Survey of Curve Squeal Occurrence for an Entire Metro System

The current work presents a statistical analysis based on data collected during approximately 1.5 years of regular operation by two vehicles equipped with an on-board noise monitoring system on the Stockholm metro. Data covers 379,776 passages through 143 curves with radii up to 1000 m. Binary logistic regression is used to investigate the importance of curve radius, vehicle speed, relative humidity, air temperature, rail grinding and vehicle individual on curve squeal. Curve squeal occurrence shows an inverse proportionality with respect to curve radius. This trend is particularly pronounced for curve radii below 600 m. The two vehicles accounted for in the study show differences in propensity to generate squeal. The influence of temperature and relative humidity, and their interaction, on curve squeal is described by an estimated response surface. Results show the occurrence of curve squeal to increase after rail grinding. No strong relationship between curve squeal occurrence and vehicle speed is identified.