3. Elasticity calibration

Linear elasticity is the foundation on which every nonlinear material model is built. Before plasticity, hyperelasticity, or damage parameters can be trusted, the elastic stiffness tensor must be identified accurately and its uncertainty must be characterized. This chapter covers the practical calibration of isotropic and orthotropic linear elastic constants from mechanical tests, and shows how to propagate measurement uncertainty into the finite element model.

3.1 The Elastic Stiffness Tensor and Its Engineering Constants

For a general anisotropic linear elastic material, the stress–strain relation is written in Voigt notation as:

σ_i = C_ij · ε_j     (i, j = 1..6)

where C_ij is the 6×6 stiffness matrix. For an isotropic material, C reduces to two independent constants, most commonly the Young's modulus E and Poisson's ratio ν. The shear modulus G and bulk modulus K are then derived:

G = E / (2(1 + ν))
K = E / (3(1 − 2ν))

For an orthotropic material — such as a unidirectional composite lamina, a rolled metal sheet, or a 3D-printed part with raster orientation — nine independent constants are required: three Young's moduli (E1, E2, E3), three Poisson's ratios (ν12, ν13, ν23), and three shear moduli (G12, G13, G23). The compliance matrix S = C⁻¹ is symmetric, and the engineering constants must satisfy thermodynamic stability constraints (positive definiteness of C).

3.2 Isotropic Calibration from Tensile and Shear Tests

The most direct route to E and ν is a uniaxial tensile test on a standard specimen (e.g., ASTM E8 dog-bone). The specimen is loaded along its axis, and the axial strain ε_axial and transverse strain ε_trans are measured with an extensometer or biaxial strain gauges. The elastic constants are obtained from the initial linear portion of the stress–strain curve:

E = Δσ / Δε_axial
ν = − Δε_trans / Δε_axial

For a shear modulus measurement, a torsion test on a cylindrical or thin-walled tube specimen is preferred. The shear modulus is:

G = τ / γ = (T · L) / (J · θ)

where T is torque, L is gauge length, J is polar moment of inertia, and θ is twist angle. Alternatively, a planar shear test (e.g., Iosipescu or V-notched rail shear) can be used for sheet materials.

A consistency check is essential: for an isotropic material, the three measured moduli must satisfy G = E / (2(1+ν)). If the measured G deviates by more than a few percent, the material is either not isotropic or the test has systematic errors (e.g., fixture compliance, strain measurement misalignment).

3.3 Orthotropic Calibration: Directional Tensile and Shear Tests

For orthotropic materials, the calibration matrix is built from directional tests. A common workflow for a unidirectional composite lamina is:

• 0° tensile test → E1, ν12
• 90° tensile test → E2, ν21 (with ν21 = ν12 · E2/E1 as a consistency check)
• ±45° tensile test → G12 via the shear response
• Through-thickness tests (e.g., short-beam shear or direct compression) → E3, ν13, ν23, G13, G23

For the ±45° tensile test, the apparent axial modulus E_x is related to the lamina constants by:

1/E_x = (1/4)(1/E1 + 1/E2 − 2ν12/E1 + 1/G12)

Solving for G12 gives a practical estimate. However, this test is sensitive to edge effects and fiber rotation, so a direct shear test (e.g., ±45° Iosipescu) is preferred when accuracy matters.

3.4 Uncertainty Quantification for Elastic Constants

Every measured elastic constant carries uncertainty from three sources: (1) measurement noise (strain gauges, load cell, extensometer), (2) specimen geometry variability (thickness, width, gauge length), and (3) test setup compliance (fixture deformation, grip slip). A rigorous calibration should report not just a nominal value but a confidence interval.

A practical UQ workflow for isotropic constants:

1. Repeat the tensile test on n ≥ 5 specimens.
2. For each specimen, compute E_i and ν_i from the linear regression of stress vs. strain over the elastic range (e.g., 0.05%–0.25% strain).
3. Compute the sample mean and standard deviation: Ē = mean(E_i), s_E = std(E_i).
4. Report the 95% confidence interval: Ē ± t_{0.025, n−1} · s_E / √n.
5. Propagate to derived quantities (e.g., G, K) using the first-order Taylor expansion (delta method) or Monte Carlo sampling.

For orthotropic constants, the same procedure applies per direction, but correlations between constants (e.g., E1 and ν12 from the same test) should be retained in the covariance matrix. A Monte Carlo simulation with 10,000 samples drawn from the joint distribution of the constants can then be used to propagate uncertainty into the FE model — for example, to bound the predicted natural frequencies or buckling loads.

3.5 Worked Example: Isotropic Aluminum 6061-T6

Consider a set of five tensile tests on Al 6061-T6 dog-bone specimens (gauge length 25 mm, width 6 mm, thickness 3 mm). The measured axial and transverse strains over the 0.05%–0.25% range yield the following:

Mean Ē = 68.86 GPa, s_E = 0.45 GPa; mean ν̄ = 0.3304, s_ν = 0.0023. The 95% CI for E is 68.86 ± 0.56 GPa (t = 2.776 for n−1 = 4). The derived shear modulus G = E/(2(1+ν)) = 25.9 GPa, with propagated uncertainty ±0.3 GPa (dominant contribution from E).

These values can now be entered into the FE model. If the model is later used for nonlinear analysis, the elastic constants serve as the baseline; any plasticity or damage parameters must be calibrated on top of this baseline.

3.6 Visual Summary of the Calibration Workflow

The diagram above summarizes the complete calibration chain: from specimen preparation and raw data acquisition, through per-specimen extraction and uncertainty quantification, to the final FE-ready stiffness tensor. The two branches (isotropic and orthotropic) share the same UQ framework but differ in the number of independent constants and the required test directions.

3.7 Summary

Elastic calibration is the first and most fundamental step in material model calibration for nonlinear FEA. For isotropic materials, tensile and torsion tests provide E, ν, and G, with a consistency check to validate isotropy. For orthotropic materials, directional tensile and shear tests populate the nine independent constants, with symmetry and positive-definiteness constraints enforced. Uncertainty quantification — through repeated tests, confidence intervals, and Monte Carlo propagation — ensures that the FE model reflects not just nominal values but the real variability of the material. These elastic constants form the baseline for all subsequent nonlinear calibration, including plasticity, hyperelasticity, and damage models.