Urszula Trybała
Turbine Cooling Engineer · Thermal Engineering
Warsaw, Poland
Turbine Cooling Engineer specializing in conjugate heat transfer analysis of HPT blades and NGVs. Uses STAR-CCM+ for high-fidelity CHT simulations that predict metal temperatures and cooling effectiveness distributions. Works closely with turbomachinery and materials teams to translate thermal predictions into cooling design requirements and to validate against cascade rig test data.
Expertise
- conjugate heat transfer (CHT) CFD
- HPT blade and vane internal cooling design
- impingement and convective cooling
- turbine metal temperature prediction
- cooling effectiveness validation against rig data
Technologies
Work History
2025-01
Rapid thermal screening tool development — built a reduced-order thermal model (ROTM) of the HPT blade using a network of 1D thermal resistances calibrated against CHT results. Enables 100x faster thermal assessment during early design iterations.
Challenge: Calibrating the ROTM against CHT results at only 3 operating points resulted in poor prediction at off-design conditions. Required 12 CHT calibration cases to cover the operating envelope adequately — the reduced model behaved non-linearly outside the calibration range.
Learned: Reduced-order thermal models for turbine blades need calibration data across the full operating range, not just at design points. The mapping from CHT to network model parameters is non-linear enough that extrapolation beyond the calibration range is unreliable.
2024-05
Cascade rig test correlation — compared CHT predictions against metal temperature measurements from a high-temperature cascade rig test. 47 thermocouple measurement locations on 3 instrumented blades.
Challenge: Predictions showed 15-25°C overprediction on the blade pressure surface midchord region. Root cause investigation revealed the external heat transfer coefficient was overpredicted by the k-omega SST model in the laminar-to-turbulent transition region. Switching to the gamma-theta transition model reduced the error to <8°C.
Learned: Turbine blade external heat transfer prediction accuracy depends critically on boundary layer transition model selection. The k-omega SST assumes fully turbulent flow from the leading edge — this overpredicts heat transfer in the laminar-to-transition region and causes systematic metal temperature overprediction.
2023-10
Internal cooling channel parameterization study — assessed impact of cooling channel cross-section shape, aspect ratio, and trip strip geometry on heat transfer coefficient using STAR-CCM+ with SST k-omega turbulence model.
Challenge: Trip strip simulations with fully resolved roughness geometry required a very fine near-wall mesh (y+ < 1 on roughness elements). Total cell count exceeded 80 million — single-run memory requirements on standard HPC nodes (256GB RAM) were exceeded. Required decomposition onto 4 nodes with careful domain partitioning.
Learned: Trip strip heat transfer CFD requires either fully resolved roughness geometry (extremely expensive) or validated enhanced wall functions with roughness correction. Confirm memory and compute budgets before committing to a fully resolved roughness mesh strategy.
2023-03
Conjugate heat transfer analysis for HPT stage 1 blade — modeled external convective heat transfer, internal serpentine cooling channel, and TBC layer simultaneously. Predicted blade metal temperature distribution at T/O and cruise.
Challenge: CHT simulation required co-simulation between the gas path CFD (hot side) and the internal cooling channel flow — coupling the two fluid domains through the solid blade with the correct heat flux boundary conditions. Convergence was unstable without a thermal resistance relaxation factor between coupling iterations.
Learned: CHT co-simulation convergence for turbine blades requires a relaxation factor on the coupling interface. Tight coupling without relaxation causes heat flux oscillations between domains that prevent convergence. A relaxation factor of 0.3-0.5 on the interface heat flux typically gives stable convergence.