Recently, I read an interesting piece about the reasoning process used by the man behind Tesla and SpaceX, Elon Musk: first principles reasoning. In this mode of thinking, espoused by Aristotle among others, one begins from the “first principles” or foundations of the problem to make the case. Musk himself has used the method to calculate the minimum costs of rockets and batteries respectively based on nothing more than their material costs: if a rocket or a battery contains such and such amounts of such and such materials, and the market prices for those materials are such, then the minimum cost for the rocket or battery can be easily calculated. As a result, Musk says he figured out that in principle a rocket needs to cost only about 2 percent of what it costs now; and battery packs could be had for $80 per kilowatt hour instead of current $600. In the article, Musk makes an important point: analogies and comparisons to what exist are fundamentally limited approaches, and in analysis they should be supplemented with other approaches, such as first principles thinking.
Which is what I’m now going to do to analyse what nuclear power might cost, Elon Musk style. The accepted wisdom, certainly among anti-nuclear activists, is that nuclear power is expensive and will forever remain so. (No matter that their preferred alternatives are often more expensive, when all the costs for equivalent end result are tallied.) This is somewhat puzzling, since in principle nuclear power shouldn’t be that expensive: while it requires large facilities, these facilities can pump out enormous quantities of energy from very, very small material inputs, with very low volumes of waste that can be controlled relatively easily.
So let’s begin the first principles analysis and start thinking how cheap nuclear power could be. As a starting point, I’ve used a widely available Environmental Product Declaration from Swedish Forsmark nuclear power plant. It conveniently lists the material and energy inputs required in building, operating and dismantling a nuclear power plant, distributing its electricity and disposing of its waste. Furthermore, it provides the details in an ISO 14000 certified manner per kilowatt hour produced (see Table 7 on p. 22). After some hours googling for raw material prices and making some conservative assumptions whenever exact data was difficult to find, I was able to come up with an answer:
Price floor of electricity could be as low as 2.1 US cents per kilowatt hour, or $21/MWh (19.2 €/MWh).
Certainly not too cheap to meter, but even if we throw in something to account for the unexpected, Muskian nuclear energy still comes out as, well, rather cheap. (For the spreadsheet I used in the calculation; Materials required for nuclear electricity and their costs, XLSX spreadsheet. Feel free to play around with it, although dropping a comment here would be nice if you use it to publish somewhere.)
Now, of course this is somewhat of an absurd calculation, but as Elon Musk himself noted, first principles reasoning is still an informative exercise. It is hard to believe nuclear electricity could be much cheaper than this; and one can suspect prices very much above this are a result from factors not inherent in the technology itself. Indeed, as a research review commissioned by none other than Friends of the Earth UK found out (PDF link), nuclear electricity costs are heavily dependent on terms of financing: using lower discount rates that Her Majesty’s Treasury recommends for decisions that have long-term societal implications — like climate change mitigation — the cost of nuclear electricity could possibly be halved from current estimates. Although policies required for this to happen are not realistic right now, such calculations illustrate how electricity costs are very much influenced by factors other than technology itself.
Furthermore, first principles reasoning provides a valuable comparison to figures thrown around by Elon Musk and others. If batteries have a price floor of $80 per kWh, and — I’m making assumptions here — solar electricity has a price floor of
$10 $0.1 per kWh produced, the lowest possible cost for equivalent service from a solar panel-battery combination would still be anything from $30 to $90 $20 to $80 or more, depending on assumptions on how much energy storage is required. (Thanks to Proteos for catching this very dumb mistake.)
Without technological innovation, these price floors are unlikely to budge and about as unlikely to be reached in practice. When unabated coal electricity can be had in practice for $30 per megawatt hour, it is easy to see why there is still need for more innovation so that even poor countries can provide enough power for their peoples without having to resort to fossil fuels.