3. Material characterization for crash models

Crash simulation credibility rests on the assumption that the constitutive and failure models inside the solver reproduce the real behavior of the materials they represent. That assumption is only defensible when every material card can be traced back to a defined test campaign, a documented specimen geometry, a controlled acquisition chain, and a reproducible post-processing path. Material characterization is therefore not a single experiment but a chain of custody: from the coupon in the testing machine to the parameter set embedded in the LS-DYNA or PAM-CRASH input deck. This section describes how that chain is built for metals and composites, what data must be captured at each link, and what pedigree evidence a reviewer should expect before accepting a material model as credible.

Metals: From Quasistatic Coupons to Rate-Dependent Cards

For ductile metals used in body-in-white and chassis structures, the characterization campaign typically combines tensile tests at multiple strain rates, notched or shear specimens for stress-state sensitivity, and sometimes hydraulic bulge or torsion tests to extend the strain range. A representative campaign for a DP780 steel might include quasistatic tension at 0.001 s⁻¹, intermediate rates at 1 and 10 s⁻¹ on a servo-hydraulic frame, and high-rate tension at 100–1000 s⁻¹ on a split Hopkinson pressure bar. Each specimen carries a unique ID that links the raw force–displacement record, the measured cross-section, the digital image correlation (DIC) field, and the extracted true stress–true strain curve.

Acquisition details matter more than many teams assume. Load cells must be calibrated against a traceable standard, and the acquisition rate should be high enough to resolve the yield transition and any serrated flow—typically 1–10 kHz for high-rate tests. For DIC, the speckle pattern, subset size, and virtual extensometer gauge length must be recorded, because the post-uniform elongation and the onset of necking depend strongly on the measurement basis. The engineering-to-true conversion should be documented explicitly: true stress from σ_true = σ_eng(1 + ε_eng) and true strain from ε_true = ln(1 + ε_eng) are valid only up to necking; beyond that, the local DIC strain in the neck region must be used, or the curve must be extrapolated with a declared hardening law.

A practical calibration workflow for a rate-dependent metal card proceeds as follows. First, fit the quasistatic hardening curve (Swift, Voce, or a combined Hockett–Sherby form) to the DIC-derived true stress–strain data up to the point of validity. Second, fit rate sensitivity using a Cowper–Symonds or Johnson–Cook multiplier against the measured flow stress at each rate. Third, calibrate the fracture locus (e.g., a Johnson–Cook or Modified Mohr–Coulomb surface) using notched tensile specimens with different stress triaxialities, extracting the equivalent plastic strain at fracture from DIC. Finally, run a single-element or coupon-level verification simulation and compare force–displacement and fracture location against the test. A mismatch in fracture location usually indicates a stress-state calibration error rather than a hardening error, and this diagnostic distinction should be recorded in the pedigree.

Composites: Anisotropy, Coupons, and Layup Traceability

Composite characterization is fundamentally more demanding because the material is defined by fiber, resin, layup, and cure, not by a single alloy designation. A credible campaign for a carbon-fiber/epoxy system used in a crash-relevant component typically includes: 0° and 90° tension and compression coupons per ASTM D3039 and D3410 (or the equivalent ISO standards), in-plane shear via ±45° tension or the V-notched rail shear method, and interlaminar fracture toughness via double cantilever beam (mode I) and end-notched flexure (mode II) specimens. For crash modeling, the ply-level elastic constants, strength values, and fracture energies feed a MAT54/55, MAT58, or MAT162-type card, and the calibration of the softening parameters is where most pedigree failures occur.

The critical acquisition requirement for composites is full traceability of the physical layup. Each panel must carry a record of prepreg batch, fiber areal weight, ply orientation sequence, autoclave cure cycle (temperature ramp, dwell, pressure), and the resulting fiber volume fraction from acid digestion or microscopy. Coupons cut from the same panel share a panel ID, and the coupon ID encodes the ply orientation and test type. Without this, a validated material card cannot be reused on a different panel even if the nominal material specification is identical.

During testing, strain measurement should combine extensometers or DIC with back-face strain gauges to detect bending artifacts in compression fixtures. For compression, the fixture alignment and the gauge length are dominant error sources; a misalignment of a few tenths of a millimeter can reduce measured strength by more than 10%. For interlaminar toughness, the crack length must be tracked optically and the compliance calibration (or corrected beam theory) documented, since the extracted fracture energy is sensitive to the crack-length measurement method.

Pedigree Requirements: What Must Travel With the Data

A material model is only as credible as the evidence attached to it. The pedigree record should allow an independent reviewer to reconstruct the test, re-run the calibration, and understand every assumption. The table below summarizes the minimum evidence set for metals and composites; it is a working checklist, not an exhaustive standard.

The figure below shows how these elements connect into a traceable chain from raw coupon to solver-ready card, and where the most common pedigree breaks occur.

Verification and Acceptance Criteria

Before a material card enters a full vehicle model, it should pass a coupon-level verification. For metals, this means simulating the tensile coupon with the calibrated card and comparing the force–displacement curve and the necking location; a reasonable acceptance band is within 5% on peak force and within 10% on displacement at fracture, with the fracture location matching the test. For composites, a single-element or small coupon model should reproduce the measured elastic modulus within 5% and the failure mode—fiber tension, matrix compression, or delamination—qualitatively. If the failure mode differs, the softening parameters or the fracture energy input is suspect, and the pedigree record should note the discrepancy rather than silently adjusting parameters to match the curve.

Finally, the pedigree must be versioned. When a material card is updated—because a new batch was tested, a hardening law was refitted, or a solver version changed the failure criterion implementation—the previous card, its supporting data, and the reason for the change should remain retrievable. A crash model that cites a material card without a retrievable pedigree is, from a credibility standpoint, an unverified assumption. The discipline of attaching evidence at every link, from coupon ID to solver card, is what allows a simulation result to be defended in a design review or a regulatory submission.