GUIde 4: 2009-2013

GUIde 4 Consortium:  Project Descriptions and Deliverables

Seven research projects were conducted under the GUIde 4 Consortium between 2009 and 2013.  Below are the technical summaries of each project.


Entire Bladed Disk Damping through Dampers on only a few Blades - Fail-Safe Optimization

Marc Mignolet, PhD
Arizona State University

This project focuses on the prediction of vibratory response for an entire bladed disk (with intentional and random mistuning effects) when damping treatments are applied to only a small fraction of the individual blades.  This study is motivated by the recognition that one of the primary factors limiting the implementation of damping treatments is their significant additional cost.  It is proposed to investigate the feasibility of minimizing this cost impact by applying the damping treatment on only a limited number of blades.  The research should provide insight into the conditions when this type of approach to controlling vibration levels will and will not be successful.  Efficient techniques will be developed to optimize intentional mistuning patterns and to select the damped blades.  Both linear and nonlinear damping treatments will be considered.  The optimization will also be investigated when the damping treatment on some of the damped blades may fail due to excessive wear or foreign object damage Another potential application of this technology is to model the effects of damping variation, especially with wear.  As such, the procedures could potentially be used to investigate this aspect of the robustness of damping applications.

Deliverables:

1)    Detailed reports including technical approach, algorithms, and results
2)    Damper optimization computer codes

Aeromechanics Response in a High Performance Centrifugal Compressor Stage
Choon Tan, PhD
MIT

The main objective is to develop understanding of excitation mechanisms in highly loaded, transonic impellers associated with unsteady impeller-diffuser interactions.   The approach is to make extensive use of CFD analyses to analyze, two configurations of industry selected impellers.  Emphasis is placed on the impact of impeller aerodynamic loading due to unsteady interactions between the impeller blades and the diffuser vanes.  The key tasks are to: (1) develop understanding of flow physics of the unsteady aerodynamics,  (2) determine key design parameters that guide excitation strength, (3) determine importance of aerodynamic coupling between impeller blades and diffuser vanes, and (4) evaluate parameters for two configurations which have varied diffuser designs.  The benefits to the Air Force and industry is to lower risk of impeller and diffuser HCF, thus reducing the potential for unacceptable reliability and reducing cost
Deliverables:
1)      Delineation of parameters that set impeller blade aerodynamic and    structural response
2)      Identification of conditions under which one would expect an upstream manifestation of aeromechanics difficulties instead of a downstream manifestation due to a downstream stimulus associated with impeller-diffuser interactions
3)      A criteria defining the upstream (impeller leading edge) and downstream (impeller blade trailing edge) manifestations of unsteadiness with a downstream stimulus associated with impeller-diffuser interactions
4)      Definition of what constitutes an adequate characterization of impeller blade system response so that it can be used to develop guidelines to avoid aeromechanic difficulties in centrifugal compressor stages

Turbomachinery Aeroelastic Analysis Using a Continuation/Proper Orthogonal Decomposition Method 
Paul Cizmas, PhD
Texas A&M University

The objective of this project is the development of a Reduced Order Model (ROM) for use in turbomachinery aeroelastic analysis. The ROM will be based on the combination of two separate modeling efforts. The first is the application of Proper Orthogonal Decomposition (POD) techniques to the 3D Navier-Stokes equations. POD extracts the principal fluid eigenmodes from a set of solutions for the full Navier-Stokes equations. The full solution set is developed to span the key design parameters that need to be considered in the aeroelastic analysis. Once the fluid eigenmodes have been extracted they are used as the basis functions for constructing full solutions of the Navier-Stokes equations at selected design parameters. The POD approach will be combined with the continuation method to produce an efficient method for searching the design parameter space for flutter boundary identification. The main benefit to industry is to reduce the amount of time needed to perform the large number of unsteady CFD simulations needed to perform forced response and flutter boundary predictions.  The reduced time will be brought about without simplifying the underlying unsteady aerodynamics model.
Deliverables:
1)      Computer code that implements the POD reduced-order model of the 2D and 3D Navier-Stokes equations
2)      User manual and input/output cases for code verification and validation
3)      Assessment of the accuracy of the reduced-order model as a function of the number of POD modes
4)      assessment of the sensitivity of the solution to the number of POD modes for each variable
5)      Evaluation of two indicators of reduced-order model accuracy: the energy spectrum and Hausdorff distance
6)      Computer code that integrates the open-source continuation code AUTO2000 and the POD code

High Fidelity Detached Eddy Simulation of Turbulence to Study Unsteady Fan / Compressor Non-Synchronous Vibration
GeCheng Zha, PhD
University of Miami

This project proposes to investigate non-synchronous vibrations (NSV) --a subject of significant interest among industry members--with a CFD code featuring the next leap forward in turbulence modeling:  Detached Eddy Simulation (DES).  Nearly all CFD analyses of rotating machinery solve the Reynolds Averaged Navier-Stokes (RANS) equations; turbulence is typically modeled using the Boussinesq eddy viscosity assumption, which assumes isotropic turbulence.  The flow features thought to contribute to NSV (shedding and/or interaction between tip clearance flow and the mean flow) usually include significant large-scale anisotropic turbulence.  As a result, RANS solvers often struggle to accurately calculate the aerodynamics behind NSV.  DES solvers, on the other hand, are specifically designed to capture large-scale turbulence.  In this project, a DES CFD code will be validated and then applied to rotors known to exhibit NSV.  This work is expected to provide insight into the fundamental physics behind NSV.  The code, which also includes fully coupled fluid structure interaction capability, will be provided to the industry members.
Deliverables:
1) multi-stage steady state code based on RANS turbulence model
2) unsteady full annulus or a sector multi-stage rotor-stator interaction code with rigid blades using URANS or DES
3) unsteady full annulus or a sector multi-stage rotor-stator interaction code with blade vibration using URANS or DES
Objectives:
1) Study the fundamental mechanism of NSV: rotating instability or acoustic resonance
2) Develop and deliver high fidelity DES CFD code for turbomachinery aeroelasticity


Flutter and Forced Response of a Bladed Rotor with Geometric Mistuning

Alok Sinha, PhD
The Pennsylvania State University

The main objective of this proposal is to develop software for flutter and forced response prediction of a bladed rotor with geometric mistuning. The unique feature of this software will be that it will be based on a highly accurate structural model using Modified Modal Domain Analysis (MMDA). In fact, the first step will be to show beyond any doubt that MMDA indeed is a general and accurate method for the development of reduced-order model. It is proposed to compare MMDA results with bench tests of IBRs. It will be shown that MMDA can accurately predict both frequencies and mode shapes of not only an isolated family of modes, but also of high-order modes. In addition, it will also be demonstrated that MMDA is valid in the presence of a large amount of geometric mistuning; for example, blending of airfoils. It is generally agreed that modeling aerodynamics is a more difficult task than modeling structures. But, the fact is that there is no results showing a match between theoretical and experimental mistuned mode shapes in the presence of many closely spaced modes and high degree of modal interactions, which are found in advanced jet engines. Therefore, before considering aerodynamic models, we must select a reduced-order model, which captures all the important physics of mistuned structures. The author of this proposal strongly believes that the MMDA is the correct approach for mistuning analysis, and has the ability to accurately predict time-varying stresses in blades, and hence the fatigue-life of blades.

Deliverable:
A design code will be developed to obtain optimal mistuning pattern and predict the acceptability of a bladed rotor from the CMM data directly.


Unsteady Aerodynamics and Aeromechanics Methods

Robert Kielb, PhD and Kenneth Hall, PhD
Duke University

The Duke efforts are focused on developing methods with strong engineering relevance to current high-cycle fatigue (HCF) design issues and failures. Successful conclusion of our proposed projects will help to conduct rapid screening and design evaluation methods and reduce the number of CFD runs required to perform aeromechanics evaluations. These methods would allow advanced aeromechanics calculations to be performed early in the design phase, so that potential problems can be addressed when they are most easily fixed.
Duke proposed a collaborative interaction with GUIde members in which the specific tasks are jointly selected, and adjusted as needed. The Duke proposal includes several possible topics related to unsteady aerodynamics of turbomachinery which are aligned with the needs described by industry:  developing adjoint version of the MUSTANG code, designing for Aerodynamic Instabilities, and designing for Aerodynamic Asymmetries.
Deliverables:
1)    Develop two versions of MUSTANG II: forward harmonic balance (HB) and adjoint/optimization.
2)    Use a model problem base on a Van der Pol oscillator to understand the physics of NSV with emphasis on the lock-on effect. This knowledge will then be used to develop the capability to conduct NSV design within MUSTANG 2.
3)    The 2D configuration D case will be used to study the effect of blends on forced response. The knowledge from this and other studies will be used to install the aerodynamic asymmetry capability into MUSTANG 2.

Forced Response in a Multistage Axial Compressor
Nicole Key, PhD
Purdue University

The objective of this proposed research is to provide detailed forcing function and response data on the embedded stage of a multistage compressor at/near a rotor resonance.  The data will be made available to the GUIde IV Consortium to provide an opportunity to calibrate existing CFD tools for rotor forced response predictions.  The objective will be met by a thorough experimental campaign measuring the unsteady aerodynamic forcing function using detailed flow field traverses and measuring the rotor response using an NSMS (Non-Intrusive Stress Measurement System) tip timing system.  This will represent one of the few databases available at resonance on multi-stage compressors in the open literature.  This proposal addresses the specific RFP request for model validation capability for Campbell diagram crossings from airfoil responses due to rotor-stator interactions.

There is a dearth of data available on multistage compressors, especially at/near rotor resonant conditions.  Rotor resonant response predictions are important to both the aircraft engine and heavy-duty industrial gas turbine applications.  However, data are required to gain a better understanding of the flow physics and are necessary for the calibration of newly developed models. 
   
The research vehicle, the Purdue Three-Stage Research Compressor, is a unique research facility in that it models the rear stages of a modern high-speed compressor, matching Mach number and Reynolds number to actual aircraft engine operating conditions.  The facility is conducive to detailed flow measurements in the pitchwise direction because each vane row can be individually indexed past stationary probes.  Most of the data available in the open literature are acquired from vane leading edge instrumentation.  However, traversing the vanes allows for blade row interaction effects, such as potential fields, to be captured.  There is a significant and measurable total pressure rise per stage, and blade heights are 2 inches, allowing sufficient space for detailed flow measurements without probe blockage issues.  This facility represents a technology readiness level (TRL) of 4, and thus, a dataset from this facility will be very relevant to the rear stages of modern core compressors.