Stored ply properties bypass Micro. Mixed stacks keep both paths. Gray blocks are not selected. Geometry, process schedules and other model records are omitted here for clarity; a real run must include all required references.
Fatigue uses intact CLT endpoint stresses and measured S–N calibration. It is evaluated separately from Run and progressive failure. The dashed handoff is not active: optimization still uses current properties until a degradation law is supplied.
Materials → Micro → Laminate.
Where to find it
In the CLT structural response workspace, open Response → Failure → Fatigue. Run the populated connected simulation first. Supply measured effective-ply fatigue calibration for every active source and stress channel. Fatigue is evaluated separately; the ordinary simulation Run does not evaluate cyclic life automatically.
Upstream coupling
Stored material plies and solved Micro plies use their existing elastic properties, angles and individual thicknesses. The laminate recovers bottom, middle and top ply stresses at two mechanical endpoints. The full six-component Load vector is used at one endpoint and Rload times that vector at the other. Progressive-ramp checkboxes are not fatigue amplitude selectors.
Selected saved thermal, cure and moisture histories contribute fixed residual strain at both endpoints. Available cure-conditioned stiffness is reused through the existing process-state pathway. Changed inputs require a new connected run. The fatigue calibration must independently cover the effective composition, orientation, process state, environment and cycling frequency; elastic homogenization does not create fatigue coefficients.
Model conventions
All stresses are in MPa and N is cycles, not reversals. U is a calibrated curve intercept, not an automatically inferred fiber strength. For peak-based curves, S is the dominant signed endpoint magnitude. Compression-dominated channels reverse sign before forming local R; calibrate with that same convention. Basquin uses half the stress range. No automatic Goodman or other mean-stress correction is applied.
| Model | Implemented relation | Inputs |
|---|---|---|
| Kim–Zhang | N = N₀ + U⁻ᵝ [(S/U)¹⁻ᵝ − 1] / [α(β−1)] | U, α > 0, β > 1, N₀ > 0 |
| Sendeckyj | S = U [1 + C(N−1)]⁻ˢ | U, C > 0, s > 0; controls use α=C and β=s |
| Weibull S–N | S = L + (U−L) exp[−α(log₁₀ N)ᵝ] | U, 0 ≤ L < U, α > 0, β > 0 |
| Kohout–Vechet | S = U [(1 + N/B)/(1 + N/C)]ᵇ | U, 0 < B < C, b < 0 |
| Basquin | Sₐ = A Nᵇ | A > 0, b < 0. Convert reversal-based coefficients before entry. |
The Kim–Zhang stress curve is obtained by algebraically inverting the stated life equation with N−N₀, preserving S=U at N=N₀. Weibull uses log base 10 explicitly: coefficients fitted with natural logarithms must be converted. The Weibull S–N curve is not a probability distribution and supplies no reliability percentile.
Calibration and persistence
- Select the effective ply source, σ1, σ2 or τ12 channel, and dominant sign.
- Select a model and enter its coefficients, published/test source, local stress-ratio bounds and tested cycle range. No material-specific coefficients are prefilled.
- Check that the calibration covers the current composition, process, temperature, moisture and frequency. Confirm applicability; changing inputs resets confirmation.
- Use Store fatigue inputs in CASES, then save the database. The setup follows that case record. Re-run the simulation after storing inputs, then Evaluate fatigue.
Reading results without a false pass
The table reports local endpoint stresses and R, estimated cycles when inside the calibrated domain, and requested cycles divided by estimated life. Missing calibration, zero-amplitude creep-only channels or an endpoint that fails the static ply check are not evaluated. Stresses above the fitted range and lives beyond it are reported separately, with no invented numeric life. Completed does not mean passed; no overall pass is issued, even when requested cycles are below estimated life.
Scope and verification
This is intact, in-plane CLT constant-amplitude screening. It does not update progressive fatigue damage or evolve temperature/cure during cycling. It does not assess delamination growth, adhesive/core fatigue, out-of-plane/FSDT shear, multiaxial fatigue interaction, variable-amplitude accumulation, creep or self-heating. The FSDT, sandwich, joint and layerwise solvers are not replaced by this panel. Frequency is a recorded calibration condition, not a numerical life correction.
Implementation checks cover forward/inverse equations, cycle conventions, monotonicity, numerical stability, thickness/load scaling, fixed residual stresses, missing data, static failure guards and saved-input round trips. These checks are not experimental benchmark validation. No existing teaching material is marked fatigue-qualified by adding this module.
Sources
- Burhan & Kim (2018), S–N Curve Models for Composite Materials Characterisation: equations 8, 11 and 22; model conventions and comparison.
- Sendeckyj (1981), Fitting Models to Composite Materials Fatigue Data: equivalent static strength and fitted fatigue data.
- Kohout & Věchet (2001), A new function for fatigue curves characterization and its multiple merits: full-range curve.
- A generalization of the fatigue Kohout–Věchet model: equivalent parameterizations.
- Kim & Zhang (2001), Fatigue Damage and Life Prediction of Glass/Vinyl Ester Composites: original model study.
- Weibull (1961), Fatigue Testing and Analysis of Results: original fatigue-curve reference.
∑ CLT and residual-strain recovery · ∑ Monotonic progressive failure · ∑ Process coupling
Workbench availability: released models, inputs and compatible study paths. The wider theory library includes reference formulations not available in every Workbench solve.