Mitigating Noise and Vibration on Rail Transit Systems

by Jeff Tuzik

Noise and vibration-related issues may not be the first concern that comes to mind in considering wheel/rail interaction. But in the rail transit industry, noise and vibration are king. Complaints about noise, in particular, can put immense political and financial pressure on transit agencies, so mitigating these phenomena is critically important to both the ridership and those living and working near the track—and, thus, to the transit agencies themselves.

Noise and vibration are closely related; they are both particle oscillations. As Briony Croft, Director of Sahaya Consulting (Canada) and Acoustic Studio (Australia), explained at the WRI ’25 Principles of Wheel/Rail Interaction: Noise is the oscillation of particles through a fluid medium (air), while vibration is the oscillation of particles through a solid medium. Vibration can be felt, and vibrating structures can in turn re-radiate noise, she said. In an applied sense, as a wheel runs over the rail, they both vibrate and transmit that vibration through the track structure and through the ground, all the while radiating noise.

Rail transit noise and vibration can be grouped into distinct frequency bands. Croft outlined the frequencies and their context within the rail transit environment:

• Very low frequency (less than 100 Hz) is typically associated with the effect of vibration on sensitive equipment and human-perceptible vibration.
• Low frequency (≈100 – 300 Hz) is typically associated with groundborne noise, re-radiated noise and the rumbling of floating slab track.
• Mid frequency (≈300 – 2000 Hz) is associated with rolling noise, corrugation, and wheel impact noise.
• High frequency (greater than ≈2000 Hz) is associated with wheel squeal, flanging, and brake squeal.

“It’s important to remember that these frequency ranges aren’t fixed or necessarily standardized, but rather indicative of general findings,” Croft said.

In discussing noise, it’s also important to have a basic understanding of decibel noise levels. “There are a lot of kinds of decibels,” Croft said. The decibel (dB) describes the loudness of a sound. The A-weighted decibel (dBA), which is commonly used in the industry, also describes the loudness of sound, but with an adjustment factor applied for human noise sensitivity (see Figure 1). In the rail industry, dBA measurements are also often described by time variance, since transit environment noise and vibration levels vary over time.

For example, Video 1 shows a passing train emitting highly varied sound levels. The maximum noise level (or L_Amax) is 86 dBA at the recording location. The equivalent average (L_Aeq) noise level here during the passby is 81 dBA. The equivalent average (L_Aeq) noise level for one passby in a 9-hour night cycle at this location is only 46 dBA. And the day/night (L_DN) noise level for a total of 31 passbys is 61 dBA. “There are lots of different statistical descriptors of noise and they’re not all directly comparable,” Croft said. “So you have to be careful when saying a train passby is this loud, you have to make sure you are clear that you are describing the noise during a single passby, and not something else, such as a 24-hour average noise level.”  

 Rolling Noise and Mitigation

Rolling noise is a ubiquitous and unavoidable part of railroading. This is broadband (non-tonal) mid-frequency noise (≈300 – 2000 Hz), Croft said. The characteristics and loudness of rolling noise is dependent on wheel and rail roughness, wheel design (including size and shape), track design (fasteners, other components, support stiffness), and train speed. Audio 1 is a recording of typical rolling noise.

“Corrugation noise is essentially a special case of rail roughness,” Croft said. “It’s rolling noise that’s dominated by a particular tonal frequency.” That frequency can be a function of train speed and corrugation wavelength. Corrugation is a common phenomenon; it can appear on tangent track and in curves and can have wide variation in wavelength. There are multiple mechanisms that can give rise to corrugation, she said. “There are so many resonances in the wheel/rail system that it’s hard to pin down or mitigate corrugation from a system-resonance standpoint.” Eliminating one corrugation-inducing resonance may simply shift corrugation to a different resonance and wavelength.

Video 2 depicts an example of corrugation noise—the “howling” characteristic heard on the passby of the second train.

In attempting to mitigate rolling noise, there are many factors to consider and many mitigation methods that are viable at certain stages of the noise-generation and transmission pipeline, as seen in Figure 2. Certain mitigation techniques, such as installation of sound barriers or façade treatments and sound-damping windows at trackside residences are less than ideal since they fail to address the cause or source of the rolling noise itself, Croft said. “The real focus on transit systems now is on reducing wheel and rail roughness, adjusting component design, and adding damping to reduce the noise radiated from specific components.” Barriers and track absorption can be very effective at limiting noise propagation in specific situations. Generally speaking, mitigation techniques further to left in Figure 2 are the more cost-effective strategies.

The use of rail dampers, for example, can reduce noise for the residents of high-rises overlooking sections of open track. “Rail dampers are a very specific mitigation technology,” Croft said. “They’re not going to work in every location, and their efficacy is strongly affected by the characteristics of the rolling stock and the track components.” But in the right location, they can provide a ≈4 to 5+ dB noise reduction.

Rail grinding and milling (and to a lesser extent, wheel retruing) are tools of the trade for reducing wheel/rail roughness. Figure 3 shows a comparison of two rails, one of which (left) has been ground aggressively with very hard, low grit grindstones, while the other (right) was ground with a softer, higher-grit grindstone to produce a much smoother rail surface—a relatively new technique (in the U.S.) known as acoustic grinding.

Aside from grinding and milling, there are less immediate, longer-term methods of corrugation and roughness control that include optimizing wheel and rail profiles, using harder rail steels and TOR friction modifiers, and designing (or redesigning) the system to avoid specific resonances, she said.

Flanging Noise, Wheel Squeal and Mitigation

Audio 2 is an example of flanging noise—something that anyone who has ridden or lived near a transit line has surely heard. This is a high-frequency noise (greater than 2000 Hz that results from the wheel flange coming into contact with the gage face of the (typically) high rail in curves. Excessive flanging in curves can indicate an issue with truck steering or angle of attack, Croft said. “Flanging noise can also occur in tangent track, where it typically indicates some other issue or combination of issues at the wheel/rail interface.”

Wheel squeal, like flanging, is more common in curves and is more severe on tighter curves. As in flanging, wheel squeal can be indicative of poor steering dynamics. Unlike flanging, wheel squeal produces a single frequency or pure tones, Croft said (listen to Audio 3 for an example of wheel squeal noise frequencies). “Wheel squeal is also very difficult to predict or model because it’s an unstable resonance,” she said.

Techniques for mitigating flanging and squeal noise share much in common.

• From a design perspective, maximizing curve radii is an effective but often impractical choice.
• Truck design and vehicle steering characteristics can be designed for improved angle of attack and curve performance. “Systems that have steerable bogies, for example, tend not to have the same level of flanging and squeal noise,” Croft said.
• Wheel-mounted noise dampers can reduce the intensity of squeal noise.
• Vehicle-based wheel-flange lubrication, and wayside gage-face lubrication can help to mitigate both flanging and wheel squeal noise.
• TOR friction modifiers can also help to mitigate flanging and wheel squeal noise levels by improving truck steering characteristics.

Impact Noise

The characteristic “clack-clack” of a train running over a switch is a sound nearly anyone could identify (listen to Audio 4 for an example). As when passing over any rail defect or joint, the noise that is generated is proportional to the size of the discontinuity, Croft said. Modelling and predictions typically add a 10-dB adjustment at each discontinuity (i.e. switch).        

Wheel flats/skid flats are another common noise contributor. As on the rail side of the equation, the noise and vibration are again proportional to the discontinuity (listen to Audio 5 for an example). Wheel flats can dramatically increase rolling noise, especially if there are many such defects in the fleet. “According to a recent study, this scenario can increase rolling noise by up to 12 dBA,” Croft said.

Mitigating wheel flats depends heavily on the use of wheel impact load detectors (WILDs) to be able to identify wheels that need to be reprofiled, Croft said. Optimizing friction management can also reduce the propensity for wheel skids. Newer vehicles have begun to incorporate automatic skid protection systems. Taking this a step further, driverless trains have been shown to have much fewer wheel flats.

Mitigating Vibration and Groundborne Noise

The most effective and efficient approach noise and vibration mitigation is focused on “source control”—mitigating noise and vibration as close to the source of its generation as possible, Croft said. And it’s also incumbent on developers to design new buildings near existing transit systems with noise and vibration mitigation in mind.

Groundborne vibration and the associated noise is an issue requiring careful attention for transit systems. “The propagation path of noise and vibration is typically through the wheels and rails into the ground (including both subway and surface track), through the soil layers and into the buildings nearby,” Croft said. This causes walls and floors to vibrate and re-radiate low-frequency noise (see Figure 4). “Often the re-radiation of groundborne noise is the real issue, rather the vibration itself,” she said. Accordingly, contributors to groundborne noise include:

• Longer wavelength roughness or irregularities
• Track components and design: “Track design is a very large factor in how much vibration is propagating into the ground,” Croft said.
• Impacts and discontinuities in the wheel/rail interface
• Special trackwork
• The un-sprung mass of the vehicle
• Vehicle suspension design
• Out-of-round wheels

Taking into account all these factors—some of which can only occur during the design phase of a new transit system—is critical to mitigating vibration and groundborne noise. This includes the use of resilient wheels, minimizing un-sprung vehicle mass, and designing the track with vibration isolation included. “To get the best effect, you want to increase the mass above the main resilient element in the trackform,” Croft said.

Resilient elements in the track include (on ballasted track) softer rail pads, under-tie pads, and under-ballast mats. On slab track, these elements include softer rail pads, softer baseplate pads, booted ties, and floating slab construction (which can be based on rubber bearings, coil springs, and other designs). See Figure 5 for examples of these resilient components and their configurations.  

Unfortunately, modelling vibration and groundborne noise is a difficult task. Referring to work done by Hugh Hunt at the University of Cambridge, Croft said: “He has said that’s it’s unreasonable to expect any better than a ±10dB accuracy for any prediction model for groundborne vibration.” The reason for this is that even a relatively small 15% change in soil parameter assumptions can have a 6dB effect in vibration modelling.

Sources of uncertainty in modelling the propagation of groundborne vibration include ground layers, voids, structure foundation coupling, groundwater, excitation, amplification at structure resonances, and assumptions and simplifications in the models themselves, she said.

Empirical predications based on the US Federal Transit Administration Transit Noise and Vibration Impact Assessment Manual “give a very reasonable estimate of vibration and groundborne noise,” Croft said, but added that estimates and predictive models are no substitute for actual measurements, particularly in highly sensitive areas. But taking the time and effort to perform detailed measurements during the design phase is much more cost-effective than having to retrofit mitigation methods after the system is built.

Monitoring and Compliance

“If you have a noise complaint, you need to know if you comply with the requirements that apply to your railway,” Croft said. This means the noise must be measured at the complainant location and compared to the criterion. “This will tell you if you comply or not, but it doesn’t tell you the source of the issue or what you can do about it.”

This includes recording the noise characteristics to determine the general source (flanging noise, wheel squeal, corrugation, etc). There are other factors to consider, too. It’s important to know the grinding and maintenance history of the rail, if the rail is scheduled for replacement, the status of friction management on the relevant track section, and the condition of track components, Croft said. These will all help to narrow down the mitigation strategies that will best fit the situation.

Both wayside and onboard measuring equipment (i.e. microphones and sound meters) can also help to determine if the source of a noise complaint is due to a systemic issue, or a specific location or vehicle. Onboard axle-vibration monitors can also help to diagnose noise and vibration concerns, which helps to determine the scope of mitigation efforts, Croft said. These measurement techniques and devices can also be used to assess the long-term status of the system from a noise and vibration standpoint. They can also illustrate and validate the effects of rail grinding and wheel and rail reprofiling on noise and vibration over time.

Mitigating noise and vibration isn’t just good for the people living near transit systems, it also minimizes wear and damage to the vehicles, track, and components, and maximizes asset life. Simply put, “achieving noise and vibration objectives overlaps with best practices in achieving and maintaining a state of good repair,” Croft said. These mitigation techniques aren’t always easy, but with the right knowledge, mindset, and maintenance practices a transit system can be designed from the ground up and maintained to minimize harmful noise and vibration.