Supply-side vs Load-side PV Interconnection

[Smart $]

I am not going to waste time on that debate, it is meaningless to me. Electricity does not care what the NFPA decides to call things.



For me there is no debate, safety demands that an enclosure containing unfused service conductors requires reliable bonding back to the source and an EGC sized per 250.122 is not generally up to that task.

I agree.

I'd rather see a main bonding jumper than an EGC.

 

[ggunn]

What is the overcurrent protection on the line side of that disco? There is none (or rather, it's whatever is on the utility transformer).

The available fault current on each side of that disco is NOT the same. At the very least, the 'EGC' on the line side of the disco should be sized for 250.66 and not 250.122.

I ran a few test cases and found little difference between EGC (250.122) and GEC (250.66) sizing for the same tap. Sometimes the GEC was one size larger, but sometimes the EGC was larger. Note that 250.66 references the size of the largest CCC from the tap to the disco (according to exhibit 250.30 in the 2011 handbook). If it is a small tap then the GEC is small as well irrespective of how much current the utility could feed to a fault.

Unless I am misreading it, of course.

 

[Smart $]

I ran a few test cases and found little difference between EGC (250.122) and GEC (250.66) sizing for the same tap. Sometimes the GEC was one size larger, but sometimes the EGC was larger. Note that 250.66 references the size of the largest CCC from the tap to the disco (according to exhibit 250.30 in the 2011 handbook). If it is a small tap then the GEC is small as well irrespective of how much current the utility could feed to a fault.

Unless I am misreading it, of course.

If we were talking feeder taps, the tap EGC minimum size is either same as the largest ungrounded tap conductor or sized per 250.122 based on the feeder OCPD rating, whichever is smaller.

IOW, tap EGC sizing is not based on the OCPD on the load end of the tap.

 

[iwire]

I ran a few test cases and found little difference between EGC (250.122) and GEC (250.66) sizing for the same tap. Sometimes the GEC was one size larger, but sometimes the EGC was larger. Note that 250.66 references the size of the largest CCC from the tap to the disco (according to exhibit 250.30 in the 2011 handbook). If it is a small tap then the GEC is small as well irrespective of how much current the utility could feed to a fault.

Unless I am misreading it, of course.

First I submit it is impossible to use table 250.122 at all because there is no OCPD ahead of this 'tap' to base it on. You do not base the size of a EGC on the over-current device it supplies.


The bonding jumper needs to be sized based on the size of the service conductors supplying the PV / service disconnect. Also the raceways, if any that contain these service conductors require bonding per service conductor sizing just like any other service raceway.

 

[ggunn]

If we were talking feeder taps, the tap EGC minimum size is either same as the largest ungrounded tap conductor or sized per 250.122 based on the feeder OCPD rating, whichever is smaller.

IOW, tap EGC sizing is not based on the OCPD on the load end of the tap.

With PV, which is the load end? Current flows one way under normal conditions and the other in the event of a fault. And according to the AHJ where I am, a supply side interconnection is not a tap, anyway.

But in any case we aren't talking about feeder taps; supply side PV interconnections are nearly always ahead of the MDP main breaker.

 

[jaggedben]

With PV, which is the load end? Current flows one way under normal conditions and the other in the event of a fault.

And it is the fault we are worried about with respect to an effective ground fault pathway.

 

[Smart $]

With PV, which is the load end? Current flows one way under normal conditions and the other in the event of a fault. And according to the AHJ where I am, a supply side interconnection is not a tap, anyway.

But in any case we aren't talking about feeder taps; supply side PV interconnections are nearly always ahead of the MDP main breaker.

As jaggedben said. Besides, grounding on the PV side is set in stone, so to speak.

If you run service conductors to the PV system disconnect that have an ampacity less than the service rating, they are taps... just not Article 240 taps.

I know we aren't talking feeder taps. I just used it as an example to show grounding conductor size is based on the the feeder OCPD... that is, based on what's on the supply side, not what's on the load side.

 

[ggunn]

As jaggedben said. Besides, grounding on the PV side is set in stone, so to speak.

If you run service conductors to the PV system disconnect that have an ampacity less than the service rating, they are taps... just not Article 240 taps.

I know we aren't talking feeder taps. I just used it as an example to show grounding conductor size is based on the the feeder OCPD... that is, based on what's on the supply side, not what's on the load side.

I get that, but looking at Exhibit 250.30 in the 2011 NEC Handbook, it shows two taps in parallel on a service drop, one with #3 current carrying conductors and one with 3/0 CCC's. The ground wire is #8 for the #3 tap and #4 for the 3/0 tap. I read that to mean that the size of the ground wire is dictated by the size of the tap conductors using Table 250.66, not by anything to do with the available fault current from the service drop. In the test cases I ran yesterday I looked at them two ways - one calculating ground wire size using Table 250.122 and the AC disco OCPD rating, and the other using the tap CCC wire sizes and Table 250.66 - and I found the results to be very similar. Am I missing something?

I talked about this yesterday with another commercial PV systems designer who has a lot more experience than I and he said, paraphrased, that there is more than one way to build a safe system.

 

[Smart $]

Yes there is a disparity in minimum grounding/grounded conductor sizing between service taps and feeder taps. And while sizing based on load end ocpd may be quite similar, grounding/grounded conductor sizing on the line side must be per requirements for service conductors.

 

[ggunn]

Yes there is a disparity in minimum grounding/grounded conductor sizing between service taps and feeder taps. And while sizing based on load end ocpd may be quite similar, grounding/grounded conductor sizing on the line side must be per requirements for service conductors.

I do not disagree in principle, but what are those requirements if they are more than what is in Table 250.66? As I said, whether you consider it an EGC sized to the OCPD of the disco via Table 250.122 or a GEC sized according to Table 250.66, there is little if any difference, and it looks to me that the differences when they show up are due to the greater granularity of Table 250.122.

I realize that I have strayed from the original question of whether the neutral and grounding conductor should be bonded in the disco (from a physical standpoint irrespective of the NEC for the moment), but it is germane in the sense that it speaks to the point of whether or not the grounding conductor is large enough to handle the fault current. It looks to me that it makes little to no difference whether you consider it to be an EGC sized to the OCPD in the disco or a GEC sized to the CCC's of the tap.

It seems to me that if you have a small tap relative to the size of the service with a grounding conductor sized to 250.66, and a phase to ground fault occurs between the tap and the disco, the tap conductors are in jeopardy irrespective of whether or not the neutral is bonded to the grounding conductor in the disco. Is the point that such a connection would enlist the participation of the neutral in handling the fault current?

 

[Smart $]

I do not disagree in principle, but what are those requirements if they are more than what is in Table 250.66? As I said, whether you consider it an EGC sized to the OCPD of the disco via Table 250.122 or a GEC sized according to Table 250.66, there is little if any difference, and it looks to me that the differences when they show up are due to the greater granularity of Table 250.122.

I understand. No matter what, you end up using one table or the other. Stands to reason both are based on potential fault current amount so they should darn near identical. I often wonder why their not consolidated so to speak, and just base the grounding (and minimum grounded) on largest ungrounded wire size and be done with it.

I realize that I have strayed from the original question of whether the neutral and grounding conductor should be bonded in the disco (from a physical standpoint irrespective of the NEC for the moment), but it is germane in the sense that it speaks to the point of whether or not the grounding conductor is large enough to handle the fault current. It looks to me that it makes little to no difference whether you consider it to be an EGC sized to the OCPD in the disco or a GEC sized to the CCC's of the tap.

If you realize it makes little to no difference why question it? I see some issues with how it is, and I realize there is an avenue for public input. The hard part is getting others to get on board...

It seems to me that if you have a small tap relative to the size of the service with a grounding conductor sized to 250.66, and a phase to ground fault occurs between the tap and the disco, the tap conductors are in jeopardy irrespective of whether or not the neutral is bonded to the grounding conductor in the disco. Is the point that such a connection would enlist the participation of the neutral in handling the fault current?

The thing about this is we're talking about service entrance conductors. On the service side of the disconnect, the [grounded] neutral and metallic parts bonded thereto are the main path for ground fault current. In this situation, we know there is another solid path through the other service conductor set(s) having a neutral to ground bond... but there is no guarantee the other path is large enough. If there is the line-to-ground fault mentioned and the neutral is not bonded to ground in the disconnect, the other paths back to the source become the main path.

 

[iwire]

Nice post Smart $.


Can I side track just a bit?

Lets assume we do have a well bonded fault path for these service conductors and a hard ground fault happens.

What happens next? My own home is overhead service supplied by a pole transformer about 400' away which also supplies about 5-8 homes. I don't expect the transformer to open the primary. My best guess is the drop between my home and the midspan tap it runs to would fuse open?

Just curious what is expected to happen. It seems the damage will be confined to outside the home which is a plus.

 

[ggunn]

Nice post Smart $.


Can I side track just a bit?

Lets assume we do have a well bonded fault path for these service conductors and a hard ground fault happens.

What happens next? My own home is overhead service supplied by a pole transformer about 400' away which also supplies about 5-8 homes. I don't expect the transformer to open the primary. My best guess is the drop between my home and the midspan tap it runs to would fuse open?

Just curious what is expected to happen. It seems the damage will be confined to outside the home which is a plus.

That has been a concern to me with line side PV taps to a service where the tap is an order of magnitude or more smaller than the service. I don't see where even tying the neutral to ground in the disco would keep the conductors from being a fuse in the event of a fault between the tap and the disco.

 

[Smart $]

Nice post Smart $.


Can I side track just a bit?

Lets assume we do have a well bonded fault path for these service conductors and a hard ground fault happens.

What happens next? My own home is overhead service supplied by a pole transformer about 400' away which also supplies about 5-8 homes. I don't expect the transformer to open the primary. My best guess is the drop between my home and the midspan tap it runs to would fuse open?

Just curious what is expected to happen. It seems the damage will be confined to outside the home which is a plus.

That's what I'd expect to happen if the primary don't open.

And hopefully the weakest point isn't immediate to the house. An added benefit to POCO running smaller conductors.

 

[iwire]

And hopefully the weakest point isn't immediate to the house.

Pfft, I have asbestos siding, bring it on.

An added benefit to POCO running smaller conductors.

Maybe even a consideration made in the past?

 

[iwire]

That has been a concern to me with line side PV taps to a service where the tap is an order of magnitude or more smaller than the service. I don't see where even tying the neutral to ground in the disco would keep the conductors from being a fuse in the event of a fault between the tap and the disco.

I have to agree, it would be the weakest point in the circuit.

 

[Smart $]

That has been a concern to me with line side PV taps to a service where the tap is an order of magnitude or more smaller than the service. I don't see where even tying the neutral to ground in the disco would keep the conductors from being a fuse in the event of a fault between the tap and the disco.

The neutral to ground bond is mostly for the load side. But without this bond, a line to ground fault in the disco has no direct return path to the source. The other indirect paths may not be sufficient for the line wires to fuse, possibly resulting in voltage gradients on exposed metal parts around the system until the problem is remedied.

 

[ggunn]

The neutral to ground bond is mostly for the load side. But without this bond, a line to ground fault in the disco has no direct return path to the source.

But it does. I have always run the EGC from the PV system back through (and bonded to) the disco to the source. I have always sized it for the OCPD of the disco, but as I have found if I had sized it as a GEC relative the largest CCC in the tap it would have made little if any difference to its size. Whether one considers it an EGC or a GEC, in my designs that wire is always there.

If the issue is that with the neutral to ground bond in the disco the neutral provides a parallel path for fault current back to the source I will add that to what I am thinking about, but no one has said that explicitly (my apologies if someone did and I missed it).

Let me interject at this point my assurance that I am not just being argumentative. If I am to change the way I am doing things, I need to understand why, and not just because a particular interpretation of the NEC implies it. Furthermore, I believe that if it is indeed the case that in supply side interconnected PV systems this is an unambiguously necessary configuration, IMO it needs to be specifically addressed in 690 or 705 (or somewhere) rather than depending on interpretations atop interpretations to get it right.

 

[Smart $]

But it does. ...

I realize you are aware of that for your particular installation. What I'm saying is the Code is written from the perspective of not knowing. What if the main bonding jumper at the service disconnect, sized to its tap conductors, is smaller than your [so-called] EGC?

Note there is no such thing as an EGC on the service side of the service disconnecting means.

Let's put it this way... regardless of whether the PV System disconnect is a service disconnect, it is by definition service equipment. 705.12(A) supports this because it refers back to 230.82(6). Note the title of 230.86 is "Equipment Connected to the Supply Side of Service Disconnect." That makes it service equipment.

Now jump over to 250.80... and while you're there also read 250.92(A) & (B).

Ultimately you have to at the very least connect the enclosure to 1) the grounded [neutral] service conductor, or 2) run a bonding jumper. Either way, the minimum size is per 250.66. So take your pick.

Where I have doubts is when the grounding conductor for the PV side is a combo EGC/GEC. If it's serving the purpose of a GEC, I believe the intent is to have it actually end up going to a GE by way of a GEC, and not by way of the service grounded conductor or a bonding jumper, right? So how does one manage that? Seems to me a GEC has to be run to the disconnect... but the Code does not specifically say you have to. The building GEC(s) can be connected anywhere to the grounded [neutral] service entrance conductor... and it may be sized for the SE taps going to the [normal] service disconnecting means....!!!

 

[jaggedben]

If the inverter requires a neutral, then I don't see why you wouldn't run a neutral to the disconnect and bond it there. It's the same as what would be done for all other service disconnects of any kind. The fact that it is PV really shouldn't change things because we are concerned with the available fault current from the utility, and the size and direction of the energy flow in normal operation is really irrelevant. The fact that some AHJ's don't seem to get this point (or don't care!) is disturbing, regardless of how they decide to enforce an effective ground fault pathway.

If the inverter doesn't require a neutral, then perhaps the issue is a little stickier, since particularly on larger systems it could be an unnecessary burden to run two conductors (neutral and GEC) when only one (combined PV GEC/EGC) might be a sufficient ground fault path. Probably the one conductor is sufficient, but I'd say it ought to be terminated in the same enclosure where the tap is made. And this only makes sense if the GEC and EGC requirements both have the same destination. That is, if the GEC is going straight to the electrode from the disco, rather than passing by the tap, then you might as well call the conductor going to the tap a neutral, for consistency's sake.

Whether you tape it white or green, whichever conductor is providing the the effective ground fault pathway back to the utility needs to be bonded to the disco and should be sized to 250.66 not 250.122 (however small in consequence that may be).