2006ISMApast.Isma-isaac.be/downloads/isma2006/ISMA2006_abstract_book.pdf10:05 A Future for...

Programme and

Book of abstracts

September 18-20, 2006Katholieke Universiteit Leuven, Belgium

ISMAInternational Conference on

Noise and Vibration Engineering

2006

Monday September 18, 2006

PAPER SUMMARIESOpening Session

9:15 WelcomePaul Sas, Conference Chairman, Katholieke Universiteit Leuven, Belgium

9:30 WelcomeLouis Tobback, Mayor of Leuven

9:50 OpeningLudo Froyen, Dean of the Faculty of Engineering, Katholieke Universiteit Leuven, Belgium

10:05 A Future for Experimental Structural Dynamics (ID 676)David J Ewins, Imperial College London, United Kingdom

Structural dynamics remains a high priority feature in the design of almost every structure and machine,essentially as a result of the adverse consequences on acceptability, reliability and integrity of a failure toget the dynamics ‘right’. Commercial competition and growing demands for safety, reliability and minimalenvironmental impact all combine to impose more demanding expectations on the structural dynamicist. Inresponse to these demands, major strides have been made in structural dynamics technology over the past 2– 3 decades. In particular, increasing demands for more and more predictions (and less and less developmenttesting) have led to dramatic improvements in computational techniques so that quantitative predictions ofthe dynamic properties of the most complex of structures are now not only possible, but even routine. How-ever, in spite of these developments, the need for experimental confirmation that the predictions are correctremains a crucial issue although one which is increasingly challenged because of its relatively high cost.

It is also clear that experimental technology has not kept pace with numerical methods. While there havebeen developments in transducers, and in the electronics that are used to condition and to process the signalsthey generate, the overall time that is required to prepare and carry out a standard test has not shortenedcommensurately with the time taken to carry out corresponding predictions. Hence, tests tend to be relativelyslower, and thus more expensive, than their computational counterparts - and are under growing pressure tobe eliminated.

In this paper, the argument for maintaining a state of the art experimental capability is made and is basedon the thesis that analysis (or theory, or simulation) should be integrated with test (or measurement, orexperiment). At the same time, it is accepted that significant changes may be required in the specific testmethodologies in order to restore an effective balance of time and cost between the two approaches. A caseis made for the continued reliance on testing as part of the overall process of designing and maintainingcomplex and critical structures which are prone to the dynamics-related effects of vibration, shock, noiseand instability. In the course of this debate, the paper travels from an initial question of “Is there a future forexperimental structural dynamics?” through an assertion to the effect that “There is a future for experimentalstructural dynamics!” to arrive finally at the proposal that “This is a future for experimental structuraldynamics.”

10:50 Auralization of structure-borne sound (ID 675)Michael Vorlander, RWTH Aachen University, Germany

Structure-borne sound, its modeling of generation, transmission, radiation and effect is the daily task ofacoustic engineers. When prediction models are extended towards an approach of signal theory, the problemcan be described in terms of signals, filters and signal processing. This way, the technique of auralization canproduce audible results which can be used for demonstration of effects and for convincing clients or localauthorities to invest in noise and vibration control.

1


Auralization in case of airborne sound produced by of vibration sources is interesting for

• building acoustics,

• car industry,

• machine diagnosis,

just to name three examples. In all three examples, audible sounds help to interpret the problem and to findgood solutions for better performance of complex systems related so sound and vibration.

2


FRIENDCOPTER Rotorcraft noise and vibration – FCOP1Room 1 – Chairman: A. Vecchio

13:00 Energy ratio approach for SEA model path identification (ID 130)Gianluca Diodati, Pasquale Vitiello, Italian Aerospace Research Center, Italy

Statistical Energy Analysis (SEA) is a methodology for solving vibro-acoustic problems in the mediumhighfrequency range to reduce at minimum the computational effort required to analyse any complex system.A typical SEA system is described by using an energy balance equations set, which can be inverted to de-termine the energy associated to the subsystems modelling the structure, but no indication can be obtainedconcerning the preferential paths along which energy is transferred. This information would be extremelyuseful to identify and design the most efficient control techniques able to reduce interior noise in a closedenvironment.

Contrary to existing techniques, the approach followed herein is based on the detection of the main energytransmission paths by taking into account the actual direction of Energy Flow (power flows from subsystemwith higher modal energy density to subsystem with lower modal energy density). To this purpose, theoriginal SEA system is transformed into an equivalent suitable digraph with a number of nodes equal to theSEA subsystems and connections with weight factors reflecting the actual energy flow ratio coming out fromeach node.

A path identification algorithm based on Eppstein shortest path detection method has been developed inMatlab environment to extrapolate all possible paths in the global system, and to classify the main pathsaccording to the total energy transferred to a referenced subsystem along the path. The technique has beenapplied within Friendcopter European Project to the SEA model of Agusta A109 helicopter mock-up.

13:25 Experimental Noise Transfer Path Analysis on Helicopters (ID 572)Antonio Vecchio, LMS International, BelgiumCarlo Urbanet, University of Rome “La Sapienza”, ItalyFausto Cenedese, AgustaWestland, Italy

An experimental Transfer Path Analyse is for the first time applied to the helicopter Agusta A109. Theobjective is to assess the noise contribution of the helicopter gearbox to the overall cabin noise level and tosimulate a number of realistic noise reduction scenarios, including active noise control, in view of achievinga sensible cabin noise reduction. Issues related to the application of a standard automotive technique to amore complex system such as a helicopter gearbox are discussed with special focus on test data completenessand model validation. The analysis focuses on the connecting points between the gearbox and the helicoptersframe and shows that noise comfort improvement can be achieved with active control systems acting on theanti-torque plate hosting the gearbox.

13:50 Active solutions for reduction of the transmission of structure - borne noise to thecabin structure on a A109 mockup (ID 601)L. Dozio, W. Corbetta, E. Vigoni, A. Forghieri, G.L. Ghiringhelli, Politecnico di Milano, ItalyFausto Cenedese, AgustaWestland, Italy

The paper presents some preliminary results of active control strategies aimed at reducing the transmission ofstructure-borne noise to the cabin structure on a AgustaWestland A109 helicopter mockup. The experimentswere performed on a full-scale mockup consisting of a gearbox housing connected to the fuselage via twofront and two rear struts and an anti-torque plate. The mockup is provided with two electric motors thatappropriately drive the gearbox at the nominal rotating speed. A suitable aerodynamic brake is attached tothe rotor must in order to reproduce the load effect of the rotor blades. Peaks at the gear meshing frequenciesreveal amplitudes very close to those acquired during in-flight measurements. Rear gearbox struts and theanti-torque plate were equipped with piezoelectric actuators and some control algorithms were designed and

3


implemented to change the transfer function at the connection between the gearbox and the cabin. Vibrationand acoustic results inside the cabin are presented and critical issues of the adopted solutions are discussed.

14:15 Simulation and Evaluation of Noise and Vibration Reduction Techniques in a Heli-copter Cabin Using SEA (ID 605)A. Perazzolo, Fausto Cenedese, AgustaWestland, Italy

Increasing acoustic comfort inside the helicopter is demanded especially by VIP customer. Noise reductionis a big challenge, due to the high acoustic levels generated by the main sources (gearbox, engines, rotors,hydraulic and oil pumps) and requires the synergy of all available technologies.

A mock-up built using a real helicopter is used like test-bench in order to apply and to validate noise andvibration reduction systems. In parallel, a SEA analytical model has been developed to reproduce the testresults both in term of noise transfer path and of acoustic levels in running condition. The validated digitalmock-up in “bare” configuration will be then used to compare several trimmed configurations to select onlythe most promising to be tested, saving time and costs required by experimental tests.

14:40 Hybrid Modelling of a helicopter gearbox using Inverse Boundary Elements Method(ID 608)Antonio Vecchio, LMS International, BelgiumFausto Cenedese, AgustaWestland, ItalyCarlo Urbanet, University of Rome “La Sapienza”, Italy

A hybrid modelling technique is presented that uses test data measured on an industrial test-bench to charac-terize the acoustic behaviour of a helicopter’s gearbox in operating conditions. Experimental data, includingboth acoustic and vibrations, are acquired on a dedicated gearbox test-bench where the boundary conditionscan be controlled on a very limited extent. The proposed approach uses inverse technique to compute thegearbox surface velocities leading to the identification of hot spots on the gearbox surface and paving the wayto noise source localization. Direct boundary element technique is then used to propagate back the pressuresand simulate the gearbox acoustic performance in different operating conditions.

15:05 Modeling and Characterisation of the Vibration Dynamics Modeling of AnisotropicPorous Foam Materials (ID 632)Remi Guastavino, Peter Goransson, KTH Kungliga Tekniska Hogskolan, Sweden

Accurate prediction of low to medium frequency surface vibration and sound radiation behavior of foammaterials, requires improved means of estimating the dynamic elastic and damping properties of the foam.This need arises due to the geometric anisotropy in the foam cell microstructure, where the foam cellsand struts are elongated in the rise and injection flow directions of the manufacturing process. To reachacceptable levels of accuracy, this inherent anisotropy needs to be correctly represented in the acousticalnumerical simulation methodology. The present paper presents a methodology built on a hybrid combinationof experimental deformation and strain field mapping, and physically based porous material acoustic FiniteElement (FE) simulation modeling, allowing for the anisotropic dynamic elastic coefficients and dampingproperties of the foam to be correctly estimated. This new methodology of model-based porous materialcharacterization is demonstrated here for a simplified seismic mass configuration.

4


Non-linearities: identification and modelling – NL1Room 1 – Chairman: J. Schoukens

15:55 Experiments, measurement procedures and curve-fitting of a post-buckled vibratingbeam (ID 65)Or Yogev, Izhak Bucher, Miles Rubin, Technion - Israel Institute of Technology, Israel

This paper discusses an experimental investigation of a dynamically buckled structure. In the experiment, theend of a beam with a narrow rectangular cross-section is oscillated by a motor that creates variable amplitudeand acceleration patterns. The dynamic shear force causes out-of-plane buckling above a critical value ofload. Continuously varying excitation frequencies and time-frequency distribution analysis has revealed thebuckling threshold. A tracking and scanning laser measurement set-up was developed to extract the in-planeand angular deformations. Fast video photography was image processed to show interesting period doublingand nonlinear behaviour. The experimental and identification procedures are described and analysed.

16:20 Experimental identification of non-linear parameters over a scaled multi-storeybuilding (ID 512)Luigi Garibaldi, Ermanno Giorcelli, Stefano Marchesiello, Politecnico di Torino, Italy

In this paper the identification of a laboratory structure is performed by means of well-assessed linear andnon-linear techniques. The structure considered is a multi-storey building subjected to different levels ofexcitation, chosen with particular care so as to put in evidence differences between linear and non-linear be-haviour and to allow a better estimation of the system parameters. A numerical linear model of the structureis performed by using a series of vertical beams also subjected to axial load. The non-linearity of the structureis due to a buffer mounted on a lower deck with a certain clearance, introducing a non-linear asymmetricalbehaviour in case of high excitation levels.

The inertance of the underlying linear system, identified by means on the Nonlinear Identification throughFeedback of the Outputs method (NIFO), is compared with that obtained by applying the Stochastic SubspaceIdentification method (SSI) to the response with low excitation, and some suggestions to practically applyboth methods are given.

The experimental results comparison reported in this paper is highlighting the key advantage of NIFO withrespect to other previously developed techniques: the capability of dealing with multi-degree-offreedomnon-linear systems holding different types of non-linearities.

16:45 Nonlinear Structure Analysis Using the Best Linear Approximation (ID 533)Lieve Lauwers, Johan Schoukens, Rik Pintelon, Martin Enqvist, Vrije Universiteit Brussel, Bel-gium

In this paper, we propose a method to distinguish between some nonlinear system structures using the bestlinear approximation (BLA) of the system in order to select an appropriate model structure. The main ideaof the method is to apply a Gaussian input signal and to vary the root mean square (rms) value and the powerspectrum of this signal. Depending on the resulting changes of the amplitude and phase characteristics ofthe BLA, an appropriate model structure for the Device Under Test can be selected. A theoretical analysis ofthe method is presented for some block-oriented structures.

17:10 Comparison of two different nonlinear state-space identification algorithms (ID 426)Johan Paduart, Johan Schoukens, Vrije Universiteit Brussel, BelgiumKris Smolders, Jan Swevers, Katholieke Universiteit Leuven, Belgium

In this paper, a comparison between two models for nonlinear systems is made. Both models have a statespace nature, but there are some differences in the identification approach and the model structure. The firstmodel that we will discuss is a discrete time model that uses input-output data for the identification. The

5


second model uses explicit measurements of the states of the system and some physical insight to model therelationship between the states. The similarities and differences between the two models are discussed, andtheir performance is compared utilizing data from an experimental setup.

17:35 On Nonlinear Parameter Estimation (ID 505)Martin Magnevall, Andreas Josefsson, Kjell Ahlin, Blekinge Institute of Technology, Sweden

The industrial demand on good dynamical simulation models is increasing. Since most structures showsome form of nonlinear behavior, linear models are not good enough to predict the true dynamical behavior.Therefore nonlinear characterization, localization and parameter estimation becomes important issues whenbuilding simulation models. This paper presents identification techniques for nonlinear systems based onboth random and harmonic excitation signals.

The identification technique based on random excitation builds on the well known reverse-path method de-veloped by Julius S. Bendat. This method treats the nonlinearity as a feedback forcing term acting on anunderlying linear system and the parameter estimation is performed in the frequency domain by using con-ventional MISO/MIMO techniques. Although this method provides a straightforward and systematic wayof handling nonlinearities, it has been somewhat limited in use due to the complexity of creating uncorre-lated inputs to the model. As is shown in this paper, the parameter estimation will not be improved withconditioned inputs and the nonlinear parameters and the underlying linear system can still be estimated withpartially correlated inputs.

This paper will also describe a parameter estimation method to be used with harmonic input signals. Byusing the principle of harmonic balance and multi-harmonic balance it is possible to estimate an analyticalfrequency response function of the studied nonlinear system. This frequency response function can, inconjunction with measured nonlinear transfer functions, be used to estimate the nonlinearity present in thesystem. This method is also applicable on nonlinear systems with memory, e.g. systems with hysteresiseffects.

The above mentioned methods are applied to multi-degree-of-freedom and single-degree-of-freedom systemswith different types of nonlinearities. Also, techniques for locating nonlinearities are discussed.

18:00 Nonlinearity Characterization for Nonlinear Dynamic System Identification Usingan Expert Approach (ID 188)Grigorios Dimitriadis, Gareth Vio, University of Manchester, United Kingdom

The identification of nonlinear dynamic systems can be rendered significantly more parsimonious if thenonlinearity present in the system is known. While there are many successful non-parametric nonlinearsystem identification methods, the resulting models do not describe the nonlinearity in physical terms andare difficult to obtain due to the large number of candidate terms that must be examined. In this paperan expert approach towards the characterization of nonlinearities in a dynamic system is presented. Themethodology is based on simulations of dynamic systems with a variety of commonly occurring nonlinearfunctions. The responses of such systems to various types of excitation are analysed and rules are developedas to what nonlinearity is likely to be present in a system given the dynamic characteristics of measuredresponses.

6


Dynamic modelling – DMOD1Room 2 – Chairman: E. Balmes

13:00 Model validation for heavily damped structures. Application to a windshield joint.(ID 335)Etienne Balmes, Ecole Centrale de Paris / SDTools, FranceMathieu Corus, ONERA / Ecole Centrale Paris, FranceSylvain Germes, PSA Peugeot Citroen, France

The use of damping augmentation concepts is spreading, thus motivating the need for design and validationmethodologies. The frequency and temperature dependence of viscoelastic materials pose the challenge ofcorrelating the response at multiple operating conditions. The fact that each mode is affected differently bythe viscoelastic treatments is another difficulty. The paper first presents test results, at multiple temperatures,on a structure representative of a windshield with a damped joint. Then after a short reminder of numericalmethodologies used to obtain response predictions, issues with the use of pole frequency and damping ratiofor model correlation are illustrated.

13:25 A Linear Programming Approach to the Optimal Design of Spline-Based MotionSystem Inputs (ID 487)Jan De Caigny, Bram Demeulenaere, Jan Swevers, Joris De Schutter, Katholieke UniversiteitLeuven, Belgium

For motion systems such as cam-follower mechanisms and loads driven by servo motors, this paper considersthe design of system inputs that are continuous up to their M-th derivative and minimize some design criterionsubject to user-defined constraints. This problem is tackled by optimizing a piecewise-linear, continuousparametrization (based on a large number of second-order B-splines) of the M-th derivative of the trajectory,which guarantees the continuity up to derivativeM. As an application, the system input of a linear dynamicsystem is optimized to reduce the residual vibrations. Both non-robust and robust solutions of the resultinglinear optimization problem are presented. For high-speed applications a trade-off in the degree of continuityat the boundaries is shown: for robustness against small plant perturbations this degree needs to be low, whilefor robustness against unmodeled higher dynamics this degree needs to be large. The numerical optimizationresults furthermore confirm and outperform earlier results for the considered benchmark.

13:50 Dynamic Behaviour of Elastic Bodies Coupled by Extended Friction Contacts (ID8)Alexander Genzo, Volkswagen AG, GermanyWalter Sextro, Technical University of Graz, AustriaLars Panning, University of Hannover, Germany

In many technical applications elastic bodies are coupled through extended contact interfaces by means ofbolted joints forming an elastic structure. The housing of an internal combustion engine built up of thecrankcase, the oil pan, the cylinder head and various covers can be regarded as a practical example foran elastic mechanical structure. To understand the influence of extended friction contacts on the dynamicbehaviour of elastic structures, the forced vibrations of a test structure consisting of two elastic half tubes jointtogether in two bolted flanges will be analysed. The contact interfaces in the flanges can be considered asdry and extended friction contacts. Since the test structure is loaded with a harmonic external force, relativedisplacements occur in the contact interfaces. Depending on the normal pressure distribution and magnitudeas well as on the contact surface characteristics these relative displacements lead to microslip effects affectingthe dynamic behaviour of the test structure.A calculation method accounting for these effects is presentedand experimentally validated by comparing measured and calculated frequency response functions (FRF) ofthe test structure.

7


14:15 Prediction of the influence of structural modifications including viscoelastic materi-als (ID 336)Benjamin Groult, EDF R&D, France

EDF is in charge of a large industrial fleet of electric power stations. Because of ageing and strong envi-ronmental conditions, unexpected vibration problems can appear on machines like pumps or electric motors.These problems have to be fixed rapidly (by structural modifications) in order to avoid electricity shortages.Stiffness or mass additions are often used to solve the problem when only a few modes of the defective struc-ture are involved in the faulty behaviour. However, the use of such modifications is sometimes irrelevant. Insuch cases, damping modifications could be an efficient alternative to reduce the amplitude of in operationvibrations. This work deals with the estimation of the influence of modifications including viscoelastic mate-rials on the dynamic behaviour of a structure. Current structural modification methods basics and reductionmethods of complex problems are first presented. Then some numerical results of prediction of the effect ofdamping modification are shown through a mock up example.

14:40 Modelling the Control Strategies for Riding a Motorcycle (ID 469)Stuart Rowell, Atanas Popov, Jacob Meijaard, University of Nottingham, United Kingdom

Motorcycle rider-control to accomplish a simple lane-change manoeuvre is modelled, combining symbolicrepresentation of vehicle equations of motion with optimal control representing the human rider. A sim-plified motorcycle is considered for which the non-linear equations of motions are derived, leading to adiscrete-time simulation of the motorcycle’s motion. Linear approximations to the non-linear equations ofmotions are calculated for each iteration step, allowing significant angles of roll in particular to be consid-ered. Furthermore, a detailed tyre model considering a toriodal-profile tyre with appropriate force-generatingcharacteristics is applied, completing the extension of previous models to allow for non-small angles for thedegrees of freedom. The motorcycle dynamics are coupled with an optimal controller representative of ahuman rider, thus allowing path following to be modelled, based on previously published works.

15:05 Determining the dynamic properties of joints in piping systems (ID 111)Jen Muggleton, Tim Waters, Brian R. Mace, University of Southampton, United Kingdom

The vibration modelling of piping systems is relevant to many engineering applications. Although there aremany prediction techniques available, at present a fundamental weakness can undermine all of them; thisweakness arises from inadequate modelling of joints due to uncertainty in joint properties. An approachto joint parameter estimation from vibration measurements, which makes use of wave models of pipingsystems, has been proposed previously. In order for this approach to be successful, measurements of thereflection and transmission properties of the joint in question are required. In a previous paper, an approachto determining the reflection and transmission coefficients of a joint in a finite piping system was described.Here, measurements are made on a right-angled bend in a finite pipe undergoing axial excitation, in orderto demonstrate the procedure, highlight when it can and cannot be applied, and bring to light some of thedifficulties inherent in making the measurements.

OPTIMESS Optical measurement techniques – OPT1Room 2 – Chairman: S. Vanlanduit

15:55 Refracto-Vibrometry for Visualizing Sound Radiation from Vibrating Objects (ID544)L. Zipser, F. Wachter, H. Franke, HTW University of Applied Sciences Dresden, Germany

A novel method for non-interacting measurement and visualization of air borne sound excited by vibratingobjects is presented. The measuring effect is based on the phase difference between two interfering laser

8


beams. One of them passes through a transparent measuring medium whose spatially distributed opticalrefractive index n = n(t, x, y, z) is changed by an invisible mechanical phenomenon, e.g. by sound waves p= p(t, x, y, z) radiated from vibrating objects. The other beam serves as reference beam. The method can beconveniently implemented with an usual laser-Doppler scanning vibrometer. Such a high-sensitivity devicepermits multipoint-sampling of the small vibrations of the optical refractive index n = n(t, x, y, z) in themeasuring medium, which are stimulated by an invisible sound field p = p(t, x, y, z). These index vibrationsare visualized by instructive two-dimensional animations which represent a realistic projection of the soundfield itself.

16:20 Study of surface acoustic and guided acoustic wave interaction with defects by useof scanning and full field stroboscopic interferometry (ID 545)Jozefien Goossens, Xiaodong Xu, Osamu Matsuda, Philippe Leclaire, Katholieke UniversiteitLeuven, BelgiumLoıc Martinez, IUP GE, Universite de Cergy, FranceChrist Glorieux, Katholieke Universiteit Leuven, Belgium

Laser ultrasonic detection is a versatile and highly sensitive tool for the observation of surface and guidedwaves and their interaction with defects. We present results for the dynamic spatio-temporal motion patternof different systems using laser excitation in combination with stroboscopic full field detection and scanninginterferometric detection in the kHz-MHz range. In the presence of defects, depending on their wavelength,waves are diffracted. The experimentally imaged wave propagation field can be temporally and/or spatiallyFourier transformed to determine the spatial and wavevector pattern and thus to analyse sub-surface defectproperties. The results show that dynamic stroboscopic interferometry and scanning laser vibrometry arepowerful techniques to analyse wave propagation as such and to determine physical properties of materialsand structures.

16:45 MIMO testing on a car body using a three-dimensional scanning vibrometer (ID546)Francois Bouteille, Matthias Schussler, Polytec GmbH, Germany

Experimental modal analysis on a car body requires multiple excitations and several hundred response de-grees of freedom (DOFs). The applicability of a three-dimensional scanning vibrometer to the measurementtask has been proven in this work. A three-dimensional scanning vibrometer measures the geometry andvibration data at all optically accessible points. The optical heads are moved to several positions around thecar body, from each position the three-dimensional vibration data is acquired at some hundred points. Forthe described test, three shakers have been used simultaneously. Each shaker was driven with a non-coherentrandom signal. The transfer-functions from each shaker to each response point were calculated by principalcomponent analysis. After the test, the results from the different laser positions were stitched, in order todisplay the operational deflection shapes of the entire car body. Then the data was exported via universal filefor the purpose of performing a modal analysis.

17:10 A review of 10 years of continuous-scan LDV developments (ID 550)Anthony Stanbridge, David J. Ewins, Imperial College London, United Kingdom

The Vibration Group at Imperial College has been exploring applications of Laser Doppler vibrometers, usedin a continuous scanning mode, over a period of more than ten years. This review summarises techniquesfor straight-line, circular and circular-arc scans, and for rectangular and spiral area scans, and presents theprocessing algorithms in a unified style. Vibration responses can be represented directly by Fourier or poly-nomial series coefficients. Measurement techniques, of more limited application, are described for measuringthese coefficients as frequency responses, so that natural mode shapes can be extracted. Other applicationsinclude measurements on rotating discs, on cylindrical surfaces, 3D point vibration measurements and de-tection of structural defects.

9


17:35 Four channel high resolution heterodyne interferometer based on birefringent op-tics (ID 551)Henk van Elburg , Joris Dirckx, W.F. Decraemer, Universiteit Antwerpen, Belgium

The Mach-Zender design for a laser velocimeter was adapted for making simultaneous measurements atseveral, arbitrarily chosen points. The optical design is based on a single, 7 mW He-Ne laser source andtwo acousto-optic modulators, which generate a common heterodyne carrier frequency in all measurementchannels. By using two modulators, one to increase the frequency of the reference beam with 80+fb MHz,and a second one to decrease the frequency with 80 MHz, the heterodyne beat signal can be adjusted to anyfrequency fb of choice, thus allowing adaptation to the input of commercially available FM demodulationchips or avoiding noisy parts of the laser noise spectrum. The apparatus is intended for simultaneous vi-bration measurements at several points on instable objects, such as biological specimens or MEMS, so thatinstantaneous phase relationships and amplitude ratios (transfer function) can be determined. So far, instru-ment performance has been tested with measurements on two and four points of interest. The dual beamversion was realized using beamsplitter-mirror combinations to double the reference and object beam. Thequadruple beam version was realized using two birefringent beam displacers in tandem within the referenceand object beams. In a multipoint system channel cross talk is a major point of concern. Our optical designensures very good channel separation: even with object points less than 0.4 mm apart, channel separationwas determined to be better than 78 dB at all frequencies in a dual beam system and better than 60 dB in aquadruple beam system. The velocity calibration of the system is determined by the laser wavelength and thesensitivity of the FM detection circuitry only. With a maximal velocity amplitude of 50 mm/s, the velocityresolution and the detection limit of our system were determined to be (2.8±0.2) micro m·s(-1)Hz(-1/2) inall channels.

18:00 Acoustic source identification using a continuous scanning LDV (ID 552)Joris Vanherzeele, Steve Vanlanduit, Patrick Guillaume, Vrije Universiteit Brussel, Belgium

This paper shows how a scanning Laser Doppler Vibrometer (LDV), an instrument designed to measurevibrations of structures or objects, can be used in a non-traditional fashion to identify acoustical sources. Thisis achieved by measuring the changes in the optical path induced by local fluctuation of the air refractionindex to which the LDV is sensitive. The acoustical signal used is sinusoidal and may be recovered byscanning at a uniform rate over a subject area (continuous scan) parallel to the source axis and demodulatingthis signal. Due to the fact that the measured scan area is in fact a line integral over a measurement volumebetween the laser head and a rigid object needed to reflect the laser beam, multiple view planes around theaxis of the acoustic source are usually measured. These are then passed through a tomographic algorithm,thereby reconstructing the full sound field. In this article however, only one view plane is measured, but theacoustic source is placed on a rotating surface with fixed rotational frequency, thereby imposing a modulationon the measured spectrum. Demodulation will allow reconstruction of the three dimensional sound field.

Active noise control – ANC1Room 3 – Chairman: R. Boonen

13:00 Comparing a Filtered-X LMS and an H2 Controller for the Attenuation of theSound Radiated by a Panel (ID 320)Lazaro Donadon, D.A. Siviero, Juan Camino, Jose Roberto Arruda, Universidade Estadual deCampinas, Brazil

This work investigates the performance of two control strategies used for structural acoustic isolation, wherethe emphasis is on controlling structural vibration that is responsible for sound radiation. The controllersare required to attenuate the sound pressure that is transmitted trough a plate under structural vibration. Theproposed setup consists of a lexan plate clamped on a rigid baffle. The exogenous disturbance that causes

10


the vibration of the plate is provided by a point force driven by a shaker. The control input used to attenuatethe sound pressure inside a semi-anechoic room is provided by a flexural moment driven by a piezoelectricpatch attached to the plate. The first control strategy is a normalized Filtered-X LMS algorithm that assumesfull information of the exogenous disturbance as reference. For the second strategy, a standard H2 optimaldiscrete-time feedback controller is designed. As opposed to the feedforward LMS design, the H2 feedbackcontroller posses no information of the disturbance. The sound pressure is measured by two microphones.The first microphone, located near the plate, provides the error sensor for both controllers. The second micro-phone, located far from the plate is used for performance evaluation. Computer simulations are carried outto compare the performance of both designs. A real-time implementation of the H2 controller on a dSPACEboard is developed and experimental results are presented.

13:25 Further modeling and new results of active and passive noise reduction using elasto-poroelastic panels (ID 278)Tomasz Zielinski, Polish Academy of Sciences / Ecole Centrale de Lyon, FranceMarie-Annick Galland, Mohamed Ichchou, Ecole Centrale de Lyon, France

The paper presents further development in modeling of active elasto-poroelastic sandwich panels. In fact, anew design of a demi-sandwich panel is proposed and analysed. A numerical model of panel is implementedin COMSOL Multiphysics environment using the most fundamental but very flexible Weak Form PDE Mode.Various physical problems are modeled using Finite Element Method: the wave propagation in acoustic andporoelastic medium, the vibrations of elastic plate, the piezoelectric behavior of actuator. All these problemsinteract in the examined application of active panel. The presented results of FE analysis and some analyticalsolutions prove the necessity of modeling the panel’s interaction with an acoustic medium. Again, confirmedis the fact that an active control is necessary for lower resonances while for the higher frequencies the passivereduction of vibroacoustic transmission performed by a well-designed poroelastic layer is sufficient.

13:50 Active Structural Acoustic Control of repetitive impact noise (ID 352)Gregory Pinte, Rene Boonen, Wim Desmet, Paul Sas, Katholieke Universiteit Leuven, Belgium

This paper discusses the effectiveness of an Active Structural Acoustic Control (ASAC) system for the re-duction of impact noise, radiated by structures with a high modal density in the controlled frequency range.Although there is a significant difference in nature between periodic and transient noise, up till now nospecific research on ASAC of transient noise was reported. This paper focuses on the design of control al-gorithms, adapted to the specific properties of impact noise. Since many industrial impact noise problemsinvolve successive impacts with a repetitive behaviour, control algorithms with a learning behaviour are dis-cussed. The efficiency of these Iterative Learning Control (ILC) algorithms is demonstrated extensively inthis paper. The developed ASAC strategy has been verified on a thick steel plate, which is excited by succes-sive impacts. The obtained results show that ASAC can be a very efficient transient noise control techniquein certain industrial applications (e.g. presses, punching machines,...).

14:15 Force transmission loss as a mechanism for the active control of noise using an activebearing: Preliminary results (ID 1)Simon Hill, Katholieke Universiteit Leuven / Flanders Mechatronics Technology Center, BelgiumBert Stallaert, Katholieke Universiteit Leuven, BelgiumWim Symens, Flanders Mechatronics Technology Center, BelgiumPaul Sas, Katholieke Universiteit Leuven, Belgium

Traditionally, active control of planar structural radiators has approached the problem through integratingsensors and actuators on or into the structure. In such an approach, the goal is to modify the systemsresponse to disturbance, rather than influence the disturbance path to the radiating structure. Many practicalissues exist, which ultimately limit the effectiveness of such an approach. A general problem in industry isnoise radiation from a structure housing a rotating device, where the rotating device creates the disturbance.

11


In this project we have attacked the problem by seeking to reduce the force transmission in the path, whichis the root cause of the sound radiation, rather then the vibration on the radiating surface itself. It is hopedthat higher levels of attenuation may result. Presented here are some of the design requirements to integratean active bearing, possible control approaches and preliminary experimental data.

14:40 Enhancement of a Passive Acoustic Liner by Active Control of Intensity (ID 15)Angelot Minotti, Valerie Pommier-Budinger, Joel Bordeneuve-Guibe, ENSICA-DAS, FranceFrank Simon, ONERA, France

The context of this work is the noise reduction issue in Aeronautics, precisely in-duct propagation with flow(ventilation system...) is of interest. Classically, passive acoustic liners are used: quarterwavelength behavior,perforated sheets backed to honeycombs (SDOF, DDOF)... The main problem is that their absorption islimited in frequency. A possible solution is to carry out a passive / active hybrid liner. The active control aimsto improve the absorption of a liner at frequencies lower than those corresponding to its passive behavior.For the present study, active control is based on the measurement of the instantaneous acoustic intensity (twoerror sensors) downstream from the liner. Two configurations of active control of intensity were carried out,one in the near field of the liner, the other in the far field. Results show the improvement of the absorption insome cases that are detailed.

Medium and high frequency techniques – MHF1Room 3 – Chairman: W. Desmet

15:30 Wave based modelling methods for steady-state interior acoustics: an overview (ID323)Bert Pluymers, Wim Desmet, Dirk Vandepitte, Paul Sas, Katholieke Universiteit Leuven, Belgium

This paper discusses the use of wave based prediction methods for the analysis of steady-state interioracoustic problems. Conventional element based prediction methods, such as the Finite Element Method(FEM), are commonly used, but are restricted to low-frequency applications. TheWave Based Method(WBM) is an alternative deterministic technique which is based on the indirect Trefftz approach. The WBMis computationally very efficient, allowing the analysis of problems at higher frequencies. The efficiency ofthe WBM is most pronounced for problems of moderate geometrical complexity. A hybrid finite element-wave based method combines the strengths of the two methods, namely, the high computational efficiencyof the WBM and the ability of the FEM to model problems of arbitrary geometrical complexity. Numericalvalidation examples show the enhanced computational efficiency of the WBM for problems of moderategeometrical complexity and of the hybrid method for real-life engineering problems.

16:20 Sound transmission loss calculations of different structural subsystems using“OpenSourceSEA” (ID 138)Soren Callsen, Otto von Estorff, Hamburg University of Technology, GermanyAlexander Peiffer, EADS Corporate Research Centre, Germany

Following the growing demand for an open platform which allows the common development of SEA basedcomputational algorithms, a JAVA tool called OpenSourceSEA has been developed. It is accessible via theinternet (http://opensea.mub.tu-harburg.de). In the current paper, first the state of this software is described.Then the setup of an adequate SEA model for the computation of the sound transmission loss is considered.The implemented approaches are discussed and the computational results are verified by an element basedapproach as well as by applying the software package AutoSEA2. Additionally, the computations are com-pared with measured data. The regarded structural components are a plane isotropic panel and a structureconsisting of a composite laminate.

12


16:45 Coupled vibro-acoustic analysis in 3D using a wave expansion finite difference tech-nique (ID 207)Gabriel Ruiz, Harman/Becker Automotive Systems, United Kingdom

The ”mid-frequency gap” or ”twilight zone” is what many authors refer to as the frequency band for whichthere is yet no established efficient modelling tool in vibro-acoustic analysis. This band usually extendsfrom a few hundred hertz to about 1KHz depending on the particular application, between the upper limitefficiency of traditional modelling techniques like finite elements, and the lower accuracy limit of non-deterministic approaches like SEA.

The main challenge to alternative computational schemes, is to reduce the nodal density required to modelthe wave propagation in the fluid region whilst being robust and well conditioned at the same time. In thispaper, a highly efficient wave expansion technique is used to model the fluid region in combination with astandard finite element approach for the structure. The technique is applied to a 3D conceptual car cabinmodel, showing to be much more efficient than finite element procedures.

17:10 Acoustic convected radiation using a partition of unity formulation (ID 281)Tanguy Mertens, Philippe Bouillard, Universite Libre de Bruxelles, BelgiumJeremy Astley, Pablo Gamallo, University of Southampton, United KingdomL. Hazard, Universite Libre de Bruxelles, Belgium

An Infinite Partition of Unity Method (IPUM) is implemented to analyse axisymmetric radiation applica-tions. The formulation is based on the convected wave equation such that it takes into account the effectof non-uniform flows on acoustic propagation. Convected IPUM has been developed to reach better perfor-mances while analysing turbofans inlet noise applications. The paper describes the formulation and comparesthe performance of the developed method with analytical solutions and results obtained with the commercialsoftware ActranTM.

17:35 On the application of the Wave Based Method in the case of an exterior 3D domain(ID 184)Jevgenijs Jegorovs, Jan Mohring, ITWM, Fraunhofer-Institut fur Techno- und Wirtschaftsmathe-matik, Germany

This paper discusses the possibility to apply a novel numerical technique, i.e. Wave Based Method, to solvesteady-state acoustic problems in the case of an exterior 3D domain. Three examples, which corroborate ourtheory, are also presented.

18:00 Test strategy for aero-engine structural dynamic model validation (ID 464)Jose Garcia, Rolls-Royce plc., United KingdomDavid J. Ewins, Imperial College London, United Kingdom

Finite Element Models have been used for many years to predict the structural dynamic behaviour of aero-engines and to influence their design accordingly. The potential savings anticipated by such methods canonly be realised if predictions are reliable, incorrect predictions might mislead the designers with seriousconsequences. Currently, a reliable validation of an aero-engine assembly model is only feasible when thefirst engine prototype is ready for Modal Testing. This paper presents a methodology, based on sensitivityanalysis, to reduce the validation of complex assembly models to the simpler task of validating a reducednumber of its components. It also gives guidance for the selection of optimal laboratory configurations thatprovide the appropriate modal properties for the validation of such individual components. Significant im-provements in the timescales and cost for the validation of assembly predictions can be easily foreseen. Theproposed methodology is demonstrated on an aero-engine assembly.

13


Operational modal analysis – OMA1Room 4 – Chairman: N. Maia

13:00 A Low Order Frequency Domain Algorithm for Operational Modal Analysis (ID209)Shashank Chauhan, Raymond Martell, David Brown, Randall Allemang, University of Cincinnati,United States of America

Most of the algorithms for Operational Modal Analysis work in the time domain and there are very few fre-quency domain based algorithms. One of the reasons for this is the poor numerical characteristics associatedwith higher order frequency domain algorithms like Rational Fraction Polynomial (RFP). These limitationscan be improved by using methods such as frequency normalization and use of orthogonal polynomials inthe traditional experimental modal analysis set up. However estimating modal parameters in the frequencydomain using output-only response data still remains a challenge. In this paper a low order frequency domainalgorithms is proposed for Operational Modal Analysis. This algorithm is derived using the Unified MatrixPolynomial Approach (UMPA) and is evaluated using a theoretical and an experimental study.

13:25 Parameter identification of ship structures using classical and operational modalanalysis (ID 135)Sven-Erik Rosenow, Gunther Schlottmann, University of Rostock, Germany

A precise prediction of the dynamic behaviour of ship structures is important to avoid resonances and highvibration levels under operation conditions. For model correlation purposes and forced vibration calcula-tions experimentally determined natural frequencies, corresponding mode shapes and damping values arenecessary. To identify modal parameters using classical modal analysis, much effort is required to achievean adequate excitation of these big structures. Good results were obtained using a big excitation mass (600kg) in combination with sensitive seismic accelerometers. For several reasons this classical modal approachwith its artificial excitation is restricted to the use at a ship yard only. To obtain modal parameters of shipstructures under real draft, ballast, loading and surrounding water conditions, a suitable testing techniqueis needed. To evaluate the suitability of operational modal approach both, classical and operational modalanalysis were applied under ship yard conditions as well as operational modal analysis during several testtrails. Under ship yard conditions both testing techniques have shown adequate results. For the identificationof the lowest elastic modes of the ship structure operational modal analysis is the preferable tool. Applyingoperational modal analysis under test trail conditions, harmonic vibrations, induced mainly by main engineand propeller blade excitation, pollute the recorded Power Spectral Densities and reduce the frequency rangesfor modal parameter estimation.

13:50 Using Enhanced Frequency Domain Decomposition as a Robust Technique to Har-monic Excitation in Operational Modal Analysis (ID 170)Niels-Jørgen Jacobsen, Bruel & Kjaer Sound & Vibration Measurements A/S, DenmarkPalle Andersen, Structural Vibration Solutions A/S, DenmarkRune Brincker, University of Aalborg, Denmark

The presence of harmonic components in the measured responses is unavoidable in many applications ofOperational Modal Analysis. This is especially true when measuring on mechanical structures containingrotating or reciprocating parts. This paper describes a new method based on the popular Enhanced FrequencyDomain Decomposition technique for eliminating the influence of these harmonic components in the modalparameter extraction process.

For various experiments, the quality of the method is assessed and compared to the results obtained usingbroadband stochastic excitation forces. Good agreement is found and the method is proven to be an easyto-use and robust tool for handling responses with deterministic and stochastic content.

14


14:15 Considerations in the Application of Spatial Domain Algorithms to OperationalModal Analysis (ID 210)Shashank Chauhan, Raymond Martell, David Brown, Randall Allemang, University of Cincinnati,United States of America

The Complex Mode Indicator Function (or CMIF) is a popular spatial domain modal parameter estimationtechnique that utilizes the singular value decomposition of the frequency response function matrix for esti-mating the modal parameters of the system. Due to several advantages like identification of closely spacedmodes, this technique is extremely popular for modal parameter estimation purposes. In recent times, theFrequency Domain Decomposition (FDD) technique was developed that extends the CMIF algorithm to theoperational modal analysis framework. The FDD technique works on the power spectrums unlike workingon frequency response functions as in conventional modal analysis. Normally the FDD is followed by theEnhanced Frequency Domain Decomposition (eFDD) to complete the overall parameter estimation proce-dure. In this paper an alternative to the eFDD, the previously introduced Enhanced Mode Indicator Function(EMIF), is reviewed and extended to the operational modal analysis framework. This algorithm differs fromthe eFDD in that the parameter estimation is carried out in the frequency domain. Further the paper analyzesthe application of spatial domain algorithms to operational modal analysis framework in more detail. It dis-cusses the critical issues and limitations associated with the application of spatial domain algorithms to theOMA framework under different excitation scenarios and proposes a simple tool, Singular Value PercentageContribution (SVPC) plot to deal effectively with them.

14:40 Modal Parameter Identification From Operative Responses (ID 586)Alessandro Agneni, Giuliano Coppotelli, University of Rome “La Sapienza”, Italy

Modal parameter estimates of a structure, vibrating under actual loadings and operational conditions, haveaddressed the interest of researchers during the last years. For all the developed approaches, structures neednot be excited by expensive devices, because the modal signature of the structure can be achieved fromthe knowledge of the system time responses only, derived from the natural excitation due to the ambientloads. Thus natural frequencies, damping ratios, and mode shapes can be estimated using techniques bothin the frequency and in the time domain only by the measured responses of the dynamic system. In thispaper, a methodology to estimate the modal model of a dynamic system is proposed. The biased frequencyresponse functions could be identified from the measured time responses by a developed technique basedon the properties of the Hilbert transformation applied to the auto power spectral densities of the outputsignals. Then, trough the identification of the state space matrix, the modal parameters could be derived.The accuracy of the proposed approach has been investigated through several experimental investigationscarried out on different aerospace structures.

Parameter estimation – PE1Room 4 – Chairman: R. Allemang

15:30 A Complete Review of the Complex Mode Indicator Function (CMIF) with Appli-cations (ID 459)Randall Allemang, David Brown, University of Cincinnati, United States of America

Over the last twenty years, the Complex Mode Indication Function (CMIF) has become a common numericaltool in processing experimental data. The literature contains many references to the original and subsequentdevelopments of the CMIF including the enhanced mode indication function (EMIF) and the use of CMIFtogether with the enhanced frequency response function (EFRF) to form spatial domain modal parameter es-timation methods. Another development was the estension of the single degree-of-freedom (SDOF) aspectsof the CMIF method to include a limited number of modes in what is known as the Enhanced Mode Indi-cator Function (EMIF) method. This paper brings all of the development together in a single reference withcommon nomenclature and gives examples of uses of the CMIF, EMIF and EFRF in various applications.

15


16:20 Experimental modal analysis using blind source separation techniques (ID 81)Fabien Poncelet, Gaetan Kerschen, Jean-Claude Golinval, Universite de Liege, Belgium

Recently, statistical and empirical signal processing techniques such as the proper orthogonal decompositionand the Hilbert-Huang transform have shown promise for structural system identification. In the presentstudy, experimental modal analysis is carried out by employing blind source separation techniques and byinterpreting the response of a mechanical system as a static mixture of sources. Specifically, it is shownunder which circumstances the normal coordinates of the vibration modes may be interpreted as virtualsources. The advantages and limitations of the proposed method will be discussed, and the procedure willbe demonstrated using numerical applications.

16:45 Identification of a Global Model Describing the Temperature Effects on the Dynam-ics of a Smart Composite Beam (ID 230)John Hios, Spilios Fassois, University of Patras, Greece

The identification of a global model describing the dynamics of a smart composite beam under various tem-peratures is addressed. The problem is treated within a novel Statistical Functional Pooling Frameworkfeaturing global, stochastic Functionally Pooled (FP) models with explicit functional dependence on temper-ature. This framework circumvents the disadvantages associated with conventional multi-model approachesin which a customary model is identified for each temperature, with no explicit dependence on temperaturebeing directly provided. In addition it offers a compact global model and optimal statistical accuracy. Aglobal Functionally Pooled Vector AutoRegressive with eXogenous excitation model (FP-VARX model) de-scribing the dynamics of the considered beam is then identified using experimental data records. Its analysisindicates that the beam’s natural frequencies decrease with increasing temperature in a somewhat nonlinearor approximately linear fashion, while the dependence on temperature seems weaker, but of potentially morecomplicated nature, for the damping factors. The global model characteristics are confirmed as being in goodagreement with those obtained by conventional multi-model analysis.

17:10 Identification of Flexibility Parameters of 6-axis Industrial Manipulator Models (ID176)Jonas Ohr, ABB AB, Corporate Research, SwedenStig Moberg, ABB Robotics / Linkopings universitet, SwedenErik Wernholt, Linkopings universitet, SwedenS. Hanssen, ABB Robotics, SwedenJ. Pettersson, ABB AB, Corporate Research, SwedenS. Persson, ABB Robotics, SwedenS. Sander-Tavallaey, ABB AB, Corporate Research, Sweden

A method for identification of flexibility parameters of a 18 DOF (degrees of freedom) robot prototype modelis proposed. Experiments show the strength of the method and the results indicates that flexibilities in thebearings and the arms, taken together, are of the same order as the flexibilities in the gears.

17:35 Rigid Body Properties of An Actual Diesel Engine by Experimental Spatial MatrixIdentification Method (ID 80)Toshiyuki Uyama, YANMAR CO.,LTD., JapanMasaaki Okuma, Atsushi Oyama, Tokyo Institute of Technology, Japan

In this paper, the authors present the results of experimental identification of the rigid body properties of anactual engine structure using an experimental spatial matrix identification method. The method identifies aset of spatial matrices using experimentally measured single-input-and-multiple-output frequency responsefunctions within a frequency range of interest including the first natural frequency. The sets of spatial ma-trices can represent not only dynamic characteristics but also the rigid body properties of test structures. Inthis paper, an actual three-cylinder diesel engine is tested to verify the method especially focusing on the

16


identification of its rigid body properties. The engine is complex so that many natural modes exist in thefrequency range of interest. Nevertheless, the method successfully identified a set of spatial matrices thatcan represent the dynamic characteristics of the engine. Then, all the rigid body properties are derived fromthe identified mass matrix with practical accuracy.

18:00 Mechanical System Identification by Multivariate Time Series Models (ID 561)Roberto Pedro Baldeon Amaro, Universidade Estadual de Mato Grosso, BrazilPaulo Roberto Gardel Kurka, Universidade Estadual de Campinas, Brazil

The work presents a Multivariate Time Series Models in the identification of mechanical systems. Themaximum likelihood technique (ML) is applied to estimate the parameters of the time series models in orderto improve the precision in the estimation of modal parameters. Spliid’s algorithm is used to estimate initialvalues to start the iterative process of the ML technique. The performance of the ML technique is verified ina three degree freedom simulated system with two inputs and two outputs. Stochastic noise is added to theoutputs in order to verify the performance of the time series model in the presence of stochastic noise.

Flite Eureka 2 – FLI1Room 5 – Chairman: J. Cooper

13:00 Current Status and Challenges for Flight Flutter Testing (ID 417)Asim Abbassi, Jonathan Cooper, University of Manchester, United Kingdom

A brief review of Flight Flutter Testing, its current status and some future challenges, is made particularlywith reference to data analysis issues. A number of improvements to the interpretation of stability plots andthe Flutter Margin are made, employing Backwards Least Squares and statistics of Least Squares estimators.Some initial results on a simulated binary flutter system show that the performance of the proposed methodsis encouraging.

13:50 In-flight modal analysis - a comparison between sweep and turbulence excitation(ID 565)Bart Peeters, LMS International, BelgiumTim De Troyer, Erasmushogeschool Brussel, BelgiumPatrick Guillaume, Vrije Universiteit Brussel, BelgiumHerman Van der Auweraer, LMS International, Belgium

At the end of the development cycle, a new aircraft is certified by means of in-flight flutter tests. These testsconsist of flying the aircraft at different airspeeds and measuring the accelerations at a limited number oflocations on the aircraft structure. The scope is to open the flight domain by verifying that the aircraft doesnot suffer from aero-elastic instabilities such as flutter.

In this paper, some modern frequency-domain modal parameter estimation methods are applied to in-flightdata of a large aircraft. Traditional sine sweep excitation was applied at the control surfaces. However, dur-ing the test the aircraft passed through a turbulent zone. The sweep excitation was immediately stopped, butthe on-board data acquisition system continued to record the aircraft vibration response. After quitting theturbulent zone, the sweep test was reinitiated. The present data thus allows for a comparison between artifi-cial and natural excitation. More specifically, aspects such as data pre-processing, easiness of the parameterextraction process and the accuracy of the results are investigated.

17


14:15 An experimental investigation of the subsonic stall flutter (ID 113)Jing Li, Nikolaos Andrinopoulos, Grigorios Dimitriadis, University of Manchester, United King-dom

This paper reports on experimental investigations of the subsonic stall flutter of a wing in a wind tunnel.Stall flutter (also know as Dynamic Stall) is a LCO phenomenon occurring when all or part of the flow overa wing separates and re-attaches at least once during a full oscillation period. In order to investigate theaeroelastic effects of the nonlinearity introduced by the boundary layer growth and flow separation and theensuing stall-induced LCO, a wind tunnel model of a wing undergoing stall flutter is designed, built andtested. The model is a rectangular wing with constant cross-section free to move in the pitch and plungedirections, restrained by torsional and linear springs respectively. The motion of the wing is measured usinglaser displacement probes. Two types of stall flutter are observed and measured: 1. Non-symmetric stallflutter, where the flow separates over one side of the wing only, and 2. Symmetric stall flutter, where the flowseparates over both sides of the wing (deep stall). The bifurcation behaviour of the wing is very complex andboth types of LCO can be observed during a single response history.

14:40 Detection of turbulence during flutter tests (ID 116)Pierre Vacher, Alain Bucharles, ONERA, France

Flutter tests are a crucial phase of the flight test program of a new aircraft. One of the main goals is toprove the absence of unstable aeroelastic modes throughout the flight domain. This is achieved by applyingcalibrated excitations to the aircraft structure. Undisturbed measurements are required for this operation.Hence turbulent conditions are avoided as much as possible. In the framework of a research program incollaboration with Airbus, a toolbox was developed by ONERA for processing flight test data. In this article,we present the tool that was designed to detect the occurrence of turbulence gusts during flight tests evenwhen excitations are applied to the aircraft. This paper details the structure and the implementation of thisturbulence detector. It also describes an innovative and efficient formulation of an adaptive bandstop filter.

15:05 Fast Derivation of Uncertainty Bounds for On-line Flight Flutter Testing (ID 437)Tim De Troyer, Erasmushogeschool Brussel, BelgiumPatrick Guillaume, Rik Pintelon, Vrije Universiteit Brussel, BelgiumBart Peeters, LMS International, Belgium

One of the most demanding applications of modal analysis is flight flutter testing. The classical approachis to expand the flight envelope of an airplane by performing a vibration test at constant flight conditions,curve-fit the data to estimate the resonance frequencies and damping ratios, and then to plot these estimatesagainst flight speed. The damping values are then extrapolated in order to determine whether it is safe to pro-ceed to the next flight test point. The modal parameter estimation algorithm must thus be fast, able to handlevery noisy data and yield accurate estimates of the damping ratios. The extrapolation process can be greatlyenhanced through the use of statistical uncertainty bounds on the poles estimates. However, the derivation ofthese bounds requires additional calculations and thus time. In this contribution, a new technique to calculateuncertainty bounds will be discussed, tested and evaluated. Special attention will be paid to the calculationof uncertainty bounds on estimates resulting from poly-reference frequency-domain leastsquares based esti-mators. These algorithms allow to determine the uncertainties with only minor additional calculations. Thisfast approach can be of great importance for future on-line flight flutter testing.

18


Flite Eureka 2 – FLI2Room 5 – Chairman: T. Uhl

15:55 Quantification of Asymmetric In-Flight LCO Response and Prediction using NeuralNetworks (ID 123)Kenneth Dawson, Daniel Maxwell, TYBRIN Corporation, United States of America

A static artificial neural network is developed to predict limit cycle oscillation (LCO) amplitude levels andfrequency of external store configurations on a fighter aircraft. This work is similar to previous networksgenerated by the authors with the inclusion of continuous analytic descriptors of the aeroelastic modes andthe flutter eigenvectors. The current work however uses the analytic descriptors of three normal modes, andallows the neural network to determine the effect of the coupling modes. In addition, asymmetric externalstore configurations are examined in this neural network, similar only by the missile configurations on thewingtips. A neural network with two layers, one with 41 nodes and the second with 17 nodes, is used forthe prediction of the two outputs, LCO amplitude and LCO frequency. The methodology for the calculatingthe descriptors is presented, as is a comparison of the analytic and flight test descriptors. In addition, samplenetwork input training data is provided. The results of the neural network using the analytic descriptors showa capability to predict trends for both the LCO onset speeds and the overall amplitude levels. This neuralnetwork was also capable of predicting an observed trend of LCO amplitude levels decreasing partially orcompletely after a certain airspeed.

16:20 When is a pole spurious? (ID 137)Ivan Markovsky, Jeroen Boets, Bart Vanluyten, Katrien De Cock, Bart De Moor, Katholieke Uni-versiteit Leuven, Belgium

The stabilization diagrams, used in modal analysis, rely on the intuitive notion of a spurious pole. In thispaper, we give a definition of a spurious pole, based on the most powerful unfalsified model (MPUM) ofthe data, i.e., on an exact model for the data. The poles of the MPUM are by definition physical and a polethat is not physical is by definition spurious. Our definition does not make assumptions about the data, apartfrom the basic postulation of the linear time-invariant model class. In this sense it is unprejudiced. Since theMPUM can be constructed from the data, one can compute the physical poles and thus answer the questionin the title. If, however, one knows a priori that the data correspond to a noise corrupted trajectory of a truedata generating system or that there is an unobserved process noise acting on the true system, one should usethis knowledge. In this case, the MPUMconcept has to be modified to allow for approximation. Methods forapproximate system identification and model reduction are reviewed and applied for spurious pole detectionof simulated data.

16:45 CUSUM test for flutter monitoring of modal dynamics (ID 193)Rafik Zouari, Laurent Mevel, Michele Basseville, Institut de Recherche en Informatique etSystemes Aleatoires, France

This article presents an extension of a flutter detection procedure based on the flutter margin commonlyused for flutter prediction. The flutter margin is a stability parameter based on the Routh’s criterion thatmeasures the interaction between two modes. A simplified aeroelastic model of a pitch-plunge wing sectionis studied and time series are simulated for different flow velocities in order to illustrate the flutter behavior.A CUSUM test, designed for monitoring the flutter margin variation, is applied to simulated measurementwith increasing velocities. The results show a good reaction with the flutter margin decrease and the flutteronset can be detected.

19


17:10 Recursive subspace identification for in-flight modal analysis of airplanes (ID 136)Katrien De Cock, Katholieke Universiteit Leuven, BelgiumGuillaume Mercere, Laboratoire d’Automatique et d’Informatique Industrielle, FranceBart De Moor, Katholieke Universiteit Leuven, Belgium

In this paper recursive subspace identification algorithms are applied to track the modal parameters of air-planes on-line during test flights. The ability to track changes in the damping ratios and the influence of theforgetting factor are studied through simulations.

17:35 The use of wavelet transform for in-flight modal analysis (ID 432)Tadeusz Uhl, Andrzej Klepka, AGH - University of Science and Technology, Poland

The paper presents a method for modal analysis of mechanical structures for systems with varying parame-ters. Such a case can be observed, for in-flight test of airplane if some flight parameters are changed. Dueto these reasons model parameters which depend on aeroelastic properties of a system varying and system isno stationary. Proposed approach is based on system response measurements (acceleration) and filtering ofmeasured signals with use of wavelet based filter. The Morlet type of wavelet function in applied to separateparticular modes of the structure. A possibility of decoupling of natural modes in vibration structures byusing Morlet wavelet is proved. The MATLAB package implementation of the algorithm is shown. Imple-mented software has been tested on simulated and experimental data. Successful application to real systemmodal analysis is discussed on example of the airplane in-flight test.

18:00 Output-only Technique for Estimation of Nonlinear System Parameters for In-flightTest Application (ID 433)Joanna Iwaniec, Tadeusz Uhl, AGH - University of Science and Technology, Poland

Classical nonlinear system identification algorithms consist of two main steps. In the first step, the consid-ered system is excited at an operating amplitude, where its dynamic behaviour is assumed to be linear. Onthe basis of the measured excitation and system responses, linear system parameters are estimated. In thesecond step, nonlinear system parameters are evaluated. Unfortunately, the behaviour of a wide class of me-chanical systems is not linear around any operating point while exciting force measurements are frequentlydifficult or even impossible to carry out. Therefore the field of applications of the classical nonlinear systemidentification methods is rather limited.

The paper concerns the method combining restoring force, boundary perturbation and direct parameter esti-mation techniques that can be used for the purposes of parameter identification of nonlinear systems workingunder immeasurable operational loads. The results of method application to parameter identification of theSkytruck airplane landing gear are presented.

Modal testing and FRF estimation – MT1Room 6 – Chairman: N. Okubo

13:00 Development of a New Automated Modal Test Planning Algorithm (ID 83)Phil Daborn, P.R. Ind, Alun Tribe, Kevin Garraway, AWE, United Kingdom

This paper discusses the development of an automatic test-planning algorithm to aid the design of modaltests. Essentially, the paper presents the amalgamation of two existing test planning techniques, the optimumexcitation point calculations, such as the Optimum Drive Point (ODP) as given by ICATs documentation,and the more recent Min-MAC algorithm of Carne et al. The resulting “OptiSet” technique provides analternative to the commonly used Effective Independence (EI) calculation, which is occasionally found tocluster test points and which may run for many hours when used on large models. The OptiSet techniqueresults in analytical AutoMAC matrices with no high-valued off-diagonal elements, using an optimal num-

20


ber of measurement DOFs. Numerical and actual experiments are used to demonstrate the effectiveness ofthe OptiSet technique. The paper also discusses the improvement in computational efficiency of the OptiSettechnique over the EI method.

13:25 A Review of Experimental Modal Analysis Methods with respect to their Applica-bility to Test Data of Large Aircraft Structures (ID 174)Marc Boswald, Dennis Goge, Ulrich Fullekrug, Yves Govers, Deutsches Zentrum fur Luft- undRaumfahrt e.V., Germany

In this paper, different modal analysis tools are investigated with respect to their applicability to test datafrom ground vibration tests (GVT) of large aircraft. Three different modal analysis methods are investigated.These are the Frequency Domain Direct Parameter Identification (FDPI), the Least-Squares Complex Expo-nential method (LSCE), and the Least-Squares Complex Frequency Domain method (LSCF or PolyMAX).Test data from GVT of a large aircraft usually have features like noise contamination and weak non-linearityof some modes. Thus, the influence on the results of experimental modal analysis of statistical errors (suchas noise) and systematic errors (such as non-linearity) in the FRFs are investigated. Simulated FRFs ofan analytical system contaminated with different levels of artificial noise are used to study the influenceof statistical errors, whereas simulated non-linear FRFs obtained from an analytical system with a friction-type non-linearity is used to study the influence of systematic errors. In addition, the aforementioned modalanalysis methods were applied to a dataset acquired during the GVT of the Airbus A380. The modal analysisresults obtained with the different analysis methods are compared and evaluated based on the results of theanalytical study of the influence of statistical and systematic errors.

13:50 Measurement plan definition on Ariane V launcher (ID 378)Alexis Macaire, EADS SPACE Transportation, FranceFrancois Dubois, CNAM, France

This paper deals with the dynamic behavior of Ariane 5 launcher during the flight. The studies performedfor the production of Ariane 5, concerning the dynamic levels analyses, consist of a flight prediction anda measurements post flight analysis based on several sensors placed on the launcher. Studies are carriedout to make the best use of flight data to improve the models representativeness. This paper concerns thedefinition of sensors (number, placement, orientation, etc.) allowing an efficient identification of the modalcharacteristics of the launcher. A robust design approach has been tested in order to take into account themodeling errors. The result is sub-optimal with respect to a theoretical approach without any error, butit is supposed to be robust to modeling discrepancies. A second algorithm is also proposed to assess thevalues of certain parameters (for instance the stiffness of certain components), taking into account flightsmeasurements. A simple case has been treated to validate the feasibility. The sensor placement step has beenperformed for a booster ground test that should occur before the end of 2006.

14:15 Estimation of Powertrain Inertia Properties via an In-situ Method (ID 579)David Johnson, Jeffrey Van Karsen, Jason Blough, Mohan Rao, Michigan Technological Univer-sity, United States of AmericaTony Ge, Ford Motor Company, United States of America

The objective of this paper is to develop a method that can be used to determine the rigid body dynamicproperties of a powertrain while it remains in the vehicle. The method relies on the Equation of Motion ofa rigid body and uses the measured acceleration of the body, along with the measured and estimated forcesapplied to the body, to calculate the ten rigid body dynamic properties of the powertrain: mass; x, y, z whichdescribe the position of the center of mass; the three moments of inertia, Ixx, Iyy, Izz; and the three productsof inertia, Ixy, Ixz, Iyz.

This paper will explain the procedure and theory behind the proposed method. Results from using themethod on both a Ford 500 powertrain, and a Test Mass that has dynamic characteristics similar to a front

21


wheel drive powertrain are presented. It is concluded from the test results that the method can be used toaccurately estimate the rigid body dynamic characteristics of a powertrain in a vehicle.

14:40 Practical Guidelines for Optimising the Measurement of Frequency Response andCoherence Functions with Digital Spectral Analysers (ID 639)Jerome Antoni, Universite de Technologie de Compiegne, FranceJohan Schoukens, Vrije Universiteit Brussel, Belgium

In modern digital spectral analysers, frequency response functions and related quantities such as the coher-ence function are measured by means of the Weighted Overlapped Segment Averaging (WOSA) technique.There are surprisingly very few results concerning the optimal setting of the parameters in WOSA and theirimpact on the precision and accuracy of the measurements. In this paper, we provide the optimal windowshape and the optimal percentage of overlap to be used with WOSA, together with the bias and variance for-mulae of frequency response and coherence functions in the general case. The distinction is made betweenthe H1, H2 and H3 estimators. These results are illustrated in the context of experimental modal analysis.

15:05 Investigation of Boring Bar Mode Shape Rotation by Experimental Modal Analysisin Correlation with Finite Element Modeling (ID 658)Tatiana Smirnova, Henrik Akesson, Lars Hakansson, Ingvar Claesson, Blekinge Institute of Tech-nology, SwedenThomas Lago, Acticut International AB, Sweden

Productivity degrading vibration problems are traditional in metal cutting, in particular in internal turning,when the boring bar is long and slender. The high levels of vibrations arise at the boring bar’s naturalfrequencies corresponding to its fundamental bending modes. The vibrations are dominating at the boringbar’s eigenfrequency in the cutting speed direction, since the cutting force has the largest component in thisdirection. The negative effects of vibrations, e.g. poor surface finish, reduced tool life, high sound pressurelevels in the working environment etc., can be eliminated for instance by using passive tuned damper oractive control. The level of success implementing any of these methods is dependent on the knowledge ofthe dynamic properties of tooling structure - the interface between the cutting tool or insert and the machinetool - involved. Results from experimental modal analysis on the boring bar clamped in a lathe reveala rotation of the mode shapes with respect to the cutting speed and cutting depth direction. The paperaddresses boring bar mode shape rotation phenomenon and discusses its possible sources based on resultsobtained from experimental modal analysis of two systems: boring bar with original boundary conditionsand boring bar with linearized boundary conditions in correlation with results obtained from corresponding3-D finite element models.

Damage monitoring and identification – DMI1Room 6 – Chairman: D. Inman

15:55 Vibration based SHM : comparison of the performance of modal features vs featuresextracted from spatial filters under changing environmental conditions (ID 94)Arnaud Deraemaeker, Universite Libre de Bruxelles, BelgiumEdwin Reynders, Guido De Roeck, Katholieke Universiteit Leuven, BelgiumJyrki Kullaa, Helsinki Polytechnic Stadia, Finland

One critical issue in vibration based SHM is to be able to differentiate the effects of variability inherent tothe system and its environment from a potential damage to be detected. In this paper, using a numericalmodel of a bridge subject to changing environment, we study the damage detection problem using (i) theeigenproperties of the system extracted by means of stochastic subspace identification, and (ii) peak indi-

22


cators extracted from the output of spatial filters. Both techniques use ambient (output-only) vibrations. Ina first step, sensitivity of the features to the environment is studied. In a second step, based on long termmonitoring of the undamaged structure under changing environmental conditions, factor analysis is appliedto the two different kinds of features in order to remove the effects of environment. A comparison of theperformance of the different features is presented.

16:20 A Study on the Correlation between PZT and MFC Resonance Peaks and DamageDetection Frequency Intervals Using the Impedance Method (ID 92)Daniel Peairs, Pablo Tarazaga, Daniel Inman, Virginia Polytechnic Institute and State University,United States of America

Impedance-based structural health monitoring uses collocated piezoelectric transducers to locally excite astructure at high frequencies. The response of the structure is measured by the same transducer. Changes inthis response indicate damage. Frequency range selection for monitoring with impedance-based structuralhealth monitoring has, in the past, been done by trial and error methods, or been selected after analysis byengineers familiar with the method. For future applications it is desirable to be able to automatically selectfrequency ranges, perhaps even before installing the system. In this study, analysis of the measurementchange through a damage metric is examined and related to characteristics of the measurement. Specifically,an outlier detection framework was used to statistically evaluate the sensing ability of the transducers atvarious frequency ranges. The variation in undamaged measurements is compared to the amount of changein the measurement upon various levels of damage. Testing was performed with both solid piezoceramictransducers and macro-fiber composite (MFC) piezoelectric devices of different sizes bonded to aluminumand fiber reinforced composite structures. The results indicate that frequency ranges containing a resonanceof the actuator are more suited for structural health monitoring.

16:45 Determination of Total Hip Replacement stem insertion endpoint and stability as-sessment by vibration analysis: first experiences with per-operative measurements(ID 357)Leonard Pastrav, Siegfried Jaecques, Katholieke Universiteit Leuven, BelgiumMichiel Mulier, University Hospital Gasthuisberg, BelgiumGeorges Van der Perre, Katholieke Universiteit Leuven, Belgium

The assessment of the primary stability of cementless implants still remains a subjective factor in total hipreplacement (THR) and, as a consequence, the excessive press-fitting of a THR component can be the causeof intra-operative fractures. Objective information about the stability of implant-bone structures can be ob-tained using methods based on vibration analysis. After extensive in vitro studies, a per-operative protocolwas designed to detect the insertion endpoint and/or to assess the stability of custom made hip prostheses.The experiments were performed on cementless hip stems and on hybrid hip stems that were partially ce-mented distally. This paper presents the frequency response function evolution during the hip stem insertionin the femur, in per-operative conditions.

17:10 Sensor fault identification and correction in structural health monitoring (ID 283)Jyrki Kullaa, Helsinki Polytechnic Stadia, Finland

The scope of this study is to identify and correct a faulty sensor using solely the measurement data. Themethod uses training data from the full undamaged sensor set on the structure to build a correlation modelbetween the sensors. This correlation model is used to detect, isolate, and correct the faulty sensor. Signalsfrom the faulty sensor can be reconstructed using those of the healthy sensors. The method is based on themissing data analysis using the time history data, e.g. accelerations, without a feature extraction process.This straightforward approach makes it possible to distinguish the sensor fault from the structural damage.Experimental multichannel acceleration measurements were used to verify the proposed method. Differentsensor faults were studied by modifying one of the sensor signals. It is anticipated that more structures in

23


the future are equipped with a sensor array having numerous sensors. This fact makes the proposed methodvaluable and available for practical applications.

17:35 Detection of corrosion damage in RC beams using modal testing (ID 10)Zubaidah Ismail, P.L. Tan, Hashim Razak, University of Malaya, Malaysia

This paper describes the determination of location of damage due to corrosion in RC beams using modaltesting. Experimental modal analysis was performed on a control beam and the beam with induced acceler-ated corrosion. The location of damage due to corrosion was determined using an indicator. The indicatorabs(lambda4) was obtained by rearranging the equation for free transverse vibration of a uniform beam, andapplying the fourth order centered finite-divided difference formula to the regressed mode shape data. Theequation is an eigenvalue problem, and the value of abs(lambda4) would be a constant. Differences in thevalues of this indicator indicated stiffness changes, and the affected region indicated the general area of dam-age, the exact location being around the center of the region. The proposed algorithm on the mode shapecould form the basis of a technique for structural health monitoring of damaged reinforced concrete struc-tures. This investigation provided a useful insight on the use of modal analysis to detect corrosion damagein the appraisal and assessment of structural concrete elements.

18:00 Damage identification for structural health monitoring using modal-based methodand neural network classifier (ID 576)Assunta Sorrentino, Ignazio Dimino, Vincenzo Quaranta, E. Quaranta, CIRA SCpA, Italy

The goal of this research is to develop a computational model for interpreting time series data and extractinguseful information for structural health monitoring.

Damage detection is treated as a pattern classification problem. Artificial neural networks are employed forstructural health monitoring thanks to their strong capabilities of pattern recognition and classification, datainterpretation and function approximation.

A modal-based damage detection algorithm for the structural health monitoring of a reinforced panel is pre-sented. Two neural networks have been designed: one to identify the damage location and one to recognizeits extent from the measured modal properties. The results show that the system is able to identify the damagepresence, level and location with a very high percentage of successful classification.

Application of (new) vibroacoustic methods to complex systems –AVM1Room 7 – Chairman: A. Sestieri

13:00 Introduction to AVM sessionAldo Sestieri, University of Rome “La Sapienza”, Italy

13:25 A review of the scaling procedure for the analysis of the vibroacoustic responses (ID616)S. De Rosa, F. Franco, University of Naples “Federico II”, Italy

This paper presents a review of the scaling procedure defined and applied for the predictive vibroacousticresponses. Large part of the most interesting results was obtained in the activities planned for a nationalresearch project, where a common benchmark was assembled. The scaling procedure, now named ASMA(Asymptotical Scaled Modal Analysis), allows representing the original system by using a reduced modalbase and a scaled domain so that the computational efforts can be significantly reduced. Here, the mostimportant results obtained applying ASMA to several test cases will be recalled, paying also attention to theproper formal definition of the overall procedure. Fundamental aim of the present paper is the theoretical

24


review of the ASMA methodology and the discussion of some numerical results on a standard vibroacousticconfiguration.

13:50 Efficiency estimation of a benchmark SEA model (ID 613)Antonio Culla, Aldo Sestieri, University of Rome “La Sapienza”, Italy

Statistical Energy Analysis (SEA) is the most acknowledged technique to solve high frequency problems.However, SEA can be able to provide reliable results only if particular conditions are satisfied. Thus, somebasic hypotheses define the range of applicability of the method. The coefficients of the SEA equationsdepend on the coupling loss factor (CLF), internal loss factor (ILF) and modal densities. Theoretical rela-tionships allow to predict the CLFs and ILFs of simple subsystems, but very often the theoretical estimateof these parameters is incorrect and their values may be rather obtained experimentally. In this paper math-ematical relationships representing the basic hypotheses involving the SEA parameters are presented, andtheir ability to provide a significant a priori evaluation of the SEA efficiency is tested on the case of a bench-mark made of three coupled plates. The energy in the system is experimentally determined, and this result iscompared with that of a SEA model: the compliance with the hypotheses is checked.

14:15 Application of the complex envelope vectorization to a benchmark made of threecoupled plates (ID 614)Aldo Sestieri, Oliviero Giannini, Antonio Carcaterra, University of Rome “La Sapienza”, Italy

The complex envelope vectorization (CEV) is an alternative technique to SEA and other recent energy meth-ods developed to study high frequency problems. Notwithstanding the large efforts devoted to this topic formany years, there is not yet an appropriate technique to tackle it efficiently. In the framework of an Ital-ian research project “Novel methods for the analysis and control of vibro-acoustic systems” the Universityof Naples and the University of Rome have by their own developed two methods (ASMA and CEV, re-spectively) that have been applied quite successfully to a benchmark made of three plates coupled together,providing quite interesting results. In this paper the CEV method is described and a critical analysis is car-ried out on the limits and advantages of this procedure. Although it is shown that CEV can be efficientlyapplied in principle to systems characterized by a relevant damping, the application of the method to theconsidered benchmark shows that the quality of results is very attractive even when a standard structuraldamping is considered. Thus, one can now begin to look at CEV as a promising approach to the solution ofvibro-acoustic problems.

14:40 Evaluation of the SEA parameters of a benchmark by the power injection method(ID 612)Antonio Culla, Aldo Sestieri, University of Rome “La Sapienza”, Italy

In this paper an experimental analysis on the energy transmission in vibrating structures is considered. TheStatistical Energy Analysis (SEA) is, at present, the most acknowledged theory for the solution of highfrequency vibroacoustic problems. The coefficients of the SEA equations depend on the coupling loss fac-tor (CLF), the internal loss factor (ILF) and the modal densities. For complex systems, the values of suchparameters cannot be provided by analytical relationships and it is rather necessary to determine them exper-imentally to provide a reliable solution. In this paper the case of three coupled plates is considered with theaim of determining the SEA parameters (CLF, etc). By forcing the system with a white random excitation,the acceleration of the structures is measured and the energy stored on each plate is determined. The solu-tion of an inverse problem provides the searched SEA parameters. These values are compared with thoseobtained by a SEA based method, and the SEA results are compared with experimental data.

25


15:05 Considerations on the Modelling of a Benchmark Structure and Applicability ofStatistical Energy Analysis (ID 615)Niccolo Baldanzini, P. Citti, Marco Pierini, A Scippa, Universita degli Studi di Firenze, Italy

The present work is concerned with modeling high frequency vibration in a simple structure through Sta-tistical Energy Analysis (SEA). The structure is the benchmark used in the Italian National Project “Novelmethods for the analysis and control of vibro-acoustic systems”. The objective of this work is twofold: totest the sensitivity of the model response to different modeling criteria for a benchmark structure; to assessthe importance of the subsystem damping related to the diffusivity of the vibroacoustic field in order to verifythe relation between the kind of the vibrational field and the modal overlap factor. The study is performedwith numerical tools and an extensive experimental activity, which includes also measurements on otherstructures similar to the benchmark. The results demonstrate that the modal overlap factor is not adequateto characterize completely the nature of the vibrational field and that damping doesn’t indefinitely improvesthe applicability of SEA.

15:30 A hybrid formulation for vibro-acoustic analysis of structures in the mid-frequencydomain (ID 471)Alessandro Pratellesi, Universita degli Studi di Firenze, ItalyMichel Viktorovitch, Rieter Automotive AG, SwitzerlandNiccolo Baldanzini, Marco Pierini, Universita degli Studi di Firenze, Italy

The mid-frequency field is a hybrid domain for which assembled structures exhibit simultaneously low- andhigh-frequency behaviours, depending on the material and geometrical properties of the different subsys-tems. Thus, dealing with the mid-frequency field requires simulation methods which are able to account thedifferences of behaviour of the different subsystems.

A hybrid formulation that is able to correctly predict the vibro-acoustic response of a mid-frequency structureis presented in this paper, it is named the FE-SIF. This formulation is based on the coupling of two differentformulations, the Finite Elements (FEM) for the low frequency behaving sub parts and the Smooth IntegralFormulation (SIF) applied to the high frequency subsystems. The FEM-SIF hybrid method enables to cor-rectly predict the deterministic response of the low frequency parts which is not affected by randomness, andthe smooth trend of the contributions of the high frequency parts.

The method therefore provides a relevant solution to the mid-frequency problem. The effectiveness of theformulation is demonstrated in the article with a numerical application to a 2-dimensional structure.

Active vibration control and smart structures – AVC1Room 7 – Chairman: N.B. Roozen

16:20 A fast converging adaptive control scheme for suppression of broadband distur-bances on a multi-variable vibration isolation setup (ID 390)Gerard Nijsse, Johannes van Dijk, Ben Jonker, University of Twente, The Netherlands

For research and development, a single reference input multiple actuator input multiple error sensor out-put vibration isolation setup is realized. The objective of the setup is to investigate if the flexible receiverstructure can be isolated from the rigid source structure by six piezo-electric actuators (serving as a hybridvibration isolation mounts), such that disturbances stemming from the source structure are reduced at thereceiver structure. Vibration isolation is established by minimizing signals from six acceleration sensor out-puts and by steering the piezo-electric actuator inputs. The first contribution of this paper is that by makinguse of an adaptive multi-variable feedforward controller which is updated on the basis of a computationallyefficient filtered error least mean square algorithm, broadband vibration isolation control is achieved with anaverage reduction of up to 9.4 dB in the error sensor outputs (between 0-1 kHz). The second contribution

26


is that the slow convergence problem of the adaptive controller is solved by incorporating the multi-variablestate space based inverse outer factor model in the control scheme. The latter is obtained from an inner/outerfactorization of the state space model of the transfer path between the actuator inputs and error sensor out-puts.

16:45 Electrical Coupling between Sensors and Actuators integrated in a Micro ActiveSuspension (ID 105)Thierry Verdot, Yann Meyer, Manuel Collet, FEMTO-ST, FranceJacek Baborowski, Paul Muralt, Ecole Polytechnique Federale de Lausanne, Switzerland

Electronic components bonded on a moving structure endure some intense dynamical solicitations due torigid body movements and vibrations of the structure at fixation points. Vibrations have some disturbingeffects on the frequential stability and the use-life of sensitive components. Consequently, it becomes crucialto protect them from the vibrating environment. The MEMS device we propose is a micro active suspension.It can be hybridized directly on electronic cards to isolate sensitive components in situ. So energy necessaryfor active control can be provided by a surrounding electronic source. We expect a low consumption levelbecause of the small weight of components to isolate. The needed actuators and sensors are included in themicro suspension from a piezoelectric layer (PZT) spread on a common reference electrode. We made anexperimental setup to validate the active functions. The results we have obtained show anomalies when weuse simultaneously actuators and sensors. We suspect that electrical coupling between the transducers occursthrough the common reference electrode. A complete model has been developed by adding to a mechanicalmodel the Kirchhoff’s current law occuring at reference electrode level. The computed results are quitesimilar to experimental ones.

17:10 Acoustic excitation of mechatronic systems by diffuse acoustic sound fields; Numer-ical predictions and measurements. (ID 181)Bert Roozen, Boudewijn Verhaar, Philips Applied Technologies, The NetherlandsMichiel Vervoordeldonk, Delft University of Technology / Philips Applied Technologies, TheNetherlands

With the accuracy of metrology frame applications entering the nanometer-range, the necessity arises totackle all types of disturbances. In the process of estimating the relative importance of the different types ofdisturbances on the machine accuracy, also called dynamic error budgeting, acoustic excitation has gainedmore importance, mainly because of the relatively high noise levels of air-conditioning systems in cleanrooms (“fabs”). To estimate the dynamic error budget claimed by acoustic excitation already in the designphase of the machine, there is a clear need to predict this type of excitation beforehand. This paper presentsa numerical procedure to predict the errors due to acoustic excitation efficiently.

To validate the numerical procedure, tests were done on a relatively simple structure consisting of a cantileverbeam and a base plate that represents the typical structural dynamics of a practical highprecision mechatronicsystem. The sensitivity of this structure to diffuse acoustic sound field excitation was predicted and measured.A diffuse field excitation was chosen so as to represent the acoustic field present in clean rooms (“fabs”) dueto air-conditioning systems. The numerical simulations were performed using a coupled structural-acousticsimulation, where we developed a novel computational procedure to determine the structural response dueto diffuse acoustic field. The simulations are compared with measurements in a reverberant room.

17:35 Active Structural Damping in high-precision equipment (ID 264)Michiel Vervoordeldonk, Jan van Eijk, Delft University of Technology / Philips Applied Technolo-gies, The NetherlandsBert Roozen, Philips Applied Technologies, The Netherlands

This paper discusses a novel approach to actively dampen structural modes of high-precision equipment,based on the presence of active vibration isolators. For both the evaluation of the achievable damping and

27


the controller tuning, an alternative analysis method is introduced. Initial experimental results - based on a3D scale model of an active isolated platform - are presented.

18:00 The use of dynamical recurrent neural network for identification and active vibra-tion control of a laminated thin plate (ID 521)Solmaz Kouhi, Javad Poshtan, Iran University of Science and Technology, Iran (Islamic Republicof)

In this paper the use of neural network for identification and active vibration control of a thin plate is investi-gated.Five piezoelectic actuators are employed to suppress the vibrations of the plate based on the commandvoltages provided by neural controller.A feedforwad neural network identifier has been used to train the plantmodel ”off-line”.Next the plant model is used by the dynamical recurrent neural network controller to de-termine the future performance.The performance of the hybrid controller tested by simulation of the simplysupported thin plate subject to impulse disturbance.To study the robustness performance of the neural con-troller the effects of structural parameter variation is also examined. The result shows the excellent robusnessof the neurocontroller.

Maduse – MAD1Room 8 – Chairman: D. Vandepitte

13:00 A Wave-Based Substructuring Approach for Fast Modification Predictions and In-dustrial Vehicle Optimization (ID 660)Stijn Donders, Rabah Hadjit, Luc Hermans, Marc Brughmans, LMS International, BelgiumWim Desmet, Katholieke Universiteit Leuven, Belgium

In the vehicle development process, design decisions are increasingly based on virtual prototypes (for exam-ple using Finite Element (FE) models), as time-to-market must be reduced, while more and more variantsmust be designed and assessed. Substructuring and Component Mode Synthesis (CMS) methods are usefulto speed up the time of local design modifications. Traditional CMS methods involve coupling the systemmatrices of each substructure along all interface degrees of freedom (DOFs). A Wave-Based Substructuring(WBS) approach has been developed, that aims to find a set of basis functions that describe the dynamicbehavior of the coupling interfaces. The interface displacements in the assembled system are then writtenas a linear combination of these basis functions. As the number of basis functions is typically much lowerthan the number of interface DOFs, the procedure reduces the number of variables in the matrix equationsand the size of the interface description. This greatly facilitates the model reduction procedure and results infaster structural and vibro-acoustic predictions. The WBS approach has been implemented in an integratedmulti-attribute virtual simulation environment, where it complements the assembly definition and trimmingfunctionality as it allows speeding up modification predictions and thus alleviates the computational burdenof industrial vehicle optimization. This is demonstrated on the basis of three test cases taken from automotiveindustry.

13:25 An approach to sensitivities-based prediction methods within uncertainty modelingof the dynamic behavior of structures. (ID 627)M. A. Schoen, Centro Ricerche FIAT, Italy

For the dynamic behavior modeling of complicated structures, such as a full-scale automotive model, predic-tive techniques that employ several deterministic models in order to consider for variations in robust designare, in general, costly by industrial standards. For example, the computational expense in designing forthe robustness of an automotive component scales exponentially with the number of uncertainty parametersconsidered in manufacturing. Moreover, it is ideally desired to develop a behavior profile based upon anindefinite number of uncertainty parameters. Thus, this work looks at eliminating the need for deterministic

28


modeling by utilizing Eigenvalue and Eigenvector sensitivities, calculated by Nastran design optimization so-lution SOL 200, for predicting dynamic responses to variations in specific uncertainty parameters. The studywill look at uncertainty variables for physical dimension for a simplified spot-welded automotive component.Sensitivity-based behavioral predictions will be presented for a full range of input values and compared withthe FE modeling of respective deterministic models.

13:50 Fuzzy Uncertainty Analysis in Automotive Crash Simulation (ID 587)Simone Turrin, Michael Hanss, Lothar Gaul, Universitat Stuttgart, Germany

Automotive crash simulation is a topic characterized by a high level of complexity. It deals with nonlin-ear dynamics, sophisticated models of materials and geometry, models of human occupants, as well as anearly unmanageable of possible boundary and initial conditions. In this complex field, uncertainty manage-ment plays an essential role. The lack of information to adequately determine the pertinence of the modelassumptions, the inevitable geometrical and material differences between nominally identical cars and thehuge number of possible boundary and initial conditions give rise to uncertainties of different origin. Theseuncertainties are reflected in model parameters of uncertain value, which have to be handled efficiently toobtain a validation of the overall crash simulation. A very practical approach to the modeling and simulationof uncertain systems is the numerical implementation of the uncertain parameters as fuzzy numbers, a specialclass of fuzzy sets, and then, the use of fuzzy arithmetic based on the transformation method.

14:15 Modelling of a vehicle windshield with realistic uncertainty (ID 574)Roberto d’Ippolito, LMS International, BelgiumUmut Tabak, Maarten De Munck, Katholieke Universiteit Leuven, BelgiumStijn Donders, LMS International, BelgiumDavid Moens, Dirk Vandepitte, Katholieke Universiteit Leuven, Belgium

In automotive industry, design decisions are increasingly based on virtual prototypes, such as numericalFinite Element (FE) models. This reduces the need for expensive physical prototypes, so that design costand time-to-market can be reduced. These FE models are deterministic, implicitly assuming that all designparameters are precisely known and that the manufacturing process produces identical structures. This istypically not valid. Uncertainty exists in the manufacturing process, in the physical and geometrical proper-ties and in the environmental conditions. A single deterministic analysis is not sufcient to fully understandtheir effect on the structural and vibro-acoustic performance. Including possibilistic models of design vari-ables into the mechanical simulation process can lead to an improved design process so as to guarantee therobustness of the design with a fuzzy approach. In this paper, the benets of this approach are demonstratedon an industrial windshield structure. Measurements of the dynamic response of the windshield have beencollected in free-free conditions for nominally identical windshields at different temperatures. An FE modelhas been validated in the frequency range of interest. Uncertainty in its physical and geometrical parametershas been obtained from test data and has been introduced in the numerical model in order to assess the ef-fects on the dynamic response. The outcome of the analysis consists of a possibilistic model of the frequencyresponse function, given the the uncertainty in the input parameters. This analysis not only provides a bet-ter insight into the effects of uncertainty in the dynamic response prediction, but also provides sensitivitymeasures of the design parameters on the nal performance of the structure. In addition, guidelines to furtherimprove structural designs and manufacturing processes are also obtained.

14:40 Possibilities and limitations of a reanalysis-based procedure for structural dynamicanalysis of uncertain systems (ID 501)Luigi Martini, Etienne Arnoult, Universite de Technologie de Compiegne, FrancePascal Lardeur, Renault / Universite de Technologie de Compiegne, France

In modern structural analysis, taking into account the variability of material and geometrical parameters isbecoming of higher and higher interest. This led engineering researchers to develop and improve stochasticmethods especially in Finite Element Analysis. Unfortunately, most of them show limitations concerning

29


the computational time or the uncertainty level of random variables. These are unacceptable inconveniencesfor industrial problems. This article sets the basis for a structural dynamic method which purpose is thestatistical moments computation of a dynamic output parameter (in this case the eigenfrequencies). It willexploit the well-known property of weak sensitivity of modal shapes to uncertainties in comparison with aconsiderable variation of eigenfrequencies. Two numerical examples are shown and the results are comparedwith a classic Monte Carlo Simulation technique.

15:05 Analysis of structures with uncertain parameters using component mode synthesis(ID 324)Lars Hinke, Brian R. Mace, Tim Waters, University of Southampton, United Kingdom

Component mode synthesis (CMS) is a well established method for the analysis of built-up structures. Un-certainty can be introduced at component level, in physical or modal coordinates. The fixed-interface (Craig-Bampton) method is preferable for the numerical analysis of models with uncertain properties, because ofthe special structure of the global mass and stiffness matrices. This paper concerns the propagation of uncer-tainties from component physical to component modal coordinates and subsequently to global coordinates.Several assumptions and approximations to reduce the numerical cost are discussed. These include neglect-ing uncertainties in modal properties, neglecting correlations and the use of perturbation methods. A possiblecombination of probabilistic and possibilistic approaches is proposed. Variations in global eigenfrequenciesand the frequency response are calculated for a numerical example.

Uncertainties in structural dynamics and acoustics – UNC1Room 8 – Chairman: G. Lombaert

15:55 Inverse problem for the identification of Chaos representations of random fieldsusing experimental vibrational tests (ID 340)Christophe Desceliers, Christian Soize, University of Marne-La-Vallee, FranceRoger Ghanem, University of Southern California, United States of America

This paper deals with the experimental identification of the probabilistic representation of a random fieldmodeling the Young modulus of a non homogeneous isotropic elastic medium by experimental vibrationtests. The random field representation is based on the polynomial chaos decomposition. The coefficients ofthe polynomial chaos are identified setting an inverse problem and then in solving an optimization problemrelated to the maximum likelihood principle.

16:20 Robust design optimization for uncertain complex dynamical systems (ID 59)Evangeline Capiez-Lernout, Christian Soize, University of Marne-La-Vallee, France

This paper deals with the design optimization problem of a structural-acoustic system in presence of un-certainties. The uncertain vibroacoustic numerical model is constructed by using a recent nonparametricprobabilistic model which takes into account model uncertainties and data uncertainties. The formulation ofthe design optimization problem includes the effect of uncertainties and consists in minimizing a cost func-tion with respect to an admissible set of design parameters. The numerical application consists in designingan uncertain master structure in order to minimize the acoustic pressure in a coupled internal cavity whichis assumed to be deterministic and excited by an acoustic source. The results of the design optimizationproblem, solved with and without the uncertain numerical model show significant differences.

30


16:45 A hierarchical approach to study the intra and inter variability of structure bornenoise in vehicles (ID 118)Cecile Lionnet, Renault, Technocentre, FrancePascal Lardeur, Renault / Universite de Technologie de Compiegne, FranceFabien Vieuille, Universite de Technologie de Compiegne, France

This paper presents a hierarchical approach for the analysis of the variability of structure borne noise invehicles. A terminology is introduced to clearly distinguish between intra variability and inter variabil-ity respectively corresponding to a lack of robustness toward environmental conditions and manufacturingprocess. Experimental results are presented to illustrate the contribution of the engine, the engine mountingsystem and the body to booming noise intra variability. Tests were also performed to assess booming noiseinter variability: it turns out that its level is slightly lower than the intra variability one.

17:10 An automated response surface based optimisation technique for the calculation offuzzy envelope FRFs of models with uncertain properties (ID 299)Maarten De Munck, David Moens, Wim Desmet, Dirk Vandepitte, Katholieke Universiteit Leu-ven, Belgium

The finite element method is a useful tool to predict the behaviour of a structure under static and dynamicloads. Reliable finite element analyses can reduce the need for prototype testing and thus reduce the designvalidation cost and time. In many real life situations however, a deterministic analysis is not sufficientto assess the quality of a design. In a design stage, some physical properties of the model may not bedetermined yet. But even in a design ready for production, design tolerances and production inaccuraciesintroduce variability and uncertainty. In these cases, a non-deterministic analysis procedure is required,either using a probabilistic or a possibilistic approach. In the former case, Monte Carlo simulation is bestknown. In the latter case, interval arithmetic and global optimisation can be used. Unless precautions aretaken, the conservatism of interval arithmetic approaches and the computational cost of global optimisationapproaches are prohibitively high for practical applications.

The authors developed a hybrid (global optimisation and interval arithmetic) interval finite element procedureto predict the bounds on frequency response functions (FRFs) of problems with interval inputs. In a first step,the bounds on the modal parameters are determined using a global optimisation approach. In a second step,the bounds on the FRF are calculated using an interval arithmetic approach. This hybrid approach reduces theconservatism compared to a full interval arithmetic approach and reduces the computational cost comparedto a full global optimisation approach.

Still, the optimisation of the modal parameters is by far computationally the most expensive step of thehybrid algorithm. Therefore, highly efficient optimisation algorithms are necessary to perform analyses onindustrial sized applications. Response surface based optimisation algorithms take advantage of the fact that- using a standard modal FE solver - the computational cost to calculate all modal parameters for all modes ofinterest is almost equal to the computational cost to calculate only one modal parameter for only one mode.This paper discusses the application of an automated response surface optimisation algorithm in the contextof interval and fuzzy finite element analysis.

17:35 The fuzzy FE approach to assess the uncertain static response of an industrial vehi-cle (ID 537)Laszlo Farkas, David Moens, Katholieke Universiteit Leuven, BelgiumStijn Donders, LMS International, BelgiumDirk Vandepitte, Wim Desmet, Katholieke Universiteit Leuven, Belgium

In a virtual prototyping context the importance of physical uncertainty modelling has become more evident.The increased computational performance over the last years enlarges the potential in the use of large models(more DOF’s) and in the application of more advanced numerical methods. This results in more accurate

31


models from the numerical point of view. Realistic numerical modelling however, besides flawless numericalmodelling, requires high-fidelity from the physical point of view. The method of Fuzzy FEM is appropriateto deal with the different physical parameter uncertainties as dimensional tolerances, scatter in materialproperties, structural design parameters. Interval analysis applied with the -cut strategy is the basis for afuzzy analysis. Main concern in the proposed interval analysis methods is to obtain conservative intervalresults using reasonable computational time. The fuzzy framework is applicable in different mechanicaldisciplines as e.g. dynamic analysis, static analysis. In the fuzzy analysis, the focus is on the investigationof the influence of different uncertainty model parameters on important performance measures. This paperpresents the application of the fuzzy principle on the static analysis of an industrial sized finite elementmodel. The problem stated for the fuzzy frame is of black-box type, with inputs the uncertain parameters,and outputs the displacements and stresses. The proposed problem is used to compare and discuss differentmethods for interval analysis. On the one hand the classical implementations are considered: the vertexmethod and the global optimisation approach. On the other hand, two newly proposed techniques are used:the reduced optimisation, and the reduced response surface method. Furthermore the use of the fuzzy analysistechnique is demontstrated as a large-scale design sensitivity tool.

18:00 Fuzzy Finite Element Method: An Experimental Modal Analysis Comparison (ID179)Thierry Tison, Karine Ruffin, Franck Massa, Bertrand Lallemand, Universite de Valenciennes etdu Hainaut Cambresis, France

This paper presents a comparison of numerical and experimental modal solutions in which geometrical andmaterial imperfections have been taken into account. First, several experimental modal analyses were com-pleted for specific design variable values and the results were aggregated to fuzzy numbers in order to con-stitute a fuzzy experimental reference. Second, a fuzzy finite-element model was built and the imperfectionswere propagated on the eigensolutions using a dedicated method. Finally, the fuzzy numerical quantitieswere compared with the experimental quantities to highlight the efficiency of non-deterministic models forpredicting the test structure’s behavioural modifications.

Poster session – POS1ARoom 9

13:00 Application of strain integral damping to engine mounts (ID 3)Ugo Tornar, PSA Peugeot Citroen, France

Stop and Start systems are going to play a more and more important role in decreasing fuel consumptionof urban vehicles in the years to come. Vehicle jerking on engine start become frequent and often uncon-trolled. Mathematical modeling is a powerful tool in studying such dynamic phenomena and powertrainmount damping is a first order issue. ”Strain Integral Damping” proved to be an effective damping formu-lation that is easy to implement in any time step integration code. This paper explains how such a dampingwas implemented in our start up model and the advantages it provides . The use of Strain Integral Damp-ing is extended to non-linear stiffness; a specific modeling is suggested in order to save modeling time andcomputational resources, when modeling complex systems such as a rotating machine

13:00 Vertical vibration control of multi-track High-Speed railway bridges under resonantconditions retrofitted with Fluid Viscous Dampers (ID 147)Maria D. Martinez-Rodrigo, University Jaume I, SpainEmma Moliner Cabedo, Pedro Museros, University of Granada, Spain

The reduction of inadmissible vertical vibrations in railway bridges under resonant conditions due to thecirculation of High Speed trains by means of Fluid Viscous Dampers is the main topic of this research. An

32


orthotropic finite element model is used to simulate the deck dynamic behaviour and modes other than theflexural plane ones which contribution may not be neglected in multi-track and skewed bridges are accountedfor. The aim of the investigation is to prove that the resonant response of the bridge may be drasticallyreduced with this type of devices and to provide some insight regarding the selection of the optimal retrofitdesign. First, the optimal parameters for the dampers which minimise the bridge deck dynamic responseassociated to a certain mode are obtained in closed form under harmonic excitation. Afterwards the adequacyof these optimal expressions to real bridges subjected to railway traffic is demonstrated over a wide range ofcirculating velocities. Finally the bare and optimally retrofitted bridge performances are compared and theeffectiveness of the proposed solution is shown.

13:00 Experimental Study on Vibration Reduction System with Granular Materials (ID242)Taichi Sato, Makoto Tanishima, Atsushi Ban, Tokyo Denki University, JapanKihachiro Tanaka, Saitama University, Japan

We investigate the reduction of vibration through the use of a granular material in a vibration system witha single degree of freedom excited in forced vibration by foundation motion. To clarify the mechanismof damping by the granular material, we propose a new concept of “equivalent added mass” in granularmaterial. We also produce a vibration model in which the granular material container also serves as themass part, and we use a hydraulic shaker to excite this vibration model to forced acceleration of the basepart. We measure the damping characteristics of the granular material while varying the magnitude of thefoundation acceleration. We also investigate the effects of varying the mass and diameter of the granularmaterial. We then discuss the relationship between the magnitude of the “equivalent added mass” and thedamping characteristics as determined from our experimental results.

13:00 Acceptance test for damped railway wheels (ID 401)Andrea Bracciali, Universita degli Studi di Firenze, ItalySteven Cervello, Lucchini Sidermeccanica S.p.A., Italy

In this paper a procedure for quality acceptance of low-noise railway wheels is presented. The low-noisefeature is obtained by applying a constrained layer damping through the use of a viscoelastic polymer and apre-formed constraining plate.

The proposed procedure is based on the analysis of the vibration amplitude in resonance conditions anddoes not use any acoustical measurement, whose use is particularly not advised in industrial workshops.Frequencies to be considered have been chosen with both the analysis of the sound power emitted by thewheel in a semi-anechoic chamber and the analysis of all acoustic frequency responses measured duringtype tests.

The methodology, verified on a reduced set of damped wheels, has shown a sufficient sensibility and provedto be potentially able to highlight the changes in the vibroacoustics characteristics of wheels due to theintrinsic repeatability of the application of the damping panel.

13:00 Thermoelastic Damping in MEMS Resonators - The role of Thermal and StructuralModes (ID 516)N. Jaroesawat, Stewart McWilliam, C.H.J. Fox, University of Nottingham, United Kingdom

The classic work of Zener for predicting thermoelastic damping is largely restricted to simple beam-likestructures. For more complex structures it is necessary to use the Finite Element Method (FEM) to predictthe damping, based on the fundamental equations of thermoelasticity. The main disadvantages of the FEMare that: i) solving the coupled thermoelastic problem can be numerically expensive; and ii) it does notprovide much physical insight into the problem. This paper considers an alternative method based on a modalapproach in which mechanical deformation and temperature are expressed using conventional structural andthermal modes. The damping is found to be dependent on the coupling between the thermal and structural

33


modes, which in turn are dependent on both the spatial and frequency considerations. A simple solid beamis used to illustrate the concept. The damping prediction shows reasonable agreement with results obtainedusing the fully-coupled FEM.

13:00 On the vibro-acoustic behaviour of high-speed train wheels (ID 90)Alfredo Cigada, Stefano Manzoni, Marcello Vanali, Politecnico di Milano, Italy

Since many years the problem of noise emission from railway wheels has been deeply studied and analyzed.Physical phenomena, linked to this noise emission, are well known and many solutions, aimed at reducingthe global disturbances, have been proposed. Moreover the continuous development of the technologies usedin the railway field asks constantly for new solutions.

The aim of this paper is to analyze the vibro-acoustic behaviour of a high-speed train wheelset performingboth experimental activities and numerical simulations; particularly an ETR 500 wheel has been chosen.The starting point for the experimental study is the analysis of the suspended free wheel, which can, then,be exploited to understand the wheel noise emission under operational conditions. This study is part of theresearches about noise emission of train wheels carried out in the European Project called InMAR.

13:00 Determination of Operational Modal Parameters and its applications in EngineNVH Development (ID 381)Qiu Luo, Laurent Polac, Renault SAS, France

The modal vibration behaviour of a powertrain in real operational conditions on bench or on vehicle maydiffer so significantly from that obtained during well-controlled laboratory conditions, suspended or clamped,by using hammer or shaker excitation. From that arises the need to identify powertrain’s modal behaviour inreal operational condition. However, the domination over structural response of engine’s orders componentin the response measurement may disrupt extraction of powertrain’s structural modal parameters. The paperwill present the testing and analysis procedure of Operational Modal Analysis (OMA) in our powertrainNVH development. The results obtained during the development of the new Renault 2.0 diesel engine showthat OMA, combined with other operational data analysis, such as Operational Deflection Shape Analysis(ODSA) and Coherence Function Analysis (CFA), can help improving the efficiency of powertrain’s FEmodel updating, of critical vibration phenomena diagnostic or other NVH development activities.

13:00 A new physical approach to generate a seismic excitation (ID 178)Sylvain Lignon, Louis Jezequel, Ecole Centrale de Lyon, France

Predicting earthquakes is currently impossible. A great deal of research is conducted on the modelizationof earthquake faults to understand the origin of earthquakes and then try to predict seismic occurrence, butdetailed models of seismic source mechanisms are complex to establish. In this paper, we focus on thebehavior of a subduction area: two tectonic plates are sliding on each other. Models are constructed byassuming that the tectonic plates behave as homogeneous elastic bodies. The fault surface is then analyzedas a friction interface between two elastic bodies.

Most of the models currently used to represent an earthquake fault are based on the simple Burridge-Knopoffmodel. This model consists of an elastically coupled chain of masses in contact with a moving rough surface.It is currently used because of its simplicity to represent the stick-slip phenomenon, which is often said tobe the source of earthquakes: each block is sticking until the shear force is high enough and allows it to slip.Some more elaborated finite element models have been developed, but based solely on the stick-slip effect.A new kind of friction instability is intoduced here by developing a new model.

34


13:00 Soil induced vibrations due to high-speed trains considering dynamic interactionand ballast effects (ID 282)Pedro Galvın, Jose Domınguez, Universidad de Sevilla, Spain

Numerical models to obtain ground vibrations induced by the passage of high-speed trains, based on finiteelement method and boundary element method, have been developed in the last years. A few models havebeen validated, comparing the numerical results and those obtained from experimental measurements. Mostof these models contain important simplifications, i.e; the ground properties and the surface geometry aroundthe track is not considered or close structures can not be taken into account. Other models only produce thevertical displacements of the ground. In this paper, a three-dimensional boundary element model basedon time domain formulation to solve wave propagation problems is presented. The soil is assumed to be auniform half-space with viscoelastic properties. The geometry effects can be considered easily. The existenceof ballast with its actual geometry on the soil surface is taken into account. From the proposed numericalmodel, ground vibrations induced by the passage of high-speed trains have been obtained. The influenceof the train speed has been studied. The numerical results obtained from the proposed model have beencompared with experimental results published in the literature. The agreement between both sets of resultsis quite good.

13:00 Seismic damage detection in existing buildings by finite element model updating (ID350)Pelin Gundes Bakir, Edwin Reynders, Guido De Roeck, Geert Degrande, Katholieke UniversiteitLeuven, Belgium

Earthquake events of moderate size may induce damage in structures in urban areas. For buildings designedaccording to the capacity based philosophy, the damage resulting from an earthquake is distributed through-out the structure, in regions specifically designed to dissipate energy (plastic hinges). After occurrence of theearthquake, an ambient vibration test will reveal altered modal properties of the building, from which damagecan be identified by inverse modeling techniques. The aim of this study is to develop vibration based damageassessment of buildings after a moderate earthquake for which the occurred damage is hardly detectable byvisual inspection. A typical building from the region north of the Sea of Marmara in Turkey is selected andused subsequently for a numerical simulation. The building is analyzed with a finite element model and thedamaged regions are selected as the elements adjacent to the beam-column joints. The stiffness values ofthese elements are decreased to simulate damage. The damage parameters are updated iteratively by mini-mizing the differences between the modal properties from the FE-model and those identified on the damagedbuilding model. This paper deals with the sensitivity-based FE model updating method and its applicationto damage detection. The extent of the decrease in the stiffnesses of the elements will be a good indicationfor the severity and the location of the damage. The same procedure is repeated in the presence of noise inorder to investigate the sensitivity of the updating procedure to the presence of noise.

13:00 Measurement of human-structure interaction in vertical and lateral directions: astanding body (ID 319)Ernesto Duarte, Tianjian Ji, University of Manchester, United Kingdom

This paper experimentally identifies human whole-body models in structural vibration for a standing bodyin vertical and lateral directions. A simple single degree-of-freedom system test rig is designed for theidentification purpose. Seventy individuals attend the tests by standing on the test rig. An impact is appliedon the test rig in the vertical and lateral directions when it is empty and when an individual stands on therig. Both accelerations and displacements are recorded and the corresponding frequency spectra show theresonance frequencies of the bar rig and the rig with a standing person. It is observed consistently from themeasurements that the human occupied test rig has a higher resonance frequency than that of the bare testrig in the vertical direction and that two resonance frequencies are measured from the occupied rig in thelateral directions. These observations clearly indicate that a standing body acts as at least a single degreeof-freedom system in both vertical and lateral directions.

35


13:00 The Finite Element Modeling of Engine Block and Response Analysis by Piston Slap(ID 262)Nobutaka Tsujiuchi, Takayuki Koizumi, Shinya Uemura, Doshisha University, Japan

This paper presents an analytical model for the prediction of piston secondary motion and the vibration due topiston slap. For the modeling of piston slap phenomenon, cylinder liner is modeled as a several spring-masssystem that are connected by modal characteristics, and lubricant film between the piston and the cylinder ismodeled as reaction force vectors which excite resonant mode of them. By comparing experimental resultsand analytical ones, the validity of the proposed model has been confirmed. Furthermore, it is appliedthis analysis to evaluate the vibration of transverse-mounted 4-cylinder engine, and is revealed the transientresponse to cylinder wall by piston slap.

Poster session – POS1BRoom 10

13:00 Optimization of a complex structure using a modal approach (ID 175)Sebastien Besset, Louis Jezequel, Ecole Centrale de Lyon, France

Optimization of complex structures often leads to high calculation costs. Indeed, the structure has to befrequently reanalysed in order to update the optimization criteriums. We propose an optimization methodbased on effective modal parameters. These parameters are close to the modal matrices used for the modalanalysis of a structure. Thus, once the structure has been analysed, their become is very easy. First, wewill explain the modal analysis that we will use in this paper. A modal model will be used to analyse thehollow parts of the structure. The modal analysis of the whole structure will be made using substructuringand “double modal synthesis” proposed by Jezequel. Secondly, we will explain the obtaining of effectivemodal parameters and their use for optimization. At least, we will show the efficiency of these parametersthrough the optimization of a complex structure.

13:00 Research on Several Envelope Analysis Methods (ID 61)Qin Shuren, Liu Xiaofeng, Bo Lin, Chongqing University, China

The aim of this paper is to summarize several popular envelope analysis approaches including the use ofHilbert transform, Teager energy operator and wavelet transform. We present the principles of these ap-proaches respectively, then focus on their enveloping ability or performance in detail. We show that theseapproaches yield their own problems which are illustrated by different simulation data analyzing. In addi-tion, Comparison between each approach is carried out by the enveloping analyzing of a signal artificiallygenerated.

13:00 Study on the sensitivity of gear stiffness to tooth crack length and location for diag-nostic applications (ID 62)Costantino Carmignani, Paola Forte, Gabriele Melani, University of Pisa, Italy

A numerical simulation of the dynamic behaviour of the system is surely a convenient means to perform pre-liminary testing of a diagnostic system, to support the interpretation of experimental results and to investigateon a wider range of operating conditions and defects. In this work the gear overall stiffness was evaluatedstatically as function of the gear angular position by means of FE models of the mating wheels and gearstructure, taking into account the deformation of the teeth under load and the presence of a tooth crack ofdifferent length and at different locations. The so calculated stiffness was then employed to simulate the geartransmission dynamic behaviour in steady conditions with a lumped non-linear model in a Matlab/Simulinkenvironment. Standard and non-standard diagnostic vibration indicators have been calculated for the syn-chronously averaged virtual acceleration signal of the progressively faulted gear in order to compare theireffectiveness in fault identification.

36


13:00 Vibration of a coated plate with absorbing material produced by an acoustic source.Theoretical-experimental correlation (ID 159)Xabier Sagartzazu, Juan Manuel Pagalday, Agustın Ugarte, Saioa Villodas, Ikerlan S. Coop.,SpainMaria Jesus Elejabarrieta, Escuela Politecnica Superior de Mondragon Unibertsitatea, SpainLuis Hervella, Universidade da Coruna, Spain

This paper deals with the calculation of vibrations in a coupled fluid-structure system that is subject to asource of sound excitation. The system consists of a thin plate with a coating of absorbing material coupledto a rigid air cavity containing a punctual acoustic source. A coupled fluid-structure model is used under wallimpedance conditions. The characterisation of the sound source and of the impedance of the coating materi-als used is also presented. The model has been implemented in Matlab. A correlation is presented betweenthe eigenvalues obtained with this model and the eigenvalues obtained using commercial programs like An-sys and Sysnoise for the coupled problem case but without a coating. Finally a comparison is presentedbetween the theoretical results and the experimental results.

13:00 Some considerations on fluid-structure interactions problems. (ID 215)Orlando Andrianarison, Roger Ohayon, I.S.A.T. Nevers, France

It is proposed in this paper to investigate formulations of fluid-structure interaction problems. One usu-ally looks at fluid-structure interaction in terms of “structural acoustics” or “hydro-elasticity” according tothe frequencies range. Structural acoustics concerns inviscid, homogeneous, compressible and weightless(without gravity effects) fluids in contact with a structure part whereas hydro-elasticity deals with inviscid,homogeneous, incompressible and heavy (gravity effects acting on its free surface) liquids.

We present in this paper formulations where both compressibility and gravity effects are taken into account.This coupling between compressibility and gravity effects can occur for instance in studies of launcherswhere cryogenic liquids are used as combustive. In this respect, local governing equations must be reviewedand we present in this paper new equations taking into account the aforementioned coupling. In the otherhand various variational formulations are also derived based on these new local equations and finally somereduced order models based on substructure techniques are constructed.

13:00 Autoregressive based diagnostics scheme for detection of bearing faults (ID 491)Suguna Thanagasundram, Fernando Soares Schlindwein, University of Leicester, United Kingdom

An investigation into the vibration characteristics of a ‘Roots and Claws’ based dry vacuum pump under dif-ferent operating conditions was conducted. An AutoRegressive (AR)-based condition monitoring algorithmwas developed and tested on both a fault-free and a pump with an implanted ceramic bearing with an innerrace defect at the High Vacuum (HV) end. The investigation provided some in-depth understanding of theeffects of different operating conditions such as speed and load on the vibration of the pump. The first keystep in the fault detection scheme was accurate determination of the running speed of the pump. It was ob-served that the rotating speed of the pump’s rotor shaft on which the bearing case was directly connected towas often less than the set speed of the pump due to rotor slip. The second step was envelope demodulationof the time domain vibration signals where the resonance excited by the faultinduced impacts was identifiedand the vibration signal were bandpass filtered around the resonant peak. The third step is spectral estimationusing parametric-based method of AR modelling. The advantage of the AR method is that it can work withsmaller sample sizes and sampling rates compared to the more traditional approach of FFT (Fast FourierTransform) and achieve far superior resolution capabilities. The analysis results showed that the effect ofactual speed was predominant in the detection of bearing faults as this was the speed that was used in thecalculations of the bearing defect frequencies and had to be determined very accurately. Initial results showthat the fault diagnostic scheme is very promising and the bearing fault could be accurately determined at allspeeds.

37


13:00 Practical Formulas for Oscillating Flow and Sound Traveling in Pipe (ID 495)Yoshichika Sato, Hiroshi Kanki, Kobe University, Japan

This paper shows practical solutions for compression wave and for steady state oscillating flow, pressuredriven in a pipe with a circular cross section. This paper also shows their applicable range and suitabilitythrough comparisons between the simplified solutions and strict analytical solutions, for the engineers anddesigners. The oscillating flow in a pipe has the boundary layer and the particle velocity profile due to thewall-viscosity. Analytical solutions for the profile are expressed by the hyper geometric functions, such asa Bessel function. Those strict but complicated solutions may become a stiff obstacle to treat much morecomplicated phenomena, such like the heat-fluid related oscillation problems. The practical formulas willcontribute to simplify such matters.

Pressure and flow velocity are directly treated in analytical solutions. Not only either of them but also flowrate may be chosen as an input and outputs by easy rearrangement of the formulas. This paper enables us toreduce computation time and analysis cost on not only duct acoustics but also water hammer and combustioninstabilities. Of course, our formulas may contribute to designing of micro devices such as MEMS stronglyinfluenced by wall-viscosity.

13:00 Coupling Factor in Vibro-Acoustic Problems and Sound Pressure Minimization byStructural Modification (ID 507)Kohei Furuya, Takuya Yoshimura, Tokyo Metropolitan University, JapanA. Suto, S. Narikuni, HONDA R&D Co., Japan

In this paper, the Vibro-Acousitc Coupling Factor (VAC) is introduced and a novel structural optimizationapproach using VAC is proposed to minimize the sound pressure in an enclosure. The presented approachcan give the information of the status of coupling between the structural vibration and the enclosure sound,which is useful to accomplish the global optimization. The proposed optimization is applied to the structuralmodification of a FE model, and the optimized structure is compared to that obtained by the conventionalapproach where the sound pressure sensitivity is utilized. It is shown that the globally optimized structure isobtained by the presented approach.

13:00 Predicting structurne-borne sound in enclosures with different techniques (ID 524)Attila Balazs Nagy, Budapest University of Technology and Economics, HungaryWim Desmet, Katholieke Universiteit Leuven, BelgiumFulop Augusztinovicz, Budapest University of Technology, HungaryBert Pluymers, Katholieke Universiteit Leuven, Belgium

This paper investigates how different techniques can be used to predict structure-borne sound in enclosures.As reference, the Boundary Element Method is taken and the performance of other methods (Wave BasedMethod, Statistical Energy Analysis, Rayleigh-based method) is compared to it. Computation efficiency,accuracy and the frequency range of applicability is discussed.

13:00 Seismicity induced by excavation activities: monitoring system problems (ID 235)Paolo Berry, Stefano Capponi, Anna Fabbri, University of Bologna, Italy

The interest in vibrations phenomena analysis to prevent the risk of damage on civil structures, caused byexcavation works for tunneling, quarrying and mining, has always been an issue of great actuality.

The above-mentioned risk is estimated by the peak particle velocity (PPV) and the frequency content, but,in several cases, it can be more convenient to measure the peak particle acceleration. As a consequence, theseismicity monitoring is a decisive facet of damage prevention, so it is necessary to use a proper calibratedvibrations measuring chain to get reliable and repeatable results.

This paper looks into preliminary results obtained by laboratory tests of a research program carried out tostudy the influence of different transducer mounting configurations on sensor responses both in timedomain

38


and frequency-domain; in particular the first experimental results show that the choice of seismometer cou-pling must be driven by the expected dominant frequency.

13:00 The application of spatial vision-based measurement for modal analysis (ID 556)Piotr Kohut, Piotr Kurowski, M. Szwedo, AGH - University of Science and Technology, Poland

In this paper the possible applications of vision techniques for modal analysis are presented. The follow-ing issues were investigated: a) automation geometry mapping and measurement points’ localization inthe preliminary phase of modal experiment; b) vibration measurement of selected elements of a construc-tion; c) executing a modal test. For this purpose procedures and algorithms were developed on the basisof vision-technique methods. Applying and developing passive 3-D vision techniques resulted in achieving3-D structure of the construction. The passive techniques, as opposed to the active ones, carry out depthmeasurements based on supplied image-sequences from one or more cameras. Stereovision algorithms weredeveloped and used, based on a single camera-image acquisition. The amplitude of vibrations was calcu-lated for selected construction points. Each point was represented by a spatial marker. The markers werecharacterized by symmetrical spatial structure. Methodology and algorithms based on model and symmetryproperties (some regularities) of measurement points were developed to compute the vibrations amplitude.Algorithms developed were implemented and tested in MATLAB programming environment. Necessaryvision data were received from high-speed digital camera “X-Stream Vision” in the form of “avi” files thatconstituted input data for developed algorithms.

39

Tuesday September 19, 2006

AMS1 - Multibody dynamics for multi-physics applications – AMS1Room 1 – Chairman: H. Van Brussel

9:00 Welcome to the AMS workshopHendrik Van Brussel, Katholieke Universiteit Leuven, Belgium

9:15 Multiphysics modeling of mechatronic multibody systems (ID 654)Paul Fisette, Universite Catholique de Louvain, BelgiumOlivier Bruls, Universite de Liege, BelgiumJan Swevers, Katholieke Universiteit Leuven, Belgium

Modeling mechatronic multibody systems requires the same type of methodology as for designing and pro-totyping mechatronic devices: a unified and integrated engineering approach. Various formulations arecurrently proposed to deal with multiphysics modeling, e.g. graph theories, equational approaches, cosim-ulation techniques. Recent works have pointed out their relative advantages and drawbacks, depending onthe application to deal with: model size, model complexity, degree of coupling, frequency range, etc. Thispaper is the result of a close collaboration between three Belgian laboratories, and aims at showing thatfor ”non-academic” mechatronic applications (i.e. issuing from real industrial issues), multibody dynamicsformulations can be generalized to mechatronic applications, for the model generation as well as for thenumerical analysis phases. Model portability being also an important aspect of the work, they must be easilyinterfaced with control design and optimization programs. A global ”demonstrator”, based on an industrialcase, is discussed: multiphysics modeling, control design and mathematical optimization are carried out toillustrate the consistency and the efficiency of the proposed approaches.

10:30 Concurrent simulation of mechatronic systems with variable mechanical configura-tion (ID 149)Maıra da Silva, Katholieke Universiteit Leuven, BelgiumOlivier Bruls, Universite de Liege, BelgiumBart Paijmans, Katholieke Universiteit Leuven / Flanders Mechatronics Technology Center, Bel-giumWim Desmet, Hendrik Van Brussel, Katholieke Universiteit Leuven, Belgium

The dynamic behavior of a mechatronic system may be largely dependent on the mechanical configura-tion. This inevitably affects the performance and the stability of any controller designed using the classicallinear control theory. Major improvements are expected if the controller is designed using the theory oflinear parameter varying systems, but one difficulty is then to provide a low-order model which captures theconfiguration-dependent dynamics. This paper presents a methodology to build such a model in two steps:(i) for several representative configurations, a local linear model is derived, (ii) the parameter-dependentmodel is constructed by interpolation in the configuration space. Moreover, a co-simulation methodologyis proposed for mechatronic systems by combining LMS Virtual.Lab Motion and Matlab/Simulink. Themethodology is applied to an industrial pick-and-place machine.

10:55 Optimal synthesis of planar mechanisms via an extensible-link approach (ID 655)Jean-Francois Collard, Universite Catholique de Louvain, BelgiumP. Duysinx, Universite de Liege, BelgiumPaul Fisette, Universite Catholique de Louvain, Belgium

This paper presents a novel approach to optimize the design of planar mechanisms with revolute joints forfunction-generation or path synthesis. The proposed method is based on the use of an extensible-link mech-anism model whose strain energy is minimized to find the optimal rigid design. This enables us to get ridof assembling constraints and the use of natural coordinates makes the objective function simpler. The opti-mization strategy is divided into two stages: the first one relies on multiple partial optimizations and provideshot starting point for the second stage which involves all the variables and all the energy contributions. The

40


question of finding the global optimum is reviewed. Instead, a simple algorithm is proposed to explore thedesign space and to find several local optima among which the designer may choose the best one taking othercriteria into account (e.g. stiffness, collision, size,...). Two applications are presented to illustrate the wholeprocess.

11:20 An extended flexible multibody dynamics software for mechatronic applications (ID656)Olivier Bruls, P. Duysinx, Jean-Claude Golinval, Universite de Liege, Belgium

This paper presents some extensions of flexible multibody formalisms for mechatronic applications. Thesoftware implementation is realized in a general purpose finite element code, and three specific subjects arediscussed: (i) nonlinear model reduction techniques for flexible multibody systems, (ii) integrated modellingof mechatronic systems, (iii) time-integration algorithms for coupled problems. The relevance of those de-velopments are demonstrated for the dynamic analysis and the control design of an experimental lightweightmanipulator.

AMS2 - Control and optimisation – AMS2Room 1 – Chairman: J. Swevers

14:00 Gain-scheduling control for mechatronic systems with position-dependent dynamics(ID 29)Bart Paijmans, Katholieke Universiteit Leuven / Flanders Mechatronics Technology Center, Bel-giumWim Symens, Flanders Mechatronics Technology Center, BelgiumHendrik Van Brussel, Jan Swevers, Katholieke Universiteit Leuven, Belgium

This paper presents a gain-scheduling-control technique for mechatronic systems with position-dependentdynamics. The proposed method fits in the framework of traditional gain scheduling, where several con-trollers designed for fixed operating points are interpolated to construct a global gain-scheduling controller.A new interpolation approach is proposed starting from an affine interpolation between the poles, zeros andgains of the local controllers as a function of the varying parameter, resulting in a polynomial state-spacerepresentation. The presented method is applied on an industrial pick-and-place machine which has position-dependent dynamics. Experimental results show the benefit of the proposed method.

14:25 A new semi-active method for the damping of a piezoelectric structure (ID 297)Bruno de Marneffe, Mihaita Horodinca, Andre Preumont, Universite Libre de Bruxelles, Belgium

This paper investigates the so-called ”negative capacitance” method for the damping of a piezoelectric trussstructure. This method mainly consists of shunting the electrodes of a piezoelectric actuator with an elec-tronic circuit simulating the behavior of a negative capacitance. First, different classical methods for thedamping of a truss structure are re-examined. They are compared with each other and key parameters arehighlighted. Then, the negative capacitance method is introduced using the same formulation. It is shownthat a piezoelectric transducer acting in parallel with a negative capacitance can be viewed as an equivalenttransducer with enhanced electromechanical coupling factor. Two different implementations are introduced;their stability is examined and conditions under which they can be implemented are highlighted. Finally,experimental results are presented.

41


14:50 Optimal Attenuation of Known Periodic Disturbances: a Convex Control DesignApproach (ID 511)Goele Pipeleers, Bram Demeulenaere, Joris De Schutter, Jan Swevers, Katholieke UniversiteitLeuven, Belgium

This paper discusses the design of optimal linear controllers for the attenuation of known periodic distur-bances in linear systems. A general framework is developed allowing for a variety of control configurations,robustness and performance specifications, and input constraints. The optimal control design is formulatedas a convex optimization problem. To robustify the framework for uncertainty on the disturbance period, twostrategies are proposed and compared to each other for a numerical example: (i) a worst-case strategy and(ii) a sensitivity-based approach.

15:15 Tracking Performances of Cascade and Sliding Mode Controllers with Applicationto a XY Milling Table (ID 634)Zamberi Jamaludin, Hendrik Van Brussel, Jan Swevers, Katholieke Universiteit Leuven, Belgium

Design and implementation of robust tracking controllers for linear feed drive high speed machine toolsare the primary objectives of this work. Cutting forces and friction forces limit tracking accuracy and theirinfluences are minimised through robust controller design. This paper discusses the design and trackingperformances of a traditional cascade and a sliding mode controller for a X-Y milling table. Circular testsare performed at selected tracking speeds and circle radii for tracking performance analysis. The trackingerror for cascade P/PI controller is proportional to the reference tracking speed and is inversely proportionalto the velocity gain value of the position loop controller. Speed and acceleration feedforward further reducethis tracking error. Quadrant glitches, a product of nonlinear friction at the points of velocity reversal,are observed. Sliding mode controller exhibits better tracking accuracy and dynamic stiffness and yields asignificant reduction of the quadrant glitches.

AMS3 - Applications – AMS3Room 1 – Chairman: D. Reynaerts

16:05 Design and control of a novel piezoelectric drive module for application in a multi-DOF positioning stage (ID 356)Wim Van de Vijver, Dominiek Reynaerts, Hendrik Van Brussel, Katholieke Universiteit Leuven,Belgium

Recent advances in measurement systems and machine tools require positioning systems that combine highstiffness with a high positioning accuracy. Moreover, a high positioning speed is favourable. Current solu-tions of multi-DOF positioning systems are often based on a stacked construction of linear systems consistingof an electromechanical drive, a transmission and a linear guiding system. Due to the series connection ofcomponents the resulting stiffness is diminished and the positioning errors are accumulated. However, byintegrating the bearing, transmission and drive functions into a multi-DOF positioning system, an accurateand at the same time rigid positioning system can be achieved. This paper presents the design of the piezo-electric drive modules which will be integrated into the proposed multi-DOF positioning system. Due to thedifferent operation modes of the module, a maximum speed of 0.3m/s can be achieved, while a positioningresolution of 10nm is shown. It is also shown that due to an alternative control strategy, an accurate controlof the drive modules is possible.

42


16:30 A dynamic vibration absorber for torsional oscillations of metro wheel sets (ID 478)Christophe Collette, Renaud Bastaits, Mihaita Horodinca, Universite Libre de Bruxelles, Belgium

In this paper, torsional oscillations of the wheel set are first studied with free boundaries conditions. Adynamic vibration absorber tuned to the first torsional resonance of the wheel set has been designed, andits numerical efficiency is compared with experimental results. The second part of the paper studies thesame phenomenon in rolling conditions. The experimental set-up is a quarter-scaled roller rig from the NewTechnologies Laboratory (INRETS-France). Conditions under which torsional vibrations of the wheel setare excited are discussed as well as differences from the first system. The efficiency of the dynamic vibrationabsorber is again evaluated experimentally and compared with outputs of the multi-body model of the rollerrig. The multi-body model includes a finite element representation of the wheel-set. From our results, railcorrugation is evaluated in terms of longitudinal and lateral frictional power dissipated in the contact patch.

16:55 Development of a numerical modelling methodology for the NVH behaviour of elas-tomeric line connections (ID 510)Andrea Stenti, David Moens, Paul Sas, Wim Desmet, Katholieke Universiteit Leuven, Belgium

The modelling of the NVH behaviour of elastomeric line connections, such as car door weather strip sealsand car windshield connections, involve the computational burden of including a detailed model of the jointinto the system model and of including variability and uncertainty, typical for elastomeric joints, in a fullscalesystem model.

In the general framework of developing engineering tools for virtual prototyping and product refinement,a methodology is being developed which allows simplified equivalent joint models to be readily includedin full-scale NVH models at a reasonable computational cost. The methodology is based on a three-levelmodelling approach involving a material, a component and a system level. This study focuses on the dynamicmodelling of car door weather strip seals and illustrates the methodology through a numerical case study.

17:20 Damping of spray boomstructures with non-linear dampers (ID 602)K. Engelen, Herman Ramon, Jan Anthonis, Katholieke Universiteit Leuven, Belgium

The potential of non-linear dampers for vibration suppression of spray boom structures is investigated bytime domain simulations. The power-law damper model is considered for this purpose. Experimental resultsare shown of the optimization of the dynamic behavior of a spray boom structure with automotive shockabsorbers.

17:45 Rotordynamic behaviour of a micro-turbine rotor on air bearings: modelling tech-niques and experimental verification (ID 571)Tobias Waumans, Peter Vleugels, Jan Peirs, Farid Al-Bender, Dominiek Reynaerts, KatholiekeUniversiteit Leuven, Belgium

Current trends in micro-turbomachinery stress the need for adequate rotordynamic models. These modelsshould allow accurate prediction of critical speeds, imbalance response and stable operation range of micro-turbomachinery rotor-bearing systems. This paper gives an overview of the total rotordynamic modellingprocess of a micro-turbine rotor supported on aerostatic bearings. A both accurate and efficient modellingtechnique is outlined to obtain static and dynamic air bearing properties. These bearing coefficients serveas input for a rotordynamic model yielding damped natural frequencies, unbalance response and stabilitylimits. Experimental verification confirms a good agreement with the predicted critical speeds.

43


Non-stationary sound and vibration analysis and applications – NSV1Room 2 – Chairman: S.D. Fassois

8:50 Parametric Time-Domain Methods for Non-Stationary Random Vibration Identi-cation and Analysis: An Overview and Comparison (ID 97)Aggelos Poulimenos A.G., M. Spiridonakos, Spilios Fassois, University of Patras, Greece

An overview and comparison of parametric time-domain methods for non-stationary random vibration mod-elling (identification) and analysis based upon a single vibration signal realization is presented. The consid-ered methods are based upon Time-dependent AutoRegressive Moving Average (TARMA) representations,and may be classified as unstructured parameter evolution, stochastic parameter evolution, and deterministicparameter evolution. The main methods within each class are presented, and model “structure” selection isdiscussed. The methods are compared, via a Monte Carlo study, in terms of achievable model parsimony,prediction accuracy, power spectral density and modal parameter accuracy and tracking, computational sim-plicity, and ease of use. The results confirm the increased accuracy and performance characteristics of thedeterministic, as well as stochastic, parameter evolution methods over those of their unstructured parameterevolution counterparts.

9:15 Cyclostationary modeling of reciprocating compressors and application to valvefault detection (ID 387)Rafik Zouari, Institut de Recherche en Informatique et Systemes Aleatoires, FranceJerome Antoni, Jean-Luc Ille, Universite de Technologie de Compiegne, FranceMichel Willaert, Burton Corblin, FranceMarcel Watremetz, Centre Technique des Industries Mecaniques, FranceMenad Sidahmed, Universite de Technologie de Compiegne, France

This paper presents a methodology for identifying events and valve faults in reciprocating compressors basedon cyclostationary modeling of their vibration. It consists firstly in angular resampling the time signals andthen in decomposing them into periodic and random parts. Secondly, the second-order cyclostationarity ofthe random part is exploited to construct an angle-frequency map of the vibration energy by means of theWigner-Ville spectrum based on cyclic averaging. From this energy representation, a correlation procedurebetween sensors is followed in order to identify the angular and spectral bands where coherent events occurin the compressor cycle. Additionally, it allows the extraction of pertinent and simple indicators for detectingfault signatures in valves; these can be used in condition monitoring.

9:40 Non stationary analysis of a railway bridge scaled model under operational condi-tions (ID 626)Luigi Garibaldi, Stefano Marchesiello, Politecnico di Torino, Italy

Among the various time and frequency domain models for identification purposes one may adopt to estimatethe dynamics of a bridge deck, lot of care has to be taken to use the right one; in fact, when we considera vehicle crossing a bridge, the response is often treated as a stationary random process, which describesan “almost” constant parameter dynamic system; for railway bridges, on the contrary, this statement doesnot hold because the train passage gives origin to parameter changes during the transit, which cannot beneglected.

This is due to the ratio of the train/deck masses, but is also caused by the superposition of different inputsdue to the roughness of the wheel-rail surface, and their constant spatial location over the system, while theiractivation is normally a function of the train dimension and speed.

Under this point of view, an interesting rig has been developed and many tests conducted to demonstrate thestrong parameters variation and the models to be adopted.

In these experiments, a simply supported bridge deck has been monitored during the deterministic transit

44


of a given weight crossing train, at constant speed, with many variable parameters. Looking at the simplestparameters, i.e. the eigenfrequencies of the beam, one can feel how important is their fluctuation and theconsequences on the dynamic identification. Some techniques, such as a time variable Stochastic approachand a input filtering would be suitable in this case.

10:05 Don’t ignore nonstationarity: use it to advantage (ID 635)Frederic Bonnardot, LASPI, FranceR. Roustany, Universite de Technologie de Compiegne, FranceAli Ibrahim, LASPI, FranceK. Sabri, LASPI / University of New South Wales, FranceJerome Antoni, Universite de Technologie de Compiegne, FranceM. El Badaoui, LASPI, France

Most signal processing techniques were originally designed to work in the stationary context. However, thesignals encountered in many fields are clearly non-stationary either by construction like man-made commu-nication signals, or due to the time-varying geometry of the transmission path as in mechanics. The propertyof non-stationarity seems at first glance inconvenient since it limits the effectiveness of the classical tech-niques, unless it is exploited to design more advanced techniques. This paper shows how to make use of thenon-stationary but cyclic (cyclostationary) nature of the signals issuing from rotating machinery. It providesthe pre-treatments to be performed in presence of speed fluctuation in order to resample the vibration signalswith respect to the shaft angle instead of time and thus accentuate the property of cyclostationarity. It thenillustrates through examples how to exploit the extra information provided by cyclostationarity to effectivelyperform signal separation and diagnostics of complex industrial systems.

Non-stationary sound and vibration analysis and applications – NSV2Room 2 – Chairman: S.D. Fassois

10:55 Time-frequency domain decomposition for modal identification using ambient exci-tation tests (ID 313)Thien-Phu Le, Patrick Paultre, University of Sherbrooke, Canada

Time-frequency technique has been applied in modal identification since the end of the 90s. The modal pa-rameters, i.e, frequencies, damping ratios and mode shapes are often extracted by the use of time-frequencytechniques based on free decay response. From ambient excitation tests, the free decay response can beobtained by the random decrement technique. Recently, Le and Paultre have proposed to use correlationfunction of recorded signal of ambient excitation tests to avoid the random decrement technique as a pre-processing step. They also showed that, using correlation function, the time-frequency technique based oncontinuous wavelet transform gives good results for modal identification and makes the computation moreefficient. In this study, the extension of this technique is proposed in order to give a consistent procedure.The redundant information is removed using singular value decomposition. Numerical examples and resultsobtained from tests performed on a 3-dimensional truss structure are used to validate the method.

11:20 Choi-Wiliams time frequency analysis of a three degree of freedom non-linear sys-tem (ID 322)Andreas Kyprianou, University of Cyprus, CyprusWieslaw Staszewski, Keith Worden, University of Sheffield, United Kingdom

The response of a non-linear mechanical oscillator, depending on the type of non-linearity, to an excitationof particular harmonic might contain sub- and super-harmonics, transients and chaotic components. Theconventional frequency response function, used to completely characterise the input-output behaviour of a

45


linear mechanical oscillator, not only does not capture the above characteristics but also, in case of broadbandexcitation, fails to attribute them to individual frequencies. In this paper we apply the Choi-Williams timefrequency distribution to simulated data of a single and three degree of freedom Duffing oscillator in orderto assess its input-output relationship.

11:45 Modal Parameters Identification Using Bilinear Joint Time-Frequency Analysis (ID257)Alireza Roshan-Ghias, Mohammad Mobed, Mohammad Bagher Shamsollahi, Mehdi Behzad,Sharif University of Technology, Iran (Islamic Republic of)

In this paper, a new method for modal parameter estimation using time-frequency representations is pre-sented. Smoothed Pseudo Wigner-Ville representation which is a member of the Cohen’s class distributionsis used that reduces cross-terms in time-frequency plane dramatically and holds resolution as well. Themethod was applied to systems with high damping, and the results were superior to other conventional meth-ods.

Damping – D2Room 2 – Chairman: J. Mottershead

16:30 Auxetic Materials and Structural Damping (ID 472)John House, M. Greaves, M. Swan, QinetiQ, United Kingdom

This paper describes part of a research programme that sought to provide a lightweight material that couldbe applied for structural damping applications. In particular this paper describes the findings of a study intoa range of novel visco-elastic, open-celled foam materials with either positive or negative Poisson’s ratio(i.e. auxetic). It was hypothesised that auxetic materials may offer additional strain energy dissipating mech-anisms, leading to more effective structural damping than that offered by the conventional constrained orfree-layer technologies that use materials with a positive Poisson’s ratio. The materials fabricated for thisstudy were able to provide significant damping when applied to thin steel plate and it was shown that thePoisson’s ratio can affect the levels of damping obtained. The structural damping advantages attributed tomaterials with both negative and positive Poisson’s ratio are discussed together with the strain energy dissi-pating mechanisms considered to be invoked. The study showed that substantial structural damping could beattributable to changes in thickness resonant effects controlled by the through-plane wave velocity and theapplied foam layer thickness. Peak (thickness effect) damping occurs at e/4 and odd multiples off e/4 (i.e.(2n+1)e/4) resonant frequencies and the Poisson’s ratio is shown to control the through-plane wave velocityin this case. That in turn modifies the thickness resonant frequency and the frequency at which peak damp-ing occurs but the actual level of peak damping remains the same. Thickness effect damping can be veryefficient over a narrow band of frequencies but any weight advantages over conventional damping materialswill depend very much on the frequency range for which damping is required. It is suggested that control ofthe Poisson’s ratio may be beneficial in certain circumstances in allowing similar damping performance butin a thinner layer.

16:55 Rheological and restoring force models regarding belt tensioner dymamic behavior:prediction & experiment. (ID 400)Jerome Bastien, Universite de Technologie de Belfort-Montbeliard, FranceGuilhem Michon, Valeo Electrical Systems, FranceLionel Manin, Regis Dufour, INSA de Lyon, France

The objective of this paper is to compare the Masing and modified Dahl model efficiency regarding theprediction of the hysteretic behavior of a belt tensioner used for automotive engines. A first experimentalstudy with deflection imposed on the tensioner is carried out to identify hysteresis loop parameters for the

46


two models. The models are then implemented in the general motion equations modeling the behavior of abelt - tensioner - mass system. The comparison beteen numerical and experimental results show that thesetwo models perform satisfactorily and that the modified Dahl model is a little more efficient.

17:20 Modeling of Assembled Combustion Engine Parts under Considertation of MicroSlip Effects in the Connection Flanges (ID 504)Carsten Schedlinski, ICS Engineering GmbH, GermanyAlexander Genzo, Bernard Laer, Volkswagen AG, GermanyLars Panning, University of Hannover, Germany

This paper addresses the modeling of assembled combustion engine parts. Here multiple flange connec-tions are found that contribute to the stiffness and damping characteristics. By example of a four cylindercombustion engine a finite element modeling technique will be introduced which focuses on a suitable andproper representation of the flange regions between crankcase and crankshaft main bearing cap. Especially,a numerical method will be presented and applied that accounts for the nonlinear damping and stiffness char-acteristics of the system. The numerical analysis results will be compared to experimental data in order toprove the effectiveness of the method.

17:45 Numerical Simulation of Rubber Devices Dynamics by Discrete Modelling (ID 84)Matteo Lancini, Andrea Magalini, David Vetturi, Universita degli Studi di Brescia, Italy

This contribution deals with approaches to the problem of rubber to metal devices dynamic behaviour char-acterization; which were undertaken by the authors. To take in account the peculiar properties of rubber,in terms of non-linearity, viscoelasticity as well as frequency, temperature and static strain dependence, anumerical analysis was developed, where materials constitutive equations modelling is operated by a neuralnetwork, working directly on numerical data supplied by experimentation. The proposed method avoids theimplementation of finite elements models, recurring to discrete models based on a multi-body, particle-basedapproach: several models of such a kind have been taken into account, from discrete concentrated parametersmodels to a chosen hybrid particle based model, where local stiffness is computed directly by the continuumconstitutive equations. Simulative results where then compared with experimental data.

18:10 Damping Model Uncertainty in Structural Dynamics (ID 486)Sondipon Adhikari, University of Bristol, United Kingdom

The characterization of uncertainty in the damping forces in a vibrating structure is of significant interestin structural dynamics. The most common approach is to use a viscous damping matrix with uncertaincoefficients. Viscous damping is not the only damping model within the scope of linear analysis. Any modelwhich makes the energy dissipation functional non-negative is a possible candidate for a valid dampingmodel. There can be many functional forms which can be used as damping models. In this paper twoapproaches are proposed to quantify such ‘model-form uncertainty’ associated with the use of the viscousdamping model (not only the model parameters). The first approach is based on an ensemble of equivalentdamping functions and the second approach is based on random matrix theory. The results obtained from thetwo methods are compared using numerical examples. It was observed that the random matrix theory can beused to quantify damping model uncertainty.

47


Active noise control – ANC2Room 3 – Chairman: J.R. Arruda

8:50 Experimental and numerical investigations of a smart acoustic box (ID 372)Friedemann Laugwitz, Jean Lefevre, Gerald Schmidt, Tamara Nestoroviæ, Ulrich Gabbert, Otto-von-Guericke University of Magdeburg, Germany

The objective of the paper is to demonstrate the potential of applying the smart structure technology foran adaptive reduction of vibration and noise, which are caused if thin-walled structures are excited. Thevibration and noise control can be managed by attaching the structure with distributed piezoelectric ceramicpatches acting as sensors as well as actuators. For the simulation and the design of such smart vibro-acousticsystems recently a numerical approach based on the finite element method has been developed, which in-cludes the main functional parts as well as the control algorithm. First, this approach is briefly presented.The second focus of the paper is on the experimental verification of the numerical approach. For this reasonan experimental setup has been developed, which consists of a wooden box inside lined with steel sheetsand covert with a smart aluminium plate. For vibration and noise reduction experiments, a velocity feedbackcontroller is used. The results of the numerical analysis are evaluated by the experimental results and a goodagreement is shown.

9:15 A Virtual Prototyping Approach to the Design of Smart Structures Applications (ID680)Herman Van der Auweraer, LMS International, BelgiumLeopoldo P.R. de Oliveira, Maıra da Silva, Katholieke Universiteit Leuven, BelgiumSven Herold, LBF, Fraunhofer Institute for Structural Durability, GermanyJan Mohring, ITWM, Fraunhofer-Institut fur Techno- und Wirtschaftsmathematik, GermanyArnaud Deraemaeker, Universite Libre de Bruxelles, Belgium

In order to bring the research results on intelligent materials to the level of industrial use in real applications,the related design processes have to become part of the complete product creation process. This requires thatthe product functional performance simulation models, which are the cornerstone of today’s design process,must be capable of supporting the specific aspects related to advanced materials, active systems, actuators,sensors and control and integrate these into system level virtual prototype models. More specifically, thisinvolves developing modeling capabilities for the intelligent material systems, sensor and actuator compo-nents, for the control systems as well as for their integration in systemlevel application designs. The finalresult will then be a multi-attribute optimization approach integrating noise and vibration performance withreliability, durability and cost aspects. It is clear that no single integrated solution will be able to fulfillall requirements of the various material and control approaches, therefore the focus of the research is onsupporting as much as possible the use, combination and extension of existing codes and tools.

9:40 A CAE modeling approach for the analysis of vibro-acoustic systems with distrib-uted ASAC control (ID 167)Leopoldo P.R. de Oliveira, Katholieke Universiteit Leuven, BelgiumArnaud Deraemaeker, Universite Libre de Bruxelles, BelgiumJan Mohring, ITWM, Fraunhofer-Institut fur Techno- und Wirtschaftsmathematik, GermanyHerman Van der Auweraer, LMS International, BelgiumPaul Sas, Wim Desmet, Katholieke Universiteit Leuven, Belgium

The aim of this paper is to present a CAE methodology for simulating coupled vibro-acoustic systems andinclude these models into a closed loop control simulation. This tool provides a reliable simulation procedurethat takes into account the vibro-acoustic phenomena as well as the control system since sensors/actuatorsmodels and the control algorithms can be included. The focus of the present paper is on the active structural-acoustic control, using piezoelectric patches as sensors and actuators. An initial structural uncoupled finite

48


element model is modified to account for the placement of piezo-patches. The modified structural modalbase is then used together with an uncoupled acoustic modal base to derive the fully coupled vibro-acousticfinite element model, which is reduced and formulated as a state-space model. The electro-mechanical cou-pling between the structure and the sensor/actuator pair is calculated and included in the state-space model.Eventually, a feedback controller is implemented. This smartstructure modeling approach is illustrated forthe case of the active reduction of the sound transmission of a firewall between the engine and the passengercompartments of a concrete car-like demonstrator.

10:05 Evaluation of the Performance Characteristics of the FxLMS and FxGAL Algo-rithms for ASAC applications (ID 578)Prasanth Vengala, Amanjot Dhaliwal, Jason Blough, Gordon Parker, Michigan Technological Uni-versity, United States of America

Active Structural Acoustic Control, ASAC, was applied to reduce the road noise on a front wheel drivepassenger car. ASAC was implemented to reduce the amplitude of a tire/suspension response at 215 Hz. Theperformance of the implementation was evaluated on a chassis dynamometer and shown to provide globalnoise cancellation with up to 15 dBA of attenuation. In addition to the vehicle implementation and as a resultof lessons learned, simulations were performed to understand the convergence characteristics of both theFxLMS and FxGAL algorithms. Algorithms were also developed with the ability to estimate the requiredsecondary path filters in an automated online fashion. The paper concludes with a brief summary of lessonslearned and recommendations for ASAC implementation on passenger vehicles.

Active vibration control and smart structures – AVC2Room 3 – Chairman: A. Preumont

10:55 Damping measurements on a carbon fibre reinforced laminate (ID 500)Diogo Montalvao, Escola Superior de Tecnologia de Setubal, PortugalAntonio Ribeiro, Nuno Maia, Instituto Superior Tecnico, PortugalJ. Duarte Silva, R.A. Claudio, Escola Superior de Tecnologia de Setubal, Portugal

By performing force-displacement tests in a universal testing machine it is possible to accurately measurethe damping of a material. By providing accurate estimates of the damping for a set of given conditions,it should be possible to determine which model is more adequate to describe damping: viscous, hystereticor a combination of both. On the other hand, by changing the amplitude it should be possible to assess thelinearity of the material behaviour. In this work, the authors present the experimental results of the dampingmeasurements on a specimen of a quasi-isotropic Carbon Fibre Reinforced Laminate, for different testingparameters, like frequency and amplitude. The setup is fully described and results are presented in detail,giving a special attention to the experimental problems, so that this work can become a useful contributionto further studies.

11:20 Active Vibration Control of Space Truss Structures: Power Analysis and EnergyDistribution (ID 582)Paulo Goncalves, Stephen Elliott, Michael J. Brennan, University of Southampton, United King-dom

This work concerns the analysis of power flow in truss structures, such as those used in space, when activecontrol is used to attenuate vibration levels at some point in the structure. A truss structure consisting of93 aluminum beams connected by 33 joints is studied. A model of the structure is obtained by the dynamicstiffness method using the exact solution for wave propagation in the beams; the model is also comparedto a finite element model of pin-jointed bars to illustrate the localized dynamic behaviour of the structuralmembers of the truss. Joint masses are used in the model, however dissipation is considered to occur only in

49


the beams. Reaction forces for individual beam ends can be calculated after solving the equation of motionfor the overall system. Knowing the forces and velocities on a beam end makes it possible to calculate thepower flow at that end. The sum of the coupling power at both beam ends gives the power dissipated by thatbeam. It is found that around 80% of power dissipated is by bending motion of the beams and only 20%of power is transmitted by bending. The principal mechanism of power flow is by longitudinal motion ofthe beams. Overall power flow and dissipated power is analyzed before and after multichannel feedforwardcontrol is implemented in order to help understand the physical mechanisms of this form of active vibrationcontrol. Results show that on average, the controller reduces the primary source input power and suppliesenergy to the system. The amount of energy dissipated by the controller can be neglected, which leads toconclusion that the controller is not dissipative.

11:45 Vibration Testing Techniques for Active Membranes (ID 125)Eric J. Ruggiero, GE Global Research, United States of AmericaDaniel Inman, Virginia Polytechnic Institute and State University, United States of America

The next generation of satellite bus technology will consist of ultra-flexible, ultra-lightweight structuralcomponents with frequency content ranging from less than 1 Hz to the KHz frequency bandwidth. Thesestructures, typically referred to as gossamer structures in the literature, possess ideal space launch character-istics but have undesirable low frequency dynamics that are easily excited by space debris impact, satellitemaneuvers, and even thermal transients. To overcome these challenges, a proper foundation of ground test-ing techniques must be developed and documented to help see this novel space satellite technology come tofruition.

The key structural component of one particular gossamer satellite bus design is a membrane mirror. Themembrane mirror will be held under tension and is expected to maintain a surface accuracy on the order of awavelength of light. To achieve such a stringent surface accuracy in the presence of undesirable system dy-namics, one proposed solution is to embed piezoceramic materials near the boundary of the lens. The activepiezoceramic bimorphs can be used to sense detrimental vibration and provide active feedback control tobring the mirror surface back into focus. For such a control system to be designed, proper testing techniquesof the augmented system’s dynamics must be well developed.

In the present work, membrane samples augmented with piezoceramic (type H4) bimorphs are tested undera variety of tensile loading conditions and in both ambient and vacuous settings. Due to the unique natureof the membrane materials, innovative testing techniques needed to be developed to excite and measure thesystem dynamics without any adverse effects. The proposed active portion of the system, the piezoceramicbimorph, was used to excite the out-of-plane modes of the sample. A laser vibrometer was used to measurethe local velocity of the membrane samples. The non-contact nature of the laser vibrometer ensured that thetrue system dynamics were unaffected during the experimentation. The presence of air around the membranesamples is shown to cause a significant mass-loading effect in the frequency response of the system. Theresults of a finite element analysis modeling the response of the active membrane samples in vacuum arecompared to the experimentally measured response within the bandwidth of 0 - 500 Hz. Over a large domainof tensile loads, the response of the system is predicted to within 6%. Good agreement is also found in themode shapes of the model and experimental results. The presented work provides a detailed summary ofproper testing techniques for capturing the dynamics of these novel structural components.

12:10 Generating 1D and 2D directional vibration waves (ID 64)Ran Gabay, Izhak Bucher, Technion - Israel Institute of Technology, Israel

Traveling waves in finite vibrating structures are often ignored since they propagate much less power andhave much smaller amplitudes than standing waves. Standing waves occur naturally in lightly damped struc-tures, under many types of excitation patterns. Traveling waves, on the other hand, necessitate specialboundary conditions and often more than one excitation source, to exist. For this reason, understanding theneeded conditions for traveling waves to prevail along with the ability to quantify the amount of wave travels

50


within the vibration is important. This work deals with the generation and identification of pure progressivewaves in finite structures a task that seems to require special methods.

Medium and high frequency techniques – MHF2Room 3 – Chairman: B. Pluymers

14:00 On the coupling of Wave Based models with modally reduced Finite Element modelsfor structural-acoustic analysis (ID 385)Bert Van Genechten, Bert Pluymers, Caroline Vanmaele, Dirk Vandepitte, Wim Desmet,Katholieke Universiteit Leuven, Belgium

This paper presents a newly developed hybrid simulation technique for coupled structural-acoustic analysis,which applies a wave based model for the acoustic cavity and a modally reduced finite element model forthe structural part. The resulting hybrid model benefits from the computational efficiency of the wave basedmethod, while retaining the finite element method’s ability to model the structural part of the problem in greatdetail. Application of this approach to the analysis of a cavity-backed plate assembly shows the improvedcomputational efficiency as compared to classical finite element procedures and illustrates the potential ofthe hybrid method as a powerful tool for the analysis of coupled structural-acoustic systems.

14:25 The Vibro-acoustic Responses of Structures with Uncertainties (ID 288)Geoffrey Lucas, N.J. Kessissoglou, The University of New South Wales, Australia

Modelling the vibro-acoustic responses of vehicles in the mid to high frequency range is particularly difficult.This is due in part to the structural and acoustic responses of sections of a vehicle being highly sensitive touncertainties arising from the assembly process and manufacturing tolerances. For example, the dynamicresponse of body panels can be greatly altered by small variations in spot welds. This leads to significantvariation in interior noise levels of successive vehicles from a production line. This paper investigates the useof techniques for predicting the responses of structures with inherent uncertainties in the medium frequencyrange. The dynamic characteristics and responses of an ensemble of coupled plate structures are examined,where uncertainty is generated by adding small masses at random locations. A measure of the uncertainty isobtained by observing the variation in the natural frequencies of an ensemble member from their mean valueacross the ensemble. An ergodic hypothesis is used to compare the frequency averaged response of a singlemember in the ensemble with the ensemble averaged response. Finally, structure-borne sound pressure levelsobtained experimentally are compared with results obtained from an SEA model.

14:50 Acoustical method for radiation calculation of complex structures (ID 425)Marianne Viallet, Ecole Centrale de Lyon / PSA Peugeot Citroen, FranceGerald Poumerol, Olivier Sauvage, PSA Peugeot Citroen, FranceOlivier Dessombz, Louis Jezequel, Ecole Centrale de Lyon, France

In the automotive industry, like in other transport industries, predicting noise during design cycle is a nec-essary step. Well-known methods exist to answer this issue in low frequency domain. Among these, Fi-nite Element Methods, adapted to closed domains, are quite easy to implement whereas Boundary ElementMethods are more adapted to infinite domains, but may induce singularity problems. In this article, a newmethod based on the well-known elastodynamics Substructure Deletion Method is presented. Analogies be-tween acoustical and seismic problems are made to transpose the method. It consists in dividing a complexunbounded problem into two easier ones to solve finite and infinite problems. Instead of considering a geo-metrically complex structure, a prismatic bounding volume is first studied. Then a classical Finite Elementcomputation is performed on the volume left between the box and the considered structure. The advantageof this technique is that when testing and comparing several geometries contained in such a box, only oneboundary element calculation is needed. Efficiency of this method is discussed in the present document.

51


15:15 Boundary Element Energy Method for the acoustic design of vehicles shields (ID374)Michael Thivant, Pascal Bouvet, VIBRATEC, FranceN. Blairon, VOLVO 3P, France

Acoustic design requires precise and reactive prediction tools. The numerical resolution of Helmholtz equa-tion is efficient in the low frequency range, but reaches its limits at higher frequencies. The method proposedhere is based on a light/sound analogy, and the resolution is based on Boundary Elements. Absorbing mate-rials are characterized by diffuse absorption coefficients. The frequency based meshing criterion is relaxed:only the geometry must be considered. The resolution can be made every third-octave. Computation timeis dramatically reduced. This energy method is dedicated to industrial acoustic issues in the mid and highfrequency range, preferably with complex geometries, broadband and distributed sources. A successful com-parison with measurements is presented, concerning the engine compartment of the Low-noise Urban Truck(LUT) prototype, developed by Renault Trucks for the PREDIT VPE (Vehicule Propre et Economique -Clean and Economical Vehicle) program. Other Industrial applications concerning the prediction of theefficiency of the shields around diesel engines are in progress.

15:40 Application of the Fast Multipole Method to the Variational Boundary ElementMethod for Large Acoustic Radiation Problems (ID 441)Stephane Paquay, Open Engineering S.A., BelgiumMichel Geradin, Joint Research Centre, European Commission, Italy

The Fast Multipole Method (FMM) has been successfully implemented in the multiphysics simulation soft-ware OOFELIE and applied to Variational Boundary Element Method (VBEM). The method allows to ad-dress acoustic radiation and diffraction problems with very large wave number (several tens of thousandunknowns on the radiation surface). It is demonstrated that the computation procedure is not only feasiblebut also runs quite fast on ordinary computers.

Medium and high frequency techniques – MHF3Room 3 – Chairman: H.-H. Priebsch

16:30 Modeling of structural sound transmission in train structures using hybrid FE-SEAand EFM analysis (ID 410)Ulf Orrenius, L. Baures, Bombardier Transportation, SwedenVincent Cotoni, ESI-US R&D, United States of America

A theory has been developed to rigorously combine FE and SEA approaches in the description of the dy-namic behavior of complex vibro-acoustic systems. The new technology, called Hybrid FE-SEA, has beenapplied to analyze transmission of structural sound in train floor structures with focus on the transmissionfrom the bogie and entrance floor regions. Certain input parameters to the hybrid model are derived using theEnergy Flow Method (EFM). After a short review of the underlying theory, the modeling and sub-structuringstrategy is described. The benefits of the hybrid approach for vehicle structure-borne noise analysis are ex-plored with reference to traditional SEA, full FE as well as EFM. Comparison between results from hybridFE-SEA analysis and analytical models as well as measured data show good agreement. A discussion ofcandidate problems for application of hybrid FE-SEA analysis for railway vehicles as well as a discussionon meshing strategies for maximal computation efficiency is also provided.

52


16:55 Prediction of the ensemble mean and variance of the response of uncertain struc-tures using the Hybrid FE-SEA method (ID 503)Robin Langley, University of Cambridge, United States of AmericaVincent Cotoni, ESI-US R&D, United States of America

A Hybrid FE-SEA method was recently derived to rigorously couple the finite element and statistical energyanalysis methods. The FE method is used to describe the components of a system that have a few modes(or a long free wavelength when compared to the dimensions), and that consequently exhibit a fairly robustdynamic behavior. Alternatively, the SEA method is used to describe the uncertain components (with manymodes or short wavelength). The result yielded by the method is the dynamic response averaged over anensemble of uncertain structures. Recently, the prediction of the ensemble mean has been completed by theensemble variance, so that the deviation from the mean response can be computed. The theory underlyingthe Hybrid FE-SEA variance derives from a recently derived diffusefield reciprocity relation and was shownto be coherent with the variance theory developed in the context of SEA only. Numerical simulations areprovided validating the new formulation.

17:20 Application of the Wave Based Prediction Technique for structural problems withstress singularities (ID 353)Caroline Vanmaele, Wim Desmet, Dirk Vandepitte, Katholieke Universiteit Leuven, Belgium

The Finite Element Method is a commonly used prediction method for dynamic simulations of plate bendingproblems. A major disadvantage of this method is its practical frequency limitation in that the computationalloads become prohibitively large at higher frequencies. A newly developed Wave Based prediction techniqueaims to relax this frequency limitation through an enhanced computational efficiency. This paper discussesthe application of the Wave Based Method for the particular case where stress singularities appear in oneor more corners of a polygonal plate domain. In this case the conventional set of field variable expansionfunctions is extended with some special-purpose functions which incorporate the corner point singularities.The beneficial convergence rate of the Wave Based Method as compared with the Finite Element Method isverified for various validation examples.

17:45 On the convergence of the WBM solution in certain nonconvex domains (ID 183)Jevgenijs Jegorovs, ITWM, Fraunhofer-Institut fur Techno- und Wirtschaftsmathematik, Germany

This paper discusses the possibility to apply a novel numerical technique, i.e. Wave Based Method, to solvesteady-state acoustic problems within 2D non-convex with smooth concave parts. We find a domain criterionwhich helps to analyse the convergence of the WBM solution. With the aid of our criterion one is able tosubdivide arbitrary 2D domains such that the number of subdomains is minimal, WBM may be applied ineach subdomain and the geometry is not altered, e.g. via polygonal approximation. Three examples, whichcorroborate our theory, are also presented.

18:10 Numerical Solution of High-Frequency Oscillations of a Gas Column Using a Mod-ified Lax-Wendroff Scheme (ID 60)Arash Ghasemi, Sharif University of Technology, Iran (Islamic Republic of)

Recently, several high-order accurate methods have been developed for solving nonlinear equations govern-ing in the high-frequency oscillations of a fluid column. If we assume that the value of Reynolds and Prandtlnumbers are high enough then it is possible to neglect the effect of Stokes Layer and it has been previouslyshown that the calculated pressure profile obtained by solving three-dimensional axisymmetric compressibleNS (Navier-Stokes) equations is in good agreement with one-dimensional adiabatic model of Ilinskii. Forthis particular flow which is the typical of engineering flows inside acoustical compressors, it is clear thathigh-resolution simulation of NS equations in higher dimensions is quite expensive. For this reason, a quickmethod based on the lax-wendroff scheme is used in this paper for solving the one-dimensional adiabaticmodel. The numerical method used here is simpler in implementation than previous lagrangian and spectral

53


approaches and is considerably more efficient than three-dimensional solutions. The results agree well withprevious multi-dimensional solutions.

Rotating machinery:dynamics – RMD1Room 4 – Chairman: M. Friswell

8:50 Analysis of Dynamic Loads in Multi-Megawatt-Drive-Trains (ID 95)Berthold Schlecht, Tobias Schulze, Thomas Hahnel, Thomas Rosenlocher, Technical Universityof Dresden, Germany

As a general problem by the construction and dimensioning of large drive trains, the possible conjunctionof drive train and structure natural frequencies emphasizes. It is especially good to see in the field of windturbines. Due to the system a complex, elastic and vibratory complete object results under turbulent sto-chastic input conditions (wind speed). The same tendencies can be found in other fields of application also.In the meantime even similar damages can be seen interbranch. This leads to the not new conclusion thatit is absolutely necessary to have an exact knowledge of the dynamic behavior of the entire drive systemand its surrounding already in the construction phase. However, that requires that for the illustration of thevibratory systems appropriate methods of modeling are used for the individual problem. At first, it seems asif a simple torsional model is enough for the determination of the first torsional natural frequencies, but atthe latest when using an elastic MBS-model it can be seen that even this simple mode shape can be overlaidwith further mode shapes (e.g. bending of the housing). To reach a complete and significant characterizationof the dynamic system it is not enough to do a torsional vibration analysis by large drive trains with elasticelements. Instead, reduced FE-models after the method of the Guyanreduction or the modally reduction areintegrated in the MBS-environment and analyzed as elastic MBSmodels.

9:15 Pressure phenomena in dynamic analysis of external gear pumps (ID 192)Giorgio Dalpiaz, Emiliano Mucchi, Gianluca D’Elia, University of Ferrara, ItalyAlfonso Fernandez del Rincon, University of Cantabria, Spain

A previous non-linear lumped kinetoelastodynamic model for the study of the dynamic behaviour of externalgear pumps has been further developed in order to improve the accuracy of the simulation results. In partic-ular, this paper deals with the modelling and analysis of the phenomena bound to the pressure distributionaround the gears, since they have the most important effect on the dynamic behaviour. Phenomena as thepressure variation in the inlet and outlet chambers have been taken into account. Special attention was givento the pressure in the trapped volume which has been modelled in a more accurate way in order to understandpossible vibration sources. The pressure phenomena are shown through some simulated results. The modelcan be used in order to foresee the influence of working conditions and design modifications on vibrationgeneration.

9:40 Difficulties in Predicting Vibrations in Turbomachinery with Hydrodynamic Bear-ings (ID 240)Ningsheng Feng, Eric Hahn, The University of New South Wales, AustraliaWenlong Hu, The University of Technology, Sydney, Australia

Turbomachineries with hydrodynamic bearings are generally statically indeterminate rotor bearing systems.To predict their vibration behaviour, one needs to correctly specify the relative lateral bearing alignment. Thisalignment, which determines the bearing reaction forces (which in turn influence the stiffness and dampingof the bearing supports), is generally unknown and its determination usually requires a knowledge of themean bearing reaction forces at some speed at which the system response is periodic. These reaction forcesare presently evaluated from the Reynolds equation using measured instantaneous rotor motion values. Itis shown via numerical experiments that these forces, when evaluated from measured rotor motion data via

54


Reynolds equation, can be very sensitive to the accuracy of this data; so that insofar as determination ofthe relative bearing alignment relies on these forces, presently existing software is likely to be deficient incorrectly predicting the system vibration behaviour.

10:05 Unbalance Response of Statically Indeterminate Rotors Supported on Hydrody-namic Journal Bearings: Use of the 32 Coefficients Bearing Model (ID 50)Demetrio C. Zachariadis, Polytechnic School - University of Sao Paulo, Brazil

Linear vibration analyses of statically indeterminate rotors supported on hydrodynamic journal bearingsdemand large amounts of numerical work in order to calculate the static equilibrium positions of the journalsrelative to the respective bearings at different values of rotating speed and the corresponding stiffness anddamping dynamic coefficients.

As shown by the author in a previous paper, all that work might be useless if the wrong bearing model isadopted in the unbalance response calculation. In that paper, it was verified that the traditional 8- coefficientbearing model is inadequate for the unbalance response calculation of single spam rotor systems supportedon hydrodynamic journal bearings placed close to nodal points of excited modes of vibration. In suchsituations, one cannot neglect the time varying tilt angle between journals and bearings, the consideration ofwhich leads to the adoption of a 32-coefficient bearing model.

A similar analysis is conducted in this work, focusing on statically indeterminate rotors. Numerical results,related to a large turbo set, indicate that the differences between vibration amplitudes calculated using bothbearing models can be greater than 80%, while discrepancies in the predicted stability thresholds are small.The conclusions of the study are coherent with previously published theoretical and experimental results.

Rotating machinery:dynamics – RMD2Room 4 – Chairman: M. Friswell

10:55 Structural analysis of a wind turbine and its drive train using the flexible multibodysimulation technique (ID 597)Joris Peeters, Hansen Transmissions International, BelgiumDirk Vandepitte, Paul Sas, Katholieke Universiteit Leuven, Belgium

This article demonstrates the application of a generic methodology, based on the flexible multibody simu-lation technique, for the dynamic analysis of a wind turbine and its drive train, including a gearbox. Theanalysis of the complete wind turbine is limited up to 10 Hz, whereas the study of the drive train includesfrequencies up to 1500 Hz. Both studies include a normal modes analysis. The analysis of the drive trainincludes additionally a response calculation for an excitation from the meshing gears, a Campbell analysisfor the identification of possible resonance behaviour and a simulation of a transient load case, which occursas a sudden torque variation caused by a disturbance in the electrical grid.

11:20 The response of rotating machines on viscoelastic supports (ID 330)Michael Friswell, University of Bristol, United KingdomJerzy Sawicki, Cleveland State University, United States of AmericaDaniel Inman, Virginia Polytechnic Institute and State University, United States of AmericaArthur Lees, University of Wales Swansea, United Kingdom

Damping in the stator of a rotating machine is able to reduce the unbalance response, and increase thespeed where the stability limit is reached. A convenient method to introduce damping is to support thebearings on viscoelastic elastomeric supports. This successfully adds damping but makes the analysis con-siderably more difficult. Viscoelastic materials have properties that depend on frequency and temperature.

55


The determination of the run-up or run-down response of a linear machine at a constant temperature is quitestraightforward based on measured material characteristics. However the introduction of transient and non-linear effects, temperature changes and model uncertainties makes the analysis difficult. This paper usesthe internal variable approach to model the viscoelastic material for the transient and non-linear dynamicresponses. For balancing, robust estimation methods are demonstrated to provide satisfactory balancing ofthe machine over a range of environmental conditions. Examples of flexible machines on rolling elementbearings supported on elastomeric mounts are given.

11:45 Vibration of Gears: a Global Optimization Approach (ID 422)Giorgio Bonori, Giorgio Scagliarini, Marco Barbieri, Francesco Pellicano, University of Modenaand Reggio Emilia, Italy

The importance of studying dynamic aspects of gear transmissions has been well established since manyyears; however, recently a new interest on this topics grew up because of significant improvement in geardesign technology. The literature can supply different approaches to evaluate the vibrations response ofsimple and complex gears system. Numerical and analytical methods can provide important information onresonances, non linear behaviours and combined dynamic interactions in existing systems. In particular thereduction of vibrations is strictly related to minimize the value of the peak to peak of the static transmissionerror in a mesh cycle, through modifications of the tooth profile. The present paper combines a numericalmodel to study spur gear pair dynamics, with a genetic optimization technique to define the best profilemodification capable to reduce the vibration. The numerical model is able to take into account gear geometry,profile modifications and profile manufacturing errors. The genetic optimization routine can provide the bestset of profile modification considering the peak to peak value of the static transmission error or a genericnumber of its harmonic components. In the last part of this work some testcases are investigated and theimprovement in the dynamics is shown.

12:10 Dynamic Modelling and Testing of a Representative Aero-engine Test Rig with Ad-justable Nonlinear Bearing Supports (ID 583)Sen Huang, David A. Robb, David J. Ewins, E. Petrov, Imperial College London, United Kingdom

A representative model of aero-engine test rig has been built, comprising the following basic components:rotor, shaft, casing, flexible support structure and bearing housings. One of the bearing housings is designedin such a way that the stiffness along one radial direction is adjustable, and can be set between linear andstrongly nonlinear. In the latter case, it has the characteristic of snap-through springs, which is somewhatanalogous to some magnetic bearing configurations. The adjustability of the bearing support stiffness makesit possible to study the dynamic properties of an engine rig with different boundary conditions. A notableaspect of this test rig is that all the major flexibilities in an operating aero-engine have been included, e.g.flexible support structure, casing etc.

In the field of structural dynamics, mathematical models are widely used, especially at the design stage of aproduct when the effect of physical modifications on the total dynamic response of the structure is requiredbefore the real fabrication is carried out. In addition, highly accurate and efficient structural mathematicalmodels are required for the emerging SMART machine concept, in which real-time machine diagnosis andprognosis methods demand fast and accurate decision-making based on the results of data processed usingthose models. The problem we are facing now is that, as the structure becomes more complicated, and con-sists of more segments and joints, many of which can be strongly nonlinear, the accuracy and efficiency of themathematical model deteriorates rapidly due to the difficulties in modelling the joints and the nonlinearitiesexisting in those joints.

In this paper, a Frequency Response Function (FRF) coupling scheme, together with the Multi-HarmonicBalance Method (MHBM), is used to model the assembled test rig. The basic FRF coupling method hasbeen well documented and its accuracy and efficiency in linear structure assemblies has been recognised inmany papers and books. MHBM is widely used to solve nonlinear problems in the frequency domain. Theadvantage of this combined methodology is shown in the comparison between test and simulation results.

56


OMAX Operational Modal Analysis in presence of exogeneous inputs– OMAX1Room 4 – Chairman: P. Guillaume

14:00 Continuous-time operational modal analysis (ID 45)Rik Pintelon, Patrick Guillaume, Johan Schoukens, Vrije Universiteit Brussel, Belgium

In operational modal analysis the modal parameters of a structure are identified from the response of thatstructure to unmeasurable operational perturbations. It is based on the assumption that the operational pertur-bations are white at the sampling instances. Although the operational perturbations are actually bandlimitedcontinuous-time signals, till now the observed power spectra are modelled as filtered discrete-time whitenoise. This introduces model errors that can jeopardize the physical interpretation of the results. Here wepropose to model the response as continuous-time filtered band-limited white noise. An in depth analysisof the imperfections of the framework is made and some remedies are proposed. In addition to the modalparameters the presented technique also provides uncertainty bounds. The theory is illustrated on real mea-surements on a bridge.

14:25 OMAX - A Combined Experimental-Operational Modal Analysis Approach (ID649)Patrick Guillaume, Vrije Universiteit Brussel, BelgiumTim De Troyer, Erasmushogeschool Brussel, BelgiumChristof Devriendt, Gert De Sitter, Vrije Universiteit Brussel, Belgium

In experimental modal analysis (EMA), all input and output signals are measurable. In operational modalanalysis only the output signal are measurable; the operational forces are unknown and usually assumed to bewhite noise sources. The OMAX framework is a unifying approach combining experimental and operationalmodal analysis. OMAX stands for Operational Modal Analysis in presence of eXogenous inputs. In otherwords, the OMAX framework is an operational modal analysis approach where unknown operational forcesas well as measurable input forces are available. The measurable input forces can be deterministic as wellas stochastic. In this contribution the OMAX framework will be elaborated paying special attention to thecurrent state-of-the-art OMAX estimators and their possible applications in mechanical and civil engineering.It will be shown how the modal parameters can be estimated from output-only (OMA) as well as input/output(OMAX) measurements done at operating conditions.

14:50 Load spectrum estimation from output-only measurements applied to a spray boommodel (ID 650)K. Engelen, Herman Ramon, Jan Anthonis, Katholieke Universiteit Leuven, Belgium

A finite element model is constructed representing the dynamic behavior of a spray boom. The model isupdated with experimental vibration data obtained from field measurements. The spectra of the input forcesare identified from these measurements and based on a parameterized spectrum model, a classification ismade between normal and rough spraying conditions.

15:15 Identification of damage in brittle materials using mode shapes obtained with a laserscanner (ID 647)Athanasia Petreli, Hugo Sol, Patrick Guillaume, Vrije Universiteit Brussel, Belgium

Mechanical and environmental loadings cause crack development in brittle materials. In beamlike specimens,crack damage can be translated into a reduction of the local bending stiffness values. This paper presentsan identification method to evaluate the local stiffness values of thin beams using mode shapes measuredwith a laser velocity scanner. In a first step, the measured mode shapes are first averaged and next curvefitted by Hermitian polynomials to remove the measurement noise. The curve fitted mode shapes togetherwith the (assumed) known mass distribution and resonance frequencies are used to compute the shear force

57


and bending moment. Next a special purpose beam element, a ‘scan element’ with only one degree offreedom, is programmed to ‘scan’ the stiffness distribution. The method requires no scaled mode shapes.Numerically generated data with artificial noise are used to validate the proposed method. A real experimenton a composite beam with a cement matrix is shown as illustration.

15:40 Operational modal analysis of time varying systems using an exogenous multi-sineexcitation (ID 652)Steve Vanlanduit, Rik Pintelon, Vrije Universiteit Brussel, BelgiumTim De Troyer, Erasmushogeschool Brussel, BelgiumPatrick Guillaume, Vrije Universiteit Brussel, Belgium

A basic assumption when using modal analysis parameter identification procedures is that the structureunder test is time invariant. In many practical structural dynamics applications, however, these assumptionsare violated (because either the boundary conditions or the structural parameters change in time). In thispaper we will introduce a new and simple method to estimate the modal parameters of a system togetherwith their variation in time during the experiment. The method uses measurements of a structure underoperational excitation combined with an exogenous multi-sine excitation. The initial modal parameters ofthe structure are identified from the operational response signals and the time variation is estimated bytracking the amplitude of the force and the response related to the exogenous multi-sine excitation. Themethod will be validated on a computer simulation and on three experimental cases: (a) a water tank with adecreasing water volume coupled on a beam, (b) a cracked U-profile and (c) a scale-model airplane placedin a wind tunnel.

OMAX Operational Modal Analysis in presence of exogeneous inputs– OMAX2Room 4 – Chairman: G. De Roeck

16:30 Reference-based combined deterministic-stochastic subspace identification for op-erational modal analysis with deterministic inputs (ID 164)Edwin Reynders, Guido De Roeck, Katholieke Universiteit Leuven, Belgium

In classical operational modal analysis (OMA), the modal parameters of a structure are determined from itsdynamic response to ambient forces. The application of this technique to civil engineering structures is par-ticularly interesting since heavy and expensive artificial excitation devices, required for experimental modalanalysis (EMA), are not necessary. However, disadvantages of OMA are the limited frequency content ofthe ambient excitation and the fact that mass-normalization of the mode shapes is not possible. A possi-ble solution is to use operational modal analysis techniques that allow deterministic or exogenous inputs(OMAX). The difference between OMAX and EMA lies in the fact that with OMAX, the ambient forces arenot considered as noise, but as part of the excitation. As a result, OMAX requires only small and relativelycheep artificial excitation devises, but it requires special combined deterministic-stochastic system identifi-cation methods. In this paper, the reference-based combined deterministic-stochastic subspace identificationmethod for OMAX is discussed. The theory is illustrated on a benchmark problem: the determination of themodal parameters of the Z24 bridge. With the presented method, the best benchmark results reported so farare obtained.

16:55 Challenges in operational modal analysis of suspension bridges (ID 600)Kirk Grimmelsman, Emin Aktan, Drexel University, United States of America

Operational modal analysis is being increasingly used as an experimental technique to identify the dynamicproperties of many types of large-scale constructed systems. The writers conducted such an experiment toidentify the dynamic properties for the masonry towers of a landmark suspension bridge with the objective of

58


providing data to validate and improve the reliability of a seismic evaluation and retrofit investigation. One ofthe towers was instrumented by a dense and stationary array of sensors, and the other tower and spans wereinstrumented with a limited number of reference sensors. The spans had been tested and characterized earlierby different researchers using ambient and forced excitation techniques. The authors did take advantage ofthose results in designing the instrumentation to characterize the towers. This paper describes some of thechallenges encountered in trying to identify the dynamic properties of the towers using this approach, andthe techniques the writers used in an attempt to overcome these challenges. Recommendations for futureapplications with this type of structure are also discussed.

17:20 A Comparison of Experimental, Operational, and Combined Experimental-Operational Parameter Estimation Techniques (ID 354)Tom Lauwagie, Robby Van Assche, Johan Van der Straeten, Ward Heylen, Katholieke UniversiteitLeuven, Belgium

A modal analysis aims at the identification of the modal parameters of a test structure from the measured vi-bratory behaviour. Traditionally, both the input forces and the resulting responses are measured. However, inmany applications it is not possible to measure (all) the input forces. During the last decade, two new classesof modal parameter estimation techniques have been developed to overcome this problem: the operationaltechniques and the combined experimental-operational techniques. Operational modal analysis techniquescan identify the modal parameters from the responses of the structure; they do not require the input forces.The combined experimental-operational techniques, only require a part of the input forces to estimate themodal parameters.

The work presented in this paper is part of the evaluation process of these new modal parameters estimationtechniques; it compares the modal parameters provided by the operational and combined experimentaloper-ational modal analysis techniques with the modal parameters obtained with the experimental modal analysistechnique.

17:45 Office floor vibrations: modal parameter identification and vibration monitoring(ID 651)R Cantieni, rci dynamics, Switzerland

Several people working on the 2nd floor of an office building complained about disturbing floor vibrations.The building consists of five floors. In the contrary to the other floors, the 2nd floor slab has no floor-to-ceiling secondary walls neither beneath nor on top of the slab. The modal parameters of four floors wereidentified using ambient vibration testing (AVT)-technology and a 5-kg-medical ball as a vibration generator.To monitor the vibration intensity and to identify the source of the disturbing vibrations, a triaxial velocitysensor was subsequently mounted in a critical point of the 2nd floor slab. The vibrations were monitored fortwo months using a newly developed internet-accelerograph. This allowed online checking of the vibrationsand downloading of the data on an external server on a daily basis. Processing of these data yielded that noother source of the vibrations could be identified than people walking on the floor.

59


Self-excited vibrations – SEV1Room 5 – Chairman: P. Vanherck

8:50 Brake squeal: a multi parametric phenomenon assessed in terms of robustness (ID133)Guillaume Fritz, Jean-Marc Duffal, David Perez-Recio, Stephane Bonnet, Renault, SpainJean-Jacques Sinou, Louis Jezequel, Ecole Centrale de Lyon, France

Brake squeal is a friction induced instability phenomenon which must be taken into account in the brakedevelopment and validation processes. In order to assess squeal propensity upstream from experiments, afinite element based method, called complex eigenvalues analysis is used. Thus the brake stability can becomputed with respect to the parameter which is the root cause of squeal: the friction coefficient. This pa-rameter analysis highlights the phenomenon referred to as mode coupling. Then, the method has been usedto compare two caliper designs. The numerical results turns out to be well correlated with experimentalones. Finally, the effect of the disc young modulus has been investigated to show that brake squeal must beconsidered as a multi-parametric phenomenon. In order to illustate this point, the brake behavior has beensynthesized by stability charts.

9:15 Effects of track parameters and environmental conditions on tramcar inducedsqueal noise (ID 204)Roberto Corradi, Alan Facchinetti, Stefano Manzoni, Marcello Vanali, Politecnico di Milano,Italy

Rail systems noise is an area of continuing research interest. Among the different noise generation phenom-ena, squeal noise play an important role when dealing with tramcar operation. In fact, urban track are usuallycharacterised by the presence of very narrow curves, which lead to a high chance of squeal. Moreover, sincetramcar operate in a urban environment, near office and/or residential buildings, noise issues assume a par-ticular relevance. The Mechanical Engineering Department has started a noise measurement and analysisprogram specifically concerned with the problem of tramcars squeal noise. The experience made duringexperimental tests and parallel numerical simulations is illustrated in this paper, whose objective consists notin providing theoretical analysis on squeal noise generation mechanism, but in pointing out some significantfeatures of this phenomenon which are very important for understanding how to control it.

9:40 A Higher Order Harmonic Balance Approach for an Aeroelastic Airfoil with aFreeplay Control Surface (ID 566)Nicholas Amuyedo, Jonathan Cooper, University of Manchester, United Kingdom

In this paper, the nonlinear dynamic behaviour of an aeroelastic system comprising an airfoil with a trailingedge freeplay control surface mounted in an incompressible flow is considered. The true nonlinear aeroelas-tic behaviour of the system is solved for using numerical integration. A Higher Order Harmonic Balance(HOHB) Method is then derived and implemented to predict the Limit Cycle Oscillations (LCO) that thesystem may encounter beyond bifurcation airspeeds. The influence of the higher harmonics on the accuracyof the predicted LCO behaviour was investigated and it was demonstrated that when five or more harmonicswere retained in the HOHB solution, the predicted results were in excellent agreement with the limit cycleoscillations obtained by the numerically integrating the non-linear equations of motion.

10:05 Determination of the Cutting Performance and Dynamic Behaviour of UniversalMilling Machines by means of a Stability Model (ID 438)Jokin Munoa, IDEKO Centro Tecnologico S.Coop., SpainMikel Zatarain, TEKNIKER Fundazioa-IK4, SpainInigo Bediaga, IDEKO Centro Tecnologico S.Coop., Spain

Self-excited or chatter vibrations are well known among milling machine builders and users. Nowadays,

60


chatter is one of the most important restrictions of the milling process. Self-excited vibrations prevent ob-taining the required accuracy in workpiece, reducing the lifetime of the cutter and the mechanical compo-nents of the machine. Due to chatter vibrations, the dynamic stiffness of the structure of the milling machinelimits the productivity in steel and cast iron roughing. Hence the stability of this kind of process is an indi-rect way to measure the dynamic stiffness of the design. This paper presents different stability models andproposes the most suitable for the determination of the cutting performance of universal milling machines.Finally the cutting performance and the dynamic behaviour of the milling machine is determined combiningthe measurements of the dynamic stiffness of the structure, the stability models, cutting tests and modalanalysis.

Transfer path analysis and source identification – TPA1Room 5 – Chairman: C.-P. Fritzen

10:55 Online estimation of external loads from dynamic measurements (ID 196)Maksim Klinkov, Claus-Peter Fritzen, Universitet Siegen, Germany

For purposes of monitoring and damage prognosis it is important to know the external loads which act on astructural system. In many practical applications it is not possible to measure the forces e.g. resulting fromwind loads or traffic directly. Therefore, these forces are determined indirectly from dynamic measurementswith an observer-based concept. In this work, a new method of simultaneous state and input estimation for aclass of nonlinear systems, which is used in the area of control engineering, is adapted to online indirect forcemeasurement. This method allows for the online simultaneous reconstruction of unknown force functionsand system states such as displacement and velocity. The load reconstruction is illustrated in the exampleof a structure. The robust observer is designed on the basis of an analytical model. The load identificationwas done with the help of strain and accelerometer transducers, which were used as input signals for thecalculated observer. The observer performance was verified by using a force sensor.

11:20 Engine contribution analysis using a noise and vibration simulator (ID 156)Andreas Schuhmacher, Dmitri Tcherniak, Bruel & Kjaer Sound & Vibration Measurement A/S,Denmark

In this paper a pure time-domain version of source-path-contribution analysis is investigated using a control-lable source, a noise and vibration simulator installed into a trimmed vehicle. Both airborne and structure-borne inputs are investigated and the matrix method in the time-domain is used to calculate source contribu-tions as sounds at a listeners’ position inside the cabin. Operating data from a simulated run-up/run-downand sets of transfer functions (FRFs) are firstly used to estimate the strength of some defined point sources,acoustically and mechanically. Secondly the operating source strengths are combined with acoustic or vibro-acoustic FRFs to predict contributions at a receiver. In this work it is attempted to make the airborne andstructure-borne models as simple as possible, and predicted contributions are validated against actual mea-sured data. All measurements were conducted on the vehicle with and without the engine simulator installed.

11:45 Acoustic Equivalent Point Source Models of Tires using Combinatorial Optimiza-tion (ID 152)Jakob Mørkholt, Andreas Schuhmacher, Bruel & Kjaer Sound & Vibration Measurement A/S,DenmarkErnst-Ulrich Saemann, Continental AG, Germany

A method for determining position-optimal acoustic equivalent point source models is presented. The pointsource positioning problem inherent in acoustic equivalent point source modeling is formulated as a combi-natorial optimization problem, based on iterative selection of indicator transfer functions from a large poolof pre-determined candidate transfer functions. The candidate transfer functions are calculated numerically,

61


using a reciprocal Boundary Element formulation. Solutions to the optimization problem are sought using acombination of Genetic Algorithm and Steepest Descent Local Search methods. Results are presented fromapplying the method to determine acoustic equivalent point source models of varying accuracy of a real tirerunning on a dynamometer.

12:10 Identification of Excitation Force Generated From Automobile Engine in High Fre-quency Range (ID 31)Nobuyuki Okubo, Satoru Shimamura, Takeshi Toi, Chuo University, Japan

In order to design the sound quality in interior and exterior space of a car and improve the vibration comfort,the identification of excitation force generated from the engine is very important and inevitable. Sincethe direct identification of force under operating condition is difficult to conduct, the indirect identificationbased on FRF measurements, that is, the inverse FRF times the response result the force, becomes popularbut requires the excitation at force generation points such as main bearings, which is very hard to installthe exciter. As alternative way, by use of reciprocity, the FRF may be replaced by exciting the monitoringresponse point and detecting the response at force generation point. But it is still very time consuming toexcite several points.

Instead of that, the ordinal modal testing in which only a few easy access point are stationary excited whilemany responses can be measured within short time and a little effort, is expected to predict the FRF be-tween the force generation point and the monitoring response point under operating condition. After curvefitting, the prediction is possible so called as synthesized FRF. However in high frequency range, the curvefitting only provides very limited number of mode and the synthesized FRF may be far from the correct oneespecially at frequency valleys that are very sensitive to be inverted in force identification process.

In this paper a new approach is proposed in which the number of mode is successively expanded in thevicinity of high frequency range of interest and the differences of synthesized FRF during such expansionare taken into account to determine the appropriate monitoring response points. This approach is verified bya numerical FE and experimental model of T plate and finally an actual V6 automobile engine is tested toidentify the excitation force.

Ground vibrations – GRV1Room 5 – Chairman: G. Degrande

14:00 A numerical model for ground-borne vibrations and reradiated noise in buildingsfrom underground railways (ID 334)Shashank Gupta, Katholieke Universiteit Leuven, BelgiumPeter Fiala, Budapest University of Technology, HungaryMohammed Hussein, University of Cambridge, United KingdomH. Chebli, Ecole Centrale de Paris, FranceGeert Degrande, Katholieke Universiteit Leuven, BelgiumFulop Augusztinovicz, Budapest University of Technology, HungaryHugh Hunt, University of Cambridge, United KingdomDidier Clouteau, Ecole Centrale de Paris, France

An efficient and modular numerical prediction model is presented to predict vibrations and re-radiated noisein buildings due to metro trains. The approach comprises of three weakly coupled problems: the dynamicvehicle-track-tunnel-soil interaction problem, the dynamic soil-structure interaction problem and the cal-culation of the acoustic response inside the rooms. The three-dimensional dynamic tunnel-soil interactionproblem is solved with a subdomain formulation, using a finite element formulation for the tunnel and aboundary element method for the soil. The periodicity of the tunnel and the soil in the longitudinal direc-

62


tion is exploited using the Floquet transform, limiting the discretization effort to a single bounded referencecell. The Craig-Bampton substructuring technique is used to efficiently incorporate a track in the tunnel.The track-tunnel-soil interaction problem is solved in the frequency-wavenumber domain and the wave fieldradiated into the soil is computed. This incident wave field is used to solve the dynamic soil-structure in-teraction problem on the receiver side and to determine the vibration levels in the structure. The receiverside dynamic soil-structure interaction problem is solved by means of a 3D boundary element method forthe soil coupled to a 3D finite element method for the structural. A weak coupling between the structuraland acoustic vibrations is accounted for and an acoustic spectral finite element method is used to predictthe acoustic response. To demonstrate the efficiency of the approach, the Bakerloo line tunnel of LondonUnderground has been modelled using the coupled periodic FE-BE approach. It is a deep bored segmentedtunnel, with a cast iron lining, embedded in the London clay. The response in the free-field due to a movingvehicle on an uneven rail is predicted. Subsequently, the re-radiated noise in a hypothetic multi-story portalframe office building is estimated. This numerical model enables to investigate the inherent physics of thegeneration and propagation of vibrations and re-radiated noise in buildings from underground railways.

14:25 Identification of random field models for elastic moduli from spectral analysis ofcompression wave experiments (ID 408)Maarten Arnst, Quang Anh Ta, Didier Clouteau, Ecole Centrale de Paris, FranceMarc Bonnet, Ecole Polytechnique, France

This paper addresses the identification of random field models for elastic material properties of heteroge-neous solids. The data used in the inversion are measured frequency-dependent phase velocities of elasticwaves travelling through heterogeneous specimens. The unknown field to be identified is parameterized by asmall set of dispersion parameters and spatial correlation lengths. A mathematical model of the mechanicalbehaviour of the specimens is built to predict the measurements. The maximum likelihood principle is thenused to set up an optimization problem for the estimation of the parameters of the random field model.

14:50 An iterative coupled boundary-finite element method for the dynamic response ofstructures (ID 474)Stijn Francois, Hamid Masoumi, Geert Degrande, Katholieke Universiteit Leuven, Belgium

Dynamic excitations in the built environment as caused by earthquakes, heavy traffic and pile driving mayresult in structural damage, which is determined by the constitutive behaviour of building materials andfoundation soils under cyclic loading.

This paper presents a coupled finite element-boundary element approach for the calculation of the dynamicresponse of structures due to dynamic excitations. Both the non-linear constitutive behaviour and the dy-namic interaction between the soil and the structure are accounted for. A time domain finite element formu-lation is used for the structure, as the non-linear constitutive behaviour of the structural materials requires adirect time integration procedure. The soil is assumed to be linear elastic and a boundary element method isused to fully account for dynamic soil-structure interaction.

As both the finite and the boundary element method impose different conditions on the time integration stepfor reasons of stability and accuracy, an iterative coupling scheme is proposed that allows for a different timestep in both subdomains. An interface relaxation technique is employed in order to speed up convergence.Instead of selecting a constant value, an optimal relaxation parameter is computed using Aitken’s method,resulting in a non-stationary Richardson iteration. The numerical behaviour of this scheme is studied indetail.

The method is applied to the problem of a circular foundation resting on a linear elastic halfspace and tothe calculation of the response of a structure due to traffic induced vibrations, where the dynamic interactionbetween the soil and the structure is fully accounted for.

63


Material characterisation based on dynamic measurements – MAT1Room 5 – Chairman: G. Pavic

15:40 Pipe thickness identification by modal order analysis (ID 342)Goran Pavic, INSA de Lyon / University of Salford, FranceHusar Calin-Ioan, University of Transylvania, Romania

Vibrations of a cross section of a pipe are composed of a multitude of sinusoidal deformations of orders0, 1, 2... The vibration waves related to orders higher than 1 can propagate only above a certain ”cut-on”frequency which is different for each order. The cut on frequencies depend on several parameters one ofwhich is the pipe thickness. The values of cut-on frequencies of an infinite pipe can be computed. The paperdescribes a method to identify the cut-on frequencies by measurement and to extract out of measured valuesthe actual pipe thickness. The identification is done via a decomposition of vibration signals coming froman array of sensors into circumferential orders. To demonstrate the method the results of both numericalsimulations and measurements are presented.

16:05 Identification of damage using low frequency harmonics in trusses and beams (ID143)Andrzej Swiercz, Przemyslaw Kolakowski, Jan Holnicki-Szulc, Institute of Fundamental Techno-logical Research, Polish Academy of Sciences, Poland

The aim of this work is to present a damage identification method dedicated to truss and frame structures. Thecore of the approach is the Virtual Distortion Method, which is a fast reanalysis method successfully appliedto damage identification. Loss of mass and stiffness are modelled by virtual distortions and modificationsof the parameters are calculated as a result of a sensitivity-based minimisation. In this paper we deal with asteady-state problem i.e. low frequency, non-resonance harmonic excitation induces a static-like structuralresponse with virtual distortions (design variables) modelling parameter modifications.

16:30 Test method development for determination of dynamic properties of shock andvibration absorbing materials (ID 554)Markku Juntunen, VTT Smart Machines, FinlandKari Ojala, Nokia Research Center, FinlandJarkko Keinanen, Kalle Vehvilainen, VTT Smart Machines, Finland

Materials with nonlinear dynamic properties and high damping properties have become more common toreduce shock and vibration loads of electronic equipment. The dynamic properties of such materials shouldbe known for proper design and modelling work. Thus, a practical mathematical model with compatiblematerial data is needed. However, reliable information on the dynamic properties of such materials is oftendifficult to find. This lack of information is even more evident with new materials and more demandingloading conditions. In this work, simplified dynamic testing and modelling methods were developed forpractical design work and as a basis for future research. The experiments discussed in this paper wereconducted on a cellular urethane material. On the basis of the results, the test set-up and the approachdeveloped here are seen to be useful for practical engineering work and to have potential for more advancedfuture measurements and dynamic models.

16:55 An evaluation of a mixed experimental/numerical method using FRF for the identi-fication of viscoelastic materials (ID 200)Aurelien Moreau, Emmanuel Pagnacco, Didier Lemosse, Dan Borza, INSA de Rouen, France

The work presented deals with a methodology to identify viscoelastic material parameters using a mixednumerical/experimental procedure involving a single test. The identification is based on a non parametricapproach where viscoelastic material parameters are complex, frequency dependent, numbers. Numericalpart of this procedure is based on a specific finite element and a displacement weighted force residue, whileexperimental data is a frequency response field.

64


The aim of this study concerns identification parameters influence on the procedure capabilities. Basedon simulated data, the study deals with sensitivity to noisy measured data and defines parameters boundsallowing optimum identifications. Next, a real structure made from a PVC plate is handled to demonstratethe procedure applicability with the chosen parameters.

17:20 Experimental assessment of a non-destructive method for measuring the elasticproperties of thin isotropic plates (ID 224)Marco Alfano, Leonardo Pagnotta, Gianni Stigliano, University of Calabria, Italy

In the present paper, a vibration based analytical method, for determining the elastic properties of isotropicrectangular plates has been assessed using cold rolled aluminum alloy (6082-T6) thin plate samples. Theprocedure, presented by the authors in a previous work, is based on the measurement of, at least, two ofthe first four resonant frequencies of the samples. It allows to obtain the elastic constants directly by usingtable sets numerically calculated. The uncertainty associated with the procedure are estimated and the resultsobtained are compared with reference values reported in literature.

17:45 Material properties’ identification using full field measurements on vibrating plates(ID 141)Baoqiao Guo, Alain Giraudeau, Fabrice Pierron, Stephane Avril, Ecole Nationale Superieured’Arts et Metiers (ENSAM), France

The paper presents an experimental application of an inverse method leading to the identification of theelastic and damping material properties of isotropic vibrating plates. The theory assumes that the searchedparameters can be extracted from curvature and deflection fields measured on the whole surface of the plate attwo particular instants of the vibrating motion. The experimental application consists in an original excitationfixture, a particular adaptation of an optical full field measurement technique, a data preprocessing givingthe curvatures and deflection fields and finally in the identification process using the Virtual Fields Method(VFM). The principle of the deflectometry technique used for the measurements is presented. First resultsof identification on acrylic plates are presented and compared to reference values. Results are discussed andimprovements of the method are proposed.

18:10 Evaluation of Strain Dependent Material Properties of Hard Coatings (ID 458)Shad Reed, Anthony N. Palazotto, Air Force Institute of Technology, United States of America

A novel vibration experiment consisting of a free-free boundary condition, an electromagnetic excitationsource, a vacuum chamber, and a laser vibrometer based surface measurement system has been developedthat permits high levels of excitation on highly damped specimens with a minimal amount of unwantedsystematic error. While some of the aspects of this experiment are not unique, when combined with anprocessing technique that accounts for the nonlinearities present in the system, this experiment permits, forthe first time, accurate measurement of strain dependent stiffness and damping properties of hard coatingsat high strain levels. This procedure has been demonstrated using a titanium beam that has been coatedwith a free-layer damping treatment of Magnesium Aluminate Spinel. The results indicate that MagnesiumAluminate Spinel has both nonlinear stiffness and damping properties. The stiffness reaches a minimumvalue around 200 microstrain and the damping is a maximum that around 100 microstrain.

65


Vibro-acoustic analysis – VAA1Room 6 – Chairman: F. Marulo

8:50 Impedance Identification out of Pressure Data’s with a hybrid Measurement-Simulation Methodology up to 1kHz (ID 618)Achim Hepberger, Stefan Volkwein, F. Diwoky, ACC Acoustic Competence Center G.m.B.H.,AustriaHans-Herwig Priebsch, ACC Acoustic Competence Center G.m.B.H. / Graz Technical University,Austria

For the vibro-acoustic simulations of interior noise, it is necessary to take the damping behavior of relevantacoustic materials into account. On the one hand, deterministic methods like the well known Finite ElementMethod or the recently developed Wave Base Technique, consider this as a normal impedance boundarycondition. On the other hand, statistical methods like the Statistical Energy Analysis are using absorptioncoefficients, which can be determined for example out of the normal impedance. This paper discusses amethodology to identify the normal impedance out of measured pressure data. For this, a suitable mathemat-ical optimization problem will be introduced by using deterministic simulations methods and optimizationprocesses. First validation results will illustrate the potential of the proposed methodology to identify theimpedance out of measured pressure data.

9:15 Acoustic measurements in a small transmission loss facility (ID 539)Francesco Marulo, Tiziano Polito, Universita degli Studi di Napoli “Federico II”, ItalyAntonio Paonessa, Alenia Aeronautica, Italy

The paper collects the first results obtained in a small transmission loss facility built for acoustic measure-ments on coupons of composite panels candidate for aircraft fuselage applications. The use of compositematerials is employed for their better structural efficiency (strength over weight) than metallic alloys, whichhave played, so far, a predominant role in aeronautical application, mainly for commercial airplanes. Theselection of candidate materials for aeronautical application, however, does not have to take into account thestructural efficiency alone, but many other characteristics such as the fatigue behaviour, the environmentalinteraction, the manufacturing costs and so on, just to name a few. One of the additional item parameterwhich arises to an important factor is the passenger interior comfort, considering both environmental para-meters as well as vibration and acoustic main figures. In view of continuing improved knowledge of thedynamic characteristics of composite panels, both numerical simulation and laboratory testing are the neces-sary elements which are the backbone of the mentioned research program. One of the aspects pursued in ourDepartment has been the development of a small transmission loss facility for testing different solutions andproposals of treatments on composite panels. Several advantages are linked to a small transmission TL facil-ity in comparison to the standard big facilities, even if much care must be used when dealing with results inthe low frequency region where such facilities exhibits their main drawbacks. The paper will present the newsmall TL facility developed in our Dept., including the possibility to consider low temperature testing, theinitial qualification tests and some preliminary results on coupon panels used for the first testing campaign.

9:40 In-situ estimation of acoustic impedance on the surfaces of a room for inverse soundrendering (ID 460)Gabriel Pablo Nava, Yosuke Yasuda, Yoichi Sato, Shinichi Sakamoto, University of Tokyo, Japan

We present a method based on the inverse boundary element method (IBEM) for the in-situ estimationof normal acoustic impedances of the surfaces in a room. As input to the inverse process, this techniqueuses the geometry of the interior space, the position and strength of the sound source, and a set of soundpressures measured at random positions in the acoustic field. Previous work in inverse radiation problemshas employed computational expensive techniques such as singular value decomposition (SVD) in orderto obtained satisfactory results. In contrast, the numerical simulations of the present work show that the

66


least-squares approach can give satisfactory solutions when the linear system of the formulation derivedhere is sufficiently overdetermined with sound samples taken at random locations in the interior. The use of3D stereo vision tracking is suggested to enable the practical measurement of large sets of field pressures.Currently, inverse sound rendering for virtual reality (VR) is a targeted application of our method.

10:05 Calibration of the two microphone transfer function method to measure acousticimpedance in a wide frequency range (ID 540)Rene Boonen, Paul Sas, Wim Desmet, Walter Lauriks, Gerrit Vermeir, Katholieke UniversiteitLeuven, Belgium

In many acoustic simulations, particularly when using lumped parameter models or electrical analog cir-cuits, the acoustic impedance of a component needs to be determined accurately. A widely used acousticimpedance measurement method is the ”two microphone transfer function method”, which is standardized inISO-10534-2. When the acoustic impedance is needed over a wide frequency band, for example from 10Hzto 10kHz, this method faces some limitations. In this paper, a new calibration method will be proposed suchthat acoustic impedances can be measured with high accuracy over a wide frequency range. The estimationof the speed of sound has been eliminated. Using the measured transfer functions between the two sensorsat two different reference sections, the sensor positions will be accurately calibrated. The calibration of thesensor mismatch has become superfluous, so interchanging positions of the sensors is not necessary. A recur-sive procedure has been proposed to maximize the microphone position accuracy. The resulting calibrationprocedure has been reduced to the accurate determination of the sensor positions.

Damping – D1Room 6 – Chairman: A. Carcaterra

10:55 Mode based prediction of vibrations in highly damped structures (ID 399)Krister Dovstam, Dovstam Innovation, SwedenPeter Goransson, KTH Kungliga Tekniska Hogskolan, Sweden

Traditionally, mode based techniques are used for prediction of sound and vibrations. This is motivatedby the possibility of using reduced modal models instead of very detailed FE models. For highly damped,composite structures the effect of reduced model size on the accuracy of predicted sound and vibration lev-els is often not known. Problems connected with application of modal methods towards highly dampedmaterials, systems and multilayer treatments using conventional modal vibration theory have recently beenhighlighted by the authors. In the present paper a simple, damped Oberst beam example is presented, nothitherto satisfactorily treated by modal techniques. It is pointed out that the contact forces at internal inter-faces between different materials have to be treated in order for a modal solution to be found. It is also shownthat non-physical tractions at free unloaded boundary surfaces have to be eliminated.

11:20 On combined viscous and dry-friction modelling in kinematically excited vibrationisolation systems (ID 78)George Juraj Stein, R. Zahoransky, Institute of Materials and Machine Mechanics of Slovak Acad-emy of Sciences, Slovakia (Slovak Republic)

The dry-friction influence on the properties of mechanical system is of permanent interest. The paper dealswith kinematically excited single DOF oscillatory system, incorporating both viscous damping and dryfric-tion. By means of simulation the combined viscous damping and dry-friction force influence on systemproperties will be analysed, e.g. the influence on the acceleration transmissibility course. Based on com-parison of field measurements and simulation research of models of different complexity a suitable modelwill be developed and analysed. These results will be used also for simulation analysis under random kine-

67


matical excitation. Reasonable agreement between simulation results and real experimental data from fieldtests has been attained, which illustrates the suitability of this approach in engineering practice. Moreoveran approach to dry-friction force value optimisation will be shown, giving the designer a further mean forvibration control systems optimisation.

11:45 Experimental investigation of the dynamic performances and durability of a col-loidal damper (ID 51)Claudiu Valentin Suciu, Fukuoka Institute of Technology, Faculty of Engineering, Japan

Dynamic performances and durability of an absorber based on the surface dissipation (colloidal damper)are experimentally investigated. Dynamic contact angle hysteresis induces the energy loss, which occurswhen water is forced to penetrate and then naturally exudes from a nano-porous silica gel, modified tobecome water-repellent. Being oil-free, the colloidal damper might be considered as an environmentfriendlyapplication of nanotechnology in the field of mechanical engineering. Test rig represents a compression-decompression chamber, which allows dynamic and endurance experiments under a certain temperature(-10 50 ◦C) and frequency (0 10 Hz), selected according with specific applications (vehicle suspensions,anti-seismic dampers, etc). Influence of the temperature, piston frequency, pre-pressurization and maximumapplied pressure on the energy loss is found for different types of silica gels. Endurance tests are performedand measures to increase the durability of the colloidal damper are discussed.

12:10 Dynamic Properties and Damping Prediction for Laminated Plates (ID 258)Bohdan Diveyev, Lviv National Polytechnic University, UkraineMalcolm Crocker, Auburn University, United States of America

Numerical evaluations obtained for vibrations in isotropic, orthotropic and composite laminated plates havebeen used to determine the efficient displacement field for economic analysis of vibrations in laminatedcomposite plate. The numerical method developed follows a semi-analytical approach with analytical fieldapplied in longitudinal direction and layer-wise displacement field employed in transverse direction. Iden-tification of elastic properties of laminated plates from the measured eigenfrequencies has been performed.Elastic constants of laminates have been determined by using an identification procedure based on experi-ment design, and multi-level theoretical approach.

Multi-body dynamics and control – MB1Room 6 – Chairman: H.-H. Priebsch

14:00 A Geartrain Model for the Dynamic Analysis of a Motorbike Timing System (ID 73)Alessandro Rivola, University of Bologna, ItalyMarco Milandri, Emiliano Mucchi, University of Ferrara, Italy

This work deals with the elastodynamic model of the geartrain that drives the cam-valve mechanisms of amotorbike engine timing system. The gear torsional vibrations are simulated by means of a lumped para-meter model which takes into account the parametric excitation due to the time-varying meshing stiffness,the tooth separation phenomenon and the lubricant squeeze effect between approaching teeth. In order toget insight into the dynamics of the timing system, and to help the development of the geartrain elastody-namic model, several experimental tests were carried out. The experimental procedures are here illustratedtogether with the signal processing techniques. The comparison between numerical results and experimentalmeasurements shows that the effectiveness of the model is satisfactorily assessed. The model can predictand interpret the actual dynamic behaviour of the system; therefore, it can be used as a tool for the designoptimisation of the motorbike timing system.

68


14:25 Simulation of Engine Dynamics by using MBS/FEM Composite Structures (ID 361)Harald Riener, W. Witteveen, M. Fisher, MAGNA Powertrain, Engineering Center Steyr GmbH &Co KG, Austria

The main aim in engine development is to implement new, optimized engine concepts regarding function,design space, weight and costs in very short development times. The trend towards increasing power densityby means of supercharging, high speeds and downsizing leads to higher load on crank train componentswith a reduction of load reserves. Fail-safe design and optimization of these high-load components can besignificantly promoted by the efficient use of transient simulation technologies based on MBS algorithmsusing elastic structures. This paper demonstrates as a primary application the dynamic analysis of an AUDIAG V8 crank train based on a run-up simulation. The aim is to demonstrate the great potential of this methodby showing interesting possibilities using optimization tools (e.g. FE-Design’s TOSCA) to the acousticssimulation of the complete engine-gearbox assemblies.

14:50 Suspension system optimisation of an articulated dump truck to reduce whole bodyvibration (ID 564)Johannes C. Kirstein, Johannes L. van Niekerk, Stellenbosch University, South Africa

Drivers of articulated dump trucks (ADTs) are exposed to whole body vibration levels that exceed the newEU whole body vibration legislation action values in most cases and sometimes the limit values. Thispaper presents a reduced order simulation model used to optimise the passive suspension systems of aforty ton ADT to reduce the vibration exposure of the driver. A 24 degree-of-freedom, three dimensionalADAMS/VIEW vehicle simulation model with linear parameters was developed and compared to experi-mental data as well as with simulation results from a more complex 50 degree-of-freedom ADAMS/CARmodel. The ADAMS/VIEW model correlated in some aspects better with the experimental data than thehigher order ADAMS/CAR model and was therefore used in the suspension system optimisation study. Aroad profile was generated from a spatial PSD (power spectral density) of a typical haul road. The weightedRMS (root mean squared) and VDV (vibration dose value) values were used as the objective function forthe optimisation study. The optimisation was performed by four different algorithms and an improvementof 30% in ride comfort for the worst axis was achieved on the specified haul road. The improvement wasrealised by softening the struts and tires and hardening the cab mounts. The results were verified by simulat-ing the optimised truck on different road surfaces and comparing the relative improvements with the originaltruck’s performance. In all cases the optimised suspension resulted in reduce whole body vibration exposureof the driver.

15:15 Flexible Multi-body Dynamics Simulation - A Powerful Method for Prediction ofStructure Borne Noise of Internal Combustion Engines (ID 577)Gunter Offner, AVL List GmbH, AustriaHans-Herwig Priebsch, ACC Acoustic Competence Center G.m.B.H. / Graz Technical University,Austria

The numerical simulation of contacting engine parts is a challenging target during the engine developmentprocess. Contrary demands like increasing use of light weight materials and tighter limits of admissiblestress and strain increase the difficulty of designing low vibrating and low noise engines. The appropriatesimulation tool has to meet conflicting demands in efficiency and accuracy, in addition. Moreover the predic-tion method must be capable to represent engines with arbitrary size and application area at all operating andloading conditions. Various excitation effects have to be covered. Besides the combustion, the excitation thatresults from impacts of body surfaces is an important source of structure borne noise and makes an accuraterepresentation in a simulation methodology necessary. In addition to the vibro-acoustic excitation, these oillubricated contacts between the surfaces mainly influences also friction and wear.

This paper outlines a flexible multi-body dynamics simulation methodology that considers linear elasticbodies considering highly non-linear contacts. The contact models are both non-linear spring damper ap-

69


proaches and advanced mixed lubrication models that consider effects of mass, elasticity, surface roughnessand hydrodynamics. The numerical integration of the resulting non-linear mathematical model is done intime domain.

The first part of the paper discusses the mathematical formulation of the flexible multi-body simulationmodel. Both the representation of the linear elastic bodies and the highly non-linear contacts are outlined.In the second part, the paper focuses on the comparison of simulation results and measurement results toshow the wide range of applications and the achievable result quality. In particular results on piston slapexcitation, valve train and timing drive excitation, crankshaft motions, engine mount vibrations and structureborne noise simulation considering all excitation effects are considered.

Vibro-acoustic modelling and prediction – VAM1Room 6 – Chairman: A. de Boer

16:05 Orthogonal radiation patterns for three-dimensional noise sources based on thespherical harmonics of sphere(s) (ID 365)Simon Hill, Katholieke Universiteit Leuven / Flanders Mechatronics Technology Center, Belgium

The control and thus sensing of radiation from three-dimensional noise sources has become a topic of stronginterest as investigation moves away from the laboratory. Furthermore, the advance in signal processingpower has facilitated the real-time fusion of a large number of sensor signals, which are typically required togive an accurate estimate of global error signals. Current state-of-the-art designs have typically been basedupon a BEM or FEA analysis, which are as complicated as the structure under investigation. However, theexact nature of a global set of acoustic orthogonal patterns and their practical measurement from large three-dimensional sources remains elusive. In this work we examine the spherical vibration modes of sphere(s)as a basis function for describing orthogonal radiation patterns with respect to radiated sound power. Thespherical harmonics form an orthogonal set of vibration patterns on the surface of the sphere, which we showcan be used to formulate a set of orthogonal (to sound power radiation) sound pressure patterns. The novelwork contained in this paper has two aspects; a development of orthogonal radiation patterns based on morethan one sphere; and a development based on spheres in both the acoustic free field and a half-space. Theresults have implications in the area of acoustic modelling and also in the area of sensing system design foractive noise control.

16:30 Robust Acoustic Vehicle Body Design Using Evolutionary Algorithms (ID 339)Andreas Kropp, BMW AG, Germany

The numerical evaluation of interior acoustic responses is currently a well-established tool in vehicle bodydevelopment. However, as recently as five years ago, an acoustic evaluation of a simplified simulation modelrequired on the order of a couple of weeks of computation time on a supercomputer.

Today, complex FE-simulations which have been successfully validated with test data can run overnight ona Linux cluster. This dramatic acceleration in computational speed has only been possible through the use ofthe Automated Multi-Level Substructuring (AMLS) algorithm, running on a parallel-processing machine.

This capability now available for acoustic design suggests an automatic design improvement process inacoustic body design, analogous to the numerical gradient-based optimization algorithms, which have beenwidely used in the past in structural design. However, such classical optimization strategies implemented foracoustic level reduction have generally failed because of the highly nonlinear boundaries of the acoustic de-sign space, as discussed by the author in an earlier paper. Alternatively, the use of self-adaptive multi-criteriaevolutionary algorithms has proven very successfully. The resulting Pareto front optimization algorithm,discussed in this paper, has been successful because it takes into account the relation between acoustic levelreduction and additional mass.

70


Two factors however complicate the acoustic optimization process considerably: Firstly, Evolutionary Algo-rithms require many hundreds of evaluations with a total turn-around time of several weeks. Secondly, anoptimized acoustic structure does not generally constitute a robust design.

This paper discusses one approach, currently being used by BMW to overcome these obstacles. The ineffi-cient computational performance is being circumvented using new numerical approaches such as the modalcorrection method (MCM). And a robust acoustic optimization is being realized through a multi-criteriaoptimization, which uses the additional objective function requirement of minimizing the acoustic levelvariance. Implementation of both these approaches together can be used to achieve a robust optimizedacoustic vehicle design virtually overnight.

16:55 Transmission Loss analyses on fuselage panels: approach, numerical results andvalidation (ID 52)Frank Grooteman, National Aerospace Laboratory (NLR), The Netherlands

In this paper a new solution strategy will be presented to solve structural-acoustic problems in a more effi-cient way. Both the structure and fluid domain are modelled with the Finite Element Method (FEM). Thenew approach reduces the computational effort compared with the more conventional FEM approach, byintroducing non-coinciding fluid and structure mesh and separate solution of both domains, but in a fullycoupled fashion. This is necessary in case of large models, existing for realistic aircraft structures such asfuselage panels or a fuselage barrel, which are the type of structures analysed within the framework of theEuropean programme FACE (Friendly Aircraft Cabin Environment). One of the main objectives within thisproject was to predict the sound transmission through aircraft fuselage panels. The solution strategy hasbeen implemented in the in-house finite element program B2000. Numerical results obtained for a largeAluminium fuselage panel are presented and compared with experimental data (modal and transmissionloss data) to validate the approach. The following panel configurations have been analysed: bare, damped(viscous constraining layer) and furnished (glass wool).

17:20 Fluid-structure interaction applied to ovalling oscillations of silos (ID 555)David Dooms, Guido De Roeck, Geert Degrande, Katholieke Universiteit Leuven, Belgium

Circular cylindrical shells, like silos, are perceptible to wind induced ovalling oscillations, an aeroelasticphenomenon, where the cross section deforms as a shell without bending deformation of the longitudinalaxis of symmetry. A fluid-structure interaction analysis aims to predict the ovalling onset flow velocity.An approximate analysis will be performed by reducing the structure, using the finite strip method, to twodimensions and by coupling it with a two dimensional flow.

First, the mode shapes and eigenfrequencies of a three-dimensional finite element and a finite strip model ofthe silo structure are compared.

A transient 2D turbulent air flow around a single silo at a Reynolds number of 1.24 x107 is computed usingthe SST turbulence model and the results are compared with the pressure coefficients in Eurocode 1 and withexperimental data. Unsteady simulations are performed for the flow around a group of 8 by 5 silos. Thegroup configuration drastically changes the time-averaged pressure distribution around the silos.

The fluid and the structure are sequentially coupled, using interfield iterations to fulfill equilibrium and con-servation of energy on the interface. The coupling procedure is validated by means of available experimentalresults of wind tunnel tests. Preliminary results of a transient fluid-structure interaction calculation at a windspeed of 7 m/s are reviewed.

71


17:45 A fast interpolation method for multi-frequency acoustic analysis by boundary ele-ment method (ID 93)Xianhui Li, Dalian University of Technology, China

A fast interpolation method based on rational Krylov projection is proposed to overcome the computationaldifficulty in multi-frequency acoustic analysis by boundary element method. It recasts acoustic quantitiesinto transfer functions of linear time-invariant systems. Then a matrix-free method is employed to reducethe original systems in the frequency band of interest. Interpolation of the acoustic quantities is realizedby the corresponding transfer functions of the reduced order models. Different interpolation schemes arepresented to investigate acoustic radiation of a pulsating sphere. Numerical results demonstrate that theproposed method can significantly speed up the multi-frequency acoustic analysis with guaranteed compu-tational accuracy.

18:10 Vibro-acoustic modelling of a rail vehicle bodyshell and resilient Floor (ID 48)Man Liu, X. Lu, A. Barlow, Areva T&D Technology Centre, United Kingdom

This paper presents the stages of modelling and testing of a rail vehicle resilient floor to determine itsstructure-borne noise performance. A series of finite element, boundary element, and statistical energyanalysis models have been developed using a suite of proprietary software packages including ANSYS,SYSNOISE and SEADS. Coupled cavity FE models have been created and several iterations of acousticanalyses have been conducted. Modal tests and analyses were also undertaken for use in the developmentand validation of the finite element models. By performing a sensitivity study the significance of certaindesign features have been evaluated. A number of SEA analysis have been performed and the simulationresults are compared against the test result for a particular bodyshell and floor configuration.

Vehicle noise and vibration (NVH) – NVH1Room 7 – Chairman: F. Augusztinovicz

8:50 Measurement of vibration and noise of railway wheels on track (ID 87)Ludek Pesek, Frantisek Vanık, J. Vesely, Jan Cibulka, AS CR Institute of Thermomechanics, CzechRepublicTomas Bohac, Bonatrans a.s., Czech Republic

In the contribution a method of measurement and dynamic analysis of the recent operational tests of themonobloc wheel with/without damping elements is presented. Dynamic behavior of the wheel is describedon basis of time-frequency analysis of its operational vibration and experimental modal analysis. The vibro-acoustic coupling is analyzed from the comparisons of vibration and sound pressure spectrograms. At pas-sage the bow track, the formation and modal-frequency composition of squealing noise is observed and thereare identified vibration modes that mostly contribute to noise emission into surrounding. For quantitativecomparison of individual wheel solutions the third octave analysis is used.

9:15 Energy minimization criterion as a methodological guide to the dynamic analysisand optimization of fully trimmed and equipped vehicles. (ID 162)Sara Lorea, Francesca Avenati Bassi, F. Tinti, Rieter Automotive Fimit, Italy

Customer demand for increased vibro-acoustical comfort performances of vehicles, coupled with a reduceddevelopment time frame, requires a continuous improvement process in the engineering design phase. Theseneeds call for fully comprehensive FEM models, in order to simulate vehicle performance from the very earlystages of the project: for this purpose, a huge amount of detailed information is necessary, as the “virtualvehicle” should be as close as possible to reality and thus complete of trim and equipment. The validatednumerical model should be used in an efficient and systematic way in order to outline criticalities, find out

72


the important transmission paths and optimise the performances: an innovative approach for this goal hasbeen conceived, focussed on total energy distribution and minimisation criteria as a guideline in the selectionof crucial structural improvements. The paper describes the methodology applied during a development of anew D-segment car for the Asian market.

9:40 New NVH optimization procedure applied to weight reduction of an automotiveengine (ID 177)Stephanie Malcuy, Renault SAS, FranceStephane Perrin, ANSYS France SAS, France

NVH is one of the most important customer requirements integrated at each step of the design of a newengine, and obviously, the weight is an important constraint.

As NVH simulation procedure is able to represent the global behavior of the powertrain (for structural bornnoise, and also for radiation noise), this paper shows a way to implement a new methodology of optimization,based on a variational approach. This innovative method realizes an optimization on physical parameters,but also shapes and topologic parameters.

Various studies are described, corresponding to different steps of a new powertrain project, and also differentrequirements of the design office : general guidance of new powertrain, detail drawing purposes, optimizationof acoustic performance, but always with only one object : reducing the weight of the powertrain.

One strength of variational technology is to open up a host of possibilities for design, through multiple com-binations that cannot be processed by classical computation technologies. Besides, variational technologyis adapted to existing computation processes, i.e. computation tools and models used in industry. Thus, theresults of each study are usable directly by design offices.

Such analysis improves the know-how of RENAULT and contributes to design ever optimal structures forthe future generation of 4 in-line cylinders and V6 cylinders Diesel engines within a minimum number ofiterations.

10:05 Noise sources balancing on vehicle development to improve customer satisfaction(ID 456)Tatiana Cristina Giorjao, Elson Lima Albuquerque, Andre Luis Cherman, Ford Motor Company,Brazil

Noise Vibration and Harshness (NVH) is one of vehicle attribute that most impact customer satisfaction.In this area three main noise and vibration sources must be considered: powertrain (due to the engine,transmission, differential and respective components), road NVH (due to vehicle roll over kinds of surfaces,tires, suspension system and body excitations) and wind noise (due to air flow around the vehicle surface).These sources may be treated independently for components development, but customer will have a globalperception. The objective of this work is to define a manner to understand what customer feels regardingNVH, introducing the concept of interior quietness to characterize the comfort sensation. The desired interiorquietness will be achieved by balancing the 3 main sources contribution. Due to facilities, timing and productchosen, only road NVH and wind noise were considered on this first assessment. Data was acquired withartificial head at front passenger seat. Facilities used include wind tunnel and different road surfaces onproving ground.

73


Vehicle noise and vibration (NVH) – NVH2Room 7 – Chairman: F. Augusztinovicz

10:55 Investigation on airborne and structure borne balance for vehicle interior noise (ID359)Francesco Sbarbati, Massimo Martorelli, L. Petrella, P. Liccardo, ELASIS S.C.p.A., Italy

In order to make mass production cars more comfortable, the manufacturers are searching ways to isolate thepassenger compartment as much as possible from the noise and vibrations produced by the internal combus-tion engine. For this purpose, it is useful to know the share of airborne noise and the share of noise that goesinto the passenger compartment through the structural path. This paper describes the research concerningthis matter.

To achieve this goal, a test procedure has been set up. Sound pressure measurements were performed on anormal production car during a run up at wide open throttle. This was then repeated with an acoustic shieldaround the engine made for this purpose (its effectiveness in lowering the airborne noise was also evaluated).In this way, the sound pressure level measured in the passenger compartment was only the amount transmittedthrough the structural path. The results for the vehicle tested are shown at the end of this paper.

Basically, it was proved that the structure borne noise is responsible for noise and vibrations in the passengercompartment up to about 600-700 Hz.

11:20 A new method for aircraft noise synthesis (ID 373)Dries Berckmans, Katholieke Universiteit Leuven, BelgiumKarl Janssens, LMS International, BelgiumPaul Sas, Wim Desmet, Katholieke Universiteit Leuven, BelgiumHerman Van der Auweraer, LMS International, Belgium

This paper presents a method to synthesize aircraft noise. The method provides designers with a tool forsound quality evaluation with which target sounds for future aircraft design can be developed. Quick andeconomic evaluations concerning the quality of sounds of different design alternatives or improvements onexisting aircraft become possible. The method has the potential to become a crucial tool in the determinationof the primary factors that determine the quality of aircraft sound.

11:45 Automatic extraction of noise annoyance features from vehicle run-up sounds (ID423)Karl Janssens, Antonio Vecchio, Herman Van der Auweraer, LMS International, Belgium

An automatic noise annoyance detection algorithm was developed to extract resonances, masking effects,order non-linearities, booming phenomena and amplitude modulations from an in-vehicle run-up sound andto visualize these noise annoyance features on top of the time-frequency spectrogram of the sound. This al-gorithm is very useful in closed loop with a Virtual Car Sound (VCS) synthesis tool. Based on the identifiednoise annoyance features, well-oriented sound modifications can be applied in VCS untill a lownuissancetarget sound is designed. An important benefit of this closed-loop approach is that various sound modifica-tions can be objectively characterized and assessed in a short period of time without the need for extensivejury testing.

12:10 Investigation of binaural transfer functions; reciprocity in free field and in-vehicle(ID 584)Peter van der Linden, Patrick Van de Ponseele, Gunther Coppens, LMS International, Belgium

The investigation of the noise contributions, transfer paths, has become a standard procedure both for diag-nosis of noise effects and for target setting in product development. The combination of operational testsand (vibro) acoustic transfer function measurements allows insight in noise contributions, interface forces

74


and system sensitivities. With the extension to binaural recording of the operational sounds and transfer pathanalysis based binaural noise synthesis of noise contributions the impact of systems and interfaces on theperceived sound quality has become technically possible.

Correct binaural reproduction/synthesis of noise contributions requires correct binaural transfer functionmeasurements (along with the other system measurements). The paper discusses the investigation of binauraleffects on transfer function measurements in relation to reciprocity. Volume acceleration sound sourceshave been built into a head-torso simulator to investigate such. Measurements on persons are compared tosimulator measurements. And the effects in free field condition are compared to persons and simulators in avehicle.

Vehicle noise and vibration (NVH) – NVH3Room 7 – Chairman: P. Sas

14:00 An efficient approach to include interactivity in the vehicle NVH evaluation process(ID 266)Roger Williams, Mark Allman-Ward, Sound and Vibration Technology, United KingdomKevin Bernard Ginn, Dmitri Tcherniak, Bruel & Kjaer Sound & Vibration Measurement A/S,Denmark

Interactive evaluation of vehicle NVH data using an NVH Simulator offers two major benefits over the tra-ditional methods of listening to fixed sounds or examining numerical or graphical data. Firstly it enables abroad range of assessors to make confident judgments about the sound quality of real or virtual cars. Theseinclude non-expert decision makers and real vehicle customers as well as NVH experts. Secondly, a sig-nificant part of the full driving envelope, ranging from idle, through part load and constant speed cruisingto full throttle sweeps can be efficiently evaluated under free-driving conditions using a single data model.Furthermore, by expanding the model to include sound sources and physical component contributions, it canbe used for cascading targets and developing or validating new designs. This paper describes an approach forminimizing the time, resources and techniques needed to build an interactive NVH Simulator model withoutcompromising its validity or usefulness for specific applications. This ensures that the significant benefitsoffered by using such a model in a vehicle development programme are supported by an efficient and costeffective data collection and preparation strategy.

14:25 Test setup for tire/road noise caused by impact road excitations: first outlines (ID368)Peter Kindt, Filip De Coninck, Paul Sas, Wim Desmet, Katholieke Universiteit Leuven, Belgium

This paper describes the design and first experimental results of a novel test setup to measure the impactresponse of a rotating tire. The test setup is based on a tire on tire principle, which requires less space than aclassic drum (rigid rotating cylinder). The setup is used to analyse mechanisms of tire/road noise during roadimpact excitations, such as a cobbled roads, joints of a concrete road surface, railroad crossings,... The noisethat is radiated from a tire strongly determines the passenger acoustic comfort and the traffic noise. Besidesthe radiated noise, the spindle forces induce vibrations that generate noise in the passenger compartment.Typical for a road impact excitation are a short duration tactile spike (shock) and a short duration noise spike(impact boom) experienced by a passenger. A series of test are performed with different driving speeds,cleat dimensions and inflation pressures. The test results are used to identify and quantify the mechanismsof noise generation during impact excitations with respect to the different parameters. These analyses willcontribute to the future development of tire models to simulate the acoustic and structural response to roadimpact excitations.

75


14:50 Towards a robust optimization of spot weld design in automotive structures (ID 416)Morvan Ouisse, Scott Cogan, University of Franche Comte, FranceBenoıt Guillaume, PSA Peugeot Citroen, France

The design of a resistance spot weld distribution for an automobile body-in-white has a strong impact notonly on the global system performance but also on the robustness of this performance with respect to uncer-tainties due to assembly defects and fatigue failures. A quantitative methodology is presented that providesdecision-making indicators that allow the analyst to insure a given level of system performance at the cost ofperforming a quality control of a limited number of welds coming off the assembly line as well as reinforcinga set of critical welds in order to improve the robustness to fatigue failure. In contrast with existing sample-based robustness analyses, the proposed methodology gives visibility to the compromise between improvedrobustness and higher assembly and quality control costs. In other words, this methodology provides a toolto guide the analyst in the next step to improving robustness while giving an estimation to the cost of thepredict ed improvement. All examples are presented on a full BIW structure (1,000,000 dofs and 6,500 spotwelds).

15:15 The influence of suspension and tyre dynamics on Driveline NVH (ID 369)Nawazish Zaidi, Romax Technology, United KingdomAtanas Popov, Seamus Garvey, University of Nottingham, United Kingdom

The automotive driveline is a source of Noise, Vibration and Harshness (NVH) in the vehicle caused by anumber of dynamic interactions. The interaction of suspension-tyre dynamics with the driveline is investi-gated through a full vehicle model development and excited by road irregularities. The integrated systemsmodelling techniques are employed to assemble the full vehicle model from its various subsystems. It isshown that the dynamic behaviour of any component within the system can be predicted and the complexinteractions are highlighted. The vibration responses at key locations are analysed. It is found that the fre-quencies of subsystems can change with each subsystem connection. In vertical vehicle dynamics, only thelow modes of vibration due to suspension and tyre vertical resonances affect the driveline behaviour.

15:40 Transmission Loss Modeling of Trimmed vehicle Components (ID 514)Abderrazak Omrani, Lassen Mebarek, Mohamed A. Hamdi, ESI Group, France

This paper presents a numerical BEM-FEM model allowing calculation of the acoustic Transmission Loss(TL) of a double walls trimmed panels of arbitrary shape, involving in particular the use of porouselasticmaterials (foams, fibers, etc..) between structural outer panels with or without air gaps. The resultingTransmission Loss module is implemented in RAYON-VTM (Vehicle Trim Modeler) software developed byESI-Group. Numerical results are given in the particular case of double wall flat aluminum panels enclosingporous-elastic layers and air-gaps.

Aeroacoustics and flow noise – AA1Room 7 – Chairman: C. Schram

16:30 Wind turbine noise from an industrial point of view (ID 351)Gabriel Alvaro Matesanz, Gamesa Eolica, Spain

This paper deals with the description of the activities in which the wind turbine industry is taking part toreduce the noise emission of future wind turbines. The noise produced by wind turbines could risk the de-velopment of this technology in populated areas and it has become one of the most important environmentalimpacts of wind energy. The numerical study of the wind turbine behaviour is one of the most challengingproblems a designer could face. The turbulent nature of the incoming flow with sudden changes of directionand velocity, the presence of the terrestrial boundary layer, the wind turbine elastic deformation and the mov-

76


ing geometry makes the problem difficult to solve. In this context the only reasonable alternative for industryis to find semi-empirical methods that could be applied to reduce noise with the time constraints imposedby the wind energy market. The beginning of this paper introduces the most relevant noise sources in thewind turbine and their theoretical treatment, then the description of the strategies that are used to reducenoise levels are discussed, and finally the current research projects where the industry is participating withthe objective of noise reduction are presented.

17:20 CAA as a Postprocessing-Step within CFD-Computations (ID 435)Manfred Kaltenbacher, Max Escobar, Christian Hahn, Irfan Ali, Stefan Becker, Friedrich-Alexander-University of Erlangen-Nuremberg, Germany

We propose a hybrid approach for the computation of flow induced sound by applying a Large Eddy Sim-ulation (LES) to the flow field and solving the inhomogeneous wave equation according to Lighthill withina postprocessing step. The inhomogeneous part of the wave equation is already calculated during the flowcomputation on the fine fluid grid and the so called acoustic nodal loads are saved for each time step. In apostprocessing step, we transform these acoustic nodal loads to the frequency domain, perform an interpola-tion to the acoustic grid and compute for each frequency of interest the sound pressure field by applying theFinite-Element-Method (FEM).

17:45 Influence of geometry on the generation of aeroacoustic noise (ID 436)Christian Hahn, Manfred Kaltenbacher, Reinhard Lerch, Stefan Becker, Franz Durst, Friedrich-Alexander-University of Erlangen-Nuremberg, Germany

The flow induced tonal noise of a square cylinder in cross flow and the aeroacoustic broadband noise of aforward facing step are investigated in an aeroacoustic wind tunnel. Both cylinder and step are varied ingeometry and the effect on the generated sound is studied. In the case of the cylinder the lowest amount oftonal noise among the subset of geometrical variations presented here is found when a short wedge is addedin front of the cylinder. Concerning the forward facing step already significant noise reductions are obtainedwhen the edge of the step is slightly rounded.

18:10 On the Use of Filtering Techniques for Hybrid Metods in Computational Aero-Acoustics (ID 640)Wim De Roeck, Gustavo Rubio, Wim Desmet, Katholieke Universiteit Leuven, Belgium

Hybrid CAA-approaches are commonly used for aeroacoustic engineering applications. In this kind of com-putational techniques, the numerical domain is split into a noise generating region, where an aerodynamicfield generates the acoustic sources, and an acoustic propagation region. Nowadays a large variety of hy-brid approaches exist differing from each other in the way the source region is modeled; in the way theequations are used to compute the propagation of acoustic waves in a non-quiescent medium; and in theway the coupling between source and acoustic propagation regions is made. The coupling between sourceand propagation region is usually made using equivalent sources (acoustic analogies) or acoustic boundaryconditions (Kirchhoff’s method). For certain applications both coupling approaches tend to give erroneousresults: acoustic analogies are inaccurate if the acoustic variables are of the same order of magnitude as theflow variables, which is the case for flow-acoustic feedback phenomena such as cavity noise or when acousticresonance occur which happens for duct aeroacoustics applications; acoustic boundary conditions are sen-sitive to hydrodynamic pressure fluctuations when a vortical flow passes through the Kirchhoff’s surface.These inaccuracies can be avoided by using appropriate filtering techniques where the solution in the sourcedomain is split into an acoustic and a hydrodynamic part. This paper illustrates the need for such filteringtechniques for CAA-applications and starts with the theoretical development of a new filtering techniquebased on an aerodynamic-acoustic splitting.

77


Maduse – MAD2Room 8 – Chairman: P. Lardeur

8:50 Verification and validation of finite element models for the vibro-acoustic behaviorof a windscreen in presence of variability (ID 119)Pascal Lardeur, Renault / Universite de Technologie de Compiegne, FranceMarco Scionti, R. Scigliano, Renault, France

The objective of this paper is to apply the verification and validation methodology in order to assess thecapability of finite element models to predict the natural frequencies of an acoustic windscreen, in presenceof intra variability. In this case, variability of polymers’ elastic properties due to temperature variation leadsto variability of the frequencies. For the verification stage, the main result is that the multilayer shell modelapproach is valid at low temperature, when polymers are relatively stiff. On the contrary, at high tempera-ture, polymers are very flexible and shell models lead to significant errors because transverse shear effectsare considerable. Finally, a solid finite element model is necessary. For the validation stage, two validationmetrics are proposed. They assess respectively the mean value and the standard deviation of the frequencies.The solid finite element model leads to very satisfactory results for the mean value of the frequencies. Thegeneral trends of the experimentally observed intra variability are also well reproduced. Nevertheless, fur-ther investigations are necessary to improve the predictive capability of the numerical model that currentlyunderestimates the experimental intra variability.

9:15 Experimental and numerical study of the intra/inter variability of an acoustic wind-screen (ID 120)Marco Scionti, Renault, FrancePascal Lardeur, Renault / Universite de Technologie de Compiegne, France

The European project MADUSE deals with the emerging problem of variability and uncertainty in the struc-tural dynamics engineering. As to variability, the notions of intra and inter variability can be introduced todefine 2 phenomena. The intra variability describes the differences of the dynamic response of a physicalstructure due to environmental changes. The inter variability refers to the variation of the dynamics of nom-inally identical systems due to unavoidable inaccuracies of the manufacturing process. Next to variability,the uncertainty is related to the lack of knowledge on the exact values of the properties to introduce in thenumerical model during the product virtual design process. In this framework, the paper focuses the attentionon the study of the dynamics of an acoustic windscreen. An experimental and a numerical study are carriedout to verify whether the system is characterized by an intra/inter variability and to assess to which extentsuch behaviour may be predicted through numerical simulations.

9:40 An application of response surface methodology in the field of dynamic analysis ofmechanical structures considering uncertain parameters (ID 445)Alberto Gallina, Adam Martowicz, Tadeusz Uhl, AGH - University of Science and Technology,Poland

The paper presents an application of coupling between Response Surface Methodology and Monte CarloSimulation in the field of dynamic analysis of mechanical structures. The error introduced by using anapproximated metamodel instead of the real model for the Monte Carlo Simulation is here analyzed and asolution is proposed to overcome the problem.

78


10:05 Analysis of the uncertain dynamic behavior of an automotive control unit using thecomponent mode transformation method (ID 430)Oliviero Giannini, University of Rome “La Sapienza”, ItalyUte Gauger, Michael Hanss, Universitat Stuttgart, Germany

The component mode transformation method is a novel approach, developed to analyze the dynamics ofsystems affected by uncertainties in their dynamic parameters. This method provides a reliable improvementof the computational performance of the original transformation method, which represents a practical tool togenerally analyze systems with uncertain parameters. The uncertain parameters of the system are modeledby fuzzy numbers in contrast to random numbers used in stochastic approaches. As a result of this fuzzyarithmetical analysis, a quantification of the overall uncertainty of the system outputs, including a worst-casescenario, is provided.

The component mode transformation method uses the component mode synthesis approach to drasticallyreduce the number of model evaluations required in the framework of the transformation method, so that asignificant reduction of the overall computing time is obtained.

As an example, the dynamic behavior of an automotive control unit, consisting of a constrained circuit board,is studied using both the original transformation method and the component mode transformation method.Finally, advantages and drawbacks of the proposed method are discussed.

10:30 Extending the Prediction Horizon of Uncertain-weighted Regression Models (ID557)Graeme Manson, Daley Chetwynd, Keith Worden, University of Sheffield, United Kingdom

In a recent paper by the authors, it was demonstrated that the issue of bounds explosion due to dependency,which is known to often arise when using interval arithmetic, limited the usefulness of interval-valued neuralnetworks for regression problems. Whilst these networks proved very capable at reducing the one step aheaderror in an NARX regression model of a pre-sliding friction process, when calculating the bounds on themodel predicted output they were found to diverge after a relatively small number of sample points. Asthe model predicted output is generally of more interest than the one step ahead prediction, this work aimsto try to considerably increase the usefulness of uncertain-weighted neural networks via the extension ofthe model predicted output prediction horizon. This will be achieved using affine arithmetic, an approachwhich overcomes the dependency issues of interval arithmetic, and its extension, quadratic arithmetic. Thesetechniques will be demonstrated on a simple linear system.

Non-linearities: identification and modelling – NL2Room 8 – Chairman: L. Garibaldi

11:20 Nonlinear Modelling and Identification of Torsional Behaviour in Harmonic Drives(ID 70)Tegoeh Tjahjowidodo, Farid Al-Bender, Hendrik Van Brussel, Katholieke Universiteit Leuven,Belgium

The demand for accurate and reliable positioning in industrial applications, especially in robotics and high-precision machines, has led to the increased use of Harmonic Drives. The unique performance features ofharmonic drives, such as high reduction ratio and high torque capacity in a compact geometry, justify theirwidespread application. However, nonlinear torsional compliance and friction are the most fundamentalproblems in these components and accurate modelling of the dynamic behaviour is expected to improve theperformance of the system.

79


This paper offers a model for torsional compliance of harmonic drives. A statistical measure of variationis defined, by which the reliability of the estimated parameters for different operating conditions, as wellas the accuracy and integrity of the proposed model, are quantified. The model performance is assessed bysimulation to verify the experimental results.

Two test setups have been developed and built, which are employed to evaluate experimentally the behaviourof the system. Each setup comprises a different type of harmonic drive, namely the high load torque and thelow load torque harmonic drive. The results show an accurate match between the simulation torque obtainedfrom the identified model and the measured torque from the experiment, which indicates the reliability of theproposed model.

11:45 Identification of Structural Free-play Non-linearities using the Non-Linear Reso-nant Decay Method (ID 338)Zhi Yang, University of Manchester / Tianjin University, United KingdomGrigorios Dimitriadis, Gareth Vio, Jonathan Cooper, Jan Wright, University of Manchester,United Kingdom

Structural non-linearities are becoming of increasing importance in determining the performance of a rangeof vibrating mechanical structures. As a consequence, the identification of systems with non-linearities isstarting to become a necessary part of vibration testing procedures. NL-RDM (Non-Linear Resonant De-cay Method) is an approach for the identification of non-linear multi-degree of freedom systems in modalspace on a mode by mode basis, using an appropriated sine excitation to isolate modes or groups of modes.However, the application of NL-RDM to a multi-degree of freedom system with a discontinuous free-playnon-linearity has not been attempted yet, except to treat it using high order polynomial terms. The difficultyof using NL-RDM when seeking a discontinuous free-play model lies in the choice of the so-called under-lying linear model to be used in uncoupling the linear equations. In this paper, a simulated two degree offreedom lumped parameter system with a free-play non-linearity

12:10 Nonlinear MDOF system characterization and identification using the Hilbert-Huang transform (ID 348)Gaetan Kerschen, Universite de Liege, BelgiumAlexander Vakakis, National Technical University of Athens, GreeceYoung Lee, D. Michael McFarland, Lawrence Bergman, University of Illinois at Urbana-Champaign, United States of America

The Hilbert transform is one of the most successful approaches to tracking the varying nature of vibrationof a large class of nonlinear systems thanks to the extraction of backbone curves from experimental data.Because signals with multiple frequency components do not admit a well-behaved Hilbert transform, it isinherently limited to the analysis of single-degree-of-freedom systems. In this study, the joint application ofthe complexification-averaging method and the empirical mode decomposition enables us to develop a newtechnique, the slow-flow model identification method. Through numerical and experimental applications, wedemonstrate that the proposed method is adequate for characterizing and identifying multi-degree-offreedomnonlinear systems.

80


Uncertainties in structural dynamics and acoustics – UNC2Room 8 – Chairman: D. Moens

14:00 An efficient fuzzy calculation algorithm with application to finite element modelupdating (ID 325)Daan Degrauwe, Edip Ozer Arman, Edwin Reynders, Guido De Roeck, Geert Lombaert,Katholieke Universiteit Leuven, Belgium

Fuzzy numbers are an uncertainty modeling tool with growing importance in different engineering areas.In practice, calculations on fuzzy numbers are executed by first transforming them to a set of intervals, thesocalled alpha-cuts. When subsequently evaluating these -cut interval expressions, a very specific situationarises: the same function has to be maximized and minimized several times, each time with different bounds.The Gradual -level Decreasing (GD) algorithm proposed in this paper exploits this situation and turns out tobe an efficient yet robust optimization method to perform fuzzy number calculations. In this study, the GDalgorithm was applied to a damage detection problem of a reinforced concrete (RC) beam by a FEM updatingmethod. The measured modal data on which uncertainty is considered consist of 4 natural frequencies and124 modal displacements. The propagation of the fuzzy uncertainty in the FEM updating process is studied,resulting in a fuzzy representation of the occurred damage.

14:25 Modal Analysis of Composite Lightweight Structures with Manufacturing Uncer-tainties (ID 44)Manfred Zehn, Gangadhar Machina, Otto-von-Guericke University of Magdeburg, Germany

Thin walled composite structures show fluctuations in Experimental Modal Analysis (EMA) results owingto variation of fibre diameters and locations, constituent properties, variation of layer thickness, etc. Theseuncertainties can often be attributed to the manufacturing process. The paper presents a novel approach formodelling of such random parameters spatially correlated in nature, suitable to be included in finite element(FE) modal analysis. A Semivariogram type material property model has been introduced to predict thespatially distributed material property over the entire structure. Local material properties are calculated fromstatistically homogeneous Representative Volume Element (RVE). The spatial arrangement of fibres in theRVE is obtained by using image processing based upon measured fibre distributions in crosssectional samplesand statistical modelling of fibre distributions. To exemplify the proposed approach in modal analysis, weconsider a thin-walled composite aircraft panel made from carbon fibre reinforced material.

14:50 Sensitivity analysis of frequency response function envelopes of mechanical struc-tures with interval uncertainties (ID 383)David Moens, Dirk Vandepitte, Katholieke Universiteit Leuven, Belgium

This paper introduces an interval sensitivity procedure that calculates the sensitivity of the envelope responsefunction in the outcome of the interval FRF analysis to each individual interval model uncertainty. Theprocedure focuses on the calculation of the sensitivity of the bounds defining the FRF response range tothe width of each individual uncertain input parameter. The approach differs from the classical sensitivityanalysis in the fact that it does not calculate local changes on the output resulting from local changes in theinput. The interval sensitivity result describes the change of the response interval width, taking into accounta change in the parameter interval width. The paper first introduces the concept of interval sensitivities. Itthen describes the theoretical background of the interval FRF procedure, on which the interval sensitivityanalysis is applied. Finally, the method is illustrated on a numerical example.

81


15:15 FE Model Updating with uncertain parameters (ID 285)Stefano Gabriele, University of ”Roma Tre”, ItalyClaudio Valente, University of Chieti-Pescara “G. d’Annunzio”, ItalyDaniele Spina, Department of Civil Protection, Italy

The selection of meaningful models and model parameters is the key issue for a successful model updating.The detection of false responses is of primary importance to get correct models, moreover, errors in theexperimental data have the potential to destroy the convergence to true solutions. In the paper the methodnamed INTerval Intersection Method (INTIM) is presented. It is formulated in the framework of intervalanalysis and is of global minimization type. Both these features allow to check for modelling errors and todeal with error affected measurements. Methods derived from the literature are used to construct uncertainfinite element models and to solve the interval eigenvalue problem. The concepts of models and parametersadmissibility are illustrated through a simple mechanical example but hardly identifiable that concerns thetransversal vibrations of a beam with unknown elastic properties and unknown boundary conditions.

15:40 Variation of structure modal parameters due to the structured uncertainties (ID398)Kheirollah Sepahvand, S. Marburg, H.J. Hardtke, Technische Universitat Dresden, Germany

This paper addresses the structured uncertainty investigation in structural modal analysis. The methods of un-certainty representation and general techniques of uncertainty formulations for modal analysis are reviewed.Variations of modal properties are investigated due to uncertain parameters. Numerical examples will showvalidity of the techniques.

Uncertainties in structural dynamics and acoustics – UNC3Room 8 – Chairman: W. d’Ambrogio

16:30 Dynamic Interval Analysis of FE models with uncertain substructures (ID 499)Hilde De Gersem, David Moens, Wim Desmet, Dirk Vandepitte, Katholieke Universiteit Leuven,Belgium

The exponential growth of the computational capacity has, amongst other things, enabled the design engineerto include non-deterministic properties in the numerical simulation and validation of new designs. In thiscontext, the interval concept has been introduced for the description of incomplete data, subjective knowl-edge and modelling uncertainties in a non-probabilistic manner. The use of the interval concept in the finiteelement context has led to the development of the interval finite element method for eigenfrequency andfrequency response function analysis. For industrially sized models with a large number of uncertainties, thecomputation time of these methods can be considerable. A reduction in calculation time can be achieved bythe substructuring of large models into substructures, which are then independently processed and reduced,and afterwards recombined.

The aim of the presented work is to combine the interval finite element method for dynamic analysis withthe Craig-Bampton component mode synthesis substructuring technique, in which the static and dynamicbehaviour of each substructure are represented by a set of component modes. Special attention is paid tothe consequences of uncertain parameters on the numerical representation of an uncertain substructure. Anumerical case study is presented to illustrate the concepts used in this paper.

82


16:55 Fuzzy finite elements: combination of Guyan reduction and a new method to solvelinear fuzzy system (ID 599)Annelies Vroman, Glad Deschrijver, Universiteit Gent, BelgiumMaarten De Munck, David Moens, Katholieke Universiteit Leuven, BelgiumEtienne E. Kerre, Universiteit Gent, BelgiumDirk Vandepitte, Katholieke Universiteit Leuven, Belgium

In practice it is sometimes very difficult and in many cases even impossible to define correct and unique inputdata for structural mechanics applications. Fuzzy numbers can represent the uncertain input for those cases.As a consequence fuzzy arithmetic, based on the extension principle can be applied to solve finite elementproblems with uncertain parameters. Application of fuzzy arithmetic directly to the traditional techniques forthe numerical solution of finite elements however turns out to be impracticable, especially solving systemsof linear equations. Here we present a new method to solve systems of linear fuzzy equations combined withGuyan Reduction. Our conclusions are confirmed by a simple static problem.

17:20 Coupling of substructures with in-tolerance uncertain dimensions (ID 160)Walter D’Ambrogio, Universita de L’Aquila, ItalyAnnalisa Fregolent, University of Rome “La Sapienza”, Italy

In this paper, the effect of component variability (due to dimensional tolerances) on the dynamics of an as-sembled structure, is analysed using Monte Carlo simulation. Dimensional tolerances are usually prescribedon individual components, whose dynamics is substantially unaffected by the variability within the toler-ance field. However, the dynamic behaviour of the assembled structure might be much more sensitive tocomponent variability. A random assembly process is performed by assuming a normal distribution for di-mensional uncertainties, and the probability density functions of natural frequencies and MAC are estimatedfrom Monte Carlo simulation. Mode shapes and CoMAC of each assembled substructure are evaluated too.Efficient ways to reduce the dynamic variability of the assembled structures are finally discussed.

17:45 Using Design of Experiments to model the effect of uncertainties in substructurecoupling (ID 403)Walter D’Ambrogio, Universita de L’Aquila, ItalyAnnalisa Fregolent, University of Rome “La Sapienza”, Italy

In this paper, the effect of component variability (due to dimensional tolerances) on the dynamics of an as-sembled structure is modeled using procedures derived from Design of Experiments (DOE). Specifically, thepossibilities offered by factorial design, in order to identify a regression model of the effect of uncertaintiesand of their interactions, are explored. Of course, the number of numerical experiments, required to fit a re-gression model, is much less than the number of realisations required for the implementation of Monte Carlosimulation, presented in the companion paper. The regression model can then be used instead of the physicalmodel to evaluate the dynamic behaviour of the assembled structure. The procedure is verified by comparingthe output of the regression model with results of the physical model. Furthermore, the percent contributionsof different uncertainties are evaluated, allowing to select which tolerance fields should be narrowed first inorder to reduce the dynamic variability of the assembled structure.

18:10 Analysis of Sources and Quantification of Uncertainty in Experimental Modal Data(ID 280)Yves Govers, Marc Boswald, Ulrich Fullekrug, Dennis Goge, Deutsches Zentrum fur Luft- undRaumfahrt e.V., GermanyMichael Link, University of Kassel, Germany

The use of uncertainty propagation analysis in conjunction with the finite element method has become moreand more popular in recent years due to increasing computer power. Among other things, uncertainty prop-agation methods are utilized to ensure the robustness of the computational model, or respectively, of the

83


structural design. The determination of fuzzy response data relies heavily on the assumptions made for thevariability of the uncertain parameters, mainly because it is impossible to express the scatter of uncertainparameters, like for example joint stiffness parameters, in a mathematical way. Consequently, an inverseapproach is required to quantify parameter uncertainties in a finite element model. Computational ModelUpdating (CMU) techniques offer the possibility to identify model parameters from measured response data.However, statistical test data is required for uncertainty identification by using finite element models andcomputational model updating techniques. Thus it is necessary to analyze the impact of different possiblesources of uncertainty on the test data variability prior to CMU.

This paper comprises the results gained from multiple vibration tests performed on a replica of the wellknown GARTEUR SM-AG19 benchmark structure at the German Aerospace Center (DLR) and subsequentextractions of modal parameters. In particular, the different sources and the corresponding magnitudes of un-certainty are illustrated. This means that special care is taken to uniquely quantify the variability introducedon modal data with respect to different operators, different modal analysis methods, structure dependentscatter, and differences in phase separation testing and phase resonance testing. As a result, a comparisonis made to express the variances of the different sources in the experimental modal analysis process whichallows for a classification of the error sources (ranking with respect to the relative impact on the variability ofthe test data). The aim of the identification of the different variances of the miscellaneous sources is to pro-vide a realistic database for subsequent stochastic finite element model updating and to support the selectionof meaningful stochastic model updating parameters.


8:50 Fuzzy Finite Element Method: Computer Aided Design Application (ID 186)Franck Massa, Karine Ruffin, Bertrand Lallemand, Thierry Tison, Universite de Valenciennes etdu Hainaut Cambresis, France

A Computer-Aided Design methodology is presented in this paper to facilitate the design and optimization ofstructures with uncertain parameters. The methodology includes both parametric uncertainties and designerjudgment and/or experience, described by fuzzy numbers. The aim of this study is to determine the validitydomain of fuzzy design variables with respect to the non-boolean restrictions applied to fuzzy solutions. Theprocess is organized in three steps: propagation of uncertainties, adjustment of design variables and optimi-sation of fuzzy solution sets. A numerical application to an industrial case will show the capabilities of themethodology.

8:50 Continuous wavelet transform and entropy criterion in modal analysis (ID 110)Merzak Dahmani, Amar Bouazzouni, Universite Mouloud Mammeri de Tizi-Ouzou, Algeria

In this work, wavelet method is used for the identification of the modal parameters of linear structures. Theanalyzed signal is supposed to be the free response collected from an accelerometer placed at one selectedposition along the structure. In order to reach optimal frequential and temporal resolutions for all the modesof vibration, a parameter N is introduced into the Morlet wavelet. The value of N which minimizes theentropy of the wavelet coefficients for each mode is the optimal value of analysis. That allows an adequatedistribution of the resolutions that gives a more precise estimate of the modal parameters of the structure.Numerical examples show the effect of the variation of the N parameter with the mode of vibration on themodes coupling and on the border effect.

84


8:50 Modal Cluster Decomposition on Two-Dimensional Frequency Domain and ItsReal-Time Filtering (ID 249)Yuichi Matsumura, University of Shiga Prefecture, JapanNaoki Hosoya, Saitama University, Japan

This paper describes a basic theory for modal cluster decomposition on two-dimensional frequency domainand a design method of real-time filter to extract each decomposed modal cluster. At first, the concept ofmodal cluster decomposition was introduced. Then, an objective function was derived to determine thesensor placement to realize the modal cluster decomposition. The sensor placement is optimized to placeeach modal cluster on any selected line in two-dimensional frequency domain. By using the two-dimensionalFourier transform of the array sensor outputs, which are measured at selected optimal sensor locations,desired modal cluster decomposition is realized. In order to extract the desired modal cluster on real-time,we also derived a design method of two-dimensional FIR filter. This filter computes the simple summationof current signals for a spatial direction, and hence there is no time delay to extract the signal of a desiredmodal cluster.

8:50 Direct and simple estimation of physical residual effect for measured and unmea-sured DOFs in experimental modal analysis (ID 296)Primoz Cermelj, Miha Boltezar, University of Ljubljana, Slovenia

When estimating the modal properties of a certain structure the residual effect of unmeasured modes outsidethe frequency band of interest usually can not be neglected. Although there have been many investigationsin the past on how to compensate for the residual effects, there is still a common belief that residual termscan not be estimated for the unmeasured part of the structure’s response model. Therefore, in this paperwe demonstrate the usage and the applicability of the simple, but most frequently used, physical residualmodel in terms of a direct estimation of the residuals for the whole FRF matrix, even for the unmeasuredDOFs. By using the unified matrix polynomial approach in the frequency domain for the estimation of modalparameters, the results on a beam case-study reflect the usability of the approach, while some influencesfor the determination of the residual terms and the overall results in the modal estimation process are alsopresented.

8:50 Accounting for glue and temperature effects in Nomex based honeycomb models (ID344)corine Florens, ONERA / Ecole Centrale Paris, FranceEtienne Balmes, Ecole Centrale de Paris / SDTools, FranceFranck Clero, Mathieu Corus, ONERA / Ecole Centrale Paris, France

To predict the effect of active control on aircraft or helicopter trim panels, made with honeycomb sandwichcomposite, one approach consists in modeling the panel by Finite Element Method. FEM with shell elementsfor the laminate and volume elements for the core is classically used in industry. The aim of the present studyis to determine the limits of the Shell-Volume-Shell model, in case of honeycomb core, through numericalcorrelation with a very detailed three dimensional model. More precisely, the influence of glue stiffness bynumerical simulation has been considered, and the temperature influence has been observed through modaltests inside a controlled environment chamber.

8:50 Wave Propagation Modelling in Composite Plates (ID 563)Pawel Kudela, Polish Academy of Sciences, PolandWieslaw Ostachowicz, Gdynia Maritime University, Poland

This paper presents results of a numerical simulation of the transverse elastic wave propagation in compositeplates. The problem has been solved by the use of the Spectral Element Method. The wave propagationand scattering phenomena has been analysed from damage detection point of view. In order to approximateaccurately A0 mode of the Lamb wave, numerical model based onMindlin’s plate theory has been developed.

85


36- and 100-nodes spectral plate elements have been considered. Numerical calculations have been carriedout for various orientations of reinforcing fibres within the plate as well as for various fibre volume fractions.Also a procedure for calculation of theoretical group wave velocity has been presented.

8:50 Waveguide finite element modelling: numerical issues (ID 172)Yoshiyuki Waki, Brian R. Mace, Michael J. Brennan, University of Southampton, United Kingdom

Wave motion in uniform structures in one direction can be modelled using the waveguide finite element(WFE) method. The transfer matrix of a section of a waveguide is found from the mass and stiffness ma-trices, which may be formed by conventional FE packages. The wave properties can be obtained from theeigenproblem resulting from applying a periodicity condition to the transfer matrix. Numerical issues inthe WFE method are described in this paper. Numerical errors are categorised into three groups: (1) finiteelement discretisation errors, (2) round-off errors and (3) errors induced by numerical illconditioning. Thelength of the finite element affects these errors. Methods including singular value decomposition are appliedto improve numerical errors. Wave motion and numerical issues are illustrated and discussed for simplewaveguides for which analytical solutions are available.

8:50 Noise reduction of a laser printer using SEA and FEM (ID 263)Atsuo Mori, Nihon ESI K.K., JapanToru Yamazaki, Katsuhiko Kuroda, Kanagawa University, JapanFumiyasu Kuratani, Wakayama University, Japan

Office machines (printers, copy machines, fax, etc) are essential tools in both business and private scenes.The features required for these machines are high quality print images, high speed printing, small size,and low noise, etc. In recent years, the noise of office machines is especially undesirable. To reduce thenoise of the office machines, many approaches have been used. Most of them are FEM and experimentalmodal analysis for vibration problems. Those approaches are effective, while they are difficult to be appliedto the cases that connections among mechanical parts are weak and properties of parts are irregular. Inthis paper, SEA (statistical energy analysis) is applied for noise reduction of a laser beam printer. In ourprocess using SEA, first, the laser printer was subdivided into some parts (SEA subsystems). Next, theexperimental SEA(ESEA) was carried out to obtain the SEA parameters (CLFs and ILFs). After that, theacoustic contribution factor for each SEA subsystem was evaluated by using the sound measurement and theESEA parameters. The factors indicate how vibration energy of each subsystem contributes to the soundpower level at the evaluation positions. Then, the perturbation method was adopted in order to decide whichSEA parameter should be changed to effectively reduce sound power level of the laser printer. By usingthis method, the CLFs between the two pairs of subsystems were extracted and it was shown that these CLFsshould be reduced to smaller values. In order to verify the validity of the processing above, we modified someparts of the laser printer in order to reduce the values of the two CLFs extracted by the perturbation method.The experiment showed 4.1dB reduction of sound power level at target peak in the frequency domain. Inconclusion, we reached the result that the combination of the SEA approach and the perturbation method iseffective for noise reduction of a laser printer. This combination can be also applicable for noise reductionof the other kinds of office machines. Then we also confirmed the change of the CLFs due to the structuralcould be predicted by ‘PAM-VA One’.

8:50 Advanced modeling of periodic structures in SEA (ID 502)Vincent Cotoni, ESI-US R&D, United States of AmericaRobin Langley, University of Cambridge, United States of America

The SEA properties of a periodic structure are computed from the FE analysis of a single periodic cell.The periodic theory is used in conjunction with FE so that any geometry can be considered. Some efficientalgorithms have been implemented to get the subsystems intrinsic properties (modal density, damping, andequivalent mass), as well as the coupling properties of the subsystem with acoustic subsystems (radiationand transmission). Comparisons with analytical results validate the method.

86


8:50 Identification of a Nonlinear State Space Model for Control using a Feature SpaceTransformation (ID 427)Kris Smolders, Maarten Witters, Jan Swevers, Paul Sas, Katholieke Universiteit Leuven, Belgium

This paper presents a new grey-box state space model structure for nonlinear systems together with its identi-fication procedure. The model structure is linear in the input and a combination of a linear state space modeland a feature space transformation. Because of the first two properties this model structure is interesting forcontrol design. The identification procedure for the model structure assumes that all the state variables canbe measured directly or accurately estimated from measurement data. The presented identification methodis a combination of well established identification methods for linear and nonlinear black-box identificationapproaches, respectively linear least squares and neural training. Physical insight is required to first select theappropriate state variables, secondly to select the right type and number of features and finally to reduce thenumber of parameters and improve the model accuracy. Therefor this modeling can be considered as a grey-box approach. The modeling approach has been applied successfully on a quarter car test setup containing anonlinear magneto-rheological semi-active damper.

8:50 Morlet Wavelet-based Parameter Identification in Different Stimulation (ID 34)Lin Yue, X.Q. Liu, College of Mechanical and Electric Engineering Nanjing University of Aero-nautics & Astronautics, China

In the dynamic testing for example flight flutter testing the input and output data are limited and heavy noise-polluted. At the meantime there are close modes in the frequency response functions (FRFs). So it is verydifficult to identify system parameters accurately. We investigate Morlet-based wavelet temporal frequencyfilter and rational fraction orthogonal polynomials (MRFOP) according to different stimulation signals forexample sweep frequency function, impulse excitation to get fine FRFs by modifying parameters of basiswavelet and reducing noise in the temporal frequency domain. Numerical simulations are conducted usingGARTEUR aircraft model, three close mode system, excited with the sweep frequency and a four modesystem of the two-dimension frame excited with the impulse respectively. The input and output data withwhite noise are added to the numerical simulation. Results show that accuracy of the system parameteridentification is improved evidently.

8:50 Resonant-Based Identification of the Elastic Properties of Ceramic Coatings (ID355)Tom Lauwagie, Ward Heylen, Katholieke Universiteit Leuven, Belgium

The elastic properties of the individual layers of layered materials can be derived from the resonant frequen-cies of a set of test specimens comprising samples with different layer configurations. Based on this, a mixednumerical-experimental framework for the identification of the elastic properties of layered materials hasbeen formulated and a number of different identification routines were developed.

In this work, the elastic properties of an yttria-stabilised zirconia top coat of an air plasma sprayed thermalbarrier coating are identified with the different mixed numerical-experimental routines that were developed.The obtained results are critically evaluated, and some recommendations on the use of the considered routinesfor the identification of coating properties are presented.

87



14:00 Adaptive Landing Gear: optimum control strategy and improvement potential (ID112)Lukasz Jankowski, Grzegorz Mikulowski, Institute of Fundamental Technological Research,Poland

An adaptive landing gear (ALG), as considered in this paper, is a landing gear (LG) capable of active adap-tation to particular landing conditions by means of controlled hydraulic force. The objective of the adaptivecontrol is to mitigate the peak force transferred to the aircraft structure during touchdown, and thus to limitthe structural fatigue factor. The paper investigates two control strategies (semi-active and active) and thepotential for improvement. As a reference a standard, passive landing gear is considered (nose LG, I23 air-craft, Institute of Aviation, Warsaw, Poland) and the calculated results are evaluated statistically, in terms ofthe mean and the median peak strut force. Additionally, a general strategy of rebound height mitigation isproposed. Accuracy of the model is verified by comparison of simulations with measurements taken duringlaboratory test of the reference landing gear. The concept has been verified experimentally using a laboratorytest stand.

14:00 Semi-active damping of vibrations. Prestress Accumulation-Release strategy devel-opment. (ID 148)Anita Orlowska, Arkadiusz Mroz, Institute of Fundamental Technological Research, PolandJan Holnicki-Szulc, Institute of Fundamental Technological Research, Polish Academy of Sci-ences, Poland

New method for semi-active control of vibrating structures is introduced. So-called Prestress Accumulation-Release (PAR) strategy aims at releasing of the strain energy accumulated in the structure during its defor-mation process. Numerical simulations as well as some experimental results prove that the strategy can bevery effective in mitigating of the fundamental mode of the structure. An example of a simply supportedbeam indicates that ca 95% of the fundamental mode of vibration can be mitigated after two cycles. In muchmore complex practical problems smaller portion of total energy can be released from the system in eachcycle, nevertheless the strategy could be applied to mitigate the vibrations of, for example, pipeline systemsor pedestrian walkways.

14:00 Modelling and tuning the SMA absorber (ID 259)Wojciech Klein, Arkadiusz Mezyk, Eugeniusz Switonski, Silesian University of Technology,Poland

In order to reduce excessive vibrations in mechanical systems there are used different absorber types. Theshape memory alloys (SMA) are more and more frequently used in semi-active vibration control systems.Such an absorber changes its stiffness when heating or cooling. This type of material has nonlinear char-acteristics, which causes difficulties in modeling process. This paper presents the process of modelling anddetermination dynamic characteristics of the SMA vibration absorber. The numerical model of primary sys-tem with the absorber was created in APDL language, which is an internal ANSYS language. The SMA wasmodelled as nonlinear material, which is very important to obtain high-quality results.

14:00 Application of dither control for automotive wiper squeal (ID 679)Bert Stallaert, Frederik Doucet, Johan Rys, Katholieke Universiteit Leuven, BelgiumAlpha Diallo, Sebastien Chaigne, Renault, FranceJan Swevers, Paul Sas, Katholieke Universiteit Leuven, Belgium

This paper describes a feasibility study on dither control for automotive wiper squeal. Wiper squeal is a tonalnoise caused by an unstable vibration due to the friction between wiper blad and windscreen. Dither control

88


is the superposition of a high-frequent signal, in this case the superposition of a high-frequent vibration, tostabilize a low-frequent unstability in a system. A finite element model of the wiper has been developedto aid the choice and design of an actuator system; piezo patch actuators are applied on the wiper to applythe dither signal. First experimental results show that wiper squeal can effectively be suppressed by dithercontrol, as soon as the dither signal reaches a certain threshold value.

14:00 Noise reduction of handheld vacuum cleaners by geometric optimization of compo-nents (ID 24)Hani Ashrafi, Mohammad Mahjoob, University of Tehran, Iran (Islamic Republic of)

High noise levels of vacuum cleaners are partly associated with the design of air canals of different compo-nents. An experimental investigation is conducted here to study the effect of the components geometry onthe noise emission. A typical charged type vacuum cleaner is selected for noise treatment. Each part is opti-mized geometrically to affect the air flow and reduce the overall noise level. This is achieved by consideringdifferent sources generating aerodynamic noise in the vacuum cleaner. The redesigned parts are prototypedand tested in different conditions and the variation of vacuum power and noise level are fully investigated.The shape treatment has no adverse effect on the machine performance and the vacuum power is improvedover the course of this noise treatment.

14:00 The influence of the sharp transients introduced by the force window when appliedto signals with DC components (ID 106)Adrian Florin Nicula, Carsten Mohr, Andreas Anstatt, LuK GmbH &Co.oHG, Germany

For validation and updating of Finite Element Models their accuracy is measured in relation to experimentaldata (Experimental Modal Analysis vs. Computational Modal Analysis). A very fast method of obtainingthe Modal Parameters used for validating the Computational Modal Analysis results involves instrumentedhammer excitation of the structures. The frequencies and mode shapes can be obtained very accurately andthe two results can be used for improving the finite elements model. If the force and excitation signals havevery large DC components the use of the force window may raise some problems, such as the variationof modal parameters. This article will investigate the effects of such problems on the estimated modalparameters of a simple structure.

14:00 Vibration Testing of Large Spacecraft Structure (ID 386)Yaoqi Feng, Song Yang, China Academy of Space Technology (CAST), China

China has launched successfully Shenzhou manned spaceship, this paper introduces briefly the three mod-ules of the Shenzhou manned spaceship and its functional subsystems, and then describes methods andtechniques used to assure the success of the vibration test of large spacecraft structures and to validate thedynamic behavior of the spaceship structure. The vibration test technique developed in this paper involves theimplementation of multi-points average vibration control, notching control, and the evaluation techniques ofthe dynamic characteristics of the structure, etc. The procedures and results analysis using these techniquesfor the Shenzhou manned spaceship structure is presented.

14:00 The poly-reference LSCF estimator optimised for vibro-acoustic systems (ID 525)Gert De Sitter, Patrick Guillaume, Vrije Universiteit Brussel, Belgium

Recently the frequency-domain poly-reference LSCF estimator (also known as the LMS PolyMAX esti-mator) has introduced an important improvement in the field of modal analysis. The poly-reference LSCFestimator produces very clear stabilisation diagrams and can deal with highly damped structures. In thiscontribution the poly-reference LSCF estimator will be optimised for vibro-acoustic systems. In the first partof the paper a recap of the theory of the LSCF estimator will be given as well as the required generalisationto cope with vibro-acoustic problems. In the second part of the paper, the theoretical results will be validatedby means of an experiment.

89


14:00 Magnetorheological landing gear design: A feasibility study for small and large-scale aircraft (ID 88)David Batterbee, Neil Sims, University of Sheffield, United KingdomZbigniew Wolejsza, The Institute of Aviation, PolandArnaud Lafitte, Messier Dowty, France

Aircraft landing gears are subjected to wide ranging excitation conditions during touchdown, which results inconflicting damping requirements. This is due to the unpredictable and large variations in an aircraft’s massand sink velocity, which serves to increase passenger discomfort and reduce the structural fatigue life. Anovel solution to this problem is to implement semi-active damping using magnetorheological (MR) fluids.

In this paper, an MR landing gear sizing methodology is described, which enables an effective assessment offeasibility to be made for many aircraft. Essentially, the methodology aims to maximise the range of landingimpacts that can be optimally damped, whilst adhering to the stringent packaging constraints of conventionaldevices. This methodology is demonstrated by performing feasibility studies for a variety of aircraft, rangingfrom small-scale utility aircraft to large-scale commercial jets.

14:00 A Study on Modal Characteristics of a Commercial Vehicle Rear Door (ID 492)Emre Dikmen, Ipek Basdogan, Koc University, Turkey

The interior noise level has been an important design criterion lately in automotive industry. Most of theinterior noise is generated by the vibrating parts of the vehicle body, mainly the doors. The vibration char-acteristics of the door panels are affected by the weatherstrip seals. The weatherstrip seals exhibit nonlinearbehavior with changing frequency, compression amplitude, temperature and previous load history. Thesetypes of materials are very difficult to model and their effect must be studied experimentally. In the antic-ipation of these needs, we built a modal analysis testing system for determining the dynamic properties ofa rear door of a commercial vehicle. Two different configurations are used to perform the experiments. Inthe first configuration, the door is hung from a supporting frame with elastic cords to simulate the free-freeboundary conditions. In the second configuration, doors are mounted to the vehicle body that is supportedby air springs. Frequency shifts and damping contribution of weatherstrip seals on the door is investigatedby performing experiments with and without the seal using the second configuration. Additionally, havingtwo different configurations for the same structure makes it possible to determine the effect of the boundaryconditions on the dynamics of the rear door.

14:00 Vibration decay using on-off stiffness control (ID 226)Diego Francisco Ledezma-Ramirez, Neil Ferguson, Michael J. Brennan, University of Southamp-ton, United Kingdom

A relatively new approach for shock and vibration isolation is to use semi-active or fully-active elements,which are designed to provide better vibration isolation over a broad range of excitation inputs. Severalvibration isolation strategies have been proposed and developed using variable damping elements, variablestiffness elements and different actuators. An interesting, novel control strategy is to have an isolator whosestiffness can take two values in an on-off manner. In this paper such a system is described and modelled, andin particular the free vibration decay behaviour is quantified and discussed.

This paper also describes the benefits of the variable stiffness isolator and quantifies the effective dampingin terms of simple model parameters.

90



8:50 Frequency response function as machinability indicator (ID 169)Alain Giraudeau, Mohamed El Mansori, Ecole Nationale Superieure d’Arts et Metiers (ENSAM),France

The paper presents a new methodology which exploits the cutting system vibrations to identify the machin-ability rating or index. The idea is to compute the frequency response function of the tool work-workpiecestructure, under the machining conditions of interest, for a given material as it is sheared by a single-pointcutting tool with the objective of assessing its machinability. To describe this approach, the design and theexperimental validation of a specic and original device for vibration excitation are rst presented. Subse-quently, the extracted modal parameters and some qualitative results of preliminary rating of machinabilityfor different materials are discussed. Implications of these results and further developments close the paper.

8:50 Identification of a Bolted Lap Joint Parameters Using Response Surface Method (ID30)Hamid Ahmadian, Hamed Hadad, Iran University of Science and Technology, Iran (Islamic Re-public of)

Mechanical joints have significant influence on the response of compound structures due to their localizednon-linear stiffness and damping effects. Considerable efforts have been expended to assign appropriate pa-rameters for the joints in assembled structures such that the overall model represents the dynamical behavioraccurately. Recently a nonlinear generic element model for bolted lap joints and interfaces is proposed bythe first author representing the dominant physics involved in the joint such as micro/macroslip. The jointgeneric element parameters are identified by minimizing the difference between the model response and theobserved behavior of the structure. Due to the difficulty arising from the nonlinear nature of the involvedinverse problem, the response surface methodology (RSM) is employed in optimization procedure whichdoes not require sensitivity analysis in extracting joint parameters from measured responses. The RSM is acollection of procedures including design of experiments (DOE), model selection and fitting, and optimiza-tion on the fitted model. Based on the nature of the objective function a quadratic polynomial function formis adopted for the Response Surface Approximation (RSA). The RSA is built from DOE and model fitting.The proposed method is demonstrated using an actual test case in which the joint parameters of a beam likestructure with a single bolted lap joint are identified using RSM. It is demonstrated that the response surfacemethod achieves the optimum solution much faster and with less computational efforts compared to the othernon-sensitivity based models methods.

8:50 Analytical methods applied to pyrotechnic shock modelling (comparison of modelresults with shock test results) (ID 55)Massimo Palladino, Techspace Aero (Safran group), Belgium

Different types of shock events can occur in spacecraft, either generated by launcher separation events or byinternal deployment events. These shocks are known as pyrotechnic shocks.

Because of the very high frequency character of pyrotechnic loads, classical normal mode analysis and finiteelement method (FEM) models are not effective for predicting structural responses to pyrotechnic loads.

In parallel, methods based on statistical energy analysis (SEA) are appropriate but are not relatively easy toaccomplish.

An analytical approach aiming at predicting structural responses to pyrotechnic shocks is presented here.This method has the advantage of being relatively easy to accomplish. The method has been used to estimatethe satellite’s 1 Newton thrusters flow control valve structural responses during pyrotechnic shock.

91


The model results have been compared with shock test results. This comparison has shown that the proposedanalytical method provides reasonably accurate results. Moreover, the method allows us to estimate safetymargins that are of a great interest in developing new space projects.

The analytical approach is based on an analytical estimation of structure fundamental frequency, estimationof the peak acceleration at the structure centre of gravity using 1 degree of freedom (DOF) model, numericalintegration of the wave propagation equation and non - linear stress calculation using Neuber ‘s algorithm.

8:50 Damping of floating yacht harbours determined by experimental modal analysis (ID274)Santiago Uhlenbrock, Gunther Schlottmann, University of Rostock, Germany

The prediction of the dynamic behaviour of floating harbours is necessary to guarantee the integrity of thestructure and moored ships and to offer more comfort for yacht owners. Accurate prediction of forcedvibration amplitudes is dependent on the model selected for energy dissipation. In engineering applicationslinear damping models are commonly used and are sufficiently accurate if the working conditions are takeninto account. However, besides the unknown boundary conditions due to the mooring line and uneven seafloor, the nonlinear characteristics of the fluid also influence the damping properties of floating structures.Within the scope of this study the determination of damping for a large floating marina with approximately200 anchorages is presented. The overall length of the analysed section is approx. 300 meters so that aclassical impact excitation is difficult to accomplish. Therefore, damping values were identified using naturalinput modal analysis techniques. Other very important advantages of using this output-only techniques isthat all influences, such as current, waves as well as mooring line and wetted surface fouling are implicit inthe identified modal damping factors. Stochastic subspace techniques were applied for damping estimationsusing a high damping stability in the stabilisation diagram. Also the complexity of the identified modes andthe repetition of damping values in the different datasets was used to estimate damping more accurately. Thedamping values identified lay between 2.5 % and 4.2 %. The associated amplitude error due to damping wasestimated to be less than 15 % for all identified modal dampings.

8:50 Sensibility analysis of the dynamic behaviour the violin plates in function of theirgeometry (ID 316)Mario Razeto, Claudia Staforelli, G. Barrientos, Universidad de Concepcion, Chile

The study of dynamic and vibratory behavior of violins considers the effect of the different parts that composeit. This study is centered in their vibratory behavior of the top and back plates, since the ideal form of themodeshapes and frequency of violins of greater prestige is known, according to bibliography. Experimentaltests determine the modeshapes with their resonant frequencies of the violin plates. A numerical model of thetop plate is made by means of the finite elements method, to determine the influence of the different geometricparameters on its main vibrational modes and natural frequencies, this analysis considers the orthotropy ofthe mechanical properties the wood. A sensibility analysis is applied on a finite elements model of the topplate, to determine the influence of the thickness in different zones on its natural frequencies, and to study theeffect produced by the thickness variations of the bass bar. With the final objective of obtaining the expectedacoustic response, making precise changes in their structure.

8:50 Simulation of forced response in linear and nonlinear mechanical systems using dig-ital filters (ID 444)Kjell Ahlin, Martin Magnevall, Andreas Josefsson, Blekinge Institute of Technology, Sweden

There exist many methods to calculate forced response in mechanical systems. Some methods are slowand the errors introduced are unknown. The paper presents a method that uses digital filters and modalsuperposition. It is shown how aliasing can be avoided as well as phase errors. The parameters describingthe mechanical system are residues and poles, taken from FEA models, from lumped MCK systems, fromanalytic solutions or from experimental modal analysis. Modal damping may be used. The error in the

92


calculation is derived and is shown to be only a function of the sampling frequency used. When the methodis applied to linear mechanical systems in MATLAB it is very fast. The method is extended to incorporatenonlinear components. The nonlinear components could be simple, like hardening or stiffening springs, butmay also contain memory, like dampers with hysteresis. The simulations are used to generate test data fordevelopment and evaluation of methods for identification of non-linear systems.

8:50 Influence of the Structural Modification on the Stiffened Plate Energy Balance (ID538)Marek Iwaniec, Joanna Iwaniec, AGH - University of Science and Technology, Poland

Introduction of stiffeners set slightly increases mass of the stiffened plate and significantly changes systemdynamic properties such as natural frequencies, spatial distribution and velocity magnitude, stress and me-chanical energy. Stiffeners can be treated as an additional subsystem that not only modifies the plate stiffnessbut also accumulates and dissipates a part of mechanical energy.

The paper concerns modal analysis carried out on the basis of energy balance for a family of similar stiff-ened plates treated as a linear connection of a homogenous plate and spatially spanned sets of beams. Systemnatural frequencies were calculated by comparison of maximal kinetic and strain energies while shape func-tions were assumed to be series of base functions. On the basis of participation factors energy dissipated bya homogenous plate and a system of stiffeners was determined. Natural frequencies and participation factorswere computed by the use of the Nelder-Mead optimization algorithm and genetic algorithm. Software per-formance and estimation accuracy were assessed by the comparison of obtained analytical results, results ofFEM computations and results presented in literature.

Theoretical computations of plate and stiffeners energies distribution were verified by computations car-ried out on the basis of frequency responses in nodes of FEM models. The influence of geometrical andmaterial properties on natural frequencies and energy distributed in plate and the system of stiffeners wasinvestigated. Introduction of sound radiation coefficient concept made it possible to formulate a model ofacoustical radiation and energy distribution in the system consisted of plate, stiffeners and acoustical volume.

8:50 Structural Dynamics of a Mobile Substation during Transport (ID 446)Christof Devriendt, Patrick Guillaume, Vrije Universiteit Brussel, BelgiumJ. Lopez-Roldan, Pauwels Contracting, Belgium

Mobile substations can be defined as completely equipped electrical substations. There is a lack of existingscientific basis in the mechanical design of the structural components of the mobile substation. In concretethere are no capabilities to determine the dynamic behavior during transport and service conditions. Improperdimensioning of the structures results in an important degree of mechanical failures during transport. Thedynamic response of structures on a mobile substation to transport-motions depends on their strength of con-struction, ductility, and dynamic properties. Lightly damped structures that have one or more natural modesof oscillation within the frequency band of transport excitations can experience considerable amplification ofboth the forces and deflections. Thus, items of mobile substation equipment whose natural frequencies lie inthe normal frequency range of transport motion are particularly vulnerable to damage and fatigue. Thereforewe are interested in analyzing the natural frequencies, damping ratios (modal parameters) and level of accel-erations of those components. The modal identification methods recently developed at the Vrije UniversiteitBrussel, allow for estimating these modal parameters from operational road testing data. Finally the dynamicbehavior of interconnected equipment will be discussed and some design recommendation will be proposed.

93


8:50 Modal Parameter Identification and Correlation of a Launch-Vehicle Interstage Us-ing “Output-Only” Experimental Data (ID 559)Giuliano Coppotelli, Franco Mastroddi, Gian Mario Polli, University of Rome “La Sapienza”,ItalyLaura Cospite, Avio S.p.A. - Propulsione Aerospaziale, Italy

The Output-Only analysis typically allows to identify the modal parameters of a given structure on the baseof the data achieved by the operative conditions or environment of such a device. In the design developmentof a Launch Vehicle (LV) the vibration tests for launch qualification are performed with the main objective toverify that the acceleration levels of the structure, attached systems, and payloads, do not exceed a prescribeddesign level. Thus, these kind of tests are performed by sensoring the output responses (i.e., the accelerationlevels) only. In the present paper, the modal parameter identification and correlation of a LV inter-stagebased on Output-Only (OO) experimental data will be presented. The practical advantage arisen by usingthis technique will be pointed out. It has essentially consisted of utilizing the same data collected duringqualification tests. Therefore, this has reduced the overall cost required for setting up a further experimentaltest that should be explicitly dedicated to the modal parameter identification. Moreover, the obtained modalestimates, based on the OO approach, have been also used for correlating the numerical Finite-Elementmodel of the LV inter-stage that have been previously used for design purposes.

8:50 A model reduction method for determining modal parameters of a fluid-loadedstructure (ID 86)Xianhui Li, Sheng Li, Dalian University of Technology, China

A model reduction method is proposed to determine modal parameters of a fluid-loaded structure. Based on amatrix-free formulation of rational Krylov projection, a reduced order model for the fluid-loaded structure isconstructed from forced responses at selected interpolation frequencies. Due to the small size of the reducedorder model, eigenpairs of the associated eigenvalue problem are available at a very low computational cost.Resonance frequencies, modal damping ratios, and mode shapes of the original system are recovered fromthe eigenpairs of the reduced order model and the forced responses at the interpolation frequencies. Criteriabased on modal damping ratio and condition number of dynamic stiffness matrix are proposed to filter outthe non-physical modes introduced in the model reduction process. Numerical efficiency of the method isdemonstrated on a water-loaded plate clamped in a baffle.

8:50 Transfer function modeling and control of a flexible gear train (ID 54)Irit Peled, Yoram Halevi, Technion - Israel Institute of Technology, Israel

The paper considers the problem of dynamic, infinite dimension, modeling of a transmission. An accurateLaplace transfer function matrix of the system that consists of flexible shafts connected by, rigid or flexible,gears is found. The first step is deriving the transfer functions for a single uniform link. The building blocksof those transfer functions are time delays, representing the wave motion, and low order rational expressions,representing the boundary phenomena. The next step is combining these individual transfer functions intoa model for a multilink flexible transmission. This is done by making use of the natural feedback betweenneighboring links in the system. The outcome is a generalized, infinite dimension, dynamic model with ageneralized state vector. The explicit and highly structured form of the transfer functions, allows physicalinsight into the system, which is lacking in finite dimension approximations, and exact calculation of naturalfrequencies. It also enables the construction of exact simulation schemes built from standard blocks, and theapplication of simple, dedicated control laws.

94



14:00 Energy Flow Method Applied to the Detection of Structural Damages (ID 115)Ahcene Amziani, Amar Bouazzouni, Universite Mouloud Mammeri de Tizi-Ouzou, Algeria

In this paper, the exploitation of local variations of energy density and power flow to detect structural dam-ages is considered. The presence of defaults deteriorates the mechanism of dissipation of energy in a struc-ture, because of increase in modal damping ratio in the vicinity of these defaults. As damage affects dampingmore than it does for structural stiffness, it may be a potential parameter that may help distinguish damagedstructures from those that are not.

Changes in damping properties are in general due to local modifications of structural characteristics; conse-quently, variations in power flow and energy density of structures may be exploited to elaborate a methodfor detecting and locating defaults that may have occurred in the structure, The proposed procedure has beenvalidated on several numerical test cases.

14:00 Damage Identification based on Vibration Measurements Applied to MasonryStructures (ID 238)Luıs F. Ramos, University of Minho, PortugalGuido De Roeck, Katholieke Universiteit Leuven, BelgiumPaulo B. Lourenco, University of Minho, PortugalAlfredo Campos-Costa, LNEC, Portugal

The present paper aims to explore damage assessment in the masonry structures at an early stage by vibrationmeasurements. Two replicates of historical constructions were built in virgin state: one arch with 1.5 mspan and one shear wall with 1.0 m2. Afterwards, progressive damage was applied and sequential modalidentification analysis was performed at each damage stage, aiming to find adequate correspondence betweendynamic behavior and internal crack growth. Accelerations and strains in many points were record in thereplicates. Eigen frequencies, mode shapes and modal strains were derived from the dynamic measurements.Environmental effects of the temperature and relative humidity on the dynamic response were studied. Afirst updating process was performed on the results of the undamaged arch to tune a finite element model.Moreover, the tests were repeated with added masses to scale the mode shapes. Finally, a brief analysis ofthe results of the several damage scenarios are presented in the paper

14:00 Experimental Damage Detection Using Observability Grammian Matrix (ID 306)Douglas Domingues Bueno, Clayton Rodrigo Marqui, Vicente Lopes Jr., Universidade EstadualPaulista, BrazilMichael J. Brennan, University of Southampton, United Kingdom

Structural Health Monitoring (SHM) has diverse potential applications, and many groups work in the devel-opment of tools and techniques for monitoring structural performance. These systems use arrays of sensorsand can be integrated with remote or local computers. There are several different approaches that can beused to obtain information about the existence, location and extension of faults by non destructive tests. Inthis paper an experimental technique is proposed for damage location based on an observability grammianmatrix. The dynamic properties of the structure are identified through experimental data using the eigensys-tem realization algorithm (ERA). Experimental tests were carried out in a structure through varying the massof some elements. Output signals were obtained using accelerometers.

95


14:00 Aeroelastic Analyses of Low Aspect Ratio Wings Based on Non-Classic Plate TheoryIncluding Transverse Shear Deformation and Rotary Inertia (ID 314)Saeed Shokrollahi, Malek.E.Ashtar University of Technology, Iran (Islamic Republic of)

A modified aeroelastic analysis of low aspect ratio aircraft wings modeled as thick plates in an incompressibleflow is developed. The structural model takes into account non-classic effects such as transverse sheardeformation and rotary inertia. Three dimensional unsteady vortex lattice method is used to model the flowfield about the wing. The results for limited cases including low thickness plates are in good agreement withprevious works, which were based on classic thin plate theory.

14:00 Squeal prediction on a simplified brake system by complex eigenvalues analysis (ID508)Francesco Massi, University of Rome “La Sapienza” / INSA de Lyon, ItalyLaurent Baillet, University of Joseph Fourier, FranceOliviero Giannini, University of Rome “La Sapienza”, Italy

Brake noise is an example of noise caused by vibration induced by friction forces. During brake operation,the friction between the pad and the disc can induce a dynamic instability in the system. The onset of squealis supposed to occur in linear conditions, during braking phase. A complex eigenvalues analysis of the finiteelement model of a simplified brake apparatus is here adopted to investigate the squeal occurrence. Severalexperimental tests are performed to reproduce different squeal frequencies and to study the dynamics of thesystem in function of driving parameters. The paper shows a good agreement between the dynamic behav-iour predicted by the parametrical complex eigenvalues analysis on the model and the dynamic and squealbehaviour measured on the experimental set-up. The simple dynamics of the system allows distinguishingthree main substructures: disc, caliper and pad. A clear distinction between squeal events involving themodes of the caliper or of the pad is highlighted.

14:00 Vibration Signature Analysis for Engine Condition Monitoring and Diagnosis (ID604)Mohammad Mahjoob, Alireza Zamanian, University of Tehran, Iran (Islamic Republic of)

In this paper, the vibration signals taken experimentally from a typical 4-cylinder IC engine are analyzed.An LMS Pimento analyzer is used along with appropriate sensors (accelerometer and tachometer). Differentpoints were chosen for acceleration (velocity) pick-up (e.g., points adjacent to each cylinder, a point on thecylinder head and another on the engine mount). Valve maladjustment and ignition failure were introducedand the effects were investigated. The test results of healthy/faulty engines were recorded and then processedand analyzed in time-frequency domain. The results showed that correlations could be found between engineperformance and the output signal spectra, STFT (Short Time Fourier Transform), and CWT (ContinuousWavelet Transform) patterns. These facts demonstrate the possibility of using vibration signals for enginemonitoring and diagnosis.

14:00 The Fuzzy Parameterization Method for Model Updating: Application to WeldedJoints (ID 166)Omar Ait-Salem Duque, Polytechnic University of Madrid, INSIA, SpainMaarten De Munck, Andrea Stenti, David Moens, Katholieke Universiteit Leuven, BelgiumA.R. Senın, Nebrija University, Spain

The use of partly rigid beams for modelling welded joints has been used and accepted by many authors butsome major problems that encounter this type of parametrization, like the determination of the initial valuesof the offsets, still haven’t been solved. A new method for subselecting the parameters and its correspondingnominal values is presented in this paper. The novel aspect of this approach is that the procedure makes useof the Fuzzy Finite Element method as a reliable tool for finding the best possible initial conditions of theparameters before updating an FE model, aiming to cope with two of the main problems of model updating:

96


The undetermined problems occurring when the number of parameters outweigh the number of available testdata in one hand, and, avoiding the convergence of the solution to a local minimum on the other hand. Themethod is applied to welded joints which are always a source of uncertainty and that are commonly used inbig automotive structures like the body frames of buses.

14:00 Identification and Characterization of the Non-linear Dynamic Behavior of a Labo-ratory Aircraft Structure (ID 592)Demian Gomes da Silva, Paulo Varoto, University of Sao Paulo, Brazil

Identification as well as characterization of nonlinearities has been a topic of major interest specially fortesting personnel in most aircraft companies. Structural nonlinearities can arise in different situations anddue to several major sources, including but not limited to geometry and material. A special and importantclass of nonlinear vibrating systems arise when the structure under test is parametrically excited. In this casethe excitation appears as a time varying coefficient in the system’s equation of motion. The major goal of thispaper is to present the results of an investigation carried out on a wing type structure specially designed andbuilt to investigate and better understand parametrically excited aircraft structures. It is shown that nonlineardynamic responses are observed from test results for different excitation levels. Particularly, for a giventest condition a very interesting phenomenon of transfer of energy between mode shapes, commonly calledmodal interactions in the literature is experimentally observed.

97

Wednesday September 20, 2006

Human vibrations – HUM1Room 1 – Chairman: J. Blough

8:50 Measurement and identification of the vibration characteristics of motorcycle riders(ID 72)Vittore Cossalter, Alberto Doria, Davide Fabris, Massimo Maso, University of Padova, Italy

This paper shows the experimental responses of five riders to roll and steer oscillations. The analysis is car-ried out to identify the properties of a rider multi-body model, which can be useful for simulation. The ridershave different physiques and riding experience and are excited by means of the motorcycle riding simulatordeveloped at DIM. A stepped sine testing is performed. The results of the steer tests are presented in terms ofFRFs between the angular acceleration of the steer and the steering torque. A multibody model is used to fitthe experimental data. The results of the roll tests are presented in terms of FRFs between the accelerationsmeasured at three point of the body of the rider (head, shoulders, torso) and the acceleration of a point placedat a fixed distance from the oscillation axis. To clarify the dynamic response of the rider, a relative motionFRF is calculated as well: it is the ratio between the relative tangential acceleration and the acceleration atthe same point caused only by the roll angular acceleration.

9:15 Identification methodology of the human pelvis for Low Back Pain diagnosis (ID 89)Nadine Conza, Daniel Rixen, Delft University of Technology, The NetherlandsChris Snijders, Erasmus MC University Medical Center, The Netherlands

The underlying assumption of our research on musculoskeletal disorders assumes that pain in the low backregion, in absence of a diagnosed pathology, might be caused by abnormal biomechanical properties of thesacroiliac joints. The main objective is the development of a medical apparatus for the assessment of theseproperties by means of vibration analysis of the pelvic bones. A dynamic model of the human pelvis has beenbuilt as a first step to establish a methodology for the identification of the sacroiliac joint properties startingfrom the measured vibration. In parallel, a measurement technique to detect the bones’ vibration basedon ultrasound technology has been developed. This paper describes the establishment of the methodologythrough modeling, experimentation, and the development of the apparatus.

9:40 Effect of automobile operating condition on the subjective equivalence of steeringwheel vibration and sound (ID 236)Marco Ajovalasit, Joseph Giacomin, Brunel University, United Kingdom

The research described in this study was performed to define curves of subjective equivalence betweensteering wheel rotational vibration and sound using stimuli from different automobile operating conditions.The steering wheel acceleration stimuli were summarised in terms of the unweighted and Wh weightedr.m.s. values, while the sound stimuli were summarised in terms of the unweighted sound pressure levelin decibels, the A weighted sound pressure level in decibels, the Stevens Mark VI loudness in sones, theStevens Mark VII loudness in sones and the Zwicker loudness in sones. The results suggest that both thestatistical properties of the stimuli, and the choice of metric, effect the shape of the equivalence curve. Nosingle combination of sound and vibration metric produced a family of curves which were separated by lessthan a single psychophysical just noticeable difference.

10:05 Subjective and Objective Vibration Characterization of User Interface Locationsaboard Snowmobiles and ATVs (ID 581)Justin Keske, Jason Blough, Michigan Technological University, United States of America

Comfort and quality have been an increasing concern for manufacturers of recreational vehicles. Advancesin technology have given rise to more sophisticated and luxurious machines. This has changed customerexpectations from the current market. To remain competitive the manufacturers must meet demands forcomfort and quality that customers of the current market have come to expect. Operator vibration exposureis one area of concern.

98


This paper utilizes sensory jury testing to characterize vibration levels perceived by the operator, with re-spect to levels measured using instrumentation. The objective being, to develop a tool for the evaluation ofvibration at the operator interfaces, to be used in the future development of vehicles.

10:30 Optimization of Child Restraint System for Crash-Safety with Child FE HumanModel (ID 250)Takayuki Koizumi, Nobutaka Tsujiuchi, Takashi Nagatani, Doshisha University, JapanHirosuke Horii, CD-adapco JAPAN Co., LTD., Japan

Child restraint system (CRS) has great influence in crash-safety of children. Therefore, much further im-provement of the safety of CRS is requested. The objectives in this study were clarifying the effect ofCRS parameters and exploring the optimums for the crash-safety by using numerical simulation. The nu-merical simulation was carried out by coupling a crash simulation tool, MADYMO with a multiobjectiveoptimization tool, modeFRONTIER. The neck injury risks in frontal and side crashes were set as the ob-jective functions to minimize. Firstly, significance analysis of the design parameters using dummy modelswas conducted. From the result, non-effective parameters were removed. Secondly, CRS optimization forthe safety in both frontal and side crashes was performed using dummy models. Finally, using CRS withthe optimal design, behaviors were compared between dummy models and 3-year-old finite element humanmodel.

Instrumentation – I1Room 1 – Chairman: L. Bregant

11:20 High Channel Count Systems Architecture for Noise and Vibration Measurements(ID 2)Thierry Debelle, K. Veggeberg, National Instruments, United States of America

Modular instrumentation is being widely used in noise and vibration measurement systems that demandhigher channel counts and the wider dynamic range that 24-bit delta-sigma ADCs make available at lowercosts. This is an overview of how flexible modular instrumentation employing the latest software technologycan be used in making high precision noise and vibration measurements where higher sampling rates, higherchannel counts, increased dynamic range, and distributed architectures are needed in smaller packages. Anexample where this is being used is in acoustic beam forming in aircraft pass by noise tests to measure anddistinguish engine and airframe noise sources.

11:45 Non contact sensors comparison for dynamic testing of lightweight structures (ID171)Luigi Bregant, Universita di Trieste, ItalyM. Bressanutti, APE Research, ItalyLorenzo Valent, Antonio Vecchio, LMS International, Belgium

With dynamic testing, engineers describe different activities focused on the identification of some proper-ties of vibrating structures. The acquisition of the necessary data, requires for the measurements on thecomponent of inputs (excitation) and outputs (responses), commonly applying the necessary sensors directlyon the test article. The used load cells and accelerometers, modify the mass and stiffness distributions andsubsequently the eigen-properties of the structure. These errors become more evident when testing lightweight structures. This paper shows the results of tests performed on a small car turbine with the purpose ofidentifying the blade mode shapes. The acquisitions were performed using standard accelerometer and twodifferent contact-less systems. One being a particle velocity sensors, the other being a optical displacementtransducer. The paper shows how both the contact-less approaches can provide very good results even if thesize of the turbine is very small and the blades are extremely light.

99


12:10 Two complementary Microflown based methods to determine the reflection coeffi-cient in situ (ID 661)Hans-Elias de Bree, Microflown Technologies / HAN University, The NetherlandsEmiel Tijs, Microflown Technologies, The NetherlandsTom Basten, TNO Science and Industry, The Netherlands

Two complementary methods have been developed that are capable to determine the reflection coefficient ofacoustic absorbing materials in situ. The methods are compared with a measurement in the Kundt’s tube.

The first method is based on the measurement of the surface impedance. Here a prototype device basedon this method is presented that for calibration and impedance determination uses of a well defined soundsource. The surface impedance method proves to work best for materials with a not too high reflectioncoefficient.

A second (novel) method is developed for materials with a high reflection coefficient. It uses of a monopolesound source that is placed close by the acoustic reflecting material. A Microflown is placed close to themonopole source in such way that its sensitive direction is aiming at the acoustic reflecting material and itsnon sensitive direction is aiming at the source. This way it is only measuring the ‘mirror source’: the reflectedimage of the monopole sound source. It is therefore called ‘the mirror source method’. From the transferfunction between the reference velocity sensor inside the monopole source and the Microflown outside thesource, the reflection coefficient can be determined. First results of this method are presented.

The results of both methods are compared to a Kundt’s tube measurement. Traditionally this method workswith two pressure transducers but it is also possible to use a combinational pressure and velocity transducer(the PU probe). This PU method has already been reported. The calibration should be very accurate to beable to measure highly reflective materials in a reliable way. In this paper an improved calibration procedurefor this method is presented.

Dynamic behavior of civil structures – DCS1Room 1 – Chairman: W. Heylen

14:00 Output-only modal identification of Luiz I Bridge before and after rehabilitation(ID 42)Alvaro Cunha, Filipe Magalhaes, Elsa Caetano, Faculty of Engineering of the University of Porto(FEUP), Portugal

Luiz I Bridge is a double deck metallic arch bridge over Douro river at the city of Porto, in Portugal, builtin 1885 by Eiffel’s former engineer Theophile Seyrig. Recently, the local authorities decided to use theupper deck of this bridge to install a line of the new light metro of Porto, which required the developmentof significant rehabilitation works, involving the modification of the upper deck and the replacement andstrengthening of a significant number of structural components, due to the important level of deteriorationobserved. This paper describes the experimental work developed before and after rehabilitation, aiming theaccurate modal identification of the bridge at two different stages. The data collected in the second testwas analysed using three different output-only modal identification techniques: Peak-Picking, PolyMax andcovariance driven Stochastic Subspace Identification (SSI-COV).

14:25 FE Modelling and Updating of Unique Fink Truss Footbridge (ID 56)Vitomir Racic, Stana Zivanovic, Aleksandar Pavic, University of Sheffield, United Kingdom

Recent improvement of mechanical characteristics of structural materials and fast development of the finiteelement (FE) based computational structural analysis are resulting in innovative solutions in footbridge de-sign. Modern footbridges are often very attractive structures with unusual structural forms. Because of their

100


inherent slenderness, they are often prone to vibration due to human-induced excitation. During design ofthese structures to dynamic load, their dynamic properties are required and are usually obtained from an FEmodel. However, because of structural complexity, the FE modelling of footbridges is a challenge for everydynamic analyst and is often prone to errors. The best and quite often the only way to evaluate the reliabilityof the FE modelling is to compare the FE obtained dynamic properties (natural frequencies and mode shapes)with those measured experimentally. In this way the source of errors in the FE modelling can be identifiedand the FE model can be updated based on the experimental results. The updated model can then be used infurther dynamic analysis.

Recent improvement of mechanical characteristics of structural materials and fast development of the finiteelement (FE) based computational structural analysis are resulting in innovative solutions in footbridge de-sign. When doing research into dynamic analysis of these structures, their dynamic properties are requiredand are usually obtained from an FE model. However, because of structural complexity, the FE modellingof footbridges is often prone to errors due to modelling uncertainty. The best and quite often the only wayto evaluate the reliability of the FE modelling is to involve modal testing and FE model updating of foot-bridge structure with the aim to match test results. Using this approach, a complex and rather unique Finktruss structure of Royal Victoria Dock Bridge in London was analysed. Eleven measured modes of vibrationwere identified via an ambient vibration survey and then compared with their counterparts from an initial FEmodel developed by best engineering judgment. In this initial FE model the maximum difference betweentwo paired natural frequencies was 29%. In a subsequent updating exercise it was found that uncertaintiesin main beam and crosshead geometry, as well as the inherent simplicity of the fully symmetric FE model,were the main source of the modelling error.

14:50 Time domain iterative procedures for vehicle-bridge dynamic interaction (ID 77)Anna Feriani, Universita degli Studi di Brescia, ItalyMaria Gabriella Mulas, Politecnico di Milano, ItalyClaudio Aliprandi, Universita degli Studi di Brescia, Italy

This work, based on the adoption of separate and decoupled mechanical models for the vehicle and thebridge to be analysed in the time domain, aims at developing two different iteration strategies, one on thewhole time history (procedure WTH) and the other on the single time-step (procedure STS). The former canbe easily applied with any commercial code and might deal with slight non linearity; the latter, still underimplementation, can be further developed to include the effects of any non linearity in both the structureand the vehicle. Parametric studies are performed, making use of the WTH procedure, on a 3D model of areinforced concrete simply supported bridge 30.3 m long, and two travelling vehicles modelled as 7-degreesof freedom (DOFs) systems; the effect of road roughness, vehicle mass, speed and frequency are considered.Moreover, the precision and rate of convergence of the WTH procedure are investigated: while appealing forits generality, a theoretical proof of its convergence properties is in fact missing.

15:15 Global vertical modes of floor vibration in multi-storey buildings (ID 630)Zoran Miskovic, Aleksandar Pavic, Paul Reynolds, University of Sheffield, United Kingdom

The paper presents a case study of a real floor structure on the sixth level of a multi-storey office building.‘Global’ modes of vertical floor vibration were identified experimentally. In the ‘global’ modes the wholeor large part of the floor area move with little bending of floor elements such as slab and beams. Detailedpost-test Multi-input-multi-output analysis of the frequency response function data revealed the presenceof the ‘global’ mode having the lowest frequency of all floor vertical modes, and the lowest magnitudesin the frequency response functions. It should be mentioned that similar modes were not identified on thelower (third and fourth) tested levels of the same building. This is probably because of the lower magnitudeof ‘global’ vertical movement at these lower levels due to shorter and less axially deformable columns.Numerical investigation of this behaviour is presented by several finite element models of the tested building,indicating how this behaviour can be analytically studied.

101


15:40 Vibration of the railway track-viaduct system under moving vehicles taking intoaccount the interaction effect (ID 542)Constanca Rigueiro, Institute Polytechnic of Castelo Branco, PortugalCarlos Rebelo, L.S. Silva, University of Coimbra, Portugal

The main purpose of this paper is to investigate the influence of the interaction on the dynamic behaviourof a simply supported single span railway concrete viaduct for which measurement results are available,taking into account the railway track. We compare the computed acceleration response of the track-bridgesystem with and without interaction considering the dynamic characteristics of the real moving vehicles andtaking into account the available modal identification and the acceleration response measurements carriedout during the field tests.

Furthermore, this paper analyse the dynamic behaviour of the railway track and viaduct under the action ofhigh speed vehicles according to EN1990. The moving vehicles are modelled as a series of two degree offreedom mass-spring-damper systems at the axle location The track modelling consists of beam elements forthe rails and a series of springs and dampers for the ballast and the sleepers.

16:05 Design of lightweight floor system with minimized vibration (ID 161)Sander Zegers, Frans van Herwijnen, Technical University of Eindhoven, The Netherlands

During the last four decades floor systems used in housing and office-buildings in the Netherlands weremostly made of stone-like materials, and can be characterized as heavy. In recent years, in light of sustain-able building methods, the trend is to reduce the use of materials and thus build lighter. Lightweight floorstructures are however often found to be more susceptible to vibrations than heavier floor structures. Thevibrations are caused by dynamic actions such as walking persons or vibrating machines such as a washingmachine.

This paper focuses on a beam as a representation for a floor system, supported by hinges with variable ro-tational and translational springs. The influences of the parameters involved are described. An analyticalapproach is used which results in an new approximation formula that can be used in practice. The analyt-ical results are compared to the results found in literature and from numerical calculations. Finally designrecommendations are given for the design of lightweight floor systems.

Damping – D3Room 2 – Chairman: H. Rice

8:50 Influence of the number of parameters of a rubber isolator viscoelastic model on thepredicted dynamic behaviour of a suspended mass (ID 473)Nere Gil-Negrete, Marıa Jose Garcıa Tarrago, Jordi Vinolas, University of Navarra, Spain

Elastomeric joints appearing in many industrial applications are often modeled through a single linear springwith a unique dashpot in parallel to take some damping into account. Nevertheless, this simple viscoelasticmodel, known as Kelvin-Voigt model, does not accurately and fully represent the frequency dependent char-acteristics of the isolator in a frequency range. A fact that, in the long run, affects the simulation results fora whole structure.

The present work compares the predicted vibration behaviour of a suspended mass mounted upon four rubberbushings modeled through viscoelastic models of different number of springs and dashpots. The frequencydependent behaviour of the isolators is experimentally characterized and the parameters of the models arefitted from these results. It is concluded that a generalized model of 5 parameters improves the prediction ofthe transfer function in a frequency range when compared to simpler models.

102


9:15 Hysteretic behaviour of tape-spring actuators: Influence on the deployment of anhexapod (ID 453)Gwenaelle Aridon, Didier Remond, Ahmad Al Majid, INSA de Lyon, FranceLaurent Blanchard, Alcatel Alenia Space, FranceRegis Dufour, INSA de Lyon, France

This paper focuses on a deployable telescope based on an hexapod equipped with tape-spring coiling devices.It aims at understanding the effects of parameters uncertainties on the deployment. Stability during the de-ployment can be better investigated if localized nonlinearities are identified. The deployment analysis startswith the establishment of a single degree of freedom (DOF) force-deflection model for a single deployablecoiling device, the parameters being identified experimentally. In a second step, this actuator model is imple-mented in a simplified bipod model similar to a planar three-DOF mechanism giving a first approach of thedeployment behaviour. Therefore, the two coiling devices are modeled as parallel phenomenological mod-els. In the last part, six actuators models are used to perform a complete model of the deployable hexapod.Finally, the hexapod response is predicted by modifying the restoring force of one of the six actuators.

9:40 Dynamics of rigid bodies with rough contact surfaces and run-in wear (ID 315)Janko Slavic, Miha Boltezar, University of Ljubljana, Slovenia

In this paper we briefly present the force-based theory of plane dynamics of multibody systems with uni-lateral contacts. The theory is already widely used for simulating local contact properties (coefficient ofrestitution, friction, roughness, etc) and we show that it is also well suited for simulating detailed multicon-tact situations of (geometrically) rough contacting surfaces.

Under real-life conditions the outstanding contact asperities are quickly worn away. To include the wear atdynamical loads and to achieve a more realistic contact shape a wear model is introduced. The wear modelbases on the local loss of mechanical energy. The numerical simulations show that after several re-shapings(i.e. wear) a dynamically stable steady state contact shape can be found.

Coefficient of friction was measured at the rim of a wheel (rotating body). Via the multiple-contact dynamicsof rough contact surfaces with the wear-in numerical analysis the experimentally observed change in themeasured coefficient of friction of up to 30% for only slightly altered experimental conditions is successfullyexplained.

10:05 Metal-polymer-ceramic constrained layer damping system (ID 63)Liya Bochkareva, United Institute of Informatics Problems NAS of Belarus, BelarusMaksim Kireitseu, University of Sheffield, United KingdomS. Panin, Institute of strength physics and mesomechanics of materials SB RAS, Russian Federa-tion

The focus in this paper is directed toward to the investigation into hybrid metal-polymer-ceramic dampingcoatings. Sandwiched coating’s layers have been manufactured by advanced technology based on combina-tion of thermal flame spraying of foamy aluminum or its alloy and viscous-elastic polymeric (polyamide)layers, including 1-4 wt.% reinforcing additives of carbon nanotubes. Technological aspects of manufactur-ing are being discussed in terms of vibration damping properties and strength of the materials. The CNT-reinforced material is formed as a constrained-layer damping coating system (CLD) showing the greatestrange over which damping have excellent energy absorbing/damping properties over bulk materials. Modalloss factors obtained at 60-80% partial-coverage of tested specimens was higher than that of maximumdamping obtained in the case of 100% specimen coverage. FEM-based computer modeling was used to pre-dict the damping of multilayer materials by a strain energy method. An effective damping design involvesselecting a proper combination of coverage area, relative thickness and stiffness values of the CLD config-uration. The coating could be recommended for applications in sliding and rolling bearing units, low-speedand light-weight gearing drives and other machine parts.

103


Damping – D4Room 2 – Chairman: H. Rice

10:55 Investigation of the sloshing effects of an aeroplane fuel tank without and includingthe new eXess-aluminium mesh (ID 23)Norbert Niedbal, Guido Rosenhager, Fachhochschule Bielefeld, Germany

The European research project “AEROSAFE” was designed to test a new aluminium mesh or, more specifi-cally, a new aluminium honeycomb structure (to increase the in-flight safety of fuel tanks by means of newfire prevention properties and the reduction of the sloshing effects of the fuel of aeroplanes. This project dealswith the measurement and the analysis of sloshing effects in fuel tanks of aeroplanes. This report includesthe design and the construction of a suitable test device, the performed vibration tests and the evaluation ofthe test data.

11:20 On the Modelling of Damping in Structural Vibrations (ID 523)Antonio Ribeiro, Nuno Maia, Julio Silva, Instituto Superior Tecnico, Portugal

The viscous damping model is a well established model, widely used both for free and forced vibrations,though particularly suited for the free response analysis. On the other hand, the hysteretic damping modelhas only been used when studying forced structural vibrations. Its simplicity and accuracy are its mainadvantages, but one easily encounters real difficulties - from the mathematical point of view - when dealingwith free responses.

In a recent paper, the problem of modelling the free response of a single degree-of-freedom using hystereticdamping has been addressed by the authors, who proposed a solution capable of circumventing the mathe-matical inconsistencies.

In this paper the authors propose a more general approach, which encompasses both damping models, a kindof unified approach. Solutions for the single degree-of-freedom system are explored and discussed.

11:45 Fractional damping Induced by a slave system attached to a master oscillator (ID532)Antonio Carcaterra, Oliviero Giannini, University of Rome “La Sapienza”, ItalyAdnan Akay, Carnegie Mellon Unyversity, United States of America

This paper illustrates a class of mechanical systems exhibiting a damping mechanism based on fractionalderivatives. In particular it is shown how a set of small oscillators of appropriate characteristics attached toan oscillator of larger mass produces an effect that globally can amount to a fractional derivative dissipationmechanism. More precisely the fractional derivative model depends on the choice of frequency distributionselected within the set of the attached small oscillators. On this basis a general theory that provides alink between fractional damping and damping induced by a slave system attached to a primary structure ispresented.

104


Passive control of noise and vibration – PNVC1Room 2 – Chairman: W. Lauriks

14:00 Study on sound absorption of a multi-leaf microperforated panel (ID 465)I Made Miasa, Gota Kishimoto, Masaaki Okuma, Tokyo Institute of Technology, Japan

A theoretical method is presented for calculating the absorption coefficient of a multi-leaf microperforatedpanel absorber. In this work, the possibility of using multi-leaf type of microperforated panels without solidbacking panel is investigated. The number of panels varies from 2 to n (¿¿2). The sound absorption coef-ficient is calculated based on electro-acoustic analogy by taking the sound transmission coefficient due tothe absence of solid backing panel into consideration. The optimum performance of leaf microperforatedpanel absorber is also investigated as the function with respect to the design parameters such as perforationratio, diameter of holes and distance between panels. The numerical results show that the configuration ofleaf of microperforated panels can increase absorption coefficient at low frequency region. This fact willgive an advantage to the multi-leaf microperforated panel over conventional microperforated panel arrange-ments. That is, to solve the problem of quite narrow frequency band absorption effect, especially in mid-highfrequency region corresponding to their resonant frequencies, that the conventional microperforated panelarrangements inherently possess. Some feasible application ideas of the multi-leaf type of arrangement arealso presented.

14:25 Control of sound transmission through double wall partitions using optimally tunedHelmholtz resonators (ID 593)Stanislaw Pietrzko, Qibo Mao, Empa - Materials Science & Technology, Switzerland

An arrangement of Helmholtz resonators (HRs) used for the passive control of sound transmission througha double wall partition is investigated. By using the modal expansion method, the partitions walls, acousticcavity field as well as Helmholtz resonators are described by a set of partial differential equations, lead-ing to a full analytical model. The Helmholtz resonator sound transmission control mechanism is basedon adding damping to the acoustical resonances of the cavity inside the double wall partitions. Firstly, amethod to compute optimal damping ratios for the whole arrangement of Helmholtz resonators guaranteedbroadband improvement of sound isolation is presented. Secondly, due to the strong coupling between thecavity acoustic modes and structural modes of the partition walls around the mass-air-mass resonance fre-quency, tuning the Helmholtz resonators exact to the mass-air-mass resonance frequency cannot guaranteeimprovement on transmission loss. Optimal natural frequencies of Helmholtz resonators were found as a so-lution of optimization problem with a global cost function defined as frequency averaged sound transmissionloss in the desired frequency range (around mass-air-mass resonance frequency). Finally, some simulationresults are presented. The simulation results confirm that using optimally tuned arrangement of Helmholtzresonators the sound transmission loss of a double wall partitions can be improved significantly in the desiredfrequency range. In presented case an overall improvement 8 dB in the 50 - 150 Hz band is obtained by using1 resonator (requiring only 2.8% of the double plate cavity volume), and a 10 dB improvement obtained byusing 3 resonators (requiring 8.4% of cavity volume).

14:50 Novel Passive Vibration Isolators (ID 201)Alexander Forrester, Andy Keane, University of Southampton, United Kingdom

We present the design of a novel anti-vibration mounting. The mounting takes the form of a triangulartruss which is ‘folded’ such that each section extends in the opposite direction to the previous section. Thegeometry of this complex, compact structure is optimised to exploit the reflections that occur in vibrationalwaves travelling through the structure to provide significant levels of vibration isolation. In this study wehave achieved a 15dB reduction in vibration energy compared to a baseline regular structure.

105


15:15 Transmissibility of Forces in Multiple-Degree-of-Freedom Systems (ID 370)Nuno Maia, Mihail Fontul, Antonio Ribeiro, Instituto Superior Tecnico, Portugal

The present paper addresses the problem of force transmissibility, i.e., the relation between transmittedand applied forces. In the basic linear vibration theory the ratio between transmitted and applied force isusually deduced for the single-degree-of-freedom system, such a ratio being called transmissibility. It is alsocommon to deduce the relation between output displacement and a given input displacement and to showthat both transmissibility expressions so obtained are, mathematically, the same. The authors, in previousarticles, have extended the transmissibility of displacements to the multiple degree of freedom system case,both for periodic and random signals. In the present article the objective is to pursue a similar generalization,this time relating two sets of forces, or more precisely, to deduce a formulation that enables us to calculate thetransmitted forces from a set of applied ones. Such a generalization is achieved, some discussion is providedand some simple examples help illustrating the new developments.

15:40 Rubber isolating elements to prevent vibro-acoustic propagation (ID 412)Stefano Manzoni, Marcello Vanali, Politecnico di Milano, ItalyCarlo Carbone, Studio di architettura Carlo Carbone, Italy

This paper deals with the vibro-acoustic isolation of the rooms used for the daily radio broadcasts. Thebasic studio structure is composed by two rooms divided by a wall with a window. One of the rooms isoccupied by the anchorman and the other one is dedicated to the broadcast producer. It is very importantthat the room exploited by the anchorman is very well isolated both from the external environment and fromthe room in which the producer works. In the paper it is shown that the isolation does not only depends onthe studio acoustic design but also on the hosting building structure. In the considered case the reinforcedconcrete floor under the two rooms was particularly sensible to vibration phenomena, allowing an anomalousvibration propagation towards the anchorman studio through a mechanical short-circuit. In this situation anew isolation between the studio floating floor and the supporting structure had to be studied, employing twolayers of different materials.

16:05 Static Analysis of a Quasi-Zero-Stiffness Vibration Isolator (ID 455)Alessandro Carrella, Michael J. Brennan, Tim Waters, University of Southampton, United King-dom

The frequency range over which a mount can isolate a mass from a vibrating base (or vice versa) is oftenlimited by the mount stiffness required to support the weight of the mass. This compromise can be mademore favourable by employing non-linear mounts with a softening spring characteristic such that small ex-cursions about the static equilibrium position result in small dynamic spring forces and a correspondinglylow natural frequency. In this paper the force-displacement characteristic of a so-called quasi-zero-stiffness(QZS) mechanism is studied. Herein the system comprises a vertical spring acting in parallel with two furthersprings inclined at an angle to the horizontal. A QZS characteristic can be obtained if the system’s parameters(angle of inclination and ratio of spring stiffnesses) are opportunely chosen. Finally, the forcedisplacementrelationship is approximated by a polynomial of third order which reduces to the Duffing oscillator when thesystem is optimally tuned.

106


EDSVS – EDSVS1Room 3 – Chairman: N.

8:50 Experimental investigation on the MRTLCD for vibration control of slender struc-tures (ID 607)Shane Colwell, Andre Jakob, Technische Universitat Berlin, Germany

An experimental investigation into the effectiveness of Magnetorheological Tuned Liquid Column Dampers(MRTLCDs) for the reduction of structural accelerations in a cantilevered (SDOF) structure subjected to si-nusoidal and random excitation is carried out. A tall, slender annular pipe is used to model the tubular towerof a flexible structure and a pre-determined steel weight was placed at the top of the tube in order to deliverthe required fundamental natural frequency. An MR fluid, which possesses a relatively low viscosity anddensity, is used as the residing liquid in the MRTLCD. The response of the structure- MRTLCD system tosinusoidal and random excitations with various magnetic flux densities applied transversely to the fluid flowin the MRTLCD is investigated. The advantages of MRTLCDs and the practicality of utilising MR fluids inTLCDs are examined and concluded upon.

9:15 Decentralised feedback control systems in double panels (ID 642)Neven Alujevic, P. Gardonio, University of Southampton, United Kingdom

This paper presents a theoretical study of active structural acoustic control on a double panel vibroacousticsystem. The system consists of two vibrating panels, which are structurally and acoustically coupled viaelastic mounts and the air in the cavity between the panels. The “source” panel is excited by an acousticplane wave, while the “radiating” panel radiates sound into free-field. A mathematical model for the problemhas been developed, which considers the fully coupled configuration. Two decentralised MIMO feedbackstrategies for the control of the low frequency noise transmission, active damping and active mass, have beeninvestigated and contrasted. Simulations of the sound transmission through the controlled system have beenperformed, and the results discussed.

9:40 Direct hybrid coupled finite element - wave based technique for 3D steady-stateacoustic analysis (ID 619)Petra Silar, Achim Hepberger, ACC Acoustic Competence Center G.m.B.H., AustriaHans-Herwig Priebsch, ACC Acoustic Competence Center G.m.B.H. / Graz Technical University,AustriaH. Pramberger, AVL List GmbH, AustriaT. Bartosch, MAGNA STEYR Fahrzeugtechnik AG & Co KG, Austria

As alternative to the well known finite element method (FEM) the recently developed wave based technique(WBT) has proven to be a computationally efficient tool for steady-state acoustic analysis. To overcome themoderate geometrical complexity restriction of WBT a hybrid coupling method between FEM and WBThas been developed. This coupling approach benefits from the advantages of both methods combining theflexibility of FEM for complex geometries and of the computational efficiency of WBT.

This paper describes the extensions necessary for the 3D hybrid coupling method originally developed for2D problems. Specifically, the characteristic of the coupling matrices for the direct coupling method of FEMand WBT is described. First results of application examples considering different boundary conditions arediscussed. The results of the hybrid coupled method are validated by results of pure WBT simulation and bymeasurements.

107


10:05 A Method to Characterise Structure Borne Noise of a Source from In-Situ Measure-ment (ID 631)Andrew Elliott, University of Salford, United KingdomGoran Pavic, INSA de Lyon / University of Salford, FranceAndy Moorhouse, University of Salford, United Kingdom

Due in part to the difficulties encountered when characterising structure borne sound sources, a potentiallysignificant contributor to overall product noise is often overlooked. A simple useable method for structureborne source characterisation would therefore be very valuable. For structure borne sound a source wouldgenerally be characterised by the mobility and free velocity at its connection points. These quantities canbe difficult to measure accurately or representatively and generally require the physical separation of sourceand receiver which may be difficult or not realistic in some cases. This paper is concerned with the develop-ment of an in-situ method which allows mobilities and free velocities to be found from measurements on asource and receiver in a coupled state. The feasibility of the method is explored and the results from somepreliminary experiments are presented.

10:30 Presenting Noise Maps in a 3D Environment (ID 483)Francesco Pilla, Henry Rice, Trinity College of Dublin, Ireland

The project this paper deals with concerns finding a new approach to the study of environmental noisepollution with the purpose to present solutions in a more comprehensible and accessible way, to increasepublic participation to this matter. A 3D model of the area of interest is created using either LI.D.A.R. dataor a pair of stereo photos, to extract from it images with the purpose of giving a quick and comprehensibleshot on how an area could change with the insertion of barriers, allowing the prediction of the solution witha lower environmental impact. The resulting 3D model is imported into Predictor and a noise map is created:both are put on software that allows a user fly-through, to have a 3D environment in which the user is free to“go around” and have a sort of immersion in the manner of noise pollution.

The final objective of this work is to have a high quality of data concerning the noise map as that given byPredictor, but allowing a much easier understanding of the outcomes for the general uninformed public.

10:55 Reduced modal models and negative concept modifications in dynamic analysis (ID13)Emiliano Mucchi, University of Ferrara, ItalyStijn Donders, Rabah Hadjit, LMS International, BelgiumWim Desmet, Paul Sas, Katholieke Universiteit Leuven, Belgium

In this paper a concept CAE approach to efficiently modify the properties of beam-like sections and jointconnections of a vehicle body is presented. Negative concept modifications in the beams and in the jointsare analyzed using reduced models. Standard beam elements are used to implement the modifications in thebeam-like section, where the joint modifications are considered through Guyan superelements. Examplesare presented for beam thickness and joint stiffness modifications, but one can consider any properties: itwill be shown that by adding beams to the beam model and by applying Guyan superelements to the jointmodel one thus obtains a much smaller models for fast modification analysis. The proposed approach is thendemonstrated on an industrial vehicle model to quickly and accurately optimise the low-medium frequencybehaviour.

108


Substructuring and coupling – SC1Room 3 – Chairman: B. Mace

14:00 Damping Allocation in Automotive Structures using Reduced Models (ID 102)Nicolas Roy, Top Modal, FranceSylvain Germes, Benjamin Lefebvre, PSA Peugeot Citroen, FranceEtienne Balmes, Ecole Centrale de Paris / SDTools, France

The use of passive damping has shown to be very effective in reducing low frequency mechanical vibrationsin passenger vehicles. It is therefore necessary to account for damping at the design level by the introductionof damping specifications related to the vehicle’s structural components or substructures.

This paper describes an ongoing development at PSA of a methodology and software tool for assessingand optimizing the allocation of damping in complex industrial structures. The tool makes extensive use ofsubstructuring and mode superposition techniques offering computational efficiency and physical insight inorder to quickly determine and understand the influence of the components on system behavior.

14:25 Assignment of receptances by added mass-spring absorbers (ID 173)Jianfeng Zhu, John Mottershead, University of Liverpool, United Kingdom

This paper addresses the problem of receptance assignment to a multiple degree of freedom system by oneor more added simple mass-spring absorbers. The added masses and spring stiffnesses may be determinedusing receptances of the original system and are generally obtained by solving a system of multivatiatepolynomials.. The analysis is illustrated by a series of numerical examples. Realistic solutions requirepositive values for the absorber parameters, so that there may be no acceptable solution, a unique solution orthere may be finitely many acceptable solutions.

14:50 Dual Assembly of substructures and the FBS Method: Application to the DynamicTesting of a Guitar (ID 195)Daniel Rixen, Delft University of Technology, The NetherlandsThibaut Godeby, Emmanuel Pagnacco, INSA de Rouen, France

Dynamic substructuring techniques allow to assemble the dynamic response measured for subcomponents(in terms of frequency response function) and to build the experimental dynamic model for the completesystem. The most straightforward and commonly used approach is the Frequency Based Substructuringtechniques. Although the theory underlying the method is well understood, experimental substructuringtechniques usually fail to provide accurate global models due for instance to the high sensitivity of the accu-racy of the model to measurement errors and to the difficulty in measuring rotational degrees of freedom onthe interfaces. In this paper we will show that, at least for simple structures, the Frequency Based Substruc-turing approach leads to accurate models if clean measurements are done. Also we discuss the case study ofa substructuring exercise performed on a guitar that allowed to better understand the interaction between thewooden instrument (body and fret board) and the strings.

15:15 On the reduction of quasi-cyclic disk models with variable rotation speeds (ID 397)Arnaud Sternchuss, Ecole Centrale de Paris, FranceEtienne Balmes, Ecole Centrale de Paris / SDTools, France

New methods for the dynamic analysis of tuned and mistuned bladed disks are presented. A new techniquefor disk substructuring using parameterized superelements is first described. It allows individual blade mis-tuning to be taken into account as slight variations in the reduced stiffness matrix of the bladed sectors or asshifts in fixed blade frequencies. A parametric multi-model reduction approach is then introduced to allowpredictions of the evolution of vibration characteristics with respect to the rotation speed. Using prestressedsolutions at three rotations speeds, this method allows a very accurate reanalysis of modes and forced re-sponses of mistuned disks at all intermediate speeds.

109


Rotating machinery:dynamics – RMD3Room 4 – Chairman: J. Antoni

8:50 Modeling the structural dynamics of horizontal axis wind turbine (ID 231)Mariusz Pawlak, Arkadiusz Mezyk, Silesian University of Technology, Poland

As the horizontal axis wind turbines are getting larger, their dynamic behavior is becoming more important.Dynamic analysis gives a knowledge how to improve efficiency and safety also in small wind turbines. Thisarticle describes FEM model of upwind, three-bladed wind turbine. Geometry of each component is gener-ated separately and then assembled together by transformation matrices. Material of the blades is composite,the tower is assumed to be made of steel. The blades are modeled by shell elements. The model takes intoaccount aerodynamic load of blades and tower, inertia forces due to rotation of the rotor. The aerodynamicloads, calculated according to the modified Blade Element Momentum theory are attached to aerodynamiccenters. The model is ready to be used in transient analysis and also for investigation of material and geo-metric modification. The transient analysis contains calculation of deflection and stress distribution in thewind turbine.

9:15 Steam whirl stability margin in a power unit (ID 476)Nicolo Bachschmid, Paolo Pennacchi, Ezio Tanzi, Andrea Vania, Politecnico di Milano, Italy

In addition to the geometric properties of the seals, the characteristics of the steam flow play a basic role inthe generation of steam-whirl instability phenomena in rotating machines. The hardening of the seal stiffnesscaused by a raise of the steam pressure and flow often provides a significant contribution to the generation ofunstable vibrations. With regards to model-based investigations the increase of the crosscoupled coefficientsused to model the seal stiffness can cause important changes in the eigenvalues and eigenmodes of themathematical model of the rotating machine. In particular, the real part of the eigenvalue associated with thefirst flexural normal mode of the turbine shaft may become positive causing the conditions for the occurrenceof unstable vibrations. The threshold level of the steam flow that causes instability conditions can be usedto define the stability margin of the power unit. This paper shows the results of investigations carried outwith model-based techniques by means of which the dynamic behavior of a large power unit affected bysteam-whirl instability phenomena have been studied.

9:40 Counterweight balancing for machine frame vibration reduction: design and ro-bustness analysis (ID 470)Myriam Verschuure, Bram Demeulenaere, Jan Swevers, Joris De Schutter, Katholieke UniversiteitLeuven, Belgium

By addition of counterweights to the moving links of a linkage, supported by an elastically mounted frame,it is possible to reduce the frame vibration induced by the resulting forces and moments of the linkage onthe frame. Determining the counterweights that yield a maximal reduction in frame vibration is a nonlinearoptimization problem. This paper shows that this optimization problem can be reformulated as a convexproblem, i.e. a nonlinear optimization problem that has a unique (global) optimum. This methodology isgeneral but developed here for a planar four-bar linkage. For the particular example considered here, arobustness analysis shows a significant reduction of frame vibration even for drive speeds and mountingparameters other than those considered during the counterweight design.

10:05 Vibro-acoustic measurements for the identification of incoming stall in axial com-pressors (ID 375)Giorgio Dalpiaz, Gianluca D’Elia, Emiliano Mucchi, University of Ferrara, ItalyAlessandro Rivola, University of Bologna, Italy

This work addresses on a complete vibro-acoustic characterization of an axial compressor with the aim toforesee the rotor instability. The tests were performed on a turboshaft Allison 250-C18. The compressor is

110


composed of six axial stages and one centrifugal stage. Four vibration signals were simultaneously measuredby means of accelerometers, while the acoustic signals were measured by means of two microphones. Twodifferent kinds of tests have been carried out on the compressor that operates at constant speed: in the courseof the first test the six signals were acquired at different positions of the throttle opening, whereas duringthe second test, the signals were acquired while the throttle was gradually opened. The test results show asensitive increase of the sub-synchronous activity in the accelerometers spectrum map, moreover, closingthe throttle, the amplitude of the spectrum components increases. These phenomena can be related to therotating stall behavior.

10:30 Modal Balancing of turbo-generator rotors using simplified spatial models (ID 289)Jorge Enrique Aguirre-Romano, Instituto de Investigaciones Electricas, MexicoJeily Cervantes Contreras, Enrique Simon Gutierrez Wing, Centro Nacional de Investigacion yDesarrollo Tecnologico, Mexico

Existing spatial models can be the basis for the modal balancing of large turbo-generators. This paperaddresses the question of whether through simplification such models can be prepared and used in field bal-ancing activities, in spite of time limitations, pressure and information restrictions that are often encounteredin field balancing tasks. A methodology is presented for the elaboration of simplified models which showresults that may be similar to those obtained with detailed models.

Aeroacoustics and flow noise – AA3Room 4 – Chairman: M. Baelmans

11:20 Jet Flow induced noise computed using Large Eddy Simulations (ID 220)Yves Detandt, Gerard Degrez, Universite Libre de Bruxelles, BelgiumStephane Caro, Free Field Technlogies, Belgium

Aero-Thermo-Mechanics department of ULB is currently developping a code based on a mixed Spectral/Finite Elements discretisation offering a natural decomposition of a fully three-dimensionnal into a set oftwo-dimensional problems. The reduced simulation time compared to complete three-dimensional code of-fers new fields of investigation for the turbulence modeling. In the present paper, we demonstrate the interestof Large Eddy Simulations for aeroacoustic computations.

11:45 Vibro-acoustic interactions with mean flows. Coupling of Galbrun’s and Biot’sequations. (ID 214)Emmanuel Redon, Orlando Andrianarison, Roger Ohayon, I.S.A.T. Nevers, France

Noise reduction is a crucial factor while designing vehicles, especially in the automotive and aircraft indus-tries.

This paper addresses the problem of predicting the sound attenuation in ducts lined with absorbing mate-rials. Such problems often deal with using absorbing materials such as fibrous media (for example glassfiber) or elastic porous materials (such as foams). We focus our study on the modelling of porous liningmaterials. Generally speaking, absorbing materials can be modelled either by using local wall impedance orby explicitly taking into account the poroelastic material behaviour.

In this paper, theoretical formulations of vibro-acoustic interactions in lined ducts conveying flow is thusproposed. Galbrun’s equations (used in linear aeroacoustics) are chosen to describe the flowing fluid domainwhereas the porous material is modelled through Biot’s equations (generally used for sound absorption andvibration damping).

The coupling between Galbrun’s equations and Biot’s poroelastic equations allows us to propose a mod-elling tool with consistent variables (displacement and pressure). Appropriate variational formulations aredeveloped and guidelines for their numerical implementation are given.

111


12:10 Airflow and acoustic optimization of cooling devices using 3D particle image ve-locimetry (ID 208)Jose Cordova, Intel Tecnologia de Mexico, MexicoMark Trautman, Marco Beltman, Intel Corporation, United States of America

This paper outlines an experimental setup developed to measure the flow field generated by electronicscooling fans and a method to analyze the interactions between fans and near field obstructions such as strutsand heat sinks. A test facility equipped with 3D laser particle image velocimetry diagnostics was developedfor this purpose. The fan impeller was rigidly mounted while the fan housing and struts were affixed to atraverse mechanism so that the position of these components could be independently prescribed. The setupis also equipped with microphones to measure radiated noise so that the spatial proximity effects on soundgeneration can be characterized in a controlled manner, allowing for optimization through experimentalanalysis.

Rotating machinery: monitoring and diagnostics – RMM1Room 4 – Chairman: B. Randall

14:00 Using of Matching Pursuit and Genetic Algorithms for Bearings’ Fault Detection,Diagnosis and Prediction (ID 667)Florin Ionescu, HTWG-University of Applied Sciences-Konstanz, GermanyDan Stefanoiu, University Politehnica of Bucharest, Romania

Signals encoding information of the existing defects or possible failures in a system are sometimes difficultto analyze because of various corrupting noises. Such signals are usually acquired in difficult conditions,far from the place where defects are located and/or within a noisy environment. Detecting and diagnosingthe defects require then quite sophisticated methods that should be able to make the distinction betweennoises due to defects and another parasite signals, all mixed together in an unknown way. Such a methodis introduced in this paper. The method combines a genetic algorithm with time-frequency-scale analysis ofmechanical vibration.

14:25 Identification of the Running-State of Railway Wheelsets (ID 241)Lars Reicke, University of Hannover, GermanyI. Kaiser, German Aerospace Center (DLR), GermanyM. Kroger, University of Hannover, Germany

This contribution deals with the identification of the running-state of railway wheelsets by means of accel-erations measured at the axle-boxes. The interaction between rail and wheel causes rigid-body motions aswell as structural vibrations. First of all the accelerations are decomposed into their symmetric and antimet-ric components, followed by a time frequency analysis using the short-time Fourier transform. Simulationsand experiments on a full-scale test rig underline the potentials of this approach for the identification of thewheelset’s bending modes. The torsional modes cannot be directly observed at the axle-boxes, they have tobe estimated by the dynamic interaction between the wheelset’s yawing and lateral motion.

14:50 Combustion diagnosis for internal combustion engines with real-time acquisitionand processing (ID 513)Stephan Ker, Fabrice Guillemin, Laurent Duval, IFP, France

The use of vibration signals for internal combustion engines diagnosis is now well established for severalapplications such as knock detection or valve mechanics behaviour assessment. If standard knock detectioncan be performed in real time, diagnoses are generally carried out off-line due to algorithm complexity. Ouraim was to get some insights on the time-frequency properties of the vibration signal in order to improve

112


off-line techniques and propose adapted real time algorithms for combustion diagnosis. This paper addressesthe problem of the identification of combustion related vibrations among others noise sources. In order tohelp this identification, pressure traces from in-cylinder sensors are also recorded and processed. Thus,pressure forces that generate the “thermal noise” in the vibration signal can be analyzed separately. The“thermal noise” associated events can be more easily isolated from the mechanical noise in the vibrationsignal mixture.

15:15 Modelling dynamic interactions of rolling elements in bearings for condition moni-toring (ID 22)Joao Guimaraes, RWTH Aachen, Germany

Bearings are one of the most common components in machines. Their movement and dynamic contributesto the overall vibration in a machine and is also influenced by other moving parts in it.

Due to the difficulty to exactly measure this excitation of the source, one has to try to physically model it.This was done adapting the theory of rough rolling contact to the case of rolling bearings. The model takesinto account not only the type and geometry of the bearing, but also the dynamic of this element and theradial load applied to it. In this way, it is possible to predict and simulate the time evolution of the excitationgenerated by the bearing’s movement - dependent on the condition of the surfaces - that will be imposed tothe machine by this source.

This work describes the physical model of the rough rolling contact in bearings and the procedure to simulatethe excitation produced by it. The main interest lies on the early stages of operation, where no severe damageon the bearing took place. This means that excitations due to impact are not the leading mechanism ofstructure-borne sound generation, but the contact between rollers and races.

15:40 Modal Control of Vibration in Rotating Machines and Other Generally DampedSystems (ID 99)Paul Houlston, Seamus Garvey, Atanas Popov, University of Nottingham, United Kingdom

Second order matrix equations arise in the description of real dynamical systems. Traditional modal controlapproaches utilise the eigenvectors of the undamped system to diagonalise the system matrices. A regret-table consequence of this approach is the discarding of residual off-diagonal terms in the modal dampingmatrix. This has particular importance for systems containing skew-symmetry in the damping matrix whichis entirely discarded in the modal damping matrix. In this paper a method to utilise modal control usingthe decoupled second order matrix equations involving non-classical damping is proposed. An example ofmodal control sucessfully applied to a rotating system is presented in which the system damping matrixcontains skew-symmetric components.

16:05 Condition Monitoring of Rotating Machinery using Active Magnetic Bearings (ID414)John Penny, Aston University, United KingdomMichael Friswell, University of Bristol, United KingdomChunliang Zhou, Harbin Engineering University, China

The concept that changes in the dynamic behaviour of a rotor could be used for general fault detectionand monitoring is well established. Current methods rely on the response of the machine to unbalanceexcitation and are mainly based on pattern recognition approaches. However these methods are relativelyinsensitive and the crack must be large before it can be robustly detected. Active magnetic bearings (AMB)have been used in high speed applications or where oil contamination must be prevented, although theirlow load capacity restricts the scope of applications. Recently AMBs have been proposed as an actuator toapply force to the shaft of a machine. The presence of the crack generates responses containing frequenciesat combinations of the rotor spin speed and applied force. This paper discusses some of the issues to beaddressed to enable this approach to become a robust condition monitoring technique for cracked shafts.The approach is illustrated with a number of simulated examples.

113


Aeroacoustics and flow noise – AA2Room 5 – Chairman: G.A. Matesanz

8:50 Application of a Boundary Element reformulation of an aeroacoustical analogybased on incompressible flow data (ID 641)Christophe Schram, Michel Tournour, LMS International, Belgium

We propose an implementation of the hybrid approach based on Curle’s analogy, which formally describesthe mechanism of sound generated by body-turbulence interaction. The acoustic propagation is carried outusing an acoustic code based on the Boundary Element Method (BEM), in which scattering by arbitrarygeometries can be computed, including impedance effects where needed. This was proved to give reliableresults for cases where the body dimensions are fairly acoustically compact, and/or when the CFD model isaccurate enough to capture the acoustical information over the turbulence-body interaction region. Difficul-ties arise however when the spatial extent of this region becomes comparable to, or larger than the acousticalwavelength, and when the flow data is provided by an incompressible model. A previous study has shownthat a straightforward application of Curle’s analogy to non-compact geometries, and in which acousticsource terms (in particular surface source terms) do not account for compressibility effects, can yield a poorestimation of the acoustical far-field. An innovative procedure is therefore developed and demonstrated onsimplified cases.

9:15 Experimental identification of a boundary pressure field induced by a turbulent flowfrom plate vibration measurements (ID 187)Charles Pezerat, Nicolas Totaro, INSA de Lyon, FranceMarc Pachebat, PSA Peugeot Citroen, FranceJean-Louis Guyader, INSA de Lyon, France

Boundary pressure induced by a turbulent flow acting on a structure is an important vibration source thatgenerates broad-band noise inside cars. The classical way to identify the exciting pressure is the measurementof boundary layer pressure in order to characterize correlation length to be used in a Corcos-like model forboundary pressure cross spectrum. This is a complicated and time consuming experimental procedure, thatneeds a dedicated set-up. Here, an indirect measurement of the wall pressure fluctuation is proposed. Theprinciple is based on a method, called RIFF, which was developed in the last years in case of mechanical oracoustic excitations. This method is based on the measurement of the vibration field of the structure. Theadvantage of this method is that the force distribution applied to the structure is reconstructed locally andthus does not depend on the conditions located at the boundaries of the window of measurement. In thispaper, the possibility of reconstructing by RIFF the boundary pressure field produced by obstacles placed inflow is investigated.

9:40 Jet noise modelling for shielding calculations using RANS solution source localisa-tion and WEM propagation (ID 420)Ciaran O’Reilly, Henry Rice, Trinity College of Dublin, Ireland

In this paper, a method to determine the jet noise source input for shielding calculations is described. Themethod involves three steps - localisation, propagation and simplification. Firstly, for the frequency of in-terest, a small localised source region is determined, using Goldstein-Rosenbaum’s acoustic model, with anumerical RANS solution providing local flow properties. A number of distinct incoherent noise sourcesmay be located in this region, using the turbulence length scales. In the second step, each source is propa-gated out through the jet flow using the wave expansion method (WEM) to solve the Helmholtz equation,with local Mach values acquired from the RANS solution. The third step is to define a Kirchhoff surface out-side the jet flow, from which the sound may be propagated further by simply using the Helmholtz-Kirchhoffintegral, which can account for any mean flow present. Although the method is quite crude, it is relativelyrobust as it is a shielding factor or ratio, rather than an absolute value, which is of interest in design evalu-

114


ation. Preliminary results, presented in this paper, provide encouragement that this new three-step jet noisesource modelling method can be used to provide a source input for use in airframe shielding calculations,that is equivalent - in terms of source frequency, distribution and directivity - to the noise produced by a jetflow.

10:05 Influence of irregular grids on the accuracy of a Quadrature-Free DiscontinuousGalerkin Method for solving the LEE equations (ID 294)Yves Reymen, Martine Baelmans, Wim Desmet, Katholieke Universiteit Leuven, Belgium

This paper assesses the effect of highly irregular meshes on the accuracy of the Quadrature-Free Discontin-uous Galerkin Method for solving the Linearized Euler Equations(LEE) equations. The method has beenapplied to two test cases and compared with analytical reference solutions on a large number of meshes,ranging from Cartesian to irregular meshes with random node distributions.

Active vibration control and smart structures – AVC3Room 5 – Chairman: J. Swevers

10:55 Active Vibration Control in a Centerless Grinding Machine (ID 190)Iker Garitaonandia, Joseba Albizuri, Maria Helena Fernandes, Jesus Marıa Hernandez, Universityof the Basque Country, SpainXabier Sabalza, IDEKO R&D Centre, Spain

In this paper, a method to simulate active vibration control in a centerless grinding machine is presented. Theobjective is to design an effective control scheme to reduce instability problems due to regenerative chatter.

Using as reference the results obtained from the experimental modal analysis (EMA), the finite elementmodel (FEM) of the machine was validated using correlation techniques. The FE model was updated usinga sensitivity-based parameter estimation indirect technique.

The modal truncation technique was used to obtain a reduced order state space model. This reduced ordermodel was used to predict the dynamic behaviour of the centerless grinding machine. The mathematicalmodel of the regenerative cutting process was used to perform a numerical stability analysis.

Using the reduced order model, an active control strategy was simulated. It is shown that the control strategyenhances the structural damping of the machine, providing a stabilizing effect on chatter.

11:20 Adaptive vibrations control of a turbine with magnetic bearings (ID 243)Zbigniew Ogonowski, K. Plaza, Silesian University of Technology, Poland

The paper presents two-layered control system for active vibration attenuation of a turbine with magneticbearings. The lower layer consists of two two-input and two-output directly adapted weighted minimumvari-ance controllers. The upper layer optimizes the performance of the system by adjusting weighting factorsof the lower-layer controllers. Adjustment is done by the parameter-path optimization (gain-scheduling) orby on-line optimization. Non-linear model of the shaft’s ends mean displacement versus weighting factorsis identified in the upper layer. The overall control system is designed to minimize shaft vibration especiallyclose to resonant modes. This allows safe acceleration of the turbine up to required angular speed withpossible minimization of the energy consumption according to magnetic bearings.

11:45 Piezoceramics-based Devices for Active Balancing of Flexible Shafts (ID 451)Peter Sloetjes, Andre de Boer, University of Twente, The Netherlands

This paper focuses on vibration control of flexible shafts by means of rotor-fixed piezoelectric materials.The target is to realize compact solutions for the suppression of problematic resonant vibration at so-called

115


flexural critical speeds. For analysis, parametric finite element models of flexible rotors with piezoceramicsheets and strain or displacement sensors are developed, where the number of degrees of freedom is kept low.Several mechanisms which can destabilize flexible rotors are quantisized, such as rotor material damping,dissipation of currents induced in rotor-fixed piezoceramics and active feedback control proportional to rotorstrain rates. The effectiveness of low frequency feedback and feedforward control for the suppression of theunbalance response is demonstrated using analytic and experimental results. Emphasis is on the interactionbetween the dynamics of the rotor and that of the connected electronic circuits. The experimental setupwhich is used for validation is a flexible shaft equipped with piezoceramic sheets and strain sensors. Aslipring assembly is used to simplify measurements with, and control of, the sensors and actuators on theshaft and to facilitate the development of compact drive electronics.

12:10 FE analysis of a PZT-actuated adaptive beam with vibration damping using an elec-tric multiple-mode shunt system (ID 104)Cu-Hai Nguyen, Stanislaw Pietrzko, Empa - Materials Science & Technology, Switzerland

This work aims to simulate via FEM an aluminium cantilever beam actuated by a PZT patch, whose flexuralvibration is transmitted to a collocated PZT sensor, which is connected to an electric multiplemode shuntsystem damping the vibration.

First, starting from a single-mode parallel R-L shunt circuit, optimal tuned values of the resistor and inductorfor the shunt circuit are calculated at a targeted harmonic resonance vibration frequency. Extensive numericalresults are then obtained in using the FEM code ANSYS Multiphysics, which has piezoelectric couplingcapabilities, and also electric finite elements used to simulate explicitly the R-L circuit. Calculated dampinglevels of the beam vibration amplitude, due to the shunt circuit, attain high values. Furthermore, the shunteffect consists in an increase of the sensor resulting voltage.

Calculations are then extended to a multiple-mode shunt system, where each branch consists of a parallelR-L shunt circuit tuned to each of three eigenfrequencies, in series with two C-L circuits “blocking” thetwo other resonance frequencies. In modifying the shunt system, the number of antiresonant circuits can bereduced. Here also, high FEM calculated values for the beam vibration damping are obtained, now at thethree eigenfrequencies simultaneously.

Transfer path analysis and source identification – TPA2Room 6 – Chairman: P. Van der Linden

8:50 Blind source separation and system identification for convolutive mixtures with wellseparated modes (ID 341)Robert Randall, Liam Holley, The University of New South Wales, Australia

In recent years a number of techniques have been developed for Blind Source Separation (BSS) and BlindSystem Identification (BSI) where there are a number of inputs to a system, and only response measurementsare available. In most mechanical applications, the relationship between sources and responses is by “convo-lutive mixing”. This paper introduces a new solution where the system under study has well separated modes(for example the low order modes of a railcar) which can be separated by bandpass filtration. The common(SDOF) impulse response function (IRF) can be deconvolved from the bandpassed mixture in each band,giving a simple (multiplicative) mixture, and it is then possible to use Independent Components Analysis(ICA) to separate the mixture into the contributions from the independent sources. This paper demonstratesthe procedure based on analytically simulated mixtures, and also on a simple free-free beam excited by twoshakers with different force signals, but in the latter case cross-talk via structural coupling impedes completeseparation.

116


9:15 An Analysis of Friction Noise (ID 431)Stephen Smyth, Henry Rice, Trinity College of Dublin, Ireland

The noise generated by friction is measured and analysed using Nearfield Acoustic Holography (NAH).An experimental rig is used to generate sliding contact between a smooth disk and a small solid block.The material, normal force applied and the sliding velocity of the block can be varied. Rubbing noise inparticular is investigated in this paper. In addition to the sound field measurements the surface vibrationsof the block are measured using a Laser Doppler Vibrometer (LDV). Signal conditioning techniques areapplied to investigate the correlation between the sound field and the surface vibrations. A numerical modelof the system is also presented where a cantilevered beam is excited at its free end by a frictional contact. Adynamic friction model known as the bristle model is used to excite the beam. The beam is modelled usingfinite elements and its response is related to the sound field by a point source acoustic model. Experimentalresults show that the sound source for rubbing noise is at the leading edge of the contact while numericalresults are promising showing that both the stick-slip and rubbing cases can be modelled.

9:40 Reconstruction of loading forces from responses data measured in-flight (ID 531)Tadeusz Uhl, Krzysztof Mendrok, AGH - University of Science and Technology, Poland

The aim of the authors was to identify the force acting on the airplane wing. The source of this force is theadjustable spot light mounted on the bottom surface of the wing. Authors decided to utilize a modal modelfor the force reconstruction purposes. A well known formula, which relates input forces vector with outputsignal vector via pseudo-inverted frequency response functions (FRFs) matrix was applied. The FRFs weresynthesized from the modal model. This technique requires the modal model with scaled modal vectors.Such modal vectors can be obtained from experiment with measured excitation, however the authors wereinterested in the in-flight forces in described region. That is why the operational test was performed. For theoperational model scaling, the scaling factors from the ground vibration test model were used.

Model updating and correlation – MU1Room 6 – Chairman: M. Link

10:30 Model Updating using Bayesian Estimation (ID 25)Cristinel Mares, Brunel University, United KingdomBruno Dratz, Ecole Centrale de Lille, FranceJohn Mottershead, University of Liverpool, United KingdomMichael Friswell, University of Bristol, United Kingdom

Variability in real structures, which could arise from manufacturing processes, and the modelling assump-tions and limitations require the creation of a statistical model of the relationship between experimental andmodel predictions and the quantification of the uncertainty of this estimate. In this paper Markov-ChainMonte Carlo theory (MCMC) is discussed and applied to model updating in the case of multiple sets ofexperimental results by using frequency responses functions. The MCMC method allows the solution ofcomplex problems in a unifying framework, by integrating over high dimensional probability distributions inorder to make inferences about the model parameters. A simulated three degree-of-freedom system is usedto illustrate some aspects of the method, allowing for practical assumptions to be tested on a simple examplewithin the WINBUGS environment (Bayesian inference Using Gibbs Sampling).

117


10:55 The QCM approach for model updating from noisy data (ID 53)Pablo Tarazaga, Virginia Polytechnic Institute and State University, United States of AmericaYoram Halevi, Technion - Israel Institute of Technology, IsraelDaniel Inman, Virginia Polytechnic Institute and State University, United States of America

The paper investigates the Quadratic Compression Method (QCM) method for model updating and its noisefiltering capabilities. The method uses a parametric setup with pre-specified principal sub-matrices mul-tiplied by unknown scalar parameters. The optimal parameters are obtained by minimizing the error in asquared down version of the eigenvalue equation, thus with reduced computation. QCM is shown to belongto the class of Minimization of the Error in the Characteristic Equation (MECE), with a particular choiceof the weighting matrix. Analysis of the propagation of the noise into the identified parameters reveals thatQCM has desirable noise filtering properties. The paper presents also a weighted version of the method,called WQCM, which is motivated by further reducing the effect of measurement noise. In addition to thetheoretical analysis, the superior robustness to noise properties of QCM and WQCM are demonstrated bysimulations and experimentally.

11:20 Using Complex Modes for Model Updating of Structures with Non-ProportionalDamping (ID 562)Michael Link, University of Kassel, Germany

The inverse sensitivity approach is most often used in computational model updating to adjust selected stiff-ness and mass parameters of large order finite element models by minimizing the test/analysis differencesof natural frequencies and real modes. However, in the case of non-proportionally damped structures theexperimental modal analysis yields complex eigenvalues and modes. One way to use these complex modaldata for model updating is to extract the underlying real modes from the complex modes by approximatemathematical procedures, which, however, are prone to unavoidable approximation errors. The straightfor-ward way would be the direct utilization of the complex modal data in the updating process. In this casethe analytical model to be updated would have to include a damping matrix which, however, is usually notknown in physical finite element coordinates.

In a previous paper a method was presented allowing to include an analytical non-proportional damping ma-trix in the computational updating procedure. The method utilizes the factors of local substructure dampingmatrices as additional updating parameters (so-called Rayleigh factors) which results in a physical non-proportional damping matrix. The definition of the substructures is user’s choice and could, for example,be related to a local discrete damper or to a larger substructure with uniform damping. In contrast to theclassical procedure the Rayleigh factors are not used in the objective function to be minimized but are fitteddirectly to the experimental modal damping ratios within each iteration step. The uncertain updating para-meters are only related related to stiffness and mass properties. They are identified as usual by minimizingan objective function containing the test/analysis differences of the complex eigenvalues and modes. In thepresent paper the theory behind this approach is summarized and additional experiences with the method arereported from an application to a laboratory test structure with a local damping device designed to generatenon-proportional damping behaviour. The prediction quality of the updated model was checked by com-paring not only the analytical and experimental modal data used in the objective function for computationalparameter updating but also by comparing the analytical and experimental frequency response functionswhich were not used during the numerical updating process.

11:45 Updating of the Myriade Platform FE Model using Microvibration Test Data (ID103)Nicolas Roy, Top Modal, FranceStephane Mary, CNES - Centre Spatial de Toulouse, France

Microvibration tests have been performed by the CNES on the Myriade microsatellite platform using anon-board reaction wheel as the disturbance source. These tests had the double objective of characterizing

118


the dynamic response levels and updating the finite element model. This paper presents the updating studycarried out by TOP MODAL, with a special focus on the methods and procedures that were adapted to thecontext of microvibration measurements.

12:10 Model Updating of Ship Decks loaded with Welding Stresses (ID 212)Andreas Hanke, Gunther Schlottmann, University of Rostock, Germany

The vibration behaviour of local ship structures like ship decks is highly influenced by residual stresses.Measurements have shown that differences of 50 % in natural frequencies due to welding stresses are possi-ble. A detailed FE-modelling of the initial welding stresses is feasible, but leads to high computational costsand is time consuming. Therefore, a simplified model for the influence of welding stresses on the modalparameters is aimed.

In this paper, it is shown that a good agreement of measurement and computation can be obtained by applyingan updating model using eigenstrains. Due to the modelling based on eigenstrains, less updating parametersare necessary. For welded structures a high advantage over methods using “usual” updating parameters(density, stiffness ...) was achieved. Furthermore, the model updating leads to a model that reflects thephysical properties of the reality much better and the gained parameters can be used further for similarstructures.

Vehicle noise and vibration (NVH) – NVH4Room 7 – Chairman: H. Van der Auweraer

8:50 Expanding the bandwidth of MDOF Road Reproductions (ID 327)Filip De Coninck, Wim Desmet, Paul Sas, Katholieke Universiteit Leuven, Belgium

This paper describes the use of a high-frequency 6-DOF shaker table for the reproduction of multiple degreeof freedom (MDOF) road input measurements in the frequency range up to 300 Hz. Results are discussed for3-DOF tire patch excitation and 3-DOF spindle excitation. It is shown that the typical bandwidth from testson a fourposter or axle testrig, can be expanded into this range by use of the shaker table if, depending on thetarget spectrum, measures are taken to ensure sufficient accuracy trough out the frequency range. Especiallythe typical low energy band (165-220 Hz) in the tire response spectrum requires further work.

9:15 Measurement of dynamic loads at wheel center for road excitations (NVH applica-tion) (ID 367)Pascal Bouvet, C. Braguy, VIBRATEC, FranceLaurent Gagliardini, PSA Peugeot Citroen, France

This article presents an experimental approach to estimate the dynamic loads (spindle loads)at wheel centreof an automotive vehicle in operational conditions. In a first step, these dynamic loads (including forces andmoments) are estimated using an inverse method. In a second step, the contribution of these loads to internalnoise is assessed by combining them with vibro-acoustic transfer functions.

The methodology relies on operational measurements (accelerations measured on the spindle) and transferfunctions measurements (matrices relating the forces and moments at the wheel centre to the accelerationson the spindle, and to the pressure levels inside the vehicle). The measurements are carried on a Citroen C3vehicle, for different rolling conditions (tyre-wheel couple, speed, road roughness).

Three forces and two moments are obtained in the frequency range 50-500 Hz, simultaneously on two wheels.Two techniques are assessed for the force estimation : a preliminary decomposition in principal componentsimproves considerably the results quality, by suppressing most of the measurement noise contained in theoperational responses. One obtains a good reconstruction of noise levels inside the vehicle, with some

119


punctual overestimations. These calculations emphasize the strong contribution of the moment excitationsto the internal noise levels..

9:40 A new approach to model tyre/road contact. (ID 526)Ysbrand Wijnant, Andre de Boer, University of Twente, The Netherlands

In the Structural Dynamics and Acoustics group at the University of Twente, we aim to develop a quantitativetyre/road noise model. An essential part of this model is an accurate contact algorithm which is fast enoughto simulate tyre vibrations up to the acoustic frequencies. In this paper we present a contact algorithm,describing the contact between a tyre and a road surface, which has the potential to be made very fast usingthe multigrid techniques developed in the field of elasto-hydrodynamic lubrication. For the development ofthe algorithm a flexible ring model is used to describe the tyre. The friction model is based on Coulomb’sfriction. We present (quasi-)static results obtained from the algorithm for various friction coefficients, aswell as frictionless results for a rotating tyre. The vibrations of the tyre obtained by this model have beenused to calculate the radiated sound field by means of a boundary element program (BEMSYS).

10:05 Two-way Coupled Structural Acoustic Optimization via Model Order Reduction(MOR). (ID 611)Srinivasan Puri, Denise Morrey, Andrew Bell, John Durodola, Oxford Brookes University, UnitedKingdomEvgenii Rudnyi, Jan Korvink, University of Freiburg, Germany

A reduced order model is developed for low frequency, fully coupled, undamped and constantly dampedstructural acoustic analysis of interior cavities, backed by flexible structural systems. The reduced ordermodel is obtained by applying a Galerkin projection of the coupled system matrices, from a higher dimen-sional subspace to a lower dimensional subspace, whilst preserving some essential properties of the coupledsystem. The basis vectors for projection are computed efficiently using the Arnoldi algorithm, which gen-erates an orthogonal basis for the Krylov subspace containing moments of the original system. A computa-tional test case is analyzed, and the computational gains and the accuracy compared with the direct methodin ANSYS. Further, the reduced order modelling technique is applied to a two-way coupled vibro-acousticoptimization problem, with stacking sequences of the composite structure as design variables. The optimiza-tion is performed via a hybrid search strategy combining outputs from Latin Hypercube Sampling (LHS) andMesh Adaptive Direct Search (MADS) algorithm. It is shown that reduced order modelling technique resultsin a very significant reduction in simulation time, while maintaining the desired accuracy of the optimizationvariables under investigation.

Vibro-acoustic modelling and prediction – VAM2Room 7 – Chairman: Ph. Bouillard

10:55 Acoustic Theoretical x Experimental Comparison of the Brazilian SatelliteLauncher Vehicle (VLS) Fairing (ID 91)Rogerio Pirk, Luiz Carlos Sandoval Goes, Aerospace Technological General Command, Brazil

Fluid-structure interactions are always present in real life dynamic systems, during operations. However,analysts, due to the complexity of building a coupled vibro-acoustic model and also because sometimes thecoupling effect has no significance, often apply “one way” or uncoupled analysis. Nevertheless, sometimes itis important to consider the mutual influence of the vibro-acoustic system, where the acoustic and structuralmatrices are coupled and the influence of the structural displacement on the fluid domain, as well as theacoustic pressure of the fluid on the structural body, are accounted in one coupled matrix.

During take off, launchers and its payloads are submitted to severe acoustic loads, generated by the reflected

120


noise, during the boosters’ combustion. Typical values of the OASPL (overall acoustic sound pressure level),achieved next the engines, lie within 150-180 dB, which are estimated at the upper parts of launchers within140-160 dB. Such strong sound pressure excitation levels, which have random and large spectral distribution,require that the coupled vibro-acoustic behaviors of space systems be studied.

In this framework, low frequency or deterministic coupling techniques were used to calculate the Brazil-ian Satellite Launcher Vehicle (VLS) fairing behavior. The fairing is the structural compartment where thepayload or satellite is, during the launcher mission. This structure has as function to give adequate aerody-namic shape to the launcher as well as protecting the payload. The structural FEM/fluid FEM and structuralFEM/fluid BEM techniques were applied to model the fairing body and its acoustic cavity. The acousticexcitation was applied and coupled calculations were done, yielding the low frequency acoustic and skinresponses.

For the high frequency prediction analysis, it was applied the Statistical Energy Analysis (SEA) technique.The equivalent air-borne excitation was applied and the mean value responses of the acoustic and structuralsubsystems, using the referred fairing SEA model were calculated.

A very important test campaign was done to measure the acoustic noise inside the fairing cavity, intendingto compare experimental x calculated results. In this test, the fairing structure was submitted to 145 dBOASPL in a 1,200 m3 acoustic reverberant chamber and microphones were positioned inside its cavity. Themeasured acoustic pressure levels are compared with the calculated acoustic responses.

This work describes the deterministic and statistical modeling procedures applied for the prediction of theacoustic environment of the VLS fairing as well as its acoustic test campaign, performed in the referredacoustic reverberant chamber. Acoustic Theoretical x Experimental comparison is done and some conclu-sions and future works are described.

11:20 A boundary integral method for noise shielding analysis in non-conventional aircraftconfigurations (ID 407)Alessandro Randazzo, Trinity College of Dublin, IrelandU. Iemma, University of ”Roma Tre”, ItalyHenry Rice, Trinity College of Dublin, Ireland

This paper deals with the development of a boundary integral method for the analysis of the shielding effectsinduced by the aircraft surfaces on the acoustic field generated by the engines. The methodology is based onthe Kirchhoff-Helmholtz integral formula for the velocity potential function, written in the Laplace domain,and numerically solved by means of a Boundary Element Method (BEM). In order to include the wakeeffect on the sound propagation, the scattering bodies are considered to be surrounded by a quasi-potentialnon-uniform flow, i.e., a flow where the vorticity is confined within a domain of zero measure (the wake).The integral formulation is derived from the non-homogeneous Helmholtz equation for the velocity potentialand associated to homogeneous boundary conditions. Starting with the assumption that the velocity potentialconsists of a mean and a fluctuating part and making use of the Bernoulli equation, the pressure acoustic fieldis computed. The pressure field obtained includes the direct contribution of the sources emission (so-calledincident field) and scattering effect of the boundary surfaces. The method is applied for the aeroacousticshielding analysis of non conventional aeronautic configurations.

11:45 Viscothermal wave propagation in a circular layer with a partially open and par-tially closed boundary (ID 424)Ronald Kampinga, Ysbrand Wijnant, Andre de Boer, University of Twente, The Netherlands

The so called low reduced frequency model has been shown to be both an accurate and a relatively simpledescription of wave propagation in narrow tubes or layers, under small signal conditions. In this paper,the low reduced frequency model will be applied on a circular layer between a fixed surface and a rigidlytranslating plate. The outer circumference of the layer is partially closed (Neumann boundary condition) and

121


partially open (Dirichlet boundary condition). A semi-analytical solution for this problem is used to calculatethe volume flow that is generated by the squeezing motion of the plate, and the resulting force on the plate.The volume flow per unit force is evaluated for several boundary conditions.

12:10 Vibro-acoustic analysis of fluid-loaded structures using component mode synthesis(ID 163)Xianhui Li, Dalian University of Technology, China

The work focuses on efficient vibro-acoustic analysis of fluid-loaded structures in the low to medium fre-quency range. A coupled finite element and boundary element formulation is used to describe the interactionbetween the structures and the fluid. Three component modal sets are proposed, including in-vacuo structuralmodes describing the free vibration of the structures, interface modes capturing the prominent dynamics ofthe fluid-structure interfaces, and adjoint modes representing the acoustic loadings from the fluid. Com-ponent mode synthesis is applied to the coupled systems to accelerate the prediction of the vibro-acousticresponses. The performance of the method is illustrated on a water-loaded stiffened plate.

122

room 1 room 2 room 3 room 4 room 5 room 6 room 7 room 8 room 9 room 10FCOP1 DMOD1 ANC1 OMA1 FLI1 MT1 AVM1 MAD1 POS1A POS1B

NL1 OPT1 MHF1 PE1 FLI2 DMI1 AVC1 UNC1

room 1 room 2 room 3 room 4 room 5 room 6 room 7 room 8 room 9 room 10AMS1 NSV1 ANC2 RMD1 SEV1 VAA1 NVH1 MAD2 POS2A POS2BAMS1 NSV2 AVC2 RMD2 TPA1 D1 NVH2 NL2

AMS2 IRC1 MHF2 OMAX1 GRV1 MB1 NVH3 UNC2 POS3A POS3BAMS3 D2 MHF3 OMAX2 MAT1 VAM1 AA1 UNC3

room 1 room 2 room 3 room 4 room 5 room 6 room 7 room 8 room 9 room 10HUM1 D3 EDSVS1 RMD3 AA2 TPA2 NVH4

I1 D4 AA3 AVC3 MU1 VAM2

DCS1 PNVC1 SC1 RMM1

AA Aeroacoustics and flow noiseAMS1 AMS Special seminar - Multibody dynamics for multi-physics applicationsAMS2 AMS Special seminar - Control and optimisationAMS3 AMS Special seminar - ApplicationsANC Active noise controlAVC Active vibration control and smart structuresAVM Application of (new) vibroacoustic methods to complex systemsD DampingDCS Dynamic behavior of civil structuresDMI Damage monitoring and identificationDMOD Dynamic modellingEDSVS EDSVSFCOP FRIENDCOPTER Rotorcraft noise and vibrationFLI Flite Eureka 2GRV Ground vibrationsHUM Human vibrationsI InstrumentationIRC Special Session on Technology Partnering: IRC Technology SeminarMAD MaduseMAT Material characterisation based on dynamic measurementsMB Multi-body dynamics and controlMHF Medium and high frequency techniquesMT Modal testing and FRF estimationMU Model updating and correlationNL Non-linearities: identification and modellingNSV Non-stationary sound and vibration analysis and applicationsNVH Vehicle noise and vibration (NVH)OMA Operational modal analysisOMAX Operational Modal Analysis in presence of exogeneous inputsOPT Optical measurement techniques OPTIMESSPE Parameter estimationPNVC Passive control of noise and vibrationRMD Rotating machinery:dynamicsRMM Rotating machinery: monitoring and diagnosticsSC Substructuring and couplingSEV Self-excited vibrationsTPA Transfer path analysis and source identificationUNC Uncertainties in structural dynamics and acousticsVAA Vibro-acoustic analysisVAM Vibro-acoustic modelling and prediction

Date post:	02-May-2020
Category:	Documents
Upload:	others
View:	1 times
Download:	0 times

2006ISMApast.Isma-isaac.be/downloads/isma2006/ISMA2006_abstract_book.pdf10:05 A Future for...

Documents