Underestimated Results of Cleaning Panels

[sensij]

It might get confusing if the other system owner also decides to clean his

Yeah, there are a few details that can get in the way of the comparative analysis I offered. One aspect of it that I don't like is that I couldn't find any systems that were very close to the same kWh / kW output as the OP's system. OP's system seems to have a better view of the horizon than most, maybe a higher elevation than nearby systems. The correlations look good, but the existence of a common mode offset between them makes it less convincing than it might otherwise be. The step change in the correlation is there, and 10% is a rough guess at an expected value, but the 95% confidence interval would probably be something more like 5% - 15%.

[J.P.M.]

Thanks for all the replies and research noted above. This board has some very intelligent people for sure and I love to learn from each of you. Below is a little chart using Weather Underground data. I was looking for any other clues. The only thing I notice of note is the recent weather has a been a bit cooler. Averaging the best pre-cleaning days (in yellow), which I am assuming was clear skies equates to 32.28 KWh per day. Post cleaning (in blue) the average is 36.20, so as noted by someone above, a 10% increase. Although the average temp for these comparisons was 89 degrees to 77 degrees.

The good and bad news is that we are expecting temps to again creep back into the low 90's, so I'll have some more data later this week. Maybe some of the difference post cleaning is coincidental temperature variance. I will report back though.

And I didn't even think about getting shocked. Dope! See, I'm not as smart as you all for sure...
ScreenHunter_616 Sep. 26 11.09.jpg

Save me a bit of work and pardon my laziness, what's your zip ?

[sensij]

Save me a bit of work and pardon my laziness, what's your zip ?

95670

Here is a previous thread of some relevance

https://www.solarpaneltalk.com/forum...atts-estimates

In that thread, the system was 21 panels (5.88 kW), and here we are talking about 22 panels (6.16 kW), but maybe there was a misunderstanding somewhere in there.

[ButchDeal]

Yeah, there are a few details that can get in the way of the comparative analysis I offered. One aspect of it that I don't like is that I couldn't find any systems that were very close to the same kWh / kW output as the OP's system. OP's system seems to have a better view of the horizon than most, maybe a higher elevation than nearby systems. The correlations look good, but the existence of a common mode offset between them makes it less convincing than it might otherwise be. The step change in the correlation is there, and 10% is a rough guess at an expected value, but the 95% confidence interval would probably be something more like 5% - 15%.

simpler test is to perform partial washing on a solarEdge system. Know some one that might have a SolarEdge system with a public monitor site?

[sensij]

simpler test is to perform partial washing on a solarEdge system. Know some one that might have a SolarEdge system with a public monitor site?

Yeah, no kidding. I'm feeling like a jacka__ for not having my weather station hooked up on this house yet to provide supporting data for a more tightly controlled experiment.

Edit: I might do the test anyway. If the panels look anything like the rest of the unwashed surfaces around my house after what has been a dry summer, there ought to be some detectable gain available. On my system, panels 1.1.7 and 1.1.10 are functionally twins. Cleaning one and not the other should provide an interesting look at daily energy, and also panel current and voltage.
Panel comparison.JPG

[J.P.M.]

95670

Here is a previous thread of some relevance

https://www.solarpaneltalk.com/forum...atts-estimates

In that thread, the system was 21 panels (5.88 kW), and here we are talking about 22 panels (6.16 kW), but maybe there was a misunderstanding somewhere in there.

Thanx. Chased an 08/06/17 thread that referenced some output data. Looking for any nearby pyranometer data.

[J.P.M.]

Yeah, no kidding. I'm feeling like a jacka__ for not having my weather station hooked up on this house yet to provide supporting data for a more tightly controlled experiment.

Edit: I might do the test anyway. If the panels look anything like the rest of the unwashed surfaces around my house after what has been a dry summer, there ought to be some detectable gain available. On my system, panels 1.1.7 and 1.1.10 are functionally twins. Cleaning one and not the other should provide an interesting look at daily energy, and also panel current and voltage. Panel comparison.JPG

Others have been wishing you'd hook it up too for comparison purposes.
No pressure however.

FWIW, after adjusting for my shade, our systems seems to be running close to the same in terms of kWh/STC kW. I haven't touched/cleaned the array since 06/21/2017.
Excluding today, I'm at 110.37 kWh/STC kW for the month with ~~ 4-5 % shading, or ~ 110.37/.955 = 115.6 kWh/STC kW after shading. Looks like you're shadeless at 115.5 kWh/STC kW Orientations are similar. Off topic but more anecdotal proof that Sunpowwer systems (mine) performs about the same as other quality stuff (yours).

Edit: More snooping, looks like your startup happened the same day of my last cleaning. As of now, my array's performance is running about 5-6 % fouled and seems to have leveled off at about that level for the last 3 weeks or so. No precip., but a few days of measurable morning dew. Point is, depending on how long your panels sat on the roof before startup, we may be similarly fouled as some similar f(time). Or, if our fouling patterns and rates are somewhat similar, as I suspect, and our outputs seem similar as the seem to be, at least for Sept., your fouling might be something like 5-6 % or so. If you do some differential panel cleaning, I'll get a bit of perhaps more than anecdotal confirmation/not of my fouling measurement techniques.

[J.P.M.]

simpler test is to perform partial washing on a solarEdge system. Know some one that might have a SolarEdge system with a public monitor site?

Makes me slightly wish I had gotten a solar edge system, but then I might not have had the impetus to root around under my array measuring all the panel's temps. well over 100 times.

All the fun I would have missed.

[inspron]

For future reference, if anyone really curious and want to see the effect of before and after cleaning reliably - simply clean all your panels except for 1 or 2 that, that will be you control. See how the cleaned panels perform immediately the next day compared to the unclean panels. You can track the cleaning effectiveness over time as well. Really simple controlled experiment. Don't bother with PVoutput or any other guesswork. - of course, only do this if you have a Solaredge or similar system.

[max2k]

For future reference, if anyone really curious and want to see the effect of before and after cleaning reliably - simply clean all your panels except for 1 or 2 that, that will be you control. See how the cleaned panels perform immediately the next day compared to the unclean panels. You can track the cleaning effectiveness over time as well. Really simple controlled experiment. Don't bother with PVoutput or any other guesswork.

if you have panel level monitoring - yes. PVOutput also works for central inverter based system as you can pick your control from one of the nearby systems which performs similarly to yours.

[J.P.M.]

For future reference, if anyone really curious and want to see the effect of before and after cleaning reliably - simply clean all your panels except for 1 or 2 that, that will be you control. See how the cleaned panels perform immediately the next day compared to the unclean panels. You can track the cleaning effectiveness over time as well. Really simple controlled experiment. Don't bother with PVoutput or any other guesswork. - of course, only do this if you have a Solaredge or similar system.

I believe butch covered that. Also, Sensij's suggestion of finding 2 panels with identical characteristics is a very good additional idea.

[sensij]

I believe butch covered that. Also, Sensij's suggestion of finding 2 panels with identical characteristics is a very good additional idea.

and an idea that we'll see in practice tomorrow. 1.1.7 got a hose down and scrub tonight (using a swiffer pad), with some overspray towards the end of the row but 1.1.10 stayed dry. I don't have words to describe the concern on my wife's face as I explained to her that I was stringing together 100 ft of hose to haul onto the roof in the dark... to clean one panel.

A completely separate topic, maybe worth revisiting at some point... in the light of the flashlight, I could pick out in the deposition patterns of the dirt where the optimizers were located.. My hot take on this is that the optimizers create a warm spot on the panel that creates enough difference in local convection to encourage the dirt to settle elsewhere. It is consistent with some of the differences in production that I'm seeing... I laid the array out with two optimizers under some panels and none under others, and those with the extra optimizers appear to underperform. I'm only commenting on it because I've been vocal about some of the possible advantages of optimizers in a shade free installation (dealing with thermal mismatch, etc), but haven't talked much about the possible performance hit that occurs from the heat rejected by the optimizer (and impairment to airflow where it is mounted). I'd guess that the benefits and costs more or less cancel out, but a trip onto the roof with an IR camera will be needed eventually to dig into this more deeply.

[AzRoute66]

[...] but haven't talked much about the possible performance hit that occurs from the heat rejected by the optimizer (and impairment to airflow where it is mounted).

Wow. That would let a lot of air out of a lot of balloons. Might even lead to a whole look at optimizer mounting practices.

[ButchDeal]

and an idea that we'll see in practice tomorrow. 1.1.7 got a hose down and scrub tonight (using a swiffer pad), with some overspray towards the end of the row but 1.1.10 stayed dry. I don't have words to describe the concern on my wife's face as I explained to her that I was stringing together 100 ft of hose to haul onto the roof in the dark... to clean one panel.

A completely separate topic, maybe worth revisiting at some point... in the light of the flashlight, I could pick out in the deposition patterns of the dirt where the optimizers were located.. My hot take on this is that the optimizers create a warm spot on the panel that creates enough difference in local convection to encourage the dirt to settle elsewhere. It is consistent with some of the differences in production that I'm seeing... I laid the array out with two optimizers under some panels and none under others, and those with the extra optimizers appear to underperform. I'm only commenting on it because I've been vocal about some of the possible advantages of optimizers in a shade free installation (dealing with thermal mismatch, etc), but haven't talked much about the possible performance hit that occurs from the heat rejected by the optimizer (and impairment to airflow where it is mounted). I'd guess that the benefits and costs more or less cancel out, but a trip onto the roof with an IR camera will be needed eventually to dig into this more deeply.

Way to take one for the team.
Can you recharge your public solaredge site for everyone?

we should watch it till the clean ones rematch the dirty ones to see how long the cleaning lasts.

on the optimizers, were they mounted flat side up or the electronics up? About what is the gap from optimizer to the back of the modules?

[J.P.M.]

and an idea that we'll see in practice tomorrow. 1.1.7 got a hose down and scrub tonight (using a swiffer pad), with some overspray towards the end of the row but 1.1.10 stayed dry. I don't have words to describe the concern on my wife's face as I explained to her that I was stringing together 100 ft of hose to haul onto the roof in the dark... to clean one panel.

A completely separate topic, maybe worth revisiting at some point... in the light of the flashlight, I could pick out in the deposition patterns of the dirt where the optimizers were located.. My hot take on this is that the optimizers create a warm spot on the panel that creates enough difference in local convection to encourage the dirt to settle elsewhere. It is consistent with some of the differences in production that I'm seeing... I laid the array out with two optimizers under some panels and none under others, and those with the extra optimizers appear to underperform. I'm only commenting on it because I've been vocal about some of the possible advantages of optimizers in a shade free installation (dealing with thermal mismatch, etc), but haven't talked much about the possible performance hit that occurs from the heat rejected by the optimizer (and impairment to airflow where it is mounted). I'd guess that the benefits and costs more or less cancel out, but a trip onto the roof with an IR camera will be needed eventually to dig into this more deeply.

Now we're getting somewhere.

If misery loves company, I'm less unhappy as a result of the looks you got. I've endured the looks and comments from my neighbors and all the golfers just off my backyard (not to mention what my bride thinks) when I would duck under my array (yea - I can do that ~~ 10-11.5 " clear by design. Kind of like being crammed into a 18.75 deg. tilted toaster) and measure and record each of 16 panel's temp. in four places over about 6-8 minutes, then duck into the garage and record (string) inverter voltages, string outputs and inv. output for 2 minutes, then rush back to the roof and repeat the temp. measurement process, the whole process taking about 16 - 18 minutes, centered around the minute of minimum solar incidence angle on the array. Lots of asses and elbows for those 16-18 minutes.

On the separate topic - great observation - some thoughts/speculation:

I bet the optimizers are changing the dirt deposition patterns, and, agreeing with you, having observed the same pattern on my panels at the wiring junction, I'd speculate that's happening due to the optimizers increasing the local panel top surface temp. around the optimizer, either from the optimizers possible (likely ?? - Butch ??) internal heat generation and/or increases in local temps when air circulation around the optimizer is reduced or cut off by the optimizer.

As for some possibilities for the deposition mechanisms involved:

If your overnite/mornings are as foggy as mine often tend to be, and I bet they are, or more so, your array will often be wet with dew until maybe 8 -8:30 hrs. or so. A lot of that dew will bead up and tend to run down the panels (and, BTW, that run off being what's responsible for the bathtub ring at the lower 6 - 12" or so of a panel that shows up after a period of time).

I'm thinking that if an optimizer both generates some small amount of heat on its own, and also acts as a local insulator to the rear of a panel, and thus raising the local panel temp. slightly, both top and bottom, any standing dew over the optimizer, maybe before the condensation quantity/mass develops enough to have gravity overcome surface tension and surface adhesion effects over the optimizer, will be at a (slightly) higher temp. and thus tend to evaporate more quickly by virtue of the increased delta T between the condensate over the optimizer and dew point, vs. the condensate temp. over the rest of the panel and the dew point. That will tend to fix the dirt in place that was acquired over the prior day over the optimizer much like the dirt that makes up the aforementioned bathtub ring, by not running off as much and tending to fix the dirt in place a bit more. Over time a small additional residue will accumulate and become more noticeable.

A side bar: As that dirt accumulates all over the plate, it will tend to block more of the irradiance. Since that irradiance raises panel temps in direct proportion to it's intensity, the larger local dirt deposition rate will lower local irradiance and lower the local plate temp. That will occur in theory and practice. How much the effect of local differential dirt deposition will have on local temp. is left as an exercise for the (local) student.

I've noticed the dew on my array often tends to evaporate shortly after the sun hits the panel, primarily as f(dew point vs. panel temp.), but if the panel at/over the optimizer is at a higher temp., that will tend to help any standing water over the optimizer evaporate sooner than the surrounding water and perhaps the "fixing" a bit of the dust/dirt acquired over the prior day before it runs off with the lower temp. condensate (with some of that runoff being what tends to get slowed down and perhaps stopped by the ~ 1/32" to 1/16" lip that is the edge of most panels).

A bit off the dirt discussion: As for how much of a temp. hit an object/obstruction may have on local panel temp.: An IR camera will give a picture. I found an IR thermometer (the one I measure panel backside temps with) gives a decent and more useful quantitative temp. estimate. The frontside temp. of my panels over the box bolted to the backside runs hotter than the rest of the panel. That delta T, as you'd expect, varies from a deg. C or so to ~~ 4-6 C., mostly as f(POA irradiance, wind vector, panel's location in the array relative to wind direction). I'm not certain what quantitative effect(s) that delta T may have on individual panel performance, but since measuring all the array's 16 panel temps. "quasi simultaneously" under clear skies as described above well over 10 times, and seeing about the same ~~ 1- 6 C. panel temp. diff. over the entire array, leading to trailing edge relative to the wind vector, I'm not sure what the effects of local temp. elevation on a panel will do to output in a way that's measurable, and thus quantifiable. Not saying local panel temp. differentials (as differentiated from gross temp. effects) will/won't have an effect on individual panel output, just that I'm not sure it's measurable (without removing/separating the obstruction from the panel), or significant given that I've observed and measured how the temp. over an entire array can vary about as much as local plate temps. vary in the region of things bolted to the back of it.

Fascinating subject. Like working on stuff I used to get paid to do.