Triangular o-ring grooves

[dgallup]

We have a part that uses a triangular o-ring groove similar to what is shown in Parker O-ring Handbook design chart 4-6 on page 4-21. We have been using this particular design for over 5 years without problems. There seems to be some change in manufacturing process as this seal has worked quite well but we have not found anything out of spec so far. The o-ring is a 70 durometer fluorosilicone material in an oddball size. I've compared the groove fill to the Parker dimensions and we are quite close. The Parker sizes above and below ours are nominally 90.2% and 90.9% groove fill and ours is 89.5% so just a little lower.

This seal gets leak checked a number of times, once in an outlet pressure rise automated test in production of short duration, ~ 6 seconds. Second in a visual submerged bubble leak test also of ~6 seconds. Both of these apply 9 bar differential pressure to the seal. If the parts come to engineering we do a much longer leak test with a flow meter at 14 bar differential, about 60 seconds. Our customer assembles 4 to 10 of our product into a larger assembly and does a pressure decay test at 9 bar differential for about 5 seconds. Our customer's customer does a finished system test that is much longer.

Recently, occasionally (maybe 0.1% of the time) one of these seals that has passed multiple short leak tests will start to leak at the end customer after a couple of minutes. We use the same o-ring material in 4 radial seals in this product and I have never seen one of them display this time dependent failure mode. I'm trying to understand what is going on internally in the triangular o-ring groove that allows this to happen.

My thought is that the pressurized o-ring initially takes a shape inside the triangular groove with approximately equal radii in each of the three corners but as pressure is applied it gradually creeps down so the low pressure corners have very small radii and the high pressure corner has a much bigger radii. Eventually the contact with one of sides with the pressure differential is lost.

It's hard to understand as when we certified this product there were 3 minute leak observations at -40C, room temp and +120C and we never saw this behavior and we do periodic confirmation of production tests as well.


----------------------------------------

The Help for this program was created in Windows Help format, which depends on a feature that isn't included in this version of Windows.

 

[Compositepro]

If the problem occurs 0.1% of the time there is not a systemic cause due to the design. There is some random improbable cause like dirt or a defective o-ring.

 

[dgallup]

We have ruled out dirt and defective parts. The assembly is done in a clean room and failed parts have been cut open, microscopically examined and measured.

My question really is why does it take a relatively long time for these leaks to manifest? What is changing during the leak test?

----------------------------------------

The Help for this program was created in Windows Help format, which depends on a feature that isn't included in this version of Windows.

 

[Compositepro]

That is a more interesting question. It seems to me that in your seal design the o-ring is acting more as a gasket than an o-ring. For sealing to occur the contact pressure between rubber and the sealing surface must be greater than the pressure that is being sealed. In a gasket the contact pressure is determined by the clamping force and the area of the gasket. In contrast, an o-ring is energized by the pressure to press harder against a sealing surface. In your design the o-ring gets highly deformed creating a large contact area with the sealing surface. The contact pressure is determined mainly by the hardness of o-ring. But rubber will creep with time, and if the contact pressure drops below the gas pressure, you will have a leak. In your design it looks like fluid pressure will not work very effectively to actually energize the o-ring to seal tighter as pressure increases.

Are you lubricating the o-ring with a compatible lubricant?

 

[dgallup]

I agree that this is not a self energizing design. Once a leak starts, it will continue leaking at lower pressure. We use this to pinpoint the exact leak location as once the seal "blows out" we drop the pressure enough that there is just a tiny stream of bubbles.

Lubrication is an interesting topic. Manufacturing stopped lubricating this o-ring right when the "big spike" (0.1%) failure rate started. However, the entire device is lubricated internally after assembly. We do see engine oil on the o-ring after disassembly. The o-ring lubricant was a very light oil, the assembly lube is 15W40 engine oil.

This is what Parker has to say about triangular grooves and pressure:

This type of crush seal is used where cost and ease of machining are important.
The O-Ring is confined in a triangular recess made by machining a
45 degree angle on the male cover. The OD of the ring should be about the
same as the recess diameter. The same sealing principle applies to crush
type seals used in recesses of straight thread tubing bosses. Pressures
are limited only by clearances and the strength of the mating parts. The
O-Rings are permanently deformed.

I must say, I have never actually seen the permanent deformation mentioned even after long thermal cycles but we do use a compound that has very little compression set.

----------------------------------------

The Help for this program was created in Windows Help format, which depends on a feature that isn't included in this version of Windows.