minutes of the phone meeting on 22.Sept.2004
In the phone meeting yesterday, 2 variants on Roger's suggestion were
Recall that the premise is to leave 20-30 l of LN2 in the magnet during a beam pulse.
1. Vent the activated LN2 into the TT2A tunnel immediately after a beam pulse, to let the activity die down somewhat before the N2 reaches the outside world. This requires boiling the residual LN2 away at the START OF THE NEXT COOLING CYCLE, before an new, unactivated LN2 is added. Hence, this scenario lengthens the cooling cycle.
2. Start the next cooling cycle immediately, with new LN2. This implies that the activated LN2 will boil off fairly soon and be vented to the outside via our standard ventline. This would release more activated N2 into the environment than variant 1.
However, Peter Titus reminded us that we have been advised by CERN Safety officers NOT to have LN2 exposed to radiation because of a risk of explosion due to liquid oxygen that might be trapped in the vessel.
See "Explosion Risks in Cryogenic Liquids Exposed to Ionising Radiation" http://www.hep.princeton.edu/~mcdonald/mumu/target/CERN/cern_at9506.pdf
After this reminder, we appear to have decided not to pursue any further the
option of leaving some LN2 in the magnet during a beam pulse (without having
clarified whether the release of activated N2 is or is not a safety issue).
We then discussed a variant on the use of the drain:
a. Don't install the "dam" that has been proposed by Peter in a different cooling scenario.
b. Fill the magnet with 200-300 l of LN2 at the start of a cooling cycle. Only about 100 l of this will be boiled away before the magnet reaches 70K. However, we should fill the magnet with more than 100 l in order to flood the upper cooling channels, so that cooling occurs over the entire magnet, and not just the lower portion. Recall that the inner coil contacts no LN2 until the magnet is about 40% full of LN2.
c. In the baseline design of the magnet, simply flooding the magnet with LN2 would lead to the problem of trapped bubbles in the horizontal cooling channels between the magnet coils. Peter notes that if we cut circumferential grooves across these channels (which we must do in the next week or 2, or it's too late), then the N2 bubbles could clear by rising around the grooves.
d. When the magnet is at 70K, stop the inflow of LN2, pressurize the magnet with He gas, open the drain valve, open a bypass line back into the LN2 storage vessel on the surface, and drive the residual LN2 back into the storage vessel.
This is a simpler version of the drain scenario laid out by George Mulholland some 2 years ago, http://www.hep.princeton.edu/~mcdonald/mumu/target/mulholland/E951_Operating_ProceduresB.pdf
IF it is technically viable, and affordable, it might be a good solution. We
encourage the RAL folks to make a technical evaluation of this scenario in the
very near future.
Recall that in previous versions of this scenario, an auxiliary storage tank was included to capture the LN2 that is flushed out of the magnet. Then, one must do something with the liquid that accumulates in this tank during a cooling cycle: either boil it and vent it, or try to reintroduce it back into the magnet. This got so complicated that the scenario priced itself out of consideration.
? Can we really pump the excess LN2 back into the original supply dewar? This will require 4-5 atm of pressure -- which is OK so far as the magnet and transfer lines are concerned. But can a standard LN2 supply dewar be used in this manner? If not, would we need a custom supply dewar? At what cost???
? We would now have both a 0.1 atm and 5 atm pressure in the magnet during parts of the cooling cycle. How quickly could we make the transitions between these pressures? That is, do these transition times lengthen the cooling cycle more than the 10 min or so that would be required to boil off the residual 30 l of LN2 left in the magnet in Peter's scenario that uses a "dam"?