How to Think About Car Aerodynamics: A Very, Very Basic Overview

The study of aerodynamics is complicated. If anyone tries to tell you otherwise, run the other direction—it’s a sure sign they don’t know what they’re talking about.
 
Over the last two years especially, my thinking about aerodynamics and appreciation of its complexity has changed dramatically—a result of my going back to school to get another bachelor’s degree, this time in aerospace engineering where a good working knowledge of airflows is required and education in not just general fluid mechanics but also aircraft aerodynamic design forms a core part of the technical curriculum. I'm in the midst of my last semester now and to clarify my thinking at this point I decided to put some things in writing in the hopes they might help someone else as well as myself, specifically focused on car aerodynamics.

A word of warning: I've tried to minimize the amount of math below, but some mathematical relations are unavoidable if you want to build an understanding of fluid flows. If anything is unclear, feel free to email me under "Contact Us" in the sidebar.

Modern cars, such as this Jeep Wagoneer S electric SUV, have dramatically different shapes than cars even just a few years ago. How does this geometry influence the aerodynamic drag that is generated as the car moves? To answer that, we have to start with the basics.

Momentum and Force
 
First, some definitions so that we’re all on the same page.
 
Momentum is a property of matter in motion. For a body of mass m and velocity v, its momentum is proportional to both:

Since mass is a scalar (that is, it has magnitude but no direction), the product of mass and velocity gives another vector that always acts in the same direction as velocity. This will be important later on, where we will see that consequently, vector decomposition of the velocity gives the same decomposition of momentum.
 
If you’ve taken a physics class, you might remember that force is described by the relation F = ma. This is a simplification of Newton’s Second Law which in its original formulation states that force is proportional to the alteration of motion ("mutationem motus"), which we understand to mean the time rate of change of momentum. By the product rule of derivation, this gives:

Consider a rocket: as propellant is expelled, both the mass of the rocket and its velocity change with time, and the force (thrust) on the rocket is dependent on both. (This is true of combustion-powered cars as well, but the mass expelled is much smaller! In fact, I read an article earlier this year that mentioned the new Corvette ZR1 produces 37 lb of thrust at its top speed of 233 mph due to mass expulsion through its exhaust—about 4% of total thrust required for the car to attain that speed).

 
Why is this important? Well, in formulating descriptions of fluid behavior, we apply the conservation of properties such as mass and momentum to arrive at mathematical "governing equations." The change in momentum of a fluid through a selected "control volume" gives us the force exerted on the fluid by the surface of the control volume and in turn the force exerted on the control surface by the fluid (this momentum balance gives us a set of relations known as the Navier-Stokes equations, after their discoverers). In other words, aerodynamic forces are caused by momentum transfer between air and (arbitrary) boundary; if we select a solid body as the boundary, this momentum transfer gives rise to a force acting on the body because its momentum changes with time (the definition of Newton's Second Law, remember). This is the first thing you must understand if your goal is to manipulate or alter airflow over your car to produce lower drag, lower lift, or both. More on this below.
 
The Origin of Aerodynamic Force
 
In the literature, you’ll read about all sorts of aerodynamic phenomena: turbulence, wake size and shape, trailing vortices, entrainment, separation and attachment, etc. It is easy to get caught up in these, especially as a beginner, without really understanding their importance or even what they mean sometimes. But I would argue, you don't need to concern yourself with these at all if you're just starting out or even have some experience with aerodynamics. Fundamentally, the aerodynamic force that acts on a body traveling through air—such as a car—arises from (and solely from) the pressures that act on the body and the shear stresses that act on the body, everywhere it touches air. That’s it. There are no other mechanisms for the fluid to transfer momentum to the car body (or vice versa, depending whether you’re conceptualizing air moving past the car, as in a wind tunnel, or the car moving through air, as on a road). No magic. Just pressure—or normal (perpendicular to the surface)—stress, and shear—or tangential (along the surface)—stress, and these stresses act everywhere the car’s surface is in contact with air. Daniel Raymer writes in Aircraft Design: A Conceptual Approach (6th edition, 2018):

"...There are only two ways in which the air mass and the airplane can act upon each other. One is friction caused by shear layers, and the other is pressure. Friction is always tangential to the surface, and pressure is always perpendicular to it. All of our terminologies and coefficients are just different combinations of these two." ("Terminologies and coefficients" refer to things like induced drag, leakage and protuberance drag, and interference drag—terms I've seen plenty of people get hung up on without understanding that these are essentially bookkeeping classifications).

Charles Jobe wrote in Thrust and Drag: Its Prediction and Verification (1985):

"The resultant aerodynamic force caused by a flight vehicle's motion with respect to the atmosphere is the summation of the pressure, or normal forces, and the tangential, or skin-friction forces, acting on the vehicle's surface."

John Anderson, in Fundamentals of Aerodynamics (6th edition, 2017), uses the example of a car specifically:

"Similarly, the aerodynamic resistance on an automobile traveling at 55 mi/h on the highway involves a complex interaction of the body, the air, and the ground. However, in these and all other cases, the aerodynamic forces and moments on the body are due to only two basic sources:

1. Pressure distribution over the body surface

2. Shear stress distribution over the body surface

No matter how complex the body shape may be, the aerodynamic forces and moments on the body are due entirely to the above two basic sources. The only mechanism nature has for communicating a force to a body moving through a fluid are pressure and shear stress distributions on the body surface" (emphasis original).

That surface or "wetted area" is important, because the stresses exerted by the airflow integrated (added up) over wetted area are the physical mechanism that produces aerodynamic force. This is why wetted area is used in many empirical and theoretical relations in aerospace engineering, especially in the estimation or calculation of drag coefficients. For example, we can make a basic or initial prediction of the parasitic drag coefficient, C_D0, of an aircraft with the relation

where C_fe is an "equivalent friction" coefficient (that is, an estimated effect of shear stresses—itself a function of the two flow similarity parameters, Mach number and Reynolds number) multiplied by the ratio of wetted area to reference (wing) area. Plug this into the equation for aerodynamic drag and you will find that drag area is some percentage of wetted area, specifically wetted area multiplied by the above friction coefficient—a reasonable model since aircraft, unlike cars, are dominated by friction drag rather than pressure drag (in aircraft with significant pressure drag, such as the C-130 with its upswept tail, we account for this by simply adding a pressure drag term—an example of Raymer's "combinations"). For a more refined prediction, we can do this component-by-component, building up a total drag coefficient based on the wetted area, friction coefficient, and interference factor for each part of the aircraft including fuselage, engine nacelles, wing and tail surfaces, and any external stores or ordnance.

Of course, where it becomes complex is in how all the aerodynamic phenomena you've heard of—trailing vortices, yawed flow, onset turbulence, transient (changing over time) flow conditions, wake behavior, etc.—affect body pressures and body shear stresses. All of them change the pressure and shear distribution, and as you gain more experience working with airflows you should start to develop a sense for how these attributes and flow structures alter aerodynamic forces (or result from the generation of aerodynamic force by pressure and shear stresses, a sort of "chicken or egg" situation). But as a home modifier, changing an existing car to achieve some goal such as reduced drag or reduced lift—usually in service of another goal such as improved lap times or better fuel economy—the simple fact is you don’t need to worry about how these flow features change pressure and shear distributions, only their effects (I’m serious: even if you have absolutely no knowledge of any of these mechanisms, you can still trial-and-error your way to effective modifications with a little time, a little thought, and a lot of measurement). And those effects are relatively easy to characterize, since there is one (cheap and easy) test type that will give you information about both: pressure measurement.

Don't misunderstand my argument: aerodynamic structures, such as the trailing vortices on this Camaro made visible in water spray, are absolutely important to understand, and there are techniques available for visualizing them or measuring their strength. But in terms of their effect, all you need to know are the pressure and shear distributions on the body; that's why they are important. (Image credit: The Drive).

Pressure
 
If you take a fluids engineering class, one of the concepts you will be introduced to is the control volume derivation of the governing equations (I find this finite volume method easier to understand than finite difference, which gives the differential form of these equations). Basically, by identifying a constant volume in space through which a fluid passes, you can apply known conditions of physics such as conservation of mass, momentum, and energy to describe changes to the fluid properties within that volume mathematically, even as that volume moves in space or deforms over time. For low-speed flows like the air moving around and through a car, conservation of momentum (which produces the Navier-Stokes equations) is a powerful tool because it directly relates to the force acting on a fluid and, by Newton’s 3^rd Law, the reaction force the fluid exerts on whatever body causes the fluid’s momentum change:

This reaction force is exerted on the body by—you can probably guess, right?—pressure and shear stresses (you can see these terms on the right-hand side of the equation above as forces due to pressure and friction; “body force” includes gravity and any electromagnetic forces, which are usually considered negligible in atmospheric airflows). Momentum has both magnitude (i.e. size or strength) and direction (encapsulated in the velocity terms above; remember I mentioned this would come back. Momentum always “points” in the same direction as velocity, and when the equation above is separated into its x, y, and z components so is momentum). Any time you change that magnitude or direction there will be changes in pressure and shear on the body that is causing the change in momentum. For example, a spoiler deflecting air upward, even slightly, will increase pressure in the flow upstream; the force exerted on the air to increase its upward velocity component (and thus, its upward momentum) results in an equal force pushing the car down (since momentum must be conserved), and that force is measurable partly as a change in pressure.
 
Consider a stream of air flowing along a flat surface such as this board taped to the roof of my car:

Bend that flow upward instead of letting it continue flat and it increases pressure along the length of the board. Remove the board to bend the flow in the other direction (which follows the curve of the roof due to the Coanda effect, the tendency of airflow to continue along a surface that deviates from its path) and it decreases pressure on the roof ahead.

No board.

Flat board.

Inclined board. All values given are static gauge pressure (i.e. difference from freestream), two-way average at 80 kph. With the board bent upward, pressure increases along its length. Both the flat board and bare roof with its downward curvature achieve the same pressure in the rearmost measurement location, but the pressure upstream and at the roof peak are more strongly negative in the unmodified configuration. In each case, the "bend" of the airflow changes the pressure distribution—a result of momentum transfer between air and car.

Since a real car must have a shape that grows from some small surface at the very front to an area which can encapsulate passengers and cargo before terminating at some point behind the cabin, forward-facing surfaces tend to be inclined upward relative to the freestream flow and rearward-facing surfaces tend to incline downward. Given the tendency of inclinations such as these to subsequently produce high pressures on forward-facing surfaces and low pressures on rearward-facing surfaces—exactly the opposite of what we want for low drag!—we need to change the surface shapes to try and alter these pressure characteristics. This has been happening for decades now on production cars as manufacturers chase lower drag and better fuel economy; pick just about any car model and you can see at a glance what changes have been made. For example, the Toyota Corolla used to look like this:

(Image credit: Toyota Canada).

Upright nose, flat hood, upright windshield, sharp corner from windshield to roof, upright backlight. Now contrast this with the current Corolla:

(Image credit: Toyota).

Note the rounding of the nose, relaxed hood to windshield angle, continuous curve in the windshield/roof/backlight line, and short, tapered trunk with a small spoiler. These surfaces have been engineered, as much as possible within the confines of the styling concept, to reduce pressures on forward-facing surfaces, increase them on rearward-facing surfaces, and avoid suction peaks (for reduced lift) or strong pressure increases (which can cause the flow to detach and increase pressure drag; more on that below). The results are evident in the drag coefficient of each of these: 0.36 for the 1988 model reduced to 0.29 for the 2025. (Note that the size of the car grew several inches in each direction, however, and that drag area—the true measure of aerodynamic efficiency—only fell from 7.23 ft² to 6.78 ft²).
 
To reduce the pressure drag on your car, which is the major component of its profile drag, you can take a cue from OEMs and make the same sorts of shape changes: round front surfaces, avoiding sharp edges as much as possible; gently taper rear surfaces, avoiding premature separation; fit separation edges or spoilers at trailing edges to avoid suction peaks and increase upstream static pressure; fit smooth panels under the car with a gently sloped diffuser, etc. But above all, don’t simply copy an OEM shape or device (or a "template" you may encounter online)—test candidate shapes on your car to ascertain their effectiveness in meeting your requirements by measuring the pressure changes that result from alterations to body shape.
 
Shear
 
Shear, or the tangential component of stress on the car body, may be possible to measure directly on a real car with a hot film sensor (although commercial sensors are ridiculously expensive)—I haven't tried this myself but plan to in future. However, we can also deduce information about shear stress from pressure measurement.

 
Shear stress arises from the fact that fluid moving past a surface, like air past the body of a car, effectively "sticks" to that surface (on a molecular level, this isn't true but on a continuum level it is). This is a phenomenon we call the "no-slip condition," and it is a characteristic of all bodies in contact with any fluid flows (of sufficient scale and/or density to satisfy the continuum approximation): submarines and boats through water, cars through air, wastewater through your home’s drainpipes, supersonic fighter jets in the atmosphere, a spoon through honey, etc.
 
Because the air "sticks" to your car’s surface while moving at a different speed further out from the body, a layer of slower-moving air called the boundary layer develops, in which the velocity increases with distance from the surface; outside this layer, the flow can be approximated as inviscid but within it viscous effects dominate the fluid's behavior. Because all fluids have viscosity, a tendency for their molecules to also "stick" together due to intermolecular forces, this means your car must "pull" on the air in order to move through it, and that the air subsequently "pulls" back. This pull exerts a force on the surface tangential to that surface: shear. Shear is proportional to the velocity derivative with respect to distance from the body, evaluated at y = 0; that is, at the body surface, the rate at which the velocity increases as you move away from the body determines the shear stress exerted there. Mathematically:

...where tau is shear stress, mu is the fluid's viscosity, and du/dy is how much the velocity increases with height (I've illustrated this in the figure below). There are two types of boundary layer (broadly; because the magnitude of turbulence can vary, these are not discrete and there can be a transitional boundary layer over a wide range of Reynolds number): laminar, in which the velocity profile is smooth and steady, and turbulent, in which the velocity profile is chaotic. On average, the velocity profile of each type looks like this:

Nicolai, L. and Carichner, G. Fundamentals of Aircraft and Airship Design, Vol. 1, Fig. 2.5 (emended).

You can see from the above that a laminar boundary layer profile—in which the increase of velocity with height is smaller than in the turbulent profile with its larger du near the body surface (for the same boundary layer height in both cases)—will exert less shear stress on the body. Further, shear stress can be reduced to 0 by separating the boundary layer from the surface, as shown on the right of the figure above. And finally, shear stress can be minimized by slowing the external flow and reducing the slope of the velocity gradient at the surface.

 
From this we can conclude that to minimize shear stress (and thus, the force that arises from shear stress) we should:
 
1) Maintain a laminar boundary layer as far back on the car as possible.
 
2) Induce flow separation.

3) Slow the local flow to a speed less than freestream.
 
Let’s look at these one at a time, and why you may or may not want to encourage them. First, to maintain a laminar boundary layer will only be possible on the forward part of the car. As flow passes further along the car its Reynolds number (ratio of inertial force to viscous force) increases, and at some point, the increase in inertial force results in instabilities that produce turbulence. There’s nothing you can do about that except slow down drastically or shrink your car, neither of which will make it very useful to you. But you can try to avoid "tripping" the flow prematurely into turbulence by keeping surfaces at the front smooth: no sharp corners, no body seams or gaps, no dirt buildup, no rough textures, no edges (even stickers and decals). It’s difficult but not impossible. See Goro Tamai’s The Leading Edge on using flow visualization paint to determine if you have any laminar flow over your car and, if so, how much.* Just a warning: this is probably not worth even investigating on a road car, as you will need absolutely smooth surfaces free of even the smallest bumps and imperfections and will, at a minimum, require complete replacement or reshaping of the front bumper cover. But you can still try if you have a mind to.
 
(*The FSAE team here told me that they’ve successfully used a dilution of Tide laundry detergent in water for flow visualization but I haven’t tried it myself. Poke around online and in books to find other options, most of which are oil-based).

Second, causing the boundary layer to separate may reduce shear drag but it comes with a huge caveat. You might think that if we slow down the flow somewhere on the car’s body to exactly match the vehicle speed—0, if you consider a stationary car in a wind tunnel—then we will eliminate any velocity gradient in the boundary layer at that point and consequently eliminate shear stress. This is the definition of stagnation—temperature, pressure, and other properties of the fluid at (isentropically-decelerated) rest relative to an inertial reference frame—and typically happens at the very front of the car over a small area (which you can measure and visualize with tufts).

Without the splitter, the streamlines diverge from about the top of the license plate to the bottom of the grill, where you can see tufts pointing down and wrapping under the car; this is the stagnation area (the grill opening has been positioned here to take advantage of the high pressure). With the splitter in place, the upper line of the stagnation area is moved down to below the license plate—as shown by the tuft directions in each image.

Outside of this stagnation area and over the rest of the car, however, as the inviscid flow slows and the fluid gains pressure, the lower part of the boundary layer close to the surface can actually reverse direction and start to flow backward. This can happen if the pressure rises too quickly in the direction of flow (we call this an adverse pressure gradient; air "wants" to flow from high to low pressure, so forcing it to go from low to high pressure is against its nature) and when it does, the boundary layer separates from the surface and, consequently, the average velocity derivative at the surface is zero. Over the rear part of a car where surfaces tend to taper inward (such as over a steeply raked backlight) or front part of a car where surfaces tend to be angled outward (such as at the base of a steeply raked windshield), this separation can increase pressure drag—which is why we typically don't want airflow to separate despite the beneficial reduction in shear stress.

(Note the qualifier above; not all separation is created equal, and sometimes causing airflow to separate somewhere on the car can result in less drag, less lift, or both compared to allowing it to stay attached to the surface. A classic example of this is the first generation Audi TT: as initially built, the rounded rear end of the car created enough lift to cause several high-speed crashes, one of which killed a professional driver. A recall was issued to fit the cars with rear spoilers; causing the air to separate in the middle of the trunk lid disrupted the fast-moving, low pressure flow there, reducing the dangerous lift the car produced before).

 
So, what we actually need to do is maintain a favorable pressure gradient over as much of the front of the car as possible (and keep it laminar if you can), as well as slow down the inviscid flow (that is, the flow at the upper edge of the boundary layer) as much as possible over the rear part of the car while keeping it attached to the car’s surface by avoiding too strong a pressure gradient. You can check this by taping tufts to the surface and measuring pressures. Because the boundary layer is isobaric (i.e. pressure does not vary through it), slower inviscid flow at the boundary layer's upper edge brings a commensurate increase in static pressure according to Bernoulli's principle.

Even the flat hood and fairly steep windscreen of my truck aren't enough to separate the flow more than a little just around the wipers. Here, I have artificially induced separation by fitting a mock bug deflector (right).

Taping this spoiler in place at the leading edge of the backlight separates flow over the window, dropping pressure down the centerline and increasing pressure drag there. I learned this by first taping tufts to the window and then measuring static gauge pressure over it with and without the spoiler in place.

Third, over the rear surfaces of a car, gradually decelerating the flow (thereby reducing its dynamic pressure and, since stagnation pressure is approximately constant for constant vehicle speed, increasing its static pressure) while keeping it attached to the surface may reduce friction drag. This qualitative relationship between adverse pressure gradient and friction coefficient has to do with the displacement of air upward in the boundary layer (we nondimensionalize this into a single "shape factor" called H which can predict separation—the last plot in the figure below); if the gradient is too strong, the boundary layer may separate.

Boundary layer model equation numerical solutions, Thwaites' Method and Head's Method. Notice the jump in local friction coefficient, c_f (fourth chart), where the flow transitions from laminar to turbulent.

The trick here is introducing an adverse pressure gradient but one that isn't too strong. If you accidentally induce separation, friction drag will be at its lowest but pressure drag could increase a lot. Keeping the boundary layer attached compared to letting it separate can greatly reduce pressure drag. If faced with this choice—reducing friction drag at the expense of increasing pressure drag vs. reducing pressure drag at the expense of increasing friction drag—it is almost always better to go for a pressure drag reduction, since pressure drag is responsible for ~80% or more of the aerodynamic drag force acting on your car.

Guidelines
 
Condensing all the above, you can see that some general guidelines follow. First, if you’re only going to invest in and use one test method to measure airflow properties over your car, it should be pressure measurement. That on its own will tell you more about your car’s aerodynamics than any other single test. Second, trial modifications to your car that:
 
1) Smooth the front surfaces to preserve laminar flow as much as possible.

Not much chance of it here. Even something as small as a bug splat can disrupt laminar flow. The grill openings, engine air intake duct, lights, seams, license plate, emblem, and surface debris all prevent that here.

2) Gently curve front surfaces to encourage low or negative pressures as much as possible.

Centerline pressures measured on my car; most locations show negative gauge pressure, a result of the gentle curvature of the hood and windscreen. Later measurement of pressures side-to-side across the windshield gave me a better idea of the extent of negative gauge pressures contributing to thrust.

3) Gently taper rear surfaces to keep flow attached while increasing pressure as much as possible.

The sloped tail extension here has been designed to have attached flow across its entire upper surface; pressure measurement showed an increase in static pressure toward the rear of the board compared to the front. Pressure on the trunk (base) under the board increased as well, further reducing pressure drag.

4) Reverse the taper or add a small spoiler at trailing edges to increase pressure upstream over rear-facing surfaces.

With the spoiler "reflex" attached, pressure increase was significantly higher than without it, achieving positive gauge pressure at the rear of the board.

5) Add smooth surfaces under the car with an upward-sloped diffuser panel at the rear.

Adding a smooth panel over the rough factory undertray reduced pressures from just above atmospheric to just under, decreasing front lift.

Measured pressure on the upward-sloped diffuser panel increases front to back (trading kinetic energy in the flow for pressure as the air decelerates), matching base pressure at its trailing edge.

Of course, ultimately you should not limit yourself to one test method! If you were able to measure the pressure distribution over your entire car (that is, every single point on its surface) or, better yet, log it over time, there would be little need for other measurement methods. Since this is not possible, other techniques such as throttle-stop testing, ride height measurement, flow visualization, vortex strength measurement, coast down testing, etc. become useful for quantifying the aerodynamic effect of geometry changes to your car. But as a starting point, begin with pressure measurement. The necessary equipment is inexpensive, and the measurements are illuminating and will tell you a lot about your car’s aerodynamic characteristics. Use those measurements to make effective, real changes that will reduce your car’s drag and lift.

Conclusion
 
Finally, compare these guidelines—which we have derived from fundamental principles and observations—to typical suggestions for good aerodynamic design in the literature. For example, here is Barnard’s list in Road Vehicle Aerodynamic Design, 3^rd ed.:
 
1. Smooth unbroken contours with favorable pressure gradients as far back as practical should be used.
2. Strongly unfavorable pressure gradients at the rear should be avoided; some taper and rear-end rounding should be used.
3. The form should produce negligible lift in ground proximity.
4. If a hatchback configuration is required, the backlight angle should not be in the region of 30°, and if a notchback is to be used, the effective slope angle Θ_eff should not be in this region.
5. The underbody should be as smooth and continuous as possible, and should sweep up slightly at the rear.
6. There should be no sharp angles (except where this is necessary to avoid crosswind instability).
7. The front end should start at a low stagnation line, and curve up in a continuous line.
8. The front screen should be raked as much as is practical.
9. All body panel lines should have minimal gap.
10. Glazing should be flush with the surface as much as possible.
11. All details such as door handles should be smoothly integrated within the contours.
12. Excrescences should be avoided as far as possible; windscreen wipers should park out of the airflow.
13. Minor items such as wheel trims and wing mirrors should be optimized using wind-tunnel testing.
14. The cooling system needs to be designed for low drag.
 

Books on vehicle aerodynamics are usually aimed at engineers and designers tasked with building a new car where they have some influence over things like door handle design and radiator placement; as home modifiers, these decisions are typically outside our purview. But otherwise, there is a lot of overlap between our derived guidelines and this list from an experienced aerodynamicist. That should not be surprising, as all this advice derives from the same fundamental physics. Go get yourself a manometer, some pressure patches, and have at it; you can explore the physics yourself!

You don't need a wind tunnel, of course, but sometimes we measure static pressures in the laboratory too. Here, we calculated lift coefficient as a function of angle of attack using measured pressure distribution. Just as on the road, static pressure on the airfoil is measured relative to average static pressure from a "ring" of taps upstream of the test section. Look closely and you will see the pressure taps in the surface of this modified Clark Y-14 airfoil.