O-Ring Face Seal - Diameter to Use to Calculate "Piston Area"

10%? Is there so little safety margin that 10% is that important? Who designed it?

Fair question. This is a very old design that does have a safety factor. The issue is whether the safety factor is just above 1.5:1 or just below1.5:1 - based on calculations.

Ok, thanks.

Sorry, I have nothing useful to offer here regarding the details; not my field - anything I add would just be speculation.

Don't know where you're from, but I suspect you have some kind of local standard that can be applied. With the unusual arrangement (only a counterbore in one flange), I've got a nasty feeling^[1] that unless it can be found described in exact detail in a spec, any Safety consultant will cover their butt by going for the worst case. It may be down to you to dream up justification for using your options 1 or 2.

[1] By analogy with odd set-ups I've come across.

It was after hours when I replied about the 10% difference in piston area meaning the difference between a set of Grade 2 bolts have either just below or just above a 1:5:1 SF. I failed to clarify that the grade of bolts could be changed to Grade 5 - but the resulting safety factor is 1.94:1 & our client would prefer an upgrade that provides a full 2:1 SF. Grade 8 bolts are too brittle in my opinion for this application, and having paper calculations on file to justify a full 2:1 SF with grade 5 bolts is the ultimate "goal".

Something doesn't register here. Pressure vessels are designed to a 4:1 safety factor, and I have seen piping safety factors around 5:1 to 8:1. Thus 2:1 surprises me, as does the mention of Grade 2 bolts (unless they are generously sized).

Uncompressed O-ring:

Compressed O-ring in position:

Using nominal PD sounds more realistic. Notice the nominal PD gets reduced on compression. This means, PD is slightly more than 'Compressed PD'. So using nominal PD as such, inbuilt some extra factor of safety.

You have to assume the o-ring is squeezed into the "white space" in your diagram.

Because; when the bolts stretch or flange buckles the o-ring moves to occupy the gap by adopting the vertical version of your horizontal 'squish'.

Or your calculation should assume "PD" at the least is the 'piston diameter'.

However - if you can fit a (metal) ring to reduce the o-ring diameter, closer to pipe ID, then you reduce the 'piston diameter'.

But frankly Fs 2 ain't smart and if you are at 1.5 ± the section difference effect - go buy a paddle.

Why are grade 8 "too brittle" is the Cryo or something?

This picture is theoretical. The pressure radially pushes the o-ring to the wall therefore it actually becomes a square section form and even penetrates to the radial gap between the top and bottom flange. To my opinion the outside diameter should be taken just in case that the o-ring has some imperfections or even the gas permiates through the o-ring and may cause the gas to reach the outside diameter or the o-ring resulting in catastrophic burst. The safety factor of 1.5 or 2 is for military/aerospace use. The 4 safety factor is for commercial and laboratory use, see MIL-STD-1522A.

Absolutley correct! 4:1 safety factor and take the largest diameter or live very dangerously.

Oliver Dunthorne

Hydraulic Engineer

Yes, any elastomeric O-ring will behave as a viscous fluid and fill the outboard atmosphere corner of the packing groove, at any substantial fluid pressure. Therefore it is normal to take the outer diameter as the "piston" diameter, since the system fluid and the elastomer behave as one unit in transmitting pressure.

There are other kinds of seal, subject to suitability of the surface finish of the faces and subject to other pertinent criteria, that could help you.

Consider a metal seal such as a hollow metal O-ring (vent on the ID) or a metal V-seal. Given suitable choice for pressure, these DO NOT yield to pack the outer corner. Especially useful to you could be the metal V-seal, which ceases transmitting system pressure to the outer face much beyond its ID. This may give you the effective diameter reduction you seek. There are lots of makers including Parker.

Mark Bingham
Relativity PL

Relativity PL is correct the o-ring acts as a fluid and the outside diameter is the correct place to calculate the stress.

Others have observed that a 4:1 safety factor is standard. ANSI flanges are normally used with "B7" grade bolts, hydraulic valves and flanges are almost always spec'd out for "Grade 8 or better". Grade 8 bolts are not brittle and are used in high stress area frequently. There are even system "L9" bolts that are stronger than Grade 8.

The question becomes-- is the flange of sufficient thickness and strength to accommodate the proposed pressure.

A smaller metal locating ring inserted into the flange may allow the use of a smaller diameter ORing and solve your problem.

Well it does rather depend on the pressure involved and the o-ring durometer.

But still, pressure enough to make them 'fully square', still won't move the "piston diameter" to the 'square corners' - will it?

So 'correct place' is not quite correct, just a 'safer assumption'.

Fair enough in the OP's first request context; but not hugely useful in finer tolerance problem the OP then outlines.

Or why I wrote what I wrote.

It is not piston area it is bore area.

You are confused as how piston comes here, right? The referred is the biggest possible area on which internal fluid act as ' hydraulic or pneumatic piston' toward separating the flanges apart.

The force acting on the o-ring is the fluid pressure x the area of the o-ring exposed to the pressure. The latter being derived from the slot area in which the o-ring sits - which is obtained from the actual inner and outer diameters.

The fluid tries to push the o-ring out of the slot into the clearance gap between mating surfaces.

Extrusion design pressures for o-rings are provided by the o-ring manufacturers - when coupled to design clearances in accordance with various standards.

But whatever the 'safe' design, at some point, a rising pressure will be enough to cause the o-ring to 'flow' or extrude into into the clearance gap, and perhaps through the gap - at which point it leaks. Which might be aided by the bolts stretching or stripping, or the flange bending.

If you a using non-standard designs - then testing will be required.

I hope this helps.

Option A is the present orientation under discussion.

Option B is proposed by MIKE L's post 13. If flanges tents separate and generate more gap, O-ring gets more gap through which it tents to flow/extrude.

Option C is looks even better. Even flanges tents separate and generate a gap, O-ring faces the same gap through which it tents to flow/extrude.

Interesting. 2 is not from 7

But as it now seems you are designing from scratch - read the second part of 9

Meanwhile be aware neither 1 or 2 are positively retained against 'suction' - be that velocity or vacuum induced.

3 you aught to consider your machining tolerances - which is why you will find this approach is usually taper based.

Not that you will read this.

You are correct in that reply #7 first proposed the metal ring to decrease the diameter therefore getting to the "safety factor needed".

Unfortunately your assumption that diagrams 1 & 2 are not subject to vacuum is incorrect. Neither maintains a positive land to prevent the o-ring from being sucked into the fluid flow. Diagram #3 does retain the o-ring but clearance diameters are very critical depending on pressure.

Diagrams #1 and #2, if the flange is machined for a face to face metal seat outside of the o-ring (or on the metal spacer ring) , then until the bolts stretch or the flange deforms there is no extrusion gap and the ORing can seal to ????? 10,000 psi plus.

Diagram #3 while it may seal during bolt stretch (not good because of fatigue issues) is really dependant on the side clearance as to how much pressure it may retain

"Diagram #3 while it may seal during bolt stretch (not good because of fatigue issues) is really dependant on the side clearance as to how much pressure it may retain"

How about a taper as indicated in post 17?

"Unfortunately your assumption that diagrams 1 & 2 are not subject to vacuum is incorrect."

Read my post again.

Rephrasing me to make me appear wrong then verbosely re-authoring what I said, is hardly useful to the OP.

You say it is a very old design. One can then assume that you are either attempting to verify it meets current standards, or you are attempting to increase the pressure being applied.

If it is an old design , and is to be used the same way is has been in the past, and there have not been any failures due to inherent design deficiencies, then you can make the case that the design has been validated experimentally.

If you are attempting to increase the pressure or apply the design to a new process, then use the O.D. and the desired Safety Factor to calculate the allowable pressure. Good idea to do this regardless of the pressure being applied, just so it is known.

In any case, you must use the counterbore O.D. to determine the area to which the pressure is applied when calculating the force trying to separate the flanges.

The suggestion of using a metal ring to reduce the O.D. of the counterbore is valid only if the ring is welded or bonded into the counterbore. If the ring is loose, the pressure will be applied between the ring and the flanges, and will be acting over the area of the original O.D.

"If the ring is loose".

I guess it's is possible that the ring 'floats up' potentially sealing the crack growing crack between the flanges. But it would only increase the 'piston area' if the looseness is not an adequate leak in itself.

For instance if the ring was rigid enough not to expand under O-ring force, (and a precision sliding fit in the bore OD), it may rise at 'mid o-ring' position, and leakage on the underside of the o-ring first, would drive the ring up to seal against the upper flange, so sealing the gap between flanges.

Perhaps the solution is not to make it a 'beautiful sliding fit', but "loose and leaky", or as one generally does with 'corrective inserts' - a press fit in the counterbore.

But there is always Loctite if press fit tolerances are a problem.

"The suggestion of using a metal ring to reduce the O.D. of the counterbore is valid only if the ring is welded or bonded into the counterbore."

Jumping in both feet first (I'm not qualified in this field), I'd suggest that the only way such a metal ring could effectively change the situation (regarding the area on which the pressure is acting, trying to separate the two pieces) is if the ring were fitted oversize (in thickness rather than diameter), welded/brazed or whatever, then faced off to the flange surface, effectively reducing the area of the joint exposed to pressure.

I'd draw some pictures, but I CBA right now. If anyone is that interested, I'll give it a go.

Hi JD, to an extent how o-rings work/fail is anti-intuitive so misunderstood.

The faces that seal in the above OP's diagrams, are the top and bottom 'flats'. The external diameter is just physical containment.

And as you will see, nor is it particularly desirable for it to be a sealed face.

Generally, outside the o-ring is - until failure - regarded as atmosphere or 'dry'. I.e. we assume if the flanges move apart - the 'o-ring flats' are still even in width and maintained by the line pressure squishing the o-ring section evenly, so the seal between flanges is maintained.

But; if the outer o-ring containment is also a seal, then the o-ring will rotate or roll on the gas (or fluid) trapped in the corner - causing wear - and if pressure equalises (by o-ring 'fluidity'), in that corner, to line pressure, then the lower seal is lost, and the ring is now sealing between outer circumference and top flange. Or the ring is now trying to 'extrude' solely into the gap between the flanges.

This is why in many rod and piston applications, an o-ring is fitted with a 'backing ring', or 'anti-extrusion ring', so that a 'leakage path is provided for that 'entrapped' pressure to vent.

So in a high stress application, you want the 'backing ring' to leak (equally 'top and bottom') of the o-ring.

But is this one? I've no idea. Not enough OP data. Mixed messages about goals.

This is why in many rod and piston applications, an o-ring is fitted with a 'backing ring', or 'anti-extrusion ring', so that a 'leakage path is provided for that 'entrapped' pressure to vent.

So in a high stress application, you want the 'backing ring' to leak (equally 'top and bottom') of the o-ring.

The anti extrusion ring is used to increase the allowable pressure applied to the o-ring. the anti extrusion ring as "named" is to help the ORing from extruding thru the clearance gap.

Leakage path has nothing to do with it. the idea is to seal, not to allow leakage paths.

"The anti extrusion ring is used to increase the allowable pressure applied to the o-ring"

I wonder if you can explain how it increases that allowable pressure - without just rephrasing what I said?

The anti-extrusion ring is used to help prevent the ORing from flowing into the extrusion gap on designs that have a gap. Parker Oring's has a nice diagram explaining how the ring seals the extrusion gap from the ORing. The increase in sealing pressure can be from 150% to 200%. In some designs two backup rings are used for systems that have pressure acting both ways on the seal (piston). An extension of the ORing with backup led to the design of "T seals" which are much used in aircraft.