Annual Report Department

2Research activities

2.1Structural design and analysis

2.1.1A fuzzy finite element procedure for modelling uncertainties in structural dynamic behaviour

A fuzzy finite element approach for the dynamic analysis of imprecisely defined structures is being developed. The approach is applicable to systems that are partially described in linguistic terms or by incomplete information. It handles certain types of imprecisely known data more realistically compared with the existing stochastic procedures. Furthermore, the fuzzy method intends to substantially reduce the computational effort compared to these stochastic procedures.

The development of the methodology starts with the basic concept of fuzzy numbers and fuzzy arithmetic. It implements fuzzy calculus concepts for the derivation, manipulation and solution of the finite element equations. However, the basic finite element scheme remains unchanged.

Different strategies are developed for the solution of the fuzzy finite element problem. In the interval arithmetic approach, fuzzy procedures based on interval arithmetic are used to calculate fuzzy system matrices. The basic equations for the solution of the fuzzy finite element analysis are assembled using these matrices. An alternative approach based on global optimisation considers the deterministic finite element problem as part of a black-box goal function. Based on these concepts, a hybrid approach has been developed for the calculation of fuzzy frequency response functions. It proves to be an efficient procedure yielding a close approximation of the exact fuzzy solution.

In the current research, the hybrid algorithm is being refined in order to enable the inclusion of a damping concept that enables independent specification of uncertainty on the damping properties of the structure. Next to that, special attention goes to the numerical implementation of the developed algorithms. The fuzzy modelling strategies are being automated in view of their use for larger industrial models by linking the code to the finite element solver Nastran. The applicability of the method is being tested through different test cases.

Current research focuses also on the use of component mode synthesis (CMS) techniques within the fuzzy finite element framework. In this approach, a large amount of degrees of freedom are expressed in terms of their active degrees of freedom. By limiting the actual calculation to these active degrees of freedom, the calculation time can be drastically reduced. The CMS method also allows substructures to be developed and optimised in a parallel way. The aim of the research is to develop a procedure that can handle the combination of deterministic and fuzzy substructures in a single analysis, yielding fuzzy results.

Related projects: IAP-P5/06 AMS, TAP31

Publications and reports: 2003PP077, 2003PP106, 2003PP132, 2003PP133

Scientific staff: D. Vandepitte, W. Desmet, D. Moens, H. De Gersem, M. De Munck, A. Stenti, L. Farkas, P. Neven

2.2Noise and vibration engineering

Measurement and analysis

2.2.1Vibration based material identification techniques

For the efficient development and production of layered materials, it is necessary to know the properties of the individual layers. However, a generally applicable methodology to measure these properties is not yet available. This is the aim of the GRAMATIC-project: to develop techniques for the identification of the stiffness and damping properties of the individual layers, by combining modal analysis experiments performed on laminates with mixed numerical-experimental identification techniques (MNET). The proposed MNET uses a finite element model to simulate the vibratory behavior of the specimen(s). The values of the unknown material parameters are updated, in an iterative way, in order to achieve an optimal correlation between output of the numerical model and the experiment. The MNET based approach for layered materials has been successfully validated on a purpose build brass-steel bimetal. The developed identification method has also been extended with an uncertainty analysis routine, which allows to estimate the uncertainty on the obtained material parameters from the uncertainty on the input parameters.

Related projects: IWT project no. ADV/000163/KUL-TW-MTM-FYS: GRAMATIC,

Publications and reports: 02PP088, 03PP061, 02PP089, 03PP081, 03PP060

Scientific staff: W. Heylen, T. Lauwagie

2.2.2Characterization of aerodynamic noise in automotive muffler applications

The objective of the research is to characterize flow-induced noise in subsonic confined flows and in particular in automotive muffler applications. For the current experimental research an aero-acoustic test rig is designed to generate low Mach number flows (0.1
Publications and reports: 2003PP147, 2003PP176

Related projects: IWT-WINDY

Scientific staff: W. Desmet, P.Sas, M. Baelmans, R.Boonen, W. De Roeck, W. Derkinderen

2.2.3Identification and characterization on nonlinear mechanical systems

Two common types of nonlinearities may be found in mechanical systems, namely: backlash and friction.

A simple but effective technique for the identification of backlash has been employed. The skeleton curve technique, which utilises Hilbert transformation and Wavelet transformation is shown to be an efficient way to identify the presence and size of the backlash component when the system is well behaved. However, once the system reveals non-periodic response (e.g. chaotic response), this technique is no longer applicable. In the latter case, chaos quantification methods, such as Lyapunov exponent and Fractal Dimension, can be used to estimate the degree of backlash in the mechanical system. Theoretical simulation studies using realistic system parameters as well as experimental studies on a mechanical system with backlash have been carried out in the framework of this investigation.

Friction behaviour of mechanical systems and components has been widely investigated in recent years. Conventional models, such as Coulomb's, are shown to be inadequate for modelling and identification, especially in the presliding regime (around velocity reversals). Some recent models of friction such as LuGre, Leuven and Generalize Maxwell-Slip (GMS) have been proposed to overcome this problem. Identification of friction, obtained from dedicated test set-ups, has been carried out utilizing several models, which range from Coulomb friction to the most recent GMS model. The results show that the latter model improves identification result, in particular in the presliding regime. An extension of this model is being also proposed, in order to capture position dependency of the friction.

Related projects: VW-Stiftung

Publications and reports: 2003RP001, 2003PP057

Scientific staff: F. Al-Bender, H. Van Brussel, T. Tjahjowidodo

2.2.4Linear and non-linear of road noise in passenger cars, including strategies for active control

Many mechanisms, like car suspensions, consist of multiple (rigid) links that are connected through relatively flexible elements. The dynamic behaviour of such a system is determined by the dynamic characteristics of the different components. For rigid links these characteristics are mostly linear, while the flexible links are characterised by more complex non-linear behaviour. This research systematically investigates the influence of these non-linearities on model accuracy, using numerical simulation and laboratory experiments under simulated operating conditions. Non-linear models, resulting from this work, are used for the development of active and passive methods for vibration reduction along the transmission paths. Passive methods include the optimisation of the dynamic characteristics of connection elements, addition (or removal) of local material mass, addition of tuned dampers. Active methods influence the vibration transmission using actuators placed along the vibration path. The research focuses on the accurate modelling of non-linear components under operating conditions for the use in global system models. An important application is the reduction of structure borne road noise in passenger cars, which is investigated as a case study. In 2003 the research has focused on accurate road reproduction experiments, the influence of preloads and amplitude effects.

Related projects: IAP P5/06-AMS

Publications and reports: 2002PP125, 2003PP085, 2003PP087, 2003PP166

Scientific staff: P. Sas, W. Desmet, F. De Coninck

Numerical modelling

2.2.5Validation of finite element and boundary element methods for vibro-acoustic modeling and analysis

A wide range of numerical techniques, based on finite element and boundary element formulations, are readily available in commercial software for the prediction of the global vibro-acoustic behaviour of complex systems. The methods range from simple purely acoustic models to fully coupled vibro-acoustic models, incorporating the interaction between the acoustic field and the vibrating structure. Due to the huge computational loads involved with finite element and boundary element models, their practical use is restricted to low-frequency applications. Research is targeted at further development and verification of these prediction tools, mainly to extend their use for a frequency range as wide as possible. Special emphasis is also put on appropriate modelling of acoustic insulation materials and on modelling of exterior acoustic radiation.

At present, the element based prediction techniques are evaluated for the analysis of several vibro-acoustic systems. A first system consists of the driver’s cabin of a harvester, for which a favorable (low-frequency) vibro-acoustic driver’s comfort has become a major design specification. A second validation project involves the modelling of the exterior acoustic radiation of electromagnetic systems with special focus on transformer noise.

Related projects: FWO-project G.0161.02, “IWT-TRICARMO”

Publications and reports: 2003PP030

Scientific staff: W. Desmet, P. Sas, W. Heylen, B. Pluymers, R. Masti, C. Vanmaele

2.2.6A wave based approach for vibro-acoustic modelling

Numerical simulation of vibro-acoustic systems is usually done by finite element or boundary element methods. Both deterministic techniques are based on an element discretization of the problem domain or its boundary surface. The dynamic variables within each element are expressed in terms of simple (polynomial) shape functions, which do not satisfy the governing dynamic equations. These element based methods are well suited for the dynamic analysis of arbitrarily shaped (vibro-acoustic) systems, but their use is practically restricted to low-frequency applications. At higher frequencies, structural and acoustic wavelengths become so small that a prohibitively large number of elements and computational effort would be required to get reasonable prediction accuracy. In order to extend the applicability of numerical prediction techniques towards vibro-acoustic analysis at higher frequencies, the PMA division has developed a wave based method (WBM).

The WBM is a deterministic technique, based on the indirect Trefftz approach. Instead of using locally defined element shape functions, the WBM applies globally defined wave functions, which do satisfy the governing dynamic equations. The vibro-acoustic response of the system at a certain frequency is expressed as a summation of wave function contributions, which result from an integral formulation of the problem boundary conditions. The WBM exhibits better convergence properties than the element methods resulting in smaller model sizes and computational efforts. However, the WBM is most efficient for systems of moderate geometrical complexity. The PMA division has proven the theoretical feasibility in previous studies.

The research activities in 2003 have focused on three items. A first item involves the extension of the wave modelling concept towards exterior acoustic radiation modelling in unbounded domains. In a second item, research focused on the dynamic modelling of three-dimensional mechanical structures consisting of arbitrary flat plate assemblies. A third item is the development of a hybrid method, which is based on a coupling between the WBM and finite element based methods. This hybrid approach aims at combining the benefits of both techniques, namely the high computational efficiency of the WBM and the geometrical flexibility of finite element methods.

Related projects: FWO-project G.0123.01, “EDSVS”

Publications and reports: 2003PP031, 2003PP032, 2003PP033, 2003PP034, 2003PP067, 2003PP069, 2003PP070, 2003PP072, 2003PP097, 2003PP098, 2003PP149, 2003PP151, 2003PP171

Scientific staff: W. Desmet, D. Vandepitte, P. Sas, B. van Hal, B. Pluymers, C. Vanmaele, R. Masti, P. Silar

2.2.7Numerical description of the sound propagation in subsonic confined flows

Convection and refraction effects have an important influence on the propagation of sound waves in a moving medium. These flow effects are not all taken into account in the classical or the convected wave equation. The Linearized Euler Equations (LEE), which are commonly used in the field of Computational Fluid Dynamics (CFD), can be used to describe the sound propagation in the presence of a non-uniform mean flow. Due to the specific character of acoustic waves, numerical schemes for solving the LEE should have, as compared to classical CFD-computations, a large accuracy, the dispersion and dissipation errors should be as low as possible and boundary conditions may not introduce spurious reflections back into the computational domain. Numerical solution schemes for the LEE equations are being developed using both a finite difference (2D) and discontinuous Galerkin (3D) formulation. These schemes can be driven with conventional acoustic source terms, with aero-acoustic analogy source terms and with (aero-)acoustic boundary conditions. In this way, the LEE can be used both to solve acoustic propagation applications in a moving medium and to solve aero-acoustic applications. Besides the LEE, the research also focuses on the validation of other formulations, similar to the LEE, to describe the sound propagation in subsonic confined flows. Alternative formulations include the Acoustic Perturbation Equations (APE) and the Expansion about Incompressible Flow (EIF) formulation.

Publications and reports: 2003PP147, 2003PP176

Scientific staff: W. Desmet, P.Sas, M. Baelmans, W. De Roeck, Y. Reymen

2.2.8Aeroacoustic analysis of subsonic flows in turbomachinery

The objective of the research is the numerical analysis of flow noise generation and propagation in turbomachines and their inlet and outlet ducting. The direct computation of noise is too expensive for industrial flows as encountered for instance in turbomachinery applications. To overcome this, a hybrid approach is adopted in which the calculation of flow and acoustic source information is separated from the calculation of the noise propagation. Different methods are viable for both types of calculations. For the flow and sources, the tonal noise can be computed using unsteady Reynolds Averaged Navier-Stokes equations, while broadband noise analysis requires Large Eddy Simulations. For the propagation, aero-acoustic analogies assume a uniform medium (at rest), whereas Linearized Euler Equations (LEE) allow a non-uniform flow and also include refraction on passive walls.

The main focus of the current research is the development of numerical tools for the simulation of the propagation of acoustic waves in the presence of subsonic confined flows. Efforts are made to implement a three dimensional LEE code based on the Discontinuous Galerkin Method (DGM). The code will allow to take into account the convection and refraction effects of a non-uniform mean flow in an arbritrary geometrical configuration. Input information regarding the flow and noise generating sources will be provided by state-of-the-use analytical, empirical and numerical methods for flow and acoustic source characterisation.

Scientific staff: W. Desmet, M. Baelmans, Y. Reymen, W. De Roeck, G. Rubio

2.2.9SEA modelling of car floor panels

Swages are introduced in a car floor for crash and stability reasons but they influence also the vibro-acoustic behaviour of the floor. When analysing the high-frequency vibro-acoustic behaviour of car floor panels through a Statistical Energy Analysis (SEA) model, the swage effects should be accounted for in such a way that the basic SEA assumptions are not violated. Introducing a coupling beam of V- or U-shape representing the swage is one possibility. For that case the transmission coefficients and coupling loss factors between the two plates coupled via a coupling beam are calculated with a general wave approach and the phenomenon of hypersensitivity is investigated. The results are compared to coupling loss factors obtained with common practice SEA routines. A second modelling approach involves the representation of a swage as a plate strip. Using plate strip theory the transmission coefficients are again calculated and compared to the results of the coupling beam. These comparisons allow to draw conclusions and derive guidelines for the modelling of swages in a SEA context.

Related projects: FWO-project G.0123.01, “EDSVS”

Scientific staff: W. Desmet, S. Nintzel

2.2.10Dynamic modelling of uncertainties in joints

Mechanical built-up structures include joints such as spot welds, bolted joints, gaskets, sealings etc., whose physical properties, like stiffness, thickness and damping, often vary substantially from one structure to another. Uncertainties in these physical properties lead to significant uncertainties in the dynamic behavior of the structure, and there is therefore an interest in predicting the statistics of the response given the statistics of the joint properties. Joints are commonly modelled using finite element models. Typically the level of refinement needed to investigate the joint effects on the dynamic behavior of the global structure is very high. This makes it difficult to include a complete description of the joint in a full finite element model.

This research is focused on the static and dynamic characterization of a car door weather sealing. Modelling such a type of joint is a highly non-linear problem. Non-linearity is due to the geometry (large displacement and large strains), the material behavior (rubber-like materials) and the contact problems (both contact and self contact).

The primary interest will be to develop an equivalent linear model more suitable to be included in a full finite element model. Then the interest will turn on the identification of the major sources of uncertainties typical of the problem and on how to include these uncertainties in the equivalent model. The goal will then be to investigate the extent to which the equivalent model can be used as a reliable alternative to predict the response statistics of the complete structure. The research will involve both numerical analysis as well as experimental validations.

Related projects: IAP P5/06- AMS

Scientific staff: W. Desmet, P. Sas, D. Moens, D. Vandepitte, A. Stenti

2.2.11Research on the dynamics of the drivetrain of a wind turbine.

Wind power is the most advanced and commercially available of renewable energy technologies. In recent years it has been the world’s fastest growing energy source. This expansion involves continually growing wind turbines. The classic concept of the drivetrain causes enormous loads for a power of 3MW. A totally new design is necessary and will change the dynamics of the whole system. Nevertheless the operator wants the power to be guaranteed. The project aims to research the structural integrity and behaviour of a new concept for the drivetrain. In a first period numerical models are built to predict the dynamic behaviour of the system. In a next stage these models will be validated on experimental set-ups. In the beginning the research will focus on the low-frequency domain (structural vibrations). A next task contains the research of the acoustic behaviour of the machine (high-frequency domain). Finally, methods and algorithms for condition monitoring of the wind turbine in operation will be examined and their practical feasibility will be validated.

Related projects: Hansen Transmissions Project

Publications and reports: 2002PP075, 2002PP106, 2003PP013, 2003PP014

Scientific staff: D. Vandepitte, J. Peeters

ACTIVE NOISE AND VIBRATION CONTROL

2.2.12Adaptive feedback anti-sway control for the load of a tower crane

Controlling the sway of the load of a tower crane increases safety, simplifies its operation, and increases the speed of load positioning, thereby saving time and money. The development of this anti-sway controller consists of several parts: modelling the dynamic crane behaviour, selection of an appropriate control structure, tuning of the control parameters and validation based on the developed crane model, implementation of the controller on a crane, fine tuning and experimental validation of the performance.

The development of a non-linear dynamic crane model was completed in 2002. Based on that model, an appropriate control structure was selected, tested and improved. The selected control structure consists of two independent adaptive controllers (one for the tower, and one for the trolley) consisting of model-based feedforward and 2dof sway angle measurement feedback. The control parameters are independent of the load, but vary with the length of the cable. Due to the limited availability of a crane, only the feedforward controller could be implemented successfully. The feedforward controller is able to reduce the sway angle by a factor of 3 or more, which is significant. The main advantages of this feedforward controller is its simplicity and robustness, and the fact it requires no sensors. Its main disadvantage is that it can only avoid load sway induced by the motion of the crane, and cannot react on disturbances, such as wind. This requires a feedback controller that will be further developed and implemented in 2004.

Related projects: IWT project 010463 “ARCOMET”

Scientific staff: J. Swevers, H. Van Brussel, K. Smolders

2.2.13Optimal decoupling for improved multivariable controller design

MIMO-identification and robust MIMO-controller design are cumbersome. As a result, MIMO-systems are often controlled by decentralized control systems, which consist of independent SISO-controllers based on the diagonal elements of the system. The neglected off-diagonal elements limit however the performance. This research investigates a design approach that combines decentralized control design with an input/output decoupling transformation yielding higher closed loop performance.

The first approach consists of a procedure to find a static transformations of the inputs and the outputs such that the relation between the transformed inputs and outputs is as diagonal as possible. Validation results on an automotive durability test rig simulation shows that the controller designed with decoupling yields better performance than a decentralized controller. In the second approach dynamic transformation filters are used (so-called inverse-based control). This approach is tested and validated on an industrial vibration test rig. Comparison with other MIMO-controllers showed that the inverse-based controller is superior with respect to design-complexity and obtained performance.

Publications and reports: 2003PP007, 2003PP004, 2003PP079

Related projects: IAP P5/06-AMS

Scientific staff: J. Swevers, P. Sas, D. Vaes

2.2.14Robust control of an active suspension of a quarter car test-rig

Comfort and road handling performance of a passenger car are mainly determined by the damping characteristic of the shock absorbers. Passive shock absorbers have a fixed damping characteristic determined by their design. Semi-active and active suspension systems offer the possibility to vary the damper characteristics along with the road profile e.g. by changing the restriction of one or two current controlled valves. An active shock absorber has the additional advantages that negative damping can be provided and that a larger range of forces can be generated at low velocities.

Linear car and suspension models were derived from an experimental quarter car test-rig using frequency domain identification techniques. The modelling errors caused by noise and nonlinearities are taken into account as model uncertainty. Based on these linear models, a robustly performant controller was synthesized using mu-synthesis. The controller was successfully validated on the quarter car test-rig, yielding an improvement of the passenger’s comfort, measured as the acceleration of the car body, of 50%.

Further research involves the extension of the developed techniques to a full car equipped with semi-active shock absorbers.

Related projects: IWT project AUT/000525 “ACOCAR”, IWT project 030391 “ACSAS”, IAP P5/06-AMS

Publications and reports: 2003PP111, 2003PP112, 2003PP110, 2003PP055

Scientific staff: J. Swevers, P. Sas, Ch. Lauwerys

2.2.15Design of a lightweight, electrodynamic, inertial actuator with integrated velocity sensor for active vibration control of a thin lightly-damped panel

Force actuators are advantageous for collocated, decentralised active structural acoustical control (ASAC).

In practice force actuators should be lightweight and usually react against ground or a reaction mass.

In order to obtain a small reaction mass large strokes are required for a sufficient transmitted force. Piezoelectric actuators can supply a large actuation force, but they just have a small stroke. Electrodynamic actuators can be designed with a sufficient stroke and large actuation force.

This research in the framework of the European Doctorate for Sound and Vibration Studies (EDSVS) network focuses on the design of a lightweight inertial electrodynamic actuator for ASAC. An innovative feature is the integration of an internal velocity sensor that allows adding internal damping to the actuator using collocated control.

A detailed analysis of design concepts has been carried out that allowed generating a design fulfilling the necessary requirements. An electromagnetic finite element analysis (FEA) has given actuation forces as function of the actuator weight for an optimised magnetic circuit design. A dynamic FEA has allowed shifting the higher order modes of the actuator structure outside the frequency band of interest. Finally the resulting actuator is being manufactured and its specifications will be determined. Then a set of these actuators will be used to damp vibrations of a lightweight aircraft panel.

Related projects: FWO-project G.0123.01, “EDSVS”

Scientific staff: P. Sas, R. Boonen, Ch. Paulitsch

2.2.16Application of active noise and vibration control methods to machining systems

This research forms part of a project that is geared towards the development of advanced machining systems for environmentally friendly manufacturing. The goal is to reduce the noise emission levels generated by the machining process. Whilst passive means are suitable for attenuating the high frequency content, active control methods are more effective at reducing the low frequency content.

Active control methods include Active Noise Control (ANC), whereby the disturbance noise is cancelled by an equal and opposite sound field generated by a control source. The latter needs to exhibit a similar sound field distribution in space and therefore a ring-shaped loudspeaker that fits around the tool holder of a high speed milling machine has been developed. In order to reduce the structure-borne noise, another active control method is employed, namely Active Structural Acoustic Control (ASAC). One of the key elements in the latter investigation concerns the optimal placement and configuration of the control actuators used on the structure. In a final stage, both active control methods can be incorporated into a hybrid solution to target the required noise reduction levels.

Collaboration: EC-Growth Project GRD1-2000-25828, ECOSYSTEMS: Advanced Machining Systems for Environmentally Friendly Manufacturing

Application area: high speed milling machines, machining processes.

Related projects: Growth project GRD1-2000-25828 “ECOSYSTEMS”, Growth Project, GRD1-2001-40674 “Noiseless”

Scientific staff: P. Sas, R. Boonen, C. Micallef, G. Pinte,

2.2.17Reduction of transient noise emission on machine tools

Noise pollution, caused by industrial activities, is an increasing environmental problem. Especially in machine halls with working machines such as punching machines, presses and, generating impact noise, the radiated noise level is too high to meet the regulations for noise emission. The objective of this project is to develop silent machine tools and taking measures over the machine structure, tools and components of punching machines and metal working presses. To attain this objective, innovative concepts in acoustic structural control (ASAC) and active noise control (ANC) will be used to develop new devices combining passive elements with active components.

The program started with the identification of the noise sources and transfer paths on the machines. This was carried out by means of measurements and FEM simulations. Once the main noise sources were identified, the development of noise attenuation devices started. New algorithms, specific for transient noise, were developed to control the devices. The optimal spatial configuration of the chosen sensors and actuators, which has great influence on the efficiency of the control system, is determined. This work is carried out on a small scale representative demonstrator. Based on the obtained results, the devices will be redimensioned for the large scale machines and evaluated.

Related projects: Growth Project, GRD1-2001-40674 “Noiseless”

Scientific staff: W. Desmet, P. Sas, G. Pinte, R. Boonen,