My system's performance - Using PVGIS to get an accurate generation?

144 Posts

Hi all, this is really just a brain dump with a few questions, I may have just got particularly lucky with my install, it's performing spectacularly and I'm not sure if it's 'normal'. I would like other users opinions, especially those with a few years of data under their belt, and also wonder what others put into PVGIS to get an accurate prediction.
My system was installed in September 2014, by the end of this year I'll have 8 years of full data, although one of the years will need to be omitted, 2018, as we had a 'tin hat' covering the roof for the summer while we had our loft converted.
This is my total generation for this period. this year has been particularly amazing, currently at 4646kWh. One thing to note, is that this data I'm certain is pre-conversion to AC as it doesn't tally with my generation meter. If I compare, my generation meter is 97.8% of these numbers so fits in perfectly with the 97% efficiency stated in the specs for my inverter by SMA. So this year I've actually generated 4543.79kWh of useable leccy currently. like I said, from a 4kW system pretty amazing!

As you can see the prediction in Sunnyportal is way off, the details were inputted by the installer and I'm sure uses 3.68kW and my location and not much else.
My system in detail:
4kW Solarworld panels (16x250W All Black), 2 strings top and bottom.
3600tl Sunnyboy inverter
slightly southeast facing bungalow
no shading most of the year, minimal shading in winter (lamp post shadow goes across the bottom corner of the array), it's annoying but doesn't cause much loss in output.
We used to have an open field in front of us but that was built on in 2018 with a massive warehouse. so now the sun only hits us from about 8.30am onwards at this time of year.
in 2014 I used PVGIS to generate an estimate, I used the correct azimuth (-20°) and pitch (35°) for my system but left the 'system loss' at default, currently 14%, not sure what it was in 2014. this gave me a yearly average of 3892 kWh. Every year I've always exceeded that, which although a great position to be in, it gets tiring seeing year after year the prediction becoming less and less accurate.
If I set the system loss to 7.8% I get very close to my current average (from the meter not invereter). What do other users put in this box? How close are you all to your predictions? Has this last 7-8 years just been an anomaly and the predictions just don't stand?
Did I just really luck out and get a panel/inverter combination which really works? My inverter was in the loft for the first 4 years, but is now in a room so suffers no massive swings in ambient conditions anymore. It's also about 1.5m from the panels so there would be minimal loss from cabling while in DC.
Needless to say, my system paid for itself in FIT payments alone by the end of year 6 (would have been earlier if not for the loft conversion), so I'm more than happy.
I didn't have a lot of options at install, the only other options were Canadian Solar panels, slightly cheaper, I liked the idea of an all German system, and Solar Edge, which at £2k more wouldn't have made sense with my lack of shading. I did like the idea of panel level monitoring, I'm a scientist by profession so the more data the better but that still couldn't justify the cost difference.
That's a lot of writing just to ask how you set up your PVGIS prediction, but I thought some background would be nice to read.
Thanks all.
My system was installed in September 2014, by the end of this year I'll have 8 years of full data, although one of the years will need to be omitted, 2018, as we had a 'tin hat' covering the roof for the summer while we had our loft converted.
This is my total generation for this period. this year has been particularly amazing, currently at 4646kWh. One thing to note, is that this data I'm certain is pre-conversion to AC as it doesn't tally with my generation meter. If I compare, my generation meter is 97.8% of these numbers so fits in perfectly with the 97% efficiency stated in the specs for my inverter by SMA. So this year I've actually generated 4543.79kWh of useable leccy currently. like I said, from a 4kW system pretty amazing!

As you can see the prediction in Sunnyportal is way off, the details were inputted by the installer and I'm sure uses 3.68kW and my location and not much else.
My system in detail:
4kW Solarworld panels (16x250W All Black), 2 strings top and bottom.
3600tl Sunnyboy inverter
slightly southeast facing bungalow
no shading most of the year, minimal shading in winter (lamp post shadow goes across the bottom corner of the array), it's annoying but doesn't cause much loss in output.
We used to have an open field in front of us but that was built on in 2018 with a massive warehouse. so now the sun only hits us from about 8.30am onwards at this time of year.
in 2014 I used PVGIS to generate an estimate, I used the correct azimuth (-20°) and pitch (35°) for my system but left the 'system loss' at default, currently 14%, not sure what it was in 2014. this gave me a yearly average of 3892 kWh. Every year I've always exceeded that, which although a great position to be in, it gets tiring seeing year after year the prediction becoming less and less accurate.
If I set the system loss to 7.8% I get very close to my current average (from the meter not invereter). What do other users put in this box? How close are you all to your predictions? Has this last 7-8 years just been an anomaly and the predictions just don't stand?
Did I just really luck out and get a panel/inverter combination which really works? My inverter was in the loft for the first 4 years, but is now in a room so suffers no massive swings in ambient conditions anymore. It's also about 1.5m from the panels so there would be minimal loss from cabling while in DC.
Needless to say, my system paid for itself in FIT payments alone by the end of year 6 (would have been earlier if not for the loft conversion), so I'm more than happy.
I didn't have a lot of options at install, the only other options were Canadian Solar panels, slightly cheaper, I liked the idea of an all German system, and Solar Edge, which at £2k more wouldn't have made sense with my lack of shading. I did like the idea of panel level monitoring, I'm a scientist by profession so the more data the better but that still couldn't justify the cost difference.
That's a lot of writing just to ask how you set up your PVGIS prediction, but I thought some background would be nice to read.
Thanks all.
4 Kwp System, South Facing, 35 Degree Pitch, 16 x 250W Solarworld Panels, SMA Sunnyboy 3600 Inverter, Installed 02/09/14 in Sunny South Bedford - £5600
Growatt AC Coupled SPA3000tl and 6.5kWh battery Installed Apr 2022
Growatt AC Coupled SPA3000tl and 6.5kWh battery Installed Apr 2022
2
Latest MSE News and Guides
Replies
PVGIS calc at the time 3.4GWh pa, actual in range 3.51 - 3.89, average 3.77.
I think they may have tweaked the PVGIS calculator since I did my figures just after install?
Just checked, there seem to be more options than I remember when I did mine but my results correspond to Sarah2,14% (the defaults)
This year has been stellar & currently stands at 4.23.
Solar generation is a bit like the stock market: estimates are not a sound predictor of solar panel performance in any given year.
Yours is 4646kWh for 4kWp = 1161 kWh per kWp.
Both are very good outputs for the UK but I've also found the PVGIS results to be an underestimate of what can be generated. Maybe that's deliberate so they're not accused of overstating the benefits of solar.
But even my lowest year up to 2022 has been 103.4% of PVGIS.
2015 - 112.4%
2016 - 111.7%
2017 - 104.9%
2018 - N/A
2019 - 115.1%
2020 - 115%
2021 - 103.4%
There's off and there's off, I just wondered whether I lucked out with a super efficient system or PVGIS is just garbage.
Growatt AC Coupled SPA3000tl and 6.5kWh battery Installed Apr 2022
I suspect from some of the reports I have seen on the forum that those who installed their systems a few years ago might have calculated their PVGIS projection using the older version and haven’t re run the calculation on the later version. Any reference to performance against PVGIS without mentioning the version is meaningless.
System 1 - 14 x 250W SunModule SW + Enphase ME215 microinverters (July 2015)
System 2 - 20 x 330W Jinko Panels + Enphase IQ7+ microinverters (Jan 2022) + Givenergy AC Coupled inverter + 2 * 8.2KWh Battery (May 2022) + Mitsubishi 7.1 KW and 2* Daikin 2.5 KW A2A Heat Pump
[Edit - forgot to say, as I recall, PVGIS revised their figures when they were able to include local weather impacts, as well as average solar levels. In some areas, such as coastal for example, cooling can boost gen, whereas an equally sunny area nearby, may not get such a benefit. Also, the running 10yr average for solar levels has been rising.]
PVGIS has, as far back as I can recall (2012 maybe) used the default 14% figure for losses. This has always been seen as being a bit high, and more so as the last decade went on, and inverter efficiency rose a few percent.
My suggestion (on threads on here) has always been to leave the defaults alone (unless you really find it interesting), and that way you should get a buffer, and probably more generation than you expect ..... albeit a silly psychological trick.
When I had my install done, the installers were only allowed to use the UK SAP figure (later updated to different SAP figures across the UK). As this was based on research at Sheffield University, it meant a strange situation, where everyone south of Sheffield could expect better gen, and those north, worse. Installers weren't allowed to say (in my case) that gen would be better, but they could hint, and show charts of the UK with different irradiance levels*, to 'help' guide you. I think the south facing SAP figure for my house was about 850kWh/kWp, PVGIS is about 1,010. But I have a nagging thought that the SAP figure was supposed to be an average, and therefore had a built in degradation for the (original) 25yr FiT period.
* something like this
For general PV advice please see the PV FAQ thread on the Green & Ethical Board.
Looking at the sunshine for 2022 in S. Wales were I am, it was a great year v's the 1961-1990 figures, about 110-115%, but only 105-110% of the 1981-2010 average, but close to average v's the 1991-2020 average.
For general PV advice please see the PV FAQ thread on the Green & Ethical Board.