4. High-speed tensile testing

Airbag fabric does not fail under the same conditions as a seat belt or a structural composite. During a deployment event, the cushion inflates in roughly 20–50 ms, and the fabric experiences strain rates on the order of 101 to 103 s−1. Quasi-static tensile data measured at 1–100 mm/min (≈10−4–10−2 s−1) therefore cannot be extrapolated directly to deployment. High-speed tensile testing exists to close that gap: it produces force–displacement and stress–strain data at rates representative of inflation, and it exposes rate-dependent mechanisms such as yarn crimp interchange, weave locking, and inertial loading that are negligible at low speed.

Test systems and their rate envelopes

Two hardware families dominate airbag fabric characterization. Servo-hydraulic systems use a closed-loop hydraulic actuator with a high-response servo valve and a fast data acquisition chain. They typically cover 0.1–20 m/s crosshead velocity, which for a 50 mm gauge length corresponds to nominal strain rates of roughly 2–400 s−1. Their advantages are programmable velocity profiles, large stroke, and the ability to hold a constant extension rate through the test. Their limitations are actuator compliance, hydraulic ringing, and load-cell resonance that can contaminate the first few milliseconds of the signal.

Drop-weight systems use a falling mass to pull the specimen through a slack adapter or a guided carriage. They reach 5–25 m/s and higher, but the velocity is not constant: the mass decelerates as the specimen loads, so the strain rate falls continuously through the test. Drop towers are mechanically simple and inexpensive, but they require careful mass selection and often a slack rope or trigger mechanism to avoid pre-loading the specimen before impact.

System Typical velocity Nominal strain rate (50 mm gauge) Key limitation
Servo-hydraulic 0.1–20 m/s 2–400 s−1 Load-cell resonance, actuator compliance
Drop-weight 5–25 m/s 100–500 s−1 Non-constant velocity, mass tuning
Split Hopkinson (fabric) 10–40 m/s 500–2000 s−1 Wave dispersion, specimen gripping

Strain-rate regimes and what changes across them

Airbag fabric response is not a single curve but a family of curves. At low rates, the fabric behaves as a crimp-interchange network: yarns straighten, the weave opens, and the initial modulus is low. As rate increases, the yarns have less time to rearrange, so the crimp-interchange plateau shortens and the modulus rises. At the highest rates, the response is dominated by yarn tensile stiffness and by wave propagation along the yarn, and the failure strain often decreases.

A practical way to organize the data is to define three regimes for the material model:

  • Quasi-static (10−4–10−2 s−1): baseline stiffness and failure strain, used for model initialization.
  • Intermediate (10−1–101 s−1): transition region where rate sensitivity is strongest; most airbag inflation events fall here.
  • High (102–103 s−1): deployment-relevant regime; used to calibrate rate-dependent strength and failure criteria.

For a typical 470 dtex nylon 6.6 plain-weave, an illustrative (not measured) set of calibration targets might be: initial modulus 0.8 GPa at 10−3 s−1, 1.6 GPa at 100 s−1, and 2.4 GPa at 102 s−1; failure strain 28%, 22%, and 18% respectively. These numbers are example values for illustrating the calibration workflow, not test results.

Inertial effects and signal interpretation

At high speed, the measured force is not purely the specimen's material response. The load cell sees the sum of specimen force, the inertia of the grip and fixture mass, and wave propagation effects. The inertial contribution scales with acceleration, so a sudden velocity change at the start of the test produces a force spike that can be mistaken for a material yield point.

Three practical countermeasures are standard:

  1. Use a slack adapter or compliant link between the actuator and the specimen so the specimen is loaded gradually rather than impulsively.
  2. Measure grip displacement directly with a non-contact extensometer or high-speed camera, rather than deriving strain from crosshead position, which includes machine compliance.
  3. Apply a low-pass filter matched to the load-cell natural frequency, and report the filter corner frequency with the data.
Do not compare force–displacement curves from different systems without stating the gauge length, specimen width, grip type, and filter settings. A 50 mm gauge length on a servo-hydraulic frame and a 100 mm gauge length on a drop tower will produce different apparent moduli even for the same fabric.

Data reduction workflow

The goal of data reduction is to convert raw force, time, and displacement into engineering stress–strain curves that can be used for material model calibration. The following sequence is representative.

1. Acquire raw signals: F(t), d(t), t at ≥ 100 kHz for high-rate tests.
2. Correct for system compliance: d_specimen = d_measured − F / k_machine.
3. Compute engineering strain: ε = d_specimen / L0.
4. Compute engineering stress: σ = F / (w · t_fabric).
5. Identify the linear region (typically 0.5%–2% strain) and fit modulus.
6. Extract failure stress and failure strain at the maximum force point.
7. Bin results by strain rate and fit a rate-dependent model (e.g., Cowper–Symonds or tabulated plasticity).

For airbag fabric, a common calibration target is a tabulated rate-dependent model in which the yield stress and failure strain are functions of strain rate. The calibration input is a set of stress–strain curves at three or more rates, and the output is a table that the solver interpolates. The quality of the fit is judged by comparing the simulated force–displacement response of a single-ply strip test against the measured curve at the same rate.

Practical verification and troubleshooting

Before trusting any high-rate dataset, run a verification specimen with known behavior, such as a metal foil or a rubber strip, and check that the measured modulus matches the reference value within 10%. If the initial slope is too high, suspect inertial ringing or a pre-loaded specimen. If the failure strain is unexpectedly low, check for grip slippage or stress concentration at the jaws. If the curve is noisy, verify the load-cell mounting torque and the cable shielding.

Record the specimen orientation (warp vs. weft), the number of plies, and the conditioning history (temperature and humidity) for every test. Airbag fabric is hygroscopic, and a 2% moisture change can shift the modulus by several percent.
High-Speed Tensile Test Data Reduction Chain Raw Acquisition F(t), d(t), t ≥ 100 kHz sampling Compliance Correction d_spec = d_meas − F/k_m k_m from calibration Stress–Strain ε = d_spec / L0 σ = F / (w · t) Rate Binning 10⁻³, 10⁰, 10² s⁻¹ 3+ rates required Inertial Effects to Remove • Load-cell resonance spike at t = 0 • Grip and fixture mass acceleration • Wave propagation along yarn • Machine compliance in displacement Calibration Output • Modulus vs. strain rate table • Failure stress vs. strain rate • Failure strain vs. strain rate • Interpolation for solver input Verification Simulate single-ply strip test at same rate Compare simulated vs. measured F–d curve Accept if peak force within 10% and failure strain within 15%

The workflow above is iterative. If the verification step fails, the most common causes are an incorrect machine compliance value, a missing inertial correction, or too few calibration rates. Adding a fourth rate near the deployment condition usually resolves the last case. Once the model reproduces the single-ply strip test across the rate range, it can be used in the airbag deployment simulation with reasonable confidence.

← Previous 📖 Contents Next →