceptable way to fabricate a 'soft' key has been found) Detailed studies on the third option gave peak stress levels as shown in Table Table Effect of Key Lengths on Keyway Stresses Peak/Cyclic Stress 4 Full Keys % % MPa in Keyway 1 964/ / /249 Keyway 2 (lower) 683/ /216 Keyway 3 (lower) 288/ /286 Keyway 4 (lower) 448/ /227 These changes in key length did not produce significant changes in the case stress or the loads on the outer intercoil structures. The third option with % key lengths was therefore adopted as reference Outer Intercoil Structure The use of the precompression rings at the inner intercoil structure has a significant impact on the outer intercoil structures (OIS) due to the changes in the radial expansion of the outboard coil leg under the magnetic loads. In normal operation conditions, it appears that the upper and lower OISs are practically redundant. The IIS poloidal keys carry the out-ofplane loads, thus unloading the upper and lower OISs, and the precompression rings prevent the development of hoop tension in these OIS sections. The upper and lower OIS are nevertheless maintained because they are used for coil positioning during assembly and their structural function is limited to providing local out-of-plane strength and stiffness in fault conditions (for example, a short circuit of a PF coil). The intermediate OISs, directly above and below the equator, are required to support the out-of-plane forces on the outboard part of the coil since the out-of-plane loads are not much affected by the inner intercoil structures. However, the precompression rings cause a significant reduction (by a factor of more than 3) in the hoop tension carried by these structures due a reduction of the radial expansion of the coil cases. The tensile toroidal loads on these structures after energisation of the TF coils are summarised in Table Design Description Document 11 Magnet Page 29

31 Table Tensile (Toroidal) Bolt Loads on the Intermediate OIS Sections in MN (for friction joint, box-type OIS is similar) Upper Equatorial OIS Lower Equatorial OIS With rings Without rings The poloidal extent of the two equatorial OIS belts is essentially determined by the size of the gaps left between the vacuum vessel ports. The OIS should, ideally, act as shear panels between adjacent coils, but due to their limited poloidal extent, significant bending stresses are present and relatively high stress concentration occurs at the re-entrant corners where the OIS connects to the TF coil case. The main design problems arise from the need to combine an acceptable assembly procedure with transmission of both shear and tensile loads through the structure and the inclusion of an insulating barrier. There are two options for the upper and lower intermediate sections of the OIS. The first option is a "friction joint" type of OIS. This OIS is assembled after the TF coils (and the vacuum vessel supports) have been installed, by welding a panel (with a pre-assembled friction joint which incorporates an insulating break between "fingers" compressed by bolts) to the TF coil case on each side. The second option is a "box" type of OIS with insulated bolts, pins and keys forming the joints. Both options have advantages and disadvantages. The "friction" type of OIS requires two major structural welds where distortion is reduced and NDT inspection is simplified if access is available from both sides during assembly (access from the inside is expected to be difficult). However, it provides a good shear capability for shear loads in both poloidal and normal (normal to poloidal) directions. The "box" type of OIS has a more complex geometry and requires bolts and shear keys able to contain both poloidal and normal shear loads. Its advantage, as compared to the friction joint, is to eliminate in-situ welding. It requires, however, precision machining of the keyways. The "box" type of OIS is also associated with much thicker material sections and there may be advantages in considering casting for making these. The high cost of the "box" OIS if forgings or plate welding have to be used means that the viability of this type of OIS is strongly coupled with the feasibility of casting. There are also concerns about the uncertainty (and orthotropicity) of the elasticity modulus of castings, as values between 130 and 170GPa were measured in the R&D programme (compared to GPa for forgings). Inspection of the castings, especially final closure welds, is a major concern for components with high cyclic loads, as generally the coarse grain structure does not permit ultrasonic inspection for defects. The peak Tresca stress for both design options is about the same. The location where the maximum stress intensity occurs is at the attachment between OIS and casing at the equatorial port. The peak alternating stress and the resulting maximum crack size is slightly worst for the "box" type, possibly as a consequence of the smaller and unsymmetric shear carrying structure versus the TF coil casing. The worst location is, in both cases (box and FJ), in the fillet transition between the OIS and the TF case. The initial maximum flaw size is ~110 mm 2 for the friction joint and ~45mm 2 for the "box" (in both cases fatigue life and fracture toughness appropriate for cast material was assumed in this fatigue analysis). Design Description Document 11 Magnet Page 30

32 Even with a preload of the bolts of 5MN (20% larger than planned), the "box" OIS flange opens when the TF coils are energized, along the plasma side of the flange. The opening is in the order of a fraction of a mm. Due to this cyclic opening between the OIS segments, the highest loaded shear key is not compressed into the key slot. A small gap occurs and is made worse by the separating force caused by the shear. These effects cause some cyclic bending stress in the bolts in the order of MPa. Based on the stress problems associated with opening of the OIS joint with the box configuration, and the uncertainty about the material quality of castings, the friction joint configuration has been chosen Nuclear Heating in the TF Coils The thickness of nuclear radiation shielding required for the inboard leg of the TF coils is determined by the local heating that the inner surface (plasma side) of the TF coil case and the first 2-3 turns of the conductor can tolerate. This is, to some extent, dependent on the acceptable cryogenic penalty. It is possible to remove nuclear heat loads from the case and winding of several tens of kw if very high helium flow rates are used, but the pumping power loss can exceed the nuclear heating and the overall cryoplant becomes unacceptably large if regular nuclear operation is required. In addition, if the helium cooling channels become too large, the resultant drop in coil current density becomes unproductive, and it is more cost effective to use the space to increase the nuclear shielding. The design decision has been made to limit the nuclear heating from the reference 500MW plasma to levels comparable with that of the AC losses, in the range 10-20kW when averaged over a 1800s plasma cycle. This prevents it dominating the cryoplant size and assists in smoothing the cryoplant loads (as AC losses concentrate in the ramp up and ramp down times when there is no nuclear heating). At the same time, options are available to operate with extra nuclear heat loads if nuclear heating predictiosn turn out to be inaccurate. This is achieved by adjusting the cryoplant operating conditions and/or the plasma pulsing rate (see section ) PF and CC Coil Maintenance Strategy The PF coil maintenance strategy has a major impact on the PF conductor and coil design. Because of the operational reliability requirements, especially for the electrical insulation, and the difficulty in replacing a coil, the PF coils are designed to meet two requirements: The conductor is provided with double insulation. This is in accordance with the design criteria of section and allows an internal short to be detected before significant damage occurs. Each coil should use a modular design. In the case of the PF coils, the module unit is formed by a double pancake. In the event of a fault in a module, it should be possible to by-pass it and continue operation at full current with the remaining modules. This implies that the PF coils include redundant ampere-turn capacity. The double turn insulation consists of two insulation layers with a thin metal screen in between. By monitoring the voltage of the intermediate screen, it is possible to detect an incipient short, defined as a short in only one of the two insulation layers. To function Design Description Document 11 Magnet Page 31

33 properly and allow the detection of all types of internal shorts, this system requires the use of two-in-hand winding. It should be noted that the two-in-hand arrangement gives significantly higher turn-turn voltages in the coil. This drawback is, however, considered a minor one compared to the advantage of being able to detect shorts. This arrangement (double insulation and two-in-hand winding) allows detection of an incipient short, before it develops into a full short (involving both insulation layers) which would then result in significant damage to the coil and, as a consequence, in the need for a major coil repair or replacement. In the event of the detection of an incipient short in a double pancake, the faulty double pancake must be disconnected and by-passed and plasma operation should be able to continue at full performance. This implies that the remaining pancakes are operated at a higher current. This mode of operation is referred to as the backup mode. Since all PF coils include 8 double pancakes (except coil PF2 which includes only 5 double pancakes), the backup mode involves operation at a current which is 8/7 of the nominal current. In the case of coil PF2, the current required for the plasma scenario and control is significantly lower than the conductor capacity (the cable design has been kept the same as for the PF3 and PF4 coils for reasons of standardization), and the backup mode requirement can be met with four instead of 5 double pancakes. The by-pass of a double pancake implies that the joints between double pancakes are accessible on the coil surface. These joints cannot, therefore, be embedded in the coil ground insulation but must, instead, be external to it. The by-pass operation involves cutting the conductor termination of the faulty double pancake and reconnecting the adjacent healthy double pancakes with external busbar links which are provided pre-installed on the coil surface. This operation is to be carried out hands-on and requires man access inside the cryostat to the joint regions at the outer diameter of the coils. Access requirements for this type of maintenance activities are described in the Remote Handling Procedures Document. It is concluded that the use of double turn insulation and the ability to continue operation with a by-passed double pancake should make a major PF coil repair or replacement unnecessary, or very unlikely, throughout the life of ITER. Should, however, such major repair be required, the following strategy is planned: The upper coils, PF1 and PF2, can be relatively easily removed from the cryostat. For them, major repair work, or rewinding, should be carried out outside the cryostat. For the lower coils, PF5 and PF6, major repair work, including rewinding, should be carried out under the machine inside the cryostat. The PF3 and PF4 coils are trapped by the vacuum vessel ports and are the most difficult to access and repair. For this reason, their resistance to faults has been enhanced by using double pancakes with individual ground insulation. For the Correction Coils, there is inadequate space to allow for redundant pancakes to be built in to the coils. The complete integration of the CCs into the TF coils and the surrounding structures make it practically impossible to consider replacement of one of the side or lower coils. Replacement of one of the upper coils may be possible. However, coil functions can still to some extent be performed by 5 coils instead of 6, by modification of the remaining coil current patterns (and of course by reconfiguration of the feeders within Design Description Document 11 Magnet Page 32

34 the cryostat to remove the failed coil from the series connection with the opposite coil). There is therefore some failure tolerance available in the system Conductor Joint Options The conductor Nb 3 Sn joint options are based on the joints developed in the ITER model coil programme. The model coil tests provided both statistics on manufacturing variability and also data on the joint performance under ITER-relevant conditions (although some extra testing would required for some joints to fully qualify them for ITER use in some situations: the butt joint under conditions of tensile strain across the contact surface is the only example that is part of the reference design). In the early stages of the ITER EDA, attempts were made to standardise the Nb 3 Sn joint layout. Requirements were formulated so that the joints would be formed (including all welding operations) before heat treatment of the conductor, and closure of helium containment would always be by welding. This would have allowed full weld inspection and leak testing and provided a common layout for terminal and interpancake joints. It would have resulted however in the elimination of some new technologies such as the butt joint, and prevented copper-steel joining by brazing or the use of soldered connections. The participants in the model coil program preferred to develop designs individually, with attention only to the ITER resistance specification and AC loss requirements. Earlier versions of the ITER coils used a single pancake winding so that joints were present at the coil inlet as well as outlet. Since the joints are cooled in series with the conductor (to avoid excessive numbers of cooling channels) this gave a strict requirement on the joint heating (both resistance and AC loss). In the present design all coils have a double pancake arrangement with a helium inlet but no joint at the midpoint (nearest the high field) and joints at the helium outlet. This relaxes the allowable heating to what can be tolerated in terms of the overall cryogenic load, which is a 'soft' limit. The joint types are described in more detail in section 2.1, Annex 6. The model coil types include 1) lap joint with a steel box and a copper sole, explosion bonded to the steel (TFMC type) 2) butt joint (cable ends highly compacted before heat treatment and afterwards sintered under high pressure and heating to a thin copper interface sheet) (CSMC type) 3) lap joint with a copper tube and either a copper-steel brazed or copper-monel-steel welded end connection (3 different CSMC variants) In operation, all of these achieved an operating resistance of under 3nΩ (mostly under 2nΩ) which is within the ITER specifications based on the allowable cryogenic loads (TF: 2nΩ, CS 4nΩ, PF 5nΩ) - the CSMC joints in particular did not aim to achieve a very low resistance as this is associated with high AC losses. The ITER Nb 3 Sn coil joints have therefore been selected from these types. The reference for the TF coils (both terminals and interpancake joints) is the lap TFMC type (steel box and copper sole). However, any of the model coil joints would be satisfactory, as long as suitable quality control processes for brazing are available. The reference for the CS coil interpancake joints is the butt type (from the CSMC outer module). Due to the need to minimise radial space, this is the only joint that can be used with the present coil configuration. Similarly, due to the lack of radial space, the reference Design Description Document 11 Magnet Page 33

35 for the CS terminal joints at the coil-busbar extension joint is the butt type. The joint between the busbar extensions and the NbTi busbars is the lap type (since it must be easily demountable). Either the CSMC or TFMC lap type would be satisfactory. For NbTi joints, only one ITER-relevant sample has been tested. This was similar to the TFMC lap type, using a steel box and a copper sole plate. The performance was satisfactory. The PFI test coil will use the CSMC lap type (copper tube with brazed end piece), with in addition a soldered contact between strands and copper tube (the inside of the tube is presoldered and the strand surface, after removal of the Ni coating by brushing, is silvered, and the contact is made by heating after compaction). Assuming that this performs satisfactorily, and that the brazing quality can be adequately controlled, either TFMC or CSMC lap type would be suitable for the PF, Correction coils and within the feeders Superconducting Cable Configuration The Nb 3 Sn superconducting cables used in the coils are the result of development work since the start of the EDA in They have the follow general characteristics - they are built up of successive cable stages each with between 3 and 5 units. The final stage is a twist of 6 'petals' around a central channel - the final cable stage (the petals) have a partial foil wrap (inconel in the model coils) that acts as a barrier to coupling currents while allowing helium transfer into the cable - the central channel is formed from an open spiral that allows helium transfer into the petals - the void fraction in the petals is around 33-34% (locally within the petal, not including petal corner radii) - the cable has an overall outer foil wrap that maintains dimensions during spooling and provides protection during jacketing operations There is now substantial experimental verification of these conductors from both conductor samples and the model coils. In the process of the development, minor adjustments to the configuration have been made and there is improved understanding of some of the limitations of the concept (especially in the areas of AC losses, helium distribution and mechanical behaviour). It may be possible to make more extensive design changes that would at least theoretically relax some of these limitations but it will not be possible to build up a complete database for new conductors in advance of construction. Modifications to the cable configuration are therefore kept minor relative to the model coil configurations. The conductor void fraction (locally in the cable space) is now in the range 33-34% (noting that tolerances on cable and jacket dimensions, plus dimension changes during winding, will give a range of about +/-2% on the nominal value). The model coils were 36% (CSMC and inserts) and 34% (TFMC). Reducing the void fraction has clear advantages as regards a higher overall current density and better strand support. On the other hand, it reduces the helium cooling of the strands, increases AC losses and may lead to increased thermal gradients between the central channel and the cable region. The central cooling channel is maintained. In some coils (TF and PF) it is required to reduce the overall pressure drop. In the CS the diameter is reduced but the central tube also provides stability during the final cabling operations and it is difficult to eliminate it without substantial changes to the petal configuration, and new cabling trials. After destructive examination of the model coil cables, more controls have been introduced on the shape of the central tube. The open surface is not more than 30% of the surface area and the width of Design Description Document 11 Magnet Page 34

36 the spiral opening is limited to about 3mm, with rounded edges on the corners facing the strands. Tests have shown that cabling operations and magnetic forces can otherwise cause local strand damage. The petal wraps have been maintained to control the AC losses. They also have a role in stabilising the cable under the mechanical loads. The material has been changed to steel (which reduces the resistance by almost a factor of 3). Within the TF and PF coils, the coverage of each petal has been reduced from the 80% of the model coils to 50%, to improve helium flow into the petals. Within the CS, the high coupling losses at plasma initiation require a lower coupling time constant to avoid short term resistive behaviour (short term because the coil is rapidly discharged and the impact, beyond a local electric field, is anyway expected to be negligible), and the 80% wrap coverage is maintained. The Nb 3 Sn cables include pure copper strands at the level of the first triplet (i.e. one strand out of the 3 is pure copper, with the same dimensions as the superconducting strands). The Nb 3 Sn strands have now been chosen with a fixed Cu:nonCu ratio of 1.0. The variation of copper in the cable required for thermal protection is provided by extra copper strands, as with the model coils. However, to provide flexibility, some cables may include extra copper strands within the higher cable stages (with the same diameter as the strands). If this option is not used, then there is a significant penalty in current density and/or strand quantity Design Criteria This section provides a brief overview of the most important design criteria for the magnet design. Full details of the criteria are given in documents attached to DRG Conductor Design Criteria The results from the ITER CSMC, insert coils and TFMC have shown some behaviour characteristics that require adjustment of the Nb 3 Sn design criteria previously used for conductor design in the 2001 FDR report. The most important is the low 'n' value of the cables relative to the strand (where n is the power index in the empirical resistive transition relation E = E c (I/I c ) n with E as the electric field and Ec conventionally taken as 10μV/m). 'n' is dependent on the critical current, dropping to 1 as the critical current approaches zero. For Nb 3 Sn strands it tends to be lower than with NbTi, but in cables it is found to be about one half of the strand values, sometimes under 10 at the ITER coil operating conditions. The (relatively) slow transition means that with adequate cooling, Nb 3 Sn conductors can operate well beyond the traditional current sharing point (defined by E = 10μV/m) and the significance of this point as being representative of the maximum achievable operating conditions is much reduced (as an example the TFMC reached local average conditions of E>100μV/m before quench and the CSMC operated up to over 40μV/m without quench). The low 'n' has also a significance below the critical conditions as resistance exists in the conductor at well over 0.5K below the critical current, sufficient to be of concern for the overall cryogenic load. The low 'n' also changes the dynamic stability of the conductor, giving a more gradual build up of resistance at a local disturbance that can allow current redistribution within the cable (if the extra resistive heating can be tolerated). When designing conductors with Nb 3 Sn, the total strain level in the cable has to be known, due to the strain sensitivity of the Nb 3 Sn superconducting properties. The total strain is largely dependent on the thermal strain (due to differential contraction with the jacket and Design Description Document 11 Magnet Page 35

37 coil structural material from the reaction heat treatment temperature) and the operating mechanical strain. The Nb 3 Sn model coil results suggest that large thermal compression on the strands in a cable is relaxed (probably due to the cabling effect) and that there is an apparent transverse load related (i.e. field x current) extra drop in the superconducting performance, probably related to either local bending strains or local transverse compression. The assessment of the total strain level depends on the parametric model used to represent the variation of critical current and temperature in the strands as a function of strain. It is only recently (i.e. late 2003) that direct measurements on Nb3Sn strands with compressive strains in the range typical of those expected from steel jackets have been made. Previously, extrapolations based on tensile measurements have been used, and discrepancies between these extrapolations and direct measurements have been observed, causing differences in the assessment of the total strain within a cable. To avoid errors in the assessment of the expected performance, it is essential that any assessment of total cable strain in operation is associated with the strand parametric model used to derive it. Testing of NbTi conductors for the PF coils has shown a fundamentally different superconducting behaviour, much closer to that of isolated strands. The cables show high 'n' values, close to those measured in strands. In some operating regions, 'unstable' quench is seen (where the cable does not show a measurable resistive development before the quench runaway) Temperature Margin For Nb 3 Sn, instead of working with the traditional temperature margin between operating temperature and the current sharing temperature, we use instead the runaway quench temperature as determined by a simulated ramped temperature test on the actual conductor configuration (i.e. modelling the actual conductor length at high field and the actual mass flow rate, with all the 'steady state' heat loads and in particular the heating multiplication effect due to the resistive behaviour of the cable). The helium inlet temperature at which quench occurs must be >1K above the normal operating inlet temperature during plasma operation. The 'steady state' heat loads are defined as heat loads that are applied for longer than the transit time of the helium through the high field part of the conductor (about 50s in both CS and TF coils). The cable 'n' value should be taken as one half of the strand value (and include the dependence on critical current). For Nb 3 Sn the temperature margin is to provide adequate headroom against unforeseeable effects such as non-uniform current distributions and temperatures and (indirectly) to provide a stability margin. It is not to provide margin against unforeseen cable performance effects (e.g. more negative thermal strain). For NbTi, the temperature margin to the current sharing temperature defined by E = 10μV/m must be >1.5K. Further considerations on the factors controlling the thermal runaway temperature are given in section Copper and Helium Contributions to Stability Strand stability is determined by the 'temperature margin' and also by the copper fraction in the strand and the heat transfer to the helium (which represents a far larger heat sink than the Design Description Document 11 Magnet Page 36

38 metal enthalpy directly available within the strand). The definition of temperature margin affects also how it impacts on stability. For temperature margin relative to the current sharing point, low 'n' Nb3Sn conductors have been shown to operate stably even with a negative margin. Using the definition proposed in , the stability margin at the runaway quench point is of course zero. Copper can contribute both as a heat sink and as an alternate current path when the strand is locally resistive. For low 'n' strands, the direct influence of copper on the early resistive behaviour is weak (although there may be an indirect influence through the value of 'n'). The ratio of copper to non copper (essentially superconducting) material cross sections in the strand is important also for manufacturing considerations as the copper can provide 'lubrication' in the drawing process. The Cu:nonCu ratio will be selected for Nb 3 Sn from the range 0.5 to 1.5, based on existing experience. If cable designs would benefit from values beyond this range, assessments must be made of the manufacturing capability of specific suppliers. NbTi manufacturing experience exists over a much wider range and the copper has less impact to the drawing process. The Cu:nonCu can practically be chosen without manufacturing considerations. With Nb3Sn the familiar well-cooled, ill-cooled 'Stekly' criterion is not an adequate method to select the copper fraction in the strand, as it does not properly model the extended low resistance region above the critical temperature that occurs with low n conductors. It tends to overestimate the influence of the copper. It is possible to derive improved models for the transition region, based on the strand 'n' value and use them to predict stability in the neighbourhood of the thermal runaway point (see also section ). Heat transfer coefficients for use with the traditional model are essentially fit factors based on experimental results, largely with NbTi, which are as high as 1000W/m 2 K. The use of a more realistic resistance model makes these coefficients incorrect and values have to be rederived from appropriate experiments. These are generally lower than the fit factors, in the range W/m 2 K (close to measurements of steady state heat transfer values for wires in Helium at the flow velocities expected in the cable bundle). Such modelling can confirm that the strand has an adequate stability margin but does not provide a route to select a copper fraction. Because of this, the lowest allowable Cu:nonCu fraction in the Nb3Sn strands has been chosen as 1.0. This is an empirical limit, based on present experience. Lower values would probably be acceptable but experience on strand and cable stability performance would have to be built up before cables could be confidently designed. With NbTi, the Stekly well-cooled, ill-cooled criterion is still used as a basis for the design with a 'fit' based heat transfer coefficient of 600W/m 2 K in NbTi. If copper strands are present within the cable (to provide thermal protection) they are considered not to contribute to the stability. This is inadequate and it would be preferable to develop an improved model based on the local peak electric field and 'physical' values of strand to helium heat transfer (see also ). At present the available data is too scattered to allow a reliable model to be developed. With both NbTi and Nb3Sn, the final design must be supported by stability analysis using 1- dimensional modelling of the helium flow and considering appropriate timescales for the transient heating. The overall criterion for stability is that the predicted margin must be twice the calculated maximum energy input from an electromagnetic disturbance. Design Description Document 11 Magnet Page 37

39 Current Uniformity The current in a cable made of parallel-connected strands is most unlikely to be uniformly distributed in each strand. The non-uniformity can be driven by resistive variations in the strands at the joints, or by inductive coupling variations between strands along the cable length. Transition between these two drivers is controlled by the time constant of circulating currents in the cable, which is of the order of 1,000 to 10,000s. In the ITER coils, the CS and PF coils are expected to have inductance-dominated current distributions and the TF coils to have resistance-dominated distributions. The non-uniformity is limited within the cable by the resistive transition of the strands, which produces a voltage that transfers the current out of high current strands into low current ones. Current non-uniformity is not a problem in itself; only if it leads to degradation of the thermal stability level of the cable does it need to be avoided. In some fast pulse coils (which are not typical of the ITER operating conditions or coil design), current nonuniformity are characterised as the ramp rate limit. Also in short samples under near steady state conditions, when the separation between joint and high field may only be a few tens of cms, joint non-uniformity can cause critical current degradation of the overall cable at resistance levels typical of the ITER joints. The ITER coils have various levels of current uniformity control: i) The cables are designed to be fully transposed with the strands in predictable positions around a central annulus, so that a uniform inductance can be expected. ii) The cables have a minimum level of transverse conductance between the strands in each of the final substages. This conductance has to be carefully controlled through the cable void fraction (and hence through the jacket manufacturing tolerances). Too high a conductance leads to a high AC loss, too low and fast current redistribution of current during thermal disturbances cannot occur. iii) The separation between joint and high field is very long, >100m The joints are designed to give uniform contact resistances at the level of the final substage (one sixth of the cable, with about strands), to try to avoid gross current variations within the cable. However, results from the CS insert (where detailed measurements were possible) gave evidence that current non-uniformity at the level of the final substage could be substantial (with current ratios between final substages over a factor of 2 before current sharing). This non-uniformity appeared to be resistively driven (i.e. it originates from the joints or from a series of joints). Fast current ramp tests on the CSI coil did not show any premature quench effects that could be attributed to inductive current non-uniformity effects, although it is not possible to know if such effects were present but had no performance impact (due to current redistribution within the cable) or that the cable transposition is sufficiently good that there are no significant inductive non-uniformity effects. Overall, the CSI results confirmed that the transverse conductance of the cable is sufficiently high to allow it to tolerate current non-uniformity without loss of performance (defined by stability and current sharing margin) Hot Spot The maximum temperature that may be reached locally inside the conductor in the event of a quench has been determined by the differential expansion between conductor materials. Below 150K, materials have a low thermal expansion and 150K is therefore selected as the maximum temperature that may be reached by the conductor jacket. The cable inside the Design Description Document 11 Magnet Page 38

40 jacket may reach up to 250K on a transient basis, as it is much more flexible than the jacket. For conductor design, the limit of 250K has been used together with the conductance and thermal capacity of the cable materials only. When the thermal capacities of part of the jacket and helium are added, the limits are met Electric and Magnetic Field The cable in operation has a significant magnetic field gradient across it, over 1T in the case of the TF conductor. The local electric field on the strands therefore varies substantially. The cable electric field is determined by an average, noting that the cable twist ensures that all strands move into all possible positions in the cross-section. The expression used is: E = Ec A A J op da J c B T (,, ε ) n with E c =10μV/m Jop is the uniform current density, Jc the local critical current density, assuming a uniform temperature, and A is the cable area. The effective magnetic field on the cable is defined as the magnetic field that has to be applied uniformly to a single strand of the cable to get the same electric field as the cable electric field. Due to the low 'n' of the strands in the cable, Nb 3 Sn cables can experience a significant resistance even at normal operating conditions. The heat generated by this may be sufficient to be a design consideration. For the ITER TF coils a maximum cable electric field at any point along the length of 2μV/m is allowed at the nominal operating point. This generates heating of about 0.13W/m, about 1/100 of the peak nuclear heating in the first turn (Fig ), giving a comfortable margin against a thermal runaway where the resistive heating generates more downstream heating due to a temperature rise of the coolant. This limit is clearly a 'soft' limit where even 3μV/m is not likely to cause a significant operating problem. For the CS coils, where the dwell time at peak field is only a few seconds, higher resistive heating could be tolerated. For the CS coils, the resistive heating is always negligible and is not a design consideration. The local electric field at this point is >10μV/m but stable operation is maintained because of the high critical temperature of Nb3Sn which allows resistive heat removal from the strand through a (relatively) large temperature difference to the helium. In the ITER model coils, stable DC operation was possible up to average electric fields above 40μV/m. NbTi cables maintain a much higher 'n' value in operation and a limit on the average electric field is not appropriate since the thermal runaway point can occur in advance of a measurable average electric field. The thermal runaway is (as with Nb3Sn) dominated by the local electric field in the high field part of the cable but, due to the much lower Tc of NbTi, much lower electric fields can be tolerated. This means that the design is limited by the thermal runaway (rather than an average electric field as with Nb3Sn) and the prediction of this is uncertain, dominated by the local strand to helium heat transfer Total Strand Strain in Cable and Transverse Load Degradation Recent experiments on Nb3Sn strands in cable-in-conduit conductors have shown that the cables can display unexpected performance degradation relative to results expected on single strands in isolation. The strain sensitivity of Nb3Sn is well known and the design Design Description Document 11 Magnet Page 39

41 criteria have always included an allowance for degradation caused by the high negative total strain caused by a steel jacket material. However, the present levels of degradation substantially exceed what can be explained by thermal contraction. With the steel jackets used as the reference in both CS and TF coils, the longitudinal strain state is dominated by the differential contraction between the strands and the jacket, from the strain equilibrium state at the reaction heat treatment to 4K. The difference in the thermal contraction coefficients is 0.8% but this can be relaxed by two factors. The steel jacket in the CS is much stiffer than the cable, and practically all the differential contraction is applied as compressive strain on the cable. In the TF, the jacket is thin and is stretched by the cable, giving a relaxation up 0.05%. The cable itself is elastic, and longitudinal strain applied to the cable does not appear fully on the strands (there is a 'spring' effect). The direct longitudinal strain is relaxed and is replaced by a pattern of cyclic bending and transverse compression on the strands. Analytic and experimental assessments of this relaxation suggest that for large (ITER type) cables it is about 80% (Assessment of Conductor Degradation in the ITER CS Insert Coil and Implications for the ITER Conductors (ITER_D_24MG2T v1.0). This value applies to the differential contraction between a complete bare strand (including the copper) and the jacket, not between the filament material itself and the steel jacket. The operating strain of the coil also contributes to reduce the cable compressive strain. The operating strain is typically in the range 0.05 to 0.2% and this again does not appear directly on the strands but is relaxed by the 'spring' effect to about 80% of the differential strain. The local transverse magnetic loads on the cables produce a degradation of the strand superconducting performance. This can be interpreted either as an extra compressive strain on the strands or as a loss of superconductor transport capability, see ITER_D_22FT5C v1.0 (Operating strain effects in Nb3Sn cable-in-conduit conductors), and ITER_D_24L5B4 v1.0 (Transverse load optimization in Nb3Sn CICC design; influence of cabling, void fraction and strand stiffness). The strands carry two loads. One is a distributed magnetic load, and the other is a cumulative load, as the overall transverse magnetic force is transmitted through the cable to one side of the jacket (about 80t/m in the TF conductor). These loads produce a combined pattern of cyclic bending and transverse compression in the strands, both of which can be expected to affect the local filament superconducting performance. The tensile strains appear sufficient to cause filament breakage. Current can locally transfer around these regions of low critical current in the copper/bronze matrix, at the cost of a small electric field. Overall, the cable degrades, diplaying an earlier and more gradual superconducting transition. Bending tests on individual strands show a substantial variation in the sensitivity to applied load and displacement, and the strain at which irreversible degradation occurs ( Critical current and strand stiffness of three types Nb3Sn strand subjected to spatial periodic bending (ITER_D_24MG4S v1.0) ). Other recent results have shown that the effect occurs also in small cables using recent high current density strands (so called 'advanced strands'). It appears likely that the trend to high jc strands is also increasing the strand sensitivity to transverse loads, probably because of enhanced filament cracking (the larger fraction of Nb3Sn material and mechanical bridging between filaments could account for this). There is at present insufficient data to predict these effects and the degradation of a particular strand and cable combination has to be assessed by measurements on the cable. Design Description Document 11 Magnet Page 40

42 In the present design, these strain related effects are accounted for by an extra equivalent compressive strain on the cable and a lower effective n value than found with individual strands. A brief summary of the values used is given in Table Table Strain Assumptions for TF and CS Nb3Sn Cable Design Component TF % CS % Thermal strain of isolated strand Thermal strain due to jacket (80% of thermal strain difference between jacket and isolated strand) Stretch of TF jacket by cable Extra effective compression due to transverse loads % of mechanical strain on conductor in coil The values of 'extra' effective compression due to transverse loads are typical of those found in the Iter model coils when scaled according to the applied magnetic loads, but are lower than those found in more recent cables using advanced strands. None of these data are for cables with the present Iter configuration. The strand values of 'n' are reduced by a factor of 2 for cable design, typically to around 7 at the critical point of the Iter conductors Hydraulic Friction Factors and Inlet Pressure Drop The helium flow passage within the conductor consists of two parallel, interconnected paths. The flow in the central spiral has a higher velocity due to the large hydraulic diameter; the mass flow rate however depends also on the flow through the cable interstices since here the flow velocity is lower but the flow area much larger. The pressure drop in the central channel is a function of the form of the spiral forming the tube and particularly of the size of the side void gaps. Two channels used in the TF model coil have been measured, one known as 'Showa' with a gap fraction (side gap length)/(side gap length + strip length) of 28% and one known a 'Cortaillod' with a gap fraction of 45% and an outer diameter of 12mm. The correlations for the friction factor in the central channel and outer annulus are given in (S. Nicollet, J.L. Duchateau, H. Fillunger, A. Martinez, S. Parodi, Dual Channel Cable in Conduit Thermohydraulics: Influence of Some Design Parameters, IEEE Trans App Sup, vol 10, 1, March 2000, ). More recently, spirals with a smaller diameter (OD 10 and 8mm) in the range proposed for the ITER coils have been measured (S. Nicollet, CEA report AIM/NTT Results of ITER type central spiral friction factor measurements in the OTHELLO facility and application for ITER Coils, April 2003). These small diameters have a significantly higher friction factor than those of the model coils and and it is clear that the friction factor correlations are dependent on the diameter. Friction factor is defined by f in ( p/ x) = - f Pe (dm/dt) 2 /(8ρA 3 ) Central channel For 'Showa' type central spiral OD 12mm f = / Re Design Description Document 11 Magnet Page 41

43 For 'Cortaillod' type central spiral OD 12mm f = / Re ITER with OD 8mm f = 0.54 / Re 0.03 ITER with OD 9mm (average 10 and 8mm) f = 0.45 / Re Annulus f = ( ( 19.5/Re ) ) / ( V ) (for Reynolds numbers from 1,000 to 6,000 and void fractions in the range 32-40%) where p : pressure x : length Dh : hydraulic diameter = 4 A/Pe dm/dt: mass flow rate A : Helium flow area A wrap Total wrap area (overall plus substage) Pe : Wetted perimeter Dcond : internal diameter of conduit Dtube : external diameter of central tube Nstr number of strands (total in cable) N sub number of final substage petals with wraps L perimeter of one substage petal (non-circular) dstr : strand diameter θ average cabling angle V : void fraction of annulus μ: dynamic viscosity (Pa.s) Re: Reynolds number = 4 (dm/dt) / μ Pe The friction factor of the central channel is calculated using a hydraulic diameter equal to the outer diameter Dtube. The flow area of the channel and the wetted perimeter must also be defined using the outer diameter, for consistency. For the cable annular region, the flow area is given by π D 2 cond /4 - N str π d 2 str /4/cos (θ) - π D2 tube /4 -A wrap wetted perimeter is defined by : Nstr π p av str + Nsub L a = d str / 2 b= d str / 2 / cos (θ) p av str = (π /2) (3 (a+b) - 2 (ab) 1/2 ) with a and b being the minor and major axes of the ellipse into which the strands on average deform (noting that a factor of 5/6 on the strand contribution, occasionally seen in this expression, is not included here) Manufacturing variations are expected to create a variation of not more than +/-20% in the pressure drop of individual lengths for the same mass flow rate, at normal operating conditions. The inlets to the conductor (where the He flow has to pass from a penetration in the jacket wall through to the cable annulus and the central channel) can also contribute a significant pressure drop unless the flow can take place over a sufficient length of cable (i.e. the radial Design Description Document 11 Magnet Page 42

44 flow area needs to be several square cms). Measurements are summarised by S. Nicollet ITER_D_23CNEA v1.0, Review of Pressure Drop for ITER Coils. Based on these results, the maximum acceptable pressure drop for the conductor Helium inlets is set as equivalent to the pressure drop over a 20m length of the conductor. It appears possible to achieve values 25% lower than this relatively easily for a range of inlet designs. The He outlets are combined with a joint in the TF and PF coils and the pressure drop is negligible. In the CS, the quad/hexa-pancake winding requires also a penetration through the jacket wall. In this case (where the stresses are low), an equivalent pressure drop of 10m of conductor is permitted Cable AC Losses The coupling loss model of the cable used to calculate AC losses includes two features found in recent experimental measurements on full size cables. Firstly, the coupling time constant of the cable depends on the local magnetic forces (these press the strands together and reduce the cable transverse resistance) and the load cycle history. Secondly, the effective time constant of the cable depends of the rate of change of field, due to screening effects. The experimental measurements are largely those made in the University of Twente cryogenic press, where 1 twist pitch length of the final cable (40-45cm) is placed in a dipole and subjected to a mechanical transverse load. Details of the interpretation of the measurements and the derivation of the coupling loss model is given in E. Zapretilina ITER_D_22FR8D v1.0 (Assessment of AC losses with LOSS code (2005) Data used to derive the models comes from: A.Nijhuis, Yu.A. Ilyin, and W.Abbas, Electromagnetic and Mechanical Performance of eight Prototype ITER NbTi Fullsize CICC s under Transverse Loading up to cycles, Final report, Tasks 3,4,5,and 6, Contract:EFDA-99/502, No:UT-EFDSA , March, 2003 A.Nijhuis et al, Performance of an ITER CS1 Model Coil Conductor under Transverse Cycling Loading up to cycles, Presented at 18 th Conf. on Magnet Technology, 23 October, 2003, Morioka, Japan A.Nijhuis, et al, Parametric Study of Coupling Loss in Sub-Size ITER Nb 3 Sn Cabled Specimen, IEEE Transaction on Magnetics, vol32, no4, July1996,pp A.Nijhuis, et al, Contact Resistance and Coupling Loss in NBTi CICC s with Various Strand Coating and Geometry for the PF R&D Program, Final Report, Contract No.:EFDA_01/597, NO UT-EFDA, , June 25, A.Nijhuis, et.al., Change of Interstrand Contact Resistance and Coupling Loss in Various Prototype ITER NBTi Conductors with Transverse Loading in the Twente Cryogenic Cable Press up to 40000Cycles, Accepted Cryogenics paper, 22 January, Benchmarking was performed on overall loss measurements from the CS model coil and from Sultan measurements with a pulse coil. Basic loss expressions are Design Description Document 11 Magnet Page 43

45 Loss Power P = nτ 2 μ B Ast with nτ = 2* τ eff and τ τ + τ k k *exp( χ / α ) eff = inf 0 hist load k hist k load are terms depending on cycle history and magnetic load χ =db/dt α is a constant determined from experiments = 1T/s 2.5 NbTi k hist = ξ *2.65*(1 exp( N /( ξ *16500))) *exp( N * ξ / 3) k k k 1) * exp( 4ζ ) load = st max ( st max k st max = (1 exp( N /10000)) 0.2 exp( N Nb 3 Sn 2 k ( N, ξ ) = f ( ξ )*(1 + ( k ( N,1) 1)* ξ 1/ ) hist k hist hist / 300) ( N,1) = * exp( N / 6500) * exp( N / 80) * exp( N / 3.5) * exp( N 4.72 f ( ξ ) = 1/ * ξ k k k 1)*exp( 4ζ ) load = st max ( st max st _ max N, ξ) = 1+ ( kst _ max( N,1) 1) * k k st max ( ξ 1/ 2 / 0.1) = *exp( N / 6500) *exp( N / 3000) 0.45*exp( N /15) ξ is the normalised local maximum load, and ζ is the normalised currently applied load Ii,max * B ξ = I max * Bmax Ii ( t)* Bi ( t) ζ = I * B max i,max max N is number of full load cycles The values recommended for the ITER cables are TF nτ(0)(unloaded)=90ms, τ inf =5ms, τo=40ms CS nτ(0)(unloaded)=45ms, τ inf =5ms, τo=20ms NbTi (PF size) nτ(0)(unloaded)=70ms, τ inf =5ms, τo=30ms These loss scalings have been derived for cables that include wraps on the final cable substage (50% coverage for TF and PF, 80% for CS) and which, for the CS and TF cables, have a void fraction of 36% rather than the 33% of the present reference designs. The NbTi strands have a Ni coating, the Nb3Sn have Cr. The recommended values include a correction for the void fraction based on (limited) experimental data which requires verification. The measurements in the University of Twente press are also limited to one twist pitch and may underestimate losses if longer coupling loops can occur in the actual conductor. Design Description Document 11 Magnet Page 44

46 The effective 'single value' time constant for the AC loss assessment are frequently quoted rather than the multiple time constant values given above and are defined as follows: All effective time constants are reported as the quantity measured/estimated at : a) zero frequency, zero db/dt, b)after a certain number of loading cycles, c) under full load : for the CS cable: nτ(db/dt=0, N=100, full load ) =75 ms, (CS1 figure); for the TF cable: nτ(db/dt=0, N=100, full load ) =200 ms (scaled from the figure for the CS cable, taking into account the conductor size and a factor of 2 for the reduced wrap coverage); for the PF cables: nτ(db/dt=0, N=10000, full load ) =150 ms (figure close to the one measured at UT for the cable without wrap.) The hysteresis loss model for the strands uses 3 loss expressions (defined in the Superconducting Design Criteria, Annex to DRG1) depending if field changes are >>B pen, >B pen or < B pen where B pen is the penetration field defined by B pen ( μ o π ) jcf Deff ( B) = [T] where jcf is the filament critical current density, i.e. the critical current over the Nb3Sn cross section and D eff the effective filament diameter. Hysteresis losses are defined in Tables and Structural Design Criteria The structural design criteria have been derived to cover some special features of the ITER magnets and structures that are not adequately covered by a single existing design code such (the ASME pressure vessel code being a well-known example). These features can be briefly summarised The magnets operate in the range 77K to 4K, with most operational life close to 4K. Compared to high temperature components, yield and ultimate strength are increased but fracture toughness remains approximately similar. Fracture is relatively more important as a failure mode than plastic yielding (and safety factors are adjusted accordingly). Yield stress is increased well above the typical level of residual fabrication stresses. Limiting of stress peaks through local plasticity is much less extensive than at room temperature and above. Secondary stresses are more significant and have lower allowables than at higher temperatures. In-service inspection of the magnets is not possible, and operational monitoring of structural behaviour (displacements, for example) is insensitive. Defect assessment based on component inspection during manufacturing is therefore required. Extensive use is made of non-metallic bonding between structural components. These bonding surfaces are generally 'failure tolerant' as regards the mechanical behaviour, but mechanical failure can lead to electrical breakdown At the global level of design, there are 5 sets of criteria that impact directly on the magnet, and hence the overall machine dimensions. Design Description Document 11 Magnet Page 45

47 Static Yield Criteria (Tresca stress limit) for Metallic Components The maximum allowable primary Tresca membrane stress in the material (base metal) is defined as 2/3 times the 0.2% yield stress. In normal operation, the allowable value is increased by a factor of 1.3 for combined membrane and bending stresses and 1.5 (i.e. up to yield) for combined primary and secondary. For welds, in sections characteristic of the coil cases (i.e. a thickness over 150mm), these two values are decreased by a factor of Fracture Criteria for Metallic Components A fast fracture assessment is performed. The design stress intensity factor, Km, should be compared with the fracture toughness K IC at the design temperature. Km is calculated using the expression (in normal operation): Km = Yσ (π a) 1/2 < K IC /1.5 where σ is the maximum principal tensile stress, Y is a stress intensity factor and a is the crack size calculated including growth due to fatigue effects. The factor of 1.5 is a safety factor. Safety factors are also applied to the initial crack size and the rate of crack growth, as defined below Fatigue Criteria for Metallic Components There are two methods of fatigue assessment. The first is material based and uses S-N fatigue life curves to establish the characteristics of a component and then applies a safety factor to define the allowable life. The second is defect based and uses linear elastic fracture mechanics to predict crack growth from a pre-exisiting defect. Both procedures must be applied where there is the possibility of manufacturing defects (i.e. in massive forged components with large welds such as the TF coil cases and OIS, where only small material samples can be tested for fatigue life at 4K. Such samples are unlikely to contain the defects characteristic of the whole section) and the components are not 'failure tolerant' (i.e. there is the possibility of failure by fast fracture and/or the result of failure is major damage to the magnets, either irreparable or resulting in repair times of several years). The fatigue life (SN) procedure is applicable on its own where many identical components are involved that are individually failure tolerant (and where developing failure can be detected by monitoring before major damage occurs). Representative samples of the components can be properly tested (so that the samples include any characteristic manufacturing flaws) to establish an SN curve. Fatigue life without defect assessment can be typically applied to items such as the bolts and keys used in the OIS, and to the individual fingers in the finger joints of the OIS. The first fatigue assessment method uses SN curves established by component testing. The curves are preferably measured for the standard +/- (R=-1) cycle with zero mean stress (not R=0 often used at 4K), if not they have to be converted using empirical scaling rules linking fatigue life to yield/ultimate stress such as Goodman or Soderberg. After scaling from the SN +/- curve for mean stress effects, and correcting for multi-axial cyclic stress components, a safety factor is applied either to the cyclic stress, of a factor of 2, or to the number of cycles, of a factor of 20, using whichever gives the most conservative cyclic stress allowable. Design Description Document 11 Magnet Page 46

48 The second fatigue assessment method is based on linear elastic fracture mechanics (LEFM). The Paris law is used to integrate the growth of an initial defect until it either penetrates the whole component thickness or reaches the critical size and propagates catastrophically. The main uncertainties in the procedure are: i) the material properties, since the crack growth rate factors used in the Paris law are dependent on the applied mean stress level as well as the material fabrication route, ii) the local stress intensity factors that should be applied to the initial defect. These depend on the geometry of the component and the location of the defect, iii) the size and shape of the initial defect. With in-service inspection, new defects that are detected are known to be cracks. During manufacturing, it can be difficult to distinguish between voids and cracks. To be conservative, it will be assumed that all components contain cracks up to the maximum size determined by the sensitivity of the NDT inspection (i.e the agreed accept/reject threshold) To cover the uncertainties, three safety factors are applied to the LEFM procedure. The first is applied to the estimate of the initial defect area that can be detected with a probability of 95%. A factor of 2 is applied to this area (which takes the detection probability to over 99%). After evaluation of the fatigue life, a reduction factor of 2 is applied to the number of cycles. A reduction factor of 1.5 is applied to K IC to calculate the fatigue life of components that can fail catastrophically Yield Criteria for Coil and Structural Insulation The yield stress limits for insulation are applied to the maximum allowable compressive stress normal to the glass reinforcement plane of the insulation. The coils contain two distinct types of insulation. High voltage insulation is applied to the electrically active components of the coils and typically has to withstand from 100V to 30kV. This insulation always incorporates an electrical barrier. Low voltage insulation is put between passive metallic structures to act as an eddy current barrier. Voltages are always under 10V and usually less than 2V. The mechanical limits on the high voltage insulation are more restrictive since microscopic mechanical damage such as cracking can create voids that lead to electrical degradation when the operating voltages exceed the Paschen minimum for gas breakdown. As long as low voltage insulation maintains a separation between the metallic surfaces, cracks are not significant. Low voltage insulation may work up to 1/2 of the compressive strength with direct stresses and 2/3 with secondary stresses. This gives working stress levels of 600 and 800MPa, respectively. High voltage insulation, vacuum/pressure impregnated with epoxy resin, subject to fatigue cycles may take stresses up to 1/3 of the compressive strength, typically giving working stresses up to 400MPa. In situations where manufacturing constraints limit the glass fraction that may be achieved (for example, in filling the TF winding pack - case gap), the allowable compressive stress may be lower Limiting Shear Stress for Coil Insulation As well as acting as a strong filler, transmitting compressive loads, the insulation acts as a bond between conductors or radial plates. Its integrity is nowhere critical for structural support, but insulation failure leading to large scale debonding could lead to an associated electrical failure and may also result in conductor movement inside the winding pack, causing quench due to frictional heat loads. Design Description Document 11 Magnet Page 47

49 The shear failure stress of epoxy-glass-kapton insulation systems is related to the local compression. Very limited tension can be tolerated (practically only local secondary tension which would not cause crack propagation if local debonding occurred. The value is limited to a strain of about 0.02% (about 10MPa). At zero normal strain the shear strength is known as the bonding shear strength. With compression, the shear strength increases proportionally up to a limiting value. The required bonding shear strength is 45MPa (static test), at the full radiation dose if appropriate (the allowable static operating bonding shear stress levels are one half of this) Allowable Pre-tension in Precompression Rings The precompression rings provide a force on the TF coils that reduces fatigue loads on the inner poloidal keys but they do not have an essential structural function. The peak loads occur at assembly on pre-tensioning at room temperature, and are reduced on cool down and operation due to differential contraction and the effect of the magnetic loads. Under these conditions a rather high pre-loading stress limit could be considered, perhaps up to 1/3 of the ultimate stress. This is twice the level allowed in the ASME X code for glass fibre - epoxy pressure vessels (although pressure vessels work under conditions of biaxial stress and the ASME values include an allowance for fatigue cycles) and has to be qualified by tests on model rings Electrical Design Criteria The following principals are applied: i) All high voltage insulation must incorporate a true electrical barrier that could be tested before application of a filler material such as epoxy resin. Glass-epoxy on its own is not adequate as it may contain voids (which cannot be detected during manufacture) and insulation is then provided only by the component separation. ii) Critical components, defined as those most difficult to replace (this includes all main coils except the CS), have a turn insulation screen that is tolerant of a single short to the screen (in the TF coils, this function is provided by the radial plates, in the PF by an extra metallic wrap around each conductor over an initial insulation layer). Currents due to a single short to the screen are limited by resistors. iii) The allowable maximum voltage to ground and between terminals is 20kV in normal and anticipated upset operation, including the possibility of a single ground short. iv) All surfaces exposed to the cryostat vacuum have a hard ground connection. Currents flowing in potential ground shorts are limited by the grounding system of the power supply. v) Generally coils will undergo proof testing at high voltage by the manufacturer. These proof voltages will not be applied once the coils are installed in the machine, because of the risk of damage if breakdown occurs (particularly in an auxiliary component such as the instrumentation). Proof testing will be both AC (especially to test turn-turn within the coil) and DC. AC test voltages and frequencies must be selected according to the coil layout. The basis for the selection of the test voltages will be the maximum of (2x the maximum voltage in normal operation + 1kV) or (the maximum voltage under anticipated upset conditions + 1kV). The factor of 2 is commonly used in transformer and generator design. It also conveniently corresponds to the factor of the voltage increase under the most severe anticipated upset fault conditions for several of the coils (because of the pairing used in the power supply and resistor connections). Design Description Document 11 Magnet Page 48

50 Nuclear-Related Design Criteria The two main nuclear-related design criteria both limit the total fluence to the magnets and do not act directly on the operating conditions. The most direct impact of the radiation from a burning plasma (and the factor that determines the shielding for the coils) is the heating of the coil case and inner winding pack turns in the inboard leg of the TF coils. The level of heat that can be tolerated is a design decision, not a criterion Insulation The insulation can receive a total radiation dose of 10 7 Gy Copper The copper resistivity increases with the neutron fluence it receives, but the increase can be very largely recovered on warming the coils to room temperature. A warm up is required after a fluence of 2x10 21 neutrons/m 2 has been accumulated. This corresponds to about 1x10-4 dpa Operation Interfaces Nuclear Heat Loads on the Magnets The ITER reference scenario specifies a total fusion power of 500MW for 400s and a repetition time of 1800s. Additional requirements, where the repetition time may be extended to allow coil recooling and cryoplant buffer refill, are that 700MW may be reached with plasma currents up to 17MA for 200s burns, 400MW with 13.8MA for 1000s burn and 360MW and 9MA in quasi-steady state conditions (burn ~3600s) (DRG1). In the magnets, nuclear heat is deposited in a non-uniform pattern around the circumference of the TF coil, with the major part in the inboard legs and with local 'hot spots' in the vicinity of the ports. At these port locations it is also deposited in the PF coils. Besides direct radiation from the plasma, there is heat from gamma radiation due to the activation ( 16 N) of the cooling water of the in-vessel components where the pipes pass through the cryostat. Calculations of nuclear heating in the coils were originally given in report NAG FDR in 2001 at the end of the EDA. The main model used for the reference calculations is a 3-D Monte Carlo calculation of a single TF coil segment. This model (referred to as the 'basic model') is not sufficiently refined to include local details of the blanket flexible supports and the heterogeneity of the vacuum vessel. Correction factors of 1.2 for each of these effects were obtained by using separate local 3-D models and are applied to all the heating and radiation results from the basic model. The original model used as reference 2cm horizontal and vertical gaps between the blanket modules. Specific models for port plugs were created and inserted into the basic model to provide the nuclear heat in the magnets around the ports, particularly for the TF and PF coil heating and the impact of the N 16 decay in the cooling circuit. The total nuclear heat in the TF coils and structures from the original model was 12.4kW. The design changes made since the 'basic model' calculations have resulted in changes to the nuclear heating. The most important of these is the change to the reference gap width between the blanket segments. The vertical gap is now 1.4cm, and the horizontal gap is 2cm Design Description Document 11 Magnet Page 49

51 with the exception of 2 modules near the equatorial plane where it is 1cm. The total heat load to the TF coils and structures is now calculated (NAG-240 'Re-assessment of Nuclear Heat in the TF coil for design specification, H. Iida, 6Jan2004) as 10.5kW (split 8.1kW on the inboard leg and 2.4kW elsewhere). The prediction of the magnet heating and radiation levels is associated with a number of uncertainty factors. The principal ones are (from NAG FDR): (i) The accuracy of the nuclear data and the Monte Carlo code (MCNP) itself (a factor of +/- 30%). (ii) The fine structural details within the blanket modules (except for the flexible joints which are already included) (a factor of +20%/-0%) (iii) A correction factor to allow for the effect of access holes and diagnostics, which are not presently included (+10%/-0%) Geometric detail in particular results generally in an increase in heating, not a decrease. A 'safety factor' of 1.2 x 1.1 = 1.32 has been applied to all the heat load and radiation results reported below, to allow for the geometric effects. This gives a total heat load on the TF coils and structures of 13.9kW (corresponding to the design level of 14kW specified in DRG1). The heat loads calculated by this process are given in Figure and 2 and Table The distribution between winding pack and case, and around the coil perimeter, is taken from NAG-240. The distributions shown in the figures correspond to the 3 TF coils adjacent to NBI ports. The remaining 15 have each a heat load that is lower by 80W (total 1200W), distributed 90% in the outer case and side walls and 10% in the winding pack. The maximum radiation doses to the coil components are given in Table , corresponding to the nuclear heating of 14kW (i.e. with a safety factor of 1.33 applied) and for a maximum local fluence at the first wall of 0.5MWa/m 2. Two further 'off-normal' cases of nuclear heating and radiation will be considered by applying further factors of 1.3 and 1.5 (corresponding to two levels of the calculation uncertainty) to the values in the figures and tables below (i.e. the values of nuclear heating to be considered in the coil design are 14 x 1.3 = 18.0kW and 14 x 1.5 = 21.0kW). Operation in either of these cases would require extended dwell times between plasma pulses and/or cryoplant/helium circulation adjustments (probably operation at a lower inlet temperature will prove more effective than increased cooling flow rates) and they are not intended to act as design drivers. The range of plasma scenarios may also need to be restricted (for example, 700MW with a high safety factor may not be possible). An extended operation for the fusion power of 700MW, but with a shorter burn time (at least 200s) and, if necessary, a long dwell time, is also an operational requirement. In this case, the value to be used for the TF coil nuclear heating is obtained by applying a factor of (700/500) = 1.4 to the figures and tables to give 19.4kW total. A further requirement is for a steady state burn, with 360MW of fusion power, lasting for at least 3600s (with an extended recool period after each burn, the pulse repitition time is 12000s). The nuclear heat is given by applying a factor of (360/500) = 0.72 to the figures and tables, to give 10.0kW total. Extra, off-normal, safety factors of 1.3 and 1.5 can then be applied to this scaled heating Design Description Document 11 Magnet Page 50

52 Front case 1st turn 2-11 turns side case (2) Back case Power W/m 1 0,1 0, Perimeter length m (zero is helium inlet) Figure Conductor Nuclear Heat (W/m for each coil) clockwise around the coil perimeter, for the main coil subunits, for each of the 3 coils adjacent to NBI ports, 14kW total in coils 1.0E-03 Heat density (W/cc) 1.0E E E Turn number Figure Nuclear Heat Density(W/cm3) variation at the inner equator for each turn, for the 3 coils adjacent to NBI ports (volume is total smeared local volume including s/c, insulation, helium, jacket, radial plate), 14kW total in coils Design Description Document 11 Magnet Page 51

53 Table Reference Nuclear Heating in the TF Coils (kw) for the 15 MA Reference Scenario with a Total Fusion Power of 500 MW Inboard Behind N Around Total Leg Divertor Ports Coil Case Winding Pack (first turn total) 6.1 (1.9) 6.1 Total Table TF Insulator Dose and Fluence (normalised to 0.5MWa/m 2 at the first wall and 13.9kW total heat in the TF coils) Nuclear Responses (Units) Epoxy insulator dose (Gy)* Fast neutron fluence in winding pack* (n/m 2 ) Cumulative damage to Cu stabiliser (dpa) Design Limits 1x10 7 1x10 22 anneal after 1x10-4 ITER Values (upper value with uncertainty) 5x10 6 (6.5x10 6 ) 7x10 21 (9.1x10 21 ) 3x10-4 * these values occur at the inner leg equator. The distribution around the coil perimeter can be found by scaling these values according to the nuclear heat in the first turn given in Figure The distribution in the winding pack thickness is given approximately by a decay factor of 0.62 for each turn The nuclear heat and radiation levels in the CS modules are negligible (at least 3 orders of magnitude below those in the TF winding packs), calculated as 600Gy and 8x10 17 n/m 2. There are two significant heat sources in the PF coils. The nuclear heat on the PF coils due to fast neutrons is summarised in Table The winding pack heat can be considered as all being deposited in the first layer or pancake directed towards the source as indicated in the table. Only approximate calculations are available for the N 16 gamma heating of the PF coils. The P2 coil receives almost all of this, due to the proximity to the outlet water pipes in the upper ducts (the shielding design has yet to be optimised). An additional 100W should be allocated, distributed in the same proportions as the 14W shown for P2 in Table Table Nuclear Heating in the PF Coils due to Neutrons and Prompt Gamma Rays(W) Ground Winding Clamps Sum Insulator Packs PF 1 Facing towards central solenoid PF 2 Under upper port * PF 3 Between upper and horizontal ports PF 4 Under divertor port PF 5 Below divertor port PF 6 Facing towards central solenoid Total 480 (more detailed allocations of the coil heat loads are given in section 2.1) * add 100W for N 16 gamma radiation Design Description Document 11 Magnet Page 52

54 The maximum radiation doses in the PF coils have been estimated from the nuclear heating and are given in Table , including both fast neutron and N 16 components. The cumulative damage to the copper stabiliser is negligible. Table PF Insulator Dose and Fluence (normalised to 0.5MWa/m 2 at the first wall and 13.9kW total heat in the TF coils) Nuclear Responses (Units) Epoxy insulator dose (Gy)* Fast neutron fluence in winding pack* (n/m 2 ) Design ITER Values Limits 1x10 7 ~0.02x10 6 1x10 22 ~8x PF Coil Current Scenarios A reference 15MA PF scenario is defined on the basis of Design Scenario 2 (PID) for the derivation of quantities such as AC losses, forces and fatigue life to confirm that the magnets can achieve their operating requirements. The coil currents for the reference scenario are shown on Table and the associated magnetic fields in Reference is made to various scenario times and states which are defined as follows: initial magnetisation (IM) or start of discharge (SOD) is the time when the circuit breakers are opened to initiate the plasma discharge. At this time the CS is fully energised and the PF coil currents are ready for the pulse; X point formation (XPF) is when the transition is made from a limiter plasma to a divertor plasma; start of flat top (SOF) iswhen the maximum plasma current is reached and additional heating is applied. In some scenarios, heating is started before peak current is reached, this point is SOH. start of burn (SOB) is the time when the nominal thermonuclear power generation starts; for scenario 2 this is 400MW but generally 500MW has been used to cover also scenario 1 (15 MA with heating at plasma ramp up and 500MW in the burn) end of burn (EOB) is when plasma cooling is started, with some current ramp down; end of current (EOC) is when the plasma cooling is finished and faster current ramp down starts; end of plasma (EOP) is the time when the plasma current has decayed to 0; dwell is the state between pulses where none of the PF coils and CS are energized; In addition to the reference scenario, the magnets must be capable of confining a range of plasma internal properties at this current. Coil currents for a limited range of these scenarios (only li and beta variations) are shown on Tables and 8 for Design Scenario 2 for SOF, SOB and EOB. The ramp up and ramp down phases for these plasmas (from IM to XPF and from EOC back to IM) can be taken as the same as for the reference scenario currents in Tables and 6. Linear interpolation can be used from XPF to SOF and from EOB to EOC. An extended 17MA, 700MW scenario (Design Scenario 5) is defined in Table and 10 for the points SOH, SOB and EOB. The ramp up and ramp down phases for these plasmas (from IM to XPF and from EOC back to IM) can be taken as the same as for the reference scenario currents in Tables and 6 (with a suitable time transposition to Design Description Document 11 Magnet Page 53

55 EOC and extension of the zero current time as required to recool the coils). Linear interpolation can be used from XPF to SOH and from EOB to EOC. The burn duration (time from SOB to EOB) is 100s, the time from SOH to SOB is 40s.. A long 3000s burn 9MA 360MW non-inductive scenario (Design Scenario 4) is given in Tables and 12. This scenario does not reach limiting conditions in the CS or PF coils. The tables give coil current variations from SOD to EOB. Current ramp up in the coils can be considered to be linear over a period of 100s, with a 10s flat top before SOD. Current ramp down after EOB can be considered to be linear to zero current in all coils over 200s. The minimum repitition time for this plasma is 12000s (as specified in PID). Currents for a long burn hybrid scenario (1000s burn, 13.8MA plasma current, 400MW nuclear heating, Design Scenario 3) are not available. However, for the purposes of AC loss calculations and thermohydraulic simulations, the coil currents for Design Scenario 2 can be used with an extension of the time from SOB to EOB to 1000s, with a repitition time of at least 4000s. The field values in these tables are based on 'smeared' currents over the active winding pack area (including the ground insulation). A more detailed field calculation including the actual cables will give peak field values between 0.05 and 0.2T higher (generally >0.1T for PF coils, <0.1T for CS). Table Reference Scenario (Design Scenario 2)-Currents Reference scenario: Inductive operation 15 MA, without heating during current ramp-up. li=0.85 during current flat top, currents in MA, magnetic axis coordinates R ax and Z ax in m Time s beta Ip Rax Zax CS3U CS2U CS1U CS1l CS2L CS3L 0 SOD XPF Design Description Document 11 Magnet Page 54

56 100 SOF SOB EOB EOC EOP SOD Time s beta Ip P1 P2 P3 P4 P5 P6 0 SOD XPF Design Description Document 11 Magnet Page 55

57 100 SOF SOB EOB EOC EOP SOD Table Reference Scenario (Design Scenario 2)-Fields Time s beta Ip CS3U CS2U CS1U CS1l CS2L CS3L 0 SOD XPF SOF Design Description Document 11 Magnet Page 56

58 SOB 530 EOB EOC Time s beta Ip P1 P2 P3 P4 P5 P6 0 SOD XPF SOF SOB EOB EOC Design Description Document 11 Magnet Page 57

59 Table PF Flexibility within Design Scenario 2 - Currents All currents in MA, coordinates in m Time li β Ip Rax Zax CS3U CS2U CS1U CS1l CS2L CS3L SOF SOF SOF SOB SOB SOB SOB EOB EOB EOB EOB Time li β Ip Rax Zax P1 P2 P3 P4 P5 P6 SOF SOF SOF SOB SOB SOB SOB EOB EOB EOB EOB Table PF Flexibility within Design Scenario 2 - Fields in T Time li β CS3U CS2U CS1U CS1l CS2L CS3L SOF SOF SOF SOB SOB SOB SOB EOB EOB EOB EOB Time li β P1 P2 P3 P4 P5 P6 SOF SOF SOF SOB SOB SOB SOB EOB EOB EOB EOB Design Description Document 11 Magnet Page 58

60 Table PF Design Scenario 5 - Currents All currents in MA, coordinates in m Time from SOH to SOB 40s, time from SOB to EOB 100s Time li β Ip Rax Zax CS3U CS2U CS1U CS1l CS2L CS3L SOH SOH SOH SOB SOB SOB EOB EOB EOB Time li β Ip Rax Zax P1 P2 P3 P4 P5 P6 SOH SOH SOH SOB SOB SOB EOB EOB EOB Table PF Design Scenario 5 - Fields in T Time li β CS3U CS2U CS1U CS1l CS2L CS3L SOH SOH SOH SOB SOB SOB EOB EOB EOB Time li β P1 P2 P3 P4 P5 P6 SOH SOH SOH SOB SOB SOB EOB EOB EOB Design Description Document 11 Magnet Page 59

61 Table PF Design Scenario 4 - Currents All currents in MA, coordinates in m Time s beta Ip Rax Zax CS3U CS2U CS1U CS1l CS2L CS3L 0. SOD XPF SOH SOB EOB Time s beta Ip P1 P2 P3 P4 P5 P6 0. SOD XPF SOH SOB EOB Table PF Design Scenarios 4- Fields in T Time s beta Ip CS3U CS2U CS1U CS1l CS2L CS3L 0.SOD XPF SOH SOB EOB Design Description Document 11 Magnet Page 60

62 Time s beta Ip P1 P2 P3 P4 P5 P SOD XPF SOH SOB EOB Correction Coil Current Scenarios The correction coils operate in two modes (i) Error field correction (ii) Resistive Wall Mode (RWM) stabilisation (section ) The current magnitudes required for correction of the expected error fields are given in Table (from Y. Gribov, A. Kavin, ITER_D_22GZ7N v1.0 Currents in Correction Coils). A further 130kAT are provided in the side coils for stabilisation. Table Currents in CC required for correction of the error fields due to different sources (in ka turns). Source of error field Top CC Side CC Bottom CC Comment NBI MFRSs, 5TBMs & Correction to 0 Coil joints, feeders, terminals TF, CS and PF Coils Misalignment Correction from 11.9x10-5 to 5x10-5 All sources Simple summation Nominal currents The scenario currents in CC are specified by the following an artificial scenario that is relatively demanding from the point of view of the AC losses. At the start of central solenoid discharge, currents in the Top and Bottom CC are negative and their values are the nominal currents for these coils: -0.14MAturn and -0.18MAturn. The current in Side CC is equal to 0.15MAturn, which is close to the maximum value of this current required for correction of the expected error fields (see Table ) At the first phase of scenario, which includes plasma initiation, current ramp-up and heating to a driven burn, we assume linear ramp-up of the current in the correction coils over 45s. At the end of this phase, the currents in Top and Bottom CC become positive and their values are the nominal currents for these coils: 0.14MAturn and 0.18MAturn. The current in Side CC becomes 0.15MAturn in Scenario 1 and zero in Scenario 2. This rather fast variation of the currents (during 45s) is chosen to get an upper estimate of the AC losses Design Description Document 11 Magnet Page 61

63 produced by the scenario currents. The scenario currents at this phase are shown in Figure (left). At the second phase of scenario, during a burn lasting from s, we assume constant values of the scenario currents. At the last, third, phase of the scenario, the CC currents linearly vary during 45s until reaching their initial values at the start of central solenoid discharge. This phase of scenario currents, in the case when the burn state lasts 300s, is shown in Figure (right). During the dwell period (until the start of the next pulse) the CC currents can be left constant or reduced to zero. "Scenario" currents in CC "Scenario" currents in CC Top CC Bottom CC Side CC (Scenario 1) Side CC (Scenario 2) Top CC Bottom CC Side CC (Scenario 1) Side CC (Scenario 2) t, s t, s Figure Scenario Currents in the Correction Coils (MAT) During Plasma Ramp Up and Ramp Down Plasma Control with the PF Coils There is a range of plasma disturbances/events whose occurrence requires a response from the PF coils determined by plasma current, position and shape control system. According to [Y. Gribov, A. Kavin ITER_D_22JYAK v1.1 (Input for the AC losses study: Control actions )] the list of disturbances is as follows: Minor disruption is defined as an instantaneous drop of two plasma parameters l i and β p, by 0.2(l i0-0.5) and 0.2β p0, respectively, without recovery for l i and with 3 s exponential recovery for β p. The minor disruption is described as a single disturbance with the highest probability to occur during XPF(29.37s) and EOC(590s) phases of the plasma design scenario 2. H to L Mode Transition at the End of the 1 st Cooling Phase in design scenario 2 (580s). The event is defined as a linear β p reduction from 0.45 to 0.1 over 3 seconds. Unless operation ends with major disruption, a single event must occur at EOC1. Edge Localised Modes (ELMs) are described as repetitive disturbances occurring during plasma burn from S0B(130s) to EOB(530s) in design scenario 2: Compound ELMs are defined as an instantaneous drop of two plasma parameters l i and β p, by 0.06(l i0-0.5) and 0.03β p0, respectively, with 1 s linear recovery for l i and 0.2 s linear recovery for β p. Estimated repetition time for the event is 10 s. The type 1 ELMs are defined as instantaneous β p drop of 0.03β p0 followed by 0.1 s linear recovery. Estimated frequency of event occurrence is 2Hz Design Description Document 11 Magnet Page 62

64 Induced ELMs are defined as instantaneous β p drop of 0.01β p0 followed by 0.1 s linear recovery. Estimated frequency of event occurrence is 4Hz In addition, the plasma current, position and shape control system may react to noise in dz/dt diagnostic. For the analysis purposes, the noise was defined as a white nose with a cut off frequency 110Hz, which after filtering has the RMS (root mean square) value at the level of m/s. To make the signal look more realistic, a noise with characteristic RMS of 0.02m/s was superimposed on the weakest actual disturbance, Induced ELMs. This combined signal has been used for the AC loss calculation along with Minor Disruptions, H to L Transition, Compound ELMs and ELMs type1. The control currents have been derived using the VS1 vertical stability controller which connects the P2, 3, 4, 5 coils in an up - down reverse series connection, in addition to the individual coil convertors and switching units (section ). Details of current variations for the control actions and the corresponding analysis are given in E. Zapretilina ITER_D_22FR8D v1.0 (Assessment of AC losses with LOSS code (2005)). An additional controller has been added for the inductive scenarios, which connects two modules of the CS coils (CS2U and CS2L) in reverse series. Control scenarios using VS2 are given in A. Kavin, ITER_D_22GZ9P v1.0 (Input for the AC losses study: Control actions with second stabilizing loop) Resistive Wall Mode Control with the Correction Coils Resistive wall mode stabilisation affects only the side correction coils and is only applied during non-inductive scenarios (i.e. design scenario 4). Specifications for the control currents are given in Y. Gribov, A. Kavin, ITER_D_22GZ7N v1.0 Currents in Correction Coils. The coils are connected with opposite coils (i.e. coils separated by 180degrees) in opposite series, with 3 independent power supplies. The control currents in the Side CCs are caused by disturbances in the RWM control during the phase of driven burn. Two types of disturbance are assumed for the AC losses analysis: 1) Diagnostic Noise and 2) Large RWM Event. The Diagnostic Noise is low frequency oscillations of the poloidal magnetic field on the sensors detecting the RWM. It is caused by low frequency plasma oscillations other than RWM and Type 1 ELMs. This disturbance is assumed during the whole period of driven burn (3000 s in the steady state scenario). The Diagnostic Noise was simulated as a white noise filtered by an elliptic filter with the cut off frequency 500 Hz. This filtering does not deteriorate control of even highly unstable RWM (C β 0.8) [3]. The noise parameters (power and sampling time) were fitted to get the quasi-stationary RMS (root mean square) value of the filtered white noise equal to about 0.5 mt. The Large RWM Event is a single event in RWM control. In the simulations of this disturbance, the RWM evolves freely starting from a small value of the signal on the sensors detecting the RWM, B (B << 1 mt). The feedback stabilization is switched on when B Design Description Document 11 Magnet Page 63

65 exceeds a specified value B 0. At this time the Diagnostic Noise, B noise, is added to the diagnostic signal. The total signal is filtered with the cut off frequency of 500 Hz. For the moderately unstable RWM (C β = 0.52) the value of B 0 was assumed 2 mt, for the highly unstable RWM (C β = 0.83), it was 1.5 mt. The results of AC losses analysis should establish the tolerable number of the Large RWM Events during the phase of driven burn. Figure shows an example of the coil currents. Figure RWM feedback control for the plasma with C β = 0.52 (5 s t 10 s). Large RWM Event (B 0 = 2 mt) with Diagnostic Noise (RMS = 0.5 mt). Solid blue line coil current (MAT) in one pair of coils A first assessment of the allowable number of large RWM events is given in Mitchell, Gribov, Zapretilina, ITER_D_22FJXH v1.0 (Allowable Contral Currents in the Side Correction Coils) Cryogenic Heat Loads During operation, heat from external sources and from the magnets themselves is deposited in the magnet system. Thermal radiation from the 80K cryostat thermal shield (including the thermal shield labyrinths) is deposited on the outer surface of the TF coil cases and structures and the PF coils. Thermal radiation from the vacuum vessel shield is deposited on the TF coil case inner surfaces. Heat is conducted from the machine gravity supports, the vacuum vessel supports and the vacuum vessel thermal shield supports. Nuclear heat is the dominant heat load for the TF coils and is essentially deposited along the inner surface of the inboard legs. Nuclear heat is also deposited in the PF coils in the vicinity of the vessel ports. There is also heat from gamma radiation due to the activation ( 16 N) of the cooling water of in-vessel components. This heat load is small and is deposited mostly in PF coils and feeders in the vicinity of the heat transfer system pipes. Internal heat loads due to the operation of the magnets include AC losses in conductors, eddy current losses in structures and resistive losses in joints. For any set of coils, the heat exchanger bath may operate from 4.3K to 3.8K to allow the capability to correct possible faults or design/analysis errors discovered in operation (for example, an underestimate of the nuclear heating). The assumption of regular continuous Design Description Document 11 Magnet Page 64

66 pulsing implied by the following tables applies only for 'normal' operation when the heat exchanger bath is at or above 4.15K. TF CS PF CC 10Structure Feeders,CTBs Reference Scenario Magnet Heat Loads (total) Scenarios 1,2 TF CS PF CC 100Structure Feeders, CTBs Reference Scenario Magnet Heat Loads (total) Scenarios 1, Heat Input in MJ 6 4 Heat Input in kw Initiation RampUp Burn RampDown Dwell 0 Initiation RampUp Burn RampDown Dwell Figure Contributions to Cryoplant Load and Distribution Over 1800s Reference Pulse with 500MW nuclear power (pump work not included) Table summarises the total heat loads in the magnets for the reference scenario and Figure gives a graphical presentation. The table also includes the losses associated with the magnet feeders. The distribution of the losses through the 5 main phases of the scenario are given in Table Table Heat Loads in the Magnets and Feeders (15 MA Reference Scenario with a Total Fusion Power of 500 MW, 400s Burn, 1800s Repitition Time) Overall Total 21.2kW Heat loads TF structures (+PF supports) 7.8kW in burn TF conductor CS structures CS conductor PF conductor CC conductor Total (average over 1800s)** Nuclear Rad 6.1kW in 0.6kW kW burn in burn AC losses * 1.72MJ 8.46MJ 1.60MJ 0.03MJ 6.56kW Eddy currents Joints Thermal Rad Cryostat Vacuum vessel Thermal Cond Gravity supports Thermal shield**** 5.34MJ 1.41MJ 3.74kW kw 0.76kW 0.46kW 2.3kW 0.2kW kW average 0.07kW average 0.1kW 0.01kW 1.08kW 1.33kW 2.5kW He feeders, 0.77kW 0.37kW 0.66kW 1.8kW Design Description Document 11 Magnet Page 65

67 SC bus bars and CTBs*** He Cryolines from ACBs and CTBs 0.05kW 0.45kW 0.2kW 0.3kW 1.0kW * AC loss assumptions are conservative, based on conductor parameters above the design targets: TF conductor based on jc 1000A/mm2, hys 1000mJ/cc, ntau 200ms PF conductor based on jc 2900A/mm2, eff fil 6mm, ntau 150ms CS conductor based on jc 800A/mm2, hys 600mJ/cc, ntau 150ms CC base losses on maximum removable heat (although not used for RWM control in inductive scenario) **The average values are for a 400 s burn with a 1800 s pulse repetition. ***based on 50W per CTB (thermal radiation, cold end heat load of current lead from Table ) and 0.5W/m of feeder ****Source for thermal loads, DDD 2.7 (August 2004 version) Gravity support conduction section VVTS support conduction table Radiation loads Table and section 2.5 (with correction for removal of VV ropes) Table Distribution of Electromagnetic (AC losses, Eddy Currents) in 15 MA Reference Scenario with a Total Fusion Power of 500 MW and 400s Burn, in MJ Losses in MJ TF Conductor CS Conductor PF Conductor CC Conductor Structure Eddy Currents Nuclear Scenario Control Scenario Control Scenario Control Scenario Control Initiation 0-1.6s Ramp Up s Burn s (inc control) 5.76 Ramp Dwn (plasma and coils) s Dwell s The losses for design scenario 4 (9MA plasma, 3000s extended burn with 360MW nuclear heat) have been estimated by scaling from the 15MA case as direct calculations are not yet available. Table summarises the total heat loads in the magnets and Figure gives a graphical presentation. The distribution of the losses through the 5 main phases of the scenario are given in Table The scalings applied are Assume side CC work with maximum heat extraction (390W) for 3000s burn Initiation (0-1.6s) same as reference Ramp up ( s): scenario losses scale by (9/15), control same Flat top and burn ( or -3130): scenario losses scale by (400/3000), control losses by (3000/400) Design Description Document 11 Magnet Page 66

68 Plasma and PF ramp down: scenario losses scale by (9/15), control same Dwell and pre-magnetisation: scale by (9/15) Table Heat Loads in the Magnets and Feeders (9 MA Design Scenario 4with a Total Fusion Power of 360 MW, 3000s Burn and 12000s Repitition Time) Overall Total 11.9kW Heat loads TF structures (+PF supports) 5.62kW in burn TF conductor CS structures CS conductor PF conductor CC conductor Nuclear Rad 4.39kW in kW 0 burn in burn AC losses * 1.25MJ 6.58MJ 2.28MJ 1.2MJ Eddy currents 6.54MJ 1.73MJ Joints kW 0.01kW average Thermal Rad Cryostat 0.76kW 0 Vacuum 0.46kW 0 vessel Thermal Cond Gravity supports Thermal shield**** He feeders, SC bus bars and CTBs*** He Cryolines from ACBs and CTBs 2.3kW 0.2kW 0.04kW average 0.1kW 0.01kW Total** (average over 12000s) 2.61kW 0.94kW 0.69kW 1.05kW 1.33kW 2.5kW 0.77kW 0.37kW 0.66kW 1.8kW 0.05kW 0.45kW 0.2kW 0.3kW 1.0kW * AC loss assumptions are conservative, based on conductor parameters above the design targets: TF conductor based on jc 1000A/mm2, hys 1000mJ/cc, ntau 200ms PF conductor based on jc 2900A/mm2, eff fil 6mm, ntau 150ms CS conductor based on jc 800A/mm2, hys 600mJ/cc, ntau 150ms CC base losses on maximum removable heat **The average values are for a 3000 s burn with a s pulse repetition. ***based on 50W per CTB (thermal radiation, cold end heat load of current leads from Table ) and 0.5W/m of feeder ****Source for thermal loads, DDD 2.7 (August 2004 version) Gravity support conduction section VVTS support conduction table Radiation loads Table and section 2.5 (with correction for removal of VV ropes) Table Distribution of Electromagnetic (AC losses, Eddy Currents) in 9 MA Design Scenario 4 with a Total Fusion Power of 360 MW and 3000s Burn, in MJ Losses in MJ Initiation 0-1.6s Ramp Up s Burn s Ramp Dwn (plasma and coils) s Dwell s Design Description Document 11 Magnet Page 67

69 TF Conductor CS Conductor PF Conductor CC Conductor Structure Eddy Currents Nuclear Scenario Control Scenario Control Scenario Control Scenario Control (same power as scenarios 1,2) TF CS PF CC 10Structure Feeders,CTBs Reference Scenario Magnet Heat Loads (total) Scenarios 1,2 TF CS PF CC 70Structure Feeders,CTBs Reference Scenario Magnet Heat Loads (total) Scenario Heat Input in MJ 6 4 Heat Input in MJ Initiation RampUp Burn RampDown Dwell 0 Initiation RampUp Burn RampDown Dwell Figure Contributions to Cryoplant Load and Distribution Over 12000s Design Scenario 4 with 360MW nuclear power (pump work not included) Table gives the load factors for the CS and PF coils (rms current divided by the maximum current, usually 45kA) for scenario 2. These are useful for working out joint and current lead average heat loads. Table Load Factors Based on RMS Currents for Reference Scenario (Table ) Coil Load Factor CS3U 0.26 CS2U 0.49 CS1U 0.60 CS1l 0.60 CS2L 0.44 CS3L 0.28 P P Design Description Document 11 Magnet Page 68

70 Disruptions P P P P The plasma is expected to undergo two levels of disruptive disturbance. In the first, a minor disruption, there is a loss of internal energy which produces a change in the internal current distribution. The plasma current however continues to flow and control actions maintain the plasma configuration, within about 100s. This type of disruption is classed as a control event for the magnets. In the second type, a major disruption, the plasma current terminates in a few ms, possibly preceeded by a large scale vertical movement of the plasma (and power supply actions to attempt to control the movement). The field and current changes in the coils are slowed by coupling with the vacuum vessel and the whole electromagnetic event lasts 1-2s. A major disruption has 5 effects on the coils: 1) Eddy currents are generated in the TF coil case (see section 2.1, 2.3 and 2.5 for details) by inductive coupling with the poloidal and toroidal plasma current. The total energy is about 13MJ. 2) AC losses in the conductor (mostly coupling loss, see section 2.1). The total losses in all conductors are about 0.8MJ and the maximum local heating is 230mJ/cc (of strand material) in the central CS modules with ntau of 50ms. 3) Current changes in the PF and CS coils due to inductive coupling with the toroidal plasma current (see section 2.1) up to 3MA in the central CS modules and about 1-1.5MA in the PF coils 4) Current changes in the TF coils due to inductive coupling with the poloidal plasma current, up to about 100kA/coil. The AC losses are negligable. 5) Dynamic vertical loads transmitted from the vacuum vessel through the magnet gravity support to the PF coils. These arise from a vertical disruptions (VDEs) and, due to halo currents in the plasma and in-vessel components, may be non-symmetric around the toroidal direction. The asymmetry of the electromagnetic forces on the supports is reduced by the structures (both magnet and vessel) and for design purposes the values in Table can be used, distributed toroidally symmetrically on the magnet gravity supports. Detailed analysis of non-symmetric effects is required when plasma simulation studies have been completed. The vertical loads in the table are categorised (Category I, II, III) according to the expected probability of occurence (category III being very rare, perhaps once in the machine life). The direction of the vertical displacement (U upward, D downward) affects the loads, as does the speed of the event (S slow, F fast). The duration of the loads is the time of the eddy current decay and can be considered as 1s for design. Further information is given in DDD1.5 (Vacuum Vessel), section It is a design requirement that the coils do not quench or undergo fast discharge after a disruption. For design purposes, there are two critical effects: 1) Eddy current heating of the TF coil case, especially in the thin sidewalls of the inner straight leg where there is a near-instantaneous temperature rise of 13K (section 2.3). This heat diffuses through the ground insulation into the side pancakes over about 100s (section 2.1) and reduces the temperature margin by about 0.3K (which is acceptable as there is no plasma operation). Design Description Document 11 Magnet Page 69

71 2) Electromagnetic impulses to the quench detection system which have to be reliably distinguished from a quench signal. An additional load case arises from the possibility of a fast discharge of the PF coils after a disruption (the disruption may also be caused by the start of the fast discharge or the discharge may immediately follow the disruption). In this case, the total energy in the TF coil cases rises to 18MJ, although almost all of the extra 5MJ goes in the upper and lower curved region and outer part, not the inner leg. It is again a design requirement that the TF coils do not quench. Table Vertical Displacement Event (VDE) Loads on the Magnet Gravity Supports Event Category Direction Speed Vertical Load (total on 18 supports, +ve is upward) MN I D S 43 I U S -32 I D F 19 I U F -20 II D S 54 II U S -41 II D F 23 II U F -26 III D S 72 III U S -54 III D F 31 III U F Internal Helium Pressure Although a negligible part of the magnet structural loads, the coil conductors, helium piping and headers are all subject to internal pressure by the helium coolant (this is also discussed in sections and The design pressures for the various components are given in Table The values for the quench inside the coil are conservative, based on a very fast quench of a long length of conductor a long way from the inlet and outlet. Table Design Pressures for Magnets and Auxiliary Components in Primary Helium Circuit Inside Coil Joints and headers Feeders and CTBs adjacent to coil Normal pulsed 1.0MPa 1.0MPa 1.0MPa operation Quench and faults 20MPa 2.0MPa 1.8MPa Physical Interfaces The magnets are located within the cryostat which provides the thermal insulation for the 4.5K superconducting coils from the ambient heat load. Vacuum within the cryostat Design Description Document 11 Magnet Page 70

72 eliminates the conductive and convective heat loads, and intermediate thermal shields at 80K intercept the bulk of the thermal radiation from the cryostat and the vacuum vessel. Feeders to the coils carry the superconducting busbars, cryogen service lines and instrumentation cables, for example Drawings 1101F1_ , 11G , 11R1T1_ These feeders run from individual coil terminals inside the cryostat, through cryostat feedthroughs (CF) which include the S bends to allow relative coil-building motion, and into coil terminal boxes (CTBs) or structure cooling valve boxes (SCVBs). These boxes are located outside the cryostat and bioshield, in the tokamak building galleries which are accessible for hands-on maintenance. The interfaces between the magnet system and the power supplies and the cryoplant are at the room temperature terminals of the current leads (within the dry boxes) and at the helium inlet/outlet flanges to the valve boxes Power Supplies The power supplies have two physical interfaces to the magnets: i) The current supplies and discharge circuits for the coils. ii) The magnet structure grounding scheme. For the TF coils, there is one current supply for the 18 series connected coils. The supply is able to provide a slow charge or discharge of the TF coils in about 30minutes. The fast discharge circuits provide the required discharge time constant of 11s. As already explained in section , the TF coils are connected in series in pairs inside the cryostat and there are 9 discharge circuits interleaved with these coil pairs. Each PF coil has its own power supply, switching network units and discharge circuit. The CS modules have one power supply and switching network and discharge circuit each but the two modules nearest to the equatorial plane are connected in series with interleaved power supplies and discharge circuits so as to minimize the ground voltage. There are two vertical stability power supplies. In VS1, four of the PF coils are connected in up-down reverse series to a single controller. In VS2, the CS2U and CS2L are connected in up-down reverse series to a single controller. For error field correction, the CCs have one power supply and discharge circuit for each anti-series connected coil pair (2 coils at 180degree toroidal separation), giving a total of 9 supplies. No fast discharge resistors are needed for the correction coils as the busbars provide the required resistance and thermal inertia. The side coils when used for RWM control (non-inductive scenarios only) have 3 separate and additional power supplies again connected to pairs of coils in anti-series. The interface between power supplies and the magnets occurs at the coil terminal boxes (CTB). Water cooled aluminium busbars carry the current through the torus pit and are connected to the current leads within the dry boxes (which provide a heated airflow to the ends of the current leads to prevent the build-up of ice). The magnet structures are all connected to ground through the busbar containment pipes to the cryostat wall. The connection of the cryostat into the overall machine grounding scheme is part of the power supplies. Details of the coil grounding scheme are given in Annex Cryoplant In operation, the cryoplant provides supercritical helium to the four sets of coils (TF, CS, PF and CC), their superconducting busbars, the magnet structures, and the current leads. The Design Description Document 11 Magnet Page 71

73 magnets return supercritical helium from coils, busbars and structures and gaseous helium at room temperature from the current leads. Each of the primary circuits (i.e. for the three main coil sets and the structures) consists of a closed flow loop that circulates supercritical helium through a primary heat exchanger. Each circuit can be fitted with an adjustable bypass valve at the heat exchanger that would be controlled to limit the heat extraction from the coils to match the cooling power setting of the cryoplant, although at the moment it appears that adequate load smoothing can be achieved with a bypass valve only on the TF circuit. The coil inlet temperature is allowed to vary (in the range 4.4 to 4.6 or 4.7K) to provide headroom for the cryoplant control. The structure inlet temperature is allowed also to vary, over a somewhat larger range. The primary cooling loop is closed and the pressure during operation increases from about the fill conditions (0.35MPa at 4.3K) at the start of regular plasma pulsing to about 0.6MPa once repeatable cyclic conditions are established. The mass flow rate in the primary circuit is not foreseen to be feedback controlled but is adjusted for each cycle to match the expected heat loads. The pumping power load on the cryoplant is a significant fraction of the total cryogenic load and a minimum mass flow rate is chosen compatible with the coil heat loads and duty cycle. A summary of the mass flow rates in the different cooling circuits is given in Tables and The correction coils are assumed to be supplied from the PF cryoline and the conductor has been designed with the same pressure drop. After a coil quench (including the TF coil quench which occurs after a fast discharge), the primary circuit is vented initially to a cold holding tank at 80K and then to tanks at 300K if the capacity of the cold storage is exceeded. The vent opening pressure in the primary loop is 1.8MPa. Some further details on the TF fast discharge are given in section The boundary between the cryoplant system and the magnet helium manifolding occurs at the coil terminal boxes (CTB) and structure cooling valve boxes (SCVB). These contain the adjustable valves that allow the proper distribution of the helium flow into the magnet components. The mass flow rate and inlet temperature provides the main adjustment to compensate for extra unexpected heat loads in the coils, although with a higher mass flow rate there is a rapid increase in the pumping power load to the cryoplant. To provide a further backup in the event of unforeseen heat loads in operation, or lower conductor superconducting performance, the cryoplant is also designed to operate with the primary heat exchanger at 3.7K (rather than above 4.1K). This can be used to lower the coil operating temperature, or remove extra heat, at the cost of a reduced duty cycle or extra installed cryoplant capacity. In the event that a PF coil has to operate in backup mode (with the loss on one pancake), the inlet temperature to the PF and CC systems must be limited to 4.4K instead of 4.7K. This restricts the use of the bypass valve to the primary heat exchanger for this circuit, reducing the flexibility for cryoplant load smoothing and possibly requiring extra cryoplant capacity or liquid He storage. Table Mass Flow Rates in Coils for the Reference 500MW Plasma (400s burn, 1800s total pulse duration) Coil Total Winding pack mass flow rate Conductor average mass flow rate (g/s) Helium transit time in coil (s) Conductor Flow path length (m) conductor Pressure drop (bar)** Design Description Document 11 Magnet Page 72

74 (kg/s) TF CS PF PF PF PF PF PF Upper CC Side CC Lower CC ** includes He inlets and outlets, see An extra pressure drop contribution in the coil circuits from cryolines, manifolds and valves (including the heat exchanger) of 0.01MPa should be included in addition to the pressure drop values in Table In the structural cicuits, the control valves make up a significant part of the pressure drop. Each structural circuit has two valves (one manual one at inlet and one on each cooling loop at outlet for control, see section ) and a pressure drop of 0.01MPa is allocated to each of these (in operation they are close to the fully open condition). Pressure drop values in the piping are calculated using a friction factor x2 that of a smooth pipe and a total of 0.003MPa is estimated (including coil, feeders and cryolines). Full data is given in 'Pressure drop in feeder pipes' (ITER_D_22GT39 v1.1) with a summary in Table The pressure drop and lengths in this table are those from the lower Structure Cooling Valve Box (SCVB), inlet pipe, pipe in structure, outlet pipe and to upper SCVB. Circuit Table Mass Flow Rates in the Structures Number of Loops No. parallel circuits/coil Length (m) Pipe ID (mm) Flow total (kg/s) Pressure drop total (kpa) Inside inner wall Inside outer walls Outside outer leg (PF supports) ** CS Structure ** mainly for cooldown, minor contribution to operational cooling The current leads are cooled with gaseous He at 50K and 1.6MPa supplied from the cryoplant. The mass flow rates are given in Table and the associated text. Design Description Document 11 Magnet Page 73

75 Coil Alignment The specification for the allowable error field from the coils is given in DRG1, section 1.1. Approximately, the allowable departure from axisymmetry of either toroidal or poloidal field is about 3x10-4 T. 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 directions) which have the largest impact on the plasma. The procedure has been an iterative one. Initial estimates of manufacturing and assembly tolerances have been assessed and the correction coil currents required to correct them (considering the full range of possible tolerance combinations) have been calculated. The space required for the CCs is limited and the best possible manufacturing and assembly tolerances are required to achieve the error correction within the CC current limits defined in Tables and The error field analysis is reported in Appendix E to 'Control System Design and Assessment', G45FDR R1.0 (and updates), which was part of the 2001 FDR. The various contributions to the error fields, including those arising from the various manufacturing tolerances, are summarised in 'Contributions to Plasma Error Fields from the CS, PF and TF Coils' ITER_D_23DVQU version 1.3. There are two main sources of error fields, those arising from the busbars and those arising from positioning errors in the centreline of the coils. The main source of the busbar error field is the unpaired lengths of busbars that are required to connect the coil terminals to the main cryostat feeders. These are detailed in the first section. The second source arises from coil positioning errors and is more difficult to define. There are two main contributions, those arising during manufacturing and those arising during assembly. In this new assessment, the assembly and manufacturing errors are considered as independent and therefore additive. In order to achieve reasonable manufacturing costs but still meet the required positioning accuracy, it will to correct some modes of manufacturing error during assembly of the TF coils into their cases. The values of tolerances given in ITER_D_23DVQU version 1.3 are generally based on experience from the ITER model coils and recent coil manufacturing studies. They are intended to be slightly conservative and to be compatible with a reasonably fast winding time (achieving higher accuracy will result in longer manufacturing times and higher cost). The need to develop a new manufacturing procedure for the TF coil case has lead to a new overall specification for the procedure defined in ITER_D_23HNWN version 1.2, Case Insertion and Assembly Procedure Specification and Tolerance Requirements for the TF Coils, see also section The philosophy of the fabrication and assembly processes foreseen is based on the following points (see also Fig ): i). Tight tolerances required on one critical part of the coil (external wall of the wedged surface of sub-assembly AU) and on the flat inner surface of the straight section A2 ii) Gaps allowed between case and WP on insertion in AU to align the CCL and compensate, if required, CCL fabrication misalignments. iii) Absorb other manufacturing tolerances at the WP-case interface by filling variable gaps with reinforced resin by VPI. Design Description Document 11 Magnet Page 74

76 iv) Avoid unless indispensable re-machining after TF final assembly. The 3 critical datum points used during the installation phases (upper and lower OIS and gravity support) will be adapted to nominal dimensions by allowing space for precise shimming, with shims determined in the factory after completion of the case. Fig Main sub-assembly and sections of the TF case. The penalty for allowing increased tolerances is that high strength case material is lost, replaced by weak filler material. A margin of 10% is allowed in the design between maximum allowable stresses and those calculated by stress analysis for the ideal geomtery, to allow for geometric imperfections Magnet Displacements There are a number of physical interfaces between the magnets and the surrounding components, either because of clearance gaps that must be maintained in operation or because the magnets are used for support. Such components are (for example) the feeder ducts that are connected at one end to the coils and at the other to the building basement, or the thermal shield, which must maintain a clearance to the coils. For design purposes, the movements of the magnets relative to the building are defined here. For normal operating conditions (cooldown and plasma pulsing), the displacements of the TF coils can conveniently be defined by the movements of the PF coil support points around the outer coil perimeter from near the top (P1) to near the bottom (P2). Table gives the support motion in 3 directions and the additional radial movement of the PF coils. Design Description Document 11 Magnet Page 75

77 For faulted operation, two conditions are considered 1) Seismic These results are taken from 'Preliminary Comparative Seismic Analysis of the ITER Tokamak with SSI and the building simulator in the candidate sites' July 2003, V. Sorin, ITER Garching. Seismic analyses have been performed assuming different seismic spectra and different seismic isolation (as seismic excitation is site related). The calculated data are considered preliminary and might be subjected to changes if some modifications are introduced (i.e. stiffness of the TFC/VV toroidal links, etc.). For conservatism, and to avoid later design changes, the following values of the displacements relative to the building at both the top and bottom of the TFC in SL-2 conditions are defined: Radial displacements: 8 mm Vertical displacements: 2 mm Toroidal displacement: 2 mm Components have also to be analysed under dynamic conditions (this applies particularly to the feeders). 2) Off-normal PF coil current combinations. The behaviour of the TF coils under exceptional (i.e. non-scenario related) PF coil current combinations is analysed in section 2.4 and in particular Annex 2a to section 2.4. The outof-plane movement of the coils can exceed 10mm at a considerable proportion of the perimeter, depending on the PF coil current combinations. However it never exceeds 20mm. For design (in fault conditions), a maximum out of plane movement of 20mm (in either direction and at any location) will be specified, as a conservative condition Diagnostics Four of the TF coils contain a Rogowski coil for the measurement of plasma current, as part of the plasma diagnostics. These coils are identical in size and outer surface material to the cooling tubes inside the case. An extra groove (on the outer surface, furthest from the plasma) is provided to contain the coil. The diagnostic coils are prepared in two sections with the cooling tubes, in the inner and outer case pieces. Each coil section is a double spiral (go and return). The coils stop each side of the upper transverse case closure weld and are connected at the lower cooling pipe case penetration. The lower structure feeders contain two connection wires (low voltage, 100V) for each of the 4 coils Assembly The ITER assembly is described in the document AP (Assembly Plan) attached to the Plant Design Description. This section is a brief summary of the main assembly steps. The lower PF coils (P5 and P6) and the lower Correction Coils (CCs) are the first coils to be placed in position, on temporary supports at the bottom of the cryostat. Various elements of the support structure are also put in position at this time (such as the lower precompression rings and the gravity supports). The feeders for the side and lower correction coils (the lower header rings) are also positioned at this time (Drawings 11R and 11TCC1_000430) The TF coils are assembled in pairs in the assembly hall at the ITER site as part of the single vacuum vessel sector. The tightly toleranced interface regions between the two coils will Design Description Document 11 Magnet Page 76

78 have been precision machined in the factory but additional shimming can be provided as required. The upper and lower OIS acts as the reference surface and joint to link the coils for this operation. Each coil has a thin 'sandwich' of insulation, contained between two steel sheets, that is fixed to one of the coil wedging surfaces (welded or bonded) during manufacturing, and shims can be attached by welds to the outer sheet, Drawing 1101CA_ Slots for the shear keys are oversized to allow the use of slot liners which are custom machined to achieve a tight fit of the shear keys in those liners. As mentioned above the TF coils pairs are fitted together onto a 1/9 vacuum vessel segment complete with thermal shield. The machine segments, i.e., the vacuum vessel sectors, are pre-assembled before transferring them to the pit in the tokamak hall. Each one is composed of two TF coils, with the upper and lower friction joint panels in the middle already welded, and the corresponding vacuum vessel and thermal shield segments. The vault structure of the TF coil inboard legs is built up circumferentially, with correction of the accumulated assembly error by shims every 3 pairs if necessary (although the intention is that the coils are machined accurately enough to avoid shimming at assembly). The upper and lower OIS structures again act as reference surfaces for the assembly operation. Once the vault structure is complete, the remaining upper friction joint panels on the toroidal edges of the vacuum vessel sector - are inserted between the outer legs of the coils by welding, while the remaining lower ones are welded only after having finalised the tightening of the toroidal links of the vacuum vessel support system. Space for this closure weld is limited due to the proximity of the vacuum vessel and thermal shield. However, all horizontal ports, including NBI ones, may be mounted after welding of the friction joint panels and there is sufficient space for it to be built up from front and back, minimising the weld distortion (which can create both high residual stresses especially in the joint fingers due to bending and coil misalignment). The bolts of the friction joints are left slightly loose during welding to allow for thermal expansion as well as weld shrinkage, and are tightened once upper and lower panels have been completed. After the friction panel structure linking the outer legs is complete, the upper and lower precompression rings are installed. The procedure that is being considered is to assemble ring pairs on their back flange (36 pieces) and pre-tension the flange/ring assembly with a specially designed circular frame. This pre-tensioning operation can be carried out by the ring manufacturer and the rings delivered already pre-tensioned on their frame. The on-site work is then essentially limited to tightening the radial bolts between the ring back flange and the coil case. When this is done, the pre-tensioning frame for each ring pair is removed. On the first warm up after the coils are charged (and subsequently if needed), the ring pretensioning can be adjusted to compensate for any settling effects. Retensioning, if necessary, can be performed with the CS in-situ, using the radial bolts (Drawing ). It is essential that the outer intercoil structures (all four sets, including the friction joint panels) are completed before the rings are tightened, as the rings also produce a substantial lowering of the tensile (toroidal) loads carried by these. Table shows the effect of the rings on the reduction of toroidal tension in the friction joint when the precompression is applied after completion of the panels (see also Structural Assessment of the Friction Joint at Two Different Assembly Procedures, C. Jong, Garching, 26 April 02) Design Description Document 11 Magnet Page 77

79 All Corrections Coils (lower, side and upper) are then attached to the TF coil cases. The PF coil installation starts with a survey of the support positions on the TF coil cases. If required, appropriate shims are prepared and installed before the PF coils are attached to the TF coils. Coil current centre alignment is done only on the basis of the geometry of the winding pack. This is measured at the final stages of manufacturing and appropriate fiducial marks are permanently attached to the TF coil case. The accuracy of the coil tolerances which cover the combined manufacturing and assembly tolerances are given in 'Contributions to Plasma Error Fields from the CS, PF and TF Coils' ITER_D_23DVQU version 1.3 (see also section ). The feeder installation (apart from the CC header ring) is the next step in the magnet assembly. Those under the machine are either lowered into position through the bore of the TF coils (Drawings 11H and 11G ) or brought in through the side ports (Drawings 11G ). Finally, the CS is assembled, complete with its preload structure and all terminals and cooling pipes and manifolds, in the assembly hall. It is then lowered into the inner bore of the TF coils and hung from the supports at the top of the TF coil inboard legs Operation Sequences and Flexibility Cool Down and TF Coil Charging For cooldown, the cryoplant is switched to provide directly helium gas from the cryoplant to the magnets and structures. The primary heat exchanger of the coils is bypassed. Temperature monitoring of the outlet gas temperature is used to control the cryoplant to maintain a temperature difference of 50K between inlet and outlet to limit thermal stresses to acceptable values. The cooldown rate is an average of about 0.4K/hr between 300K and 4K. Additional time is required to cool the thermal shields and fill the magnets with Helium (see DDD3.4). Additional temperature sensors on remote parts of the structure may also be required to confirm temperature uniformity. Once the coil outlet temperature is below about 20K, the cryoplant is switched to liquefaction mode and the magnets are supplied with supercritical helium through the primary cooling circuit. The TF coils are generally maintained in the cold condition even if not charged. Charging (and discharging) takes place over a timescale of about 30 minutes through the power supplies. Before charging is started, all instrumentation and fast discharge systems are activated Vacuum Vessel Conditioning During vacuum vessel conditioning, electron cyclotron discharge cleaning is applied to the walls using the auxiliary heating systems. A toroidal field is required to enable the RF antennae to couple to the plasma, in the range of 4-5.7T. There is no poloidal field. Design Description Document 11 Magnet Page 78

80 Hydrogen Plasma Pulsing In the initial stages of plasma operation, hydrogen plasmas will be used. The machine activation will be low and full hands-on maintenance of the coils will be possible. This operation phase will be characterised by variable plasma currents and short flat top periods as the plasma control and auxiliary heating systems are characterised. The main coil heat loads will arise from AC losses and the pulse rate may (as far as compatible with the cryoplant) be above the 2 pulses per hour of the reference scenario. Frequent plasma disruptions can be expected DT Plasma Pulsing There will be a phase during which the in-vessel systems are characterised for burning plasmas. The plasma current will generally be high (in order to obtain the necessary confinement) but the full range of scenario flexibility will be investigated. Burn lengths will be variable, from short (<100s) to long burn non-inductively driven (>1000s) and there will be a moderate frequency of disruptions. In the later stages of operation, a reference scenario will be selected and the machine operation will consist largely of repetitive pulses with this scenario. Disruptions will be rare Scenario Flexibility Scenario flexibility with D-T plasmas is an important feature of the machine. The main characteristics of this flexibility are: i) Although plasma current ramp up follows a single rate of current rise, there are variations in the point at which the divertor configuration is formed that affect the PF coil currents. ii) For the reference plasma current of 15MA, the PF coils are able to confine plasmas with a range of li and poloidal beta while providing a minimum burn time of 400s and a repetition rate of 2 pulses per hour. iii) The PF coils and TF coil heat removal systems are able to drive 700MW burning plasmas with currents up to 17MA and with a limited range of li and beta for at least 200s, with no requirement on the repetition rate. iv) The PF coils and TF coil heat removal systems are able to drive a steady state 360MW burning plasma with non-inductive current drive for a few thousand seconds, and a repetition rate of 1 pulse every 2-3 hours Warm up and Cryostat Opening During opening of the cryostat to air the coils may be heated to not more than 350K (measured by the He gas inlet temperature) to prevent water condensation on the magnet surfaces Faults and Safety Because of the severe effect of magnet faults on the overall machine availability, and the difficulty of repair, there are multiple monitoring and protection systems built into the design. These include inherent features, detection/monitoring systems (that operate continuously while the coils are charged), and testing systems (that are applied periodically when plasma pulsing is interrupted or when the magnets are discharged). Particular attention Design Description Document 11 Magnet Page 79

81 is paid to reducing the probability of potential cascade sequences, where the existence of an initial fault increases the probability of others (for example, heat from a short degrades a protection barrier or increases local voltages), and common mode faults where several components (due for example to a common manufacturing error) have the same initial fault. Details of these systems are given under the component descriptions but can be summarised: All coil insulation is designed (and proof tested) for operation at the voltage level caused by a single ground fault; The TF coil radial plates enable insulation quality monitoring and prevent damage in the event of a single short; The PF coil insulation wraps similarly enable insulation quality monitoring but cannot limit short currents; The quench detection systems consist of a primary (resistive voltage based) and back-up (flow based) detection systems; The TF discharge resistors and switches are failure tolerant, so that failure of 1 out of 9 still allows a near-normal coil discharge; The structures are redundant so that full catastrophic failure of one component (such as the TF coil case) does not cause structural failure of the whole coil; Surfaces facing the cryostat are hard grounded to prevent arcing. The magnets have only a limited tolerance to faults in the power supplies and, therefore, the quench detection systems and fast discharge switches have high redundancy. In the absence of electromagnetic disturbances (particularly plasma disruptions) and nuclear heating the magnets can tolerate a loss of helium circulation for many hours (as the high field regions are well protected from thermal radiation). The initial fault and damage mechanism assessment for the magnets (see section 2.4) shows that: All potential damage mechanisms that can affect the nuclear components (either the cryostat, the vacuum vessel or the pipework and ducting within the cryostat) are associated with molten material produced by arcing. Contrary to general expectations, shorts that develop inside coils do not lead to significant arcing and the damage can be confined to the coil. External shorts on the CS and PF coil busbars (superconducting or normal) can potentially lead to molten material generation in the coils themselves (not significantly in the busbars) due to the coupling of extra energy into the coil, followed by a quench that cannot be discharged. Due to the thin protection cases on the PF coils, a substantial fraction of this molten material could enter the main cryostat (1,000kg would be a conservative estimate). Due to the location of the PF coils, most of it would be deposited onto the vacuum vessel, although distributed over several square metres around the toroidal direction. Although this molten material from the coil could cause some local melting of the thermal shield and outside of the vacuum vessel by heat transfer, the material will be sufficiently distributed that there is no possibility of failure or leaks. Arcing within the cryostat outside the electrical components is not possible, and busbar arcs will transition through the cryostat in a few seconds. Molten material production from busbar melting is small and mostly confined within the busbars. Most arc accidents will be associated with helium leakage from the affected coil into the cryostat and with failure of the feedthroughs in the cryostat wall. Design Description Document 11 Magnet Page 80

82 1.2 Component Descriptions TF Coil The 18 TF coils are D shaped and consist of a winding pack contained in a thick steel case. The TF coil case is shown on Drawings 1101CA_ , 1101CA_ and 1101CA_ The winding pack is a bonded structure of radial plates (which contain the conductor) with an outer ground insulation. Cross sectional views of the TF coil at the equatorial plane are shown in Drawings 1101CA_ and 1101CA_ The coil terminals project out of the case on the lower curved part of the case, Drawings 1101WP_ and 02; all other double pancake joints (i.e. joints linking radial plates) are contained within a local extension of the case, Drawing 1101WP_ Some of the coil physical parameters are summarised in Table and inductances are given in Table Table TF Coil Parameters TF Coil parameters Length of the coil centre line (m) 34.1 Cross section of the steel at inboard mid plane (jacket, radial plate, case) (m 2 ) Cross section of the steel at outboard mid plane (jacket radial plates, case) (m 2 ) Materials (in the form of plates and forgings): Example Class 1 TF inboard leg JJ1 steel Class 6 Upper and lower OIS regions ANSI 316LN steel OIS bolts, IIS poloidal keys, PCR bolts Inconel TF Conductor Table TF Magnet Inductances Coil Inductance (H) Full TF magnet (18 coils) 17.7 Mutual between coil 1 and coil N Coil 1 (self) Coil 2 or Coil Coil 3 or Coil Coil 4 or Coil Coil 5 or coil Coil 6 or Coil Coil 7 or Coil Coil 8 or Coil Coil 9 or Coil Coil The conductor is a circular Nb 3 Sn cable-in-conduit with a central cooling channel, cooled by supercritical helium, as shown on Drawing 1101WP_ The parameters are shown in Table The conductor current of 68kA is determined by the fit of the conductors into the winding pack and the maximum allowable value of 70kA. In addition to the cable size limits due to concerns about the transverse loads (section ), it is also important Design Description Document 11 Magnet Page 81

83 for the out-of-plane strength of the coil case that the minimum case thickness at the ends of the inner leg does not drop below 7-8cm. This gives a constraint to the acceptable range of values of the conductor diameter if the maximum number of turns are to be fitted in the available winding pack area while keeping the required groove spacing in the radial plates. Table TF Conductor Total number of turns per coil 134 Jacket type circular steel Number of turns per pancake (from side pancake to middle pancake) Type of strand Nb 3 Sn Operating current (ka) Nominal peak field (T) 11.8 Operating temperature at peak field (K) 5.0 Total operating strain (%) Equivalent discharge time constant (s) hot spot 15(*) n at operating point 7 Tcs (Current sharing temperature) (K) 5.7 Iop/Ic (Operating current/critical current) 0.78 Cable diameter (mm) 40.5 Central spiral outer x inner diameter (mm) 9 x 7 Conductor outer diameter (mm) 43.7 Jacket material 316LN(modified variant) SC and Cu strand diameter (mm) 0.82 SC strand cu : non-cu 1.0 Cabling pattern ((2s/c+1Cu)x3x5x5+core)x6 SC strand number 900 Cu core at 4th stage 3x4 Cu wires 0.82mm Wrap coverage on final substage (0.05mm SS) 50% Local void fraction (%) in strand bundle 33.2 SC strand weight/m of conductor (kg/m) 4.50 (**) (*) note the use of non-linear discharge resistors, see section (**) cosθ =0.95, d strand 9 kg/dm 3 The strands, 0.82mm in diameter, are chromium coated to provide a chemically stable outer surface, with an acceptable conductivity to adjacent strands, which does not diffuse into the underlying pure copper during heat treatment. They are cabled to a 5 stage cable with the final 6 fifth-stage subunits cabled around a central cooling spiral (a tube formed from a steel strip to give good access for helium to cross between central channel and cable). The central channel is essential to minimise the pressure drop while permitting an adequate mass flow rate to remove the nuclear heat from the first two turns. The local cable space void fraction is about 33%, to give an acceptable level of transverse conductivity. The cable incorporates pure copper strands (with the same diameter as the superconducting strands to give better cabling compatibility) to build up the copper section necessary for quench protection, and the final substage has a 50% wrap of steel foil (0.05mm thick) to control AC losses (see Drawing 1101WP_ ). The jacket is assembled by butt welding circular seamless tubes that are about 1.5-2mm oversize on the final conductor dimension. The cable is jacketed by the pull-through method through the finished unit length of empty jacket. The Design Description Document 11 Magnet Page 82

84 cable has a final wrap of steel tape (0.08mm, 50% overlapped) to protect it and maintain its dimensions during the pull through process. The jacket is then compacted onto the cable by passing it through a set of rollers that reduce the jacket dimensions in a single pass. As indicated in section , the discharge time constant is not the minimum compatible with the allowable coil voltage, but has been selected to keep the stresses in the vacuum vessel within accceptable values. The table shows a discharge time constant of 15s which is equivalent (in terms of the I 2 t integral) to a time of 2s for quench detection and implementation, followed by a current discharge with a time constant of 11s. The material used for the conductor jacket is stainless steel, modified with very low carbon content to avoid embrittlement during the heat treatment to form the Nb3Sn. The basis for this choice is given in section The cooling inlet sections of each double pancake are located at the inner surface of the coil, at the double pancake cross-over regions, as shown on Drawing 1101WP_ The cold helium quickly reaches the high field region where most of the nuclear heating is concentrated, cools the rest of the pancake and exits through the joints located on the outer surface of the coil at the bottom curved part. The conductor design point is where the combination of maximum field and nuclear heating creates the minimum temperature margin, which occurs towards the end of the TF coil inboard leg in the innermost turn TF Double Pancakes, Radial Plates and Joints Winding and Conductor Terminations The winding uses one-in-hand conductor (about 800m long) with a double pancake configuration. If the conductor-plate transfer operation (see below) proves excessively complicated, a single pancake winding could be used at the cost of doubling the number of joints. The extra joints would be required on the inside of the winding pack which is a moderate field level location (5-6T). If these joints were also cooled in series with the conductor, the joint heating would create a further K temperature rise in the high field region and might require a small increase in the conductor design operating temperature. The conductor is wound into a 'mould' to hold the shape of the conductor during the reaction heat treatment. The mould must carefully control the conductor dimensions but does not require the complexity of a radial plates. During the TF Model Coil manufacture, it has been found that the conductor experiences a small change in longitudinal dimensions during the heat treatment, measured to be an elongation of 0.05%. During heat treatment, it is held flat by clamps but is allowed to move radially. The initial winding dimensions must take account of this change in length after heat treatment, and the change must be confirmed by trial heat treatments of a few turns of superconductor. During winding the conductor jacket at the inner transition region is opened and the helium inlet formed. The jacket and outer cable wrap (both outer wrap and subcables) are removed by hand over a length of about 25mm. Two semi-circular shells are then welded over this opening, one of them containing the helium inlet. The inner surface of the shells contains grooves to allow helium distribution into the cable, Drawings 1101WP_ and 1101WP_ The conductor terminations are formed after winding but before heat treatment of the conductor. The reference designs of the terminations and the joints (between double Design Description Document 11 Magnet Page 83

85 pancakes) are illustrated for a lap joint option (section ) on Drawings 1101WP_ and 1101WP_ Each joint consists of a box machined from an explosion bonded steel-copper plate, Drawing 1101WP_ The steel side of the box is initially open, leaving a U-shape whose base is copper. The cable (with outer wrap and internal subcable wraps removed on the outer surface but maintaining the subcable wraps inside the cable, and a reinforced central channel) is pressed into this box, reducing the local void fraction to about 25%. The box lid is then welded shut. Joint connections are made, as shown in Drawing 1101WP_ , after impregnation of individual double pancakes but before the final impregnation of the winding pack assembly (see Section ). The butt joint option would have a similar configuration for the joint box but the internal layout would correspond to Drawings 1101WP_ and 1101WP_ , with a U section of superconducting cable being used to connect the two conductor ends. After winding of the conductor and completion of the joints, the conductors are heat treated (about 650 C for 200hrs) in a furnace with a controlled inert atmosphere. After heat treatment, the conductor is 'unsprung' from each side of the mould using a special handling tool. The mould is then withdrawn from between the two unsprung conductor layers and the radial plate is inserted, see 1101WP_ This is a complex operation because the two conductor pancakes are linked by the transition region and it is essential to avoid any conductor strain above about 0.1%. The conductor is then returned to the grooves of the radial plate. The conductor is then locally lifted from the radial plate (lifting about 1 turn gives the required vertical movement of 0.5m) and a 1.25mm layer of dry insulation is applied to the conductor as the unspringing proceeds along the pancake. The conductor must be carefully supported during this unspringing to avoid strains above +/-0.1% that could permanently damage the superconducting cable. The insulation consists of typically three layers of half lapped polyimide film interleaved with glass, followed by one or more layers of glass. The insulation includes two thin (0.1mm) steel strips with a width of about 6mm, enclosed between two polyimide films, which are each wound around the conductor with a pitch of about 40cm (the same as the final cable twist pitch and in the same direction to reduce differences in electromagnetic coupling between the plasma and cable/strip in the event of a plasma disruption) to act as a voltage based quench detection system (one is for redundancy), see section The insulation layout is illustrated on Drawings 1101WP_ and 1101WP_ The correct alignment between conductor and plate has to be maintained during this process, using marks and adjusting the number of outer glass wraps, to prevent a progressive build up of misalignment Radial Plates The insulated TF conductors are wound into grooves on each side of a total of seven flat steel radial plates, Drawing 1101WP_ The radial plates are of three types (3+9 turns on one side, Drawing 1101WP_ ; turns in the middle, Drawing 1101WP_ ; 9+3 turns on the other side). The wedge shaped side pancakes are essential to maintain the minimum case thickness that is needed for out-of-plane support while minimizing wasted space (i.e. space that does not contain conductor or the functional part of the radial plate with the grooves) in the winding. These side pancakes have a different conductor length. There are flow throttles in the helium supply to maintain the same flow rate as in the central pancakes. The turn to turn and pancake-to-pancake conductor transitions, the cooling lines, joints and current feeders are all located at the lower curved region. Design Description Document 11 Magnet Page 84

86 The material used for the radial plates is SS316LN with two levels of yield stress requirements, see Table It is planned to machine the high strength material out of solid forged plates and to form the low strength sections by welding extruded profiles. The plate is assembled by butt welding the various sections, with laser welding to avoid distortion that would require a further machining step. Outside the transition region near the joint region, the grooves are parallel and constant width, with the exception of a local region above and below the inner straight leg, Drawing 1101WP_ Here there is an adjustment in the spacing to allow a thicker tooth between the grooves in the nose in the inner leg, to better support the wedging forces. The outermost grooves (i.e. those nearest the nose of the coil) also have a thicker tooth in the inner leg than those further in. In the outer leg, all teeth have the same thickness. The width of the grooves in the outer leg is larger (by 2mm) compared to the inner leg. This allows some space to absorb tolerance effects during the conductor winding, by adjusting the position of the conductor in the groove. Table Radial Plate Material Classes Class Yield Strength at 4K Fracture Toughness at 4K 7A (inboard) >950MPa >125MPam 1/2 7B (outboard) >700MPa >125MPam 1/ Final Insulation Each jacketed conductor is contained in its groove by means of a cover plate that is laser welded in place. These cover plates have small holes every 20cm to allow epoxy resin flow during the vacuum impregnation process. In the grooves, there is a total space of 4mm (on the diameter) for insulating material between the conductor and the radial plate groove. The electrical turn insulation has a nominal total thickness of 2.5mm (over the diameter). The remaining 1.5mm (on the inner leg) or 3.5mm (on the outer leg), plus whatever can be obtained by careful compression of the dry insulation wrap, has been left as assembly gap. This gap is later to be filled with epoxy resin and additional layers of glass insulation. Once the groove is closed, the radial plate is then covered with a 2.1mm (nominal thickness) layer of insulation to complete a double pancake. This insulation thickness may be locally reduced to a minimum of 1mm to accommodate the tolerances on thickness and flatness of the radial plates. This insulation consists of an interleaved polyimide and dry glass wrap. The whole plate assembly including the conductor turn insulation is then vacuum impregnated with epoxy resin and cured. The final tolerances on the double pancake are therefore those of the impregnation mould, not the radial plate TF Winding Pack Seven double pancakes are stacked together to form the winding pack of one TF coil, as shown on Drawing 1101WP_ In this operation, the radial plates are aligned along the front edges of their inboard legs so as to provide a flat and smooth bucking surface. A 0.8mm layer of glass sheets is placed between double pancakes to provide a path for the epoxy resin during impregnation and allow a proper bonding of the double pancakes. The ground wrap has a minimum thickness of 7 mm and includes multiple wraps of a polyimide film electrical barrier material interleaved with glass. Design Description Document 11 Magnet Page 85

87 The joints are placed outside the main winding (in a praying hands configuration) but within the containment provided by an extension to the TF coil case. Where each conductor leaves the plate there is a stop welded to the jacket, Drawing 1101WP_ This prevents debonding of the conductor turn insulation in the exit region due to differential thermal contraction between conductor and jacket (even though the jacket and plate are both steel, the cable causes the jacket to elongate relative to the plate during cooldown). Drawings 1101WP_ and 02 show the joint layout. The most favourable orientation for the lap joint options (to reduce eddy currents across the contact surface) is for the contact surface to lie perpendicular to the plane of the coil. This is, however, not so convenient for manufacturing reasons and the reference design uses contact surfaces parallel to the plane of the coil. The joints between double pancakes are connected during the stacking process as accessibility is much better. There is sufficient flexibility in the lengths of conductor leading to the joints to allow for small movements of the plates as they are compressed into the impregnation mould. The lap joints will be made by soldering before impregnation. Any misalignment between the contact surfaces will be taken up by the use of copper shimming plates (and overlap by local machining). After this step, the stack of double pancakes together with the ground insulation is vacuum impregnated and cured. The cooling lines to the helium inlets on the inside of the coil are brought round the outside of the coil winding packs and into the joint region, where the insulating breaks (section ) are located. The cooling lines are at high voltage and are included within the final ground insulation, using filler pieces to provide a smooth surface and load protection, Drawing 1101WP_ Casing and finishing operations The winding pack is inserted into four case components (an inner and outer U section each with a corresponding cover plate on the inner surface, as shown in Figure The case surrounds completely the winding pack and is closed by welding. The inner surface of the coil case is coated with an anti-adhesive material which prevents the winding pack from bonding to the coils case during the final insertion process as the remaining case - winding pack gap is filled with resin. A bond would fracture in operation as the winding pack pushes out against the case outer surface, allowing abrupt movement, whereas slip reduces shear stress in the winding pack and also creates a small gap for thermal insulation against nuclear heat conduction from the inner case surface along the inner straight leg. The TF coil case design and manufacture are described in section Design Description Document 11 Magnet Page 86

88 Fig Main sub-assembly and sections of the TF case Insertion of the winding pack into the case is shown diagrammatically in Drawings and is described in detail in ITER_D_23HNWN version 1.2, Case Insertion and Assembly Procedure Specification and Tolerance Requirements for the TF Coils. It starts with the positioning of the inner U section horizontally (with the nose on the ground and the U pointing upwards). The critical point is the final accuracy (position and shape) of the wedged shape of AU for the 18 TF coils. This is achieved at machine start-up as long as the toroidal geometry of the nose of each individual coil is within tolerances and the coil noses are placed at the same radial position (again within tolerances) at assembly. The coils will move inwards by the same amount simultaneously and wedge into an accurate circle on energisation. To maintain acceptable stresses in the nose regions of individual coils, and to avoid interference during assembly, tight tolerances are required on the TF coils 20 deg wedge surfaces. To ensure the accuracy of the current centre line in the TF final assembly in the machine, the WP has to be placed with tight tolerances relative to the nose of the coil sub-assembly AU. This is achieved by the accurate machining of the inside flat surface of section A2 and by the 7 mm lateral gaps foreseen between the WP and the case; which allow sufficient space to position the WP accurately relative to the nose region (even tilting it slightly if required to align the outer leg), without being constrained by other parts of the case. The insertion process process procedes as follows: Step 1 WP insertion in sub-assembly AU Coat inner flat surface of sub-assembly AU with 0.5 mm of resin to smooth irregularities. Lower the winding pack into the section AU. Align to give the vertical positioning. The winding is adjusted to compensate for the out of plane tolerances of the winding pack, so that it could be placed slightly 'skewed' along the length to minimize as much as possible the influence on the error fields. If perfectly aligned, the gap between the internal lateral walls Design Description Document 11 Magnet Page 87

89 of the subassembly AU and the WP is 7mm. The maximum allowable errors at F and G can be compensated by using 4mm of this. The WP current centre at sub-assembly AU can be moved toroidally within this space. It is essential that the WP plane is perpendicular to the flat surface of AU. This can be achieved by moving the WP in the toroidal direction and/or by 'tilting' the WP in the out of plane direction. A 1mm out of plane movement at the outer leg can be obtained approximately by an extra 0.2mm gap at the flat surface of the nose which can easily be filled by the resin coating. The WP must remain at least 3 mm clear from the lateral walls of the case at the top in order to allow for welding shrinkage during closure plate AP welding. Step 2 BU sub-assembly installation Possibly machine the weld preparation of BU-AU butt weld on section BU (probably not required but can be used to ensure the 10mm case-wp gap in the outboard region once the WP alignment is finished). Coat the inner surface of BU with an antiadhesive agent so that the filler material sticks to the WP and not to the case. Insert section BU over the WP. The 10mm gap all around the WP in the outboard leg BU makes safe the delicate BU installation onto the WP and ensures a sufficiently big gap to protect the WP from excessive weld distortions in step 3. Weld shrinkage of the case walls is expected, the WP must remain free with no contact with BU after the butt welds to AU are made. Local protection to the WP needs to be placed at the back of the butt weld position just before the BU is positioned for welding (space should be available for 3mm, the protection will then remain inside the case). Step 3 BU-AU weld Make the butt welds to section AU. This may cause local distortion of the case or a global movement of BU (probably twisting out of plane) but as long as the WP position is unchanged because of the free gap between WP and the case, this is acceptable (i.e. there are no tight tolerances required on the external case shape). Step 4 Fixation of WP in case and CCL benchmark trasnfer The coil weight is still supported on the sub-assembly AU and it must remain in this position unless a more rigid support at the sides and especially at the poloidal inner surface of BU is provided (at this point there is a gap between WP and case everywhere but in the flat surface of AU). Different options are under consideration to fix the WP toroidal position in BU and AU. One is based on the installation of rigid shims at the sides every 50cm or so; these have to hold the poloidal position also against any differential pressure that develops during filling of the gap with resin. Another option considered is a set of epoxy filled bladders that are pre-placed on the outer poloidal face and sides of the WP within BU (and the sides of AU) and are filled at this point, using the access from the inner bore (closure plates AP and BP are not yet in place). Probably 6 bladders at both sides and outer locations around BU and 4 along both sides of AU would be adequate. This resin would have to be self curing (i.e. mixed with a hardener before injection and left to cure at room temperature). At this point the WP datum marks must be transferred (by accurately defined displacements) to the outer walls of the case. The reference points of the WP are chosen so that the transferred positions in the case are well away from any weld distortion caused by step 5. Step 5 Weld of cover plates AP and BP Put in the inner bore cover plates AP and BP and weld in place, welding AU first. Again, possibly use an antiadhesive agent on the inner surface. The side distortion on AU is expected to be acceptable as long as it is localized on the back half of the coil, i.e. it should Design Description Document 11 Magnet Page 88

90 not be necessary to re-machine the case after this welding. If the bladders are used in Step 5, at this point the coil could be inverted in order to make the closure weld of BP in the 'down' position, which would be easier; however not moving the coils during the assembly phases is the preferable solution. If the bladders are not used, the coil must remain in the same position until the case-winding gap is finally filled and the epoxy cured; in this case the BP closure weld must be made in the 'up' position. Step 6 Filling of the WP-TF CC case It is probably necessary to use a charged epoxy filler (resin with chopped glass to increase the resistance to compressive stresses) as it is very difficult otherwise to get a sufficient filler content in the resin (there is no access once the case is closed and the gap cannot easily be densely packed with glass before). A VPI process is needed to improve the impregnation and avoid voids. The areas around the inner cooling pipe penetrations can be used for resin inlet and an extra hole might be needed at the outer equator to monitor filling. Step 7 Finishing of the TF coil. Final machining of the assembled TF coil should be avoided whenever possible (although may be required for example on the first coil where manufacturing experience is still being gained). The 3 interface surfaces P, Q and R are used to define the coil positions (see Figure 1.2-2) during the installation phase; hence their accuracy is important. Location holes for dowels (and bolts) in the 4 surfaces PL,R and QL,R are pre-drilled in the TF CC; these holes provide the final alignment of the coils and are therefore critical; if required a re-machining shall be performed but this is not expected. An accurate survey of the surfaces P, Q and R will allow the design of shims to match the nominal dimensions within tolerances. The 4 shims required are machined such to achieve a flatness tolerance within +/-0.5mm with angles +/-10 around theoretical plane OAB with the additional requirement that tolerance on the toroidal distance between points of surfaces P1 and P2 and Q1 and Q2 should be neither bigger than +/-0.5mm. The surfaces must be insulated and the final dimensions should allow for a sheet of 1 mm of composite G10, though this thickness could vary and serve as shimming between pairs of coils at the assembly stage. The holes are drilled in the shims at fixed distances on the P and Q planes from the C and F reference points with (R, Z) position errors of +/-0.2mm. The gravity support plane has a similar tolerance although it does not contribute to the centre line positioning. This tolerance is mainly for assembly convenience. Step 8 Poloidal Keyways An accurate survey is required in this stage to determine the keyways shape. The poloidal keyways must be locally accurate in shape and straightness (order +/-0.2mm) but a larger uncertainty in the absolute position on the coil is acceptable as this can be corrected on assembly (order +/-1-2mm in poloidal location). There is an uncertainty on the effect of the case closure weld AU-AP on the keyways as they extend to the back of the coil (the case walls are thicker in this section than in the inner leg and should resist better the weld shrinkage). Nevertheless, if this is found to be a problem the keyways could be left undersized (or not machined at all) with final shaping with a mobile milling machine at the end of Step 5. Final possible coil-coil local misalignments may be also compensated at level of keyways by machining, after accurate survey, of the C-shaped 7-mm thick intermediate elements inserted in the keyways to house the cylindrical keys themselves. Step 9 Insulation of Wedge Surfaces Design Description Document 11 Magnet Page 89

91 Attach the insulation sandwich to one coil lateral wall of sub-assembly AU (either bonding or tack welding through small holes in the surface). This insulation sandwich sheet is 6 mm thick (ss-g10-ss) and it is not a shim; hence it does not need machining to compensate the TF coils tolerances. However, with the first coil, if a final machining of the nose is found to be required, the insulation sheet thickness could be increased to compensate the lost material (see final section). 1 Interface surface P 2 C 7 5 J Magnet Cen ter G 6 3 F 4 Interface surface Q Interface surface R Machine & Magnets OZ axis J R C, R FR P, R QR G R J L C, L FL P, L QL Fig Definition of reference assembly surfaces P, Q and R and transfer of CCL reference points to the external walls of the cases Tolerances for the case insertion procedure are shown in Drawing 1101CA_ A 10% margin on the present stress analysis predictions is allowed to accommodate extra stresses due to geometric imperfections. G L Design Description Document 11 Magnet Page 90

92 TF case design and manufacture The centring force on each TF coil is reacted by toroidal hoop pressure in the central barrel vault formed by the straight inboard legs of the coils. The front part or nose of each coil case is thickened to take part of the load. In operation, the coil cases are wedged over their full thickness and about half of the centring force is reacted toroidally through the winding pack part of the coil, while the other half is reacted by the case. During operation, the winding pack pushes outward all round the case and tends to shrink slightly in the toroidal direction due to the Poisson effect. A gap opens on the inside (facing the plasma), of the order of 0.5mm in width (see section 2.2 Appendix 1) and may locally close due to torsion of the coil under out-of-plane forces. A positive consequence of this gap is to reduce the heat conduction from the case to the winding pack. This thermal barrier has (conservatively) been reinforced with a thin insulating layer on the back of the winding pack, and the case cooling pipes on the inner surface also act as a screen, with spacing chosen to minimise heat transfer to the case even in the case of full thermal contact and the failure of the thermal barrier. In the inboard leg, the wedging pressure tends to suppress the side gap. On the outboard leg, there is a small amount of slip through the plasma pulse as the out-of-plane forces change. In the inboard curved regions above and below the straight leg, the case thickness has to increase rapidly to provide adequate out-of-plane support. Case sections are shown in Drawing 1101CA_ , 1101CA_000548, and 1103CA_ The peak stresses tend to occur at the point where the straight leg ends and are sensitive to the minimum wall thickness of the case at this point (this determines its torsional rigidity which supports the loads as the wedging stops). Thus the minimum case wall thickness permissible in the inner leg is 7-8cm. Each leg of the TF coil case is formed of a U-section and the closure plate. The resulting four main TF coil case sub-assemblies are shown in Drawings 1101CA_ A. The U-based sub-assemblies, inner and outer, are made up of a total of 7 poloidal basic (3 inner and 4 outer) segments joined by means of butt welds. The subassemblies are expected to be formed by forging and/or welding of plates. The use of castings is generally not possible because of the yield strength requirements and also because of concerns about defect detection in fatigue loading condition. The subassemblies have different material property requirements depending on their location (see Drawing 1101CA_ ) These fabrication butt welds must be placed outside the peak stress regions in particular the upper and lower inboard curved regions. Analysis shows that at the ends of the straight leg, the butt weld will be close to, but not coincident, with the peak stress. Butt welds between poloidal sections (with the exception of the final transverse and poloidal closure welds, see section ) are expected to be a combination of Electron Beam (EB) welding for the first 50 mm of thickness followed by multiple passes of Submerged Arc Welding (SAW) for the remainder. This requires the weld pool to be on a horizontal surface during welding which has implications on the handling tooling required for manufacture of the sections. The highest stress sections of the case, where the highest material yield stresses are required, are the inboard leg and the regions where the intermediate outer intercoil structures (the friction joint assemblies) are attached. These will be made from high strength forged Design Description Document 11 Magnet Page 91

93 austenitic steel, using electro-slag refining, to the ITER 'class 1' material specification. An example is the special purpose cryogenic steel JJ1. This has a higher manganese content than 316LN which gives a slightly different thermal contraction but finite element simulations show that the thermal stresses produced by joining it to 316LN sections are negligible. The upper and lower inner curved regions have high tensile stresses and require a high fatigue resistance but have lower yield stress requirements. These will be made from high strength steel to the ITER class 3 specification. An example is the EU material EK1 (essentially a 316LN steel) used for trial fabrications of case sections. The remainder of the coil will be made to ITER class 5 and 6 specifications, with 6 being achievable with a standard 316LN steel. The use of multiple material specifications is required in order to reduce costs and also to stay within the available capacity for ESR facilities. The specifications are summarised in Table and Fig Various options exist for forming the case sections. They can be forged as a square tube before cutting into two U-sections as was done in the TF coil case R&D programme. Another option, under evaluation, is to forge U-sections as complete rings. As discussed in , castings are not acceptable for use in the TF coil case or associated structures because of problems in achieving adequate levels of defect detection. Table TF Case Material Classes Yield Strength at 4K Fracture Toughness at Class 4K 1 >1000MPa >200MPam 1/2 2 >900MPa >200MPam 1/2 3 >850MPa >200MPam 1/2 4 >750MPa >200MPam 1/2 5 >650MPa >200MPam 1/2 5A and at RT >290MPa 6 >550MPa >200MPam 1/2 Design Description Document 11 Magnet Page 92

94 Material Class Color Yield Strength 1 Red >1000MPa 3 Blue >850MPa 5 Light Grey >650MPa 5A Dark Grey 6 Green >550MPa Fig Material Yield Stress Requirements The case will be equipped with a set of cooling channels manifolded in the joint area to intercept nuclear and eddy current heat loads before they reach the superconductor in the winding pack. The cooling channel layout is illustrated on Drawings 1101CA_ These channels are placed on the inner surface of the case and run poloidally around the circumference. There is a single inlet manifold at the bottom of the coil and two outlet manifolds at the top, all external to the coil case. The manifolds are connected to the external valve boxes to allow control of the helium flow, one circuit cooling the inside inner wall and the second the 3 other inner surfaces. Each of the two circuits has two branches, Design Description Document 11 Magnet Page 93

95 one covering the inner leg of the coil and the other the outer, both flowing from bottom to top. The cooling channels are tubes located in slots machined on the case inner surface and held in position by a thin welded cover. The cover weld does not need to be leak tight and contraction of the weld helps press the tube into the slot, achieving good thermal contact. The use of single tube lengths without butt welds except at the external manifolds should greatly reduce the risk of helium leaks inside the case. On the inner surface of the inboard leg (where the nuclear heating is concentrated), the spacing between the tubes is about 40mm and the tube inner diameter about 10mm (see Drawing 1101CA_ ). Some limited R&D may be required to verify the thermal contact which can be achieved between the tubes and the case and investigate the effect of cool-down/warm-up cycles on it. If required, a fine copper mesh can be wrapped around the cooling pipes before they are pushed into the slot, to improve contact. An additional outer circuit (see Drawing 1101CA_ is required to cool down the massive parts (e.g., the intercoil structures and PF supports) on a reasonable timescale without excessive temperature differences, as they are not close to the inner circuits or winding packs. The outer circuit also cools local hot spots on these components. The outer pipes (inner diameter about 10mm) are held by C clamps every ~20cms welded to the case and again using copper mesh around the pipes to improve thermal contact. There are 3 parallel circuits covering essentially the two sides of the coil and the back, Drawing 1101CA_ The overall cooling schematic is shown in Details of the thermohydraulic parameters for these circuits are given in Table The case is not a vacuum-tight containment due to the terminal lead penetrations in the joint region, but any leakage of internal gases will be very slow. Vent channels and a bursting disk are planned to be provided to release any internal gas pressure to the cryostat in the event of an internal helium leak Structures The ITER magnet system requires mechanical structures to provide support against magnetic forces and to provide gravitational support. In the TF coils, these structures link the coils to withstand the out-of-plane loads during a machine pulse. In the PF coils, they link the coils to the TF cases to withstand the vertical forces and net horizontal forces Inner Intercoil Structures At the top and bottom of the inboard leg of the TF coils there are two sets of four poloidal shear keys. The design is illustrated on Drawing 1101CA_ These keys help support the overturning moment through torsion of the central vault but also act to stabilise the wedged area of the coils against slipping. When the TF coils are energized, these keys tend to de-wedge radially outward and become loose. This effect increases the local stress concentrations on the shear key insulation and key slot and reduces the life of these components. The length of the keys has been optimised as described in section to distribute the shear loads uniformly, so that the innermost keys are shorter (50% and 75% of the coil thickness respectively) and do not penetrate to the inner bore. This arrangement also keeps the peak stress region away from the final closure weld of the TF case. Design Description Document 11 Magnet Page 94

96 The keyways are machined oversize and lined with two half-shells made of high strength G10 epoxy glass. The shells are machined based on a precise survey of the keyways after assembly so as to absorb any mismatch between the coils and provide an inner perfectly circular slot for the key. They also provide the low voltage insulation between the coils. The keys are segmented into 4 units (so that the first keyway contains 2 units and the second 3). In order to increase the initial pre-load on the shear keys and reduce somewhat the opening of the key slots when the TF coils are energised, the use of pre-compression rings is foreseen, as shown on Drawing 1101CA_ Three rings at the top and bottom curved regions are installed during initial assembly of the machine and pre-loaded with typically eight Inconel 718 bolts per TF coil, to provide a radially inward pre-load at each curved section of one TF coil of about 30-35MN, corresponding to a radial extension of about 25mm for the ring. The radial pre-load is applied to thick pre-compression flanges welded to the top and bottom of the straight leg of the case. Details of the mechanism to apply the pre-tension, and to re-tighten if needed, are given in section The required pre-compression of 30-35MN has to be achieved within the space constraints in radial direction (due to the presence in the inner bore of the Central Solenoid and its pre-load structure) and vertical direction (not to move out the adjacent PF coils number 1 and 6). These constraints limit the maximum area available for the rings to about 0.22m 2 total. This space restriction means that the hoop stress in the rings must be at least 400 MPa, applied at room temperature. The load on the precompression rings is transmitted to the TF coil itself partially through the two gussets that run back underneath the PF1 and 6 coils. The conflicting requirement of PF coil supports, the gussets and the keyways for the inner poloidal keys make space assignment in this region quite critical The manufacturing procedure for the rings is based on a unidirectional fibre composite wound in hoop (toroidal) direction. The fibre is coated with resin, wound and then cured. More details on the precompression rings are given in 'The Pre-compression System of the Toroidal Field Coils in ITER' (ITER_D_24MLHB v1.0) It is not possible to bring replacement rings under the machine in the case of lifetime failure of the lower PCRs once the magnets have been assembled. However, the rings themselves can be accessed for replacement by the removal of the central solenoid which is a relatively limited disturbance to the magnet system. Therefore replacement rings are placed and stored underneath the machine at assembly Outer Intercoil Structures There are four OIS structures as illustrated on Drawings 1101CA_ The upper and lower OIS are located respectively above the upper ports and below the divertor ports of the vacuum vessel. The upper and lower intermediate OIS are located respectively above and below the equatorial ports. There are two types of OIS. The first one, used in the upper and lower OIS, is based on box structures consisting of two main shear panels linked midway between the coils by insulated expansion bolts, which are oriented in the toroidal direction (see Drawings 1101CA_ and 1101CA_ These expansion bolts (the largest one is based on a M120 pin) carry simultaneously toroidal tension and shear loads. Design Description Document 11 Magnet Page 95

97 The second type, used in the upper and lower intermediate OIS is a friction joint based on a single shear panel, with a thickness of 100mm, protruding from the side wall of the case and supported by webs. This is shown for the upper IOIS one on Drawing 1101CA_ to which is welded the shear panel as in Drawing 1101CA_ and for the lower IOIS one as 1101CA_ to which is welded the shear panel as shown on Drawing 1101CA_ The transmission of the shear loads from shear panel to case is improved by connecting the shear panel to the inner (plasma side) and outer (PF coil side) walls of the case through a Y-shaped base. At the top and bottom re-entrant corners where the panel connects to the case side wall, a large fillet radius is provided to mitigate stress concentration effects.. The shear load transmission across the insulated joint in the panel is accomplished by multiple-finger friction joints which are welded to the two adjacent shear panels. The friction joints are pre-loaded by two rows of insulated bolts acting on 6 fingers separated by insulated washers. This multiple finger arrangement allows to multiply by a factor of 5 the number of friction interfaces and therefore the shear load capability of the joint. The friction joint panel is assembled on site, once the TF coils are in position, as described in section 1.1.6, by a weld on each side to the shear panel extension attached to the TF coil cases. A potential issue is shrinkage when this in-situ weld is performed. This shrinkage is expected to reduce the toroidal dimension of the OIS. This toroidal shrinkage is accommodated by temporarily releasing the bolt pressure to allow some sliding between the fingers of the joint. Shrinkage can also induce bending deformations which can be controlled only if the weld is done from both sides. Access to the inside of the shear panel is, therefore, essential Gravity Supports The machine gravity supports are shown on Drawings 1101GS_ and 1101GS_ They are placed under the outer curved region of each TF coil between PF4 and PF5. A low voltage electrical insulation (at sheet of G10) is placed between each TF coil and its support, with load transmission through insulated bolts and shear keys. The machine gravity support pedestals are equipped with parallel vertical flexible plates, so that they can deflect in the radial direction to allow thermal expansion of the magnet system, but they are rigid versus out-of-plane bending caused by TF coil torsion or seismic motion. They are sufficiently robust to transmit out-of-plane bending moments to a lower supporting ring, which is an integral part of the cryostat structure. This ring resists the bending moments transmitted by the pedestals but transfers horizontal seismic loads to the building through horizontal tie plates. The gravity load is transferred to the building through 18 cylindrical support columns. The thermal load from the room temperature supporting ring is intercepted by a thermal anchor consisting of cooling channels for 80K helium gas (Drawing 1101GS_ ). These channels are welded on the sides of the flexible plates at a distance of about 600 mm below the top connecting flanges of the pedestals. The supports also react dynamic vertical loads from the vacuum vessel/plasma due to vertical disruptions. These loads are defined in section Design Description Document 11 Magnet Page 96

98 PF Coil Supports The PF coils are self supporting as regards the radial magnetic loads. These loads are reacted by hoop tensile stresses in the conductor jacket. The coil vertical supports are shown on Drawings 11P1SU_ , 11P2SU_ , 11P3SU_ , 11P5SU_ and 11P6SU_ The vertical loads on each PF coil are transmitted to the TF coil cases. Load transmission is through flexible plates (for the PF2 and PF5 as shown in Drawing 11P2SU_ and for PF3 and PF4 as in Drawing 11P3SU_ ) or sliding supports (for the PF1 and PF6 coils where space is limited) with fibreslip, Drawing 11P6SU_ , to allow the radial expansion of the PF coils. The vertical and lateral loads on each PF coil, plus bending moments (due to the coil loads caused by the residual toroidal field ripple) are transmitted through these plates to the TF coil cases. In addition, bending moments due to the TF coil tilting motion are transmitted from the TF coils to the PF coils through these plates. The supports for PF2-PF5 consist of a set of 18 clamps while PF1 and PF6 have only 9 clamps. At each PF coil support, the winding pack is clamped between a pair of clamping plates linked by tie-rods. The clamping plates and tie rods can be used for lifting/handling purposes during PF coil fabrication and installation and they are also used in some of the PF coils to support the coil terminals, outer protection cases and break boxes. The clamping plates have been designed to minimize stresses in the winding pack insulation so the toroidal edge of the clamping plate is provided with a triangular shaped extension. This extension provides a soft ending to the clamping plate and avoids bearing stress concentration on the ground insulation of the winding pack. The clamping plates are also curved in the radial direction, with the plate curved away from the inner and outer edges of the coil. The avoids stress concentrations at the sides, on the coil ground insulation. For the PF2 to PF5 coils, flexible plates provide the link between one of the clamping plate and the support posts which are welded to TF coil case, Drawings 11P2SU_ and 11P3SU_ These are two flexible plates on each side of the coil which are welded at one end to one of the clamping plates and at the other to flanges which are bolted to the support plate. The support plates are placed on the side of the coil nearest the machine equatorial plane so that the vertical forces, mostly directed towards this, compress the attachments. The P3 and P4 coils are linked to each other by a strut connecting the support posts of each coil. Each strut is linked to the adjacent TF coil just below the P3 coil by a circular dowel. This supports the resultant vertical forces but allows free out of plane rotation of the TF coils, Drawing 11P3SU_ For the PF1 and PF6 coils, sliding supports have been provided due to space limitations. This is because the gap between the TF coil case and the PF coil winding is too small for a clamping plate of adequate bending stiffness. When the vertical loads are directed towards the TF coils, sliding occurs between the clamping plate and the TF coil case. When vertical loads are directed away from the TF coils, sliding occurs between the other clamping plate and a rigid support plate which, together with support posts, form a frame around the Design Description Document 11 Magnet Page 97

99 winding. For these sliding interfaces, a low friction material ( fibreslip ) is placed directly between the clamping plates and the TF coil case support pad. At the attachment between the PF coil supports and the TF coil support posts, shims allow to adjust the height and keys allow some limited horizontal adjustment of the PF coil position. This should allow to achieve the tight geometrical tolerances required for the PF coils. Table summarizes the main sizes of the PF coil flexible plates. PF Coil Table PF Coil Clamp Geometrical Characteristics No. of Flexible Plates [1] Thickness of each Flexible Plate (mm) Flexible Plate Free Length [2] (mm) Flexible Plate Toroidal Width [2] (mm) PF ,000 PF ,000 PF ,000 PF ,000 [1] This is the number of flexible plates in one flexible group, to be multiplied by 2 for the two sides of the winding pack, and then multiplied by 18. [2] Free length as per double built-in end beam The direct loads to be experienced by the supports are summarised in Table These are considered also to include disruption loads (see also section ). Table Normal Operating Conditions for PF Coil Supports Vertical Force : Coil + : away from equator, - : towards to equator, (MN) PF1 +15/-141 PF2 +0/-17 PF3 +0/-71 PF4 +0/-62 PF5 +0/-95 PF6 +115/-118 A moment is applied to the supports as a result of the ripple field (with radial and vertical components) on the poloidal coils arising from the toroidal field coils. This is given in detail in section 2.2. It produces a 360 degree sine wave form of field in both radial and vertical directions, which is zero at each TF coil (and so produces no net force component). The component is largest on the P2 - P5 coils, reaching about 0.5T in the radial direction on P4 and slightly over 0.5T in the vertical direction on P2 and P5. The corresponding vertical force reaches about 4MN/m on p4, 2.5MN/m on P2 and 5MN/m on P5. The displacements and rotations of coils and supports are given in Tables and 4. The PF coil displacement are estimated based on a uniform hoop stress. The support movement at TF on in the toroidal direction is zero, in the vertical the same as EOB. Design Description Document 11 Magnet Page 98

100 Table PF Coil Supports and PF Coil Displacements in mm Coil Cooldown TF On EOB values for supports, PF maximum during scenarios Support radial Support vertical PF Coil radial** Support radial PF Coil radial Support radial PF Coil radial Support Toroidal Support Vertical P P P P4* P P *the P4 is attached to the same support as the P3: displacements are for the dowel on the TF coil ** thermal contraction on cooldown for PF assumed to be the same as the TF Table PF Coil Support Angular Rotations (in degrees about the 3 axes) Coil TF on Initial Magnetisation End of Burn radial toroidal vertical radial toroidal vertical radial toroidal vertical P P P P4* P P *the P4 is attached to the same support as the P3: rotations are for the dowel on the TF coil Evaluations of the PF coil supports are given in detail in section 2.2 and the associated annexes. The stresses from the extra bending loads due to the ripple field have not yet been considered but are expected to be acceptable. The PF coils may be positionally unstable under some operating conditions (i.e. there is a net radial force on the coil due to a misalignment of the coil axis with the other PF and CS coils), and this force acts to increase the misalignment. These forces must be reacted by the PF coil supports. Preliminary values are given in section Correction Coil Supports The correction coils are supported by clamps directly on to the TF coil cases, see Drawings 11BC and 11BC (for the bottom correction coils), 11SCC1_ to 11SCC1_ (side correction coils) and 11TCC1_ (top correction coils). Each support consists of two brackets clamped across the CC case by bolts, with an L flange and bolts attaching it to the TF coil case and structures. There is an insulating cover (G10) on the outside of the case underneath the brackets. Each of the lower correction coils has 10 supports (Drawing 11BC ), the side have 15 each (Drawing 11SCC1_ ) and the top 10 (Drawing 11TCC1_ ). Design Description Document 11 Magnet Page 99

101 Each correction coil has a leg in the toroidal direction that bridges the gaps between 3 toroidal coils, and there is a clamp on the leg at each TF coil. The loads on the supports are caused by the CC magnetic forces and the movement of the TF coils, especially in the out of plane (toroidal) direction. Magnetic loads are complex due to the alternating nature of the fields along the CC legs, and the local distribution dominates the clamp loads rather than the overall total on the coil. They are given for the lower coils in section 2.2 Annex 19 and for the side coils in 'Structural Analysis for Side Correction Coils, June , JAERI superconducting magnet laboratory report (although for a coil current of 200kA). The movement of the TF coils is also specified in these analyses but can also be deduced from the movement of the PF coil supports as defined in Table and 4. The top CC corresponds to the PF1 support, the side to PF3 to PF5 and the lower to PF6. The loads on the clamp bolts to the TF cases (and the case stresses, see section ) are dominated by the TF coil displacements, which cause bending as well as direct stresses Central Solenoid The CS consists of a stack of 6 independently operated circular coil modules, as shown on Drawing 11CS The energy stored in the coils during normal operation is shown on Figure The associated inductances (also with the PF coils) are given in Table The force variations on the coils are defined together with those on the PF coils in section Table Mutual Inductance Matrix for One Turn (in μh) for CS CS3L CS2L CS1L CS1U CS2U CS3U CS3L 2,601E+00 CS2L 8,079E-01 2,601E+00 CS1L 1,765E-01 8,079E-01 2,601E+00 CS1U 6,065E-02 1,765E-01 8,079E-01 2,601E+00 CS2U 2,703E-02 6,065E-02 1,765E-01 8,079E-01 2,601E+00 CS3U 1,420E-02 2,703E-02 6,065E-02 1,765E-01 8,079E-01 2,601E+00 PF1 3,768E-02 6,100E-02 1,067E-01 2,053E-01 4,393E-01 9,810E-01 PF2 1,343E-01 1,945E-01 2,854E-01 4,148E-01 5,715E-01 6,973E-01 PF3 2,663E-01 3,390E-01 4,143E-01 4,747E-01 4,990E-01 4,769E-01 PF4 4,541E-01 4,957E-01 4,926E-01 4,462E-01 3,748E-01 2,989E-01 PF5 6,823E-01 5,537E-01 4,010E-01 2,765E-01 1,893E-01 1,313E-01 PF6 9,575E-01 4,599E-01 2,248E-01 1,198E-01 6,960E-02 4,342E-02 N. turns Design Description Document 11 Magnet Page 100

102 Figure Magnetic Energy Variation (in GJ) in CS and PF Coils CS Conductor The CS conductor is a Nb 3 Sn cable-in-conduit, similar to that used for the TF coil, except for the substage wrap. Due to the very high coupling losses at plasma initiation, this is increased to 80% coverage. The conductor current is chosen to be in the range 40-45kA as a compromise between structural issues (the larger the conductor, the larger the stress concentration factors on the jacket) and cable current density (the larger the current, the lower the amount of copper required for thermal protection). The superconductor for the CS is shown on Drawings 11CSU1_ and 11CSU1_ and the parameters are summarised in Table The steel square jacket, similar in size to that used for the CS model coil, is provided as a single square section. The material is JK2LB, a 316LN type of steel modified with manganese to increase carbon solubility to avoid precipitation during the Nb3Sn heat treatment. The composition is shown in Table The material has a lower thermal contraction than 316LN between RT and 4K which makes precompression of the CS stack easier to achieve. The jacket is first hot extruded and then (if necessary) cold drawn to achieve the final dimensions. The mass of the billet used for the extrusion is limited by the extrusion press capacity to about 150kg and the target length of the extrusion is 8 to 10m to reduce the number of butt welds required. This length is achievable for the CS jacket which has a structural cross section of 1,560mm 2 and would require a billet mass of about 125kg. Design Description Document 11 Magnet Page 101

103 However the extrusion length may also be limited by the tolerance requirements on the position of the central hole in the section, as the central forming die position tends to move off centre during the extrusion. Manufacturing optimisation is required to achieve the best balance between the number of butt welds and the acceptable tolerances. A previous CS jacket design used a smaller extruded section with a co-wound reinforcing strip. The main advantage of this design was to reduce the cross sectional area of the butt welds joining the jacket lengths. However, the co-wound strip complicated the winding process and either required a longitudinal weld to the square jacket or, if not welded, requires additional insulation which reduces the current density. The single square section was selected. Due to the short conductor length between an helium inlet and the outlet (about 150m), the central cooling channel can be reduced in size as the helium pressure drop is low. For the 7mm central channel given in the table, a mass flow rate of 8g/s/channel requires a pressure drop of about 0.05MPa. The conductor has two design conditions. At Initial Premagnetization (IM), the CS modules are all energised with the same current to a peak field of 13.0T. This field is achieved with a conductor current of 40.0kA. At End of Burn (EOB), the current is concentrated in the central CS modules and the plasma and PF coils act to reduce the overall field. The result is that the peak field is set lower, at 12.4T, but also that a higher current, 45kA, is required to reach this. The conductor design is a compromise between these two operating conditions. The discharge time constant of 11.5s is equivalent (for the I 2 t integral) to a quench detection and action time of 2s, then a current discharge with a time constant of 7.5s. Table CS Conductor Number of turns per module 549 Jacket type Square steel Type of strand Nb 3 Sn Operating current (ka) IM/EOB 40.0/45.0 Nominal peak field (T) IM/EOB 13.0/12.4 Operating temperature (K) 4.7 Operating strain (%) Equivalent discharge time constant (s) 11.5 n at operating condition 7 Tcs (Current sharing temperature) 13.0T 5.4 Iop/Ic (Operating current/critical current) IM 0.80 Cable diameter (mm) 32.6 Central spiral outer x inner diameter (mm) 9 x 7 Conductor outer dimensions (mm) 49 x 49 Jacket material JK2LB SC strand diameter (mm) 0.83 SC strand cu : non-cu 1.0 Cabling pattern (2s/c+1Cu)x3x4x4x6 SC strand number 576 Wrap coverage on final substage (0.05mm SS) 80% Local void fraction (%) in strand bundle 33.2 SC strand weight/m of conductor (kg/m) 2.95 (*) (*) cosθ =0.95, d strand 9 kg/dm 3 Design Description Document 11 Magnet Page 102

104 Spec. Target Table Specification for CS Jacket Material (JK2LB) C Si Mn P S Cr Ni Mo N B with fracture toughness > 130 MPam 0.5 and 0.2% yield strength > 800MPa at 4.2K, after aging The modules are wound as hexa-pancakes (6 pancakes with a single conductor length) and quad-pancakes (4 pancakes with a single conductor length). Helium inlets are at the crossover regions on the inner bore between each double pancake and outlets are at the cross-over regions and joints on the outside. The high field region is therefore cooled by the coldest helium CS Winding Pack Design and Manufacture The CS stack consists of 6 electrically independent modules, arranged in a vertical stack as shown in Tables and 5 and Drawing 11CS It is free standing and supports all the magnetic loads through structural material (the conductor jacket) within the winding. The main loads are the magnetic hoop force, which creates hoop tension in the structural material, and the vertical inward pressure from the outer modules at initial magnetisation, which creates vertical compression. There may be a small resultant vertical force component in some scenario conditions, which is reacted through the vertical support structure. Forces on the coil modules are given in the section on the PF coils in Tables and 3. The resultant forces on the stack are given in section Table CS Winding Configuration for Each Module Pancake Type Conductor Length (m) Conductors in Hand Number of Units Number of Turns Nr x Nz Total Number of Turns Hexa x6 82 2/3 Quadra x4 54 1/3 Total in Module Table Location and Size of CS at 4K CS Module Position of Coil Centre Coil Size without Ground Insulation Rc (m) Zc (m) dr (m) dz (m) CS3U CS2U CS1U CS1L CS2L CS3L Design Description Document 11 Magnet Page 103

105 The modules are each made up of 6 hexa-pancakes and 1 quad-pancake (see Table ), as shown on Drawing 11CS As compared to the conventional double pancake winding arrangement, the hexa and quad-pancake winding arrangement minimizes the number of joints at the outer diameter, and therefore, it reduces the complication associated with the joint configuration which includes the joint itself with its mechanical clamps, the helium pipes and the structure which carry the mechanical hoop tension. All these components must fit in a narrow space between the CS and TF coils. On the other hand, this winding configuration requires more complicated manufacturing processes and tools, in particular for the winding of a conductor when the winding starts from the outer diameter. Industrial studies suggest that the hexa or quad-pancake winding arrangement will not imply serious technical difficulties. The conductor is wound at constant radius with a turn to turn transition region over a toroidal width of 40 o within each pancake, Drawing 11CS The pancake to pancake transitions at the inner and outer bore are similarly confined to a 40 o region. The location of the transition regions is rotated for each pancake so that the coil modules have toroidal uniform mechanical properties. The full orientation of the pancake transitions within a CS module is shown in Drawing 11CS Insulation is applied after the conductor is reacted, by unspringing the pancakes vertically, Drawing 11CSU1_ The turn insulation has a nominal thickness of 1.0 mm around each conductor (2mm total separation) and is a hybrid system composed typically of two 50% overlapped layers of interleaved glass and polyimide film and one 50% overlapped dry glass layer outside, Drawings 11CS and 11CS Additional dry glass layers will be inserted between pancakes (1.0mm thick) and between turns (0.5mm thick) to help absorb tolerance effects (such as conductor keystoning, twisting and deviations in flatness and bending radius). The butt joints between hexa and/or quad-pancakes are also made at the stage of application of the turn insulation. After making the joint, the structural element which carries the conductor hoop load is also installed. Good access around the joint area is provided when the conductors ends are unsprung vertically into a helical shape, Drawing 11CSU1_ There is an additional insulating plate inserted between the hexapancake units as they are stacked together, Drawings 11CSU1_ and 11CSU1_ This sheet is 2.1mm thick, of GFRE, and includes the quench detection pick-up coils. Each sheet contains 4 separate coils (2 are redundant, one is used for the hexapancake unit above and one for the unit below), with each coil containing about the same number of turns as the hexapancake module. The 4 coils are co-wound with a 0.1x0.6mm conducting strip. Details of the quench detection method are given in At the ends of the module winding, a structural element is required to carry the conductor hoop load. This element consists of a full dummy turn (equivalent to a jacket without cable) which is welded to the end of the conductor jacket and closed on itself by means of insulated bolts and shear keys, as shown in Drawing 11CSU1_ The ground insulation has a total thickness of 10mm with the external 2mm having a essentially a mechanical protection function. The insulation is composed of a sandwich of Design Description Document 11 Magnet Page 104

106 0.25mm thick glass, 0.05mm thick polyimide film (barrier), and 0.25mm thick glass wrapped to a total of nine 50% overlapped layers. The ground wrap extends to the leads and joints. After application of the ground insulation, the whole module is impregnated with epoxy resin by a vacuum pressure impregnation (VPI) process and cured. The 6 CS modules are identical up to this point. Before assembly into the stack, an interface plate, made of high density GFRE, is bonded onto one surface of each module, Drawings 11CS to 11CS These interface plates are individual to each coil module because they contain grooves for cooling pipes (see below) and radial keys (see section ) that have to match the toroidal locations of the helium outlets. The upper three modules are placed in the stack so that the top two have a rotation of 120 degrees about the vertical axis relative to the coil below, Drawing 11CSHH_ The lower three coils are similarly rotated but are placed 'upside down' relative to the upper three. The terminal leads, running outside the coils (up for the top 3 modules and down for the lower 3) are therefore distributed uniformly around the outer perimeter (between the vertical support flanges), Drawing 11CS , to help reduce the stray field effects from the leads, Drawing 11CS The helium supplies to the modules run in the inner bore of the stack (there is insufficient space outside) between the modules. The outlets from each cooling path are collected together and pass into the inner bore through 6 pairs of grooves cut in the interface plates, to the insulating breaks and manifolds in the inner bore, Drawings 11CSHH_ and 11CSHH_ (details of the insulating breaks are given in section ). Because of the vertical pressure on the module, the cooling pipes with long vertical lengths on the outside of the module require bends to allow flexing, to prevent stress concentrations at the grooves or helium outlets from the conductor. The outer surface of the coil is crowded and it is important to restrict the vertical support flanges to the minimum possible area. Drawing 11CSHH_ shows the overall arrangement. Within a vertical 'slot', it is not possible for a busbar extension to pass above a set of helium outlets, due to the radial space requirements, which determines the relative toroidal orientation of the modules CS Joints, Terminals and Helium Inlets/Outlets The CS joints are located at the coil outer diameter and are embedded within the winding pack, as shown on Drawings 11CSU1_ and 11CSU1_ The coil terminal joints are placed above each other at the top and bottom pancakes to align the busbars, Drawing 11CSHH_ The conductor exits the pancake and is turned through 90 o before the joint, into the vertical direction. Two types of joints are used: the overlap type for the top and bottom terminal connections at top and bottom of the CS stack, see section 1.2.6, and the butt type between pancakes and between pancakes and busbar extensions at the level of the individual modules. The pancake to pancake butt joint is an integral part of the winding pack and is subjected to the jacket operating stresses. The joint housing transmits the hoop force between the two conductors. The module terminals are placed close to the coil and there is then an extension lead (formed of Nb 3 Sn conductor, similar to the CS conductor) extending to top or bottom of the stack, where there is a connection to the superconducting busbars. The leads are paired for most of their length (to reduce net forces) and cooled in series with the end pancake of the modules. Supports to the CS vertical precompression structure must allow relative vertical movement of the leads and the structure, due to the magnetic forces. Design Description Document 11 Magnet Page 105

107 The support system for the busbar extensions is shown in Drawings 11CS , 11CS and 11CS A series of clamp plates (vertically segmented into 20-30cm units) is fitted around the insulation of the feeder extensions and bolted together. On each side they contain tongues that fit into vertical slots formed on the edge of the vertical tie plates on each side of the feeder groove. These slots are closed by the insertion of welded edge pieces once the clamps are in place, Drawing 11CS to 07. The scheme allows free vertical motion between coil surface, vertical tie flanges and the busbars in operation. Special fit pieces have to be made to cover the varying relative position of the two feeders as they come together at the top/bottom of the module, at the butt joint. It is possible that optimisation of the shape of the clamps, with a fit of the back part into one groove and the front part into the opposite groove followed by clamping, might allow a fully closed pre-machined groove to be used in the vertical flanges, eliminating the welded closure lengths. The overlap type of joint would require more radial space than the butt joint and is not compatible with the present space allocation, either for the inter hexapancake joints or the coil to busbar extension joints. The overlap joint is however easily dismountable and is therefore used at the joint between the busbar extensions and the NbTi feeders. This joint uses a cable compacted inside a circular copper tube. The copper tube is connected to the steel jacket through a welded steel-monel-copper end piece so the final closure weld is monel to steel. The contact shoes that link the two tubes (in the form of a double U) are divided along the length of the joint to reduce eddy current losses. Further details are given in sections and 2.1. In forming the butt joint, the jacket is removed from a short (20cm) length of cable which is then highly compacted into a copper tube (to reduce the void fraction to about 20%). Helium access to the central channel is improved by the removal of the wraps before the start of the compacted section. After heat treatment, the reacted compacted cable end is cut flat as shown on Drawing 11CSU1_ Two cables are then pressed onto each side of a thin copper separator plate (1mm) and sintered under pressure to create the joint. The successive steps are shown in Drawings 11CSU1_ , 11CSU1_ and 11CSU1_ , with more details given in section 2.1. The joint containment box, helium outlets for the cables on each side and internal support are then built up round the cable joint. There are three helium inlets for each hexa-pancake and two for each quad-pancake, shown in Drawings 11CSU1_ and 11CSU1_ These inlets are located at the CS inner diameter, at the cross-over regions between pancakes, where tensile stresses are highest. The helium inlet region requires, therefore, a local reinforcement to allow the opening in the conductor jacket without excessive stress intensification. The inlet must also provide a good distribution of helium in the 6 sub-cables of the conductor. The vertical transition between pancakes is extended over about 1m to avoid the creation of bending stresses. An elongated narrow slot, about 0.5m long, is then machined into the jacket for the helium inlet, Drawing 11CSU1_ The outer cable wrap and the inter-petal wraps exposed at the bottom of the slot are also removed (by hand to avoid strand damage). This slot is long enough to allow direct access for the helium to several sub-cables. A structural reinforcement, which carries the helium inlet pipe and provides a tapered channel extending about 0.25m in each direction from the inlet along the conductor, is then welded above the slot. Finite element analysis confirms that this layout has low stress concentration factors Design Description Document 11 Magnet Page 106

108 (see section 2.2), considering also the partial penetration of the closure weld of the slot (to avoid damage to strands). R&D is still required to verify the fatigue behaviour of the helium inlet region under cyclic strain and to establish reliable inspection techniques for the welds (specially for the butt welds of the reference design). Helium outlets are at the CS outer diameter, Drawing 11CSU1_ Two of these outlets are at the conductor ends, at the joints between hexa-pancakes. The other outlets (two outlets for hexa-pancakes and only one for quad-pancakes) are at the cross-over regions between pancakes. The design of these outlets is simpler than that of the inlets as the requirements on stress and flow distribution are less severe. A similar arrangement of slots is used but the radii on the outlet pipe support can be smaller, making it more radially compact CS Preload Structure The CS coil stack is self supporting against the coil radial forces and most of the vertical forces, with the support to the TF reacting only the weight and net vertical components resulting from up-down asymmetry of the oloidal field configuration. The preload structure is provided to contain forces internal to the CS stack. The field curvature at the ends of the CS stack creates vertical forces on the modules. At initial magnetization (IM), these forces are towards the centre of the stack, whereas at some other equilibrium configurations during the scenario, the end modules carry opposite currents to the central ones and are repelled. This means that a vertical support structure is required. This structure applies axial pre-compression to the coil stack so that the modules remain in contact during all operating conditions, although local gaps between the coils are acceptable. The coil conductor uses a modified 316LN steel that has a significantly lower thermal contraction coefficient from RT to 4K compared to 316LN (approximately 0.1% less). The precompression structure is fabricated in 316LN steel and any initial precompression is therefore enhanced on cooldown. In addition to the separating vertical loads on the end modules, there is also a significant difference in radial loads, and therefore in hoop strain, between the end modules and the adjacent ones. If no sliding is allowed at the interface between modules, higher vertical tensile stresses appear within the end modules. These internal stresses which may, in the long term, lead to delamination and damage to the insulation should be essentially eliminated. This can be achieved by providing a vertical compression larger than these stresses. Allowing sliding at the module interfaces reduces the necessary pre-compression. A fibreslip sliding interface is expected to be able to achieve a friction factor of less than 0.2 and is adopted in the reference design. The fibreslip is applied to the outside surface of the module spacers attached to one end of each module, Drawing 11CS to 04. Under some operating conditions, it may be possible for some of the solenoid modules to show lateral position instability i.e. while a perfect vertical alignment of the magnetic axes of the individual modules gives zero force in the radial direction, a small departure from this condition produces a positive radial force, which tends to increase the initial misalignment. Limited investigations of this condition have been done (by imposing a 2.5mm radial axis offset on each coil in turn for 3 points in the reference scenario) and the instability force gradients (i.e. radial force per metre of axis offset for the coil concerned) are defined in Table (this covers also the PF lateral instabilities). Assuming realistic values of axis misalignment (about 2-3mm, section 1.1.6), the radial force must be reacted by friction/keys Design Description Document 11 Magnet Page 107

109 at the module interfaces created by the vertical precompression structure. A full instability assessment is required for all possible operating scenarios to establish the maximum values. Table Lateral Instability Force Gradients on CS and PF Coils (blank entry indicates no laterial instability Coil SOD MN/m SOF MN/m EOC MN/m CS3U CS2U CS1U CS1L CS2L CS3L 42.3E 18.6 PF PF PF3 PF4 PF PF6 The CS Pre-load Structure consists of the lower flange, the upper flange, a set of 12 tieplates, buffer plates, wedges and connecting bolts (see Drawing 11CS and 11CS for some details of the tie plate wedges and buffer zone). The flanges are split into 9 sectors linked by electrically insulated bolted joints, as shown on Drawing 11CSHH_ (details on 11CS ) to reduce AC losses during pulsing of the machine. The induced voltage across the insulated joints is low (<10V). The tie plate region needs to have at least 30% open space in the toroidal direction for the joints, current leads and helium pipe arrangement. Although not shown on the drawings, the end flanges and vertical tie plates require cooling pipes. There will be 9 branches which will run over each sector of the lower flanges, up the outside of each tie plate (in a groove in the outer surface) and over the upper flange segments. The 9 pipes are contained in one of the lower structural feeders (together with some of the TF structure circuits) with a manifold (with insulating breaks) to supply each of the 9 branches. At the top the 9 circuits remain separate and exit through one of the upper structural feeders (see also section ). The CS assembly is hung from the top of the TF coils through its pre-load structure, as shown on Drawings 11CS There are flexible plates at the top supports to allow relative radial displacements between the CS and the TF coils. Due to rotation of the top flange due to uneven radial forces on the coil modules, these flexible plates need to be positioned near the mid-radius point of the flange. There is a bottom spring loaded centring system to provide a locating mechanism and support against dynamic net horizontal forces, as shown on Drawing 11CS This connection allows relative vertical displacements and toroidal rotation between the CS and the TF coils. The resultant vertical forces on the assembly in normal operation (excluding disruptions) are +39MN to -40MN (in the reference scenario) and +18MN to -26MN in the scenario flexibility cases (positive being upwards from L to U ). The preload structure is designed so that it can restrain the maximum vertical separating load of 75MN acting on the end modules of the stack as well as the maximum inward compression when all coils push towards the axis, without overall gaps developing between coils or coils and end flanges. To obtain uniform axial compression, tie-plates running axially along the CS are provided at both inner and outer diameters and connect to rigid flanges at top and bottom. The required axial tension in the structure is achieved partly by pre-tensioning at room temperature and partly during cooldown. The variable hoop forces Design Description Document 11 Magnet Page 108

110 on the modules tend to create a vertical bending of the solenoid stack which can lead to loacal areas of insulation tension and gaps. Gaps are defined interfaces (i.e. between the coils) are acceptable, but local tension within the insulation has to be eliminated. Pre-tensioning at room temperature of the CS modules is accomplished through a system of wedges placed under the upper flanges (the lower flange region cannot be used as the wedges support the gravity weight of the coil and cannot be moved by the screws). The tieplates are heated and the wedges are pulled into the resulting gap by screws, providing a defined extension of the tie plates once they cool down again to room temperature. A temperature of 90C provides the required pre-compression. Details of the analysis are given in Pre-compression Requirements for the ITER Central Solenoid (ITER_D_22FVN3 v1.0). The preload corresponds to an increase in vertical height of the wedges of about 13mmAs explained in section , the combination of JK2LB as a jacket material and stainless steel for the tie-plates provides the required differential contraction. A set of 6 radial keys is provided at each of the interfaces between modules and between modules and the upper and lower flanges of the pre-load structure, Drawing 11CS These keys prevent ratcheting of the relative positions of the modules, over successive pulses, which may eventually lead to damage of the cooling tubes and busbars that run on the outer surface of the coils. These keys also resist any horizontal load between CS modules due, for example, to seismic events. Materials used in the pre-load structure are specified in Table The 316LN is grade 2 with a 4K yield stress of 900Mpa (see also section ). Table Materials Used in the Preload Structure Material Specification Tie Plates, Key block, End Flanges Buffer zone 316LN JK2LB PF Coils The 6 PF coils are placed around the outboard legs of the TF coils as shown on Drawing 11P1WP_ Coil dimensions are given in Table (at 4K). The operating forces acting on these coils are given in Tables a-b (reference scenario) and (including reference and extra scenarios). The PF coils are self supporting as regards the radial magnetic loads and the vertical loads transmitted to the TF coil cases by supports designed to allow a free radial expansion of the coils (see section ). Under some load conditions the PF coils can be laterally unstable, this is discussed and quantified in section The scenario used to derive these forces is one of the 15MA options, without heating during current ramp-up. For the scenario flexibility table, the extreme values have been selected. The coil magnetic energy is shown in the section on the CS (Figure 1.2-2). The PF inductance matrix is given in Table Design Description Document 11 Magnet Page 109

111 Table Location and Size of PF Coils at 4K PF Coil Coil Size without Ground Position of Coil Centre Insulation & Protector Rc (m) Zc (m) ΔR (m) ΔZ (m) PF PF PF PF PF PF Table a Radial Forces on PF and CS Coils (Total on Perimeter) in MN Time (sec) IM XPF SOF SOB EOB EOC Ip (MA) li(3) CS3U CS2U CS1U CS1L CS2L CS3L PF PF PF PF PF PF Table b Vertical Forces on PF and CS Coils (total on perimeter) in MN positive being upwards from L to U Time (sec) IM XPF SOF SOB EOB EOC CS3U CS2U CS1U CS1L CS2L Design Description Document 11 Magnet Page 110

112 CS3L PF PF PF PF PF PF Table Range of Vertical and Radial Forces on the CS and PF Coil for Scenario Flexibility in MN positive being upwards from L to U Force Radial Vertical coil Fmin Fmax Fmin Fmax CS3U CS2U CS1U CS1L CS2L CS3L PF PF PF PF PF PF Table PF Coil Single Turn Inductance Matrix (μh) PF1 PF2 PF3 PF4 PF5 PF6 PF1 1,145E+01 PF2 3,971E+00 3,638E+01 PF3 2,193E+00 1,126E+01 5,189E+01 PF4 1,165E+00 5,254E+00 1,429E+01 5,220E+01 PF5 4,534E-01 1,898E+00 4,561E+00 9,838E+00 3,319E+01 PF6 1,362E-01 5,390E-01 1,210E+00 2,364E+00 4,799E+00 1,129E+01 N. turns The field on the PF coils is given in Tables ,8,10 and 12. Further detail on the location of the maximum field and also the db/dt variations for reference scenario 2 are given in the report ITER_D_24MVAJ v1.0.the following tables showing the maximum db/dt on the P6 coil are taken from this report. The field at the maximum db/dt can be found approximately and conservatively by scaling the maximum field by the current ratio between the maximum db/dt and the maximum B. More accurate values are given in the report Design Description Document 11 Magnet Page 111

113 Table Maximum Field and Rate of Change of Field on the PF Coils (Scenario2) PF1 Panc Nr Turn Nr Time (s) Current (ka) Bn peak (T) dbn/dt max (T/s) < t < < I < dbn/dt min (T/s) < t < > I > PF2 Panc Nr Turn Nr Time (s) Current (ka) Bn peak (T) dbn/dt max (T/s) < t < > I > -10 dbn/dt min (T/s) < t < > I > 2.55 PF3 Panc Nr Turn Nr Time (s) Current (ka) Bn peak (T) dbn/dt max (T/s) < t < > I > dbn/dt min(t/s) < t < > I > 1.78 PF4 Panc Nr Turn Nr Time (s) Current (ka) Bn peak (T) dbn/dt max (T/s) < t < > I > dbn/dt min (T/s) < t < > I > 1.19 PF5 Panc Nr Turn Nr Time (s) Current (ka) Bn peak (T) dbn/dt max (T/s) < t < > I > dbn/dt min (T/s) < t < > I > 0.37 PF6 Panc Nr Turn Nr Time (s) Current (ka) Bn peak (T) dbn/dt max (T/s) < t < < I < dbn/dt min (T/s) < t < > I > PF Conductor The PF coils will use NbTi superconductor, cooled by supercritical helium. This gives a substantial cost saving compared to Nb 3 Sn and the elimination of a reaction heat treatment greatly simplifies the insulation of such large diameter coils. The cable configuration is similar to that used with Nb 3 Sn with 6 sub-cables arranged around a central cooling space. The conductors have a heavy walled stainless steel jacket and an isometric view is shown in Drawing 11P1WP_ The cable is placed in the jacket by a pull-through, roll down procedure, as with the TF and CS coils. As the general stress levels in the PF coils are lower than the CS, it would be possible to fabricate the (oversize) jacket from U channels, as an alternative to seamless circle-in-square tubes. Another alternative is to jacket the cable with a thin circular jacket (similar to the TF conductor) and then apply two U sections around this by welding after the circular jacket has been compacted onto the cable. The external shape of these three conductor options is the same and the choice has no impact on the PF coil manufacture, except for minor considerations concerning the preparation of helium inlet pipes and conductor joints/terminations. The final selection can be made on the basis of cost, after manufacturing studies. Design Description Document 11 Magnet Page 112

114 Non-uniform current distribution can be a more sensitive issue with NbTi cables than with Nb 3 Sn cables because of the lower value of the critical temperature T c. To ensure a uniform current distribution in each sub-cable while maintaining AC losses at acceptable level, the transverse resistivity within the cable must be controlled. Strand with bare copper surfaces is unacceptable as the transverse resistance is dependent on the oxidisation state and therefore a coating (nickel is the reference candidate) is used on the strand. Chromium coating is not used as it tends to give relatively high values (it is used for Nb 3 Sn as it can go through the heat treatment without contaminating the stabilising copper). With nickel, transverse resistances between strands of the order of 0.1microOhm.m are expected to be achieved (between strands not separated by the final substage wrap). Coupling currents are controlled by the wrap on each sub-cable (0.05mm, 50% coverage, stainless steel) to limit the coupling time constant (nτ) to 50ms and the joint layout must provide a uniform contact to each of these sub-cables. The losses in the coils are, at a maximum, about 25% due to coupling, so if necessary a significantly higher coupling time constant could be tolerated (the sub cable wraps could be removed) if supported by test data. The cable has an outer 0.08mm wrap, 50% overlapped, for protection during the manufacturing. The conductor current is selected as 45kA (up to 52kA in backup mode) because of the similarity with the CS conductor and the R&D database (note that in normal operation the PF2 requires only 41kA, see footnote to Table ). No advantage has been found in using higher current conductors as the copper fraction is limited by stability rather than thermal protection. Three different conductor designs corresponding to three field values have been selected for the PF conductor design. For the PF2, 3 and 4 coils, the field is up to 4T, for the PF5 coil, up to 5T and for the PF1 and 6 coils, up to 6T (Table ). The thermal discharge time constant is equivalent (in terms of the I 2 t integral) to a detection and implementation time of 2s followed by a current discharge with a time constant of 14s. As explained in section , the PF conductors are designed with the capability to operate in the backup mode, at a higher current, to compensate for the loss of one doublepancake, while still maintaining the same total current capacity in each coil. In such a case, not only the conductor current increases but also the peak field. Both normal mode and backup mode operations can be achieved within the field limits for the two lower grades, while the operating field upper limit of the high grade is increased up to 6.4 T in backup mode and requires an operating temperature of 4.7K instead of 5K (sub-cooling is required with the coil He inlet at 4.4K instead of 4.7K). The three different conductor designs able to accommodate the various combinations of operating current and field in both normal and backup modes are shown in Table The conductor parameters are shown in Table and the conductor configuration in Drawing 11P1WP_ The selected operating temperatures include a 0.3K temperature increase (due mainly to AC losses) within the coil up to the maximum field and an inlet temperature of <4.7K, allowing a cryoplant window of K for the PF coils. Table Three Types of PF Conductor Coils Design Current (ka), Peak Field (T), Operating Temperature (K) Normal mode Backup mode PF1 & PF ka, 6.0 T, 5.0K 52.0 ka, 6.4 T, 4.7K Design Description Document 11 Magnet Page 113

115 PF2, 3 & PF ka, 4.0 T, 5.0K 52.0 ka, 4.0 T, 5.0K PF ka, 5.0 T, 5.0K 52.0 ka, 5.0 T, 5.0K Table PF Conductors PF1 & 6 PF2, 3 & 4 PF5 Coolant normal/backup inlet 4.7K/4.4K inlet 4.7K inlet 4.7K Type of strand NbTi NbTi NbTi Operating current (ka) normal/backup 45/52 45*/52 45/52 Nominal peak field (T) normal/backup 6.0/ Operating temperature (K) normal / backup 5.0/ Equivalent discharge time constant (s) hot spot Tcs (Current sharing temperature) (K) 6.5/ / /6.51 normal/backup Iop/Ic (Operating current/critical current) 0.127/ / /0.305 normal/backup Cable diameter (mm) Central spiral outer x inner diameter (mm) 12x10 12x10 12x10 Conductor outer dimensions (mm) 53.8x x x51.9 Jacket material 316L 316L 316L SC strand diameter (mm) SC strand cu :non-cu Cabling pattern (+ is Cu core) 3x4x4x5x6 ((3x3x4+1)x4+1)x6 ((3x3x4+1)x5+1)x6 SC strand number Cu core 2/3/4 stage (mm) 0/0/0 0/1.8/3.5 0/1.2/2.7 Local void fraction (%) in strand bundle SC strand weight/m of conductor (kg/m) * although the P2 conductor is designed for 45kA, the operating current in normal conditions is 41kA. This is because this coil contains only 5 double pancakes and in back-up mode, with one DP lost, the current cannot exceed 52kA. Therefore, the normal operating current has to be limited to 41kA PF Winding Pack Design and Manufacture For PF1, PF2, PF5 and PF6, the winding pack consists of a stack of double pancakes enclosed in a common ground insulation wrap, as shown on Drawing 11P1WP_ By contrast, PF3 and PF4 have double pancakes with individual ground insulation and separator plates in between, as shown on Drawing 11P1WP_ The reasons for this design have been explained in section Design Description Document 11 Magnet Page 114

116 The first step of the winding pack manufacture is the winding of the double pancakes. This is to be done on a rotating winding table with two conductor feed lines since the double pancakes are wound two-in-hand, Drawing 11P The conductor is provided with double turn insulation as illustrated on Drawing 11P1WP_ This double insulation consists in an inner insulation layer, a thin metal screen and an outer insulation layer. The turn insulation consists of two insulation layers, each 1.5mm thick, with a thin 0.2mm metal screen in between. Each insulation layer includes two layers of half overlapped interleaved polyimide film and dry glass. The metal screen is sandwiched between layers of dry glass tapes and may require small perforations to facilitate epoxy penetration during vacuum impregnation. During the winding adjustable turn-to-turn and pancake-to-pancake insulation is added to compensate the conductor tolerances and control its position. After winding, the conductor terminations are installed and the joints internal to the double pancakes (between the two hands) are made. For PF1, 2, 5 and 6, the double pancakes are individually impregnated. During this process, the conductor ends, which carry the terminations to be used for the joints between double pancakes, must be kept unbonded to the winding pack over a length of typically 5m. This is to provide the required flexibility when the joints are made. These conductor ends are, therefore, provided with an anti-adhesive wrap before impregnation. This wrap is to be removed after impregnation. Each double pancake is then wrapped with a thin insulation layer, the double pancakes are stacked and the ground insulation with a compacted thickness of 8mm (uncompacted 9.9mm) is applied to the whole stack. The orientation of the pancakes to distribute the joints around the perimeter between the support clamps is shown in Drawings 11P3WP_ and 11P3WP_ for the P3 coil and 11P6WP_ and 11P6WP_ for P6. The ground insulation is composed of a sandwich of 0.25mm thick glass, 0.05mm thick polyimide film (barrier), and 0.25mm thick glass wrapped to a total of nine 50% overlapped layers. The ground wrap extends to the leads and joints and is vacuum pressure impregnated in a second impregnation step. In addition to the ground insulation, a 2mm thick protective cover made of pre-cured G10 glass epoxy composite is applied. The joints between double pancakes are made at this stage. The stack of double pancakes is then impregnated with epoxy resin and cured. Winding of the large PF coils is expected to be at the ITER site. Details of the layout of the fabrication line are shown in Drawing 11P For PF3 and 4, the manufacturing process is slightly different. The double pancakes are individually impregnated with an individual ground insulation. The outer surface of each double-pancake is covered with ground insulation with a thickness of 8 mm. During this impregnation, the conductor ends are kept unbonded to the winding packs as already described for the other coils. After impregnation, the double pancakes are stacked together with G10 separator plates. The joints between double pancakes are made at this stage. The stack of double pancakes is then wrapped with an outer layer of insulation and then impregnated with epoxy resin and cured. Details of the coil assembly are shown generally for P3 and/or P6, as examples. The other coils are similar. The following operations consist of installation of the prepared jumper leads (to be used to bridge pancakes in the event of coil failure), the installation of the cooling pipes and instrumentation cables, the completion Design Description Document 11 Magnet Page 115

117 of the coil terminals and helium headers, the assembly of the coil clamps (Drawing 11P3WP_ ) and finally the installation of the protection cover plates. The distribution of the jumper cables around the coil outer surface is shown in Drawings 11P3WP_ for coil P3 and 11P6WP_ for P6. The helium inlets are situated on the inner bore and the pipes are brought over to the outer surface to a header before running around the coil perimeter to a feeder connection point, Drawings 11P3WP_ and 11P3WP_ for P3 and 11P and 11P for P6. The insulating breaks, Drawing 1101WP_ , for the cooling lines are placed near the inlets to the coil. The instrumentation cables for the quench detection are connected to the conductor through a resistor (Rs in Drawing 11P ) on the high voltage side of the insulation breaks as shown in Drawing 11P (for the location on the coil surface) and 1101WP_ (for details of the connection and resistor). Other resistors are required for the short detection system using the conductor screens, which have yet to be located. After completion of the external pipework and cabling, the clamp plates (corresponding to each PF coil support point) are added, Drawing 11P2SU_ Finally the external cover plates are mounted. These are steel boxes, 5mm thick, that surround the coil between the clamp plates. They are assembled by bolting/welding and are attached to the top clamp plate, Drawing 11P , by insulated bolts (a direct connection may be used on one side to ensure better thermal contact for cooling). They ensure that the full outer surface of the coil facing the cryostat is at ground potential. As well as providing mechanical protection for the coil insulation, helium pipework and cables, in the case of PF3 and PF4, the cover plates also act as an intermediate thermal shield so as to reduce the heat load radiated by the labyrinths between the vacuum vessel thermal shield and the port thermal shield. If found necessary by more detailed assessment, some of the cover plates will be provided with cooling pipes for 4.5K helium, but at present they are cooled by conduction to the supports. Table summarizes the winding configuration, including the total conductor length and number of turns for each coil. All PF coils are wound with two conductor-in-hand. Table indicates current and voltage requirements. Table gives the conductor design values in the normal and the backup modes. PF Coil Conductor Length (m) Table PF coil winding configuration Conductor Unit Length (m) Conductors in Hand Number of Conductor Number of Turns Nr x Nz, Total PF x PF x PF x PF x PF x PF x Design Description Document 11 Magnet Page 116

118 Coil Requirement (MA) Table Current and voltage requirements Iop (A) Normal mode Actual Current Capacity (MA) Max. one turn Voltage (V) Max. Terminal voltage (kv) Winding Current Density at Iop (MA/m 2 ) PF PF PF PF PF PF Table Conductor design values DP s Capacity Normal Redundancy (MA) Iop (ka) Bop (T) Top (K) Iop (ka) Bop (T) Top (K) PF PF * PF PF PF PF * operating current is 41kA, see footnote to Table PF Joints & Cooling Pipes The general layout of coil PF3, used here as a typical example, including joints between double pancakes and feeders is shown on Drawing 11P3WP_ The joints, including lead in and lead out, have to be fitted between the coil clamps and toroidal space is therefore limited, Drawing 11P3WP_ (for P3) and 11P6WP_ (for P6). The conductors at the ends of each double pancake are brought out of the winding but remain in the toroidal direction. A structural element is required to transfer the operating hoop load on the conductor around the joint area. A tail on the conductor distributes the jacket load into the turn behind through shear in the turn insulation, Drawing 11P6WP_ and 11P6WP_ The stiffness of the tail is controlled by use of a hollow section to avoid stress concentrations, Drawing 11P6WP_ The joint contact surfaces to the next double pancake above or below are aligned parallel to the plane of the coil since this is the most convenient for manufacturing reasons. The joint type shown here is that of a copper tube with details shown in Drawing 11P6WP_ The copper tubes from the two conductors make contact through a double U saddle piece and are compressed into position by a series of bolted clamps along the coil. The copper-copper surface contact is improved by silvering and indium mesh (solder cannot be used because the thermal expansion damages the insulation of the pancakes). Details of the clamps are shown in Drawing 11P3WP_ Once the joint is closed, two half cylindrical shells are welded around it and to the conductors at each end. These shells provide a surface for the ground insulation and outer screens. Design Description Document 11 Magnet Page 117

119 After insulation, the joints are supported by a series of G10 blocks placed around the box, filling the gap between box and coil and between box and adjacent joints (at the same toroidal location), Drawing, 11P3WP_ These blocks, and the joints, are then clamped to the coil surface by a series of steel straps running around the coil and welded in place under tension, Drawing 11P3WP_ The coil terminal layout is illustrated on Drawing 11P6WP_ The conductors are brought out of the top and bottom pancakes and brought round to the vertical direction. The joint is the same as the interpancake joint. Each double pancake is to be supplied with supercritical helium at the inner diameter of the coil through a coolant inlet pipe located at the cross-over region of each of the two conductors, as shown on Drawing 11P The inlet pipe arrangement is similar to that used for the CS coil (section ). The cooling pipe to the middle of the conductor in a double pancake or the cooling pipe to a joint or termination will operate at the local electrical potential of the coil and must be ground insulated. Each insulated cooling pipe is routed to an insulation break, Drawing 1101WP_ and section , and then into a header pipe which runs to the coil terminations, Drawings 11P3WP_ and 11P3WP_ Correction Coils Eighteen multi-turn CCs are used to compensate field errors arising from misalignment of the coils and winding deviations from the nominal shape as a result of fabrication tolerances, joints, leads and assembly tolerances. 6 of the coils (the side CCs) are also used for feedback control of plasma resistive wall mode (RWM) disturbances. The operational current scenarios for the coils to be used for design are given in section There are 6 top CCs, 6 side CCs and 6 bottom CCs, arranged toroidally around the machine, outside of the TF coils but inside the PF coils, as shown on Drawings to Each of the CCs covers a 60 sector and spans three TF coils in the toroidal direction. The top and bottom coils are essentially planar while the side coils lie on a cylindrical surface. The radial or vertical sections of each coil are placed to coincide with a TF coil leg and the coil support is provided by clamps to the TF coil cases (section ). The CC orientation relative to the TF coils (i.e. the choice of the 6 TF coils used for support) is determined by the space requirements for assembly of the neutral beam ducts. Space for the coils is very limited due not only to the in-vessel access ducts but also to the TF and PF auxiliary supplies. Each CC consists of a bonded winding pack enclosed in a thin steel case. Pairs of diametrically opposite CCs are electrically connected in series inside the cryostat. There are, therefore, 9 independent circuits, each with its feeders and external power supply CC Conductor The CCs use two versions of a 10kA cable-in-conduit conductor using NbTi superconductor, as shown in the Drawing for the side correction coil and Drawing for the upper and lower correction coils. The cable for the upper and lower coils includes 300 strands and does not have any central cooling channel or Design Description Document 11 Magnet Page 118

120 subcable wraps. The cable for the side correction coils has 288 strands and consists of 8 final stage subunits cabled around a central cooling spiral. The cable for the upper and lower coils is formed circular and jacketed into a circular tube before forming to a square shape, with jacket thickness 2.1mm thick. The cable for the side coils is circular, contained in a circle-in-square jacket preformed by butt welding short sections. The field on the conductors is dominated by the adjacent PF coils. The maximum is estimated as 3.4T (although not all PF scenarios have been checked) which is used for design. There is considerable variation along the conductor length. Table CC Conductor side cc top & bottom cc Coolant inlet (K) Type of strand NbTi NbTi Nominal peak field (T) Maximum operating temperature (K) Maximum operating current (ka) Temperature margin at peak temperature (K) Equivalent discharge time constant (s) S/C strand diameter (mm) Central spiral outer x inner diameter (mm) 9 x 7 no Cable layout 3x3x4x8 3x4x5x5 Cable dimension (mm) x 14.8 Jacket dimension (mm) 24.6 x x 19.2 Local void fraction (%) Strand Cu:nonCu ratio Total non copper area in strands (mm 2 ) CC Winding Pack The CCs are wound as one in hand double pancakes, with joints on the outer surface and helium inlets on the inside. The cooling channel length corresponds therefore to one pancake, Drawings and The current capacity of top and side CC is kA at 3.4T. The current capacity of the lower CC is 180kA at 5T. Detailed parameters are given in Table The CCs are self supporting for loads along the conductor, but are supported for lateral loads by the TF Coil case. Table Operating Condition of the Correction Coils Top coil Side coil Bottom coil Max. operating current (ka) Max. current capacity per coil (ka) Max. operating voltage to ground and between terminals (kv) ±3 ±10 ±3 The turn, layer and ground insulation for each CC will use multiple layers of glass interleaved with polyimide film. The ground insulation thickness will be 8mm. The winding Design Description Document 11 Magnet Page 119

121 pack is vacuum impregnated with epoxy resin and, after impregnation, is enclosed in a protective steel outer casing 20mm thick. The residual gap between winding pack and case is then filled with epoxy resin CC Case The case design, for the lower correction coil (the other sets are similar in concept) is shown on Drawing 11BC In order to reduce induced currents in the CC casing, it must be split electrically along its perimeter an with insulating break, and the supports must be insulated from the TF coil cases. The break is formed by an insulating sheet of G10 placed at a flange closed by bolts, Drawing 11BC The CC winding parameters are given in Table Table Major Parameters of the Correction Coils Top coil Side coil Bottom coil Number of coils Average turn length (m) Turns per coil Nr x Nz 7x2 7x4 9x2 Total conductor length per coil (m) Number of conductors per coil dr (winding section) (m) dz (winding section) (m) Coil case thickness dr (coil section) (m) dz (coil section) (m) Winding weight of each coil (kg) Total weight of each coil (kg) Number of in-cryostat feeder pairs Number of terminal joints The helium and current inlets/outlets are situated towards the middle of each coil side, as shown for the side correction coils in Drawing 11SC Auxiliary Systems The auxiliary systems consist of the coil and structure feeders, the coil terminal boxes (which include the current leads, the coil and structure valve boxes, the coil instrumentation and the coil grounding scheme. The electrical insulating breaks between the high voltage and grounded parts of the conductor helium supply lines, although mounted at the coils, are a general component used in all coils and are also described here. More detailed descriptions are provided in the annexes to section In-cryostat Feeders Each in-cryostat feeder is a subassembly that connects a coil or structure to the end of a Cryostat Feedthrough (CF) located just inside the cryostat, as shown on Drawings Design Description Document 11 Magnet Page 120

122 11F , 11G and 11H for the coils and 11F for the structures. Each has a robust steel outer containment that provides protection to the cryostat components in the event of a short within the current supply busbars. For a coil, it consists of the feed and return current supply busbars (using NbTi superconductor), the return and supply helium lines and instrumentation lines, as shown on in Figure For the structures, the feeders contain a bundle of cooling pipes (since control of the supply to each of the 3 TF case cooling circuits is required). The upper structural feeders contain the 3 outlet supplies from the inside of the case as well as the outer surface outlets. The lower structural feeders contain a common inlet supply for each coil for all 3 circuits. Support Clamp 1st Ground Insulation 2nd Ground Insulation Cooling Pipes Inlet (4.5K) Separator N 11 GR F 2 Conduit Cooling Pipes Outlet (6.5K) Coil Vacuum Superconducting Busbar with Ground Insulator Co-wound Tap for Quench Detection Intermediate Screen (Metaric Tape) External Screen (Metallic Tape) Instrumentation cables KY, , DDD11JDE.cv35 R f Ground Insulation Monitor Rf R cg Figure Diagram of a Coil Feeder Cross-section (with Insulation Monitoring System) Table shows the number of magnet feeders, the number of joints in the in-cryostat feeders required for assembly, and the estimated total length of the in-cryostat feeders for all coils. Table summarises the principal features of the NbTi busbars used for the magnet feeders of the main coils. The busbars for the CCs will use a smaller conductor. Each busbar has a double insulation system that will allow detection of a developing bus bar-to-ground short circuit and allow operation to be stopped before the short occurs, providing protection against further damage (see also section 1.2.7). The scheme consists of a double layer of insulation with a metallic shield within the double layer. The insulation will be several half-lapped layers of glass/polyimide, followed by one or two layers of metallic tape with glass in between (to prevent eddy currents), and finally several layers of half-lapped glass/polyimide to complete the turn insulation. An over-wrap of steel tape will be applied for mechanical protection in handling, and for grounding when installed so as to serve as an electrostatic screen. This screen serves as extra protection against a short circuit between coil terminals in addition to the steel separator plate, Figure Design Description Document 11 Magnet Page 121

123 Coils Table Characteristics of the Feeders for Each Coil Incryosta t Feeders Estimated Length per Feeder (m) Total Length of Feeder (m) Total Length of Busbar (m) In-cryostat Busbar Joints per Feeder TF Coil PF Coil (PF1) PF Coil (PF2) PF Coil (PF3) PF Coil (PF4) PF Coil (PF5) PF Coil (PF6) Correction Coil (Top) Correction Coil (Side) Correction Coil (Low) CS Coil Structure Cooling (Up) Structure Cooling (Low) Instrumentation Total 35 Table ANSI Pipe Diameters Pipes (ANSI Schedule 5) Nominal Pipe Size Outer Diameter (mm) Inner Diamater (mm) ½ ¾ ¼ ½ ½ Total of In-cryostat Busbar Joints Each feeder contains a pair of superconducting busbars separated by a steel plate, two or more cooling pipes, and high and low-voltage instrumentation cables. The steel separator plate extends along the entire length of the conduit, as additional protection against a short circuit between busbars. The cooling pipes and the external conducting screens on the busbars are electrically connected to the separator, the separator is connected to the conduit, and the conduit is connected through its cryostat feedthrough to the grounded cryostat wall. The vacuum inside the conduit is part of the cryostat vacuum for all coils. Table summarizes the numbers and sizes of the conduits and the cooling pipes they contain for the coil and structure feeders (Table gives conversions to mm for ANSI pipe sizes). Details of the pressure drop estimated for the feeders are given in Pressure drop in feeder pipes (ITER_D_22GT39 v1.1) Design Description Document 11 Magnet Page 122

124 In-Cryostat Feeders Table In-cryostat Feeder Conduits and Cooling Pipes No. of In- Cryostat Feeder (IF) No. of Busbar in IF No. of Pipes in IF** Pipe size* OD of Feeder Duct (mm) TF /4 500 PF Correction Coil ½ 350 CS Structures TF Inlet CS Precompression Structures Outlet TF front inside TF sides and back inside TF outside CS Precompression (1 feeder only) (9) 9 (2) (5) 2 ½ 1-1/2 1-1/4 ½ ½ Instrumentation Total Pipe size is defined by NPS in ANSI B36.19, see Table ** ( ) shows valve number The superconducting cable for the busbars is designed to have more thermal protection than the coils, so that the coils can still be properly discharged in the event of a busbar quench. The time constant of the current decay is taken as 26s, approximately twice that of the coils. The operating current is taken as 68kA (i.e. the maximum for all coils) which provides some overcapacity on the PF coils (52kA in back-up mode). This has a negligible cost impact and allows a standard cable design. The conductor specifications are shown on Table Table Conductor for the Superconducting Busbars for the TF and PF coils and CS Parameters Large coil busbars CC busbar Type of strand NbTi NbTi Operating current (ka) Nominal peak field (T) 4 4 Operating temperature (K) normal Equivalent discharge time constant (s) hot spot Cable diameter (mm) Central spiral outer / inner diameter (mm) 8x6 0 Conductor outer diameter (mm) Jacket material SS316LN SS316LN SC strand diameter (mm) SC strand cu : non-cu Cabling pattern 3x4x5x5x6 3x4x5x5 SC strand Nr Cu core 2/3/4 stage (mm) 0/0/0 0/0/0 Local void fraction (%) in strand bundle SC strand weight/m of conductor (kg/m) Cryostat Feedthrough The Cryostat Feedthough (CF) includes a straight length from the cryostat wall to a S-Bend Box, as shown on Drawings 1101F1_ & -04 and 11G (the CS is Design Description Document 11 Magnet Page 123

125 similar). The S-bend box is also part of the CF and contains S-shaped bends in the busbars and the cooling lines. These bends accommodate the movements of the in-cryostat feeders and the coils, relative to the fixed S-bend box and Coil Terminal Boxes (CTB), due to differential thermal contraction and expansion, electromagnetic forces, and earthquakes. Table shows the number and length of the CF for each coil. The relative movement to be accommodated between magnets and building (to which the CTBs are attached) is summarised in section Detailed assessment of the S bend and the feeder stresses has yet to be completed. Table Number and Length of the Cryostat Feedthrough for each Coil Coils Number Length of CF Length of S- Total Length (m) bend (m) (m) TF Coil PF Coil (PF1, PF2, PF3) PF Coil (PF4, PF5, PF6) Correction Coil (Top) Correction Coil (Side, Bottom) CS Coil (CS1U, CS2U, CS3U) CS Coil (CS1L, CS2L, CS3L) Structure Cooling Instrumentation Total 31 Vacuum-tight bellows are provided to connect the straight length to the S-bend box to accommodate cryostat movements with respect to the fixed S-bend box and CTB. The CF also contains a thermal radiation shield between cryogenic parts and room temperature surface. This shield is designed similarly to the vacuum vessel and cryostat shields and uses silver plated steel panels cooled with 80K helium gas. The vacuum barrier at the flange connecting the CF to the CTB is also a feedthrough, which allows the busbars, cooling lines and instrumentation lines to pass into the CTB. High voltage wires penetrate directly the CF through high voltage feedthroughs. The busbar feedthrough is shown in Drawing 1101WP_ The NbTi busbars include terminations for joints to the in-cryostat feeder at one end, and terminations for joints to the CTB current leads at the other end. An additional joint is provided (for assembly reasons) between the S-bend box and the straight length. Cooling lines and instrumentation lines must be similarly connected. These joints can use either of the NbTi joint options described in section The grounded separator plate between the busbars inside the in-cryostat feeder conduit is extended and continued throughout its CF and CTB, as additional protection against a short circuit between busbars. The CF for the structure cooling pipes are similar to the coil CF (including the radiation shield), but they contain no busbars, a greater number of cooling pipes, no inner bellows, and they are connected to structure cooling valve boxes instead of CTB. Design Description Document 11 Magnet Page 124

126 Coil Terminal Box The Coil Terminal Boxes (CTB) provides the housing for interconnection of the magnet systems with the Cryoplant, the Power Supplies and the Data Acquisition System and they also house the local cryogenic control components. An internal thermal radiation shield covers all surface of the CTB. The internal layout is shown diagrammatically in Figure and in Drawing 1101F1_ Valves in the CTB control the mass flow rate of helium for each coil and current lead. These valves are also used during cool-down and warm-up operations to control thermal gradients. Figure Functional Diagram of CTB Current Leads ITER has a total of 60 current leads (18 for the TF, 24 for the CS/PF coils and 18 for the CC) with a total capacity of MA, Table The original reference design used conventional copper current leads cooled with 4.5K supercritical helium, based on the 1998 ITER baseline design. Since this time there have been considerable advances in current lead technology. Initially Nb3Sn was used to reduce the resistance of the low temperature section up to about 20K, providing some improvement over the oloidal nal leads. However, the discovery of High Temperature Superconductors (HTS) with superconducting capability up to over 90K has brought a much larger improvement. It is now possible to design leads that are cooled entirely by conduction at one end to the 4.5K superconductor and at the other by conduction to a heat exchanger with a cold end He input in the range 30-70K. If ITER were to use HTS current leads cooled with 50K He, the expected operational saving of the Design Description Document 11 Magnet Page 125

127 cryoplant is about 2.5 MW (over 6 months about 0.44kIUA with electricity at 40Euro/MWhr) compared to conventional low temperature leads. After reviews about the maturity of HTS technology for application to ITER, it is clear that HTS technology is already mature enough (being used in the EAST tokamak and for LHC) to be applied to the ITER current leads and that further advances are possible in the 5 years or so before the leads need to be built. The recommended HTS material is at present BSCCO 2223, metal (Ag-Au) stabilised. HTS-CLs consist of a HTS section and a conventional copper section that are connected together through an electrical joint at the 65~80K location, as shown in Fig The cold end of the HTS section is maintained around 5K by conduction to the main superconductor cooling circuit. The heat input from the warm end of the HTS section to the 4.5K refrigerator is determined by the thermal conductivity of the HTS section. Although cooling of the HTS section (by a feed from the 4.5K circuit) is possible (allowing a shorter HTS section and reducing overall cooling power) it introduces an undesirable complication into both the lead design and the cyoplant. This will be avoided in the ITER design. The warm end of the HTS section is maintained in the range 40~75K by thermal contact with a copper heat exchanger section, which is actively cooled with a gas flow (Helium or liquid nitrogen) in the range 50-80K. The outlet temperature of the coolant from the warm end of copper section is about 290K. The total cooling consumptions for HTS type current leads are a sum of requirements at both the 4.5K and 40-75K temperature levels. The main free design parameter to be selected for HTS leads is the coolant inlet temperature at the warm end of the HTS stack. The main options are liquid nitrogen, at either 77K or subcooled (at sub-atmospheric pressure) to 70-75K, or gaseous He in the range 30-65K. Generally sub-atmospheric operation is undesirable as the temperature can only be maintained by active pumping. Higher coolant inlet temperatures reduce the cryoplant power but also result in a larger requirement of HTS material. The power advantage is relatively small at 70K with LN2 and 77K appears marginal with the present HTS material. Sensitivity studies have been limited to cooling with gaseous Helium. Critical superconducting performance issues with the leads are the joints to the HTS section, especially at the cold end, where resistive heat has to be extracted by the 4K circuit. The ITER voltages to ground (Table ) are also very high and probably represent the largest design challenge. Coils Table Number and Capacity of Current Leads Number of pairs Maximum Current (ka) Max. Voltage (kv) normal/fault TF Coil /15 PF Coil /17 Correction Coil /15 CS Coil /20 Design Description Document 11 Magnet Page 126

128 Figure Main Components of a Low Temperature Current Lead (from Fzk, 70kA HTS lead) Another issue with HTS current leads is their behaviour in the event of a loss of coolant accident. The stacks of superconducting Bi-2223/AgAu tapes need to be mechanically supported by a low thermal conductivity tube (for example stainless steel), which also is an electrical shunt if the HTS material becomes resistive. The additional support tube considerably enhances the heat capacity of the HTS section. Due to the larger heat capacity a prolonged time is also available before the critical temperature of the superconductor is reached in the case of cooling flow loss. In principle, the same effect can be reached by an enlarged HTS cross-section. However, additional HTS is much more expensive than stainless steel and would lead to a more obvious increase of the heat leak to 4.5 K. The reason is that the ratio of heat specific to thermal conductivity of Ag-Au alloy is much smaller than that of stainless steel. Consequently, an over-dimensioned stainless steel support may be used to reach the ITER safety requirements (the HTS lead must not quench for 300s after a loss of coolant event and must allow a discharge time >25s in the event of an HTS quench). Sensitivity studies have been performed to select the basic HTS lead parameters. Details of these are given in Thermodynamic Optimisation of Binary HTS Current Leads (ITER_D_24L3WA v1.0) and Conceptual Design of Current Leads for the ITER TF, CS, PF and CC Coils (ITER_D_24L3VV v1.1). The following sections are taken from these reports. The design of the HTS lead proceeds in steps. The first is the consideration of the HTS properties. The base assumption is shown in Fig The HTS material is sensitive to the local magnetic fields, mainly the self field of the conductor. A critical current of 12 A/cm2 at 77K and zero field is considered, corresponding to currently available commercial material. For these studies, constant HTS and heat exchanger sections are considered. The length of the copper heat exchanger section is a function of the inlet temperature, the temperature at the cold end (T Cu ) and the heat exchanger performance, as shown in Fig The temperature of the cold end of the heat exchanger affects the heat conduction through the HTS section. For Fig and 7, a total lead length of 1.5m is chosen as being a reasonable value (sensitivity studies have also been performed on this) and the effect of HTS jc and He inlet temperature is considered. Fig shows that the minimum refrigerant power is reached with a He inlet temperature of around 60K using the base HTS material. Design Description Document 11 Magnet Page 127