Marianna Pisarska

2025-02

Critical plane fatigue analysis using Smith-Watson-Topper (SWT) parameter — searched all planes at the notch root to find the plane maximizing SWT, compared with simpler normal stress maximum principal approach.

Challenge: Critical plane search is computationally expensive for fine-resolution plane sweeping (1-degree increments over hemisphere). Implemented vectorized plane search in MATLAB that runs in seconds instead of minutes per analysis point.

Learned: SWT critical plane criterion is more accurate than maximum principal stress for non-proportional loading but requires the full stress-strain history on each plane. Pre-computing the plane stress history in vectorized form before searching the critical plane reduces run time by 50x.

2024-10

Rainflow cycle counting on measured strain gauge data from engine ground test — 300-second test segment, multi-channel data, rainflow counting for each gauge, damage accumulation using Miner's rule per ASTM E1049.

Challenge: Strain gauge data had noise spikes that created spurious small cycles in the rainflow count. Applied a range gate filter (removed cycles below 5 microstrain peak-to-valley) to eliminate noise. Filter threshold calibration required comparing histograms with and without filtering.

Learned: Rainflow counting on noisy measured data requires noise filtering before counting. The filter threshold must be justified — too aggressive removes real small cycles; too lenient inflates damage predictions with noise artifacts.

2024-04

Multiaxial notch correction using Hoffmann-Seeger method — extends Neuber correction to biaxial stress state. Applied to turbine disk hub region under combined centrifugal + thermal loading.

Challenge: Hoffmann-Seeger method requires the ratio of principal stresses to remain constant during loading (proportional loading). At the disk hub, the centrifugal and thermal loads have different spatial distributions — proportionality assumption violated. Had to use Mróz-Garud non-proportional correction.

Learned: Non-proportional loading significantly affects notch plasticity and fatigue life — cyclic hardening under non-proportional paths can reduce life by 2-5x compared with proportional loading at the same strain amplitude. Always assess proportionality before choosing a correction method.

2023-12

Fatigue notch factor Kf derivation — from Kt (geometry effect) and notch sensitivity q (material effect using Neuber's material constant a and notch root radius). Compared with direct test data for the alloy.

Challenge: Neuber's material constant 'a' for Ti-6Al-4V is reported with wide scatter in the literature — values from 0.025 to 0.076 mm depending on surface finish and heat treatment condition. The resulting Kf uncertainty propagated to ±15% life prediction uncertainty.

Learned: Kf from Kt is sensitive to the material constant. For life-critical components, deriving the material constant from companion specimen tests with the actual material and surface finish is more reliable than using literature values.

2023-08

Stress concentration factor Kt extraction from FEM — compared ANSYS FEM Kt values with Peterson's charts for common geometries (shoulder fillet, circular hole, U-notch). Documented where Peterson charts are accurate vs where FEM is needed.

Challenge: Peterson's charts cover specific geometry ranges — Kt for a stepped shaft fillet with d/D > 0.9 is not tabulated. FEM was required for these cases. The resulting Kt database is now shared within the stress team as a reference.

Learned: Peterson's Kt charts are invaluable for common geometries but have explicit range limitations. Any geometry outside these ranges must be analyzed by FEM. Always check the applicability range before using a chart value.

2023-03

Neuber rule application for elastic-plastic notch stress correction at compressor disk bolt hole — local stress and strain at notch root from elastic FEM + Neuber hyperbola intersection with cyclic stress-strain curve.

Challenge: Neuber's rule is non-conservative for plane strain (thick sections) — Molski-Glinka energy method gives more accurate notch root strain in triaxial stress fields. Compared both methods for the disk bolt hole geometry — Molski-Glinka gave 12% lower notch strain (less conservative, but more accurate against strain gauge data).

Learned: Neuber's rule is conservative but can be significantly so in triaxial stress fields. Molski-Glinka (strain energy density) is more accurate for thick sections. When test data is available, validate both methods before selecting one for routine use.