Ground Fault Protection

[ggunn]

A source of some confusion for me is the idea of GFP for inverters interconnected on the load side of a service entrance OCPD.

705.32 says: Where protection is installed in accordance with 230.95 the output of an interactive system shall be connected to the supply side of the ground-fault protection.

Exception: Connection shall be permitted to the load side of ground-fault protection if there is ground fault protection for equipment from all ground-fault current sources.

One consultant tells me that that means we need GFP on our load side interconnection breaker and another tells me that the SMA inverters we are using have GFP, so we do not. The data sheet for the inverter says it has ground fault monitoring: Riso/Differential current, but I am not sure that it means that it has GFP on the AC side so 705.32 is covered; the line on the data sheet does not say if the ground fault monitoring is on the AC side, the DC side, or both.

What say you?

 

[don_resqcapt19]

Is there an OCPD between the inverter output and the service conductors? If so, I will read 230.95 as requiring the ground fault detection to open that OCPD.

 

[ggunn]

Is there an OCPD between the inverter output and the service conductors? If so, I will read 230.95 as requiring the ground fault detection to open that OCPD.

I may not have made myself clear. My question is this: When interconnecting PV via a backfed breaker in a panel on the load side of OCPD that has GFP, does the interconnection breaker need to have GFP?

705.32 appears to say that I do, but if it does, does GFP integral to the inverter(s) (specifically the SMA Core 1 series) satisfy the requirement?

 

[jaggedben]

I'd say that GFP on the DC side of the inverter does not count.

 

[BBEE]

I don't believe a GFP breaker is required if it's built into the inverter. All the systems that I've seen that have interconnected on the load side of a GFP main breaker, have used a normal breaker.

Sent from my Pixel 4a (5G) using Tapatalk

 

[BBEE]

I don't believe a GFP breaker is required if it's built into the inverter. All the systems that I've seen that have interconnected on the load side of a GFP main breaker, have used a normal breaker.

Sent from my Pixel 4a (5G) using Tapatalk

Also, in reading the exception to 705.32 it says you can connect on the load side of GFP if there is ground fault protection for equipment from all ground fault current sources. My interpretation of that would be the main GFP breaker would protect from utility side and the GFP inside the inverter would protect from the PV side.

Sent from my Pixel 4a (5G) using Tapatalk

 

[jaggedben]

Also, in reading the exception to 705.32 it says you can connect on the load side of GFP if there is ground fault protection for equipment from all ground fault current sources. My interpretation of that would be the main GFP breaker would protect from utility side and the GFP inside the inverter would protect from the PV side.

Sent from my Pixel 4a (5G) using Tapatalk

Can you provide any documentation that the inverter has any GFP that detects faults on the AC side? I'm suspicious that your premise here is false.

 

[jaggedben]

Is there an OCPD between the inverter output and the service conductors? If so, I will read 230.95 as requiring the ground fault detection to open that OCPD.

I'd say that might be one way to meet the code requirement, but it also may not work, and the code language doesn't require it per se.

It seems to me the key question about your method is whether the grid-side GFP is assured to properly detect a ground fault in a backfeeding situation. If it does, then opening the grid side breaker will cause the interactive inverter shut down, effectively fulfilling the exception's requirement. But if it doesn't, then your solution simply doesn't function?

 

[BBEE]

Can you provide any documentation that the inverter has any GFP that detects faults on the AC side? I'm suspicious that your premise here is false.

I'd like to understand this better if you have a different interpretation of current sources. I only see two of them in this scenario. One from the utility and one from the PV. The exception doesn't say whether the GFP needs to be on the DC or AC side of the inverter. That's why I'm thinking an additional GFP isn't required on the AC side.

I've never had an inspector call us on this. But it doesn't come up very often since it's mainly supply-side connections in a switchboard.




Sent from my Pixel 4a (5G) using Tapatalk

 

[jaggedben]

I'd like to understand this better if you have a different interpretation of current sources. I only see two of them in this scenario. One from the utility and one from the PV. The exception doesn't say whether the GFP needs to be on the DC or AC side of the inverter. That's why I'm thinking an additional GFP isn't required on the AC side.

Very simply, DC side ground fault detection cannot detect a ground fault on the AC side. (That is the one thing in this discussion that I'm pretty certain of, I'll defer to others on the rest, generally.) Therefore if there is a ground fault in some a piece of equipment - let's say a load, or a piece of distribution equipment in between the inverter and the GFP - there is no independent GFP from the inverter as a source of current.

You have a point that the 'for equipment' language of the exception is quite vauge as to which equipment and where it might be located. However, ground fault detection requirements for the DC side of things are pretty clear in 690. And this is article 705. That strongly implies we're talking about the AC output current of inverters and equipment on the AC side.

I've never had an inspector call us on this. But it doesn't come up very often since it's mainly supply-side connections in a switchboard.

Perhaps the inspectors aren't aware of the code requirement, or don't understand it better than those of us here. This is relatively rare, as you state, and also rather esoteric and into-the-weeds of how this stuff works.

 

[ggunn]

Perhaps the inspectors aren't aware of the code requirement, or don't understand it better than those of us here. This is relatively rare, as you state, and also rather esoteric and into-the-weeds of how this stuff works.

Connecting a commercial PV system on the load side of a panel or switchboard that has an OCPD with GFP is not rare in my world; this is not an academic question for me. Ever since our consultant recommended that we specify GFP on backfed breakers in such a situation I have been doing so, but an outside design house we are using for some projects is pushing back on it.

 

[GoldDigger]

Connecting a commercial PV system on the load side of a panel or switchboard that has an OCPD with GFP is not rare in my world; this is not an academic question for me. Ever since our consultant recommended that we specify GFP on backfed breakers in such a situation I have been doing so, but an outside design house we are using for some projects is pushing back on it.

My only concern is that the breaker may have line and load sides identified, in which case it would not be permitted for backfeeding even if the source of the backfeed is a line interactive inverter and the GF function would still work properly.
If identified with LINE and LOAD, you may need a document from the manufacturer stating that it is suitable for backfeeding despite that labelling.

 

[PWDickerson]

I have run into this twice, and in both instances we installed a separate GFP device between the backfeed breaker and the inverter.

As JB has stated, solar inverters have ground fault protection on their DC side, but not on their AC side, so the built-in GFP in the inverter cannot be used to satisfy the 705.32 exception.

Seems to me that if a ground fault were to occur, the main breaker would trip, which would cause the solar inverter to shut down, and it wouldn't be a source of current into the fault, and so should be excepted from the requirement.

I am curious to see the solutions presented by others that routinely design larger systems and so run into this problem frequently.

 

[pv_n00b]

Yes, the inverter OCPD requires GFP if the main OCPD has GFP. It's not that big a deal, unless you are using smaller CBs where GFP is just not available. In that case, it falls under NEC 90.4 and the AHJ can make an allowance for equipment that is not available.

 

[ggunn]

I have run into this twice, and in both instances we installed a separate GFP device between the backfeed breaker and the inverter.

As JB has stated, solar inverters have ground fault protection on their DC side, but not on their AC side, so the built-in GFP in the inverter cannot be used to satisfy the 705.32 exception.

Seems to me that if a ground fault were to occur, the main breaker would trip, which would cause the solar inverter to shut down, and it wouldn't be a source of current into the fault, and so should be excepted from the requirement.

I am curious to see the solutions presented by others that routinely design larger systems and so run into this problem frequently.

Here is the respnse we got from SMA:

"...our CORE1 inverters (and probably most non-isolated inverters) detect ground faults by differential current that arises whether the fault is on the DC or AC side of the inverter. This should meet the intent of 705.32 that generation equipment connected on the load side of a GFP device should itself have ground fault protection to prevent it from feeding current into a ground fault."

 

[pv_n00b]

Someone at SMA proves again that they have never talked to a skeptical AHJ. If their opening line with the AHJ is "This should meet the intent of..." they have already lost.

 

[ryan_mayfield]

I was involved in one of the original conversations that preceded this thread and thought I'd update those interested in what I found since it wasn't necessarily easy info to gather.

Per information from SMA and in their attached block diagram, the differential current measurement used to detect ground faults happens in the AC power unit (the white box in the diagram), not the dc side. This is why the inverter will also see AC ground faults. To add to that, a somewhat dated paper from the Solar ABCs (bottom of page 13) talks more about this ground fault detection method:

"When the inverter is operating, Iso measurements cannot be taken because of the continually changing voltage reference on each pole to ground. Therefore, a residual current measurement is made, generally on the AC side, to look for any imbalance in current going in versus current going out. This will detect faults within the inverter as well as ground faults in the DC system. Because the fault current is sourced on the AC side, the residual current detection can be highly sensitive to ground faults. This two-pronged approach to identifying faults both before and during operation is one of the more robust GFDI detection methods available for any system topology."

From this, my conclusion is this meets the 705.32 Exception for the GFP requirement when interconnecting on the load side of a GFP main since the inverter is providing protection from the PV source and the main breaker is providing protection from the utility source.

 

[jaggedben]

I really think that ultimately misses the point. That current differential detection still only detects ground faults on the DC side. I really think that protecting a portion of the inverter AC internals is irrelevant to this discussion, since the code essentially considers the entire inverter to be a black-box piece of equipment, and moreover that is a far less likely location for a ground fault than field wiring. Ok, the inverter protects itself from a ground fault sourced from the DC side. Whoop de do. Unfortunately the code exception is vague about which equipment requires protection, and doesn't limit that merely to the solar system equipment.

Consider what would actually happen if the inverter is backfeeding a feeder with GFP and one of the hots faults to ground, but not hard enough to trip overcurrent devices. The current will take the route from hot to ground, then to the main (or system) bonding jumper, then the neutral to the system transformer, and then from the transformer back to the inverter on one of the other hot legs. From the inverter's point of view the ground fault is no different than a line-to-neutral load. Moreover, a GFP breaker on the inverter circuit won't notice anything either, as far as I can tell, if the fault is on its supply (utility) side. So I don't see what good that does.

It seems to me that the GFP at the supply (utility) end of the feeder still has a opportunity to detect such a fault, although I honestly have no idea if it will do so properly. It seems to me that this is the crux of the issue. If the feeder GFP won't operate properly when backfed, I struggle to see how the exception can be met. Period.

 

[synchro]

I would think that a 3-phase GFP breaker would need to be equally sensitive at any phase angle of the common-mode current that it is sensing. That's because combinations of L1-G, L2-G, and L3-G fault currents can produce a common-mode current at any arbitrary angle over 0-360° relative to the line voltages (even if the faults are all resistive, all capacitive, etc.) Backfeeding a breaker would create a 180° shift of the current vs. voltage, but that shouldn't matter with respect to sensing common-mode current when the GFP already works over 0-360°.

Perhaps single-phase breakers would also be insensitive to the phase of the common mode current, because synchronous detection (for example with a mixer) would be needed to respond to phase shifts. And that would add some unnecessary cost and complexity for no reason.