[home] [Personal Program] [Help]
tag
16:30
30 mins
A Strongly Coupled Frequency Domain FSI Solver for Turbomachinery Blade Vibration Analysis
Johann Gross, Christian Berthold, Christian Frey, Malte Krack
Session: Rotary aeroelasticity 2
Session starts: Tuesday 18 June, 16:00
Presentation starts: 16:30
Room: Room 1.6


Johann Gross (University of Stuttgart)
Christian Berthold (German Aerospace Center)
Christian Frey (German Aerospace Center)
Malte Krack (University of Stuttgart)


Abstract:
The efficiency requirements of modern turbomachinery lead to light and flexible blades that promote potentially damaging vibrations. Specifically, turbine blade vibrations and their interactions with the surrounding flow can be significantly nonlinear. In order to reliably predict the nonlinear behavior and achieve endurable blade design, strongly coupled FSI solvers are essential. In this work, we present the first nonlinear Frequency Domain Fluid-Structure Interaction solver for blade vibration analysis. The solver is based on the Harmonic Balance method applied to fluid and structure domains. We utilize the cyclic symmetry of the problem to reduce the multi-physics computational domain to only one sector, which is validated with full annulus simulations in the time domain. Based on a realistic aeroelastic model of a bladed disk with interlocked shrouds we demonstrate the computational performance of the solver. We perform flutter simulations, where the initial exponential amplitude growth is bounded by dry friction between shrouds. Amplitude-dependent change of the aeroelastic vibration shape and frequency are accounted for, which in contrast to conventional methods enables accurate limit cycle prediction and even detection of the nonlinear instability phenomenon, i.e. a situation, where the initially stable equilibrium becomes unstable due to a sufficiently strong perturbation (e.g. impact or forced excitation). Further, we propose two variants of numerical path continuation framework for the solver. The first variant is a straight forward coupling along the sequential computation of dynamic equilibrium points. The second variant relies on an already available or easily computed prediction of the solution curve, e.g. by initially considering the aerodynamic influence in a linearized manner. Subsequently, a fully coupled refinement is performed on each of the relevant solution points in parallel. Both variants are applied to a forced response computation of the same model, the advantages of each variant are assessed. We are convinced that the proposed method unlocks the potential towards the necessary increase of endurable turbine blade design.