Planning a Solar PV and Battery system

[Rob99]

After a bit of a false start on this last year I’m looking again at a PV system for our house.

 

We have East and West roof elevations suitable for panels and I intend to max out the kWp of the array as I just want a “one hit” roof install without revisiting later to add any more.

 

First step I’ve been looking at the PVGIS database and modelling output on East and West. For initial purposes I’m using 5.39kWp (14x385W panels) on the East and 3.08kWp (8x385W panels) on the West.

 

I’d always thought East and West were poor cousins to South but PVGIS shows East as generating 4,647kWh (86% of array max) per year and West as 2,297kWh (75% of array max) per year.

This seems high to me but do people find PVGIS outputs to be a good match to real world outputs of installed systems?

I’m going to start modelling generation against usage so I can size the invertor and battery correctly but don’t want to just blindly use the PVGIS data if it’s over-estimating.

 

I've attached the PVGIS results for both.

West-PVGIS.pdf East-PVGIS.pdf

[Conor]

Just one year of data for me, but PV GIS estimate of 4.5MWh, is pretty close to our actual output of 4.26MWh. not looked at it on a monthly basis.

[Rob99]

That's good to know. Do you have any shading at all on your array which might account for the slight reduction?

 

Oh, and what size (kWp) is your PV system?

Edited April 22, 2023 by Rob99

[Dillsue]

The MCS forecast for our original 4kw system was lower than any real annual generation since install in 2015. I assume MCS used the same data as PVGIS but if there were meaningful differences between PVGIS and MCS it would well known.

 

With a split array you will always have shading, with the east facing array in the sun in the morning and the West not in direct sun, assuming a reasonably pitch roof. The opposite will be true in the afternoon. Both arrays will generate when they are not in direct sun. To get round that problem you can wire each array to its own MPPT controller or use optimisers/micro inverters.

 

To work out what total max output will be for any given time you can use the PVGIS hourly forecast figures and crunch them in excel. Find your local highest annual yield from the last 10 years then download the hourly figures for that year for each array, separately. Put the date/time/kw output for each data set next to each other then sum the kw outputs for every time period. Then sort to find the highest summed output. Use that max figure to size your inverter  bearing in mind that some inverters can take significantly more DC input than their rated value- a Solaredge 3.68kw inverter can have up to 5.6kw of DC input from panels.

[Conor]

3 hours ago, Rob99 said:

That's good to know. Do you have any shading at all on your array which might account for the slight reduction?

 

Oh, and what size (kWp) is your PV system?

 

No, that kind of variance is purely weather /climate related. It's a 50% south, 25% west, 25% east array, 5.1pkW gross, 4.5pkW output at the inverter. Limited to 3.8kW.

[Rob99]

15 hours ago, Dillsue said:

The MCS forecast for our original 4kw system was lower than any real annual generation since install in 2015. I assume MCS used the same data as PVGIS but if there were meaningful differences between PVGIS and MCS it would well known.

That's quite interesting. I wonder if the MCS data makes some assumptions which isn't part of the PVGIS data. Also wonder whether there's a commercial element in this too whereby MCS don't want installers or themselves to be criticised for PV generation estimates which don't materialise and therefore use much more conservative figures.

 

15 hours ago, Dillsue said:

With a split array you will always have shading, with the east facing array in the sun in the morning and the West not in direct sun, assuming a reasonably pitch roof. The opposite will be true in the afternoon. Both arrays will generate when they are not in direct sun. To get round that problem you can wire each array to its own MPPT controller or use optimisers/micro inverters.

Yes, that's true with shading although I have been quite surprised by how much sun actually falls on each roof slope as the sun moves around. The pitch is 35deg so not huge (its why we never ended up doing a loft conversion 😞) Each array will need to be on a seperate string so will have to make sure I source an inverter with at least 2 MPPT controllers. We also have a bit of shading from trees which shades between 15-20% of the East array in the first couple of hours of sun and around 10% of the West array is shaded very late in the afternoon for no more than an hour. Having looked into optimisers I don't think they wil be worth having as panels and invertors these days seem to be designed to manage shading really well.

 

15 hours ago, Dillsue said:

To work out what total max output will be for any given time you can use the PVGIS hourly forecast figures and crunch them in excel. Find your local highest annual yield from the last 10 years then download the hourly figures for that year for each array, separately. Put the date/time/kw output for each data set next to each other then sum the kw outputs for every time period. Then sort to find the highest summed output. Use that max figure to size your inverter  bearing in mind that some inverters can take significantly more DC input than their rated value- a Solaredge 3.68kw inverter can have up to 5.6kw of DC input from panels.

I've downloaded the hourly data from PVGIS for the whole 16yr period and dumped it into excel (all 280,000 rows 😱) and will now spend some time pivotting and analysing the data.

 

13 hours ago, Conor said:

No, that kind of variance is purely weather /climate related. It's a 50% south, 25% west, 25% east array, 5.1pkW gross, 4.5pkW output at the inverter. Limited to 3.8kW.

Your generation of 4.26MWh seems pretty good if your inverter limits output from your PV arrays to 3.8kW

[Rob99]

Quick question on PVGIS data.....

 

Can I assume that the hourly data output uses UTC and not local time?

 

The PVGIS user manual doesn't seem to say one way or the other.

[S2D2]

34 minutes ago, Rob99 said:

Quick question on PVGIS data.....

 

Can I assume that the hourly data output uses UTC and not local time?

 

The PVGIS user manual doesn't seem to say one way or the other.

From memory it states UTC in the downloaded spreadsheet somewhere, that's what I used it as when predicting system performance.

[Carrerahill]

On 22/04/2023 at 16:47, Rob99 said:

After a bit of a false start on this last year I’m looking again at a PV system for our house.

 

We have East and West roof elevations suitable for panels and I intend to max out the kWp of the array as I just want a “one hit” roof install without revisiting later to add any more.

 

First step I’ve been looking at the PVGIS database and modelling output on East and West. For initial purposes I’m using 5.39kWp (14x385W panels) on the East and 3.08kWp (8x385W panels) on the West.

 

I’d always thought East and West were poor cousins to South but PVGIS shows East as generating 4,647kWh (86% of array max) per year and West as 2,297kWh (75% of array max) per year.

This seems high to me but do people find PVGIS outputs to be a good match to real world outputs of installed systems?

I’m going to start modelling generation against usage so I can size the invertor and battery correctly but don’t want to just blindly use the PVGIS data if it’s over-estimating.

 

I've attached the PVGIS results for both.

West-PVGIS.pdf 223.26 kB · 1 download East-PVGIS.pdf 227.62 kB · 0 downloads

Why are you using such low power density panels? Mid 400's are common and 500-520W are common now too. 

 

2 ways to go here, about same no. of panels (they are a bit bigger), much greater generation, or less panels and same level of generation. 

Edited April 25, 2023 by Carrerahill

[Rob99]

4 minutes ago, Carrerahill said:

Why are you using such low power density panels? Mid 400's are common and 500W are common enough too.

 

2 ways to go here, same no. of panels, much greater generation, or less panels and same level of generation. 

Initially I just needed something to use as an estimated array size in PVGIS and those panels are what I was quoted for by an installer previously.

 

Having looked at lots of panels in the past few days I know there are lots of higher power ones now but they are also different sizes. Some of the larger output panels won't fit on my roof in 2 rows so wouldn't be suitable.

 

I've also noticed that even the higher power panels generally seem to produce the same power per m2, usually around 210-220 W/m2 so it'll be all about matching panel size to the roof to maximise the kWp output. I just wish you could get triangular panels to fill in the corners too....!!!

 

I'll be doing a lot more research when I look at panels as I also want something completely black if possible. The Perlight Delta panels look very nice. I think they are also the new "shingle" type ones too.

[Rob99]

24 minutes ago, S2D2 said:

From memory it states UTC in the downloaded spreadsheet somewhere, that's what I used it as when predicting system performance.

Couldn't find any reference to UTC on my downloads but when I look at the historic data for today (Apr25) the PVGIS data shows no solar generation after 7pm whereas sunset is at 8pm (BST) so it must be using UTC.

[Carrerahill]

1 hour ago, Rob99 said:

Initially I just needed something to use as an estimated array size in PVGIS and those panels are what I was quoted for by an installer previously.

 

Having looked at lots of panels in the past few days I know there are lots of higher power ones now but they are also different sizes. Some of the larger output panels won't fit on my roof in 2 rows so wouldn't be suitable.

 

I've also noticed that even the higher power panels generally seem to produce the same power per m2, usually around 210-220 W/m2 so it'll be all about matching panel size to the roof to maximise the kWp output. I just wish you could get triangular panels to fill in the corners too....!!!

 

I'll be doing a lot more research when I look at panels as I also want something completely black if possible. The Perlight Delta panels look very nice. I think they are also the new "shingle" type ones too.

Aim for about the 430W mark, they are the same size as the old 350-375W panels. 

[S2D2]

1 hour ago, Rob99 said:

Couldn't find any reference to UTC on my downloads but when I look at the historic data for today (Apr25) the PVGIS data shows no solar generation after 7pm whereas sunset is at 8pm (BST) so it must be using UTC.

Yes not in the spreadsheet sorry, it's here: https://joint-research-centre.ec.europa.eu/pvgis-online-tool/pvgis-tools/hourly-radiation_en

 

Quote

time [UTC]

[SteamyTea]

When analysing PV output, UTC, which is just GMT, makes the most sense.

Not as if the altitude and azimuth of the sun will suddenly jump to match the stupid clock change.

Noon should be Solar Noon, no matter where you are.

[S2D2]

2 minutes ago, SteamyTea said:

When analysing PV output, UTC, which is just GMT, makes the most sense.

Not as if the altitude and azimuth of the sun will suddenly jump to match the stupid clock change.

Noon should be Solar Noon, no matter where you are.

Interestingly it does need factoring in when modelling the Flux tariff as the cheap/peak periods do follow the clock changes. A good chunk of my export has moved into the peak period since they went forwards.

[Rob99]

1 hour ago, S2D2 said:

Interestingly it does need factoring in when modelling the Flux tariff as the cheap/peak periods do follow the clock changes. A good chunk of my export has moved into the peak period since they went forwards.

Yes, that's also true as I've been measuring hourly consumption over the past month when it's all been on BST. I now have a pretty accurate model of our hourly, daily and monthly usage.

 

To ensure accuracy with matching usage to generation and timed tariffs like flux I have shifted all the PVGIS generation figures forward by an hour between April and October so that they match BST during that period. It's actually made a difference to the modelling, especially with Flux. Not a huge difference but my OCD wouldn't allow me not to change it 🤣

Edited April 25, 2023 by Rob99

[S2D2]

41 minutes ago, Rob99 said:

Yes, that's also true as I've been measuring hourly consumption over the past month when it's all been on BST. I now have a pretty accurate model of our hourly, daily and monthly usage.

 

To ensure accuracy with matching usage to generation and timed tariffs like flux I have shifted all the PVGIS generation figures forward by an hour between April and October so that they match BST during that period. It's actually made a difference to the modelling, especially with Flux. Not a huge difference but my OCD wouldn't allow me not to change it 🤣

Might as well be accurate! I've mentioned it before but if you have a smart meter you can get 13 months worth of historic data via Bright.

[Nickfromwales]

On 23/04/2023 at 10:39, Rob99 said:

Also wonder whether there's a commercial element in this too whereby MCS don't want installers or themselves to be criticised for PV generation estimates which don't materialise and therefore use much more conservative figures.

Yes to that. Every single PV install my chap has done ( for my private clients and a bunch of members on here ) have all met or exceeded the stated expectations, so there is "definitely something in the MCS water" .

 

6 hours ago, Rob99 said:

I just wish you could get triangular panels to fill in the corners too....!!!

You can, but you'll spend 20 years paying them off. There's a company that takes full panels and cuts them down / reinstates the circuitry / re-frames them to suit.

The price is £OUCH

[Rob99]

5 hours ago, S2D2 said:

Might as well be accurate! I've mentioned it before but if you have a smart meter you can get 13 months worth of historic data via Bright.

That would've been great but I've been one of those cynical types suspicious of the motives of energy suppliers so always refused a smart meter until it would benefit me.

 

Seems like now is that time 👍

 

Just on that - should I get one fitted by my current supplier now, or would it make more sense to have it done after I sign up to Octopus? Is there a lead in time for meter changes?

 

Also, as my meter installation has never been touched since the house was built in 1995 it doesn't have a DNO cutout and I'm guessing I'll have to have one fitted.

 

Will the meter people sort that or will they turn up, sharp intake of breathe then "sorry mate can't fit your meter as no DNO cutout" and then scarper?

 

 

[Nickfromwales]

7 hours ago, Rob99 said:

 

Also, as my meter installation has never been touched since the house was built in 1995 it doesn't have a DNO cutout and I'm guessing I'll have to have one fitted.

 

Will the meter people sort that or will they turn up, sharp intake of breathe then "sorry mate can't fit your meter as no DNO cutout" and then scarper?

Everything before the meter is DNO responsibility, so you won't ask anything of the fitters other than "would you like a cup of tea?" Do not start off by saying "please fault my sub-standard home!".  Just leave them to it.

[ProDave]

8 hours ago, Rob99 said:

Also, as my meter installation has never been touched since the house was built in 1995 it doesn't have a DNO cutout and I'm guessing I'll have to have one fitted.

Go on, post a picture, I find it hard to believe a 1995 install does not have a DNO cutout.  Lets at least see the bodge someone has done.  If it was you, don't admit to doing it.

[Rob99]

Here you go.

Pretty much as it was when we bought the house in 1995, except for the upgraded meter tails the electrician did a few years ago when we extended and had a new consumer unit installed (hence the missing seals!)

[PhilT]

1 hour ago, Rob99 said:

Here you go.

Pretty much as it was when we bought the house in 1995, except for the upgraded meter tails the electrician did a few years ago when we extended and had a new consumer unit installed (hence the missing seals!)

I hope that black thing is a DNO cutout otherwise I don't have one either!

[Rob99]

11 minutes ago, PhilT said:

I hope that black thing is a DNO cutout otherwise I don't have one either!

Yes, that's technically a DNO cutout, it's the incoming main with a 100A supply fuse.

 

I was referring to the type of DNO cutout which is a double pole isolator installed after the meter so that fixed wiring/consumer units etc can be isolated without touching the meter/supply fuse (which technically you're not supposed to as it's not yours.

 

[Nickfromwales]

After the meter it's yours.

Before the meter it's the DNO's.

Simple.