2. Material characterization workflow

Airbag fabric characterization is not a single tensile test performed on a convenient strip of material. It is a controlled chain of custody that begins at the roll and ends at a calibrated material card in an explicit solver. Every link in that chain — sampling, specimen preparation, test matrix design, and data traceability — determines whether the resulting constitutive model predicts deployment behavior or merely fits a curve. This section describes how to build that chain so that the numbers entering the model can be defended.

2.1 Sampling from the Fabric Roll

Coated and uncoated airbag fabrics are anisotropic, and their properties vary along the warp (machine direction, MD), the weft (cross-machine direction, CD), and through the roll width. A specimen cut from the roll edge may differ measurably from one cut at the center. Sampling must therefore be planned, not improvised.

Establish a roll identification record before any cutting. Record the manufacturer, fabric designation (for example, a 470 dtex nylon 6.6 plain weave with silicone coating), roll number, lot number, coating batch, width, and length. Then define a sampling grid: divide the roll width into at least three zones (selvedge-near, center, selvedge-near opposite) and sample from at least two positions along the length. This captures both within-roll and along-roll variability.

Cut specimens with a die or a CNC cutter, never by hand with scissors, because frayed edges and non-parallel sides introduce stress concentrations that lower the measured failure strain. A typical specimen for a strip tensile test is 50 mm wide with a 200 mm gauge length, or 25 mm wide with a 100 mm gauge length for coated fabrics where grip slip is manageable. Mark the MD and CD orientation on every specimen with a permanent marker or a punched code, and record the cutting position relative to the roll edge.

2.2 Specimen Preparation and Grip Considerations

Airbag fabric is slippery, especially when coated. Grip slip is the most common source of invalid tensile data. Use serrated or rubber-faced grips, and consider tabbing the specimen ends with a thin aluminum or cardboard tab bonded with cyanoacrylate or a pressure-sensitive adhesive. The tab transfers load into the fabric without crushing the weave.

For biaxial testing — which is essential for capturing the coupling between warp and weft — prepare cruciform specimens with arm widths matched to the fabric's crimp interchange behavior. A common configuration uses a 100 mm × 100 mm central gauge area with 50 mm wide arms. The arms must be long enough that the grip region does not constrain the central gauge area. Mark the central gauge area with a contrasting ink grid or apply a speckle pattern if optical strain measurement (digital image correlation, DIC) is used.

For tear testing (trapezoidal or tongue tear), prepare specimens with a pre-cut notch of defined length and geometry. The notch tip radius matters: a sharp razor cut produces a different tear path than a punched notch. Standardize the notch preparation method and record it.

2.3 Designing the Test Matrix

The test matrix must cover the deformation modes and rates the airbag experiences during deployment. A practical matrix for a coated nylon 6.6 fabric includes:

This matrix is a starting point, not a universal prescription. If the airbag design includes seams, add seam shear and seam peel tests. If the fabric is coated, add coating adhesion and coating friction tests. The principle is to cover every deformation mode that the constitutive model must reproduce, and to include enough replicates to quantify scatter.

2.4 Data Traceability and the Material Card

Traceability means that any number in the material card can be traced back to a specific specimen, a specific test run, and a specific raw data file. Implement this with a unique specimen ID that encodes roll, position, orientation, and test type. For example, R047-C-MD-T100-03 identifies roll 047, center position, machine direction, 100 mm/min tensile test, third replicate.

Store raw force–displacement data, not just processed results. The raw curve reveals grip slip, premature failure, and machine compliance artifacts that a summary table hides. Process the raw data with a documented script that applies the same filtering, gauge length correction, and failure criterion to every specimen. Record the script version and the processing parameters.

# Example processing record (YAML)
specimen_id: R047-C-MD-T100-03
raw_file: data/roll047/C_MD_100_03.csv
gauge_length_mm: 200.0
crosshead_speed_mm_min: 100.0
filter: savgol(window=21, polyorder=3)
failure_criterion: max_force
modulus_range: 0.5% to 2.0% strain
processed:
  modulus_MPa: 412.3
  strength_MPa: 68.7
  failure_strain_pct: 24.1
operator: J. Chen
date: 2025-03-14

The material card itself should reference the test matrix and the processing script. In LS-DYNA, for example, the *MAT_FABRIC card requires elastic moduli, Poisson's ratio, and failure strains. Each of these values should be accompanied by a source reference: which test, which specimen set, which statistical treatment (mean, median, or a conservative percentile).

2.5 Workflow Summary

The complete workflow is a sequence of controlled transfers: roll → sampling plan → specimen preparation → conditioning → test execution → raw data capture → processing → material card → model validation. At each transfer, record the identity and state of the material. The SVG below maps this chain and highlights the traceability links that must be preserved.

The workflow is only as strong as its weakest traceability link. A material card with excellent tensile data but no record of which roll or which conditioning state produced it is not a calibrated model — it is an unverified fit. Build the chain deliberately, and the model will have a defensible foundation for validation against sled or drop-tower tests.