March 31, 2009
Renewables Face Harder Time With Fossil Fuels Price Dip

An article by Matthew Wald in the New York Times takes a look at costs for different ways of generating electricity. Wald reports that declines in the prices of coal and natural gas make the renewables (wind, solar) a much harder sell.

The cost of solar thermal electricity, made by using the sun’s heat to boil water and spin a turbine, would be nearly three times that of coal and more than twice that of natural gas. (It would be almost double the cost of wind energy, too.)


A modern coal plant of conventional design, without technology to capture carbon dioxide before it reaches the air, produces at about 7.8 cents a kilowatt-hour; a high-efficiency natural gas plant, 10.6 cents; and a new nuclear reactor, 10.8 cents. A wind plant in a favorable location would cost 9.9 cents per kilowatt hour. But if a utility relied on a great many wind machines, it would need to back them up with conventional generators in places where demand tends to peak on hot summer days with no breeze. That pushes the price up to just over 12 cents, making it more than 50 percent more expensive than a kilowatt-hour for coal.

Solar photovoltaics (PV) on home rooftops costs more than solar thermal (aka concentrating solar). Plus, Brian Wang argues that PV is more dangerous than nuclear and wind because roofers get injured and killed in falls. But he says PV is still safer than coal. My guess is it depends on how tightly regulated the coal plants are. If only coal plants weren't allowed to emit particulates and mercury their damage to our health would go down substantially.

In the US half our electricity comes from burning coal. The 2000 terawatt-hours (twh) of electricity generated each year in the United States from coal electric would cost 3 cents more per kilowatt-hours (kwh) if generated from nuclear power. So what would that cost us? 2000 terawatts is 2 petawatts or 2 times 10 to the 15th power of watts. A kilowatt is to the 1 times 10 to the 3rd power watts. So we are talking 2 times 10 to the 12 power kilowatts times 3 cents extra per kilowatt. Or 6 times 10 to the 10th power of dollars. Am I correct in thinking that is $60 billion per year extra? Seems pretty small.

Of course the cost of converting to nuclear power would be higher than the price difference since lots of existing coal electric plants would need to be phased out before end-of-life. What is the cost of all existing coal electric power plants?

Share |      Randall Parker, 2009 March 31 12:09 AM  Energy Electric Generators

Vincent said at March 31, 2009 12:26 AM:

Who cares about the roofers? Certainly not the roofers themselves, who would welcome the extra business.

Dale said at March 31, 2009 6:18 AM:

I like Vincent's comment. The government is not a parent to a toddling population. Choice is the essence of freedom.

Also, $60B only seems small in light of recent expenditures. It's a large annual cost. Still, it would pale in comparison to the capital investment.

2E15 Watts required
1000MW typical nuclear power plant output = 1E9 Watts
2E15/1E9=2E6 or 2,000,000 nuclear plants
$2B (probably a low estimate) per plant
$4,000,000,000,000,000 startup costs for complete replication of current US power demand using nuclear power plants. That's four quadrillion dollars, or 73.6 times the world's GDP for 2007 (World Bank est. of $54.35E12). Wow, is that right?

So, maybe we just replace dying coal reactors with wind and nuclear power as we move forward.

Tim said at March 31, 2009 9:05 AM:

According to
World energy usage is 16TW, 1.6E13. 1000MW = 1E9. 1.6E13/1e9= 16,000 nuclear reators.
16,000 X 2billion = 32Trillion dollars, about 3X times current US GDP. Alot to be sure, but spread out over the likely 50+years it would take to do it not beyond the pale. And I am talking about the whole world not just the US, and total energy usage from all sources. Obviously nuclear wouldn't realistically have to do it all.

bbartlog said at March 31, 2009 10:37 AM:

Yeah, the thing with comparing the roofer deaths to those from coal is that many of the coal deaths are externalities (environmental mercury, radiation, other pollution) while for the roofers the costs are assumed to be fully captured by the wages that need to be paid to the roofers. Of course this is really more an argument for trying to achieve true cost accounting for coal power...

The other thing though - why assume that all photovoltaic solar is *rooftop*? I can well imagine that the deaths per TWH would indeed be pretty substantial if every damn panel had to go on someone's roof, but for construction of larger-scale projects I daresay that most photovoltaics would end up in large ground installations (like empty desert acres).

Nick G said at March 31, 2009 3:53 PM:

Wald's article appears similar to ones I've seen recently that point out that hybrid sales are falling, while failing to notice that they're falling less than the overall light vehicle market.

Here's a reply to Wald's article:

"I have known the New York Times energy reporter Matt Wald for 15 years, and generally think he is pretty good. But he has published perhaps the most flawed, inaccurate, and indefensible article in his career.

Wald's piece could also be a poster child for award-winning journalist Eric Pooley's searing critique of the media's coverage of climate economics (see How the press bungles its coverage of climate economics -- "The media's decision to play the stenographer role helped opponents of climate action stifle progress").

And, a little calmer:

Brian Wang said at March 31, 2009 5:31 PM:

A misquote from my article on deaths per TWH (terawatt hour)
Rooftop solar is several times more dangerous than nuclear power and wind power. It is still safer than coal and oil, because
those have a lot of air pollution deaths.

Yes, non-rooftop solar can be safer [0.44 up to 0.83 death per twh each year).
Coolearth mylar balloons for concentrated solar. If the rooftop solar
is part of the shingle so you do not put the roof up more than once and do not increase maintenance then that is ok too.
Or if you had a robotic system of installation.

World average for coal is about 161 deaths per TWh.
In the USA about 30,000 deaths/year from coal pollution from 2000 TWh. 15 deaths per TWh.
In China about 500,000 deaths/year from coal pollution from 1800 TWh. 278 deaths per TWh.

Wind power proponent and author Paul Gipe estimated in Wind Energy Comes of Age that the mortality rate for wind power from 1980–1994 was 0.4 deaths per terawatt-hour. Paul Gipe's estimate as of end 2000 was 0.15 deaths per TWh, a decline attributed to greater total cumulative generation. By comparison, hydroelectric power was found to to have a fatality rate of 0.10 per TWh (883 fatalities for every TW·yr) in the period 1969–1996

Nuclear power is about 0.04 deaths/TWh.

Randall Parker said at March 31, 2009 5:45 PM:


Sorry for the misquote. I read it in a hurry. I'll fix the post.


You are confusing watt-hours with watts. The 2000 TWH is the total over a year. Divide by 365*24 to get an idea of how much capacity would be needed. But you can get at it more easily by noting that nukes provide 20% of US electricity and coal provides 50%. So build 250 nukes along with the 100 we already have. Those 250 nukes are probably at most $6 billion a piece. Probably substantially less in large quantity. So we are talking less than the US budget deficit in 2009.

Engineer-Poet said at March 31, 2009 6:26 PM:
World energy usage is 16TW, 1.6E13. 1000MW = 1E9. 1.6E13/1e9= 16,000 nuclear reators.
Apples and oranges; you're comparing total raw input with nuclear electric (not thermal) output.  World electric power consumption is about 1.6 TW, so 1.6*10³ reactors.

The cost of the reactors is calculated based on current designs for advanced PWRs.  If we went to molten-salt reactors, the cost would be only 30-50% of the PWR cost, we'd have far more material to call fuel (thorium is a biggie), thermal efficiency could skyrocket (the SSTAR proponents claim 44% from supercritical CO2 turbines), and the price of power would probably fall below coal-fired even without considering externalities such as mining damage, ash disposal, heavy metals and atmospheric CO2.

Fat Man said at March 31, 2009 8:56 PM:

What is the cost of all existing coal electric power plants?

The question should be what is the unamortized cost of existing coal plants. Utilities are required to depreciate their existing plants. A plant that is 30 or 40 years old should be fully depreciated.

Dale: You are off by a couple of orders of magnitude.

The existing 104 n-plants already produce 20% of the US electric supply, or ~800 TWh of electricity annually Link.

New light water plants are about 1GW in capacity. At 90% operation, a 1 GW plant produces ~8 TWh per year. Total US production is ~4200 TWh. So let's say it is about 525 plants or 421 new plants.

I have seen more recent estimates of $6/W construction costs for new nuclear, but lets use $10 to make the math easier. 421 plants * 10G$ = 4.2 T$ which is 1/1000 of your estimate.

Of course plants do have to be replaced. Let us assume we want to have a fleet of 600 1 GW plants. A LW reactor and power plant should last at least 60 years. There are several in the US that are being licensed to do that. So we will have to build 10 plants per year, which will cost us 10* 10 G$ or 100 G$/yr. That is not enough to make a congressman sweat.

Nick G said at March 31, 2009 11:32 PM:

So, we're all agreed that eliminating coal wouldn't, in the larger scheme of things, be that expensive?

Given that coal is 50% of US CO2 emissions, that suggests that dealing with global warming would also be not that expensive, right?

Fossil Fuels said at April 1, 2009 4:09 AM:

nice article. I'm regular reader of you blog.

Paul F. Dietz said at April 1, 2009 7:17 AM:

If we went to molten-salt reactors, the cost would be only 30-50% of the PWR cost,

I find this claim to be unbelievable, particularly since it involves paper reactors that would have highly radioactive primary loops and complicated systems for online reprocessing not found in conventional reactors.

Engineer-Poet said at April 1, 2009 10:46 AM:

Much of the cost of a PWR is expensive on-site labor and interest costs during construction, with a subtantial fraction in specialized hardware.  MSRs get around this:

  • The MSR reactor vessel can be small enough to be factory-built and trucked to the site; this slashes both cost and the schedule.
  • The MSR requires no pressure vessels or containment building capable of holding a steam explosion.
  • Not using steam as a working fluid allows the turbine systems to be made smaller and run at higher pressure.  Smaller turbines are cheaper.
I'm not sure the radioactive loop is an issue.  The Russians are working on modular lead-bismuth cooled reactors, and Po-210 isn't exactly benign but they appear to have dealt with it.

Paul F. Dietz said at April 1, 2009 11:25 AM:

The radioactive loop is a big issue. With LWRs, the reactor operators go to great lengths to avoid or remove even traces of radioactivity in the primary loop, since they increase operating costs. Each nuclear worker is limited in how much radiation he can receive in a year, and once that limit is reached you have to keep paying him even though he can't work on the reactor.

Liquid metal cooled reactors have always had problems with inspection. This is an issue facing any reactor with an opaque coolant, and would affect MSRs as well.

The Russian experience with Pb/Bi cooled reactors does not include making them commercially competitive.

Smallness of reactors is not a plus, actually. There are substantial fixed, power-independent costs associated with reactors, many related to licensing and security. New LWR plans are involving larger reactors, not smaller, for just this reason.

I doubt MSRs in the US (or any western country) could be licensed without containment buildings.

The putative advantages of gas turbines with nuclear reactors would be easier to swallow if high temperature graphite moderated reactors had not been commercial failures.

And I mention once again the issue of the online reprocessing plant that would have to be present with each MSR. This is a complicated system not present in today's reactors. Do reactor operators really also want to run their own onsite reprocessing plant? I certainly wouldn't want to.

Bill Woods said at April 1, 2009 12:52 PM:

One large power station can have numerous small reactors, so those site-level costs can be shared. Anyway, MSRs can be large, if that's what's wanted.

The containment structure can be smaller, since it doesn't have handle the steam from a breached reactor vessel.

The bulk of the reprocessing doesn't seem that complicated. Remove the noble gases as they come out of solution, fluorinate uranium, and distill the fluoride salts off the waste. Further processing of the fission products could be handed off to specialists.

Engineer-Poet said at April 1, 2009 2:55 PM:

The MSRE had a heat exchanger and a clean salt loop going to the radiator; such a heat exchanger could be located inside a "swimming pool" reactor.  I don't see why a system using a supercritical CO2 turbine couldn't have a single heat exchanger, especially if the salt used no lithium (no tritium generation).  Worker exposure can be addressed.  The MSRE was designed for remote servicing of moving parts, and that was over 40 years ago; if we can't do better now, we've left this technology to sit for FAR too long.

Liquid metal cooled reactors have always had problems with inspection.
MSRs do not.  If you want to inspect something, you drain it.  There is no solid fuel full of fission products that will melt down if it doesn't have continuous cooling; the fuel is the coolant, and it may have only a few days' or weeks' worth of fission product inventory anyway.

Removing fission products is a major advantage.  Conventional reactors cannot shut down and restart easily because of xenon poisoning; once they're down for a while, they stay down until the xenon decays.  MSRs bubble out the xenon in the circulation pump; the MSRE staff weren't paid for 24/7 coverage, so it was drained to shut it down on Friday and restarted by refilling each Monday.

The Russian experience with Pb/Bi cooled reactors does not include making them commercially competitive.
Yet they're working on the SVBR to do just that.  One of the goals is to repower existing thermal powerplants.  There's no reason why this couldn't be done with MSRs of some kind.

LFTRs would have cost as little as 1/10 as much to develop as the liquid-metal fast breeder.  Had it not threatened the LMFBR's gravy train 35 years ago, we might already be standardized on them and we wouldn't be having this conversation ("Pressurized water?  Are you kidding?  They had so much trouble with the pressure vessels corroding and getting brittle; that belongs back in the dark ages!").

Smallness of reactors is not a plus, actually.
You're confusing physical size with installation capacity.  You can have multiple reactors per installation to share the fixed costs, and the smaller the physical size the easier it is to build at a factory rather than on-site.  If e.g. large coal plants are being re-powered, smaller reactor sizes make it easier to fit the thermal capacity to the existing steam systems.
I doubt MSRs in the US (or any western country) could be licensed without containment buildings.
Oh, you'd still have them.  You just wouldn't need them to be anywhere near the size (no need to hold large volumes of steam, and a much smaller reactor to begin with) or strength (no pressurized fluids involved outside the power-conversion section).
The putative advantages of gas turbines with nuclear reactors would be easier to swallow if high temperature graphite moderated reactors had not been commercial failures.
Yet the pebble-bed developers are still at it, proving that people with money at risk disagree with you.
And I mention once again the issue of the online reprocessing plant that would have to be present with each MSR.
That actually depends on the particular options chosen; one design manages reactivity by adding low-enriched fissionables over time and dumping the salt every decade or two.

However, the MSR has not stood still.  Reprocessing a molten salt is trivial compared to reprocessing an oxide.  Development has even yielded a molten-bismuth exchange system for a single-fluid LFTR which electrolytically pulls protactinium out of solution and isolates fission products continuously.  Would I want one of these at my plant?  Considering the headaches of managing unseparated fission products in large volumes of fuel rods in cooling pools, I would jump at the chance to try something better.

Engineer-Poet said at April 1, 2009 3:00 PM:

Oh, on physical size:  the power densities being talked about for some MSR's are in the region of 400 kW(th) per liter, so even a 1 GWe core unit would fit in a large living room.  You do not need large size to obtain high power; we are literally talking about units that can be shipped on flatbed trucks.

Post a comment
Name (not anon or anonymous):
Email Address:
Remember info?

Go Read More Posts On FuturePundit
Site Traffic Info
The contents of this site are copyright ©