Zane SelvansJigar Shah (current head of the US DOE's Loan Programs Office) recently wrote a long enthusiastic thread about the role he sees for #nuclear power in the US energy system. I've been surprised and kind of confused by this, given the cost overruns and delays associated with recent attempts to build new nuclear in the US, and the relative immaturity of smaller factory-built reactors that some hope can get onto technology learning curves and so reduce costs over time.
Some of his nuclear enthusiasm seems to come from pessimism about the prospects for clearing the interconnection queue traffic jam (~1 TW of renewables is currently waiting) and the difficulty of permitting and building out new transmission at the scale required by scenarios for rapid decarbonization through electrification. Which... yeah. Fair enough. Interconnection & transmission are currently very hard.
What surprises me is that he thinks nuclear would be *easier* -- it's got plenty of its own regulatory and permitting baggage!
He briefly talks about the problem in a more generic way in this podcast: On a zero-carbon grid *something* ultimately has to manage the non-dispatchability of renewables, and almost by definition, that thing is going to be utilized at a low capacity factor, and will raise costs relative to bulk renewables e.g. solar PV at $20/MWh.
We can build a continent-spanning super grid to provide geographic and resource diversity, and connect it to load centers, but that'll be expensive and slow and often that grid capacity won't be fully utilized.
Long-term electricity storage close to remote generation can improve transmission utilization and load following, but also won't be fully utilized.
Large banks of batteries, or pumped hydro, or any other capital intensive energy storage that are only on a seasonal basis, or during extreme gaps in renewable generation end up adding huge costs on a per MWh basis.
Engineered/Enhanced geothermal probably also needs transmission, and would also be pressed to operate at a high capacity factor to amortize up front costs over lots of MWh.
In theory nuclear is attractive because you can imagine dropping it in to replace existing centralized supply (coal or gas plants) without needing to build anything else, and get clean, dispatchable replacements. But again in a world where low-cost renewables dominate on an energy basis, even ignoring permitting, those will be some very expensive MWh, since they'd probably be load-following and not operating flat-out like most nuclear plants do today.
(though there are some designs for coupling thermal storage with nuclear so the reactor can run at constant capacity, while the electricity generation ramps up and down each day)
Webinar – Natrium: Latin for Sodium, Big for Advanced Nuclear
But then there's also "dispatchable load" or "virtual power plants" AKA demand response: Anything that lets us control when & where electricity is used such that it follows renewable energy supply instead of the other way around. This is an almost entirely information-based strategy, that centers on coordination rather than infrastructure. What's the potential scale in MWh? On what timescales will it operate? How cheap can it be?
Pushing dispatchable loads into industrial processes kind of just displaces the underutilized capital investments into a different domain. If you've got a bunch of fixed investment sunk into a factory or a refinery you want to run it at full capacity -- just like a power plant -- otherwise those fixed costs are amortized over a lower volume of output.
But the residential, transportation, and maybe to a lesser extent commercial sectors are just flagrantly wasteful and overbuilt.
Nobody is even pretending to care how much we spend on cars given how infrequently they're used. Lots of US houses have a giant tank of water that's constantly being heated regardless of whether anybody intends to take a shower or do laundry.
The combined total power output of all US cars is ~10x that of all US power plants.
Anyway, I don't really know where I'm going with this.
There's a fundamental problem to be solved if we're going to rely heavily on non-dispatchable low-cost energy, which I think does mean delivered energy will end up costing more than the cheapest bulk energy. But how much more?
And it seems like a very open question what will end up being the winning combination(s) in terms of economic, technological, and political feasibility.
Chris Nelder 🤘@ZaneSelvans There's a whole lot to unpack there... In any case, I have not seen any studies that could actually answer those questions. And even if they existed, they'd have to make some highly speculative assumptions about future costs and capacities. IOW, I don't see us taking a techno-economic approach to the question, which leaves political feasibility as the driving force. As the nuke proponents know very well.
@chrisnelder It seems like a free-for-all right now (which isn't necessarily bad!). I wouldn't be surprised if we end up with wildly different zero-carbon solutions in different countries. Or even different states. And I wonder what the consequences of different regions taking different paths will be.
@ZaneSelvans Well I guess the theory--at least in the good ol' federalist republic known as the US of A--is that the several states are laboratories of grid power. But if you were going to forecast it, you could do worse than just taking a straight political approach. At least, it looks to me like only the red states are interested in building more nukes.
@chrisnelder The southeastern ones certainly don't seem overly concerned about electricity costs!
@ZaneSelvans As red as red gets... Just some good ol' boys, never meanin' no harm.