A snap-fit is the cheapest joint in plastics: no screws, no inserts, no adhesive, moulded as part of the wall. It is also the joint most often got wrong, because the instinct that serves everywhere else in mechanical design — work in stress, keep it under yield — is the wrong instinct here.
The one idea: design to strain, not stress
Deflect a snap beam far enough to clear its ledge and linear bending theory will tell you the stress is far above the material's yield strength. Yet the part comes off the line, snaps together, and works. It is not that the calculation is wrong; it is that the event is brief and the material is non-linear. Plastics tolerate a short excursion to high strain that they would never tolerate as a sustained stress.
So snap-fits are dimensioned to a permissible strain. And when you do need a force — how hard is this to push together? — you must use the secant modulus at that strain, not Young's modulus. Young's modulus is the tangent at the origin; at 1.5% strain a typical engineering plastic has already softened, and using the initial modulus overpredicts both the stress and the assembly force by the same proportion. Ticona's worked example makes the point plainly: a 1.6 × 10⁶ psi initial modulus predicts 24 000 psi at 1.5% strain, where the real curve says 18 000 psi.
A snap-fit cantilever is breaking on assembly, so you make the beam twice as thick to strengthen it. At the same snap depth, what happens to the strain in the beam?
Strain is ε = 3 t y / 2L², so it is linear in thickness. Doubling t at the same deflection y doubles the strain and the beam cracks on the first assembly. This is the single most common snap-fit mistake, and it comes from designing to a stress the way you would in metal. A snap fit is designed to a strain: to fix a beam that breaks, make it longer, or taper it so the strain spreads along the beam instead of piling up at the root.
Try it — beam proportions vs strain
Strain goes as 1/L² — drag the length and watch how fast it falls. Width never appears in the strain at all; it only changes the force.
How much strain is permissible?
There is no single number, but there is a clear rule, and it depends on what kind of plastic you have:
- Semi-crystalline (PP, PE, POM, PA, PBT) — for a single brief assembly, almost the full yield strain.
- Amorphous (PC, ABS, PS, PMMA, PVC, SAN) — about 70% of the yield strain.
- Glass-fibre reinforced — no distinct yield point, so about half the elongation at break.
As orders of magnitude: unfilled materials land around 6% and filled ones around 1.5%. And if the joint has to come apart and go back together more than once, take 60% of whatever the above gives you. That factor is the difference between a battery cover that survives its service life and one that snaps off on the fourth opening.
Geometry: the two ratios worth memorising
Ticona gives two reference beams worth carrying in your head. A 6% strain beam is one whose thickness is 20% of its length and whose deflection is also 20% of its length — a 5:1 length-to-thickness ratio. A 1.5% strain beam is 10% and 10%, a 10:1 ratio. If your beam is far outside those proportions, something is wrong before you calculate anything.
Beyond that, length is your friend: strain goes as 1/L², so a beam 40% longer is half as strained at the same deflection. Thickness works against you linearly. Width does not appear in the strain equation at all.
Taper the beam — the cheapest win available
A parallel beam concentrates its bending stress at the root, where the moment is highest and the section is no bigger than anywhere else. Taper the thickness toward the snap foot and the stress spreads along the length instead. Ticona tabulates the gain as a factor K on permissible deflection; Bayer, in a completely separate document, says tapering to half thickness "increases the permissible deflection by more than 60%". Those are the same statement: K = 1.636 at hL/h₀ = 0.5.
It costs nothing. The mould is no harder to cut, the part uses less material, and the beam takes 64% more deflection for the same root thickness. If you take one habit from this page, take this one.
Try it — what the taper buys
Same root thickness, same length, same permissible strain — the tapered beam simply takes more deflection. The shading is bending stress: concentrated at the root when parallel, spread along the length when tapered.
Angles: one for going in, one for staying put
The lead angle governs assembly. Both the deflection force and friction have to be
overcome, and the mating force is W = P·(µ + tan α)/(1 − µ·tan α). Use
15–30°. Above about 45°, with a realistic friction coefficient, the parts may simply
refuse to go together.
The return angle governs retention, and the same equation applies with the return
angle substituted. For a joint meant to be separable, use 30–45°. Watch the
denominator: when µ·tan α = 1 the force goes to infinity. That is the
self-locking angle, and past it no pull will separate the joint — something has to
break instead.
Try it — angles, friction and self-locking
Push the angle up and the multiplier climbs toward infinity at the self-locking angle. Raise the friction and that wall moves closer.
The failure mode nobody expects: roll-off
It is widely believed that a 90° retaining face makes a snap-fit unbreakable — that the foot or the ledge would have to shear. It is not true, and two of the source guides say so independently.
On a hook-type snap foot the separation force acts off the beam's neutral axis. That offset is a bending moment, and it can simply roll the foot back off the ledge, releasing the joint with nothing sheared and nothing visibly broken. This is why a joint that passed a static pull test comes apart in a drop test.
There are two fixes, and they are the same fix wearing different names. Ticona's snap loop and First Mold's sleeve-type retainer both put the force line through the neutral axis, so there is no moment to roll anything. The cost is a weld line where the two flow fronts meet around the loop — thicken the section there, or move the gate so the weld line lands somewhere it does not matter.
Details that decide whether it survives
- Fillet the root. An inside radius of about half the beam thickness, and never below 0.5 mm. A sharp inside corner concentrates stress exactly where strain is already highest — the commonest cause of snap-fit breakage there is.
- Do not leave it assembled under stress. A snap held deflected will creep, the retention will relax, and it will quietly stop holding. Design so the beam returns to zero strain once engaged.
- Mind the mould. Most snap features need a side action or a lifter unless you design a through-hole under the foot. That is a real cost, and converting a snap so it can be moulded straight up is often worth more than any calculation on this page.
- Guide, then position, then lock. A good joint leads the parts in, locates them precisely, and only then engages. Without positioning features the snap takes loads it was never sized for.
- Hermetic seals and snap-fits do not mix. If the beam or the ledge relaxes at all, the seal load goes with it.
- It is irreversible. A broken snap cannot be repaired — the part is scrap. That argues for a generous strain margin on anything expensive.
Which type to use
Cantilever for almost everything — easy to mould, easy to calculate, easy to tune. Annular for anything rotationally symmetric: caps, bezels, pen barrels, ball-and-socket joints; permissible interference is simply ε·d, and doubles if both parts flex. Torsional for a press-to-release catch, where travel comes from twisting a long thin bar rather than bending a beam — low operating force and far better cycle life.
Size them on the Cantilever Designer, Annular Snap-Fit and Torsional Snap-Fit tools, and keep the strain, friction and K-factor tables to hand.