Sound insulation with tailored properties: a multiphysics simulation approach for acoustic metamaterials

语音或音乐形式的声音为人类互动铺平了道路。然而，噪音会导致严重的健康问题，并对我们的健康和环境产生重大的负面影响。日益增长的城市化和交通负荷使得实现当前的噪声保护目标更具挑战性，需要创新和有效的隔音技术，如声学超材料。这些包括局部共振微结构阵列，调整以消除不希望的振动声学模式，如发动机噪声。然而，目前的设计方法要么计算成本高，要么不合理地假设完美的周期性，缺乏边界层效应。这些假设通常会导致在功能和制造过程方面的不正确或次优设计。本研究项目的目标是发展一种用于声学超材料研究和设计的新型数值方法。假设是有可能设想和验证一个多物理场模拟模型，该模型可以准确地解释流固相互作用和边界层效应，例如粘热损失，同时足够有效地处理大型非周期模型。本文提出了一种基于边界元法（BEM）的无流声学方法。粘热不连续配置边界元法考虑了边界层效应及其吸收势。进一步扩展了流固耦合的双向耦合条件。最后，快速多极方法有效地求解了现代异构硬件架构下的方程组。这三种方法的结合将使具有定制隔音性能的声学超材料的鲁棒性和准确性建模成为可能。

Sound in form of speech or music paves the way for human interaction. Noise, however, can cause severe health issues and has a significant negative impact on our well-being and our environment. Increasing urbanisation and traffic loads make reaching current noise protection goals even more challenging and require innovative and effective noise insulation technologies like acoustic metamaterials. These comprise arrays of locally resonant microstructures tuned to cancel undesired vibro-acoustic patterns such as engine noise. However, current design methods are either computationally expensive or unwarrantedly assume perfect periodicity and a lack of boundary-layer effects. These assumptions often result in incorrect or sub-optimal designs regarding their functionality and manufacturing process.This research project targets the development of a novel numerical method for the investigation and design of acoustic metamaterials. The hypothesis is that it is possible to conceive and validate a multi-physics simulation model that accurately accounts for fluid-structure interaction and boundary-layer effects, e.g. viscothermal losses, while being efficient enough to handle large aperiodic models.We pursue a novel approach based on the boundary element method (BEM) for flow-free acoustics. The viscothermal discontinuous collocation BEM accounts for the boundary-layer effects and their absorption potential. The method is further extended by bi-directional coupling conditions for fluid-structure interaction. Finally, the Fast Multipole Method efficiently solves the system of equations on modern heterogeneous hardware architectures. The combination of these three methods will enable the robust and accurate modelling of acoustic metamaterials with tailored sound insulation properties.