O-ring design

In one lineCheck squeeze at the worst tolerance corner and the coldest temperature, with the joint gap at its widest under pressure — not at nominal.

Before thisCase sizing (the joint that the seal lives in)

Why this matters

O-rings are the cheapest parts on a motor and some of the most safety-critical. Case joints, the nozzle-to-case joint, igniter bosses and closures all rely on them to hold hot, high-pressure gas. A leak path past a seal lets combustion gas cut through metal in seconds. The worst-known example is the loss of the Space Shuttle Challenger in 1986, traced to a booster field-joint O-ring that could not seal on a cold morning.

How an O-ring seals

How an O-ring seals — radial gland section

An O-ring is a pressure-energised seal. Squeeze gives it a starting contact stress; pressure then pushes it into the low-pressure corner, so contact stress rises with the pressure it is holding. The limit is the clearance gap: push hard enough and the rubber flows into it.

Think of it like…

A plug-type aircraft door

An airliner’s passenger door is slightly bigger than its frame and fits from the inside, so cabin pressure pushes it harder into the frame the higher the aircraft climbs. An O-ring works the same way: the squeeze you build in gives it a starting seal, and system pressure then pushes it harder into the low-pressure corner of the groove. The seal gets tighter as the pressure rises — until the pressure is high enough to push rubber through the clearance gap.

Rubber behaves like a very stiff liquid in bulk and a very soft solid in shape: it changes shape easily but hardly changes volume.

Think of it like…

A water balloon in a box

Push down on a water balloon in a box and it bulges sideways — its volume can’t change, only its shape. If the box is too small it has nowhere to go and the pressure inside climbs enormously. That is why groove volume (gland fill) matters as much as depth: the ring must have room to change shape, plus room to grow when it is hot or swells in fluid.

The four numbers

For a ring of cross-section $d$ in a gland of depth $g$ (the radial or axial space it is squeezed into) and width $w$:

Quantity	Definition	What goes wrong
Squeeze	$s = \dfrac{d - g}{d}$	Too little: no initial seal, especially cold. Too much: overstress, set, high assembly force
Gland fill	$\phi = \dfrac{\pi d^2 / 4}{w\,g}$	Over ~90%: no room to expand — hydraulic lock, cracked groove lands, extrusion
Stretch	$\varepsilon = \dfrac{D_\text{groove} - ID}{ID}$	Too much thins the section and shortens life; negative leaves the ring loose to roll
Extrusion gap	Diametral clearance at the low-pressure side	Too wide for the pressure and hardness: rubber is nibbled into the gap

Stretching a ring onto a groove thins its cross-section. Assuming constant volume, the stretched section is

$$d' = \frac{d}{\sqrt{1 + \varepsilon}}$$

so 2% stretch loses about 1% of the section — and about 1% of squeeze with it.

Nominal is not enough

Every one of those numbers changes across the tolerance band, and the extremes are what fail. A 5.33 mm section has a tolerance of about ±0.13 mm. On its own, that moves squeeze by about ±2.5 percentage points. Add the groove, bore and piston tolerances, then let the parts sit off-centre (the gland is shallow on one side and deep on the other), and a nominal 19% squeeze can range from about 13% to 25%.

Then add the things that move after assembly:

• Temperature. Elastomers expand about ten times more than steel (around $1.6\times10^{-4}$/K). Cold rings shrink and lose squeeze; hot rings grow and fill the groove.
• Fluid swell. A few percent of volume growth in service is normal and must fit in the groove.
• Compression set. Over time and temperature the ring takes a permanent set and stops pushing back as hard. Budget it into minimum squeeze.
• Joint opening. Under pressure, case joints rotate, bores dilate and flanges lift. The gap the ring has to fill gets bigger exactly when the pressure arrives.

The Challenger lesson: the seal must follow the joint

A seal is a spring, and cold springs are slow

In a Shuttle booster field joint, ignition pressure made the joint rotate slightly and open the gap the O-rings sealed. A warm O-ring springs out fast enough to follow. On the cold morning of 28 January 1986 (around 2 °C ambient, with the joint colder still) the fluoroelastomer rings recovered too slowly. Hot gas got past before they sealed, and the joint failed about a minute into flight.

During the Rogers Commission hearings, Richard Feynman showed the effect with a sample of the O-ring rubber clamped in iced water: released from the clamp, it stayed squashed instead of springing back. Two design rules come straight from that:

1. Size the minimum squeeze with the joint opened under pressure, at the coldest temperature — the calculator’s “joint opening” input does exactly this.
2. Do not rely on rubber resilience to follow motion near its low-temperature limit. Specify the compound by its low-temperature recovery (TR-10 and resilience tests), heat the joint, or design the joint so it does not open. The redesigned booster joint added a capture feature to limit rotation, a third O-ring and joint heaters.

Two more joint lessons:

• Redundancy only counts if both seals work. A primary and secondary O-ring with a leak-check port between them is standard. Show that the secondary can still seal when the joint is opened and the primary has already let gas past.
• A leak check pushes the primary O-ring the wrong way. Pressurising between two rings drives the primary towards the motor side of its groove, so at ignition it must first travel across the groove before it can seal. Leak-check pressure and procedure are part of the seal design.

Calculator

O-ring gland — worst-case checkSuggest groove for 20% squeeze / 75% fill

Arrangement

Parts can sit fully off-centre (no guide or pilot)

O-ring (ISO 3601-1 class A)

CS tolerance ±ID tolerance ±

Hardware — min / max

minmaxBore ØPiston ØGroove root ØGroove widthFlange gap

Service

Min temp (°C)Max temp (°C)Pressure (MPa)Hardness (Shore A)Joint opening under pressure (mm)Elastomer CTE (×10⁻⁶/K)Volume swell (%)Compression set by end of life (%)

Gland verdict

—

Nominal squeeze

—

Worst-case squeeze

—

Fill nominal / worst

—

Compounds against this temperature range and medium

Compound	Range °C	Fit	Notes

Screening values: squeeze bands, the fill limit and the extrusion table are typical guidance, not ISO 3601-2 dimensions. Stretch thinning assumes constant volume. Confirm against ISO 3601-2/-4, the seal supplier's tables, and a compatibility test with your actual fluid at temperature.

Use Suggest groove to size a groove for 20% nominal squeeze (14% for dynamic) and 75% fill around the selected ring, then edit the min/max values to your real tolerances and check the worst case.

Worked example — a nozzle-to-case radial seal

The calculator’s defaults: a static radial seal in a piston-type closure, 150 mm bore at 10 MPa, steel housing, service range −40 to +60 °C. Ring: 5.33 mm section, 139.29 mm stock ID. Groove root Ø141.45–141.55, width 7.0–7.2 mm; bore Ø150.000–150.063, plug Ø149.860–149.900; parts free to sit off-centre.

1. Nominal. Gland depth $(150.03 - 141.50)/2 = 4.27$ mm. Stretch onto the root is $(141.5 - 139.29)/139.29 = 1.6\%$, which thins the 5.33 section to 5.29 mm. Squeeze $= (5.29 - 4.27)/5.29 = 19.3\%$; fill 73%. Comfortably in band.
2. Worst corners. Fully off-centre, the gland ranges from 4.16 to 4.41 mm. Maximum squeeze — largest ring, hot, swollen 5% — is 25%. Minimum — smallest ring, most stretch, −40 °C — is 13.5%, and 10.8% after 20% compression set. That is right at the edge of the 10% screening floor.
3. Extrusion. The clearance can reach 0.203 mm. At 10 MPa a 70 Shore ring wants about 0.08 mm, and even 90 Shore only about 0.15 mm. A back-up ring is needed, or a tighter fit and a harder compound.
4. Compound. Standard FKM is typically limited to about −20 °C. At −40 °C it fails the temperature check — use a low-temperature fluoroelastomer grade or another compound, and confirm the TR-10.
5. Joint opening. Enter 0.25 mm of radial opening under pressure: minimum squeeze falls to 6.9% and the cold-resilience check fails. That is the Challenger situation in miniature — fix the joint stiffness, not the seal.

Choosing the compound

Pick the compound for the coldest, hottest and chemically worst moment — in that order. The table in the calculator checks every compound against your temperature range and medium. The traps that catch experienced engineers:

• EPDM with petroleum grease. EPDM is excellent in water, steam and phosphate-ester fluids, but swells and fails in mineral oil — including the assembly lubricant. Specify the lubricant on the drawing next to the compound.
• Standard FKM when cold. Superb hot and in fuels, but standard grades stiffen around −20 °C. Low-temperature grades exist; specify one by its TR-10.
• Silicone in dynamic duty. Wide temperature range, but poor tear and abrasion strength. Static seals only.
• Oxygen service. Ordinary elastomers can ignite in oxygen. LOX and high-pressure oxygen seals use specifically tested materials and controlled cleanliness.
• Brittle point is not the limit. Sealing is lost far above the brittle point. Use TR-10 or resilience data.

Groove details that decide the outcome

• Surface finish. Static sealing faces typically around Ra 0.8 µm (finer for gases and vacuum); dynamic counterfaces Ra 0.2–0.4 µm; groove walls about Ra 1.6–3.2 µm. The machining lay should not cross the seal.
• Lead-in chamfers of 15–20°, deburred and blended, on every edge the ring passes over. More O-rings are cut on assembly than fail in service, and a cut ring cannot be seen once installed.
• Groove corner radii small enough for the ring to seat, never sharp.
• Back-up rings on the low-pressure side, or on both sides if pressure reverses. Add their width to the groove.

Where this stops

Screening, not a seal qualification

The squeeze bands, fill limit and extrusion table are typical guidance, not ISO 3601-2 dimensions. Real joint seals are qualified by test at temperature, with the joint deflecting as it does in service, and the compound’s low-temperature recovery measured directly.

References and further reading

• BS ISO 3601-1 to -5 — O-ring sizes and tolerances, housing dimensions, quality, back-up rings, material suitability.
• SAE AS568 — the inch-series O-ring sizes that ISO 3601-1 class A is based on.
• Parker Hannifin, O-Ring Handbook (ORD 5700) — groove design, extrusion limits and compound data.
• Report of the Presidential Commission on the Space Shuttle Challenger Accident (Rogers Commission, 1986), including R. P. Feynman’s Appendix F.

Note

Educational and screening material. Final groove dimensions and tolerances come from ISO 3601-2, the seal manufacturer, and your programme’s qualification testing.