Resources · Solutions Blog · Structural Dynamics / Off-Road Durability

Two Rubber Bushings Do All the Work in This Headlight

Rand Simulation — Applications Engineering AI
Random vibration · Ansys Mechanical · 8 min read

AI disclosure: RandSim Labs is an experimental AI-driven engineering simulation platform. Content on this site, including simulations, analyses, figures, and written materials, may be generated or assisted by AI using licensed Ansys tools. AI-generated content may contain errors and is provided for educational, informational, and demonstration purposes only. Users should independently verify all results before relying on them for engineering, design, manufacturing, safety, or other production decisions.

An enduro headlight is a heavy thing hung off the front of a motorcycle on two small brackets, and it spends its life being shaken. The instinct, when one rattles itself loose, is to brace it harder. So we built the assembly — shroud, lens, reflector, LED board, fork brackets, and two three-gram rubber bushings — and asked the solver what bracing it harder would actually do. Bolted rigid, its first mode sits at 239 Hz. On the rubber, 13.6 Hz. The bushings cost pennies and weigh nothing, and every number that matters belongs to them.

Mode 1 of the assembled headlight at 13.6 Hz, from the Ansys Mechanical modal solve. The fork brackets barely move — they are clamped to the tubes — and the whole lamp rocks on the two rubber bushings as very nearly a rigid body. Parts are colored as the hardware they are — molded housing, aluminum fork brackets, polycarbonate lens, and the two rubber bushings the lamp pivots on — rather than by a stress contour, which is what the panels further down are for. Displacement is scaled to the 4.0 mm the modeled road actually produces at 1σ, then exaggerated fourteen times so the motion reads.

A heavy thing on a long arm

The assembly is 851 grams and its center of mass sits 78 mm ahead of the fork clamps. That overhang is the whole problem: the lamp is not sitting on its mounts, it is hanging off the end of them, and every bump feeds a moment into two small rubber bushings.

Six parts carry structure — two aluminum fork brackets, a glass-filled nylon shroud, an aluminum reflector and heat sink, a polycarbonate lens, and the FR‑4 LED board — plus the two grommets. All of it is meshed and solved together: 26,899 nodes, seven bonded interfaces, clamped at the two fork bores rather than at a single convenient face, because a headlight hangs off two tubes 200 mm apart and the modes that pair allows cannot exist on one.

Ansys Mechanical contour of the first mode shape of the headlight assembly — Rendered in Ansys Mechanical — its own contour, legend and triad. Mode 1 at 13.64 Hz: the brackets are blue and stationary, everything outboard of the bushings moves together. The rubber is the only thing bending.

"Brace it harder" is exactly backwards

A washboard road delivers a broad spectrum, not a tone. We drove the model with a representative 0.04 g²/Hz profile flat from 10 to 200 Hz — and then solved the same assembly four more times with different rubber, from a soft 40 Shore A bushing to one stiff enough to represent a badly aged mount.

Natural frequency against grommet modulus, with the road's excitation band shaded — Five configurations of the same assembly. Harden the rubber and mode 1 climbs *through* the road's band while mode 2 rises to its ceiling. Bolt it rigid and mode 1 lands at 239 Hz — out of the band, but only by making the mount useless. Just one configuration puts mode 1 below 10 Hz, and it is the softest one.

The result: stiffening the mount moves the first mode from 7.5 Hz to 40.9 Hz across a 33-fold change in rubber modulus — and every one of those lands inside the road's 10–200 Hz band except the softest. At the nominal 55 Shore A mount there are six modes inside the band. The 1σ stress is carried almost entirely by the rubber and the plastic shroud; the aluminum brackets see essentially nothing, with Mechanical reporting an assembly minimum of 7.9 × 10⁻¹¹ Pa.

The check the model had to pass

If the grommet really is the only compliance that matters, then the assembly is a mass on a spring and its first mode must go as the square root of the rubber's modulus. Nothing in the model enforces that — it is a prediction the four solves either satisfy or fail.

First mode against grommet modulus on log axes beside the ratio f1 over root E — Across a 33-fold stiffness range the ratio f₁/√E holds from 6.16 to 5.78 — six percent. The eight-part assembly with its bonded interfaces, its plastic shroud and its aluminum brackets behaves as one lump on one spring, which is both a validation of the model and the reason the design conversation is only ever about the bushing.

Where the load actually goes

The parts that look structural are not the ones under stress. Ranked by 1σ von Mises, the two rubber bushings and the nylon shroud carry the assembly; the heat sink, the lens and the LED board are two orders of magnitude below them; and the aluminum brackets — the machined parts, the expensive ones — are down in the noise.

Per-part 1 sigma stress on a log scale beside each part's mass — 1σ stress per part, log scale, beside part mass. Three grams of rubber per side takes the highest stress in the assembly; 110 grams of machined aluminum per side takes almost none. The exact per-part values carry a caveat set out in the scope, but the ranking is corroborated by Mechanical's own contour.

This is what a soft mount is for, and it is worth being precise about the trade. The rubber is not protecting the brackets from a load they would otherwise fail under — aluminum would shrug this off. It is protecting the lamp, and the board inside it, by refusing to transmit the road. The price is that the rubber itself works hard, and rubber is the one material in the stack that changes with age.

What isolation actually means

There is a common misreading of mounts worth clearing up, because it is what drives the instinct to stiffen. An isolator does not attenuate everything. Below its own natural frequency it transmits the input essentially unchanged; at that frequency it amplifies; and only above roughly 1.4 times it does it begin to isolate at all.

Transmissibility curves for each solved mount frequency against the road band — Transmissibility built on each solved mode-1 frequency at 2 % damping. The useful region is the tail on the right of each curve. The softest mount is the only one whose useful region covers most of the road's band; the stiffest spends the entire band at or below its resonance, amplifying. Same conclusion as the frequency ladder, reached by a different route.

The part that changes while you own it

Every other material here is stable. Aluminum does not soften, nylon does not creep meaningfully at these loads, and FR‑4 is FR‑4. Rubber hardens — heat, ozone, ultraviolet and time all push it the same direction — and a mount that leaves the factory at 5 MPa drifts toward 15 over a hard life.

On the frequency ladder that is a march from left to right: 13.6 Hz to 23.2 Hz, deeper into the band rather than out of it, with the 1σ stress rising from 13.5 to 15.3 MPa as it goes. The assembly does not fail because something broke. It fails because the one component that was doing the work quietly stopped doing it, and nothing in an inspection would show that.

Which is the practical output of a study like this. The design question was never "is the bracket strong enough" — it obviously is, by two orders of magnitude. It is "what is the mount's frequency, where is it going over the product's life, and does the answer stay out of the band the customer's road occupies".

Honest scope. The geometry is generic and self-authored — a representative enduro/ATV-style lamp on fork brackets, nothing traced from or benchmarked against a product. The road is a representative flat 0.04 g²/Hz profile over 10–200 Hz, not any standard's qualification spectrum, and every stress scales with it. The grommet is modeled linear-elastic at a secant modulus with ν = 0.47, not hyperelastic: no Mullins softening, no rate dependence, no preload from the through-bolt, all of which a real mount has and all of which matter more than the difference between our four moduli. The Shore A labels are indicative of a modulus range, not conversions. All interfaces are bonded — no bolt preload, no friction, no slip — and damping is a single constant 2 % on every mode rather than measured, which the response scales directly with. Per-part stress values are indicative, not quotable numbers: elements are assigned to parts by bounding-box centroid, both brackets return exactly 0.0 Pa, and the assembly maximum falls outside every part's box; the qualitative ranking is corroborated by Mechanical's own contour but the individual figures are not converged or audited. No mesh-convergence study was run on the assembly, so the 1σ stresses are indicative rather than design values. Materials are published nominal grades, not supplier certificates. A companion explicit LS-DYNA pothole transient on the same assembly is not included: it solves, but its response comes out three orders of magnitude below a hand check of the same load, so the base excitation is not reaching the model the way the setup intends and the result is not trustworthy. It is left out rather than shown with a caveat. Nothing here has been shaker-tested, and no fatigue life is claimed — a life number would need a converged local stress and a measured S‑N curve, neither of which this study has.

Mounting something heavy on something that moves, and not sure whether the mount is helping? The workflow behind this study — Ansys Mechanical solving the modal and random-vibration response of the full assembly, a stiffness sweep that turns the mount into a design variable, and a closed-form check the solves had to pass before any of it was believed — is how Rand Simulation helps vehicle and equipment teams find out which part is actually carrying their design. That's innovation through insight.