Thanx again for the info.
Before I start, I'll say I'm becoming more to the opinion that dew, some of it heavy, from marine layer and cooler radiant night sky temps. has the effect of some "cleaning" of my array to a greater degree than I might have once thought or suspected, with that dew, which often amounts to some small but measureable precipitation (~ 0.01" - 0.02 " when heavy), while not reducing the fouling level too much, tending to hold the fouling penalty at what might be described as a non increasing level of, for the period from ~ 09/26 to the present at about 6 % or so.
For most of the summer, since 06/20 and after 70 continuous daily A.M. cleanings to get an estimate of array performance deterioration as I do 1X/yr., as well as an estimate of clean array performance from 36 clear sky instantaneous observation procedures as described here previously, my array's (uncleaned since 06/20) fouling tends to increase at a rate of ~ 1 % per week as before when it doesn't rain or if, as I seem to observe, there's no/not much overnight dew. With dew however, the fouling penalty, which has had single day values as high as 9 % or so, seems to have leveled off at an 11 day average value of ~ 5% - 6% or so, same as observed previously.
Doing a heat/energy balance on my array and using estimated sky temps., wind data, dew points., ambient temps. and some convective and radiant teat transfer correlations yields an estimate that condensation rates on my array (Both top/bottom as observed BTW) that dew can amount to an equivelant of something like ~ 0.01" - 0.03" of measureable precip. overnight. My Davis rain gauge on the roof has measured 0.01" - 0.02 " during such events.
With that said, if it was me, and it ain't, after looking at your data, I'd consider a couple of things.
1.) I'd do a linear least squares fit of the ratios from your first cleaning on/about 09/26 (?) to now. I think that might give some indication/clue of how the cleaned panel fouls in relation to an already dirty panel, and/or how performance penalty from dirt on an array might, in your (and my BTW) case tends to level off due to some cleaning of the type I seem to have observed on my array as described above. That is, your 1.1.7. panel started cleaner (higher ratio) and the ratio on the graphs seems to be decreasing as f(time) in what looks like a somewhat consistent if not linear downward trend, but maybe flattening out lately as a higher order fit might tend to show, or maybe not. One of the things that might be telling/showing/pointing us to is that dirt on an array can, and in this case does appear to be somewhat transitory with the elements, with some of the pre 09/26 dirt being somewhat washed away by dew/precip. from the non cleaned 1.1.10 panel, while both panels were equally cleaned and fouled by weather since 09/26. I guess that may be saying no more than something like 12 months from now, the ratio will be zero (DUH !?), but the slope of that least squares fit and/or the shape of higher order fits may have some fouling rate information buried in them.
2.) I think I understand the logic of reversing the cleaned vs, dirty panel. But, given all the other variables, and the problems of separating small fouling differences from the noise of about the same level as the date differences, as well as not seeing a lot to be gained by doing so, I'd skip that part.
3.) FWIW, and as much of a PITA as it might be, I'd consider cleaning the array on a regular basis (1X/month ?) with the dirty panel remaining uncleaned (covered ?) during the cleaning process. That way, you'll have the data of an uncleaned panel that's continuously exposed only to added fouling (and cleaning) from the environment, for comparison to a panel and an array that's cleaned on a regular basis. If one goal of this exercise is to see how/if/what regular array cleaning will influence array performance, that process will probably yield good and useful information.
Are we having fun yet ?
Before I start, I'll say I'm becoming more to the opinion that dew, some of it heavy, from marine layer and cooler radiant night sky temps. has the effect of some "cleaning" of my array to a greater degree than I might have once thought or suspected, with that dew, which often amounts to some small but measureable precipitation (~ 0.01" - 0.02 " when heavy), while not reducing the fouling level too much, tending to hold the fouling penalty at what might be described as a non increasing level of, for the period from ~ 09/26 to the present at about 6 % or so.
For most of the summer, since 06/20 and after 70 continuous daily A.M. cleanings to get an estimate of array performance deterioration as I do 1X/yr., as well as an estimate of clean array performance from 36 clear sky instantaneous observation procedures as described here previously, my array's (uncleaned since 06/20) fouling tends to increase at a rate of ~ 1 % per week as before when it doesn't rain or if, as I seem to observe, there's no/not much overnight dew. With dew however, the fouling penalty, which has had single day values as high as 9 % or so, seems to have leveled off at an 11 day average value of ~ 5% - 6% or so, same as observed previously.
Doing a heat/energy balance on my array and using estimated sky temps., wind data, dew points., ambient temps. and some convective and radiant teat transfer correlations yields an estimate that condensation rates on my array (Both top/bottom as observed BTW) that dew can amount to an equivelant of something like ~ 0.01" - 0.03" of measureable precip. overnight. My Davis rain gauge on the roof has measured 0.01" - 0.02 " during such events.
With that said, if it was me, and it ain't, after looking at your data, I'd consider a couple of things.
1.) I'd do a linear least squares fit of the ratios from your first cleaning on/about 09/26 (?) to now. I think that might give some indication/clue of how the cleaned panel fouls in relation to an already dirty panel, and/or how performance penalty from dirt on an array might, in your (and my BTW) case tends to level off due to some cleaning of the type I seem to have observed on my array as described above. That is, your 1.1.7. panel started cleaner (higher ratio) and the ratio on the graphs seems to be decreasing as f(time) in what looks like a somewhat consistent if not linear downward trend, but maybe flattening out lately as a higher order fit might tend to show, or maybe not. One of the things that might be telling/showing/pointing us to is that dirt on an array can, and in this case does appear to be somewhat transitory with the elements, with some of the pre 09/26 dirt being somewhat washed away by dew/precip. from the non cleaned 1.1.10 panel, while both panels were equally cleaned and fouled by weather since 09/26. I guess that may be saying no more than something like 12 months from now, the ratio will be zero (DUH !?), but the slope of that least squares fit and/or the shape of higher order fits may have some fouling rate information buried in them.
2.) I think I understand the logic of reversing the cleaned vs, dirty panel. But, given all the other variables, and the problems of separating small fouling differences from the noise of about the same level as the date differences, as well as not seeing a lot to be gained by doing so, I'd skip that part.
3.) FWIW, and as much of a PITA as it might be, I'd consider cleaning the array on a regular basis (1X/month ?) with the dirty panel remaining uncleaned (covered ?) during the cleaning process. That way, you'll have the data of an uncleaned panel that's continuously exposed only to added fouling (and cleaning) from the environment, for comparison to a panel and an array that's cleaned on a regular basis. If one goal of this exercise is to see how/if/what regular array cleaning will influence array performance, that process will probably yield good and useful information.
Are we having fun yet ?
Comment