How to Measure Changes in Drag

For a lot of people interested in modifying their car’s aerodynamics, drag reduction is the primary goal (perhaps more so now that gas prices have shot up).

In a bit of good news, my license plate is relevant again.

Drag is any force that resists the car’s motion along its longitudinal axis i.e. an imaginary line that runs from back to front along the centerline. Mechanical drag is, as its name suggests, produced by the inertia of and friction between the various mechanical components of the car and/or the road. Aerodynamic drag is the resistance generated by air as it flows over, under, and around your car.

Because of this focus on aerodynamic drag reduction, measuring changes in drag becomes an important task—perhaps THE most important in any home modifier’s development program. Fortunately, you don’t need access to a wind tunnel or a supercomputer running CFD simulations to measure drag. There are some simple test methods you can use on the road to measure the drag changes which result from modifications.


First, we’ll need to temper our expectations. Don’t think you’re going to be able to conduct some on-road testing and then calculate a drag coefficient to four decimal places. That sort of sensitivity depends on extremely accurate measurements—of both the forces acting on the car and parameters such as its frontal area—that aren’t realistic for us at home (in fact, this is actually outside the sensitivity of even the most advanced wind tunnels, which can reliably calculate CD only to about +/- 0.001). Some aerodynamic parameters, such as the rate of change of yawing moment, are practically impossible to measure outside of a wind tunnel, so you should restrict yourself to things that are actually measurable like drag and lift. Finally, because of the natural variability of testing on the road, there will be some scatter in results. Be as consistent as possible and you can minimize those errors but realize you won’t be able to directly measure small changes in drag e.g. from cracking the windows or removing the antenna. So, when you test go big—try a tall spoiler rather than a short one, or a deep air dam rather than a small one. Larger changes will likely show up in test results where small ones may not.

Test Techniques

Now let’s look at some specific techniques: constant-throttle testing, constant-speed testing, and coast down testing. There are benefits and drawbacks to each.

Constant-Throttle Testing

This technique was developed by Julian Edgar, former editor of Autospeed and author of Modifying the Aerodynamics of Your Road Car and Car Aerodynamic Testing for Road and Track
. I highly recommend both books.

Basically, this technique limits engine power so that your car has a “new” top speed, which is determined by the drag force acting against it. Place a “stop” (a wood block, metal bracket, or similar—you can go as basic or fancy as you want) under the throttle pedal or on the throttle body to limit its travel so that your car tops out at a sane speed, say 60 or 65 mph. Make a change to the body/aerodynamics and test it again, with the throttle stop still in place. If the speed goes up, drag has been reduced. If it goes down, drag has increased. And the percentage change in drag is proportional to the ratio of the squared speeds, i.e.

For example, say you get a base speed of 109 kph, make a change, and get a new speed of 112 kph. Your change in drag is

or a 5.58% reduction in drag (I’ll cover significant figures and how to use them in a future post; for now, just know that your percentage can only have as many numbers in it as your smallest speed measurement).

Benefits: This is perhaps the easiest and most repeatable method for measuring drag changes. Because you’re fixing the throttle position, there’s less chance for error than, say, trying to maintain a constant speed. You will need some sort of throttle position display, whether on a monitor connected to your car’s OBD port or hard-wired in to ensure that throttle position is constant, but other than that there’s not much to it.
Drawbacks: Because this test involves a change in speed, there is some error associated with it. Mechanical drag can change with speed and engine torque output may vary as well. But the changes in these should be very small, likely much smaller than the change in drag you’re trying to measure (another reason to go big).
The other drawback has to do with your car: this doesn’t work on cars with electronically-controlled throttles (where the computer can vary the throttle opening regardless of throttle pedal position), or on hybrids which can add torque from an electric motor based on what speed the ECU predicts the driver wants. So, if you have one of these cars (like my Prius, unfortunately—it’s a hybrid with ECT), this probably won’t work. However, some drivers of EVs have reported successfully using the throttle-stop method to measure changes in drag.
Constant-Speed Testing
Alternatively, we can keep the speed constant and measure something that tells us how much power the engine is making to see if that goes up or down. Possibilities include measuring MAF sensor voltage, throttle position sensor voltage/TPS readout from the car’s computer via the OBD port, or possibly even engine load or calculated horsepower from a computer such as a Scangauge.
Since this involves keeping to a set speed, it will probably be easier with cruise control—but be careful because even cruise control systems can allow some variation in speed. On a flat stretch of road, drive at a constant speed and note the MAF sensor voltage, throttle position, load, or horsepower. Then make a change and drive the same road again at exactly the same speed. If the parameter decreased, drag has gone down; if it went up, drag has increased.
Benefits: The main benefit of this technique as opposed to constant-throttle testing is the removal of the change in speed, which means that mechanical drag and engine RPM are now constant (assuming you have a transmission with fixed gear ratios, not a CVT).
Drawbacks: There are a few drawbacks, however. No matter how conscientious you are about it, there’s always the possibility that speed will vary slightly over the test. This method also depends on the sensitivity of your MAF sensor or TPS, so small changes in drag may not show up. The problems with electronically-controlled throttles and hybrid systems apply here too. And crucially, even if you have a combustion-engine car with fixed gear ratios and a mechanical throttle connection you won’t be able to easily calculate the percentage change in drag like you can with constant-throttle testing.
Coast Down Testing
Finally, coast down testing is one of the oldest (and most problematic) methods for measuring drag changes.
Most people are familiar with this one: accelerate your car to a certain speed, then put the transmission in neutral and let it coast, measuring the speed at regular time intervals. Graph the speed as a function of time and you should get a decreasing parabolic curve.
Benefits: The most obvious benefit of this method is the removal of engine power as a variable. Without that, you’re really just looking at an acceleration determined by the force (combined mechanical and aerodynamic drag) acting against the car’s motion. To ensure that aerodynamic drag predominates, you have to test at very high speeds; too low and aerodynamic drag has little influence on the car’s deceleration.
Drawbacks: In reality, this method doesn’t yield consistent results on the road; a single errant gust of wind can screw up your results for an entire run. I’ve tried it on three different occasions—taking care to minimize error as much as possible by testing in the middle of the night with no traffic and low wind, on the same stretch of flat road—and never got reliable data. If you can get it to work, more power to you. But I see little reason to attempt it again when methods like throttle-stop testing are easy to do, more reliable, and take far less time.
No matter which method you decide to try, it’s a good idea to check it against a known change in drag. The easiest is to simply roll down all the windows, which usually increases drag by a few percent. Run windows up against windows down and that change should show up in your results. If it doesn’t, ensure you’re minimizing potential sources of error and try it again, or use a different method.
I should also note that there are other ways to quantify changes in drag, from measuring fuel economy to using sensors that attach to a driveshaft and measure torque directly (this method was used by Ford as early as the 1960s!). However, the former is subject to wild inaccuracy and the latter is very expensive. Still, the sky is the limit—use your imagination, and you might be able to come up with another way of measuring drag.