2. Vehicle body-in-white modeling for rollover

Rollover and roof crush simulation places demands on the body-in-white (BIW) model that differ substantially from frontal or side impact. The loading path is dominated by quasi-static roof intrusion, large plastic deformation of thin-walled pillars and roof rails, and stability of the occupant survival space. Because the roof structure is a shell-dominated assembly, the quality of the shell mesh — not the sophistication of the material card — usually determines whether the simulation predicts the correct collapse mode. This chapter covers the practical workflow from CAD geometry to a PAM-CRASH-ready BIW shell model, with concrete quality targets and verification checks.

Geometry preparation before any meshing

Start from the BIW CAD assembly in STEP or CATIA format. The goal of geometry preparation is to produce a clean, connected, midsurface-ready model. Work through the following in order:

  • Remove non-structural content. Delete trim, seals, wiring, fasteners, badges, and any part that does not carry roof crush load. Keep the roof panel, roof rails, A/B/C/D pillars, rockers, floor, cross members, and the header and bow structures.
  • Repair topology. Use the CAD healing tools in ANSA, HyperMesh, or SpaceClaim to close gaps, remove slivers, and fix duplicate surfaces. A common defect is a 0.3–1.0 mm gap between the roof rail and the pillar where they should be welded; this must be closed or bridged before midsurfacing.
  • Defeature. Suppress beads smaller than 3 mm depth, holes below 6 mm diameter, and fillets below 3 mm radius unless they are on a known crush initiator. Retain beads and stiffeners that are part of the load path.
  • Establish a consistent coordinate system. PAM-CRASH uses a right-handed system. Define X forward, Y lateral (left positive), Z vertical (up positive), with the origin at the vehicle centerline at the front axle. Document this convention and keep it identical across all subsystems.
Keep a "defeature log" listing every suppressed feature and its reason. When the roof crush result looks wrong, the first thing to check is whether a load-bearing bead was accidentally removed.

Midsurface extraction

BIW parts are thin-walled stampings, typically 0.6–2.0 mm thick. Shell elements are the correct idealization. Midsurface extraction converts a solid or double-sided surface model into a single surface at the mid-plane of the sheet.

In HyperMesh, use Geom > Midsurface with the automatic extraction method, then manually repair the result. In ANSA, use Mid-Surface > Offset. The practical rules are:

  • Extract midsurfaces part by part, not on the whole assembly at once. Assembly-level extraction produces tangled surfaces at joints.
  • After extraction, assign thickness. In PAM-CRASH this is done through the /SHELL property with THICK or through a /PART referencing a shell property set. Thickness must match the drawing nominal, not the CAD solid thickness, which often includes coating.
  • Check for overlapping or duplicated midsurfaces at flanges. Two surfaces occupying the same space will produce spurious contact forces.
  • Verify that the midsurface normal direction is consistent. In PAM-CRASH, shell normals affect contact and hourglass control. Use Orient to make normals point outward from the closed section.

A useful sanity check: for a closed section such as the B-pillar, the extracted midsurfaces should form a closed loop when viewed in section. If there is a gap, the section will not carry bending correctly.

Mesh quality targets for roof crush

Roof crush is a bending-dominated problem. Element quality directly controls the predicted collapse load. The following targets are appropriate for a BIW model intended for PAM-CRASH roof crush per FMVSS 216a or ECE R66-style loading:

MetricTargetAcceptable limitWhy it matters
Element size (pillars, rails)5 mm3–8 mmResolves plastic hinge length
Element size (roof panel, floor)8–10 mm≤ 12 mmReduces cost without losing mode
Aspect ratio< 3< 5Bending accuracy
Warpage< 10°< 15°Prevents artificial stiffening
Jacobian> 0.7> 0.6Integration accuracy
Quad interior angle45°–135°30°–150°Shear locking
Trias in load path< 5%< 10%Trias are stiffer in bending
Min. length> 2 mm> 1.5 mmStable time step

Element size also sets the explicit time step. For steel with E = 210 GPa, ρ = 7.85e-9 t/mm³, the stable step for a 5 mm shell is approximately:

dt = L * sqrt(rho / E)
   = 5 * sqrt(7.85e-9 / 210000)
   ≈ 9.7e-7 s  (0.97 µs)

A 3 mm element reduces this to about 0.58 µs, roughly doubling runtime. This is the main reason not to over-refine the roof panel.

Connections and the load path

Roof crush load enters through the roof rail and pillars. The connections between these parts must be modeled explicitly:

  • Spot welds: use /SPOTWELD with a rigid or plastic beam. Place one weld per 25–40 mm of flange, matching the actual weld schedule. Do not merge nodes across flanges — this over-stiffens the joint.
  • Structural adhesive: model with solid or shell elements with a cohesive or elastic-plastic material. Adhesive in the roof rail is common in modern BIW and significantly raises collapse load.
  • Hemmed flanges: represent with a single row of shell elements sharing nodes, or with a rigid connection if the hem is not in the crush zone.
A frequent error is to merge coincident nodes at the pillar-to-roof-rail joint. This creates a rigid joint that does not rotate, and the roof crush force will be over-predicted by 15–30%. Always check the joint with a section plot before running.

Verification before the solver run

Run these checks in order and fix failures before submitting to PAM-CRASH:

  1. Free edges and T-connections: every free edge in the load path should be intentional. Use the free-edge finder and review each one.
  2. Duplicate elements: remove them; they cause contact instability.
  3. Normals: consistent and outward.
  4. Thickness assignment: every shell property has a non-zero thickness.
  5. Mass check: total BIW mass should be within 2% of the physical body mass. A large deviation indicates missing or duplicated parts.
  6. Penetration check: run a contact initialization and confirm no initial penetrations above 0.5 mm.

The SVG below summarizes the workflow and the decision points where the model is either accepted or sent back for repair.

BIW Modeling Workflow for PAM-CRASH Roof Crush 1. CAD Preparation Remove trim, seals, fasteners Heal gaps, defeature < 3 mm 2. Midsurface Extract per part Assign nominal thickness 3. Shell Meshing 5 mm pillars / 8–10 mm panels Quads, aspect < 3 4. Connections Spot welds, adhesive No node merging Quality Gate — Mesh Checks Jacobian > 0.7 · Warpage < 10° · Trias < 5% · Min length > 2 mm Pass? Mass ±2%, no penetration Repair & remesh Pass PAM-CRASH Ready /SHELL, /PART, /SPOTWELD Contact, material cards Key parameters carried into the solver • Shell thickness from drawing nominal (0.6–2.0 mm typical BIW) • Stable time step ≈ 0.97 µs for 5 mm steel shell — controls runtime • Spot weld spacing 25–40 mm along flanges; adhesive modeled separately • Coordinate system: X forward, Y left, Z up, origin at front axle centerline

The model is now ready for the material and contact definitions covered in the next stage of the workflow. The critical discipline at this stage is to treat mesh quality as a load-path property, not a cosmetic one: a 5 mm quad with a Jacobian of 0.75 in the B-pillar will predict a plastic hinge location that a 12 mm warped element will miss entirely.