报 告 人：Steffen Marburg 教授
Steffen Marburg教授1998年于德国德累斯顿工业大学获博士学位。2010至2015年担任德国慕尼黑联邦武装部队大学工程动力学讲席教授。2015年7月起担任慕尼黑工业大学汽车与机械振动噪声研究所讲席教授。他的研究兴趣包括振动噪声模拟及模拟方法、不确定性量化、参数及参数分布的实验测定、缺陷检测，以及其他振动噪声相关问题。Marburg教授是欧洲声学学会计算声学委员会主席之一，Journal of Theoretical and Computational Acoustics（之前称为Journal of Computational Acoustics）期刊共同主编，The Journal of the Acoustical Society of America和Acoustics Australia两个期刊副主编。他目前已发表100余篇学术论文和8本学术著作章节，他还作为Editor之一出版了学术著作——Finite and Boundary Element Methods in Acoustics，以及作为Guest Editor负责了JCA/JTCA期刊的8个special issues.
Finite element models have been used for virtual prototyping for more than four decades. The setup of a digital twin is one of the major goals of engineering since it allows to avoid expensive construction of real prototypes and may even be used for optimization purposes. While this may sound very reasonable, the real world of virtual prototyping in vibroacoustics happens to be an extremely complex one. It is widely accepted that the finite element model of a sedan body, a vehicle power train or even an entire car are only suitable and valid in a low frequency range while energy based approaches such as the statistical energy analysis are well suited for high frequency simulation. A reliable finite element model of a whole vehicle may cover the frequency range of up to 200....600 Hertz. Assuming that such a reliable finite element model is available, it can be parameterized and optimized by a simple numerical parameter optimization.
This presentation will discuss finite element modelling techniques together with experimentally based identification, uncertainties and numerical optimization. For this, the theoretical foundation of finite element modeling and numerical optimization are briefly revisited. The author will discuss strategies and problems of realistic finite element models. A number of challenges is identified and three of the will be illuminated in some detail. These three challenges are related to structural damping by sound radiation, effect of gravity in case of modal analysis of plates and box structures and uncertainties of material parameters of composites. For optimization purposes, an objective function is required. An appropriate choice of the objective function can be crucial for the success of optimization. In vehicle industry, it is common to use the sound pressure at one or a few points within the cabin to assess the acoustic quality. A weighting over the frequency range is finally required to end up with a scalar value to assess the entire NVH quality of a vehicle in an optimization process. This strategy will be briefly discussed. Another crucial problem of optimization consists in efficient numerical algorithms. In this context, the author will briefly discuss a few approaches to substantially increase the efficiency of evaluation of the objective function and for efficient optimization. Among them are the use of acoustic transfer vectors and efficient sensitivity analysis. Finally, the author will present a few optimized designs of sedan body structures and discuss experimental verification of optimization results.