Odd differences between two (nearly) identical PV systems

I've had my 18-panel (two groups of 9) PV (solar) electric system in service for about a year and recently I decided to expand it a bit after realizing that I could do so, myself, for roughly $1/watt, after tax incentives.  An so it was done, with a bit of help from a friend of mine who is better at bending conduit than I:  Another inverter and 18 more solar panels were set on the roof - all done using materials and techniques equal to or better than that which was originally done in terms of both quality and safety.

Adding to the old system:

The older inverter, a SunnyBoy SB 5000-TL, is rated for a nominal 5kW and with its 18 panels, 9 of each located on opposite faces of my east/west facing roof (the ridge line precisely oriented to true north-south) would, in real life, produce more than 3900 watts for only an hour or so around "local noon" on late spring/early fall summer days that were both exquisitely clear and very cool (e.g. below 70F, 21C).  I decided that the new inverter need not be a 5kW unit so I chose the newer - and significantly less expensive SunnyBoy SB3.8 - an inverter nominally rated at 3.8kW.  The rated efficiencies of the two inverters were pretty much identical - both in the 97% range.

Figure 1:
The installed 3.8 kW inverter in operation with the 2kW
"SPS" (Secure Power System) island power outlet shown below.
Click on the image for a larger version.

One reason for choosing this lower-power inverter was to stay within the bounds of the rating of my main house distribution panel.  My older inverter, being rated for 5kW was (theoretically) capable of putting 22-25 amps onto the panel's bus, so a 30 amp breaker was used on that branch circuit while the new inverter, capable of about 16 amps needed only a 20 amp breaker.  This combined, theoretical maximum of 50 amps (breaker current ratings, not actual, real-world current from the inverters and their panels!) was within the "120% rule" of my 125 amp distribution panel with its 100 amp main breaker:  120% of 125 amps is 150 amps, so my ability to (theoretically) pull 100 amps from the utility and the combined capacity of the two inverters (again, theoretically - not real-world) being 50 amps was within this rating.

Comment:  The highest total power that power that I have seen from my system has been about 8000 watts - 3900 watts from the SB3.8 and just over 4100 watts from the SB 5000 for a maximum of about 36 amps at 220 volts (abnormally low line voltage!) or about 33 amps total with a more typical 240 volt feed-in - well under the "50 amp" maximum.

For the new panels I installed eighteen 295 watt Solarworld units - a slight upgrade over the older 285 watt Suniva modules already in place. In my calculations I determined that even with the new panels having approximately 3.5% more rated output (e.g. a peak of 5310 watts versus 5130 watts, assuming ideal temperature and illumination - the latter being impossible with the roof angles) that the new inverter would "clip" (e.g. it would hit its maximum output power while the panels were capable of even more power) only a few 10s of days per year - and this would occur for only an hour or so at most on each occasion.  Since the ostensibly "oversized" panel array would be producing commensurately more power at times other than peak as well, I was not concerned about this occasional "clipping".

What was expected:

The two sets of panels, old and new, are located on the same roof with the old array being higher, nearer the ridge line and the new being just below.  In my situation I get a bit of shading in the morning on the east side, and a slight amount in the very late afternoon/evening in mid summer on west side and the geometry of the trees that do this cause the shading of both the new and old systems to be almost identical.

With this in mind, I would have expected the two systems to behave nearly identically.

But they don't!

Differences in produced power:

Having the ability to obtain graphs of each system over the course of a day I was surprised when the production of the two, while similar, showed some interesting differences as the chart below shows. 

Figure 2:
The two systems, with nearly identical PV arrays.  The production of the older SB5000 inverter with the eighteen 285 watt panels is represented by the blue line while the newer SB3.8 inverter with eighteen 295 watt panels is represented by the red line:  Each system has nine east-facing panels and nine west-facing panels.  The dips in the graph are due to loss of solar irradiance due to clouds.  Because the data for this graph is collected every 15 minutes, some of the fine detail is lost so the "dip" in production at about 1:45PM was probably deeper than shown.
The total production of the SB3.8 system (red line) for the day was 27.3kWh while that of the SB5000TL system (blue line) was 25.4kWh - a difference of about 7% overall.
Click on the image for a larger version.

In this graph the blue line is the older SB5000TL inverter and the red line is the newer SB3.8 inverter.  Ideally, one would expect that that the newer inverter, with its 295 watt panels, would be just a few percent higher than the older inverter with its 285 watt panels, but the difference, particularly during the peak hours, is closer to 10%, particularly during the peak times when there is no shading at all.

What might be the cause of this difference?

Figure 3:
 The two parallel east-facing arrays, the older one being closer to
the (north-south) peak of the roof.
Click on the image for a larger version.

Several possible explanations come to mind:

1. The new panels are producing significantly more than their official ratings.  A few percent would seem likely, but 10%?
2. The older panels have degraded more than expected in the year that they have been in service.
3. The two manufacturers rate their panels differently.
4. There may be thermal differences.  The "new" panels are lower on the roof and it is possible that the air being pulled in from the bottom by convection is cooler when it passes by the new panels, being warmer by the time it gets to the "old" panels.  If we take at face value that 3.5% of the 10% difference is due to the rating - leaving 6.5% difference unaccounted, this would need only about a 16C (39F) average panel temperature difference, but the temperature differences do not appear to be that large!
5. The new panels don't heat up in the sun as much as the old.  The new panels, in the interstitial gap between individual cells and around the edges are white while the old panels are completely black, possibly reducing the amount of heating.  Again, there doesn't seem to be a 16C (39F) difference.
6. The new inverter is better at optimizing the power from the panels than the old one.

It's a bit difficult to make absolute measurements, but in the case of #2, the possibility of the "old" panels degrading, I think that I can rule that out.  In comparing the peak production days for 2016 and 2017, both of which occurred in early May (a result of the combination of reasonably long days and cool temperatures) the best peak was about the same - approximately 28.25kWh on the "old" system even after I'd installed the "new" panels on the east side.

I suspect that it is a combination of several of the above factors, probably excluding #2, but I have no real way of knowing the amount of contribution of each.  What is surprising to me is that I have yet to see any obvious clipping on the new system on the production graphs even though I have "caught" it pegged at about 3920 watts on several occasions during local noon, so it seems that my calculation of "several dozen of hours" per year where this might happen is about right.

I'll continue to monitor the absolute and relative performance of the two sets of panels to see how they track over time.

* * *

Update - June, 2018:

The differences noted above still persist.  In comparing the overall power production results between this year and last, there are no obvious changes in the two systems.


[End]

This page stolen from "ka7oei.blogspot.com"