Acoustic immersive experimentation through real-time control of boundary conditions

Abstract

Acoustic waves play an important role in society: they form the basis for oral communication and acoustic echo- and source-location, and are exploited for non-invasive imaging of the interior of media in both research and industry. Imaging applications range from the millimeter scale in medical diagnostics to kilometer scale in seismic exploration. Recently, new techniques have been developed to control such waves as they propagate. In particular, the method of immersive boundary conditions (IBCs) allows one to design and produce acoustic wavefields on demand, and to create a dynamic coupling between the real world and a virtual environment, such that waves propagate seamlessly between the physical and virtual domains. This thesis explores practical aspects of a physical implementation of IBCs in a wave propagation laboratory. The immersion of a physical experiment within a virtual environment can be realized by implementing the following key ingredients: [1] an (ideally) sound-transparent sensor surface, [2] a Kirchhoff-Helmholtz-type extrapolation of the recordings from these sensors to the boundary of the physical domain, including all interactions with the virtual domain, [3] injection of the predicted boundary wavefield by means of sources along the boundary. With this latter step, the correct immersed wavefield is radiated as a boundary condition into the experimental domain. To ensure a dynamic coupling in real time, the predicted boundary conditions are updated and injected at every time step of the experiment based on the current and past wavefield recordings. As part of this thesis, the ability to impose arbitrary boundary conditions on an experimental domain is demonstrated in a variety of one-dimensional experiments in air. A broadband cloaking experiment shows that scattering objects can be made acoustically undetectable, even without prior knowledge of the incident wavefield. Moreover, acoustic illusions are created: virtual objects are emulated where they are not present – even objects exhibiting properties that do not occur in nature such as media in which energy increases during wave propagation. The results provide a proof of concept that IBCs can be used to account for all higher-order, long-range interactions of physically propagating waves with an arbitrary virtual environment in practice. Additionally, the development of a 3D underwater IBC laboratory involving advanced, custom- built hardware is discussed. A low-latency data acquisition, compute and control system is capable of extrapolating wavefield recordings simultaneously from 800 sensors to 800 sources. The real-time nature of IBCs poses strict requirements on the custom-built sensors and sources. In contrast to conventional acoustic experimentation, undesired hardware effects such as frequency- dependent transfer functions cannot be corrected post the experimentation, but must be included in the real-time extrapolation in the form of compact finite impulse response filters. This thesis provides a comprehensive acoustic characterization of the sensors and sources, which is a fundamental preparatory step for the first underwater IBC experiments. Finally, a method to determine experimentally the frequency- and angle-dependent reflection coefficients of planar interfaces is presented. The method relies on the injection of recorded wavefield quantities into a numerical simulation using so-called multiple-point-source injection. Thereby, the recorded wavefield quantities are separated into their incident and reflected components, and independently redatumed as if they were recorded at the reflecting interface. This method is particularly useful for the characterization of reflection coefficients of the source surface in an IBC experiment, since the effect of that reflection must be included in the real-time extrapolation.