My experience with Vacuum is from taking part on a process for producing fish oil enriched DHA/EPA capsules. We had two vacuum pumps (1 HP each) to empty a 7 cubic meters boiler. We had an ethanol / hydrolized fish oil fraction separated by ion exchange mixture and we recovered the ethanol from it boiling it at 15° C to keep the oil from spoiling. I recall we worked a very low pressures to be able to do that. All our valves were swagelok, they are kind of the industry standard, you close one and it keeps the reactors at the same internal pressure for weeks.
Curbina -- this is medium to high vacuum and thus likely a lot higher vacuum than what you needed for your fish oil capsule manufacturing. The volumes are lower so a smaller pump will not take excessively long to reach the vacuum required. Yes, it will use Swagelok -- that's the standard and they work at these temperatures. The only question is how high of a vacuum is needed, how long to get it there, and how to measure it. I think this intermediate approach with a cheaper turbopump (backed by a rougher pump) is the best way to get to the 10^-4 Torr level which I believe is enough to prove that the rig doesn't leak and to vacuum out the volatile components. Overall, I think the rig will cost around $10K with the majority spent on the turbopump/backer pump. This assumes the experiment can be done reliably without RGA or Mass Spec, which I view as primarily needed for debugging rather than verification. The important measurements are vacuum (i.e. partial pressure) and temperature. I think if the unit is making 2 to 3 kW, we can go with rough natural convection + radiation calorimetry as that is a lot of power, but not so much that we cannot measure it with thermocouples. For those who would criticize Rossi's use of same, I remember than he needed a thermal camera to measure surface temperature. We don't need that. He also had a continuous H2 supply hooked up to his rig. We will have that valved off, and for good measure, we can disconnect the tube going to the tank to eliminate any chance of D2 resupply during the run. I think of the final unit like a vacuum tube that is sealed. If this is indeed what Mizuno is running, and that is what I think Jed had indicated, a functional sealed tube putting out 2 to 3 kW at 0.3 to 0.5 kW input will be all the proof we need. As anyone can then replicate same, it becomes proof positive.