31 October 2001

Presentations

Minutes

The weekly project meeting was held on October 31.  The focus of the meeting was to status progress generating a buildable coil set.  The requirement for a buildable coil set is driven by the need to perform manufacturing studies and engineering analyses on a design that is realizable, or at least describable.  Coil designs to date have entailed overlapping windings, degenerate shell surfaces, and insufficient space for elements not yet in the model.  These designs did not even pass the describable test.  Perhaps characterizing what we need as a describable coil set would be more appropriate than a buildable coil set.  The determination of whether it is buildable will be made in the course of the manufacturing studies, but we need to start with something describable.  The measure of whether the design is describable is the development of a Pro/E model which has good surfaces (smooth, without creases) for the integral shell, the tee and root features, and the winding packs; which has adequate space for adding clamps to hold the winding pack on the structure; and which has adequate space for assembly, relative thermal expansion, cooling tubes, and thermal insulation between the modular coils and the vacuum vessel (no interferences).

David Williamson reported on manual adjustments to the 1018a2 modular coil set which was generated on the fixed winding surface.  Initial efforts to modify the coil geometry in (u,v) space were stymied by our inability to identify the regions of tight curvature and minimum coil-to-coil separation in (u,v) space.  Straight lines in (u,v) space might actually be tight curves in 3D, so Williamson reverted to moving the points in 3D.  The Rhino code that allows moving points on a surface appears to be quite helpful in this regard.  Williamson will continue with the manual manipulations to get to a describable design.

Mike Cole is developing the geometry of the integral shell, located outside the modular coil windings.  Cole was unable to expand (thicken) the winding surface far enough to encompass the region where the shell would be located, so he has to generate a new, smooth surface that satisfies the minimum offset requirement.  Cole indicated that this surface should be available next week.

Art Brooks commented that the trouble in expanding the winding surface (and perhaps meeting the minimum radius of curvature requirement) might be related to the geometry of the winding surface which itself has a minimum radius of curvature of 4cm, far less than the target value of 14cm.  The minimum radius of curvature in the winding surface appears in the vertical ridge on the inboard side near v=0.5.  Tom Brown agreed to provide a new winding surface created by smoothing the existing winding surface in this region only, without violating the required stand-off from the vacuum vessel.  He indicated that the new winding surface should be available by the end of this week (November 2).  Other participants (e.g., Williamson, Strickler, and Brooks) can use this improved winding surface if they find it advantageous to do so.

Dennis Strickler reported on recent results using CoilOpt.  The 1018a2 coil case was designed using the fixed winding surface that Tom Brown generated and Art Brooks fit with Fourier harmonics.  It resulted in fitting errors of 0.9% average and 3.7% maximum.  The minimum radius of curvature was 9.6cm (v. a target value of 14cm) and the minimum coil-to-coil separation was 11.8cm (v. a target value of 16cm).  Strickler investigated using a spline representation (1029b5) instead of a Fourier representation (1018a2) on the same winding surface.  There was virtually no improvement.  Allowing the winding surface to vary resulted in an improved fit, with an average error of 1.2% and a maximum error of 4.7%.  The coil-to-coil separation increased to 14cm and the minimum radius of curvature increased to 11cm.  The coil-to-plasma separation increased ever so slightly to 19.8cm.  However, the crucial coil-to-vessel distance was not calculated.

Art Brooks reported that he has a code that is operational (based on the TiltOpt code) that might be helpful in improving the fit and smoothing the coils generated by the manual changes made by Williamson.

In Williamson's presentation, he noted that the minimum radius of curvature constraint was violated over a fairly large region in the M1 coil.  The minimum radius of curvature constraint is related to how tightly we can bend an insulated conductor with crinkling the insulation or keystoning the conductor.  The current criterion is 5X the conductor thickness, which results in a required minimum bend radius of ~14cm.  Heitzenroeder brought in a recent conductor sample and wrapped some glass insulation around it.  He showed that it could be bent to a tighter radius than we would have allowed, although keystoning was clearly visible and would have to be accounted in the coil build in regions of tight curvature.

The minimum radius of curvature and coil-to-coil separation constraints are different in character.  If two coils are too close to each other, they will overlap or there will be insufficient space for clamping the conductor to the structure.  We cannot even build a valid Pro/E model if the coils violate the minimum coil-to-coil separation constraint.  Our target value is 16cm.  This would allow the coils to be in any orientation with bumping into each other.  If the coils were perfectly parallel, we might be able to get by with as little as 12cm.  The true practical minimum probably lies between these two extremes.  The radius of curvature constraint is different in that we can build a valid Pro/E model with a tighter radii of curvature as long as it does not get so small that it puts a kink in the coil.  The manufacturing studies and our own R&D should tell us more definitively how tight we can go.  Heitzenroeder will discuss with Nelson near term R&D that can be conducted to more definitively establish a minimum radius of curvature constraint.

In order to expedite development of a describable coil design, Williamson was directed to relax the minimum radius of curvature constraint from 14cm to 9-10cm but move the coils far enough apart (14cm?) so that we can grow tees off the structural shell to support the windings; not have the tees or windings bump into one another; and have adequate space for staggered clamps.

The day of this meeting, October 31, was the target date for having a buildable coil design and that did not happen.  The following revised dates leading to the start of the manufacturing studies are proposed:

Please forward any comments or corrections to reiersen@pppl.gov

(last edited on 11/01/2001 12:37 PM )