4. Material modeling for battery enclosure structures

The battery enclosure is the primary load path in a PAM-CRASH electric-vehicle crash model. It carries the pack mass, transfers impact forces into the body-in-white, and shields the cells from intrusion. If the enclosure material model is wrong, every downstream result — intrusion depth, bolt rupture, cell crush — becomes unreliable. This section covers the three modeling layers you must calibrate for enclosure structures: elastic-plastic metal behavior, strain-rate dependence, and failure criteria.

Elastic-Plastic Metals: From Tensile Coupon to PAM-CRASH Card

Enclosure components are typically made from aluminum extrusions (6000-series), stamped steel (DP590, DP780), or occasionally magnesium alloys. In PAM-CRASH these are modeled with shell elements using material law 103 (elastic-plastic with isotropic hardening) or law 104 (kinematic hardening for cyclic loading). The input is a true stress–true plastic strain curve, not the engineering curve from the tensile test.

Conversion from engineering to true values follows the standard relations. For engineering stress σeng and engineering strain εeng:

σ_true = σ_eng * (1 + ε_eng)
ε_true = ln(1 + ε_eng)
ε_pl   = ε_true - σ_true / E

where E is Young's modulus. For a DP780 steel coupon with σeng = 820 MPa at εeng = 0.12 and E = 210 GPa, the true stress is 918 MPa, true strain is 0.1133, and plastic strain is 0.1090. This plastic strain value is what enters the hardening table.

A practical calibration workflow for an aluminum extrusion (e.g., 6061-T6) looks like this:

  1. Obtain tensile data in the rolling/extrusion direction and at 45° and 90° if anisotropy matters. For extrusion, the longitudinal direction dominates.
  2. Convert to true stress–plastic strain and truncate at the point of diffuse necking (Considère criterion: dσ/dε = σ). Beyond necking, the curve must be extrapolated — use a power-law or linear extension, and document the assumption.
  3. Enter the curve into the PAM-CRASH material card as a piecewise-linear hardening table. Keep the first point at yield (zero plastic strain) and space points densely in the high-curvature region near yield.
  4. Validate with a single-element or coupon-level simulation before running the full enclosure model.
For aluminum extrusions, the difference between the longitudinal and transverse flow curves can exceed 15% at 10% plastic strain. If your enclosure has extrusions loaded in multiple directions, consider law 104 with an anisotropic yield surface rather than a single isotropic curve.

Strain-Rate Dependence

Crash events load the enclosure at strain rates from 100 to 103 s−1. Both steel and aluminum exhibit positive strain-rate sensitivity: flow stress increases with rate. Ignoring this makes the enclosure appear softer than reality and can over-predict intrusion.

In PAM-CRASH, strain-rate effects are added via the Cowper–Symonds model (law 103 with rate flag) or a tabulated rate-dependent yield stress. The Cowper–Symonds relation is:

σ_dynamic / σ_static = 1 + (ε̇ / D)^(1/p)

where D and p are calibrated constants. For DP780, typical values are D = 40,000 s−1 and p = 5. For 6061-T6 aluminum, D ≈ 6,500 s−1 and p ≈ 4. These are starting points — you must calibrate against split-Hopkinson pressure bar (SHPB) data or published rate-dependent curves for your specific alloy and temper.

Calibration procedure:

  1. Collect quasi-static (10−3 s−1) and dynamic (102–103 s−1) flow curves.
  2. At several plastic strain levels (e.g., 2%, 5%, 10%), compute the ratio σdyn/σstat.
  3. Fit D and p by least squares on the log-linearized Cowper–Symonds equation.
  4. Verify that the fitted curve reproduces the dynamic flow stress within 5% across the rate range of interest.

If SHPB data is unavailable, use published values for the same alloy family and temper, but flag this as an assumption in your report. The sensitivity of intrusion depth to p is typically stronger than to D, so prioritize p accuracy.

Failure Criteria Calibration

Enclosure failure modes include ductile tearing at bolt holes, shear fracture at extrusion joints, and buckling-induced rupture. PAM-CRASH offers several failure models; the most common for enclosure metals are:

  • Equivalent plastic strain (EPS) criterion: element deletion when εpl exceeds a threshold. Simple but mesh-sensitive.
  • Johnson–Cook damage: accounts for stress triaxiality and strain rate. Better for mixed-mode loading.
  • GISSMO: incremental damage accumulation with regularization. Recommended for shell elements where mesh size varies.

For a DP780 enclosure, a practical GISSMO calibration uses a fracture locus in stress-triaxiality space. You need at least three coupon geometries: smooth tensile (η ≈ 0.33), notched tensile (η ≈ 0.5–0.7), and shear (η ≈ 0). The fracture strain at each triaxiality is extracted from the tests, then fitted to a curve. In PAM-CRASH, this is entered as a table of equivalent plastic strain at fracture versus triaxiality.

Mesh regularization is critical. GISSMO uses a characteristic length to scale the damage evolution so that the energy dissipated per unit area is mesh-independent. Set the regularization flag and provide the characteristic element length. Without this, a 5 mm mesh and a 2 mm mesh will predict different failure locations.

Do not calibrate failure criteria on the same coupon data used for hardening calibration without independent validation. Use a separate set of tests (e.g., different notch radii or a component-level test) to verify that the calibrated model predicts failure at the correct location and displacement.

Integrated Calibration Workflow

The three layers are not independent. Strain-rate sensitivity affects the flow stress that feeds into the damage model, and failure strain itself can be rate-dependent. A defensible workflow is:

  1. Calibrate quasi-static hardening from tensile tests.
  2. Calibrate rate sensitivity from SHPB or published data.
  3. Calibrate failure locus from multi-geometry coupon tests at quasi-static rate.
  4. If rate-dependent failure data exists, add a rate-scaling factor to the fracture strain.
  5. Validate on a component-level test — for example, a three-point bend of an extrusion segment or a bolt-pull-through test — before committing to the full vehicle model.

The SVG below summarizes the data flow from coupon tests to the PAM-CRASH material card and the validation loop.

Enclosure Material Calibration Data Flow Coupon Tests Tensile (quasi-static) SHPB (10²–10³ s⁻¹) Notched / shear Component bend test Data Reduction Eng → true stress/strain Necking truncation Rate ratio σ_dyn/σ_stat Fracture strain vs η PAM-CRASH Input Law 103/104 hardening Cowper–Symonds D, p GISSMO fracture locus Mesh regularization Validation Component test Force–disp match Failure location Intrusion depth Iterate if force–displacement or failure location deviates >10% Typical Calibrated Parameters (Example Values) Material Yield (MPa) D (s⁻¹) p Fracture ε_pl (η=0.33) DP780 steel 450 40,000 5.0 0.35 6061-T6 Al 275 6,500 4.0 0.18 DP590 steel 340 35,000 4.8 0.40 Values are illustrative starting points; always calibrate against your specific alloy, temper, and supplier data.

Practical Checks Before Running the Full Model

Before committing to a full vehicle crash simulation, run these checks on the enclosure submodel:

  • Energy balance: Hourglass energy should be below 5% of internal energy. If it exceeds this, refine the mesh or adjust the hourglass control coefficient.
  • Element deletion: Verify that deleted elements are in physically expected locations (bolt holes, joint fillets) and not scattered randomly. Random deletion indicates a failure criterion that is too aggressive or a mesh quality issue.
  • Force–displacement correlation: Compare the simulated force–displacement curve of a component-level test against the physical test. Peak force within 10% and displacement at failure within 15% is a reasonable target for a calibrated model.
  • Strain-rate check: Confirm that the maximum strain rate in the enclosure during impact falls within the calibrated range. If it exceeds 103 s−1, your Cowper–Symonds fit may not be valid.

These checks catch the most common modeling errors before they propagate into the full vehicle result. A well-calibrated enclosure material model is the foundation for trustworthy intrusion and cell-crush predictions.

← Previous 📖 Contents Next →