Hiding stuff under the panel.

[sensij]

For me, the strongest argument against SolarEdge is the fact that their system is so proprietary. Only SolarEdge optimizers work with a SolarEdge inverter, and vice versa. If eventually someone reverse engineers their products and produces a "Solaredge compatible" optimizer, I doubt I would even consider a string inverter again. (I'm still specifying string inverters for a 500 kW commercial system I'm involved with, for this reason).

[foo1bar]

I mis-wrote earlier - it's 190 modules, not 196.

I wouldn't and I don't care

Maybe you wouldn't care that 1 module isn't producing.
But I do.
That module cost about $830 to be installed.
It was installed because we want/expect it to produce electricity and reduce our utility bill. (about $120/year of electricity.)

I think your expectations are a little unrealistic

What part is "unrealistic"?
Determining that one module of the 190 wasn't connected to an optimizer (or was DOA)?
Expecting the installer to connect every one of the modules we paid for?

This is a real life install. And I can actually see the evidence that 1 of the 190 is not producing. So I think that part is very realistic.
And I think it's reasonable to expect the installer to properly install everything we pay him for.
I think it's also realistic to expect that mistakes are made when installing 190 modules, 100 optimizers, and 4 inverters. Which is why I checked that all the optimizers are reporting (they all are) and that we had the proper amount of power coming out (We should have had 10 optimizers at half the power of their siblings. Instead we have 11 at half-power)

[foo1bar]

exactly one- non producing module would degrade entire string. Besides strings can't be made longer that 14 panels in series due to 600V input voltage limit.

Hm - I wonder how this would show up on string inverters...

With 190 modules it'd probably be 6 strings of 13 and 8 strings of 14.
I think most string inverters (even 3-phase ones) have two mppt inputs.
With 14 strings, that'd probably mean 2 strings on some inputs but not others.
So - if one of the strings was short a module, what would happen?
I think if it was in parallel on the same mppt as a proper length string it'd be noticable as "inverter 3 is producing like it's only got 2 strings instead of 3" (or 3 instead of 4) - very easily noticed.
If it wound up with the short string on one mppt input by itself and the other mppt input had the pair of strings then it'd be unlikely to be detected (ex. you wouldn't notice the difference between 1x13x300W+ 2*14*300W (12kW DC) vs. 1x12x300W+2*14*300W (12.3kW DC)

I also wonder if the installer might go "Oh, this string has voltage like it's 13 long but I have it paired on an mppt with a 14-long? I must have made a mistake in my wire labeling - let me just put it with the 13-long on the other input." (You'd hope they'd be more diligent, but I think that could really happen.)

Ah - well, all fun for a thought exercise...

bottom line - OP asked what the benefits are. This is a real case where it'll make a >$100/year difference to the organization and I think there's a good chance it wouldn't have been caught with a string inverter.


Also I think they were able to save some in wiring/equipment between the panels and the inverters... I'd have to look at how many pairs of wires they ran from the inverters to the arrays.
At most it's 10 pairs (vs. 14 pairs for string inverters). Might even be less - not sure if they combined some by the arrays.
(But it's probably not a huge difference in cost.)

[max2k]

right, so you have strings in parallel in a big system like that. Now, bypass one panel out of one of those strings. By itself, that string's voltage would fall by an easily detectable amount. In parallel with other strings, the voltage drop will be harder to see, since the mppt still wants to maximize power output, and those parallel strings aren't going to like having their voltage pulled down much.

of course, this only works in the case of exactly one failure. If you have a second failure before the first one was detected, and it happens to be in the other mppt that you are using for your reference voltage, now you've lost the ability to easily detect either failure.

I have never said that individual panel monitoring does not have advantage in case of problems. What I said is even without individual level monitoring those problems are not hard to troubleshoot. Given panel failures don't happen very often I can't justify SE application just for the troubleshooting reasons. Another member solarix who sounded like installer reported alarming # of optimizers he had to replace over last year on his own installations. Just using his stat it looks to me SE system helps in troubleshooting the problems it created in the first place .

I don't know how commercial scale installations are done but on residential installs even my 3.8 kW inverters have 2 individual MPPT inputs each and higher rated models have 3, there's no need to put strings in parallel. Besides even one string is already producing current close to MPPT input max rating, it simply can't take another string..

IMO your example of simultaneous failure of 2 panels in 2 different strings is not very realistic one. With increasing number of strings in the system this probability will be even lower as you need only 1 string to serve as 'reference'.

[max2k]

...I'm still specifying string inverters for a 500 kW commercial system I'm involved with, for this reason.

infidel

[max2k]

Hm - I wonder how this would show up on string inverters...

With 190 modules it'd probably be 6 strings of 13 and 8 strings of 14.
I think most string inverters (even 3-phase ones) have two mppt inputs.
With 14 strings, that'd probably mean 2 strings on some inputs but not others.
So - if one of the strings was short a module, what would happen?
I think if it was in parallel on the same mppt as a proper length string it'd be noticable as "inverter 3 is producing like it's only got 2 strings instead of 3" (or 3 instead of 4) - very easily noticed.
If it wound up with the short string on one mppt input by itself and the other mppt input had the pair of strings then it'd be unlikely to be detected (ex. you wouldn't notice the difference between 1x13x300W+ 2*14*300W (12kW DC) vs. 1x12x300W+2*14*300W (12.3kW DC)

I also wonder if the installer might go "Oh, this string has voltage like it's 13 long but I have it paired on an mppt with a 14-long? I must have made a mistake in my wire labeling - let me just put it with the 13-long on the other input." (You'd hope they'd be more diligent, but I think that could really happen.)

Ah - well, all fun for a thought exercise...

bottom line - OP asked what the benefits are. This is a real case where it'll make a >$100/year difference to the organization and I think there's a good chance it wouldn't have been caught with a string inverter.


Also I think they were able to save some in wiring/equipment between the panels and the inverters... I'd have to look at how many pairs of wires they ran from the inverters to the arrays.
At most it's 10 pairs (vs. 14 pairs for string inverters). Might even be less - not sure if they combined some by the arrays.
(But it's probably not a huge difference in cost.)

it was my misunderstanding of your system layout, I thought you were referring to inter- module production level difference of 0.5%.

[sensij]

I have never said that individual panel monitoring does not have advantage in case of problems. What I said is even without individual level monitoring those problems are not hard to troubleshoot. Given panel failures don't happen very often I can't justify SE application just for the troubleshooting reasons. Another member solarix who sounded like installer reported alarming # of optimizers he had to replace over last year on his own installations. Just using his stat it looks to me SE system helps in troubleshooting the problems it created in the first place .

I don't know how commercial scale installations are done but on residential installs even my 3.8 kW inverters have 2 individual MPPT inputs each and higher rated models have 3, there's no need to put strings in parallel. Besides even one string is already producing current close to MPPT input max rating, it simply can't take another string..

Solarix is talking about optimizers from the first generation SolarEdge equipment, from several years ago, not more recent models. His recent thread is related to failures of Suniva brand solar panels, exactly the kind of failures that panel level monitoring can help detect.

Commercial scale inverters have just two mppts, and typically put 2-3 strings in parallel. SMA's Tripower series, from 12 kW to 30 kW, allows 33 A per mppt tracker, and strings up to 1000 V (rated mppt range up to 800 V). Strings in parallel are necessary for the inverter to perform as intended.

Detecting a bad panel in a large system like that is not as easy as you seem to think.

[Mike90250]

That's why it's handy to have a combiner on parallel strings. I can easily test the balance of string 1 vs 2 & 3 and it only takes a couple minutes to see if there is an issue. More than 20 watts difference between strings would upset me.

[wienerdog]

What actually happens when a panel goes bad? IE if I had 2 strings of 11 panels on 2 channels of the inverter does 1 string have less voltage than the other when the same sunlight is applied to both? If so how much of a difference is bad?

[max2k]

What actually happens when a panel goes bad? IE if I had 2 strings of 11 panels on 2 channels of the inverter does 1 string have less voltage than the other when the same sunlight is applied to both? If so how much of a difference is bad?

typically 60 cell panels consist of 3 substrings of cells. If one goes bad (open circuit) bypass diode takes over and the panel voltage (and consequently its string voltage) drops by around 10V. If you have another 'reference' string to compare against it is easily noticeable. If you don't then it's much less obvious as you'd need to know the 'expected' voltage with the drop of only 3% for 11 panel string. The 'good' news is if it continues failing in this fashion finally you'll notice drop of 6% and start troubleshooting. The expected voltage is Vmp at given temperature multiplied by number of panels in the string.