Miscellaneous Grounding Questions

[Grouch]

Hi.
Back with more questions! .
So I'm stuck on the 2008 NEC code... however I do glance at the 2017 code, at article 690, to see if it's a little clearer. A lot of things were re-organized. Some questions I still have:

1. I see diagrams (google ), and drawings from engineers, where:

a) each PV string (say 10 modules in a string) has its own EGC, and then all the EGC's connect together to one EGC that goes to the inverter.
b) I also see diagrams where each PV module in adjacent strings are bonded together, and then only one EGC, on the side of the array, that goes back to the inverter.

Which is the correct way? Or are they both allowed? From reading section 690.45, the EGC is sized based on the OCPD serving that string. So to me it sounds like only a) above would be allowed. But I know b) is allowed also.

2. NEC 2008 section 690.47(C)(3). I cannot picture what this section is trying to say. is there a diagram somewhere that can help?

3. NEC 2008 (or 2017) section 690.46. why is this a separate section? it seems to only talk about the size of the EGC. Shouldn't this be just a sub-section of 690.45? I don't understand why it's titled 'Array Equipment Grounding Conductors', if it's talking about the EGC size from section 690.45. Is this something else entirely?

4.

 

[Buck Parrish]

This might get you close. I think it's NEC 14 though.

 

[Grouch]

Thanks for the video. I did actually watch this several weeks ago. But for example, section 690.47(C)(3) in the 2014 NEC, as the video shows in 5:05... is that the same thing as 690.47(C)(3) in the 2008 code? That's where i'm not sure. If yes, that would answer my 2nd question.

 

[jaggedben]

Hi.
Back with more questions! .
So I'm stuck on the 2008 NEC code... however I do glance at the 2017 code, at article 690, to see if it's a little clearer. A lot of things were re-organized. Some questions I still have:

1. I see diagrams (google ), and drawings from engineers, where:

a) each PV string (say 10 modules in a string) has its own EGC, and then all the EGC's connect together to one EGC that goes to the inverter.
b) I also see diagrams where each PV module in adjacent strings are bonded together, and then only one EGC, on the side of the array, that goes back to the inverter.

Which is the correct way? Or are they both allowed? From reading section 690.45, the EGC is sized based on the OCPD serving that string. So to me it sounds like only a) above would be allowed. But I know b) is allowed also.

2. NEC 2008 section 690.47(C)(3). I cannot picture what this section is trying to say. is there a diagram somewhere that can help?

3. NEC 2008 (or 2017) section 690.46. why is this a separate section? it seems to only talk about the size of the EGC. Shouldn't this be just a sub-section of 690.45? I don't understand why it's titled 'Array Equipment Grounding Conductors', if it's talking about the EGC size from section 690.45. Is this something else entirely?

4.

1. It doesn't matter. Just make sure that all circuits are accompanied by an EGC that goes back to a groundbar where the circuits terminate. Same as with AC wiring. If the circuit is in conduit the EGC should be in the same conduit. If the circuit wires are dancing around an array to to reach solar modules then the EGC can dance around the array to bond the module frames and racking.

2. They cut all this stuff out later because they realized none of it mattered. In '08, it's basically this: if your inverter has a grounding electrode terminal, you can run a GEC to a new electrode which you bond to the existing AC electrode(s), or you can just run it to the AC electrodes. With today's codes and equipment you generally just run an EGC back to the groundbar at the AC point of interconnection.

3. That is specifically addressing exposed EGCs on arrays, and saying they have to be #6 even if the otherwise applicable requirement is smaller. I don't know why it isn't just a subsection of 690.45 but I wouldn't ascribe much to that.

 

[Grouch]

1. It doesn't matter. Just make sure that all circuits are accompanied by an EGC that goes back to a groundbar where the circuits terminate. Same as with AC wiring. If the circuit is in conduit the EGC should be in the same conduit. If the circuit wires are dancing around an array to to reach solar modules then the EGC can dance around the array to bond the module frames and racking.

2. They cut all this stuff out later because they realized none of it mattered. In '08, it's basically this: if your inverter has a grounding electrode terminal, you can run a GEC to a new electrode which you bond to the existing AC electrode(s), or you can just run it to the AC electrodes. With today's codes and equipment you generally just run an EGC back to the groundbar at the AC point of interconnection.

3. That is specifically addressing exposed EGCs on arrays, and saying they have to be #6 even if the otherwise applicable requirement is smaller. I don't know why it isn't just a subsection of 690.45 but I wouldn't ascribe much to that.

1. ok, that I follow, that it can be done either way. So based on my items a) and b) in my original post above:

a) if an EGC is routed for each string, then I understand that each EGC is sized based on Table 250.122. How do you then size the one EGC that they all connect to (the EGC that lands in the groundbar where the circuits terminate)? Does its size match the EGC size for each string?
b) Assuming all PV modules are bonded together, and you have only one EGC on the side of the array going back to the groundbar, how is that EGC sized? Does it just match the EGC size for each string, assuming each string had one (which they don't in this case)?

2. ok, got it.

3. not following regarding the size... the EGC can still be smaller than #6 correct? I don't see where it says #6 is the smallest size allowed.

 

[jaggedben]

1) same rules as in article 250. If you have multiple circuits in a conduit then you can have one EGC that is sized to the largest required for any of those circuits.

3) look at the part of 250 it's referencing. You either go with #6 or put it in conduit. The point is, you might have an EGC that can be #10 in your conduit to the roof, but when you splice to a bare solid copper to bond the racking out in the open your AHJ may require a #6 for that part. That said most of the AHJs I've worked with do not enforce this for rooftop arrays. Maybe they're just not aware or maybe they don't consider it subject to damage if it's under the panels.

 

[Grouch]

1) same rules as in article 250. If you have multiple circuits in a conduit then you can have one EGC that is sized to the largest required for any of those circuits.

3) look at the part of 250 it's referencing. You either go with #6 or put it in conduit. The point is, you might have an EGC that can be #10 in your conduit to the roof, but when you splice to a bare solid copper to bond the racking out in the open your AHJ may require a #6 for that part. That said most of the AHJs I've worked with do not enforce this for rooftop arrays. Maybe they're just not aware or maybe they don't consider it subject to damage if it's under the panels.

I see. ok, I understood both parts now! Thanks for the help.

 

[Carultch]

Hi.
Back with more questions! .
So I'm stuck on the 2008 NEC code... however I do glance at the 2017 code, at article 690, to see if it's a little clearer. A lot of things were re-organized. Some questions I still have:

1. I see diagrams (google ), and drawings from engineers, where:

a) each PV string (say 10 modules in a string) has its own EGC, and then all the EGC's connect together to one EGC that goes to the inverter.
b) I also see diagrams where each PV module in adjacent strings are bonded together, and then only one EGC, on the side of the array, that goes back to the inverter.

Which is the correct way? Or are they both allowed? From reading section 690.45, the EGC is sized based on the OCPD serving that string. So to me it sounds like only a) above would be allowed. But I know b) is allowed also.

2. NEC 2008 section 690.47(C)(3). I cannot picture what this section is trying to say. is there a diagram somewhere that can help?

3. NEC 2008 (or 2017) section 690.46. why is this a separate section? it seems to only talk about the size of the EGC. Shouldn't this be just a sub-section of 690.45? I don't understand why it's titled 'Array Equipment Grounding Conductors', if it's talking about the EGC size from section 690.45. Is this something else entirely?

4.

1. both are code-compliant, but option (b) is most likely going to be more cost-effective. Option (a) you certainly are permitted to do, but you don't necessarily need to do it that way.

The reason why (b) is correct, is that the EGC is not going to be carrying simultaneous faults of all the strings at once. It just needs to be sized for the single circuit's fault it will realistically carry. Any serious short to ground will be a temporary condition, that will stop once it clears the OCPD and shuts the circuit off.

The EGC gets sized to the worst-case-scenario circuit, and 250.122 rules that apply to it, within any given conduit, or wiring method in general.

The purpose of the EGC is as follows:
1. Eliminate the voltage difference between all electrically inactive metal and the surrounding ground, so that it remains as de-energized as possible.
2. Provide the return path back to the source when there is a fault to what should be electrically inactive metal, so that it clears the OCPD before it takes a worse path back to the source. Such as through a human body.

 

[Grouch]

The reason why (b) is correct, is that the EGC is not going to be carrying simultaneous faults of all the strings at once. It just needs to be sized for the single circuit's fault it will realistically carry. Any serious short to ground will be a temporary condition, that will stop once it clears the OCPD and shuts the circuit off.

Thanks. Yes, this makes a lot of sense. Realistically you won't have shorts everywhere, so the EGC wouldn't have to be sized for all the strings. Understood!

 

[Grouch]

1. I see diagrams (google ), and drawings from engineers, where:

a) each PV string (say 10 modules in a string) has its own EGC, and then all the EGC's connect together to one EGC that goes to the inverter.
b) I also see diagrams where each PV module in adjacent strings are bonded together, and then only one EGC, on the side of the array, that goes back to the inverter.

In b), how are the bond wires between the adjacent strings sized? In this case, would they all match the size of the EGC on the side of the array?

 

[jaggedben]

In this case think of the array as being akin to a raceway that contains all the circuits on it. Whatever circuit requires the largest EGC that will be your minimum requirement. Don't count circuits in raceways that happen to be attached to the array, just the circuits with exposed conductors or cables on the array. This is leaving aside the additional possible requirement that it be #6 minimum if exposed.

Typically all your string sizes will have the same EGC requirement so any exposed EGCs on the array will have the same size requirement as your EGC coming to the roof in your conduit. Again, leaving aside the possible additional requirement that exposed EGCs be minimum #6.

 

[Carultch]

In b), how are the bond wires between the adjacent strings sized? In this case, would they all match the size of the EGC on the side of the array?

For a group of circuits, you size according to the worst case among them. The same size ground that can work for 1 source circuit, can work for a bundle of 20 source circuits. Once you electrically combine the strings, then your size increases, and your size becomes based on the combiner output circuit. But if all you do is mechanically collect the wires in the same raceway, the EGC is only sized according to the largest among them.

The EGC is sized according to the OCPD (whether it be a fuse or breaker) and table 250.122. If an OCPD is not applicable, you first calcuate 1.25*Isc, and then select an EGC according to it, per 690.45. It is a little non-intuitive that the factor isn't 1.56*Isc, since that's what would ordinarily govern the OCPD rating (when used) and subsequently the EGC. The other factor that governs the sizing is #6 Cu or #4 AL being the minimum size allowable outside of a raceway, which you inevitably have to do to connect to your racking system.

 

[Grouch]

If an OCPD is not applicable, you first calcuate 1.25*Isc, and then select an EGC according to it, per 690.45. It is a little non-intuitive that the factor isn't 1.56*Isc, since that's what would ordinarily govern the OCPD rating (when used) and subsequently the EGC.

I read 690.45 in the 2017 if there is no OCPD... if there is no OCPD, it refers you to 690.9(B)... that says not less than 125% of the max current... isn't the factor then 1.56 * Isc?

So according to you and jaggedben, the size of those bonding conductors just matches the size of the largest EGC... did I follow correctly?

 

[Carultch]

I read 690.45 in the 2017 if there is no OCPD... if there is no OCPD, it refers you to 690.9(B)... that says not less than 125% of the max current... isn't the factor then 1.56 * Isc?

So according to you and jaggedben, the size of those bonding conductors just matches the size of the largest EGC... did I follow correctly?

You are correct. This rule has changed over the years. In 2011, it was total Isc with no factors. In 2014, it was 1.25*Isc, and that was the last one I remembered. As of the onset of NEC2017, single combiner systems where this would govern a design have become a lot more rare. It makes a lot more sense for it to be the same factor that sizes it when there is an OCPD, as when there isn't.

Yes, you did follow myself and Jaggedben correctly. As an example, given a 40A circuit and a 10A circuit both sharing a conduit routed together in the same wiring method, you size it according to the 40A circuit and you can share it between the two of them.

 

[Grouch]

Great. Understood on both now. Yeah I'm not surprised that factor changed as the code cycles went on.... most of 690 kept getting revamped as I looked through the cycles. It's a bit of a mess. Now I'm just flipping between 2008 and 2017 only.

 

[Grouch]

Thanks for the help guys!

 

[ggunn]

An underlying issue is that PV is still relatively new to the NEC and the current flow is in the opposite direction from what it was written to address. You sometimes have to stand on your head when you are reading the code and applying it to PV.

 

[pv_n00b]

An underlying issue is that PV is still relatively new to the NEC and the current flow is in the opposite direction from what it was written to address. You sometimes have to stand on your head when you are reading the code and applying it to PV.

PV has been a part of the NEC since 1984, 37 years in the code is not relatively new to the code. It's always been hard for people to make the code work for PV because it's not a simple source and load situation. That's why we see huge changes in the NEC around PV in almost every code revision. If NFPA can't figure it out after 37 years I don't have a lot of confidence that it's going to settle out any time soon.

 

[jaggedben]

Arguably the first 20 or so years don't count because almost no one was interested in it and the technology also changed massively afterward.

 

[ggunn]

PV has been a part of the NEC since 1984, 37 years in the code is not relatively new to the code. It's always been hard for people to make the code work for PV because it's not a simple source and load situation. That's why we see huge changes in the NEC around PV in almost every code revision. If NFPA can't figure it out after 37 years I don't have a lot of confidence that it's going to settle out any time soon.

The NEC has been around since 1897, so yes, 1984 is relatively recent