Optimization of large solar park DC and AC collectors

[ters]

Has anyone dealt with optimization of number and size of inverters for large scale solar projects. For example, if a 10MW solar park is located on a land parcel with regular dimensions, is it more economical to use 20 500kW inverters each covering 5% of the area, or 10 1MW units each covering 10% or 5 2MW units (provided that they exist at all) each covering 20%.

It is obvious that that 5 large units with cost much less than 20 small ones. However, in case of 2MW units, each would be serving a large area of 5-7 acres (2.5-3 hectares), where DC cable will be extremely long and unless heavily oversized (up to 3 times nominal current capacity) losses will be way too high. So factoring in tens of kilometers of very thick cable and increased losses on DC side, any savings achieved buying big inverters seems to melt.

With smaller inverters serving much smaller area, DC cables are becoming much shorter (and as such don't have to be extremely thick). On the other side, HV collector cable becomes longer, but the lenght is still very reasonble and losses on the HV side are much lower anyway.

I know that this depends a lot on the DC collector voltage, type of PV panels, number of combiner boxes, inverter+transfomer cost, etc, but just wondering if anyone on the forum have dealt with this before.

 

[Skogsgurra]

From what I have seen - not so many different plants, I must admit - there seems to be a preference for highest practical DC voltage and inverter units in the 1 - 2 MW range. At 690 V DC, the cables needn't be too heavy and losses are kept reasonable low.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[ters]

On a particular project where I’m trying to come with some preliminary idea as how to run the collector, somebody else already selected 2MW units and 600V DC collector, as a concept which might change.

But after reviewing cable lengths, from what seems to be the most optimal locations of both, combiner boxes and inverters, it appear that average length of circuit cable for both poles would be in the range of 300 meters (both ways), resulting in over 50km of such cable, which is to carry about 140 Amp. Then, if cable sufficient to curry current is selected, losses will still be over 2%, just on that cable-combiner-inverter...

Increasing the DC collector voltage to 900V would help a lot, but that comes with another offset – that voltage level is already classified as “high voltage” (local jurisdiction code), where there are other cost implications – the whole solar park area would be treated as a HV facility.

 

[ters]

And also another question re big vs. smaller units is reliability and availability. Inverters are known for higher failure rates, so it seems still more optimal to loose say 1MW at the time, than 2MW.

 

[itsmoked]

There may be some intangibles too. With smaller units you may be able to put smaller sections onto the grid sooner as the systems are installed. This can get you testing the first install which might help you refine the installation of the rest of the groups. This could help get the cash flow in the right direction faster. Testing, if any is needed, would be easier with smaller inverters.

I never stopped to think about stringing solar panels together up to voltages like 600V! Seems a little scary to me when I think about all those little scrawny traces in the solar cells and what, any kind of fault would do to them.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[ters]

That scares me too. And how about 1000V strings. Imagine a short circuit in a combiner box...

Your idea is definitely worth exploring, but that requires the substation to be built and commissioned first, and looks like the present practice is that it usually gets done last. The commissioning is very extensive, every string has to be tested and defective panels identified. Depending on the panel voltage, there may be 3000-4000 strings for 10MW capacity...

These projects are more political than are based on some technical sense and rely on huge incentives from the tax payer pocket, being me ...

 

[Skogsgurra]

Hey Smoked - this is what they look like. This one uses around 600 V DC. The voltage doesn't cause any problems.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[dpc]

There's a local outfit who is building inverters for 1000 V dc for solar systems. I figured this must be some kind of sweet spot for the power electronics in terms of cost versus capacity?

David Castor

http://www.cvoes.com

 

[ters]

I'm sure there is a sweet spot, but what is it . Seems to depend on several variables, type and ratings of panels, kW/acre (or hectare) density of panels, inverter design, DC collector voltage, local jurisdiction code, etc.

Generally, it seems to me that units over 1MW are to be avoided. In case of 2MW units, one way or the other, power from the most remote panels, which can be as far as 250m from the inverter has to come to the inverter somehow in which case I see no simple way that losses can be low at 600V, unless wires are really big. But it may be debatable what are “low” losses. Seems that at 600V, DC cables will still be of reasonable size if losses of about 1.5-2% are allowed. But 2% may be a bit high just for DC wire and cable losses. And by the time all other DC an AC losses are added, derating may get to 10%.

 

[itsmoked]

Thanks Gunnar.

dpc; The maximum common switching voltage for semiconductors is presently 1,600V. So that seems to make 1000V about the top spot before you have to start doing a lot more.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[dpc]

Keith,

Thanks - that makes a lot of sense.

Dave