Engineering Answers / Production & Quality
Hardness tells you how hard the rubber is on the day it ships. Compression set tells you whether it will still be pushing back a year later — and that is the property a seal is actually hired for.
Compression set is the permanent deformation a rubber part keeps after being compressed at elevated temperature for a period of time. A seal with high set stops pushing back, contact stress drops, and the leak starts slowly rather than at installation. Hardness does not predict this: two 50A compounds can differ by a factor of two in set depending on cure system, post-cure and filler. For any part held under load, set should be specified with a test condition, not inferred from durometer.
A seal that leaks on day one is a manufacturing defect. A seal that leaks in month fourteen is usually a compression set problem, and it is much more expensive: the product is in the field, the batch is untraceable, and the failure looks intermittent because it depends on temperature and pressure.
Compression set is the deformation a rubber part retains after being compressed for a period at temperature. Test it by compressing a sample to a fixed deflection, holding it at temperature for a set time, releasing it, and measuring how far it fails to return.
The mechanism is straightforward. A seal works because the rubber pushes back against the mating surfaces. That push-back is contact stress. As the network relaxes and crosslinks rearrange under heat and load, some of the stress never comes back. The seal is still the same shape on the outside; it simply is not pushing any more.
Hardness is an instantaneous measurement. Press a durometer into a part and it reads the resistance at that moment, at room temperature, at that point on the surface. It says nothing about what happens over 1,000 hours at 120 °C.
Two compounds can both read 50A and behave completely differently:
| Property | Compound A | Compound B |
|---|---|---|
| Hardness | 50A | 50A |
| Cure system | Platinum addition, post-cured | Platinum addition, no post-cure |
| Set after 22 h at 175 °C | Low | Roughly double |
| Service life in a hot seal | Years | Months |
Both are legitimate mouldings. Both pass a durometer check. Only one is a seal.
Cure completeness. Under-cured silicone keeps reacting in service, and the network continues to change shape under load. Cure time and temperature are process parameters that should be validated, not estimated from a data sheet.
Post-cure. A post-cure cycle drives off volatiles and completes the crosslink network. It adds a day to the schedule and is routinely skipped by suppliers optimising for delivery — and it is invisible in the incoming inspection, because the part measures correctly on the day it ships.
Service temperature. Set accelerates with temperature. A compound rated for 150 °C continuous service will take a set much faster at 180 °C, and the seal may still look fine when it is removed.
Compression ratio. More compression generates more stress but also more relaxation. Seals are usually designed in the 15–30% compression range; beyond that, the material is being asked to do work it will not sustain. Groove depth and gland fill are part of the set story.
Filler and formulation. Fumed silica loading, polymer molecular weight and any custom compounding for oil or fuel resistance all change set. This is why "silicone" is not a specification.
For automotive and industrial programmes, set is measured as part of the property test set alongside tensile and tear strength, and it is paired with thermal ageing — for example 200 °C for 1,000 hours — so that the material is evaluated at the temperature the application actually sees rather than at a comfortable bench condition.
For a medical or consumer seal, the same logic applies with different numbers: state the temperature, the duration, the deflection and the acceptable limit, and require the report to reference the test method. A supplier who cannot state the condition cannot be compared to another supplier.
"Compression set ≤ X% after Y hours at Z °C at 25% deflection, per ASTM D395 Method B, on the finished part or a plaque from the same lot." Without a condition, the number is marketing. With it, the number is testable — and it is the only number that predicts whether the seal still works next year.
Even a good compound loses some force, so the design should not depend on 100% retention.
Hardness is what the part is on the day it ships. Compression set is what it will be after a year in service. Specify set with a test condition, require post-cure, and design the groove so the material is not over-compressed. If a seal fails slowly, in the field, at temperature — the first thing to ask for is not a hardness certificate. It is a set report with the condition printed on it.
It depends entirely on the test condition, so a number without a condition means nothing. A common specification for a sealing compound is a set below roughly 20–25% after 22 hours at 175 °C, measured per ASTM D395 Method B at 25% deflection. A compound that reports 15% under that condition may report 40% at 200 °C over a longer period, which is why the RFQ has to state the temperature and duration the seal will actually see.
No. Hardness changes how much force the seal generates at a given compression; it does not change how much of that force is lost permanently. A harder compound with poor set will still relax and leak — it will simply leak at a slightly higher residual pressure. Set is a material and cure property, not a stiffness property.
Three things dominate: incomplete cure or the wrong cure system, no post-cure (or the wrong post-cure cycle), and service temperature above what the compound was formulated for. Post-cure is what drives off volatiles and completes the crosslink network; skipping it to save a day in the schedule is one of the most expensive shortcuts in silicone moulding, because the part looks and measures correctly on the day it ships.
Yes, wherever the part is held under sustained load. A silicone foot on an appliance, a suction base, a damping pad, a button dome — all of them take a permanent set and all of them change function when they do. The test is the same; only the acceptable limit changes.