Here are three videos that describe the passive cooling mechanisms in the AP1000 during a blackout (total loss of onsite and offsite power); passive core cooling[1], passive containment cooling[2] and passively cooling spend fuel[3].
Basically the reactor will cool itself without operator intervention and without power for 72 hours. After 72 hours the operator must use various on site equipment and coolant supply to maintain cooling and/or provide coolant from some off site source.
72 hours... is certainly a lot more than the 12 or so they had at Fukushima. But it's really not that much considering just how many things can go wrong at once.
I wonder what prevents them from capturing the steam from the spent fuel storage and running it through a condenser that is connected to a ground heat exchanger (basically tubes run into the earth)
Everything is natural convection or gravity-fed in a blackout station scenario, so there's no way to pump the steam back down into the ground. Also, the ground wouldn't dissipate the heat from the steam fast enough, and would quickly reach a hot equilibrium temperature. It's basically a low capacity heat sink quickly saturated.
True, but space is generally not a problem at nuclear plants because there is a huge area around them to protect against the accidental release of radiation. Fourier's law for heat conduction would suggest that if you had enough surface area in just steel pipes you could dump a megawatt or more of excess heat into the surrounding atmosphere. In an unconstrained atmosphere (like your heat exchanger structure is outside) that would induce an air current that would further enhance the ability to dump heat into the air passively.
If you had a larger water reserve you could also build the equivalent of a Watt steam pump that could use the heat in the steam to pump water from the reserve into the cask pond. Time to run some simulations to see whether that is doable at all.
Exactly what I was thinking as I read the article, why not just slap the equivalent of a giant passive CPU cooler on it. At that scale you should be able to generate real airflow from the convection currents, making the whole thing more efficient. And you've got a basically unlimited atmosphere to dump the heat into.
Build it on the coast and put the heat sink in a shallow tidal pool fed from the ocean and you could have a massive evaporative cooler. But then you have the very real possibility of a tsunami damaging/disabling the system somehow (e.g. covering it with debris that make it less efficient, washing part of it away.)
If you had a very large, very deep lake though, that should be incredibly efficient. Or a river with a good flow rate.
I'm not sure how much heat a typical commercial nuclear reactor designed for power generation produces, but the reactors at Hanford and Savannah River (used for plutonium production, not power[1]) generated heat measured in gigawatts[2]. That's why they were built next to large rivers - because of the massive cooling requirements.
Maybe you were taking that into account, or maybe heat production drops by a factor of 1000 during an unexpected shutdown, but I read your comment as being that the design you have in mind is limited to a few megawatts.
I was specifically commenting on the spent fuel pool which generates much less heat (at most about megawatt as far as I can find in the literature). However those pools have been a source of issues at Fukishima as there were some indications that the pool could reach criticality under the loss of cooling. (which rapidly escalates the heat production!) The referenced video talks about cooling the spent fuel passively for 72 hrs and then having to turn on a pump to add more water. (it doesn't say where the steam is going, presumably its being vented) So that got me thinking about letting the steam go through pipes forming a heat exchanger (either in the ground or in the air, and after this discussion the air would be better) so that as it condensed it could be returned to the spent fuel pond rather than lost.
[1] https://www.youtube.com/watch?v=FCorzfw5liQ [2] https://www.youtube.com/watch?v=ghy9aba3kHU [3] https://www.youtube.com/watch?v=jzBe_kwIs28
Basically the reactor will cool itself without operator intervention and without power for 72 hours. After 72 hours the operator must use various on site equipment and coolant supply to maintain cooling and/or provide coolant from some off site source.