Sizing Photovoltaic Output Circuit Currents

[illini1022]

My primary area of expertise is medium/high voltage power systems, so this is a bit new to me. I'm doing some cursory checks on the sizing of DC conductors at a large solar facility. I'm not the design engineer or independent engineer - I'm really just trying to learn by reviewing these calculations.

Let's say I have modules with a short circuit current rating (at STC) of 9A. Perhaps I'm combining 20 strings in a combiner box and taking it into the input of a large PV inverter. That means the output cable from that combiner box (aka the "PV Output Circuit Current) has an Isc (at STC) of 180A. My read of NEC 690 tells me that I need to take the following steps.

• Maximum Circuit Current (690.8A1) = 180A * 1.25 = 225A. This 125% factor is primarily intended to account for the fact that under certain conditions (irradiance/temperature higher than STC), the effective short circuit current may be higher.

Determining Conductor Ampacity

• ​Per 690.8B
  1. "PV system currents shall be considered to be continuous." I interpret "continuous" to mean that no load/demand factor is applicable? This seems to be a point of some contention in the industry?
  2. "Circuit Conductors shall be sized to carry not less than the larger of 690.8B1 or 2." This tells me we need to perform two specific checks and choose the larger conductor of the two.
     1. ​For 690.8B1, am I correct in saying that you now must choose a conductor from the NEC tables that can carry 125% of the maximum circuit current calculated above. Effectively - you are using 156% of the STC ISC and picking an appropriate conductor without using any adjustment/correction factors. Presumably, this is to establish a bare minimum conductor size that also aligns with the fact that you'll choose an over-current device rated at, at least 156% of ISC.
     2. For 690.8B2, you must use the appropriate adjustment/correction/derating factors but the current you compare to remains 125% of the ISC - there is no additional factor of 125% needed for this check.
        1. Here is where I start to get a little confused. First off - is my interpretation correct?
        2. Secondly, for complex large setups, it seems more appropriate/accurate to model the exact cable configuration in a software such as Cymcap or Etap. If so, I think the correct input current to use would be ISC*1.25. The NEC derating factors alone don't necessarily account for complicated duct-bank type setups with multiple neighboring conduits containing multiple conductors, etc. Is it appropriate and per code to use a cable simulation software instead of derating tables? Is there some kind of blanket exemption where "engineering guidance" trumps the exact written code for situations like this?
           1. Digging into the cable simulation scenario a bit more - these software packages will typically ask you to define a load profile or load factor. Obviously these solar cables won't be loaded to 100% 24 hours a day, so if you are making a realistic simulation to check cabel temperature I would think it's technically sound to define a worst case loading profile that simulates the cables running at a very high load for daylight hours and then cooling back down during the night. This loading profile/factor can have a major impact on cable sizing, especially when you are evaluating a choke point such as the entry to an inverter with many DC inputs. When dealing with large cable sizes such as these (500-1000kcmil), the cooldown periods definitely affect conductor ampacity. However, the NEC does say that PV currents are "continuous" so does that present any problems to this logic?

Sorry for making this a rather wordy post, but I'd appreciate any input/opinions. Thanks.

 

[jaggedben]

I think you have pretty much nailed it. Don't see anything I think is wrong.

... Secondly, for complex large setups, it seems more appropriate/accurate to model the exact cable configuration in a software such as Cymcap or Etap. ...

I don't recall the exact language, but I believe the 2017 NEC is going have allowances for this sort of thing on large systems.

 

[illini1022]

I think you have pretty much nailed it. Don't see anything I think is wrong.



I don't recall the exact language, but I believe the 2017 NEC is going have allowances for this sort of thing on large systems.

Just looked up the draft of NEC 2017, that's quite interesting, thanks for the observation. It looks like there will be an entirely new article 691 for large scale (>5MW) PV plants and the current language seems to address these very issues by allowing for thorough engineering to take place, rather than just following 690.

Now I just need to convince myself what an appropriate load factor would be to use for solar cable sizing.

 

[jaggedben]

Now I just need to convince myself what an appropriate load factor would be to use for solar cable sizing.

I would think that it's not appropriate to use one, since during the summer time you are likely to see constant maximum current for several hours in the middle of the day, day after day, if there's no weather. However, I don't know much about theory behind load factors and cable sizing.

One thing that could be considered, though, is that in actuality the current will be limited by the inverter input rating during normal operation. The inverter is not going to pass more current than it's firmware allows, even if the array has the potential to deliver it. So it's really not that realistic that 690 requires us to use the maximum current theoretically possible from the array. This is especially true on larger systems where the DC to AC ratio tends to be higher.

 

[illini1022]

I would think that it's not appropriate to use one, since during the summer time you are likely to see constant maximum current for several hours in the middle of the day, day after day, if there's no weather. However, I don't know much about theory behind load factors and cable sizing.

One thing that could be considered, though, is that in actuality the current will be limited by the inverter input rating during normal operation. The inverter is not going to pass more current than it's firmware allows, even if the array has the potential to deliver it. So it's really not that realistic that 690 requires us to use the maximum current theoretically possible from the array. This is especially true on larger systems where the DC to AC ratio tends to be higher.

Understood on the 2nd point, although there is always the possibility that the firmware/inverter malfunctions and doesn't properly limit the input current I suppose.

Regarding the 1st point - in something like Cymcap you could actually establish a 30 day loading profile, input the maximum array current and simulate a worst case scenario of hourly loading and monitor the cable temperature and ensure it never exceeds its rating. I think you can justify something like this from an engineering perspective. 100% agreed that the cable needs to be capable of carrying the maximum current for many consecutive hours without exceeding the temperature rating.

 

[Andrew445]

Understood on the 2nd point, although there is always the possibility that the firmware/inverter malfunctions and doesn't properly limit the input current I suppose.

Regarding the 1st point - in something like Cymcap you could actually establish a 30 day loading profile, input the maximum array current and simulate a worst case scenario of hourly loading and monitor the cable temperature and ensure it never exceeds its rating. I think you can justify something like this from an engineering perspective. 100% agreed that the cable needs to be capable of carrying the maximum current for many consecutive hours without exceeding the temperature rating.

I would say 50% load factor is a conservative but fair assumption. I messed around a bit with the transient thermal stuff in ETAP in order to get a feel for it. I'm just starting to get more into CYME which appear to me more complex. The appropriate selection of RHO value appears to play a significant role as well.

P.S. Can Cyme model DC cables?

 

[illini1022]

I would say 50% load factor is a conservative but fair assumption. I messed around a bit with the transient thermal stuff in ETAP in order to get a feel for it. I'm just starting to get more into CYME which appear to me more complex. The appropriate selection of RHO value appears to play a significant role as well.

P.S. Can Cyme model DC cables?

I feel safe using a load profile in Cyme for these purposes, but I'm unsure of how exactly ETAP treats a 50% load factor, for example? The cables do need to be sized to carry 100% design current for multiple hours at times, so I'd hope that load factor doesn't result in a complete haircut of the maximum for sizing purposes?

Cyme can handle DC just fine. RHO makes a huge difference. You need to make sure you are using the correct value and also consider the backfill compaction specification (for trenching) as
well as the accurate dry-out characteristics. Equally important is using the appropriate ambient earth temperature. You also need to decide whether to use a uniform or multi-zone soil model.

Regarding the load profile in Cyme - I attached a snapshot of how this works over a 48 hour period. We also check much longer duration. http://i.imgur.com/QfgKyCE.jpg

 

[Andrew445]

I feel safe using a load profile in Cyme for these purposes, but I'm unsure of how exactly ETAP treats a 50% load factor, for example? The cables do need to be sized to carry 100% design current for multiple hours at times, so I'd hope that load factor doesn't result in a complete haircut of the maximum for sizing purposes?

Cyme can handle DC just fine. RHO makes a huge difference. You need to make sure you are using the correct value and also consider the backfill compaction specification (for trenching) as
well as the accurate dry-out characteristics. Equally important is using the appropriate ambient earth temperature. You also need to decide whether to use a uniform or multi-zone soil model.

Regarding the load profile in Cyme - I attached a snapshot of how this works over a 48 hour period. We also check much longer duration. http://i.imgur.com/QfgKyCE.jpg

View attachment 14342

Thanks for the response! Agreed that the cable needs to carry the full current for peak solar hours, but recall that a cable's ampacity is basically it's temperature limit. In reference to your graphic, how do the cable peaks compare to the temperature reached at steady-state full load? I would imagine the transient peaks are quite a bit less. If an ETAP "haircut" was set at a value corresponding to a steady-state temperature that is greater than that reached by the transient peaks, then a conservative result could be found. All that aside, I think using the load profile is awesome, and I will look into how to go about that with Cyme once I get more familiar. I'm hoping PVSyst .csv output files could be directly translated to a load profile.

In regards to thermal modelling, are you assuming 0% moisture exists directly around the cables, then maybe 3% further out, etc.? Secondly, how do you arrive at earth ambient temp.? Most geotech. reports provided to me do not include these values.

Thanks again!

 

[illini1022]

Thanks for the response! Agreed that the cable needs to carry the full current for peak solar hours, but recall that a cable's ampacity is basically it's temperature limit. In reference to your graphic, how do the cable peaks compare to the temperature reached at steady-state full load? I would imagine the transient peaks are quite a bit less. If an ETAP "haircut" was set at a value corresponding to a steady-state temperature that is greater than that reached by the transient peaks, then a conservative result could be found. All that aside, I think using the load profile is awesome, and I will look into how to go about that with Cyme once I get more familiar. I'm hoping PVSyst .csv output files could be directly translated to a load profile.

In regards to thermal modelling, are you assuming 0% moisture exists directly around the cables, then maybe 3% further out, etc.? Secondly, how do you arrive at earth ambient temp.? Most geotech. reports provided to me do not include these values.

Thanks again!

I think I mostly follow what you are saying in the first part there. You're right - the transient peaks are quite a bit less and the end results aren't drastically different... Using a load factor of .5 and no load curve, we got a cable temperature of 72C. Extrapolating out the load curve for 6000 hours the daily maximum cable temperature becomes roughly 80 degrees. So I do think you need to be careful not to low-ball it when using a load factor.

I forget the name of the database off-hand but we rely on historical USGS type data for earth temperature. Usually do maximum observed temperature + 5C for buffer or something along those lines. Using the maximum on record bakes in some conservatism right there. Sometimes we'll use 0% moisture uniform soil model which is obviously quite conservative. Other times we'll go two-zone and assume 0% moisture for the trench width (perhaps 1') and in-situ thermal rho for outside the trench.

I work for an owner/operator so I'm generally thinking long term and don't mind a healthy dose of conservatism. Large cables help us a bit on losses as well.

If you talk to an EPC their approach might be a bit different.