3. Crashworthiness simulation workflow

A crashworthiness simulation is not a single action but a chain of decisions in which every stage constrains the next. The quality of a frontal offset or side impact prediction is largely determined before the solver is launched: by how the mesh captures spot welds and folded sheet metal, by how contacts are defined, and by whether the material card actually represents the strain-rate behaviour of the steel in question. This chapter walks through the three linked stages — model build, solver setup, and result interpretation — using a representative full-frontal rigid-barrier impact of a body-in-white (BIW) as the running example. All numerical values below are illustrative teaching parameters, not measured results.

Stage 1 — Model build: geometry, mesh, connections, and materials

The build stage begins with geometry cleanup. CAD assemblies contain overlapping flanges, small fillets, and fastener holes that are irrelevant to crash energy management but expensive to mesh. A practical rule is to remove features below roughly 5 mm in a 10 mm target element size model, while preserving every bead, rib, and fold line that carries load. Mid-surfacing is appropriate for thin-walled stampings; solid meshing is reserved for cast nodes, suspension knuckles, and battery enclosures where through-thickness stress matters.

Mesh quality in crash models is governed less by Jacobian ideals than by stability and contact behaviour. A common starting point for a BIW is a target element size of 8–10 mm on closures and rails, with 5 mm refinement in the front longitudinal members and hinge pillars where folding is expected. Hourglass control must be enabled for reduced-integration shells; without it, under-integrated elements can produce non-physical zero-energy modes that inflate intrusion predictions. The table below summarises typical build parameters for the example model.

Connections deserve particular attention. Spot welds modelled as rigid beams over-stiffen the joint and suppress the peeling failure that drives many real-world separations. Deformable beam welds with a defined failure criterion, or solid hexa-clusters tied to both sheets, give more realistic load transfer. Adhesive bonds and structural rivets require cohesive elements or tied contact with a damage law. Material cards should be sourced from coupon tests at multiple strain rates; a single quasi-static curve will under-predict the crash force of high-strength steels by a significant margin.

Stage 2 — Solver setup: contacts, boundary conditions, and control

With the mesh complete, the solver deck is assembled. For an explicit crash code the critical settings are contact definitions, initial velocity, rigid barrier, and termination control. Contacts should be defined as a global single-surface with a penalty formulation, plus dedicated contact pairs for the barrier-to-bumper and tyre-to-ground interfaces. A friction coefficient of 0.1–0.3 is typical for painted steel-on-steel; the exact value should be calibrated against a sled or component test rather than assumed.

*CONTACT
  global single surface, penalty, friction 0.15
  barrier-to-front-end pair, friction 0.30
*INITIAL_VELOCITY
  all nodes, 56 km/h along -X
*RIGIDWALL
  planar, fixed, normal +X, offset at bumper face
*CONTROL_TERMINATION
  end time 120 ms, timestep scale 0.9
*DATABASE_BINARY_D3PLOT
  output interval 2 ms

The initial velocity of 56 km/h (15.6 m/s) corresponds to a regulatory frontal impact; the direction convention here is that the vehicle travels along −X and the barrier normal points along +X. This sign convention must be stated once and held consistently through post-processing. The timestep is governed by the smallest element and the material sound speed; a scale factor of 0.9 provides margin against instability. Mass scaling should be used sparingly and audited — adding mass to the front rails to raise the timestep will change the deceleration pulse and invalidate the result.

Stage 3 — Result interpretation: reading the pulse, intrusion, and energy balance

Interpretation starts with the energy balance, not the pictures. In a valid explicit run the total energy should be approximately conserved, hourglass energy should stay below about 5% of internal energy, and added-mass energy should be negligible. If hourglass energy climbs through the event, the mesh or hourglass control is inadequate and the intrusion numbers are suspect.

The primary engineering outputs are the barrier force–time and vehicle deceleration pulse, the occupant compartment intrusion, and the deformation mode of the front structure. A well-designed frontal system produces a relatively flat force plateau: the longitudinal members fold progressively, the bumper beam spreads load, and the firewall intrusion stays within the target envelope. A sharp initial spike followed by a deep trough indicates an over-stiff first fold or a contact that engages too late. Intrusion is best read as a time history at defined measurement points — pedal box, toe board, steering column base — rather than as a single peak value.

The figure above shows the three stages and, importantly, the two feedback paths. Interpretation is not the end of the chain: a force spike traced to a late-engaging contact sends the engineer back to solver setup, while a fold mode that differs from the physical test sends them back to geometry or weld definition. The validation gate at the bottom is where simulation meets the physical crash test, and it is the subject of the chapters that follow.

In practice, the workflow is iterative and the boundaries between stages blur. A change to weld representation alters contact behaviour; a change to the material card alters the timestep. The discipline that keeps the process trustworthy is traceability: every result should be reproducible from a documented deck, with the build parameters, solver settings, and interpretation criteria recorded alongside it. That record is what allows a simulation to serve as evidence in a validation programme rather than as an illustration.