Title: Mechanical Realignment of Modular Coils to Mitigate Field Errors from As-Built Coils
1Mechanical Realignment of Modular Coilsto
Mitigate Field Errors from As-Built Coils
2Overview
- Objective
- To adjust alignment of each Modular Coil to
balance field errors induced from deviation in
current center of as-built coils. - Explore various alignment schemes for
effectiveness and required alignment - Method
- Generate as-built coils file from metrology data
- at present 12 coils are wound, A1-A4, B1-B3,
C1-C5 - Calculate Target Field Errors (b) from VACISLD
code - Used 3/5, 3/6 (symmetric) and 1/2 sine and cosine
modes on M50 ref plasma - Generate family of coils files with unit
displacements (1 mm) and rotations (1 millirad)
from as-built coils files - dr, dth, dz, rx, rth, rz
- Calculate Coupling Matrix (A) of each
displacement with each mode for each coil - Solve SVD problem Axb for coil displacements
(x) - Construct final coil transform from displacements
and verify coil performance
3As Built Coil Data
- There are presently 12 of 18 Coils wound and
current centers measured - A1 thru A4
- B1 thru B3
- C1 thru C5
- Design Goal of /- .020 achieved over most of
coils but not everywhere. - Coils wound to match current center achieved on
prior wound coils to minimize symmetry breaking
field errors.
4Modular Coil CastingsClamp Numbering
A
B
C
5Modular Coil Winding Cross Section
Current Center determined by averaging Casting
Measurements (prior to winding) and Bundle
Measurements (after winding) at each turn.
6As-Built Coil GeometryA1 thru A4 Lateral Current
Center Shifts
7As-Built Coil GeometryA1 thru A4 Radial Current
Center Shifts
8As-Built Coil GeometryB1 thru B3 Lateral Current
Center Shifts
9As-Built Coil GeometryB1 thru B3 Radial Current
Center Shifts
10As-Built Coil GeometryC1 thru C5 Lateral Current
Center Shifts
11As-Built Coil GeometryC1 thru C5 Radial Current
Center Shifts
12Field Errors from As-built Coils
- Differential coil files constructed by adding
current center deviations to original coil files
and subtracting original coils file (ie reversing
current) - Differential field added to VMEC field for
reference plasma configuration (full current,
full beta) within VACISLD code and normal
(radial) field on resonant surfaces computed.
Island size estimated from Bmn. - VACISLD code originally written as a module for
Stellopt, the suite of codes used to design
Modular Coils and Plasma
13Li383 Targeted Resonances
3/7 near axis and 6/9 near edge not targeted
14Island Width Evaluation used in VACISLD using
VMEC data
15Field Error from All Coils not much different
than worse coil taken separately
Winding Error Only Assumes Coil Perfectly
Positioned
16Some Possible Alignment Schemes
- Align each coil to target its own field errors
- Pro No Impact on assemble schedule
- Con Requires larger displacements and can target
fewer resonances (at most 6, more practically
only 4) - Align all coils of each type to target collective
field errors - Pro More dofs can target more resonances with
smaller displacements - Pro All coils of same type carry same current
- Con Requires all coils wound before first coil
can be assembled (schedule impact) - Align all coils of a given period to target
collective field errors - Pro More dofs (than single coil) can target more
resonances with smaller displacements - Pro No Schedule Impact (coils are assembled by
period) - Con Coil types carry different currents.
Effectiveness may drop as currents change
17Coil Displacements and Rotations madeRelative to
Local Coordinate Systems
y
3rd Period
18Effect of Displacements and Rotations on Sine
Phase of 3/5 Resonance for C1 as-built Coil
Similar Behavior Found with Other Coils And Modes
Coupling fairly linear, but not independent Some
displacements ineffective
19Field Error Reduction for Each Coil Independently
Correcting their Own Errors
A1-A4
Note Only 4 modes targeted due to
ill-conditioning
20Field Error Reduction for Each Coil Independently
Correcting their Own Errors
C1-C4
Note Only 4 modes targeted due to
ill-conditioning
21Field Error Reduction for All Type C Coils
Collectively Correcting the Net Errors
Note Targeting 6 modes with 24 dof More modes
may be feasible. Not yet explored
22Required Realignment MovementsWith Each Coils
Targeting Its Own Field Error
Movements are large relative to coil winding
tolerance of /- 0.5 mm Makes Coil Assembly
(flange, shim mate up) more difficult Also
induces fear of the unknown (what else is it
doing)
23Required Realignment MovementsEach Coil Type
Targeting Their Own Field Error
Movements are significantly smaller than when
aligning coils independently. Comparable to coil
winding tolerance of /- 0.5 mm
24Required Realignment MovementsAll Coils in
Period Target Collective Field Error
Movements are much smaller than when aligning
coils independently. Much less than coil winding
tolerance of /- 0.5 mm Minimizes impact on coil
assembly. Results for Period 1. Need B4 before
Period 2 can be evaluated.
25Points for Discussion
- What is the preferred Mechanical Alignment
scheme? - Can we tolerate aligning full period together?
- If the required motions are very small, should we
just not bother? - Are the right modes being targeted?
- Are the near resonance a concern (2/3)?
- Other plasma scenarios with different iota
profiles? - How to assess damage caused by non resident
fields (ie at boundary)? Or do we care? - Max field 20 Gauss or higher if individual coils
targeted - Is Mechanical Alignment a reasonable starting
point for the Magnetic Alignment (if Project
Approves) or is it better to do nothing? - Not clear that they target the same thing.