E-scooter frame and fork engineering: loads, alloys, fatigue

Scootify 18 min read

The article «Frame, handlebar, and folding mechanism of an e-scooter» describes the typology of the load-bearing assembly (5 components — deck, stem, hinge, handlebar, fork), the four folding-mechanism types (lever-latch / multi-point hinge / twist-and-fold / trigger-pin), historical failure modes (the official Xiaomi M365 recall of 10 257 units in 2019 due to a screw backing out of the folding apparatus; deck cracks in early Lime/Okai sharing models), and a market matrix of 10 models with their frame materials. This article is an engineering deep-dive into the physics of the structure itself: why a 2-mm wall in a 50-mm-diameter tube delivers 8× the bending stiffness of the same 2-mm wall in a 25-mm tube; why 6061-T6 loses half of its yield strength in the heat-affected zone of a weld (276 MPa → 138 MPa) and why designers add welded gussets to compensate; why 7075, with a yield strength of 503 MPa — almost twice that of 6061 — is not used as the basis of a fully welded frame; why aluminum, unlike steel, has no endurance limit and its fatigue curve keeps falling forever, only with different slopes; and why the Xiaomi M365 hook broke precisely at the weld toe in a stress-concentration zone with K_f ≈ 4–6, where high-cycle fatigue (HCF > 10⁴ cycles) accumulated damage per Miner’s linear rule to the critical D = 1. This is the eighth engineering-axis deep-dive (after protective gear engineering, lithium-ion battery engineering, brake system engineering, motor and controller engineering, suspension engineering, tire engineering, and lighting engineering) — it adds the structural axis as the integrator of every other load: everything the motor creates, the brake dissipates, the suspension isolates, and the tire transmits to the road passes through the frame.

Prerequisite — understanding material and folding-mechanism types, post-crash inspection, and pre-ride safety check.

1. Why the frame is a structural integrator, not «a rigid beam»

A bicycle or scooter frame is a 3D space frame under combined loading. One static scenario — the rider stands on the deck at rest — is a balanced-moment problem about the front-wheel contact point. The dynamic reality is more complex: a curb impact of 5 cm at 25 km/h drives an impulse of 1.5–2 kN through the front wheel over 5 ms, multiplying the static load by a factor of 2–3 depending on how forces decompose along the stem tube. Under a front-brake deceleration of 0.8 g on a 100-kg scooter+rider system, the center of mass shifts forward and the normal load on the front wheel becomes N_front = m·g·(1 + 0.8·h/L) with h/L ≈ 0.5 → the normal force rises by 40 %, and that force is transmitted through the fork and headset into the stem tube as combined bending + axial + torsion simultaneously.

Three fundamental loading modes per Euler-Bernoulli beam theory:

1) Bending from side loads on the handlebar (cornering, wind) and vertical wheel impacts:

σ = M · c / I

where M is the bending moment (N·m), c is the distance from the neutral axis to the outer fiber (m), and I is the section’s second moment of area (m⁴). For a round tube I = π(D⁴ − d⁴) / 64 — a quartic function of diameter. This is the fundamental reason a LARGE diameter with a thin wall is always stiffer than a small diameter with a thick wall: doubling the diameter at the same wall thickness multiplies I roughly eightfold.

2) Torsion from asymmetric handlebar loads (one hand stronger than the other), during cornering:

τ = T · r / J

where T is torque (N·m), r is radius from center (m), J = π(D⁴ − d⁴)/32 is the polar second moment of area (m⁴ — same quartic law).

3) Axial (compression/tension) from the rider’s vertical weight through the stem:

σ = F / A

where A = π(D² − d²)/4 is the cross-sectional area.

All three superpose simultaneously. The yield criterion for 3D stress state per von Mises:

σ_v = √(σ² + 3τ²) ≤ σ_y

— when σ_v reaches the material’s yield strength, plastic deformation begins and irreversible defect accumulation starts. In reality the calculation is even more complex — at a folding hinge the geometric stress concentration K_t is added (see § 6), and in a weld HAZ the yield strength is already reduced (see § 4). So designers don’t simply compute σ_v under nominal load; they multiply by K_f · safety_factor / knockdown_HAZ — typically a 4–6× margin above the simple balanced-load calculation.

2. Beam mechanics: why wall thickness loses to diameter in the thin limit

Treat a scooter stem as a cantilever beam fixed at the deck and loaded by a force F at the top (at the handlebar). The bending moment at the fixity:

M_max = F · L

where L is the stem length (typically 0.8–1.2 m on adult scooters). Tip deflection:

δ = F · L³ / (3 · E · I)

Here we already see that bending stiffness 3EI/L³ is proportional to E·I: only E (the material’s Young’s modulus) multiplies I (the geometry of the section). Two strategies for raising stiffness:

  • Increase E — switch from aluminum (E ≈ 70 GPa) to steel (E ≈ 205 GPa) → 3× stiffer at the same geometry, but 2.9× heavier.
  • Increase I — thicken the tube or grow its diameter.

For a round tube, compare four D × t variants with the same mass (same weight):

Outer DWall tArea A (mm²)I (mm⁴)Stiffness EI for 6061
25 mm4.2 mm~27314 6001.01 × 10⁶ N·m²
32 mm2.9 mm~26527 9001.92 × 10⁶
40 mm2.3 mm~27251 1003.52 × 10⁶
50 mm1.8 mm~27281 0005.58 × 10⁶

— a fivefold stiffness difference at the same weight, purely from distributing mass further from the neutral axis. This is a universal principle: under a weight constraint, a large diameter + thinner wall always yields higher EI, until the wall becomes critically thin (local wall crippling/buckling). The empirical rule for aluminum frames — D / t ≥ 25 approaches the stability limit in bending regions, so manufacturers don’t push beyond D / t ≈ 22 (Xiaomi M365 — 32 mm / 2.5 mm = 12.8; NAMI Burn-E — 50 mm / 3 mm = 16.7; Wolf King GT — 56 mm / 3.5 mm = 16).

That explains why modern top off-road models have 50–60 mm stem diameters, while budget urban models stay at 30–40 mm: load F × impact velocity scales differently → off-road requires roughly 5× the stiffness of a city scooter.

3. Materials: Young’s modulus, yield strength, specific quantities, and Ashby selection

Engineers choose materials not by absolute strength but by specific quantities — properties normalized by density:

  • E / ρ — specific stiffness (m²/s² or MJ/kg), for stiffness-limited problems (truss, beam without geometry loss)
  • σ_y / ρ — specific strength, for strength-limited problems (a bar in tension)

Michael Ashby formalized this in the classic «Materials Selection in Mechanical Design» (Butterworth-Heinemann, 4th edition 2010) through two-axis charts of log(E) vs log(ρ) and log(σ_y) vs log(ρ). On these charts, straight lines of slope +1 connect materials with equal E/ρ (for tension), +2 — equal for beam bending, +3 — equal for plate bending.

Summary of the main candidates for PLEV frames:

MaterialE (GPa)σ_y (MPa)σ_UTS (MPa)ρ (g/cm³)E/ρ (GPa·cm³/g)σ_y/ρ (MPa·cm³/g)Weldability
Al 6061-T668.92763102.7025.5102excellent (GTAW AC)
Al 7005-T6722903502.7825.9104good (auto-age post-weld)
Al 7075-T671.75035722.8125.5179poor (hot cracking)
Al 6082-T6702603102.7025.996excellent
Steel 4130 Cr-Mo2054606707.8526.159excellent (GMAW/GTAW)
Mg AZ91D (cast)451602401.8124.988requires SF₆ shield
CF UD T700S (along fiber)135—4 900 (σ_t)1.5587.11645molded layup, no welding

Key observations:

  1. All aluminum alloys deliver essentially the same specific stiffness E/ρ ≈ 25.5. This is not a coincidence: E is set by interatomic forces driven by the fcc crystal lattice, and alloying elements Cu/Mg/Zn/Si change σ_y via precipitation hardening (phases like Mg₂Si in 6xxx or MgZn₂ in 7xxx), but don’t change E itself. So there is no reason to pick 7075 over 6061 for stiffness — only for strength.

  2. Steel 4130 has the same specific stiffness as aluminum but worse specific strength (59 vs 102 in 6061 and 179 in 7075). Counter-intuitive but true: steel is 3× heavier and 3× stiffer → identical specifically. Steel only wins on frames where absolute stiffness is critical within a constrained envelope (BMX, downhill MTB — where geometry is regulated, not weight).

  3. Unidirectional T700S carbon along the fiber — 1645 specific strength — an order of magnitude better than metals. But only along the fiber direction: transverse, carbon behaves like a brittle polymer with σ_t ≈ 50 MPa. So a real carbon frame demands 12–20 plies of tape layup at different fiber orientations, which wipes out most of the advantage (effective specific strength of a quasi-isotropic layup 400–600 MPa·cm³/g — still 3–4× better than metals, but no longer by an order of magnitude).

  4. Magnesium is the lightest, but σ_y/ρ matches 6082. The win is modest, especially in AZ91D cast components (sand- or die-cast), which are brittle and require anti-corrosion coating. Magnesium is used in cheap compact models (Inmotion L8/L9) mostly for the marketing pitch «lightest», not real engineering benefit.

Why 6061-T6 is the universal default on consumer scooters: an optimum of (σ_y/ρ + weldability + corrosion resistance + price-per-tonne + market availability). 7075 is twice as strong, but does not weld in thin-wall frames — it’s used only as a CNC-machined part at high-load points (stem hook, fork crown reinforcement in Mantis King GT and Dualtron Storm) bolted onto a 6061 chassis. 7005 is closer to 7075 in strength and welds better (precipitation hardening re-activates naturally after welding, without post-weld heat treatment), so it’s used in Trek/Specialized/Cannondale sport bikes — but rarely seen in scooters due to higher cost.

4. Welding metallurgy: why HAZ halves the yield strength

Aluminum 6061-T6 in the as-supplied condition has a yield strength of 276 MPa. In the heat-affected zone (HAZ) just beside a weld, the yield strength drops to ~138 MPa — exactly half. This is not a process defect; it’s a fundamental metallurgical inevitability.

Mechanism. 6061 is a precipitation-hardened alloy (often confused with «heat-treated»; the proper name is aging). The initial T6 condition is produced as follows:

  1. Solution heat treatment — heating to 530 °C, where all alloying elements (Mg, Si) dissolve into an α-Al solid solution.
  2. Quenching — rapid water cooling → a supersaturated solid solution (metastable state).
  3. Artificial aging — hold at 175 °C × 8 hours → nanoscale β'-Mg₂Si precipitates (10–100 nm) form and block dislocation movement in the crystal lattice, giving high yield strength.

During welding, the temperature in the HAZ reaches 300–500 °C, which destroys the β'-Mg₂Si precipitates — they redissolve into α-Al solid solution (overaging → solid solution → near-annealed condition). Post-weld cooling is much slower than the initial quench, so a supersaturated state doesn’t reform — yield strength remains at the T4 level (138 MPa) or even lower (annealed O-temper, ~55 MPa in the crystallized core).

Per AWS D1.2:2014 «Structural Welding Code — Aluminum» and Aluminum Association ADM (Aluminum Design Manual 2020), engineers are advised to design welded joints at 50 % of the base material’s yield strength. That’s a knockdown factor of 0.5 — a critical parameter in structural calculations.

How this is managed:

  1. Filler metal selection. Three main filler wires for 6061:

    • ER4043 (Al-5Si, melting point 575 °C): low cracking susceptibility (silicon lowers the coefficient of thermal expansion), but not aging-responsive — strength stays at 95–125 MPa. Default for cosmetic/non-critical welds.
    • ER5356 (Al-5Mg): higher strength 165–200 MPa with post-weld natural aging (Mg continues to precipitate over time). Default for structural frames of bicycle and scooter type.
    • ER4047 (Al-12Si): a brazing/casting filler with no aging response, the highest fluidity but the lowest strength. Not used in frames.

  2. Welded gussets — additional 6061-T6 plates at high-stress points, welded over the main tubes. A gusset increases local I and reduces nominal stress in the HAZ to a level where the 50 % knockdown becomes tolerable. Visible to the naked eye on quality scooters (Apollo Phantom V3, NAMI Burn-E 2 — around the stem mount and at the folding zone).

  3. Post-weld heat treatment (PWHT) — re-solution treatment at 530 °C × 30 min + quench + artificial aging at 175 °C × 8 hours — restores T6 strength across the entire frame. This is aerospace-grade processing; it significantly increases manufacturing cost and warps geometry (thermal distortion). Series-production scooters don’t use PWHT; only custom builds (Deyman, Magnumix) occasionally do.

Why 7075 is unweldable in frames. 7075-T6 has the Zn-Mg-Cu precipitate η-MgZn₂. Unlike Mg₂Si in 6xxx, this precipitate doesn’t fully recover even with PWHT — Cu in the strain field provokes hot cracking in the heat-affected zone (the Al-Zn-Mg-Cu solidification range is too wide). 7075 in thin-wall frames cracks rapidly in the HAZ. So 7075 is used only as a CNC-machined solid part (such as the Mantis King stem hook), bolted onto a 6061 frame mechanically, not welded.

6082-T6 — practically identical to 6061 in weldability and aging response, but has slightly better corrosion resistance through lower Cu content (≤0.1 % in 6082 vs 0.15–0.40 % in 6061). NAMI Burn-E and Apollo Air use 6082 mostly as an «aerospace-grade» marketing label; the engineering difference is minimal — 5–10 MPa in σ_y and slightly lower pitting corrosion susceptibility.

Bulk corrosion resistance is settled not at alloy-chemistry level but at surface-engineering level — anodize Type II / Type III, conversion coatings, powder coating, and galvanic isolation at the interface with steel fasteners. The details are in Aluminium surface treatment engineering: MIL-PRF-8625F + ISO 7599:2018 standards, the Cr(VI) → Cr(III) regulatory transition under RoHS + REACH Annex XIV, salt-spray testing ASTM B117 / ISO 9227, and the critical fatigue debit of 20-50 % from Type III hardcoat on 7075-T6 (Cirik & Genel 2008) that forces the design engineer to derate cyclic stress allowables in hardcoat zones.

5. Fatigue: why aluminum has no endurance limit

Fatigue failure is the accumulation of microscopic defects (slip bands → micro-cracks → macro-crack → fracture) under cyclic loads with amplitudes below the yield strength. The engineering curve is described by the Basquin equation (Basquin O. H., 1910):

σ_a = σ'_f · (2N_f)^b

where σ_a is the stress amplitude (half-range), σ'_f is the fatigue strength coefficient (a characteristic material stress), N_f is the number of cycles to failure, and b is the fatigue exponent (typically from −0.05 to −0.12 for metals). A plot of log(σ_a) vs log(N_f) is a straight line of slope b.

For 6061-T6: σ'_f ≈ 478 MPa, b ≈ −0.083 per ASM Handbook Vol. 19 (Fatigue and Fracture). For 4130 Cr-Mo: σ'_f ≈ 950 MPa, b ≈ −0.076. For UD T700 carbon: σ'_f ≈ 2 200 MPa (anisotropic), b ≈ −0.06.

This describes HCF — high-cycle fatigue, N_f > 10⁴ — the regime of normal operational frame loading (vibration from the road, cyclic loads from each step in the ride rhythm). For LCF — low-cycle fatigue, N_f < 10⁴ the Coffin-Manson equation with plastic-strain amplitude is used — less relevant for scooter frames, where the primary regime is HCF.

The critical metal-vs-metal difference:

  • Steels 4130 / 4140 / Cr-Mo have an endurance limit — a horizontal S-N asymptote at N → 10⁷. If σ_a < σ_endurance ≈ 0.5·σ_UTS, cyclic stress accumulates no damage. Physically this is explained by the body-centered cubic lattice of steel, with Lüders bands — discrete dislocation pinning, where slip bands don’t activate below threshold.

  • Aluminum (face-centered cubic lattice) has NO endurance limit. Per ASM Handbook Vol. 19 and ISO 12107:2012 «Metallic materials — Fatigue testing — Statistical planning» all aluminum alloys show a continuous decrease in σ_a even at N = 10⁹. This means given enough time, an aluminum frame will fail under any cyclic load, however small it may be. In practice, engineers define a «conditional fatigue limit» σ_f(5×10⁸) — the stress level at which the frame survives 500 million cycles. For 6061-T6 this is ≈ 96 MPa, for 7075-T6 ≈ 160 MPa.

This fundamentally changes the design approach. In a steel frame you can design a «forever frame» — a frame where σ_a < σ_endurance, in theory infinite life. In an aluminum frame there is no such regime — it always has a finite life N_f, after which it cracks. So standards EN 17128 § 6.5 and ISO 4210-3 specify a concrete cycle count (50 000 or 100 000), not an endurance limit. A scooter is designed for a life cycle of 5–10 years at 5 rides per week (typically 2·10⁶ — 4·10⁶ cycles for main structural members), and if the frame survives that window — it gets replaced.

Mean stress effect — Goodman / Soderberg / Gerber diagrams. The Basquin equation describes fully reversed loading with R = σ_min/σ_max = −1. In reality the frame carries non-zero mean stress — σ_m > 0 from the rider’s static weight. This lowers the allowable σ_a for the same N_f:

  • Goodman line (linear): σ_a/σ'_f + σ_m/σ_UTS = 1 — most conservative
  • Soderberg line (linear): σ_a/σ'_f + σ_m/σ_y = 1 — even more conservative (for plastic-deformation avoidance)
  • Gerber parabola: σ_a/σ'_f + (σ_m/σ_UTS)² = 1 — best fit to test data but less common in engineering practice

For a scooter with typical 30–50 MPa mean stress in a welded joint, Goodman correction reduces the allowable σ_a by 20–30 % relative to uncorrected Basquin.

Miner’s linear damage rule. Real loading is variable amplitude (different cycles at different σ_a), not constant. The Palmgren-Miner linear hypothesis (Miner M. A., 1945):

D = Σ (n_i / N_i)

where n_i is the actual number of cycles at amplitude σ_a,i and N_i is the cycles-to-fracture from the S-N curve at the same amplitude. Fracture is expected when D = 1. This lets engineers combine different regimes (rough pavement, curb hits, smooth riding) into a single life-time predictor.

6. Stress concentration K_t and where frames break

The theoretical stress concentration factor K_t describes the local stress amplification at geometric discontinuities (notches, holes, fillets, weld toes) above the far-field nominal stress. For an infinite plate with a circular hole under tension per Pilkey W. D. («Peterson’s Stress Concentration Factors», 3rd edition 2008):

K_t = 3.0    (circular hole in infinite plate under tension)
K_t = 2.0    (semi-circular notch in plate under tension)
K_t ≈ 3–6   (sharp fillet at weld toe, depends on radius/width ratio)
K_t ≈ 1.5–2.5 (fork crown to steerer tube transition)

In fatigue, K_t is modified by the notch sensitivity factor q:

q = 1 / (1 + a/r)
K_f = 1 + q · (K_t − 1)

where a is Neuber’s material constant (a ≈ 0.5 mm for aluminum, ~0.1 mm for steel) and r is the notch radius. If r → 0 (sharp notch), q → 0, K_f → 1 — aluminum would have no notch effect, which is counter-intuitive. In practice, the Topper modification keeps K_f close to K_t for sharp notches due to defect segregation.

The critical stress-concentration locations on a scooter frame (highest K_f):

  1. Stem base weld toe — where the vertical stem tube meets the deck through a welded gusset. K_f ≈ 3–5 from the combination of section change (tube → plate) + 50 % HAZ knockdown + a typical 1–3 mm weld-toe radius. The most frequent fatigue-crack initiation site on the Xiaomi M365 (this is where the 2019 crack nucleated that drove the recall).

  2. Folding hinge pivot pin location. A pin sliding through holes in deck and stem. K_t = 3.0 for a circular hole + K_f = 3–4 with notch sensitivity. On the M365, that’s exactly where the screw that backed out lived — once it backed out, local stress on the pin spiked, it sheared (shear failure), and the stem broke free.

  3. Fork crown / steerer tube transition. Section change from a thin steerer tube (28.6 mm) to a wide fork crown (50–80 mm) gives K_f ≈ 2–3. On bicycles this is a classic rupture site (historical cases on Cannondale CAAD in the 1990s). On scooters it’s rarer but appears as a failure mode in off-road models after 1+ meter jumps.

  4. Deck-stem joint weld. Where the stem tube meets the deck plate at an 80–90° angle. K_f ≈ 2.5–4 from the section change + weld geometry + HAZ knockdown. This is where the first Lime/Bird sharing scooters cracked in 2018–2019 (deck cracks).

  5. Quick-release lever bolt hole on the folding hinge. A hole through a 6061-T6 plate, K_t = 3.0. If thread engagement is less than 5 pitches (the ISO 5855 minimum), the bolt can bend under load and further concentrate stress → a fatigue crack at the hole edge.

  6. Handlebar T-joint — where the horizontal handlebar meets the vertical stem. K_f ≈ 2–3. Unlike a bicycle, where the handlebar is held by a stem clamp (no welded joint), on a scooter this is a welded joint with HAZ effect.

Engineering mitigation:

  • Increase fillet radii (r → ∞ → K_t → 1). Visible on the NAMI Burn-E 2 — fillet radius around the stem socket of 8–10 mm vs 2–3 mm on the Xiaomi M365.
  • Increase wall thickness locally through gussets and reinforcement plates.
  • Eliminate welded joints where possible — top models partly carry the frame as a monolithic CNC-milled block of 6061-T6 with no weld at all (NAMI Burn-E 2 steerer / Wolf King GT center bracket).
  • Shot peening of weld toes — surface compressive residual stress of 100–300 MPa, blocking fatigue-crack initiation. Standard in aerospace, rare on consumer scooters.

7. Kinematics and mechanics of folding locks

A folding mechanism is a single-degree-of-freedom hinge mechanism with a locking device. Three main types from a mechanical standpoint:

Type 1. Lever-latch with hook (Xiaomi M365 family, Segway Ninebot Max). A steel hook is pressed by a lever pivoting about O. Moment balance:

F_hook · L_arm = F_lever · L_lever

with L_arm / L_lever ≈ 0.2 — 0.3 → 3–5× mechanical advantage. With a 50 N lever force (average finger force) the hook holds 150–250 N. If the stem applies through the hook a 0.5 g deceleration with a CoG 0.9 m above the deck, F_hook ≈ m_rider · 0.5 g · 0.9 / 0.1 = 220 N for a 50-kg rider. The net margin is only 2–3× — and that’s precisely why the Xiaomi M365 hook is the first to fail when worn.

Type 2. Multi-point hinge (Apollo City Pro, Phantom). A 3-bar linkage with three contact points. Load is distributed: at each pin the friction force F_pin = N · μ ≈ 0.1 · N complements the main locking force. Total reserve capacity 2–3× vs a single hook. Costlier to manufacture, more complex.

Type 3. Twist-and-fold with threaded sleeve (NAMI Burn-E lock taper). A conical thread interface, similar to a chuck-jaw mechanism in a lathe. Thread engagement ≥ 5 full thread pitches per ISO 5855 and Machinery’s Handbook (29th edition 2012) — the recommended minimum for full strength in aluminum threads. Self-locking design: thread lead angle α < tan⁻¹(μ_static) ≈ 6–10° for steel-on-steel means the thread won’t back out from vibration alone — an active torque is required.

Type 4. Push-button trigger-pin (Mantis King GT, certain Dualtron models). A spring-loaded pin shoots through a hole. Pin shear strength:

F_shear = π/4 · d² · τ_y

For an 8-mm 4140 steel pin with τ_y ≈ 0.577 · σ_y ≈ 0.577 · 655 = 378 MPa:

F_shear = π/4 · 0.008² · 378 × 10⁶ = 19 000 N

A comfortable margin over rider load. The weakness — the pin can bind from dust and grit with poor sealing, or the spring corrodes in damp environments.

Defense-in-depth via a secondary safety pin. On high-power models (Dualtron Storm, NAMI Burn-E 2) the primary release lever is paired with a secondary manual safety pin that further locks the stem. That’s single-point failure mitigation: if the primary lock fails through fatigue, vibration, or operator error, the secondary pin holds the stem from sudden folding.

Bolt preload and vibration loosening (Goodman screw fatigue). Bolt-tightening preload F_pre = T / (k · d) where T is torque (N·m), k ≈ 0.2 for unlubricated steel, d is bolt diameter. For an M6 bolt at 8 N·m: F_pre = 6 700 N. Vibration loosening occurs when external loading exceeds preload + friction → the bolt slips axially. Standard countermeasures — threadlocker (Loctite 243 medium-strength, breakaway torque 12 N·m at room temperature) or a lock washer (Belleville spring washer maintains compressive preload). On the M365 2019 recall this exact mechanism failed — the screw lacked adequate threadlocker, vibration loosening caused bolt slip leading to catastrophic stem separation.

8. Steering geometry: trail, wheel flop, headset bearings

The front fork connects to the frame through the headset via angular contact bearings — typically two conical bearings in a 36° / 36° or 45° / 45° configuration (semi-integrated headset, e.g., FSA Orbit, Cane Creek). Angular contact handles axial + radial loads at once, critical for scooters (vertical wheel impact = both axial steerer load + side-to-side bearing load from cornering).

The headset bearing is its own engineering discipline with its own standards (ISO 281 L₁₀ dynamic life, ISO 76 C₀ static load rating, ABEC/ISO 492 precision class) and a failure mode characteristic for scooters (false brinelling from impact load > C₀/4 on a curb strike, which plastically indents the raceway and produces a notchy steering feel well before visible wear). The full control-engineering treatment is in the rolling-element bearing engineering deep-dive (ISO 281) — §6 (C₀ static rating and true brinelling), §7 (ABEC precision classes and their role in headset preload tolerance), §11 (failure modes including impact-induced indentation).

Mechanical trail t — the horizontal distance between the projection of the steering axis on the road and the wheel-contact point:

t = (R · cosα − r_offset) / sinα

where R is the wheel radius, α is the head angle (complementary to the stem rake), and r_offset is the fork offset (rake). On scooters:

ModelR (mm)head angleoffset (mm)trail t (mm)
Xiaomi M365110 (8.5″)78° (≈12° from vertical)530
Segway Max G30127 (10″)76°032
Apollo Phantom152 (12″)73°556
NAMI Burn-E 2152 (12″)70°075
Dualtron Thunder 2140 (11″)68°860

— the trend: higher trail → more stable at speed, harder to turn at low speed. Bicycle trail is typically 50–60 mm; on scooters it’s slightly larger precisely because of the short wheelbase (1000–1100 mm on scooters vs 1000–1200 mm on MTB) and higher h/L mass-distribution ratio.

Wheel flop factor Wflop — a related metric describing the stabilizing or destabilizing effect of deviating from straight-line:

W_flop = t · sinα · cosα

High W_flop makes the wheel «fall» into a turn (autoturning tendency), useful for low-speed handling but increasing oscillations at high speed (shimmy / speed wobble — a resonant instability that can start at 30–40 km/h on scooters with low trail). So off-road off-models use high trail (75 mm) and low W_flop — stability outweighs agility.

Steerer tube shear stress under braking impulse. The front fork takes pure shear τ = F_brake / A near the fork crown when the front brake engages. For a 0.8 g deceleration on 100 kg total mass: F_brake = 0.8 · 100 · 9.81 = 785 N longitudinal through the front wheel. This is transmitted via the steerer to the frame as the moment M = F · h_wheel = 785 · 0.3 = 235 N·m. In a circular steerer tube 28.6 mm OD × 25.4 mm ID:

J = π · (28.6⁴ − 25.4⁴) / 32 = 35 700 mm⁴
τ_max = M · r / J = 235 000 · 14.3 / 35 700 = 94 MPa

— below the 6061 base material yield of 276 MPa, but close to the HAZ knockdown of 138 MPa, especially with impact amplification × 2–3 on a curb strike. That’s why fork-crown welds are class-critical.

9. Frame and fork strength standards — full comparison matrix

StandardPublisherScopeKey requirements
EN 17128:2020 § 6.4CEN/TC 354 (AFNOR, FR)PLEV — frame impactDrop test 22 kg × 180 mm via front wheel; frame must not separate from deck, no catastrophic failure
EN 17128:2020 § 6.5CEN/TC 354PLEV — frame fatigue50 000 cycles × 1.3 dynamic factor over static rider load; no visible crack growth
ISO 4210-3:2014ISO/TC 149Bicycle frame+fork100 000 cycles vertical 1 200 N + horizontal forward 600 N; horizontal forward fatigue 50 000 cycles 1 200 N; impact falling mass 22.5 kg × 180 mm
EN 14781:2005CEN/TC 333Racing bicycle (frame)Stricter fatigue tests than ISO 4210, specific to UCI-class racing — 100 000+ cycles
ASTM F2641-15ASTM Subcommittee F08.18Recreational Powered Scooters ≤ 32 km/hStatic load 2× max payload; impact test from defined drop height; no separation under load
ASTM F2711-08ASTM Subcommittee F08.18Trick scooters (non-powered, BMX-style)Frame deflection limits; weld penetration verification; static load 1.5× design load
DIN 79014:2014DIN (Germany)City Bike additional requirementsStricter than ISO 4210 in some clauses, specific to urban commuter use
JIS D 9301:2024JISC (Japan)Bicycle Frame StrengthStatic load test; fatigue test 100 000 cycles
UL 2272:2016UL (US)E-mobility structural + electricalImpact test; vibration test; required for retail sale in some US states

EN 17128:2020 remains the core PLEV standard in Europe — the same one we saw in suspension engineering (§ 6.4–6.5 applies to the frame in general). Concrete test parameters for frame impact (§ 6.4): drop test 22 kg × 180 mm via the front wheel in vertical orientation — the frame must not separate from the deck, no catastrophic failure, no visible cracks initiated. This simulates a curb impact at roughly 25 km/h (energy = 22 · 9.81 · 0.18 = 38.9 J).

ISO 4210-3:2014 is a bicycle-specific standard, but is often applied by analogy to scooter frames, especially in jurisdictions without a PLEV-specific standard. Test parameters: 100 000 cycles vertical 1 200 N + horizontal forward 600 N (combinations to simulate combined loading); horizontal forward fatigue 50 000 cycles 1 200 N — the same test rig EFBe has applied for over 30 years to bicycle frames (Sheldon Brown documentation lists 12 high-end frames tested by EFBe — Cannondale CAAD, Trek 8500, Specialized M2 — a usable benchmark for consumer hardware).

ASTM F2641-15 covers recreational powered scooters with speed limited to ≤ 32 km/h, which includes most consumer scooters. It does not cover on-road PLEVs — for those, UL 2272 applies on the electrical side + state-specific regulations for the structural part. So a consumer scooter sold in the US must meet UL 2272 + ASTM F2641, while in the EU it must meet EN 17128.

ASTM F2711-08 — for trick scooters (BMX-style without a motor), has even stricter impact requirements because it anticipates jumps and stunts. Some off-road powered scooters (Mantis King GT, NAMI Burn-E 2) are voluntarily tested to F2711 for the marketing positioning «engineered for jumps».

Why the standards are fragmented: PLEVs are a new category (legal status in the EU since 2019 in DE, FR; in the US — state-by-state since 2018), and standards aren’t yet unified. Top-model manufacturers often voluntarily test to bicycle-grade ISO 4210-3 + EN 14781 on top of PLEV-specific EN 17128, because bicycle standards have been historically stricter.

10. Engineering ↔ symptoms — diagnostic matrix

SymptomProbable engineering causeTest/inspection
Stem wobble (horizontal «play» of the stem)Wear in fold-hook pivot or loose folding latch boltTighten bolt, verify thread engagement ≥ 5; if wobble persists — replace hook assembly
Cracking sound at the stem-to-deck joint under loadIncipient fatigue crack in weld HAZ (50 % knockdown zone)Visual inspection with magnifying glass at the weld toe; dye-penetrant testing (Magnaflux Spotcheck SKL-SP2)
Headset creaking when steeringBearing race wear or loose bearing preloadRe-grease + reload via top cap; if not silenced — replace bearings
Sudden deck cracking under your feet at speedCatastrophic fatigue failure in deck weldSTOP IMMEDIATELY, no further riding. Likely Miner’s D = 1 reached
Loose folding bolt (gradually loosens over rides)Vibration loosening — insufficient preload or no threadlockerClean threads + Loctite 243 + retorque to spec (typically 8 N·m M6)
Hum/vibration at 30–40 km/h that grows with speedSpeed wobble — instability of trail/wheel-flop geometryCheck tire pressure; if it doesn’t disappear — geometry problem, requires stem/fork replacement
Stem tilts out of vertical when scooter stands stillLatch not fully engaged or wear in hookFold/unfold, verify hook fully captures the frame
Silvery radial lines from a weld toeStrain-hardening lines from K_f stress concentration — a precursor to crackingSTOP IMMEDIATELY, photograph, replace the frame
Rust/corrosion at the weld toeHAZ is more susceptible to pitting corrosion through altered microstructureClean, apply anti-corrosion primer + paint; if deep pitting — replace
Bent stem after a relatively minor fallYielding of the HAZ zone (knockdown to 138 MPa)If bend > 5 mm — frame compromised, replace; bend ≤ 2 mm after straightening — use with elevated caution
Spongy feel in the folding mechanism (doesn’t click crisply)Pin wear, spring failure, or slop in the hinge axleDisassemble, inspect pin radii, replace if wear > 0.5 mm

Recap in 8 points

  1. The frame is a structural integrator, carrying loads from every other subsystem (motor, brake, suspension, tire). The physics is bending + torsion + axial simultaneously, with yield criterion σ_v = √(σ²+3τ²) ≤ σ_y per von Mises. A round-section tube has I = π(D⁴−d⁴)/64 — a quartic function of diameter — so a large diameter with a thin wall is always stiffer than a small one with a thick wall.
  2. All aluminum alloys deliver the same specific stiffness E/ρ ≈ 25.5 GPa·cm³/g — E is set by the crystal lattice, not by alloying. Choosing 7075 over 6061 is for strength alone (specific strength 179 vs 102 MPa·cm³/g), not stiffness. 6061-T6 is the universal default for the combination (weldability + corrosion resistance + price).
  3. HAZ knockdown of 50 % is a fundamental metallurgical inevitability for 6xxx alloys. At a weld toe in 6061-T6 the σ_y of 276 MPa falls to 138 MPa. Designers compensate via welded gussets or post-weld heat treatment (PWHT, aerospace-grade). Per AWS D1.2 and the Aluminum Design Manual the standard practice is a 50 % knockdown factor.
  4. 7075 is unweldable in thin-wall frames due to hot-cracking susceptibility of Cu-Mg-Zn precipitates. It’s used only locally as a CNC-machined part bolted to a 6061 chassis. 6082 ≈ 6061 with slightly better corrosion resistance — marketing «aerospace-grade», engineering-wise minor.
  5. Aluminum has no endurance limit, unlike steel. The fatigue curve keeps falling per the Basquin equation σ_a = σ'_f · (2N_f)^b. Standards therefore specify a concrete cycle count (EN 17128 § 6.5 — 50 000 cycles × 1.3; ISO 4210-3 — 100 000 cycles 1 200 N), not an endurance limit. Engineering life is 2·10⁶ — 4·10⁶ cycles for main elements, equivalent to 5–10 years at 5 rides/week.
  6. Stress concentration K_t at geometric discontinuities + K_f = 1 + q(K_t−1) with notch sensitivity. Critical hotspots — stem base weld toe (K_f 3–5), folding hinge pivot (K_t 3.0), fork-crown transition (K_f 2–3), deck-stem joint weld (K_f 2.5–4). The 2019 Xiaomi M365 hook failure sat exactly in a high-K_f HAZ zone with Miner’s D = 1 reached after vibration-induced bolt loosening (recall of 10 257 units).
  7. Folding locks — single-DOF mechanisms with a lock. Lever-latch (M365) gives a 3–5× mechanical advantage with a slim 2–3× margin against rider load — the main failure point on consumer scooters. Multi-point hinge (Apollo), twist-and-fold (NAMI lock taper with ISO 5855 thread engagement ≥ 5 pitches), and trigger-pin (Mantis) offer wider margins. A secondary safety pin as defense-in-depth is standard on top models.
  8. Standards are fragmented by jurisdiction. EU PLEV → EN 17128:2020 § 6.4–6.5 + ISO 4210-3:2014. US → UL 2272:2016 + ASTM F2641-15 (recreational) + ASTM F2711-08 (trick). Japan → JIS D 9301:2024. Germany → additionally DIN 79014:2014. Top-class brands (NAMI, Apollo) voluntarily test to bicycle-grade EN 14781 + ISO 4210-3 on top of PLEV-specific standards because bicycle standards have historically been stricter.

Sources

A list of ENG-first sources behind §§1–10. Grouped by thematic clusters; in parentheses — the brief context of what exactly was used from the source and in which section of the article.

§1–§2 Beam mechanics, Euler-Bernoulli theory, section modulus, von Mises criterion:

  1. Wikipedia — Euler–Bernoulli beam theory (the fundamental σ = M·c/I formulation for §1 bending mode + δ = F·L³/(3·E·I) cantilever deflection for §2 stem-as-cantilever).
  2. Wikipedia — Second moment of area (I = π(D⁴ − d⁴)/64 for a thin-wall tube — the fundamental cause of the quartic dependence in §2; polar J = π(D⁴ − d⁴)/32 for §1 torsion mode).
  3. Wikipedia — Section modulus (Z = I/c as a measure of elementary bending capacity, used in §1 for derivation σ_max = M/Z).
  4. Wikipedia — Von Mises yield criterion (σ_v = √(σ² + 3τ²) ≤ σ_y for combined bending + torsion stress state — §1 yield criterion).
  5. Gere, J. M., & Goodno, B. J. (2018). Mechanics of Materials, 9th ed. Cengage Learning. ISBN 978-1-337-09334-7 (the canonical undergraduate engineering textbook for cantilever beam analysis, combined loading, von Mises criterion — §1–§2 foundation).
  6. Hibbeler, R. C. (2017). Mechanics of Materials, 10th ed. Pearson. ISBN 978-0-13-431965-0 (parallel reference for beam mechanics, torsion of circular shafts, principal stresses — §1 derivation of combined-loading stress state).

§3 Material science, specific stiffness/strength, Ashby selection methodology:

  1. Ashby, M. F. (2017). Materials Selection in Mechanical Design, 5th ed. Butterworth-Heinemann. ISBN 978-0-08-100599-6 (the canonical text for E/ρ vs σ_y/ρ charts; performance indices M = σ_y^(2/3)/ρ for a stiffness-limited beam; methodology for narrowing material candidates — §3 entire framework).
  2. Wikipedia — 6061 aluminium alloy (chemistry Mg-Si, T6 temper σ_y = 276 MPa, E = 68.9 GPa, weldability rating — §3 6061-T6 row of the comparison table).
  3. Wikipedia — 7075 aluminium alloy (chemistry Zn-Mg-Cu, T6 σ_y = 503 MPa, HAZ hot-cracking susceptibility, η-MgZn₂ precipitate — §3 7075-T6 row + §4 unweldability rationale).
  4. Wikipedia — 7005 aluminium alloy (chemistry Zn-Mg without Cu, auto-aging post-weld response — §3 7005-T6 row).
  5. Wikipedia — 6082 aluminium alloy (corrosion-resistance comparison vs 6061, lower Cu content ≤0.1 % — §3 6082-T6 row + §4 NAMI/Apollo “aerospace-grade” context).
  6. Wikipedia — 41xx steel (Cr-Mo low-alloy steel, E = 205 GPa, σ_y ≈ 460 MPa, GMAW/GTAW weldability — §3 steel row).
  7. Wikipedia — Magnesium alloy § AZ91 (die-cast AZ91D properties, SF₆ shielding requirement during welding — §3 magnesium row).
  8. Wikipedia — Carbon-fiber-reinforced polymer (unidirectional T700S tensile strength 4900 MPa along fiber direction, anisotropy 50 MPa transverse — §3 CF row + quasi-isotropic layup reasoning).
  9. Aluminum Association (2020). Aluminum Design Manual 2020. ISBN 978-0-9826308-7-7 (industry-standard reference for aluminum structural design including HAZ knockdown factors and allowable stress derating — §3 + §4 + §5).

§4 Welding metallurgy, HAZ, filler-metal selection:

  1. American Welding Society (2014). AWS D1.2/D1.2M:2014 — Structural Welding Code — Aluminum. AWS. ISBN 978-0-87171-836-2 (the canonical North-American standard for structural aluminum welding; 50 % yield-strength knockdown factor across HAZ — §4 foundational reference for the knockdown rationale).
  2. Wikipedia — Heat-affected zone (microstructural definition of HAZ, overaging mechanism in precipitation-hardened alloys — §4).
  3. Wikipedia — Gas tungsten arc welding (GTAW process, AC mode for aluminum oxide-film disruption, electrode types — §4 introduction).
  4. Wikipedia — Precipitation hardening (β’-Mg₂Si precipitate physics in 6xxx, η-MgZn₂ in 7xxx, T4/T6/O temper definitions — §4 mechanism narrative).
  5. Polmear, I., StJohn, D., Nie, J.-F., & Qian, M. (2017). Light Alloys: Metallurgy of the Light Metals, 5th ed. Butterworth-Heinemann. ISBN 978-0-08-099431-4 (the canonical textbook for aluminum metallurgy including 6xxx precipitation kinetics, HAZ effects, filler-metal selection for ER4043/ER5356/ER4047 — §4 detailed mechanism).
  6. The Aluminum Association — Filler Metal Selection Chart for Aluminum Alloys, 2018 (the industry-canonical chart for matching filler to base alloy for 6061/7005/6082; references ER4043/4047/5356 properties — §4 filler-metal selection table).

§5 Fatigue: Basquin, Coffin-Manson, Miner’s rule, no-endurance-limit Al:

  1. Basquin, O. H. (1910). The exponential law of endurance tests. Proceedings of ASTM, 10, 625–630 (foundational S-N curve formulation σ_a = σ'_f · (2N_f)^b — §5 cite as original source).
  2. Miner, M. A. (1945). Cumulative damage in fatigue. Journal of Applied Mechanics, 12(3), A159–A164. DOI 10.1115/1.4009458 (linear damage hypothesis D = Σ(n_i/N_i), fracture at D = 1 — §5 variable-amplitude framework).
  3. Wikipedia — Fatigue (material) (S-N curves, endurance-limit absence in non-ferrous metals, HCF vs LCF regimes — §5 introduction).
  4. Wikipedia — Goodman relation (mean-stress correction σ_a/σ'_f + σ_m/σ_UTS = 1; Soderberg + Gerber alternatives — §5 mean-stress effect).
  5. ASM International (1996). ASM Handbook, Volume 19: Fatigue and Fracture. ASM International. ISBN 978-0-87170-385-6 (industry-canonical fatigue data including 6061-T6 σ'_f ≈ 478 MPa, b ≈ −0.083 — §5 specific values).
  6. Stephens, R. I., Fatemi, A., Stephens, R. R., & Fuchs, H. O. (2000). Metal Fatigue in Engineering, 2nd ed. Wiley. ISBN 978-0-471-51059-9 (canonical graduate-level fatigue reference, Basquin/Coffin-Manson treatment, Miner’s rule — §5 derivation depth).
  7. ISO — ISO 12107:2012 Metallic materials — Fatigue testing — Statistical planning and analysis of data (standardised methodology for S-N curve construction, conditional fatigue-limit definition σ_f(5×10⁸) — §5 statistical-fatigue framework citation).

§6 Stress concentration K_t, notch sensitivity K_f:

  1. Pilkey, W. D., & Pilkey, D. F. (2008). Peterson’s Stress Concentration Factors, 3rd ed. Wiley. ISBN 978-0-470-04824-5 (the canonical reference for K_t values across discontinuity geometries — circular hole, semi-circular notch, fillet, weld toe — §6 numerical values).
  2. Neuber, H. (1958). Theory of Notch Stresses. J. W. Edwards Publisher (foundational notch-sensitivity K_f = 1 + q·(K_t − 1), material constant a for aluminum vs steel — §6 fatigue-modified concentration factor).
  3. Wikipedia — Stress concentration (K_t definition, sharp-notch vs blunt-notch distinction — §6 introduction).

§7 Folding mechanism kinematics, bolted-joint mechanics, thread engagement:

  1. ISO — ISO 5855-1:1999 Aerospace — MJ threads — Part 1: General requirements (thread-engagement length specification ≥5 pitches for full thread strength in aluminum — §7 NAMI lock-taper twist-fold reference).
  2. Industrial Press — Machinery’s Handbook, 31st ed. (2020). ISBN 978-0-8311-3733-7 (the industry-canonical reference for thread-engagement-length recommendations, bolt-tightening torque, K-factor methodology — §7 bolt preload F_pre = T/(k·d)).
  3. Henkel — Loctite 243 Threadlocker Technical Data Sheet (medium-strength threadlocker, 12 N·m breakaway torque at 25 °C — §7 vibration-loosening countermeasure cited for M365 recall context).

§8 Steering geometry, trail, wheel flop, two-wheeled dynamics:

  1. Wikipedia — Bicycle and motorcycle geometry (mechanical trail t derivation, head angle, fork offset (rake), wheel flop factor — §8 geometric parameters).
  2. Meijaard, J. P., Papadopoulos, J. M., Ruina, A., & Schwab, A. L. (2007). Linearized dynamics equations for the balance and steer of a bicycle: a benchmark and review. Proceedings of the Royal Society A, 463(2084), 1955–1982. DOI 10.1098/rspa.2007.1857 (canonical 4-DOF linearized bicycle model with reproducible parameters — §8 dynamic-stability framework, foundation for speed-wobble cross-link).
  3. Cossalter, V. (2006). Motorcycle Dynamics, 2nd ed. ISBN 978-1-4303-0861-4 (canonical reference for motorcycle steering geometry, trail effects, weave-wobble eigenvalue analysis — §8 contrast e-scooter vs motorcycle geometry).
  4. Wilson, D. G., & Schmidt, T. (2020). Bicycling Science, 4th ed. MIT Press. ISBN 978-0-262-53880-4 (foundational reference for bicycle frame engineering, steering geometry, energy budget; cross-reference for §8 trail values vs bicycle conventions).

§9 Standards matrix for PLEV frame and bicycle frame:

  1. CEN — EN 17128:2020 Light motorized vehicles for the transportation of persons and goods and related facilities and not subject to type-approval for on-road use — Personal light electric vehicles (PLEV) — Requirements and test methods (the canonical European PLEV standard, § 6.4 frame impact + § 6.5 frame fatigue — §9 primary EU reference).
  2. ISO — ISO 4210-3:2014 Cycles — Safety requirements for bicycles — Part 3: Common test methods (vertical 1 200 N + horizontal 600 N fatigue tests, impact falling mass 22.5 kg × 180 mm — §9 international bicycle standard often applied to PLEV by analogy).
  3. ASTM — ASTM F2641-15 Standard Consumer Safety Specification for Powered Scooters for Use by Children (recreational powered scooters ≤32 km/h, static load 2× max payload, impact test — §9 US standard).
  4. ASTM — ASTM F2711-08 Standard Test Methods for Bicycle Frames (trick-scooter equivalent test methods, frame deflection limits — §9 US reference for stunt/jump-rated frames).
  5. UL — UL 2272 Standard for Safety: Electrical Systems for Personal E-Mobility Devices (electrical + structural test requirements for e-scooters; NYC Local Law 39 mandatory reference — §9 US safety standard).
  6. JISC — JIS D 9301:2024 Bicycles — Frames (Japanese static-load and 100 000-cycle fatigue test — §9 Japan reference).
  7. DIN — DIN 79014:2014 Pedelecs — Safety requirements and test methods (German city-bike additional requirements layered on ISO 4210 — §9 Germany reference).
  8. CEN — EN 14781:2005 Racing bicycles — Safety requirements and test methods (racing-grade fatigue requirements stricter than ISO 4210, voluntarily adopted by top PLEV brands — §9 EN-racing reference).

§10 Diagnostic matrix, manufacturer recalls, fastener failure modes:

  1. Australian Competition & Consumer Commission — Product Safety Australia: Mi Electric Scooter Recall Notice (Xiaomi M365 stem-folding-mechanism safety advisory — §6 + §10 cite for the canonical real-world hook-failure case context).
  2. Magnaflux — Spotcheck SKL-SP2 Dye Penetrant Technical Data (industrial-standard dye-penetrant testing material referenced in §10 weld-toe crack inspection).

The frame is not «a rigid beam» holding the rider’s mass. It is the structural integrator carrying every other subsystem’s cyclic load (motor → axial vibration; brake → impulsive shear; suspension → resonant vibration; tire → vertical impact). Engineering quality is not described by «6061-T6 aluminum» as a marketing price-tag — it is described by section geometry (I, J, Z as functions of D and t), the welding process and HAZ knockdown factor, stress-concentration design at geometric discontinuities, the endurance or conditional fatigue limit, and defense-in-depth in the folding mechanism. An owner has no way to optimize these parameters after purchase — but they can identify them through verification of CE marking referencing EN 17128:2020, the presence of a multi-step folding mechanism with a secondary safety pin, visible welded gussets at high-stress points, and the absence of K_f-critical geometric discontinuities (sharp fillets, vivid stress-concentration points). If you see a frame with smooth 8–10 mm radii at the stem joint, gussets around the folding hinge, and no thin-wall regions with D/t > 22 — that’s a sign of serious engineering. If not — it’s a scooter built «for weight», cutting corners where there is no right to cut them.

Read next

Consultation