sizing inverter and DC-to-AC ratio

[RShackleford]

I've always assumed that, since generally the panels themselves are where most of your money is going, that you would choose an inverter that is guaranteed to be able to handle the maximum DC power the panels can deliver.

However, looking at the "advanced" option in PVWatts, there is this confusing language describing the "DC to AC size ratio":

The DC to AC size ratio is the ratio of the inverter's AC rated size to the array's DC rated size. Increasing the ratio increases the system's output over the year, but also increases the array's cost. The default value is 1.20, which means that a 4 kW system size would be for an array with a 4 DC kW nameplate size at standard test conditions (STC) and an inverter with a 4 DC kW / 1.2 = 3.33 AC kW nameplate size.

For a system with a high DC to AC size ratio, during times when the array's DC power output exceeds the inverter's rated DC input size, the inverter limits the array's power output by increasing the DC operating voltage, which moves the array's operating point down its current-voltage (I-V) curve. PVWatts® models this effect by limiting the inverter's power output to its rated AC size.

Confusing, because in the first sentence it appears they're talking about AC-to-DC ratio, but they call the input DC-to-AC ratio ("X-to-Y ratio" means X/Y, not Y/X); the then in the example, it again appears they're talking DC-to-AC ratio.

Anyhow, if you figure you have the space and money to install a certain number of watts of panels, WHY would you choose an inverter which might clip the output of those panels under maximum power output conditions ? Why would you choose a DC-to-AC ratio greater than 1 ?

The language suggests cost of the inverter is the reason. In my situation, I'm going to install 4860 watts of panels, and use an SMA SunnyBoy inverter. PVWatts seems to suggest I'd want to choose their 3.8kw model, but my inclination would be to choose the 5.0kw model. The price difference is a whopping $70 (out of about $1000). Plus the larger one gives me 3 MPPTs, which might be nice for my 18 panels.

Am I missing something, is there some other downside in choosing the 5.0kw unit ?

[scrambler]

Hopefully the real experts will chime, in, but I will volunteer what I gathered so far, like that they can correct me if I am wrong

First there is the Inverter efficiency curve. The inverter efficiency is maximum towards its maximum power rating. At lower than max, the efficiency drops slightly. Given the biggest part of the day has your panels producing below their maximum, there is a sweet spot where what you may loose in clipping during the few hours of max production could be compensated by what you gain in efficiency during lower production hours.

Add to that, the fact that in most cases, your panel will rarely if ever produce their max rated power. If you look at the PVWatts hourly data, and depending on your configuration/location, you will find out that they will likely never produce more than 90% or even 80% of their max power. This combined with the above, and having an inverter with a max power rating of 90% of the panel power rating would lead to a better global efficiency.

Of course the flip side is future proofing in case there is a possibility you may need to add more panels.

[RShackleford]

Oh yeah, the maximum hourly input that PVWatts gives my 4.86kwh system (which is the STC rating) is 4050 watts. The data sheet (https://files.sma.de/dl/33902/SBxx-US-DS-en-34.pdf) for the inverters says the 3.8kw has max AC output power of 3840 watts, which is 95% of that 4050 watts. OTOH, if you look at the efficiency curve, it's a bit higher as the power drops from max to 0.4 of max. What matters a lot more is the string voltage; it's seems to fall a little as string voltage drops from 480vdc to 220vdc, and it's "rated" from 195-480 vdc. This all suggests to me that it'd be best to arrange my 18 panels as two string of 9 instead of 3 strings of 6, but that I should probably use the 5kw inverter.

[Ampster]

My installer used a 3.8 kW inverter with 5 7 kW of panels. That is close to the max allowed by the inverter manufacturer. Despite clipping during summer days I am getting close to forecasted output and above the guaranty the vendor gave me.

[bcroe]

I've always assumed that, since generally the panels themselves are where most
of your money is going, that you would choose an inverter that is guaranteed to be able to handle the
maximum DC power the panels can deliver.

That IS where the science started. Solar cells/panels were so precious that no available output energy
should be wasted. But the semiconductor industry kept advancing, and the ability to produce excellent
quality 6 X 6 inch cells very cheaply, combined with huge market volume, made panels suddenly cheap.
I am pretty sure my optimized mounts are a lot more costly than the panels on them.

For a system with a high DC to AC size ratio, during times when the array's
DC power output exceeds the inverter's rated DC input size, the inverter limits the array's power output
by increasing the DC operating voltage, which moves the array's operating point down its current-voltage
(I-V) curve.

That assumes all panels reach their peak output capability at the same time. Shading might cancel
out this or that panel at varied times, so extra panels help get closer to the inverter capability. Or
panels may take on multiple orientation (E + S + W for example) so it is impossible for all panels to
reach max output at the same time. And add enough panels to cancel DC transmission losses.

Anyhow, if you figure you have the space and money to install a certain number
of watts of panels, WHY would you choose an inverter which might clip the output of those panels under
maximum power output conditions ? Why would you choose a DC-to-AC ratio greater than 1 ?

Nice blackboard exercise, but in the real world there are lots of reasons (beyond cost) of limiting the
size of an inverter. The net metering contract may set a limit for the class of license. The electrical
feed in facilities may be limiting the power that can be safely fed in. In my case an existing 600 foot
AC loop of 4 gauge was trenched in place long before solar. Running 58A AC caused an undesirable
loss of more than 3%, and more current would push the the wire and circuit breakers to an unworkable
state. As it was a 9V loss across that wire added to high line voltage to sometimes cause over line
voltage trip out.

As others mention, the amplitude and time product of a smaller amount of over paneling will actually
cause quite a small loss. Seen another way, adding some panels can help a lot on all but the peak
time of day, and all the time on less than perfect sun days.

Finally the AC output of a 96% efficient inverter requires a larger DC input of 1 divided by 0.96 times
the AC just to reach that limit. Bruce Roe

[RShackleford]

I should point out that the 3rd and 4th paragraphs of my OP are just quoting PVWatts' explanation of the the DC-to-AC ratio input.

Panels are cheap, yeah, in fact I'll probably pay $0.36/watt for my panels; but I'm not willing to go bigger than 18 panels.

Nice blackboard exercise, but in the real world there are lots of reasons (beyond cost) of limiting the
size of an inverter. The net metering contract may set a limit for the class of license. The electrical
feed in facilities may be limiting the power that can be safely fed in. In my case an existing 600 foot
AC loop of 4 gauge was trenched in place long before solar. Running 58A AC caused an undesirable
loss of more than 3%, and more current would push the the wire and circuit breakers to an unworkable
state.

Understood. I don't think any of this is a concern for me though, as the 5.0kW model maxes out at 21 amps and it's no problem for me to run 10awg to a 30amp breaker; the distance is 50ft max, about 2 volts dropped.

So if I may ask, which inverter (SunnyBoy 3.8kW or 5.0kW) would you suggest I use for my 4.86kW of panels ? For my situation, doesn't seem like there's much downside to going bigger, other than the $70 price difference. A minor upside of not getting clipped on "perfect" days. And a bigger upside of having the option of running 3 strings instead of 2, if I decide that makes sense (though I'm kinda thinking I won't, but I'd appreciate your opinion on that as well).

[bcroe]

So if I may ask, which inverter (SunnyBoy 3.8kW or 5.0kW) would
you suggest I use for my 4.86kW of panels?

Have no experience here to make that recommendation, my ancient Fronius inverters have
been doing great for half a dozen years. Bruce Roe

[scrambler]

That is between under sizing the inverter by 22% versus oversizing by 3%.
22% undersized seems a bit extreme to me.

[RShackleford]

That is between under sizing the inverter by 22% versus oversizing by 3%.
22% undersized seems a bit extreme to me.

Well, it's really not undersizing by that much, because even though my panels are 4860 watts (18 * 270), these are STC test conditions, which are considered pretty unrealistic. Under the much more stringent NOCT conditions, the panels are 199 watts for a total of about 3600 watts. They are not rated at the intermediate PTC test conditions. PVWatts' hourly output puts the maximum DC power at 4492 watts (this is with the default system loss parameters, shading only 3%). Still 15% undersized though.

There seems to be a conentional wisdom that it's ok to see a little clipping during the few hours when the array is producing maximum power (the 4492 watts), because there is a correspoding advantage during the lower-production hours. Bruce gives some arguments above (post #5), but I don't think they apply to me. I've read that inverters are more effiicient near full capacity, but the data sheet (https://files.sma.de/dl/33902/SBxx-US-DS-en-34.pdf) for the line I'm considering does not bear that out. Bruce also points out that with an inverter efficiency below unity, the DC array can be oversized by that much without loss; for the SunnyBoy, at something like 97.5%, that means the array can out out somethijng like 3900 watts without the 3.8kW model losing anything. But that's still a lot.

I am still leaning towards the 5.0kW model.

[scrambler]

I think that in the end, you are right in the middle of both scenarios, and either way, I doubt it will make a significant difference (and yes, that is a feeling, not a scientific assessment ).

If your shade is not all the time but only in some specific hours of the day, I would run PVWatts number without shade impact, as PVWatts probably applies the shade reduction globally, so the max production number it gives you does not reflect what your array will produce at max when there is no shade.

[RShackleford]

If your shade is not all the time but only in some specific hours of the day, I would run PVWatts number without shade impact, as PVWatts probably applies the shade reduction globally, so the max production number it gives you does not reflect what your array will produce at max when there is no shade.

The number for peak DC power of 4492 watts (which I quoted above) is with all the system losses set to the default, except the shading set to zero (so total of 11.42% losses).

So I could still get some pretty significant clipping with the 3.8kW model.

I remain mystified as to why there's this conventional wisdom (at least it appears so, from PVWatts and from some of the comments here) that the inverter should be a bit smaller than the maximum DC output power of the panels. The arguments seem to be as follows, and all appear not to apply for my situation:

1. Bigger inverter is more expensive: yes, but only $1100 instead of $1030.

2. Bigger inverter's max AC output may stress downstream elements of the system (post #5 above). In some systems, perhaps so; for my choice, it just means going from a 20amp breaker and 12awg wire to a 30amp breaker and 10awg wire - not a hardship.

3. Inverter may operate more efficiently near 100% loading. Maybe for some types, but that's not the case with these models, per the data sheet: https://files.sma.de/dl/33902/SBxx-US-DS-en-34.pdf
There's actually a very slight drop, though less than 1%, in efficiency in going from 0.4 times to full output power.

[RShackleford]

Inverter may operate more efficiently near 100% loading. Maybe for some types, but that's not the case with these models, per the data sheet: https://files.sma.de/dl/33902/SBxx-US-DS-en-34.pdf
There's actually a very slight drop, though less than 1%, in efficiency in going from 0.4 times to full output power.

Actually though, the efficiency does drop fairly precipitously below about 10% of rated output power (see graph below). This would occur more often for the larger 5.0kW model.

Screen Shot 2019-11-12 at 8.09.23 PM.png

Actually, PVWatts does seem to model this effect. If I look at DC output versus AC output for my system (4.86kW peak DC output), with DC-to-AC size set to 1.2, and sort the hourly results by descending DC power, the AC output peaks at 4050 watts, which is exactly 4.86/1.2. As the DC power gets less, the observed AC-to-DC ratio quickly increases to 96%, which is PVWatts' default value for inverter efficiency. But then as the DC output drops below 1kW or so, the observed AC-to-DC ratio starts to droop below 96%, down to only 90% at 300 watts DC power, and dropping sharply below that.

So I guess what I really need to do is to compute my own column of AC output power, modulating the inverter's efficiency per the curve for the SMA models.

[bcroe]

Below 10% power, the total amount of energy involved is getting so small that the drop
in efficiency counts for too little to matter.

My Fronius is actually a large and a small inverter, so it can be efficient over a very large
range. Bruce Roe