Documentation · Open method

Trail model: derivation and validation

By Carlos Eduardo Ravello Joo · ORCID 0009-0007-5631-7436 · BikeLab Studio · 2026
The method behind the Trail Calculator. Licensed CC BY 4.0.

This document defines, reproducibly, the physical steering-trail model used by our calculator. It is not an explainer: it is the method record — the derivation, the formulas, the dynamic model with its sign, and how it was validated. Anyone can reproduce every figure.

1. Convention

The head angle θ is measured from the horizontal (the geometry-chart convention: 64° is slacker than 73°). 2D model in the central plane, rigid frame, point contact, trail at zero steering. It is the same convention manufacturers use to compute the trail in their charts, so results are directly comparable.

2. Closed form

ground trail = (R·cos θ − offset) / sin θ mechanical trail = R·cos θ − offset wheel flop = ground trail · sin θ · cos θ

with R = wheel radius to the ground and offset = fork rake. Mechanical trail is the perpendicular distance to the steering axis —the one that generates the self-centering torque—; ground trail is what appears in charts. The wheel-flop factor follows Tony Foale's convention (Motorcycle Handling and Chassis Design). Reference case (enduro: R 370, offset 44, θ 64°): ground trail 131.5 mm, mechanical 118.2 mm, flop 51.8.

3. Radius: not invented

Radius takes two labeled modes. [MEASURED]: from the real rolled circumference, R = C/2π. [ESTIMATED]: R = BSD/2 + width, with the ISO bead-seat diameter (622 for 29"/700c, 584 for 27.5", 559 for 26") and tire height approximated by width. An estimate is never presented as a measurement.

4. Analytical sensitivities

The exact derivatives of ground trail (the citable asset; they yield reproducible magnet figures):

∂trail/∂offset = −1 / sin θ ∂trail/∂R = cot θ ∂trail/∂θ = (offset·cos θ − R) / sin²θ (per radian; ×π/180 per degree) ∂trail/∂D = ½ cot θ (per mm of diameter)

In the reference case: +1° slacker → +7.6 mm of trail; +7 mm of offset → −7.8 mm; +10 mm of diameter → +4.9 mm; going from 27.5" to 29" (Δdiameter 38 mm, same width) → +9.3 mm.

5. Dynamic model (with the corrected sign)

When the fork compresses the A2C (axle-to-crown) drops; the front falls, the head angle steepens (θ increases) and, since ∂trail/∂θ < 0, the trail drops. This is the opposite direction to overforking (a longer fork, larger A2C, makes the bike slacker).

Δθ ≈ arctan( ΔA2C·sin θ / wheelbase ) (steeper, +)
ScenarioθTrailReading
Front compression (braking)64° → 66.0°131.5 → 116.7 (Δ −14.8)steering gets nervous when you grab the brakes
Balanced SAG (seated, full)64° → 63.8°131.5 → 133.2 (Δ +1.7)≈ no change: seated ≈ chart

At balanced SAG the front (steepens) and the rear (slackens) nearly cancel: the net change is negligible. The dramatic number lives in the front-only transient (braking), and points to steeper / less trail.

6. Validation

The model was validated against an independent numerical oracle implemented from scratch. Two checks:

Derivative self-check
Each analytical derivative is compared with its finite difference (±1e-5 perturbation). They match to three decimals → the derivatives carry no typos or crossed signs (∂trail/∂θ = −7.58 mm/° by both routes).
Sign guard
Permanent assertion: under front compression, θ must increase and trail must drop. If an implementation inverts the dynamic sign, the assertion fails. The calculator ports this same test and self-verifies on load.

The linearization of the dynamic term (arctan ≈ x) is exact up to ~3° of change; the error versus a full trigonometric recompute is on the order of 0.01 mm of trail across the entire realistic SAG range.

Use the modelTry the Trail Calculator or understand what you feel with each number in the geometry cluster.

License of the method and this document: CC BY 4.0 — use and cite it with attribution. © 2026 Carlos Eduardo Ravello Joo · BikeLab Studio.

BikeLab Studio · Trail model / Derivation and validation / Carlos Eduardo Ravello Joo · Trujillo, Peru