Cleaning panels doesn't really matter - measurement and photo

[J.P.M.]

JPM, I meant PVOutput and have corrected the post. I'm a bit North of you and in high elevation. My array is almost due South, but I do have some shading in the early morning from trees and again in the late afternoon from my chimney. Earlier in September I was generating closer to 40kWh per day prior to this haze coming in. I think my best day was in May of this year when I generated 46kWh.

Thank you.
Elevation here = 1,480 ft. above mean sea level.
Best production day here 06/12/2017, 35.47 kWh = 6.78 kWh/installed STC kW. The array was ~ 4 yrs. old at the time. Cool clear day. Ave. ground level amb. temp. was ~ 5-6 C. below average for the day. Max. cell temp that day was 50.8 C. That daily max for my array seems to jibe w/your best of 46 kWh/6.7installed STC k w = 6.87 kWh/day per installed STC kW.

Anecdotally, for yesterday, I noticed the sky north of me seemed cloudier yesterday and I was almost on something of a demarcation line of clouds and grey as happens 1X/awhile when stuff drifts down out of L.A.

At 40 kWh/day/6.7 STC kW = 5.97 kWh/installed STC kWh, your early Sept est. of output is higher than mine for the period 09/01 to 09/06/2020. Mine for that period was 28.65 kWh/day/5.232 installed STC kW = 5.48 kWh/day average per installed STC kW for that period.


A bit off topic: The similar result between our arrays, while not a smoking gun, is similar to other data I've gathered since 2014 that tends to confirm a statement I've been making for years:
Arrays in the same location and similar orientation tend to produce about the same annual as well as short term output regardless of panel type.
The primary reason I purchased Sunpower equipment was to attempt to verify for myself whether or not the Sunpower claims of "highest efficiency" and higher output per installed kW had any substance or were and still are B.S.
IMO, the claims are B.S.
Beyond some basic quality level, panels are, and for the most part always have been, a commodity. Folks who spend more time/effort finding a quality installer will get more bang for their buck and a better installation than those who worrying/swallow the B.S. advert. hype. Caveat Emptor still applies.

[PugPower]

JPM, I meant PVOutput and have corrected the post. I'm a bit North of you and in high elevation. My array is almost due South, but I do have some shading in the early morning from trees and again in the late afternoon from my chimney. Earlier in September I was generating closer to 40kWh per day prior to this haze coming in. I think my best day was in May of this year when I generated 46kWh.

[J.P.M.]

So you guys inspired me and I decided to do my own un-scientific experiment. I cleaned my panels last night after 7PM and decided to compare the output from yesterday to today. By cleaning I mean I hosed them off from a stepladder. Took all of 10 minutes. I had not cleaned them since last year and they looked pretty dusty. I have a 6.7kW system consisting of 1 Solar edge 6000HD inverter and 20 LG 335w panels. Yesterday the high temp according to my personal weather station was 98.6 F, little to no wind, mostly clear with few clouds, and a light haze. According to PVoutput I generated 31.353kW. Today the high was 102.0, little to no wind, mostly clear with few clouds, and light haze (I honestly could't tell if it was worse or better than the day before, maybe a little better.). Today according to PV watts I generated 33.054kWh. A difference of 5.4% on a day which was 3.4 degrees hotter and maybe slightly clearer than the day before. Basically this un-scientific experiment confirms that cleaning your panels will not result in a big increase in production and like JPM says a hosing off is probably all that is needed and/or worthwhile.

Pug: I don't know what your zip is or where you are in S,D. county but I'm in 92026. The data below sort of agrees with your observations with respect to atmospheric clearness.
FWIW, and FYI, PVWatts won't tell you what you generated today or any other day. It uses synthetic data for all weather and insolation. Was your reference to PVWatts a typo ?
PVWatts is a decent model, but using its modeled output for any day's actual output is useless.
PVoutput will probably get you a reasonable number for generation.
What's your array's orientation ? Got a lot of shade ? 31.353 kWh/day per 6.7 installed STC kW seems low compared to my output.

The last two days at my location:

Yesterday, 09/16/2020:
Temp. corrected daylong total measured GHI: 5.94 kWh/m^2
Average amb. air temp. at the array, 0800 to 1700 P.D.T.: 34.1 C. (93.4 F.)
Average amb. air temp. at ground level 0800 to 1700 P.D.T.:31.2 C. (88.2 F.)
Average wind velocity at the array: 1.8 m/sec (4.1 M.P.H.)
Total array output for the day: 27.85 kWh (5.32 kWh/installed STC kW)
Highest array measured cell temp: 63.3 C., 28.4 C. above amb. air temp. on roof. P.O.A. irradiance at that point was = 963 W/m^2
Modeled max. clear day output for my array using TMY temp. and wind data and clear sky irradiance model: 32.02 kWh/day (6.12 kWh/day per installed STC kW)

Today, 09/17/2020:
Temp. corrected daylong total measured GHI: 6.00 kWh/m^2
Average amb. air temp. at the array, 0800 to 1700 P.D.T.: 35.1 C (95.2 F)
Average amb. air temp. at ground level 0800 to 1700 P.D.T: 30.5 C (86.9 F)
Average wind velocity at the array: 1.5 m/sec (3.4 M.P.H.)
Total array output for the day: 28.21 kWh (5.39 kWh/installed STC kW)
Highest array measured cell temp.: 67.0 C, 30.9 C. above amb. air temp. on roof. P.O.A irradiance at that point was = 988 W/m^2
Modeled max. clear day output for my array using TMY temp. and wind data and clear sky irradiance model: 32.18 kWh/day (6.15 kWh/day per installed STC kW)

So today was a bit clearer (6.00 kWh/m^2 vs.5.94 kWh/m^2 GHI), and a deg. C or so warmer.

Measured by the array's recent output compared to clear sky output is improving. The worst day for output recently was 09/10 at 61 % of the max. possible value for production under a completely clear sky, with a more/less steady improvement to today's (09/17) value of 88% of the max. possible value for production for this date under a completely clear sky.

Depending on how far apart our locations are, my insolation data may/may not be useful to you.

[PugPower]

So you guys inspired me and I decided to do my own un-scientific experiment. I cleaned my panels last night after 7PM and decided to compare the output from yesterday to today. By cleaning I mean I hosed them off from a stepladder. Took all of 10 minutes. I had not cleaned them since last year and they looked pretty dusty. I have a 6.7kW system consisting of 1 Solar edge 6000HD inverter and 20 LG 335w panels. Yesterday the high temp according to my personal weather station was 98.6 F, little to no wind, mostly clear with few clouds, and a light haze. According to PVOutput I generated 31.353kW. Today the high was 102.0 F, little to no wind, mostly clear with few clouds, and light haze (I honestly could't tell if it was worse or better than the day before, maybe a little better.). Today according to PVOutputs I generated 33.054kWh. A difference of 5.4% on a day which was 3.4 degrees hotter and maybe slightly clearer than the day before. Basically this un-scientific experiment confirms that cleaning your panels will not result in a big increase in production and like JPM says a hosing off is probably all that is needed and/or worthwhile.

[J.P.M.]

Sure... but a 10mph breeze could increase PV output by almost 5% by cooling the panels. There's just A LOT of variables. I barely see a difference between clean and dirty let alone method of cleaning...

Regardless; assuming your numbers are accurate the other variables are frequency as you mentioned and oversizing. If you get a good rain right after you washed your panels then you didn't really get much return for your labor. And washing once a month means they're dirty at the end of the month so some production is already lost. You're not gaining a 10% increase in your annual production you're gaining ~10% the day after you clean them.

There's also oversized arrays. Instead of cleaning once a month I would just add 10% more panels. I've installed two arrays that are oversized by 50%. Between ~11 and ~3 most days it doesn't matter how dirty they are since the inverters are saturated and no amount of dirt that's able to stick to the panels seems able to change that

To be clear, I don't care if arrays get cleaned or not. My purpose here is to describe what I think I've learned or confirmed from what studied. The two seem to have produced similar results in terms of an array's output as f(inputs).

By my calcs, which tend to agree well with a lot of empirical heat transfer correlations as well as text book stuff, lead me to believe that in common PV situations, a 10 MPH (~ 4.47 m/sec.) boost in wind velocity, will increase the convective heat transfer coeff. from the array from ~ 35.0 w/m^2*deg. C. to ~ 59.1 W/m^2 deg. C.

That improvement will lower the array temp. However, in doing so, the (now lower) radiative heat transfer will be reduced by virtue of the panel temp. being closer to the surrounding temps. The net result will be a change in the array temp. - under common, summer, clear sky, midday conditions that will be less than the reduction in the convective heat transfer coeff. would lead one to believe. The actual drop in temp. will be something like ~ 5 to 6 deg. C. or so. So, using a bottom line ballpark for improvement in array output at ~ 0.5 % improvement per deg. C. drop in cell temp.: 0.005*5.5 = ~ 2.8 % improvement, not the 5 % you guessed at.
Doing a bit more involved stuff, my array's change in temp. per 1 m/sec. change in wind velocity is about 0.40 %/m/sec., or ~ 2% over 5.5 deg. C change in array temp. I believe my lower change/deg. C. may have something to do with the larger than normal array to roof clearance I have.

But unless I had a reliable source of wind blowing over the array, I'd not count on it too much. A clean array is something I have some control over - the wind, none and it's usually not blowing that hard in many locations.

As for frequency of cleaning, A decent 1/4" or so of rain seems to produce about the same effect on the cleanliness of my array as the simplified cleaning method I described of 5 minutes with a hose. If I lived in a place that got rain on a semi regular basis, I might never need to clean my array - depending on local sap/leaf/blossom conditions and the size of the local bird population and the bird's affinity for my array.

If my panels foul at 0.75 %/week and I use my method of simple hosing that removes ~ 2/3 of the fouling, I'll wind up with a more/less continuous fouling of 4.5 % after a few months. The math will change with the per week ave. fouling and cleaning schedule but the fouling penalty will level of dirt will level off. Your results will vary.

The choice between adding surface/capacity to an array vs. cleaning is an individual one.

IMO only and NOMB, I believe oversizing arrays is not the best way to implement sensible and cost effective alternate energy solutions any more than nuclear power is a sensible long term solution to meeting energy needs.

[azdave]

If I'm super bored I'll wash the panels but the production gain is lost in <2 weeks.

If I have to go up on the roof for other maintenance I'll wash them off while I am up there (soft water too) but the risk factor of a bad fall and a hospital bill could wipe out 10 or more years of savings in a heartbeat. I've washed mine 3 times now in 7 years.

[nwdiver]

As for margin/confidence in my numbers and methods, see the ret of my prior post if you interested.
As for the time/effort a simple hosing takes me under normal conditions (whatever that means/becomes in the future), that's why I will probably spend 5 minutes on my roof and ~ 15 to 20 gal. of H2O about 1X/month if it doesn't rain.

Sure... but a 10mph breeze could increase PV output by almost 5% by cooling the panels. There's just A LOT of variables. I barely see a difference between clean and dirty let alone method of cleaning...

Regardless; assuming your numbers are accurate the other variables are frequency as you mentioned and oversizing. If you get a good rain right after you washed your panels then you didn't really get much return for your labor. And washing once a month means they're dirty at the end of the month so some production is already lost. You're not gaining a 10% increase in your annual production you're gaining ~10% the day after you clean them.

There's also oversized arrays. Instead of cleaning once a month I would just add 10% more panels. I've installed two arrays that are oversized by 50%. Between ~11 and ~3 most days it doesn't matter how dirty they are since the inverters are saturated and no amount of dirt that's able to stick to the panels seems able to change that

[J.P.M.]

Ah... I get it now. So if production was down 10% he got back 6 - 8% not 10% and he's not losing 66 - 75% from dirt. ok. IMHO you're into 'margin of error' territory there.

That's why I just wait for rain. If I'm super bored I'll wash the panels but the production gain is lost in <2 weeks.

As for margin/confidence in my numbers and methods, see the ret of my prior post if you interested.
As for the time/effort a simple hosing takes me under normal conditions (whatever that means/becomes in the future), that's why I will probably spend 5 minutes on my roof and ~ 15 to 20 gal. of H2O about 1X/month if it doesn't rain.

[nwdiver]

That is not what he wrote. To say it another way, he got back 2/3 to 3/4 of the few percent of production he was down due to dirty panels. With a simply spray of a garden hose he recovered nearly all of the partial production that was lost to dirt on the glass. A more through scrubbing and rinse gets you back to the best output possible.

Ah... I get it now. So if production was down 10% he got back 6 - 8% not 10% and he's not losing 66 - 75% from dirt. ok. IMHO you're into 'margin of error' territory there.

That's why I just wait for rain. If I'm super bored I'll wash the panels but the production gain is lost in <2 weeks.

[J.P.M.]

Your output increased 3x by cleaning them? Were your panels covered in 3 layers of water soluble paint? I’ve cleaned panels so dirty they were more brown than blue and got <10% increase. And unless you have REALLY good water you need to squeegee the water off to avoid scale buildup.

I wish. No. Reread what I wrote. Note the part about restoring ~ 2/3 to 3/4 of the output that was lost to fouling.

A (recent) example may help make things clearer.

On 09/04/2020, after my summer session of ~ 65 daily cleanings for reasons loosely related to this topic, I gave my array what I thought would be one final cleaning until next May - and I mean a very thorough one, soaping 2X, rinsing 3X and squeeging. Details on why the 65 daily cleanings if you want.

Anyway, smoke/haze started for me on 09/06/20. I left the array alone for a few days but keeping an eye on input, output and array performance, on 09/11/2020 I noticed a lot of visible fouling and measured the fouling to have caused a performance deterioration of ~ 0.054 or ~ 5.4 % from what I believe were the clean conditions of 09/04/2020.

Under normal conditions, I expect a performance deterioration of ~ 0.75% to maybe 0.90% fouling penalty/week. A blinding flash of the obvious led me to conclude things are not normal.

The next morning, 09/12/2020, at 0800 P.D.T., I got up on the roof with my trusty hose and hosed off the array at a rate of ~ 3/4 gal./panel. No wiping/scrubbing/squeeging. Just a hosing.
On 09/12/2020 on the 9 minutes centered around the time of minimum incidence angle on the array for that day, I took measurements in my regular way and usual time that allow me to estimate array fouling.
The fouling rate I measured for that day was 0.012 or a bit more than 1 %.

So, the percentage improvement in lost output, or restoration of output that was lost due to fouling amounted to (0.054-0.012)/0.054 or ~ 78 %.

A comment on how accurate I think those fouling numbers might be: Having done this (measured array performance - both clean and fouled) several hundred times for going on 6+ years, I estimate the methods I use and the numbers I get for fouling are probably good to +/- 0.90 % or so at the 95 % confidence level. So, for this example, I would say my estimate of the dirty fouling to be between maybe 4.5 % and 6.3 %, and the post hosing fouling between maybe 0.3 % and 2.2 %.

A final comment: When I began all this array performance measurement/cleaning stuff in 2014, I compared before/after performance for my array cleaned in 4 different ways:
- Soaped, brushed, rinsed, wiped w/distilled water and then squeeged.
- Soaped/brushed/rinsed w/tap water but not wiped w/distilled water and not squeeged, just air drying.
- Not soaped but hosed w/ tap water, brushed and hosed again w/ tap water, no squeeging, and air drying only.

- Simple hosing from above the array at a rate of ~ 3 l/panel, with no other treatment. That method seems to restore ~ 2/3 to 3/4 of the array output that was lost to fouling as I wrote.

After doing each method at least 30 times (which took about 18 months or so) and measuring performance on most days when the sky was cloudless for at least an hour or so before the 9 minutes around the minute of minimum solar incidence angle for that day (which is the time when I take irradiance, weather and array measurements), I was unable to measure any significant difference in the level of post cleaning fouling attained by any of the first 3 methods described.

My conclusions:
- While there may be a difference in % of performance restoration after each of the first 3 thorough cleaning methods described above, the methods I use to measure array fouling and its effect on array output show little if any difference in clean performance. If there is a difference, which I don't doubt, I can't measure it and it's masked by the variance in my measurements. If I can't measure it, and given the accuracy of most every method I know of or have read about, I'm not worrying about it. Hell, even pyranometers kept in very tight calibration are +/- 1 % if you're lucky.
- Water spots left from not squeeging an array after cleaning make no difference in clean output using the methods I describe.
- If I ever stop fooling around w/my array, if I want to keep the array's fouling penalty below about 3 % or so average, the easiest way for me to do so will be to get up on the roof ~ 1 X/month if it doesn't rain and hose the 16 panels with about 4-5 minutes of simple tap water.

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

[azdave]

Your output increased 3x by cleaning them?

That is not what he wrote. To say it another way, he got back 2/3 to 3/4 of the few percent of production he was down due to dirty panels. With a simply spray of a garden hose he recovered nearly all of the partial production that was lost to dirt on the glass. A more through scrubbing and rinse gets you back to the best output possible.

[nwdiver]

For my array, I've got numbers that point to a 5 minute or less simple hosing restoring about 2/3 to 3/4 of the array output that was lost to fouling. If you or anyone else can get at their array with a hose, it might be worth considering.

Your output increased 3x by cleaning them? Were your panels covered in 3 layers of water soluble paint? I’ve cleaned panels so dirty they were more brown than blue and got <10% increase. And unless you have REALLY good water you need to squeegee the water off to avoid scale buildup.

[J.P.M.]

I live in a dusty semi-rural area. I cleaned my panels once, only noticed a nominal increase in production. Within a week or so my panels were dirty again. I have't bothered with cleaning again.I believe this was discussed in an earlier thread and the general consensus was that cleaning panels is not necessary unless they are REALLY dirty to the point where they are completely obstructed from sunlight.

For my array, I've got numbers that point to a 5 minute or less simple hosing restoring about 2/3 to 3/4 of the array output that was lost to fouling. If you or anyone else can get at their array with a hose, it might be worth considering.

[PugPower]

I live in a dusty semi-rural area. I cleaned my panels once, only noticed a nominal increase in production. Within a week or so my panels were dirty again. I have't bothered with cleaning again.I believe this was discussed in an earlier thread and the general consensus was that cleaning panels is not necessary unless they are REALLY dirty to the point where they are completely obstructed from sunlight.

[J.P.M.]

The best way to actually measure the difference would be on an array with individual panel monitoring, cleaning half of them only and looking at the panel production of both side at the same time.
From what I have read, people report for 2% to 10% depending how dirty the panels were.

A thin layer of dust probably does nor make a huge difference. A thick layer of ashes like we have had recently in CA and or lots of bird dropping probably creates a bigger difference.
It will all depend on the opacity of the layer.

Agreed. If I was of a mind, I'd engineer a PV fouling measuring gadget consisting of 2 PV panels, a pyranometer, a data logger and some wiring/ancillary equipment.

A workable alternative might be a pyranometer and a fixture that allows two glass lites' transmission to be compared, one constantly exposed to the atmosphere and not cleaned with the other cleaned before measuring how much sunlight gets transmitted through it.

As for the rate of panel fouling, it varies a lot by location and application. For my location and application only, after measuring a bunch of stuff for 6 years, including irradiance, and other weather variables w/a Davis Pro II + weather station, under normal conditions, without rain or cleaning, my array fouls at a rate such that the output drops somewhere between about 0.75% and 0.90%/week. I believe that rate may become asymptotic after 8 weeks or so, but I've not been rain free long enough to get data to confirm/refute. The max. fouling I've measured doing daily readings on days under clear skies is ~ 6+ % or so and the rate of fouling seemed to be leveling off the few times I had rainless periods that long. More anecdotal than provable. I suppose I could check windshields in junkyards with a handheld pyranometer.