In addition to an often-increased noise level, trains passing through a curve frequently have additional salient features in the pass-by noise which are typically tonal (curve squeal) or transient covering a wide frequency band (flanging noise). The main mechanisms behind these two sources of noise are well known: the lateral stick-slip effect which produces squeal and the flange hitting the rail thus producing the typical flanging noise.
Recently, two comprehensive studies, BEGEL and ESB, were concerned with the factors influencing the frequency of occurrence, the spectral composition, the detection, and the effect on the pass-by level of such curve-specific noise. The investigation also considered factors such as the track-decay rate, the radius of the curve, the cant, and the pass-by speed among others. Much less is known about how the increase of annoyance caused by such salient acoustic feature changes with respect to spectral composition, duration, or variations in time.
The current study aims at investigating the influence of such emission parameters on annoyance in a perception test in the laboratory with 30 listeners. For this, measurements of three different railway curves with radii less than 500 m are planned. The pass-by noise will be recorded at different distances using conventional microphones. Furthermore, a head-and-torso-simulator will be used to perform binaural recordings in two emmission distances (25 m and, depending on the topography, up to 50 m from the centre of the track).
Based on the data acquired in the two preceding projects at 7.5 m from the track, a classification of typical curve noise will be performed which forms the basis for the selection of typical curve pass-bys in the newly acquired data. On the one hand, this selection process will yield corresponding binaural recordings of curve noise and, on the other hand, typical pass-bys without squeal or flanging noise will be selected. The latter will be superimposed with either modified or synthetically produced squeal or flanging noise samples to systematically produce conditions for the perception test that were not recorded binaurally.
The use of the results of the perception test based on free magnitude estimation is twofold. Firstly, suitable acoustic and psychoacoustic features shall be identified that best model the changes in annoyance due to variations in curve pass-by noise. Secondly, correction factors for the use in practice shall be derived that properly model the effect of squeal and flanging noise on the annoyance. Using these factors and the results of BEGEL and ESB, recommendations will be provided and illustrated using exemplary calculations using noise mapping tools.