Michel Villot

This paper is about the history of the acoustic activities at CSTB, French Center for building Science and Technology. It is an improved and extended version of a paper presented in a structured session on the history of acoustic activities in Europe, at Forum Acusticum 2023 in Torino, Italy. The CSTB Acoustic Department was created in the 1960s around an acoustic test laboratory. From the beginning CSTB has been involved in improving acoustic performance for building occupants. It played an important role in technically supporting the French acoustic regulation (first one in 1968 and last one in 1995) and in developing solutions to reach mandatory building performances in cooperation with the industry. The acoustic department grew fast, mainly supported financially by the government (more than 60% of its budget at the beginning), thus becoming the largest acoustic team in France and worked on 3 main domains: room acoustics, environmental acoustics and building acoustics, in which physical models and calculation tools have been developed to bring science to construction. From the start, CSTB has been involved in standardization at French, European and International levels.

The sound radiated by the ground vibration generated by trains has been estimated by calculation in a case study and compared to the total sound measured on-site. A 2.5D ground numerical model, calibrated from on-site vibration measurements, has been used to estimate the ground vibration field, from which the ground-borne sound radiated has been estimated using the Rayleigh Integral Method. According to the calculations, the estimated ground-borne sound radiated is dominant at low frequencies (below 125 Hz) and otherwise rather negligible compared to the sound directly emitted by the train/track system. Moreover, a small parametric study performed using a 2D source (vehicle and tracks) model coupled to the same ground model shows that the ground-borne sound radiated is ground dependent. The ground should be taken into account as a separate sound source when dealing with railway sound and should be added to the other known sources (track and vehicle components).

Noise measured inside buildings close to railway lines is often a mixture of ground-borne noise and airborne noise. Predicting ground-borne noise from vibration measurements and comparing the result to noise measurements may be useful to identify ground-borne noise. In this paper, the prediction is performed using the building acoustics theory and a power-based parameter called radiation efficiency. Due to the low-frequency range of ground-borne noise where both building elements and rooms have strong modal behaviour, the usual values of radiation efficiency – either measured or calculated under diffuse field conditions – cannot be used anymore. However, this parameter can still be measured or calculated using numerical models, but it becomes situation dependent. In this paper, the values of radiation efficiency useable at low frequencies are calculated and their spectra discussed. Rough approximations are then proposed, thus leading to simplified formulae, useable at the engineering level for prediction but only valid under certain assumptions.

A method for characterizing and predicting vibration emission from railway lines in tunnels is presented. The characterization method consists in measuring both the tunnel vibration response during train pass-by events and the tunnel effective mobility (mobility combining point and transfer mobilities), from which the line of uncorrelated forces characterizing the source can be calculated. The results from field measurements of commuter train operations are presented and improved using a numerical tunnel/ground model. The transfer of an existing situation to a new one is performed using two different train/track/tunnel models. The situation change is expressed in terms of difference in the force density applied to the tunnel invert between the original and the new situation. The track models are first calibrated from the characterization field measurements and then used to estimate a given situation change. The results of the two models/procedures compare rather well, thus cross-validating the approaches.