Troubleshooting

Common Mistakes and Troubleshooting in Gasket, Seal, and O-Ring Manufacturing

The mistakes that turn a passing gland into a field leak or a molding run into scrap, with the symptom, the root cause, and the fix for each.

Most seal failures trace back to a handful of repeatable errors, not exotic material science. The pattern is always the same: a symptom on the floor or in the field, a root cause upstream, and a fix that moves a number. A gland that leaks at 150C, an O-ring extruding a nib at the parting line, a cure press missing its build quantity by 400 parts, and a blank yield sitting 2 points under target all have a diagnosable cause. This guide walks the costliest recurring mistakes so you can catch them before compound, tooling, and cure time are already spent. The tools referenced here quantify each fix.

The most expensive dimensional error is applying the wrong shrink rate to the cavity. Symptom: parts pass first article but land undersize on a critical ID after the compound stabilizes. Root cause is treating all rubber as one shrink value. Nitrile runs near 2 percent linear, while silicone frequently exceeds 3 percent, so a cavity cut for 2 percent shrink running a silicone compound leaves a part roughly 1 percent small on every dimension. On a 20 mm ID that is 0.2 mm, enough to fail the print. Confirm the compound's published shrink in the Material Shrinkage calculator before cutting steel, and re-verify after any filler loading change.

Groove overfill is the quiet killer because it passes at install and fails in service. Symptom: an O-ring that seated fine on the bench extrudes or takes a permanent set weeks later. Root cause is checking fill against as-installed cross-section and ignoring fluid swell and thermal expansion. Design fill should sit in the 60 to 85 percent band, but a compound that swells 10 percent in the service fluid can push a 78 percent gland past 90 percent, where the seal has nowhere to go. Run Seal Groove Fill twice, once at nominal and once with swollen dimensions, and hold the swollen case under your upper limit.

Thin compression set margin gets released as passing and leaks in the field. Symptom: a seal qualified at a comfortable-looking number that leaks after a season at temperature. Root cause is validating margin at the wrong temperature. Compression set climbs steeply with heat and dwell, so a 22 percent set measured at 70C tells you nothing about a 150C application where the same compound may take 40 percent or more. A 30 percent limit with only a 1 to 2 point margin is inside normal batch scatter. Use Compression Set Margin against the actual service temperature and dwell, and hold several points of headroom, not one.

Capacity overcommitment is a planning error that shows up as a missed ship date. Symptom: the press was quoted at cavity math but delivered hundreds fewer good parts. Root cause is planning off gross capacity, cavities times cycles, and ignoring the two real losses. A 64-cavity mold at 38 cycles is 2,432 gross, but 88 percent uptime and 96 percent first-pass yield strip roughly 292 and 86 units, leaving about 2,055 shippable. That 16 percent gap is normal for multi-cavity molding. Plan builds off the good figure from Cure Press Capacity, and never commit a lead time against the theoretical ceiling.

Flash blowout looks like a molding defect but is usually a cost and tooling problem. Symptom: rising deflash labor and a per-part flash cost climbing past 0.20 dollars when precision O-rings should sit in the low single-digit cents. Root cause is a worn flash land, an oversized parting-line gap, or cavity overpacking, all of which force more trim labor and media. Track flash cost per molded part run over run in Flash Scrap Cost; a rising trend means land wear or overpacking, not operator variation. The fix is a tighter flash land or balanced fill, which shows up immediately as lower cost per part.

Mixing loss sources into one yield number hides the real problem. Symptom: a single yield figure trends down but nobody can name the cause. Root cause is folding preform and blanking losses together with molded defects. Blank yield captures off-weight preforms and cutting registration before the mold; molded first-pass yield captures flash, undercure, and voids after curing. A blank can pass at 95 percent and still fail later. Keep Rubber Blank Yield and molded yield separate so a 92.8 percent blank yield points you at preform weight, not at cure. Combining them wastes days chasing the wrong step.

Averaging over unstable runtime distorts every rate you set. Symptom: a molding standard that the cell can never hit despite no obvious problem. Root cause is computing cavitation output across a slow startup ramp blended with steady running, or using total shift time instead of active mold runtime. A tool making 1,850 parts in 8 hours reads 231 units per hour raw, but at 92 percent effective cavitation efficiency it sustains only about 213. Segment the data in Mold Cavitation Output and use active runtime only. Set standards off the effective rate, since the raw number is a ceiling the tool cannot hold cycle after cycle.

Unit and reference errors quietly poison otherwise correct math. Symptom: a groove fill or margin result that is wildly off or that swaps in-spec for a failure. Root cause is mixing units between seal and groove area, or entering measured set above the limit so a positive margin masks a real reject. Keep both areas in the same unit basis, mm squared to mm squared, and confirm the allowed limit is the ceiling and measured set is the result before reading the margin. A negative margin is a failure, not a formatting quirk. Sanity-check every result against the expected band before you release a lot or cut a gland.

Published 2026-07-01.