Temperature Considerations in PV Design

[wwhitney]

What computations require knowing a maximum and minimum ambient temperature? I'm thinking max Voc on the DC string requires maximum temperature, maximum Isc on the DC string requires minimum temperature, and conductor ampacity requires maximum temperature.

If you use ASHRAE historical data for a given location, you can get numbers like the mean 1% high temperature (1% of the time it was hotter, so for 87.6 hours per year), the mean 0.4% high temperature (35 hours per year it was hotter), the mean annual extreme (50% of the years the record high for the year was hotter, 50% of the years it was lower), or you can compute a likelihood of exceedance at any level for any length of time (e.g. a 10% chance that the highest temperature seen over 20 years exceeds the value, and a 90% chance it doesn't).

Which of those temperature levels would you use?

And lastly, since we've had a warming trend for the last 100+ years, if the ASHRAE data is purely historical (is that right?), when designing a project with an expected lifetime of 20 years, do you make a prediction about how much warmer things will be in 20 years, and use that prediction for determining the maximum temperature to use? If so, what source do you use for the future temperature prediction?

Cheers, Wayne

 

[ggunn]

What computations require knowing a maximum and minimum ambient temperature? I'm thinking max Voc on the DC string requires maximum temperature...

You have it backwards. Minimum temperature makes for maximum Voc. Temperature differential on Isc rarely makes any difference to the design.

 

[wwhitney]

You have it backwards. Minimum temperature makes for maximum Voc.

Thanks for the correction. So how do you choose what temperature to use?

Cheers, Wayne

 

[ggunn]

Thanks for the correction. So how do you choose what temperature to use?

Cheers, Wayne

I use the ASHRAE extreme minimum temperature and the beta number (volts or percentage of Voc per degree C below 25 degrees C) on the module data sheet to determine the maximum voltage, and then I divide the maximum input voltage for the inverter by that voltage, rounded down to the nearest integer, to get the maximum string length.

 

[wwhitney]

I use the ASHRAE extreme minimum temperature

ASHRAE publishes mean extreme annual minimum temperature and the standard deviation of that, so you can come up with an extreme temperature for any time period and any likelihood. Are you just using the 50% 1 year extreme minimum?

Cheers, Wayne

 

[ggunn]

ASHRAE publishes mean extreme annual minimum temperature and the standard deviation of that, so you can come up with an extreme temperature for any time period and any likelihood. Are you just using the 50% 1 year extreme minimum?

I use this: https://energyresearch.ucf.edu/solar-certification/solar-reference-map/ and the extreme min for the nearest weather station. If there are a few stations about the same distance away, I look at them all and make a judgement call if they show different numbers for Tmin.

 

[electrofelon]

I am not too worried about it. I admit I may be less conservative than others. An ultra extreme low temperature is nearly impossible to occur during high irradiance. Also it is not like the equipment is going to explode at 600.000001 volts (or whatever the rating is).

 

[ggunn]

I am not too worried about it. I admit I may be less conservative than others. An ultra extreme low temperature is nearly impossible to occur during high irradiance. Also it is not like the equipment is going to explode at 600.000001 volts (or whatever the rating is).

Voltage isn't very dependent on irradiance; early on a cold morning when the inverter is off and the cell temperature is near where it was just before dawn is when the DC voltage will typically be at its highest. Also some (many? all?) inverters have a register where they store the highest DC voltage that the inverter has experienced, and if it is higher than the published maximum that the inverter can tolerate, any subsequent warranty issues can be made more complicated.

 

[jaggedben]

I believe voltage is dependent enough on irradiance that I think we can say that if the design temperature only ever occurs in the dead of night then the calculated max Voc won't ever occur. That is, you need more irradiance than the moon to get full Voc. (But perhaps the design temperature will be seen at dawn. That's where the judgment call comes in.)

I think electrofelon's other point was that while it may be a warranty issue if the equipment sees 601V when barely any power is available, it's not a saftey issue. As a residential installer I never spent over a residential inverter's equivalent of time worrying about how I picked my design temp. If you're designing a utility scale project you might want to spend a residential inverter's equivalent of time on protecting your much more expensive inverter.

 

[ggunn]

I believe voltage is dependent enough on irradiance that I think we can say that if the design temperature only ever occurs in the dead of night then the calculated max Voc won't ever occur. That is, you need more irradiance than the moon to get full Voc. (But perhaps the design temperature will be seen at dawn. That's where the judgment call comes in.)

I think electrofelon's other point was that while it may be a warranty issue if the equipment sees 601V when barely any power is available, it's not a saftey issue. As a residential installer I never spent over a residential inverter's equivalent of time worrying about how I picked my design temp. If you're designing a utility scale project you might want to spend a residential inverter's equivalent of time on protecting your much more expensive inverter.

Actually, what I do is pretty simple. I consult the ASHRAE numbers and use the min temp with the beta value for the module to set my string length so that the thus adjusted string Voc does not exceed the inverter's maximum allowed DC voltage. What I don't spend time worrying over is the likelihood and under what conditions the maximum Voc I calculated will ever be seen by the inverter.

 

[electrofelon]

Voltage isn't very dependent on irradiance; early on a cold morning when the inverter is off and the cell temperature is near where it was just before dawn is when the DC voltage will typically be at its highest. Also some (many? all?) inverters have a register where they store the highest DC voltage that the inverter has experienced, and if it is higher than the published maximum that the inverter can tolerate, any subsequent warranty issues can be made more complicated.

Right, VOC is not "very" dependent on irradiance, but if it's a couple volts per module, you are at a couple dozen volts for a string and
That's easily enough to not stress out about finding or using a 100 year or record low or something like that that. The numbers I use here in central NY are "damn cold" but not records. Sure there is a chance they could see 10-15 degrees lower in their lifetime, but whatever, lack of irradiance will cover that, it will be fine

 

[electrofelon]

I use the ASHRAE extreme minimum temperature.

So what exactly are "extreme min" and mean extreme min". And how do they compare with the record low temperature?

 

[wwhitney]

So what exactly are "extreme min" and mean extreme min". And how do they compare with the record low temperature?

Pretty sure that "extreme min" = "mean extreme min" = average of the annual mins over the observation period.

ASHRAE publishes mean annual extreme and standard deviation of annual extreme. So if you assume that the annual extremes are normally distributed, you can calculate say a "10% likelihood 20 year extreme", where there's a 10% chance that the 20 year extreme will be more extreme, a 90% chance it will be less extreme. Or any other choice of likelihood and time period.

Cheers, Wayne

 

[electrofelon]

Pretty sure that "extreme min" = "mean extreme min" = average of the annual mins over the observation period.

ASHRAE publishes mean annual extreme and standard deviation of annual extreme. So if you assume that the annual extremes are normally distributed, you can calculate say a "10% likelihood 20 year extreme", where there's a 10% chance that the 20 year extreme will be more extreme, a 90% chance it will be less extreme. Or any other choice of likelihood and time period.

Cheers, Wayne

Perhaps it varies by geographic area, but at least here it seems kind of like splitting hairs. In other words, here, I don't think there is much difference between any of these figures. A quick Google said the absolute record low here was -34 set in 1934. That record low isn't really that much colder than it generally gets. Just going by memory, I would say every 3 years it gets down to -15, maybe -20, so 14° from the all time record low, not very significant in terms of PV voc is it?

It's been a while since I have had to think about being anywhere near to Max VOC, but just as an example of my philosophy, I would probably do a calc at negative 10 and let the lower irradiance give me a dozen or so volts to cover the extremely unlikely chance of a lower temp.

 

[PWDickerson]

Find your extreme min temp here:

Solar Reference Map - FSEC®

[Last updated: 4/28/2023]

energyresearch.ucf.edu

Take the difference from 25C
Multiply by your TCVoc found in module datasheet
Add your Voc in datasheet to the result
Devide this value into your max string voltage
Done!

Other than consider temperature corrections for conduit fill, that is all I do. I used to pay attention to typical operating voltage of the strings to make sure it was within the inverters MPPT range, but that is not really an issue with modern inverters any more.

 

[wwhitney]

Find your extreme min temp here:

Solar Reference Map - FSEC®

[Last updated: 4/28/2023]

energyresearch.ucf.edu

You're the second person to link to that page. It says "NOTE: This page uses outdated ASHRAE weather data. We are working to get the data updated and permissions from ASHRAE."

Since it uses ASHRAE data, pretty sure the extreme min temp is the average minimum for a year.

Cheers, Wayne

 

[pv_n00b]

We usually use SAM to find the Vmax and Imax for a system. I don't recommend using the Solar Reference Map site. It has not been updated for a long time. It's only still up because FSEC hosts it.

 

[ggunn]

Right, VOC is not "very" dependent on irradiance, but if it's a couple volts per module...

I don't think it is, but Voc and Isc are based on Standard Test Conditions (STC) which assume 1000W/m^2 irradiance.

 

[electrofelon]

I don't think it is, but Voc and Isc are based on Standard Test Conditions (STC) which assume 1000W/m^2 irradiance.

You don't think it's a couple volts (per module) at say 500 vs 1000?

 

[ggunn]

You don't think it's a couple volts (per module) at say 500 vs 1000?

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.