Panels efficiency differential increasing

[scrambler]

As I was doing my monthly solar reporting, I noticed the difference between my best panel and my worst panel seemed higher that I remembered.
Best and worst have been fairly consistent so far and other panel spread in between these two, with several panels closer to the worst, and several panels closer to the best.

So I went back 3 years from installation, and checked the January and July month at mid day, and indeed the difference between my best panel and worst panel is slowly increasing.
Differential is also larger in the summer.

Here are the numbers:
% are mid-day, worst panel production over best panel production, every day, and averaged for the whole month

• January 2020: 100% -- July 2020: 97.6%
• January 2021: 98.3% -- July 2021: 96.5%
• January 2022: 97.8% -- July 2022: 96%
• January 2023: 97.6% -- July 2023: 95%

It is too early for a trend, so I was wondering if anyone with older arrays checked that, and how it evolved over a longer period.
Like did it keep getting worse, or did it taper after a number of years?

Thank you for your experience

[J.P.M.]

Without knowing more information such as the panel temps. when measurements were done as well as other parameters, it's not possible to draw any information from what you present.

Some comments/thoughts:

First off, to maybe answer your question, I didn't check my array's performance in the same way you did, but I did monitor my array's performance and did what I think was a rather extensive and perhaps innovative investigation of my array's fouling rate and fouling characteristics rather extensively for about 7 years after the 1st year's anniversary of startup. During that time and after the1st year's burnin, my array's efficiency seems to have deteriorated at a rate that's a bit less than what Sunpower said it would. The lowest year/year deterioration was ~ 0.22%/yr. of STC efficiency. The highest was ~ 0.42%/yr. The spec sheet listed deterioration rate is 0.40 %/yr.
After 7 years, the estimated total system deterioration over that time was ~ 0.021 or 0.021/7 ~= 0.003 or ~ 0.3%/yr. as an average rate.
Because of required significant figure discipline, only the last figure of 0.3% of has any real validity and that's most likely only slightly better than a dart throw.

I didn't get serious about measuring my array's fouling rate and system efficiency until the 2d year of its operation so I'm ignorant of what that 1st year burnin loss was but from the last set of measurements I took and extrapolating back to new (and assuming new) spec sheet STC efficiency, that 1st year burnin was probably close to a 1% loss and the panels (which are all from the same batch with consecutive serial numbers) probably averaged about maybe 2% or so over nominal STC spec. STC warranty from the spec sheet was -3% under/+5% over.
I stopped measuring array efficiency about 3 years ago as it's a real PITA and I kept getting data that seemed to confirm and repeat what I'd been gathering for several years. Besides, I'm getting too old for this stuff, even if it was rewarding, fun and educational.

One thing I confirmed over the several years and several thousand panel measurements (and was not surprised by it) is that panel temps. over my array varied in a manner very similar to the temperature distribution over a heated (or cooled) flat plate that has air flowing over it in a manner that is described in any undergraduate heat transfer text.
If an array is more or less uniformly irradiated by the sun such as mine or yours, the individual panel(s') temperature(s') at any one point in time over the entire array will be lower (or closer to the ambient air temp.) for the panels at the leading edge of the array (the panels that are most "upwind") than for the "downwind" or panel(s) - the ones at the "trailing " edge of the array.

Depending on wind vector and gustiness (and somewhat on natural convection in the cases of zero wind vector), one on the things I found was that the temperature distribution of my array under conditions of full sun and under clear skies when the cosine of the beam irradiance on the array was > 0.99 was that the temperature difference between the warmest and coolest panels of my array was somewhere between maybe 1 and 3 degrees C and nonlinear with respect to distance from the leading edge with usually about half that temp. difference occurring between a leading edge panel and the one next immediately adjacent to it. Variation in wind vector including magnitude and direction will change that distribution but prevailing winds will tend to set a pattern over time that averaging data will tend to show.

Now, we all know that a panel's efficiency is partially dependent on its cell temperature.
So, if your array behaves in a manner similar to mine, your panel efficiencies will vary all over the array by maybe something between 0% to maybe close to 1% or a bit more. If so, that may be responsible for about half the efficiency difference you note.

One other thing to consider is that arrays will often foul at different rates as f(location within the array). The most common example is that of the bathtub ring that appears at the bottom edge of panels. It's possible and likely that panels within an array will also see differential fouling rates based partially on where they are located within that array. How often is your array cleaned, either by you or mother nature ?

There is also the likelihood, although I suspect the effect won't be very large and may therefore be masked by other effects, of slight manufacturing differences from one run of panels to the next.

Also, there may be some differential effects on efficiency caused by changes in albedo when the surroundings are changed by such things as growing (or removal of vegetation).

You didn't say, or maybe I didn't notice, but are your best panel and worst panel always the same panel ?

I also noticed that your July production ratio from any year is always less than that of the next future Jan. ratio. That would lead me to think that maybe temp, has something to do with the differences between best/worst panels or some differential heating (or cooling) of the panels as I described above.

As for the Jan. ratios being higher than the prior July ratios, perhaps that has something to do with the idea that winter usually sees more precipitation than summer in many regions which would point to array fouling as a contributor to the ratios being different.

Just some thoughts. There's a lot of other possible candidates to explore things.

But without more data, particularly individual panel temps for a start, and if your looking for some ways to measure or estimate panel or array performance deterioration using the best/worst panels' relative output, even if they are always the same two panels, it'll be hard to quantify what the overall annual array performance deterioration might be, not only for those two panels, but the rest of the array as well.

Take what you want of the above. Scrap the rest.

[scrambler]

Thanks,
This was not so much about array global deterioration, but more about the weaker panels appearing to deteriorate faster than the stronger ones (increasing their differential).

As I think I mentioned, which panels are weaker and which are stronger appears to be fairly consistent, so much that I could select one panel which is always the weakest and one panel that is always the strongest, to look at their differential. I have three rows of panels, and the weakest appears to be the top left and the strongest the bottom left.
Location should normally not be a factor, as it is always the same, so whatever impact location has on making one the weakest and the other the strongest, that parameter is not changing over time, when the differential appears to increase.

I made calculations for January and July, because the differential appears to be less when the panels are producing less (low sun elevation) and I suppose cooler (winter). So I don’t see much need in comparing January value of one year with July value of the other, but only in comparing all the January, OR all the July.

• If I look at the differentials in January of each year, I can see a regular increase in differential.
• If I look at the same in July of each year, I see the same just bigger, which again I would attribute to temperature and power due to sun angle.

So I guess I am trying to anticipate what the future may hold.

• If panels should deteriorate at approximately the same speed, then that differential may just be a burn in period and should stop increasing.
• If weaker panels are actually a sign of defects, then differential may continue to increase leading to some panel dying before other.

Just wondering.

[J.P.M.]

Thanks,
This was not so much about array global deterioration, but more about the weaker panels appearing to deteriorate faster than the stronger ones (increasing their differential).

As I think I mentioned, which panels are weaker and which are stronger appears to be fairly consistent, so much that I could select one panel which is always the weakest and one panel that is always the strongest, to look at their differential. I have three rows of panels, and the weakest appears to be the top left and the strongest the bottom left.
Location should normally not be a factor, as it is always the same, so whatever impact location has on making one the weakest and the other the strongest, that parameter is not changing over time, when the differential appears to increase.

I made calculations for January and July, because the differential appears to be less when the panels are producing less (low sun elevation) and I suppose cooler (winter). So I don’t see much need in comparing January value of one year with July value of the other, but only in comparing all the January, OR all the July.

• If I look at the differentials in January of each year, I can see a regular increase in differential.
• If I look at the same in July of each year, I see the same just bigger, which again I would attribute to temperature and power due to sun angle.

So I guess I am trying to anticipate what the future may hold.

• If panels should deteriorate at approximately the same speed, then that differential may just be a burn in period and should stop increasing.
• If weaker panels are actually a sign of defects, then differential may continue to increase leading to some panel dying before other.

Just wondering.

I'd anticipate that the future will hold an array performance deterioration at about the same rate as shown on the panel data sheet with annual system variation pretty much is sync with the daily and 30 day GHI.

What you did may make you feel good, but it won't say much about your array's past or future performance deterioration in any way that's quantitatively useful.