Solar power myth buster

It never fails; whenever an article about solar power is posted on a system that allows comments, someone will eventually post a series of bologna statements that will almost certainly include…

• Solar increases the use of dirty power
• Solar panels require more energy to manufacture than they produce in return
• Solar panels release more greenhouse gasses during production than they offset in return
• Solar panels release all sorts of nasty chemicals
• Solar panels last only 25 years

So here’s a handy refutation guide.

• Using solar or wind means more dirty fossil fuels are burned for power

The basic argument here is that since PV and wind can’t be relied on, due to clouds, night, etc., you need to back up every watt of PV with a watt of some other power source. And since these power plants can’t turn on and off as fast as solar panels, you have to keep them in a “warm” state so they can spin up quickly, so they’re burning all the time. And while they are being spun up, they run inefficiently, so they burn more fuel, and dirtier, so the total emissions go up.

Whenever you’ll see this argument you should notice one obvious thing: no numbers. And whenever you see some sort of argument with no numbers, you can pretty much be sure it’s wrong. One can construct all sorts of scenarios, some will burn more fuel, some less. But which one is right?

Well now the numbers are in. And they show, clearly and overwhelmingly, that this problem does not exist. It turns out that it is trivially easy to know when to start up and shut down the extra power, and moreover, the PV system keeps the turbines turned off so long that the savings are always worth it.

NREL concluded a series of studies, and they all showed that the fuel savings absolutely dwarf any drawbacks, even when you include things like wear and tear on the systems due to start/stop cycles. And dwarf is the right word, the basic outcome was install as much PV and wind as you can.

• Solar panels require more energy to manufacture than they produce in return

Solar cells are produced through a heavy-chemical process similar to making soap or fertilizer. During the processing the feedstock has to be heated to about 300 Celsius. That’s hot, but not that hot.

So in the end, it’s trivially easy to measure how much energy goes in, and how much goes out. And that clearly demonstrates that panels will produce many, many times as much power as went into them. We’re talking 10 to 20 times as much. Even the worst-case design, traditional polysilicon panels, pay back their energy in 3.5 years. They’re expected to last over 30, so that’s a 10x payback, for the worst case. For the newer thin-film designs, payback is in about  year.

The NREL FAQ on the topic can be found here. Note that their numbers include not only the panels, but everything else in the system as well – inverters, racking, etc.

• Solar panels release more greenhouse gasses during production than they offset in return

At first glance it might seems this is the same argument as the one above – its the GHG of the energy needed to make the panels, ship them around etc.

But that’s the not the case here, this argument focusses primarily on the use of SF6 and similar chemicals during the production of panels. This is the primary argument against deploying PV in a recent book called Green Illusions.

However, the industry had made enormous strides in controlling the release of these chemicals, reducing them almost to zero. You can read an in-depth report on this from 2007, which talks about these newer processes as they were coming out. In the following five years they became almost universal.

I should point out, that this evolution was complete when Green Illusions was written, yet he made this claim anyway.

If you’re wondering why this happened, and so quickly, it’s largely because the new processes are cheaper. Companies were racing each other to get these methods into production.

But back to the GHG issue…

NREL did an extensive study on this issue, and the conclusions are clear: solar panels release about 20 times less GHG, everything in, than you would get from burning enough coal to produce the same amount of power. It’s also lower than the amount of GHG you’d get from natural gas, wood, or practically any power source that uses fuel.

It’s also common to mix arguments (see below for another example) and mention a chemical that’s used in one process without clearly stating that it’s not used in another. For instance, this post complains about the use of NF3 in terms of “solar panels”, without noting that NF3 is only used in a particular kind of panel which is not common. Moreover, hundreds of times that amount was used in the production of flat-panel TV’s, and in both cases its use has since been dramatically reduced, or eliminated outright. Any one of these points is important, but none of them are mentioned in these criticisms.

• Solar panels release all sorts of nasty chemicals

Solar panel production uses a number of caustic chemicals during the processing, and after. However, these are rarely released to the environment, and, like the item above, have been continually pushed out of the production systems over time. After all, if you leak expensive chemicals you’re basically leaking money. No matter how much of a cynic you are, you’ll probably agree that there’s incentive to fix that problem.

It’s also sometimes claimed that there are other nasty chemicals involved, like cadmium. But these claims are disingenuous. Like the GHG issue above, this refers to a particular type of solar cell that represents a tiny fraction of production. There’s no cadmium in the “solar panel” you’re thinking of. Even in this case, the “CdTe thin-film panel”, it’s questionable whether there is any route for those chemicals to enter the environment, any more than there is for the deadly chlorine gas to leak out of your table salt.

• Solar panels last only 25 years

I’m not sure I understand why I even have to write this, but I have seen this come up as some sort of complaint. No one complains that their television doesn’t last 25 years, but for some reason this is a problem for solar panels? Why?

My car has a 80,000 km power train warrantee, but I expect it to last 250,000 or more. Likewise, panels are warranted for 25 years, but are expected to last much longer. How long? Well, we don’t really know, because they’ve only been around 40 years and they’re still running.

You’ll often find claims that panels are expected to degrade slowly over time, and that’s what the warrantee is based on. Normally the panel gets a free replacement if the output drops below 80% of its original value in that 25 year window, which seems like you might expect a little under 1% degradation a year. But they just don’t do that. What really happens is that you lose 1 or 2% in the first couple of months, and then nothing more for decades.

Arco started serial production of solar panels in the early 1970s. Most of these went to navigation buoys and pipeline rust protection systems, the vast majority of which were later scrapped even though they were still working. But a number of them ended up in other roles, and they are producing the same amount of power today as they were in the 1970s. Here’s a perfect example, a panel from 1980 that was pulled down and found to be producing 100% of its rated power after 30 years.

Panels do fail, however, which is a different issue. If you buy 100 panels and wait 25 years, most of them will be producing 100% of their original power rating, but some small number of them will have failed outright. So on the whole the array is producing less power then when you bought it. To you, there’s no difference between an array with 5% failed panels and one with no failures but 5% degradation. But there are those that quibble.

So what’s the actual number here? There’s two big studies to point to, NREL’s from the late 1990s (published early 2000s) and  a Swiss array running since 1982 that showed an initial degradation of up to 5%, and then basically none after that (here’s an overview). Both of these show the same thing, and conclude that the expected lifetime is 30 years or more, and the “degradation” rate is either 0.7% (NREL) or 0.23% (LEEE-TISO).

So there you have it. If you have any other arguments you’d like to see added, by all means, fire away in the comments. And I’m always looking for more links and studies too!