[paulsutter]

Could you clarify the "huge problem of load-following" for nuclear power plants? My understanding is that it's mostly a licensing issue rather than a technical issue, and in practice hasn't had a big impact on CF[1]

[1] https://www.oecd-nea.org/nea-news/2011/29-2/nea-news-29-2-lo...

"Most of the currently operating Generation II nuclear reactors were designed to have strong manoeuvring capabilities. Nuclear power plants in France and Germany operate in load-following mode."

"The economic consequences of load-following are mainly related to the reduction of the load factor... In France, the impact of load-following on the average unit capacity factor is sometimes estimated at about 1.2%."

"Licensing of load-following is specific to each country. In France and in Germany, for instance,load-following is considered early in the licensing process, and no further authorisation needs to be obtained by the utility to operate in manoeuvring regime. In other countries, load-following restrictions apply: for example in the United States, automatic load-following is not authorised"

[barney54]

Load following is a technical issues--nukes are limited in the amount they can load-follow. But winds cannot load follow at all and solar can help some, but doesn't necessarily help cover peak daily load. For example, yesterday in California the system peak load was at 5:49 pm, but solar PV peak was at 9:44 am. data: http://content.caiso.com/green/renewrpt/20151102_DailyRenewa...

Nukes aren't great for load following, but they can load follow some--which is better than wind and solar. To really load follow, you need hydro or nat gas turbines (though nat gas combined cycle can load follow).

[tempestn]

Another possibility would be to add massive storage to the grid. Maybe it's pumped hydro or this kind of thing: https://news.ycombinator.com/item?id=8646787. Or perhaps more realistically, if someday soon there are millions of electric vehicles connected to the grid at any given time, especially during current off-peak hours, that storage could be leveraged for load shaping.

[filmor]

Most wind turbines can be regulated very fast in negative direction, i.e. you can change the angle that the blades have to the wind to reduce the force. This is needed for the turbine to survive storms but it can also be (and is) also used to "help the grid".

[datenwolf]

There are also physical limitations with load following. In a fission reactor you're dealing with a steady state of influx of fission products (by the primary nuclear reaction) and the delayed decay of those products. In fact a large portion of the heat generated by a fission reactor comes from the delayed decay of the fission products. Furthermore some of the fission products are very efficient neutron catchers; notably Xe135 which is so efficient in catching neutrons, that its buildup in a reactor is called "xenon poisoning"; the accumulation of neutron catchers in a fission reaction is called "neutron poisoning".

What this comes down to is, when you ramp down a fission reactor, there's some internal inertia in its internal workings, that will actively prevent it for some time from being able to ramped up in a safely manner. The quicker the shutdown, the larger the amount of neutron poisoning and the longer you have to wait before ramping it up again. This leaves you with a nice second order differential equation coupling the power output modulation factor with the period of that power modulation.

The period of the power modulation is 24h, following the daily load swings, so for a given reactor that gives you only so much load following capacity to stay within safe margins.

[DennisP]

That's one advantage of molten salt reactors: the Xe135 bubbles right out, making load following a lot easier. The Transatomic white paper talks about this on page 44 (pdf): http://www.transatomicpower.com/wp-content/uploads/2015/04/t...

[PolLambert]

Load-following is both license and technical related in a chicken-and-egg-way. You need to incorporate load-following into your reactor design. More specifically: the speed with which a reactor can modulate its output while staying within nominal operational bounds is a design parameter. If licensing doesn't allow load-following, you're not going to give your reactor that feature.

[yk]

Thing is to reduce the electrical output of a reactor, you can either reduce the thermal output, but then the entire reactor core needs to cool, or you have to reduce the efficiency of the electricity generation. In the first case we are talking about a timescale of days and in the second case you are burning fuel for nothing. So NPRs are not designed to do much of the second.

[olau]

The real problem is that to get anything resembling economical use out of a nuclear plant, you need to output 100% almost all the time. New plants are too damn expensive to just let them sit there.