Joanna Lewandowska

2025-02

CFD-FEM workflow automation for thermal analysis — Python pipeline that takes Fluent solution files, maps HTC to ANSYS mesh, runs thermal FEM, exports temperature to structural FEM, and generates PDF summary report.

Challenge: The pipeline had 5 steps with different file formats — Fluent CAS/DAT, ANSYS RST, ANSYS MAPDL commands, and a custom HTC transfer format. Each format required a specific parser. The most fragile step was the ANSYS batch command file generation — APDL syntax errors caused silent failures.

Learned: Automation pipelines for multi-physics workflows are highly valuable but require robustness at every hand-off point. Each step must verify its output before passing it to the next step. Silent failures in intermediate steps produce wrong final results with no error message.

2024-11

Thermo-mechanical fatigue (TMF) assessment of HPT blade airfoil — in-phase vs out-of-phase TMF loading, temperature and mechanical strain cycling, TMF damage parameter using Ostergren energy criterion.

Challenge: TMF life data for the alloy was available only at strain ratios of -1 (fully reversed) and 0 (zero-to-max). Blade operating loading had positive mean strain (IP-TMF with mean tensile strain). Had to extrapolate to positive mean strain using a modified Smith-Watson-Topper equivalent.

Learned: TMF is a mode that combines thermal and mechanical loading with complex phase relationships. IP-TMF (in-phase — max T at max mechanical strain) is usually more damaging than OP-TMF for most alloys. Test data must cover the actual phase angle of the component loading.

2024-05

Thermal barrier coating (TBC) stress analysis — three-layer model: substrate (Ni superalloy), bond coat (MCrAlY), thermally grown oxide (TGO), ceramic topcoat (YSZ). Computed peel stress at TGO-topcoat interface during thermal cycling.

Challenge: TGO layer is only 2-8 microns thick — requiring extremely fine mesh that created a massive element count model. Used 1D analytical approximation for TGO oxidation growth stress to derive a correction factor on the 3D FEM without explicit TGO meshing.

Learned: Thin-layer stress analysis (TGO, adhesive films, coatings) is best handled with analytical solutions or reduced-order models rather than explicit meshing when the layer thickness is 3+ orders of magnitude smaller than the substrate. FEM of thin layers gives poor accuracy unless the aspect ratio is very carefully controlled.

2023-12

Creep analysis of HPT blade using Norton-Bailey power law — multi-step analysis with takeoff (high T, high stress) and cruise hold times. Computed creep strain accumulation per flight cycle and integrated over design life.

Challenge: Norton-Bailey parameters for the Ni superalloy were fitted at 800°C and 900°C but the blade operates across 700-950°C depending on location. Temperature-interpolated parameters were required — but the power law exponent n changes significantly with temperature, making interpolation non-trivial.

Learned: Norton-Bailey creep law parameters must be interpolated with respect to temperature using physically appropriate functions — exponential interpolation for the rate constant A, linear for the exponent n. Simple linear interpolation of A can give orders-of-magnitude errors.

2023-08

Transient thermal-structural analysis for engine start-up and shut-down cycles — differential thermal expansion between disk and casing during rapid temperature transient. Identified critical interference condition during cold re-start.

Challenge: Transient thermal analysis requires small time steps during rapid heat-up — disk and casing have very different thermal masses. Adaptive time stepping with minimum step 0.1s and maximum 60s was necessary. Total of 4000+ timesteps for a 20-minute start-up simulation.

Learned: Thermal transient analyses for engine components require significant thought on time step strategy. The governing thermal time constant is different for each component — disk boreis much slower than blade trailing edge. Time step must resolve the fastest-changing component.

2023-02

Steady-state thermal analysis of HPT disk using ANSYS Mechanical — temperature boundary conditions from CFD disk cavity analysis (ANSYS Fluent), conduction through disk, radiation from hot gas surfaces.

Challenge: Mapping CFD heat transfer coefficients from Fluent unstructured mesh to ANSYS structural mesh required interpolation between non-matching grids. Built Python script using scipy.interpolate with natural neighbor interpolation — reduces heat flux error vs bilinear to <2% on the disk bore.

Learned: CFD-to-FEM thermal boundary condition mapping quality directly affects temperature prediction accuracy. Natural neighbor interpolation preserves local heat flux gradients better than distance-weighted average, especially near film cooling holes where the gradient changes rapidly.