Monika Szymanska

2025-02

Mistuning analysis of HPT bladed disk — intentional mistuning pattern optimization to reduce forced response amplification factor below 1.3 (target), using sector FEM with cyclic symmetry.

Challenge: Cyclic symmetry analysis assumes tuned (identical) sectors. Mistuning breaks this assumption — required full 360-degree model with individually perturbed blade properties. Full model had 8M DOF; solved overnight on HPC cluster.

Learned: Even small mistuning (±0.5% frequency scatter) can amplify forced response by factors of 2-3 in unlucky engine orders. Intentional mistuning can counteract this but requires careful optimization of the pattern.

2024-08

Test-analysis correlation using frequency response functions from vibration rig test — measured FRF with LMS Test.Lab vs predicted FRF from ANSYS, MAC matrix and FRAC values for mode pairing.

Challenge: Several measured modes had MAC values < 0.7 against predicted modes — unacceptable for certification correlation. Root cause: missing bolted joint contact in the FEM. Adding friction contact at all bolted interfaces improved MAC to > 0.85 for all modes.

Learned: Bolted joints are the primary source of discrepancy between FEM and test modal results. Modeling them as rigidly connected is almost always non-conservative. At minimum, use spring elements with calibrated stiffness.

2024-03

Component Mode Synthesis (CMS) for large engine assembly — each major component reduced to Craig-Bampton superelement, assembled for system-level modal analysis. Reduced model from 4M DOF to 12K DOF.

Challenge: Superelement interface DOF selection is critical — too few interface modes and the reduced model is inaccurate at component boundaries. Ran convergence study with 20, 40, and 80 interface modes to identify minimum required set.

Learned: CMS accuracy depends on correctly identifying which interface DOFs are dynamically active. Static condensation interface modes are mandatory; the number of dynamic interface modes is problem-specific.

2023-10

Random vibration analysis for avionic box qualification — input PSD from DO-160G profile, output acceleration PSD at mounting points, 3-sigma stress for fatigue damage assessment.

Challenge: The spectral density response was dominated by a mode that fell exactly at the PSD breakpoint — small errors in frequency prediction caused large changes in spectral response. Required test correlation before relying on fatigue predictions.

Learned: Random vibration fatigue using Miles equation assumes narrowband response. For broadband excitation with multiple contributing modes, the more conservative Dirlik method gives more accurate fatigue damage estimates.

2023-05

Harmonic response analysis of engine mount bracket — swept frequency 20 to 2000 Hz, peak response identification, fatigue assessment at maximum acceleration location.

Challenge: Modal damping was unknown — used assumed damping ratio of 2% initially. Sensitivity study showed that the peak response at 847 Hz was very sensitive to damping: 1% vs 3% damping changed peak stress by 3x. Escalated to get measured damping from test.

Learned: Harmonic response results are extremely sensitive to damping near resonance. Assumed damping values must be conservative (low) and sensitivity studies are mandatory when test-measured damping is unavailable.

2023-01

Modal analysis of fan blade assembly — extracted first 20 mode shapes and natural frequencies, constructed Campbell diagram to identify potential resonance crossings with engine excitation orders (EO1-EO24).

Challenge: Two bending modes were within 5% of each other in frequency and had similar mode shapes — difficult to assign labels (1F vs 2F) unambiguously. Used Modal Assurance Criterion (MAC) matrix to establish mode tracking across operating speeds.

Learned: Campbell diagram resonance crossings must be assessed against the mode shape, not just the frequency. A crossing at low excitation amplitude is benign; the same crossing with high aerodynamic forcing is a design concern.