Remove ads

How Does an RBMK Reactor Explode

Dartz said:
Make it safe, make it reliable, and stick it in a box it can't get out of incase you're wrong about the first two.
Click to expand...
Click to shrink...
I think this merits some expansion.

Fundamentally, water is just not a friendly material to put into the core of a large nuclear reactor in any capacity. It's always trying to boil, and you're always trying to fight it. If you get too much heat, the water can boil or react with the metal structural materials at the same time as those materials have been weakened by heat. This leads to cascading failures in an accident scenario, and provides a mechanism for dispersing radioactive material.

Thus, for providing safe nuclear power cheaply, even if another coolant is more of a pain under normal conditions, it's a big improvement if the accident scenario isn't as bad. Ideally you also use more passive mechanisms to avoid accident scenarios. Modern reactor designs do both.

For example, in a salt-fueled molten salt reactor, the fission products are chemically bound to the salt, so that even if the reactor literally melts into a puddle they mostly don't manage to go anywhere. Consequently, in the worst beyond all physics reactor accident, the minimum safe distance (by which I mean less than a CT scan) is typically less than 1 km. Unless the reactor vessel is built below grade (as is usually planned), in which case the minimum safe distance is... anywhere outside containment, because there's no mechanical pressure anywhere to actually cause the stuff to escape.

Terrapower's Natrium sodium-cooled design (for which they recently got US DOE funding for a prototype) is similar, in that they're planning to use metal fuel in a sodium pool. Everything around the reactor is metal, so there's no real dispersal mechanism. Not quite to the same degree as a molten salt, because it's still possible to have a sodium fire if everything goes wrong, but that's not comparable to a pressure vessel rupture. On the other hand, this design is based on the EBR II reactor, in which they just... stopped cooling it, as a demonstration and it politely declined to melt down.

Nupower's small modular reactors are still water-cooled, but small pressure vessels are tougher than large pressure vessels, and small reactors are easier to cool. AFAIK Nupower's design also achieves its safety primarily through passive mechanisms.

Actually, the bigger issue with current nuclear reactors is that they're (mostly) old designs (because only a few countries are building new ones in quantity). There are so many ways to improve on them, particularly in the areas of making them cheaper, and building the safety into passive, robust features instead of large pressure vessels and an unbroken string of careful operators.

To zeroth order, the cost of a nuclear reactor is all capital cost. In the USA, the 60s-70s wave of pressurized/boiling water reactors have safety-critical features of a giant steel pressure vessel and an even bigger reinforced concrete pressure vessel, constructed by local workers who haven't built reinforced concrete pressure vessels to safety-critical nuclear standards before (and so they have to fix it until they get it right, which is a slow process). Also, since construction takes a while, and NIMBYs are a thing for literally any construction project of sufficient scope, it has a chance to be delayed. This is all fantastically expensive.

If you can do your safety-critical stuff in a factory, the budget for unobtainium can be very high and still much cheaper. That's why and how various players around the world are trying for new nuclear power. The question is always whether they can develop something commercially viable before they run out of money. I hope so.
 
Factory construction was the original end-goal of the RBMK project. The hardest part of building the reactors were the welds at the interface between the zirconium alloy channels and steel header pipework connecting to the steam seperators and discharge headers. In order to fit, the pipes are actually tied into the side of each other - it's not a 90-degree elbow over at the top. The welding on the nuclear circuits was subject to stringent quality control - even though this was much harder in the field compared to in a controlled factory environment.

The RBMKp 2400 and 4800 project was to build a modular reactor which could be factory assembled in completed blocks, and then assemble them like lego onsite. In which case, the only limit to reactor size is your ability to keep it under control, either by having multiple operators or more advanced control computer. This was possibly the big hidden strength of the RBMK design - it was no longer limited by the size of the pressure vessel you could build to hold it - you could just extend it arbitrality in either direction.

These things would've been massive. 2.5GW electric for the smaller model, and nearly 5GW for the larger - and would've featured nuclear steam superheating to improve efficiency.

A chernobyl-like accident on one of these would have been frightening. There is the feasible situation where half the reactor has left the building, while any undamaged remnants of the core - now open to atmosphere - would be in such a condition that they could undergo uncontrolled fission and spew fresh hot fission products over a wide area.

This was, for a long time, the real potential nightmare of Chernobyl. Nobody knew what was still inside the reactor vessel, or whether any combination of it could inadvertently reach a renewed criticality through rainwater seepage or material changes
 
Dartz said:
The RBMKp 2400 and 4800 project was to build a modular reactor which could be factory assembled in completed blocks, and then assemble them like lego onsite. In which case, the only limit to reactor size is your ability to keep it under control, either by having multiple operators or more advanced control computer. This was possibly the big hidden strength of the RBMK design - it was no longer limited by the size of the pressure vessel you could build to hold it - you could just extend it arbitrality in either direction.
Click to expand...
Click to shrink...
That is a nifty feature I didn't know about. The plans for extending it, I mean. Doesn't sound like a great idea, for the reasons you cite.

However, US practice is to put a pressure vessel around a pressure vessel, so a hypothetical American RBMK wouldn't have had that advantage; the containment building wouldn't have been small enough to build in a factory (probably, although maybe it would be possible to do something interesting with precast concrete if you're just trying to beat down the civil engineering costs?). I assume France, Korea, and Japan have similar practices although I'm frankly not as well-read as I'd like on the history of nuclear power.

I would say that the more important strength of the RBMK design was burning natural uranium or similarly cheap material in a power reactor is pretty cool in the first place. That's why so many of the new reactor designs are fast reactors--if you have unobtainium that can stand up to the fast neutrons in the first place (signs point to yes), that's a way to do things like burning Spent Nuclear Fuel, or your enriched uranium is only for startup, etc.

There are three big components of the capital cost of a nuclear power plant, which makes up most of the cost in general. Nuclear steam (or other working fluid) supply, turbomachinery, civil engineering. Modern nuclear power plants get killed with expensive civil engineering and turbomachinery. The steam supply isn't so bad. They also have fuel costs, which I find completely disgusting due to the intrinsic energy density.

The civil engineering is expensive because large pressure vessels are assembled on site. These can be reduced by making everything smaller and minimizing what you have to do on site. Martingale's Thorcon design, for example, plans to build the whole damn power plant in a shipyard, float it out to the site and ballast it down with concrete. Presumably part of their eventual business model is that if the customer plays games with the schedule they can just send the power plant somewhere else that will actually pay for it, but that's speculation on my part.

The other big theme of popular reactors is that they work at higher temperatures and reduce fuel cost. Most steam supplies (eg coal fired, solar thermal) achieve higher temperature steam. The turbines are not "off the shelf" but they are readily available. If you're generating 300 C steam, and you have fuel costs that mean you need to squeeze every joule of work out of it, Those are specialized and much larger turbines. They are consequently much more expensive.

The fuel costs come from making a complicated assembly of fuel that's also a primary containment barrier. The complication partially comes from using oxide fuel. Uranium oxide is a ceramic. It's thermal conductivity is poor assuring heat transfer to the cladding is tricky, and requires high precision manufacture. In contrast, metal fuel forms (such as for sodium-cooled fast reactors) are castable, and tolerate much higher burnup (much longer life in the core before refueling). Molten salt fuel "fabrication" is a chemical treatment, with no dimensional stability required. TRISO fuel, although expensive for other reasons, is at least not nearly so demanding of precise fabrication, and it's a much better containment barrier.

Anyway, so if you have a small, higher-temperature (>500 C) reactor that uses cheaper fuel that doesn't need to be replaced very often, you may have a winner that can compete with solar and natural gas on Levelized Cost of Energy. See for example Terrapower's Natrium and Molten Chloride Fast Reactor designs, Elysium's Molten Chloride Fast Reactor design, Martingale's Thorcon and Terrestrial Energy's Integral Molten Salt Reactor.

I think Terrapower, Martingale, Nuscale and Terrestrial Energy are the most actionable US reactor builders, but there's a DOE program to fund technology risk reduction activities for several designs at ($30M/yr? I forget) and two prototype reactors at $80M/yr. Terrapower's Natrium design and some TRISO-fueled gas-cooled reactor whose name I can't remember got prototype reactor funding. The risk reduction awards could/should bring some other contenders to higher readiness. I'm not completely clear what Martingale's current activities are, but they have an agreement with Indonesia to build a Thorcon power plant by 2026.
 
Dartz said:
Factory construction was the original end-goal of the RBMK project. The hardest part of building the reactors were the welds at the interface between the zirconium alloy channels and steel header pipework connecting to the steam seperators and discharge headers. In order to fit, the pipes are actually tied into the side of each other - it's not a 90-degree elbow over at the top. The welding on the nuclear circuits was subject to stringent quality control - even though this was much harder in the field compared to in a controlled factory environment.

The RBMKp 2400 and 4800 project was to build a modular reactor which could be factory assembled in completed blocks, and then assemble them like lego onsite. In which case, the only limit to reactor size is your ability to keep it under control, either by having multiple operators or more advanced control computer. This was possibly the big hidden strength of the RBMK design - it was no longer limited by the size of the pressure vessel you could build to hold it - you could just extend it arbitrality in either direction.
Click to expand...
Click to shrink...

Were they ever successful? How embarrassingly cheap would they be compared to normal nuclear power or coal?
 
Cloak&Dagger said:
That wasn't the question, I meant the past plan described in the quotes, but since you brought it up, which reactors made today are cheaper per kw/h?
Click to expand...
Click to shrink...
South Korea has (had?) a good track record of building PWRs on time and on budget. ~$2000/kWe most recently IIRC. One of their secret forbidden techniques is that not only to they finalize the design before they start construction, but also they build them the same way every time! If you look at the list of operational reactors in SK, almost all of them are OPR-1000 or its successor APR-1400. Also they only have a handful of sites and most of them have 4 or more GW-class reactors.

Here in the USA, conventional practice is to start with a design that's supposed to cost ~$4-5000/KWe. Not a complete detailed design, mind because you're also going to pick a brand new site out in the middle of nowhere. So the general design may or may not be the same, but the detailed design is site specific. Then, while your engineers make reams of paper specific to the general construction and plumbing of putting a reactor here instead of there, you round up the local construction firms. They haven't built a nuclear reactor before, but under the watchful eye of regulators they give it their best shot using designs received just in time from the engineering team.

They make mistakes of course, but they're going to fix them until they get it right, and the result will be nuclear grade. In the context of something like a containment building, this means things like assembling a choked morass of rebar and dabbing a little concrete on it, which is a tricky product to produce even for the most experienced concrete layers but never mind that. And the assembly of the rebar is done by hand, no more productively than it was done 60 years ago. Did I mention the American construction worker is paid a lot more than the Korean?

Well, there are also mistakes in the design. It's understandable, because it's in many ways a new design, just for this site. But it's a nuclear power plant and they have to get it right, so they make changes until it's right. This may cause other rework of course, and the rework may cause careful schedules to fall apart so that just-in-time contractors instead sit on their thumbs. Hopefully the local government doesn't change its mind about whether they want a power plant there at all during this drawn-out process; after all, they never had a nuclear power plant on that site before, and they'll get on fine if it never gets built.

The result of this process is typically that a 5 year $4500/KWe build turns into a 10 year $10,000/KWe build. Which nobody actually finds cost effective. The US civilian nuclear industry will change or it will die.

So the cost of the power plant actually has almost nothing to do with the cost of the reactor. And the power plants that can be made cheaper today are the ones built according to a vaguely rational process and designed for manufacturability. In particular:
  1. Don't touch the US construction industry with a 10 foot pole. Build literally every possible component of the power plant in a workshop and ship it to the field to be put together. If the parts can go together like legos, do that. Modularity, modularity, modularity.
  2. If you could build the whole power plant in a sane working environment and ship the whole power plant, that's even better. This is the thought process behind floating nuclear power plants, ship-deployed nuclear power plants etc. While the US shipbuilding industry is smaller and less productive than South Korea's shipbuilding industry, it is much healthier and able to deliver things on time and on budget than the US construction industry.
  3. On land, save on shielding and put the reactor below grade. It's tempting to think that a dumb lump of concrete for shielding isn't such a difficult thing, but that's not always the world we live in.
  4. If you need a containment building, don't use such dense rebar. Higher strength steel, more overall thickness, anything more manufacturable.
  5. Take advantage of the various features of advanced reactors to reduce the construction work you need to do. Depends on the details of the reactor. One typical advantage is being able to use more normal steam turbines instead of boutique giant turbines for extracting work from 300 C steam.
  6. Make the detailed design site-agnostic, so if you're so lucky as to build more than one reactor, the next can be much the same as the first (and consequently cheaper).
Alternatively, buy a power plant from South Korea, if that's diplomatically possible.
 
Cloak&Dagger said:
Were they ever successful? How embarrassingly cheap would they be compared to normal nuclear power or coal?
Click to expand...
Click to shrink...

Modular construction got as far as an early 1986 propeganda video promising a future of unlimited power by extending the reactors. Technically, the project would've probably worked - the Soviets had built RBMK-style reactors with nuclear steam superheating using supercritical steam before and it worked fine. But a lot of the technical debt built up over the previous generations of RBMK would also be coming due.

There were upgraded RBMK proposals that included a proper containment, and other modifications to the cooling circuit - but construction was stopped at the 90th percentile of completion.

Generally, nuclear power is expensive because of each and every failsafe that is being added. What if an earthquak happens, after a coronal mass ejection, while an airliner crashes into the containment and a crazy-person is fucking with the controls. The consequences of an accident are so nightmareish - almost beyond reason - that the entire kitchen sink is thrown at preventing them.

Which, of course, introduces further complexity and may actually lead to shit like 'normal' accidents.

It's possible to overcomplicate things to to.
 
Dartz said:
Modular construction got as far as an early 1986 propeganda video promising a future of unlimited power by extending the reactors. Technically, the project would've probably worked - the Soviets had built RBMK-style reactors with nuclear steam superheating using supercritical steam before and it worked fine. But a lot of the technical debt built up over the previous generations of RBMK would also be coming due.

There were upgraded RBMK proposals that included a proper containment, and other modifications to the cooling circuit - but construction was stopped at the 90th percentile of completion.

Generally, nuclear power is expensive because of each and every failsafe that is being added. What if an earthquak happens, after a coronal mass ejection, while an airliner crashes into the containment and a crazy-person is fucking with the controls. The consequences of an accident are so nightmareish - almost beyond reason - that the entire kitchen sink is thrown at preventing them.

Which, of course, introduces further complexity and may actually lead to shit like 'normal' accidents.

It's possible to overcomplicate things to to.
Click to expand...
Click to shrink...

So there was nothing in the way except the usual red tape deluge?
 
Dartz said:
Which, of course, introduces further complexity and may actually lead to shit like 'normal' accidents.

It's possible to overcomplicate things to to.
Click to expand...
Click to shrink...
This is definitely a problem in construction; I wouldn't know about operation. The containment building rebar thing is a good example. Normally an engineer starts to look askance at you for proposing a reinforced concrete structure with more than 6% rebar by volume. It makes it progressively harder to ensure the concrete actually penetrates everywhere it should. Reactor containment buildings have sometimes been designed way past even that limit, out of a tendency to overdesign things for worst-case scenarios. And to be fair, a PWR has a great big pressure vessel full of stuff that's supposed to be contained.

If everything goes as planned, such a containment building is merely expensive. In a US PWR power plant construction project, going over budget by a factor of 2 is the default outcome lately, and difficulties in constructing such an object are one of the factors.

Proposed remedies for future containment buildings include using much stronger rebar but less of it in order to achieve the same design strength without so much congestion, or using prefab steel sandwich panels to both relieve congestion and greatly reduce the assembly labor. It's not just the containment buildings, of course. More far-reaching remedies propose giving up on building a nuclear power plant in the field--instead you build it in a shipyard and transport it by sea. I think that one has a lot of promise, because US construction and civil engineering capabilities are in bad shape, and counting on them to improve will leave you waiting a while.
 
You must login or register to reply here.
Remove ads
Back
Top