Miniproposals

Miniproposal: Date Filed: Title: First Author: Session Leader: Joseph Snipes (shots 1-32)

Engineering Operator Run Comment

Lengthen pulse and work on position control

Session Leader Plans

Physics Operators Plans

Session Leader Summaries

Entered: Jul 7 2004 03:49:07:477PM

Author: To Be Determined

Summary of run of Thursday, 5/27/93 Physics Operators: Wolfe,Snipes Engineering Operators: Gwinn,Daigle The idea of today's run was to extend yesterday's results toward longer discharges with better position control. Main focus was on use of the R_CUR controller. The predictor for R_CUR was the FILAMENT predictor with SCALE_BIT set to 1.e-4. Plans to reduce the offset errors in this and Z_CUR by customizing the BTOR coefficient were never carried out. The first shot of the day was based on the programming from the last shot of yesterday, modified to account for the fact that the engineers fixed the gain problem in the OH1 supply. It lasted 340msec, with current up to 260kA, decaying after 100msec. The radial position showed larger excursions, with the plasma spending much of the time hard up on the inner wall. The fact that the first shot of the day was also the best shot of the day says something about how the run went. The RCUR controller at the beginning of the day employed OH1, OH2U&L, and EF3. On shot 14, we added the EF2U&L, with M=2e-2. On shot 20 the M coefficient corresponding to EF3 was dropped from .1 to 4e-2. The OH1 and OH2U&L coefficients were maintained at -1.e-2 and -2e-3, respectively. A PID scan from 1 to 5, and then down to 0.5, was carried out between shots 21 and 32. The results were as follows: As the gain was increased to 5., the R_CUR position was progressively better controlled. The actual plasma position, as represented for example by the x-rays, also seemed to be better controlled; the FFIT:RGEOM value, which seems to be pretty consistent with the xray centroid, also showed the plasm to be well positioned at the higher gain, while the excursions at the lower gains were over 5cm. However, the higher the RCUR gain, the earlier the plasma started experiencing disruptions. I suspected that it might be getting too elongated, and tried manually reducing the EF1 current, which had been extended earlier in the run. This did not seem to help, although further reductions are possible. The last change in the current demand waveform for EF1's was not realized due to too extreme a voltage programming and insufficient gain (100ms effective time constant). The last shot of the day was taken with the RCUR PID gain reduced to 0.5, and recovered the main characteristics of shot 1, lasting until 320msec, and with poorly controlled position. The conclusion is that this controller can control the position, but not in a way the plasma likes. The problem may be the shape, although this looks pretty good, or interaction with the wall. In connection with this possiblity, Ian's reconstructions indicate that the disruptive cases touch the outer limiter shortly before they disrupt. The FFIT reconstructions on the other hand show a large gap to the limiter. There is some supporting evidence from H-alpha and bolometric profiles in favor of the FFIT edge location, but these are not yet conclusive. Problems: The OH1 exhbited a 60 cycle problem, putting large amplitude (500a) current ripples on the OH1 and OH2 coils, beginning on shot 8. The supply was running in 12 pulse mode for the first 8 shots and the first 7 were fine. Changing to the rear cabinet only made the problem worse; changing to the front cabinet only seemed to alleviate the problem. This pathology has appeared before, and needs to be hunted down and snuffed. The shot-to-shot irreproducibility of the current rise is still with us. Of 32 total shots, there were 17 good discharges, 12 fizzles, and 3 no breakdowns. Several attempts were made to tweak fields to improve reliability, without success. The Filament R_CUR predictor is very insensitive. The \FFIT:R_GEOM position moves by 5 to 10 times as much as the predictor value. The predictor is more or less consistent, up to small offsets, with the FFIT:R_CUR position, which also exhibits small excursions. The problem may be that the model can match the measurements by putting too much of the m=1 currents in the vessel and structure, which manages to get the outer flux surfaces right but the current centroid wrong.

Physics Operator Summaries

Session Leader Comments
May 27 1993 09:14:04:480AM		Joseph Snipes	Start of run 5/27/93 Multiplied M matrix R_CUR OH1 control by 4. Multiplied OH1 voltage pre-programming by 4 from commutation onward to take into account hardware changes in the gain. Z_CUR control of OH2 also reduced from 6.25 x 10**-3 to 6 x 10**-3. Fill pressure set to 6 x 10-5.
May 27 1993 09:26:02:960AM		Joseph Snipes	Start of run 5/27/93 Supply Start Close Open Crow Bar Invert Commutation R (Ohm) EF1 L -1.6 -1.55 0.0 0.05 0.35 0.18 EF1 U -1.6 -1.55 0.0 0.05 0.35 0.18 OH1 -1.45 -1.40 0.0 0.05 0.35 0.10 OH2 L -1.6 -1.55 0.0 0.05 0.35 0.02 OH2 U -1.6 -1.55 0.0 0.05 0.35 0.02 EF2 L -1.0 0.35 0.045 EF2 U -1.0 0.35 0.045 EF3 -0.1 0.35 EF4 -1.6 0.35 TF -1.65 0.35
May 27 1993 09:43:48:440AM	930527002	Joseph Snipes	Shot 1 Decent shot. Lasted 344 ms, but started disrupting about 230 ms. About 260 kA.
May 27 1993 09:45:58:790AM	930527002	Joseph Snipes	Shot 2 Cut in voltage from EF3 from previous shot for the pre-programming. Cut in the B_z from the previous shot into S programming.
May 27 1993 09:54:56:930AM	930527002	Joseph Snipes	Shot 2 Cut in voltage from EF3 from previous shot for the pre-programming. Cut in the B_z from the previous shot into S programming. Shot not as good.
May 27 1993 10:01:23:610AM	930527003	Joseph Snipes	Shot 3 Extended R_CUR proportional gain from 0.2 to 0.4 s. Eliminated B_z feedback that was on from 0 to 0.05 s at a gain of 2.0. Dud!
May 27 1993 10:16:44:720AM	930527004	Joseph Snipes	Shot 4 Try again. No breakdown!
May 27 1993 10:30:28:560AM	930527005	Joseph Snipes	Shot 5 Put back the B_z proportional gain from -0.01 to 0.05 s at a gain of 2. O.K.
May 27 1993 10:48:35:430AM	930527006	Joseph Snipes	Shot 6 EF1 voltage pre-programming changed from -200 something at -0.25 s to -100 V. Decent shot, but still disrupting early.
May 27 1993 10:52:05:190AM	930527007	Joseph Snipes	Shot 7 Held voltages on EF1 at -100 V from 50 ms to 250 ms. Extended the EF1 current waveform to 350 ms. Wimper.
May 27 1993 11:06:00:570AM	930527008	Joseph Snipes	Shot 8 Try again. No breakdown!
May 27 1993 11:19:41:960AM	930527008	Joseph Snipes	Shot 8 Try again. No breakdown! Oscillations on OH2. Changing to rear cabinet only on OH1. It seems that the oscillations on OH2 are induced from OH1, due to problems in the front power supply cabinet.
May 27 1993 11:52:10:600AM	930527008	Joseph Snipes	Shot 9 O.K., but imbalance in EF1's.
May 27 1993 11:55:38:140AM	930527010	Joseph Snipes	Shot 10 OH1 switched to front power supply cabinet. Turned on proportional gain to 1.0 from 0.05 s to 2 s on EF1U + L. Wimper.
May 27 1993 12:25:54:040PM	930527011	Joseph Snipes	Shot 11 Try again. O.K.
May 27 1993 12:29:36:490PM	930527012	Joseph Snipes	Shot 12 From 0.05 to 0.3 s OH2 voltages changed from -45 V to -30 V. Similar to 11.
May 27 1993 12:50:36:560PM	930527013	Joseph Snipes	Shot 13 Turned off PID proportional gain on OH1 current after commutation. OH1 voltage programming changed from -200 V to -150 V from 0.05 to 0.3 s. Changed pulsed gas increased to 15 V from 0.1 to 0.3 s.
May 27 1993 12:56:48:200PM	930527013	Joseph Snipes	Shot 13 Turned off PID proportional gain on OH1 current after commutation. OH1 voltage programming changed from -200 V to -150 V from 0.05 to 0.3 s. Changed pulsed gas increased to 15 V from 0.1 to 0.3 s. We have been running on the gas line without the plenum all day. Valve was closed. Opened for this shot. Much better with gas!
May 27 1993 01:06:37:710PM	930527014	Joseph Snipes	Shot 14 Added EF2U + L to the M in R_CUR at a value of 2 x 10**-2. Wimper!
May 27 1993 01:16:54:860PM	930527015	Joseph Snipes	Shot 15 Try again. Another wimper.
May 27 1993 01:47:25:070PM	930527016	Joseph Snipes	Shot 16 Problem with the two-color-interferometer offset.
May 27 1993 01:50:13:200PM	930527016	Joseph Snipes	Shot 17 Adjusted EF2 voltage programming to try to match the feedback. Wimper!
May 27 1993 02:01:25:400PM	930527017	Joseph Snipes	Shot 17 Wimper!
May 27 1993 02:05:48:760PM	930527018	Joseph Snipes	Shot 18 Changed PID proportional gain on EF2 from 1 to 0 after 0.04 s. Was going out to 0.2 s. Wimper!
May 27 1993 02:23:53:050PM	930527019	Joseph Snipes	Shot 19 Raised EF4 voltage waveform at 1 ms from 740 to 790 V. Much better! Still disrupts after 150 ms.
May 27 1993 02:34:43:700PM	930527020	Joseph Snipes	Shot 20 Changed M matrix for EF3 on R_CUR from 0.1 to 0.04. Wimper!
May 27 1993 02:46:02:870PM	930527021	Joseph Snipes	Shot 21 Delayed onset of the proportional gain on R_CUR from 24 ms to 40 ms. Looked reasonably good. Breakdown conditions seem to be marginal.
May 27 1993 03:05:49:890PM	930527022	Joseph Snipes	Shot 22 Proportional gain on R_CUR increased from 1 to 2. Looked a little better.
May 27 1993 03:09:17:940PM	930527023	Joseph Snipes	Shot 23 Proportional gain on R_CUR increased from 2 to 5. Moved the time at which the B_z gain goes to 0 from 0.045 to 0.04 s. No breakdown!
May 27 1993 03:41:16:470PM	930527025	Joseph Snipes	Shot 25 Reduced EF1U + L current waveforms from 0.05 s onward. Wimper!
May 27 1993 04:08:50:270PM	930527027	Joseph Snipes	Shot 27 Delay crowbar on OH1U + L from 0.05 to 0.06 s. Wimper!
May 27 1993 04:23:50:660PM	930527028	Joseph Snipes	Shot 28 One more time! Still disrupting at 120 ms.
May 27 1993 04:30:49:090PM	930527029	Joseph Snipes	Shot 29 Ran the EF1 currents (S waveform) down to 100 A at 99 ms. Wimper!
May 27 1993 04:47:40:390PM	930527030	Joseph Snipes	Shot 30 Dropped EF2 current by 30 A at 17 ms. Last 250 ms with disruptions.
May 27 1993 04:57:54:410PM	930527031	Joseph Snipes	Shot 31 Dropped proportional gain on R_CUR from 5 to 0.5. Wimper!
May 27 1993 05:15:27:190PM	930527032	Joseph Snipes	Shot 32 Last shot of the day. Very similar to shot 1. Back to where we started. Plasma duration about 330 ms. First disruption around 240 ms. Good place to start tomorrow. Punt!

Physics Operator Comments

Engineering Operator Comments
Shot	Time	Type	Status	Comment
1	09:21:06:760AM	Plasma	Ok	350 msec good shot
2	09:43:20:160AM	Plasma	Ok	200 =msec good
3	09:56:28:940AM	Plasma	Bad
4	10:06:17:890AM	Plasma	Bad
5	10:18:16:790AM	Plasma	Ok	200ka 250 msec--tf com fail conv 4 in inversion
6	10:37:54:170AM	Plasma	Ok	Plasma Current 260 KA
7	10:50:07:760AM	Plasma	Ok	Plasma Shot 60 KA.
8	11:04:10:290AM	Plasma	Bad
9	11:23:56:850AM	Plasma	Ok	260 ka
10	11:53:43:770AM	Plasma	Ok	40ka -- oh1front converter only, no 60Hz
11	12:04:37:150PM	Plasma	Ok	270ka plasma
12	12:27:19:640PM	Plasma	Ok	280ka plasma
13	12:47:00:510PM	Plasma	Ok	PLasma shot 260KA.
14	01:04:50:790PM	Plasma	Ok	PLasma shot Low current.
15	01:14:40:840PM	Plasma	Ok	Plasma Shot Low current 30 KA.
16	01:28:06:900PM	Plasma	Ok	Plasma shot 250KA.
17	01:48:33:210PM	Plasma	Ok	Plasma shot low current.
18	02:02:42:080PM	Plasma	Ok	Good Plasma SHot
19	02:16:14:440PM	Plasma	Ok	Good Plasma shot 250KA.
20	02:32:49:530PM	Plasma	Ok	60ka
21	02:42:04:380PM	Plasma	Ok	Plasma shot 250KA.
22	02:52:47:690PM	Plasma	Ok	Plasma current 240KA.
23	03:07:20:660PM	Plasma	Ok	plasma shot Very low current.
24	03:18:59:360PM	Plasma	Ok	Plasma shot 230KA.
25	03:39:26:710PM	Plasma	Ok	Plasma shot Low current.
26	03:48:20:320PM	Plasma	Ok	Plasma shot 240kA.
27	04:07:30:300PM	Plasma	Ok	Plasma shot Low current.
28	04:16:29:030PM	Plasma	Ok	Plasma shot 240KA.
29	04:28:22:520PM	Plasma	Ok	plasma shot Low current
30	04:43:05:440PM	Plasma	Ok	Plasma current 230KA.
31	04:54:51:950PM	Plasma	Ok	plasma shot low current.
32	05:08:59:510PM	Plasma	Ok	good shot