Aneta Kaczmarek

2024-11

Fatigue delamination growth analysis using Paris law for composites — da/dN vs dG relationship calibrated from fatigue DCB tests, integrated to predict delamination front progression under cyclic loading spectrum.

Challenge: Paris law scatter for composites is very large — coefficient of variation for the Paris law slope m can exceed 30% between specimens. Monte Carlo analysis with material scatter was necessary to bound the fatigue life prediction.

Learned: Composite fatigue delamination data is inherently scattered. Deterministic Paris law integration gives a median prediction; upper bound life requires using upper bound Paris law constants from the scatter band. Probabilistic analysis is necessary for certification.

2024-05

Adhesive bonded joint analysis for composite panel repair — scarf joint, film adhesive modeled with cohesive elements, static failure load prediction and peel/shear stress distribution.

Challenge: Adhesive elastic modulus in peel direction was measured from thick adherend shear test (TAST) specimens — modulus under peel loading was 30% lower than under shear. Using isotropic adhesive properties significantly overestimated peel strength.

Learned: Film adhesive material properties differ significantly in shear and peel directions. Always use orthotropic adhesive properties derived from appropriate test configurations — TAST for shear, climbing drum peel for peel direction.

2023-12

Cohesive zone model calibration for Mode I and Mode II interlaminar fracture — DCB (Double Cantilever Beam) for Mode I, ENF (End Notched Flexure) for Mode II. Extracted GIc and GIIc from test, validated CZM in Abaqus.

Challenge: CZM mesh sensitivity is critical — the cohesive zone must contain at least 3 elements. Cohesive zone length for the tested CFRP was 0.8mm (Mode I) — requiring element sizes below 0.25mm at the crack front. Very fine mesh and large models resulted.

Learned: CZM calibration from DCB/ENF tests is reliable for simple crack modes but mixed-mode ratios in real structures differ from the calibration tests. Mixed-mode CZM calibration (B-K or Power law criterion) requires additional mixed-mode fracture tests.

2023-07

Double-lap shear joint analysis with progressive failure — ply-by-ply failure tracking using Hashin criterion in Abaqus, stiffness degradation model after first ply failure.

Challenge: Abaqus stiffness degradation after Hashin failure uses instantaneous (sudden) degradation by default. This caused numerical instability at failure initiation — the global stiffness matrix became near-singular. Implemented gradual stiffness degradation via USDFLD subroutine.

Learned: Sudden stiffness degradation in progressive failure models causes numerical instability. Gradual degradation (exponential or linear) or a viscous regularization parameter avoids the instability while maintaining physical accuracy.

2023-01

Bolted composite joint analysis for nacelle attachment — double-shear joint with 8 titanium bolts, combined in-plane and out-of-plane loading. HyperMesh for meshing, Abaqus for nonlinear analysis with contact.

Challenge: HyperMesh-to-Abaqus contact definition export was not clean — contact pairs were missing or duplicated for some bolt-hole interfaces. Wrote a Python post-processing script to verify and correct contact pair definitions in the Abaqus input file.

Learned: HyperMesh's Abaqus interface is reliable for most model entities but can struggle with complex contact definitions. Always verify the exported Abaqus input file for contact pairs, especially in models with many small-diameter fasteners.