4. Vehicle body-in-white geometry preparation for side impact

Side impact simulation quality is decided long before the first LS-DYNA run. The barrier-to-door-to-occupant load path passes through a small number of thin-walled stamped parts, and any geometric defect in those parts — a missing flange, a duplicated surface, a hole that should have been filled, or a midsurface offset by 1.5 mm — propagates directly into contact stiffness, intrusion velocity, and finally into the injury criteria you report. This chapter covers the geometry preparation chain for the four most critical side-structure assemblies: the doors, the B-pillars, the rockers (sill), and the floor. The goal is a clean, watertight, properly connected shell mesh that runs stably at the target time step.

4.1 Reference Frame and Modeling Conventions

Fix the coordinate convention before touching geometry, because every offset, projection, and contact thickness depends on it. Use the vehicle global frame: X positive forward (longitudinal), Y positive to the left side of the vehicle, Z positive upward. The side impact barrier approaches the driver-side (left) door along −Y. Shell normals must point consistently outward from the enclosed sheet-metal volume; LS-DYNA uses the normal direction for single-surface contact and for the sign of the offset in *SECTION_SHELL. A mixed-normal mesh on a rocker will produce spurious penetrations and negative volume errors within the first few milliseconds.

Adopt a single unit system throughout — the standard for crash work is mm, ton, s, N, MPa. In this system, steel density is 7.85e-9 ton/mm³, Young's modulus is 210000 MPa, and a 1.2 mm shell with a 5 mm characteristic length gives a stable time step near 0.7 µs. Mixing units here is the single most common cause of an "exploding" model.

4.2 CAD Cleanup

Start from the OEM-supplied assembly, usually a CATIA or NX file with the full BIW. Isolate the side structure and delete non-structural content: seats, trim, wiring, glass, and fasteners. What remains for this chapter is the door inner and outer panels, the door intrusion beam, the B-pillar inner and outer, the rocker inner and outer, and the floor pan with its cross members and tunnel.

Cleanup is a sequence of defect removal operations, and the order matters:

  1. Heal gaps and slivers. Stamped parts often carry 0.01–0.05 mm gaps between adjacent faces from the export. Use the CAD heal function with a tolerance of 0.05 mm. Gaps larger than this are real design gaps and must not be closed.
  2. Remove duplicate and coincident faces. Overlapping surfaces from mirrored or patterned features create double shells that double the local stiffness. Delete one of each pair.
  3. Fill or retain holes deliberately. Weld-access holes, drain holes, and locating holes smaller than roughly 8 mm should be filled — they force tiny elements and collapse the time step. Holes larger than 20 mm that carry structural or crash significance (for example, the B-pillar belt routing slot) must be retained and meshed with a washer ring.
  4. Repair fillets and flanges. Small radii below 3 mm on the rocker and B-pillar should be removed or replaced with a sharp edge, because a midsurface cannot be extracted reliably from a tight radius and the resulting mesh will be degenerate.
  5. Check for free edges and non-manifold edges. Every free edge on a structural panel is either a real trim boundary or a defect. Run an edge-check and resolve each one.
Do not skip the edge check. A single unhealed free edge on the B-pillar outer will later appear as a contact leak, allowing the barrier to pass through the pillar without resistance and producing unrealistically low intrusion.

4.3 Midsurface Extraction

Stamped BIW parts are thin shells, so the solid CAD must be reduced to a midsurface. The extraction tolerance controls how faithfully the midsurface follows the original wall. A practical setting for automotive steel is 0.1 mm; looser tolerances (0.5 mm) lose the subtle beads and stiffening ribs that dominate side-impact performance.

After extraction, verify the thickness attribute on every surface. The midsurface tool reads the original wall thickness, but it frequently misreads at flanges, at hemmed door edges, and where two panels overlap. For the door outer, the B-pillar outer, and the rocker outer, confirm the nominal values — typically 0.7–0.8 mm for door outer, 1.2–1.6 mm for B-pillar, 1.4–2.0 mm for rocker, and 0.7–1.0 mm for floor. Where the CAD thickness is wrong, override it manually and record the value in the part table.

Two extraction problems deserve special attention on side structures:

  • Door hem flange. The outer panel is folded over the inner panel at the door perimeter. Midsurfacing this hem produces two nearly coincident surfaces. Model the hem as a single shell with the combined thickness, or as a short rigid connection — do not leave two overlapping shells.
  • Tailored blanks and patch reinforcements. Where the B-pillar or rocker uses a tailored blank, the thickness changes across a weld line. Split the midsurface at the weld line and assign each region its own thickness in *PART / *SECTION_SHELL.

4.4 Shell Mesh Generation

Mesh the four assemblies with a target element size of 5 mm on the main panels and 3 mm in the B-pillar and rocker impact zones. Use quad-dominant meshing with a maximum of 5% triangular elements; triangles are stiffer in bending and should be confined to transition regions. Maintain an aspect ratio below 3 and a warpage below 15°.

Mesh quality targets for this chapter:

MetricTargetReject if
Element size (main panels)5 mm< 3 mm or > 8 mm
Element size (impact zone)3 mm< 2 mm
Aspect ratio< 3> 5
Warpage< 15°> 20°
Jacobian> 0.7< 0.6
Triangles< 5%> 10%

Connectivity between the four assemblies is what makes the model behave as a structure rather than a collection of panels. Use these connection types:

  • Spot welds between door inner and B-pillar, B-pillar and rocker, and rocker and floor: model with *CONSTRAINED_SPOTWELD or a short beam, placed at the CAD weld locations. Typical spacing is 25–40 mm.
  • Adhesive / structural bond along the door outer-to-inner hem and along the rocker-to-floor seam: model with *CONSTRAINED_INTERPOLATION or a solid adhesive layer if the bond is thick.
  • Continuous seams at the floor tunnel and cross-member joints: use shared nodes or *CONSTRAINED_NODAL_RIGID_BODY.

After meshing, run a connectivity audit: count free edges, check that every weld has two coincident nodes within tolerance, and confirm that no element has zero or negative Jacobian. Then run a 1 ms trial with the barrier at low velocity to confirm the model initializes without negative volume or contact errors.

Side-Structure Geometry Preparation Chain 1. CAD Cleanup Heal gaps ≤ 0.05 mm Delete duplicate faces Fill holes < 8 mm Resolve free edges 2. Midsurface Tolerance 0.1 mm Verify thickness Handle hem flange Split tailored blanks 3. Shell Mesh 5 mm / 3 mm zones Quad-dominant Jacobian > 0.7 Warpage < 15° 4. Connections Spotwelds 25–40 mm Adhesive seams Shared nodes Connectivity audit Critical Side-Structure Parts and Their Mesh Parameters Door (inner + outer) t = 0.7–0.8 mm Mesh 5 mm Hem → single shell B-Pillar (inner + outer) t = 1.2–1.6 mm Mesh 3 mm (impact) Retain belt slot Rocker / Sill t = 1.4–2.0 mm Mesh 3 mm Remove radii < 3 mm Floor Pan t = 0.7–1.0 mm Mesh 5 mm Seam at tunnel Side-Impact Load Path Through Prepared Geometry Barrier −Y direction Door Outer Intrusion beam B-Pillar Bending + shear Rocker Axial crush Floor Diffusion Units: mm, ton, s, N, MPa | Steel: ρ = 7.85e-9 ton/mm³, E = 210000 MPa | Target Δt ≈ 0.7 µs at 5 mm / 1.2 mm

4.5 Verification Before Proceeding

Before the geometry is handed to the load-case setup, run this verification sequence and record the results:

  1. Free-edge count. Should be zero on all structural panels except at intentional trim boundaries. Any unexpected free edge is a defect.
  2. Normal consistency. Display shell normals; all should point outward. Flip any that do not.
  3. Thickness audit. Compare the assigned thickness on every part against the CAD nominal. Flag any deviation greater than 0.05 mm.
  4. Weld audit. Confirm each spotweld connects two nodes within 0.5 mm and that no weld is orphaned.
  5. Time-step check. Run *CONTROL_TIMESTEP with the initial time step and confirm the smallest element does not drive Δt below 0.5 µs. If it does, locate the offending element and coarsen or remove it.
  6. 1 ms initialization run. Launch the model with the barrier at 1 m/s. The run should complete without negative volume, without contact errors, and with a stable energy balance.
Keep a part table alongside the mesh: part name, material, thickness, element size, and connection type. This table becomes the reference for every subsequent load case and is the fastest way to diagnose a stiffness anomaly later.

The geometry prepared here — clean midsurfaces, consistent normals, verified thicknesses, and audited connections — is the foundation on which the side-impact load case is built. A defect that survives this stage will not disappear in the solver; it will simply reappear as an unexplained intrusion or a contact failure that costs far more time to trace than the cleanup would have taken.

← Previous 📖 Contents Next →