Nuclear Power: Does it produce as much carbon as fossil fuels?

The following is my reply to Licia Corbella, who handles the editorials page for my local newspaper, the Calgary Herald.

I corresponded with Ms. Corbella in late December of 2009, and given her generally right-of-centre views in print, was surprised to find she had strong disagreements with the right-wing on nuclear power:

Nuclear power is, in my view, utterly illogical if you look at costs alone. It is not carbon free, when you consider the carbon needed for building the plant, mining the uranium, shipping it etc. Factor in storage and it's a disaster. Add in the potential for disaster and it's even more of a disaster.

A source Ms. Corbella's disapproval arises from a book by Dr. Helen Caldicott, "Nuclear Power is Not the Answer" which Ms. Corbella suggested I read. I promised to do so, and if I disagreed, back it up with more research than one can find at books and web sites paid for by the nuclear industry. (Tricky: most information about nuclear power is either from the industry or its harshest critics. Disinterested analyses are hard to come by.)

I believe that I found such material, at least to my satisfaction; the entire matter is so disputed that agreement is forever unlikely.

The Issue: Does the full Life Cycle of a Nuclear Plant emit similar carbon levels to a fossil-fuel plant?

That is the question I limited myself to addressing, because I consider the other nuclear issues to be well-debated and, frankly, decided. (My answer to just about all the other issues I have relegated to a separate page: What About France?)

So for me, the only issue that is new and undecided is whether nuclear energy drastically reduces carbon output per kilowatt-hour generated. This has been taken for granted for some time - the reaction itself produces no carbon - but was challenged by Dr. Caldicott's book.

Caldicott's Figures Are All From One Paper

The book, in turn, took virtually all of its argument that nuclear power does not greatly reduce carbon emissions from a
paper by Jan Willem Storm van Leeuwen and Philip Smith, entitled "Nuclear Power: The Energy Balance". The report has been highly influential on the anti-nuclear movement; I found it cited countless times in such materials on the web.

Dr. Caldicott's book, for the "energy balance" issue, credits that "most of the information" in that chapter comes from the Storm and Smith report. (I couldn't find any information in that chapter that did not come from the Storm/Smith paper.) I found her book irritating in that she quickly deviates away from offering numbers that can be added up in terms of "per kilogram of uranium" and compared to the energy output of the kilogram of uranium itself, to numbers that simply look big and scary but can't be added up. And if they had been added up properly, it wouldn't have helped: she somehow omitted the energy you get out of that kilogram of uranium.

The Sum That Caldicott Did Not Do

I decided to "fix" that omission and spent over a week of spare hours summing up all of Caldicott's energy assumptions to find the grand sum, and how it compares to the total energy produced by the plant that she completely failed to do in her own book. While I think that was intellectual sloppiness of the highest order, she was intellectually honest to confess at the end of the smoke-and-mirrors show of scary numbers, that the final sum was still one-third of the carbon emissions of a natural gas plant.

With typical nuclear plants at current efficiencies (better ones are to come) you get about 167 GJ of delivered electric energy out of a kilogram of uranium: over 52 thousand kilowatt-hours, enough for five people for a year. Your upkeep as a member of industrial civilization entails burning 10 tonnes of coal per year, or 200 grams of uranium. (The conversions between joules and kilowatt-hours and all the rest, I have summarized in a page on Energy and Power).

That number can be disputed up or down my several percent, but is mostly conservative. Canada's "CANDU" nuclear plants are all more efficient than that, and newer ones more so.

By comparison, when I summed up of the figures that Dr. Caldicott didn't, I got about 8 GJ/kg of uranium for mining the ore and making enriched fuel from it, if you have rich ores in soft rock. The worst-case, with low-grade ore in hard rock, still only 15 GJ/kg.

The Paper Uses An Estimation Technique a Layman Can Debunk

Probably the most-disputed numbers used by Storm and Smith, and echoed by Dr. Caldicott (save that when they expressed a range, she generally quoted the upper end as the "most likely") concerned the total energy debt incurred by building the plant and later by decommissioning it. Storm and Smith's numbers for this are so large as to dwarf the other components of the sum; if they are correct, there's hardly any need to add up the energy emitted by mining and milling and so forth.

But these, alas, are the numbers that even laymen can understand the critique of. Storm and Smith calculated the energy-costs of construction and deconstruction by a model. They took the cost of the work and multiplied by a nation-wide figure for the average energy-consumption per dollar of GDP.

Even I know that nuclear plants are not ridiculously expensive because of the energy that goes into making their steel and concrete and piping; they are expensive because of the human time put into them. Far more design time and checks of designs. Far, far, far more inspection and checking of every weld and wire; far more testing and certification.

It's like assuming that those $400 hammers and $600 toilet seats for which the Pentagon is infamous must have had twenty times the oil and coal expended on their manufacture than ordinary hammers and toilet seats.

The most-independent critique of these calculations - so academically comprehensive as to dwarf Storm and Smith's own effort - can be found on the web here - at an archive for the University of Sydney. It in turn cites similar work by the University of Melbourne.

Anti-nuclear activists unsurprisingly treat all the industry's own studies as propaganda, but that charge is harder to level at University professors.

Universities Disputed The Paper and The Discussion is On-Line and Quite Readable

Page 44 of the above-cited study contains the major objection to the Storm & Smith construction estimates which I have summarized with the "Pentagon" remarks. Page 63 summarizes all of their disputes with Storm and Smith. To repeat, however, they do not dispute all of their numbers, or only about small degrees of accuracy.

If you will concede that the University of Melbourne physics department could take a balanced view of the issue, then you may give credence to their many "facts for the public"-style articles at their web site,

They offered critiques of Storm and Smith, and published two rounds, no less, of rebuttal and counter-rebuttal with those authors. The discussion was mostly kept non-technical enough for laymen to follow. Storm & Smith's rebuttals did not convince me of their side.

A Hard-to-Dispute Estimate From a Real-World Case Study

The Australian papers, and others, frequently cite a very different analysis conducted by the Swedish utility, Vattenfall, which produces electricity and heat by just about every method possible; they own a large number of power plants.

Vattenfall built the Forsmark nuclear plant, at 3090 MW (3.1 GW) about half the size of Canada's Bruce station, in the 1970's and it produces about a fifth of Sweden's power. As part of an approval process in 1999, they had to submit an "EPD", "Environmental Product Declaration" you will find cited and linked to from the "" page above, and just about every other source that disputes the Storm and Smith study. Unlike the latter, Vattenfall had to actually add up all the inputs - the tonnes of concrete times the energy-costs of concrete, the same with steel and all the other raw materials; the energy-costs of manufacturing raw materials into plant parts, and the construction effort.

The concluded with numbers that are mostly a small fraction of those of Storm and Smith. Most importantly, they added up 8 PJ to to decommission a plant, vs Storm and Smith estimating 80-240 PJ.

A Petajoule is Just a LOT of Energy

To get a grip on what the heck "Petajoule" meant, I worked out that a petajoule was the energy consumed by a crew 100 of the largest-size bulldozers in use, the awesome Caterpillar D-10 - working 12 hour shifts, 5 days a week save for statutory holidays, for two years. That assumes they are putting out their full 700HP about a third of the minutes on-shift, which I would still call conservative given the time I've spent on heavy construction sites.

It's pretty hard for me to believe that the actual construction effort comes to over 200 crew-years for such a large crew. It's much easier to believe the numbers are in the single-digits for "PJ", perhaps as low as the two PJ that Vattenfall avered (four crew-years!), on top of the 6 PJ energy costs of their plant parts, for a total of 8 PJ.

So, for my summing-up of Dr. Caldicott's numbers, I did use the huge numbers of "petajoules" in the Storm/Smith paper for construction and demolition of the plant - but I used their lower-end estimates because even those are nearly ten times the numbers that practically everybody else agrees upon. And I got about 21 GJ/kg, for a total under 30 GJ/kg of input-energy to the nuclear industry lifecycle, when 167 GJ/kg comes out of it. Even using their numbers, the nuclear industry generates nearly six times as much energy as it consumes. Even if all were carbon-intensive, it emits one-sixth the carbon of a coal plant, one-third the carbon of a gas plant.

The Dispute is Between "one-sixth" and "one-twentieth" the Carbon of Coal

If you use the Vattenfall numbers, or most of the others offered by the Australian academics, nuclear power emits between 2 and 5 percent of the greenhouse gases to get the same electrical energy as a coal-fired plant. The exact numbers depend on your choice of type of nuclear plant, your chosen technology for enrichment, and how good your uranium ore is. (Canada's is among the best, by the way: some 2% of that north-Saskatchewan rock is uranium, in contrast to Dr. Caldicott's concerns that 0.01% uranium in the worst ores will make nuclear energy unaffordable. The great majority of uranium reserves today are more than ten times as rich. The bulk of Canadian reserves are the best in the world: at over 2% uranium, they are 200 times that rich.)

Another Wikipedia page one might go to for an overall summary of the issue is the page you get if you search on either "Emission Factor" or "Emission Intensity". It gives the carbon emissions of every kind of power generation. And the intensity for nuclear is given as 65 g/kWh - that is, about 6.5% of the 1000 g/kWh that is the rule-of-thumb for coal-fired power, barely 10% of that for gas. This is a very little higher than the 2%-5% cited by the pages of actual nuclear companies, and it comes from the same University of Sydney study I link to above.

This very low figure for nuclear emissions is undisputed on the "talk" page for the article.

Few Activist Besides Caldicott Use Her Numbers

"Wikipedia don't get no respect", of course, but perhaps it's a little more telling that the recent New York Times article on Obama's decision to support the first nuclear reactors in 35 years stimulated one of the Times' "Room For Debate" pieces on the topic. And nobody in the whole debate, pro or con, made the claim that nuclear power did not greatly reduce carbon emissions. What was very interesting to me is that this is one issue that is not divided along right/left partisan lines. The right-wing idealogues of the Cato Instiute wrote "Stop Nuclear Welfare" against it; the right-wing nutjobs (my opinion) of the American Enterprise Institute were guarded about the financial risk, but assigned it all to "bureaucratic and regulatory" interference and on the whole were all for it as "reliable clean energy". But none of the "anti" opinions attempted the "it doesn't save on carbon" argument.

So, I suspect that anti-nuclear activists who must argue their points in open forums, rather than their own books and private web pages, are backing away from the thesis that Dr. Caldicott embraced; it all appears to come from the one paper, and that paper has proven very hard to defend against other studies.