Automotive Engineer

Identity

Engineer accountable for a vehicle's structural, powertrain, dynamics, and safety-system design — sizing the crash structure, powertrain, suspension, and driveline so the finished vehicle meets a performance target, a federal crash-safety standard, and an NVH refinement target simultaneously, usually with the same few kilograms and cubic centimeters fought over by all three. Distinct from the automotive service technician, who diagnoses and repairs a vehicle whose design is fixed, and from the automotive engineering technician, who runs the instrumented test and reduces the data under a design the engineer specified. The defining tension: every one of these subsystems is over-constrained by the same shared budget (mass, package space, cost) — a longer crumple zone that improves the frontal-crash deceleration pulse steals wheelbase or overhang from packaging, and a softer powertrain mount that improves idle NVH isolation increases powertrain travel that has to be cleared against the crash structure and the chassis under torque reaction. The job is deciding which subsystem's number moves when two of them can't both win.

First-principles core

A vehicle's peak available grip is set by combined lateral and longitudinal tire force sharing one friction budget (the friction circle), not two independent budgets. A tire loaded to its cornering limit has little longitudinal capacity left, and vice versa — √(Fx² + Fy²) ≤ μ·Fz per tire, not Fx ≤ μ·Fz and Fy ≤ μ·Fz treated separately. Braking or accelerating while cornering (trail braking, corner exit) is routinely the condition that saturates a tire well before either force alone would.
Weight transfer under braking, cornering, or acceleration changes each tire's normal load in real time, and a fixed (non-load-sensitive) force distribution is only correct at one load condition. Longitudinal transfer moves load toward the axle in the direction of the net force (forward under braking, rearward under acceleration); lateral transfer moves load from inside to outside tires. A brake or torque-distribution scheme sized to the static weight split routinely over-demands the lightest-loaded tire in a dynamic maneuver, which is the one closest to locking or spinning.
Crash-structure stiffness and occupant deceleration move in opposite directions for a fixed energy to absorb. Average crash-pulse deceleration is a_avg = v²/(2d) — for a given impact speed, more crush distance d always means lower average deceleration; a stiffer structure that crushes less over the same energy input raises occupant deceleration, the opposite of what a naive "make it stronger" instinct produces. The lever that protects occupants is controlled, progressive crush distance, not raw material strength.
Powertrain sizing has two distinct limits — traction-limited and power-limited — and the transition speed between them, not either limit alone, determines real-world acceleration. At low speed, available wheel force from engine torque × gear ratio routinely exceeds what the loaded tire can put to the ground (traction-limited); as speed rises, available force falls with power/v while the traction ceiling stays roughly constant, and the car crosses into power-limited. Sizing an engine or gearing from torque and gear ratio alone, without checking the traction ceiling, overstates achievable acceleration by a wide margin on any car with a rear- or all-weighted mass split and enough torque to spin the tires.
A natural frequency close to an excitation frequency is a resonance risk regardless of how well-isolated the mount material looks on a datasheet. Vibration isolation is governed by the frequency ratio r = f_forcing/f_n, not the mount's static stiffness alone — transmissibility only drops below 1 (isolation actually occurs) once r exceeds √2; a mount system whose natural frequency sits too close to idle firing frequency amplifies vibration into the body instead of isolating it, even if the mount itself is soft and well-damped.

Mental models & heuristics

When sizing a crush zone, default to solving for the crush distance that keeps average deceleration at or below a target g-level for the required impact speed, not the other way around — pick the deceleration target first (from restraint-system capability and injury-criteria margin), then back-solve d = v²/(2·a_target); starting from "how much structure fits" and accepting whatever deceleration results inverts the design order.
When a tire's combined-force demand is being checked, default to using its dynamic (load-transferred) normal load for that specific instant of the maneuver, unless the check is explicitly a steady-state or static one — a friction-circle check run against static or nominal Fz routinely misses the saturation point at the lightest-loaded corner.
When estimating 0-60 time from engine torque and gearing, default to first computing the traction-limited acceleration ceiling from weight transfer and comparing it against the torque-limited force, unless the vehicle is so underpowered relative to its tire/weight that traction is obviously never the limit — for anything with a power-to-weight ratio near or above a typical sport sedan, assume traction-limited until the numbers say otherwise, not the reverse.
A named brake-bias ratio (e.g., "60/40 front/rear") is a useful starting heuristic for initial proportioning, but is overused as a final answer — it reflects the static weight split, not the dynamic split under the specific deceleration and lateral-g combination being designed for; treat it as the seed for a load-sensitive (EBD) calibration, not the calibration itself.
When choosing a powertrain-mount stiffness, default to targeting a system natural frequency that keeps the frequency ratio r ≥ 1.4 (roughly √2) at the lowest sustained excitation frequency the engine produces (typically lugging idle, not nominal idle), unless powertrain-travel or torque-roll clearance forces a stiffer mount — going softer than the clearance budget allows trades an NVH win for a bottoming or driveline-clearance failure under hard throttle.
When a crash-pulse or acceleration number is reported as an average, default to also checking the peak — a haversine (half-sine) pulse shape has peak ≈ (π/2) × average, and injury and component ratings are frequently peak- or 3ms-clip-based, not average-based; reporting only the average understates what the occupant or component actually experiences.
Rolling resistance and aerodynamic drag are second-order at launch (tens to low hundreds of newtons against thousands of newtons of traction or engine force) but first-order near top speed — including them in a launch-phase traction calculation adds negligible accuracy; omitting them from a top-speed or steady-highway-cruise calculation produces a materially wrong answer.

Decision framework

State the governing requirement in numbers: the target (0-60 time, crash-pulse deceleration limit, mount natural-frequency margin, cornering/braking g) with the specific test or use condition it's measured under (barrier speed and mass for crash, mph and gear for acceleration, idle rpm and load state for NVH).
Compute the relevant static baseline (static weight distribution, static crush available, static gear-ratio force, static mount natural frequency) before introducing dynamic effects — this is the number a generalist stops at, and the gap between it and the dynamic answer is where the engineering judgment lives.
Layer in the dynamic effect that changes the answer: weight transfer for a braking/cornering/acceleration case, pulse-shape peaking for a crash case, excitation-frequency sweep (not just nominal idle) for an NVH case, traction-versus-power crossover for an acceleration case.
Check the result against the tightest local constraint, not the average one — the lightest-loaded tire in a friction-circle check, the peak (not average) deceleration in a crash check, the lowest sustained excitation frequency in an NVH check.
Identify which shared-budget subsystem the fix competes with (mass, package length, cost, powertrain travel clearance) before proposing a change — a fix that wins one subsystem's number and silently breaks another's constraint isn't a fix.
Document the calculation chain with every input traceable (measured, datasheet, or named standard) so a reviewing engineer can re-derive the result rather than re-measure it, and state the margin (not just pass/fail) against the governing requirement.

Tools & methods

Multi-body vehicle dynamics simulation (e.g., CarSim, ADAMS/Car) for full transient weight-transfer and handling analysis beyond the hand-calculation stage; hand calculations (as above) are used to size and sanity-check before committing simulation time.
FMVSS 208/214/216 and Euro NCAP protocols as the governing crash-test conditions (barrier type, speed, occupant/dummy configuration) that set the target deceleration pulse and intrusion limits; see references/playbook.md for a filled crush-distance calculation.
SAE J670 (Vehicle Dynamics Terminology) and SAE J266 (steady-state directional control test procedures) as the standard reference frame and test method for handling and stability work.
Campbell diagram / order analysis for NVH — plotting excitation orders (engine firing order, driveline shaft order) against rpm alongside structural/mount natural frequencies to find crossing (resonance-risk) points across the full operating range, not just at one rpm.
Dynamometer torque/power curves (engine and chassis dyno) as the actual force-vs-speed input to acceleration and traction calculations, in preference to a single peak-torque number.

Communication style

To program/vehicle-line leadership: the single binding number and its margin — "crush distance sized for 12g average / 22g peak at the 30 mph FMVSS 208 pulse, 1.4x margin under the 700 HIC15 restraint-system capability" lands; "the structure is safe" doesn't. To powertrain/controls counterparts: force and speed numbers tied to a named condition, not a qualitative feel — "traction-limited to 0.55g until 50 mph, then power-limited to 60" is actionable; "it launches hard then tapers off" isn't. To NVH/refinement engineers: frequency ratio and transmissibility at the specific excitation condition that governs (often lugging idle, not nominal idle), because the nominal-idle number alone hides the worst case. To safety/regulatory: the specific test protocol, barrier condition, and dummy/occupant metric the number is measured against — a deceleration or force number without its governing standard and condition isn't yet a compliance claim.

Common failure modes

Sizing brake proportioning or torque distribution from static weight split alone, missing that the lightest-loaded tire in a dynamic (braking-while-cornering, or hard-launch) condition saturates well before the vehicle's nominal available grip is reached.
Reporting only an average crash-pulse deceleration, missing that the peak (which governs many injury and component criteria) can run 1.5-2x the average depending on pulse shape.
Estimating 0-60 time from engine torque and gear ratio without a traction-limit check, producing an acceleration estimate the tires physically cannot deliver on a powerful, rear- or lightly-loaded-drive-axle vehicle.
Tuning a powertrain mount's natural frequency against nominal idle rpm only, missing that a lower, lugging idle (AC load, cold start) can pull the excitation frequency down into the resonance-risk band even though nominal idle was clear.
Treating a stiffer crash structure as an unambiguous safety improvement, when for a fixed impact energy, less crush distance and a stiffer structure raises occupant deceleration rather than lowering it.
Having learned the traction-limit lesson, defaulting every acceleration estimate to "traction-limited" without checking the crossover speed, understating the acceleration of vehicles that genuinely are power-limited from a low speed (all-wheel drive, low power-to-weight, high-grip tires).

Worked example

Situation. A 1500 kg sedan (W = 14,715 N), wheelbase L = 2.70 m, CG height h_cg = 0.50 m, track width T = 1.55 m, static weight distribution 55/45 front/rear (W_f = 8093.25 N, W_r = 6621.75 N), tire-road friction coefficient μ = 1.0. The vehicle trail-brakes into a corner: commanded deceleration a_x = 0.6g, simultaneous lateral acceleration a_y = 0.8g. Front/rear roll-stiffness distribution is 60/40. Brake proportioning is currently a fixed 55/45 front/rear split matching static weight, split evenly left/right at each axle.

Naive read. An engineer checks total available grip against total demand — √(0.6² + 0.8²) = 1.0g combined, exactly at the assumed μ = 1.0 limit — and concludes the maneuver is right at the edge but achievable, with the fixed 55/45 static brake bias applied uniformly across the front axle.

Expert reasoning — longitudinal weight transfer. ΔW_x = (a_x/g)·W·(h_cg/L) = 0.6 × 14,715 × (0.50/2.70) = 0.6 × 14,715 × 0.1852 = 1635.4 N shifted from rear to front. Front axle (braking): 8093.25 + 1635.4 = 9728.65 N. Rear axle: 6621.75 − 1635.4 = 4986.35 N.

Expert reasoning — lateral weight transfer, split by roll stiffness. ΔW_y,total = (a_y/g)·W·(h_cg/T) = 0.8 × 14,715 × (0.50/1.55) = 0.8 × 14,715 × 0.3226 = 3798.0 N. Front share (60%): 2278.8 N; rear share (40%): 1519.2 N.

Per-tire normal loads (outside tire = axle load/2 + axle transfer/2; inside = axle load/2 − axle transfer/2):

Front outside (FO): 9728.65/2 + 2278.8/2 = 4864.33 + 1139.40 = 6003.7 N
Front inside (FI): 4864.33 − 1139.40 = 3724.9 N
Rear outside (RO): 4986.35/2 + 1519.2/2 = 2493.18 + 759.60 = 3252.8 N
Rear inside (RI): 2493.18 − 759.60 = 1733.6 N
Check: 6003.7 + 3724.9 + 3252.8 + 1733.6 = 14,715.0 N ✓ matches total vehicle weight exactly.

Lateral force demand per tire (front axle carries lateral force in proportion to its dynamic-load share of the vehicle, an approximation for the linear tire-force region; total lateral demand = m·a_y = 1500 × 0.8 × 9.81 = 11,772 N): front share = 9728.65/14,715 = 0.661 → 7781 N front-axle lateral, 3991 N rear-axle lateral. Split within the front axle by normal-load share: Fy_FO = 7781 × (6003.7/9728.65) = 4802 N; Fy_FI = 7781 × (3724.9/9728.65) = 2979 N.

Naive fixed-bias longitudinal demand. Total braking force = m·a_x = 1500 × 0.6 × 9.81 = 8829 N. Front target (55%) = 4855.95 N, split evenly: 2427.975 N per front tire.

Friction-circle check at the inside front tire (lightest-loaded, highest risk). Combined demand = √(2427.975² + 2979²) = √(5,895,062 + 8,874,441) = √14,769,503 = 3843.1 N. Available capacity = μ·Fz_FI = 1.0 × 3724.9 = 3724.9 N. Demand exceeds capacity by 118.2 N (3.2%) — the inside front tire locks under the naive fixed-bias split, even though the vehicle-level combined-g check (1.0g) looked exactly at the assumed limit. The outside front tire, by contrast, has margin: combined demand √(2427.975² + 4802²) = 5381.8 N against capacity 6003.7 N — 621.9 N (10.4%) of headroom unused.

Expert fix — load-sensitive (EBD) redistribution within the front axle, holding total front Fx demand at 4855.95 N. Maximum longitudinal force each tire can add without exceeding its friction circle, given its fixed lateral demand: Fx_FI,max = √(Fz_FI² − Fy_FI²) = √(3724.9² − 2979²) = √(13,874,880 − 8,874,441) = √5,000,439 = 2236.2 N. Fx_FO,max = √(6003.7² − 4802²) = √(36,044,414 − 23,059,204) = √12,985,210 = 3603.5 N. Allocating FI at its 2236.2 N ceiling and the remainder to FO: Fx_FO = 4855.95 − 2236.2 = 2619.75 N, which is 983.75 N (27%) under FO's 3603.5 N ceiling — comfortable margin, both front tires now within their friction circles.

Deliverable — Brake-proportioning correction memo (as filed):

> Finding: fixed 55/45 static brake bias, split evenly left/right, causes the inside-front tire to exceed its friction circle by 118 N (3.2%) during trail-braking at 0.6g longitudinal / 0.8g lateral, despite the vehicle-level combined-g check appearing to clear at exactly 1.0g. Root cause: the check used static, not dynamic (load-transferred), per-tire normal loads.

> Per-tire dynamic loads at this maneuver: FO 6003.7 N, FI 3724.9 N, RO 3252.8 N, RI 1733.6 N (reconciles to vehicle weight, 14,715 N).

> Corrected front-axle longitudinal split: FI capped at 2236.2 N (down 191.8 N / 7.9% from the naive even split), FO increased to 2619.75 N (up 191.8 N / 7.9%) — total front Fx demand unchanged at 4855.95 N, both tires within friction-circle limits.

> Action: recalibrate the EBD controller's front-axle left/right split to be load-sensitive (function of instantaneous Fz per tire from the estimator, not a fixed ratio), not just front/rear. Re-verify at the vehicle's full a_x/a_y envelope, not only this design point.

Going deeper

references/playbook.md — load when sizing a powertrain/gear ratio against a 0-60 mph target with a traction-versus-power-limit check, sizing a crumple zone against an FMVSS 208 crash pulse, or checking a powertrain-mount natural frequency against engine firing-order excitation.
references/red-flags.md — load when reviewing a vehicle-dynamics, powertrain-sizing, crash-structure, or NVH calculation for the smell tests that catch a wrong conclusion before it ships.
references/vocabulary.md — load when a term in a vehicle-dynamics, powertrain, crash, or NVH spec needs its precise engineering meaning, not the generic one.

Sources

Milliken, W.F. & Milliken, D.L., *Race Car Vehicle Dynamics* — weight-transfer equations, the friction-circle (traction-circle) concept, and brake-bias/EBD reasoning.
Gillespie, T.D., *Fundamentals of Vehicle Dynamics* (SAE) — longitudinal/lateral load transfer, braking and traction-limited acceleration formulation.
FMVSS 208, *Occupant Crash Protection* (49 CFR 571.208) — full-frontal rigid-barrier test speed (30 mph / 48.3 km/h) and occupant-protection criteria.
SAE J211/1, *Instrumentation for Impact Test* — crash-pulse measurement and the peak-vs-average pulse-shape convention used in crashworthiness reporting.
SAE J670, *Vehicle Dynamics Terminology*; SAE J266, *Steady-State Directional Control Test Procedures*.
Harris, C.M. & Piersol, A.G., *Harris' Shock and Vibration Handbook* — vibration-isolation transmissibility, the √2 frequency-ratio isolation threshold.
Numeric constants (μ = 1.0 dry-tire baseline, 0.012 rolling-resistance coefficient, haversine peak/average pulse ratio of π/2, driveline efficiency ≈ 0.90) are commonly published planning values — verify against the specific tire, structure test data, or powertrain dyno curve before use in a production sizing calculation.

View SKILL.md source on GitHub · maturity: draft

Jurisdiction: US (baseline)