Effect of shading on solar panels in series

[mperkel]

I'm trying to understand the readings I'm getting off my solar panels to try to determine if something isn't quite right. Hoping to find someone to answer this who has a deep understanding of solar panels and how series connections, shading, and bypass diodes work.

I have a SolarCity install that I'm monitoring and I feel like it's not producing what it should during partial shade. So I'm going to explain this in detail and hope someone can explain what's happening.

I have 14 panels arranged in 2 strings in series of 7 panels. The AAB inverter has inputs for 2 separate strings. In full sun both strings produce almost exactly the same voltage and current. Panels are made by Canadian solar, 255 W I think.

Neighbor has some big palm trees that start to shade the array mid afternoon. One string is in full sun. The other is partially shaded. The shadow covers one panel completely and 2 other panels 1/2 diagonally. So the shaded string has 4 panels in full sun and 3 panels which are probably fully dead.

The sunny string has these readings: 193 V - 5.2 A - 1003.6 W
The shaded string has these readings: 111 V - 3.2 A - 355.2 W

So - here's my question.

The voltage drop is about right on. 111 / 193 = 4 / 7 - just what you would expect with 4 panels working and 3 dead.

but ...

Why is the current down from 5.2 A to 3.2 A ?
Shouldn't I be getting 577.2 W? Why am I missing 222 W?

If I had just 4 panels in the string in full sun I should get 111 V - 5.2 A - 577.2 W

The bypass diodes in the system should bypass the shaded panels allowing the full current of the sunny panels.

So - what am I missing here? There's something I don't understand.

[inetdog]

I do not think that there is anything that you do not understand, but rather that some of the hardware is not behaving as expected.
If the bypass diodes on all of the shaded panels are working correctly, then you should see the Imp of the shaded string being essentially the same as the Imp of the unshaded string. Just the Vmp values should differ. And the Voc values should be very close since even the shaded panels will contribute voltage at zero or low current.

Can you measure the voltage from + to - on each panel? That is what it would take to diagnose possible problems with bypass diodes.
For a panel with three bypass sections (three diodes), you should see a voltage of negative (opposite polarity from normal) 1V to 3V depending on how good the bypass diodes are. If you see a substantially higher negative voltage, then one or more of the bypass diodes is open circuited and you are forcing current through one of the shaded sections. That could reduce the Imp in the way you are seeing.

You can also experiment by covering one panel at a time while the string is in full sun. That may be useful if you cannot easily measure the panel voltages.
If only one panel changes the string Imp, then that panel has bad bypass diode(s).

[mperkel]

I don't have access to the panels individually so I can't test that.

What confuses me is that the voltages are right on for what I would expect. If diodes were bad I would expect a little higher voltage (due to getting some light) and far less current (due to lack of bypass diodes).

What I'm seeing isn't what I expect if diodes were bad.

I'm wondering if the inverter isn't working in an optimal fashion.

[mperkel]

I'm beginning to wonder about the software in the inverter that does MPPT.

[inetdog]

I don't have access to the panels individually so I can't test that.

What confuses me is that the voltages are right on for what I would expect. If diodes were bad I would expect a little higher voltage (due to getting some light) and far less current (due to lack of bypass diodes).

What I'm seeing isn't what I expect if diodes were bad.

I'm wondering if the inverter isn't working in an optimal fashion.

You can still do the test of covering individual panels with a piece of cardboard and looking at the inverter readout.
It is possible that the string voltage has dropped to the minimum MPPT tracking voltage and that if the inverter pulled more current the voltage would drop another 10% or so. Check the input specs for the inverter.

[mperkel]

I think I figured it out. I think there's a design flaw in the ABB inverter. It has a minimum voltage VSTART setting of 120V. Why it requires this is a mystery but with 3 panels shaded and 4 in sun, at 27.5 volts / panel that's 110 volts. So it's in under voltage degraded mode.

I send an email to SolarCity about it. They can probably fix this in software. Bad design.

[sensij]

I think I figured it out. I think there's a design flaw in the ABB inverter. It has a minimum voltage VSTART setting of 120V. Why it requires this is a mystery but with 3 panels shaded and 4 in sun, at 27.5 volts / panel that's 110 volts. So it's in under voltage degraded mode.

I send an email to SolarCity about it. They can probably fix this in software. Bad design.

120 V is actually lower than some, with a different inverter, you could have it much worse. 7 panel strings was kind of a poor design decision, and may have been done to prevent the bad situation with shade from being worse. A SolarEdge inverter with panel optimizers would have been a much better choice, and SolarCity uses them a lot. Perhaps, for an upgrade fee, they would swap out your existing inverter for that system. You could also look into rewiring them into a single string with Tigo optimizers... that should also improve your output.

Edit: Keep in mind the sun is low in the sky right now, we just passed the winter solstice. Shadows that are bad right now may not effect you as much in the summer, when your system is generating much more energy overall. The loss may seem painful, but might not be hurting your annual production too badly.

[inetdog]

I think I figured it out. I think there's a design flaw in the ABB inverter. It has a minimum voltage VSTART setting of 120V. Why it requires this is a mystery but with 3 panels shaded and 4 in sun, at 27.5 volts / panel that's 110 volts. So it's in under voltage degraded mode.

I send an email to SolarCity about it. They can probably fix this in software. Bad design.

This is definitely not a design flaw in the AAB inverter. It is a design flaw in the system that connected a heavily shaded 7 panel string to the ABB inverter in the first place.
I very much doubt that the behavior can be changed in software, since it involves hardware optimization.

You probably would have been better off splitting the shaded panels between the two strings (or going for a single string) since the losses in the bypass diodes are not outrageous.

[DanS26]

I think I figured it out. I think there's a design flaw in the ABB inverter. It has a minimum voltage VSTART setting of 120V. Why it requires this is a mystery but with 3 panels shaded and 4 in sun, at 27.5 volts / panel that's 110 volts. So it's in under voltage degraded mode.

I send an email to SolarCity about it. They can probably fix this in software. Bad design.

Try switching the strings in the inverter and see if you get different results.

[mperkel]

In order to take advantage of getting the most amount of power from a partially shaded environment the ideal inverter would be capable of dealing with a very large range of input voltages. An inverter that could work from say 25 to 400 volts would be ideal.

Is there any inverters that have that kind of range? If not - why not?

[sensij]

In order to take advantage of getting the most amount of power from a partially shaded environment the ideal inverter would be capable of dealing with a very large range of input voltages. An inverter that could work from say 25 to 400 volts would be ideal.

Is there any inverters that have that kind of range? If not - why not?

In a properly designed system, partial shading won't drop the string below the start voltage of the inverter. There are inverters out there that can take a lower voltage input, but it is an expensive solution to what is a relatively simple design constraint. A better choice would probably have been a single string, or using equipment that handles shade much better like SolarEdge, microinverters, or another optimizer.

[mperkel]

So - what I'm hearing is that instead of 2 strings of 7 I should have 1 string of 14?

The inverter is an ABB PVI-3.0-OUTD-US.

[sensij]

So - what I'm hearing is that instead of 2 strings of 7 I should have 1 string of 14?

The inverter is an ABB PVI-3.0-OUTD-US.

Inetdog's suggestions in this post might help. If the shade is just seasonal, you might be better off just accepting the lost generation if it is expensive to make changes. If the shade is more than seasonal, a different type of inverter would be performing much better. Relying on bypass diodes daily in a 14 panel string is, in my fairly uneducated opinion, asking for trouble, since of the parts that can fail in a panel, they are probably higher on the list than most. By putting the full string power through the diodes instead of just one panel's worth, it seems like it would be a lot harder on them.

[mperkel]

Part of my questions is trying to understand the technology. I used to do a lot of electronic design years ago. But it seems to me that current inverter designs are leaving a lot of watts on the table. A lot of watts that could be easily recovered with a little design effort.

[sensij]

Part of my questions is trying to understand the technology. I used to do a lot of electronic design years ago. But it seems to me that current inverter designs are leaving a lot of watts on the table. A lot of watts that could be easily recovered with a little design effort.

I think that most of the grid tire inverters you see are attempting to optimize efficiency, cost, and reliabiliy. Very few watts are left on the table by the inverter when the system is designed properly. I would guess that the design effort you are suggesting is harder than you think, and / or there is very little incentive to design a lower turn on voltage, especially if the power conversion efficiency or the reliability takes a hit to handle the wider range.

Shade and string inverters are a bad combination. Changing the turn on voltage doesn't really fix that. Even with inetdog's suggestions, your system would still underperform a system with panel level optimization if the shade occurs all year.

Put more simply, shading is a panel level problem. Addressing it at the panel makes more sense to me than attempting to compensate for it at the end of the string with the inverter.