Some results of benchmark cases in linear acoustics

In 2014, the Technical Committee Computational Acoustics of the European Acoustics Association has launched its benchmark initiative. For this, a first paper has been published in Acta Acustica united with Acustica 101:811-820 (2015). Three initial benchmark cases have been proposed therein, i.e. the long duct, the cat's eye and the radiatterer. This talk refers to two of these benchmarks and shows how useful a careful investigation of these problems. By analysing the radiatterer, the author discovered a surprising problem in the use of the method of Burton and Miller. Analysing the long duct problem, exhibits the problem that the conventional boundary element formulation in acoustics suffers from numerical damping. Numerical damping is a problem if sharp resonance peaks are expected. However, numerical damping is also causing an additional error which increases with the length of a model and challenges the idea that a certain number of a certain type of elements per wavelength are the only parameter to control the accuracy of a boundary element model.