Beata Jankowska
CFD Engineer — Turbine Cooling · Aerodynamics & CFD
Warsaw, Poland
CFD Engineer specializing in turbine hot-section thermal analysis. Performs conjugate heat transfer calculations for HPT blades and disks, film cooling effectiveness studies, and cavity flow analyses. Provides thermal boundary conditions (heat flux maps, HTC distributions) to structural engineers for FEM thermal-structural analysis.
Expertise
- conjugate heat transfer (CHT)
- film cooling CFD
- turbine disk cavity flow
- CFD-FEM thermal coupling
- ANSYS Fluent
Technologies
Work History
2024-10
Adjoint-based optimization of cooling channel geometry — ANSYS Fluent adjoint solver provides sensitivity of thermal performance to shape parameters. Used to guide redesign of coolant supply duct.
Challenge: Adjoint solver in Fluent is much less robust than the primal solver — converged primal solutions often diverge in adjoint mode with default settings. Required significant under-relaxation and a good initial sensitivity estimate to stabilize.
Learned: Adjoint methods provide gradient information at the cost of one additional solve — extremely efficient for high-dimensional shape optimization. But they require a well-converged primal field and are sensitive to numerical stability issues near boundaries.
2024-04
Multi-stage HPT 3D CFD — full annular computation with mixing plane interface between NGV (nozzle guide vane) and rotor. Extracted NGV-rotor interaction and secondary flow structures.
Challenge: Multi-stage mesh with mixing plane required careful interface matching — small mesh misalignment caused non-physical pressure discontinuities at the interface. Implemented automated interface quality check in Python before each run.
Learned: Mixing plane convergence is sensitive to the averaging approach (area-weighted vs mass-weighted) for non-uniform inlet conditions. In highly distorted flow fields, mass-weighted averaging is more physically appropriate.
2023-11
HPT disk cavity flow analysis — pre-swirl system, bore cooling flow, parasitic leakage paths. Steady-state RANS in rotating frame with MRF (Multiple Reference Frame) for disk rotation.
Challenge: MRF approach gave inaccurate results near the rotor-stator interface due to strong swirl gradients. Switched to full transient rotating mesh (sliding mesh) for 5 seal teeth — 10x more expensive but required for accurate leakage prediction.
Learned: MRF is adequate for flows where the interface relative velocity is small. For rotor-stator gaps with high swirl (common in turbine disk cavities), sliding mesh is necessary to capture the unsteady mixing at the interface.
2023-06
Film cooling effectiveness study — compared cylindrical holes vs fan-shaped holes vs sister holes for HPT leading edge. RANS (k-omega SST) turbulence model, blowing ratios M = 0.5, 1.0, 1.5, 2.0.
Challenge: k-omega SST significantly overpredicts film cooling effectiveness for fan-shaped holes at M > 1.0 — literature-known issue. v2f turbulence model was more accurate but not natively available in Fluent. Implemented as a UDF (User Defined Function).
Learned: Turbulence model selection for film cooling CFD is not trivial. k-omega SST is adequate for cylindrical holes at low blowing ratios. Advanced models (v2f, RSM) are necessary for shaped holes and high blowing ratios. Always validate against benchmark data.
2023-01
Conjugate heat transfer analysis of HPT blade cooling — external hot gas film, internal cooling channels with impingement and film holes, full 3D CHT model in ANSYS Fluent.
Challenge: Mesh generation for internal cooling channels with shaped film holes was extremely time-consuming — 4 weeks for a single blade. Complex hole geometries (fan-shaped, laidback fan-shaped) required structured mesh blocks that could not be generated automatically.
Learned: For film-cooled blade CHT, mesh generation is the longest part of the project. Parametric mesh templates using blocking in ICEM CFD reduce setup time for geometry variants from weeks to days.