Large Industrial Interconnection Details

[solarken]

Designing a 2.15MW system at Industrial site. I need to interconnect on the customer side of the CT cabinet that the utility requires and probably cannot connect in that cabinet due to utility co rules. So looking at installing a Tap cabinet adjacent to the CT cab and connect there.

There will be 12 x 400kcmil conductors per phase in from the CT cabinet (utility feed), 12 x 400kcmil conductors per phase out to the 4000A panelboard in the building, and 8 x 500kcmil conductors per phase from the PV inverters. Conductors are aluminum.

Looking at 2023 NEC Annex B, Figure B.2(1) shows duct bank for 9 conduits (3 row, 3 col) and has ampacity for 500kcmil at 375A for RHO 60 LF 50. Then it drops severely to 240A for RHO 90 LF 100, and marginally further to 220A for RHO 120 LF 100. The notes say this is for copper conductors.

I have several challenges I hope someone can help me with:

Copper conductors for this would be cost prohibitive as the run between inverters and Tap cabinet will be about 400ft. Any suggestions on extrapolating the Annex B.2(2) Ampacities to Aluminum?

The few examples in the Annex are not granular, obviously. But for a PV inverter, we calculate continuous current as 1.25 x Max inverter current. However, everyone knows generally an inverter may output it's max for only several minutes, to a few hours max, depending on DC/AC ratio. How does LF in this context relate to the inverter current? Do I use the LF 50 table entry and compare that to the datasheet max inverter current? or compare it to the calculated inverter continuous current? And over what time interval is the Load factor calculated for solar?

And the Annex Design criteria lists Rho of 60 with LF of 50. As I understand, Rho of concrete is typically 55, and average soil is 90. But the design criteria in the Annex shows Rho Concrete = Rho Earth - 5. What does that mean? And it lists Rho of PVC, cable insulation, and cable jacket. How are we to interpret how to use that info?

I know this is a complex heat transfer problem, how have you all approached this? I requested pricing on Ampcalc, but am thinking it may be too expensive.

-Ken

 

[wwhitney]

What's the time period over which the load factor is to be determined for these Neher-McGrath calculations? Obviously the worst-case load factor will be much lower computed over 24 hours than it would be computed over 30 minutes.

Cheers, Wayne

 

[electrofelon]

Something I have always been unclear on. When do we do a duct bank calculation vs just using the tables? Do we have a choice?

 

[solarken]

Something I have always been unclear on. When do we do a duct bank calculation vs just using the tables? Do we have a choice?

I interpret it that 310.14(3) requires wiring methods involving associated conductors installed in such a way that the tables or annex B prescribe derating, that we need to do the calculation.

 

[tortuga]

Designing a 2.15MW system at Industrial site. I need to interconnect on the customer side of the CT cabinet

Interesting a large scale PV system, PV output is 2.15MW ?

that the utility requires and probably cannot connect in that cabinet due to utility co rules. So looking at installing a Tap cabinet adjacent to the CT cab and connect there.

There will be 12 x 400kcmil conductors per phase in from the CT cabinet (utility feed)

I have several challenges I hope someone can help me with:
Tap cabinet will be about 400ft.

we calculate continuous current as 1.25 x Max inverter current.

I am not sure you need to do that on a 'large scale PV system' see 691.

However, everyone knows generally an inverter may output it's max for only several minutes, to a few hours max, depending on DC/AC ratio. How does LF in this context relate to the inverter current? Do I use the LF 50 table entry and compare that to the datasheet max inverter current? or compare it to the calculated inverter continuous current? And over what time interval is the Load factor calculated for solar?

I imagine your going to need to put your stamp on the design per 691.6, I'd just follow what the manufacturer recommends.

And the Annex Design criteria lists Rho of 60 with LF of 50. As I understand, Rho of concrete is typically 55, and average soil is 90. But the design criteria in the Annex shows Rho Concrete = Rho Earth - 5. What does that mean? And it lists Rho of PVC, cable insulation, and cable jacket. How are we to interpret how to use that info?

I know this is a complex heat transfer problem, how have you all approached this?

I don't have any large scale solar experience but for a load over 2 MW at a outdoor site like a mine, campus or base I would have meeting with the utility extend primary all the way your 4000A Main distribution panel (400') and set a pad mount there, or if thats not an option I would triple confirm and annoy the POCO till they say yes.
Or look at a primary service and a primary CT meter and extend primary on the 'customer side'.
If regulations wont allow anything over 5kV I would ask for a 2400V service, at 2400V I could run two sets of 1/0 MV 90 AL cable per phase.
EDIT or @ 4160V it would be a about a 350 - 400A feeder.
(I am not sure why some places will do MV but not over 5kV but it happens.)
If they don't want any 'medium voltage' at all I'd seriously pursue a 1000Y/577V system even at 1000V a 1600A duct bank would leave some room for expansion perhaps some inverters could run on a 1000Y577 system ?

 

[solarken]

So after reviewing Annex B, they have guidance on interpolation using Figure B.3 for LF and Rho that differ from the Tables. So here is what I cam up with for this project. I know this is a busy post, but would appreciate feedback as to my methodology/approach and my conclusion. Thanks.

The Inverter I am planning on using is Chint SCH200K that has max AC current output of 253A or 241A, selectable, depending on if overhead is enabled to permit up to 210kVA for PF correction. I can disable that and limit to 200kVA if needed as a last resort.

Here is interpolation Figure B.3 for reference:



Here are my calculations:

Duct Bank Ampacity from 2023 NEC Annex B
Start with LF = 100, with rational that the inverter could possibly operate at it's max current for several hrs around noon while clipping, since DC/AC ratio is above 1.2.

Since we are talking about Load Factor, want to use actual Inverter Imax current, not the calculated x 1.25 continuous current.
In this case, Imax = 253A for SCH200K when it has 210kva enabled
Use Rho = 60, and specify concrete fill

From B.2(2) Table with 9 ducts and 3 copper cables per duct:
500kcmil ampacity for Rho 60, LF 50 is 375A per cable
500kcmil ampacity for Rho 90, LF 100 is 240A per cable
Interpolate using B.3:
For Rho 60 LF50, I1 is 375 A
For Rho 90 LF100, I2 is 240 A
I2/I1 is 0.64
At bottom of B.3 chart, locate Rho 60, and follow up to intersection of LF100 and get equivalent Rho=60.
Follow Rho 60 into upper chart to intersection of 0.64 ratio to get F = 0.77
The corrected Ampacity is 0.77 x 375 = 288.75
Now this is for copper. To get ampacity for AL, look at Table B.2(5) for 9 ducts, Detail 4
500kcmil copper Rho 60 LF 50 545
500kcmil AL Rho 60 LF 50 430
AL to Cu Ratio 79%
500kcmil copper Rho 90 LF 100 387
500kcmil AL Rho 90 LF 100 305
AL to Cu Ratio 79%
So the Corrected Ampacity after applying Al to Cu Factor is
288.75 x 79% = 228.11

This is below the desired rating. So let's take LF into account, since the Inverter will not generally operate at 100% continuously
Using PVWatts, I found that the highest Load Factor over the 8hrs of 9am thru 4pm is 89%
Using this, will interpolate again:
At bottom of B.3 chart, locate Rho 60, and follow up to intersection of LF 89 and get equivalent Rho=46.
Follow Rho 46 into upper chart to intersection of 0.64 ratio to get F = 0.83
The corrected Ampacity is 0.83 x 375 = 311.25
Now this is for copper. To get ampacity for AL, use the Cu to AL Factor calculated previously:
So the Corrected Ampacity after applying Al to Cu Factor is
311.25 x 79% = 245.89
This answer 246A is just shy of 253A.

But will only have 8 conduits instead of 9 with CC conductors, and can put the N and EGC in the center conduit, so no heating will be generated in that worst case center position

Also, we are using conductors rated for 90C, not 75C, which gives additional conservative buffer
So using 500kcmil AL in outer 8 ducts run a 9x9 bank with concrete encasement might be sufficient.
Alternatively, increasing to 600kcmil AL and going back to Rho=60 and LF = 100, try to interpolate 500 to 600kcmil using Table B.2(2):
From B.2(2):
350kcmil with Rho=60 & LF=50 310 350
500kcmil with Rho=60 & LF=50 375 500
Interpolate 600, as 167% of the way from 350 to 500: 600 1.67
So 167% of the way from 310 to 375: 418
350kcmil with Rho=90 & LF=100 200 350
500kcmil with Rho=90 & LF=100 240 500
So 167% of the way from 200 to 240: 267

Repeat Interpolation using B.3, but with the above 600kcmil interpolated current values instead of for 500kcmil:
For Rho 60 LF50, I1 is 418 A
For Rho 90 LF100, I2 is 267 A
I2/I1 is 0.64
At bottom of B.3 chart, locate Rho 60, and follow up to intersection of LF100 and get equivalent Rho=60.
Follow Rho 60 into upper chart to intersection of 0.64 ratio to get F = 0.77
The corrected Ampacity is 0.77 x 418 = 321.9
Now this is for copper. To get ampacity for AL, use the Cu to AL Factor calculated previously:
So the Corrected Ampacity after applying Al to Cu Factor is
321.9 x 79% = 254.3 A
This answer is above the 251A desired value. So even when using LF = 100, 600kcmil AL has sufficient ampacity
In conclusion, 600kcmil AL has sufficient ampacity using Rho=60, LF=100, and the following just add more buffer:
Load Factor will be less than 100 over the course of an 8 hour day
With only 8 ducts populated with CCC's, the heating will be less than the Annex examples
The conductor temp rating of 90C will also provide additional buffer vs the 75C used in Annex B.

 

[solarken]

Interesting a large scale PV system, PV output is 2.15MW ?


I am not sure you need to do that on a 'large scale PV system' see 691.
I imagine your going to need to put your stamp on the design per 691.6, I'd just follow what the manufacturer recommends.

Thanks for your reply. In the NEC a large scale system as stated in 691.4(7) must have an inverter generating capacity of 5MW, so with 1.6MW of inverter in this project, it falls under 690.

I don't have any large scale solar experience but for a load over 2 MW at a outdoor site like a mine, campus or base I would have meeting with the utility extend primary all the way your 4000A Main distribution panel (400') and set a pad mount there, or if thats not an option I would triple confirm and annoy the POCO till they say yes.
Or look at a primary service and a primary CT meter and extend primary on the 'customer side'.
If regulations wont allow anything over 5kV I would ask for a 2400V service, at 2400V I could run two sets of 1/0 MV 90 AL cable per phase.

This is a behind the meter project, and interconnection needs to occur at their 480V service, after the meter, but on the grid side of the service disconnect.
I considered 800vac inverters and a transformer near the interconnection point to step down from 800 to 480, but this is not a great option due to space and construction sequence factors. If we were connecting at MV, I would definitely go with 800vac inverters which would help with the conductor sizing for the long run from the inverters to the connection point. Basically I am stuck with 480V.

EDIT or @ 4160V it would be a about a 350 - 400A feeder.
(I am not sure why some places will do MV but not over 5kV but it happens.)
If they don't want any 'medium voltage' at all I'd seriously pursue a 1000Y/577V system even at 1000V a 1600A duct bank would leave some room for expansion perhaps some inverters could run on a 1000Y577 system ?

I am not aware of any 1000V inverters but there are 600v and 800v inverters from several manufacturers.

 

[tortuga]

This is a behind the meter project, and interconnection needs to occur at their 480V service, after the meter, but on the grid side of the service disconnect.

I see well I'd still try a few more times with the POCO to get a 'primary meter' and move the service point up the chain, then extend primary to the PV system. Around here the first person I get is 'Mr No' The answer to everything odd is 'no that cant be done we've never done that and if we did we wont do it again'.
then a few more inquires gets mr 'yes but you gotta pay for that'.

In the NEC a large scale system as stated in 691.4(7) must have an inverter generating capacity of 5MW, so with 1.6MW of inverter in this project, it falls under 690.

Ahh thats big

 

[pv_n00b]

I do a lot of cable ampacity studies so here are my suggestions.
AmpCalc is fairly cheap and has no maintenance fee. Buy it and you own it. It works well for smaller projects under 50MW. Go above 50MW and I recommend CYMCAP, super expensive but it's the industry standard like PVSyst is for production analysis.
There is no average RHO for native soil. It's highly variable from site to site and the only way to know what you have is to get a geotech report on the soil. I can say a RHO of 90 is low for soil, I very rarely see it in a geotech report. If I were doing a preliminary ampacity report with no geotech I use a default RHO of 125 for the native soil to be reasonably conservative.
Without any other input we use a PV system default LF of 70%. I have not seen a detailed analysis comeback higher than 70% but a LF might get down to 65% in a detailed analysis. A LF of 50% is unreasonably low for a PV system, a 2 hour BESS would get down that low or less but not a PV system.
The maximum soil temperature is again highly variable site to site. I'll use a default of 30C if I have to, but I can usually get the soil temperature at the depth of the cable from a USDA Natural Resources Conservation Service (NRCS) regional SCAN site at https://nwcc-apps.sc.egov.usda.gov/imap/

 

[wwhitney]

Without any other input we use a PV system default LF of 70%. I have not seen a detailed analysis comeback higher than 70% but a LF might get down to 65% in a detailed analysis.

Can you comment on what time period is used for calculating LF? Does it depend on the rest of the system parameters, like rho? I could see those affecting the value of the time constant of the system.

Cheers, Wayne

 

[solarken]

I do a lot of cable ampacity studies so here are my suggestions.
AmpCalc is fairly cheap and has no maintenance fee. Buy it and you own it. It works well for smaller projects under 50MW. Go above 50MW and I recommend CYMCAP, super expensive but it's the industry standard like PVSyst is for production analysis.
There is no average RHO for native soil. It's highly variable from site to site and the only way to know what you have is to get a geotech report on the soil. I can say a RHO of 90 is low for soil, I very rarely see it in a geotech report. If I were doing a preliminary ampacity report with no geotech I use a default RHO of 125 for the native soil to be reasonably conservative.
Without any other input we use a PV system default LF of 70%. I have not seen a detailed analysis comeback higher than 70% but a LF might get down to 65% in a detailed analysis. A LF of 50% is unreasonably low for a PV system, a 2 hour BESS would get down that low or less but not a PV system.
The maximum soil temperature is again highly variable site to site. I'll use a default of 30C if I have to, but I can usually get the soil temperature at the depth of the cable from a USDA Natural Resources Conservation Service (NRCS) regional SCAN site at https://nwcc-apps.sc.egov.usda.gov/imap/

Ampcalc is $2100. Not really cheap unless there are a lot of studies to be done. Thanks for the insight into your approach!