Shrinkage Allowance Calculations for Precision Industrial Silicone Seals

Paste 2% shrinkage on a platinum RTV job and you will cut the cavity wrong.

The internet band about 1.5% to 4% is mostly heat-cured silicone (LSR injection and millable HCR) plus some stacked vacuum-cast processes. It is not a law. Platinum RTV mold rubber is typically ~0.1% or less after a proper cure. Tin-cure RTV shrinks more and keeps shrinking in storage. ISO 3302-1 even warns that some silicones’ larger shrink makes classes M1 and M2 hard to hold. Precision industrial seals care about that split more than they care about a round number on a blog.

This is how to put allowance into a steel cavity for LSR/HCR seals, and how not to contaminate that math with RTV prototype data.

The percentage is not a constant

Shrink here means: the part comes out smaller than the cold cavity (after cool and after any post-cure), so the cavity is cut larger.

What moves the number:

• Polymer and filler. More filler, generally less shrink. Different VMQ/FVMQ/PVMQ grades are not one percentage.
• Hardness as a proxy for filler, not as a magic table. Do not use Shore A as a shrink calculator.
• Cure temperature and time. Hotter, longer - more lock-in of the network, and more thermal excursion.
• Post-cure. A second shrink. Seals that are post-cured for volatiles will move again after T1 if T1 skipped the oven.
• Constraint. Rubber bonded to a metal ring cannot shrink with the ring the way a free lip can. The lip still tries.
• Section thickness. Thick and thin do not always shrink alike (see ribs/sink in this cluster).
• Direction. Anisotropic effects are usually smaller than in filled thermoplastics, but long, thin seals can still surprise you in ID vs width.
• Batch. The reason you do not tattoo 2.00% into a family of tools forever without a recheck when the compound lot family changes.

Do not publish a fake universal table. Ask the compound supplier for a shrink range on that grade in that process, then verify.

LSR and HCR: how to put allowance in the cavity

A workable sequence for a precision seal (gasket, gland, custom lip - not a standard O-ring that belongs on an O-ring spec):

1. Freeze the compound and the process. Grade, hardness, color (some pigments move shrink slightly), LSR vs HCR, post-cure yes/no and the actual oven profile.
2. Get a supplier starting factor or a shop history on a similar section. Treat it as a center, not as a promise.
3. Scale the CAD for the cavity. Uniform scale is the default: cavity = nominal × (1 + s), where s is the decimal shrink (2% → 0.02). For a 50.00 mm ID and s = 0.02, the steel ID is 51.00 mm if the ID is a free dimension formed in steel.
4. Do not scale inserts. Metal and plastic inserts stay at 1:1. Only the rubber volumes grow. Hybrid bodies need a cavity that is grown relative to a fixed insert nest - that is a CAD skill, not a uniform “scale part 2%.”
5. Decide F vs C. Closure heights in compression will not obey a pretty scale factor if flash thickness walks. LSR injection can treat more dims as fixed if the tool actually shuts (ISO 3302-1 note).
6. Cut a single-cavity or a shrink coupon in the real steel temperature, real post-cure.
7. Measure after 16 h (ISO 3302-1), conditioned, undistorted. Soft seals squash in calipers; use the method you will use in production.
8. Adjust the insert, not the process window, for the remaining error. A 0.3% miss on a 50 mm ID is 0.15 mm - inside some M2/M3 bands, fatal for a precision gland.

Worked numbers (illustrative, not a grade promise):

• Target seal ID 40.00 mm, free rubber, assumed s = 0.018 (1.8%).
• Cavity ID = 40.00 × 1.018 = 40.72 mm.
• First-article ID after post-cure = 40.10 mm → actual shrink was 0.0152 on that cavity, not 0.018.
• Recut toward 40.00 × 1.0152 = 40.61 mm, or polish/EDM the difference if the steel still has meat.

That recut loop is the calculation. A spreadsheet without the coupon is a guess.

If you ignore post-cure: T1 passes at 40.02 mm. After the customer’s 4 h oven, the ID is 39.7 mm and the gland leaks. The allowance was for press-shrink only.

Platinum RTV is a different number

If the “seal” is a prototype poured in platinum RTV, or the mold is platinum RTV, do not import 2%.

Platinum RTV: near-zero shrink, so the cavity (or the part, if the part is RTV) tracks the master. Make the master to the nominal you want, after accounting for any master-print error. SLA masters that are 0.2 mm fat make 0.2 mm fat RTV seals. The rubber will not shrink you back to CAD.

Tin-cure RTV: higher shrink, continues over weeks. Illegal as a source of LSR steel factors, and illegal as a precision seal process if the ID must hold a month later.

Vacuum-cast PU in an RTV tool has its own shrink stack (tool + PU). Do not use that stack as LSR allowance.

Constrained seals and bonded hardware

A seal vulcanized to a stainless ring:

• The rubber over the bond shrinks and goes into tension or peels (see metal overmolding in this cluster)
• The free lip shrinks toward the ring
• ID may grow or shrink depending on whether the lip is inside or outside the constraint

Uniform scale of the whole CAD will put the groove in the wrong place relative to the ring nest. Model the metal at 1:1. Grow only the rubber, then expect to tune the lip after a bonded coupon. ISO 3302-1’s note on non-rubber parts affecting shrink is this case.

What to write so the shop does not pick 2% from a forum

RFQ sentence: *Process [LSR/HCR], compound grade if known, post-cure [profile], shrink to be established on a single-cavity coupon; do not apply RTV or 2% default. Unspecified dims ISO 3302-1 M[n]; seal ID/OD/height are criticals. Inserts at 1:1.*

If the vendor answers “we always use 2.5%,” ask for the last coupon on a similar section. Always is how precision seals miss by a percent and still look “in family” on a loose class.

The test that ends the math: coupon, real oven, 16 h wait, optical or low-pressure measurement of ID, OD, and section. Recut from that. Until those three numbers exist, you do not have a shrinkage allowance. You have a rumor scaled in CAD.

FAQ

Why is pasting 2% on a platinum RTV seal the wrong cavity?

The 1.5-4% band is mostly heat-cured LSR and HCR. Platinum RTV is typically near 0.1% or less after a proper cure, so the part tracks the master. An oversized SLA master stays oversized. Tin-cure keeps shrinking in storage and is not a source of LSR steel factors.

Do I scale the metal insert with the rubber?

No. Inserts stay at 1:1. Only rubber volumes grow. A uniform "scale part 2%" on a hybrid body puts the groove in the wrong place relative to the nest.

Why wait to measure a precision seal after 16 hours?

ISO 3302-1 expects conditioned, undistorted measurement. Soft seals also squash in calipers, so use the production method. T1 that skipped post-cure will move again in the customer's oven; allowance for press-shrink only is how a gland leaks later.

Can an RTV prototype ID become the LSR shrink factor?

No. Vacuum-cast PU in an RTV tool has its own stack. Platinum RTV near-zero shrink is a different number from heat-cured LSR. Freeze grade, process, color, and post-cure, take a supplier center, then coupon in real steel temperature.

How does a stainless ring change seal shrink?

Rubber over the bond goes into tension or peel; the free lip still tries to shrink toward the ring. ID may grow or shrink depending on whether the lip is inside or outside the constraint. Uniform CAD scale will misplace the groove; grow only rubber, then tune the lip on a bonded coupon.