The Dark Future of Solar Electricity

Guest Post by Willis Eschenbach

The “Annual Energy Outlook” for 2011 is just out from the US Energy Information Administration. The section called “Levelized Cost of New Generation Resources” looks at what are called the “levelized” costs of electric power from a variety of sources. Their study includes “renewable” sources like solar, although I’ve never found out exactly how they plan to renew the sun once it runs out. The EIA data in Figure 1 shows why solar will not be economically viable any time soon.

Figure 1. Levelized costs of the different ways of generating power, from the EIA. Blue bars show the capital costs for the system, while red bars are fuel, operations, and maintenance costs. Estimates are for power plants which would come on line in five years. Operation costs include fuel costs as appropriate. Background: HR diagram of stars in the star cluster M55 

“Levelized cost” is a way to compare different electrical generation technologies. It is calculated by converting all of the capita costs and ongoing expenses for the project into current dollars, and dividing that by the amount of energy produced over the lifetime of the plant. For the mathematically inclined there’s a discussion of the various inputs and calculations here. Levelized cost is the all-up cost per kilowatt-hour of generated power. The levelized costs in Fig. 1 include transmission costs but not the costs of backup for intermittent sources.

So why is this chart such bad news for solar electricity? It’s bad news because it shows that solar won’t become cheap enough to be competitive in the open market any time in the near future. Here’s why.

Now, please don’t get me wrong about solar. I lived off the grid for three years on a houseboat with solar power in Fiji, collecting sunshine and drinking rainwater. I am a solar enthusiast and advocate, there are lots of places where it is the best option.

But not on the grid. It’s too expensive.

Yes, it’s true that the sunshine fuel is free. And the operations and maintenance is cheap, 2 cents a kilowatt-hour. And as backers are always claiming, it’s the only technology where the capital cost is falling rather than rising, as the price of solar cells drops.

But here’s the problem. Solar cell prices have already fallen so far that only about thirty percent or so of the cost of an industrial-sized solar power plant is solar cells. The rest is inverters, and wiring, and racks to hold the cells, and the control room and controls, and power conditioners, and clearing huge areas of land, and giant circuit breakers, and roads to access the cells, and the site office, and half a cent for the transmission lines from the remote locations, and labor to transport and install and wire up and connect and test all of the above, and …

That means that out of the twenty cents of capital costs for solar, only about six cents is panel costs. Let us suppose that at some future date solar panels become, as they say, “cheap as chips”. Suppose instead of six cents per kWh of produced power, they drop all the way down to the ridiculous price of one US penny, one cent per kilowatt-hour. Very unlikely in the next few decades, but let’s take best case. That would save five cents per kWh.

The problem is that instead of 22¢ per kWh, the whole solar electric system at that point would have a levelized cost of 17¢ per kWh … and that is still two and a half times the price of the least expensive option, an advanced combination cycle gas turbine.

Finally, this doesn’t include the fact that when you add an intermittent source like solar to an electrical grid, you have to add conventional power for backup as well. This is so you will be sure to still have power during the time when the sun doesn’t shine. Even if you never use it, the backup power will increase the cost of the solar installation by at least the capital cost of the gas plant, which is about two cents per kWh. That brings the levelized cost of solar, IF panels dropped to a levelized cost of only one penny per kWh, and IF the backup generation were never used, to 19¢ per kWh … and that’s way more than anything but offshore wind and solar thermal.

However, it gets worse from there. The cost of fuel for the gas advanced cycle power plant is only about 4 cents per kWh. So even if gas prices triple (which is extremely unlikely given the advent of fracking), the gas plant cost will still only be about 14¢ per kWh, which is still well below even the most wildly optimistic solar costs.

And that means that the dream of economically powering the grid with solar in the near future is just that—an unattainable dream. The idea that we are just helping solar get on its feet is not true. The claim that in the future solar electricity will be economical without subsidies is a chimera.

w.

PS—On a totally separate issue, I suspect that the maintenance costs for wind power are underestimated in the report, that in fact they are higher than the EIA folks assume. For example, both wind and water are free, and the EIA claims that wind and hydro have the same operation and maintenance cost of about one cent per kWh.

But with hydro (or almost any other conventional technology) you only need to maintain one really big generator on the ground.

With wind, on the other hand, to get the same amount of power you need to maintain dozens and dozens of still plenty big separate generators, which are stuck way up at the top of really tall separate towers … and also have huge, hundred-foot (30 m) propeller blades whipping around in the sky. You can imagine the trek you’ll have when you forget to bring the size #2 Torx head screwdriver …

Do you really think those two systems, both feeding the same amount of power into the grid, would cost the same to maintain? Check out the windfarms and count how many of the fans are not turning at any given time …

Get notified when a new post is published.
Subscribe today!
3 2 votes
Article Rating
317 Comments
Inline Feedbacks
View all comments
December 6, 2011 1:09 am

Willis writes “I objected that you had left out the other incremental cost of wires and the upsizing of the alternator and the additional rack to mount the panel and the labor to do all of that …”
Thats fair enough there are certainly initial installation costs involved but not all panels will always require an increase in the inverter capacity if the existing inverter still has capacity and at any rate they’re paid for over much less than the life of the panel.
There are concrete examples today actually. I have several friends who have solar PV installations and so far their projected payback period is about 8 years. Probably less if the cost of energy increases as is likely. The solar rebates are decreasing in Australia solar instllations are still affordable and popular.
I disagree with your general suggestion that the cost is about as low as its going to get. PV solar is still far from a consumer product in the same way a computer is now. Computers have dropped to a tiny fraction of what they were once worth and the same will happen to any technology that gets mass adoption. Much lower than you think is possible I would suggest.

December 6, 2011 1:18 am

Willis writes “Although I understand the words, I fear that I don’t believe what they say. Why will the running costs (presumably of solar) reduce? ”
Because there are virtually no running costs for PV solar. Once installed they pretty much take care of themselves and over time their reliability will improve too.
Compare this fundamental feature of the enrergy production to that of oil where there are ongoing costs to produce that oil. Exploration, development, extraction, processing, distribution. All those costs are always with us and always increasing.
Now you might argue that you can make more money by “investing” rather than spending on PV solar with a return over many years. And you may well be right but thats irrelevent This isn’t about how you can make the most money, its about how we can best cater to our future energy needs and “investing” sure doesn’t do that.
You can “invest” in fossil fuels and sure you’d make more money. Great if thats the goal to make money but its not the path I prefer.

December 6, 2011 3:14 am

Willis writes “It means lots of small switchgear, which is more expensive than single large switchgear. Same is true for inverters.”
Its that kind of reasoning that leads to statements like ““I think there is a world market for maybe five computers – Thomas J. Watson .””. Reality has an electric drill in every household and on average its used only a few times ever. People dont mind buying and owning stuff even when they could simply borrow someone else’s drill or hire one. Thats a fact of life.
Willis writes “Redundancy also means a whole lot more generating units that need to be reliably disconnected from the grid when there is a power outage, to keep from frying the repair personnel.”
You’re dissing redundancy? Distribution companies spend a fortune on redundant feeders and switchgear to manage them. Live line maintenance is common and in the worst case, one only needs to disconnect a bit upstream and downstream of the fault to isolate it.

Cyrus P. Stell, P.E., CEM
December 6, 2011 6:38 am

: You have pointed out the inherent weakness of all future casting. We make assumptions, based on the available evidence, and then run the numbers and see how it pans out. My analysis MUST include a life-cycle-cost analysis or it’s worthless. To make that analysis, we decide in advance what we expect future cost increases (or decreases) to look like. EIA projects such numbers, and I use them despite my misgivings (there was another thread, here or somewhere, about how EIA consistently over-estimates future cost of fossil fuels, and under-estimates future costs of “renewables” as defined by legislation). Here’s the thing, though, when we do an engineering analysis we are comparing alternatives. i.e., do I keep what I have, do I buy more of what I have, do I install more of option A (maybe that’s solar PV) do I install more of option B (call that solar thermal) or option C (maybe that’s wind). From that analysis you derive Levelized Costs (gee, where have I heard that term before? Oh, yeah, refer to the title.)
Now if we make the wrong assumption about, say, the future costs of electricity, in many cases it won’t matter if we’re comparing something that just uses different amounts of electricity, the same “wrong” costs are in all of the life cycle equations, so it still gives a fairly accurate picture of the best choice. But when we’re comparing different technologies, we have to be a bit more certain of our future-casts. But, you start off assuming no consideration was given to a wrongly forecast future price, and in fact, it was carefully considered. As I originally stated, we assumed 35 years ago that our assumptions (double assumption? I’m leaving it in there) were likely wrong in one direction (future costs of capital would decline while future costs of fossil fuels would rise at some rate) and it turned out we were wrong in the other direction (future costs of fossil fuels in inflation-adjusted $ actually declined for a long time, while the future cost of capital equipment, at least the equipment we were selecting, did not decline, or at least not as much as we had hoped, and might have even risen). So at this point, it’s not just a coin flip to determine who’s most likely right, the figures published by EIA have some analysis behind them, while your hypothesis is just that, a figment of your wishful thinking. Guess who I’d put my money on?
Next point, you talk about operating costs of already-installed solar being near zero, and that’s just false. I worked at a place that had made 3 different installations of solar-thermal, and all 3 were deactivated well before the projected end of their expected useful life (the life used in the LIFE-cycle cost analysis) and 1 had been demolished and removed entirely. Why were they inoperative? One thing, well, maybe 2… They got no maintenance, not even drain-down in advance of freezing weather, so much of the tubing burst, and secondly, they were not metering the production of the solar-thermal, so when the question came up, “is it worth it to repair these?” nobody could argue that it was. One of the earlier commenters posted a link to solar installations that had received no maintenance, both solar-PV and solar-thermal, so scroll up. You need to re-evaluate that assumption.

Andyj
December 6, 2011 9:55 am

I’ve fitted a vacuum tube solar panel to my roof. Did not seek subsidies nor paid the rip-off rates for buying/fitting here in gloomy UK.
£500, paid for itself in 18 months two years ago and it looks good for the next 25 years of free hot water for 8 months of the year.
Absolute no-brainer. No energy supplier can better that.

mike g
December 6, 2011 11:34 am

So, Willis, I’m skeptical of the safety of fracking and I mispelled aquifer and I’d rather see coal burned for electricity than gas. Y’all are tough on skeptics here.

DirkH
December 6, 2011 11:51 am

TimTheToolMan says:
December 6, 2011 at 3:14 am
“Willis writes “It means lots of small switchgear, which is more expensive than single large switchgear. Same is true for inverters.”
Its that kind of reasoning that leads to statements like ““I think there is a world market for maybe five computers – Thomas J. Watson .””.”
Here, Tim has a point. We’ve only just started to seriously mass-produce inverters; there’s a lot of room for driving down the production cost. Compared to say, motor electronics, which are produced nearly 100% automated – a typical assembly line produces 5,000 of them or so in an 8 hour shift, and a factory has maybe 10 such lines – inverters are still large boxes with a lot of manually connected cables. This will change when we use one small inverter per module; they become as small and as plentiful as motor electronics boxes, in the millions of pieces range, and the same automation will be used for mass production. Shortly thereafter they might become solid state modules, and later one chip solutions. I’m fantasizing here, but that’s what usually happens when the numbers are scaled up.
” Reality has an electric drill in every household and on average its used only a few times ever. ”
Tim, you shoot yourself in the foot here – that’s exactly not an example for high efficiency ;-).

DirkH
December 6, 2011 11:59 am

Willis Eschenbach says:
December 6, 2011 at 11:46 am
“No, I’m not “dissing redundancy”, Tim. I am a realist who knows that there are problems that come along with every solution, including redundancy. For generator redundancy, one of the problems is disconnection of all of the generators in cases of system failure.”
Willis, all solar inverters used in Germany need to switch themselves off as soon as the grid frequency rises above a certain threshold – because this frequency signalizes overload. Similarly, a lot of emergency conditions are already encoded in the inverters; they have a digital signal processor that does about 15,000 cycles a second and checks these conditions.
Most inverters will only feed in when an outside grid exists; they never build up a grid by themselves! Disconnecting the grid suffices to make them all shutdown by themselves.

Louise
December 6, 2011 12:00 pm

On December 3, 2011 at 8:03 pm, Willis Eschenbach says:
“why do you have to be so snarly and ugly and nasty? All it does is make you look like a vicious, vindictive little man.”
Pot, kettle?

DirkH
December 6, 2011 1:04 pm

Willis Eschenbach says:
December 6, 2011 at 12:40 pm
“Thanks, Dirk. And would you be willing to bet your life that the disconnects all work correctly 100% of the time? I wouldn’t, I can easily envision conditions where the breakers wouldn’t pop. I’ve seen too much of “machinery gone wild” to think that safety systems are failure-proof. And repairmen are indeed betting their lives on your claims …”
I’d much rather trust a system designed under safety-critical considerations than anything else, because that’s the best bet I can make. As an additional precaution, gloves can’t harm 🙂
“In fact, one of the disadvantages is that you need (as you point out above) special switchgear to disconnect each solar rooftop installation.”
That’s exactly one of the beauties of one inverter per module – the inverter semiconductors do the disconnect. When you have a large inverter for a string of modules, you get a DC voltage of about 800 V; that can be nasty. Not so with the microinverters. You have the output DC of one module, that’s 60V or so. (I’m guessing). Even if the the emergency switchoff fails, the voltages are not that dangerous.
At the moment, efficiency lags behind big inverters and total costs are higher. We will see. I’m not trying to advertise anything, at the moment I have no business interests there.

December 6, 2011 1:17 pm

Timthetoolman has a few statements-
1) “Because there are virtually no running costs for PV solar. Once installed they pretty much take care of themselves and over time their reliability will improve too.”
2) “I disagree with your general suggestion that the cost is about as low as its going to get.”
Tim,
In response to statement 1:
My PV system has been in service for 5.5 years. In that time I have had my inverter serviced once (software was incorrect in how it calculated kwh’s- it read low by about 25%)- the upgraded software then read about 10% high per a separate kwh meter I had installed just before my PG&E E-7 net meter. After about 2 years of service my inverter started reporting negative values for one of the attribute it reports (instantaneous wattage). Rather then try to figure out what was wrong with my original inverter the unit was replaced. The new inverter has worked fairly well- it reports my kwh output about 4% high per my secondary kwh meter. The estimated mean time to failure for inverters is around 10 to 15 years, so any cost calculation needs to take this into account. Additionally, the efficiency of the inverter is not going to improve over time.
If I don’t keep a fig tree trimmed, that is located just south of a couple strings of my panels, my output drops by about 20%. Hence I have some ongoing maintenance for my system to operate at it’s rated POTENTIAL max output. Over the years I have found that without cleaning my panels (especially in the dry, dusty summer months) my overall output will drop between 8 and 10%. My weekly preventative maintenance is rather straightforward- a rinse with water from my garden hose with a fairly strong stream of water- geese fly overhead occasionally and it takes a bit of water pressure to remove their droppings from my panels).
As far as reliability goes each manufacturer of panels is required (in CA anyway) to limit (warranty) the degradation of their potential max output (STS rating) over the expected life of the panels. For my panels an expected degradation in STS max rating was something like a 20% reduction in STS max output over time (time being 20 or 30 years for my panels) of the warranty.
In response to statement 2- “I disagree with your general suggestion that the cost is about as low as its going to get.”
I concur with Willis on this one but with a change in the word cost to price for an installed residential PV system. As the balance of system (copper wire, inverter, aluminum railing system and installation labor, shipping) costs become a larger part of the overall system costs a 5 to 10% improvement in the cost of the panels are going hit the law of diminishing returns from a total price to purchase a self generation option. The price I paid to have my 6.12 Kw system installed in 2006 was $1.10 a Kw. Today the labor and misc materials (some cu wire, shut off switches) costs are between $1.30 and a $1.50 a Kw out here in CA to put a 6.12 Kw system in. The price paid by the wholesaler who I bought my panels from has gone down with the drop in panels costs. Their costs for the Al railing and their cost for the inverter haven’t come down. Their costs for shipping (primarily fuel costs related) have gone through the roof so free shipping is no longer included in the purchase of the hardware components of a PV system. The best price (before rebates, and tax credits) to have a mid sized investment grade residential PV system (4 to 10 kw) installed are likely never going to get below $3.00 to $4.00 (CEC AC rated) watt. Under my version of a best case scenario for residential PVgeneration (lets use the VERY optimistic $3.00 (installed watt cost)* 5.22 CEC Kw rating (the AC rating of my PV system)= $15,660.00 which yields a total yearly output of 9300 kwh- for a 6.12 kw STS rated system). The upfront costs in this best future scenario is $15.6K for a 5.22 kw cec rated residential system and it will yield 9300 kwh a year

DirkH
December 6, 2011 1:58 pm

Willis, you’re really having a bad day; I’ll leave after this comment. Nobody forced you to answer my comment; so blaming me for saying “Tim has a point there” with “We were discussing whether this kind of inverter is cheaper than that”, well, that’s just silly. No offense; don’t work yourself up. Bye.
http://en.wikipedia.org/wiki/Solar_micro-inverter
“Released in 1993, Mastervolt’s Sunmaster 130S was the first true micro-inverter.”
Looks pretty real to me:
http://www.smartsolarsystems.biz/wp-content/uploads/2011/05/micro-inverter-install6.jpg

December 6, 2011 2:05 pm

The chart shows 16 electrical energy sources. The costs of production do not include externalities, so they constitute a rough guide on real costs of production.
One poster noted that solar hot water, by comparison, in effect delivers energy at a much lower cost, yet this source of energy is omitted.
I also note that a British study has showed that the cost of reducing electricity demand varies from zero (in the case of turning down a thermostat, for example) to less than 1 cent per kw for many simple technologies like lagging of hot water pipes, but almost always much less than the unit cost of supply.
This endless and often bitter dogfight over preferred energy supply technologies is interesting, but it has very limited horizons.

Spector
December 6, 2011 2:46 pm

Solar power may be useful as an auxiliary power source, but eventually carbon-power is not going to be an option. (Bio-solar carbon excepted.) I am not prepared to say when that might be, but it does look like we are now consuming more carbon-power than we are discovering, at least in the case of petroleum.
I look at this as a question of whether energy from the sun can replace exhausted carbon-power in all uses. How much of the earth’s surface must be set aside so that each person can collect all the energy needed for heating, transportation, and feeding from the sun. That presumes the manufacture of synthetic transportation fuels. I am guessing that the average recoverable solar energy is less than 100 watts per square yard.
We do know that back in 1880 we had a lifestyle model and total global population that did not depend on carbon-power to the extent we do today.