AERODYNAMIC FORCING & DAMPING COMPUTATIONS IN A TRANSONIC AXIAL TURBINE OF A TURBOCHANRGER USING TIME & FREQUENCY DOMAIN FLOW TRANSFORMATION TECHNIQUES

Abstract: The current report discusses efficient and accurate numerical techniques to evaluate aerodynamic disturbance and damping forces based on investigations carried out at ABB Turbo System AG. The test turbomachine comprises a single-stage transonic axial turbine analysed with the fluid dynamic software ANSYS CFX 17.0 and post-processing tool Matlab. In the first stage, the 23 unequally spaced stators 33 rotors turbine configuration was scaled to two configurations with 22 and 24 equally spaced stators respectively and 33 rotors. These configurations permit integer reduction of blade count across adjacent blade rows and as such a reduced computational domain. An innovative time domain signal patching routine was implemented to recover the original 23 stators 33 rotors blade forces in the cylindrical direction and generalised force from the 22 and 24 stators configurations with reasonable accuracy. Computational savings of 57.5% were enabled with such an approach. To further reduce computational demand, the two configurations with 22 and 24 equally spaced stators were simulated with time domain flow field transformation methods; profile transformation, time inclination and Fourier method. A comparison of total blade and midspan forces in cylindrical direction, generalised force and local unsteady pressures demonstrated the capacity of the time inclination and Fourier methods to capture harmonic data within an accuracy of 15%. Moreover these time domain flow field transformation methods when used in conjunction with the signal patching routine allowed for computational savings larger than 93%. In the second stage, two different strategies to extract aerodynamic damping forces were compared. The first termed aerodynamic influence coefficient method (AIC) takes into account the aerodynamic influence of an oscillating blade on neighbouring blades in a blade row was introduced. This approach enables a single numerical simulation for the construction of an aerodynamic damping curve. The second approach involves simulating each travelling wave mode and extracting the corresponding unsteady aerodynamic work for damping derivation termed as the energetic approach (EA). An evaluation into the accurate set-up of the AIC method indicated the need to run a full 33 rotors simulation without the assignment of spatial periodic boundaries to limit false boundary settings. Moreover, the influence across 23 of the blades needed to be asses to obtain an accurate aerodynamic damping curve. The energetic approach was carried out using a simple transient method, the time domain flowfield transformation Fourier method and the time/frequency domain flowfield transformation harmonic balance method. Both the Fourier and harmonic balance methods predicted identical aerodynamic damping values to the transient case at three interblade phase angles with large computational savings. However, instabilities at backward travelling waves for the harmonic balance limited its application. Identical aerodynamic damping predictions with the AIC method, energetic approach with Fourier method and harmonic balance methods verified three numerical aerodynamic damping prediction approaches. A comparison of computational cost however suggests that the Fourier and harmonic balance methods are more efficient techniques for future damping computations.