Ideally, you overbuild and have some flexible consumer of electricity that can arbitrarily ramp consumption up or down to suck off any excess supply whenever inflexible basal civilization demand doesn't match supply, e.g. ramp up when everyone is asleep and ramp down if a reactor needs to go offline.

This is called demand response, https://www.energy.gov/oe/activities/technology-development/..., and it's probably as important of a component to future grid management as energy storage.

Right. A useful sink for surplus energy could be something like surplus heat or electricity to synthesize carbon-neutral fuels. We're not likely to get large electric aircraft anytime soon.

Electric aircraft doesn't really strike me as this kind of customer. Aircraft has relatively inflexible fueling demands based on customer demand that is reserved weeks ahead of time.

Bitcoin mining is commonly promoted as this kind of ideal electricity consumer. Hydrogen production could be another.

However, I think more interesting for this community would be to build some kind of demand response general computing datacenter. Basically, sell computing resources with poor availability guarantees, but at a cheaper rate than standard datacenters.

i.e. we'll run your batch jobs very cheaply, but we can't give you strong guarantees about how long it will take because we have to wait for excess grid electricity to have the energy to run them. Best case that's every night. Worst case we won't be able to run anything for weeks because Texas is going through another freak winter storm.

> Electric aircraft doesn't really strike me as this kind of customer. Aircraft has relatively inflexible fueling demands based on customer demand that is reserved weeks ahead of time.

I think you misunderstood. The parent comment implied that excess electricity could be use to produce synthetic fuels (via carbon capture) to power conventional airplanes, ships or other machinery.

I did misunderstand. Thanks!

If you have datacenters on multiple grids (and applications that are favorable to being moved around), you could direct traffic to where energy was abundant. This is not uncommon for large commercial sites; you can usually get a better rate from the utility if you commit to demand response, and it's not too hard to shift load if you're already doing multi-site for reliability/continuity)

…or maybe a censorship resistant peer 2 peer cash that could also subsidise these plants

A bit of pumped hydro storage would also satisfy those needs.

PSH could not bridge demand for long enough. Maintenance cycles on nuclear reactors can be quite long.

Make it 8 stations instead of 7 to have an excess of energy and then use the extra energy, when there's not enough demand, to produce fertilizers or other energy-intensive goods.

I suspect in today's world of maximizing capital efficiency, and being able to contract in large numbers of people, the very expensive power plant won't be offline for long for maintenance.

They'll probably shut the plant down, and have 500 workers come in to replace everything that needs to be replaced, and then power it up again within a few days, and let it run for another 6 months or so.

Everything will be planned on workplans, and most won't require specialist knowledge... Eg. "Replace pump 205 in building C with this pump. Required skills people: 2x plumber, 2x electrician".

That is exactly how nuclear outages work. Everything is planned in a resource-loaded schedule and many of those resources are just there for the outage.

Finland is actually somewhat known for running short (1 year) cycles with very very short (<10 days for refuelling is not unheard of) refuelling outages every other cycle alternating with slightly longer maintenance outages. A more typical scenario in the US would be 18 month cycles with a 30ish day outage. Some plants are moving to 24 month cycles with slightly longer outages.

Capacity factor for nuclear plants in the US is consistently around 93% (similar figures in many other countries as well), which is significantly higher than other generation sources.

Most reactor downtime is due to refueling, since that requires cooling down the core through several stages. AFAIK it's not practical to do it safely and economically in under a few weeks.

Most reactor maintenance is already done during refueling so there isn't much more room to optimize that downtime. (though that may change the older a reactor gets)

Most reactor downtime is actually for maintenance during refueling windows. As I mentioned in another comment, there is experience doing refuelling-only outages in <10 days. If everything goes well, you can startup a PWR from cold shutdown to hot full power in 3 12 hour shifts

Rushing maintenance costs more, and nuke operating cost is already radically more than renewables + storage.