Ground Fault Protection vs. Alarm Only

[Pharon]

So here's the situation -- Data Center, with a 13.8kV service feeds a main-tie-main 480V switchgear (open tie). Transformer size on both ends is 2500kVA Delta/Wye. There is ground fault detection on the 13.8kV service disconnect, but alarm only (LSIA) on the 4000A, 480V mains.

If I look at the language in 230.95 and 240.13 (NEC 2014), it clearly states that only service disconnects 1000A or more are required to have ground fault detection, and based on the design described above, it should be a Code-compliant installation, since the service disconnect is on the 13.8kV side.

But why is this the case? Aren't these 4000A mains just as susceptible to a slow-burn ground fault that would never automatically trip?

It's also been awhile since I was in the data center design industry and can't remember if this is a common practice or not.

Potential fire hazard or just paranoia?

 

[GoldDigger]

Seems to me there are also some terminology issues.
Ground detection applies to an ungrounded system and by itself only provides notification so that the fault can be repaired before a second fault results in current flow.
The terms ground fault and ground fault protection generally apply to phase faults to ground in a grounded system, which immediately carry fault current, but possibly at a low enough level that it will not trip OCPD based on phase current alone.
These generally involve some form of automatic protective action.
What mix of the two types of detection are you talking about in each case?
True ground fault detection in a grounded system (including European Residual Current Detector, RCD, as well as our GFP and GFCI) is unlikely to be alarm only.

Tapatalk!

 

[Pharon]

Yes, my apologies. I meant to say ground fault PROTECTION, not detection. These are power breakers where I would typically expect to see LSIG for the trip units. I'm not sure if it's determined by residual or zero sequence -- does that matter?

 

[steve66]

215.10 also requires ground fault protection, and would probably apply to the 4000 amp breakers, if the 480V system is solidly grounded.

If the 480 is not grounded (I'm guessing its not), then you only need the alarm. The advantage is that if one phase connects to ground, no fault current flows, and power is not lost. (Very nice for a data center.)

I'm not sure why they would have Ground Fault Protection on the 13.8KV service. Those disconnects should be less than the 1000 amp limit. (And its also above the 600 V limit for 230.95 - so its not required).

 

[Pharon]

Ah yes, I knew there was one more Code section on that somewhere. But as it only pertains to feeder OCPD's, I don't see how it would apply to the mains. Though there are 2000A distribution feeder breakers in that gear, as well. I will have to check those... But if they feed UPS distribution boards, are they considered "equipment?"

And yes, I believe the X0 is grounded at the transformer. It's a 4-wire switchgear I think, but the neutral is not used downstream.

 

[GoldDigger]

On the other hand the ground detection on an ungrounded 480 cannot be either residual or zero sequence, since no fault current will flow through the first connection to ground.
A common way to detect a grounded phase is to run a small voltage sense coil or lamp load from each phase to ground. These have a high enough impedance that it does not count as grounding the system, but tends to create a weak neutral point at ground. If any phase goes to ground (including faults to ground in the middle of a line-to-line load) it will unbalance the voltages and the voltage imbalance or lamp brightness imbalance serves as your alarm.
If you have some sort of current detector, then it is *not* an ungrounded system and the detection event is required to actually trip a disconnect.

Tapatalk!

 

[don_resqcapt19]

With the 2014 code, almost any OCPD rated 1,000 amps and more on a 277/480Y system is required to have ground fault protection. 230.95 for service OCPDs, 215.10 for feeder OCPDs and 210.13 for branch circuit OCPDs

 

[Maders]

Pharon, could you clarify whether the wye on the delta/wye system you describe above is solidly grounded or not? There seems to be some confusion.
Thanks

 

[Pharon]

Pharon, could you clarify whether the wye on the delta/wye system you describe above is solidly grounded or not? There seems to be some confusion.
Thanks

Yes, the wye is solidly grounded. The neutral is just not used.

 

[Maders]

Is the system primary or secondary metered from the utility? (e.g. metered at 13.8kV or at 480V)

 

[Pharon]

Primary metered at 13.8kV.

 

[texie]

Yes, the wye is solidly grounded. The neutral is just not used.

Based on my understanding of your description and use of the installation, I don't think you are in compliance with 230.95. I think many might consider this a pretty big safety issue.

 

[Maders]

If the installation is primary metered (as Pharon states above), I don't see how the secondary on the transformer qualifies as a service disconnect (as described in 230.95).

 

[texie]

If the installation is primary metered (as Pharon states above), I don't see how the secondary on the transformer qualifies as a service disconnect (as described in 230.95).

I think it is moot whether this is the service disconnect or not. In any case 215.10 for feeders and in the 2014 NEC 210.13 for branch circuits.

 

[Maders]

I think it is moot whether this is the service disconnect or not. In any case 215.10 for feeders and in the 2014 NEC 210.13 for branch circuits.

215.10 applies to feeders. Is the secondary OCPD of a transformer (e.g. unit substation main circuit breaker) really a "feeder disconnect" (as described in 215.10)? Also this section states "in accordance with the provisions of 230.95" which addresses services. Not trying to be difficult, just genuinely curious.

 

[texie]

215.10 applies to feeders. Is the secondary OCPD of a transformer (e.g. unit substation main circuit breaker) really a "feeder disconnect" (as described in 215.10)? Also this section states "in accordance with the provisions of 230.95" which addresses services. Not trying to be difficult, just genuinely curious.

It seems to me that anything on the load side of the transformer secondary OCPD is a feeder or a single branch circuit. In fact, it is my understanding that the reason for the addition of 210.13 for the 2014 edition was to make it clear that all circuits 1000 amps or more (assuming 150 volt or more to ground and the other required conditions) require GFPE.

 

[Pharon]

I'm not sure where you're getting that from.

 

[don_resqcapt19]

If the conductors are not service conductors or branch circuit conductors, then they are feeder conductors.

Feeder. All circuit conductors between the service equipment, the source of a separately derived system, or other power supply source and the final branch-circuit overcurrent device.

 

[Pharon]

Ah, I understand now. So hypothetically speaking, then, if I was still on the 2011 NEC, the 4000A mains would not require GF but any feeders downstream 1000A and larger would, correct? Because without being a service, and without 210.13, all that's left is the equipment protection requirement in 240.13, right?

 

[texie]

Ah, I understand now. So hypothetically speaking, then, if I was still on the 2011 NEC, the 4000A mains would not require GF but any feeders downstream 1000A and larger would, correct? Because without being a service, and without 210.13, all that's left is the equipment protection requirement in 240.13, right?

Nope, the 4000 amp mains need GFPE even under the 2011 NEC.