Mismatched PV panels in series

[Siwastaja]

Hi,

Asking for practical experiences from those of us who know what we are talking about, because Googling this seems to be an infinite source of elementary level (+ often wrong even at that) bullshit.

So I have a 3kWp rooftop system installed by the previous owner of the house. One series chain, one MPPT tracker in the inverter. Ten panels of approximately 300Wp. Last time I checked today with dirty (spring; pollen, no rain for weeks) panels, production with blue skies and nearly optimal solar angle is 2.2kW at 299V so around 7.4A. All-time high peak record is at 3700W.

The system is installed in 2017 and I doubt I could get the exact same panel.

There is room for 5 panels more on the roof (10 -> 15), and besides, the inverter is grossly oversized (5kW) so it would be just obvious to add those 5, for a 4.5kWp system.

My understanding is, of course, that I need to match
* Most importantly, rated current at maximum power point
* But also, temperature coefficient thereof
* And also, having similar response to stray light vs. direct light

The last two bullet points so that maximum power point current would be similar not only in optimal conditions, but over wider operating area. But I guess finding similar enough panels would be a true challenge - so matching rated current is likely the best I can do.

Would you agree so far?

Since 5 < 10 (i.e., less than 100% increase in system size), the old panels should dominate, and it would be a smaller monetary loss to buy panels with a tad larger current rating than the existing ones - only the new ones, 5 out of 15 would be operating at <100% current. If I buy panels with too small current rating, then the old ones - 10 out of 15 would be operating with too little current: more wasted resources. Again, would you agree with this?

MPPT obviously should be able to find the global optimum for the whole chain.

Any other practical comments?

As it is now, both roof area and inverter power capability is wasted resource, and I do want more PV. But of course if 50% of the investment goes to waste, I buy 1500Wp worth of panels but only get 750Wp more due to the whole system operating outside optimal power point, then it does not make sense to extend the system. But my gut feeling says, match the rated current and add maybe 10% and it'll be good enough?

[bdunham7]

MPPT obviously should be able to find the global optimum for the whole chain.

Any other practical comments?

Some inverters have a MPPT algorithm that can get 'stuck' at a local peak power point and miss that a better peak somewhere else.  This most often shows up in systems with shading, but I suppose a mismatched system might cause it as well.  So whether it finds the global optimum depends on whether or not it has that feature--and it isn't universal.

Assuming the panels are the same voltage, or at least the same number of similar cells, than as long as the new ones are rated for more current than the old ones I think the worst-case scenario would be an output equal to 15 of the original panels.   This is all assuming that all the panels, new and old, are the same basic type--presumably poly-silicon.  Although in this case, even mixing poly- and mono- might not be much of  a problem.  Just make sure they all have operational bypass diodes if you have any shading at all during any time of the day where you care about production.

And yes, I agree that larger is better, especially since you probably won't have to pay much more for them as solar panels keep increasing in output.

[Siwastaja]

Thanks for comments, the inverter is StecaGrid 5003 and they advertise "global MPP". It probably does a wider perturbation scan every now and then. I would guess problems with MPPT would happen in partial shadow conditions more easily than just mismatched panels.

Any particular reason you suggest same number of cells per module? Why would the number of cells in a module affect anything, in the end, all cells just end up in series.

I could imagine voltage per cell would be indicative of other possible differences, though.

I totally agree that if Impp_new > Impp_old, then the whole system should act like 15 old panels, and only 5*(Pmpp_new - Pmpp_old) power capability is wasted. So I would need to figure out the exact model of the panel I have - get on the roof and try to get a camera between the roof and the panels to read the labels - and then pick panel type with maybe 5-10% higher Impp. It would be a pity if the new type performed significantly more poorly in low or diffuse light conditions, bringing down the whole system.

My old panels are full black so they might be mono, but might be poly as well. Mixing poly and mono might or might not pose a problem because depending on who you ask, mono/poly is better in low light or diffuse light conditions, than the other one. Found this great resource about the classic myths: http://www.solarblogger.net/2014/12/mono-vs-polycrystalline-solar-cells.html   As always with components, only actual data matters, and "rules of thumbs" are usually incorrect.

[bdunham7]

Any particular reason you suggest same number of cells per module? Why would the number of cells in a module affect anything, in the end, all cells just end up in series.

No, I wasn't really suggesting you need to the use same-voltage panels in a string, just using that assumption to make the rest of the sentence an easy calculation.  Of course you have to watch the maximum string voltage--I think the MPPT algorithm will cause the higher-current panels to run very near their OCV at high power levels, depending on how much greater their current capacity is.  Here you can find 300W 72-cell mono panels cheaply on the used market, although that probably isn't the case for you.

[trobbins]

Definitely need to assess the inverter max DC voltage rating, and the worst case OCV of the proposed array at lowest temp.

[Siwastaja]

Definitely need to assess the inverter max DC voltage rating, and the worst case OCV of the proposed array at lowest temp.

Yeah, thanks for reminding. Here it's completely possible to have very optimal solar angle in late April yet like -10degC temperature, producing highest possible OCV if the inverter, for any reason, shuts down or sets current to zero, even briefly. So need to check total OCV in such conditions against the inverter ratings. I know design software take all this into account but it isn't hard to calculate manually.

[tszaboo]

My suggestion is not to change the existing installation. It has a warranty, which is done if you change it. It inverter only has one MPPT input ,so there is very little you can do about it.
Plus, inverters are cheap. A 2KW inverter costs something like 4-500 EUR. You pay a lot more for the installation.

[Siwastaja]

Fair enough.

I'm comparing between these two options:

* 5 panels
* mounting of 5 panels on the roof
* complete wiring work
* price of inverter
* installation of inverter, electrician needed

vs.

* 5 panels
* mounting of 5 panels on the roof
* minimal wiring work, just open one or two loops of the existing wiring and add new panels in series

Former is like 5-6 k€ job. I would expect the latter to be less than half of that. It's also something I could almost consider doing myself despite having all sorts of other things to do.

Of course, optimally the previous owner would have bought a larger system to begin with. I'm sure it was expensive anyway; it was sold by a company known to give false promises and then went bankrupt I think and continued the work with a new name. Anyway, thanks to the Current Things, now the false promises have realized, as I'm buying from the grid for 5.1 cents/kWh (+ 8 cents for transmission) but selling by spot price, it's been 10-50 cents,  maybe 25 cents on average this spring! So I actually have had to reverse my thinking and try to avoid using my own production, and sell as much as possible.

[tszaboo]

That 5K quote looks to be a bit expensive IMHO. I had put up 11 panels for that much last year, and that was a final quote, including paperwork done by them. Solaredge system, with EU made panels
The wiring for modern panels is minimal, they come with wires preinstalled, and they click together.
Though I had to shop around because most installers were giving me 20-30% higher quotes.

[Siwastaja]

Well, this is Finland. Cheapest full installs of 10 panels + inverter are 8000€ as we speak. Neighbor got a quote of 10000€. Of course the Current Events show already in the prices. It has never been below 6-7k, though.

This is why I DIY almost everything.

[nctnico]

My suggestion is not to change the existing installation. It has a warranty, which is done if you change it. It inverter only has one MPPT input ,so there is very little you can do about it.
Plus, inverters are cheap. A 2KW inverter costs something like 4-500 EUR. You pay a lot more for the installation.

I agree. Just get an extra inverter for the 2nd installation.

15 panels in series likely exceeds the voltage for the inverter anyway.

[bdunham7]

I agree. Just get an extra inverter for the 2nd installation.

15 panels in series likely exceeds the voltage for the inverter anyway.

If he has and uses 60-cell panels, which might be correct given his numbers, then he's looking at ~600V max OCV, so not an issue.  72-cell and he might be approaching 750V, but the inverter is rated for 1000V max and 800V operating, so still not an issue.

As for how to do it, I personally wouldn't worry about warranties (if he even has any) and I'd be pretty hawkish about simplicity and capital costs.  My own system has paid for itself 2X and a big part of the reason for that is I was ruthless about not paying large profits to installers at the time.  And now that the electric company here has managed to persuade the regulatory authorities to gut the value of solar power, I'm darn glad I'm no longer worried about the payback time.  Adding on would be a matter of properly bolting on 5 panels and switching a few wires.  Adding an inverter means a whole second set of cables, a second inverter somewhere and all the expenses related to that.  Not a good plan for something that is premised on a financial return.  And don't know what his house looks like or how it would be laid out, but it sounds aesthetically yuck.

[Siwastaja]

It would be a weird inverter if it did not support the typical number of panels to get the rated power.

More specifically, similar panels are rated to OCV of 40V at +25degC and -0.273%/degC so +11% or ~45V at -15degC.

15 panels is 675V max OCV, then. Spec for the inverter says maximum input voltage 1000V, and operating input voltage range 250V .. 800V.

Currently the problem is that I lose some production in partial shading or poor conditions because the inverter is so much oversized, it's clearly designed for at very least 12 60-cell panels, 10 is too few. I quickly drop below that 250V in partial shading.

Maybe I did not emphasize the oversizing of the inverter enough.

[Siwastaja]

As for how to do it, I personally wouldn't worry about warranties (if he even has any) and I'd be pretty hawkish about simplicity and capital costs.  My own system has paid for itself 2X and a big part of the reason for that is I was ruthless about not paying large profits to installers at the time.  And now that the electric company here has managed to persuade the regulatory authorities to gut the value of solar power, I'm darn glad I'm no longer worried about the payback time.  Adding on would be a matter of properly bolting on 5 panels and switching a few wires.  Adding an inverter means a whole second set of cables, a second inverter somewhere and all the expenses related to that.  Not a good plan for something that is premised on a financial return.  And don't know what his house looks like or how it would be laid out, but it sounds aesthetically yuck.

Yes I agree with this. In fact, if I wanted another inverter and another set of cabling, I would make it another complete installation, maybe free-standing array on the ground, AND upgrade my current roof array to match the inverter - and the free area on the roof! My roof isn't that big, so it only has room for 5 more. Which also happens to be within the excess inverter rating. The less I spend money on upgrading that, the more I can spend doing something else more meaningful than trying to squeeze a few % more performance of the 5 added panels.

I'm looking for reasons that make those 5 panels way more underperforming than maybe 10-20%, but it seems there is no such problem.

[bdunham7]

I'm looking for reasons that make those 5 panels way more underperforming than maybe 10-20%, but it seems there is no such problem.

IMO the efficiency of a solar system should be calculated in kWh/$.  A system that 'clips' or otherwise 'wastes energy' is not worth worrying about.  A system that costs too much or doesn't work, that's a problem.

[nctnico]

It would be a weird inverter if it did not support the typical number of panels to get the rated power.

More specifically, similar panels are rated to OCV of 40V at +25degC and -0.273%/degC so +11% or ~45V at -15degC.

15 panels is 675V max OCV, then. Spec for the inverter says maximum input voltage 1000V, and operating input voltage range 250V .. 800V.

Currently the problem is that I lose some production in partial shading or poor conditions because the inverter is so much oversized, it's clearly designed for at very least 12 60-cell panels, 10 is too few. I quickly drop below that 250V in partial shading.

Maybe I did not emphasize the oversizing of the inverter enough.

Maybe get an inverter that is designed to deal with 2 strings then and fix 2 issues at the same time. I looked into getting solar panels myself a couple of years ago and over here there is a large number of sellers that sell used solar inverters for very reasonable prices (low enough to replace it after 5 years and still come out on top). In my case I was looking at putting 2x 7 or 8 panels in series using two strings. AFAIK keeping the voltage low also helps to deal with isolation breakdown issues in the panels. From what I've read running panels at nearly 1kV referenced to ground is not something they like.

Adding on would be a matter of properly bolting on 5 panels and switching a few wires.  Adding an inverter means a whole second set of cables, a second inverter somewhere and all the expenses related to that.  Not a good plan for something that is premised on a financial return.  And don't know what his house looks like or how it would be laid out, but it sounds aesthetically yuck.

It looks like the best option would be to get a different inverter anway and that can probably be achieved for free (sell the oversized one and get a used one in return). Extra wiring is not much of an issue IMHO. The hole to lead the cables indoors is already there.

[bdunham7]

From what I've read running panels at nearly 1kV referenced to ground is not something they like.

Yes, PID--Potential Induced Degradation.  A selling point for microinverters, at least 10 years ago.  I thought there were solutions to this issue by now--low leakage PID resistant modules, isolated inverters, grounding or neutral referencing the negative DC side in transformerless designs and so on.  But IIRC, the real problems only showed up when you started approaching negative 1kV, and in his case he probably only has about 450V normally if he has 15 60-cell panels.  So even if it the issue hasn't been mitigated elsewhere, it probably isn't an issue.  However, that got me thinking--what would a  measurement (or scope) on the DC (either side) to ground look like with a 3-phase delta transformerless inverter?