How Do Molten Salt Reactors Start and Mitigate Thermal Gradients?

[essenmein]

I've been thinking about molten salt reactors a fair bit here and there, and one thing that I keep wondering about is how do they start?

Ie How does one take a system of pipes, pumps heat exhangers etc, filled with a solid salt at room temp, to melt that salt, and then operate?

Or is the thing drained before shut down and to start it a vat heated up and then the reactor refiled?

Either way just the CTE stresses must be crazy.

My google foo has failed me... Any insight into how this is done and how they mitigate for example problems from thermal gradients would be interesting to know about!

 

[Rive]

Maybe you can try to find something for lead- or lead-bismuth coolants instead, since reactor types for those does exists.

 

[DEvens]

You can find the docs for the prototype molten salt reactor that operated at Los Alamos in the 1950s. They're all on line. I downloaded a huge swath of them some time ago, several gigs of it. It is necessary to read a long way through to find the information you are asking about, and it's often scanned docs with no character recognition. So it's not surprising you can't find it.

Basically, the piping would need heating coils. You start it up by using electrical heating to melt all the salt in the pipes, pumps, heat exchangers, and the core. Once it's molten enough to flow you can keep it molten with heat provided by the mechanical action of the pumps.

 

[artis]

@DEvens you said keep it molten with heat provided by mechanical pump action , isn't the salt heated primarily by the very chain reaction itself ? (after it has been heated up to temp to become molten )

 

[DEvens]

Once you start the reactor, yes. But between melting the salt and starting the reactor there will probably be some significant time. You will be wanting to do this long list of tests and measurements and so on.

Maybe I assumed people know as much as I do about commissioning a reactor. Between the first load of fuel into a reactor and achieving 100% normal full power, there are often many weeks of tests. Often you will operate the reactor at the very minimum power, say about 0.01 % normal, and do a long list of tests. Things like checking the control devices, shut off rods, and so on. In a water-cooled reactor there are also a bunch of chemistry tests to check that nothing is leaking, everything is operating as it should, and so on. Presumably there will be equivalent tests in a molten salt reactor.

During those tests, you sometimes will keep the temperature of the coolant where it should be by depending on pump heat. I know most about CANDU. A CANDU has four pumps, each using about 5 MW at full power. This heat winds up going into the coolant, and eventually out the usual heat exchangers and through the boilers.

(Note: I erroneously said 50 MW in an earlier version of this answer. Thank you gmax137.)

So, once you get the system operating at zero reactor power you can turn the heaters off and run on pump heat. That will allow you quite some time to do all your tests, tweak configurations, etc.

 

[artis]

right I forgot those large pumps have MW rated motors and there is plenty of heat when they work

 

[Klystron]

Powered by solar energy this molten salt project should provide methods for initializing and handling similar molten salt technology.

Inactivated by Nevada Energy for economic/political issues, Crescent Dunes plant passed integration systems tests (IST) and operated on the electrical distribution grid.

This lists current solar thermal power generation projects including some that employ molten salt technology.

 

[gmax137]

A CANDU has four pumps, each using about 50 MW at full power.

@DEvens I don't know much about CANDU reactors but that (50 MW) can't be correct. Maybe 5 MW?

 

[DEvens]

@DEvens I don't know much about CANDU reactors but that (50 MW) can't be correct. Maybe 5 MW?

You are correct. Pump heat amounts to approximately 1 percent of full reactor power.

 

[artis]

Well maybe he was thinking electrical MW instead of thermal? because as far as I know the Candu pumps are induction motors driving a centrifugal pump, hard to say their efficiency but not all electric power is lost as heat.

 

[gmax137]

Just as a point of comparison: PWR reactor coolant pumps are 80,000 to 100,000 gpm each, at about 245 ft head. That gives a water horsepower of ~5800 hp (4.3 MW). The motors are 6500 to 7000 hp (say 5 MWe).

During a PWR startup, running three of the four pumps can raise the coolant temperature at 100F/hour. The coolant mass is about 500,000 pounds; that works out to over 14 MW from 3 pumps, or 4 to 5 MW per pump.

 

[essenmein]

You can find the docs for the prototype molten salt reactor that operated at Los Alamos in the 1950s. They're all on line. I downloaded a huge swath of them some time ago, several gigs of it. It is necessary to read a long way through to find the information you are asking about, and it's often scanned docs with no character recognition. So it's not surprising you can't find it.

Basically, the piping would need heating coils. You start it up by using electrical heating to melt all the salt in the pipes, pumps, heat exchangers, and the core. Once it's molten enough to flow you can keep it molten with heat provided by the mechanical action of the pumps.

Thanks, I've read about the ORNL experiments, did not know that the docs were available on line! (this? http://moltensalt.org/)

The 650C operating temperature just blows me away, so many questions. What are the pump impellers made from? How do the bearings work? Seals?! I assume the drive motor can't handle that temp, so somehow the shaft must have a massive thermal gradient somewhere. Insulating shaft coupling? What does a 7000hp insulated coupling look like?!

I've seen what thermal strain will do going from -40C to 125C, going from 25C to 650C with a phase change must be crazy.

 

[essenmein]

Well maybe he was thinking electrical MW instead of thermal? because as far as I know the Candu pumps are induction motors driving a centrifugal pump, hard to say their efficiency but not all electric power is lost as heat.

All the input power from the pumps results in heat in the system, they are circulating a fluid in a closed system and that fluid is not doing any work external to this system, compare for example a hydraulic pump acting on a ram, which is doing work outside the fluid system.

Hence for the coolant circulation, all the mechanical work they are doing is overcoming what is essentially friction in the fluid to keep it circulating, as a consequence all of the power from the pumps ends up as heat in the fluid and/or the fluid container.

 

[Astronuc]

The 650C operating temperature just blows me away, so many questions. What are the pump impellers made from? How do the bearings work? Seals?! I assume the drive motor can't handle that temp, so somehow the shaft must have a massive thermal gradient somewhere. Insulating shaft coupling? What does a 7000hp insulated coupling look like?!

Hastelloy N (initially known as INOR-8, nominal composition 72% Ni - 16% Mo -7% Cr - 5% Fe) was developed at the Oak Ridge National Laboratory (ORNL) in the Aircraft Nuclear Propulsion (ANP) Program. Hastelloy is a Haynes International product. SMC (including INCO) also produces specialty Ni-based alloys. Haynes and INCO developed Ni- and Co-based alloys for the aerospace and nuclear industries. Cannon-Muskegon also developed Ni- and Co-based superalloys for aerospace, but I have not seen them in the nuclear industry. Most superalloy applications in aerospace are in the combustion turbine section, where the temperatures are greatest.

Hastelloy experience is discussed in http://moltensalt.org/references/static/downloads/pdf/ORNL-TM-4189.pdf

The paper mentions 'thermal convection loops', so no pumps necessary. If pumps are employed, I would expect they would be made of Hastelloy N or similar alloy.

 

[essenmein]

That FA report is an interesting read.