I am working with a piece of equipment that has a vacuum sealed chamber. The outer wall (ring) of the chamber is shown in the image below. This ring is bolted to a bottom plate and a top plate is raised up and down using pneumatics. When the chamber is sealed the differential pressure between the inside and outside is nominally maintained at 10 psid. The system also has a laminar flow meter and once vacuum is pulled and the chamber is for the most part isolated, we can read about 1.5cc/min flow (leakage) in the chamber. Maximum flow readings for all other system components other than the chamber seal is below 0.0.03 cc/min, so I am convinced the issue is with the sealing of the chamber.

In the image can be seen a gland for an o-ring. This is an existing design, not of our own, so only minimal alterations to the design may be allowed. We will be looking at the flatness and finish on the top plate, but there has always been some variability on leakage from cycle to cycle.

So, this is my question... What is the likelyhood that adding a second o-ring in series with the first, with a minimal gap between them, may reduce the total leakage rate of the chamber?

enter image description here


Simply adding another o-ring in series will not give much improvement in the leak rate. This is because the leak rate of the inner o-ring will prevent the internal space (between the two o-rings) from being brought to any reasonable vacuum level so the pressure differential will remain the same on the inner o-ring.

However, if you can use a separate pump to pump down the cavity between the two o-rings, then you will greatly reduce the effective leak rate into the main chamber. The separate pump must be able to hold a good vacuum in the space between the o-rings for this to work.

Another, perhaps quicker, solution might be to use vacuum grease to create a better seal with the single o-ring. To quote from a Kurt J. Lesker technical note

The debate about greasing o-rings is ongoing. Typically, the same vacuum level can be achieved with or without grease. However, when not using grease, the o-ring must be dust- and fiber-free, and the metal surfaces touching the o-ring must have at least a 32 rms finish. Grease permits sealing in less pristine conditions.

  • $\begingroup$ Thank you very much for your comment. I should have noted above that Molykote 111 vacuum grease is being used. Changing of o-rings and regreasing has no improvement on leakage. It was my hope that if the gap between the two o-rings were kept as narrow as possible, that the volume of the space between o-rings would be very small (nearly zero with steel on steel contact), so that even if the inner leaked some or began to leak more, that the outer would provide a reasonable backup. $\endgroup$ – ashur668 Jun 21 '16 at 16:42
  • $\begingroup$ @ashur668 A leak rate of 1.5 cc/min is an incredibly large leak. What is the material and what is the finish of the surfaces in the o-ring groove and on the cover? $\endgroup$ – Chris Mueller Jun 21 '16 at 18:11
  • $\begingroup$ Also, grease reduces O-ring wear from rubbing on the metal parts due to pressure cycling. The tendency is to use too much grease on O-rings, but the O-ring should barely feel greasy and should not have any excess on it. $\endgroup$ – DLS3141 Jun 22 '16 at 16:49
  • $\begingroup$ Both answers provided were essentially the same, so I will flag Chris's answer simply for being first. $\endgroup$ – ashur668 Jul 14 '16 at 19:06

I'm assuming that the piece with the groove and the mating pieces are metal or a dense molded plastic. If this was a 3d printed piece, I'd be worried about porosity of the parts too.

A second O-ring won't do much for you honestly. If your leak rate is that high, I'd question the design of the O-ring groove and O-ring selection. Does the O-ring sit proud of the groove when installed? If so, by how much? How much stretch is in the O-ring when installed? What's the ratio of the groove cross section to the nominal cross section of the O-ring when installed?

A common mistake is to design such a groove based on the nominal cross section of the O-ring and not accounting for the fact that this shrinks as the O-ring stretches.

The mating surfaces must be relatively flat, in order to control the gap. if there's an excessive gap and the O-ring material is too soft, the pressure can push the O-ring material into the gap and cause leaks. The best resource I've found for designing this type of static seal interface is the Parker O-Ring Handbook

I'd also like to ask; Why use an O-ring? Why not a flat gasket? Or even a liquid gasket replacement like Loctite 510 or one of the hundreds of other sealants?

  • $\begingroup$ Many excellent questions. $\endgroup$ – ashur668 Jun 22 '16 at 19:13
  • $\begingroup$ Many excellent questions: The piece shown and the plate that goes on top are both stainless steel. These parts were designed by another company, but the o-ring groove cross-section is dovetailed. An o-ring is used because the chamber is opened/closed about once per minute. The o-ring does sit proud by about 0.015". $\endgroup$ – ashur668 Jun 22 '16 at 19:24
  • $\begingroup$ The o-ring is liberally coated with grease before installing into the groove to fill in the corners of the dovetailed groove, but then all excess grease is removed from the top surface where the o-ring is exposed and top plate meets the ring. Both surfaces are then wiped every few cycles with lint free wipes. I don't know that the company designed it will share their CAD files to calculate cross-section. $\endgroup$ – ashur668 Jun 22 '16 at 19:25
  • $\begingroup$ The serial idea came from a white paper I found by a company called Seal Dynamics Inc.. I think this may now be Heico. "O-rings may be used in series in vacuum applications, preferably with a separate vacuum between them". However, I have found no other support for this and concensus is leaning toward this not being affective. $\endgroup$ – ashur668 Jun 22 '16 at 19:27
  • $\begingroup$ It will be a couple more weeks before I will have flatness/finish data on the existing plates. Visual inspection found nothing and 4 of 4 plates in one set produce the higher leakage, while plates from other sets produce leakage rates more typical to expectation, which has generally been < 0.4 cc/min total for the entire system. $\endgroup$ – ashur668 Jun 22 '16 at 19:31

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.