A groove is not a hole with a round bottom. It is a controlled volume with a controlled depth, and every dimension in it does a job: the depth sets the squeeze, the width sets how much room the swollen ring has, the corner radii decide whether the ring survives installation, and the surface finish decides whether the seal leaks through the wall instead of around it.
This note works through the decisions in the order they are usually made, using AS568-214 (0.984 in ID × 0.139 in cross-section) as the worked example. The arithmetic is the same for any size in the table.

Start from the cross-section, not the ID
Everything in gland design is expressed as a fraction of the cross-section (CS). A groove depth that gives 20 % squeeze on a 0.139 in section gives 28 % on a 0.103 in section — the same millimetre value is not the same design.
Three numbers define the starting point:
- CS — the nominal cross-section from the size table. For AS568-214: 0.139 in (3.53 mm).
- Squeeze — the percentage of CS removed by the gland, in the direction of compression.
- Fill ratio — the share of the groove volume the ring occupies once installed.
Get the first two right and the third usually follows. Get the third wrong and the seal behaves unpredictably at temperature, because a swollen ring in a full groove has nowhere to go and pushes the joint open.
Squeeze targets by seal type
| Seal type | Squeeze | Why |
|---|---|---|
| Static radial (rod or piston, no motion) | 20–30 % | Maximum tolerance to surface finish and thermal cycling |
| Dynamic radial (reciprocating) | 15–20 % | Lower friction and less wear; the ring slides, so it does not need to conform as hard |
| Static face (flange, cover) | 20–30 % | Same as static radial, but the groove is usually machined in the cover |
| Rotary shaft, low speed | 10–15 % | Friction heat dominates; squeeze is deliberately light |
| Vacuum | 20–25 % | Needs enough stored stress to hold against zero internal pressure |
For AS568-214 at 20 % squeeze the groove depth is 0.139 × (1 − 0.20) = 0.111 in. That is the nominal depth. The tolerance on it is not cosmetic: a depth at the shallow end raises squeeze and lowers service life, a depth at the deep end lowers contact stress and leaks. A common shop tolerance is ±0.002 in on depth for this size, which moves squeeze by about 1.5 points either way.
If the drawing gives the depth without a tolerance, the shop will use its default, and the seal performance will vary between suppliers for reasons nobody can see on the part.
Groove width comes from fill ratio
The ring has to fit in the groove with room for thermal expansion and for swelling in the media. The usual rule is that the installed ring should occupy no more than 75–85 % of the groove cross-sectional area. Above 85 % the ring becomes effectively incompressible and starts loading the joint instead of sealing it.
The arithmetic for a 0.139 in section:
- Ring cross-sectional area = π × (0.139 / 2)² = 0.0152 in²
- Groove area at 75 % fill = 0.0152 / 0.75 = 0.0202 in²
- Groove width = 0.0202 / 0.111 = 0.182 in
So a groove about 0.182 in wide with a 0.111 in depth holds the ring at 20 % squeeze and 75 % fill. Published gland tables for this cross-section land in the same region — which is the useful check: if your calculated width is 25 % away from the table value, you have either the wrong squeeze or the wrong fill assumption.
Width also has a mechanical floor. The groove must be wide enough that the ring does not bottom out on both side walls when it swells, and wide enough to accept the thermal expansion of the groove material. Aluminium housings grow faster than steel covers; a groove designed at zero clearance for a steel-on-steel joint can pinch the ring in an aluminium one at 100 °C.
Clearance, extrusion and when a back-up ring is mandatory
The gap between the two metal parts — the diametral clearance — is where the seal either holds or gets nibbled. Under pressure the rubber is pushed into that gap. If the pressure is high enough and the gap wide enough, the ring extrudes: it shears at the edge, and the joint starts leaking with a ring that looks chewed rather than worn.
Published design guides give a pressure-versus-clearance ceiling for each hardness. The shape of the relationship matters more than the exact numbers:
| System pressure | Maximum clearance, 70 Shore A | Maximum clearance, 90 Shore A |
|---|---|---|
| 1,000 psi | 0.012 in | 0.016 in |
| 1,500 psi | 0.008 in | 0.012 in |
| 2,000 psi | 0.005 in | 0.008 in |
| 2,500 psi | 0.004 in | 0.006 in |
Two conclusions follow. First, a 70 Shore A ring is not a high-pressure seal: above roughly 1,500 psi the clearance needed to assemble the joint is already wider than the material can bridge. Second, the fix for extrusion is a back-up ring or a harder compound, not more squeeze. Adding squeeze to a clearance problem increases the force pushing rubber into the gap.
Groove geometry that does not damage the ring
The ring has to get into the groove without being cut. That imposes four details that are routinely left off drawings:
- Lead-in chamfer of 15–20° on the bore or rod, long enough that the ring slides in rather than catching on the edge.
- Corner radii of 0.005 in minimum at the groove bottom. A sharp internal corner is a stress riser in the rubber and a place where the ring takes a permanent set.
- No burrs, no spiral marks. A turned groove with a thread-like finish will leak along the helix, because the seal cannot conform to a continuous groove that crosses its path.
- Deburr the cross-drilled ports. A drilled port breaking into the bore leaves a burr that cuts the ring on the first stroke. This is one of the most common causes of a rod seal failing in the first week.
Surface finish: the part that gets argued about
For static seals, a groove finish of 8–16 µin Ra is the usual target, with the groove floor smoother than the side walls. Rougher than about 32 µin and the rubber cannot conform at low pressure; much smoother and the surface holds no lubricant film, which raises friction and can cause stick-slip in dynamic seals.
The finish requirement is directional. Circumferential turning marks in the groove floor are tolerable; axial scratches across the sealing path are not, because they form a leak path the rubber is not asked to fill.
Stretch, and why the installed cross-section is smaller than the table
An O-ring installed on a rod is usually stretched a little to stay in place. Stretch reduces the cross-section: at 5 % stretch the section loses roughly 3 % of its height, and at 10 % it loses more than twice that. A gland designed for the nominal 0.139 in section and then assembled with a 10 % stretched ring is over-squeezed by design.
This is why the groove diameter is specified for the seal it will hold, not for the nominal bore. If the application needs heavy stretch to retain the ring, the groove depth has to be adjusted, or a different seal type used.
Checking a design before it is cut
A five-minute review catches most gland mistakes:
- Compute squeeze from the actual depth tolerance band, not the nominal depth. Report the min and max, not the middle.
- Compute fill ratio with the ring at its maximum swell in the actual media, plus thermal expansion at the maximum temperature.
- Compare clearance at maximum pressure against the table for the chosen hardness. If it fails, change hardness or add a back-up ring.
- Check that the groove width is at least 1.3 × the cross-section. Below that, the ring has no room to roll or swell.
- Check the drawing for chamfer, radii, burr callout and finish direction. If any is missing, the shop will decide for you.
A groove at maximum depth with a ring at minimum cross-section and a bore at maximum diameter is a different seal from the nominal drawing — usually the one that leaks. Design to that corner, not to the centre.
Frequently asked questions
How deep should an O-ring groove be?
Depth follows from the squeeze target: depth = cross-section × (1 − squeeze). For a 0.139 in section at 20 % squeeze that is 0.111 in. Specify the depth with a tolerance, because the tolerance band is what the seal actually sees.
What is the minimum groove width for an O-ring?
At least 1.3 times the cross-section for most static seals, and enough that the ring occupies no more than 75–85 % of the groove area once swollen. For a 0.139 in section that works out around 0.18 in.
Can I use the same groove for 70 and 90 Shore A?
No. A 90 Shore A ring in a groove designed for 70 Shore A is over-squeezed and takes a higher seating load, and it conforms less well to the surface. The groove depth for a harder compound is usually slightly greater, and the clearance allowance changes too.
Why does my O-ring keep getting cut during assembly?
Almost always the lead-in. Either the chamfer is missing, too short, or has a burr on its edge. The second most common cause is a sharp groove corner radius under 0.005 in.
Does groove depth tolerance really matter that much?
For a 0.139 in section, ±0.002 in of depth moves squeeze by roughly 1.5 percentage points. Across a production run that is the difference between a seal at 18 % squeeze and one at 21 % — small on paper, visible in leak rate at low pressure.
Should the groove be in the rod or the bore?
For a radial rod seal the groove normally goes in the bore (the housing), so the ring is supported on the rod. For a piston seal the groove goes on the piston. Face seals put the groove in the cover, which is easier to machine and to replace.