A computational analysis of the aerodynamic and aeromechanical behavior of the purdue multistage compressor

David James Winchester Monk, Purdue University

Abstract

Compressor design programs are becoming more reliant on computational tools to predict and optimize aerodynamic and aeromechanical behavior within a compressor. Recent trends in compressor development continue to push for more efficient, lighter weight, and higher performance machines. To meet these demands, designers must better understand the complex nature of the inherently unsteady flow physics inside of a compressor. As physical testing can be costly and time prohibitive, CFD and other computational tools have become the workhorse during design programs. The objectives of this research were to investigate the aerodynamic and aeromechanical behavior of the Purdue multistage compressor, as well as analyze novel concepts for reducing rotor resonant responses in compressors. Advanced computational tools were utilized to allow an in-depth analysis of the flow physics and structural characteristics of the Purdue compressor, and complement to existing experimental datasets. To analyze the aerodynamic behavior of the compressor a Rolls-Royce CFD code, developed specifically for multistage turbomachinery flows, was utilized. Steady-state computations were performed using the RANS solver on a single-passage mesh. Facility specific boundary conditions were applied to the model, increasing the model fidelity and overall accuracy of the predictions. Detailed investigations into the overall compressor performance, stage performance, and individual blade row performance were completed. Additionally, separation patterns on stator vanes at different loading conditions were investigated by plotting pathlines near the stator suction surfaces. Stator cavity leakage flows were determined to influence the size and extent of stator hub separations. In addition to the aerodynamic analysis, a Rolls-Royce aeroelastic CFD solver was utilized to predict the forced response behavior of Rotor 2, operating at the 1T mode crossing of the Campbell Diagram. This computational tool couples aerodynamic predictions with structural models to determine maximum Rotor 2 vibration amplitudes excited by both vortical and potential disturbances. A multi-bladerow, full-annulus unsteady simulation was performed to capture the aerodynamic forcing functions and understand the influence of bladerow interactions on these flow disturbances. The strength and frequency content of the S1 vortical field and S2 potential field were examined to quantify the aerodynamic forces exciting resonant vibrations. Detailed comparisons were made to experimental datasets acquired on the Purdue compressor which characterize the forced response behavior at the 1T mode crossing. Lastly, stator asymmetry was examined as a means of reducing forced response vibration amplitudes. For this study, a new Stator 1 ring was designed with a reduced vane count, creating the ability to isolate the relative contribution of the S1 wakes on R2 vibrational amplitudes. A second Stator 1 ring was then designed with asymmetric vane spacing such that two stator half-sectors of different vane counts were joined together to form a full stator ring. By joining two stator half-sectors with different vane counts, the energy of the wakes is spread into additional frequencies, thereby reducing the overall amplitudes. The aeroelastic CFD solver was again used to perform steady-state and unsteady simulations, capturing the effect of the stator asymmetry on resonant vibrational amplitudes. The resulting blade deflection amplitudes are presented and discussed in detail.

Degree

M.S.M.E.

Advisors

Key, Purdue University.

Subject Area

Aerospace engineering|Mechanical engineering

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS