From: Kirk T McDonald [mailto:kirkmcd@Princeton.EDU]
Sent: Tuesday, November 08, 2005 6:30 AM
Subject: Towards an affordable valve box for the MERIT experiment

This note, which some of you may have received already, is for discussion during our VRVS meeting Tue 8 Nov.

Background documents include:

From Friedrich Haug:

From RAL:

The preliminary advice from potential European vendors that the proposed Version A cryogenic valve box for the MERIT experiment might cost ~ 150k Sterling is 4-5 times what we expect -- and can afford!

To improve my perspective on cost issues for the valve box, I visited Tam Tu of Cryofab,

located about 1 hour from Princeton. Cryofab is the US low-cost vendor of portable LHe and LN2 dewar and related small-scale cryogenic systems. See, for example,


1. We need to specify the LOWEST working pressure possible, if we are to keep costs down.

21 bar is much too high a spec!

The pressure spec is related to item 2:

2. We need to specify a valve/pipe size <=> specify a maximum flow rate.

I propose that we specify 50 liters/min for the line from the magnet to the valve box, and 10 liters/min for the other 4 lines.

I think that this implies that 2 of the valves need to be rated for 50 l/min, while the other 3 can be only 10 l/min.

Then, the length of the lines between the valve box and elsewhere can be used to determine the maximum operating pressure.

As I understand it, the lines between the magnet and the valve box could be as long as 100' (30 m).

Once we have the pipe size spec, we can provide a spec for the control valves and the bayonet connections for the transfer lines.

The spec should permit the vendor to propose relevant valves. To maximize cost savings, we should not insist on use of a particular vendor of valves.

The bayonet connections may, however, have to be sole source -- as per CERN's recommendation.

[USA facts: 3 standard pipe sizes are
1/2" => ID actually 0.71"
3/4" => ID actually 0.92"
1" => ID actually 1.18" ]

3. The valve box should consist of a single outer wall of stainless steel, not a double wall (with vacuum between) as was perhaps suggested thus far.

There will be vacuum between the outer vessel and the 200-l inner vessel, so we don't need an additional layer of vacuum.

4. The valve box should NOT be built with a large flange pair, which would permit later disassembly. The upper dome of the outer vessel should be welded to its vertical cylindrical wall.

A large flange pair will add $5-$10k to system cost. It is needed only if we don't know what want!

A possible compromise would be to have the vendor build the system for welding, but not actually perform the final weld between the outer cylinder and outer lid. Then, the functionality of the delivered device could be checked out at CERN, modifications made, and then the final weld performed at CERN.

Of course, this means no (easy) vacuum test by the vendor before delivery.

5. A lot of the cost of the stainless vessel is in the end caps (with penetrations). So, it's better to make the vessel tall and thin, not short and wide.

In particular, the 200-l tank should be entirely below the array of valves and feed lines.

(This is the way the typical portable LHe and LN2 storage vessel is constructed.)

=> Can use a tank about 20" (0.5m) in diameter.
=> Outer vessel only 22-24" in diameter.
=> Total height might be ~ 8' -- but so what?

The 200-l tank can be entirely supported off a central tube that serves as the vent to the pressure relief valve at the center of the top of the outer vessel.

However, we can obtain additional support for the 200-l tank by welding it to 2 of the control valves..... (i.e., don't need to use coiled pipe between the valve and the inner tank.....Details perhaps best worked out by the vendor).

=> We don't need additional struts to support the inner vessel, as hinted at in the present sketches.

6. The number of instrumentation penetrations can be reduced by using a tee structure on a single penetration for this purpose. (This cost saving may be modest.....)

7. The valve box should be built to code standards (ASME if built in the USA), but should not be "code stamped".

This will save $5-10k -- and is the path we have already followed on the pressure vessel for the 15-T magnet.


The spec for the valve box should be for the general concept of Version A, but with revisions to permit cost savings of the above sort.

Then, the spec should be sent to both European and US vendors.....

The first step towards this is to get the valve/pipe size specs, the bayonet type specs, and the max pressure spec. CERN should provide this, in consultation with RAL (particularly on the pipe size spec which requires a flow vs. pressure calculation).

Then, a bid package (for only one option) should be written up and sent out vendors in both Europe and the USA.

Some subtleties as to financing remain to be discussed.