motor starting current waveformsà contact bounce? 2

[electricpete]

At the following link are powerpoint slideshows of starting current waveforms during two starts of a 2500hp motor.

http://www.geocities.com/pschimpf/fwbp21.htm

There are unusual simultaneous subcycle-duration bursts of current on all three phases prior to each start, as if perhaps all three breaker contacts were bouncing together several times prior to finally closing. Once finally closed the current waveforms are as expected with decaying dc offset. But if that is what is going on I am at a loss to explain how the current attains such a high magnitude in such a short period of time.

The system is 13.2kv low-resistance grounded system (20-ohm transformer secondary neutral grounding resistor). The motor is connected ungrounded wye and has gronded-wye surge capacitors attached to the terminals.

I have phase-to-phase voltage traces from the bus which show no anomalies (phase to ground not available).

My primary interest is to try to understand how these current waveforms discussed can occur.

I know some strange things can happen when switching capacitors but I really can't quite picture what can be going on here.

FYI - Here is overview of why we are monitoring the current during starting of this motor:

We experienced 3 trips of a 2500hp 13.2kv motor upon starting over a 2 year period.
It is not reproducible. It did not recur during the monitored starts.
This is some confusion on which protective flags were actually received.

Relays tested OK. Motor and cable meggered ok. Motor ok by partial discharge test.
CT excitation and ratio tests sat. ct/relay circuit meggered to ground sat with ct neutral ground lifted.

CT ratio is 40:1. All quantities in secondary amps.

Normal Running current (less than full load) ~ 1.8A
Full load amps ~ 2.5A
Locked Rotor Current =15A

Hi Dropout Trip setpoint 20A (if does not dropout within 0.1 sec).
Instantaneous setpoint 24A

During these two monitored starts the motor did not trip. However the peak of the bursts of current do exceed the peak value of the instantaneous setpoint (35*1.41), although apparently not long enough to cause a trip in this case. Also it seems plausible we may have been close to tripping the high dropout.

 

[electricpete]

I apologize... the files are about 350k each and the site is a little slow.

 

[electricpete]

Correction: Instantaneous trip setpoint is 35A.

 

[electricpete]

I don't know much about breakers and switching transients, but after thinking about it some more I am thinking maybe this cannot represent any real contact bounce phenomenon. What do you guys think?

There is one other far less exotic possible explanation. EMI noise pickup on our sensing leads. Onto the secondary of our permanatly installed ct's we added clamp-on probes which outputted 1 volt per 10A CT secondary current. The leads carrying these signals were twisted but not shielded. We ran them along top of the switchboard and then down the front of the open panel door to the probes. The 50A peak current we see would be only 5v peak induced voltage. Seems plausible considering voltage spikes from breaker closing coils etc.

One other thing I noticed that seems to support it is that both starts has the same very similar pattern of polarity on these early current bursts.... A phase positive, C phase negative, B phase slightly positive.

Voltage leads were routed in a similar manner but carrying a much higher signal 120vac rms so maybe they have similar induced signal which just doesn't jump out on the trace. I will try to zoom in on the voltage signal to see if it has any small ripples in that time.

 

[electricpete]

There are two other things that bother me. Referring to the file fwbp21b on slide #3:

#1 - Why is B phase current so much higher than the others (look at the peak-to-peak currents in the neighborhood of the 5th thru 8th cycles on that slide)?

#2 - Why does the dc component appear to oscillate rather than decay away?

Any thoughts on these questions or whether this can/can't be contact bounce?

 

[busbar]

I hope I didn’t overlook these if already posted.

What kind of switching/motor control is used?

This happens on only one motor? Out of a larger group?

 

[electricpete]

The motor is started direct-accross the line by closing a GE Magneblast circuit breaker.

These trips have only been experienced on one motor out of 6 identical sister motors. We also have 24 other 13.2kv motors with similar relaying and have seen tripping upon starting only once before.... that instance was attributed to binding of the sleeve bearings after long period of inactivity. The 2500hp motor which is subject of this post has been checked for binding and none found.

 

[DanDel]

I don't believe that this is contact bounce, more likely interference as you have stated. There seems to be too much time between the spikes and the starting current.

I'm not sure about your second set of questions dealing with slide #3 of fwbp21b; the slide #3 shows the 60Hz(?) current sinewaves for each phase oscillating for a short time(about 17 cycles/280ms?) at a frequency of about 5Hz. I can't make out the time intervals very well. This is probably due to your system, which is probably reactive and possibly showing a little resonance. The high 'B' phase current(blue?) you mention seems to be part of the oscillation, as I see it. If this is not the correct slide, please inform me.

 

[electricpete]

Thx Dan, Good comments. With your comments I can feel more confident in concluding we don't have contaact bounce.

I think you are looking at the correct slide. As you say it appears the dc component is oscillating with a period of perhaps 5hz. It seems a little unusual for me to see an oscillation of the dc as opposed to a decay of the dc, which was my question #2. If I think about it from a circuit standpoint, maybe it is telling me that a capacitance is playing a role in the oscillation?.... I base this on the fact that for linear circuits, RLC circuit can oscillate, RL circuit can only decay. (ok, it's not a linear circuit... but at first glance I don't see how nonlinearity could contribute to oscilaltion of the dc component either).

Back to my question #1, I still think that B phase has significantly higher ac component (ignoring the dc) in that region of the waveform. This is seen by comparing the peaks.... the peaks of the blue current waveform stick out above and below the peaks of the other current waveforms.

Yes,, it is 60hz. The first few slides have axis labeled -250msec->+250msec or 500 msec window length. If you view in powerpoint at 100% magnification (may require adjusting screen to high resolution), you can see the gridlines which have a horizontal spacing of 50 msec and a vertical spacing of 20A (ct secondary current).

 

[busbar]

A couple of idle comments...for elimination. Sure doesn’t seem to correspond to a mechanical problem.

Looks more like it could be a data-acquisition artifact, but I suppose signal terminal and shield continuity has been examined.

Seeing it on one motor doesn't make obvious sense. With the long contact travel in a magneblast, one usually associates this as possible only with vacuum breakers or contactors.

 

[DanDel]

'pete, is it possible that where you say the blue phase peak-to-peak is higher, that the green and red phase p-p's are actually less than the blue? Disregarding the DC offset, it almost looks like both the green and red p-p's attenuate in that time frame and the blue stays the same over the whole slide.
I think its extremely possible that the oscillation is due to interaction between the reactive components of your system(L & C). I've seen switching transients of extremely short duration ring for almost two complete cycles(although at a higher-than-system frequency) in a highly reactive system, so this wouldn't be unusual. Consider the DC offset as a step voltage(take the 60Hz sine wave out), and instead of decaying right away(as you mentioned), it oscillates for a short time. The difference is that the rate of oscillation is a sub-harmonic of the system frequency.

 

[jbartos]

Suggestion: Consider a different monitoring set-up since there is the time of 240ms on:

http://www.geocities.com/pschimpf/9

when the normal waveform start appearing. This time interval is too long.

 

[electricpete]

jbartos - I realize the slides are difficult to read. If you look at the first page there is a table of contents. It identifies the location of some "zoom-in" slides which slow an expanded view.

 

[jbartos]

Suggestion to the previous posting: I am addressing the measurement principle and hardware setup adequacy for the measurement. The hardware is usually calibrated and tested for more important measurements, as this one appears to be.

 

[electricpete]

jbartos - I don't see anything in the slides that indicates a measurement problem. We captured around 15 seconds of data using Yokogawa digital recorder. These are simply screen captures from the post-capture waveform viewer software. There is capability to zoom-in and "pan" to whatever time interval is needed.