PAPERS
Study on the Occurrence Conditions of Squeal Noise and High-frequency Noise in Railway Curved Sections
2025 Volume 66 Issue 2 Pages 123-130
Details
2025 Volume 66 Issue 2 Pages 123-130
Squealing (below 10 kHz) and high-frequency noise (above 10 kHz) can occur when trains pass through curved sections of track. Measurements of these noises were carried out on commercial lines to understand how they are generated. It was found that these noises vary from wheel to wheel, having large dispersion. The analysis results showed that these noises are prominent when the outer wheel flanges are in contact with the outer rail at certain passing speeds. On the other hand, these noises are reduced when a train is running at equilibrium speed in curves.
When trains run on a curved track, both wheels and rail often radiate noise with higher frequency than rolling noise which has frequency components of around 500-2,000 Hz at 160 km/h [1], for example. Noise generated in curved sections has tonal components, whereas rolling noise in straight sections usually has broadband components. The noise with higher frequencies in curved sections is generally called “squeal noise.” In some cases, the noise with much higher frequency, above 10 kHz, is also observed at curved sections.
Squealing and high-frequency noises greatly contribute to overall wayside noise and may cause complaints from residents. A lot of research and studies have been carried out on noise in curved sections [2, 3, 4, 5]. However, the conditions under which this noise occurs are not fully understood.
There are two main railway systems in Japan. One is meter-gauged railways and the other is Shinkansen. Meter-gauged railways, sometimes called “conventional railways,” have relatively small curve radii, level crossings, etc. Trains on meter-gauged railways in Japan run at 130 km/h or less, except for a few railways. Shinkansen is a high-speed railway system that has some designated features such as specialized rolling stock, large curve radii, no level crossings, and uses standard gauge, etc. Shinkansen trains generally run at 200 km/h or more. Squeal noise and high-frequency noise occur on both meter-gauged railways and Shinkansen lines.
In this report, squeal noise and the high-frequency noise were measured and their characteristics were investigated in commercial meter-gauged and Shinkansen lines. The conditions under which they occur were analyzed.
Squealing and high-frequency noises that occur exclusively in curves are generated by complex contact between the wheels and rails. In this study, these noises are categorized based on dominant frequencies as follows:
− Squeal noise (2-10 kHz) [2, 6]: Noise which is mainly generated when a train runs through sharp curves and is caused by relative slippage between wheel treads and rails. The main frequencies are from 2 kHz to 10 kHz. This type of noise is particularly noticeable on inner wheels and rails.
− High-frequency noise (above 10 kHz) [7]: This noise is mainly generated when a train runs through relatively gentle curves and is caused by contact between the wheel flange and the railhead. The main frequencies are above 10 kHz. This noise tends to be particularly salient on outer wheels and rails.
Although tendencies and frequencies of occurrence differ between meter-gauged and Shinkansen lines, squealing and high-frequency noises occur on both types of railway system as discussed in Chapters 3 and 4. It should be noted that the forms of occurrence cannot be strictly classified and may be mixed.
Figure 1 shows the outline of measurement points on a section of meter-gauged railway line. The section used for measurement is a curved section with a 404 m curve radius, superelevation of 98 mm, and no gauge widening. The track consists of ballasted track, Japanese 50-kg N type rail, concrete sleepers, and rail pad with 110 MN/m stiffness. The equilibrium speed in this curve is 68.7 km/h.
Two microphones were installed close to the track and on both the inner and outer rail sides to be roughly symmetrical across the track. The height of microphones was set roughly equal to wheel axle level at 450 mm height from the rail head on the outer rail side and -130 mm on the inner rail side. The position of the wheels relative to the rails in the transverse direction were also measured.
The trains used for the measurements were four types of commuter trains, consisting of four, six, seven, and eight cars. Measurements were made in a section located approximately 100 m from the station. Some trains accelerated through the section after having stopped at the station. Other trains passed through the section at a constant speed because they did not stop at the station. Trains which accelerated through the section are referred to as “local trains,” while those which ran through it at a constant speed are referred to as “rapid trains.” Local trains passed through the section below the equilibrium speed, while rapid trains travelled above the equilibrium speed.
All data was collected when weather conditions were either sunny or clouded, which means that the rail surface was dry.
3.2 Representative results of noise measurementsFigures 2 and 3 show the spectrograms of noise for a local train and a rapid train passing through the section, respectively. The black lines in the lower part of the figures indicate the range over which train passes through the measurement section. It should be noted that in these results, 0 dB is set to averaged overall value of the measured noise, obtained by the microphone close to the outer rail, for all results when the wheels of all trains pass through the section.
Figure 2 shows the noise spectrogram for passage of a local train. This train was one of four types described in the section 3.1 and consisted of eight cars. The leading car of the train passed at 40 km/h and the trailing car at 60 km/h. The average speed was 51 km/h. All the cars of the local train passed through this section below the equilibrium speed.
As shown in Figs. 2 (a) and (b), the sound pressure levels (SPL) below 1 kHz were higher when trains passed through. They correspond to rolling noise. The levels were also higher at around 6-9 kHz at around 5 and 8 seconds, and those at around 19 kHz at around 9-13 seconds.
The lower frequencies were considered to be squealing noise, and the higher ones considered to be high-frequency noise. The number of seconds correspond to wheel passage. The component around 19 kHz was modulated with respect to time due to the Doppler effect.
Comparing the results for the inner and outer rail sides, the tendency appears to be that larger squealing noise is observed on the inner rail side, and the greater high-frequency noise on the outer rail side. However, the noise levels on the outer rail side in the frequency range of 5-15 kHz were greater than those on the inner rail side at 10-13 seconds, indicating that there were slight variations to levels in squeal and high-frequency noise tendencies depending on time.
As shown in Fig. 2, the level of high-frequency noise around 16 kHz remains at a high level from around 13 seconds to after the train has passed, albeit at a lower level that when the train was actually passing. It is thought that this may be a result of the rail becoming a source of and radiating sound.
3.2.2 Rapid train measurement resultsFigure 3 shows the noise spectrogram for passage of a rapid train. This train consisted of eight cars, with a train speed of 76 km/h. The train type was the same as the local train descried in Section 3.2.1. Train speed was a constant since it did not stop but ran straight through the station. All cars passed through this section at a constant speed, which is above the equilibrium speed.
As shown in Fig. 3, the levels corresponding to rolling noise (below 1 kHz) of the rapid train were higher than those of the local train due to differences in speed. Squeal noise occurred at around 6 seconds (above 4 kHz) and 8 seconds (around 8 kHz).
The high-frequency noise, approximately 18 kHz, became particularly noticeable at 6 seconds, corresponding to the wheel passage considered to be the noise source. The frequency of high-frequency noise was modulated with respect to time due to the Doppler effect. Although the noise level fell after the wheel passage, a relatively high noise level continued as the train passed. This was due to the noise from the rail. The magnitudes of noise below 13 kHz, which occurred at around 6 seconds, were greater on the inner rail side than on the outer rail side.
3.3 Occurrence situations in the meter-gauged railwayThe results described in the previous section indicate the occurrence of squeal noise and high-frequency noises was not uniform, even for the same train. This tendency was seen in other trains, not reported in this paper. Therefore, averaging alone is not suitable for explaining the phenomena.
This section therefore investigates conditions where squeal and high-frequency noise occur using a histogram to verify data distribution. To this end, the top 10% of data which seemed to express these noise were used.
Figure 4 shows two-dimensional histograms of the frequency spectrum of the noise when each wheel passes in front of the microphones. These histograms were derived from 140 trains, i.e. 3,332 wheel sets.
The level value distributions were almost constant and the most counted values were in the middle of the distributions at frequencies below 4 kHz in Fig. 4. In contrast, the distributions above 4 kHz were wider, the most counted values were in the lower part of the distributions, and the values were also small. The distributions were especially wide at 6-7 kHz and 14-20 kHz, indicating that squeal noise or high-frequency noise occurred only on some wheels and varied greatly in magnitude from wheel to wheel.
Figure 5 shows the magnitude of the 6-7 kHz (representing squeal noise) and 14-20 kHz (representing high-frequency noise) components on the front wheel of each bogie. These figures were derived from the top 10% of data. The blank areas had no data. The horizontal axis shows wheel passing speed, and the vertical axis the lateral position of wheel measured by laser displacement sensor (see Fig. 1). The positions where the outer wheel flange contacts the outer rail are shown in Fig. 5 as the red lines. The outer wheel flange contacts the rail when wheel position exceeds 38 mm.
Figures 5 (a) and (c) indicate that squeal noise at 6-7 kHz is particularly noticeable when the train is running slightly below the equilibrium speed, around 52 km/h, and when the outer wheel flange is in contact with the outer rail. Sound pressure level on the inner rail side is greater than that on outer rail side, indicating that the outer wheel coming into contact with the outer rail affects the noise that occurs on both the inner and outer rail sides. However, on condition that velocities are anywhere except around 52 km/h slightly below the equilibrium speed then even with outer rail/wheel flange contact, the 6-7 kHz component is small. When there is no contact between the outer wheel and the outer rail, the 6-7 kHz component is also small whatever the speed.
Figures 5 (b) and (d) show that in the high-frequency noise component, the sound pressure level is high when the outer wheel flange is in contact and the outer rail at speeds below and above the equilibrium speed. However, at speeds closer to the equilibrium speed, the sound pressure in the high-frequency noise range is slightly lower. The generation of high-frequency noise even when the equilibrium speed is exceeded, is a point which differs from what happens with squeal noise.
Figure 6 shows the set up for taking measurements on a Shinkansen line. The chosen section is a curved section on a viaduct with a curve radius of 3,500 m, 180 mm superelevation, and no gauge widening. The track consists of slab track, Japanese 60-kg type rail, and rail pads with 60 MN/m stiffness. The equilibrium speed in this curve is 236 km/h.
Two microphones, one on the inner rail side and one on the outer rail side, were installed close to the rail. The height of microphones was 180 mm below rail head level for the outer rail side and 455 mm above rail head level for the inner rail side. Although the positions of the microphones should have been installed in the same configuration as on the meter-gauged railway line as described in Fig. 1, it was impossible to put them symmetrically. The position of the wheels relative to the rails in the transverse direction were also measured.
Five types of Shinkansen train were used for making measurements. Three consisted of sets with eight cars and two with sixteen cars. Trains passed through this section at different speeds. In this test, trains passing through the section at speeds faster than the equilibrium speed are called “high-speed trains” (approximately 240-260 km/h), and those running below this speed are referred to as “low-speed trains” (approximately 150-230 km/h).
All measurements were taken in sunny or clouded weather conditions, so the rail surface was dry.
4.2 Representative results of noise measurementsFigures 7 and 8 show the spectrograms of noise for one low-speed train and one high-speed train, respectively. The black lines at the bottom of the figures indicate the range during which trains passed through the measurement section. In these results, 0 dB corresponds to the averaged overall values of measured noise, obtained by the microphone close to the outer rail, for all results when the wheels of all trains pass the section.
Figure 7 shows the spectrogram of the low-speed train. This train consisted of eight cars, and was running at a speed of 195 km/h.
As shown in Fig. 7, the noise levels below 4 kHz, especially around 2 kHz, were higher while trains passed by. The noise levels mainly corresponded to rolling noise caused by marks left on the rail after grinding. Their wavelength was approximately 30 mm and equivalent to 1.8 kHz at 195 km/h. Squeal noise occurred at frequencies of approximately 2 kHz, 5 kHz, and 10 kHz at around 4 s. At this time, the levels at these frequencies contained both rolling and squeal noises.
High-frequency noise occurred at frequencies of 12-17 kHz with passing trains. This means that it was generated around almost all wheels. The frequency of high-frequency noise was modulated with respect to time due to the Doppler effect. The levels of these frequencies remained high before and after trains passed by. They were also radiated from the rails, similar to the measurement at the meter-gauged railway described in the previous section.
The level difference of noise between inner and outer rail sides were not compared directly because these microphones were not installed symmetrically. A previous study [7] reported that high-frequency noise is predominant on the outer side of the rail, so there is a high possibility that the noise on the outer rail side is also greater than that of the inner rail side of the rail in this section.
4.2.2 High-speed train measurement resultsFigure 8 shows the spectrogram for a high-speed train. The train consists of 16 cars, with a speed of 255 km/h.
As shown in Fig. 8, the levels below 4 kHz, especially around 2.5 kHz, were also greater when trains passed by. These correspond to rolling noise. The frequency due to the mark generated by rail grinding was 2.4 kHz at 255 km/h. High-frequency noise occurred only on some wheels at a frequency of approximately 14 kHz, unlike the low-speed trains. The components at these frequencies were modulated with respect to time due to the Doppler effect. Squeal noises were not observed in this train.
4.3 Occurrence situations in the Shinkansen lineEven on the Shinkansen, the occurrence of squealing and high-frequency noises were not uniform, even within the same train. This tendency was seen in other trains, not shown in this paper. Therefore, averaging alone was not suitable for explaining the phenomena.
In this section, as well as in the section 3.3, the occurrence conditions of squeal and high-frequency noises were investigated using a histogram to verify data distribution. Then, the top 10% of data which seemed to express these noise were used.
Figure 9 shows two-dimensional histograms of the frequency spectrum of the noise when each wheel passes in front of the microphones. Note that they were derived from 119 trains i.e. 5,435 wheel sets.
There tended to be small level variation at all frequencies because the most counted values were in the middle of the distribution and their values were high at the middle, except around 2 kHz, 4.5 kHz, and above 10 kHz. This indicated that the variation between trains was small in these frequency ranges except around 2 kHz, 4.5 kHz, and above 10 kHz. The frequency of around 2 kHz corresponded to rolling noise due to the marks of rail grinding, which varied respect to train velocity, leading to less count of the level. The frequency of around 4.5 kHz was affected by squeal noise, also leading to less count of the level. Those above 10 kHz corresponded to high-frequency noise. The distribution above 10 kHz were wide and most counted values at these frequencies were not high. This was due to that the occurrence of the high-frequency noise was varied.
Figure 10 shows the magnitudes of component of 8-10 kHz (representing squeal noise) and 10-20 kHz (representing high-frequency noise) on the front wheel of each bogie. These figures were derived from the top 10% of data. The blank areas had no data. The horizontal axis means the wheel passing speed, and the vertical axis means the lateral position of wheel, which was measured by the laser displacement sensor (see Fig. 6).
Here, the laser displacement sensor was installed on the inner rail side, unlike in Chapter 3. This meant the contact relationship was reversed. The positions where the outer wheel flange was in contact with the outer rail are shown in Fig. 10 as red lines. The outer wheel flange is in contact with the rail when wheel position is below 15-20 mm. It was noted that almost all front wheel flanges are in contact with the rail below equilibrium speed. However, some flanges of front wheels are not in contact with the rail above the equilibrium speed.
Figures 10 (a) and (c) show that noises in the frequency range of 8-10 kHz are also particularly noticeable on each side when the outer wheel flange is assumed to have come into contact with the outer rail and the passing speed was around 195 km/h, which is below the equilibrium speed. Noises in the frequency range of 8-10 kHz fell from 220 km/h to the equilibrium speed. Whereas these noises increased again above the equilibrium speed, their level remained below those of around 195 km/h. When the outer wheel flange was assumed not to have come into contact with the outer rail, the noises in the frequency range of 8-10 kHz were quieter.
Figures 10 (b) and (d) show that the level of noises in the frequency range of 10-20 kHz are great on each side when the outer wheel flange is assumed to have contacted the outer rail and passing speed was below the equilibrium speed. The noise in the frequency range of 10-20 kHz above the equilibrium speed was slightly smaller than that below the equilibrium speed. When the outer wheel flange was assumed not to have contacted the outer rail, the noises in the frequency range of 10-20 kHz tended to be quiet. As shown in Figs. 7 and 8, high-frequency noise above 10 kHz was observed in almost all bogies of low-speed trains, whereas that noise was only observed in the part of bogies of high-speed trains. This could be due to that some wheel flanges did not contact the rail when the train speed exceeded the equilibrium speed.
Based on the results described in Chapters 3 and 4, it can be assumed that the conditions in which squealing and high-frequency noise occur in curves have following tendency, regardless of whether they are meter-gauged railways or Shinkansen.
● Squealing occurs easily when two conditions are met: trains are running at speeds slightly below the equilibrium speed and the outer wheel flanges are in contact with the outer rail. Squealing is observed both on inner and outer rail sides of curves. When comparing both, the noise on the inner rail side rail is predominant.
● The train speed at which squealing occurs easily, was around 52 km/h in the meter-gauged railway and 195 km/h on Shinkansen, respectively. These speeds correspond to -0.39 m/s2 of the excess centrifugal acceleration in both the meter-gauged railway and Shinkansen.
● High-frequency noise is observed at both inner and outer rail sides of curves when outer wheel flanges are in contact with outer rail.
● The magnitudes of these noises tend to decrease when trains are running close to the equilibrium speed or when the outer wheel flanges are not in contact with the outer rail. These characteristics are common in both meter-gauged trains and Shinkansen trains.
We investigated squealing and high-frequency noise generated when trains pass through curved sections for meter-gauged trains and Shinkansen trains, respectively. Results gave insights about the conditions in which these noises are generated.
The conditions described in this report are based on estimations made from measurement results at two locations. It is necessary to obtain more samples to verify whether similar results can be obtained for all curves. It is also necessary to reveal mechanisms on these occurrence conditions. We plan to research these issues in the future.
|
Takeshi SUEKI, Dr.Eng. Chief Researcher, Noise Analysis Laboratory, Environmental Engineering Division Research Areas: Railway Noise |
|
Yasuhiro SHIMIZU
Assistant Senior Researcher, Noise Analysis Laboratory, Environmental Engineering Division Research Areas: Railway Noise |
|
Takuma NITTA
Supervisor, Head of the Kanazawa Branch Environmental Measures Office, West Japan Railway Company Research Areas: Railway Noise |
|
Kentarou TAKAI
Supervisor, West Japan Railway Company Research Areas: Railway Noise |
