Piotr Wisniewski

2025-02

Probabilistic fatigue analysis using Monte Carlo simulation — sampled material scatter (S-N curve parameters), geometric scatter (notch radius tolerance), and load scatter to generate a distribution of fatigue lives.

Challenge: Monte Carlo with 10,000 simulations and NASGRO crack growth for each was computationally expensive — 8 hours on a single machine. Parallelized with Python multiprocessing across 12 cores, reducing runtime to 45 minutes.

Learned: Probabilistic analysis reveals which input parameters dominate life scatter — in this case, material S-N curve slope (m exponent) was the dominant contributor, not geometric or load scatter. This prioritizes which parameters need tighter statistical characterization.

2024-10

NASGRO crack growth analysis for assumed initial crack scenario at disk bore — grew crack from NDE detection threshold (0.5mm) to critical size, computed inspection intervals for damage tolerance per FAR 33.70.

Challenge: NASGRO stress intensity factor (K) solutions are only available for standard crack geometries (elliptical surface crack, through-thickness crack). The disk bore had a non-standard combined geometry requiring FEM-extracted K values imported as a user-defined K table in NASGRO.

Learned: NASGRO's built-in K solutions cover the most common crack types but not complex structural geometries. FEM-extracted stress intensity factors using J-integral or interaction integral method can be used as input — this significantly extends NASGRO's applicability to non-standard cases.

2024-04

Fatigue life prediction using FEM stress field input to nCode DesignLife — full surface stress field from ANSYS imported into nCode, E-N analysis at every surface node, life contour plot identifying minimum life location.

Challenge: ANSYS-to-nCode interface required results in a specific ODB format — direct OP2/RST file import was supported but had version compatibility issues between ANSYS 2023 R2 and nCode 2022. Required intermediate CSV export as workaround.

Learned: FEM-to-fatigue post-processor data transfer is a known pain point. Maintaining compatible versions of ANSYS and nCode DesignLife requires coordination between the tools team and fatigue team — version freezes should be synchronized.

2023-12

Combined LCF+HCF interaction analysis using Crossland criterion — maps combined mean stress (LCF contribution) and alternating stress (HCF vibration) onto the Crossland fatigue limit surface.

Challenge: Crossland criterion requires hydrostatic mean stress — not just the maximum principal mean stress. Computing hydrostatic stress tensor from ANSYS results requires extracting all 6 stress components, not just von Mises or principal values.

Learned: Multiaxial fatigue criteria that use hydrostatic stress (Crossland, Dang Van) require the full stress tensor as input. Tools that only output von Mises or maximum principal stress are insufficient for these criteria. Always verify what your FEM post-processor is actually computing.

2023-08

LCF life analysis of HPT disk — strain-life approach (Coffin-Manson) with Neuber notch correction at critical locations. Mission analysis: takeoff, cruise, descent, ground. Cycle extraction by rainflow on temperature-load history.

Challenge: Disk mission analysis requires tracking stress-temperature history simultaneously — both affect life. nCode DesignLife's thermal fatigue module was used but required input in a specific format that took two weeks to set up correctly from ANSYS results.

Learned: LCF life of hot section components requires synchronized stress and temperature histories per point. Post-processing scripts that extract both at the same time step are critical. Point-by-point ANSYS result extraction for 10,000 nodes and 200 time steps requires efficient array operations.

2023-03

HCF analysis of HPT blade under engine-order resonance — vibratory stress from ANSYS modal + forced response analysis, mean stress from centrifugal and thermal loads, Goodman diagram check against Haigh diagram material data.

Challenge: Goodman diagram material data was available only for ambient temperature — blade operates at 800°C in HPT. Applied temperature knockdown factors from elevated-temperature Haigh diagram data in the material database. Factor was 0.52 on the fatigue limit at 800°C.

Learned: Temperature knockdown on fatigue strength for Ni-based superalloys is dramatic — at 800°C, the fatigue limit can be less than half the room temperature value. Always confirm that your Goodman diagram is for the operating temperature, not ambient.