Announcement

Collapse
No announcement yet.

Shade impact question

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

  • Shade impact question

    Hi there,
    In the process of getting quotes for a 5.5kW south facing array, I had (prematurely) assumed I did not have shading issues (we have had rain for two month), but with the sun being back, I realize the 3 x 50 foot tall skinny palm tree on my neighbor's yard east of me will shade the array in the first 1.5 to 2.5 hours of the morning.
    Here is the model of the Palms at time of worst impact: https://1drv.ms/u/s!At3vMAQjaOZLkRL8QLv7HiDE7KYQ

    I modelled my house and trees in Sketchup, so I geo localized the model and ran shade simulation for the various days of the year in these early hours to see how bad it will be (controlling that the shadows it gives me for the current date actually matches what I see on my roof).
    I positioned the array in a portrait orientation as it seems to allow the array to clear the shade April first. It also minimizes the time during which the shade hits the array. Depending on fire code I may need larger setbacks...

    The result can be seen with pictures in the attached file
    https://1drv.ms/w/s!At3vMAQjaOZLkRaTmepm9Iucgzuw
    And can be summarized as follow
    1. 5.5 month with no shadows (Summer Apr-1 to Sep-10)
    2. 4 month with 2 to 2.5 hours of shade in the early morning on 2/3 of the array coming from the bushy top of the palms. (Sep-10 to Nov-10 & Feb-1 to Apr-1)
    3. 2.5 month with 1.5 hours of partial shade in the early winter hours (low production), coming only from the tree trunks. (Winter Nov-10 to Feb-1)
    I am reviewing two types of configurations, one with Power Optimizers, and one without.
    1. With POs, I am I right to assume the loss would be pretty much directly proportional to the shaded surface
    2. But If I was to use a solution that does not have Power Optimizers, just a few strings (like 3), roughly how much of the array production would I loose in period B and C.
      1. In B, am I right to assume almost all of it
      2. In C where the tree trunks seem to shade about 1/3 of individual panel, would I only lose 1/3 of the whole array?
    Thank you for any feedback

    Last edited by scrambler; 03-13-2019, 07:41 PM.

  • #2
    Originally posted by scrambler View Post
    ...
    ..
    ... I am reviewing two types of configurations, one with Power Optimizers, and one without.
    1. With POs, I am I right to assume the loss would be pretty much directly proportional to the shaded surface
    2. But If I was to use a solution that does not have Power Optimizers, just a few strings (like 3), roughly how much of the array production would I loose in period B and C.
      1. In B, am I right to assume almost all of it
      2. In C where the tree trunks seem to shade about 1/3 of individual panel, would I only lose 1/3 of the whole array?
    Thank you for any feedback
    1. Yes, although the loss on any given panel may range from 1/3 of its output to its entire output. The reason is that shading of one cell will kill the output of all the cells that are in that bypass diode group. So total shade on exactly the wrong set of three cells can reduce the output close to zero even with power optimizers.
    2. Not sure about B and C, it would again depend on exactly what portion of each panel is shaded, and the depends on the internal wiring structure of the panel.
    3. If it is the right (lucky) third of each panel you might lose only 1/3 of the total output. If it is the wrong 1/3 of the cells, you could lose the whole output.
    SunnyBoy 3000 US, 18 BP Solar 175B panels.

    Comment


    • #3
      Thanks,
      Is there a standard in the way the bypass diode are organized, like each row or each column of cell?

      Comment


      • #4
        Typically they cover two vertical column's of cells in a 6 column panel, 20 vertical cells in a 60 cell panel and 24 in a 72 cell panel. If panels are in landscape orientation shadows moving across the bottom third of the panel will affect I/3 of its output. In portrait orientation , all 3 banks of cells can be affected by the same shadow. That being said, there are other diode configurations in some other, non typical panels.
        2.2kw Suntech mono, Classic 200, NEW Trace SW4024

        Comment


        • #5
          Thanks.
          As long as it is either vertical or horizontal, that should be good, as the trunk shadows will always be diagonal.

          Comment


          • #6
            Originally posted by scrambler View Post
            Thanks.
            As long as it is either vertical or horizontal, that should be good, as the trunk shadows will always be diagonal.
            Good? A diagonal shadow would cause all 3 of the banks to be affected. Zero output from that panel, except at the very beginning and end in which only one third of the output would be reduced.
            9 kW solar, 42kWh LFP storage. EV owner since 2012

            Comment


            • #7
              I see what you mean, I guess it was wishful thinking on my part...

              It makes sense, as soon as one cell is shaded in the panel, the whole bypass diode row/column is dead, similar to one panel in a string taking the whole string down.
              This means a panel will pretty much be dead as long as the trunk cast their shadows across its diagonal.
              Now I can go and estimate the average amount of dead panels in each shaded period, and estimate the number of dead strings if no POs, then compute the power reduction using the hourly PVWatts data.

              Thanks again for the clarification.

              Comment


              • #8
                Originally posted by scrambler View Post
                I see what you mean, I guess it was wishful thinking on my part...

                It makes sense, as soon as one cell is shaded in the panel, the whole bypass diode row/column is dead, similar to one panel in a string taking the whole string down.
                This means a panel will pretty much be dead as long as the trunk cast their shadows across its diagonal.
                Now I can go and estimate the average amount of dead panels in each shaded period, and estimate the number of dead strings if no POs, then compute the power reduction using the hourly PVWatts data.

                Thanks again for the clarification.
                Almost, but not quite....
                Because of the bypass diodes one panel will not necessarily take down a string, but the voltage of that string will be reduced by the Vmp of that panel plus a volt or two for the forward drop in the bypass diodes.
                If the affected string is connected in parallel with an unshaded string to a single MPPT input, the results will depend on the number of panels in the string. If the loss of one or more panels drops that string's Voc below the operating voltage threshold of the MPPT input, then you will lose the total contribution of that string regardless of how it is wired.
                SunnyBoy 3000 US, 18 BP Solar 175B panels.

                Comment


                • #9
                  Originally posted by inetdog View Post
                  Because of the bypass diodes one panel will not necessarily take down a string,
                  Thanks for adding to my confusion

                  Are you saying that in the case of an installation like Pika Energy with a Hybrid string inverter and PV Links/MPPT on each string, there would also be bypass diodes between panels?

                  EDIT: Ignore question above, it was a dumb one on my part...
                  Last edited by scrambler; 03-14-2019, 04:48 PM.

                  Comment


                  • #10
                    As each segment of a panel is shaded and it's voltage drops, with enough segments shaded, the array will eventually reach a low enough voltage that the MPPT will shut down.
                    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


                    • #11

                      Just FYI in case it helps someone, below is the video simulation of the first 2 to 3.5 hours of morning exposure, weekly, during the period where shade occurs Sept-10 to Apr-1st.
                      https://1drv.ms/v/s!At3vMAQjaOZLkRiPWJEiGDDEfwH4

                      I guestimated the reduced % of production for these hours, for both POs or non POs configurations, and put that in the PVWatts Hourly Excel sheet, to see the difference in production loss between both scenarios.
                      The PVWatts were for a 5.5kW array with a nominal production of 8,960 kWh per year

                      My assumptions were as follow
                      Part of the estimates is because often in the first and last third of the period, panels are unobstructed

                      With Power Optimizers
                      • Fall (Sept-10 to Nov-10) 33% production
                      • Winter (Nov-10 to Feb-1st) 50% production
                      • Spring (Feb-1st to Apr 1st ) 33% production
                      With these assumption, the revised production is 8,201 kWh or 8% loss

                      Without Power Optimizers
                      • Fall (Sept-10 to Nov-10) 10% production
                      • Winter (Nov-10 to Feb-1st) 33% production
                      • Spring (Feb-1st to Apr 1st ) 10% production
                      With these assumption, the revised production is 7,956 kWh or10.8% loss, or an extra 2.7% from the PO option.

                      Alternate assumptions
                      • Fall (Sept-10 to Nov-10) 0% production
                      • Winter (Nov-10 to Feb-1st) 10% production
                      • Spring (Feb-1st to Apr 1st ) 0% production
                      With these assumption, the revised production is 7,818 kWh or12.3% loss, or an extra 4.3% from the PO option.

                      What I am taking away is that a non Power Optimizer option would only lose an extra 2 to 4%, which is not a deal breaker if the configuration otherwise makes sense.

                      Happy to hear if you see any flaws in all that J

                      Comment


                      • #12
                        Originally posted by scrambler View Post
                        Just FYI in case it helps someone, below is the video simulation of the first 2 to 3.5 hours of morning exposure, weekly, during the period where shade occurs Sept-10 to Apr-1st.
                        https://1drv.ms/v/s!At3vMAQjaOZLkRiPWJEiGDDEfwH4

                        I guestimated the reduced % of production for these hours, for both POs or non POs configurations, and put that in the PVWatts Hourly Excel sheet, to see the difference in production loss between both scenarios.
                        The PVWatts were for a 5.5kW array with a nominal production of 8,960 kWh per year

                        My assumptions were as follow
                        Part of the estimates is because often in the first and last third of the period, panels are unobstructed

                        With Power Optimizers
                        • Fall (Sept-10 to Nov-10) 33% production
                        • Winter (Nov-10 to Feb-1st) 50% production
                        • Spring (Feb-1st to Apr 1st ) 33% production
                        With these assumption, the revised production is 8,201 kWh or 8% loss

                        Without Power Optimizers
                        • Fall (Sept-10 to Nov-10) 10% production
                        • Winter (Nov-10 to Feb-1st) 33% production
                        • Spring (Feb-1st to Apr 1st ) 10% production
                        With these assumption, the revised production is 7,956 kWh or10.8% loss, or an extra 2.7% from the PO option.

                        Alternate assumptions
                        • Fall (Sept-10 to Nov-10) 0% production
                        • Winter (Nov-10 to Feb-1st) 10% production
                        • Spring (Feb-1st to Apr 1st ) 0% production
                        With these assumption, the revised production is 7,818 kWh or12.3% loss, or an extra 4.3% from the PO option.

                        What I am taking away is that a non Power Optimizer option would only lose an extra 2 to 4%, which is not a deal breaker if the configuration otherwise makes sense.

                        Happy to hear if you see any flaws in all that J
                        This is not a beef or a knock, but can you describe how you did the guesstimating you write of ? It would seem to me it's not possible, for me anyway, to comment intelligently or perhaps constructively on what you did without knowing some of the details of how you estimated the shade penalty for any hour. It's tempting - but I can't do it and be objective - for me to look at a couple % loss reduction from adding optimizers that you write about and say to myself: "See - that confirms what I've suspected all along - AH HA". However, without knowing how you estimated panel output reduction(s), it's hard, for me anyway, to make a comment or draw any conclusions. You may be correct in your thesis, but I'd need more details before I'd endorse it.

                        I do however, applaud your efforts and encourage you to continue. Very interesting, to me anyway. Honest and FWIW.

                        More FWIW, I used a different method to estimate shading on my array, but it involves the use of photos of the array for each hour of the day on the solstices and equinoxes, and knowledge of the solar azimuths and altitudes at those times. All is then plugged into a spreadsheet that I did from some software I wrote that does some of the same things SAM does, and some things SAM doesn't do. Long story.

                        Comment


                        • #13
                          Originally posted by scrambler View Post
                          Just FYI in case it helps someone, below is the video simulation of the first 2 to 3.5 hours of morning exposure, weekly, during the period where shade occurs Sept-10 to Apr-1st.
                          https://1drv.ms/v/s!At3vMAQjaOZLkRiPWJEiGDDEfwH4
                          I am not sure what you used to make the assumption or the model for the video. AuroraSolar would make an accurate estimate of production given a 3D world model like that for any inverter type with or without optimizers and different string configurations but short of that, almost anything is really a WAG (Wild Ass Guess).
                          OutBack FP1 w/ CS6P-250P http://bit.ly/1Sg5VNH

                          Comment


                          • #14
                            Thank you guys.
                            I don't really want to pay for a full blown analysis, but more enjoy doing the work and getting a rough feel for how large of an impact and how much of a differential between POs and non POs

                            And I agree these are guestimates / WAGs J , especially for the Non PO configuration.

                            Given the above, no need to validate my WAGs, but here was my reasoning.

                            For the PO configuration:
                            For the fall period, as a rough evaluation of a max 3 hours period in the worst conditions
                            The first hour, only half of the panels are affected
                            Second hour, at max all the array is affected
                            Third hour half the panels affected
                            That would be: .33x.5 + .33x0 + .33x.5 = 33%
                            AT the beginning and end of the period, there is actually less shade than that, so the above should be conservative

                            For the winter period, as a rough evaluation of a max 3 hours period.
                            As the trunks pass through, I can usually count at least 50% of the panels unaffected, usually more.
                            That would be: 50%
                            In the first and last hour, shade is actually less, so the above should be conservative.

                            For the fall period, as a rough evaluation of a max 3 hours period in the worst conditions
                            The first hour, only half of the panels are affected
                            Second hour, at max all the array is affected (never quite the case)
                            Third hour half the array
                            That would be: .33x.5 + .33x0 + .33x.5 = 33%
                            AT the beginning and end of the period, there is actually less shade than that, so the above should be conservative

                            For the Non Po configuration:
                            This is clearly more of a WAG, as from what you guys have mentioned, when a string will go down is linked to way too many parameters than just shaded surface.
                            It would also depend on the actual number of strings (for 18 panels on Pika would that be 2 or three??)

                            For the Fall period, first month (half the period), at the minimum the top 40% of the array is basically unaffected, so even if after that the whole array goes down...
                            That would be: .4 x .5 = 20%
                            I counted 10% conservatively

                            For the winter period, during half of the 2.5 month period, I would say at least 40% of contiguous bottom left (could be same string) is unobstructed
                            That would be: .5 x .4 = 20% so my 33% is probably high

                            For the spring period, the last half month the top right 50% (string ?) is almost unobstructed, assuming we don't get anything from the rest of the array and remaining time.
                            That would be: .25 x .5 = 12.5%
                            I counted 10% conservatively.

                            Comment


                            • #15
                              Originally posted by scrambler View Post
                              Thank you guys.
                              I don't really want to pay for a full blown analysis, but more enjoy doing the work and getting a rough feel for how large of an impact and how much of a differential between POs and non POs

                              And I agree these are guestimates / WAGs J , especially for the Non PO configuration.

                              Given the above, no need to validate my WAGs, but here was my reasoning.

                              For the PO configuration:
                              For the fall period, as a rough evaluation of a max 3 hours period in the worst conditions
                              The first hour, only half of the panels are affected
                              Second hour, at max all the array is affected
                              Third hour half the panels affected
                              That would be: .33x.5 + .33x0 + .33x.5 = 33%
                              AT the beginning and end of the period, there is actually less shade than that, so the above should be conservative

                              For the winter period, as a rough evaluation of a max 3 hours period.
                              As the trunks pass through, I can usually count at least 50% of the panels unaffected, usually more.
                              That would be: 50%
                              In the first and last hour, shade is actually less, so the above should be conservative.

                              For the fall period, as a rough evaluation of a max 3 hours period in the worst conditions
                              The first hour, only half of the panels are affected
                              Second hour, at max all the array is affected (never quite the case)
                              Third hour half the array
                              That would be: .33x.5 + .33x0 + .33x.5 = 33%
                              AT the beginning and end of the period, there is actually less shade than that, so the above should be conservative

                              For the Non Po configuration:
                              This is clearly more of a WAG, as from what you guys have mentioned, when a string will go down is linked to way too many parameters than just shaded surface.
                              It would also depend on the actual number of strings (for 18 panels on Pika would that be 2 or three??)

                              For the Fall period, first month (half the period), at the minimum the top 40% of the array is basically unaffected, so even if after that the whole array goes down...
                              That would be: .4 x .5 = 20%
                              I counted 10% conservatively

                              For the winter period, during half of the 2.5 month period, I would say at least 40% of contiguous bottom left (could be same string) is unobstructed
                              That would be: .5 x .4 = 20% so my 33% is probably high

                              For the spring period, the last half month the top right 50% (string ?) is almost unobstructed, assuming we don't get anything from the rest of the array and remaining time.
                              That would be: .25 x .5 = 12.5%
                              I counted 10% conservatively.
                              I believe I understand. Thank you.

                              Comment

                              Working...
                              X