Zero Energy Home Friends,
My daily PV output from a 6.56 kW PV system has averaged 22.7 kWh/day over the last two years.
irrday= cumulative kWh/m2/day plane of array
pvptmp= array temperature (F) taken on back on one of the modules
obs= elapsed days (August 15th 2012 to August 15, 2014)
To illustrate, my PV output AC over the last two years. IRRDAY is daily cumulative irradiance on the plane of the PV array (kWh/m2). There are about 46 m2 of pv modules (32 ESA 205s) While my PV output looks to have gone down a bit this year, it is coming from lower irradiance primarily. We've had a rainy hot summer.
Now we increase the scrutiny: Here we plot the daily plane of array irradiance against the daily PV power production.
The plot symbols are the Julian dates over the two years (0-365) The mysterious days with high output, but lower irradiance (Julian dates 328 and 329 and 236-240) likely come from something like a well-aimed bird poop on the Licor pyranometer.
Ignoring those, there is a cloud of unusual points which start around Julian date 322 and get worse through the end of the year and also continue in early January. This is cause by shading of the extreme south end of the PV array in November - January by my avocado tree. The avos must be good because they are costing me 5-8 kWh a day during that time. As expected, the impact of the shading is worst around Julian date 340-358 (December 6th- December 24th)-- which is right at the solstice. Also, looking at the first plot, this shading has gotten worse in the last year. I think its time for some trimming. Finally, we can examine how PV performance varies with measured data over time to look for degradation. I created a dummy variable (0,1) for the shading that appears to take place from November 15th to roughly January 16th. (Julian dates 1-16 and 320-365).
The statistical model has an R-squared of 92%-- explains 92% of the variation. The shading is a very significant impact, although the daily solar irradiance is the largest factor (no surprise). PV array temperature has a negative influence (hotter array= worse performance). The variable iT is (irradiance * pvtemp) since solar irradiance is strongly correlated with array temperature in order for the influence of array temperature to properly show itself.
The elapsed days do show a negative course -0.0016 kWh lower each day-- but the term is small, equal to perhaps a kWh in two years. However, this could easily be affected by the growing avocado tree. In fact, a regression on the residual on the above regression showed that this is, in fact, the case. Most of the indicated "degradation" is almost certainly coming from a taller avocado tree in the second year as the shortfall in the prediction is coming during the shading period.
Lisa , just added a Ford Cmax Energi Plug-in to our loads so.... Time to trim! But the PV system appears to keep chugging along.
A further note: You can see most of these factors when you plot the PVoutput to grid against Julian Date and using the year as a plot symbol:
Highest output is in mid March on clear cool days. Output goes down over summer with its heat. And then, starting around November 15th (xlin), the output drops from array shading. This reaches a worse level around Julian date 342 (December 7th-- Pearl Harbor Day-- when it plummets as the output of much of the lower string is lost.
You can read much more than you likely want about the house here:
On-line data here: