3. Battery pack CAD preparation and meshing in PAM-CRASH
Impact simulation of an EV battery pack is only as trustworthy as the mesh it runs on. A pack assembly arriving from CATIA, NX, or Creo is a design artifact: it contains overlapping surfaces, tiny fillets, bolt holes, weld flanges, and mixed units. PAM-CRASH will happily accept almost any geometry you feed it, but the resulting hourglass energy, contact penetration, and mass scaling will tell you whether the model was simulation-ready. This section covers the three-stage pipeline that turns raw CAD into a validated PAM-CRASH mesh: geometry cleanup, shell/solid element selection, and mesh quality control.
Stage 1 — Geometry Cleanup Before Any Meshing
Cleanup is done in the pre-processor (ANSA, HyperMesh, or PAM-CRASH's own Visual-Environment) and should be completed before a single element is created. The goal is a watertight, topologically connected surface model with no free edges in load paths.
Work through the following operations in order:
- Unit and tolerance audit. Confirm the CAD is in millimetres and set a global geometry tolerance of 0.1 mm for a typical pack. PAM-CRASH is unit-agnostic; a model built in metres will produce a mass 10⁹ times too small and a stiffness that is meaningless.
- Remove non-structural detail. Delete logos, embossed text, chamfers under 1 mm, and cosmetic ribs. These generate sliver elements that force the explicit time step down without adding stiffness.
- Defeature fastener holes. Bolt and rivet holes below 8 mm diameter are usually filled and represented by rigid or tied connections. Leaving them meshed creates a ring of tiny shells around every hole.
- Heal surfaces. Use topology repair to stitch gaps, remove duplicate surfaces, and fix missing faces. Then run a free-edge check: any red free edge inside a module or tray wall indicates a discontinuity that will open under impact.
- Midsurface extraction. For thin-walled parts (tray, cover, busbar, cold plate), extract the midsurface and assign thickness. For thick or solid parts (module housing, foam, potting), keep the solid geometry.
Stage 2 — Choosing Shell vs. Solid Elements
The choice is driven by the through-thickness stress state, not by convenience. PAM-CRASH shell elements (type 103, 104, 105) are four-node, under-integrated, and carry membrane, bending, and transverse shear behaviour with a user-defined thickness. Solid elements (type 1, 2, 6, 14) resolve through-thickness stress and are required wherever crushing, densification, or triaxial stress governs the response.
| Pack component | Typical element | Rationale |
|---|---|---|
| Battery tray / enclosure wall | Shell (type 103/105) | Thin sheet, bending-dominated, thickness 1–3 mm |
| Top cover / lid | Shell | Thin, large area, low bending stiffness |
| Module housing | Shell or solid | Shell if thin; solid if cast or thick-walled |
| Cell jelly-roll / can | Solid (type 1/2) or homogenised solid | Through-thickness crush and densification |
| Thermal pad / foam | Solid (type 1) | Volumetric compression, near-incompressible |
| Busbar / connector | Shell | Thin conductor, bending and tearing |
| Cold plate | Shell | Thin extruded channel walls |
Two rules prevent the most common errors. First, never mix shell and solid in the same load path without a tied interface; a shell node sitting on a solid face will not transfer moment unless you define a tied contact or a rigid spider. Second, when a solid part is thin in one direction (for example a 4 mm cast rib), use a single layer of solid elements with an aspect ratio below 5, or switch to shells — do not stack three 1.3 mm solids, which will collapse the time step.
Stage 3 — Mesh Quality Criteria for Explicit Impact
In an explicit solver the stable time step is governed by the smallest element:
dt = L_min / c, c = sqrt(E / rho)
Example: steel tray, E = 210 GPa, rho = 7.85e-9 t/mm^3
c = sqrt(210000 / 7.85e-9) = 5.17e6 mm/s
L_min = 2 mm -> dt = 3.9e-7 s (0.39 us)A single 0.2 mm sliver drops the step to 0.04 µs and multiplies runtime by ten. Control the minimum size deliberately, then use mass scaling only as a last resort and always report the added mass.
Apply these targets for a pack model:
- Element size: 5–10 mm global for the tray and cover; 2–5 mm in crush zones and around mounting points; 1–2 mm only in local detail studies.
- Aspect ratio: below 5 for shells, below 10 for solids in the crush direction.
- Warpage: below 15° for quads.
- Jacobian: above 0.6 for solids; reject anything below 0.3.
- Quad angle: 45°–135°; triangles below 5% of the total shell count and never in a bending-dominated region.
- Hourglass control: use type 4 (Flanagan–Belytschko) with a coefficient of 0.05–0.10 for under-integrated shells and solids.
Verification Before the Impact Run
Run a short checks-only deck before committing to the full simulation. Confirm total mass against the CAD bill of materials (within 2%), check that no element has a time step below your target, and verify that all tied and contact interfaces are closed. A useful sanity check is a 1 ms low-velocity indentation on the tray floor: the force–displacement curve should be smooth, with no sawtooth from hourglassing and no sudden force drop from a failed tie.
The pipeline is sequential and gated: do not mesh until cleanup passes the free-edge check, and do not run impact until the quality metrics and the indentation sanity test are clean. A pack model that satisfies these gates will produce stable energy balance, predictable contact behaviour, and results you can defend in a safety review.