Effects of shading.

[Graeme Harrold]

Good news and thanks for taking the time to post a valuable update, as Im about to double the size of my home system to 8kW (before the FIT is slaughtered), and I too would be plagued with a nice big pole on the last couple of panels up to about 12am, aster this Im clear.

Total yeild so far is 1959kWhr since 23rd May on a south east roof.....

 

[Worcester]

@Series530

You've also got to consider "hard" shadows from dormers or other building edges that might knock out a few panels (not just the bypass diodes kicking in here) causing a significant drop in the string voltage, which could therefore drop below the inverter's cut in point, and where strings are in parrallel instead of into seperate MPP trackers (bad design you and I know) what effect that would have had.

Perhaps we should bear in mind that in the 'early' days of only a few years ago, panels didn't have bypass diodes and inverters were single string perhaps even without MPP technology.

I may be wrong, perhaps we are highlighting a problem that is now less true because of such developments.

Thoughts?

 

[langstroth2]

This may not be directly relevant, but I've also been monitoring the effect of a more diffuse shade caused by a silver birch tree on two right hand panels of one 8 panel string (of two strings). I knew the shading was there, but wanted to confirm that was why my new system was not performing as well as a nearby array (same orientation, v.similar size). The graph below shows the jump in output after 12pm when the 1 string that is in shade comes out into full sun (data collected on a clear sky day).



..and here's the results from that day comparing my array output with a nearby system:

	Langstroth	Other Array	Delta
	Wh	Wh	Wh
8pm to 12pm	2854	5217	2363
12 noon to 2pm	5575	5607	32
2pm to 4pm	4475	4634	159
4pm to 6pm	1759	1186	-573

As you can see, my array when not shaded is producing the same output as it's unshaded near neighbour.
The morning shading though, even though "slight", results in a near halving of output.

(The tree is due to come down btw - it's also shading too much of our garden)
(For info: system size: 4Kw. Two strings of 8 on separate MPP trackers.)

 

[series530]

This may not be directly relevant, but I've also been monitoring the effect of a more diffuse shade caused by a silver birch tree on two right hand panels of one 8 panel string (of two strings). I knew the shading was there, but wanted to confirm that was why my new system was not performing as well as a nearby array (same orientation, v.similar size). The graph below shows the jump in output after 12pm when the 1 string that is in shade comes out into full sun (data collected on a clear sky day).

View attachment 8313

..and here's the results from that day comparing my array output with a nearby system:

	Langstroth	Other Array	Delta
	Wh	Wh	Wh
8pm to 12pm	2854	5217	2363
12 noon to 2pm	5575	5607	32
2pm to 4pm	4475	4634	159
4pm to 6pm	1759	1186	-573

As you can see, my array when not shaded is producing the same output as it's unshaded near neighbour.
The morning shading though, even though "slight", results in a near halving of output.

(The tree is due to come down btw - it's also shading too much of our garden)
(For info: system size: 4Kw. Two strings of 8 on separate MPP trackers.)

To my thinking, this and what Worcester posted earlier is actually very relevant. What it says to me is that a hard shadow will knock out a set of bypass diodes, a panel or several panels and, if several panels suffer from hard shading, it may also take out a string which would lead to a partial or a complete inverter shut down.

Soft shading (due to diffused objects such as trees or structures which are some distance away) may not be enough to trigger the bypass diodes but will be enough to reduce the output of the string until a point where the bypass diodes trigger (about 10.50am).

The panel output doesn't increase then until about 11.30am when, I have to assume, the sun has moved enough to bypass the shading. At this point, we see almost a doubling of power output as the shaded array comes back on line. I don't believe that the string shut down though. Based upon 250W per panel, it seems like about 2 panels came on line at about 11.20 and the rest came back at about 11.30 (smoothed out on the graph).

Had the shading not taken place I suspect the gradual increase in harvest would have continued in a linear fashion until reaching the systems capability and we would have seen a smooth and gradual curve similar to what we saw before 10.50.

Something is also occurring between 10.50 and 11.30: the variation in power is pretty wild. I suspect that this is panels or sections of panels bypassing (or otherwise) as the shade takes more and more of a hold on the array. Smoothed out as it is, you can see jumps in power level as something like 4 panels are switching in and out and then the whole of the rest of the array suddenly comes on stream at about 11.30.

When my micro inverters finally arrive (something of a moving target) I will try to produce something similar. If nothing else, I should be able to track what is happening panel by panel as shade takes effect.

Thank for sharing this with us.

Highly interesting for me.... being an engineer, I like to understand what is happening.

 

[Scooby]

Somewhat delayed further feedback from me on this - sorry. Due to pretty cloudy conditions recently with no shadows being cast...

Took these two readings yesterday, 22nd Oct. Bright sunshine, well-defined shadows, but not completely 'blue' skies.



Left hand photo shows shading at 3pm, right hand at 4.10pm.
To recap, the array is split left & right, 8 panels each side.

At 3pm, LH array: 335W RH array: 1201W

At 4.10pm, LH array: 404W RH array: 510W

This info is for others to digest, and I have found the responses genuinely interesting.

The bottom line for me, of course, is that despite the lowish Autumn sun, the shadows in the afternoon, and the overwhelmingly cloudy & overcast weather since the install, it has still produced 192kWh in 25 days! So even under these 'worst-case' conditions (I'll have no shadow at all over the highest producing Summer months) , it is still on target! That, I find, is pretty darn impressive... :angel_smile:

 

[SolarCity]

Thanks for the info, Scooby.

These results are consistent with the tests we have done. I was surprised that the pole appeared to be having less of an impact on your previous posts but clearly the shadow cast by the pole is larger in these photos. The drop in production from 1201w to 510w as the shadow moves along it is dramatic and demonstrates the importance of getting the design right in the first place.

A quick question for the installers on this forum. What factor would you apply in your SAP calculations for this pole?

It seems that the system is actually slightly over performing. I calculate a yeild of over 195kWh over the whole of October in my software. It's been a pretty good month for solar production!

 

[Worcester]

At first glance, (before the full story of this thread) I would probably have taken "Modest" as is clearly an obstacle, and therefore taken the factor of 0.8 (20%) drop.

For me to choose "None or very little" the pole would have to have been the other side of the road / track.

 

[Scooby]

'Over-performing', eh? I'll have no over-performing in my household. I'm out to unplug a panel...

 

[series530]

Those two pictures that you took Scooby are really interesting. For me it's another good illustration of the effects of shading. I'll assume for a moment that the power produced by the sun is constant at both times. I know it isn't, but for the sake of the analysis, I'll assume that it is. Your reference is the right hand array in the first picture which is producing 1.2KW. The left hand array, at that time has one panel where the bypass diodes have probably isolated the output and three more that are marginal. Worst case, we might expect 4 of the 8 panels turned off. The power produced on the left hand panel, at just 335W is about 28% of the output of the right hand panel.

Move forward to the later picture now. In this one the effect is almost symmetrical on both panels. Theoretically, with two panels of out 8 affected on each array, the power produced should be 6/8 * 1201 KW = 900W. However, what is produced is roughly half of that.

So, what is happening and why is it different to your pictures from some time back? My theory is this: these two pictures, at a time when the light levels are lower anyway, show situations where the bypass diodes are not triggering. Taking the RH array in the first picture as the reference, in all of the other situations the effect of shading is constraining the harvest to the performance of the worst panel because the difference in harvest between the light and shade is less than 20% of the panel capability (using the Solaredge numbers). As the diodes are not triggering, the whole array is being constrained to the performance capability of the worst panel.

So, where the irradiance is higher, the difference in power between best and worst sections exceeds the 20% value and the diodes trigger. So, a nice sunny day in the middle of summer would do the trick. However, early and late in the year or early or late in the day at other times, the diodes may not trigger and the power harvest will be reduced. Luckily though, this is at a time when the harvest is low anyway - so the loss is less of an issue.

Just a hypothesis... does this make sense or are there different factors at play here?

 

[Scooby]

Hi Series. What you say makes sense to me, tho' I don't have the knowledge to confirm your hypothesis! Certainly, the LH array shows a much bigger reduction in output compared to my earlier - first - post on this.

I have to say I was pretty amazed at how little the original (last week's) effect of the shading was; it was slightly earlier - at 2.30 - and the shadow was affecting only 2 panels, one fully, and the other by less than a half, but still... The example above is half an hour later; as the picture shows, a total of 4 panels are now affected to some degree. However, the overall light levels between this one and last week's must have been pretty close in both cases as the RH array had outputs of 1201W and 1170W respectively.

I would suggest that comparing the 4.10 example with the others has to be treated with some caution as the light angle has changed a fair bit, but how much effect this has I don't know.