Two traffic representations have been tested in this paper. Comparison was achieved by confronting estimates to on-field data, collected at 8 points corresponding to different traffic situations: close to a bus station, in front of a traffic signal, down a traffic signal, and set back from the corridor. The coarse static representation often gives errors in L_(Aeq) estimation that exceed 3 dB(A). This representation overestimates noise levels, as it is based on a coarse estimation of vehicle mean speeds. Moreover, the static model fails in reproducing the spectra envelopes along the corridor. In particular, low frequencies sound levels are systematically underestimated, as they are mainly emitted by vehicle at slow speeds or accelerating vehicles, which are not reproduced by the static model. This could be problematic when achieving noise impact studies, especially in the presence of noise reducers, which often offer better results for high frequencies than for low frequencies. Results of the static model could be improved by using speed distributions instead of a mean speed when assessing noise emissions. This could be obtained by elaborating noise emission laws that correspond to real traffic situations.
展开▼