Optimizing a Tail for Low Drag: Part 3

Initial Testing

To start the process of designing a drag-reducing tail, I threw together some boards that would give me some adjustability of their angle relative to the rearmost surfaces of my car. Then I headed out on the road for initial testing.


Before you start testing, it’s important to recognize potential shortcomings and what your tests can actually show you (as well as what they can’t). In this case, separate boards at the top, bottom, and side of my car do not replicate a complete, solid tail; rather, I’m using this test to get an idea of what might be appropriate dimensions and taper angles to start my investigations of a full tail. I’ll use these data to try and predict the drag changes from the various angles and then use those as a jumping off point to design the real tail rather than just guess at a shape.
 
You will see people misunderstand this all the time online. Commonly, someone will make a change to their car and then use one tank’s measured gas mileage as “proof” it worked. Don’t fall for this! Often, the claims made from this sort of “test” are unbelievably large in comparison to the change made, and the test can’t actually show what the tester thinks it can. For instance, before I start in on this tail I added outboard diffuser extensions to my car (outermost boards/vanes here):


The first tank with these on returned a full 5 mpg higher than the previous tank—a nearly 10% increase. Success, right? Wrong! Attributing that large a change in gas mileage to this small (and as yet unknown) change to the car’s aerodynamics is bananas. It also does not account for the fact that the previous tank was almost entirely short trips to and from school and work (a distance of approximately 3 miles), whereas this last tank included 200+ miles of freeway driving. Of course the car got better mileage: the driving profile was completely different and better suited to economical driving.
 
Keeping in mind what I want to learn from these tests and how I want to use that information, I headed out on the road.
 
Pressure Measurement
 
First up, I measured aerodynamic pressure on each board toward its trailing edge as I varied its orientation. The way these boards are cobbled together didn’t really allow for fine adjustments, so the angle changes ended up being quite large—but the results were still informative.
 
I tested on a N-S road on a day with winds out of the SSW at ~10 mph. So, the yaw angle and freestream speed are different for each direction here—differences that show up in the results.
 
Top board:
 

Angle from Horizontal

South (Pa)

North (Pa)

Average (Pa)

-13°

0

-10

-10

-24°

-10

0

-10

-34°

-20

0

-10

 
Bottom board:
 

Angle from Horizontal

South (Pa)

North (Pa)

Average (Pa)

+2°

-40

-10

-30

+16°

-30

-20

-30

+34°

-40

-10

-30

+44°

-40

-10

-30

 
Side board:
 

Angle from Centerline

South (Pa)

North (Pa)

Average (Pa)

-30

-40

-40

-5°

-40

-20

-30

-10°

-60

-20

-40

-18°

-20

-40

-30

-25°

-40

-40

-40

-31°

-40

-40

-40

 
One curiosity: the spike in negative pressure at 10° inward slant (where 5° is approximately aligned with the bumper cover side). Concerned this might be an error, I ran that test again in both directions and got the same result. That sticks out as an angle I probably want to avoid in my tail design.
 
In addition to the above, I measured the base pressure on the bare car (in the center of the license plate) and found that it was -40 Pa southbound and 0 Pa northbound:
 
You would be surprised how many people ask me what OPEC means.

Comparing this to the measured pressures on the boards in each direction, this means that the bottom and side boards were essentially supporting base pressure in all positions—which might indicate that they are in separated flow regardless of angle. The top board, on the other hand, developed higher than base pressure southbound, even up to a steep angle.
 
In the meantime, I can conclude the following from these tests:
-drag reduction from pressure recovery on this tail will likely come from its upper surface more than from the bottom or sides
-any pressure recovery is likely to be a lot smaller than “theory” suggests

Unfortunately, without more pressure recovery this tail won’t even come close to my drag reduction goal. If the full mockup doesn’t develop higher pressures, I’ll have to investigate shape changes to try and increase them.
 
Tuft Testing
 
Second, I went out on another day and had a friend drive my car while I took photographs of tufts taped to the top, bottom, and side boards at various angles.
 
There was a fairly gusty wind that day, 15-20 mph out of the NW. Despite the yawed flow, the tufts showed attached flow in more configurations than I expected.
 
I couldn’t see until I reviewed the burst images at home later, but the bottom board had attached flow in its lowest position (it sagged a bit during testing). That flow separated if the board was angled upward at all:
 


The side board, which I had suspected was in separated flow in all positions, turned out to have attached flow even in the crosswind, in both directions, if the board was colinear with the body side (which is tapered inward about 5°). Slanting the board further toward the centerline of the car, the tufts started to show separation:
 



On the top, parallel(ish) to the rear window the board showed attached flow. Angling it further down (where my camera refused to focus; you can just barely make out the tufts in the image), it still had attached flow on the verge of separation, and further still resulted in completely detached flow:
 



I had debated waiting for a less windy day but decided to go ahead since it’s harder to coordinate testing with two people’s schedules. This turned out to be fine, as these conditions are more representative of typical use and it is perhaps more important that my final tail design perform well in conditions such as this (gusty crosswinds) than calm, perfect days—which are a rarity. Ultimately, reality—the real car, in the real world, in real conditions (which are almost never “perfect”)—must be the arbiter of a design: it doesn’t matter how a specific shape performs in CFD or a wind tunnel if those don’t reflect real conditions. And aerodynamic designs must be tested in the real world. Current CFD can only fully simulate very low speed, geometrically simple flows (and it takes a supercomputer to do even that); all other CFD car simulations, even those by OEMs and especially the generic ones you’ll see on Youtube videos and blogs, depend on (sometimes massive) assumptions and approximations that remove them several steps from reality. Ultimately, it does not matter if your design is perfectly optimized in a digital environment if it turns out not to perform the same way in the real world. So, go ahead and test on windy days.
 
Analysis
 
Now that I have some data to work with, I can try and predict how a tail will perform and use that as a starting point for my next test buck. To do this, I wrote a Python program to calculate total base pressure drag as a function of top board angle, top board measured pressure, bottom board angle, bottom board measured pressure, side board angle, and side board measured pressure (approximating pressure as constant over the entire surface area of each part of the tail). Pressure drag is the sum of each pressure multiplied by the surface area of each part normal to the x-direction i.e. in the yz-plane, giving the x-component of the force developed by the pressures acting on the slanted areas (I won’t bother here, but you can prove this relationship with simple trigonometry). Each of those areas changes, as well as the remaining vertical base area, with change in the boards’ orientations, and of course the measured pressures change as well—which means this is a lot easier to let a computer sort out and spit out an answer.
 

The greatest drag reduction predicted by this algorithm is about 3% better than the car with no tail—nowhere near my 15% goal. To achieve that, I will have to essentially eliminate base pressure drag. So, for the next mockup I’ll test some geometries, such as panel curvature, that might help to increase pressure over a tail extension.
 
Keep in mind, too, that these are only general estimates; I’ve made a lot of assumptions (and ignored things like shear stress drag) to try and roughly predict what a real tail will do. But that’s fine; I’ve only begun testing at this point, and as I continue to test and measure, I will refine my parameters for a final design. These initial tests are just a starting point.

Comments

Popular Posts

How Spoilers Work

Optimizing Aerodynamics of a Truck: Part 5

A Practical Guide to Aerodynamic Modification