Transformers to accompdate voltage drop

[Joulio]

Hi all,

I was hoping someone could clarify the use of transformers to mitigate voltage drop. A project that has two separate arrays (1.616MW & 1.44MW), but only one utility transformer interconnection point close to the 1.616MW array. The issue is getting the power from the 1.44MW array back to the utility transformer which is around 1000' from the interconnection point.

If anyone has thoughts on how to best tackle the issue it would be much appreciated.

Thanks

 

[Smart $]

1Ø, 3Ø?

Inverter output voltage?

POCO voltage?

 

[GoldDigger]

Run the 1000' at the highest voltage possible and use three phase if available from POCO.
And remember that although the energy loss will be the same, the voltage at the inverters will be higher than the grid voltage by the amount of the voltage drop.

mobile

 

[electrofelon]

Pretty sparse details, but my first thought is it comes down to weighing the cost of the following:
1. Weighing the cost of energy lost against the cost of conductors and finding the best long term balance.
2. Weigh the cost of the above against other options like the utility extending their line.

As GD said, run the highest voltage, which assuming 480 inverters and 1kv strings would mean run the strings back (but yikes thats a lot of strings..cable tray?). Note Aluminum PV wire is available. I hope however designed this factored this in to the cost.

Now if you are concerned strictly about the "voltage rise" bringing you out of the inverter's voltage window, I suppose you could use an autotransformer on the inverter outputs, but perhaps a much cheaper option would be to see if the utility could change the taps on their transformer to lower the secondary voltage. Either way you would have to be careful that then you dont hit the lower end of the voltage range during low output and low grid voltage conditions.

 

[Joulio]

Details - The design is driven by multiple factors so it isn't ideal. The system will use only string inverters (conductors are all aluminium where possible). All inverters are 25kw @ 480v/277 33a inverters, the array with the distance issue has 40 inverters combined using a 1200a & 800a combiner feeding 2 suitably sized AC Disconnects. Poco is 12kv at utility transformer.

Thanks for the previous response.

 

[Joulio]

Disregarding the voltage drop question. What are the main parameters for sizing a utility transformer for a solar site of roughly this size?

 

[PVfarmer]

getting the power from the 1.44MW array back to the utility transformer which is around 1000' from the interconnection point.

run the strings back (but yikes thats a lot of strings..cable tray?). Note Aluminum PV wire is available.

What are the main parameters for sizing a utility transformer for a solar site of roughly this size?

I'd say make the combiners to inverter the 1000' run, offhand.

Those size (1000 or better, 1500kVA) xfmrs are *very* expensive!
Without looking up any specifics, you might pay as much for the xfmr without its 1000' feet of (AC) conductors as you would for 160,000' of 2 AWG Al for higher voltage DC.
(160,000' based on 40 inverters, so maybe 2 inputs/4 conds. each, so 4 x 40 x 1000).

At around 1/15 the size of your project and ½ the distance, 4 inverters (20kW ea.,480/277V) going 500' so... 16 conds. total at 4 AWG/800VDC is better than going the 500' at close to 100A and 480/277VAC. (so in my example, 8000' of 4 AWG vs. 2000' of...i think 2/0? the 8000' wins...)

For me. Using copper. Hope that helps.

 

[electrofelon]

Disregarding the voltage drop question. What are the main parameters for sizing a utility transformer for a solar site of roughly this size?

The utility should be sizing that. A recent 1.3 MW I worked on, the utility sized it to the sum of our inverter outputs' KW. I was a little surprised they didnt "overload it" a bit.

 

[ggunn]

The utility should be sizing that. A recent 1.3 MW I worked on, the utility sized it to the sum of our inverter outputs' KW. I was a little surprised they didnt "overload it" a bit.

Why would they do that? Inverters are current limited devices.

 

[Smart $]

... I was a little surprised they didnt "overload it" a bit.

Why would they do that? Inverters are current limited devices.

I'm thinking he was surprised the POCO didn't undersize the xfmr compared to inverter output sum. Trannies can handle moderate overload for a few hours per day and not even blink on its MTBF.

 

[electrofelon]

I'm thinking he was surprised the POCO didn't undersize the xfmr compared to inverter output sum. Trannies can handle moderate overload for a few hours per day and not even blink on its MTBF.

Correct, that is what I was thinking. Anyone else have any solar farm utility transformer info to share for academic purposes?

 

[PVfarmer]

1

only one utility transformer interconnection point close to the 1.616MW array.

2

The system will use... Poco is 12kv at utility transformer.

3

the main parameters for sizing a utility transformer for a solar site of roughly this size?

4

The utility should be sizing that.

5

not even blink on its MTBF.

So... the PV is not installed yet- what about the utility xfmr?
The "is" in #2 makes it sound as if the POCO xfmr is there already. If so, do you mean POCO is 12kV and customer side is 480/277V? Or, which side is "at" in #2?

If you did have 12kV service, meaning customer side = 12kV, you'd have to buy *two* massively pricey xfmrs, one at the point you mention in #1 and the other 1000' away.

So about #4- the POCO should be *paying* for that! :?

And about #5... wouldn't the efficiency suffer so much when overloaded that the "customer" (side) would want to oversize, *if* any losses showed up on the meter?
Rough/smaller example- if you wanted to step 50kVA of 240V inverter output up to 480V, you might want to use a newer, "exceeds D.O.E. @ 35-65% of rated KVA" xfmr, sized 75kVA. 50 being 66% of 75.

Hypothetically, the 75kVA choice would be at 98% efficiency every day for the one/few hours during peak, while a 50kVA xfmr for 50kVA of inverters might run at X% less every day(-ish?) for those same peak hours?

 

[PVfarmer]

Correct, that is what I was thinking. Anyone else have any solar farm utility transformer info to share for academic purposes?

If you ctrl-f (search text) this for kVA, there's some info.
https://www9.nationalgridus.com/non_html/shared_constr_esb756.pdf

This part in italics below is not for all of the states (NY, MA, RI etc.) referred to in the pdf, so...?

Using software that does an hourly spreadsheet of PV output... 1000kVA of PV will almost never put out the full 1000KVA, but it will do over 900kVA/90% for quite a few hours of the year. Depends where you are, of course, and panel angle, 30 deg. is ideal where I am.

If the customer owned a PV xfmr, I'd at least look into oversizing it- say you for some odd reason wanted to use 208/120V inverters with 480/227V service and loads, or step 480/277V down for 208/120V loads- a more efficient xfmr means less PV to loads, more to grid?

If the D.O.E+ efficiency models pay for themselves just when supplying loads, the added PV shortens the payback, I believe.

And, I've mentioned this before- for SMA inverters (only 3 phase I think?), the manual says the PV kVA should be <= 90% of the step-up xfmr kVA. Of course it doesn't say why!


Inverter-based systems such as Photovoltaic (PV) Systems are limited in aggregate to 500kVA on 4 or 5kV and in aggregate of large units 500kVA and above up to 3.0MVA on 15kV class systems (this is in addition to small (e.g. residential rooftop) PV until aggregate of these exceeds 500kVA). Operating issues on EPS voltage regulation occur from the effects of cloud transients on large PV systems.

(My note: So either RI is cloudier than NY or MA, or the EPSs in RI are smaller- probably the latter?)

 

[ggunn]

I'm thinking he was surprised the POCO didn't undersize the xfmr compared to inverter output sum. Trannies can handle moderate overload for a few hours per day and not even blink on its MTBF.

Oh, right. In my haste I read it as meaning just the opposite. Never mind.

 

[electrofelon]

1

2

3

4

5


So... the PV is not installed yet- what about the utility xfmr?
The "is" in #2 makes it sound as if the POCO xfmr is there already. If so, do you mean POCO is 12kV and customer side is 480/277V? Or, which side is "at" in #2?

If you did have 12kV service, meaning customer side = 12kV, you'd have to buy *two* massively pricey xfmrs, one at the point you mention in #1 and the other 1000' away.

So about #4- the POCO should be *paying* for that! :?

And about #5... wouldn't the efficiency suffer so much when overloaded that the "customer" (side) would want to oversize, *if* any losses showed up on the meter?
Rough/smaller example- if you wanted to step 50kVA of 240V inverter output up to 480V, you might want to use a newer, "exceeds D.O.E. @ 35-65% of rated KVA" xfmr, sized 75kVA. 50 being 66% of 75.

Hypothetically, the 75kVA choice would be at 98% efficiency every day for the one/few hours during peak, while a 50kVA xfmr for 50kVA of inverters might run at X% less every day(-ish?) for those same peak hours?

Yes we were given very few details, but for a system of this size, Occam's razor says that the inverters are most likely 480 3 phase, the utility is providing service at 480, and supplying the transformer.

 

[jaggedben]

Some analyses suggest that avoiding voltage drop on the DC side is not cost effective.

http://solarprofessional.com/articl...-selection-in-pv-designs?v=disable_pagination
http://solarprofessional.com/articl...voltage-drop-conventions?v=disable_pagination

If you're actually getting 480V from the utility then running the DC 1000' may well be the most cost effective solution. (I would pay attention to electrofelon's comment about cable tray.)

I think the most important consideration, if going this route, is just to check to make sure that DC voltage doesn't drop below the inverter MPPT threshold. That could cause far more energy loss than the voltage drop itself.

 

[PVfarmer]

Yes we were given very few details, but for a system of this size, Occam's razor says that the inverters are most likely 480 3 phase, the utility is providing service at 480, and supplying the transformer.

Some analyses suggest that avoiding voltage drop on the DC side is not cost effective.

Thanks for those links, jb.
I'm still pulling for DC for 1000'.

I just tried this (with Helioscope).
With an array 1000' from a PCC... (western (straight) edge of array is 1000' east of PCC to be precise...)

Using 4 inverters (24kW and 480/277V) and 4 combiners, combiners at the western edge of the array both times....

1. With inverters at (or within 20 feet of PCC, so 4 inverters > AC combiner panel > PCC are adjacent), and 4 current carrying conductors per inverter (2 MPPT inputs per inverter), you have 16,000 feet of DC between the DC combiners and the inverters' DC inputs. The software says 1.8% line losses using 4 AWG for that 16,000 feet. (Really 1000' distance)

2. Then! If you move the inverters and the AC combiner panel 1000' so they are adjacent to the DC combiners, you are using 4000 feet (for the same 1000' dist.) of...wait for it..
500kcmil for the same 1.8%! Which costs roughly around 11x as much $$ as 4 AWG in copper.

And then maybe 4000' of EGC for #1 scenario and only 1000' for #2, but the EGC for #1 (DC) can be 10 AWG instead of...quite a bit larger for an AC run!

Thoughts? There are other cost factors of course.

Edit: Whoops! So when using 25% as much conductor, yet it being 11x as much $$, I came up with the 500kcmil would be 2.6x times a much $$ overall as 4 AWG, without other factors.

 

[jaggedben]

Thanks for those links, jb.
...

1. With inverters at (or within 20 feet of PCC, so 4 inverters > AC combiner panel > PCC are adjacent), and 4 current carrying conductors per inverter (2 MPPT inputs per inverter), you have 16,000 feet of DC between the DC combiners and the inverters' DC inputs. The software says 1.8% line losses using 4 AWG for that 16,000 feet. (Really 1000' distance)

...

You don't add together the length of multiple parallel circuits when calculating voltage drop. You do add the positive and negative together. So the correct length to use for one DC circuit would be 2000'. Amps may vary with each MPPT.

 

[PVfarmer]

You don't add together the length of multiple parallel circuits when calculating voltage drop. You do add the positive and negative together. So the correct length to use for one DC circuit would be 2000'. Amps may vary with each MPPT.

Well, the software is doing the calcs.
But yes, I did set it up as 8 DC "home runs"/circuits (2 for each combiner->inverter, 4 inverters = 8), equalling 16,000ft total.

16 strings total with 4 strings/8 wires connected to each combiner's inputs, and 2 outputs/4wires exiting each of the 4 combiners, meaning 16 total exiting, 8+ and 8- to the inverters.

Each circuit is I think 19.5 amps at STC.
So the software isn't adding them, it's just saying "line loss" for one/each 1000' foot DC circuit (pretty sure it uses 2000' there) at 19.5A with 4 AWG is 2.0%.

Check this out. I still say 4 AWG wins, but 6 AWG squeaks in too.

Is 1 extra MWh a year worth buying 4 AWG over 6 AWG? Maybe.
Is 1 extra MWh a year worth paying an extra $10,000 to get 2 AWG instead of 4 AWG? Nope.

4 AWG vs 500kcmil? Well, 4 AWG costs 3x less and puts out a hair more so... of course!

 

[ggunn]

Well, the software is doing the calcs.
But yes, I did set it up as 8 DC "home runs"/circuits (2 for each combiner->inverter, 4 inverters = 8), equalling 16,000ft total.

16 strings total with 4 strings/8 wires connected to each combiner's inputs, and 2 outputs/4wires exiting each of the 4 combiners, meaning 16 total exiting, 8+ and 8- to the inverters.

Each circuit is I think 19.5 amps at STC.
So the software isn't adding them, it's just saying "line loss" for one/each 1000' foot DC circuit (pretty sure it uses 2000' there) at 19.5A with 4 AWG is 2.0%.

But that is 2% of that circuit, not of the whole array. You do not add those 2% numbers to get the total losses for the system. If you have four strings that each lose 2%, your total losses are the same: 2%.