Zofia Kaminska

2025-02

Multiple crack interaction analysis — two adjacent surface cracks in disk bore, crack coalescence prediction using LEFM-based coalescence criteria (Melin, ASTM E2899 coplanar crack interaction).

Challenge: AFGROW does not natively support multiple interacting cracks. Implemented a Python wrapper that runs two single-crack AFGROW simulations with interaction-adjusted K values (using the ASTM E2899 magnification factor) and checks for coalescence criterion at each growth step.

Learned: Interacting cracks amplify each other's stress intensity factor — the magnification factor can reach 1.3-1.5 at close spacing. Treating adjacent cracks as independent (as many DTA codes do by default) is significantly non-conservative when their spacing is less than the combined crack size.

2024-11

NDE inspection interval determination from crack growth curve — from assumed initial flaw (0.5mm per FAR 33.70) to critical crack size, identification of detectable crack size from POD curve, inspection interval = half the remaining life at detection.

Challenge: POD (probability of detection) curves for the inspection method (fluorescent penetrant inspection, FPI) were available only for flat plate specimens, not for the curved disk bore geometry. Required engineering justification for geometry correction factor.

Learned: POD curves are specimen geometry-specific. Applying flat panel POD to curved bore geometry is not rigorous. Ideally, POD data should be acquired on geometry representative specimens. If not available, a geometric correction factor with documented uncertainty bounds is the minimum requirement.

2024-05

Residual strength analysis after assumed initial crack — computed fracture toughness KIc requirement vs available KIc from material database. Determined maximum permissible crack size at which residual strength equals limit load.

Challenge: KIc data in the material database was for base material at standard test specimen orientation. The disk material has significant anisotropy — KIc in the radial-tangential plane (where the bore crack propagates) was 15% lower than the L-T test value. Required orientation-specific toughness data.

Learned: Fracture toughness is directionally dependent in forged disks — grain flow orientation affects KIc significantly. Always confirm that the KIc value used in residual strength assessment corresponds to the actual crack propagation plane and direction.

2023-12

Damage tolerance analysis per FAR 33.70 for HPT disk bore crack — used AFGROW with user-defined K table from FEM. Propagated crack from NDE threshold to critical size, generated inspection interval recommendation.

Challenge: AFGROW's crack growth integration uses fixed load cycles — but the disk experiences variable amplitude loading across different mission profiles. Implemented spectrum loading in AFGROW using the load spectrum file format, weighted by flight hours per mission type.

Learned: AFGROW spectrum loading with variable amplitude is essential for realistic life predictions. Constant amplitude (design load only) predictions are non-conservative if the spectrum includes frequent lower-amplitude cycles that contribute to crack opening load and retardation.

2023-07

Parametric stress intensity factor database — K vs crack size (a/c ratio 0.5 to 2.0, a/t ratio 0.1 to 0.9) for semi-elliptical surface crack in a round bar under combined tension and bending.

Challenge: 120 Abaqus runs for the K database required automation. Built Python script to generate each Abaqus input file with parametric crack geometry, submit to HPC, extract J-integral, and assemble results into the K table. Three runs failed silently — detected only during K table smoothness check.

Learned: When building parametric K databases, always check the resulting K vs a/c and a/t trends for physical plausibility. K should increase monotonically with a/t and have smooth variation with a/c. Outliers indicate modeling errors in specific cases.

2023-01

LEFM analysis for assumed surface crack at HPT disk bore — modeled semi-elliptical surface crack in Abaqus using seam crack front, extracted stress intensity factor KI via J-integral for different crack sizes.

Challenge: J-integral extraction in Abaqus requires correct crack front definition and contour selection. The first contour (immediately at crack tip) is unreliable due to crack tip singularity — must use contours 2-5 and verify path independence before reporting K.

Learned: J-integral path independence is the primary quality check for FEM-based K extraction. If J values vary more than 2% between contours 2-5, the mesh near the crack tip is insufficient. Focused mesh with mid-side node placement at 1/4 position (quarter-point elements) is mandatory.