TRIPPING OF HARMONIC FILTER 2

[reactive]

We have a 6,6kV 6MW 6/12 pulse thyristor Mine Winder with a 3,1 MVAr 3rd order damped filter (Resistor and decoupling capacitor across the reactor) tuned at 4,8p.u. connected to the same busbar (no other loads). Acceleration of the winder peaks the load at about 15MVA.

Measurement at approx. 500 ms averaging shows normal harmonics for the winder with the filter out of circuit. With the filter in circuit acceleration of the winder results in immediate tripping of the filter on harmonics (the filter is rated for 90% harmonics at 5th).

Parallel resonance below the tuning frequency occurs at exactly 4p.u. We measured the filter harmonics during the trip and managed to see one set of readings as the winder accelerated before the filter tripped. The 4th harmonic was at around 350A on all phases (it could be and is likely, based on the levels that the relay saw, that the peak 4th is much higher).

The 4th without the filter in circuit is 4-5A giving an amplification factor of anywhere between 90 and 300.

Simulation of the network shows a max. possible amplification of 25. What on earth is going on?

Is it possible that some interaction is occuring between the filter and the winder and is it related to the lack of damping on the busbar?

In South Africa we have relatively high levels of harmonics due to the many mines and therefore are very experienced in the field of harmonics. However, this is difficult to explain - the max. amlification that we have seen in 20 years is 40x and that situation was theoretically explained.

 

[ScottyUK]

Reactive,

Is it possible that your voltage source is being modulated by the 4th harmonic i.e. giving you a 4th harmonic voltage component?

Can you measure whether there is 4th harmonic current contribution from the transformer, as well as from the rectifier? It seems like you have an energy source which isn't accounted for within the 4 or 5A originating within the rectifier. The passive type of filter you are describing will have a well defined Q-factor which you appear to have calculated, and this doesn't seem to account for amplification to 350A or more with the Q-factor quoted, so there must be another source of current.

I don't have the simulation tools I had at my previous employer, so I'm casting ideas forward to provoke thought.(or to be shot at!)

 

[jghrist]

Perhaps you could model the filter and the winder in EMTP to find out whether, as you suspect, the distorted voltage causes an increase in 4th harmonic currents generated by the winder.

 

[duffey777]

Reactive,

Did you finally solve your problem?

The last post on EMTP modelling is an excellent recommendation, and I also would recommend this. In my earlier years when working under a noted harmonics expert, he always recommended never allowing a first parallel resonant point sitting on the 4th. He never supported his advice, but said there were possible "control" interactions with the rectifier. I have never measured, read or seen this at work, but it appears you are living in the midst of what he spoke about. I would be interested in knowing what your actual conclusions are about your issue.

 

[reactive]

What we have discovered is that the winder is operating in so called 'sequential' mode. Apparently this is a control system developed by Alstom specifically for the South African market due to relatively weak supplies in some areas. During the winder cycle this sequential control converts the firing from 6 to 12 pulse and back to 6 pulse and it appears that the problems occur during this conversion. Our theory is that in this sequential mode the control synchronising is highly sensitive to 4th harmonic no matter how small.

We have increased the damping on the filter and it now stays in circuit. We have therefore been able to measure the disturbances. As the winder converts from 6 pulse to 12 pulse the characteristic 6 pulse harmonics rapidly diminish and large uncharacteristic even harmonics appear (with the filter in circuit). The voltage THD just before this happens is less than 2%. Strange one!

Alstom are going to site on Monday to try the winder in standard 12 pulse mode to confirm that this will solve the problem. I will let you know.

Seems to us to be a case of a rarely encountered and largely untested control system.

 

[duffey777]

Reactive,

Great info on a unique problem being debugged.

We have debugged several like this. One involved energizing a backup UPS with filters onto a backup gen. The transformer magnetizing inrush and UPS filter first parallel resonant point on the 3rd created an excessive transient over-voltage. We actually simulated the problem in EMTP to prove cause, and to determine mitigation alternatives.

If you have any waveform data captured (like a stream of three phase V and I samples for say 1 to 10 seconds), let me know. I would be interested in running them through an analysis program I have developed. If you were any closer (Oregon, USA), I would offer to do a free measurement. This is a phenomenon really worth documenting.

I would suggest that you consider (of course you would have to have time), writing a paper, magazine article, or white paper on this phenomenon. Like I said earlier, this has been talked about, but I have never seen any documentation on this interaction problem.

 

[jbartos]

Suggestion: Even harmonics are generated by half-wave rectifiers (AC-DC converters) or similar devices that use three diodes and three controlled rectifiers not six controlled rectifiers and no diode in the AC-DC conversion part.

 

[kevd]

Possible source of even harmonics is transformer in-rush. Perhaps when the driver starts the winder he closes the convertor transformers whose in-rush current interacts with the filter bank and is the source of 4th. I am unclear how a 6 or 12 pulse convertor generates 4th harmonic, unless one or more thyristor devices is mis-firing, however for such a big drive which spends so much time in inversion, this would manifest itself in the drive shutting down for other reasons.
For 12 pulse series convertors, sequence control is simply the control of each 6 pulse series bridge in such a manner as to reduce the kVAR demand from the supply at intermediate speeds. Both bridges start on the their backstops and as the drive requires voltage, one convertor fires foward, the other stays on its back-stop. When the volage demands causes one bridge to reach its front-stop, the other convertor bridge starts to fire forward. Maximum loop voltage is then reached with both convertors on the front stop. When the winder slows down the sequence reverses, with one bridge retarding its firing angle until it reaches its backstop and then the other bridge retards. This is as opposed to normal 12 pulse series control where both 6 pulse bridges operate with the same firing angle. I still do not understand how this gets 4th harmonic.

Question, has this system just been installed? or has it been working for some years and only just now giving problems. Also has someone actually checked and measured the capacitors, just in case you have lost some capacitor cans and the effective MVAR capability is much lower than designed?

Kevd

 

[reactive]

Transformer inrush is not the problem here. We are seeing the problem during the constant speed part of the cycle and not during acceleration or deceleration.

As mentioned before the problem does not occur with the filter out of circuit. However, the only harmonic that the filter can amplify is 4th and we believe that this amplification of an uncharacteristic even harmonic is somehow causing problems with the control system, even though the resultant amplified 4th is relatively small. Voltage THD just prior to the chaos is less than 2%.

Cap bank has been checked (it also has out of balance protection). Winder has been in service for years but filter has just been installed (old filter moved from another shaft).

As mentioned the control system was specifically developed for South Africa by Alstom and I wonder how well it has been tested in service.

More comments would be appreciated.

 

[jbartos]

Suggestion to the previous posting: Small traces of even harmonics can came from imperfect characteristics of various devices, e.g. switching devices. Have you tried to trace the 4th harmonic origin, e.g. by a power quality analyzer, etc.?

 

[reactive]

All converters generate some uncharacteristic harmonic.

 

[jbartos]

Comment on the previous posting: The uncharacteristic harmonics can be even and odd; however, they tend to be noticably smaller than characteristic harmonics.