High cost of preminum panels worth it?

[kb58]

I know the general rule is "no" but I thought I'd work through the numbers because one factor made me wonder: thermal derating.

For "fun" I picked two panels of similar output, with the main difference being their thermal derating. For some of the following calculations I didn't have hard numbers and guessed, so feel free to correct me.

The contenders:

Panel A: Silfab Solar SLG-345M, $267, thermal derating -.38%/C
Panel B: Panasonic VBHN330SA16, $398, thermal derating -.26%/C
Price difference, $131

Assumptions:
Panel runs 50C above room temp
Runs this way 4 hrs a day, for four months of the year.
Average power cost of $0.25/KWh

Power output at 25C:
Panel A: 345W
Panel B: 330W

Power derating due to 50C rise:
Panel A: 0.38% * 50 * 345 = 66W loss
Panel B: 0.26% * 50 * 330 = 43W loss

Power loss over time:
Power loss difference, 66 - 43 = 23W
0.023KW * 4hrs/day * 30 days * 4 months = 9.6KWh a year of lost power

Assuming an average cost of $0.25/KWh for electricity, that's a loss of roughly $38 a year (wrong: is actually $2.40)

Panel B costs $131 more than Panel A, so it'll take $131 / $2.40, or roughly 54.6 years to break even on, which means it's not worth it for thermal performance alone (granted, there are other parameters which may still be better than bargain basement panels).

[kb58]

Thinking about it, 50C may be too much, but I'll let the numbers stand and see what others say...

[J.P.M.]

Use PVWatts and skip your assumptions. The model's output will be a better representation than what you're using.
However, forget PVWatts cost numbers, they're bogus. Do your own using utility tariffs.

More later. .

[sensij]

You can use NREL's System Advisor Model to run the same kind of analysis as PVWatts, except accounting to the the thermal coefficient difference you'd noted. There is learning curve to it, but I can take a look later and see how far off your numbers look.

[kb58]

That NREL software looks very impressive, I'll check it out.

[J.P.M.]

That NREL software looks very impressive, I'll check it out.

SAM's a pretty good model, and I use it a fair amount, but I'd respectfully suggest starting with PVWatts and learning to walk before you run. Actually, SAM is somewhat analogous to PVWatts on steroids. They use similar information. SAM probably gets a bit closer to reality, but as Sensig notes, the learning curve may be a bit steep. PVWats first can flatten that out a bit. SAM's documentation can also be very informative, but not for the technically faint of heart.

[kb58]

Understood, but the site has pretty serious morning shading issues which I don't know how to model in PVWatts. Other than a "% shading" field, it doesn't seem to have a way to tell it where that shading is, but I'll look more carefully.

[J.P.M.]

That NREL software looks very impressive, I'll check it out.

SAM's a pretty good model, and I use it a fair amount, but I'd respectfully suggest starting with PVWatts and learning to walk before you run. Actually, SAM is somewhat analogous to PVWatts on steroids. They use similar information. SAM probably gets a bit closer to reality, but as Sensig notes, the learning curve may be a bit steep. PVWats first can flatten that out a bit. SAM's documentation can also be very informative, but not for the technically faint of heart. But, SAM will do a pretty fair job of estimating shading. Again just be sure you know how to make it run right and know what it's telling you.

[J.P.M.]

Understood, but the site has pretty serious morning shading issues which I don't know how to model in PVWatts. Other than a "% shading" field, it doesn't seem to have a way to tell it where that shading is, but I'll look more carefully.

PVWatts won't do shading. SAM will but as I suggested, it's not for the uninitiated without risk of misinterpreting output (or input for that matter).

Google solar shade analysis tools for other methods.

Quite honestly, with a bit of familiarity with solar motion and solar geometry, I got within a few % of SAM and a solar pathfinder. It ain't rocket science.

[sensij]

That NREL software looks very impressive, I'll check it out.

I tinkered with SAM to see how it would answer your question.

Some inputs:
Solar resource: TMY3 format created from the NSRDB using San Marco as the location
Array: 24 panels (4 strings of 6 for Panasonic, 3 strings of 8 for Silfab)
Inverter: SMA 7.7 (not too important here)
Array design: Fixed ground mount, 220 deg azimuth, 30 deg pitch

The specific panels listed in the first post are available in the CEC database, so I used those.

The CEC parameters for those panels indicate

Panasonic temp coefficient of power : -0.310% / deg C
Silfab temp coefficient of power: -0.497% / deg C

Panasonic NOCT: 43.8
Silfab NOCT: 47.3

These numbers are slightly different than what is listed on the mfg's datasheet, but still demonstrate the point.

The difference in NOCT helps make up some of the 15 W delta in panel rating. It suggests that under identical NOCT conditions, the Panasonic panel will run 3.5 deg cooler.

The model output for DC energy is 14,516 kWh for the Panasonic system (1.833 kWh / W), and 14,729 kWh for the Silfab system (1.779 kWh / W), so between the temp coefficient and the NOCT, the 4.5% higher DC rating of the Silfab panel results in only 1.5% more energy. (AC energy would include more sources of losses, including the inverter)

In terms of dollars
24 * $398 = $9552, so 1.52 kWh / $ for the Panasonic system
24 * $267 = $6408, so 2.30 kWh / $ for the Silfab system

For more detail...

Here is what the histogram of cell temperature calculated by SAM looks like (4224 hours of daytime in the year)

kb58 temp.JPG


And here is the histogram of DC power (of a single panel)

kb58 power.JPG

[JSchnee21]

Wow, thanks sensij. That's a great analysis. While it makes me cry to hear it, Panasonic, or any "premium" panel gets a whupping despite MFG claims of special performance features (better thermals, textured cells, special glass, etc.). Not to mention the fact that given the pricing you've outlined one could purchase 11 additional Silfab panels increasing the size of your array by 45% (assuming one has room to do so). Lowest panel cost per watt (assuming the panel isn't crap) really is the way to go. Of course the Silfabs are 72-cell, so they're going to take up quite a bit more space as is for the same DC array size.

[kb58]

That's pretty fantastic sw, thank you very much for showing its abilities. On the one hand I'm surprised that Trise is only ~10C but on the other hand, if it was otherwise, premium panels would have more of a foothold here. I played around with SAM a little and understand having to step lightly to get good data.

As for the big picture, I'm also going to improve things the old-fashioned way, conservation. The main pond pump and UV lamp will be put on a timer to turn them off between 4-9pm. That'll save roughly 0.5KWh, which adds up. The biological filter aeration and pond returns will be left enabled to ensure good oxygenation, but they don't take nearly as much power.

My end goals is to determine both the required panel count, and what I can expect for a yearly electric bill as a result - assuming TOU 2.0. To zoom out even further, I guess the ultimate goal is to see whether through some reasonable conservation, can we achieve a low enough bill that solar ends up being more trouble than it's worth, and right now I don't know.

(EDIT: This is now way off topic... I should probably start a "blog" thread on our installation)

[sensij]

That's pretty fantastic sw, thank you very much for showing its abilities. On the one hand I'm surprised that Trise is only ~10C but on the other hand, if it was otherwise, premium panels would have more of a foothold here. I played around with SAM a little and understand having to step lightly to get good data.

Well, you are pursuing ground mount, so the Trise is more limited. The NOCT model to calculate Tcell attempts to account for the mounting clearance. I didn't check the sensitivity to the other mounting options, but you can see that if you were to mount them with very little clearance to a rooftop, the Trise would be much more under high irradiance conditions. NOCT.JPG

[max2k]

Wow, thanks sensij. That's a great analysis. While it makes me cry to hear it, Panasonic, or any "premium" panel gets a whupping despite MFG claims of special performance features (better thermals, textured cells, special glass, etc.). Not to mention the fact that given the pricing you've outlined one could purchase 11 additional Silfab panels increasing the size of your array by 45% (assuming one has room to do so). Lowest panel cost per watt (assuming the panel isn't crap) really is the way to go. Of course the Silfabs are 72-cell, so they're going to take up quite a bit more space as is for the same DC array size.

Well, like with any model it makes certain assumptions and reflects only portion of what is going on. For example, how these 2 panels age? It wouldn't be so great anymore if Silfab loses its performance significantly faster than Panasonic or would require replacement somewhere after 10 years in service due to cheaper sealant used. That's not to say I recommend either I only trying to be level headed about this. From installer's point of view Silfab option is no brainer based on this analysis: higher initial output, more competitive price and if it fails it would lead to new sales.

[SunEagle]

Well, like with any model it makes certain assumptions and reflects only portion of what is going on. For example, how these 2 panels age? It wouldn't be so great anymore if Silfab loses its performance significantly faster than Panasonic or would require replacement somewhere after 10 years in service due to cheaper sealant used.

I would expect you would see more production losses due to dirt or foreign matter build up on the panel surface then from cell degradation.

Then again cheap cells having a shorter life then A grade cells is probably a safe assumption.