PhD Thesis - Model substructuring and interface reduction techniques for complex vibro-acoustic systems

Candidate: André Antonio Andrade Paiva, University of Liège



Modern means of transport such as marine vessels, cars, trucks, planes, trains, and others, increasingly integrate acoustic criteria in their design process. Usually, both the interior noise (which impacts the passenger’s comfort) and the exterior noise (which affects the environmental pollution) are considered. The compliance with more demanding noise standards and regulations, added to a shorter development time, leads engineers to use digital simulations to select best performing designs and technical concepts.

The expected accuracy often requires high-fidelity modeling, integrating more geometric details and analyzing a greater frequency range to predict as accurately as possible how the end-user will perveive the implemented design. This leads to very large models, that could reach tens of millions of unknowns, which can only be solved with considerable computational resources and represents a high cost, especially in time (hours, days or even weeks for the most ambitious models). This high calculation time is especially detrimental to optimization strategies, which demand iterative computations of different design configurations to reach a given objective, and to robust design which assess variability of the system response caused by the uncertainty on certain key parameters.

The design process for complex systems is also increasingly collaborative. As an example, in the automotive industry, while the manufacturer design the structure, and often, the engine, it also act as an integrator for multiple subsystems that are designed and manufactured by subcontractors. Similarly, the current design of cruise ships incorporates a large number of standardaized passenger cabins which are prefabricated modules. The design of these cabins is the heart of the shipyard business in Europe. A recurring difficulty consists in predicting the vibroacoustic response of the complete system based on assemblies of existing module numerical.

The Ph.D. project addresses these two themes by proposing to study an innovative method of modeling a complex system by coupling different subsystems. The aim is, at the same time, to be closer to the collaborative design process described above and to reduce the total computational cost, which will ease the implementation of an optimization loop or robust design.

Specifically, we propose a novel sub-model coupling strategy, based on the use of pellicular modes (developed over several years by Free Field Technologies for speeding up acoustic radiation simulations [1]). The innovation relates to the use of the pellicular modes as a projection space where the coupling relation between the different components will be enforced. These pellicular modes are fictitious sound pressure distributions that correspond to the acoustic modes of a cavity which would have the shape of the coupling surface extruded over a small thickness perpendicular to the coupling surface. If these modes do not correspond to anything physical, they have the double advantage of providing a user-selectable number of orthogonal functions on the coupling surface (regardless of the coupling surface shapes) and, by their wave variation, to be each characterized by a given wavelength which ensures the filtering of the information transmitted between the two coupling surfaces. As such, the pellicular modes consist of a promising choice concerning the subspace of representation of component’s behavior at the level of its coupling surface, which will be compared with more classical choices of coupling modes such as local modes, analytical functions and FE shape functions.

The basic milestones of this projection technique were developed by Free Field Technologies and published in [1] and [2].

The research topic described in this abstract will be fully developed in a partnership with the company FFT as part of the REMOVASCO framework (Ph.D. in company) partially funded by the Walloon Government.


[1] "A new technique for obtaining vibration data from acoustic measurement based on an inverse finite element technique," EC Project AST5-CT-2006-030814, acronym: CREDO, Cabin Noise Reduction by Experimental and Numerical Design Optimization, DWP6.3: Plan for using and disseminating the knowledge: section 2.3.8, 2009.

[2] J. Jacqmot, Y. Detandt, G. Lielens and D. Copiello, "Vibro-aero-acoustic simulation of side mirror wind noise and strategies to evaluate pressure contributions," in Proc. Of the ISMA Conference, Leuven, 2016.

Share this page Share