Ground fault on 3-wire ATS transfer? 1

[peebee]

We have experienced a strange problem at a new installation:

We have a 3-wire ATS fed from a 3-wire grounded 3-wire 480-volt service, as well as a 3-wire grounded emergency generator source. ATS is reported to have midpoint-off position, although delay setting is unknown (currently being verified). ATS loads are (4) 3-wire elevators.

As part of comissioning, the generator was manually started, and a manual transfer to emergency was initiated. The emergency breaker tripped on ground fault, the ATS automatically retransferred to utility, and utility then also tripped on ground fault.

The breakers were reset and power was restored with no other corrective action. Feeders insulation had previously tested OK.

We have not been able to recreate the problem since, transfers have been OK. However, a concern remains about future failures of the emergency power system.

Some possible explanations come to mind:
1) Intermittent motor fault
2) Intermittent feeder fault
3) Super-low setpoints on ground fault pickup (or sensor) -- we're currently checking this.
4) Bad filter or TVSS in elevator controller
These two would seem to be easy enough to test for -- any specific recommendations on specific tests?

Here's some other explanations, although they all seem less probable:
5) Harmonics -- but zero-sequence current harmonics are not possible with 3-wire system, are they?
6) Insufficient mid-point delay on ATS -- but that would only show up as an overcurrent, not as a ground fault, right?

Any thoughts on this? Any of the above that we can eliminate as impossible? Any other potential problems I might have overlooked?

 

[peebee]

Oops -- should have read "These *four* would seem to be easy enough to test for. . ."

 

[peebee]

I'm tempted to recommend the following test:

Disconnect all phase conductors at the service entrance panel. Bond all phase conductors together and perform hi-pot test to ground.

BUT -- I think we'd need to disconnect the (solid-state?) elevator controllers, and they could well be the source of the problem (due to integral TVSS or power supply filter).

ALSO -- It's not clear to me what voltage we'd want to use for this test -- 2kV?

 

[electricpete]

Just thinking out loud…
Are motor loads secured during the transfer?
Is it break before make (interval?) or break before make? Any phase check?
I don’t really see how any transfer issues give a ground fault but just wondering.

If it happened once and not repeated maybe it is related to slow or uneven movement of mechanical transfer switch contacts (doesn’t apply to electronic contacts).

Describe the ground relay monitoring scheme.

You are right that surge protection needs to be disconnected. My thought is 500vdc for megger of 480v system. Higher voltages would be considered a hi-pot. You are right that surge protection needs to be disconnected for test.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[peebee]

Re: "secured" -- please clarify -- are you asking if the motors are securely attached to the motor mounts, or if the elevators have been stopped?

I'm still digging into the midpoint-off and phase-check configurations/settings at the ATS's.

I believe that ground-fault tripping is via a typical residual system, but I'm double-checking this too, as well as sensor ratings and settings.

 

[electricpete]

Sorry, you can take the man out of the Navy but you can't take the Navy out of the man. Secured means stopped.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[peebee]

Apparently, no-one knows if the elevators were running or not. But the test occurred at 4am, so it seems unlikely that the elevators were running.

 

[rbulsara]

this may be a late input..

1. Since no breaker is tripping now, there is no existing ground fault to be found. So I do not see value of hi-pot or meggar tests.

2. since two breakers tripped on GF, they must have sensed 'GF'. This does not mean there was an actual GF.

I suspect they saw some imbalance in phase currents for whatever reason and triggered the GF. I would suspect low and too sensitive (no or little time delay) in GF setting.

The cause of imbalance could be a motor starting at the time of the event, or cap charging or Transformer inrush?. Electricpete's suspicison of imporper contacts is also likely.

If there was a real fault the fault may have burned clear, but that sounds remote to me. You may want to look closely at all equipment look for some telling signs such a burnt cap /wire etc.

Some loose CT conncection may also do it, but to happen on two breakers cast some doubt in that theory. But my prime suspect would be GF scheme and settings.

 

[jrhaus]

I believe you are experiencing back EMF from your elevators, unless you have delayed transfer on the ATS. If the generator is considered an emergency source, ground fault protection is not required for the emergency side of the ATS.

 

[busbar]

Is the ‘normal’ source solidly grounded wye? Is the ‘emergency’ source solidly grounded wye? Are there any phase-to-neutral loads served by this 480V system on either side of the ATS? Were ground-fault features in the two circuit breakers tested by primary injection prior to the trip incidents?

 

[peebee]

rbulsara -- not too late, we may be digging into this for a little while, it seems like it usually goes that way with intermittent problems. . . .

We do realize that GF is not required on the emergency source, but we'd still like to eliminate the problem on the normal side (plus, if there is an existing ground fault, we'd like to keep the GF operational on both systems until we figure out where the problem is).

Turns out that both systems are grounded wye, 4-wire. But the ATS in question is only 3-wire, no neutral conductor at all. Downstream loads are 3-wire elevators. The elevator manufacturer states that the power wiring lands directly on isolation transformers within the controllers, and that any connections they have to ground or neutral are on the secondary side of that isolation transformer. I still have not seen schematics, though, and I'm not quite convinced that there couldn't be an upstream TVSS, filter, or other grounded circuit.

On a Dy transformer, any secondary side ground fault would simply show up as a balanced overcurrent on the primary side, right?

Most of the information I'm working with so far is all coming from 2nd-hand verbal reports, I'm trying to get documentation or other confirmation on all of this.

We're still double-checking the relay settings, that would seem to be a #1 suspect. We're also trying to verify that the specified injection tests were actually performed on the relays -- it would seem unlikely that there were more than one breaker installed with mismatched/miswired sensors, but you don't know for sure with out the injection test reports.

I appreciate the recommendation against hi-pot testing, and a test engineer I've spoken with recommended the same thing. Similarly, he also DID recommend looking for visible signs of fault damage as a much easier check for ground faults. In addition, he recommended placing a PQA wired for zero-sequence (around all 3 phases) at various locations, and transferring the ATS a couple times, preferably with the elevators running in the DOWN direction (motors lifting counterweights). Those two tests seem like as good a place as any to start -- any other recommendations?

Keep the guesses coming, they're very much appreciated.

 

[peebee]

By the way, as I alluded to in that last post, the GF sensing at all breakers is residual off the phase CT's.

 

[busbar]

Sizes permitting, a jumbo clip-on CT at various points around three phases {connected to a multimeter} might expose something.

One thing about [d-y] transformer-secondary ground faults—they should manifest as overcurrent in two primary phases.

 

[peebee]

Re: "One thing about [d-y] transformer-secondary ground faults—they should manifest as overcurrent in two primary phases."

Exactly -- and the total 3-phase circuit would still be balanced -- no ground fault on the primary side. Maybe an overcurrent trip if the magnitude was really high, but no GF. Right?

 

[busbar]

Agreed with a secondary ground fault, the drive-isolation transformer primary windings will have only unbalanced phase currents, and nothing that should initiate a ground-fault trip.

 

[peebee]

Some updates on this:

1) A PQA was attached to the ATS a few weeks ago, and the ATS was transferred a couple times. The trip units were showing ground faults in the 400-600 amp range, but the PQA showed zero.

2) I neglected to mention this problem was actually happening on two ATS's. In addition to the one above, there's a second ATS, which is 4-wire, but which is feeding only (2) Dy transformers via a 4-wire distribution board. No actual 4-wire loads anywhere on this. The breaker vendor is saying that the GF trip is being caused by transformer inrush -- this seems unlikely to me -- any thoughts on that?

3) Here's an excerpt from a report from an engineer who was present during an attempt to recreate the problem last Monday. PQA was apparently NOT in place during all this, the readings stated were taken off the trip units:

We put the ATS through a series of 10 transfers to generator power and 10 retransfers to normal power this morning between 3:30 AM and 4:20 AM. We started by increasing the center off delay on ATS-ECDP02 from 3 seconds to 5 seconds. We transferred and retransferred the ATS 4 times without any problems. We did not watch the ground current for the first attempt, but on attempts 2, 3 and 4, the ground current displayed on the normal power feeder breaker was zero. On the 5th attempt, we tripped the normal power feeder breaker. This circuit breaker displayed a ground current of 406 amps.

We then installed a jumper between terminals B8 and B9 on the normal power feeder circuit breaker, disabling the zone interlock feature. We transferred/retransferred the ATS a 6th time. On the transfer to normal power the normal power feeder circuit breaker displayed a ground current of 521 amps, but did not trip. This suggests that disabling the zone interlock solves the nuisance tripping problem.

On the 7th transfer/retransfer the normal power feeder circuit breaker showed a ground current of 16 amperes. No circuit breakers tripped.

On the 8th transfer/retransfer, the emergency power feeder breaker for ATS tripped on ground fault. The circuit breaker displayed a ground current of 433 amps. We disabled the zone interlock on this emergency source feeder circuit breaker.

We transferred/retransferred two more times. On the 9th attempt, the normal power circuit breaker displayed a ground current of 51 amps, but did not trip. On the 10th attempt, the circuit breaker displayed a ground current of zero.

The ground currents displayed are considerably lower than they were a month ago. A month ago the circuit breakers we showing ground currents of up to 950 amps when they tripped. Before leaving we confirmed that the ground fault on the main circuit breaker on the normal side was set to 0.4 seconds and 1200 amps. This should be adequate to hold even, if these larger ground currents show up again.

I suggest that from here on teh ATS be exercised along with all of the other ATS's, but keep an eye on it for the next few weeks.

I will continue to work with the breaker vendor to see why the circuit breakers think there is a ground current, even though we were not able to see this on the power analyzer two weeks ago. By the end of this week I also hope to make some recommendations for modifying the settings on all of circuit breakers in the emergency-source switchboard. In all likelihood, we will want to turn off the ground fault feature on these circuit breakers.

 

[rbulsara]

Zone interlocking automatically sets the feeders breaker's (lower tier)GF delay to minimum, regardless of setting you dial in.

By the way what size (amp wise) ATS and breakers are we talking about? The amp numbers are all relative, but 400-600A GF sounds very odd. If PQA does not see a ground but trip units sees it, it makes CT (sensors in the breakers) suspect. Are they of correct ratio? Any loose sensor CT conncetion..?

I still tend to agree that transfromer inrush is causing the imbalance (which always does) and initiating the feeder CB GF which is pretty much has zero time delay (because of the zone interlocking).

I would remove the zone interlocking, set the feeder GF time delay to .3 seconds or so and try the above tests again. This should resolve the issue.

If still trip ( very unlikely) remove the GF from these breakers.

 

[peebee]

The breakers/ATS feeding the elevators are 400A/3-pole.

The breakers/ATS feeding the (2) 225kVA Dy transformers are 800 (4-pole ATS), but only 3-wire loads.

I think it's clear from the above, but the steady-state GF current is very low, just a couple amps or less. The 400-900 amp values are transients which occur only during transfer operations. And these are being recorded only at the breakers, such high GF currents were never recorded at the PQA.

I too suspect the breakers. Injection tests were specified, but I've still not seen a copy of the report despite several requests. The breaker vendor has spent some of these 3:30am Monday mornings out at the site, but generally has blamed tx inrush and has so far avoided the elevators.

 

[peebee]

Re: "If still trip ( very unlikely) remove the GF from these breakers." -- I'm scared that if we do this, it will just push the problem up to the main breaker in each switchboard -- and those are required to have GF protection.....

 

[rbulsara]

peebee:

I understand your point, but I think your main GF is set at 1200A and with good time delay (0.4 sec) which is more than sufficient.

I am pretty sure that removing the zone interlocking feature, thus making ATS feeder breakes GFP indepdent and setting them to 0.3 sec or better will solve the issue.

Its the lack of time dealy that is causing the problem, I think. Also getting different current readings for differernt successive tests is very indicative of transformer inrush being the root cause. You will always measure different inrush on susccessive (random) energizing...as you never hit the same point of the voltage wave twice in succession.

This assumes that CTs, wiring etc. is ok.