Good writeup that can be useful to many :)

For decades I've built labs primarily for chemical instrumentation, sometimes with electronic capabilities in excess of other chemical labs. This article emphasizes success in building dedicated electronic labs, sometimes having outstanding chemical abilities as an option as well.

There are a number of things in common either way

With the gas lines, don't trust contractors, do as much of the design, build, inspect, and verify your own self. This is usually an embarrassing failure otherwise.

Use Swagelok (top-shelf, having polished threads) brass fittings with straight hardened copper tubing, available in 20 foot lengths, manually cut to size and hand-dressed to outperform the near-perfect output of motorized abrasive cutoff alternatives. Don't make me laugh if you think a plumber's tubing cutter is appropriate, that's a desperate tool for when you can not physically remove particles from an emergency modification of an established system.

https://www.swagelok.com/downloads/webcatalogs/EN/MS-01-107....

Your specialty gases need to have the lengths of tubing manually cleaned with high-purity solvents or steam anyway before installation, and that washes away the particles generated during the cutting process.

Stainless steel tubing & fittings work too and can have an impressive look, but the exact same extreme effort will not result in as leak-free of a system than copper & brass, which makes a big difference with light gases like helium or hydrogen which are escape artists. For utility vacuum (or aggressive gases which might not be best centralized anyway), stainless might be needed if corrosives are being sucked in for centralized trapping, but I never would consider anything less than a dedicated local vacuum pump(s) for high vacuum work on an individual basis.

A vacuum line may benefit from larger diameter tubing, but one of the good ways of doing it for the gases is to use 0.25 inch diameter tubing rated to handle thousands of PSI. Even if you are starting out with a single gas cylinder/regulator in the supply area and delivering the typical 125 psi to the benchtop locations. With a number of further downstream workstations in use simultaneously someday, 0.25 tubing will not be able to deliver enough gas through that small diameter at 125 psi. But if everything is done correctly, conversion can be made later, to run the same gas distribution lines at higher pressures with appropriate new central tank regulators, or even at full tank pressure. Then using an additional local gas regulator at each workstation. Alternatively upsizing the tubing diameter by one notch would then require more wall thickness and then it becomes a real PITA compared to 0.25 inch.

Therefore the completed specialty gas delivery lines should ideally be leak tested over a period of days using full tank pressure of about 2500 psi. And under these conditions you should be able to develop the proper engineering technique for assembling the brass-to-copper so that it almost always shows no detectable leaks every time.

You will also need a flow damper on each main line to limit or shut down in case of major full-pressure uncontrolled escape during an unplanned lab incident. You don't want liters per second of flammable or asphyxiant gases to be able to pour into the lab on top of any other undesirable occurrences. Not even oxygen, which has its own specialized hazards including self-ignition of trace combustibles in the tubing by surprise, like oily residue or even fine metal particles. Oxygen has been known to behave safely for years before an unusual pressure shock occurs and ignites an internal hazard that was lurking overlooked the whole time.

Linux is still useless for most chemical instruments, and Windows has not been suitable for the internet for years, so I need a completely air-gapped LAN of the local lab PCs, each of which can be recovered to its original "immutable" condition from backup, without anything ever having been connected to the web. Mainly for productivity, but safety is also a major consideration for all corporate IT to be barred from labs. They're good at what they do, but their hands are more than full with office machines where it's still a lot uglier that it should be for all the work they are putting in.

It's a lot easier to train natural scientists to do everything with PC's that their particular lab is going to need, than it is to train corporate IT operators to do an outstanding job of anything a unique lab needs.