Tuesday, 1 July 2014

bimodal June

June, with its longest day, has come and gone.  So we have seen maximum possible solar power for this year, if not maximum actual.  Here is a view of the power generation per month:

The horizontal time axis runs from 3:00am to 9:00pm GMT (no correction for summer time: this is all sun time!). The vertical axis runs from zero to 8kW.  Here the black line is the mean generation at that time, averaged over the month.  The orange regions indicate the minimum, lower quartile, median, upper quartile, and maximum generation at that time, over the respective month.

The daily power production per month is:

(For an explanation of these violin plots, see the May solar power post.)  There are several things interesting about this plot.

  1. June is (slightly) bimodal.  We had mostly either dull days or sunny days, with fewer intermediate days.
  2. The June minimum is significantly higher than previous months: even the dull days weren’t that dull.  In fact, the June minimum is about the same as the January upper quartile: three quarters of the January days were duller than the dullest June day.
  3. The June maximum is less than the May maximum, which is less than the April maximum.  We never had an all clear day in June.  Here are the maximum days for each month (where the black line shows the production on that particular day):

We suspect that there is a second effect causing this reduction, too.  These later very sunny days not reaching the peak generation around noon may be due to temperature effects: solar panel power output decreases at higher temperatures.  The orange maxima shown were achieved on a range of different days with spiky production, due to scudding clouds, which will have reduced the temperature.  We await a hot summer to see if this effect continues.  We’re not sure how many years we will have to wait for a hot summer, of course…


  1. Could there be a couple more significant factors? How about the exact orientation of your nominally south facing roof and the pitch of the roof, compared with the angle of the sun in the sky at 12 noon?

    I'm sure it would be relatively straightforward to model the power of the sun incident on panels for each day of the year, physical house location, house orientation and angle of pitch of roof, which would give a theoretical maximum at any time of any day (and a nice normal curve for the day) and determine the theoretical "best day of the year" for your panels. Of course this would not take into account panel efficiency, temperature or cloud effects, but it would be interesting to see how your panels are performing against a theoretical maximum. It's now nearly 4 decades since I got my Maths degree, so I'll leave the calculations to someone else :-)



    1. I'm not sure the orientation (very slightly east of full south) is a *more* significant factor, although it does have an effect. However, and around noon the production is currently maxed out: there's a distinct "flat top" to the graphs. Yet that flat top isn't achieved on sunny days, only as the aggregate of "spiky" days. The orientation effect can be seen by the longer tail in the evening than the morning: as the sun starts to set there is actually no direct sunlight incident on the panels: all the production is from scattered light, and we get more if there is some cloud than from a perfectly blue sky.

      But yes, it's definitely not as simple as "maximum production at the solstice". I'll just say "analemma", and leave it at that :-)

      Having solar panels provides as much entertainment value as it does electricity!