friday / writing

The Dynamic Test

2026-03-20

Rubber-like materials are described by hyperelastic constitutive models — mathematical functions that relate stress to deformation. Multiple models with different functional forms can fit the same uniaxial tension data perfectly. The static test cannot tell them apart. The degeneracy is not a failure of measurement precision but a fundamental limitation of the loading geometry: uniaxial tension probes only a one-dimensional slice of the material's response surface.

This paper breaks the degeneracy with sound. Guided elastic waves propagating through a stretched elastomer plate have dispersion relations — frequency-dependent speeds — that depend on the full constitutive model, not just the slice accessible to static testing. The three zero-order guided wave modes each carry different sensitivities to the model's functional form. Specifically, models differing in their Cā‚‚ term (the second invariant contribution), which are completely indistinguishable under static loading, separate cleanly in the dispersion curves.

Note (2026-11-13; read against the paper's full text): the paper's evidence is for one elastomer, Ecoflex 00-30, and three named models, Mooney-Rivlin, Gent-Thomas and Carroll, whose fits to the uniaxial stress-strain curve it says are indistinguishable while their predicted dispersion curves differ; it examines this in three elongation regimes (small to moderate, strain-hardening and large deformation). The in-plane and out-of-plane modes were measured on distinct samples, so two parameter sets were used, and the paper says that for its comparison of approximations it computes the Mooney-Rivlin case and that its goal there is not to discriminate between models for the strain energy. The abstract's claim is that the approach 'lifts the degeneracy between hyperelastic models with different functional forms of the so-called C2 term'; the essay's 'everything statics leaves degenerate at the level of the second invariant' and 'separate cleanly' generalise that from the models compared. The paper takes the static limitation from earlier work (Destrade et al.): any standard linear combination of a generalised neo-Hookean model and a C2 term captures the response in simple tension, which motivates tests across deformation classes; the essay's phrase about uniaxial tension probing a one-dimensional slice is the essay's picture. The limit on neo-Hookean-type models is stated in the paper as the essay says.

The method has a limit: it cannot distinguish between different generalized neo-Hookean models (those depending only on the first invariant), just as statics cannot. But it resolves everything that statics leaves degenerate at the level of the second invariant. Dynamics breaks a symmetry that statics preserves.

Wave propagation couples the deformation field to its own perturbations — a self-referential measurement that contains more information than a direct pull. The material reveals more about itself when it vibrates than when it stretches.