Dissertation Proposal: Efficient Wave-base Sound Propagation Optimization for Computer-Aided Design

A standing problem in the field of acoustic engineering is the automated design of buildings and structures for the purpose of obtaining desired acoustic properties, including noise metrics and metrics related to characteristics of the acoustic impulse response. These goals are related to the usefulness of the building or structure: a lecture hall should have clear and understandable sound from the lectern, a concert hall should have the feeling of being surrounded by sound, and a school or hospital should have minimal environmental noise levels. There are various challenges in the automated design of buildings according to these requirements. First, many of the requirements are influenced by low-frequency sound. However, traditional efficient geometric techniques for sound propagation cannot easily represent some important low-frequency wave effects such as diffraction. Therefore, wave-based approaches that directly solve the acoustic wave equation and thus simulate every aspect of sound wave propagation should be used for the purpose of accurate acoustic simulation. However, the computational requirements of wave-based acoustic simulations can be extremely high. This is especially true for simulations on very large domains; as a rule, wave-based approaches scale in complexity with frequency of the simulated acoustic signal to the fourth power and linearly with the volume of the domain. The second challenge in automated acoustic design is using simulation techniques to optimize the acoustic configuration of a room or building in order to yield one that conforms to the design requirements. This problem in particular is sensitive to the first challenge: in order to optimize a function that is dependent on the way sound propagates throughout the environment, one needs to evaluate the acoustic solver. Finally, there is the direct problem of the interface between computer simulation and real-world application. Verification of simulated results is important in addition to applying the results on real-world scenes. Additionally, acoustic design optimization is a subset of multidisciplinary design optimization, where other constraints (such as structural constraints) may be in effect. Therefore, automated acoustic design needs to be usable with external constraints and cost functions.