Transient Voltage Spikes - What is normal or acceptable?

[petersonra]

I think worrying over any "grounding" issues is going to point you in the wrong direction. It is probably the least likely thing to be at issue. My guess is you could install ten ground rods at each building and not have it change a thing.

I don't see how a 60V spike is likely to cause the kind of problems you are seeing. Modern equipment can usually tolerate that kind of transient. Surge suppressors often do not even engage at that point, although I think my first point of protection would be to install surge protectors on all the panelboards and any place where I had repeated failures.

I agree with the poster who suggested an intermittent neutral. They can be very hard to find and can cause all kinds of strange symptoms.

 

[well grounded in Kansas]

I think worrying over any "grounding" issues is going to point you in the wrong direction. It is probably the least likely thing to be at issue. My guess is you could install ten ground rods at each building and not have it change a thing.

I don't see how a 60V spike is likely to cause the kind of problems you are seeing. Modern equipment can usually tolerate that kind of transient. Surge suppressors often do not even engage at that point, although I think my first point of protection would be to install surge protectors on all the panelboards and any place where I had repeated failures.

I agree with the poster who suggested an intermittent neutral. They can be very hard to find and can cause all kinds of strange symptoms.

Very possible. Since the emergency lights sense a power failure and come on and UPS units sense a failure I would not rule out a loose connection that fully opens for milliseconds or a few wave forms.

Heres a question. From the main to the separate buildings are 2 parallel 500's and the main feeders from the utility transformer are 6 parallel 400's. Looking at the feeder to building B from A would both neutral wires have to open (be intermittent) or would the loss of one cause the transient? While I am asking, could anything on a branch circuit cause it?

 

[well grounded in Kansas]

More loggings. In less than 24 hours 40 transients of 50% RMS 120VAC where captured.
Started logging at 17:15 11/14/14

1. 11/14/14 19:19:04
2. 11/14/14 19:34:29
3. 11/14/14 19:39:59
4. 11/14/14 19:40:44
5. 11/14/14 19:47:27
6. 11/14/14 19:41:41
7. 11/14/14 20:10:12
8. 11/14/14 20:10:56
9. 11/14/14 20:14:11
10. 11/14/14 20:37:57
11. 11/14/14 23:56:18
12. 11/14/14 23:56:56
13. 11/15/14 00:01:28
14. 11/15/14 00:05:44
15. 11/15/14 00:06:00
16. 11/15/14 00:08:44
17. 11/15/14 00:12:50
18. 11/15/14 00:17:01
19. 11/15/14 00:21:11
20. 11/15/14 00:21:20
21. 11/15/14 00:25:33
22. 11/15/14 00:28:39
23. 11/15/14 00:29:01
24. 11/15/14 00:39:47
25. 11/15/14 00:40:29
26. 11/15/14 12:43:34
27. 11/15/14 13:53:23
28. 11/15/14 13:57:18
29. 11/15/14 13:57:59
30. 11/15/14 14:09:01
31. 11/15/14 14:14:11
32. 11/15/14 14:14:42
33. 11/15/14 14:24:32
34. 11/15/14 14:59:19
35. 11/15/14 15:05:52
36. 11/15/14 15:06:11
37. 11/15/14 15:06:28
38. 11/15/14 15:08:41
39. 11/15/14 15:11:02
40. 11/15/14 15:11:49

Unit stopped due to the maximum it can store.

 

[junkhound]

Can you correlate any change in scheduled activity at 2:30 PM and 9 PM ??

 

[gar]

141120-0911 EST

well grounded:

A list of times with no amplitude and duration information is of little value other than to indicate something occurred.

What does 50% RMS 120 V mean? For example, is it the instantaneous peak voltage of the full wave rectified value of the measurement relative to common (base line) that exceeds 120 + 60 = 180 V. The peak of a 120 V RMS sine wave is 170 V. Thus, a transient combined with a sine wave at its +/- peak of just 10 V that increased the amplitude would trigger an event. But a transient of 10 V that reduced the peak would not trigger an event. If a transient occurred at a zero crossing, then its amplitude would need to be 180 V to trigger an event.

What is the required duration of a transient to trigger an event.

The waveforms you have shown imply a very short duration relative to the base frequency. Thus, these can relatively speaking be referred to as high frequency.

.

 

[well grounded in Kansas]

Can you correlate any change in scheduled activity at 2:30 PM and 9 PM ??

No. This has been going on since the building opened over 3 years ago and there is no pattern other than it's total randomness. It can happen at feverishly anytime and then go a day with little to nothing. One group of readings went mostly at night then the next group is only in the day.

 

[well grounded in Kansas]

141120-0911 EST

well grounded:

A list of times with no amplitude and duration information is of little value other than to indicate something occurred.

What does 50% RMS 120 V mean? For example, is it the instantaneous peak voltage of the full wave rectified value of the measurement relative to common (base line) that exceeds 120 + 60 = 180 V. The peak of a 120 V RMS sine wave is 170 V. Thus, a transient combined with a sine wave at its +/- peak of just 10 V that increased the amplitude would trigger an event. But a transient of 10 V that reduced the peak would not trigger an event. If a transient occurred at a zero crossing, then its amplitude would need to be 180 V to trigger an event.

What is the required duration of a transient to trigger an event.

The waveforms you have shown imply a very short duration relative to the base frequency. Thus, these can relatively speaking be referred to as high frequency.

.

The 50% RMS with a 120 AC being measured means nothing is logged until it sees a spike of 60VAC or more. It would be hard to upload 40 pictures, but all of these captures show a spike in the waveform of a short duration. The spike appears as straight line in the waveform and can be at the top and extend the peak of 170 to 270 say. Others are in the middle of the waveform and the spike is a straight line that goes below the sign-wave and above the sign-wave. The larger percentage are recorded at the top of the sign-wave. A sign-wave peaks at about 170 so it is not just looking 180v measurement, but anything in short duration that is 60volts or higher out of the norm.

 

[mayanees]

thoughts

thoughts

When I do PQ metering I align the transient event with the ITIC PQ curve to illustrate the deviation from an acceptable waveshape. All we have from your description is a % over nominal, without a duration.
But, I don't contest that your experiencing events given that this is a Fluke PQ meter, and your list of damaged devices is significant.

In your initial post you wrote:
The first grounding system was inadequate and was three rods bonded was added to the 1600 amp service, but two separate buildings with their own panels were not grounded until recently. That EC left them un-interconnected. All three buildings are separately grounded now, but feed from the same 1600 amp main and transformer.

Are the remote building panels that are fed from the 1600-amp main properly treated as sub-panels, with 5-wire 480 Volt service feeders? Your description casts doubt, and suggests possible issues with multiple N-G bonding jumpers.

 

[gar]

141120-1211 EST

well grounded:

If you have a positive transient spike of 61 V peak centered at a voltage positive zero crossing, then will this trigger a transient event? The peak of this transient is 109 V below the sine wave peak.

An experiment to try:
A branch circuit of about 50 ft or more of #14 or #12 wire.
Plug the test instrument into a duplex socket at the end of the branch circuit.
Find a two bulb 8' Slimline with a magnetic ballast, no noise filter, and with bulbs.
Put a two prong plug on the Slimline cord.
Rapidly oscillate the plug back and forth in the socket to quickly produce many random on off cycles of the ballast.

What are the highest peak voltages of the transients recorded? I have observed upwards of 4000 V with a scope.

Do the above test with the Slimline at the same outlet, but the test instrument at the main panel on the associated breaker output. How much is the highest transient reduced?

Other inductive loads can be used instead of the Slimline, but I don't have a specific suggestion.

.

 

[well grounded in Kansas]

When I do PQ metering I align the transient event with the ITIC PQ curve to illustrate the deviation from an acceptable waveshape. All we have from your description is a % over nominal, without a duration.
But, I don't contest that your experiencing events given that this is a Fluke PQ meter, and your list of damaged devices is significant.

In your initial post you wrote:
The first grounding system was inadequate and was three rods bonded was added to the 1600 amp service, but two separate buildings with their own panels were not grounded until recently. That EC left them un-interconnected. All three buildings are separately grounded now, but feed from the same 1600 amp main and transformer.

Are the remote building panels that are fed from the 1600-amp main properly treated as sub-panels, with 5-wire 480 Volt service feeders? Your description casts doubt, and suggests possible issues with multiple N-G bonding jumpers.

The Fluke 43b can log other things, like harmonics, but not simultaneously when catching transients. The three buildings are feed from one 1600 amp service in building 2 (then middle) Feeders are run underground to buildings 1 and 2. Since there was only two rods (no other GES) and they are inside the foundation and not to code as to distance apart* after about six months of fairly significant problems they added three bonded rods properly outside and ran a new ground wire to the service.

The problems continued and we brought in our electrician and found a large amount of the project was done without a dedicated neutral on branch circuits as the spec had required. Still that way. Our equipment in one building is connected to the GEN SET panel and has 37 hots and only 27 neutrals on the three phase panel. The rooms share a neutral and are on AFCI breakers which have tripped with no explanation.

We have thought of field bonding (intentional or accidental) and they have assured us that none exist. After stating the site was to code, we discovered that building 1 and 3 did not have their own grounds (GES). The ground was back at building 2. They then put in ground rods at 1 and 3, but disconnected them from each other. I questioned that and feel it is not how the code is detailed. 250.32 I was told the code has gone back and forth. I am going by what I read now.

So, to answer your question. The separate buildings are 4 wire (A,B,C and N. No G to the main)

 

[well grounded in Kansas]

141120-1211 EST

well grounded:

If you have a positive transient spike of 61 V peak centered at a voltage positive zero crossing, then will this trigger a transient event? The peak of this transient is 109 V below the sine wave peak.

An experiment to try:
A branch circuit of about 50 ft or more of #14 or #12 wire.
Plug the test instrument into a duplex socket at the end of the branch circuit.
Find a two bulb 8' Slimline with a magnetic ballast, no noise filter, and with bulbs.
Put a two prong plug on the Slimline cord.
Rapidly oscillate the plug back and forth in the socket to quickly produce many random on off cycles of the ballast.

What are the highest peak voltages of the transients recorded? I have observed upwards of 4000 V with a scope.

Do the above test with the Slimline at the same outlet, but the test instrument at the main panel on the associated breaker output. How much is the highest transient reduced?

Other inductive loads can be used instead of the Slimline, but I don't have a specific suggestion.

.

Gar,
I will see if I can do the experiment. The outlet we are on is dedicated to us and in a locked room. About 50 feet of MC to the Emerg. Panel we are on.
To answer your question, Yes it would catch that spike in the middle of the waveform and even though it does not extend past the ends of the signwave.
On a side note, I have been looking for an excuse to buy a scopemeter.

 

[GoldDigger]

The Fluke 43b can log other things, like harmonics, but not simultaneously when catching transients. The three buildings are feed from one 1600 amp service in building 2 (then middle) Feeders are run underground to buildings 1 and 2. Since there was only two rods (no other GES) and they are inside the foundation and not to code as to distance apart* after about six months of fairly significant problems they added three bonded rods properly outside and ran a new ground wire to the service.

The problems continued and we brought in our electrician and found a large amount of the project was done without a dedicated neutral on branch circuits as the spec had required. Still that way. Our equipment in one building is connected to the GEN SET panel and has 37 hots and only 27 neutrals on the three phase panel. The rooms share a neutral and are on AFCI breakers which have tripped with no explanation.

We have thought of field bonding (intentional or accidental) and they have assured us that none exist. After stating the site was to code, we discovered that building 1 and 3 did not have their own grounds (GES). The ground was back at building 2. They then put in ground rods at 1 and 3, but disconnected them from each other. I questioned that and feel it is not how the code is detailed. 250.32 I was told the code has gone back and forth. I am going by what I read now.

So, to answer your question. The separate buildings are 4 wire (A,B,C and N. No G to the main)

If the runs from the middle building to the other are extensions of the service (upstream of service disconnect) then the four wire feed is OK but each building needs its own GES and neutral bond.
If after the disconnect there should be an EGC according to current code (and good practice even if grandfathered.)
Whether there should/may be a neutral to EGC bond at the other buildings depends on your code cycle.