35
$\begingroup$

What would cause a nuclear power plant to break down after 2000 years, but not sooner?

The setting is a society in a dark age. Their ancestors were considerably more advanced than earth currently is. The society has been stagnate at close to the tech levels of modern earth since. For the past two millennia, they have maintained the existing power plants, but cannot create new ones.

Schematics for everything in the plant mostly still exist, but are written in dead languages spread across ancient computer systems that no one understands well.

Now one of the most important plants has broken down.

What part of a power plant could work for 2000 years, but stop at that point?

My initial thought was that the uranium was depleted. But Uranium 238 and 235 have half-lives far too long. Uranium 234 only lasts ~200K years, but that still is too long.
Then I considered some mechanical or electronic failure, but it can't be something which would happen in the first few decades, or the plant managers would have a recent record of how to fix it.

Specifics about Power Plant:

The underground plant uses a super material 'durium' that is near indestructible. The ancestors used durium for any non-flexible permanent solid in the plant, such as:

-The walls, ceiling, and main structure

-The tube holding water that gets boiled and turbine sticking into it

-The water cooling tubes

The fuel is of some kind or quantity that lasts longer than 2000 years.

Bonus points if you can think of a reason why all the power plants would break down over the same century.

$\endgroup$
  • 6
    $\begingroup$ Related: technologyreview.com/s/544211/… "While there are significant unknowns around extending the lives of nuclear plants built in the 1970s and 1980s, most people in the industry believe that the reactors can operate safely for 80 years." $\endgroup$ – Jerry Jeremiah Aug 14 at 23:22
  • 24
    $\begingroup$ In the reactor, nuclear fuel depletes at a faster rate, and 2000 years is actually too long. In fact, no part of nuclear reactor can last that long. $\endgroup$ – Alexander Aug 14 at 23:29
  • 7
    $\begingroup$ A civilization which can make nuclear fuel can also make a nuclear reactor. Nuclear power plants need refueling on very much shorter periods than 2,000 years. $\endgroup$ – AlexP Aug 15 at 0:35
  • 5
    $\begingroup$ I feel like I could write an entire story surrounding the monumental engineering efforts to make a nuclear reactor which produces energy on the timescale of the great pyramids! $\endgroup$ – Cort Ammon Aug 15 at 4:54
  • 3
    $\begingroup$ What is wrong with running out of fuel? a half life for 20K does not mean at 19999 years you still have 100% uranium and on Dec 30th half in changes to lead in an instant. nuclear fuel still needs to be replenished and most powerplants don't store a huge amount of fuel on site. $\endgroup$ – John Aug 15 at 18:20

19 Answers 19

150
$\begingroup$

Software licencing.

Software for your automated maintenance and refuelling robots was under a 999-years licence (similar to common law permanent lease) . After it ran out, the licence was automatically renewed for another 999 years, payment pending. As no payment was received during that time the licence lapsed, and the plant went into a controlled shutdown until all contractual obligations are paid in full. Which would be difficult because the old banking system is now defunct as well.

$\endgroup$
  • 4
    $\begingroup$ Minor nitpick - the software made for a secure facility like power plant wouldn't auto-renew (it shouldn't have an internet connection), so it would just lapse. $\endgroup$ – Carl Kevinson Aug 15 at 18:06
  • 23
    $\begingroup$ @CarlKevinson The software for home banking shouldn't store their client's passwords in plaintext, and we still get leaks of the sort way more often that we would expect... $\endgroup$ – T. Sar - Reinstate Monica Aug 15 at 18:24
  • 15
    $\begingroup$ @CarlKevinson not really, the way such licensing systems are currently implemented (e.g. mainframe systems in various facilities) it doesn't necessarily need an internet connection (it can be 'upgraded' by a particular keycode entered on-site) and the exact scenario described - your licence has expired but the system continues to work for another period with a warning that payment is overdue and system will stop working in X days - is exactly how certain current enterprise licensing works; you don't want accidental downtime causing losses because the licensing servers are broken, etc. $\endgroup$ – Peteris Aug 15 at 20:19
  • 59
    $\begingroup$ They bought a perpetual license but the software doesn't support perpetual licenses, so the salesperson just typed 999 in the "years" box. $\endgroup$ – user253751 Aug 16 at 0:20
  • 14
    $\begingroup$ Yeah, this is exactly how civilisation ends. $\endgroup$ – DrMcCleod Aug 16 at 15:22
60
$\begingroup$

Remember Y2K?

Software is hard, programmers are human.

If you want your readers to know exactly what went wrong, you can babble at length about data types or speculative branching or null references or race conditions.

But I propose that if your characters don't know what broke or how to fix it, then maybe your readers shouldn't either. The software has a bug, and the power plant isn't power planting.

To keep it vague but dangerous, the problem could be with a safety system. An engineer finds a single screen with the message:

Error 32801: An unsafe condition was detected and the reactor has been automatically shut down. Do not attempt to restart the reactor.

$\endgroup$
  • 13
    $\begingroup$ Perhaps you could tie in something similar to the 2038 problem: the size of the memory slot they used to store the date runs out and the date overflows into several decades ago. en.wikipedia.org/wiki/Year_2038_problem $\endgroup$ – AmbientCyan Aug 15 at 2:41
  • 7
    $\begingroup$ @AmbientCyan Not only could the date fall back to 1970-01-01 or something arbitrary like that, but there is also the much greater risk that you would overflow data into the next unit in your stack. Suddenly your variable "bolEmergencyShutDown" switches from 0 (false), to 1(true), and the whole plant activates its emergency shutdown system. $\endgroup$ – Nosajimiki Aug 15 at 15:58
  • 4
    $\begingroup$ @Nosajimiki Note that numbers in computers don't generally work like that. They have to fit in fixed-size "boxes". 1111111111111111 + 1 (in 16 bits) doesn't go to 1 0000000000000000 and put the 1 in the next location, it just goes to 0000000000000000 and discards the 1. (It may also remember that it discarded a 1, but most software doesn't check for that) $\endgroup$ – user253751 Aug 16 at 0:21
  • 2
    $\begingroup$ @immibis Computers don't "normally" work like that because most modern compilers automatically include canaries in their stack. That said, anything written in a low level language that requires manual stack management can have what is called a stack buffer overflow if you are not careful to prevent it. FYI: This kind of error is where the original Stack Exchange site got its name, "Stack Overflow". en.wikipedia.org/wiki/Stack_buffer_overflow, Also, infrastructure designers are kinda notorious for using low level programming languages they don't fully understand $\endgroup$ – Nosajimiki Aug 16 at 3:21
  • 4
    $\begingroup$ @Nosajimiki Computers don't "normally" work like that because all the instructions that you might use to add two numbers put the result into fixed-size spaces. It is not because the compiler adds additional space for safety. An integer overflow does not overwrite adjacent data in memory. It is possible for an integer overflow to indirectly cause another memory location to have incorrect data, but it is not certain. $\endgroup$ – user253751 Aug 16 at 3:44
32
$\begingroup$

You don't need any special materials to make a reactor last a long time. The Gabon natural reactor consists of uranium deposits in sandstone. Approximately 2 billion years ago they went critical. This is because U235 has a shorter half life than U238. So 2 billion years ago there was relatively more of it, and it could go critical with natural water moderator.

The deposits would operate at a few kilowatts for a short time, perhaps a half hour. Then the water would boil, decreasing moderation. Then the reactor would shut down and cool for about 3 hours. This cycle continued until the fuel was depleted. The sandstone then sat for 2 billion years until it was discovered in the 1970s.

This 1/2 hour on 3 hours off cycle continued for between 100 thousand and 1 million years. It was busy breeding, so U238 was getting converted to Pu239, which supplied additional fuel. Interestingly, when the reactor finally shutdown due to fuel depletion, the remaining Pu239 eventually alpha decayed into U235, which somewhat masked the depletion of U235 that originally tipped people that something was up in Gabon.

This has inspired a lot of designs for nuclear reactors. None has been built as yet. The idea is, you would have a block of stone about 8 or 10 meters on a side. You pump cold water in here and get hot water out there. It breeds new fuel until it goes far enough that it can't usefully do that any more. Depending on the design and the use load, it could in principle be made to last for 2000 years. Though it's far more likely to be designed to last 40 or 50 years in the first design, since that will tend to allow for new designs later.

You pump cold water in. If you pump it in at the wrong temperature, high or low, or too fast or too slow, or at too high or low pressure, the reactor just stops. And you get boiling water out. You then run a turbine.

So if these things were mass produced in a factory, then shipped all over, they could be coming to the end of their useful life. They were all designed to provide the same power for the same time. Subject to a few of them being used slightly less than the others, they will run out of fuel at about the same time.

$\endgroup$
  • 5
    $\begingroup$ en.wikipedia.org/wiki/Traveling_wave_reactor $\endgroup$ – David Tonhofer Aug 15 at 16:06
  • 5
    $\begingroup$ I like the concept of the traveling wave reactor. But, having spent the last 29 years in the industry, I'm eager for somebody else to pay to build the first one. I've now been on three separate projects of a reactor design that got a tonne of money spent only to be cancelled. Accurately predicting how the as-built system will operate is a hard problem. $\endgroup$ – puppetsock Aug 15 at 17:38
  • 4
    $\begingroup$ I believe you. The point where extreme materials science meets engineering meets bleeding edge experimentation meets economics meets politics meets changing regulation meets long-term commitment meets people's risk aversion. Not sure we are in an epoch which can sustain that. $\endgroup$ – David Tonhofer Aug 15 at 17:53
  • 2
    $\begingroup$ @puppetsock welcome to modern economics - building something is a incidental byproduct of funneling money from (usually taxpayers) to contractors. If your workers happen to build something while this important process is ongoing, then that's a bonus $\endgroup$ – gbjbaanb Aug 17 at 13:15
  • $\begingroup$ @gbjbaanb Hey, at least it's a step above funneling money from consumers to investors, because something actually gets done along the way. :P $\endgroup$ – user253751 Aug 18 at 22:30
16
$\begingroup$

This may not be what you were originally thinking, but perhaps some sort of natural disaster strikes which disables the plant. Even if the structure of the plant is made of an indestructible material, it may be vulnerable to less direct damage such as flooding.

For a real life example, we can look at the Fukushima-Daiichi accident which happened in 2011. An earthquake struck off the coast of Japan, and while the earthquake did not damage the structure of the plant, the resulting tsunami from the plant caused flooding of key components which eventually led to a meltdown:

On detecting the earthquake, the active reactors automatically shut down their fission reactions. Because of the reactor trips and other grid problems, the electricity supply failed, and the reactors' emergency diesel generators automatically started. Critically, they were powering the pumps that circulated coolant through the reactors' cores to remove decay heat, which continues after fission has ceased.[10] The earthquake had generated a 13-15 meter high tsunami that arrived approximately 50 minutes later, which over-topped the plant's seawall, flooding the basements and disabling the emergency generators. The resultant loss-of-coolant accidents led to three nuclear meltdowns, three hydrogen explosions, and the release of radioactive contamination in Units 1, 2 and 3 between 12 and 15 March. (Wikipedia)

$\endgroup$
  • 6
    $\begingroup$ It's worth noting that many nuclear power plants are unfortunately built right along fault lines. The reason is that they need water for cooling, so are often built near rivers, which often run along fault lines, if not necessarily very active ones currently. $\endgroup$ – Darrel Hoffman Aug 15 at 15:55
  • 1
    $\begingroup$ @DarrelHoffman Well, only the largest rivers have anything to do with fault lines, but it's definitely enough for the scenario the OP is asking about :) Most importantly for the scenario, while the damage to Fukushima and its surroundings was relatively small, it's also essentially irreparable - if you can't build a reactor from scratch, you can't recover a melted down reactor vessel either. And reactor vessels (even chemical reactors, much less nuclear) take very advanced metallurgy, among other things. So definitely +1 to both you and Ambient :) $\endgroup$ – Luaan Aug 16 at 7:19
12
$\begingroup$

Transmutation of the water

Both fission and fusion reactors produce copious quantities of neutrons. In fission reactors, the "fast" neutrons are slowed by moderator so that, ideally, most of the split another fuel nucleus. Fusion reactors (at least current conceptual designs), the neutrons escape the plasma, then thermalize and capture in the surrounding water bath, which produces the heat to run the turbine. A huge engineering concern is that some of those neutrons capture on your structural elements and transmute them, leading to eventual material failure.

Your structural elements are all made of durium, so presumably they are immune to neutron degradation. But suppose the designers use a closed-loop water system. The designers probably didn't think thousands of years in advance, and they would have known to just replace the water eventually. Over time, the water will transmute, in two potentially harmful ways:

1) Hydrogen eventually turns to He-3

The hydrogen in the water will first convert to deuterium, then tritium. Tritium naturally decays to He-3, which will eventually escape the system, unless your durium also contains He better than any material we have. Even if not, you'll eventually convert enough of the water to gas that it won't cover things anymore, and either you overheat (fission) or no longer transfer power (fusion). In fusion's case, this would be a gradual decrease in efficiency, whereas for fission, gradually rising heat could lead to a sudden failure at some point.

This is easy to notice if the descendants can view the water volume, and just top it off every now and then, but maybe the close-loop water system is not visible to inspection.

2) Buildup of HF

Let's say the descendants notice that the water needs topping off every now and then, but don't ever empty the whole tank. The oxygen in the water will eventually capture enough neutrons to become O-19, which decays into F-19. The fluorine will bond with hydrogen to become HF, hydrofluoric acid, which is one of the nastiest substances you can dream of. Eventually the concentration will build up enough to be a relatively high-strength acid. Maybe durium isn't quite indestructible, and HF is one of the few weaknesses. Or some part of the water system is made of something else (e.g. glass or fused silica, like a viewing window).

This sort of defect would affect all power plants on similar timescales (assuming the volume of water per power capacity is approximately constant, which seems reasonable), although the point where some component, weakened by the HF, eventually gives way will be variable.

$\endgroup$
9
$\begingroup$

In Asimov's Foundation series, the Empire's nuclear power stations were controlled using vacuum tubes and relays.

Out-of-universe, this was because solid-state transistors had not been invented when the story was written. In-universe, this might have been for lack of solid-state transistor technology, or because vacuum tubes and relays are easier to understand, make, and replace than bits of integrated circuits. Vacuum tubes and relays can also be designed to handle much larger currents, voltages, and power flows than integrated circuits.

In-universe, the Empire lost the technology to be able to replace even "a single quartz D-tube". If enough of those essentially-glass vacuum tubes were physically smashed, an entire nuclear power station would be rendered out-of-commission.

In real life, expensive vacuum tubes have cathodes with a limited life. For example, the travelling wave tubes on weather and communications satellites typically have a design life of 12 or 15 years. The limiting factor is the amount of metal in the cathode. (This does not have to be the limiting factor. Some tubes are predicted to have cathode lives of hundreds of years. But satellite parts have strict mass budgets. Every tenth of a gram counts, so only a modest amount of extra cathode metal is included.) When the tube is operating, the cathode is heated to a high temperature, and the metal slowly evaporates at a very predictable rate. The satellite will need to be decommissioned or replaced before the metal's reservoir runs out.

The vacuum tubes in the control system of the original poster's power plant were engineered to have a consistently long life. But eventually the cathodes will run out of metal, and the vacuum tubes will stop working. Many vacuum tubes will have experienced similar times at operating temperature, so that they fail after a similar length of time in all of the power plants. As gbjbaanb pointed out in a comment, the consistently long life does not need to be 2,000 years. "All you need is a box of replacements and instructions of what to do to fit them... after 2,000 years the priest-tech goes to find a new one and discovers the box of spares is empty. And, over the same century - all plants had only a similar number of repalcement parts, so will all fail around the same time, particularly if parts were shared out." This scenario is especially likely if the last vacuum tube factory was asked to make a large number of spares just before it went out of business; this would have both extended the life of the factory by a few years, and made it much more difficult to start a business to replace the factory.

In real life, the processes to make cathodes for high-efficiency vacuum tubes are notoriously finicky. It is very easy to accidentally contaminate the cathode material while manufacturing it. This is part of why there are so few companies capable of making space-rated vacuum tubes, and why the number of such companies is decreasing.

In the original poster's scenario, the vacuum tube manufacturers could well have been out of business for centuries. A great deal of technology would need to be reinvented to replace the failed vacuum tubes. Or the characters could try to get by with much more primitive vacuum tubes, and risk reliability problems in their nuclear power plants' control systems. For that matter, the original vacuum tubes might be much more space-efficient and energy-efficient. There might not be enough room or available "conditioned" power to install a jury-rigged control system.

To get an idea of how much technology might need to be reinvented, Chapter 5 (pages 47 - 102) of Scott Gilmour's Principles of Traveling Wave Tubes provides an overview of "Cathodes", with 30 scientific references. Some of the other chapters discuss "Electron Guns", "Reliability", and methods for modulating / using the outputs from "Electron Guns".

$\endgroup$
  • $\begingroup$ Incidentally, it was the transistor that stopped us from building huge space stations in Earth's orbit (vacuum tube systems with mercury-based solar power require human maintenance and large volume and energy; transistor-based systems allowed modern satellites). It might very well be that it was the lack of transistor-based systems that allowed Asimov's space exploration to really get off the ground; it would also force software engineers to focus on extremely reliable and compute-efficient systems, rather than relying on hardware "just getting better" and being easily replaced :) $\endgroup$ – Luaan Aug 16 at 7:53
  • 4
    $\begingroup$ Doesn't matter even if they don't have a 2000 year lifespan. All you need is a box of replacements and instructions of what to do to fit them... after 2000 years the priest-tech goes to find a new one and discovers the box of spares is empty. And, over the same century - all plants had only a similar number of repalcement parts, so will all fail around the same time, particularly if parts were shared out. $\endgroup$ – gbjbaanb Aug 17 at 13:19
  • 2
    $\begingroup$ @gbjbaanb -- Thank you for your suggestion. It makes the scenario much more plausible, so I have included it in the answer. $\endgroup$ – Jasper Aug 17 at 16:31
  • 3
    $\begingroup$ @Luaan From Arthur C. Clarke's "Superiority" (1953), a bit later than "Foundation" (~42-50): "The Analyzer contained just short of a million vacuum tubes and needed a team of 500 technicians to maintain and operate it. It was quite impossible to accommodate the extra staff aboard a battleship, so each of the 4 units had to be accompanied by a converted liner to carry the technicians not on duty. Installation was also a very slow and tedious business, but by gigantic efforts it was completed in six months." Broke the space fleet... $\endgroup$ – David Tonhofer Aug 17 at 20:35
7
$\begingroup$

Ran out of control medium.

While your reactor fuel may last a really really long time, those reactions are controlled by cadmium, hafnium, enriched boron, salt, or some other neutron absorber. As these materials absorb neutrons from the fuel, their ability to hold "extra" neutrons decreases. It may be that the fuel in question can last much longer, but the designers only engineered the control medium to last 2000 years. Once the control medium becomes too saturated or degraded, it will no longer be able to slow the rate of fission resulting in an eventual critical meltdown.

They would all break down at the same time because the amount of control medium was measured out for a 2000 year lifespan as per the national safety regulations for nuclear power plants that were in place when they were designed.

$\endgroup$
  • $\begingroup$ It won't cause the reactor to simply stop working, but to run out of control. It would be a very poor engineer indeed who built a reactor deliberately in this way. $\endgroup$ – puppetsock Aug 15 at 17:48
  • 1
    $\begingroup$ Nothing is designed to last forever. They were probably scheduled to be decommissioned after the 500 year mark hit, but by then civilization had already regressed and the value of a working reactor was worth more than the risk of continuing to run something that can theoretically keep going another 1500 years; so, everyone conveniently "forgot" they were supposed to shut it down. $\endgroup$ – Nosajimiki Aug 15 at 18:27
6
$\begingroup$

BUGS!

Not software bugs, but the icky kind that can get into those cracks and start chewing on those high temperature super conductive wires.

Once a few electro magnets quench, you can no longer sustain the fusion reaction.

(Fission reactors won't last half that long)

$\endgroup$
  • $\begingroup$ Well, you can have fission fuel last arbitrarily long (on human civilization timescales), but it comes with a cost to efficiency and economy. If we really wanted to build a fission reactor that lasts 2000 years, we could. That is, we could build the reactor itself - then there's all the other parts like the computer systems and turbine bearings that we couldn't. And of course, a nuclear fission reactor without computer systems and turbines are yet another thing that makes the whole thing even less efficient and economic... $\endgroup$ – Luaan Aug 16 at 7:49
4
$\begingroup$

Despite the described state being a stagnant society, the people's knowledge and skill would have continued to degrade even very slowly until finally there is no overall concern or ability to maintain the power plants any longer.

Why? Because most technology that is dependent upon the power plants would have long since failed. If there was no motivation to re-learn necessary knowledge and skill as a society before now, then their long-lasting apathy has simply turned into contentment with basic life as old technology fades away. A burned out light bulb that is no longer manufactured and costs too much money is never replaced, so they get used to a more natural daylight cycle... which is just an example/metaphor of all the other old tech. Pervasive computing devices like smartphones would have stopped functioning centuries ago. About the only thing left to power would be incandescent bulbs and basic motors.

In 2000 years of barely being able to maintain the power plants which were designed for long-term robustness, all of the devices and systems that relied on those power plants were far inferior in terms of material quality and longevity. New electronics would have ceased to be produced due to manufacturing failure long before 2000 years. Satellite systems would have failed, thereby rendering many dependent technologies nonfunctional. Basic maintenance as suggested in the question details is simply not enough to continue manufacturing necessary tech components any more than the nuclear science which has become obsolete. If a society cannot handle nuclear science/engineering, then they can't handle necessary material, quantum-mechanical, atomic, electrical, chemical sciences/engineering for all other tech.

Eventually their maintenance would simply be insufficient, especially due to depleted stores of necessary components.

If only a small number of individuals had continued their education and maintained books and past knowledge, enough to prevent the deterioration of all the other necessary tech surrounding the plant, then certainly the necessary nuclear-science knowledge would also have been maintained. In summary, 2000 years of stagnation for only maintaining nuclear power simply would not be stable. Either they would have re-learned necessary skills to maintain and replace it, or it would necessarily fall even further past the ability to maintain it.

$\endgroup$
4
$\begingroup$

Coastal erosion

A nuclear power station needs a lot of water for cooling. Almost all are on the coast for that reason - the warm water returned can't significantly warm up an entire ocean.

The problem is that coastal erosion is also a thing. Over 2000 years, the shorelines of much of the world have changed fairly significantly. Locations for power stations are chosen for stability, of course. However a change of currents could easily start to dump extra sand and silt around the power station, and with a not-properly-industrialised society they may well not able to keep the water intakes clear.

$\endgroup$
3
$\begingroup$

The AI that instructed the maintenance workers how to repair faults finally developed a fault itself.

Your dark age society could hardly be expected to maintain such an advanced piece of machinery in good working order without assistance, after all. Perhaps the AI was always part of the power plant's design, though it might have originally been only intended to supervise maintenance (to reduce the risk of human error) rather than to be in sole charge of instructing the maintenance teams; or perhaps it was a last-minute effort by a society that already realized it was falling into a dark age.

Assuming there was sufficient redundancy in the AI systems, it would even have been able to instruct the maintenance teams how to repair faults in its own systems, as would inevitably have happened many times over the centuries. But sooner or later, there was bound to be a fault in a non-redundant component, or multiple faults that took out too many components too quickly to repair in time.

After the AI failed, the power plant itself might continue to run for some time, so long as the only mechanical failures were ones that the maintenance team had encountered before. But eventually something new would go wrong, or society would gradually lose the knowledge of how to perform more than very basic maintenance; once the AI was no longer available, it would likely only take a few generations for this to happen.


Since you would like the problem to affect all the power plants over a relatively short period, you might wish to suppose that there was only a single AI that was supervising all the surviving power plants. Once it failed, the power plants would start failing, any time between a few years or a few generations later.

(In the scenario where the AI was always part of the design, this might have been a cost-saving measure; this would not be unreasonable if it was only meant as an additional precaution against human error rather than as an essential part of the system. In the scenario where the AI was created as the society fell into a dark age, they might have only had the resources to build a single AI. Or perhaps they intended to build several, but society collapsed before any additional AIs could be completed. Or perhaps all but one of the systems were cannibalized in an unsuccessful effort to keep society afloat.)

$\endgroup$
3
$\begingroup$

The plant, being built by those ancient geniouses of the past, was of course self-repairing. It has small robots (either nano bot or something bigger), that would check the whole plant regularly, immediately fixing any issues and counteracting deterioration. Now those robots cannot (of course) create matter. So they were supplied with stocks of the basic materials, from which they could build anything needed. And those materials, sadly, ran out some centuries ago.

Despite from not knowing anymore how to resupply them again, there is the issue that none of the materials can be produced in high enough quality anymore, as so much knowledge was lost. So even if the robots got fed the new supplies, they would reject it, as not suitable for fixing a plant.

(I hope you read the foundation by Asimov, if you are interested in long lasting (or not lasting) societies?)

$\endgroup$
3
$\begingroup$

I suggest you read or watch "City of Ember". It might give you idea.

In the movie/book there is a generator, made by "The Builders" (ancestors), that has broken down.

Simple reasons:

  • The generator has lasted for over its expected lifetime,
  • Lack of knowledge/tools to properly maintain the generator
$\endgroup$
3
$\begingroup$

It's a breeder reactor.

Reactors can normally operate continously for periods around a year before refuelling. In most reactors, that is a complex operation taking months, involving handling highly radioactive spent fuel rods, i.e. big headache. Some 1970s reactors (AGRs in the UK) were designed for on-load refuelling, without shutting down, but in practice most of them still shut down to refuel.

Trend since then is to move towards reactors with a service life of decades without refueling - for instance on submarines where there isn't much room to access the reactor in the first place. So, refuelling is such a huge hassle ... even today ... people avoid it if possible.

Now, based on the premise in the question, you could imagine a society perfectly capable of servicing external equipment like steam plant - as the Victorians could - but without the ability to open the core and refuel it. (Indeed, perhaps the core is fully sealed with no designed access).

All we need to get there, is a way to fuel the reactor for 2000 years instead of 30.

Enter the breeder reactor.

Only 0.7% of natural uranium (the U235) is fissile; the remainder (U238) is not. However, during normal operation, some of that U238 is converted into Plutonium, which is also fissile, and by tuning the flow of neutrons, you can control the rate of that conversion. Pu "burns" less efficiently, but it's usable and there are mixed fuel and Pu burning reactors operating today.

Thus you can extend the operational life until not only the U235 is gone, but most of the U238 has been converted to Pu, burned, and gone. Fiddle around with the numbers, and you can extend the working life from 30 years (U235) to millenia.

For example, if the original fuel was enriched to 1.5% U235 and burned in a fairly neutron-efficient reactor, slow enough to last 30 years, there is about 66 times as much U238 waiting to be converted to Pu. And so on, until after 66 times 30 years, the U238 is exhausted. The reactor simply stops because it has run out of fuel, and society has had enough of refuelling accidents and waste disposal centuries ago.

(A real nuclear scientist could nitpick many reasons why this won't work exactly this well in practice. But in principle...)

$\endgroup$
2
$\begingroup$

The concrete will decay after a couple of millenia, assuming they were able to repair/replace corroded metals. Foundation of the plant and the construct of the dome would irreperably fail.

$\endgroup$
  • $\begingroup$ True, but the containment is a safety feature - it doesn't have any other function. As it started to fail, the people would just remove the concrete dome. They probably don't even know why it's important (Chernobyl didn't have one either :P ). $\endgroup$ – Luaan Aug 16 at 7:55
2
$\begingroup$

Durium lasts for millions of years in normal circumstances... but just a couple of thousand years of hard neutron radiation makes it brittle in the parts closest to the nuclear fuel.

https://en.m.wikipedia.org/wiki/Neutron_embrittlement

Some of the atoms of the unobtainium matrix have been dislocated by high energy neutron collisions, weakening the crystal structure, and the nanotube lattice reinforcing it is riddled with nano-fractures. That durium just isn't as durable as it used to be.

Because of this, the structure of the building is still very sound, but that turbine is going to break and the pressure vessel could rupture if you look at it sideways.

$\endgroup$
  • $\begingroup$ In The Hunt for Red October, this exact scenario occurred with a valve in a nuclear reactor. The broken parts of the valve then got pulled along by fluid flows to a critical location. In effect, a nuclear reactor can "have a stroke". $\endgroup$ – Jasper Aug 17 at 16:58
2
$\begingroup$

I vote for fuel exhaustion. Just because Uranium has a long half-life doesn't necessarily mean an unreplenished supply would remain sufficiently radioactive over 2000 years to be able to generate enough heat to keep putting out usable power.

$\endgroup$
  • $\begingroup$ This, exactly. The half-life of Uranium is "If I leave this alone, it takes X time for half of it to decay." Using it in a reactor isn't leaving it alone. $\endgroup$ – Ghedipunk Aug 16 at 23:43
  • $\begingroup$ Fuel rods are rated for a decade at most, anyways, but they have a surprising shelf-life before they're used. They could just have enough of a backstock to refuel 200 times. $\endgroup$ – Ghedipunk Aug 16 at 23:45
  • $\begingroup$ The radioactivity has nothing to do with the power they can produce. Ironically, until they are spent fuel rods, which are full of short lived fission products, highly radioactive (for a few decades) and generate quite a lot of heat. $\endgroup$ – Brian Drummond Aug 17 at 13:42
2
$\begingroup$

My initial thought was that the uranium was depleted. But Uranium 238 and 235 have half-lives far too long.

Fuel depletion of a reactor is not based on the half-life of the fuel. Half-life determines how long it takes for spontaneous decay to occur. But a fission reactor does not rely on spontaneous decay. It creates a situation where there are alpha particles and free nucleons bouncing around. (An alpha particle is two protons and a neutron; basically a helium-4 nucleus.) When an alpha particle impacts a uranium atom, it creates a plutonium atom (plutonium has two more protons than uranium). The plutonium atom has a much shorter half-life. Nucleons cause different reactions, but the basic idea is the same. Particle meets fuel and forms a less stable atom.

This is why they talk about moving fuel rods in and out. They put the fuel rods inside a lead shield to protect them from the particles bouncing around. They pull them out to expose them to the particles, increasing the chance that they will start a fission reaction.

If the reactor is in active use, even if it is less use than that for which the reactor is intended, the reactor can simply run out of fuel eventually. The fission process consumes uranium and produces more stable atoms like lead. Eventually, even if all the fuel rods are out, the process will run out of excess alpha particles, nucleons, and/or fuel. No more energy will be produced.

Two thousand years seems like a long time for this, but perhaps the reactor is running well below capacity. Maybe it has only been partially exposing one fuel rod and still generating enough heat to keep things going. It's just done this for so long, it ran out of fuel rods.

This would be a rather advanced reactor compared to what we have now. It can apparently run at a much broader range of speeds than they currently do. Perhaps it has been adjusted to be not just baseload power but on demand power. So it can give more or less power as required. Apparently it has only been required to give less for a long time.

It's also possible that the reactor was designed to run for thousands of years. It has many more fuel rods than it needed for normal operation. So as they became spent, it could replace them internally, pulling them out of their lead shield.

Current fuel rods last about six years before needing to be replaced. So we can imagine a reactor with around 333 fuel rods lasting for 2000 years if one is used at a time. Or there could be 2000 rods with six used at a time. You might even be able to use fewer rods than that. Because while spent rods are no longer able to work at full capacity, they still produce both radiation and heat. So a hundred spent rods might be almost as good as one new rod. 300 rods might last two thousand years. Finally there just isn't enough heat produced to keep the turbines running.

If each power plant is running at minimum capacity for the entire time and they all have the same amount of fuel initially, they could all stop working around the same time. It's unlikely to be the same minute, but the same century seems quite feasible.

TL;DR: it takes about six years of use per fuel rod to deplete it, not anything related to the half life.

$\endgroup$
1
$\begingroup$

A time rollover problem. There are a number of them known about, besides the well known Y2K problem. You can read about them here some of them are set to give problems in the time frame you mentioned.

$\endgroup$

protected by James Aug 16 at 20:38

Thank you for your interest in this question. Because it has attracted low-quality or spam answers that had to be removed, posting an answer now requires 10 reputation on this site (the association bonus does not count).

Would you like to answer one of these unanswered questions instead?

Not the answer you're looking for? Browse other questions tagged or ask your own question.