Arguments
Sabin 33 #12 - Do solar panels work in cold or cloudy climates?
Posted on 21 January 2025 by BaerbelW
On November 1, 2024 we announced the publication of 33 rebuttals based on the report "Rebutting 33 False Claims About Solar, Wind, and Electric Vehicles" written by Matthew Eisenson, Jacob Elkin, Andy Fitch, Matthew Ard, Kaya Sittinger & Samuel Lavine and published by the Sabin Center for Climate Change Law at Columbia Law School in 2024. Below is the blog post version of rebuttal #12 based on Sabin's report.
Solar panels generate energy even in cloudy or cold conditions1 (also Ramli et al. 2016). Although cloudy weather may reduce power generation by as much as 45%, substantial energy can still be generated during those conditions (Ramli et al. 2016 and Makrides et al. 2012).
Furthermore, in most instances, cold temperatures do not reduce electricity output at all—and actually increase solar panel efficiency by increasing voltage2 (also Sarmah et al. 2023). Crystalline silicon cells, which comprise approximately 84% of the U.S. market, and cadmium telluride cells, which comprise approximately 16% of the U.S. market, actually perform better in colder weather2. Only amorphous silicon cells, which represent a negligible percentage of the U.S market, experience decreased performance in colder temperatures.
Footnotes:
[1] What happens to solar panels when it's cloudy or raining?, Solar Energy Industries Association (last visited March 25, 2024)
[2] Solar Photovoltaics: Supply Chain Deep Dive Assessment, U.S. Dep’t of Energy, Feb. 24, 2022, at iii
Skeptical Science sincerely appreciates Sabin Center's generosity in collaborating with us to make this information available as widely as possible.
This is a classic case of the rebuttal omitting key information.
While the statements in the rebuttal are factually correct, key data for understanding the full context is missing ie there remains significant decline of overall electric generation from Solar during the winter months.
The drop in electric genration from solar is well documented from numerous sources and as detailed in the monthly data from EIA, the Dec/January capacity factor for utility solar pv averages around 13.5% vs 30%-31% during June july and august.. Northern latitudes such as north of 45degrees north (minneapolis), the capacity factor during the winter months drops to around 7% -10%. Links to source data is provided below:
www.eia.gov/todayinenergy/detail.php?id=39832
www.spglobal.com/market-intelligence/en/news-insights/research/2022-monthly-us-solar-capacity-factors-underscore-winter-doldrums
www.eia.gov/electricity/monthly/epm_table_grapher.php?t=epmt_6_07_b
Um, the myth doesnt say anything about winter. Just cold and cloudy, and the rebuttal acknowleges the impact of clouds. I was surprized to see that peak output from our panels didnt vary much from summer to winter - the increased efficiency cutting in - but the shorter days and increase in cloud certainly reduce daily output. A very very long way from "dont work" however.
The snow Sunday cut my main solar to essentially nothing probably through Friday. My rooftop panels charge a Generac lithium battery sitting next to my feet. I'm currently charging from AC since it was down to zero this morning. My laptop and two phone use about 20% of the battery per night. My Starlink is normally plugged into AC, but I can switch it to the Generac or my small portable generator when I need to. Those all run the fridge when needed (normally AC).
My rooftop solar (four 100W panels) are about 35 degrees pitch. They are not attached to the roof which is 25 degrees, but attached to pressure treated lumber straddling the peak. A separate panel now about 20 years old was putting out normal power yesterday and charges AGM batteries in the basement. It is mounted at a much steeper pitch, probably about 60 degrees.
The snow on Sunday was wet, followed by a hard freeze that has continued. The wet snow froze to the four panels and even if it slid down during the snowfall, a small amount at the bottom will basically kill power output. I know this because I have two panels on the ground mounted to a large resistor to constantly heat the bathroom on sunny days. They are fine now with a very steep winter angle with no snow at the bottom.
Which brings me to my point. Yesterday morning northern MD line voltage monitored by my friend dipped to 106V as many resistive heaters kicked on and as heat exchangers switched to defrost mode or turned off. There were probably also a higher load of old resistive heaters and space heaters running.
Point 1: maintainiing gas backup is essential.
Point 2: rooftop solar at roof pitch is not ideal for snow. It can work but can be out for days. Even my greater pitch has not helped and I cannot safely walk on the roof to clean off panels. Solar on the ground makes much more sense especially if panel pitch is adjusted for the season which help with both capture and snow sliding off. Utility solar makes the most sense.
Point 3: I see more people using ground mounted home solar instead of rooftop. There is still lots of rooftop solar and more being installed, and it can be ideal in summer for both pitch and matching peak AC demand. But winter still brings these occasional snow to cold challenges. I have no grid tie whatsoever and probably never will. But if I did it would be ground mounted and I would only do it with dynamic pricing and not the current net metering here in Virginia. Net metering is a subsidy from less well-off ratepayers to more well-off solar owners. Those solar owners are supplying valuable electricity at peak summer demand which is normally our highest here in Virginia and south.
But in winter in these unusual (and getting more unusual) situations those solar panels do nothing. They obviously don't detract from the grid, but those homes are currently getting cheap power from very expensive gas peaker plants paid with summer credits. If the credits were generated in similar peak situations then great, but there's no market mechanism to incentivize that in our simplistic net metering system.
Comment 1 is a classic case of David-acct engaging in whataboutism, to distract from the rebuttal that points out that cold is not a problem, and solar panels do continue to generate power even with cloudy conditions (albeit at reduced rates).
As scaddenp points out in comment 2, David-acct is essentially going off-topic. David-acct is attempting to rebut something that isn't in the OP. David-acct acknowledges this in his comment, where he says "...key data for understanding the full context is missing..." David-acct wants to expand the context, to try to make a different point. This is a case of deflection - something politicians are good at when asked a question they don't want to answer.
David-acct's first link includes this graphic:
Wow. It turns out that you get more output from your solar panels where solar irradiance is higher. And solar irradiance is higher in clearer skies. Whodathunk.
Note that the graphic shows "direct normal solar irradiance". This is the strength of solar energy measured pointing an instrument with a narrow field of view (about the size of the sun) directly at the sun. The instrument used for this is the pyrheliometer. It does not include input from any of the rest of the sky (known as "diffuse radiation"). The sum of direct + diffuse gives total solar irradiance.
For any surface (the ground, a solar panel, the side of your house), you need to make a calculation to convert direct normal irradiance as measured pointing directly at the sun to an irradiance value at the orientation of the surface. The surface in question will only get the full "direct normal" value if that surface is pointing directly at the sun - e.g., the sun is directly overhead at that location for a horizontal surface (e.g. the ground), or the surface is tilted to point directly at the sun (a sun-tracking solar panel).
The graphic above does not indicate whether it's daily mean totals are corrected for a horizontal surface, or whether they represent the total available to a tracking system that always points at the sun. The use of the term "direct normal" implies the latter - "normal" is used in the geometric context of "at right angles to", and to stay at right angles to the sun's rays requires that you track the sun. If they mean the former, then they are sloppy in calling it "direct normal" - they should refer to it as "direct radiation on a horizontal surface".
And solar panels to use the non-direct radiation that comes in from the sky at angles other than the direct path to the sun. In a clear sky, this is the blue you see away from the sun, and is a small proportion of the total. In an overcast sky, diffuse radiation can be quite a bit higher than the diffuse radiation in a clear sky. And this is one of the points in the OP: just because it is cloudy does not mean that solar panels produce no power.
Another nuance in all this is that orientation of solar panels is important, to maximize the use of power available from the direct sun:
So, there is lots to consider in siting solar panels. And all of the above is pretty well known to people studying the development of solar energy. (Granted - the details of direct versus diffuse radiation are often not measured at a lot of locations, but that does not mean that the principles are not known.)
...and lots more "context" than David-acct alludes to. I doubt he really understands how to interpret the sources he has linked to. But we're used to that now, based on his history here at SkS. At least this time he did not call it "raw data". But it is clear that - once again - he is simply trying to throw something at the wall in hope that it sticks, to discredit the statements in the OP. Once again, he's not doing a very good job.
Eric @ 3:
Your points about rooftop slope not being optimal for solar panels in some cases relates to some of the principles I mentioned in comment 4. Mounting panels at the same slope as the roof probably reduces installation costs, but may also reduce efficiency. And many houses do not have roofs that are pointed optimally south (or whatever direction is best in that location). In a ground installation, you have more options.
...but many people do not have ground space for large solar arrays, either. Even if you have space, there is the question of sky view factors: how much of the sky is unobstructed? The higher up you can place the panels, the fewer object will cause shade issues (trees, other building, etc.) Rooftop is better in crowded urban environments - but as you state, keeping them clean from debris gets harder. Even large amounts of accumulated dust can reduce panel efficiency if you live in a climate with little rain to wash the panels.
There are always trade-offs.
I have a question for you Bob that is a little off topic, but which the readers may find interesting. It has to do with tradeoffs and generating power when it is most needed.
Specifically, wouldn't it be good to point some panels either West, or Southwest? The point being to maximize power generation in the late afternoon, when power companies are struggling to meet demand. In Minnesota, Time of Day electric rates are about 35 cents/kwh between 4-8 pm, whereas standard rates are 12 cents/kwh during the rest of the day, The 35 cents/kwh varies by the time of year and I'm sure other factors, but the point is that there are times when power is most valuable and most needed.
Do you agree that it makes sense to point some panels either West or Southwest to generate power during the heavy-use times, or do you think it still best to always point them due south and at the optimum lattitude-based angle? My assumption here is that a solar array is only generating a portion of the power needed, and that there is therefore continued reliance on some amount of grid power.
"The drop in electric genration from solar is well documented from numerous sources and as detailed in the monthly data from EIA, the Dec/January capacity factor for utility solar pv averages around 13.5% vs 30%-31% during June july and august.. "
The capacity factor drops by about half. Could you not compensate by installing twice the number of panels? This would increase costs, but for new homes it looks affordable. Costs in New Zealand for residential solar are as follows. "The Electricity Authority estimates the average set up for a 4.4kW installed system with around 10-11 panels to be around $12,000-$13,000". If you doubled the number of panels the cost is $24,000 - $26,000.
In New Zealand average home size is about 150M2, and cost around $4000 M2 to build so around $600,000 to build. Trim just 6M2 squared off the floor area, and you have paid for a solar power instillation capable of dealing with seasonal swings rather than relying on grid backup. I suppose its a question of what people value - the large homes we build now well beyond what is really needed, or a sustainable energy system that also ultimately saves them money.
Yes, on a friends solar installation with thoroughly non-optimal roof, they just put up more panels. The panel cost is now so cheap, (inverters not so much), that just piling on panels makes sense.
Evan @ 6:
On a rooftop installation, you probably end up putting panels on more than one roof section, which would be oriented in different directions. That would even out the power production through the day. But if you have lots of roof surfaces to choose from (more than you want covered in panels), then choosing which ones to use gets challenging.
The time of day question is an interesting one. Where I live, the electricity rates are broken into peak, mid-peak, and off-peak hours, and the time periods are 7pm-7am, 7am-11am, 11am-5pm, and 5pm-7pm.
...so I'd agree there are real possibilities to optimize the solar panel installation to get maximum cost savings. And hopefully the utility company has set rates so that peak rates are when the grid can most benefit from extra power. Maximizing local production during the hot part of the day in summer also means that there is less need for transmission infrastructure, as the power is produced where it is needed for A/C.
The common single-orientation leave-it-alone setup is pointing south, set at "latitude tilt". (Zero is flat. 45° works for 45° latitude. 60° is a lot of tilt, and starts to run into the fact that at 60° latitude the sun is well above the horizon for a lot of the day (and rises in the NE and sets in the NW, so something pointing due south is shaded part of the day!) In high latitudes, a flat panel works best.
Optimizing runs into more detailed calculations that simple rules-of-thumb don't do well at. I've done such calculations at a research site where we ran some instrumentation off solar-powered battery setups. It happened to be a research station where we collected the direct and diffuse radiation measurements needed to do the local optimization. (By coincidence, NW of where you are in Minnesota.)
Where I am now, we considered doing a rooftop solar installation, but in winter our south-facing roof area is partly shaded by the house next to us. Roof geometry is not good (and small yards make ground-based solar impractical).
Thanks Bob for your comments. I always learn a lot from your posts, including the interesting rate structure. Where we live in Minnesota our current rates are 6.4 cents/kwh from 8 pm to 8 am (when we charge our car), 12 cents/kwh from 8 am to 4 pm, and 22 cents/kwh from 4 to 8 pm. The 35 cents I quoted was from a different pricing structure not appropriate for our time-of-day rates optimized for car charging.
Even though we live in the country, our roof is far from optimal. We also don't like the idea of ground-based solar. Not worth going into the details here of why.
So we are planning to hang panels off our deck. We designed our deck to hold 120 psf, because we want to use it as a large, raised-bed garden. The edge where we will hang the panels is supported by an I-Beam. Hence, we have plenty of strength to hang panels off the edge.
We are planning to have a few panels facing due east for morning generation, the bulk of the panels facing south, and a few panels facing west, or southwest, to catch the afternoon sun. The idea is to have some power from sunup to sundown, and the main power during the day.
I like the idea of hanging the panels off the deck, because the panels effectively extend the reach of the deck, providing lots of covered areas to park vehicles. And the panels will be readily accessible from the deck for snow removal and periodic cleaning.
Upstream, in comment 4, I talked about some of the aspects of solar panel installation and orientation. Words are nice, but pictures are often better, so I've graphed out some data to show some difference between clear/cloudy, winter/summer, and horizontal/tilted measurements of solar radiation.
The following graphs are a continental location, at about 50°N latitude.
All radiation graphs show five different measurements:
The first graph is a clear day at the beginning of January. Direct radiation peaks at over 900W/m2, and diffuse radiation is less than 100W/m2. Because the sun is low in the sky, the global reading is much less than the direct - peaking slightly over 300 W/m2. The tilted sensor, though, peaks at over 800 W/m2 - not only is it pointing much closer to the sun, but it also sees a lot of ground that is very bright. The reflected reading peaks at over 200W/m2 - the ground is snow covered, reflecting about 75% of the global signal, so much brighter than the deep blue sky of the diffuse signal.
Clearly, a tilted solar panel would produce much more power than a horizontal one. We can see why when we look at the solar elevation angle (how high about the plane of the panel the sun is located). This graph shows the elevation above a horizontal surface (global instrument) and tilted surface. The sun is barely 20° above the horizon of the horizontal sensor, but is over 60° above the tilted sensor's "horizon". Note that the daylight period is only about 8 hours - elevation>0° for the horizontal view. Even though the titled sensor has an elevation >0° for much longer, those "extra" hours mean nothing, as the view of the sun is blocked by the earth!
The next day, cloud moved in. Direct radiation is zero, except for a brief period in early afternoon when the clouds thinned enough to let a bit of direct sun through. The four other lines are, from highest to lowest, tilted, Global and Diffuse (virtually tied), and Reflected. With no direct sun, and a snow-covered surface that reflects most of the solar radiation, there isn't much difference between the horizontal and tilted readings.
Note that for the horizontal sensor (global) the cloudy day is not much lower than the clear day. It peaks around 250W/m2, compared to a little over 300W/m2 on the clear day.
What about summer? Here is a "mostly" clear day in early July. Direct beam peaks only slightly higher than in January, but global radiation is much higher because of the higher solar elevation. The titled sensor peaks a little higher than global - it's tilt is no longer much of an advantage over the global sensor, and the portion of ground it sees is now dark (reflecting only about 20% of the global). Diffuse is again <100W/m2.
Daylight is now more like 16 hours, though, so daily totals will be quite different from January. We also see something odd in the tilted sensor - it peaks higher than the global (horizontal) sensor, but in early morning and late afternoon, it sees less than the global sensor. In fact, at the extremes it looks like it is only seeing the diffuse radiation - no direct.
We can understand this by looking at the solar elevation again. Note that for the titled sensor, the sun "rises" much later and "sets" much earlier (elevation <0°) than for the global sensor. What is happening is that the sun rises in the NE and sets in the NW, so it is actually behind the tilted sensor, not in front of it.
And lastly, we'll look at a cloudy summer day, right on the summer solstice. We do see some direct sun getting through in the afternoon, but we can see the cloudy period that covers most of the day. We see a substantial reduction in global before noon local time (compared to the clear day). After 12pm, we see a higher value for global as the cloud thins and a bit of direct radiation makes it through the clouds. During the cloudy period, the tilted sensor is not much different than the global one - both are seeing the same diffuse radiation.
So, hopefully this helps illustrate some of the complexities related to solar panel installation and orientation. To refer back to the OP - no, cloudy skies does not mean "no solar energy". The OP is correct - the myth is busted.
This is only one location, and a few days of data. And this level of data is not readily available for most locations. But it does illustrate that installation may be dependent on local factors such as amount of cloud, type of cloud, timing during the day, etc.
Evan at 6,
Where I live in Florida most of the utility solar farms are on two dimensional tracking. They face East in the morning, horizontal at noon and West in the evening. I think that is standard for utility farms in the USA. I think they get more expensive evening power and more power the entire day. I have not seen two dimensional tracking for home use.
I get ads all the time for three dimensional tracking setups ground mount at home. They claim up to 40% more power than latitude mount. I have a big yard so more panels seem more cost effective than more expensive mounts. In a small yard they might work well.
Evan,
To try to stay ‘close to topic’, but respond to your interest in the most beneficial ways to install home solar panels, I will start by saying that incorrect claims like the Myth that this Rebuttal addresses are attempts to unjustifiably discourage people from:
A very important consideration is that the governing objective is to minimize CO2, or other lasting or significant global warming impacts, produced by electricity generation.
Grid electricity generation needs to be transitioned to be truly net-zero ghg. And grid, as well as home, solar panels can be part of that transition. Even a rapid correction of a grid system to responsibly limit the harm done can take many years. And ‘natural gas peaker plants’ are likely to be the last parts removed from regular operations (natural gas peaker plants may exist long-term as an ‘emergency-emergency back-up’ that is very rarely needed).
Based on that understanding, people can help more rapidly reduce the harm of electricity generation by installing home solar systems. And they should design home solar installation to maximize the total annual generated electricity, appreciating what affects the amount of electricity generated. They should not be discouraged by the reality that many factors affect the amount and timing of electricity generated. And they certainly should ignore misleading claims like “Solar panels don’t work in cold or cloudy climate.”, or when covered by snow.
Bob Loblaw, and others, have provided plenty of very helpful information, including information about how cold and cloud affect solar panel performance.
Moving a little further from the topic to add some more considerations for home solar installations...
Variation of pricing should guide the use of electricity. However, a focus on the ‘variation of grid pricing’ may result in choices that do not minimize the overall amount of CO2 emissions.
If CO2 per Grid-generated MW-hr does not vary with time of day then minimizing the ‘net demand for grid energy’ should minimize CO2 from grid energy production. That means maximizing the total annual home solar generation with unused home generated electricity stored for later use or delivered to the grid to reduce grid generation. Building home solar for maximum generation during ‘peak price times’ would likely reduce the total annual home generated electricity and result in more CO2 generation than building to maximize total annual generation.
However, if CO2 per MW-hr is significantly higher during peak price times, because of things like the use of natural gas peaker plants to meet peak demands, then designing the home system to reduce the ‘home grid demand’ during the peak ‘CO2 per MW-hr’ times may be more helpful. However, as the grid system is improved to reduce CO2 generation the home system built to maximize generation during those current peak CO2 times becomes less helpful than a home system designed to maximize total annual generation.
The story changes if home generated electricity cannot be delivered to the grid or will not be stored for later use. Minimizing the grid energy needed would still be the objective. But maximizing the home solar generation during times of peak home electricity use becomes a significant, but complex, consideration. This could lead to potential benefits from maximizing generation during peak use times. However, that would also lead to understanding the benefits of charging the EV during peak home solar generation times.
All of the above considered may result in the conclusion that the best thing to do is build the home solar system to maximize the annual total generation and then adjust the home electricity use to minimize CO2 emissions impact of grid energy needed.
Final points regarding panel orientation:
Nigelj
Thanks for the comment - you actually hit the nail on the head with the best observation with the core issue.
Your comment " The capacity factor drops by about half. Could you not compensate by installing twice the number of panels? "
Yes you can compensate by installing twice the number of panels. That is exactly what has to be done to make a renewable system work.
The more important is the impact with utility solar.
First EIA website is a great source to cross check the data and provides a wealth of information.
The combined decrease in electric generation from wind and solar during the winter months is approx 30%-40% lower during the winter months vs the spring and fall (even before taking into account the frequent 2-5 day wind doldrums that occur during the winter).
Electric demand during the winter is approximately 30%-40 higher than during the spring and fall. As heating is converted to electric, the demand from the spring/fall vs winter demand will continue to widen. As such, gross capacity of renewables for the winter needs to be approx 2x of the needed capacity during spring and fall to cover the winter demand.
Further the idled capacity during 6 or so months of the spring and fall due to the redundancy needed for the winter months wrecks havoc on the math used in the LCOE computations. Guess what happens to the LCOE computation when significant amounts of redundancy is needed to generate sufficient electricity during the winter.
David-acct @14 ,
Yes, a perfectly good point. Costs must be viewed from several angles.
It is too broad to speak of electricity generation costs at simply peak demand times or at average annual GWh produced. "Minimum demands" must be considered, as well as back-up costs. And of course the longer term environmental and health costs.
Temporary reserves and redundancy costs also apply to gas/coal fuelled generation. LCOE has its official definition ~ but really should be looked at in the Big Picture. If only Life were simpler !
There will be some complications and expense entailed in engineering and producing a remedy for the limitation Eric focuses on, but they're not showstoppers. We have everything in hand, technically speaking.
Related example: A few years ago I put together some PV-powered WiFi repeaters for a harbor up here in the NW. These use LiFePO4 batteries for overnight operation and these batteries cannot be charged successfully below about 0C, precluding reliable operation during winter. The problem was solved by equipping batteries with heater film, with a thermal snap switch in the battery housing attaching pre-controller power output from panels to the heaters and thereby diverting up to ~20W of PV output to warm batteries, until the snap switch disconnects after temperature rise. This has worked perfectly to maintain operations.
In the case of PV panels, a high-latitude/extreme conditions variant (as with DHW thermal collectors) could be equipped with inexpensive electronics (flea-power minimal microcontroller, monitoring optical and other inputs as required) and control warming of the panels when necesssary. The chief and only hard technical optimization problem to solve here is that of how to convert a bit of juice to heat where it's needed while keeping engineering trades low. This would likely entail a slight performance drop for ultimate panel power output, if an off-the-shelf technology was used for that (microgrid or resistive film coating) so as to minimize effects on mechanical packaging of the PV panels.
The magic of the market will address this, to the extent Eric's situation is a problem sufficient to provide motivaton. That magic is the only mystery in this scenario.