Supply-Side Connection Bonding

Status
Not open for further replies.
C

c_picard

Senior Member
Location
USA
The topic of bonding the grounded conductor at the disconnect for a PV supply side connection has come up so frequently, I had to put together an FAQ document for my inspection coordinators to refer to when getting push back from inspectors. Somebody, somewhere, started the rumor that this disconnect should not be treated as a service disconnect.

The code is fairly explicit in describing the conductors that feed this enclosure as a "set" of service entrance conductors.
(230.40, exception 5). I really don't see where the confusion lies. Yet, another inspector is asking that we remove the main bonding jumper and isolate the neutral as in a subpanel, a violation of 250.24(B), 250.28.

Other relevant articles:
Art 100 - Definitions, see Service,Service Conductors, and Service Equipment.
705.12(A)
230.82(6)

Not one of these inspectors has cited a code article that prohibits the bonding of the neutral, they all parrot the same line..."it's not a service". Correct, it is not a seperate service, it is a service disconnect, and the neutral is required to be bonded.

Can one of you kind folks let me know if I'm way off base here? And if you are going to argue that it's not a service disconnect, please reference a code artricle.
 
Follow up question:

Does one need to be an NFPA member to access any archived formal/informal interpretations?

John Wiles mentions an informal interpretation on this topic in an article he wrote for IAEI (Jan2010).
 
Thanks jaggedben.

I've been digging deeper into what's been written on this topic, and I am seeing a few different arguments being made. Even the man himself, Mike Holt has an opinion that I disagree with. Here in SolarPro

http://solarprofessional.com/articl...ec-section-70512-and-utility-interconnections

"My position is that a PV system disconnecting means is not a service disconnect because the inverter?s output circuit is not a utility service, and the utility service is not affected when the disconnect is in the Off position."

The output circuit is not a service, agreed. Note that PV is connected to the LOAD side terminals. My logic is that Service Entrance Conductors terminate in a Service Disconnect. Period. What happens when you shut off the main breaker in a panelboard? Is the utility affected?
 
c_picard said:
Thanks jaggedben.

I've been digging deeper into what's been written on this topic, and I am seeing a few different arguments being made. Even the man himself, Mike Holt has an opinion that I disagree with. Here in SolarPro

http://solarprofessional.com/articl...ec-section-70512-and-utility-interconnections
Click to expand...

Note that even though Mike doesn't think a PV disconnect is a service disconnect by definition, he still says that "PV system disconnects are service rated, contain overcurrent protection and must be installed in a similar fashion to service disconnects, but PV system disconnects are not service disconnects." The point Mike is making is about the 6 handle rule, not the subject of your question. I think Mike would agree that a PV disconnect for a supply side connection should have a N-G bond.

EDIT: In fact, he makes this point explicitly in one of the graphics accompanying the article.
 
Last edited:
Exactly, I saw that graphic. I left a voicemail for the AHJ to call me back. In true Jersey style, he won't reference a code article.

I offered to send him published articles substantiating the need to bond the Grounded conductor. Inspectors do not often use email as part of their workflow, however. Leave a message, they might call you back between 4-4:30 on the second tuesday of the month. Meanwhile, the customer gets stuck waiting for permission to operate.
 
c_picard said:
The topic of bonding the grounded conductor at the disconnect for a PV supply side connection has come up so frequently, I had to put together an FAQ document for my inspection coordinators to refer to when getting push back from inspectors. Somebody, somewhere, started the rumor that this disconnect should not be treated as a service disconnect.

The code is fairly explicit in describing the conductors that feed this enclosure as a "set" of service entrance conductors.
(230.40, exception 5). I really don't see where the confusion lies. Yet, another inspector is asking that we remove the main bonding jumper and isolate the neutral as in a subpanel, a violation of 250.24(B), 250.28.

Other relevant articles:
Art 100 - Definitions, see Service,Service Conductors, and Service Equipment.
705.12(A)
230.82(6)

Not one of these inspectors has cited a code article that prohibits the bonding of the neutral, they all parrot the same line..."it's not a service". Correct, it is not a seperate service, it is a service disconnect, and the neutral is required to be bonded.

Can one of you kind folks let me know if I'm way off base here? And if you are going to argue that it's not a service disconnect, please reference a code artricle.
Click to expand...

I don't think this is even open for debate. To not bond the neutral would be a violation of 250.24(C), and I might add, create a serious safety hazard. Just imagine what the fault return path would be in the event one of ungrounded conductors from the utility were to short to the enclosure of the disconnect. Even worse, what if the installer chose to use a PVC nipple to supply the disconnect? Just imagine that fault return path!
I realize you have a neutral brought to the service but that is only part of 250.24(C). It also says that you must have a main bonding jumper.
So, yes, I think you are 100% correct. In my view to not do so is a disaster waiting to happen.
 
Last edited:
texie said:
I don't think this is even open for debate. To not bond the neutral would be a violation of 250.24(C), and I might add, create a serious safety hazard. Just imagine what the fault return path would be in the event one of ungrounded conductors from the utility were to short to the enclosure of the disconnect. Even worse, what if the installer chose to use a PVC nipple to supply the disconnect? Just imagine that fault return path!
I realize you have a neutral brought to the service but that is only part of 250.24(C). It also says that you must have a main bonding jumper.
So, yes, I think you are 100% correct. In my view to not do so is a disaster waiting to happen.
Click to expand...

The disco could be bonded with an EGC brought from the non-PV service disco instead, and then those concerns would essentially go away. But yes, you are absolutely correct.
 
c_picard said:
Correct, it is not a seperate service, it is a service disconnect, and the neutral is required to be bonded.
Click to expand...

I will have to agree with that statement...and as a new additional service disconnect to the existing service, it must have the grounded conductor brought to the service disconnect and bonded to the enclosure. I will also add that that the new added disconnect must be GROUPED with the other existing service disconnects.

shortcircuit
 
Does not need to be grouped, though it would make sense. The code does not require it. Each SET of no more than 6 needs to be grouped. There actually can't be more than 6 in any one location, so they cannot be grouped. If they are remotely located, a directory must be provided.

Update on original post: This has developed into a lively discussion over what is and is not a service. Pretty awesome that the head of the Building dept is at least willing to have the conversation. The road block seems to lie in a misunderstanding over two key definitions: "service", and "inverter output connection"

AHJ holds that PV is not a "service", so we shouldn't even be talking about article 230 at all. I was able to get us into that article via 705.12, but didn't make much headway. Citing 690.14, which in his view implies that the AC disconnect does not have to be rated as service eqipment, therefore 230 doesn't apply.

His interpretation of the inverter output circuit definition includes the AC disconnect that is between the inverter terminals and the point where the conductors "tap" the existing service entrance conductors, therefore 230 does not come into play.

If anyone would like to volunteer for a peer review, I am going to compile some articles and try to paint as clear a picture as possible. I'd love to get your feedback.
 
c_picard said:
Does not need to be grouped, though it would make sense. The code does not require it. Each SET of no more than 6 needs to be grouped. There actually can't be more than 6 in any one location, so they cannot be grouped. If they are remotely located, a directory must be provided.

Update on original post: This has developed into a lively discussion over what is and is not a service. Pretty awesome that the head of the Building dept is at least willing to have the conversation. The road block seems to lie in a misunderstanding over two key definitions: "service", and "inverter output connection"

AHJ holds that PV is not a "service", so we shouldn't even be talking about article 230 at all. I was able to get us into that article via 705.12, but didn't make much headway. Citing 690.14, which in his view implies that the AC disconnect does not have to be rated as service eqipment, therefore 230 doesn't apply.

His interpretation of the inverter output circuit definition includes the AC disconnect that is between the inverter terminals and the point where the conductors "tap" the existing service entrance conductors, therefore 230 does not come into play.

If anyone would like to volunteer for a peer review, I am going to compile some articles and try to paint as clear a picture as possible. I'd love to get your feedback.
Click to expand...
FWIW, I emailed the lead engineer for my local AHJ (Austin Energy) and asked her for her opinion on this point; I have not heard back from her yet. This is a salient point for me as I am in the process of building 11 commercial PV systems in her territory, two of which are supply side taps. I will post her response here.
 
Last edited:
ggunn, one point that he brought up that may apply to your commercial projects is GFPE for greater than 1000A.

He said he just had a case where the contractor was not installing GFPE on the premise that PV is not a service, so 240.13 would not apply. IMHO, GFPE is not required, but that is not the reason. Part of his argument is that the PV industry can't have it's cake and eat it, too. Based on a false premise, though.
 
c_picard said:
ggunn, one point that he brought up that may apply to your commercial projects is GFPE for greater than 1000A.

He said he just had a case where the contractor was not installing GFPE on the premise that PV is not a service, so 240.13 would not apply. IMHO, GFPE is not required, but that is not the reason. Part of his argument is that the PV industry can't have it's cake and eat it, too. Based on a false premise, though.
Click to expand...
None of my systems are >1000A.
 
c_picard said:
ggunn, one point that he brought up that may apply to your commercial projects is GFPE for greater than 1000A.

He said he just had a case where the contractor was not installing GFPE on the premise that PV is not a service, so 240.13 would not apply. IMHO, GFPE is not required, but that is not the reason. Part of his argument is that the PV industry can't have it's cake and eat it, too. Based on a false premise, though.
Click to expand...

PV is not a service, but it may be connected directly to its own service. If a new service were installed just for a PV production plant, per Art. 230.2(A)(5), would this inspector buy the same argument? I would think not. And if not, then why buy the same argument regarding 230.40 Exception 5?

My thoughts are this: As an installer looking to save money, I would have no problem with being exempt from all the requirements of 230. But on the other hand, as an installer wishing for consistent interpretation of the NEC from different AHJs, I prefer to look to 230 as the authority. And finally, as a rational person reading the code and considering electrical safety, I think his point of view is nonsensical.

As for Mike Holt's argument that PV isn't a service because it doesn't meet the article 100 definition, well, that isn't logically consistent with 230.2(A)(5) either. Something in the code needs to be fixed.
 
Could not agree more.

In the past few days, I have read every easily-available article published on this subject. Even in the case of Mike's writing, where his logic is screwy, the conclusion is the same: the grounded conductor MUST be bonded at this enclosure to ensure the highest levels of safety.

I've also been assembling some examples to illustrate the conflicting logic. If we consider everything between the inverter output terminals and the IPC's at the service conductors the "inverter output circuit", how do we size the circuit conductors? For example: Inverter nameplate output current 20A > OCPD required 25A > we would design for 10AWG conductors.

If it's all the same circuit, I can tap with 10AWG??? I bet New Jersey requires a minimum 60A switch!!! (which I agree with)

Many thanks for being the sounding board. And for the trolling NJ AHJ's, I mean no disrespect. I have kids, and it helps me sleep at night knowing that I'm doing everything possible to make sure we are installing safe and code-compliant PV systems wherever somebody else's kids live.
 
c_picard said:
Could not agree more.

In the past few days, I have read every easily-available article published on this subject. Even in the case of Mike's writing, where his logic is screwy, the conclusion is the same: the grounded conductor MUST be bonded at this enclosure to ensure the highest levels of safety.

I've also been assembling some examples to illustrate the conflicting logic. If we consider everything between the inverter output terminals and the IPC's at the service conductors the "inverter output circuit", how do we size the circuit conductors? For example: Inverter nameplate output current 20A > OCPD required 25A > we would design for 10AWG conductors.

If it's all the same circuit, I can tap with 10AWG??? I bet New Jersey requires a minimum 60A switch!!! (which I agree with)

Many thanks for being the sounding board. And for the trolling NJ AHJ's, I mean no disrespect. I have kids, and it helps me sleep at night knowing that I'm doing everything possible to make sure we are installing safe and code-compliant PV systems wherever somebody else's kids live.
Click to expand...

Code issues aside, I am a bit confused as to why it would be a safety issue if the ground and neutral were not bonded in the PV disco. The neutral in a supply side tap has continuity with the neutral at the meter and therefore with the neutral in the building MDP. The same is true for the ground. The neutral and the ground are connected at the service, so anywhere you check, there is continuity between neutral and ground, and neutral is never switched. Where is the hazard?
 
I see your point.

If we map out the fault current, and do the math, maybe it won't effect the ability to safely clear a fault on the load side of the OCPD all that much. I was running under the assumption that the strict requirements in place for services address the need to direct line side faults back to the source with the least resistance possible, by creating a reliable bond between the enclosure and the grounded conductor. I stand corrected if that's not the case.

If this isn't a service enclosure, then I don't need the GEC, which means I'm asking the 10AWG equipment ground to carry the fault back to the main panel. I also can install a 30A pullout A/C disconnect from home depot, tap using 10AWG romex, and call it a day.

Do you have any thoughts about why the GEC is required to bond the grounded conductor at each enclosure when there are multiple disconnects, like in an apartment buildings meter bank? Same principle, and I thought clearing faults was the driver for that requirement. I'll dig into that a little deeper also.

Thanks
 
the shorter answer/question: is there a low impedance connection between the grounded conductor and the enclosure?
 
c_picard said:
I see your point.

If we map out the fault current, and do the math, maybe it won't effect the ability to safely clear a fault on the load side of the OCPD all that much. I was running under the assumption that the strict requirements in place for services address the need to direct line side faults back to the source with the least resistance possible, by creating a reliable bond between the enclosure and the grounded conductor. I stand corrected if that's not the case.

If this isn't a service enclosure, then I don't need the GEC, which means I'm asking the 10AWG equipment ground to carry the fault back to the main panel. I also can install a 30A pullout A/C disconnect from home depot, tap using 10AWG romex, and call it a day.

Do you have any thoughts about why the GEC is required to bond the grounded conductor at each enclosure when there are multiple disconnects, like in an apartment buildings meter bank? Same principle, and I thought clearing faults was the driver for that requirement. I'll dig into that a little deeper also.

Thanks
Click to expand...
Understand as well that a fault can only be fed from the service side. An inverter is incapable of sourcing enough current to endanger the conductors and will shut down if a fault occurs between it and the service.
 
Status
Not open for further replies.
Top