Secondary OCPD With String Inverters

[bwat]

I've seen this image (or one very much like it) multiple times before. I'm trying to understand the code compliance, implications, and options with it. Specifically with the transformer secondary conductors and their protection.

Assume they're all 125kW string inverters. So the inverters (16x125kw = 2MW) aren't greater than the transformer (2MVA) rating. Because of 690 and 705, I don't need secondary transformer protection. But there's still the secondary condcutor protection requirements in 240.21(C), and then 408.36 if it's a panelboard and not a switchboard.

The only option that seems to make sense to allow multiple circuits off the secondary like this is the 240.21(C)(2), but that's limited to 10ft conductors, and there's no way these conductors in the picture would be less than 10ft when you go through the conduit path.

Or perhaps I'm reading that section entirely wrong, and when 240.21(C)(4), outside conductors, says that it needs to land on a single OCPD, it really only means that this set of conductors need to land on a single OCPD. The conductors to the other 3 panels are irrelevant and treated separately. As long as each set lands on a single OCPD that aligns with 240.21(C)(4), then you're good. Is it as simple as that?


I've even seen some places show these panels being MLO, and I'm not sure how that is able to fly. Maybe the "industrial installation" clause in 240.21(C)(3), but it would require full size secondary conductors to each panel, which could be problematic with lug connections there. 2500A+ conductors landing in 800A panel... and then there's also 408.36 panelboard protection issue. Right?

 

[jaggedben]

...

Or perhaps I'm reading that section entirely wrong, and when 240.21(C)(4), outside conductors, says that it needs to land on a single OCPD, it really only means that this set of conductors need to land on a single OCPD. The conductors to the other 3 panels are irrelevant and treated separately. As long as each set lands on a single OCPD that aligns with 240.21(C)(4), then you're good. Is it as simple as that?


...

This, I think.

I've always been told you can tap a feeder as many times as you like and I've never heard the 'single' OCPD interpeted to apply to more than the conductors for one circuit. Granted it's ambiguous, but it would lead to arguably even more weirdness and ambiguity if interpreted your way. Like, can you have both 10ft and 25ft taps or secondaries.

 

[bwat]

Like, can you have both 10ft and 25ft taps or secondaries.

I believe so. Since I posted the OP, I did some more poking around. My 2017 handbook specifically speaks to this in the commentary and talks about using the 25ft for one set, and the 10ft for another set.

 

[pv_n00b]

This here is a problem that dogs us in larger installations.
240.21 for tap conductors and XFMR secondaries concedes that fault protection comes from physically protecting the conductors and limiting the length and overload protection usually comes from putting a single OCPD on the load end of the conductor. This is in general what is happening in this article.

This is not written for current limited supplies like inverters. It's written with loads in mind to keep someone from overloading the conductor. This results in some significant restrictions on the design of larger systems with multiple inverters fed by a single transformer. Mostly cost-related since it's costly to provide that single OCPD when you are grouping inverters in an AC combiner. If you have an 800V system it might be impossible since higher current rated UL Listed 800V CBs are not available. So you end up doing what I do which is to have conversation after conversation with AHJs to convince them that there is no way for the conductors to be overloaded by current limited inverters on the XFMR secondary. So the inverters should be considered equivalent protection from overload to an OCPD. Lots of talking and emailing over this single point.

I have not had to deal with the panelboard protection question, if this were my system all these inverters would be combined in a single SWBD. Waste of time dinking around with small panelboards.

 

[pv_n00b]

Or perhaps I'm reading that section entirely wrong, and when 240.21(C)(4), outside conductors, says that it needs to land on a single OCPD, it really only means that this set of conductors need to land on a single OCPD. The conductors to the other 3 panels are irrelevant and treated separately. As long as each set lands on a single OCPD that aligns with 240.21(C)(4), then you're good. Is it as simple as that?

You are correct that each set of conductors on the secondary terminals of the XFMR is treated separately. So if each of the panelboards has a main OCPD then it's code compliant under the outside conductor section.

 

[bwat]

Thanks for the great answers. Very helpful.

Anybody know how how you could make it compliant with MLO? Image below that suggests this. I have to be able to defend myself in saying this is either not possible or not a good idea.

Options with 240.21(C) and my comments. (1) means 240.21(C)(1), (2) means 240.21(C)(2), etc

(1) If Protection by primary device. Doesn't apply unless by some odd scenario it's a D-D.
(2) If not over 10ft. This almost seems possible, but there's no way the conductors aren't over 10ft in the picture, and these would have to be switchboards and not panelboards due to 408.36. Perhaps it works in another scenario where it's less than 10ft and an MLO switchboard?
(3) If Industrial installation and not over 25ft. All secondary conductors would have to be rated for full transformer rating. Again the 408.36 though. Seems problematic, but maybe possible with this one in another scenario with an MLO switchboard.
(4) If outside conductors. Not possible since it needs to terminate on OCPD
(5) From Feeder Tapped Transformer. Not possible since it needs to terminate on OCPD per referenced 240.21(B)(3)
(6) If not over 25ft. Not possible since it needs to terminate on OCPD.



So it seems like there are only these three options if somebody really wants to do MLO.
1. Fusing at the secondary terminals of the transformer
2. Keep conductors less than 10ft and make it MLO switchboard
3. Call it industrial installation, keep conductors less than 25ft, make it MLO switchboard, and run full transformer rated conductors

Agree? Disagree?

 

[electrofelon]

This is a utility transformer, none of the transformer secondary conductor protection rules apply, what am I missing?

 

[bwat]

This is a utility transformer, none of the transformer secondary conductor protection rules apply, what am I missing?

I can't comment on whether or not the one in the picture was owned by the utility (I see the meter) since it wasn't my project, but in all the PV projects I'm involved in like this, the utility very much does not own the transformer. And that's the scenario I'm describing.

 

[electrofelon]

I can't comment on whether or not the one in the picture was owned by the utility (I see the meter) since it wasn't my project, but in all the PV projects I'm involved in like this, the utility very much does not own the transformer. And that's the scenario I'm describing.

Oh ok, I misunderstood, I thought the photo was specifically the system you were talking about.

 

[bwat]

Oh ok, I misunderstood, I thought the photo was specifically the system you were talking about.

Only as an example. It's being discussed with a client in a "why can't we do something like this?"

You bring up a good point though in that if this picture is a utility transformer, then that very much doesn't apply to my scenario, but it would still seem like MLO is not possible there since those become service disconnects.

I don't want to railroad the convo with that topic though. Please just assume I'm talking about customer owned transformer.

 

[wwhitney]

240.21 for tap conductors and XFMR secondaries concedes that fault protection comes from physically protecting the conductors and limiting the length and overload protection usually comes from putting a single OCPD on the load end of the conductor. This is in general what is happening in this article.

For utility-tied systems, the above is all true whether there's a load at the end, or an interactive source at the end. So I see no reason the tap rules should be any different for the two cases. The short circuit and ground fault protection requirements are the same either way.

So you end up doing what I do which is to have conversation after conversation with AHJs to convince them that there is no way for the conductors to be overloaded by current limited inverters on the XFMR secondary.

While that may be true, so what? That has no bearing on the application of the tap rules.

Cheers, Wayne

 

[ggunn]

Maybe I am reading this incorrectly, but 240.21(B) is only for feeders, not for service conductors. With service conductors there is already a virtually infinite amount of current available (orders of magnitude more than the ampacity of the conductors); your PV system output has no appreciable effect on the current available to the other service disconnect(s).

 

[wwhitney]

Anybody know how how you could make it compliant with MLO?

So per transformer, what is the desired number of (A) sets of secondary conductors (total, counting each set in parallel installations), (B) number of OCPD, and (C) number of PV inverters?

Obviously the easiest case is that A:B:C is 1:1:1, and then you can run each set of secondary conductors outdoors an unlimited length, landing on one OCPD, serving one inverter. Or for sufficiently large inverters, X:1:1, with the X sets installed in parallel. But presumably there are some downsides to this approach, so you'd like different ratios?

Cheers, Wayne

 

[wwhitney]

service conductors.

This discussion is about customer owned transformers, so their secondary conductors are feeders, not service conductors. (Secondary="not grid side")

Cheers, Wayne

 

[bwat]

So per transformer, what is the desired number of (A) sets of secondary conductors (total, counting each set in parallel installations), (B) number of OCPD, and (C) number of PV inverters?

Obviously the easiest case is that A:B:C is 1:1:1, and then you can run each set of secondary conductors outdoors an unlimited length, landing on one OCPD, serving one inverter. Or for sufficiently large inverters, X:1:1, with the X sets installed in parallel. But presumably there are some downsides to this approach, so you'd like different ratios?

Cheers, Wayne

I'll take a stab at answering, although I'm not sure I'm on the same page. 4:16:16

 

[bwat]

Crude sketch:

 

[ggunn]

This discussion is about customer owned transformers, so their secondary conductors are feeders, not service conductors. (Secondary="not grid side")

Cheers, Wayne

Where is such a transformer in a system, what is the primary voltage, and is there any OCP between it and the utility? If it has MV on the primary and no OCP ahead of it, then I believe its secondary conductors are still functionally service conductors.

 

[wwhitney]

Looking back at the OP, you mentioned 125 kW inverters. So are those 600V/120A 3 phase output, or 480V/150A? I'm guessing the latter since you are showing 800A panels, and 4*120*125% = 600A, which is a smaller panel size, I think. [But if you are using a transformer anyway, is 600VAC a better solution?]

That means barring any 100% rated OCPD being involved, each single line from the transformer to one of the 4 panels is a 750A feeder. So that could be (4) sets of 250 kcmil Al, for example. Which means you would have to terminate 16 sets of conductors at the transformer.

Given that, is there something prohibiting using 16 different 200A disconnects, one for each inverter, and zero panelboards? E.g. a 6 throws of the hand type rule? If not, what's the upside to grouping your 16 sets into 4 groups of 4 parallel sets and using the panelboards?

If using 16 separate enclosures is an issue, maybe there is a way to get multiple OCPD, with individual lugs on both the line and load sides, into one enclosure? E.g. a DIN mounted solution?

If there is a 6 throws of the hand type limit, I wonder if there's a solution that would effectively handle-tie 4 such OCPD within an enclosure.

Cheers, Wayne

 

[ggunn]

Crude sketch:

View attachment 2565109

There is no OCPD required between the panels and the transformer as long as the total maximum output current of the inverters does not exceed the current rating of the transformer per 705.30(C). You may need to size the conductors so that they would be protected by the OCPD on the primary side of the transformer, though.

 

[ggunn]

Looking back at the OP, you mentioned 125 kW inverters. So are those 600V/120A 3 phase output, or 480V/150A? I'm guessing the latter since you are showing 800A panels, and 4*120*125% = 600A, which is a smaller panel size, I think. [But if you are using a transformer anyway, is 600VAC a better solution?]

That means barring any 100% rated OCPD being involved, each single line from the transformer to one of the 4 panels is a 750A feeder. So that could be (4) sets of 250 kcmil Al, for example. Which means you would have to terminate 16 sets of conductors at the transformer.

Given that, is there something prohibiting using 16 different 200A disconnects, one for each inverter, and zero panelboards? E.g. a 6 throws of the hand type rule? If not, what's the upside to grouping your 16 sets into 4 groups of 4 parallel sets and using the panelboards?

If using 16 separate enclosures is an issue, maybe there is a way to get multiple OCPD, with individual lugs on both the line and load sides, into one enclosure? E.g. a DIN mounted solution?

If there is a 6 throws of the hand type limit, I wonder if there's a solution that would effectively handle-tie 4 such OCPD within an enclosure.

Cheers, Wayne

A disco on the primary side of the transformer will satisfy the six handle rule.