3. Building the full vehicle CAE model

The quality of a full vehicle crash model is decided long before the first LS-DYNA or PAM-CRASH run is submitted. A PAM-CRASH full vehicle model for frontal ODB, side pole, or rear impact sign-off typically contains 8–15 million elements, 200–600 parts, and dozens of connections. If the geometry pipeline is sloppy, the solver will either terminate on negative volume or produce energy errors that invalidate the correlation. This chapter walks through the four-stage pipeline that turns a CAD assembly into a runnable, sign-off-grade model: CAD cleanup, mid-surfacing, meshing, and subsystem assembly.

Stage 1 — CAD Cleanup

OEM CAD is authored for manufacturing, not for shell meshing. Expect to find: overlapping surfaces at hem flanges, gaps of 0.3–2.0 mm at weld seams, sliver faces narrower than 1 mm, and duplicated surfaces from mirrored parts. The cleanup goal is a watertight, single-shell-per-part representation with no free edges inside a structural panel.

A practical workflow in ANSA or HyperMesh:

1. Topology check. Run Topo > Check with a tolerance of 0.1 mm. Flag free edges, T-connections, and duplicate faces. On a typical BIW, expect 3,000–8,000 free edges before cleanup.
2. Geometry repair. Use Topo > Fix to stitch gaps under 0.5 mm. For larger gaps at rail-to-apron joints, manually extend surfaces rather than increasing tolerance — a 2 mm stitch tolerance will collapse thin flanges.
3. Feature removal. Delete beads smaller than 3 mm depth, embosses, and holes under 6 mm diameter. These features do not affect crash energy absorption but multiply element count and create poor aspect ratios.
4. Flange preparation. Extend all spot-weld flanges to a common mid-plane so that the two mating shells can be connected by a single weld entity without offset.

Stage 2 — Mid-Surfacing

Crash structures are thin-walled: 0.6–2.5 mm for BIW panels, 1.0–3.0 mm for closures, 2.0–6.0 mm for chassis castings. Shell elements are the standard representation. Mid-surfacing extracts the neutral plane of each sheet-metal part.

Two approaches dominate:

• Automatic mid-surfacing (ANSA Mid-Surface, HyperMesh Midsurface): fast, but produces offset errors of 0.2–0.8 mm at curved regions and often fails on castings.
• Manual/assisted mid-surfacing: required for A-pillar, B-pillar, rocker, and any part with a closed hollow section. The engineer offsets the outer surface inward by half the local thickness.

Thickness assignment is critical. Each property must carry the measured or nominal gauge. For a typical steel BIW:

For aluminum closures and castings, mid-surfacing is often replaced by solid or thick-shell representation. A common rule: if the wall thickness exceeds 4 mm or the part is a casting with variable section, use tetrahedral solids with at least three elements through the thickness.

Stage 3 — Meshing Strategy

Element size is a trade-off between accuracy and cost. For full vehicle sign-off, the industry baseline is:

• Front structure (rails, bumper, crash box): 5 mm target, 3 mm in the crush zone.
• BIW body panels: 8–10 mm.
• Closures (doors, hood, tailgate): 10 mm.
• Chassis / subframe: 5–8 mm.
• Powertrain and battery pack: 10–20 mm with rigid or simplified representation.

Quality targets for shell elements:

Use structured mesh on rails and pillars where possible — mapped quad meshing along the part axis gives better hourglass control and cleaner fold patterns. Free meshing is acceptable on floor pans and closures. Always run Quality > Check after each part and fix failures before assembly; a single bad element in the crush zone can trigger negative volume at 20 ms.

Stage 4 — Subsystem Assembly

Assembly is where most models fail. The order matters:

1. Body-in-white (BIW). Import all cleaned, meshed panels. Connect spot welds using *CONSTRAINED_SPOTWELD (LS-DYNA) or PAM-CRASH PLINK entities. Weld spacing should match the CAD weld map — typically 25–40 mm on structural joints, 50–60 mm on closures.
2. Closures. Attach doors, hood, and tailgate via hinge and latch representations. Hinges are usually modeled as rigid bodies with a revolute joint; latches as a tied contact or a rigid constraint that releases at a defined force.
3. Chassis subsystems. Subframe, suspension arms, and steering rack are connected to the BIW via bushings and ball joints. Use *MAT_SPRING_ELASTIC or a discrete beam with calibrated stiffness. Do not rigidly tie the subframe — this over-stiffens the front end and suppresses rail folding.
4. Powertrain and battery. Represent as a rigid body with correct mass and CG, connected via engine mounts (spring elements) and, for BEVs, a battery pack tied to the floor with a calibrated number of connections.

Verification Before the First Run

Before submitting the model, run these checks:

• Mass check: total model mass must be within 2% of the physical vehicle curb mass. A 1,650 kg sedan should produce a model mass of 1,617–1,683 kg.
• Free-free modal check: first torsional mode of the BIW should fall in the 25–40 Hz range for a typical sedan. A value below 20 Hz indicates missing connections or incorrect thickness.
• Penetration check: run a 5 ms gravity settle. Any initial penetration above 0.5 mm will cause contact instability.
• Energy check: initial kinetic energy must match ½mv² for the prescribed velocity. For a 64 km/h ODB test with a 1,650 kg vehicle, KE ≈ 260 kJ.

The pipeline above is iterative, not linear. A mass shortfall discovered at the verification gate usually traces back to missing thickness on a mid-surfaced panel or an omitted subsystem. A modal frequency below 20 Hz almost always means spot welds were merged instead of constrained, or a closure was rigidly tied. Treat each verification failure as a signal to revisit the specific upstream stage rather than patching the symptom in the assembly.

Once the model passes all four gates, it is ready for the first explicit run. The next chapter covers boundary conditions, initial velocity, contact definitions, and the control card settings that govern a stable, sign-off-grade PAM-CRASH simulation.