Acoustic shape optimization including viscothermal losses

Background
Computer-based simulation and optimization tools can provide engineers with valuable insight into the behaviour of acoustic devices. These tools can also reduce the need for expensive prototypes, which are often rebuilt in multiple iterations of manufacturing and experimental validation. Microphones and loudspeakers are acoustic transducers. The general trend is to reduce their size as much as possible so they can fit inside, for example, smartphones and hearing aids. As the size is reduced, sound waves inside such devices are modified in their propagation by acoustic losses, which are explained by the viscosity of air and the heat exchange with the device material. Including these effects into simulations with traditional methods comes at a significant computational cost, which in turn makes simulation and optimization of these small devices very challenging. 

Project
The aim of the project is to improve the computational performance of an existing Boundary Element Method shape optimization technique that incorporates losses, and extend its usability to more general 3 dimensional problems. This is achieved by investigation into fast approximate viscothermal models and development of adjoint sensitivity methods.  

Perspective
The results of the research has the potential to improve  the design of acoustic devices and include viscous and thermal losses in the optimization process.

Abstract
Computer-based numerical acoustic optimization methods have over the last two decades evolved from simple academic examples to a viable engineering tool, that can be used to improve the design of acoustic devices, such as smartphones and hearing aids. However, to accurately model such devices require the inclusion of acoustic losses, which is often neglected or simplified in the optimization process. The recent development of an acoustic shape optimization technique that incorporates viscothermal losses without simplification has shown the importance of losses when performing optimization. In this project, the computational performance of the shape optimization technique is improved by implementation of new fast viscothermal models and development of adjoint sensitivity methods. Paving the way for large scale and more realistic three-dimensional acoustic shape 

Supervisors

Vicente Cutanda Henríquez (DTU Elektro) & Niels Aage (DTU MEK)

Completed in 2019

Contact

Peter Risby Andersen
Postdoc
DTU Electrical Engineering
http://www.act.elektro.dtu.dk/Research/Postdoc-projects/Peter-Risby-Andersen
7 APRIL 2020