Anyone with WSW array in SoCal

[FFE]

More later, but for now, run PVWatts w/ a 8-10% system loss instead of the 14 % default, get the hourly output option and look for a "clear" day near the date you're seeking. It ain't perfect, but it's not too bad either.

That is sort of what I did to get 17.6 kWh except with 11.42%. If I switch to 9% and check +/- 7 days I get the following highest three of 19.4 for Nov 13, 17.9 on Nov 21 and 17.7 on Nov 16 (the day in question). If I go back to Nov 5 it is 20.4. I used 18.6 kWh in the video to give it my best guesstimate.

The 3.5 I am looking at was old school methods. Use Google Earth to find a neighbor on a parallel street and ask what his production was. He even has the exact same panels I do, but a different inverter. The closest PVOut was Aurora WH with 225 degrees, 18 tilt and no shade with a 3.4.

Thank you for all of your help!

[J.P.M.]

My post # 4 of this thread is incorrect. My 21-22 kWh clear day est. as written was for a south facing (180 deg. az.) array - NOT west (240 deg. az.).

I apologize for the error.

Now, after checking myself:

Lat. = 34.2 deg.
Long. = 118.9 deg.
Tilt = 18.4 deg.
Az. = 240 deg.

Approx. Clear day output for 11/15/2015:

6 A.M- 7 A.M - 15 Watt-hr.
7 - 8 ~ 231
9 - 10 ~ 427
10 - 11 ~1,194
11 - 12 ~2,113
12 - 13 ~2,765
13 - 14 ~3,008
14- 15 ~ 2,547
15 - 16 ~ 1,617
16 - 17 ~ 148

Day long output ~17.1 kWh.

Array size, 5.886 kW, S.P. 327's

Using TMY3 temp./wind data and HDKR model for clear sky irradiance est. at location w/no shade and about 3 % fouling.

Seems pretty close to what you SWAG'ed.

Hope that helps.

Note, even if off some for fouling and degradation, hourly values will probably be proportionate to one another.

Again, sorry for my error.

Add: check PVWatts daily output for 11/18 using Van Nuys as data source. That seems the nearest semi/mostly clear day and shows ~ 17.6 kWh for the day.

[FFE]

Thank you for your help. I'll stick with the 18.6 kWh as my theoretical output. I will probably start a new thread in case someone without a WSW array might want to see it.

I know another option would have been to put up 12 super efficient panels on the sunny side to get a 4.2 kW DC system. But every day isn't Nov 16. The picture below is from Jun 20.

[sensij]

FWIW, I'm observing that this is the time of year that PVWatts seems have the less than normal accuracy relative to live system data. For example, the OP has reported 14.4 kWh from a 5.3 kW array, 2.72 kWh / kW, with significant shading. From the previous post (JPM's swag and the PVWatts model), 17.1 - 17.6 kWh from a 5.89 kW array, 2.90 - 2.99 kWh / kW. Did the shade really cost only 10%?

Some more PVOuput.org data from 11/16
Compare the Aurora WH system which produced 12.71 kWh from 3.75 kW (3.39 kWh / kW), at SW azimuth + 18 deg tilt, with significant afternoon shading.

to the Fridrich Home, which got 43.319 kWh from 10 kW (4.33 kWh / kW) at 225 deg azimuth and 15 deg tilt, no shade. It suggests the shade on Aurora WH cost 22%.

Here is another one... Dawn-David got 17.483 kWh from 5 kW, (3.50 kWh / kW) at 275 deg azimuth + 22.5 deg tilt, no shade, with a small boost in the morning. Even the next day, which was 12 deg F warmer, and did not have the small morning boost, still produced 3.24 kWh / kW... pushing the OP's shade loss to at least 20%. PVWatts suggests the OP's 240 Az + 18.5 tilt would significantly outperform (+24%) this system's 275 deg Az + 22.5 deg tilt this time of year, so the shade loss could be even higher.


Picking on unknown PVoutput systems for deep analysis is fraught with peril, but even my own system (cited earlier and linked below) is producing more than what a PVWatts model for the location would show right now, with ambient temperatures that are very much in line with what the TMY file is using. A 3.12 kW system, 179 deg azimuth and 18.5 deg tilt, using Montgomery Field TMY and 8% loss shows clear day production of 14.1 - 14.4 kWh this time of year (4.52 - 4.62 kWh / kW). I'm actually producing 14.5 - 14.9 kWh (4.65 - 4.77 kWh / kW), with some late afternoon shade. The difference can also be seen in the hourly output, which suggests I should be peaking at around 2150 W and am in fact routinely getting more like 2300 W - 2400 W.

I'm not picking on PVWatts, or what J.P.M. posted... just trying to illustrate that driving daily model accuracy better than 10% or so this time of year may require better on-location data, from a local weather station that includes an irradiance sensor, for example (such as what J.P.M. has installed for his own system, and I'd love to do for mine eventually).

Ultimately, and perhaps more to the OP's point, the difference in output for a panel optimized system and a string system in his particular case (and in many common situations) is likely to be very small. Partial hard shade on a panel, like that produced by his chimney, will cause the circuit to bypass the shaded cells through the built-in diodes, regardless of whether that panel is in a string or has panel level MPPT. Panel level electronics may squeeze out a little extra power from the tree-shaded panels, but it isn't going to be much relative to what the unshaded panels are producing. When each string has its own MPPT, and as long as the bypassed portion does not drop the string voltage below efficient MPPT operation, there isn't going to be much difference between a partially shaded string and a partially shaded panel optimized system. Scare tactics claiming 50% loss on string systems are frequently unfounded.

Whether or not it makes economic sense to install panels that receive significant shade is a different question. The answer to that depends very much on what the total electricity cost would be with and without those panels, accounting for the upfront cost to install them... beyond the scope of the discussion in this thread so far.

[FFE]

Fridrich Home, which got 43.319 kWh from 10 kW (4.33 kWh / kW) at 225 deg azimuth and 15 deg tilt, no shade. It suggests the shade on Aurora WH cost 22%.

Here is another one... Dawn-David got 17.483 kWh from 5 kW, (3.50 kWh / kW) at 275 deg azimuth + 22.5 deg tilt, no shade, with a small boost in the morning. Even the next day, which was 12 deg F warmer, and did not have the small morning boost, still produced 3.24 kWh / kW... pushing the OP's shade loss to at least 20%. PVWatts suggests the OP's 240 Az + 18.5 tilt would significantly outperform (+24%) this system's 275 deg Az + 22.5 deg tilt this time of year, so the shade loss could be even higher.

Scare tactics claiming 50% loss on string systems are frequently unfounded.

Thank you for taking the time finding the two examples in PVout. Those didn't show up in my search. Nor did I figure out how to show one day's output graphically which would make it easy to see shading issues. Aurora showed in their profile that their system has no shade. As soon as you look at it graphically it is obvious they have issues. Strange that PVWatts shows that a 240 system would be 24% better than a 275. That would mean that a 240 degree system would out perform Fridrich's since 3.5 times 1.24 equals 4.34. That doesn't sound right. I believe my system should be somewhere between the two above examples. So, maybe my system would be close to 4.0 kWh/ kW DC without shade since my 240 degree array is between the two but closer to the higher one. Also, both of my examples for 3.5 were problematic. Aurora does have shade and I double checked Google Earth and realized that the neighbor has about 1/3 of his panels on the other side of the roof. Time to edit my video.

I agree about the scare tactics. They might be unfounded, but they are plentiful. If you google videos "effect shade solar panel" the top videos show massive power losses. Even SunPower has a video that shows one leaf causes 1/3 power loss on conventional panels compared to 11% loss on theirs. Later is shows 100% power loss from one leaf on the panel over time. I am just trying to get some real wold info out there on a complete array in real shade. Not have a chicken walk in front of a panel causing 50% power loss or a sliver of shade and a running bilge pump turns off.

[sensij]

The 1/3rd power loss due to a single leaf is very real, but that is not specific to a string inverter. Most 60 cell panels are broken into three 20 cell strings, with bypass diodes between them. If any one of those strings is not able to produce enough current to keep up with the others, the voltage biases in a way that allows the full current of the other cell strings to pass through the diode, effectively shutting down the underperforming 1/3rd the panel. It only take blocking one cell (with hard shade) for this to happen. With soft shade, the boundary between bypassed or not is fuzzier.

With respect to the influence of array orientation, here is some PVWatts data, using Van Nuys TMY3 and looking at just the average daily November plane of array insolation.

225 deg azimuth, 15 deg tilt = 4.39 kWh / sq m / day
225 deg azimuth, 18 deg tilt = 4.52 kWh / sq m / day
240 deg azimuth, 18.5 deg tilt = 4.26 kWh / sq m / day
275 deg azimuth, 22.5 deg tilt = 3.43 kWh / sq m / day

These data are based on "typical" weather, not just clear sky, so actual weather patterns during the month (morning or evening clouds) will move the numbers around some and could skew more in favor of south facing or west facing. I don't have clear sky modeling handy to post what the weather independent orientation difference would be, which would be the more germane comparison for a day like 11/16.

[wwu123]

Thank you for your help. I'll stick with the 18.6 kWh as my theoretical output. I will probably start a new thread in case someone without a WSW array might want to see it.

I know another option would have been to put up 12 super efficient panels on the sunny side to get a 4.2 kW DC system. But every day isn't Nov 16. The picture below is from Jun 20.

Well that looks like all my panels by 3 pm year-round, as I've got trees, trees, trees to the west and southwest. I also have some morning trees to the east, and until recently, some eucalyptus trees to the south that from Nov to Jan would put the panels in shade from noon onwards due to the low sun position. Miraculously, the neighbor topped those 70-80' tall eucalyptus during the summer, so whereas I'd normally be headed into the darkness this week, I'll now have a clear south through the winter.

I've also got a coast live oak that is evergreen but drops dead leaves all year-round on the roof. So I've always got usually got at least a couple of small leaves on half my panels, but amazingly, there is no visible difference in production on a panel with leaves vs one without. However, coast live oak leaves are small, each is much smaller than one cell.

But as far as the shading from the trees, yes it's not all doom and gloom. However, for sure if a panel is half-shaded, I'm not getting 50% output, I'm getting close to zero.

[J.P.M.]

FWIW, I'm observing that this is the time of year that PVWatts seems have the less than normal accuracy relative to live system data. For example, the OP has reported 14.4 kWh from a 5.3 kW array, 2.72 kWh / kW, with significant shading. From the previous post (JPM's swag and the PVWatts model), 17.1 - 17.6 kWh from a 5.89 kW array, 2.90 - 2.99 kWh / kW. Did the shade really cost only 10%?

Some more PVOuput.org data from 11/16
Compare the Aurora WH system which produced 12.71 kWh from 3.75 kW (3.39 kWh / kW), at SW azimuth + 18 deg tilt, with significant afternoon shading.

to the Fridrich Home, which got 43.319 kWh from 10 kW (4.33 kWh / kW) at 225 deg azimuth and 15 deg tilt, no shade. It suggests the shade on Aurora WH cost 22%.

Here is another one... Dawn-David got 17.483 kWh from 5 kW, (3.50 kWh / kW) at 275 deg azimuth + 22.5 deg tilt, no shade, with a small boost in the morning. Even the next day, which was 12 deg F warmer, and did not have the small morning boost, still produced 3.24 kWh / kW... pushing the OP's shade loss to at least 20%. PVWatts suggests the OP's 240 Az + 18.5 tilt would significantly outperform (+24%) this system's 275 deg Az + 22.5 deg tilt this time of year, so the shade loss could be even higher.


Picking on unknown PVoutput systems for deep analysis is fraught with peril, but even my own system (cited earlier and linked below) is producing more than what a PVWatts model for the location would show right now, with ambient temperatures that are very much in line with what the TMY file is using. A 3.12 kW system, 179 deg azimuth and 18.5 deg tilt, using Montgomery Field TMY and 8% loss shows clear day production of 14.1 - 14.4 kWh this time of year (4.52 - 4.62 kWh / kW). I'm actually producing 14.5 - 14.9 kWh (4.65 - 4.77 kWh / kW), with some late afternoon shade. The difference can also be seen in the hourly output, which suggests I should be peaking at around 2150 W and am in fact routinely getting more like 2300 W - 2400 W.

I'm not picking on PVWatts, or what J.P.M. posted... just trying to illustrate that driving daily model accuracy better than 10% or so this time of year may require better on-location data, from a local weather station that includes an irradiance sensor, for example (such as what J.P.M. has installed for his own system, and I'd love to do for mine eventually).

Ultimately, and perhaps more to the OP's point, the difference in output for a panel optimized system and a string system in his particular case (and in many common situations) is likely to be very small. Partial hard shade on a panel, like that produced by his chimney, will cause the circuit to bypass the shaded cells through the built-in diodes, regardless of whether that panel is in a string or has panel level MPPT. Panel level electronics may squeeze out a little extra power from the tree-shaded panels, but it isn't going to be much relative to what the unshaded panels are producing. When each string has its own MPPT, and as long as the bypassed portion does not drop the string voltage below efficient MPPT operation, there isn't going to be much difference between a partially shaded string and a partially shaded panel optimized system. Scare tactics claiming 50% loss on string systems are frequently unfounded.

Whether or not it makes economic sense to install panels that receive significant shade is a different question. The answer to that depends very much on what the total electricity cost would be with and without those panels, accounting for the upfront cost to install them... beyond the scope of the discussion in this thread so far.

I have no problem believing what you write here. The nature of the resource and what influences it is variable and unpredictable, I'm surprised actual output estimates and actuals are as consistently close as they often seems to be. This business is not like measuring input/output from, say, a gas turbine. Even under "clear" skies, things such as irradiance can change from minute to minute, not to mention other forcing variables like temp, wind, changing reflection patterns, ...

In spite of what folks think (or do not consider, or know), most "clear" days are not created equal, even consecutive days. Add to that the less than professional grade and accuracy of most instruments used, usually with accuracy no better than a few % if kept in calibration, and the idea that shading is variable, not only from min./min., day/day and season/season, but that shading will change the temp. profiles of an array.

For example, I'll hang my hat on my 17.1 kWh est, but I'm under no illusions that it's any better than, say, 5-10% or more off from actual output for that day. For starters, I've no knowledge of the actual array temps. or amb. temps. as affected by wind vector. Also, while 11/15 was a nice day, I'm pretty sure it varied some from my clear day model. That modeled clear day data is also and almost certainly different than TMY data, and could account for several % error by itself. Also, the array fouling is unknown. I SWAG'd 3 %. Recent rains could lower that, or morning dew could turn dust to mud and make things worse. Or ???

Too many things that can and do affect array output are unmeasured, or if measured at all, usually done so done with instruments and methods whose accuracy does not warrant the implied precision. Even the high $$ precision Eppley instruments drift around a few % with the common precision pyranometers drifting around by a % or 2 over a day.

Sort of a bottom line on panel output and shading for me: It's been my experience that most arrays where I've recorded output from several cooperative neighbors have almost always and consistently performed above expected output, usually by several %. My estimate for my array is about 4-6% lower than actual output in terms of a capacity factor on a consistent basis (Array A.C. output/P.O.A. input for both instantaneous (irradiance) and time integrated (insolation) values).

Sensij: I'm not surprised by your #'s. Seems about right - get the Davis dude ! (Float a loan or sell some plasma or something).

More anecdotally, late afternoon shading for my stuff, for which I calc'd a penalty of between 2-5 % or so depending on season, seems to be much less than that. Some day, I'll stop being so lazy and get serous about looking at that. At this time, I'd SWAG shading is about 1/3 to 1/2 as bad as I'd have thought or as SAM suggested. I've got a string inverter.

All that is something I mostly ascribe to intentionally conservative design by mfg's. which leads to happier customers and larger arrays, but works against cost effective sizing efforts. My prior life as a peddler showed me that's a pretty universal situation in most energy producing capital goods markets. No one I know of ever got fired for oversizing or selling too much equipment.

BTW, I'd again note that Sensij's array and mine have similar orientation and about equal kWh/kW production, at least on sunny days, even though I've got S.P. and he doesn't. Additionally, I suspect I've probably got a bit more insolation. So much for the S.P. "most efficient". claim, at least as best as I can figure it.


This whole measurement business is nowhere near as precise as some folks want to make it. A corollary to Murphy's law says constants aren't and variables won't. Some things in this solar business are intractable. As long as there are more variables than instruments to measure them, and thus not enough equations to deal with and estimate the unknowns to the limits of instrument accuracy, things will continue to be somewhat uncertain. Luckily, the available accuracy is usually sufficient for the task at hand given the biggest long term variable: weather.

Take what you want of the above. Scrap the rest.

[FFE]

With respect to the influence of array orientation, here is some PVWatts data, using Van Nuys TMY3 and looking at just the average daily November plane of array insolation.

225 deg azimuth, 15 deg tilt = 4.39 kWh / sq m / day
240 deg azimuth, 18.5 deg tilt = 4.26 kWh / sq m / day

These data are based on "typical" weather, not just clear sky, so actual weather patterns during the month (morning or evening clouds) will move the numbers around some and could skew more in favor of south facing or west facing. I don't have clear sky modeling handy to post what the weather independent orientation difference would be, which would be the more germane comparison for a day like 11/16.

Wow, I am surprised that the two above numbers are so similar. I would have thought that the 15 degree difference in azimuth would have made a bigger difference. I am going to be able to produce some significant power in summer. I was hoping that someone had clear sky modeling software or a link handy. But, real world data with the above information for comparison is probably better.

Thank you everyone for your help. I will edit the video with an estimated range from 20-22.3 and start a new thread.

[sensij]

For one approach to clear sky modeling of POA irradiance on a tilted plane, similar to what J.P.M. has described, check this out:

http://rredc.nrel.gov/solar/models/spectral/SPCTRAL2/

To calculate insolation over the day would require iterating for each hour (or sub-hour) and totaling the values.

[J.P.M.]

Wow, I am surprised that the two above numbers are so similar. I would have thought that the 15 degree difference in azimuth would have made a bigger difference. I am going to be able to produce some significant power in summer. I was hoping that someone had clear sky modeling software or a link handy. But, real world data with the above information for comparison is probably better.

Thank you everyone for your help. I will edit the video with an estimated range from 20-22.3 and start a new thread.

Clear sky diff. in est. output for 225 vs. 240 deg. az. at 18.4 deg. tilt is about the same as Sensij suggests, maybe something like 5-6% or so greater for the 225 az.

My swag for your clear sky production in Nov. gives about ~ 15-16 % increase in system output over my SWAG for Nov. output using TMY values. Year long the diff. is about the same. That's pretty typical for most locations in So. CA that I've done, using clear sky vs. TMY values.