1. A PROPOSED TF JOINT DESIGN FOR THE NSTX CENTERSTACK UPGRADE Robert D Woolley 25 February 2009

2. NSTX CSUG CONCERNS Reliability of TF Joints in CSUG Cost & time to implement CSUG – Costs are dominated by PPPL labor – Costs are driven by design complexity Reliability of other CSUG items – Structural supports (e.g., vacuum vessel) – Interturn insulation shear in TF central bundle – PF coils and their supports 2

3. NSTX TF JOINT DESIGNS HAVE BEEN A SOURCE OF TROUBLES – Arc and fire occurred after a TF joint of the first design opened while powered. (Cost=1 yr & $) – TF joints of the second design then failed due to inadequate potting quality (Cost=0.5 yr & $) – Third try still behaves strangely above 0.5 Tesla (Cost=No higher field operations to 0.6 Tesla) – Pitting of joint contact surfaces continues, perhaps due to localized current concentrations & arcing – A fundamental issue is support of the radials. 3

4. REASONS FOR THE PREVIOUS NSTX DESIGNS’ TROUBLES (IN MY OPINION) TF flag stiffness transmits to joints torques resulting from lateral & vertical flag forces – Two competing load paths: TF flag stiffness vs flag box potting compound – Joint lift-off occurs at full field, but occurs at lower field if potting is bad The design's right-angle turn in current direction concentrates current at joint corner – Current concentration jumps to new location when lift-off occurs (thus arcing?) Magnetic forces do not clamp joints closed, they only rock joints laterally or vertically Contact surfaces are cut by bolts which concentrate current in the heightened contact pressure zones surrounding bolt holes while also reducing overall net contact area 4

5. TF FLAG CURRENT TURNS CORNER Note that current streamlines bunch more at corner than these current direction arrows

6. THE NEW DESIGN SHOULD HAVE THE FOLLOWING FEATURES (2) – Current should have parallel (vertical) current flow on both sides, as lap-joints 6

7. TF JOINT DESIGN CONCEPT SHOULD BE REPLACED FOR THE CSUG The new NSTX operations will increase toroidal field, plasma current, poloidal field, and pulse duration, so forces and temperature rises will increase at many locations. These force and temperature increases could challenge the original design concept. 7

8. PROPOSED CONCEPT FOR TF JOINT & FLAG REPLACEMENT (IP) SINGLE-LAYER TF IN CENTERSTACK LAP-JOINT GEOMETRY PROVIDES UNIFORM CURRENT DENSITY || CURRENT SHAPING GENERATES E-M FORCES PRESSING JOINTS CLOSED FLEXIBLE CONDUCTING TF STRAPS REDUCE JOINT TORQUES AND ALLOW THERMAL EXPANSION NO BOLTS CUTTING THROUGH CONDUCTING JOINT SHEAR KEY INTEGRATED WITH ELECTRICAL JOINT

9. CT FLEX CURVES CLOSE TO OH FIELD This reduces OOP forces near centerstack where support is difficult Compromised to avoid lid and to maintain EM clamping force on joint Flexes could run closer to OH field lines if lid raised

10. ANSYS TF MODEL 36-fold rotational symmetry Multiphysics includes conduction, magnetics, structural, electromechanica l contact Includes TF Central Conductor, insulation, shearkey, joint contact, straps Outer leg not realistic PF not yet modeled No Vacuum Vessel

11. TOROIDAL FIELD

12. PROPOSED CONCEPT FOR TF JOINT & FLAG REPLACEMENT (OOP) VERSION A: WATER-JET CUT PLATE – Flexible in IP direction, Stiff in OOP direction – No OOP flexes – May not need any more OOP support (calcs ongoing) – Likely Showstopper=Assembly Tolerances (±0.25 degree OOP machined joint surface angle) Could boltplate shims rescue the Version A concept ? VERSION B: BRAIDED CONDUCTORS – Flexible in OOP direction – OOP support by leaning on a stationary interturn structure mounted on lid

13. TF JOINT APPARENT RESISTANCE P VS R MEASUREMENTS (2003) FOR SILVERED CU JOINTS LED TO THE FOLLOWING PLOT THIS WAS CHECKED BY A MARCH 04 TEST USING A SPARE TF FLAG & DIFFERENT BOLT TENSIONS. RECENT ANALYSIS SHOWED THAT FOR NSTX CSU, MINIMUM PRESSURES OF 14-20MPa (2-3ksi) ARE ADEQUATE ASSEMBLY TOLERANCE ISSUE

14. THE PROPOSED TF DESIGN (B) INCLUDES THE FOLLOWING : (72) TF Radial assemblies shaped to direct internal TF current in an optimized pattern. Each assembly includes the following copper components: A Boltplate, to be bolted to an TF outer leg terminal A VerticalBar, with a TF joint surface on the lower part of its inner vertical face, and with shearkey features registering its fit with a TF turn Flexible cables connecting Boltplate with VerticalBar (72) Lid-mounted OOP radial rib supports (2) Bucket-ring Supports (if needed) – Including screw-adjusted inclined-plane clamps 14

15. SINGLE-LAYER TF IN CENTERSTACK 36 TF centerstack conductors form a single layer, with 36-fold rotational symmetry in the toroidal direction. Existing TF outer leg terminations require additions: – Existing terminals for each end of the 12 three-conductor TF outer legs accommodate the present 2-layer version of the TF central bundle. Thus 2/3 of the existing TF outer leg terminals are at elevations closer to the plasma while 1/3 are at elevations farther from the plasma. – Vertical copper bars will be attached to the 2/3 of outer leg terminals closer to the plasma to extend them to the same elevation as the 1/3 of terminals farther from the plasma. 15

16. SINGLE-LAYER TF IN CENTERSTACK (2) Max radius 0.1925 m Each conductor subtends 10 degrees Turn width is 33.5 mm (= 1.32”)

17. COPPER BARS FOR TF OUTERLEG TERMINALS Bars bolted to lower TF outerleg terminals extend to match elevation of upper terminal Bars may also need to spread sideways – 36 equal spaces – Enough space to insert bolts

18. FLEXES SHOULD BE BRAIDED (B), NOT SOLID Water-jet cut is only flexible in vertical plane. Misalignments require horizontal flexibility OOP Support of flexes by Lid-Mounted Radial Ribs requires horizontal flexibility Flat braided cable isi commercially available in many sizes, including up to 500 kcm. Total needed is 6.4 mcm, so as few as 13 cables could suffice.

19. ADDITIONAL OOP SUPPORT An additional OOP support may be needed. The VerticalBar may need OOP support tying it to the TF central bundle of conductors. The TF central bundle may need OOP support if insulation shear is excessive without it. These can both be provided if needed via a bucket-ring structure clamped to the TF centerstack.

20. SUMMARY For the CSUG with its new higher field operations, the scheme for TF radials and joints should be replaced with a better one. A flexible radial structure with CT curve shape will eliminate competing multiple load paths. A lap-joint configuration will provide EM clamping pressure and more uniform current