3D plot of sound level vs. frequency vs. time Cycling road test: 0 to 15 to 0 MPH in 25 seconds Green is lowest sound intensity; red is highest
When cycling, air flowing around the head experiences rapid and chaotic changes in pressure, and these pressure fluctuations are perceived as wind noise. The physics of wind noise are complex, as is the way in which individuals respond to it. Wind Noise can be loud enough to be annoying and unsafe, masking other sounds, like those of an approaching vehicles.
Source: Cat-Ears testing
At Car-Ears we are fascinated by the problem of wind noise.
If you would like to learn more, please read on...
Caveat to a very short course in acoustics, aerodynamics and aeroacoustics
The brief discussion of acoustics, aerodynamics and aeroacoustics that follows must start with the proviso that air flow is a highly complex physical phenomenon that is very hard to model and predict. The mathematics are difficult, including partial differential equations, vector calculus, and tensor notation. For most real-world situations, it is not possible to directly solve the equations governing air flow, and engineers must rely on approximation and experiments, using apparatus like wind tunnels. Advances in computer power allow for better approximate solutions of flow equations; this is the basis for the field of computational fluid dynamics (CFD). The physics and mathematics that describe sound generation and sound fields - acoustics - are different than that of aerodynamics and complex in their own right. Combining aerodynamics and acoustics into a single field of study - aeroacoustics - is required in order to fully understand and predict aerodynamic wind noise. This is a very tough challenge. What follows is a very high level tutorial, absent mathematics.
What is air?
There are four phases of matter: solid, liquid, gas, and plasma. At normal temperatures and pressures, air is a gas - actually, a mixture of gasses. Air is also considered a fluid. A fluid is a substance that exhibits little to no resistance to deformation (try making a brick out of a gas or a liquid). Fluid mechanics is the study of the physical behavior of fluids, including liquids and gases. This is important because air and water move according to the same physical laws and that has important consequences for the testing that Cat-Ears performs.
What is wind?
Wind is the large scale flow of air from a higher pressure zone to a lower pressure zone. Under ideal conditions, the air moves in a uniform, predictable manner known as laminar flow. Much of the time, however, because of day/night temperature changes, and surface features like trees, wind flow is chaotic. Wind flows are reasonably predictable at large scales (your county) in terms of direction, average speed, and gusts, but not at a smaller scale (your street). Gusty winds are an indication of the chaotic flow known as turbulence.
A moving cyclist creates “wind” - obviously on a small scale - by motion through still air. On a windy day, the air flow experienced by the cyclist is a combination of bike speed and direction and wind speed and direction.
Fluids have a property known as viscosity. It is a measure of how readily layers within the fluid move relative to one another. In other words, a measure of the fluid’s internal resistance to flow. The forces between layers in a fluid are called shear forces, and viscous fluids experience higher shear forces than non viscous fluids. Water has low viscosity and flows readily; molasses has high viscosity and does not. Although air flows readily, it still experiences viscous shear forces between layers, although the effects are very small, except as noted below.
When a fluid flows over a surface (known as a boundary), the layer in immediate contact with the surface does not move with the flow. This is due to molecular forces between the surface and the fluid. The space between the surface where flow velocity is zero and the layer in the fluid where the particles are moving at 99% of the full velocity of the flow is called the boundary layer. The forces on the particles in the boundary layer are dominated by viscous shear forces, while the forces on particles outside the boundary layer are dominated by what are called inertial forces – those imparting momentum to the moving particles. The behavior of boundary layers is very important in aerodynamics.
Laminar, attached boundary layer
Two dimensional flow around a cylinder
Engineers use a number, known as the Reynolds Number (Re) to describe flow conditions. The calculation involves the density, viscosity, and velocity of the fluid, and a characteristic length of the boundary constraining the flow (such as pipe diameter or in our case below cylinder diameter). Reynolds number is very useful, since it has been determined that whenever Re is the same, the flow characteristics of different systems with similar geometries are the same. This is the basis for testing scale airplane models in wind tunnels to predict what actual aircraft will do, and for Cat-Ears testing product by substituting water flow for air flow.
Our interest is in what happens as air moves around a human head. But for simplicity let's start with air moving around a very basic shape: a cylinder.
The first few characteristic stages of air flow around the cylinder all occur at low Reynolds numbers, equating to velocities much slower than a rider experiences save for a few fractions of a second when accelerating from rest. But it is helpful to go through them, as it will be easier to understand what is happening at normal cycling speeds (please refer to the pictures at right, which are courtesy NASA). Initially, the air flows all the way around the cylinder in smooth, orderly, predictable layers. This is called laminar flow. The boundary layer remains attached to the cylinder, and the flow within the boundary layer is also laminar. Air pressure is highest where the air flow impinges on the front of the cylinder.
As speeds increase, the shear forces in the boundary layer become too high,and it can no longer remain attached to the rear of the cylinder. The boundary layer separates from the cylinder, but the flow remains laminar. Pressure drops behind the cylinder (and thus begins to create aerodynamic drag).
A further increase in speed causes the wake behind the cylinder to no longer flow in smooth layers, but rather to become chaotic with extensive mixing of what had formerly been smooth layers. This is called turbulent flow. Rotational flow patterns, called vortices, are generated and begin to alternate from side to side well behind the cylinder. At this point, speeds are still less than 1 MPH or 0.5 m/s.
As speeds increase to a couple of m/s (~5MPH), the point on the cylinder at which the flow separates continues to move forward. The boundary layer remains laminar, and the flow after separation remains turbulent. As speeds continue to increase up to the maximum that might be encountered by a cyclist, the flow conditions remain largely the same, but the turbulent flow is more energetic. A wide, turbulent wake is produced, and this is the point of maximum aerodynamic pressure drag (drag is not directly related to the noise problem we are working, but interesting none-the-less).
There is one final stage of flow, included for the sake of completeness. In this stage, the flow inside the boundary layer becomes turbulent, and by in doing so is once again able to extend around the sides towards the back of the cylinder before separating. Although turbulent flow remains, the size of the wake is reduced, and hence the drag. The effect on wind noise is unclear, as we have not been able to test this scenario. Why? The Reynolds number is above 500,000, requiring speeds of about 70 m/s, or 160 MPH.
Re < 1
Re = 1
Re = 100
Re = 20000
Re = 1,000,000
Obviously the shape of a human head is not a perfect cylinder, and the nose, cheekbones, and outer ears have an effect on air flow. However, It is best to start with the simplest possible model, understand the physics as best we can, and then attack the more complex case. And look at how complicated the air flow over a simple cylinder is!
Now that we have a basic understanding of air flow, we need to turn to sound.
Sound is characterized by sound pressure level (how loud it is) and by frequency. Sound pressure level (SPL) is measured in decibels (dB). Because the human ear can hear sounds over a huge range of levels, the dB scale is logarithmic, like the Richter scale used to measure earthquake intensity. Accordingly, 50 dB is ten times the SPL of 40 dB. 60 dB is 100 times the SPL of 40 dB. The range of human hearing is remarkable: a race car engine has a SPL roughly one billion times higher than a pin drop. Roughly speaking, the human brain perceives a 10 dB SPL increase as a doubling of loudness, with 2 dB being the minimal detectable change. However, human hearing is very complex and these numbers vary with frequency, SPL, and individual.
Noise is sound that interferes with other, desired sound signals. Engineers use a calculation known as the signal to noise ratio (SNR) to compare the level of desired sound to noise.
You might note that Cat-Ears measures wind noise in a unit called dBA. This scale is modified to reflect the fact that humans are less sensitive to sound at very low frequencies. dBA is the standard metric in the US for measuring noise.
What is noise?
Sound is energy that travels through fluids such as air and water as a mechanical pressure wave. This energy is a wave phenomenon that causes air particles to vibrate back and forth, but without net movement (after the sound wave passes, the particles are back in their original position). These air particle movements are extremely small, as are the pressures generated.
These pressure waves are converted by the complex mechanisms of the human ear into signals that the brain interprets as sound.
What is wind noise?
When air flow becomes turbulent, it swirls and changes directions and speeds very rapidly and unpredictably. These swirling, rotational patterns of flow are called vortices. You generally can’t see this in air without special test equipment. However, occasionally you can see vortices coming off the rear surfaces of an airliner wing when flying in damp air. You can also see them in the turbulent flow of smoke rising from a lit candle. Turbulent flow is a very complex physical phenomenon that even with powerful computers is very hard to precisely predict.
Many observers have noted that sound is generated under certain conditions when turbulent flow occurs. A good example may be found in the high speed air hand dryers found in public restrooms. When you place your hands in the air flow, a loud noise is perceived. Clearly, the introduction of a solid body - your hands - into the air flow greatly increases the level of noise. This noise is called aeroacoustic noise.
Wind noise varies in intensity (and to a lesser extent, frequency) depending on speed. Most of the energy in wind noise at speeds encountered during cycling is at what is considered low frequencies: 250 hertz and lower. For reference, middle C on a piano is at 262 hertz.
In the hand dryer example above, your hands are in the air stream; your ears are not. Thus, you know that the pressure waves acting on your ear drum, causing you to "hear" the wind noise, are sound waves. However, if you place your head into the dryer's air stream, you will also perceive wind noise. That noise, however, may come from another source, known as pseudo sound. Pseudo sound as perceived by the brain does not come from sound waves, but from aerodynamic pressure waves that vibrate the ear drum in a manner similar to sound waves.
Wind Noise Intensity Plot
Intensity vs. Speed
Bike road test using in-ear microphones
Source: Cat-Ears testing
Wind Noise Spectrum Plot
Intensity vs. Frequency
Approximately 13.5 MPH
Source: Cat-Ears testing
How do cyclists experience wind noise?
Cyclists experience wind noise when moving through still air. If it is a windy day, the noise level will vary depending on bike and wind speed and direction. Perceived noise levels increase as speeds increase, and as they do, the ability to discern useful sounds, like the voice of a nearby cyclist or the sound of an approaching car goes down (ie, the SNR is going down). Different individuals experience different perceived levels of noise, for a complex variety of factors that includes head shape and hearing acuity.
At zero degrees (head straight ahead), both ears will on average experience equal wind noise at speeds of a few MPH and above. Note that a side wind will alter the overall direction of the air flow relative to the cyclist.
Source: CE wind tunnel testing, bare head at 17.5 mph
As the head turns, the ear turned into the wind at first will experience a slight increase in wind noise, and then noise levels will decrease until they are substantially reduced when the head is turned 90 degrees and the ear is perpendicular to the air flow. For the ear that turns away from the wind, noise is reduced until 45 degrees, at which point wind noise no longer reduces as the head turns to 90 degrees.
What is happening?
As the air moves around the temples, it strikes the outer ear, a complex shape. As the air flow interacts with the ear, rotating air currents called vortices are formed, and it is the interaction of these vortices with the ear that generates sound waves which travel to the ear drum.
At zero degrees, both ears are exposed to a similar flow, and hence experience similar patterns and intensities of noise. As the head turns, or as a cross wind shifts the direction of air flow, each ear will experience a different flow and hence different wind noise levels.
Aerodynamic and aeroacoustic testing and simulation
In order to predict and manage wind noise, we have to understand both aerodynamic flows and the associated sound formation mechanisms. There are well established techniques for simulating and testing air flows, as well as for simulation and testing of sound, but the combination - aeroacoustic simulation and testing - is highly specialized. Much recent work has focused on the noise produced by large wind turbines. Automobile companies study aeroacoustics because they do not want exterior body design elements like mirrors to increase interior noise. Japanese researchers have done work in the field to understand noise generated by high speed trains.
Wind tunnels are a proven tool for studying aerodynamic flows. The object to be tested is placed into the air stream, enabling the measurement of the forces on it, or the visualizing of the flow around it using smoke or other means. The problem is that wind tunnels, while very good at creating well behaved air flow under controlled conditions, are noisy. So noisy that it is practically impossible to do aeroacoustical testing in one, without much effort to silence it.
A large scale wind tunnel. Image courtesy General Motors.
A quiet wind tunnel is specialized and very expensive, and only a few of them exist. Furthermore, to precisely measure sound in the wind tunnel's test chamber requires careful acoustic design to minimize sound reflection off hard surfaces. A chamber that is designed to absorb all sound reflections, as well as provide insulation from external sound, is called an anechoic chamber. Since we are interested in wind noise, we must concern ourselves with very low frequencies. To build an anechoic chamber that functions down to 20 Hz, a few calculations yield the following dimensions: height of anechoic wedges 4.2 meters; major dimension of chamber 12.8 meters; minimum chamber volume 5 cubic meters. Clearly not a small chamber..
Anechoic test chamber at Cal State Poly
Thus an aero-acoustical test facility suitable for rigorous engineering testing of objects generating low frequency wind noise is a combination of a quiet wind tunnel and an anechoic chamber, in large scale. Like that of Mitsubishi Heavy Industries, shown below. Note the relative size of the cars parked outside!
A further complication concerns measurement. The state of the art in fine resolution visualization of aerodynamic flows involves seeding the air flow with tiny reflective particles, such as titanium dioxide, and using a specialized laser and high speed camera system in a synchronized sweep across the flow field (known as Particle Image Velocimetry). Since we are not just interested in air flow but also in sound pressure wave generation, we need similarly high resolution measurement of sound wave intensity, frequency, position, and direction. Some bleeding edge work is being done using arrays containing as many as hundreds of microphones.
What about simulation? Let's start with aerodynamics. Can't we use math to predict the flows? The equations are well established, but so complex that we don't know how to solve them for all but the simplest conditions. A lot of work has been done to use ever more powerful computers to arrive at approximate solutions to the equations, and this field is known as computational fluid dynamics or CFD. There are also simulators for sound generation. Some researchers are working to link the output of CFD simulation - very detailed three dimensional information on pressure variations - with the input of a sound simulator, in order to predict aeroacoustic noise. This is a very difficult task and much work remains.