Moving sources and the 2.5D Helmholtz Boundary Element Method
* Presenting author
For perception tests on environmental noise, e.g. evaluating the effect of a noise barrier, it can become necessary to virtualize moving sources. In the framework of the 2.5D boundary element method, which is very useful for long and large structures, the full treatment of moving sources becomes computationally extremely involved. Even in the simplest case, i.e. mono-frequent emission and constant speed, higher speeds result in the necessity of calculating ever higher 2D frequencies beyond the signal’s frequency in order to perform the necessary inverse Fourier transform to acquire the time signal. For wide-band signals a 2D inverse Fourier transform needs to be performed. Adding further a non-uniform motion results in yet another dimension that needs to be integrated over. The evaluation for sampled signals of sufficiently high sampling rates and durations of a few seconds can thus become challenging. Interpolation of data points in either frequency, space, or time can help to decrease the computational burden in particular when employing different types of demodulation schemes. The aim of the work presented is to investigate for sampled source signals at which points in the calculations interpolation can be applied without introducing too high a computational error.