Don’t Guess—Test!

My truck, wearing its first, tested aerodynamic modifications.

It kills me sometimes to see people going on forums and asking questions about how to aerodynamically modify their cars. Why? Because almost inevitably they get a stream of answers that state, quite confidently and often unequivocally, that such-and-such a change will lower drag by x amount, or that a specific modification will always reduce drag, or what the dimensions of a modification should be for maximum, “guaranteed” drag reduction.
 
Often the people answering these sorts of questions will point to research on simple bodies (these are frequently used to justify guidelines on tapering of boat tails, for instance) or papers on the development of production cars (data on cooling system drag or optimum backlight angle, for example) or studies of modifications on production cars (the most famous is a 2012 investigation of changes to an Audi A2 run by Tata Motors, Jaguar Land Rover, Warwick University, and others) to back up their assertions.
 
The problem with this approach—taking results from one car or simple body and applying them to another to make predictions—is that it has limited uses; in real life you can’t take data from one car and apply it wholesale to another, expecting the same change to have the same results on a different car.
 
Why can’t we compare or predict results on one car based on a similar parameter of or change to another? Because—believe it or not—air flows differently around objects of different shapes. This is such a fundamental concept to grasp that it should go without saying, yet when it comes to aerodynamics, lots and lots of people fail to understand this.
 
And not just cars of markedly different shape and size, either. As far back as 1976, Wolf Heinrich Hucho noticed that “despite the similarities of shape, the variance of drag among cars is remarkable.” Details, sometimes small details, are enough to completely change the flow characteristics over a given car. Because of this, you cannot apply measured results of a change on one car wholesale to another or predict what a change will do to one car based on another, and you cannot intuit the flow characteristics over a car with and without a modification.
 
Predictions
 
This is not to say that tentative predictions aren’t useful. If we can’t guess at something, how will we determine what devices to try out on our cars? But these predictions have to be based on lots of data of similar changes to other cars and lots of experience reading and testing. Even then, predictions must be verified.
 
A good example of this is that Audi A2 I mentioned above. JP Howell wrote a paper on the research project (which you can find in this book), which basically took a stock A2 and modified it by blocking cooling airflow, removing mirrors, lowering ride height, etc. to see how much they could reduce its drag (this investigation was intended to inform Tata Motors’ then-ongoing development of its X-Prize car and was conducted as part of the Low Carbon Vehicle Project, a research consortium of manufacturers and universities). The A2 was first tested in the stationary-ground MIRA wind tunnel in the UK and then in the moving-ground S2A tunnel in France.
 
Removing the mirrors reduced the drag coefficient by 10 counts (about 4%) in both tunnels. But this exact change depends on the particular shape of the A2’s front end, its cooling airflow, the angle of its windshield, the rounding of its A-pillars, the shape of its hood, the paneling under the engine, the shape and size of the mirrors that had been removed—and many more parameters. You simply can’t take those results and declare that removing the mirrors on any car will reduce drag by 4% or by 10 counts because other cars are shaped differently. You can predict that removing the mirrors on your car will reduce its drag, but then you must test to verify or disprove that prediction and measure if you want to know how big the change is. You can predict that tapering a boat tail to 22° from horizontal will reduce drag more than 10° or 15° or 20° or 30°, but then you must test to verify or disprove that prediction and measure if you want to know how much the drag changes with the taper angle and what the best angle is for your car. You can predict that blocking part of your car’s grill will reduce its drag, but then you must test to verify or disprove that prediction and measure if you want to know how much it changes drag. You can predict that adding a rear spoiler lip will increase pressure on the bodywork ahead of it, but then you must test to verify or disprove that prediction and measure if you want to know how much the pressure changes and where.
 
Testing
 
Testing reveals what is actually happening with the air around your car. Without it, you are shooting in the dark. That’s fine to begin with—“Say, what happens if I block off part of the cooling air intake at the front of my car?”—but it’s only a starting point. You must test if you want to know.
 
Again, the A2 project provides a good example of this. Decades of research have shown that lowering a car usually reduces its drag. Howell tried this on the A2, in both wind tunnels, expecting to see a drag decrease as the body was lowered. However, as he wrote in the paper, “One unusual feature of the A2 was the negligible drag reduction resulting from reduced ground clearance. There is an almost universal ‘rule of thumb’ which states that drag coefficient increases by ΔCD = 0.004 for every 10mm of ride height. This rule did not apply to the A2.”
 
Take a lesson from that: even a career aerodynamicist with years of experience can be surprised by a particular car’s shape and how the air flows around it, making predictions that are shown by testing to be wrong.
 
One of These Things Is Not Like the Other
 
Here’s a good real-world example. In other posts on this website I have put up these images of the flow over the side window of my 1991 Toyota Hilux:


…compared with the flow over the side window of my 2013 Prius:


Both cars have hoods, windshields, and doors, etc.—of course—and were tested with no mirrors. Yet the flow over the windows is very different. Differences in the vehicle shapes, A-pillar rounding (the truck actually has rounder pillars; the Prius has a sharp edge running along the pillar), windshield angle (~40° from horizontal on the Prius, ~60° on the truck), hood curvature, windshield curvature (again, perhaps surprisingly, the Prius’ windshield is flatter than the truck’s), wheel/tire size, wheel arch opening, ride height, door shape, cooling air inlet size and location—all these things are different between the two vehicles. Looked at this way, it’s incredible that anyone would expect that the same change to both cars would have anywhere close to the same effect on the airflow over them.

Horizontal wire traces of the A-pillar shapes of the truck (left) and Prius (right). The windshield is at the top and side window at the bottom.

Okay, but the Prius and truck are very different shapes. What about similar shapes? What about, say, two trucks?
 
To answer that, let’s compare my truck to a friend’s 2011 Toyota Tacoma. Not only is this a truck of similar shape and size, it’s built by the same company, is the successor to my older truck (one generation removed and fattened up for the US market), and may have even been tested in the same wind tunnel when it was designed. How similar are the flows over the side windows of each truck?

Answer: not very. Less disruption from the mirror, less separation at the A-pillar, and less turbulence in general can be seen on the newer truck, tested at the same speed, on the same road, and in similar conditions.
Seemingly inconsequential differences in the shape of each truck’s hood, wheel arches, windshield, A-pillar, roof, door, mirror, and window add up to a substantial difference in the character of the airflow over them. The flow over the side window of the 2011 Tacoma is cleaner with a mirror than the 1991 Hilux without one. Only twenty years separate these two trucks, and to many people they look pretty much the same!


Take this to heart. Stop guessing. Start testing.

Comments

Popular Posts

Coastdown Testing Revisited

A Practical Guide to Aerodynamic Modification

Tuft Testing: A How-To Manual