April 26, 2001

A joint Engineering/Physics meeting was held on Thursday, April 26.  There were three topics on the agenda:

Fred Dahlgren presented the results of his buckling analysis of the NCSX vacuum vessel.  The shell thickness he used was 0.175".  Ports were included and had thicknesses ranging from 0.375" for the smaller ports to 0.475" for the larger ports and port covers.  The material was Inconel 625.  The only load was a 1 atmosphere external pressure load on the vacuum vessel.  A buckling factor of safety of 2.9 was calculated using NASTRAN.  This factor of safety was below the value of 5 recommended in the B&PV code.  The buckling mode appeared to be localized to the top (and bottom) sections of the NB port extensions in the v=0 plane (with the banana-shaped plasma cross-section).  Peak Tresca stresses from the 1 atmosphere load were around 20ksi in the regions where the large rectangular port joined the vacuum vessel shell and in the inboard region near the v=0 plane near the tips of the banana-shaped cross-section.  These peak stresses appeared to be below the allowable for Inconel 625 of 27.5ksi. 

Dahlgren repeated the calculation with a shell thickness of 0.25".  The buckling FOS increased from 2.9 to 6.7.  The peak stress dropped from 20ksi to 13.4ksi.  Dahlgren concluded that a shell thickness of 0.22" would be required if there were no other modifications.

Mike Cole questioned the shell thickness of 0.175".  His recollection was that the shell thickness should be 0.375".  From a stress and buckling standpoint, a thicker vessel would be delightful.  However, there are four concerns:

1. Electrical time constant

2. Formability

3. Uniformity of thickness

4. Stresses due to EM (disruption) loads

5. Cost

The time constant of the vessel (w/o ports) was calculated by Brooks on 3/27/01 to be 6.6ms for a 0.375" thickness of Inconel 625 (130e-8 ohm-m).  The requirement is 10ms, so we appear to be in good shape here even with a thicker vessel.  The vacuum vessel forming study recently completed by Molodikh et al (KhAI), recommended explosive forming based on a shell thickness of 6mm (0.24").  However, the report indicated that thicknesses up to 20mm (0.79") could be formed using explosive forming.  Non-uniformity of thickness is also a concern.  We would like to have a vessel for which the minimum guaranteed thickness is structurally adequate.  Stresses due to EM (disruption) loads would be additive to the stresses due to atmospheric pressure loads.  No calculation of stresses due to EM loads has been performed to date.

Cost is a major concern on NCSX.  We need to make the vessel as inexpensive as possible without compromising performance or adding technical risk.  It appears that the following steps are in order:

1. Assess cost as a function of thickness.

2. Verify that increasing the shell thickness to 0.375" does not cross the feasibility threshold for explosive forming.

3. Determine the uniformity of thickness expected with explosive forming.

4. Provide a preliminary estimate of stresses due to EM (disruption) loads.

Once these steps have been completed, it should be possible to make an informed recommendation on the vessel thickness (and cost).

Art Brooks assessed the impact of a multi-filament representation of the modular coils on the field at the plasma.  The assessment was based on the 0907 healed coils.  A 4x4 array of filaments was used represent the coil cross-section.  The coil cross-section was oriented normal to the winding surface.  Brooks examined:

Changes in the field errors at the boundary and differences in VMEC reconstructions were very small.  Changes in resonant field errors are also small, but large enough so that islands should be observable (if the plasma reconstructions from the single filament coils were indeed island-free).

PIES runs for the multi-filament coils actually looked better than for the single filament coils after 100 iterations in PIES.  However, a PIES problem was encountered shortly thereafter that terminated execution before convergence was attained.  Nevertheless, the initial results were very encouraging.  Work will continue to fix the PIES problem, run the single and multi-filament cases until convergence, and compare plasma properties using our normal suite of evaluation tools.

Two problems surfaced after the PVR regarding the PIES beta-current scans:

Runs were made with the healed coil set and compared with runs for the unhealed set.  The healed coil set appeared to improve other states as well as the reference state.

Please forward any comments to reiersen@pppl.gov

(last edited on 10/18/2001 03:39 PM )