Proceedings of the DETC'99, 1999 ASME Design Engineering Technical Conferences, Las Vegas, Nevada, USA, 12.-15. September 1999

This paper deals with the dynamical behavior of washing machines. Due to large unbalance forces of the laundry during the spinning process a mechanical model has been established from d'Alambert's principle that considers both the rigid body motion and the elastic vibrations. According to the complexity of the machine parts the results of a Finite Element mode shape analysis are used to describe the elastic behavior. The distinct bodies of the system are linked together through linear and nonlinear force elements resulting in the excitation of several machine orders. In order to verify the model, simulation results are compared with measurements.

(PDF-Datei 3870 KB)

Proceedings: MSC Anwenderkonferenz, Weimar, Germany, 21.-22. June 1999

Für die Validierung von FE-Modellen können Versuchsdaten, wie z. B. experimentelle Modaldaten, verwendet werden. Falls die Abweichungen zwischen Versuch und Analyse nicht akzeptabel sind, muss die Idealisierung des realen elastomechanischen Systems überprüft und gegebenenfalls angepasst werden. Ist die Modellstruktur hinsichtlich des Diskretisierungsgrades, der Art der verwendeten Elemente etc. in Ordnung, so kann eine Minimierung der Versuchs-/Analyseabweichungen durch Modifikation geeigneter physikalischer Parameter (Schalendicken, E-Module, Dichten etc.) erreicht werden. Werden nur einzelne Parameter betrachtet, so kann die Anpassung noch mittels Ingenieurverstand und -erfahrung erfolgen. Bei realen elastomechanischen Systemen stösst diese Vorgehensweise jedoch schnell auf ihre Grenzen, da für diese eine Vielzahl von Parametern zu berücksichtigen sind. Hier müssen Verfahren zur computerunterstützten Modellanpassung verwendet werden, die eine simultane Anpassung mehrerer Parameter erlauben.

In dieser Veröffentlichung wird die Modellvalidierung mittels eines speziellen Programmpaketes zur computerunterstützten Modellanpassung vorgestellt. Das Programmpaket erlaubt die direkte Anpassung von MSC/Nastran Modellen und nutzt zu einem grossen Teil die Analysekapazitäten von MSC/Nastran, insbesondere den Modul zur Berechnung von Eigenfrequenz- und Eigenformsensitivitäten. Hierdurch werden FE-Modelle industrieller Grössenordnung handhabbar. Die grundlegende Theorie des verwendeten Verfahrens wird zunächst vorgestellt. Danach wird das prinzipielle Vorgehen erläutert. Abschliessend wird die Effektivität der Methode an Hand eines Kfz-Getriebegehäuses demonstriert.

(PDF-Datei 632 KB)

Proceedings: XXV. FEM-Kongreß, Baden-Baden, Germany, 1998

In dieser Veröffentlichung wird die Anwendung eines Verfahrens zur computerunterstützten Modellanpassung (engl. model updating) exemplarisch an Hand des Finite Elemente Modells einer Flugzeugtriebwerkskomponente der Firma BMW Rolls-Royce AeroEngines vorgestellt. Verwendung findet hierbei ein spezielles MATLAB-Programmpaket, das am Fachgebiet für Leichtbau und Strukturmechanik der Universität in Kassel entstanden und gemeinsam mit ICS weiterentwickelt worden ist. Dieses Programmpaket nutzt zu einem grossen Teil die Analysekapazitäten von MSC/Nastran, insbesondere den Sensitivitätsmodul, wodurch die Handhabung von Modellen industrieller Grössenordnung ermöglicht wird. Die computerunterstützte Modellanpassung ist Teil eines Validierungskonzeptes für Finite Elemente Modelle, das ebenfalls erläutert wird. Als erstes werden die einzelnen Schritte des Konzeptes zur Modellvalidierung vorgestellt. Im Anschluss daran folgen die Ergebnisse für eine repräsentative Flugzeugtriebwerkskomponente, um die Effektivität der verwendeten Methode zu demonstrieren.

(PDF-Datei 388 KB)

Proceedings of the 23rd Noise and Vibration Engineering Conference, ISMA, Leuven, Belgium, 1998

In this paper the application of a computational model updating procedure to aeroengine components is presented. A special MATLAB based software developed at the University of Kassel, Lightweight Structures and Structural Mechanics Laboratory, was applied. This software utilizes as much as possible of the analysis capacities of the FE-code MSC/NASTRAN, in particular the sensitivity module which allows to handle large order FE-models. Main goal was to improve the confidence in the predictions coming from the complex ‘Whole Engine Model (WEM)’. The concept applied is based on updating the FE-models of engine components using experimental modal data of the components which allows to restrict the number of uncertain parameters of the WEM. The different steps performed during the model development are presented by example of one representative component. Then the results for the other investigated components are summarized and the correlation of the WEM to experimental modal data is presented in order to verify the effectiveness of the utilized concept.

(PDF-Datei 195 KB)

PHD Thesis, VDI Fortschritt-Bericht, Reihe 11 (Schwingungstechnik) Nr. 247, VDI Verlag, Düsseldorf, Germany, 1997

In der Luft- und Raumfahrtindustrie werden häufig Schwingtischversuche mit Fusspunkterregung durchgeführt. Hierbei werden elastomechanische Strukturen hinsichtlich ihrer Betriebsfestigkeit oder Funktionsfähigkeit unter spezifizierten Fusspunktbeschleunigungen untersucht. Darüber hinaus besteht die Möglichkeit, die modalen Parameter Eigenfrequenz, Eigenform und modale Dämpfung der am Fusspunkt gebundenen Struktur aus den Versuchsdaten zu identifizieren. Die modalen Massen können nicht gewonnen werden, da die Einspannkräfte i. a. nicht bekannt sind. In dieser Arbeit wird nun untersucht, wie die Einspannkräfte, die mittels spezieller Kraftmesssysteme erfasst werden, zusätzlich für die Identifikation genutzt werden können. Es wird gezeigt, dass in diesem Fall neben den modalen Massen der am Fusspunkt gebundenen auch die modalen Parameter der ungebundenen Struktur sowie die Trägheitsparameter (Masse, Schwerpunktslage, Massenträgheitsmomente) identifizierbar sind, und es werden Identifikations-verfahren vorgestellt, die mit simulierten und realen Versuchsdaten erprobt werden.

(PDF-Datei 138 KB)

Proceedings of the CEAS International Forum on Aeroelasticity and Structural Dynamics, Vol. 3, Rome, Italy, 1997

In this paper the methods and results of estimating the physical and modal parameters of a liquid propellant tank from base excitation on a six axes shaking table are presented. The measurement of interface forces allows not only the identification of two sets of modal data: one for the fixed/free and the other for the free/free system, but also the identification of all rigid body properties. The high consistency of the two identified parameter sets and the good correlation with analysis results show the benefits of the additional measurement of interface forces.

(PDF-Datei 387 KB)

Proceedings of the 15th International Modal Analysis Conference, IMAC, Orlando, Florida, USA, 1997

This paper presents a procedure for identification of the full set of rigid body properties (overall mass, center of gravity location and moments of inertia) of a given elastomechanical system which is based on experimental modal analysis. The approach uses measured inertance frequency response functions (FRF) up to the first elastic natural frequency of the free/free system which are obtained either by testing the system suspended in soft springs or by testing the fixed/free system with an additional measurement of the interface forces. In a first step the underlying rigid body response is extracted from the FRFs by taking the influences of the elastic modes into account. A second step then uses these data to estimate the rigid body properties.

(PDF-Datei 257 KB)

Proceedings of the 1996 ESA Conference on Spacecraft Structures, Materials and Mechanical Testing, European Space Agency (ESA), Noordwijk, The Netherlands, 1996

Base excitation testing on a shaking table is frequently used in industry to qualify mechanical systems with respect to specified base acceleration levels. The measurement of structural and interface (base) accelerations furthermore allows an identification of the natural frequencies, mode shapes and modal damping values of the fixed system. However, modal masses, mass participation factors and effective masses cannot be identified.

Now, the additional measurement of the interface forces allows not only to identify these missing data but also the modal data of the free system and the rigid body properties.

This paper describes an approach to identify the rigid body properties (overall mass, moments of inertia and location of center of gravity) from base excitation test data since the other possibilities have already been published by the authors elsewhere. The requirements, capabilities and restrictions of the approach will be discussed in detail and an analytical example will be presented.

(PDF-Datei 106 KB)

Proceedings of the 14th International Modal Analysis Conference, IMAC, Dearborn, Michigan, USA, 1996

This paper introduces an automated approach to pick-up and exciter placement for modal testing purposes and its application to a car body component. It is a two step procedure which in the first step localizes a subset of structural degrees of freedom of an analytical model as measurement points such that the linear independence of the mode shapes to be measured is maximized. In the second step a given number of exciter locations are chosen among the selected measurement points which allow an excitation of the mode shapes. The approach is based on the QR-decomposition of the modal matrix and the QR-decomposition of the product of the mass matrix with the modal matrix.

(PDF-Datei 104 KB)

Proceedings of the 1995 Design Engineering Technical Conferences/Vibration Control, Analysis, and Identification, Vol. 3 Part B, American Society of Mechanical Engineers (ASME), Boston, Massachusetts, USA, 1995

Base excitation testing is used in industry in order to qualify mechanical systems with respect to specified base acceleration levels. This type of excitation only allows to identify eigenfrequencies, mode shapes and modal damping values of the fixed/free system. Modal masses, mass participation factors and effective masses of the fixed/free system as well as the modal data of the free/free system cannot be identified because the excitation forces are unknown.

This paper introduces an approach to identify these modal data as well. For this purpose the reaction forces at the table/structure interface have to be measured also. Furthermore, the verification of the theory using a laboratory test structure will be presented.

(PDF-Datei 250 KB)

European Space Agency (ESA/Estec) Contract Report 10133/92/NL/PP, Noordwijk, The Netherlands, 1994

Over the last ten years ESA has placed contracts for four major studies on updating dynamic mathematical models. The tasks of the first study 1984 were mainly a literature review on modal identification and update, the investigation of test methods, analysis/test correlation procedures and the selection of update procedures, which have been applied to small numerical examples with 5 and 27 degrees of freedom.

Over the years the update procedures have been improved significantly, so that the final report of this study now can present update results of a 30.000 degrees of freedom model for a complete spacecraft on the basis of measured data.

The study team consisting of experienced engineers from the University of Besancon, the University of Kassel and the company Dornier, worked together in the last three studies with success. There was a very close cooperation between the contractors. The developments were partly overlapping, so that the development results could be verified and experiences be exchanged.

Emphasis of the work performed in this study was on

• - advancement of update and error localization method, e.g. with the enlargement of the knowledge space by the use of total masses, effective masses and participation factors or by introduction of artificial new boundary conditions,
• - use of force measurements and test data scatter,
• - identification of additional information as effective masses and participation factors from base excitation tests,
• - identification and update of laminate material properties and damping data,
• - use of frequency response functions and thermal deformations from tests,
• - detection of stiffness degradations and damages,
• - check of already proven methods by large scale applications with real test data,
• - search for new ideas to overcome problems as numerical stability, bad convergence, double frequencies in the sensitivity approaches and non uniqueness of results.