M. Vahdati

Modeling of Three-Dimensional Viscous Compressible Turbomachinery Flows Using Unstructured Hybrid Grids

An advanced numerical model for the simulation of steady and unsteady viscous compressible e ows for turbomachinery applications is described. The compressible Favre-averaged Navier ‐Stokes equations are used together with a one-equation turbulence model. The e ow domain is discretized using unstructured hybrid grids that can contain a mixture of hexahedral, pentahedral, tetrahedral, and triangular prismatic cells. The e ow equations are discretized using a node-centered e nite volume scheme that relies on representing the mesh using an edge-based data structure. A dual time stepping technique is applied to a point implicit formulation so that time accuracy can be maintained with large Courant ‐Friedrichs‐Lewy numbers. Nonree ecting boundary conditions are applied at the ine ow and oute ow boundaries to prevent any spurious ree ections of the outgoing waves. The model was validated against measured data for two cases. Radial proe les of pressure and temperature rise were determined from the steady e ow analysis of a rig fan blade, and these were found to be in very good agreement with the measured quantities. A rotor/stator interaction was studied next. Detailed comparisons were carried out against measured steady and unsteady e ow data and good agreement was obtained in all cases.

Aeroelasticity analysis of air-riding seals for aero-engine applications

This paper presents the results of a feasibility study on air-riding seal aeroelasticity for large-diameter aero-engines. A literature survey of previous seal studies revealed a significant amount of experimental work but numerical modeling using CFD techniques was relatively scarce. Indeed, most existing theoretical studies either deal with the structural behavior, or use simplified flow modeling. The aeroelasticity stability of a simplified air-riding seal geometry, devised for this particular feasibility study, was analyzed in three dimensions for typical engine operating conditions. Both the unsteady flow and structural vibration aspects were considered in the investigation. The boundary conditions and the seal gap were varied to explore the capabilities and limitations of a state-of-the-art unsteady flow and aeroelasticity code. The methodology was based on integrating the fluid and structural domains in a time-accurate fashion by exchanging boundary condition information at each time step. The predicted characteristics, namely lift and flow leakage as a function of pressure and seal gap, were found to be in agreement with the expected behavior. Operating seal gaps were determined from the actual time histories of the seal motion under the effect of the aerodynamic and the restoring spring forces. Both stable and unstable cases were considered. It was concluded that, in principle, the existing numerical tools could be used for the flow and aeroelasticity analyses of hydrostatic seals. However, due to large Mach number variations, the solution convergence rate was relatively slow and it was recognized that a preconditioner was needed to handle seal flows. For small gaps of about 10 microns, typical of spiral groved seals, the flow has a high Knudsen number, indicating that the Navier-Stokes formulations may no longer be valid. Such cases require a totally different treatment for the modeling of steady and unsteady aerodynamics, either by modifying the transport parameters of the Navier-Stokes equations or by considering rarefied gas dynamics.

A fully distributed unstructured Navier-Stokes solver for large-scale aeroelasticity computations

We present the development and validation of a parallel unsteady flow and aeroelasticity code for large-scale numerical models used in turbomachinery applications. The work is based on an existing unstructured Navier-Stokes solver developed over the past ten years by the Aeroelasticity Research Group at Imperial College Vibration University Technology Centre. The single-process multiple-data paradigm was adopted for the parallelisation of the solver and several validation cases were considered. The computational mesh was divided into several sub-sections using a domain decomposition technique. The performance and numerical accuracy of the parallel solver was validated across several computer platforms for various problem sizes. In cases where the solution could be obtained on a single CPU, the serial and parallel versions of the code were found to produce identical results. Studies on up to 32 CPUs showed varying levels of parallelisation efficiency, an almost linear speed-up being obtained in some cases. Finally, an industrial configuration, a 17 blade row turbine with a 47 million point mesh, was discussed to illustrate the potential of the proposed large-scale modelling methodology.

Whole-Assembly Flutter Analysis of a Low Pressure Turbine Blade

This paper reports the findings of a flutter investigation on a low pressure turbine blade using a 3D non-linear integrated aeroelasticity method. The approach has two important features: (i) the calculations are performed in a time-accurate and integrated fashion whereby the structural and fluid domains are linked via an exchange of boundary conditions at each time step, and (ii) the analysis is performed on the entire bladed-disk assembly, thus removing the need to assume a critical interblade phase angle. Although such calculations are both CPU and in-core memory intensive, they do not require pre-knowledge of the flutter mode and hence they allow a better understanding of the aeroelasticity phenomena involved.

A non-linear integrated aeroelasticity method for the prediction of turbine forced response with friction dampers

An integrated aeroelasticity model was described for turbine blade forced response predictions. An iterative procedure was developed to determine the resonance shift under the effects of both unste ...

An Integrated Time-Domain Aeroelasticity Model for the Prediction of Fan Forced Response Due to Inlet Distortion

The forced response of a low aspect-ratio transonic fan due to different inlet distortions was predicted using an integrated time-domain aeroelasticity model. A time-accurate, non-linear viscous, unsteady flow representation was coupled to a linear modal model obtained from a standard finite element formulation. The predictions were checked against the results obtained from a previous experimental programme known as “Augmented Damping of Low-aspect-ratio Fans” (ADLARF). Unsteady blade surface pressures, due to inlet distortions created by screens mounted in the intake inlet duct, were measured along a streamline at 85% blade span. Three resonant conditions, namely 1F/3EO, 1T&2F /8EO and 2S/8EO, were considered. Both the amplitude and the phase of the unsteady pressure fluctuations were predicted with and without the blade flexibility. The actual blade displacements and the amount of aerodynamic damping were also computed for the former case. A whole-assembly mesh with about 2,000,000 points was used in some of the computations. Although there were some uncertainties about the aerodynamic boundary conditions, the overall agreement between the experimental and predicted results was found to be reasonably good. The inclusion of the blade motion was shown to have an effect on the unsteady pressure distribution, especially for the 2F/1T case. It was concluded that a full representation of the blade forced response phenomenon should include this feature.Copyright

Aeroelasticity analysis of a bird-damaged fan assembly using a large numerical model

Bird strike is a major consideration when designing fans for civil aero engines. Current methods rely on impact tests and structural optimisation but it is highly desirable to have predictive numerical models to assess the aerodynamic and aeroelastic stability of bird-damaged fan assemblies. A detailed feasibility study towards such a prediction capability, consisting of a CFD solver coupled to a finite element representation of the structure, is reported in this paper. A whole-assembly model was used for both the fluid and the structural domains, an approach necessitated by the loss of cyclic symmetry due to one or more blades undergoing plastic deformation under the effect of the bird impact. It was assumed that two consecutive blades had suffered unequal amount of bird damage, the so-called heavy-and medium-damaged blades. A viscous steady-state solution of the bird-damaged assembly was computed first. The results indicated the formation of a strong wake from the heavy-damaged blade onto the downstream medium-damaged blade. It was also found that the mass flow had reduced considerably due to the blockage effect of the damaged region. A structural analysis of the fan assembly showed that the vibration modes were significantly different from those of the corresponding undamaged assembly. Viscous unsteady flow calculations with blade motion were performed for the whole assembly and the results indicated vibration instability in a torsional mode and the possibility of rotating stall, both being due to flow separation behind the heavy-damaged blade. The aeroelasticity computations required about 1Gb of memory and took about 50 days on a fast single-CPU workstation. The predictions were found to be in good agreement with available experimental data.

A Resonance Tracking Algorithm for the Prediction of Turbine Forced Response With Friction Dampers

This paper presents an iterative method for determining the resonant speed shift when non-linear friction dampers are included in turbine blade roots. Such a need arises when conducting response calculations for turbine blades where the unsteady aerodynamic excitation must be computed at the exact resonant speed of interest. The inclusion of friction dampers is known to raise the resonant frequencies by up to 20% from the standard assembly frequencies. The iterative procedure uses a viscous, time-accurate flow representation for determining the aerodynamic forcing, a look-up table for evaluating the aerodynamic boundary conditions at any speed, and a time-domain friction damping module for resonance tracking. The methodology was applied to an HP turbine rotor test case where the resonances of interest were due to the 1T and 2F blade modes under 40 engine-order excitation. The forced response computations were conducted using a multi-stage approach in order to avoid errors associated with “linking” single stage computations since the spacing between the two bladerows was relatively small. Three friction damper elements were used for each rotor blade. To improve the computational efficiency, the number of rotor blades was decreased by 2 to 90 in order to obtain a stator/rotor blade ratio of 4/9. However, the blade geometry was skewed in order to match the capacity (mass flow rate) of the components and the condition being analysed. Frequency shifts of 3.2% and 20.0% were predicted for the 1T/40EO and 2F/40EO resonances in about 3 iterations. The predicted frequency shifts and the dynamic behaviour of the friction dampers were found to be within the expected range. Furthermore, the measured and predicted blade vibration amplitudes showed a good agreement, indicating that the methodology can be applied to industrial problems.© 2000 ASME

A Numerical Investigation of Aeroacoustic Fan Blade Flutter

This paper reports the results of an ongoing research effort to explain the underlying mechanisms for aeroacoustic fan blade flutter. Using a 3D integrated aeroelasticity method and a single passage blade model that included a representation of the intake duct, the pressure rise vs. mass flow characteristic of a fan assembly was obtained for the 60%–80% speed range. A novel feature was the use of a downstream variable-area nozzle, an approach that allowed the determination of the stall boundary with good accuracy. The flutter stability was predicted for the 2 nodal diameter assembly mode arising from the first blade flap mode. The flutter margin at 64% speed was predicted to drop sharply and the instability was found to be independent of stall effects. On the other hand, the flutter instability at 74% speed was found to be driven by flow separation. Further post-processing of the results at 64% speed indicated significant unsteady pressure amplitude build-up inside the intake at the flutter condition, thus highlighting the link between the acoustic properties of the intake duct and fan blade flutter.Copyright