Memo to Steve Wolfe From Peter Titus Subject Analysis of the EF1

Document Sample
Memo to Steve Wolfe From Peter Titus Subject Analysis of the EF1 Powered By Docstoc
					Memo to Steve Wolfe
From: Peter Titus
Subject: Analysis of the EF1 Terminal wit
12kA current limit.
Date: October 7 2005

   Running a large positive current in EF1 and large
negative current in CS1/2 will produce a large launching
loads on EF1 upper and lower. The top of EF1U is
used for the LN2 cooling channels, and I am not sure
how well this detail transfers loads. I am guessing that
the cooling channels are on top of EF1L and would have
to transmit the centering loads. Having different current
directions in CS1/2 and EF1 is magnetically unstable, so
the coil will shift laterally within the clearances allowed.
I have not yet addressed any of these issues. I have only      F. 1 EF1 Analysis Model.
looked at the terminal loading, as this was the issue I
addressed after the EF3 flag failure.

    Upgrading EF1 to a positive 12 kA limit depends on the sense of
the positive current with respect to the TF field. If the current
direction and TF field are such that the region, where the current in
the strip turns, has loads that tend to collapse the end turn onto the
coil then the positive 12kA limit is probably acceptable ( The
issues in the first paragraph need to be resolved) . In the 1992
analyses I assumed a direction that bulged the end turn flag region
away from the coil. The coil current for EF1 was negative ~6ka in
this earlier scenario, and positive 12kA should collapse the                 Fig 2Current Directions that bulge out the
conductor and cover plate onto the                                           flags
coil. I think this is consistent with
the negative IP direction quoted in
Steve Wolfe’s email at the end of
this memo.        Thjs needs to be
confirmed       though because the
performance of the flags will be
very poor at 12kA if the current
direction is in the “bulging”
    The 1992 model of the upgrade
design treated only the loads
resulting from the current crossing
the TF field turning upward to the
terminal. This is the major loading
on the terminal details. Lorentz
forces from the PF scenario were
neglected. The uncompensated cross
over current was also ignored. The
hoop stress in the coil for the
scenario which damaged the EF3
coil was only about 2.5 ksi, so
ignoring all but the loads from
current crossing the TF field was not       Fig. 3 Stress in the cover plate, EF1 at 12kA, CS1/2 at their Maximums. This
a bad assumption.                           would be unacceptable, particularly at the fluid and electrical connections
                                            which are assumed to support the upper half of the bulging pressure on the
                                            end turn.
                                                                                                Elements in

          Fig. 4 1992 Analysis Model and modification. The steel backing plate is modeled as 1/16th inch thick. The
          current analysis assumes that the backing plate was used inbstalled.

  In this new model, the geometry from the old model is embedded into a model of the coil which is loaded
with the consistent Lorentz loading, and thus will impose the correct hoop tensile stress on the coil and the
terminal details. A Bo of 9 Tesla at Ro=.64 m was used to compute the toroidal field throughout the
model. The fix for the solder joint failure was to electroform an extension on the strip, which was simply
wound beyond the terminal. The termination is bonded through Nomex to the remainder of the winding
pack. The dead end is neither thermally loaded nor loaded with Lorentz forces. The capacity of the bonding
through the Nomex is difficult to predict. The bond between the flag extension and ground wrap
insulation may be better. In this analysis the bond is assumed lost, and frictional restraint from the tail and
the terminal region provides the main transfer of the first turn tension to the rest of the winding pack. If
the current directions are such that the turn region collapses onto the coil, the frictional connection between
the end turn and the rest of the coil will be increased. In these analyses, a “real” scenario was not used.
CS1, CS2, and EF1 were run at their administrative maximums to produce the largest vertical field and
thus largest hoop bursting load. The critical loading remained the loads resulting from currents crossing
with the TF field.

 Fig. 5 Radial Displacement ,. EF1 at 12kA, CS1/2 at their Maximums
  Fig. 6 Theta or toroidal displacements. EF1 at 12kA, CS1/2 at their maximums. At this level of
  loading, the turn extension which is only held with friction, has slipped.

Fig. 7 Vertical displacements. EF1 at 12kA, CS1/2 at their Maximums
        The present administrative current limits for the PF coils are
(assuming negative plasma current)

Coil       Imax (kA)
OH1      -35, +35
OH2U     -35, +35
OH2L     -35, +35
EF1U**   -7.5, +5 (proposed to change to +12)
EF1L**   -7.5, +5 (proposed to change to +12)
EF2U       0 , -5.5
EF2L       0, -5.5
EF3U*      0, +15
EF3L*      0, +15
EF4U*     -5, +5
EF4L*     -5, +5
EFCU*      0, -3
EFCL*      0, +3
Plasma     0, -2000

* The EF3 coils are connected in series, and the EFC coils are anti-
series, so their current magnitudes are always the same. The EF4 coils
are in parallel, so in principle their currents can differ, but not by
very much; in particular it is not physically possible for them to go to
opposite current limits.

** It is proposed to change the positive current limit on the EF1 coils
from +5kA to +12kA.
The EFc's will continually chop between their limits, and they
presumably have little impact on the out-of-plane load on the TF anyway,
so they shouldn't need to be varied. I expect that the most dangerous
condition is for a
disruption (plasma current to zero) and all the PF's go to one or the
current rail. That makes a total of 2^9 =512 combinations, rather than
Dave's 2^13; that might even be doable, if you can simply impose the
condition on the TF current distribution at the end of flattop of the
existing scenario. Your existing scenario probably actually exceeds the
current limits on the OH's, in which case it is pessimistic relative to
our present operational constraints.

I suggest that if you don't want to do all 512 combinations, you take
the positive current limits on the OH1, OH2U, OH2L,EF1U, and EF1L and
test with the stated symmetric limits on the EF2's (i.e. both -5.5ka or
both 0), the EF3's (both +15kA or both 0) and the EF4's (both -5kA or
both +5kA).

The other thing you might try is just look at the EF1's by themselves
(everything else zero). In fact, maybe you should try that first. If the
magnitude of the out of plane load due to the EF1 is small enough
compared to
the OH's in the plasma scenario, then we're done!

Stephen M. Wolfe
M.I.T. Plasma Science and Fusion Center
Room NW17-101
175 Albany Street
Cambridge, MA 02139

telephone:        (617) 253-5510
fax:                (617) 253-0627

Shared By: