In this study, two time-accurate Navier-Stokes analyses were obtained to predict the first-vane/first-blade interaction in a 1 and 1/2-stage turbine rig for comparison with measurements. In the first computation, airfoil scaling was applied to the turbine blade to achieve periodicity in the circumferential direction while modeling 1/18 of the annulus. In the second, 1/4 of the wheel was modeled without the use of airfoil scaling. For both simulations the predicted unsteady pressures on the blade were similar in terms of time-averaged pressure distributions and peak-peak unsteady pressure envelopes. However, closer inspection of the predictions in the frequency domain revealed significant differences in the magnitudes of unsteadiness at twice vane-passing frequency (and the vane-passing frequency itself, to a lesser extent). The results of both computations were compared to measurements of the vane-blade interaction in a full-scale turbine rig representative of an early design iteration of the PW6000 engine. These measurements were made in the short-duration turbine-test facility at The Ohio State University Gas Turbine Laboratory. The experimentally determined, time-resolved pressures were in good agreement with those predicted with the 1/4-wheel simulation.

The ability to predict levels of unsteady forcing on high-pressure turbine blades is critical to avoid high-cycle fatigue failures. In this study, 3D time-resolved computational fluid dynamics is used within the design cycle to predict accurately the levels of unsteady forcing on a single-stage high-pressure turbine blade. Further, nozzle-guide-vane geometry changes including asymmetric circumferential spacing and suction-side modification are considered and rigorously analyzed to reduce levels of unsteady blade forcing. The latter is ultimately implemented in a development engine, and it is shown successfully to reduce resonant stresses on the blade. This investigation builds upon data that was recently obtained in a full-scale, transonic turbine rig to validate a Reynolds-Averaged Navier-Stokes (RANS) flow solver for the prediction of both the magnitude and phase of unsteady forcing in a single-stage HPT and the lessons learned in that study.

No abstract