Accounting for traffic dynamics improves noise assessment: Experimental evidence

This paper compares three traffic representations for urban traffic noise assessment: (i) a coarse static calculation based on mean speeds and flow rates, (ii) a refined static calculation based on mean kinematics patterns, (iii) a whole dynamic noise estimation model that considers vehicle propagation on the network. The three methodologies are applied on real traffic situations and compared to on-field noise levels. Representation (i) is not refined enough to guarantee a precise noise assessment. Representation (ii) can be sufficient for LAeq estimation in most of cases. However, representation (iii) improves noise estimation since it considers vehicle interactions on the network. Moreover, it allows for specific descriptors to be estimated with a great accuracy, like the LAeq,1s distributions or the mean noise pattern that reproduces every traffic cycle. Finally, the dynamic noise estimation appears to be still consistent if the model is fed with data averaged on 2-h period.