The POCO said there was a problem with the voltage readings on a 480 open delta service. the readings were around 480V phase to phase. the readings from phase to ground were 486,410 and 87. The POCO said the phases should be within 5% when reading to ground. They said the low reading on one of the legs indicated a ground fault. They said this low voltage reading was an early warning and that if another leg went to ground it would be a direct short. I found an electrical engineers forum in which one of the members posted that the line to ground voltages were of no significance. I was wondering if anyone has any experience with this and could maybe submit an explanatory diagram showing a short situation as mentioned so it is easy to understand.

This is an ungrounded service? My wild a guess
The grounded point seems to be in between the 410 and 87v phases and a probably ground fault in the load. I think I would start by looking for ground current in the EGCs to see if you can find the bad one and then try turning it off.

Hey Greg
Your feedback helps.
It is an ungrounded/Open Delta service that's 40 years old. That's about all I know about it and it's the first time I've run into it.
It would be so easy if we could just start flipping breakers.
It's never that easy. It's gonna be tough to get to those grounds in a hot panel. I might have to break the bad news
to them

Is this something like a glass plant where a shutdown is devastating? That is usually where you see floating delta.
I am not sure how you will isolate this but I do think you have leakage in one of the loads, pretty close to one end.
(401:87)
It might be interesting to ground the 486v phase through a fairly high value resistor, 100k ohms or so, and see what it does to your readings. If you significantly change the readings you have a high resistance fault. That would just be a calculator exercise to figure it what the resistance of the fault was.
A very high resistance fault might be able to give you this symptom without really being a problem if you had another bolted fault but if this is the first bolted fault inside a heating element or motor (no significant change with the 100k ground), you would certainly know which one it was if you had a phase to ground fault. It would be the one smoking. Hopefully the breaker would trip.

Greg
It's actually a medical center. I just had a quick chance to look at the situation today. Monday I am going to look for the obvious: rooftop equipment, fans etc. I'll be spending a few hours educating myself over the weekend. I doubt I'll be confident enough to start grounding phases with resistors though:)I visualize some little old lady with a cane chasing me down the hallway because the power went off!I love these jobs where you get to learn something new!

If there are no plans to replace this service, you should consider installing a ground monitoring system. A ground monitor would actually be required by code if this was a new installation. 250.21(B)

Tom

Thanks.
I'll make that recommendation to the property owners.

I am not sure what a monitor would show you in this situation. It is expecting one phase to be grounded in a fault, not a ground somewhere in the middle between 2 phases.
87v might even be enough to tell the monitor things were OK.
The old school monitor was just 3 light bulbs, one per phase. If one was out, the phase was grounded. I assume they are more sophisticated these days.

In an ungrounded system the Line-Ground voltage readings are dependent upon the capacitive coupling of each phase conductor to ground. In a perfectly balance system the (3) L-G voltages should all be about 277V on a nominal 480V. If the loads are not balanced then the L-G voltages will move away from the ideal 277V. L-G current (i.e. from a ground fault) will also cause a shift. The extreme case being L-G = 0V on a solidly grounded phase.

Part of the problem you face, is not knowing what your base L-G voltages have been.

Ground detectors are not usually much fancier than the 3-light system, as Greg mentioned. I prefer to get a detector that includes a horn or at least alarm contacts.

Luckily the utility had documented the voltages in April.
They just happened to return recently for another issue and measured the voltages again. This is how we became aware that there may be a problem. Today we isolated the motor that was causing the voltage imbalance. What seemed strange was that there were 2 identical compressor motors. One was running very hot. Hot enough to discolor the plastic label on the motor. The other motor was luke warm so you could put your hand on it. Right away we thought it must be the motor that is running hot. After shutting them down individually it turned out to be the cooler motor that was probably causing the problem. We were told the hotter motor was running within temperature specs.
Next steps:
Meg the motor, check the motor connections and the contactor contacts. Anyone want to throw in their 2 cents about motor megging?

The winding between the 87 and 401v legs is shorted to the case is my bet. I bet the ohm readings are just about the same proportions too.

Greg,

The ground monitor would not help in this situation, since it is already known that there is a problem.

However, in the future, immediate attention could be given to the first ground fault. Ignoring the first ground fault can lead to more serious problems when a second ground fault occurs on another phase. These faults can have a fairly high impedance so OCPD's may not trip and fires can occur.

There is a listed product on the market that has the 3 lights and there are more sophisticated products which provide relay outputs. I, too, prefer an audible alarm since the building occupants may not realize the dangers involved. A light being out is easily ignored, but a horn sounding is not (of course, they might fill the horn with shaving cream to quiet it down).

For those interested, here's a pair of sites to visit
http://www.bender.org/irdh.aspx
http://www.atcdiversified.com/Diversified/GFD_series.htm

" (of course, they might fill the horn with shaving cream to quiet it down)."

Round here it would usually be filled with a ball peen hammer

Most common here is "Cut the red wire" (pool door alarms).

Good call gentlemen!!! The motor in question, ran nice and cool and sounded normal. However....a meg test revealed it to be faulty.
Thanks a lot!!