250.97 and ungrounded inverters

[pv_n00b]

Actually GFDI was always required. And the solution of grounding one DC conductor (but not solidly, only through a fuse) was a simple way to engineer that requirement. But it made putting in a transformer necessary, which lowered efficiency, and it wasn't really that reliable. Once manufacturers figured out how to make reliable electronic GFDI that would get listed by UL without using a grounded conductor, transformerless inverters took over quickly due to lower cost and higher efficiency. They are also safer for installation crews and others, if perhaps more prone to nuisance tripping and destructive failure in the case of faults.

I'm not sure when GFDI was first required, my NEC library only goes back to 1999, but up to the 2008 NEC it was only required on roof mounted PV systems on dwellings. I designed a lot of roof mounted commercial systems without GFDIs and no ground mounted systems needed them. It took several years for inverter manufactures to catch up after 2008 went into effect and they started putting GFDIs in the larger inverters. Up to then it was only standard on residential inverters and those were too small for commercial systems. It was questionable if GFDI was required on apartment or condo buildings and a lot of those from that time don't have GFDI. The systems were solidly grounded at the inverter DC input with a bonding jumper, but without GFDI you could do it anywhere in the array if you wanted. Good times.

I'm not sure what you mean about electronic GFDIs. Some inverter manufacture tried to get an allowance for an electronic switch to act as a GFDI several NEC versions ago but it was voted down by the CMP. The usual way to comply with 690.35(C) in an ungrounded transformerless system is to shutdown the inverter when a ground fault is detected.

 

[jaggedben]

Aren't most central inverters still grounded?

I'm going to trust you on that G. Seems like everything under 50k is ungrounded now though.

I'm not sure when GFDI was first required, my NEC library only goes back to 1999, but up to the 2008 NEC it was only required on roof mounted PV systems on dwellings. I designed a lot of roof mounted commercial systems without GFDIs and no ground mounted systems needed them. It took several years for inverter manufactures to catch up after 2008 went into effect and they started putting GFDIs in the larger inverters. Up to then it was only standard on residential inverters and those were too small for commercial systems. It was questionable if GFDI was required on apartment or condo buildings and a lot of those from that time don't have GFDI. The systems were solidly grounded at the inverter DC input with a bonding jumper, but without GFDI you could do it anywhere in the array if you wanted. Good times.

You should change your handle. :lol:

I'm not sure what you mean about electronic GFDIs. Some inverter manufacture tried to get an allowance for an electronic switch to act as a GFDI several NEC versions ago but it was voted down by the CMP. The usual way to comply with 690.35(C) in an ungrounded transformerless system is to shutdown the inverter when a ground fault is detected.

I was just kind of using 'electronic' as a short hand for anything that doesn't make the connection to ground through a fuse. My understanding is that GFDIs for ungrounded inverters use Hall effect sensors and Megger type testing and other such stuff. All of this is part of the inverter listing now as far as I see it.

 

[pv_n00b]

I was just kind of using 'electronic' as a short hand for anything that doesn't make the connection to ground through a fuse. My understanding is that GFDIs for ungrounded inverters use Hall effect sensors and Megger type testing and other such stuff. All of this is part of the inverter listing now as far as I see it.

For the most part I think they do residual current monitoring and insulation testing for ground fault testing in ungrounded systems. I wanted to point out that it only does GFD, no "I" since there is no grounded conductor in these systems to isolate in a fault. If there is a dual line to ground fault that fault current will continue to flow until someone fixes the fault. So in a way we are back to the old pre GFDI days and if there is a double fault our current sizing of ECGs is questionable to me. But then again we don't design for multiple faults.

I filled out my library with old NECs back to 1984 when 690 was added, boy was life simpler then. The GFDI requirement was added in 1990 for systems on dwelling roofs. Since that's only 16 years ago there are systems still out there in operation with no GFDI. My guess is that it was several years after the 1990 NEC came out before GFDIs were widely available and being installed. If I remember correctly the first GFDIs were standalone components with a couple of circuit breakers tied together so if the CB connecting the grounded conductor to ground opened in a ground fault it caused the tied CB connecting the inverter to the array to open. Good thing to keep in mind when servicing old systems.

 

[SolarPro]

Aren't most central inverters still grounded?

Kinda sorta. They are not solidly grounded. They are grounded through a fuse or circuit breaker, which is not the same thing. (See solidly grounded definition in Article 100.)

Similarly, what we have been calling ungrounded systems are not really "ungrounded,. The ac side of the inverter is grounded, which effectively grounds the dc side of the system when the inverter is operating.

NEC 2017 adds a new definition functional grounded PV system. This definition basically covers the vast majority of the PV systems that we install, which are neither solidly grounded or truly ungrounded. The revised Code also details a set of unified design standards that apply to all functional grounded PV systems. This will simplify things for both integrators and inspectors. No need to keep track of two different sets of design standards; no problems retrofitting non-isolated inverters; etc..

The HTML version of Bill Brooks' latest SolarPro article on NEC 2017 will drop on Sunday, May 1st. Pretty fascinating stuff. The same issue also features an article by Marvin Hamon, a PE from Alameda, CA, about how to get involved in the codes and standards making process. (Remember, if you're not at the table, you could be on the menu.)

 

[jaggedben]

Similarly, what we have been calling ungrounded systems are not really "ungrounded,. The ac side of the inverter is grounded, which effectively grounds the dc side of the system when the inverter is operating.

Except for Enphase's current products, which are ungrounded and isolated... :happyyes:

NEC 2017 adds a new definition functional grounded PV system. This definition basically covers the vast majority of the PV systems that we install, which are neither solidly grounded or truly ungrounded. The revised Code also details a set of unified design standards that apply to all functional grounded PV systems.

:thumbsup::thumbsup::thumbsup:
(And 'functional grounded' includes the Enphase design.)

 

[jaggedben]

For the most part I think they do residual current monitoring and insulation testing for ground fault testing in ungrounded systems. I wanted to point out that it only does GFD, no "I" since there is no grounded conductor in these systems to isolate in a fault. If there is a dual line to ground fault that fault current will continue to flow until someone fixes the fault. So in a way we are back to the old pre GFDI days and if there is a double fault our current sizing of ECGs is questionable to me. But then again we don't design for multiple faults.

...

My understanding is that 'I' stands for 'interruption' (690.5(A)) and the inverter very much performs interruption by functionally disconnecting the ungrounded conductors. Maybe the language changed at some point but I'm pretty sure it's been 'interruption' since 2008.

Also these inverters are very sensitive, or else I wouldn't have had to have my crews replace half the wiring on a system that, by my Ohm's law calculation, was passing a fault current of 3ma. (And only when the whether was wet, so I believe it was not a metal-to-metal-contact fault. Obviously I never found the actual location.) So we are definitely not 'back to the old pre-GFDI' days! :happyno:

And no, we don't design for multiple faults. The whole idea is predicated on someone noticing the inverter error message after the first fault.

 

[pv_n00b]

My understanding is that 'I' stands for 'interruption' (690.5(A)) and the inverter very much performs interruption by functionally disconnecting the ungrounded conductors. Maybe the language changed at some point but I'm pretty sure it's been 'interruption' since 2008.

Also these inverters are very sensitive, or else I wouldn't have had to have my crews replace half the wiring on a system that, by my Ohm's law calculation, was passing a fault current of 3ma. (And only when the whether was wet, so I believe it was not a metal-to-metal-contact fault. Obviously I never found the actual location.) So we are definitely not 'back to the old pre-GFDI' days! :happyno:

And no, we don't design for multiple faults. The whole idea is predicated on someone noticing the inverter error message after the first fault.

GFDI in only specified for grounded arrays, it is not applicable to ungrounded arrays, 690.5. The "I" is for interruption of the ground fault current only, not disconnection of the ungrounded conductors, 690.5(A)(2). This is typically done by opening the grounded conductor bond to ground by the GFDI fuse opening. I don't know of any inverters that also open the ungrounded conductors, and it would not matter anyway since the inverter is a load and not the source.

So when we are talking about ungrounded PV arrays there is no GFDI requirement and we are in the pre-GFDI days. The system ungrounded inverters use to detect ground faults is very sensitive but when the system is triggered it only shuts down the inverter and shows a ground fault error. Nothing is done in the PV array by the inverter that could stop fault current flow. The PV array is still faulted with at least one fault and there will be fault current flowing on the second fault when it happens. We depend on someone caring about the inverter being shutdown enough to get the fault repaired quickly since there is little protection built into the array. But what happens if no one cares that the inverter is shutdown and showing a ground fault? It's not like the inverter is twisting someone's arm and forcing them to fix it.

We don't design for multiple faults, mostly because standard electrical system design is fault tolerant. In a standard grid connected grounded electrical system a fault will clear because there is a large available fault current from the grid and it will flow on the first fault of an ungrounded conductor. This makes design of safe systems fairly easy, pick the right combination of conductor and OCP and you know you are protected. But we give up a lot of this certainty when dealing with PV arrays. We don't have large available fault currents to cause OCP operation and more systems are ungrounded. So we can't allow ourselves the same warm and fuzzy feeling we get applying the simple protection systems that we use with in standard grid connected electrical systems. We have to think outside the box if we want to make safer systems. If we just want to check off NEC requirements then that's easy.

Since PV arrays don't have large available fault current the only way to truly protect against fault current flow is to have fault sensing systems in the module that can disconnect the module if it sees a fault in the wiring, even before fault current can flow. I don't know if we will have that anytime soon though. Until then we just do our best and hope for the best.

 

[jaggedben]

GFDI in only specified for grounded arrays, it is not applicable to ungrounded arrays, 690.5. The "I" is for interruption of the ground fault current only, not disconnection of the ungrounded conductors, 690.5(A)(2). This is typically done by opening the grounded conductor bond to ground by the GFDI fuse opening. I don't know of any inverters that also open the ungrounded conductors, and it would not matter anyway since the inverter is a load and not the source.

So when we are talking about ungrounded PV arrays there is no GFDI requirement and we are in the pre-GFDI days. ...

You have totally overlooked 690.35(C). But your points about ignored faults are correct.

Since PV arrays don't have large available fault current the only way to truly protect against fault current flow is to have fault sensing systems in the module that can disconnect the module if it sees a fault in the wiring, even before fault current can flow. I don't know if we will have that anytime soon though. Until then we just do our best and hope for the best.

Yes I think that we'll likely see module level electronics being required in the 2020 or 2023 code, maybe just for systems on dwellings. Optimizer systems are basically safe from this problem.

 

[SolarPro]

If you know what the code is going to require four years from now, you must have 2020 vision.

 

[jaggedben]

If you know what the code is going to require four years from now, you must have 2020 vision.

<groan>

(And incidentally, I'm just repeating what I've heard others say, which I'm pretty sure includes you at the top of the list. :lol