X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Generation Calculations

    I ran across something strange. I'm getting the data from the Solar Edge cloud for 15 minute increments and the day totals they provide. The issue is I can't reconcile the totals they give per day as opposed to what I am getting from the 15 minute data points. For example, on the 21st between 8 and 9 am they give a total of 926 watts. However, when I average the 15 minute data they give for the same hour I get 738. As I only have SE, I'm not sure if this is specific to SE or if all brands have the same kind of reporting differences.

    Before I try to get an answer from SE, I was wondering if anyone here could explain the descrepancy. Perhaps there is something I'm missing...

    PS. 783 not 738.
    Last edited by DrLumen; 12-22-2019, 05:35 PM.

  • #2
    First off I would try to understand if what you are getting is instantaneous Watts from one screen and kWhrs from another. Your post wasn't clear which unit of measure you were referring too.
    9 kW solar, 42kWh LFP storage. EV owner since 2012

    Comment


    • #3
      Unless they are doing some aggregation of the previous hour(s) or projecting, I assume the 15 minute data would have to be some average of the previous 15 minutes in watts.

      15 minute data from SE.
      . 2019-12-22_153018.jpg

      Hourly data from SE.
      2019-12-22_152912.jpg

      Calculated from 15 minute data
      2019-12-22_154458.jpg
      Last edited by DrLumen; 12-22-2019, 05:50 PM.

      Comment


      • #4
        When you look at SE dashboard for today production you'll see curved line with X axis showing time and Y axis showing watts produced. KWh on that graph would be area under the curve which could be calculated using antiderivative of the equation of the curve and that's probably what SE is using. You're using simplified formula, just taking averages and if plot it against SE curve, it would be straight lines connecting data points on the graph, not curves, so you're underestimating or overstating the calculations, comparing to SE. Something similar to calculating area of semicircle using formula to calculate area of triangle fitted inside that semicircle. Sorry if this is too much math to chew on Sunday night, but that's the real reason. Btw both yours and SE calculations are estimates, with SE probably closer to actual number.
        Pete4

        Comment


        • #5
          You still have not clarified which of the above is kWhrs and what is Watts. As suggested above just use the SolarEdge calculations.
          Last edited by Ampster; 12-22-2019, 08:54 PM.
          9 kW solar, 42kWh LFP storage. EV owner since 2012

          Comment


          • #6
            I'm not sure I understand your question. When you go to SE monitoring website first page is dashboard and when you click on day, you get a graph of today's solar production. Graph has 2 axis: vertical is marked watts, horizontal is time and it has data points 15 min apart. When OP added all the watts data points, averaged it for each hour and added it up, his number was little bit different from the total SE website stated for that day. For example OP added first hour production, 32+56+456+640=1185/4=296Wh and that was different from website's 230Wh. The difference is due to OP's simple averaging, compared to SE more complicated curved averaging. To visualize it better one could plot data the way OP did it (straight lines between 15 min intervals on the graph and compare to curved lines produced by SE. The area bellow each graph is equal Wh produced. The graph with straight lines (OP) will have different area (Wh) than area under curved graph (SE way of calculating) and it's all due to the way average Wh are calculated. Either way is just estimate: OP is simple to calculate but less accurate, SE way is much more complicated to calculate but little more accurate. BTW 1 kWh is 1kW produced in an hour and kW is 1000 Watts, is that what you're asking?

            Comment


            • #7
              Originally posted by pete4 View Post
              I'm not sure I understand your question........... BTW 1 kWh is 1kW produced in an hour and kW is 1000 Watts, is that what you're asking?
              I was asking the OP to clarify or label the numbers he was posting. You clarified it very well and explained why there is a difference. That is exactly how I understand the numbers from my SolarEdge inverters.
              9 kW solar, 42kWh LFP storage. EV owner since 2012

              Comment


              • #8
                Originally posted by pete4 View Post
                When you look at SE dashboard for today production you'll see curved line with X axis showing time and Y axis showing watts produced. KWh on that graph would be area under the curve which could be calculated using antiderivative of the equation of the curve and that's probably what SE is using. You're using simplified formula, just taking averages and if plot it against SE curve, it would be straight lines connecting data points on the graph, not curves, so you're underestimating or overstating the calculations, comparing to SE. Something similar to calculating area of semicircle using formula to calculate area of triangle fitted inside that semicircle. Sorry if this is too much math to chew on Sunday night, but that's the real reason. Btw both yours and SE calculations are estimates, with SE probably closer to actual number.
                Pete4
                Yeah, that is kinda what I was thinking that they are either calculating from fitted b-splines, a bezier function or something else but did not know for sure. I was hoping someone knew the exact process and even more hopeful that someone may have the actual formula. However, using curves would also imply a non-linear production ramp for the panels. If it was one panel and there weren't much sun, the apex of the curve would be at the max produced at that moment. If full sun, that curve would change with steeper ascending/descending slopes. Perhaps that is the cause of the discrepancies. (shrugs) Again, just going off an assumption I was thinking that panels were linear and that power was a direct relation to the sunlight they get. Maybe that is not the case. If not, every panel from every manufacturer would have different power curves.

                Ampster, that is something I have to check. I always assumed that hourly or daily totals given by SE were kWh but they don't specify. I just get something like 31453 for the day and assumed it was ~31.453 kWh but you're right, it may just be total watts. I would have to think, for this data, the base they use would be the same regardless. Maybe their hour and day totals are kWh but those should be less than the total watts I calculate. Most times, my calculated totals are less than their totals. Not always but most.

                Comment


                • #9
                  Originally posted by DrLumen View Post

                  Yeah, that is kinda what I was thinking that they are either calculating from fitted b-splines, a bezier function or something else but did not know for sure. I was hoping someone knew the exact process and even more hopeful that someone may have the actual formula. However, using curves would also imply a non-linear production ramp for the panels. If it was one panel and there weren't much sun, the apex of the curve would be at the max produced at that moment. If full sun, that curve would change with steeper ascending/descending slopes. Perhaps that is the cause of the discrepancies. (shrugs) Again, just going off an assumption I was thinking that panels were linear and that power was a direct relation to the sunlight they get. Maybe that is not the case. If not, every panel from every manufacturer would have different power curves.

                  .
                  Simple answer to your original question is as I stated above, but you need to realize Inverter itself reads volts/amps constantly and updates those many times a second, it also receives data from individual optimizers, every couple of minutes, so it has much more data to work with, than what's uploaded to SE website in 15 min intervals. As far as panels outputting linear power, maybe so, if they were connected to some steady state resistor, they're not. For example my grid's voltage right now is at 249.5v, with maximum allowed 252V, another couple volts and my inverter will need to shutdown and produce 0 Watts, regardless of how much electricity panels make. As far as formula's go, inverter itself probably takes a reading every 1/10 sec or so, writes it to accumulators, then averages it. If you really wanted that formula, you could do best curve fit, but why?

                  Comment


                  • #10
                    Originally posted by pete4 View Post
                    ......but why?
                    Yeah, I am happy with my daily totals.
                    9 kW solar, 42kWh LFP storage. EV owner since 2012

                    Comment


                    • #11
                      I can understand there is more going on in the inverter than the simplistic data in the SE cloud. However, the SE cloud data should be based on the inverter data regardless of the time frames. If the 15 minute data is not entirely correct how can the hourly or daily data be correct? It all accumulates upward.

                      As to the panels giving a non-linear output, about the only thing I can think that would affect it is the ambient temperature. As the panel temp changes the output changes. That would cause a curve to the output instead of a line from 0 to max.

                      I will likely just use what I have as a reference for now. Maybe it is the touch of OCD in me that wants it to make more sense and for the calculations to be consistent and correct. <shrugs>

                      Once I get the power monitoring installed and set up then I can get more granular. I will know what is being generated as it will be actual real time data and not some type of pseudo totals from SE. Another good thing from that will be that I can use it as an outdoor light meter to trigger the lights instead of using the sunrise/sunsets times or a stand alone PV switch.

                      Comment


                      • #12
                        Originally posted by DrLumen View Post
                        ..... However, the SE cloud data should be based on the inverter data regardless of the time frames.......
                        Literally where else would the cloud get the data other than from the inverter? What the software does with it doesn't matter to me as long as it is consistent and I can use it to see the trends.
                        Last edited by Ampster; 12-24-2019, 06:55 PM.
                        9 kW solar, 42kWh LFP storage. EV owner since 2012

                        Comment


                        • #13
                          Originally posted by Ampster View Post
                          Literally where else would the cloud get the data other than from the inverter? What the software does with it doesn't matter to me as long as it is consistent and I can use it to see the trends.
                          That's really the whole point. The trends are different depending on where you look. You buy 10 widgets @ $.50 each and they say the total is $6.13. Oh well, no matter. You can use the 6.13 for your trending. Garbage in = Garbage out.

                          I'll see what SE has to say.

                          Comment


                          • #14
                            Originally posted by DrLumen View Post

                            I'll see what SE has to say.
                            Ask them if the total kWhrs reported by the portal each day is a revenue grade measurement?
                            9 kW solar, 42kWh LFP storage. EV owner since 2012

                            Comment

                            Working...
                            X