Sunday, November 18, 2007

Vacuum Pumping

I'd like to go a little deeper into the subject of vacuum pumping and easily attainable ultimate pressures.

Since this is a paper exercise we can pick any pumps we want. I'm going to look at 3 pump mfgrs. and pick their highest capacity turbo molecular pumps.

Adixen - Mag Lev [pdf]
Pfeiffer - Mag Lev
Varian [pdf]

There are two critical specifications for us. H2 pumping speed in liters/second and H2 compression ratio.

Turbo Molecular Pump Comparison
H2 Speed l/sH2 Compression Ratio
Adixen ATH-2300M12003.0X103
Pfeiffer HiMag 340028504X104
VarianTurbo V 3K-T23001.5X104

It looks like the Pfeiffer HiMag 3400 all the way.

Next lets look at a Roots fore pumps. I really like the Adixen RSV [pdf] series. The link has a great look at pumping capacities vs vacuum pressure and shows the limitations of roots blowers. Let us look at the RSV-1002. The largest standard pump. It has a pumping speed of 800 M3/hr with the model 2100 SD [pdf] roughing pump. That is about 225 l/s at a pressure of 3E-1 mbar (which for out Rough Order of Magnitude (ROM) purposes can be considered equal to 3E-1 torr). At 1E-2 mbar (where you can turn on your turbo pumps) it is about 100 l/sec. At 1E-3 mbar it is about 40 l/s. Ultimate pressure is 2E-4 mbar.

The Roots fore pumps basically stops pumping at 2E-4 torr. Given a turbo pump compression ratio in the neighborhood of 1E4 that gets us down to 2E-8 torr. Not enough. In addition the closer you get to the Roots ultimate pressure the lower your capacity. To get us lower we are going to need another turbo pump in series with the chamber pumps. Tom Ligon in an e-mail suggested using one pump to service all the chamber pumps. Brilliant idea. This should work fine as the volume of gas to be moved will be 1/10,000th of the amount (in liters) of the gas being pumped out of the chamber. Even with 10 or 20 chamber pumps you would still have a lot of excess capacity if you used a turbo pump with 1/100th the capacity of the total of all your chamber pumps. A pump of 600 l/s should be more than adequate. We leave the choice of that pump as an exercise for the reader.

So let us look at the chain of pumps. Say 6 or 10 reactor chamber turbo molecular pumps. A smaller turbo pump drawing from those pumps. A Roots blower next followed by a roughing pump. This all has to be properly sequenced to avoid damage to the pumps and systems. Then you have to sequence the various pressure reading devices depending on pressure. There should be enough work to keep the vacuum guy busy for at least a couple of weeks. Especially if s/he is budget constrained.

Update 21 Nov 007 0642z

I messed up pump volume calculations so let us go over them.

20 turbopumps X 3,000 l/s = 60,000 l/s. 1/100th of that is 600 l/sec.

Corrected in the text.


brent said...

Comments about Pumps:

I looked at the turbo Pump description for the Pfieffer HiMag 3400 vacuum pump. It’s been awhile since I learned about sizing a pump, but I seem to remember that we used to look at pump performance curves consisting of NPSH (Net Positive Suction Head) vs. Flow Rate to get an idea of how a pump behaves. In addition, I believe positive displacement pumps are used for low flow rates as well as for gritty material (e.g. augers), and centrifugal pumps are used for higher flow rates. Sometimes you have to worry about cavitation with centrifugal pumps, which is damage to the pump blades caused by expanding vapor bubbles. Obviously my memory is a little fuzzy at this point.
The Pump World website has something to say about this.

From what I’ve seen about the Pfieffer HiMag 3400 vacuum pump, it seems reasonable. The purpose is to evacuate air I presume. It’s also rated for hydrogen, which is nice.
I’ll need to look into more closely. You might want to check out the link that says Characteristic. It looks like a flow rate vs. pressure graph (I would guess this is the output flow and output pressure).

Another place to look for pumps is the Thomas Standard Products Catalog. I’ve investigated the G12/07-N pump the most closely. Larger pumps are also available here. Further details are in the e-mail I’ve sent you.

I’ve also found some additional links about pumps. It appears that if the pump performance curves are not available you can estimate the performance using the pump affinity laws. Engineering Toolbox provides a good way of estimating this.

I also found an article from Germany that describes complications for Vacuum pumps. Much to my Chagrin, it’s in German. If you want to look at it you might try Babel Fish or Google Language Tools .

A word about hydrogen: it is commonly thought of as a bear in industry. It causes embrittlement in many steels primarily by migrating into imperfections in the material and forming high pressure methane bubbles. Or at least that is my understanding from what I have seen on the Internet thus far. Hydrogen also manages to find any holes; so hermetic sealing is sometimes a good idea. For a manufacturer try going to --industries served--automotive. Also see two additional articles, one from Wikipedia and another from the International Pipeline Conference .

I’ll have to take a look at my old Plant Economics Textbook. Hopefully what I’ve said so far is helpful.

brent said...

Frogive me, the G12/07-N pump produces very small flows. It is therefore not useful for excavating large chambers. I know, it could have sounded like that.

M. Simon said...


I'm familiar with the pump issues you mention including cavitation.

Turbo molecular pumps are different. Once in their operating range the flow (for a particular gas) is constant despite declining pressure.

I'm also familiar with the hydrogen embrittlement problem.

Low pressures cause an outgasing of the hydrogen. So does the cookout prior to operation. It will warrant attention. I do not think it is a serious problem.

BTW the only above atmospheric source of hydrogen is the storage tanks and the initial regulators. The majority of the plumbing and tanks are well below atmospheric.

The only other concern is H2 exhausted from the reactor. It will be less than .1 cc/ second (for the test reactor) and could be safely vented to the lab. An outdoor vent is the wise thing to do though. It will be important to keep the vent pipe short and flooded with an inert gas (N2 or Argon) to avoid explosion problems.

brent said...

It looks like you have a firm understanding of the problems at hand. Much better than I do.
You probably know this as well, but I'll just throw it out:
Pumps in Series vs. Pumps in Parallel on Engineering Toolbox

Thanks for filling me in on what a turbo pump is. I wasn't really sure. For further clarification, are they like special
centrifugal pumps as this

M. Simon said...


Here is a look at Turbo Molecular Pumps.

Also have a look at Vacuum Pumping.

And Turbo Pump Ratings

brent said...

Did you get the Microsoft Word file about Knudsen Diffusion? If so, how would you rate it? Would you say was easy to understand, had unnecessary detail or information, was overly redundant, was too elementary, was unclear….
You can also e-mail me at ZetaAntares at aol dot com.

Douglas Eagleson said...

A vacuum system is critical to correct testing and strong vacuum was designed into the test chamber to allow evacuation in instability times. A poor reaction plasma was to be evacuated into safe configuration.

High volume pumps are required and explosive blast plates would make for the correct test chamber design. A plate blasted to open the chamber evacuation area.

Safe configuation then ensues.

I can only hope the meaning of neutron structure is unstood.