California is back to drought weather = great solar production

[steveholtam]

In Sacramento my little system generated 157% better than PVWatts estimated at zero loss conditions. 663 kWh produced compared to the 422 estimate. This from my 6.16 KW system. Wish we had some rain, but I guess I'll take the solar energy. I normally base my system losses at 10.5%, and that ended up being 176% of expected. Crazy. Anyone else enjoying the sun?


ScreenHunter_716 Jan. 02 10.55.jpg

[Sunking]

Well you can blame the recent fires on environmentalist or the Green Mafia as I call them. They have a lot of blood on their hands and billions of dollars of property losses. Most of those fires could have been contained or avoided all together. Several proposals proven to work were denied. What is worse would not have cost tax payers money. In many states the US and State Forestry departments allow companies associated with Forestry and Landscaping to go in and cut fire breaks, remove dead under brush/trees, and thin out trees in the Forrest. Then allow those companies to process that material to pay for it and make a profit. All attempts have been blocked by the Green Mafia. It is OK to cut down tress if it cast shade on a solar panel, but do not dare cut or remove anything that threatens property or human life with wild fire. Just plain stupid and ignorant policy.

Enjoy the sun through the smoke.

[azdave]

My 6.63kW system in Phoenix generated 606 kWh which was 99% of my needs however, that is only 82% of what PVWatts predicted assuming 10% system losses. 82% seems logical to me since we had a very cloudy month in December.

Of course, 30-day snapshots of data don't have much real value other than being a fun factoid.

[J.P.M.]

Dec. around here was dry this year, with no measurable precip. However, array production, while not bad, wasn't spectacular compared to the prior 4 Decembers it's been in operation.

Dec. production,
2013 = 587 kWh
2014 = 460 kWh
2015 = 585 kWh
2016 = 467 kWh
2017 = 578 kWh

PVWatts w/10% system loss parameter = 584 kWh.
SAM w/3% fouling factor and 3.13 % wiring losses = 598 kWh.

Watts and SAM outputs based on Miramar MCAS, TMY3 data.

System size = 5.232 STC kW
Lat. = 33.22 deg. N
Long. 117.10 deg. W
Alt. above mean sea level = 420 m.
Array az. = 195.3 deg.
Array tilt = 18.8 deg.

The Dec. 2017 output would have been higher by probably a few %, but the array has intentionally not been cleaned since 06/21/2017 due to my ongoing efforts at trying to figure out how, when, and in what ways the array fouls up. Dec., 2017 started with the array performance impacted by about a 6 % penalty due to dirt, and finished with about a 10 % or so penalty after no rain and no heavy dew to affect cleanliness.

A bit off topic, but that 10% - 6% = 4% increase in fouling over a month seems about in line with prior data I've taken that points to about a 1 % performance deterioration/week if it doesn't rain.

[steveholtam]

Yeah, I guess it is just for fun. I think I just got lucky with a near-perfect roof line directly aimed at the sun all day. As seen below, my monthly production for the last twelve months is only showing about a 3% loss from maximum expected. I still have overages almost every month since my family thinks we have free energy. lol.

ScreenHunter_717 Jan. 02 15.25.jpg

[morgandc]

I have been loving the Sacramento weather, but haven't been using to produce much energy... my panels are flat on the top of my trailer

[J.P.M.]

Yeah, I guess it is just for fun. I think I just got lucky with a near-perfect roof line directly aimed at the sun all day. As seen below, my monthly production for the last twelve months is only showing about a 3% loss from maximum expected. I still have overages almost every month since my family thinks we have free energy. lol. ScreenHunter_717 Jan. 02 15.25.jpg

Fun or otherwise, what's more likely is that your weather produced about 8% or so more output than the PVWatts model estimated because the actual weather, probably and mostly due to greater actual irradiance, with lesser but still some influences from ambient temp. and wind vector, produced most of the difference in output between actual output and modeled output you're showing, which is the result of TMY or SolarAnywhere data. And, BTW, with most all of the TMY data itself being modeled data. That is, most of the TMY data is not from actual measurements - it's mostly estimated or extrapolated from the 35 to 60 original SOLMET stations around the U.S., that number depending on which stations were in/out of service, and how they were maintained many years ago (1951 - 1975). See the TMY manual for details. You'll need to dig out that most of the 1,000 + station data is synthetic, and it's cryptic in reference, but it is there and confirmed by other sources.

This isn't a knock, but without knowing the actual P.O.A irradiance and temps. and wind vector at your array, you cannot know if the difference between the 10,378 kWh actual reported output and what the model estimated was due to weather or differences between actual and input parameters, or other things. Your numbers, while they may be fun, but don't mean much in terms of meaningful analysis.

FWIW, keeping a running day to day total of the ratio of my system's actual 31 day total output to the PVWatts modeled 31 day total output for same periods over the last 4+ years produces an average system output for the 1505 data points that is 99.7% of the PVWatts output. Doing all that was what led me to confirm my suspicion (and that of other's) that the 14 % default system loss parameter was too high. The system loss parameter that actually gets that 0.997 to 1.000 for my system is actually slightly lower than 10 %, but this business being less than rocket science doesn't warrant that much more accuracy.

System loss parameter used = 10%.
Average of system 31 day output/PVWatts modeled 31 day output = 0.997
Minimum ratio of system 31 day output/PVWatts modeled 31 day output = 0.602
Maximum ratio of system 31 day output/PVWatts modeled 31 day output = 1.233.
Population std. dev. = 0.099.
N = 1505.

When graphed, the ratio of actual 31 day output to modeled 31 day output looks pretty much like a normal probability density function, that is, a normal distribution. So, after slightly more than 4 years, over any 31 day period, my array's average 31 day output has a pretty high probability (~ 0.99) of being the same as the average PVWatts 31 day output for the same period, +/- ~ (2.33)*(0.099) = 0.231 or +/- 23 % or so. 2.33 being the Z-score for 0.99 confidence level.

I don't have a clue how anyone else's array might perform compared to any model, but I suspect most reasonably well designed arrays in north county San Diego, and probably a lot of arrays in CA as well, may exhibit behavior similar to mine.

Although different than what the PVWatts manual says about modeled compared to actual system output (my data seems tighter than PVWatts' +/- 30% for monthly data), the above seems to compare reasonably favorably with that +/- 30% that the PVWatts user manual suggests as a confidence level for comparing likely and expected monthly actual system outputs to monthly modeled system outputs.