Solar Production in the Winter

[NYHeel]

I installed a 10.85 kW system this past June and fully turned it on at the end of July. 4.65kW are southern facing and 6.2 kW are eastern facing. Obviously this is my first winter with solar and I'm wondering if my production is normal. I live in Northern New Jersey (Bergen County) so we do get real winter weather here.

I track my production daily both with PVOutput (from my Solaredge inverter) and by reading my production meter (revenue grade meter for SREC purposes). I also have my PVWatts projections by month (and day but I know those aren't very helpful) using a 10% loss. My system has very little shading, probably a little in the early morning and possibly a drop at the end of the day.

My PVWatts projection shows 686 kWh for December (averaging 22.1 kWh per day) and 848 kWh for January (averaging 27.3 per day). However, my production for December was only 543 kWh (21% below PVWatts) and for January I'm on pace for about half of the PVWatts number (I'm averaging about 13.7 kWh per day). I realize that days are getting longer so my January numbers might bump up a bit but I'm still likely to be at no more than 60% of the PVWatts projection.

While I've never really studied the number of cloudy/sunny days in previous winters, the overall weather here this winter has been fairly normal, if not a little better than normal. Also, we haven't had very much snow yet, just one storm of about 5 inches. With that said I'm trying to figure out if the PVWatts numbers for the winter are just crazy high. I just don't see how I could possibly average 27 kWh per day in January. I had a few near perfect sunny days a few days ago and produced 33 and then 34 kWh on those days. That means on perfect days I'm only out producing the average by 6-7 kWh. Yet on a cloudy/rainy day I'm only going to produce 0-4 kWh. For example, yesterday with rain was 2 kWh and today with just clouds and no rain is looking like it will be about 3-4 kWh. So for every cloudy day, I need about 4 nearly perfectly sunny days. That's just not realistic. Could the PVWatts numbers be wrong? That doesn't seem likely either. I put in my measurements based on what my installer had calculated. The Azimuth appears correct and the tilt is probably not off by more than 5 degrees based on my visual estimate.

Also, regarding snow, I had read that snow typically melts/slides off of a tilted array the next day if there's sunshine. I haven't found that to be true. My panels are at a 25-30 degree tilt and it took over 3 days to get melting of a 5 inch snow storm. The first day after the snow had plenty of sun but temperatures in the 20s. The panels remained nearly fully covered. Over 4 days (including the day of the snow) I got a combined 2 kWh of production because of the snow.

I'm not really trying to just complain about my performance. I just want to see if others have experienced this as well and if what I'm seeing is relatively normal. I was within +/-12% of my PVWatts numbers for each month in August - November. It's really just been December and January that my numbers seem significantly low relative to PVWatts. I have my system on PVOutput and can post a link if that would be helpful.

[bcroe]

The truth is you are totally at the mercy of the weather. Here, ice I have to let melt; snow is always cleared manually before
the next sunny day. The equipment is designed to be easier to clear. The only way to collect more energy under clouds is to put
up more panels, though the rest of the plant might be unchanged. My production in NW IL is comparable to your numbers.

PV watts doesn't have a weather man; it just guesses based on historical info. It will probably be a lot closer when you take
the average of half a dozen winters. Bruce Roe

[NYHeel]

The truth is you are totally at the mercy of the weather. Here, ice I have to let melt; snow is always cleared manually before
the next sunny day. The equipment is designed to be easier to clear. The only way to collect more energy under clouds is to put
up more panels, though the rest of the plant might be unchanged. My production in NW IL is comparable to your numbers.

PV watts doesn't have a weather man; it just guesses based on historical info. It will probably be a lot closer when you take
the average of half a dozen winters. Bruce Roe

Yeah, I hear what you're saying. The December numbers seem doable with the right kind of weather. The January numbers though, just seem too high to be plausible in any kind of reasonable scenario. But I guess I'll just have to wait 5+ years to get more credible data. 27 kWh per January day for my system just seems crazy high but I guess there's always wait until next year.

For me, manual removal of roof snow isn't really an option. I don't have a ladder that can get to my roof and, even if I got one, considering that if snow is on the roof then there's snow on the ground, going up a ladder to then reach around my roof with some kind of broom seems like a really bad idea and a way to get myself killed to try and save $10. I can't actually stand on my roof since there are panels on most of it and it's on the steep side to be on it with snow anyway.

[DanS26]

Did you run PVWatts as one calculation or two...in other words did you do a separate calc for each array orientation? Then there's shading.....

10% loss seems low the default is 14%....did you try to true up the PVWatts estimate with your actual experience for the first few summer months?

Then there is also the possibility that your system is not performing up to specs. On a perfectly sunny day is/are the inverters performing as expected?

[J.P.M.]

Pretty much what Bruce said, except that PVWatts not only doesn't have a weather man, but most of what it spits out is based on something other than actual data, particularly for the irradiance data, with most of that for most locations being modeled and not historical data (BTW, you'll only find that little factoid with some digging around on the NREL website). Long story. Still, the output from models that use TMY or Solar Anywhere data seem to produce reasonable versions of what to expect over many (say 10 or so) years. For any one year, its' +/- 20-30 % or so. For a season, good luck.

FWIW, My array went live on 10/17/2013. One of the things I keep track of is a running comparison of my array's output for 31 day periods. The immediate prior 31 day output for the period ending 01/17/2017, (12/18/2016 - 01/17/2017) was 81.5 % of the same period 1 yr. ago, 71.4 % of the output from the same period 2 years ago, and 63.1% of the output for the same period 3 years ago. The same #'s for the period 11/17/16 - 12/16/17 were 83.4%, 103.3 % and 100.3 % respectively.

The min. and max. of the running 31 day average this year (prior 365 days) against the same period last year: Min. = 72.9%, Max. = 127.1 %.
The min. and max. of the running 31 day average this year (prior 365 days) against the same period 2 years ago: Min. = 71.4 %, Max. = 128.2 %.

I don't have the same type of data for a comparison to PVWatts or SAM or TRNSYS, but the min/max #'s above seem to make some sense with the PVWatts advisement that any one month's model output from PVWatts may be no better than within +/- 30 % or so of an array's actual output.

I don't have the same type of data for a comparison to PVWatts, SAM, or Trnsys. Seemed a waste of time.

Points are:

- There's a lot of variation from one year's same period to the next.
- The weather data used by most models is mostly synthetic anyway.
- The probability of an array's actual output matching that of a model's estimated output when the weather is involved is mostly an exercise in futility, regardless of what some folks may incorrectly believe.
- Actual array output will vary a lot over short periods, with "short" having a somewhat imprecise definition.

If you think there's a problem with your array, call your installer. Otherwise, IMO, it's probably a situation of some days (or seasons) you get the bear and some days (or seasons) the bear gets you. Climate's what you expect, Weather's what you get. Same with models of solar performance dependent on the weather.

[NYHeel]

I did 2 separate PVWatts calculations, one for each array. It could be that I should use 14% instead of 10%. Also, I used the Solar Anywhere Weather data. Is that bad?

I understand and agree with what everyone here is saying. JPM, those numbers are interesting showing the volatility in 31 day production, year over year. I'll have to wait a few Januaries to see how they end up looking for me.

With that said, I still think the PVWatts January numbers are ridiculous. I think the daily numbers are somewhat telling. I understand that the daily numbers are useless for purposes of comparing to real life data with the same date. However, I would think that minimums and maximums of a period can provide some insight when comparing to one's own system. The maximum for January is 41.5 kWh with 10 days above 36 kWh and 5 days above 40. So far I have zero days above 36 kWh. The two near perfect days came to 33 and 34 kWh. Granted, I was never really hitting the maximum numbers that PVWatts shows, even in the summer, though my differential was a significantly lower in the summer. The crazy thing is their minimum performance in January is 4.8 kWh with only 2 days below 10 kWh. It's like they don't realize that it ever rains or snows. I've already had 6 days in January with 2 kWh or less. The numbers in PVWatts just seem strange. I guess only time will tell.

[J.P.M.]

I did 2 separate PVWatts calculations, one for each array. It could be that I should use 14% instead of 10%. Also, I used the Solar Anywhere Weather data. Is that bad?

I understand and agree with what everyone here is saying. JPM, those numbers are interesting showing the volatility in 31 day production, year over year. I'll have to wait a few Januaries to see how they end up looking for me.

With that said, I still think the PVWatts January numbers are ridiculous. I think the daily numbers are somewhat telling. I understand that the daily numbers are useless for purposes of comparing to real life data with the same date. However, I would think that minimums and maximums of a period can provide some insight when comparing to one's own system. The maximum for January is 41.5 kWh with 10 days above 36 kWh and 5 days above 40. So far I have zero days above 36 kWh. The two near perfect days came to 33 and 34 kWh. Granted, I was never really hitting the maximum numbers that PVWatts shows, even in the summer, though my differential was a significantly lower in the summer. The crazy thing is their minimum performance in January is 4.8 kWh with only 2 days below 10 kWh. It's like they don't realize that it ever rains or snows. I've already had 6 days in January with 2 kWh or less. The numbers in PVWatts just seem strange. I guess only time will tell.

To repeat, climate's what you expect. Weather's what you get. Modeling programs are not performance predictors, nor are the weather predictors.

If you can compare your output for similarly oriented arrays reasonably close to you using PVOutput, and see that your output is regularly lower than others when they are similar to one another, then I'd be suspicious.

I'd also check my PVWatts inputs, particularly if you've never got close to PVWatts output when comparing identical calendar days.

Maybe the shading has a bigger influence than you think.

But, it also sounds like snow retention may be having an impact as well for you, maybe a large one, and that's what's probably having an impact in the winter as well.

If still in a quandary, there's always the possibility that the system is indeed not functioning as efficiently as designed. In that case, and if you've done due diligence to uncover as much on your own as possible, and are still stumped, a call to the vendor may be in order, and completely appropriate.

Good luck,