Do all sources of shading affect panel performance the same way?

[solar_trees]

Is there a difference in shade that comes from the clouds vs shade that comes from a building (or tree) in how it affects the performance of solar panels?

 

[MurphyGuy]

Between clouds and a tree, yes.. Between a tree and a truck, not so much.

Clouds tend to let light bleed through them.. the darker the cloud, the less light bleeds through = less power at your panel.

Partial shading of a panel, if it's dark enough, can drag down the entire panel.

 

[Substrate]

Will add that as the clouds move, and the sun starts to peek through their edges, a magnifying property known as "edge of cloud effect" can occur, temporarily increasing solar panel output, sometimes surprisingly.

 

[Checkthisout]

Is there a difference in shade that comes from the clouds vs shade that comes from a building (or tree) in how it affects the performance of solar panels?

Expand your language.

Shadowing = trees/chimneys or whatever casting shadows on the panels.

Shading = reduced sunlight on the panels due to cloud cover.

If you have shadowing issues, you'll probably see higher output on overcast days than you will with the panels being shadowed on a clear, sunny day.

 

[Bellyman]

This thread is about as close as I see to what I am wondering about myself so I didn't start a new one. I'm hoping one or two of you guys have already been down this road.

I've seen Will's video comparing several solar panels and in order to "shade" them for testing purposes, I think he placed a piece of paper over the corner of each of them, or something like that. But I do not know if that gives any meaningful representation of a "cloudy" or "overcast" day.

Would the characteristics of different panels (say poly vs mono) be different when dealing with diffused light like we'd see with an overcast day vs a truck that got parked in front of a set of panels that effectively blocked 40% of their direct sunlight?

Or is there really any difference? I'm still on the fence about whether mono panels are really worth it to catch those rays on overcast days or whether poly panels of the same wattage will do the job just as well. Or maybe I'm asking the wrong questions?

 

[RoadTurtle]

I have an antenna on a painter pole above my 5th wheel. This puts a shadow across one panel until about noon. That panel gathers ZERO sunlight until the shadow clears the panel. One teeny tin y corner was shadowed by aircon. That little shadow took out 50% of that panel.

I went with 24v panels on a 12v system. It's been suggested it will collect more energy on cloudy days than does 12v panel.

 

[wattmatters]

I've seen Will's video comparing several solar panels and in order to "shade" them for testing purposes, I think he placed a piece of paper over the corner of each of them, or something like that. But I do not know if that gives any meaningful representation of a "cloudy" or "overcast" day.

No, it would not.

Partial hard shading has different effects on different panels and arrays depending on the type of panel, the nature of the array configuration, the type of inverter(s)/charge controller(s) being used, the MPPT algorithms, optimisers, etc etc. Not to mention exactly how the panel or the array is being shaded. Which part of the panel or array is shaded?

There is just no single answer to the question, because it is insufficiently defined.

You have to be very specific about all of these factors before anyone can hazard a reasonable guess as to what the impact of any hard shading would be.

Change an array from regular to half cut panels and the impact may well be quite different.
Change to an inverter with a different MPPT algorithm and the impact may well be different.
Change between string inverter and micro-inverters and the impact may well be different (and not always in ways people have been led to believe).
etc etc

As to impact of low insolation due to clouds, just go look at any one of hundreds of thousands of systems online at PVOutput.org and you can compare production on sunny v cloudy days.

Here's mine for example:

wattmatters 11.000kW

PVOutput.org - share, compare and monitor live solar photovoltaic output data

pvoutput.org

My worst output day was 1.5 kWh, heavy cloud cover and rain all day, dark and dismal even though it was a Summer day.
My best was >70 kWh, a sunny Summer day. And I've had pretty much everything in between.

 

[Bellyman]

look at any one of hundreds of thousands of systems online at PVOutput.org and you can compare production on sunny v cloudy days.

Thank you for sharing. Lots of data there. I hadn't seen that site before but I'm pretty new here.

 

[zanydroid]

poly vs mono

This one is probably easy, because the difference is localized to a single cell. Above the cell the architecture of a panel (# of bypass diodes, half cut or not) it's the same.

Anyway, I'd like to tag onto wattmatter's post above.

So the problem here is that there are a lot of locally easy to understand phenomena, but when you combine them together it's very hard to solve analytically. You would probably need simulation, but then you won't have the necessary inputs for the simulation.

IMO, it's a fool's errand to try to analyze this manually. Since you will have varying shading conditions throughout the whole year. You can kind of hand-solve a few, but you'll never be able to cover all cases. Arguably it follows that understanding the electrical engineering of shading really doesn't help you that much (since you're probably not writing a simulation software), other than improving your general intuition about PV.

• PVwatts etc will not model your hard vs soft shadow situation. How can it? It has no LIDAR map of your site.
• OK, you can use a LIDAR tool to predict PV output. Better than PVwatts. I've had it done on my house. But nobody came to actually scan the environment, they just extrapolated the obstruction geometry from aerial photos. There's not even an assessment of the "error bars" around using that approach. Garbage in Garbage out.

Which means eventually everyone uses rules of thumb, like what the payoff is for each adjustment to the architecture

EG, Non-controversial

• half cut good
• MPPT per plane
• don't block the sun in any way if you want a simple to analyze situation. there are people out there that say, take what you get with a shade situation

Controversial

• while MLPE like optimizers and microinverters are controversial on how much gain you will actually get.

(But micros and optimizers give you other advantages like removing errors in design or installation phase from messing you up).

Apart from the poly vs mono:

• Diffuse: at some point it will soften hard shadows to the point where the solar panel pops out of bypass. In my heavily hard-shadowed install, if a day is overcast over a critical threshold, I get 50% more production! This is certainly not going to show up in PVwatts. It can definitely show up in LIDAR modeling if it is included in the programming.
• Now, combining together all of the individual factors that change per-panel and per-cell performance. With shading this will create a complex looking Power-Voltage graph that the MPPT will scan across or otherwise try to optimize against. Hopefully it finds the global power maximum and doesn't get confused. The longer the string, the more complicated this can be due to combinatorics. For a single panel (EG SolarEdge / microinverter case), the set of possible shapes is more bounded, so they've probably tested it fairly extensively.

 

[wattmatters]

OK, you can use a LIDAR tool to predict PV output. Better than PVwatts.

There are also tools such as:

Login | Pylon Observer

Fast and accurate solar designs with high resolution aerial imagery

app.getpylon.com

which enable you to simulate a PV system and generate a shading profile to see the expected impact on your production throughout the year.

Here's a couple of quick videos showing an example:

Location matters.

You see a lot of guides that you can access online that give you an average amount of productive sunlight hours for solar power generation for your geographic location and time of year. They are handy but they do not tell the whole story. This came to me as I was enjoying my morning coffee out...

diysolarforum.com

 

[zanydroid]

Hmm I might try that out. I did some manual measurements from a drone of the angle to my obstructions, and some manual sun angle checks (based on ephemeris data)

That made me really doubt the accuracy of Aurora LIDAR simulation that a salesperson whipped up for me overnight. Since the winter generation numbers don’t jive with my manual checks.

 

[wattmatters]

Don't forget, these simulations also assume unlimited load, e.g. grid-tied without export limitation.

 

[zanydroid]

Took a look at your videos. Very informative.

I do see a common issue here with the Aurora I got... it assumes that the trees are round, but mine are extremely jagged and asymmetrical. I'm sure you can override it to a different species I guess.

 

[wattmatters]

I do see a common issue here with the Aurora I got... it assumes that the trees are round, but mine are extremely jagged and asymmetrical. I'm sure you can override it to a different species I guess.

You can select different basic tree shapes but expecting a simulation to get every detail correct (e.g. deciduous vs perennial) is expecting a bit much I think.

Such models are designed to help understand the general impacts and to sharpen the questions, not to be perfect emulators.

For example, here is four years or so of actual monthly production compared with the estimate built using pylon observer:



In general the estimate is a little generous, particularly in Autumn, but when I consider:

i. my trees grow and I get them lopped, so the shade profile is a continuously variable feast. When that happens (usually late Autumn) and how hard they get lopped affects the production

ii. the last two years have been well over annual rainfall averages (double) and so the weather and insolation patterns have not been typical. Prior to that we had a drought year.

Even so, the estimates (provided they are set up correctly) are pretty decent indicator of what one should expect.

 

[zanydroid]

Yeah, that makes sense. Though I think deciduous vs perennial should be an easier thing to deal with than others.

In my case, there were two takeaways that I applied to my projects.

1) I used a PVwatts only estimate (created by the installer, not me, so it's not my fault? it's my first solar install) for my existing turn-key installation, vs the Aurora estimate from a different proposal. Boy, that was a big mistake.

2) After checking the tree heights / foliage, etc with my drone, micromanaging like crazy, I decided I shouldn't take the Aurora estimate as gospel when planning out the followup system & the expected ROI for panels in different parts of the roof. We'll see how the numbers pan out next year after I get it installed and have real world info.

I think I have a better intuition for where the uncertainty is in the predictions. The ROI predictions for panels that have less impact from shading simulation should have much lower error bars than the ones that rely heavily on shading simulation.

 

[solar_trees]

MPPT per plane

What does plane mean here?

 

[zanydroid]

Roof plane