Panels vs Inverter - how much oversizing makes sense?

Qyburn

Hi,

I'm trying to get an idea on sizing, if we give up on anything more sophisticated and just stick to G98 limits. Suggestions from installers range from 4.2kW to 6.8kW, all using a 3.6kW inverter. 

Are there any tools or rules of thumb to estimate how many hours per year, or what percentage of the time, a given panel/inverter combination would be oversubscribed?
 

Netexporter

A rough rule of thumb is 20%. However, with a 3.68kW hybrid inverter and DC-coupled battery you could export/use the 3.68kW and the "clippings" would go direct to the battery.

Alnat1

I often delay filling my battery until lunchtime on blue sky days to cut down on the clipping. I can either set to only fill to 20% in the morning and change it to 100% later or I can use another option to fill the battery at a slower rate all day.

It will depend which way your roof faces though, my SW panels catch the later rays.

Qyburn

Cheers. Do you know roughly what percentage of the time a panel will generate it's "kWp" output, vs what percentage say 50% or 75% of that rating?

What I'm getting at is let's say I oversubscribe by 25% then at any time they're generating no more than 80% of their rated output, I'd be getting full value from that extra capacity.

For my simplified plan I've pretty much given up on a battery. As far as i can tell any sort would need grid monitoring and I simply don't get straight answers as to how that could be achieved or what we'd have to do to enable it to be installed.

Netexporter

The inverter limits your export to 3.68 whether it come from the battery, the panels or a mixture.

Exiled_Tyke

I really wish I could find the report I read from a PV farm operator so I could quote it accurately.  I think it said in the UK best results are achieved from an inverter around 75-80% of maximum output. 

However for domestic systems I think there are two other considerations:

1.  As @Netexporter says, a DC Battery will be able to make use of exta output in certain conditions.

2. If you have panels in different orientations then there's even less of an issue as they will peak at different times.

It is not possible to answer the question on how often clipping occurs. It will depend on different factors. 

My 3kW invertor very very occasionally produces at 3.3kW (most inverters will over produce if needed for short times) but this happens to me very rarely.  However my 3.3Kw of panels are on two different orientations (W and S) and so I get a longer day of production but lower peaks.

QrizB

Qyburn said:

Cheers. Do you know roughly what percentage of the time a panel will generate it's "kWp" output, vs what percentage say 50% or 75% of that rating?

You'll nearly-never get the rated output from a panel.

The rating is based on a particular illumination (essentially full sun), falling perpendicularly onto the panel, with the panel at 25C. If the panel is not pointing directly at the sun (and for a fixed array it'll only do that twice a year, at most) and if it's hotter than 25C (and panels are dark-coloured, so get hot in the sun, hot enough to burn your skin) it'll produce less power than the nameplate shows.

Typically the highest power output from my array is on cool, clear spring days in March or April.

ABrass

Is this all perfectly south facing? Mine (see signature) works out very well thanks to the split. It tops out at 6.6kW but thats very rare.

Qyburn

ABrass said:

Is this all perfectly south facing? Mine (see signature) works out very well thanks to the split. It tops out at 6.6kW but thats very rare.

Its 191° (11° W of South).  20° pitch.

charles_b_2

The difficulty is that there is no 'one size fits all'.  I'm looking at around 5.7KWh panels and 5Kw inverter with DC battery (basically Solaredge).  With my usage during the day, likelihood is that the battery will fill up very quickly and leave me in late Spring/ Summer with a lot of clipping if I went with the 3.6Kw inverter. Of the three DNO's, mine is SSE which all the installers groan about as G.99 can take 45 days, whereas UKPN and NG apparently can give you the say so in a couple of hours!  That made me consider 3.6 and just a G.98
But, it's a long term decision (min. 10 year inverter warranty) and I felt like I would kick myself if I didn't go with the higher rated inverter.  So I understand your predicament!

QrizB

Qyburn said:

ABrass said:

Is this all perfectly south facing? Mine (see signature) works out very well thanks to the split. It tops out at 6.6kW but thats very rare.

Its 191° (11° W of South).  20° pitch.

PVGIS now lets you download hourly historical data.

https://re.jrc.ec.europa.eu/pvg_tools/en/

For an array in the middle of England (roughly level with the Wash), with that orientation, and looking at June 2020, there were a few hours where the irradiance exceeded the standard value of 1000W/m2 but the peak output from 1kWp of panel was about 900 watts.

Taking the data and calculating how much you'll lose from an undersized inverter, for 1kWp of panel, and ignoring all losses (so assume 100% efficiency):

• 1kW of inverter: generate 1105.96kWh.
• 0.9kW of inverter: generate 1104.66kWh, lose 1.3kWh, 0.12% loss
• 0.8kW of inverter: generate 1093.85kWh, lose 12.11kWh, 1.09% loss
• 0.7kW of inverter: generate 1067.68kWh, lose 38.28kWh, 3.46% loss
• 0.6kW of inverter: generate 1021.08kWh, lose 84.88kWh, 7.67% loss
• 0.5kW of inverter: generate 952.65kWh, lose 153.31kWh, 13.86% loss
• 0.4kW of inverter: generate 856.01kWh, lose 249.95kWh, 22.61% loss
• 0.3kW of inverter: generate 727.64kWh, lose 378.32kWh, 34.21% loss

What does this mean? I think it shows that you can undersize by 30% (700 watts of inverter per kWp of panel) without worrying too much about what you're losing. Even down at 50% you're still only losing 1/8 of potential generation.

And I think it's remarkable how well output holds up even when the inverter is only rated at 30% of the panel capacity!

This won't be the same for everyone, but will be similar for many.

Turning the maths around, a 3.6kWp inverter on G98 approval shouldn't be a concern up to (3.6/0.7) ~5kWp of panel. Much beyond that and you're missing out on 5% or more of generation potential.

(Obviously you can try this for other years; PVGIS has data back to 2005. I chose 2020 because it was one of my best years for solar generation.)