In turbocharger application bleed air at impeller exit is typically used to seal bearing compartments and to balance axial thrust in the rotor. It was previously shown that this bleed air can have a significant impact on both compressor performance and stability. Experiments suggest that spike stall inception in centrifugal compressors can be formed by a vaned diffuser. To address these issues, a numerical study on an advanced, vaned-diffuser centrifugal compressor was conducted to investigate stall inception. A steady three-dimensional Reynolds-averaged Navier-Stokes simulation using a mixing plane was carried out first to evaluate the effects of bleed air at impeller exit on stage and diffuser subcomponent performance. The steady simulation was compared with experimental measurements and did not show significant changes in stage and subcomponent performance due to leakage flow as observed in the experiments, indicating the importance of unsteady flow effects in the vaneless space and adjacent bleed cavity. Next, an unsteady three-dimensional Reynolds-averaged Navier-stokes simulation was carried out on four vaned diffuser passages to investigate the response of the diffuser flow field to short wavelength inlet disturbances in total pressure. The simulation employed a new approach, using circumferentially-averaged diffuser inlet conditions obtained from the steady stage simulation, eliminating the impeller and significantly reducing the computational time. This method was capable of simulating spike-like stall precursors rotating at 66% rotor speed which formed in response to inlet flow disturbances. The results represent a first numerical simulation of rotating spike-like flow disturbances in a radial vaned diffuser, and suggest that the spike stall precursors are formed by the vaned diffuser in absence of a tip leakage flow as it can occur in the rotors of axial compressors. Thesis Supervisor: Zoltán S. Spakovszky Title: H.N. Slater Associate Professor of Aeronautics and Astronautics
32 Figures and Tables
Figure 1-1: High pressure ratio centrifugal compressor of advanced design. Note leakage location at impeller exit. Figure adopted from .
Figure 1-2: Effect of bleed air on compressor performance at 100% and 105% corrected speed; circles: compressor rig test (no bleed); diamonds: turbocharger test (bleed). Figure adopted from .
Figure 1-3: Static pressure traces measured in the vaneless space showing spikes (right) and modes (left) driven by the leakage flow. Figure adopted from .
Figure 2-1: Fine mesh projected on solid surfaces. Shroud endwalls not shown for clarity.
Figure 2-10: Measured static pressure rise of diffuser subcomponents at 100% corrected speed. Figure adopted from .
Figure 2-11: Calculated static pressure rise in diffuser subcomponents. The scale shown here occupies approximately the right third of the scale shown in Figure 2-10
Figure 2-12: Calculated momentum thickness through the diffuser passage, showing a reduction in the hub endwall boundary layer in the vaneless space when leakage flow is bled from the impeller exit.
Figure 2-13: Plot of blockage quantity through vaneless space of diffuser between impeller exit (at 100% tip radius) and mixing plane (at 107% tip radius) for an operating point near stall.
Figure 2-14: Sketch of endwall flow field with inlet normal vorticity. Figure adopted from .
Figure 2-17: Streamlines originating at 0.4% span (top) and 50% span (bottom) from the hub endwall, showing the differences in streamline curvature between boundary layer and core flow in the diffuser passage.
Figure 2-18: View of diffuser vane hub-corner for simulation cases with and without leakage flow at impeller exit. The diffuser inlet corrected flow is similar in the two cases and near the stall point.
Figure 2-2: Meridional view of computational domain, showing shroud and hub contours and key details of the compressor geometry.
Figure 2-4: Typical throttle, backpressure, and compressor characteristics.
Figure 2-5: View of leakage flow mesh set-up (top) and resultant velocity through bleed slot (bottom).
Figure 2-6: Bulk performance of simulated compressor stage at 100% corrected speed.
Figure 2-7: Comparison of pTinlet/pexit between experiments and steady CFD simulations with and without leakage flow at impeller exit.
Figure 2-8: Overall performance of compressor stage as a function of diffuser inlet corrected flow.
Figure 2-9: Qualitative depiction of change in diffuser static pressure rise due to endwall leakage flow. Figure adopted from .
Table 2.1: Summary of pertinent grid and stage geometry characteristics
Figure 3-1: Axial view of four-passage diffuser mesh used for unsteady diffuser calculations. Shown in the figure are the radial coordinates of the inlet and outlet boundaries.
Figure 3-10: Pitchwise-averaged total pressure inlet boundary condition profile projected over the quarter-circumference.
Figure 3-11: Total pressure inlet boundary condition profile from Figure 3-10 with disturbance added.
Figure 3-12: Contours of total pressure illustrating the input disturbance as it passes through the diffuser passage.
Figure 3-13: Location of a line of simulated unsteady pressure transducers used to examine the effects of the forcing input on the flow in the vaneless space.
Figure 3-2: Plots of inlet boundary conditions, illustrating the relative non-uniformity in velocity components and total quantities.
Figure 3-3: Inlet Boundary Conditions
Figure 3-4: Diffuser inlet total to exit static pressure ratio characteristic for experiment, stage simulation, and for boundary conditions used in unsteady diffuser-only calculations.
Figure 3-5: Total pressure ratio characteristic for stage calculation and experimental results without leakage flow. Stall point, as defined previously, is circled.
Figure 3-6: Diffuser static pressure rise coefficient calculated from diffuser-only, stage, and experimental data.
Figure 3-7: Diffuser subcomponent characteristics for time-averaged unsteady diffuser-only calculations.
Figure 3-8: Diffuser static pressure rise coefficient. Inset shows operating points A, B, and C used for forced disturbance response.
Figure 3-9: Axial view of diffuser showing approximate location of forced perturbation.
Download Full PDF Version (Non-Commercial Use)