The Clip That Doesn't Break
The Clip That Doesn't Break
The lid has to open for battery service and calibration, and every time it opens, the clip bends past its limit, because the clip is a spring, and the spring was sized by feel instead of by strain.
The lid has to open. The battery has to be serviced, the board has to be calibrated, and the lid has to come off, twenty-five times over a ten-year life. The clip that holds the lid on is a small plastic arm with a hook on the end, and every time the lid opens, the arm bends.
The first nine openings are fine. The tenth is the one that snaps it, and the clip lies on the bench, and the lid no longer closes, and the product that was designed to be recalibrated, not replaced, is now a product with a broken lid.
The clip did not break because the plastic is weak. It broke because the design asked it to bend further than the plastic can stretch, and the place it broke, right at the root, where the arm joins the wall, is the exact spot a single formula predicts. The clip is a spring, and the spring was sized by feel instead of by strain. And strain is a number, and the number is small, and the number is the whole design.
The simple idea first
Think of the clip as a diving board. The board is fixed at one end and free at the other, and the hook is at the free end. To open the lid, the hook has to ride over a ledge, so the board bends out of the way and then springs back behind it. While it is bent, the outside surface of the board stretches. That stretch is called strain, and it is largest where the board joins the wall, at the root.
Every plastic tolerates a certain strain before it cracks or stays bent. A clip designed without checking that number is a coin flip. (Loftsmith) The check is one formula, and the formula has four numbers in it: how thick the arm is, how far the hook has to move, how long the arm is, and how much strain the plastic tolerates. Get those four numbers right and the clip lives. Get one of them wrong and the tenth opening snaps it.
The formula: one number, three levers
For a cantilever beam of constant rectangular section, the maximum strain at the root for a tip deflection y is:
ε = 3·h·y / (2·L²)
where h is the root thickness, L is the beam length from root to hook, and y is the deflection, how far the hook has to move to clear the ledge, which is the overlap, the undercut depth. (NovaSolver; Xometry Pro, from the Bayer and BASF snap-fit design guides)
Read it like a set of levers. Strain grows with thickness and with overlap, and it falls with the square of length. So a longer arm is by far the strongest fix: make it 40% longer and the strain halves. (Loftsmith) A thicker arm, which feels like the obvious fix, makes things worse, because the strain grows in direct proportion to the thickness, and the arm that is too thick is the arm that snaps. (Loftsmith)
Bayer's guide writes the same relation the other way round, as the allowable deflection, how far the hook is allowed to move for a given strain:
y = 0.67 × ε × L² / h, for a beam of constant rectangular section.
y = 1.09 × ε × L² / h, if the beam tapers so that its thickness at the hook is half its thickness at the root. (Bayer, via Synectic; Navinyasa)
The taper is the first fix, and it is free: by thinning the beam toward the tip, you distribute the strain much more evenly along the whole length instead of concentrating it at the root, and the same strain allows 63% more deflection (1.09 ÷ 0.67 = 1.63). The rule of thumb is to reduce the thickness at the end of the beam to one-half the value at the root.
(Synectic) Bayer's own comparison says it plainly: the cantilever arm of unsatisfactory design has a constant cross section, and its uneven strain shows in the fringes, this design uses 17% more material and exhibits 46% higher strain than the optimal design, in which the thickness decreases linearly to 30% of the original cross-sectional area and the strain in the outer fibers is uniform throughout the length of the cantilever. (Bayer)
Turned around, the formula tells you the shortest arm that is safe:
minimum length = √(1.5 × h × y ÷ allowed strain) (Loftsmith)
A 1.5 mm thick PLA arm that has to move 1 mm needs to be about 11 mm long; in PETG, about 8 mm. (Loftsmith) The instinctive design, a stubby 5 mm arm, would see about 9% strain, four times what PLA tolerates. That is the crack you heard. (Loftsmith)
The material: how far the plastic can stretch
The allowable strain is a property of the material, and it is not the same for every plastic, and it is not the same for one assembly and for twenty-five. Bayer's table gives the permissible short-term strain at 23 °C, for a single assembly:
| Material | Permissible strain, single assembly at 23 °C |
|---|---|
| Polycarbonate (PC, Makrolon) | 4% |
| PC blends (Makroblend) | 3.5% |
| PC/ABS (Bayblend) | 2.5% |
| PC with 10% glass fibre | 2.2% |
| PC with 20% glass fibre | 2.0% |
| Apec High Heat PC | 4% |
(Bayer, Snap-Fit Joints for Plastics, A Design Guide, Table 2)
And the rule that decides the hub's clip: for a joint that is opened and closed often, use about 60% of the single-assembly value. (Bayer) The hub's lid opens 25 times over ten years, that is a joint opened and closed often, so the allowance is 60% of 4%, which is 2.4%. The other sources say the same thing in their own words: for repeated use, aim for less than half the single-assembly value.
(NovaSolver) And the printed version: the design strains in Loftsmith's table are conservative values for clips that are assembled more than once, a clip that snaps together only once can go a little higher. (Loftsmith)
The glass fibre is the trap. Adding glass makes the plastic stiffer and stronger, and it makes the allowable strain smaller, PC with 20% glass fibre tolerates 2.0%, half of unfilled PC's 4%. The stiffer material is the one that snaps first, because the strain is what breaks the clip, not the force.
And the temperature is the second trap: the allowable strain varies strongly with temperature, brittle when cold, and with strain rate, number of cycles, glass-fibre content, and the position of weld lines, where the resin flows merge. In particular, when a weld line falls on the tension side of the root, the allowable strain can drop to less than half. (NovaSolver) And the allowable undercut shrinks 10–15% when operating at 60 °C versus the 23 °C reference.
(NovaSolver)
The other guide's numbers agree, and they are worth knowing because they cover the materials Bayer's table does not: acetal homopolymer 5%; nylon 6/6 unreinforced 6–10%; PC unreinforced 4%; PC/ABS alloy 2.5%; PET glass reinforced 1.7%; PPO 4%; PPO glass reinforced 1%. (Qlution Mold) And the rule for the two families: for unreinforced, semi-crystalline materials like PP or nylon, stay within 70% of the yield strain; for amorphous resins like PC or ABS, be much more conservative, use only 50% of the yield strain.
Overlooking this difference is one of the most common reasons why ABS snap-fits crack during the first assembly trial. (Qlution Mold)
The root: where every clip breaks
The formula assumes an ideal beam of constant section, and it assumes the root is a right angle or a smooth radius. On a real moulded part, making the root radius too small creates a stress concentration there, and the local strain can be 1.5 to 3 times the calculated value. Almost all snap-fit failures start at the root radius.
Even with a calculated safety factor of 2, a small radius leaves you with essentially no margin. Aim for a root radius of about 0.5 times the thickness h, and always round it. (NovaSolver)
That is the biggest pitfall: calculating strain while ignoring the root fillet. (NovaSolver) The square inside corner at the root concentrates the strain right where it already peaks. (Loftsmith) Give the inside corner where the arm meets the wall a fillet at least half the arm's thickness: 0.75 mm for a 1.5 mm arm.
It spreads the peak strain over more plastic and costs nothing. (Loftsmith) The other checker's default is R = 0.4 × h, the ratio where the stress concentration curve starts to flatten, with a live stress concentration factor Kt of 1.85 at that ratio. (Xometry Pro) And the peak stress at the weak point is σmax = Kt × Es × ε, where Es is the strain-dependent secant modulus and Kt is the stress concentration factor from the root fillet.
(Xometry Pro)
The secant modulus is the part of the guide most people skip, and it is the part that makes the force numbers honest: the stress–strain behaviour of the material is already outside the linear range, and the ordinary modulus of elasticity must therefore be replaced by the strain-dependent secant modulus. (Bayer) The force numbers in the demo use Bayer's own secant modulus, 1,815 N/mm² at 2% strain for Makrolon PC, not the 2,400 MPa catalogue modulus. (Bayer, via Synectic)
The angles: easy in, hard out, or the other way round
The clip has two faces, and they do different jobs. The lead-in face is the side you push over: about 30°. Steeper, and assembly needs brute force. (Loftsmith) The retaining face is the side that holds: 90° makes a permanent joint that must be released by hand; about 45° makes one that pulls apart with force and can be reused. (Loftsmith)
The force the assembler actually feels depends on the lead angle α and the friction coefficient μ between the two parts:
P = b × h² × E × ε / (6 × L), the force needed to bend the hook to that deflection, where b is the beam width and E is the material's secant modulus at that strain. (Bayer)
W = P × (μ + tanα) / (1 − μ × tanα), the push-on force. (Bayer; NovaSolver)
W' = P × (μ + tanα') / (1 − μ × tanα'), the removal force, where α' is the return (retention) angle. (NovaSolver)
And the removal force is where the joint decides whether it is a serviceable lid or a sealed box: increasing α' raises W' sharply, and once tanα' ≥ 1/μ the denominator becomes zero or negative, giving a self-locking (permanent) joint. For μ = 0.3 this happens above α' ≈ 73°. For parts you open repeatedly, keep α' around 45° so that W' is only slightly larger than the insertion force.
(NovaSolver) For μ = 0.5, PC against PC, from Bayer's Table 3, the self-locking threshold is α' ≥ 63.4°, and at α' = 90° the joint is permanent. A battery cover you open often gets a modest α'; an internal housing clip that may stay permanent gets α' = 90°. (NovaSolver)
And the tug-of-war: "low insertion force = good design" is not true. Focusing only on lowering insertion force W reduces the deflection force P and therefore the retention force as well. A snap-fit is a tug-of-war between ease of assembly (W) and resistance to separation (W' / retention). Decide the angles and dimensions while watching all three: insertion force, retention force, and disassembly. (NovaSolver)
The verdict: usage, not strain
The checker's verdict is not the strain. It is the usage, the induced strain divided by the allowable strain for the material and lifecycle, as a percentage:
Under 75 percent is a comfortable pass. Between 75 and 100 percent is marginal, because material scatter and molding variation can consume the remaining margin. Over 100 percent exceeds the material limit. (Xometry Pro)
And the safety factor is the same number read the other way: the strain safety factor (ε_perm/ε) should exceed 1.5 for injection-molded polycarbonate or 2.0 for acetal copolymer. (NovaSolver) The hub's clip, as drawn, is at 86.8% usage, marginal, with a safety factor of 1.15, below the 1.5 minimum. The demo shows what that means, and what fixes it.
The moulding: the undercut has to come out of the tool
The hook is an undercut. Add a through-slot in the wall directly under the hook so a straight-pull core can form it. (Xometry Pro) Snap-fit undercuts need lifters or sliders unless designed with a pass-through slot, a small design change at DFM stage can remove a costly side action.
(BAOLAIHUA) And the clip lives in the same wall as everything else in Article #16: uniform wall 1.5–3.0 mm, draft 1–3° per side, bosses ≤0.6× nominal wall, and ~70% of first-submission designs contain a defect-causing feature, and a post-tooling modification costs $2,000–$20,000 and 2–6 weeks. (Article #16) The clip that is drawn without a slot is a clip that needs a lifter, and the lifter is a tooling cost that was visible on the CAD screen.
And the printed version, for the prototype that comes before the tool: size a cantilever snap arm so its bending strain stays below what the plastic tolerates, strain = 1.5 × h × y ÷ L². Aim for about 2% in PLA and 3–4% in PETG, and halve those if the arm bends across its layers. (Loftsmith) The design strains for printed arms: PLA about 2% flat, about 1% upright; PETG 3–4% flat, 1.5–2% upright; ABS/ASA 3–4% flat, 1.5–2% upright; nylon 4–6% flat, 2–3% upright.
(Loftsmith) Print the arm lying flat, give its root a fillet at least half its thickness, and use a 30° lead-in. (Loftsmith) The clearances: 0.2–0.3 mm of total clearance on the surfaces that slide past each other, and none on the hook itself, the hook's overlap is the deflection, and it is what holds the joint shut. For most clips, 0.8–1.2 mm of overlap gives a positive click without overdoing it.
(Loftsmith) And the test that costs nothing: cut one clip and its catch out of the design as a small test piece, print it, and snap it ten times. That takes a few grams and a few minutes, and it tells you more than any table. (Loftsmith)
The clip that holds the hub's lid
Example Labs is a made-up company we use for worked examples.
The Example Labs field sensor hub: 100 × 60 × 40 mm, aluminium base, polycarbonate lid (Article #47, the hub's answer is a hybrid). The lid is held by 8 × M4 screws today (Article #49) and opens 25 times over a 10-year life for battery service and calibration (Article #49's labelled assumption). The proposal is to replace four of the screws with snap-fit clips, the lid still opens, but it opens without a tool, and the screws that remain are the ones that hold the seal (Article #52, the gasket needs the fasteners at ≤40 mm spacing, and the hub's 8 screws are exactly at 40 mm).
The clip, as drawn: a cantilever arm on the lid, moulded in polycarbonate, length L = 12 mm, root thickness h = 2.0 mm, undercut y = 1.0 mm (the overlap, inside Loftsmith's 0.8–1.2 mm for a positive click), width b = 8 mm. The dimensions are the article's own worked example, labelled. The material is Makrolon PC: permissible strain 4% for a single assembly at 23 °C (Bayer), and the hub opens 25 times, so the allowance is 60% of that, 2.4%.
Step 1, the strain as drawn. ε = 3 × 2.0 × 1.0 / (2 × 12²) = 6 / 288 = 2.08%. The usage is 2.08 / 2.4 = 86.8%, marginal, in the 75–100% band where material scatter and moulding variation can consume the remaining margin. (Xometry Pro) The safety factor is 2.4 / 2.08 = 1.15, below the 1.5 minimum for injection-moulded polycarbonate. (NovaSolver) And the allowable undercut is y_perm = 2 × 0.024 × 12² / (3 × 2.0) = 1.15 mm, the 1.0 mm undercut is 86.8% of it, and the guidance is to keep the undercut to about 70–80% of y_perm. (NovaSolver) The clip as drawn is marginal on paper, and the paper is the best case, because the paper assumes a smooth root.
Step 2, the root. As drawn, the root is a square corner, no fillet. The local strain at a sharp root can be 1.5 to 3 times the calculated value (NovaSolver): 2.08% × 1.5 = 3.13%, and × 3 = 6.25%. Both are over the 2.4% allowance, and the 6.25% is over Bayer's 4% single-assembly limit for PC itself. The clip as drawn does not just fail the frequent-separation allowance at the root, at the worst end of the range it fails the material's own limit. Almost all snap-fit failures start at the root radius, and even with a calculated safety factor of 2, a small radius leaves you with essentially no margin. (NovaSolver) The fix is a fillet of about 0.5 × h, R = 1.0 mm for a 2.0 mm arm, and the checker's screening number at R = 0.4h is Kt = 1.85. (Xometry Pro)
Step 3, the taper. Taper the beam to half thickness at the hook (h_tip = 1.0 mm). Bayer's tapered formula: ε = y × h / (1.09 × L²) = 1.0 × 2.0 / (1.09 × 144) = 1.27%. The usage falls to 1.27 / 2.4 = 53.1%, a comfortable pass, under 75%. The safety factor rises to 2.4 / 1.27 = 1.88, above the 1.5 minimum. The taper alone takes the clip from marginal to comfortable, and it uses less material doing it, Bayer's optimal design uses 17% less material and 46% less strain than the constant-section arm.
Step 4, the length. Make the beam 1.4× longer: 12 mm → 16.8 mm. Strain is inversely proportional to the square of length, so making L 1.4× longer almost halves the strain. (NovaSolver) The ratio is (12 / 16.8)² = 0.51. The usage is 0.65 / 2.4 = 27.1%. The safety factor is 2.4 / 0.65 = 3.69. The allowable undercut is now 3.69 mm, and the 1.0 mm undercut is 27.1% of it. And even without the taper, the longer constant-section arm is 1.06% strain, 44.3% usage, safety 2.26, a comfortable pass on its own. The length is the strongest lever, and the taper is the cheapest: together they take the clip from 86.8% usage to 27.1%.
Step 5, the forces. The secant modulus is Bayer's own: Es = 1,815 N/mm² at 2% strain for Makrolon PC. P = b × h² × Es × ε / (6 × L). As drawn (ε = 2.08%): P = 8 × 4 × 1,815 × 0.0208 / 72 = 16.8 N. Tapered at 12 mm (ε = 1.27%): P = 10.3 N. Tapered and lengthened to 16.8 mm (ε = 0.65%): P = 3.7 N. The push-on force, with α = 30° and μ = 0.5 (PC against PC, Bayer Table 3): W = P × (0.5 + tan30°) / (1 − 0.5 × tan30°) = P × 1.51. As drawn: W = 25.5 N. Tapered: 15.6 N. Tapered and lengthened: 5.7 N. The removal force, with α' = 45°: W' = P × (0.5 + 1.0) / (1 − 0.5) = 3 × P. As drawn: 50.4 N. Tapered: 30.8 N. Tapered and lengthened: 11.2 N. And the self-locking line: for μ = 0.5, tanα' ≥ 2, so α' ≥ 63.4° locks the joint, at α' = 90° the lid never opens again. The hub opens 25 times, so α' = 45° is the angle, and the removal force is three times the deflection force, which is the tug-of-war: the tapered and lengthened clip is easy to assemble at 5.7 N and easy to remove at 11.2 N, and if the lid needs to hold harder, the answer is more clips, not a steeper angle, because a steeper angle locks the lid shut.
Step 6, the temperature. The hub's inside reaches 85 °C (Articles #50 and #52, 60 °C ambient plus the 25 °C self-heating rise). The allowable undercut shrinks 10–15% when operating at 60 °C versus the 23 °C reference (NovaSolver), so the 2.4% allowance is effectively 2.04–2.16% at 60 °C. The clip as drawn, at 2.08%, sits inside that shrunken window, at the edge, with no margin for the root fillet, the weld line, or the twenty-fifth opening. The tapered and lengthened clip, at 0.65%, sits at a third of the shrunken allowance. The temperature is not a separate problem. It is the same problem, with the allowance smaller.
Step 7, the cycles and the alternative. The hub opens 25 times, and a prototype is fine after one or two assemblies, but once a customer engages a mass-produced part dozens of times, an over-strained hook fatigues and breaks. (NovaSolver) Twenty-five is dozens. The clip as drawn, at 86.8% usage, is the prototype that is fine after one or two assemblies and breaks on the tenth opening in the field. And the alternative is Article #49's insert: the joint the clip replaces, needed when a joint is opened and re-fastened more than 5–10 times. Eight M4 inserts at ≈$0.09 each is $0.72 a hub, and the insert holds 1,180 N on pass 1 and 840 N on pass 50, while the screw in bare plastic decays from 380 N to 152 N by pass 8 and strips by pass 11. (Article #49) The clip costs nothing and holds 30.8 N of removal force when it is tapered; the insert costs $0.72 a hub and does not fatigue over 25 openings. The decision is the clip done right, tapered, 16.8 mm, R = 1.0 mm, α' = 45°, or the insert, if the arm cannot be made long enough. The clip done wrong is the third option, and it is the one that snaps on the tenth opening.
Step 8, the cost of finding it late. A post-tooling modification costs $2,000–$20,000 and 2–6 weeks (Article #16), and the snap-fit undercut needs a lifter or a slider unless the wall has a pass-through slot under the hook, a small design change at DFM stage that can remove a costly side action. (BAOLAIHUA) And the field failure is Article #40's: 4 of 10 hubs failed on days 19–23, 40% of the fleet, with the cost escalation 1× design / 5× prototype / 20× pilot / 50× production / 100×+ field. The clip that snaps in the field is a lid that does not close, on a product that is recalibrated, not replaced, and the recalibration cannot happen if the lid will not stay shut.
The decision test
Five questions. If the answer to any one of them is no, the clip is a coin flip, and the decision is to size the clip, and to test one arm before the whole box.
- What is the strain, and what is the allowance? Not the force, the strain, from ε = 3·h·y / (2·L²), against the material's permissible strain for the lifecycle: 4% for PC single assembly, 2.4% for a joint opened and closed often. If the usage is over 75%, the clip is marginal; over 100%, it exceeds the material limit.
- Is the beam tapered, and is it long enough? The taper to half thickness at the hook buys 63% more deflection for the same strain. Making the beam 1.4× longer almost halves the strain. If the arm is short and constant-section, the strain is the highest it can be for the space it occupies.
- What is the root radius? If the root is a square corner, the local strain is 1.5–3× the calculated value, and the calculated safety factor is fiction. The fillet should be about 0.5 × h, and it costs nothing.
- What are the two angles, and is the joint separable? The lead-in is about 30°; the retaining face is 45° for a joint that opens again and 90° for a joint that never opens. For μ = 0.5, the self-locking threshold is α' ≥ 63.4°, past that, the lid is sealed, and the serviceable product is not serviceable.
- Can the undercut come out of the tool? The hook is an undercut, and the undercut needs a lifter, a slider, or a pass-through slot in the wall directly under the hook. If the wall has no slot, the tooling cost was visible on the CAD screen, and the clip that needs a lifter is a clip that should have had a slot.
Objections
"The clip is fine, it snapped together in the prototype." A prototype is fine after one or two assemblies, but once a customer engages a mass-produced part dozens of times, an over-strained hook fatigues and breaks. (NovaSolver) The prototype was snapped twice. The hub opens 25 times. The prototype proves the clip can be assembled once. It does not prove the clip can be opened twenty-five times, and the strain number is the proof, not the prototype.
"We'll just make the clip thicker, thicker is stronger." A thicker arm, which feels like the obvious fix, makes things worse. (Loftsmith) The strain grows in direct proportion to the thickness, so the thicker arm bends less far for the same force, and the strain at the root is higher for the same overlap. If an arm needs to be stiffer, make it wider, not thicker. (Loftsmith) The fix is longer and thinner, not thicker, and the taper, which makes the arm thinner at the tip, is the fix that also uses less material.
"We'll use glass-filled PC, it's stronger." Glass makes the plastic stiffer and stronger, and it makes the allowable strain smaller: PC with 20% glass fibre tolerates 2.0%, half of unfilled PC's 4%. (Bayer) The stiffer material is the one that snaps first, because the strain is what breaks the clip, not the force. The glass-filled clip is the clip that breaks at half the deflection of the unfilled one, and the hub's lid opens 25 times, so the allowance is already 60% of the single-assembly value, and the glass takes half of that.
"The clip is cheaper than screws, just use more of them." More clips is the right answer for retention, for a car interior panel, engineers use long, thin, low-strain beams and make up the total holding force with the number of hooks (NovaSolver), but more clips of the same bad design is more clips that snap. The number of hooks is the lever for the holding force. The strain is the lever for whether any of them survive. Fix the strain first, then count the hooks.
"We'll just use screws, clips are unreliable." Screws are reliable, and they cost $0.72 a hub in inserts, and they need a tool, and the hub opens 25 times, which is why the insert exists, and why the insert is the fallback. But the clip done right is free, needs no tool, and holds 30.8 N of removal force at 53.1% usage with a safety factor of 1.88. The clip is not unreliable. The clip sized by feel is unreliable. The clip sized by strain is a number, and the number passes.
Sources
- Bayer, *Snap-Fit Joints for Plastics, A Design Guide* (via the guide's own PDF), "Snap joints are a very simple, economical and rapid way of joining two different components. All types of snap joints have in common the principle that a protruding part of one component, e.g., a hook, stud or bead is deflected briefly during the joining operation and catches in a depression (undercut) within the mating component. After the joining operation, the snapfit features should return to a stress-free condition."
What we used
"The joint may be separable or inseparable depending on the shape of the undercut." "The cantilever arm of unsatisfactory design has a constant cross section. The non-uniform distribution of lines (fringes) indicates a very uneven strain in the outer fibers. This design uses 17% more material and exhibits 46% higher strain than the optimal design. Bottom: The thickness of the optimal snapfit arm decreases linearly to 30% of the original cross-sectional area. The strain in the outer fibers is uniform throughout the length of the cantilever." "The stress-strain behavior of the material is already outside the linear range and the ordinary modulus of elasticity must therefore be replaced by the strain dependent secant modulus." Permissible short-term strain at 23 °C, single assembly: Apec High Heat PC 4%; Bayblend PC/ABS 2.5%; Makroblend Blends Polycarbonate 3.5%; Makrolon PC 4%; Makrolon PC (10% glass) 2.2%; Makrolon PC (20% glass) 2.0%. "For a joint that is opened and closed often, use about 60% of the single-assembly value." Worked example (Makrolon PC, L = 19 mm, b = 9.5 mm, y = 2.4 mm, lead angle 30°, permissible strain 4%, strain used = 1/2 pm = 2%): thickness h = 3.28 mm; deflection force P = 32.5 N (7.3 lb); secant modulus Es = 1,815 N/mm² (264,000 psi); friction coefficient PC against PC μ = 0.50
- Synectic, "Snap Fit Design, How To Design A Snap Fit Joint"
What we used
"Maximum snap deflection is the displacement of the beam tip required to engage or disengage the joint with the mating part without failure. It must remain below the maximum permissible strain of the plastic material to avoid permanent deformation or fracture." "Y = 0.67 ∗ ( (Ɛ ∗ L²)/ H)", "L = Snap Beam Length; H = Snap Beam Thickness at the root; Ɛ = Maximum Permissible Strain… For polycarbonate (PC), typical values range between 4% and 8%." "P = (B ∗ H² ∗ E ∗ Ɛ)/ 6L", "for example, if your material is polycarbonate, the modulus of elasticity would be ~2,400 MPa (350,000 psi)." "Using a tapered cantilever beam (thicker at the base) will provide a more evenly distributed stress along the beam to avoid concentration points that could lead to failure, as well as increase the snap deflection if desired. The rule of thumb is to reduce the thickness at the end of the beam to one-half the value at the root. In this case, you can calculate the snap deflection similarly by replacing the 0.67 value with 1.09 in the formula above." Worked example (injection moulded spring finger, L = 8 mm, H = 1.5 mm, B = 3 mm, PC, E = 2,350 MPa, ε = 4%): "Y = 0.67 ∗ ( (0.04 ∗ 8²)/ 1.5) = 1.14 mm. Applying a 10% safety factor: Y = 1.0 mm."
- NovaSolver, "Cantilever Snap-Fit Design Simulator, Strain & Insertion Force"
What we used
"For a cantilever beam of constant rectangular section, the maximum strain at the root for a tip deflection y is ε = 3hy / (2L²), where h is the root thickness and L is the beam length. Strain is proportional to thickness h and inversely proportional to the square of length L, so the basic rule for bending a hook far without breaking it is to make the beam thinner and longer." "The allowable undercut is y_perm = 2·ε_perm·L² / (3h). ε_perm is the allowable strain for a single assembly, roughly 2% for ABS and 6% for POM. For L=20mm, h=2mm and ε_perm=2%, y_perm ≈ 2.7mm. Exceeding it causes yielding, stress-whitening or cracking at the root during assembly. In practice, keep the undercut to about 70-80% of the calculated y_perm." "The most effective move is to make the beam longer. Strain is inversely proportional to the square of length L, so making L 1.4× longer almost halves the strain." "A classic field failure happens here: a prototype is fine after one or two assemblies, but once a customer engages a mass-produced part dozens of times, an over-strained hook fatigues and breaks." "For repeated use, aim for less than half the single-assembly value." "The biggest pitfall is calculating strain while ignoring the root fillet (radius). The formula ε = 3hy/(2L²) used here is for an ideal beam of constant section and assumes the root is a right angle or a smooth radius. On a real moulded part, making the root radius too small creates a stress concentration there, and the local strain can be 1.5 to 3 times the calculated value. Almost all snap-fit failures start at the root radius. Even with a calculated safety factor of 2, a small radius leaves you with essentially no margin. Aim for a root radius of about 0.5 times the thickness h, and always round it." "Verify strain safety factor (ε_perm/ε) exceeds 1.5 for injection-molded polycarbonate or 2.0 for acetal copolymer." "Acetal copolymer (Delrin) tolerates higher strain cycles (50,000+) versus polycarbonate (10,000) before stress-whitening." "Allowable undercut y_perm shrinks 10–15% when operating at 60 °C versus 23 °C reference." "Insertion force W is proportional to the deflection force P: W = P·(μ+tanα)/(1−μ·tanα). Three ways to lower it: (1) reduce the lead angle α (around 30° is typical), (2) reduce friction μ at the contact face, and (3) make the beam longer and thinner to cut P itself." "Removal force W' depends on the return (retention) angle α': W' = P·(μ+tanα')/(1−μ·tanα'). Increasing α' raises W' sharply, and once tanα' ≥ 1/μ the denominator becomes zero or negative, giving a self-locking (permanent) joint. For μ=0.3 this happens above α' ≈ 73°. For parts you open repeatedly, keep α' around 45° so that W' is only slightly larger than the insertion force." "For a car interior panel, for example, engineers use long, thin, low-strain beams and make up the total holding force with the number of hooks." "A battery cover you open often gets a modest α'; an internal housing clip that may stay permanent gets α' = 90°." "The allowable strain differs greatly between 'single assembly' and 'repeated use'." "The preset values in this tool are representative figures for room temperature and a single assembly. The real ε_perm varies strongly with temperature (brittle when cold), strain rate (more brittle the faster you insert), number of cycles, glass-fiber content and the position of weld lines (where resin flows merge). In particular, when a weld line falls on the tension side of the root, the allowable strain can drop to less than half."
- Xometry Pro, "Snap-fit Design Checker"
What we used
"Cantilever snap-fit strain calculator for injection-molded thermoplastics. Check strain usage, assembly force, and moldability against the Bayer and BASF design guide methodology across six materials." "ε = 3 × Y × h / (2 × L²)", "The classical cantilever beam relation from the Bayer and BASF snap-fit design guides. Strain falls with the square of arm length, which is why lengthening the arm is the most effective fix." "usage = ε / εallow × 100%, Induced strain divided by the allowable strain for your material and lifecycle. Under 75 percent is a comfortable pass. Between 75 and 100 percent is marginal, because material scatter and molding variation can consume the remaining margin. Over 100 percent exceeds the material limit." "σmax = Kt × Es × ε. Es is the strain-dependent secant modulus, Kt the stress concentration factor from the root fillet." "Root fillet radius (R): Default R = 0.4 × h, the ratio where the stress concentration curve starts to flatten. Current Kt: 1.85." "The hook is an undercut. Add a through-slot in the wall directly under the hook so a straight-pull core can form it." "Data based on the Bayer and BASF snap-fit design guide cantilever beam methodology, with a stress concentration factor derived from a smoothed engineering approximation pending full digitization of a published Kt chart, and material properties from typical published ranges. Values are for reference and preliminary screening only."
- Loftsmith, "Snap Fit Design for 3D Printing (With Math)"
What we used
"Size a cantilever snap arm so its bending strain stays below what the plastic tolerates: strain = 1.5 × thickness × deflection ÷ length². Aim for about 2% in PLA and 3–4% in PETG, and halve those if the arm bends across its layers. A 1.5 mm thick PLA arm that has to move 1 mm needs to be about 11 mm long; in PETG, about 8 mm." "The clip didn't break because PLA is weak. It broke because the design asked it to bend further than PLA can stretch, and the place it broke, right at the root, is the exact spot a single formula predicts." "Strain grows with thickness and with overlap, and it falls with the *square* of length. So a longer arm is by far the strongest fix: make it 40% longer and the strain halves. A thicker arm, which feels like the obvious fix, makes things worse." "minimum length = √(1.5 × h × y ÷ allowed strain)." Design strain for printed snap arms: PLA about 2% flat, about 1% upright; PETG 3–4% flat, about 1.5–2% upright; ABS/ASA 3–4% flat, about 1.5–2% upright; nylon (PA) 4–6% flat, about 2–3% upright. Worked examples: "The arm is 1.5 mm thick and the hook overlaps its ledge by 1 mm. In PLA, printed flat, we allow 2%: L = √(1.5 × 1.5 × 1 ÷ 0.02) = √112.5 ≈ 10.6 mm. So the arm needs to be at least 11 mm long. The instinctive design, a stubby 5 mm arm, would see about 9% strain, four times what PLA tolerates." "PETG tolerates more. At 3.5%, the same arm needs √(2.25 ÷ 0.035) ≈ 8.0 mm." "Give the inside corner where the arm meets the wall a fillet at least half the arm's thickness: 0.75 mm for a 1.5 mm arm." "Lead-in (the side you push over): about 30°. Steeper, and assembly needs brute force. Retaining face (the side that holds): 90° makes a permanent joint that must be released by hand. About 45° makes one that pulls apart with force and can be reused." "Two kinds of gap live in a snap fit. The surfaces that slide past each other, the walls of the lid and box, want a normal sliding clearance of 0.2–0.3 mm total so nothing binds. The *hook overlap* is not a clearance at all." "For most clips, 0.8–1.2 mm of overlap gives a positive click without overdoing it." "Cut one clip and its catch out of the design as a small test piece, print it in the filament you'll use, and snap it ten times."
- Qlution Mold, "Snap Fit Design Guide & Free Calculator"
What we used
"For unreinforced, semi-crystalline materials (like PP or Nylon), the rule of thumb is to stay within 70% of the yield strain. These materials have a more forgiving stress/strain curve, allowing them to handle the brief stress of the 'snap' even when pushed close to their limit. However, for amorphous resins (like PC or ABS), we have to be much more conservative. Since these materials lack a well-defined yield point, we generally recommend using only 50% of the yield strain." Permissible short-term strain: acetal homopolymer 5%; nylon 6/6 unreinforced 6–10%; polycarbonate unreinforced 4%; PC/ABS alloy 2.5%; PET glass reinforced 1.7%; PPO 4%; PPO glass reinforced 1%. "Go with a tapered design. By thinning the beam toward the tip, you distribute that stress much more evenly. It's a simple change that can boost your permissible deflection by 60% without adding a penny to the part cost."
- BAOLAIHUA, "Living Hinge & Snap Fit Design Guide"
What we used
"The cantilever snap fit is the most common assembly feature we mold, and the most commonly broken. The governing relationship: maximum strain at the clip root must stay below the material's allowable strain at engagement." "Allowable strain: ~2% for filled/brittle resins (PC, glass-filled PA), 4-6% for ABS, 8-10%+ for unfilled PP/PE/POM, check the datasheet, temperature matters." "Add a fillet at the root (minimum 0.5x thickness), sharp roots are where every broken clip starts." "Keep assembly angle 25-35° and retention angle per service need; 90° retention faces make parts non-demountable." "Check moldability: snap fit undercuts need lifters or sliders unless designed with pass-through slots, a small design change at DFM stage can remove a costly side action."
- Demo arithmetic verified in Python before writing: Bayer's worked example reproduces exactly from the guide's own numbers
What we used
H = 1.09 × 0.02 × 19² / 2.4 = 3.28 mm; P = 9.5 × 3.28² × 1,815 × 0.02 / (6 × 19) = 32.5 N; W at α = 30°, μ = 0.5 = 32.5 × 1.51 = 49.2 N. Hub clip as drawn (L = 12, h = 2.0, y = 1.0, b = 8, PC, allowance 2.4%): ε = 3 × 2.0 × 1.0 / (2 × 144) = 2.08%; usage 86.8%; safety 1.15; y_perm = 2 × 0.024 × 144 / 6 = 1.15 mm. Sharp root: 2.08% × 1.5 = 3.13%, × 3 = 6.25%. Tapered (1.09 factor): ε = 1.0 × 2.0 / (1.09 × 144) = 1.27%; usage 53.1%; safety 1.88; y_perm = 1.88 mm. Tapered and lengthened (L = 16.8): ε = 0.65%; usage 27.1%; safety 3.69; y_perm = 3.69 mm. Constant-section at L = 16.8: ε = 1.06%; usage 44.3%; safety 2.26. Length ratio (12/16.8)² = 0.51. Taper gain 1.09/0.67 = 1.63 (+62.7% deflection). Forces (Es = 1,815, μ = 0.5): P as drawn 16.8 N, tapered 10.3 N, tapered + lengthened 3.7 N; W (30°) = P × 1.51: 25.5 N, 15.6 N, 5.7 N; W' (45°) = 3 × P: 50.4 N, 30.8 N, 11.2 N. Self-locking: tanα' ≥ 1/0.5 = 2 → α' ≥ 63.4° (for μ = 0.3, α' ≈ 73.3°). Root screening: 1.27% × Kt 1.85 = 2.36% vs the 2.4% allowance; σmax = 1.85 × 1,815 × 0.0127 = 42.8 MPa (screening number, not certified). Temperature: 2.4% × 0.90 = 2.16%, × 0.85 = 2.04% at 60 °C. Loftsmith cross-checks: PLA L = √(1.5 × 1.5 × 1 / 0.02) = 10.6 mm; stubby 5 mm arm = 9.0% strain; PETG L = √(2.25 / 0.035) = 8.0 mm. Inserts: 8 × $0.09 = $0.72 per hub