A method for predicting train noise in subway stations

Abstract

Subway stations are typically long enclosures, where classical room acoustic theory is inappropriate [J. Kang, J. Acoust. Soc. Am. 99 (1996)]. In this research, a computer model for predicting the temporal and spatial distribution of train noise in subway stations is developed [J. Kang, J. Sound and Vib. 194 (1996)]. The train is regarded as a series of sections, and the train noise distribution in a station is calculated by inputting the sound attenuation from a train section source in the subway system (i.e., the station and tunnel). This input can be obtained by physical scale modeling. The prediction in a subway station shows good agreement with the site measurements. A series of computations in the station demonstrates that: (1) the overall level of the train noise in the area near the end walls is slightly less than the other areas; (2) some conventional architectural acoustic treatments in the station are effective when a train is still in the tunnel but not as helpful when the train is already in the station; and (3) train noise typically decreases the speech transmission index of public address systems to below 0.5. [Work supported by MTRC, ORS, and COT.]