Explainer: Tuft Testing

Tuft testing is an experimental methodology most amateur aerodynamicists are familiar with, even if the vast majority of them have never actually tried it. This is too bad, since tuft testing can give us a lot of good information about the quality of the flow over a car.

 
We all know that this involves taping bits of yarn to your car, like what the actors playing Carroll Shelby and crew do to the Ford GT in a (comically reductive) scene in Ford v. Ferrari (2019). The tufts as depicted in that film are pretty much useless for trying to determine what the "engineers" were attempting to ascertain (magnitude and direction of front lift), but in real life they can be tremendously instructive if applied with purpose and interpreted correctly. So, what is the physical explanation behind tuft testing, why does it work, and what can it actually tell us?

Stress
 
To explain tuft testing, we first need to define some terms and concepts, beginning with stress.
 
Divide force by area and, if that force is not applied directly perpendicular to a surface, we get not "pressure" but stress. "Stress" refers to a force exerted over an area in an arbitrary direction. Pressure is one component of stress, normal (perpendicular) to a surface. The other component is shear, which acts in orthogonal directions along the surface or, in other words, tangent to it.

When the force is exerted directly perpendicular to the paper, it creates only pressure stress at the point of contact (upper image). The more general case is a force exerted at some oblique angle, in which case there are normal (pressure) and tangential (shear) stresses (lower image).

Shear
 
Shear stress is the component of stress acting tangent to a surface (the dashed green and blue arrows in the second image above), and that surface can be solid or fluid. Wherever moving air touches a car body, for example, the fluid exerts stress that acts both normal and tangential to the body. Normal stress (pressure) is usually designated p; shear stress is usually designated τ ("tau"). The general variable for stress is σ ("sigma"), usually accompanied by subscripts x, y, or z. You may see normal and tangential stresses identified in this last way, with the first letter in the subscript designating the surface and the second letter indicating the direction of the stress component e.g. σ_yz acts on the plane normal to the y axis in the direction of the z axis and is thus a shear stress. These stresses are commonly represented in a stress matrix where each row is one face of a cube element (assumed in equilibrium) and each column is a direction:

The diagonal of this matrix is pressure: stress on the x face (that is, defined by a normal vector along the x dimension), pointing in the direction of x, for example. Using matrices like this, various calculations can be performed such as the determination of the principal stress direction (orange arrow in the image above) by matrix rotation; once the principal stress direction has been found, we can rotate our coordinate system to align with it to make calculations easier. This is also possible to do geometrically, using "Mohr's circle." You may have heard of this in connection with tire traction but that is not its intended application! It was developed for internal stress calculation in materials engineering.
 
Shear stress also acts between the "layers" of moving air. Wherever these layers move at different speeds, there is friction between them and the friction stress vector points in the same direction as the local flow. This friction between air layers is especially important in the boundary layer, where there is a velocity gradient (change) as the velocity increases from 0 at the surface to freestream speed some distance away, and in regions of separated flow.

(M. Clarke, AE416 Applied Aerodynamics Class Notes, University of Illinois, Fall 2025).

Tufts
 
Tufts are short lengths of yarn or string taped to a surface to show the direction of flow. How does this work?
 
We've seen so far that shear stress on a surface or in an airflow always has a magnitude (strength) but also points in a direction. The direction of shear in airflow is always aligned with the local velocity. Since velocity at the surface where the tufts are taped imposes shear on not just the body but also the yarn or string, the tufts are pulled in the same direction as the flow by the local shear stress (this is true as well if the tufts are taped to something like a fishing line or wire grid, or are otherwise left in freestream; the local shear imposed by the velocity acts on the tufts due to friction with the tufts themselves). So, tufts taped to a car body can show us the direction of flow wherever we place them. This is commonly called "tuft testing." Because tuft testing does not return a numerical measurement value, it is a form of qualitative testing but it is no less important than quantitative measurements like static or total pressure.

Tuft Testing
 
Because the tufts align themselves with the local direction of shear, we need them to be short enough that the shear direction doesn't change much over the length of the tuft.
 
When the tufts are short enough, they show "streamlines" or lines that are tangent to instantaneous velocity everywhere in the flow. In a steady flow (one that does not change over time), streamlines are the same as streaklines and pathlines, like the trails traced out by smoke in a wind tunnel. Turbulent flow (like over most of a car's surface) is not steady by definition but on average we can sometimes consider it steady—which means if you look at tuft directions averaged over some time interval we can usually consider them as representing the local instantaneous velocity and hence approximating streamlines.
 
Tuft length is important. It's common to see tuft tests conducted using pieces of yarn that are verging on too long; if you can see the yarn following the surface curvature but still twisting or bending around over the length of a single tuft, you need to cut them shorter (I've found that 2"-3" is about the right compromise between keeping the tufts short while having enough length to ensure they are visible to a camera). If your tufts are the right length, each one will point in essentially a short, straight line.
 
Why tuft test at all? What can it show us? First, tuft testing is perhaps the easiest method of flow visualization—a way of "seeing" what otherwise invisible airflow is doing. Tuft testing is used for this purpose by everyone from enthusiasts to NASA engineers:

F/A-18 Hornet research vehicle flow visualization (image credit: NASA).

Visualizing flow like this allows us to see the streamline patterns and flow direction over the vehicle's surfaces.

Second, and perhaps more importantly, tuft testing allows us to identify areas of attached and separated flow. When a boundary layer follows a surface, it is said to be attached; even if the surface gently bends or curves away from the direction of local airflow ahead of it, the boundary layer (and inviscid flow outside the boundary layer) bend to follow it—a phenomenon called the Coanda Effect.
 
However, if the boundary layer has lost enough streamwise momentum/dissipated enough energy, it no longer follows the surface curvature and can detach. If this happens, the boundary layer is said to be separated. Separation can be roughly predicted by a value called the "shape factor" H,

where δ^* is the boundary layer displacement thickness and θ is the boundary layer momentum thickness. When H becomes large enough (in other words, when enough streamwise velocity in the boundary layer has been lost), the boundary layer tends toward instability and separation. (You may recognize H as a term in the von Kármán Momentum Integral Equation, which can be used to predict friction drag).

 
Generally (there are always caveats, such as sharp edges to deliberately separate flow at the back of the car), we want to see attached flow over the car body everywhere we can—verified through observation of tufts taped to the various body panels. Attached flow tends to dissipate less energy and lose less momentum than separated flow. Decreased axial momentum loss in the flow behind your car translates directly to lower aerodynamic drag.

When the boundary layer separates, the flow near the body surface reverses direction, something that is immediately identifiable in tuft behavior. In practice, this looks like the tufts being pulled backward (against freestream flow direction) and flapping around, sometimes even standing straight up. Let's look at some examples.

 
Examples

Burst mode on a cell phone camera can be used to capture a quick succession of tuft images on a car as it passes.

Small sections can also be photographed or recorded on video. If the area of interest is a window, this is easy to do from inside the car.

Tufts can be used to check for separation and attachment. Here, the attached flow over the hood in stock configuration (upper image) has been deliberately ruined by fitting a mock bug deflector (lower image).

On the same truck, tufts taped to the bed show the recirculation behind the cab. Notice the tufts pointing forward on the bed floor and tailgate.

Choosing the right position from which to photograph can make all the difference. Here, I positioned myself at the side of the road and down low to capture tuft behavior not just around the cooling air inlet but underneath the splitter and on the engine undertray.

Tufts are useful in wind tunnel experiments as well. This Clark Y wing section shows the difference in flow at low angle of attack (6°, upper image) and high (26°, lower image). Almost the entire upper surface of the wing is in separated flow at the higher angle; this is called stall. A common misconception is that, in stall, the wing completely loses lift; this is not true at all. This wing had about the same sectional lift coefficient at both angles of attack, just over C_l = 1.0 (but much higher maximum lift between these two, at α = 14° and C_l = 1.5).

Go try tuft testing on your car!