Pump for liquid sodium 1

[moltenmetal]

We have an application for low pressure liquid sodium circulation. Temperature is only about 10 C above melting (ie around 110 C)- this is not a high temperature heat transfer application. Electromagnetic pumps are apparently often used but we have a fairly low flow (a few USGPM) and aren't at high temperature, so an EM pump isn't absolutely necessary.

Our original thought was a sub-ANSI magdrive centrifugal pump of the type we routinely use with hot oil etc. Regrettably none of the vendors we've contacted has experience with pumping sodium to be able to say yea or nay. My concern is with the liquid metal in the magdrive- are there any concerns pumping a highly conductive fluid such as a liquid metal with a magdrive, related to eddy currents or something else we're missing? From a materials, head/flow, pump size/efficiency and maintenance perspective, we're confident that the little magdrive will do the job for a lot less than we can buy an EM pump for. The sodium is continuously strained to remove solids so particulates in the circulation space between the back can and driven magnet and around the bushings etc. shouldn't be an issue.

 

[MikeHalloran]

How do you melt the frozen sodium in the magdrive for a restart?



Mike Halloran
Pembroke Pines, FL, USA

 

[moltenmetal]

We're talking about perhaps a 1x3/4x5 pump here, so picture that for scale.

http://www.magnatexpumps.com/mph.htm

We'd heat trace and insulate the pump including the outer casing around the drive magnet, as well as all the lines. The melting point isn't that high- just less than 100 C- but the material does have a substantial heat of fusion so we'll need some significant wattage available for melting, which can be turned down for heat loss maintenance. The "can" around the driven magnet would be heated by conduction only because obviously it cannot itself be externally heat traced, but the environment around the can would also be hot so it shouldn't take hours to get everything "thawed". We'd also be designing the piping such that the pump and strainer/filter can be drained or blown back to the tank on loss of power etc.

 

[btrueblood]

Molten,

Somewhere in the google universe is an article/report from a NASA contractor that built a canned-motor liquid sodium pump. I could find the abstract/ref. if you need it.

 

[btrueblood]

Wups, one more thing - a mag drive may suffer from loss of torque due to the decrease in magnet strength (coercivity? magneticity? whatever...) as temperature increases, and yes, eddy current losses may push that further. Dunno where you might find data on it. You've considered a conventional pump with a purged seal?

 

[moltenmetal]

We'd like to stay away from a mech sealed pump because of the O-rings, and cost. The pump can handle the magnetic losses due to temperature easily enough, and our flow is low relative to the run-out so we're not challenging the peak torque of the magdrive by any stretch. We've heard that a vendor has a client with a pump like this in sodium service, which suffers from some "issues" related to the durability of the SiC bushings, but that it does work. We're trying to get more data but I'd hoped somebody out there would know this from experience and be willing to share...

Canned motor is something that for some reason we rarely manage to use, again probably due to size.

 

[EdStainless]

I don't see why there would be any issues.
Sure Na is conductive, but not much more than brine solutions that I pump.
I would be more concerned about drag from the high visc and density.

The SiC may be because it is a reaction bonded grade with excess Si.
Ask if they use high density direct sintered material.

= = = = = = = = = = = = = = = = = = = =
Plymouth Tube

 

[moltenmetal]

The vendor doesn't make it clear exactly what silicon carbide they're using, but we suspect it's the Saint Gobain controlled porosity material which S-G touts as being the ideal material for the service. This is not a reaction bonded material AFAIK, as S-G touts their material's advantages against reaction bonded SiC.

http://www.magnatexpumps.com/dry-run-resistant-bearings.html

We don't have direct contact with the user to find out whose pump and what problems they were having- we're trying to find that out through back channels.

 

[moltenmetal]

Back channels have it that the pump is working fine, and the bushing failures they encountered were due to "operator error"...We're going to try it.

Viscosity isn't an issue- the melt is similar density and viscosity to that of water. Unless it isn't melted, or is full of oxides...