Temperature Considerations in PV Design

[electrofelon]

What I am saying is that we are already assuming 1000W/m^2 irradiance. We are not allowed to assume an irradiance of less than that to justify longer strings.

Sure, I am not saying the code gives an allowance or direction for considering less than 1,000w irradiance, I'm speaking more in practical real world terms. I forget what the current wording is but the code is generally vague on how exactly to determine highest VOC. Doesn't it say something like lowest "expected" temperature? I think that is the root of the whole discussion what exactly that means. I guess I am just saying I personally am okay using the radiance to give me that little bit of overhead where temperature might be a hair lower than "expected"

 

[ggunn]

Sure, I am not saying the code gives an allowance or direction for considering less than 1,000w irradiance, I'm speaking more in practical real world terms. I forget what the current wording is but the code is generally vague on how exactly to determine highest VOC. Doesn't it say something like lowest "expected" temperature? I think that is the root of the whole discussion what exactly that means. I guess I am just saying I personally am okay using the radiance to give me that little bit of overhead where temperature might be a hair lower than "expected"

I still don't know what you mean. For me the calculations are cut and dried. Maximum Voc is module Voc + (beta)(Tmin - STC), where Tmin is the ASHRAE minimum for the nearest weather station or an interpolated number between more than one station. It assumes an irradiance of 1000W/m^2. Whatever difference in Voc there may be for different irradiance values is irrelevant in the design process.

Irradiance certainly influences power, of course, and I do consider irradiance when I am determining the DC:AC ratio of a system. I would make it higher, for example, for a low slope west facing rooftop system in a northern latitude than I would for an optimally oriented (tilted to match latitude and facing due south with no shading) ground mounted system in a southern latitude.

 

[wwhitney]

I still don't know what you mean.

If I understand correctly, Voc is a decreasing function of temperature and an increasing function of irradiance. If so, you can calculate the max Voc by simultaneously using the max irradiance (STC) and the min temperature. But if those conditions can't be contemporaneous, i.e. min temperature only occurs at dawn, when irradiance is near zero, then that is (perhaps only slightly) overly conservative.

On the other hand, it would make sense to me to use the expected min temperature for the entire project lifetime. So using instead the expected min temperature for a single year would be (perhaps only slightly) under conservative.

And so electrofelon was asking whether those two errors are perhaps of comparable magnitude, providing some comfort for the under conservative choice of single year expected min temperature.

Cheers, Wayne

 

[ggunn]

And so electrofelon was asking whether those two errors are perhaps of comparable magnitude, providing some comfort for the under conservative choice of single year expected min temperature.

Do you know the mathematical correlation between irradiance and Voc? It isn't a number published on module data sheets and I don't know of any publications that discuss it. It's a non-issue for the design process, though. I would not, for example, fudge my maximum Voc numbers to enable longer strings if I expect the maximum irradiance to be less than 1000W/m^2.

 

[ggunn]

Do you know the mathematical correlation between irradiance and Voc? It isn't a number published on module data sheets and I don't know of any publications that discuss it. It's a non-issue for the design process, though. I do not, for example, fudge my maximum Voc numbers to enable longer strings if I expect the maximum irradiance to be less than 1000W/m^2.

...even though the NEC allows for calculating maximum Voc differently per 690.7(A)(3) and Informational Note 2.

 

[electrofelon]

Do you know the mathematical correlation between irradiance and Voc?

It's a bit hard to find a high resolution graph. Looking at some module data sheets more closely, the value is a little less than I assumed when I said "a couple volts". Looks like it would be about 1.5 volts going from 1000w to 400w per sq meter. I just claim that is not insignificant and perhaps not worth ignoring in certain situations. Say maybe when you come up with 17.8 modules per string , that could make you feel okay with just bumping that up to 18, if it would be significantly beneficial to have strings of 18 instead of 17. That could be 30 volts of extra overhead on a 1000 volt string system.

What exactly is the ASHRAE minimum? I don't know if I have gotten an answer on that.

 

[ggunn]

Say maybe when you come up with 17.8 modules per string , that could make you feel okay with just bumping that up to 18...

Show me where the NEC enables you to do that. "Feeling OK" about maximum string length and voltage limits is not in the code.

 

[electrofelon]

Show me where the NEC enables you to do that. "Feeling OK" about maximum string length and voltage limits is not in the code.

It does not, and certainly if you are getting a high degree of oversight from a plan reviewer or whatnot that may not be an option (although perhaps there is some fudge room depending on how the temperature is selected). Good engineering means taking all factors into considerations into account and a decision always has to be made on the conservativeness of a design and likelihood of extreme circumstances.

 

[j-rts]

What exactly is the ASHRAE minimum?

IMO, this link is a better ASHRAE data source than the FSEC. Choosing a station (Flagstaff Pulliam, for example) and then scrolling down, you can see the Extreme Annual Design Conditions:
Looking at this table, three temperatures stand out to me as possible choices for "minimum" temperatures. Mean Min =-20.4; Min, n=50 years = -27.4; or (Mean Min) x (3xSTD) = -28.5.

Do you know the mathematical correlation between irradiance and Voc?

Compare Voc at STC (1000W/m2) to Voc at NOTC (800W/m2). If we use the Silfab 520W commercial panel as an example, the Voc at STC is 47.52, vs 43.71 at NOTC. Temp coeff is -0.24. If you look at the voltage rise using Voc at STC and -20.4 vs Voc at STC and -28.5, they differ by 0.9V. If you instead look at the voltage rise using Voc at STC and -28.5 vs Voc at NOTC and -28.5, they differ by 4.2V. Which is to say, it seems like Voc is roughly 4x more strongly dependent on irradiance than it is on temperature.

Say maybe when you come up with 17.8 modules per string , that could make you feel okay with just bumping that up to 18

Maybe a better way of saying this would be to say - since it's not clear what minimum temperature to use (especially when the closest ASHRAE station is not that close to our site), we have some leeway to decide what the minimum temperature should be. Let's use that leeway to pick a temperature that allows string lengths of 18 instead of 17.

 

[ggunn]

It does not, and certainly if you are getting a high degree of oversight from a plan reviewer or whatnot that may not be an option (although perhaps there is some fudge room depending on how the temperature is selected). Good engineering means taking all factors into considerations into account and a decision always has to be made on the conservativeness of a design and likelihood of extreme circumstances.

I design systems to code whether or not my work is subject to "a high degree of oversight". YMMV.

 

[tallgirl]

It's a bit hard to find a high resolution graph. Looking at some module data sheets more closely, the value is a little less than I assumed when I said "a couple volts". Looks like it would be about 1.5 volts going from 1000w to 400w per sq meter. I just claim that is not insignificant and perhaps not worth ignoring in certain situations. Say maybe when you come up with 17.8 modules per string , that could make you feel okay with just bumping that up to 18, if it would be significantly beneficial to have strings of 18 instead of 17. That could be 30 volts of extra overhead on a 1000 volt string system.

What exactly is the ASHRAE minimum? I don't know if I have gotten an answer on that.

Vmpp changes with current based on the series resistance used in the cell model, but it is primarily a temperature issue since the internal resistance of the cell changes with temperature. That’s why Voc increases as temperature decreases.

 

[wwhitney]

What exactly is the ASHRAE minimum? I don't know if I have gotten an answer on that.

See post #13, I answered your question.

Looking at this table, three temperatures stand out to me as possible choices for "minimum" temperatures. Mean Min =-20.4; Min, n=50 years = -27.4; or (Mean Min) x (3xSTD) = -28.5.

Unless your project is going to last 50 years, I don't see any reason to use a 50 year number. But the mean min and mean 50 year min are both expected values; there's a 50% chance the actual low would be lower over the year or 50 year period.

So if your project lifetime is 20 years, say, and you want to be very conservative, I'd suggest something like a 1% likelihood 20 year extreme low, where there's a 1% likelihood of getting a lower temperature in any 20 year period. [Or you could pick 5% or 10%, your choice.] The ASHRAE data seems to indicate that the annual minimums are normally distributed.

In which case the 20 year 1% likelihood would be the mean extreme annual low plus 3.29 times the standard deviation. Your suggestion of taking the mean plus 3 times the standard deviation would give you the 2.7% likelihood 20 year extreme, or if your project lifetime is 30 years, that's the 4% likelihood 30 year extreme.

Cheers, Wayne

 

[jaggedben]

If I understand correctly, Voc is a decreasing function of temperature and an increasing function of irradiance. If so, you can calculate the max Voc by simultaneously using the max irradiance (STC) and the min temperature. But if those conditions can't be contemporaneous, i.e. min temperature only occurs at dawn, when irradiance is near zero, then that is (perhaps only slightly) overly conservative.

If I look at microinverter data I see the microinverters 'wake up' as much as an hour before dawn and start reporting a Voc of 25-50% of Voc @ STC. Maybe half an hour later they reach start voltage and start producing a watt or two of power near Vmp, suggesting that the irradiance perhaps needs not to even reach as much as 10 w/m^2 to get substantially close to Voc @ STC. In other words the Voc to irradiance curve is extremely steep at very low irradiance and then transitions to a (comparatively) extremely low slope thereafter.

That said, you need some irradiance for voltage, so I believe it is correct to say that if min temp always occurs in the darkness of night then max irradiance and min temp are extremely unlikely to be contemporaneous. The moon (or even, say, cloud reflected light from human sources) is only a small fraction of a watt/m^2. So, for example, if one had temperature data that excluded all times post-dusk to pre-dawn, I believe one would be justified and code compliant to use that.

 

[wwhitney]

Maybe half an hour later they reach start voltage and start producing a watt or two of power near Vmp, suggesting that the irradiance perhaps needs not to even reach as much as 10 w/m^2 to get substantially close to Voc @ STC.

While that all sounds fine, with a beta of -0.24 (from the example in post #29), if "near Vmp" and "substantially close to Voc" mean off by 2V, that's the equivalent of 8 degrees F. So still comparable to the calculation effect from different possible choices of minimum design temperature.

Cheers, Wayne

 

[jaggedben]

While that all sounds fine, with a beta of -0.24 (from the example in post #29), if "near Vmp" and "substantially close to Voc" mean off by 2V, that's the equivalent of 8 degrees F. So still comparable to the calculation effect from different possible choices of minimum design temperature.

I'm not sure how that speaks to the point. Which I'll summarize as saying that we ought to be able to ignore ambient temps when it's actually dark.

 

[wwhitney]

I'm not sure how that speaks to the point. Which I'll summarize as saying that we ought to be able to ignore ambient temps when it's actually dark.

That's certainly true. But my last comment refers to electrofelon's question, as detailed in post #23.

Cheers, Wayne

 

[tallgirl]

If I look at microinverter data I see the microinverters 'wake up' as much as an hour before dawn and start reporting a Voc of 25-50% of Voc @ STC. Maybe half an hour later they reach start voltage and start producing a watt or two of power near Vmp, suggesting that the irradiance perhaps needs not to even reach as much as 10 w/m^2 to get substantially close to Voc @ STC. In other words the Voc to irradiance curve is extremely steep at very low irradiance and then transitions to a (comparatively) extremely low slope thereafter.

That said, you need some irradiance for voltage, so I believe it is correct to say that if min temp always occurs in the darkness of night then max irradiance and min temp are extremely unlikely to be contemporaneous. The moon (or even, say, cloud reflected light from human sources) is only a small fraction of a watt/m^2. So, for example, if one had temperature data that excluded all times post-dusk to pre-dawn, I believe one would be justified and code compliant to use that.

Voc is caused by the forward voltage drop (I think that’s correct - haven’t thought about it in a bit and I just finished my lunch) of the ideal diode which is a solar cell. So, Voc is a fundamental physical (physics, not stuff you can touch) property of the cells.

Once enough photons with enough energy - proper wavelength start to strike any of the junctions, electrons begin to move and the voltage is always somewhere around Voc.

The equivalent circuit of a solar cell is a doide, in parallel with a resistor, and a series resistor in the output. Since V = I * R, a low irradiance and low resultant current, Vdrop is very low, so Voc will be close to whatever the name of that voltage drop is across the junction.

I’m sure someone will now tell me I’m wrong. I do not care. My cheese soup was delicious and so was the carrot cake.

 

[jaggedben]

Right, but how many photons are enough?

 

[ggunn]

So, for example, if one had temperature data that excluded all times post-dusk to pre-dawn, I believe one would be justified and code compliant to use that.

Per what article of the NEC?

 

[jaggedben]

Per what article of the NEC?

690.7(A)(1) and (2) refer to methods employing the "lowest expected ambient temperature" without further defining. The ASHRAE suggestion is an informational note which is not strictly binding. I suppose if you want to be a stickler, it doesn't say "lowest expected ambient temperature between dawn and dusk", but I think if we are debating how extreme a minimum temperature we want to design for, that could be taken into account if one is conservative with other factors. The code is not otherwise specific on how you choose the temperature.

I don't know if there are any industry standard methods (see 690.7(A)(3)) for doing this for systems over 100kW that use my suggestion, but that is potentially another option.