3. Side impact finite element model assembly

In a side impact simulation, the model is not a single monolithic mesh but an assembly of subsystems whose interfaces determine whether the load path from the moving deformable barrier (MDB) reaches the occupant correctly. The body-in-white (BIW) carries the structural load, the door trim and inner panel transfer that load into the occupant compartment, the seat positions the occupant and provides a secondary load path, and the occupant compartment mesh must be fine enough to capture intrusion without becoming computationally prohibitive. This section describes how to assemble these subsystems and how to choose meshing strategies for each.

Assembly hierarchy and coordinate conventions

Before meshing, fix the global coordinate system and keep it consistent across all subsystems. The standard LS-DYNA automotive convention is: X positive forward (vehicle longitudinal), Y positive to the left side of the vehicle, Z positive upward. Side impact is typically simulated with the barrier striking the driver or left side, so the impact direction is −Y (barrier moving in +Y toward the vehicle, or vehicle moving in −Y toward a fixed barrier, depending on the setup). State this convention explicitly in the model documentation, because a sign error in the barrier velocity or in the contact normal direction will silently produce a non-physical result.

Assemble the model as a set of include files, one per subsystem, tied together by a master keyword deck. A practical structure is:

*INCLUDE
biw_main.k
door_front_left.k
door_rear_left.k
seat_driver.k
occupant_dummy.k
barrier_mdb.k
*CONTROL_TERMINATION
  0.120
*CONTROL_TIMESTEP
  0.0  0.9
*DATABASE_BINARY_D3PLOT
  0.002

Each include file owns its own mesh, part definitions, and material cards. Contacts and constrained connections are defined in the master deck so that interface behavior is visible in one place rather than scattered across subsystem files.

Body-in-white meshing strategy

The BIW is the primary load path. For side impact, the critical regions are the B-pillar, rocker, door inner panel, roof rail, and floor cross members. Use shell elements (ELFORM 2, Belytschko-Tsay) for the stamped panels, with an in-plane target element size of 5–8 mm in the B-pillar and rocker, and 8–12 mm in less critical panels such as the roof and floor. The B-pillar should have at least four elements across its width so that bending and buckling modes are captured; a two-element-wide pillar will over-stiffen the section and under-predict intrusion.

Spot welds connect the stamped panels. Represent them with *CONSTRAINED_SPOTWELD or, for weld clusters that must fail, with a tied contact plus a failure criterion. A common practical choice is a beam or solid spot weld with a rupture strain, but for a first-pass model, rigid spot welds are acceptable if the study focuses on occupant kinematics rather than weld failure. Place weld nodes at the actual flange mid-surface; offsetting them by the sheet thickness introduces artificial moments.

Thickness and material assignment must match the drawing. Use *PART with ELFORM, SHRF, and NIP set per part, and assign *MAT_024 (piecewise linear plasticity) for mild steel and *MAT_037 or *MAT_036 for high-strength steel with strain-rate effects. The strain-rate parameter C and P in Cowper-Symonds matters in side impact because the loading is fast; omitting rate effects typically under-predicts the force transmitted to the occupant.

Door trim and inner panel

The door trim is the interface between the striking barrier and the occupant. Model the door outer panel, intrusion beam, inner panel, and trim substrate as separate parts. The trim substrate is usually plastic; use *MAT_024 with a low yield stress or *MAT_003 if you need failure. The trim should not be merged with the inner panel — keep them as separate shells with a contact between them, because the trim can separate and rotate during intrusion, and merging would suppress that behavior.

Mesh the intrusion beam with solid or thick shell elements and align the element edges with the beam axis. A misaligned mesh on a bending-dominated beam produces hourglassing; if you see hourglass energy exceeding 5% of internal energy in glstat, refine the beam mesh or switch to ELFORM 1 with hourglass control type 4 or 5.

Seat and occupant compartment

The seat frame is usually modeled with shell elements for the rails and a combination of shells and solids for the recliner and cushion pan. The foam cushion and backrest are modeled with solid elements using *MAT_057 (low-density foam) or *MAT_083 (Fu-Chang foam) with a load curve from a compression test. Foam element size of 10–15 mm is adequate; finer meshes increase cost without improving occupant response because the foam is already a homogenized representation.

The occupant compartment mesh must resolve the surfaces the dummy contacts: the door trim inner surface, the armrest, the seat, and the console. A 10 mm target size on these surfaces is a reasonable compromise. The dummy itself is typically a rigid or deformable FE dummy (e.g., ES-2re or SID-IIs) with its own mesh; do not remesh the dummy to match the compartment — instead, ensure the compartment surfaces are smooth and free of gaps that could cause contact penetration.

Contact definitions and interface checks

Define a single *CONTACT_AUTOMATIC_SINGLE_SURFACE for the vehicle self-contact, and a *CONTACT_AUTOMATIC_SURFACE_TO_SURFACE between the barrier and the vehicle, and between the dummy and the compartment. Set SOFT = 1 or 2 for the dummy contact to handle the soft foam and flesh. Use FS and FD friction values of 0.2–0.4 for trim-to-dummy and 0.1–0.2 for metal-to-metal.

After assembly, run a short initialization (0.001 s) and check the following before committing to a full run:

  • Initial penetration: contact output should show zero or near-zero penetration. Any penetration above 0.5 mm indicates overlapping geometry that must be fixed by adjusting node positions, not by increasing SOFT.
  • Mass scaling: check added mass in glstat; it should stay below 1% of total mass. If it exceeds this, the smallest element controls the time step and should be remeshed.
  • Hourglass energy: keep below 5% of internal energy. High hourglass energy in the B-pillar or intrusion beam means the mesh is too coarse or the element formulation is inappropriate.
  • Interface gaps: verify that the door trim, inner panel, and seat are not initially interpenetrating and that the dummy is seated with the correct H-point and clearance.
A practical rule for side impact assembly: mesh the load path (barrier → door outer → intrusion beam → door inner → trim → occupant) at 5–8 mm, and mesh the surrounding structure at 10–15 mm. This concentrates resolution where the load transfers and keeps the model size manageable.
Side Impact Model Assembly — Load Path and Mesh Resolution MDB moving deformable barrier v = 50 km/h, −Y Door outer 5–8 mm Intrusion beam Door inner + trim 5–8 mm Occupant compartment 10 mm target Seat foam 10–15 mm ES-2re dummy Load path: barrier → door outer → intrusion beam → door inner → trim → occupant Mesh resolution by region Door outer / inner: 5–8 mm Intrusion beam: 5–8 mm, aligned Compartment: 10 mm Seat foam: 10–15 mm Dummy: as supplied, no remesh Barrier: rigid / deformable as specified

Verification before the full run

Once the assembly is complete, run a 1 ms initialization and inspect the glstat and matsum files. The total mass should match the sum of the subsystem masses within 0.5%. The initial kinetic energy should equal the barrier mass times velocity squared divided by two. If the barrier is modeled as rigid, its velocity should remain constant until contact; if it drops before contact, there is an unintended initial penetration or a constraint error.

A common failure symptom is a sudden spike in contact force at t = 0, which almost always means the barrier and door outer panel are initially overlapping. Fix this by translating the barrier in +Y by the measured overlap distance, not by reducing the contact stiffness. Another symptom is a negative volume error in the foam elements, which indicates the foam is being compressed beyond its densification strain; check the LCS curve in *MAT_057 and ensure the unloading curve is defined.

With the assembly verified, the model is ready for the full 120 ms side impact run. The next step is to interpret the occupant response — head, chest, abdomen, and pelvis accelerations and the associated injury criteria — which depends on the quality of the interfaces established here.

← Previous 📖 Contents Next →