In article ,
Joe Strout wrote:
- How can I estimate the leak rate through the seal?

Depends greatly on the technical details. The fast answer is that it can
be made arbitrarily small if you're willing to work hard and accept
compromises. For example, you can use a double seal, with vacuum pumps
running continuously to scavenge air that leaks past the first seal before
it gets past the second.

- Can I characterize the problem only by the pressure difference between
the inside and outside? In other words, is a seal that holds 1 ATM
against a vacuum the same as one that holds 2 ATM against 1 ATM?

The pressure difference is all that the seal sees... except that vacuum
can have harmful effects on some seal materials, which is a different
issue but possibly a significant one.

- What type of seal would you expect this to be? Radial shaft?
Labyrinth? Rotating face? Something else? (I only vaguely understand
these different types of seals, so even small insights will be
appreciated.)

You wouldn't use a labyrinth seal for this, at least not as the primary
seal -- labyrinth seals are what you use when rubbing contact is
unacceptable (e.g., because of high rotation speeds) and you're willing
to tolerate a certain amount of leakage.

My thought would be to make the tube from (say) the stationary side a few
meters bigger than the tube from the rotating side, and put bearings and
seals between them. Why? So you can maintain the bearings and seals!
The large difference in diameter is so people can walk (well, sort of,
since spin "gravity" will be minimal at such a short radius) and work
between the tubes.

In particular, it's not that hard to have multiple bearings set up so that
you can replace parts of any of them without stopping rotation. The
bearings are rollers against the inner tube, supported inside the outer
tube by struts, so that you can just step between the struts to get
through the first bearing ring and work on the next one. To replace a
roller, just jack it down -- the struts telescope a little bit to permit
this -- and take it out and swap in a new one.

Seals obviously need a continuous wall supporting them, but it's still
possible to work on them if you're clever.

At the outer end, a pair of labyrinth seals; because they're non-contact
seals, they doesn't wear out or need maintenance. Normally, they have no
air pressure across them, because the sealing is being done by the inner
seals -- a pair of shaft seals, each supported against the inner tube by a
wall running up from the outer tube. The walls have doors, but they have
airtight seals and are normally shut. There are vacuum pumps scavenging
leakage air from the volume between the inner seals.

When the leakage rate gets too high, you pressurize the space between the
inner seals and the one between inner and outer seals, open the doors, and
go in and fix the inner seals. The outer seals carry the pressure
meanwhile, with the vacuum pumps scavenging from the space between them.
There'll still be some loss, but it won't be much. When you're done,
close the doors, and have the pumps clear out first the space between
inner and outer seals, and then the space between the inner seals. You're
back in business.

This sort of arrangement is how you need to think for long-lived hardware
that's going to be *permanently* in service -- it has to be maintainable.
With proper design (including provisions for replacing all parts subject
to wear) and regular maintenance, this sort of heavy equipment can have an
essentially infinite service life.
--
spsystems.net is temporarily off the air; | Henry Spencer
mail to henry at zoo.utoronto.ca instead. |