Insolation curves look strange

Problems
#1

Up until a few days ago my Insolation curves were, generally speaking, normal. A few days ago I began seeing some “blips” appearing more or less every two hours from 10 AM to 4 PM. This is yesterday’s plot:

image
image1815×700 67.4 KB

and March 29th plot:
image
image1916×614 65 KB

As you can see they occur in both arrays at the same time suggesting that the calculation method is the reason. The day was completely clear and the array sees minimal shading in early morning and mid afternoon, neither of which correlate with the blips.
Anyone else seeing this behavior?

#2

Checked my settings and found that the temperature being recorded was an amalgamation of the weather station temp and the inverter temperature. The blips correspond with the temps recorded which seems to alternate between the invert temperature and the weather station temperature. I don’t know why it does that but I disabled the Temp on the inverter page and the blips are going away. It would appear that the temperature recorded is utilized in the calculation of Insolation.

image
image1909×667 65.6 KB

Temperature profile is here:
image
image1916×716 54.3 KB

It is my guess that the temperature gaps occur because the Isolation calculation are affecting the temperature which is being recorded. I disabled the inverter temperature about 11:30. So the higher inverter temperature was depressing the Insolation value. That also explains why my Isolation values were low before I started recording my weather station data.

#3

Ah - this will explain why my insolation curves have developed a ripple since I added the temperature tracking via OpenWeatherMap recently.
Before, I wasn’t tracking temperature at all, and my insolation curves were perfectly smooth.
Now, after adding OpenWeatherMap temperature to my charts, my insolation curves have ripples:

insolationripple
insolationripple942×387 119 KB

#4

Ideally the temperature should be measured at the panel but this requires more equipment. My weather station (ambient weather WS2000) is about 40 feet away and somewhat higher than my panels. However, most of the day when the sun is shining the prevailing wind is from the West by West Northwest direction from the ocean. The ocean air temperature on the Southern California coast is fairly constant constant during the productive hours so my panels see the sea breeze most afternoons and the temperatures are fairly constant.
For the longest time I couldn’t figure out why my Insolation was so low and then I found through this analysis that the temperatuer used in the calculation is that which is displayed on the Generation page. I was using the setting which enabled the recording of the Inverter temperature and sure enough the higher it went the lower the Insolation. Disabling it solved the problem.
I just checked and now the Isolation has been oscillating since 2 PM today. Before that it was smooth.

image
image1896×765 80.2 KB

I am really not sure why because the temperatuer curve is not oscillating.
image
image1372×764 47.7 KB

Go figure!

#5

Checked again and all the squiggles shown in the previous post are gone.

image
image1938×857 89.7 KB

#6

I still have slight squiggles, but consistently only in the mornings portion of the curve. left-hand-side before around 11am-12pm squiggles, after that the curve is smooth.
https://pvoutput.org/intraday.jsp?id=64310&sid=57175&dt=20190421&gs=0&m=1 and click on ‘insolation’ to see it.

#7

These squiggles must have something to do with the way the calculation is done and it looks that bankstownbloak isn’t telling. Mine behaves differently with the squiggles appearing during the day near the current reading but disappearing completely after dark.
The only difference I see in the way our data appears is that my power and average power are almost always within a few Watts of each other all day while yours seem to differ more showing large deviations from the green power curve.

#8

If you set the panel coefficient to 0 the strange phenomena , and temperature dependency with it, will at least go away.
It looks like the “squiggles” are caused by a bug in the low-pass filtering method for the temperature ? But who knows , we would know more if the software was open-sourced :slight_smile: