Numerical Methods for Turbomachinery Aeromechanical Predictions

University dissertation from Stockholm : Royal Institute of Technology

Abstract: In both aviation and power generation, gas turbines are used as key components. An important driver of technological advance in gas turbines is the race towards environmentally friendly machines, decreasing the fuel burn, community noise and NOx emissions. Engine modifications that lead to propulsion efficiency improvements whilst maintaining minimum weight have led to having fewer stages and lower blade counts, reduced distance between blade rows, thinner and lighter components, highly three dimensional blade designs and the introduction of integrally bladed disks (blisks). These changes result in increasing challenges concerning the structural integrity of the engine. In particular for blisks, the absence of friction at the blade to disk connections decreases dramatically the damping sources, resulting in designs that rely mainly on aerodynamic damping. On the other hand, new open rotor concepts result in low blade-to-air mass ratios, increasing the influence of the surrounding flow on the vibration response. This work presents the development and validation of a numerical tool for aeromechanical analysis of turbomachinery (AROMA - Aeroelastic Reduced Order Modeling Analyses), here applied to an industrial transonic compressor blisk. The tool is based on the integration of results from external Computational Fluid Dynamics (CFD) and Finite Element (FE) solvers with mistuning considerations, having as final outputs the stability curve (flutter analysis) and the fatigue risk (forced response analysis). The first part of the study aims at tracking different uncertainties along the numerical aeromechanical prediction chain. The amplitude predictions at two inlet guide vane setups are compared with experimental tip timing data. The analysis considers aerodynamic damping and forcing from 3D unsteady Navier Stokes solvers. Furthermore, in-vacuo mistuning analyses using Reduced Order Modeling (ROM) are performed in order to determine the maximum amplitude magnification expected. Results show that the largest uncertainties are from the unsteady aerodynamics predictions, in which the aerodynamic damping and forcing estimations are most critical. On the other hand, the structural dynamic models seem to capture well the vibration response and mistuning effects. The second part of the study proposes a new method for aerodynamically coupled analysis: the Multimode Least Square (MLS) method. It is based on the generation of distributed aerodynamic matrices that can represent the aeroelastic behavior of different mode-families. The matrices are produced from blade motion unsteady forces at different mode-shapes fitted in terms of least square approximations. In this sense, tuned or mistuned interacting mode families can be represented. In order to reduce the domain size, a static condensation technique is implemented. This type of model permits forced response prediction including the effects of mistuning on both the aerodynamic damping as well as on the structural mode localization. A key feature of the model is that it opens up for considerations of responding mode-shapes different to the in-vacuo ones and allows aeroelastic predictions over a wide frequency range, suitable for new design concepts and parametric studies.

  This dissertation MIGHT be available in PDF-format. Check this page to see if it is available for download.