Announcement

Collapse
No announcement yet.

Underestimated Results of Cleaning Panels

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #61
    Originally posted by NukeEngineer View Post
    I want to play along....

    I've found two panels that are right next to each other and their Power charts seem to be identical. 1.2.7 and 1.2.8

    https://monitoringpublic.solaredge.c...lic#/dashboard

    At 3:45 eastern, I cleaned one of these two with a dry mop.

    We haven't had any rain in about 2 weeks, and there's a very light coating of dust, it appears, on the panels. I'm way away from roads, down a dirt and gravel road, so the dust from that has probably collected on my panels.


    See if you can figure out which one I cleaned!

    I'm attaching a photo of these two panels before the cleaning.

    EDIT: Crap, the public view site won't allow charting. I'll post the chart from these two after they get shaded this evening.
    Well the chart of the two, before and after cleaning, are completely on top of each other, even zoomed way in. So, not very conclusive.
    https://pvoutput.org/list.jsp?sid=54099

    Comment


    • #62
      Originally posted by NukeEngineer View Post

      Well the chart of the two, before and after cleaning, are completely on top of each other, even zoomed way in. So, not very conclusive.
      That is conclusive, just not the conclusion you were looking for
      CS6P-260P/SE3000 - http://tiny.cc/ed5ozx

      Comment


      • #63
        I would NOT use a dry mop to clean the panels.
        You won't see the micro scratches for the first year or so, but eventually the effect will show as frosted glass on the panels, and that wont be good. Like running the wipers on your windshield without water, you get scratches. And it can't be good for the AR coating..
        Powerfab top of pole PV mount (2) | Listeroid 6/1 w/st5 gen head | XW6048 inverter/chgr | Iota 48V/15A charger | Morningstar 60A MPPT | 48V, 800A NiFe Battery (in series)| 15, Evergreen 205w "12V" PV array on pole | Midnight ePanel | Grundfos 10 SO5-9 with 3 wire Franklin Electric motor (1/2hp 240V 1ph ) on a timer for 3 hr noontime run - Runs off PV ||
        || Midnight Classic 200 | 10, Evergreen 200w in a 160VOC array ||
        || VEC1093 12V Charger | Maha C401 aa/aaa Charger | SureSine | Sunsaver MPPT 15A

        solar: http://tinyurl.com/LMR-Solar
        gen: http://tinyurl.com/LMR-Lister

        Comment


        • #64
          Originally posted by ButchDeal View Post

          just lower your blades way down.... they will stop complaining about mowing infrequently enough and less need for watering...

          I am pretty sure I have never watered my lawn
          Even calling it a lawn is generous, it is really just a patch of grass that I use a weed whacker to cut. I keep it raked and weed-free, but like it a little bit longer to hide the spots that get burned by dogs. Oh well.

          I'm still working on re-doing this using integrated power instead of energy, but we had some drizzle Saturday morning (9/30) that has also affected the correlation enough to be worth posting here.

          (It looks like I can only download two days worth of data at a time and preserve the original optimizer power data+timestamps, so it is slow to just pull the data, let alone analyze it. I'm expanding the analysis though to look at the correlations of *all* panels to each other, which I think might help me answer some other questions I've been curious about with this.)



          DailyComparison2.JPG
          CS6P-260P/SE3000 - http://tiny.cc/ed5ozx

          Comment


          • #65
            Originally posted by sensij View Post
            I'm still working on re-doing this using integrated power instead of energy,
            Since I will bank on the concept that one definition of 'energy' is 'integrated power' I presume that this breaks down to different sources of the energy/power numbers, and/or to the reporting/logging functions available. Can you elaborate a bit? Is it just a matter of better resolution, or do you feel that one method will be inherently more accurate than the other? Bear in mind that I 'think' I know that enphase data is aggregated at an enphase cloud site, then passed to PVOutput, each of those having their own particular way of presenting that data, some or all of which might not even be available locally to the owner of the system - and nothing really more than that. On another tack, when I read of some people not having the right password to access their own data the way they want to it leaves me bewildered.

            Comment


            • #66
              Originally posted by AzRoute66 View Post
              Since I will bank on the concept that one definition of 'energy' is 'integrated power' I presume that this breaks down to different sources of the energy/power numbers, and/or to the reporting/logging functions available. Can you elaborate a bit? Is it just a matter of better resolution, or do you feel that one method will be inherently more accurate than the other? Bear in mind that I 'think' I know that enphase data is aggregated at an enphase cloud site, then passed to PVOutput, each of those having their own particular way of presenting that data, some or all of which might not even be available locally to the owner of the system - and nothing really more than that. On another tack, when I read of some people not having the right password to access their own data the way they want to it leaves me bewildered.
              Yeah, it's crazy alright.I wouldn't have access to the underlying data if I hadn't stayed persistent with my installer. I finally got access to the SolarEdge charting and exports but, IMHO, it still sucks. Case in point is how sensij is having to get his data out of the cloud.

              I understand why SE thinks it should be kept in the cloud as the data would accumulate rapidly. ~8 possible data points for 30 panels @ 6 samples per hour. I would still like to get easy access to the data and keep it locally though. I think sensij would too.

              Comment


              • #67
                Originally posted by AzRoute66 View Post
                Since I will bank on the concept that one definition of 'energy' is 'integrated power' I presume that this breaks down to different sources of the energy/power numbers, and/or to the reporting/logging functions available. Can you elaborate a bit? Is it just a matter of better resolution, or do you feel that one method will be inherently more accurate than the other?
                Yes, as far as I can tell, SolarEdge either measures or derives its Energy reports from something other than the data it reports in its Power measurements.

                For the *inverter*, I've been able to show that the integrating the power gives an energy generated number much closer to what a revenue grade meter will report. Instantaneous power is reported at 5 min intervals, so as you might expect, when conditions change more rapidly than that, some data isn't captured that impairs the accuracy of even that method. SolarEdge's inverter Energy report is consistently overstates production by a few percent (relative to a revenue grade meter), and appears to just be an intrinsically less accurate report.

                This is the first time I'm diving into Optimizer data this deeply, but I guess it isn't surprising to see the same behaviors. As you can see in this post, the energy data shows mid-day divergence between the two panels charted while the power data do not. We also had another thread recently where the OP's optimizer data showed truly erroneous energy data but accurate power data, so it really looks like those two numbers are determined through independent systems in their reports. Since the correlations between panels using the optimizer power data are much better than using the energy data, I'm assuming the power data are more accurate (as it is with the inverter), but haven't attempted to make calibrated measurements to confirm one way or another.

                I don't really know what the optimizer is reporting... it might be just voltage and current and letting the inverter calculate the rest, but I think it probably sends power and energy as well and the inverter just acts as the transmitter to the SolarEdge portal where the data are logged.


                I can try to wade through some of this more deeply if you'd like, but once I get my analysis programming further along some of the differences between power and energy accuracy will become more apparent, I think.




                CS6P-260P/SE3000 - http://tiny.cc/ed5ozx

                Comment


                • #68
                  Originally posted by sensij View Post

                  Even calling it a lawn is generous, it is really just a patch of grass that I use a weed whacker to cut. I keep it raked and weed-free, but like it a little bit longer to hide the spots that get burned by dogs. Oh well.

                  I'm still working on re-doing this using integrated power instead of energy, but we had some drizzle Saturday morning (9/30) that has also affected the correlation enough to be worth posting here.

                  (It looks like I can only download two days worth of data at a time and preserve the original optimizer power data+timestamps, so it is slow to just pull the data, let alone analyze it. I'm expanding the analysis though to look at the correlations of *all* panels to each other, which I think might help me answer some other questions I've been curious about with this.)

                  DailyComparison2.JPG
                  Suggest you take dog off diuretics, or (only slightly tongue in cheek), consider artificial turf.

                  Other: Comparing all/more/most of the panels may be worthwhile, but may not show much more than two reasonably matched panels will show besides maybe wind patterns and differential fouling as (panel location). If my gut est./feeling of any fouling measurement being probably no better than +/- 0.75 % to 1.0 % or so has any validity, that ~~ 3% variation as shown for the last few days is perhaps partly due to panel temp. differences from some differential fouling that can occur for lots of reasons.

                  If it's any use, I've estimated my array's average heat loss to be ~ 32 - 33 W/m^ per deg. C. temp. diff., panel to ambient air temp., with some est. of variation at ~~ +5 to + 6 W/m^2 *deg. C. (considering 2 sides, top and bottom) for wind velocity after adjustments for irradiance.

                  My ave. array temp. as measured lots of times seems surprisingly close to the temp. the Sandia method estimates, if with a somewhat wider data spread, using "a" = -3.56, "b" = -0.075 and the cell temp. < 3 C. higher than the back of the panel temp. when adjusted for P.O.A. as suggested by the Sandia method. From the Recent 36 clean trials:

                  Measured temp diff.: mean panel - amb. T. = 28.5 C.
                  min. delta T. = 23.9 C.
                  max. delta T. = 31.6 C.
                  std. dev. = 1.64
                  Sandia method estimated: panel - amb. = 28.3 C.
                  min. delta T. = 26.7 C.
                  max. delta T. = 30.1 C.
                  std. dev. = 0.88.

                  N == 36

                  I farted around with Sandia coefficients, particularly the "b", that is the wind coeff., but decided to stay with suggested coeff. from the method for the panel cofig. and take my lumps. Still, IMO, the agreement at least seems reasonable, particularly keeping in mind the panel temp. differences I've measured over the array at anywhere from ~ 0 to ~ 4 or 5 C., or so as f(wind vector) between panels at the leading to trailing edge of the array.
                  Last edited by J.P.M.; 10-02-2017, 03:26 PM.

                  Comment


                  • #69
                    Originally posted by J.P.M. View Post

                    Other: Comparing all/more/most of the panels may be worthwhile, but may not show much more than two reasonably matched panels will show besides maybe wind patterns and differential fouling as (panel location). If my gut est./feeling of any fouling measurement being probably no better than +/- 0.75 % to 1.0 % or so has any validity, that ~~ 3% variation as shown for the last few days is perhaps partly due to panel temp. differences from some differential fouling that can occur for lots of reasons..
                    Since I can see how much of the variation in power is due to current vs voltage, it is clear the difference I'm looking at here is not temperature driven.

                    A full 28x28 correlation matrix will do a few things for me, but the first question I want to attempt to answer is the thermal effect of the optimizers on the panel output.
                    CS6P-260P/SE3000 - http://tiny.cc/ed5ozx

                    Comment


                    • #70
                      I take it a treatise by SolarEdge on "The Origin of Numbers" with enough detail is not available and that is essentially what you are looking to reverse engineer, which leads me to believe that people have asked them and they [SolarEdge] have not provided a suitable answer. Not that I think they have to - how many system owners are using them as a laboratory to investigate soiling or optimizer mounting practices or 'active bypass diodes' or ______?

                      Comment


                      • #71
                        Originally posted by AzRoute66 View Post
                        I take it a treatise by SolarEdge on "The Origin of Numbers" with enough detail is not available...
                        This is what I got from SolarEdge technical support a couple of years ago when I pushed on it.

                        I'm sorry but the best answer I can find for you is just in regards to measurement accuracy of the devices used to measure current and voltage. Individually they have a tolerance of accuracy of 5%, but it would seem based on your calculation that due to our manufacturing practices that tolerance is actually quite a bit smaller, though inaccurate (at less than the tolerance) to other sensors within the same equipment which perform similar if slightly different functions (in this case calculation of E vs. P). I hope that helps explain the difference.
                        If I take the response literally, it suggests different sensors are user to perform energy and power calculations. For whatever reason, their inverter (and now optimizer) power numbers hold up under scrutiny much better than the energy numbers, but most of the portal is built around conveying energy information.
                        CS6P-260P/SE3000 - http://tiny.cc/ed5ozx

                        Comment


                        • #72
                          You got more than I would have expected. Like I said, they don't seem to be in the 'panel laboratory' business. Now, this whole business of having 'homeowner', 'installer', 'admin', and/or other equipment accounts, for which the owner of the equipment gets least access/visibility seems a bit awkward. Probably got that notion from the implementation of OBD II ports on automobiles... becoming available now, big 'dealer' secret not so long ago.

                          Comment


                          • #73
                            Originally posted by sensij View Post

                            Since I can see how much of the variation in power is due to current vs voltage, it is clear the difference I'm looking at here is not temperature driven.

                            A full 28x28 correlation matrix will do a few things for me, but the first question I want to attempt to answer is the thermal effect of the optimizers on the panel output.
                            I believe I understand what you are writing/saying. Are you looking at how does panel temp. vary over the panel, and what factors may be influencing that pattern ? And then, how might those variations be different from that of a panel that had a constant, and isotropic temp. distribution ? And, what effects might those variations have on panel power generation and energy production ?

                            I don't have much that's definitive to answer such questions, but I'm not sure such answers exist, given the variety of arrangements and panel characteristics available.

                            As for power output being influenced by panel/cell temp., well, in one or several ways, it is, along with the primary mechanism of the energy input, but in ways that cannot always be separated from other causes.

                            I wasn't explaining myself very well in that prior post. From what I've measured, temps. over the same panel can vary quite a bit, as well as even more so over an entire array of more/less identical panels with sequential serial numbers. I've found that in addition to the local panel temp. variations of usually ~ 1 C. to maybe ~ 2 C.- 3 C. over the rear face of one panel, the region on the front face of the panel directly over the junction boxes affixed to the back my panels by adhesive, and not attached to the frame, will be ~, say, maybe 5 C. to 7 C. warmer than the immediately surrounding panel surface, and a bit less so between the edge of the box and the edge of the panel frame - a distance of an inch or two. Separate from that, I suspect most of the temp. variation over the rest of an entire panel has several reasons, two of which might be cell mfg./wiring differences, and wind patterns over the array, but I can't verify that first speculation. Nor do I have any data that might relate to any temp. differences over a panel as a result of the bath tub ring effect. Maybe some day.

                            I've made the assumption that while the bathtub ring on a dirty panel will change the temp. profile over the entire panel, the fouling level and pattern on my panels seems to be independent of panel location within the array.

                            As for a fouling or dirt "shadow", over the junction boxes on my array, it is present, but not extremely obvious, being much less obvious that the bathtub ring. I'd describe it more as dirt of a different nature rather than more/less dirt, with maybe more of a finer patina of smoother sediment than say, dust that's merely decided to settle there. As I speculated, that difference may be due to the higher panel temp. over the box, and that higher temps. effect on increasing the relative rate of evaporation of what's mostly condensate (dew). Without being able to measure any difference in transmittance between the dirt over a junction box and the rest of a panel, any differential effects caused by the seeming and perhaps apparent difference in the fouling over the boxes will be unknown and speculation on my part anyway. I am of the opinion however, that the increased temp. over the boxes as I've measured is caused mostly by the decrease in available heat transfer surface caused by the insulating effects of the box being glued to the back surface of the panel.

                            In a larger sense, for you and others w/the ability to do what amounts to close side/side clean/dirty output comparisons with virtually identical panels, I'd suggest that panel local temp. distribution and profiles will tend to be similar, one panel to the next, and thus tend to cancel out, particularly for panels close to one another. However, and somewhat/mostly separate from the effects of fouling, the question of how, and in what ways, local panel temp. variation can effect performance remains, if becoming a bit more academic and less than a question carrying a burning practical importance.

                            To that (and probably more academic) end, looking at the effects of panel fouling from an energy balance standpoint, a single panel in an array will absorb most of the incident solar energy that hits it. All that is, except for that portion of the incident solar energy that is reflected by the glazing, and also reflected, or absorbed on the surface by any dirt on the array in some convoluted combination that further reduces the amount of energy that hits a cell to and making electricity.

                            Simplifying : If a panel is operating in a quasi steady state of constant environmental conditions, including reasonably constant irradiance over a period that's, say maybe 5 or 10 panel thermal time constants ( ~ 45 - 60 minutes or so), with no energy storage within the panel, all the energy that gets past the glazing reflection losses and (any) dirt, and gets absorbed by the panel cells will either be used to make electricity or be rejected as waste heat.

                            The energy not turned into electricity (the waste heat) will cause the panel temp. to rise above the temp. of the surroundings. Assuming a simplification of steady wind vector as part of the steady state requirement, the temp. of the panel will increase as f(irradiance) until the panel reaches a temp. such that the waste heat that's rejected will be equal to the product of : (the temp. differ. between the panel and the ambient air and surroundings) * (the heat transfer coeff. between the panel and the surroundings). For most 1st order analyses, that heat transfer coeff. can be treated as a constant.

                            Now, of the energy that gets headed off by the dirt before it gets to a cell, all of it will either be reflected by the dirt, or absorbed by the dirt, the proportions of each being f(nature of the dirt). Any of the additional dirt-reflected energy, being unable to get into the panel will lost and not contribute to panel heating. What gets absorbed by the dirt will be transferred to the panel via thermal conduction. Any absorbed energy will serve to increase the panel temps. while, along with the dirt reflected energy, decreasing energy available for electricity production.

                            Somewhat incorrectly disregarding any wavelength dependence so as not to muddy the waters too much, and as a simple example, if black soot and talcum powder block equal amounts of irradiance, the soot will probably tend to increase temps. more than say, talcum powder. All this is ignoring any interactions the dirt may have with the surface characteristics of the glazing, which may/may not be of importance.

                            Given the variable nature of dirt (yours is probably similar to mine, less grass clippings), while I suspect the overall effects are to increase the panel temp. and decrease electricity production - a blinding flash of the obvious maybe, I bet how much each effect, absorbance or reflectance, will have will vary with the nature of the dirt.

                            Moreover, and to perhaps repeat, the dirt will effect the energy output of a panel in a double whammy: Decreasing input to the cells and increasing panel temp. In that sense, and maybe as a secondary effect, production will be affected by the effects that fouling will have on temp.

                            From measurements I've done, much of it descibed previously, while I believe my data and methods are good and defendable within the instrument and practical limits I'm under, I can't claim I've been able to confirm any temp. effects due to fouling in a quantitative way. I see some differences in average panel temps. for several series of measurements that pretty consistently show higher ave. panel to amb. temp. differences for dirty panels compared to the sets of clean panel data, but the teemp. differences, clean to dirty are, for the most part, less than the std. deviations in the data, and more importantly inseparable from changes in irradiance levels. That's what happens when one tries to measure differences small enough to get buried in the data noise.

                            Comment


                            • #74
                              Originally posted by J.P.M. View Post

                              Given the variable nature of dirt (yours is probably similar to mine, less grass clippings), while I suspect the overall effects are to increase the panel temp. and decrease electricity production - a blinding flash of the obvious maybe, I bet how much each effect, absorbance or reflectance, will have will vary with the nature of the dirt.

                              Moreover, and to perhaps repeat, the dirt will effect the energy output of a panel in a double whammy: Decreasing input to the cells and increasing panel temp. In that sense, and maybe as a secondary effect, production will be affected by the effects that fouling will have on temp.
                              Because of the differences in time stamps, it take more work to present this in a statistically valid way. What I'm seeing graphically is that the difference in power output that was introduced between these two selected panels after cleaning one of them can be fully explained by the change in current. I'm not trying to ignore what you've written here, just making sure we are operating from the same observation.

                              By pure luck, in all the days of operation up to 9/26/17, these two panels are indistinguishable from each other by the reported power, voltage, and current. After the cleaning test the night of 9/26, the only significant response shows up in current, which affects power with any additional contribution from a response in voltage.

                              A simple hypothesis is that by cleaning the panel, transmitted irradiance increases, and therefore current and power both increase. For voltage to be unaffected, perhaps the dirt on the dirty panel is absorbing most of the irradiance that is being transmitted in the clean panel, such that the temperature of panel + dirt is approximately equal to the temperature of the clean panel.

                              Edit: Let's say the fouling blocks 100 units of POA irradiance, On the clean panel, 18 of those units might be converted to electricity, 4 reflected, and the remaining 78 absorbed and converted to heat. On the dirty panel, if the dirt reflects 22 units while the other 78 are absorbed and converted to heat, we end up at about the same temperature, with the clean panel producing more current. Is dust 22% reflective? It doesn't seem unbelievable, especially against the typical albedo numbers presented in this Wikipedia entry. (and yes, ignoring wavelength dependence in this simple example)
                              albedo.JPG


                              Reconciling "panel temperature" as expressed by mppt voltage with the temperature distribution within the panel that you've proven to exist is a more complicated electrical model than I've attempted to compute.
                              Electrical data.JPG

                              Last edited by sensij; 10-03-2017, 05:08 PM.
                              CS6P-260P/SE3000 - http://tiny.cc/ed5ozx

                              Comment


                              • #75
                                Originally posted by sensij View Post

                                Because of the differences in time stamps, it take more work to present this in a statistically valid way. What I'm seeing graphically is that the difference in power output that was introduced between these two selected panels after cleaning one of them can be fully explained by the change in current. I'm not trying to ignore what you've written here, just making sure we are operating from the same observation.

                                By pure luck, in all the days of operation up to 9/26/17, these two panels are indistinguishable from each other by the reported power, voltage, and current. After the cleaning test the night of 9/26, the only significant response shows up in current, which affects power with any additional contribution from a response in voltage.

                                A simple hypothesis is that by cleaning the panel, transmitted irradiance increases, and therefore current and power both increase. For voltage to be unaffected, perhaps the dirt on the dirty panel is absorbing most of the irradiance that is being transmitted in the clean panel, such that the temperature of panel + dirt is approximately equal to the temperature of the clean panel.

                                Reconciling "panel temperature" as expressed by mppt voltage with the temperature distribution within the panel that you've proven to exist is a more complicated electrical model than I've attempted to compute.


                                Electrical data.JPG
                                And I can pretty much agree with your finding that instantaneous power is strongly correlated with current, which is pretty much directly proportional to P.O.A irradiance. To that end, I've stated in the past that if, under clear skies and steady state conditions near the time on min. incidence angle, it's a pretty to very good first approx. to ratio instantaneous current to a panels rated current to get a good approx. of the actual P.O.A. irradiance, particularly if that current is corrected for the estimated or measured panel temp. Then, it gets real good to the point where stuff I've measured agrees very well with HDKR P.O.A. irradiance, turning an array into a pyranometer, provided the current's measurement is from a clean panel or array.

                                On a dirty panel, what I'm suggesting is that the current will drop approx. in proportion to the irradiance decrease, and thus power, and further, the voltage will change, maybe up,, maybe down, maybe not at all, but in some fashion that's dependent on the optical characteristics of the dirt. For example, a "light" color fouling layer may actually lower panel temp. a bit over a clean panel while decreasing the irradiance. That will change the location of the Voltage - current curve, sending it to the right a little bit and probably changing the current a bit, as can be mostly estimated by the small current temp. adjustment. I'm also suggesting that the opposite effect may be seen with a "dark" colored fouling layer, with a "clean" panel obviously not being subject to temp. effects of dirt.

                                The way I measure fouling is to ratio the actual array output (on both sides of the inverter) taken off the inverter readout to the theoretical output of a clean array under the same conditions of P.O.A, irradiance, and weather conditions. That's the reason I need the panel temp. estimates. After 2 set of measurements, done in 2015 that I've described previously, 1 in winter, 1 in summer, 60 trials in all, I believe I have and now use, a reasonably reliable estimate of the Voltage /deg. C. change in my array's Voltage that agrees quite well with the Sunpower spec sheet value, and very consistently with the temp. measurements I took as simultaneously and as symmetrically as possible, especially when conductor resistance/deg. C. is considered . Unfortunately or otherwise, the inverter readout (not the monitor values which are 5 minute averages) are only good to 2 significant figures, 5.5 A for example. That means I'm unable to estimate how much of the fouling penalty of, say, 5 % for example, is due to irradiance reduction, and how much of that 5 % power output decrease is due to Voltage decreases (or increases) caused by temp. effects of the fouling layer. I suspect, and probably concurring with you, that the fouling effect on my array's Voltage is probably small with respect to the current decrease effects, but I've no way at this time of guessing the optical properties of a fouling layer.

                                Bottom line: I believe current is affected more by fouling than voltage, and that affects output in a mostly linear fashion, but Voltage is affected by the absorbance of solar energy by the fouling layer and that may increase or decrease the array temp. and thus array Voltage.

                                Comment

                                Working...
                                X