Prediction of structural dynamic behaviour under uncertainties: With applications to automotive structures
Licentiate thesis, 2017

The automotive industry is moving towards shorter development cycles for new car generations. This means that less expensive prototypes can be built and tested, and that, increasingly, computer models must be used for decision making. Further, the automotive industry is producing thousands of nominally identical cars which are known to exhibit noticeable spread in their vibration characteristics. A car's noise and vibration behaviour is therefore not the same between nominally identical cars. This implies a need for structural dynamic models considering uncertainties for robust decision making. Due to the final products complexity a substructuring approach is considered in this thesis, including experimental and computational methods, where predictive models of components are created, to be assembled for a predictive system response. The first part of this thesis considers the reduction of uncertainties introduced from vibration experiments. A method for sensor placement in vibration experiments is developed, based on the method of effective independence, so that symmetric sensor positions are rejected using system gramians. Further, a measurement system is developed in MATLAB for fast and efficient stepped sine excitation. The second part considers the spread between nominally identical components and the calibration, and an associated parameter uncertainty quantification, of industrial finite element models of said components. Results are reported here for three front and one rear subframe. For model calibration, a model updating procedure is employed that uses a frequency response function based deviation metric and equalised damping. A bootstrapping procedure is subsequently used to quantify parameter uncertainties with respect to the measurement noise. Calibrations are performed for an ensemble of front subframe components. Particular care is taken in the modelling of coupling elements and for the rear subframe the elastic modulus in rubber bushings is estimated using a mass loaded bushing boundary configuration. In the automotive industry high fidelity models are common, with many interface degrees of freedom decreasing the efficiency of component mode synthesis methods. Therefore, a component mode synthesis interface reduction method is developed to speed up the process, using coarse meshes.

sensor localisation

experimental design

noise and vibrations

structural dynamics

automotive industry

model updating

uncertainty quantification

substructuring

Gamma/Delta, Hörsalsvägen 7A, Göteborg, Sweden
Opponent: Dr. Andreas Josefsson, Saab Aeronautics, Linköping, Sweden

Author

Mladen Gibanica

Chalmers, Applied Mechanics

Redundant Information Rejection in Sensor Localisation Using System Gramians

Topics in Modal Analysis & Testing, Conference Proceedings of the Society for Experimental Mechanics Series,; Vol. 10(2016)p. 325-333

Paper in proceedings

Calibration, Validation and Uncertainty Quantification of Nominally Identical Car Subframes

Model Validation and Uncertainty Quantification, Conference Proceedings of the Society for Experimental Mechanics Series,; Vol. 3(2016)p. 315-326

Paper in proceedings

Parameter Estimation and Uncertainty Quantification of a Subframe with Mass Loaded Bushings

Model Validation and Uncertainty Quantification, Conference Proceedings of the Society for Experimental Mechanics Series,; Vol. 3(2017)p. 61-76

Paper in proceedings

A reduced interface component mode synthesis method using coarse meshes

Procedia Engineering,; Vol. 199(2017)p. 348-353

Paper in proceedings

Subject Categories

Mechanical Engineering

Vehicle Engineering

Thesis for the degree of licentiate of engineering - Department of Applied Mechanics, Chalmers University of Technology: 2017:03

Publisher

Chalmers University of Technology

Gamma/Delta, Hörsalsvägen 7A, Göteborg, Sweden

Opponent: Dr. Andreas Josefsson, Saab Aeronautics, Linköping, Sweden

More information

Latest update

9/21/2018