DIY Rocker Vents

In a previous article I covered Miata fender vents in some detail, including commercially available louvers and DIY solutions. I cited Race Louvers wind tunnel data, which found that after a certain size, more vents don’t equal more downforce. What I didn’t mention was the reason for those diminishing returns.

All of this venting is overkill unless you can reach further inside the wheel well.

Take a look at a Miata unibody with the fenders removed, and you can see the problem. At the top of the fender, vents can reach just about the width of the tire, but the air has to travel all the way up there in order to be extracted. A more direct route is behind the tire, but the chassis is in the way! Moreover, air needs to take a 90-degree turn to make its way out of the wheel well. On a Miata, there’s a sharp lip here where two seams come together, which makes it even harder for air to turn the corner.

Naked Miata shows the crux of the problem.

In order to extract more air, the vents need to reach further inside the wheel well. Ideally the vent should be located low and behind the tire, because that’s the most direct exit route from the trailing edge of the splitter. And while doing that, you’d also design a much smoother exit. The end result would be quite different than the haphazard dumping and venting of air into the wheel well, and more of a purposeful and free-flowing duct from splitter diffuser to behind the wheel. Now these aren’t new ideas; you can see all of that on a McLaren M8 from half a century ago.

McLaren M8 had big vents and smooth exits back in 1971.

A more subtle solution can be seen on the underside of current NASCAR stock cars. Notice the air channels behind the tire that exit at the front of the rocker panel. This allows the cars to look stock-ish, but they perform much better and make a lot of downforce. In fact too much downforce, because NASCAR had to reduce the effectiveness by adding splitter stuffers.

NASCAR underbody aero was so effective they required splitter stuffers (the black horizontal bars in the splitter diffusers) to *reduce* front downforce.

I don’t know if these vents have a proper name, but I’m going to call them rocker vents based on their location in the rocker panels. They are a very Occam’s Razor solution to extracting air from under the splitter; of the myriad ways you might solve the problem, rocker vents are the simplest. Well, at least from the perspective of air, it’s the most direct route.

I wanted to implement something like the NASCAR rocker vents on Falconet, but hadn’t seen anyone else modify their Miata in such a manner. AJ Hartman did something like this on his Mustang, and said it was very efficient, with a L/D of 14:1.

So I knew it could be done, I just didn’t know how to do this on a Miata. I did a lot of exploratory cutting on the driver’s side, and the pictures show how I did a much cleaner job on the passenger side.

I want to thank Aussie reader/supporter Jim Grant for assistance with this project. He recognized I had no idea what I was doing and explained how cars are put together, and how to get my head out of my ass. Thanks Jim!

Step 1: Measure

My first decision was how large (height and depth) to make the vent. There are a few considerations for height, see the following image with colored lines.

The white line is the most conservative cut, because it goes just below the curved wheel tub inside the wheel well. Cutting at this height leaves two threaded holes intact, in case you want to add bolt-on fender braces.
The yellow line is about where the mid-body style line is on a NA Miata. Aesthetically, this is probably the best place to cut a NA Miata, but on a NB it wouldn’t matter. This cut goes between the two bolt holes, and so you could still use the upper hole to mount something.
The red line is to the top of the wheel tub. You could go higher than the red line, but most of the air you’re trying to extract is below the splitter, so I’m not sure if a higher cut helps that much. Might look cool tho.

Bring that rocker vent all the way to the top and it might look like a RX-7 GTO. Fucking badass.

In retrospect, if I did this again, I’d cut on the yellow line. Mostly because I think it would look better to have the duct exit on the style line. I don’t know that there’s a lot more air to extract at that height, and it could be that a vent even half this high is just fine (the NASCAR vents are quite low). In any case, I used the white line.

The next decision is how deep to make the cut. I measured and marked a line 9.5” from the inside of the wheel tub. This is where the inner sheet metal is located within the footwell, and so the cockpit remains unmolested.

I removed the white crosshatch area, it starts about 9.5” from the inner wheel well.

The rocker vent would be more effective if I had cut further inside behind the wheels, but then I’d need to weld sheet metal inside the cabin. Going further than that, there’s other unibody structure and cage tubing that would be difficult to work around. If I’d kept going deeper, I’d hit the clutch dead pedal, which could be sacrificed I suppose, but I do use it.

All said, I chose to make the rocker vent only about 3” (75mm) deep. This isn’t a lot of extra volume, but it does allow air to exit at less of an angle.

I used a sharpie to draw a line back to the hinge at my desired height. I’m not going to use the lower hinge (my doors are only half height), but there’s a vertical chassis member here and I wanted to keep that intact. So at this point, I have defined the area I want to remove.

Take a minute to pause and question if you’re really going to do this.

Step 2: Surgery

I used a grinder with a cutoff wheel to cut along the lines. There’s a horizontal shelf inside that has to be removed. It’s spot welded in various places (you can see the dots) and I tried drilling those out and in some cases getting in between with a cold chisel and breaking the welds. This little shelf is a pain in the ass.

Cut an opening so that you can remove the shelf.

I continued to cut out all of the sheet metal and spot welds until I had a rectangular-ish hole like this.

The point of no return. Or, what the fuck have I done?

The interior footwell and exterior wheel tub are two pieces of sheet metal with an air gap between them. I used a cutting wheel to make a slit in this area, and then used a reciprocating saw to cut out the sheet metal. I ended up with a cavity or “smile” between the interior footwell and the fender well.

Cut out all the sheet metal between the footwell and the wheel tub.

To get a smoother exit for air, I used a cut off wheel to cut several straight lines through the bottom sheet metal. I then bent the sheet metal upwards to close that gap.

Slits make it easy to bend the bottom upwards.

I used a hammer to tap these into a more graceful arc.

I should have cleaned off all that gunk with a wire wheel before starting. This needs to be bare metal for welding.

I then cut the front of the rocker panel off at an angle, so that air can make less than a 90-degree exit. Notice the rocker panel is made of a few layers of sheet metal, and plays a significant role in chassis structure and stiffness.

Next I’ll weld in some sheet metal to cover all this ugliness, and provide a smooth path for air to flow.

Step 3: Welding

Welding Mazda sheet metal kinda sucks. I was once a certified structural welder, but you’d never know looking at the shit job I did on this. I got the best results by spot welding pieces together, and then joining some of those to create stitches. Note that continuous welds are unnecessary, as you can fill the gaps with seam sealer.

First step was to bend in a piece of sheet metal to make a smooth curve, and then cut it to size.

Next I welded in a horizontal flat that makes the ceiling part of the vent. Otherwise there would be an open cavity that might collect gunk inside. I welded that to the curve I had cut.

Next I welded the tabs I cut in the footwell area, that I had bent upwards with a hammer. I ran a cutoff wheel into the overlapping pieces of metal and pulled the little triangles out, then smashed it all flat again. This way the seems were pretty tight and easy to weld inside and out.

Now the fun part, bending sheet metal into curves and tacking it into place. I used aviation shears to cut sheet metal into various shapes, and then tack welded it all together. This process is more art than science, and satisfying work.

Small pieces of sheet metal can be bent and spot welded to make easy curves.

The next step was to add external bracing to the shotgun panel (the frame member that connects the door hinge area to the shock tower). There are several commercially available fender braces, but I chose to DIY my own and weld it on rather than use bolts.

I used 1” square tubing and 1/8” steel spreader plates to make a triangular brace. Weight weenie that I am, I was pleased they only weighed 3 lbs apiece. I then welded the braces in several spots, choosing places where the sheet metal overlapped and was doubled in thickness. I also seam welded the entire front of the chassis for good measure.

Front view shows the 3” (75mm) of extra depth cut inside the wheel well, and rounding underneath and at the exit.

I ordered a new set of fenders, which I’ll cut artfully to expose the rocker vents. In the meantime, I took my existing fenders, which have a large fender cut, and mounted them up to see what this all looks like.

When looking at the rocker vent from the rear, you can see that air will exit much lower and smoother behind the wheel. No more 90- degree bend or having the air exit between the chassis and the quarter panel.

Big vent with a smooth exit will help the splitter make more downforce, and should cool the brakes as well.

Step 4: Finish

The final steps are to fill the gaps between the spot welds with seam sealer (automotive caulk). I have also heard that you can fill the rocker panel and front cavities with expanding foam, which supposedly adds more rigidity. I don’t know that expanding foam from a can (Great Stuff) is appropriate for this, but it would certainly be easy to do.

I’m not going to cover the rocker vents with quarter panels; the area will be entirely exposed. So it needs Bondo, primer, and paint to match the bodywork. I haven’t done all that yet, but it’s just regular bodywork, and I’m the last person you want to watch do that.

When that’s done, I’ll also add some vanes under the car. I’ll take my inspiration from NASCAR again, and guide the air out the new rocker vents with a pair of strakes.

Wind tunnel test?

To find out how much downforce and drag the rocker vents make, I’ll fabricate rigid covers that approximate the shape of the original bodywork. I’ll then remove the covers and A/B test the vents back to back in the wind tunnel.

The A2 wind tunnel has a static floor, and so it’s not as accurate as one with a vacuum to remove the boundary layer, or a rolling floor. So while the numbers won’t be 100% accurate, there will be a useful delta value, and a way to compare the rocker vents to other vents.

Caveats aside, I’ll update this article with those results sometime after 6/20.

Now it’s your turn

If I’ve inspired you to make rocker vents, consider the following:

First, do you really need rocker vents? A splitter alone makes enough downforce to offset a low-angle single-element wing. Add splitter diffusers and vent the fenders, and you may get another 50% more downforce. Add spats, side plates and canards, and you could double the original splitter’s downforce. At this point you’ll almost certainly need maximum wing angle and a Gurney flap, and even then the front aero load distribution may be too high (oversteer in fast corners).

But if you’ve already done all the tricks and still need more front downforce, then the rocker vent is perhaps the next step. But you’ll probably need to add some combination of spoiler, second wing element, and rear diffuser.

On the other hand, maybe you’re not after maximum front downforce and are simply after better efficiency, or maximizing the effectiveness of the undertray. Or perhaps you’re dodging the points taken for using a splitter, but you want similar downforce. That’s a pretty clever use of rocker vents, and I would definitely get on board with any of that.

Next question, do you have a full cage? You’re removing important structure from the unibody and a full cage is arguably a requirement for rocker vents. Without a cage, you’ll need fender braces or other supports that help support the shock tower (no, not a strut tower brace).

As a practical matter, most commercially available Miata fender braces use the two threaded bolt holes in the wheel tub for mounting, and the location of those bolts limits the size of the rocker vent. If you recall the previous image with the horizontal lines, you’ll want to cut on the white line, or lower.

If you make rocker vents, you’ll need a full cage and/or fender braces. Some bolt-in solutions shown.

Final question, if you’re racing, do your rules allow modifying the unibody structure? For any Spec racing series, that answer is certainly no. For other series, it will depend on how they evaluate such modifications.

I’m building Falconet to the NASA ST/TT 3 rules, and changes to the unibody require the “non-production chassis” mod, which is a .4 points penalty to the lbs/hp calculation. If I was building to the NASA ST/TT 4-6 rules, then this modification would be illegal, as the rules state you can’t alter the unibody.

My car also fits into SCCA Time Trials Unlimited 2 class, and this would be legal. But not in any of the Max classes (in fact Falconet isn’t legal in Max for other reasons as well).

I didn’t look up the SCCA road racing rules because it’s a 1000 page rule book. Nor did I look at the 400 page autocross rules, because this is an aero mod, and unlikely to help much at 40 mph. But there’s probably some flavor of unlimited class my car would fit into if I wanted to roadrace with the SCCA or dodge cones in a parking lot. (I don’t.)

Most of the Grid Life Trackbattle classes state that you can’t make modifications to the chassis, and that includes GLTC. But you should be able to get away with rocker vents in Street Mod, Track Mod and the Unlimited classes.

Most endurance racing rules would allow rocker vents, and with their high efficiency, it would be a good idea for AER, 24 Hours of Lemons, Lucky Dog, Northeast GT and WRL. Champcar would assign material points (2 points per square foot of metal), and that might be 4 points total. But a clever team could reuse sheet metal taken from various weight loss trimmings and/or use the lower quarter panels and do this mod for free.

If you aren’t racing or tracking your car, one could argue that rocker vents are a lot of work for Racing Inspired Cosmetic Enhancement (RICE). But if that’s the way you roll, you’ll one up everyone else’s fender vents at the local cars and coffee.

All told, this project cost me maybe $40 bucks in materials (8 feet of 1” steel tubing, welding wire and gas), so if you have more time than money, I say why the fuck not? So now it’s your turn. Let’s see some rocker vents!

From Corvette C5 Wind Tunnel Test to GLTC Win

Luke McGrew qualified on pole for the first Grid Life Touring Cup (GLTC) race at COTA this year, and then proceeded to win all the races. This isn’t super surprising, because he’s always a front runner. But Grid Life nerfed the flat-tuned cars even more this year. So how the fuck is Luke doing it in a C5 Corvette?

For starters, he’s a hell of a driver. He’s also really smart about the way he sets up his car. For example, he uses a small spoiler rather than a wing. But wait a goddamn minute, everyone knows wings work better than spoilers, right? Well, it depends on the car, and it depends on the rules.

GLTC is a pounds-per-horsepower series that allows some aero for free (small wings and spoilers, undertrays without splitters, hood and fender vents, etc), but penalizes or bans other aero parts. As such, a careful reading of the rules is important, and optimizing to those rules can confer a small advantage.

Luke knows what he’s doing, and part of that is doing the research. In that, he found an old wind tunnel test on a C5 Corvette. After reading that, he asked me to “check his math” so to speak, by running simulations in OptimumLap. After purchasing the wind tunnel report (to get the cL and cD data), I built several versions of his car, ran simulations, and verified his gut feelings were spot on.

No splitter and a spoiler instead of a wing.

There’s more backstory to this story, so let me elaborate.

The wind tunnel report

Back in 2002 a group of SCCA racers took a C5 Corvette to a wind tunnel and published a report on the results. It’s not a very long report, but the story is compelling, and the data speaks for itself. The report is available here for $37. I’m going to review some of what’s in that report, but without any specifics, because the author said not to reprint any of it without permission, and so I won’t.

The group did 26 runs in 10 hours, which is oddly the same number of runs I did at the A2 wind tunnel. They used a much larger wind tunnel at the Canadian National Research Center, in Ottawa, Canada, whuch measures 9 meters square by 24 meters long. This is quite a bit larger than A2 wind tunnel (which is 14 feet wide and 58 feet long), and so the Canadian results should be more accurate.

But how accurate is a wind tunnel compared to the real world? I don’t know. When I posted my wind tunnel data online, some internet pundit, without a shred of empathy or humility, puffed up said I made a major mistake in my report, because the wind tunnel optimizes to a constant Qrh, not V-100mph, so that my data was useless without the Qrh average for each run. I have no idea what that means, but I don’t see anything like a Qrh average column for this wind tunnel report either. And so I guess all this data from Canada is similarly worthless?

Well, I’m not a professional aerodynamicist, I’m a fuggin hack, but I’d have to think the differences from each test run are still important, even if the actual numbers aren’t 100% accurate. So let’s shove all the caveats and internet buffaloes aside and move ahead with what they tested, and the comparative data.

Drag – In the test they tried various things to reduce drag, from taping up the front grill to rounding the B pillar, to putting a hole in the license plate. Some things worked surprisingly well, some had no effect at all.
Rear wing vs spoilers – A couple different wings were tested, and since the baseline car used a spoiler, they included the data for that as well. But isolating the spoiler data is rather difficult.
Wings, end plates, and Gurney flaps – They tested three different end plates on the standard wing, and their results were somewhat similar to mine, which is that end plates are the least important part of the entire aero package.
Splitters – They tested a splitter with a flat undertray and one with diffusers. They call this a Laguna undertray for whatever reason, and I will say the design looks quite good.
Yaw – I didn’t test yaw, but they did, using both + and – 3 degrees for most of the runs, but they also tested higher yaw angles initially before settling on just 3 degrees for the rest of the tests.
Tire life – While tire life isn’t something you test in a wind tunnel, the report concludes with results from the race season, which showed tire life was considerably longer using downforce. This is something I wrote about before, that downforce increases tire life, and their experience was the same.

OptimumLap simulations

With all of this wind tunnel data in hand, I went into OptimumLap and built Luke’s exact car. I started with the basic specifications for a C5 Corvette, but used a 252 horsepower flat-tuned dyno chart instead. Detuning is what allows a Corvette to compete in GLTC, and a result of that is a very flat torque curve. This is recognized as an advantage, and flat-tuned engines are penalized for that. Cars are also penalized for aero.

To see which aero version was fastest, I created nine versions of his car, each with different aero parts. I used the coefficients of lift and drag from the report, and swapped out every version of splitter, wing, spoiler, etc. This may sound easy, but the the table that shows the coefficients has low resolution, which made isolating the individual aero components a little tricky. Anyway, I persevered and had my nine different cars, giving them different weights to match the rules.

Grid Life Touring Cup is a pounds per horsepower series, and penalizes cars for using aero, by making them heavier or less powerful. For example, if you use a spoiler or wing that’s larger than 250 square inches, there is a penalty depending on how large you go. Likewise, a splitter carries a penalty over an airdam, and there are penalties for various combinations of wings with splitters.

Because it’s easier to adjust a car’s weight than its engine tune, I simply changed the weight of each aero build to match the GLTC rules. Thus, the car would weigh between 3213 lbs (free aero) to 3371 lbs (splitter and big wing). Note that these weights may be off by a season, as GLTC again nerfed the flat tunes. And also, don’t take too much into the lap time itself, OptimumLap can’t really predict lap times without a lot of fudging, so this is just comparative data you’re seeing.

I then ran all nine cars around various the race tracks GLTC goes to, to see which would win. The winner wasn’t the same at every race track, but a few builds bubbled up to the top, and some sank to the bottom. The following image shows a speed trace and lap times of five of those builds at COTA. I’m not going to reveal which one was the fastest (that’s between me and Luke), but I will tell you which one was the slowest.

Speed trace of five C5 aero builds for GLTC.

See that black line that has the highest top speed? That’s the OEM aero version, essential a base trim model (BTM) off the showroom floor. It might have a 5-10 mph advantage on the back straight, but it posted a 154.44 lap time, which was the slowest at COTA, and also the slowest at every other track. In the end, cornering speed matters more than top speed.

To get back to what I was saying earlier, Luke uses a spoiler and not a wing. GLTC allows you to use a small wing (less than 250 square inches) for free, and so wouldn’t this be better? Not always, and it’s actually quite close. I go through this investigation in my own wind tunnel report (a $25 bargain), showing that there are times when the wing is faster, but sometimes the spoiler wins.

Miata Wind Tunnel Test Ideas

I’m going back to the A2 wind tunnel this summer and I’ll test a bunch of stuff on my Hayabusa-swapped Miata, Falconet. Wind tunnel testing is expensive, and it’s an 11-hour drive each way, so I need to be ultra prepared so that I’m not wasting time and money.

As part of that preparation, I want to know what other people are curious about. Do you have some parts to test? Send them to me and I’ll send them back when I’m done. Do you have some ideas you want to test, but can’t implement? Maybe I can cobble something together in time. Please drop me a comment at the bottom or use my contact form to email me, and I’ll do my best to test what’s important to the Miata community.

One of the benefits to wind tunnel testing is the parts don’t have to be race spec, they just need to survive a couple runs. So a lot of parts go on with the minimum number of fasteners and the maximum amount of duct tape. This allows me to do so a lot more fabricating and testing than would normally be possible.

Anyway, here are some things I’ll be testing.

Canards

I’ll be the first to admit I was wrong about canards, and so this is one area where I’ll be spending a lot of ~~time~~ money. Before I tested my Veloster N in the wind tunnel, I thought canards were poseur junk, but after finding out that changing the height by 8″ made a 700% increase in downforce, I realized I knew jack shit about canards.

I’ll test height to find out the optimal position on the lower canard. I’ll also test size, shape, angle, profile (blade vs airfoil), and end treatment (wicker sizes).

Splitter

On my Veloster I tested flat vs curved splitters and found massive gains (150% more downforce) using a splitter that curved upwards at the trailing edge. This is essentially that same thing as using splitter diffusers, but instead of diffusing air into the wheel wells, the air is diffused over the entire width of the car. So I’d like to test this vs a flat splitter with splitter diffusers.

I also added vortex strakes in front of the wheels and this reduced drag quite a bit, but because those strakes were only on one splitter, I didn’t do a proper A/B test. So I’ll test these again on the same diffuser to see how worthwhile that is.

I’ve seen some online conjecture on the drag from splitter rods, and it doesn’t make a lot of sense to me. I’ll double the number of splitter rods (they’ll be fake) and see what happens.

There’s already published data on splitter length, but I might test this if enough people are hungry for that data. Likewise splitter height has been tested and published (but not on a Miata). Just the same, it’s easy enough to put blocks under the tires and test changes to height and rake, and how that affects downforce and drag.

And I might get around to testing an airdam with an undertray and no splitter lip. I think I can get the undertray to make a lot of suction, even without a splitter lip. This test isn’t a high priority for me, because I’m not personally going to set up Falconet like this, but with enough community whining, this test could go higher up the list.

Underbody

When it comes to underbody aero on touring cars (especially Miatas), I’m a confirmed naysayer. But, just like it was with canards, I might find myself eating my own words after this test!

Some of the things I’ll be testing are a flat bottom, a partial flat bottom (trans tunnel exposed), barge boards, and at least one diffuser. Falconet uses a motorcycle transmission, and so that whole transmission tunnel is open. I’ll diffuse some air into that area and see what happens.

Vents

Al at Race Louvers has some of the best wind tunnel data on the web, and I see no reason to duplicate his efforts. But I have some ideas to get more extraction out of the wheel wells, and these haven’t been tested by him yet. I’ll also be testing a hood extractor vent, which is specific to Falconet, but the data may be interesting to others.

Wings

I’ve already tested and published wind tunnel data on five wings, four end plates, and Gurney flaps, but I have a few things still to test.

I have an oddball wing I made with a short 41″ wingspan with 16″ of chord. It seems absurd, but the additional chord was shown to be very efficient in previous testing, with a clear top-speed advantage. I want to try this as a single and dual element.

There are big wings and big wings. This is the latter.

I’ll compare that wing to a Wing Logic, and that in turn to the industry standard 9 Lives Racing wing. Wing Logic appears to be a CH10, which has less camber and thickness than a Be 123-125 (which is about what the 9LR wing measures). If both wings were the same size and had the same Gurney flap, I’m fairly certain the 9LR would outperform Wing Logic. But this isn’t apples to apples, since the latter has more chord and a built-in Gurney flap. Anyway, interesting comparison.

I may also test my MSHD wing as a dual element. It’s designed as a 3D wing, but unlike many, the trailing edge is a single flat line across the span, and so I can add a second element pretty easily. And I kinda want to make a 2D MSHD, this one will be 63.5 x 11 with a built-in 1/2″ Gurney flap.

MSHD 3D 500 sq-in outperformed all other wings in my testing.

Tops

I tested the first version of my fastback at Watkins Glen, and I’d like to correlate the results from real-world track testing to wind tunnel testing. So I’ll bring an OEM roof and trunk with me and I may as well do one run without the top as well. I wish I still had a Chop Top, that would be worth testing again.

The one time I tried my race car’s fastback on my street car, the engine dropped a valve. But notice how narrow it is at the B pillar.

Open windows

Open windows add drag and reduce downforce, and so I’d like to test various things that may help. I’d like to test a wicker or vortex generator on the A pillar, smoothing airflow out the B pillar, using a longitudinal strake along the top of the window, and large NASCAR-style window nets (which are mostly fabric and not a lot of holes).

Mirrors and mirror stalks are another thing that might affect open windows, or downforce in general. By forcing air downwards, it’s possible to move air away from the windows (and wing). Conversely, moving air upwards may add downforce. And how would these trick mirrors compare to OEM mirrors or no mirrors at all? Gotta find out.

And you?

So that’s at least $4000 worth of testing and I haven’t started on your tests yet. What’s keeping you up at night?

The Dumbest Aero Rules? Splitter Rules

As someone who is building a car to fit into multiple racing classes, I need to keep up with the aero rules in different series. It’s important to know not only what is allowed in each class, but how various aero components are weighed vs other performance modifications. I gave a broad overview of several aero rules in Aero Rules… but OMG the Fecking Rules!, but wanted to dive deeper in one area.

Through this journey, I’ve leaned that most rules are written by people who have a Childs understanding of car aerodynamics. That’s not a typo; let me explain.

I used to read the Jack Reacher series of novels by Lee Childs, and in every book the author would make a dumb technical error relating to firearms. I know firearms because I was a nerdy reloader for several years, and in this field, where things can literally blow up in your face, as Jack Reacher would say, “details matter.”

As a character, Jack Reacher understood this, but his creator did not. After the author incorrectly referred to a trio of Thompson submachine guns as “grease guns” I emailed Lee Childs and said that while I enjoyed his writing, he needed a technical editor. I offered to copyedit his next book for free.

Note #1: The M1928A1 “Tommy gun” and M3 “grease gun” fire the same ammunition and perform the same role, but they look nothing alike and even someone who knows nothing about firearms wouldn’t confuse the two. I mean, one of them looks like a tool for squirting out grease, the other one is in gangster movies.

Note #2: Writing the author directly and offering my services may seem like a bold move, but this is exactly how I became a globe trotting motorcycle journalist for Moto-Euro magazine.

I wasn’t surprised when I didn’t receive a reply from Mr Childs. And so I also wasn’t surprised when he fumbled again in a later book referring to to a rifle as a “M14 Garand”. <sigh>

There’s no such thing as a M14 Garand. I own a M1 Garand, and I’ve shot a civilian M14, and while they are similar rifles, they don’t even use the same cartridge. Calling a rifle a “M14 Garand” is as idiotic as saying that Reacher’s new car is a “Mustang Corvette.” Yes, it’s that stupid.

To make matters worse, in the same story Reacher gets a ride from a woman in an enormous pickup truck. Astounded by the size of the truck, Reacher notes that it’s a Honda. Come on now!

Honda doesn’t make an enormous pickup truck, and now I was certain about two things: Childs knows as much about firearms as he does trucks. These are man’s-man topics, as baked into Jack Reacher’s DNA as the fisticuffs he engages in. After these two gaffes I let out a guffaw and could no longer read anything from that charlatan.

Let’s bring this back to aero rules. It’s difficult to write a racing rulebook, it takes the input of specialists with specific knowledge on suspension, tires, safety, engines, and aero. For whatever reason, rulebooks, like the Reacher books, get published with a Childs understanding of aero. And this results in some silly rules.

Of all the nonsensical rules I’ve seen, from banning fastbacks on convertibles, to equating wings with spoilers, to allowing diffusers but not flat bottoms… the most Childs-like rules are the splitter rules.

Flat splitter nonsense

Splitters separate the air above and below the splitter blade. They create downforce via a high pressure zone on top of the splitter blade, and a low pressure zone below it. You know what else creates downforce in the same manner? A wing.

Just like a wing, a splitter creates much more downforce through suction than from pressure. You can vastly improve the performance of a splitter by adding camber. Most people do this by installing splitter diffusers (splitter ramps), which add curvature over a small area in front of the wheel wells. Time attack cars often curve the entire rear of the splitter’s trailing edge upwards, thereby creating a splitter diffuser across the full width of the car. And some cars also curve the front upwards as well.

However you do it, adding curvature creates a Venturi, accelerating air under the lowest part of the splitter. This in turn drops the pressure, resulting in suction and downforce.

Given that this is how a splitter works, it’s surprising how many rulebooks specify that a splitter must be flat (or horizontal, or without curvature). A rule that states that a splitter must be flat is akin to a rule that states that a wing must be flat! I think we can all agree just how well a wing like that would work.

Splitters without side plates

Because splitters behave similarly to a wing, they also benefit from some attention paid to the outer edge. Look at any wing and you’ll see end plates. Look at any pro-level race car with a splitter, and you’ll see various things on the end of the splitter, which I’ll collectively refer to as side plates. These devices trap high pressure air, change the stagnation point, promote extraction, or in other ways improve the splitter’s functionality.

AJ Hartman’s side plates add over 60 lbs of downforce at 100 mph.

If there’s room on the end of your splitter blade, side plates are a no brainer. And yet, how many racing rules state that splitters can’t have anything on the ends of them? Many of them! Just for fun, I’ll pick on the SCCA autocross rules:

Front splitters are allowed but must be installed parallel to the ground… The splitter must be a single plane with the top and bottom surfaces parallel… A front splitter and its associated features shall not function as a diffuser… Splitter fences or longitudinal vertical members that serve to trap air on top of the splitter by preventing it from flowing around the sides of the car are not allowed.

You see the same verbiage from NASA, SCCA road racing, and numerous other club racing rules. Given that the splitter rules don’t allow side plates, I find it surprising they allow wings with end plates. I mean, it’s the same damn thing.

Splitter width and length

Time attack cars don’t race wheel to wheel, and so rear wings and front splitters are often much wider than the car. For example, in the Global Time Attack rules, in the Limited class, you’re allowed to use a splitter that extends 10″ in front and is 14” wider than the body.

On the other hand, most wheel to wheel racing rules limit the span of wings and splitters to body width. This is understandable, as you don’t want aero parts to hit each other on track. However, there’s no standardization on the length of the splitter lip. When you consider how few racing rules mention the chord of a wing, it’s amazing how many rules there are on splitter length. Again, this is the same thing! Here’s a smattering of splitter lengths:

12” Champcar
6″ – NASA ST1-ST4
5″ – SCCA Time Trials Nationals, SCCA GT1, GT2
4″ – NASA ST5
3″ – Grid Life Touring Cup, SCCA STU
2″ – SCCA STO, GT3, Super Touring, T1
0″ – SCCA Street Prepared

I’ll make fun of the SCCA autocross rules one more time, because it’s such low-hanging fruit. Did you see that last item on the list? The SCCA autocross Street Prepared rules allow you to have a splitter, but it can’t stick out past the bumper. Here’s the exact wording:

“A spoiler/splitter may be added to the front of the car below the bumper. It may not extend rearward beyond the front most part of the front wheel well openings, and may not block normal grille or other openings, or obstruct lights. Splitters may not protrude beyond the bumper. “

WTF? You can have a splitter but it can’t protrude beyond the bumper? How is that possible? Perhaps when this rule was written (1973?), all cars had underbite bumpers, but show me a modern car that you can fit a splitter to that doesn’t extend beyond the bumper!

You could fit a splitter to the *Mustang Corvette* on the left, but not on the right.

I don’t want to just pick on SCCA autocross rules, because splitter rules across most car racing rulebooks show a misunderstanding of how splitters work. Whoever is writing these splitter rules could have easily written the following rule for wings: “Cars may use a rear wing, but the wing must be completely flat, installed horizontally to the ground, and with no curvature. Vertical members that serve to separate air above and below the wing are not allowed.”

I’m curious, who was the first rules lawyer that decided to castrate splitter performance? I feel like they have a lot to answer for. Some may argue that cost cutting is the reason, but you can make splitter diffusers and side plates for $10. Heck, you can make a fully curved splitter for free by selecting a warped piece of plywood!

But in the end, I guess it doesn’t matter who started this, because virtually every other rules writer copied and pasted the same absurdity into their rulebooks. That’s on all of them for having a Childs understanding of aerodynamics.

Diffusers Without Flat Bottoms

There are a lot of racing organizations that have rules which allow cars to have a diffuser, but don’t allow a flat bottom. If you don’t understand how a diffuser works, you might think this is some kind of advantage. If you understand how a diffuser works, you might take a hard pass. So how does a diffuser work?

A diffuser helps air to expand both within the chamber, and in the wake of the car. As a result of this, the air in front of the diffuser drops in pressure and increases in velocity. The result is downforce. Contrary to popular belief, the diffuser itself isn’t where downforce is made; downforce is located where the greatest restriction is, in front of the diffuser.

So, if you have a car without a flat bottom, what’s in front of your diffuser? On the minority of cars, the manufacturer has done a good job making everything smooth, and it can almost replicate a full flat bottom. However, on most cars, and certainly my cars, there’s a shit ton of stuff in the way: transmission, differential, exhaust piping, suspension components, fuel tank, hoses, brackets, exposed frame members, etc. When air hits all of those pieces under the car, it creates local flow separations and drag. Accelerating air through that maze of parts doesn’t create downforce, it creates turbulence.

Diffusers help air expand, accelerating the air in front of it. What’s in front of your diffuser?

Clean airflow makes the most downforce, and thus turbulence is the enemy. Let me give you an example using wings, because I have solid data on that. My Miata has a DIY fastback that provides clean airflow to the rear wing, and it makes 390 lbs of downforce. If I use an OEM hardtop, which has more turbulence around the sides of the canopy, the downforce drops to 300 lbs at the same speed. If I then add AirTab vortex generators and thicken the boundary layer over the roof, the wing makes only 216 lbs. And finally, if I remove the top altogether so that the car is a convertible, the wing makes just 120 lbs of downforce. (These figures come from my testing at Watkins Glen report | data).

So while you can make downforce in turbulent conditions, clean airflow is obviously better, and the situation underneath the car is no different than on top. This is why proper race cars with diffusers have a flat bottom or tunnel under the car, so that they can get clean airflow.

But on a car that must to adhere to rules that don’t allow a flat bottom, the area in front of the diffuser is often a total mess. Moreover, because the diffuser has to begin before the rear axle, any downforce you create is only on the rear tires. Unless your diffuser has a better L/D ratio than your wing, why would you do it?

Maybe that’s a difficult question, let me explain first. On a proper racecar with a flat bottom and diffuser, the throat of the diffuser (where the downforce is located) is often ahead of the rear wheels, around the middle of the car. This means that the downforce is created equally over the front and rear tires. Some diffusers extend even further forward, making more front downforce than rear. Typically front downforce is harder to attain than rear downforce, so this is sort of the holy grail of underbody aero, the way I see it.

Conversely, if your car doesn’t have a flat bottom, and your diffuser begins at the rear axle, you are creating downforce over the rear tires only. A car with a rear wing already creates a lot of downforce over the rear tires, and it does so very efficiently. So unless your diffuser is more efficient than your wing, adding more wing is often a better way to make rear downforce (more wing angle, bigger Gurney flap, more planform area, greater coefficient of lift, etc..).

Nature abhors a vacuum

Let’s play the fantasy game where, without a flat floor, you somehow manage to create a low pressure area in front of the diffuser. Fact: the air everywhere around it is at higher pressure. Because nature abhors a vacuum, the air outside wants to invade the area inside, to balance the pressures as it were.

With great suction comes great responsibility, and so defending that low pressure area becomes your life’s work. There are many ways to seal off the area under the car, such as side skirts (barge boards), or by creating vortices on the sides of the car using canards, or under the car using strakes.

Having done all of that, you also have to attend to what’s happening with your wheels and tires. The most significant problem is that as your tires roll forward, they compress the air underneath them, like a supercharger. This tire squirt sends a high pressure jet of air out both sides of the tire, directly in front of your diffuser! You’ll recall that a diffuser only works because it creates low pressure, and so a jet of high pressure air is a significant problem. It’s less important, but air also intrudes through the spokes of your wheels, and lower caketin covers are required to block this air from going under the diffuser as well.

So as you can see, if you want to make downforce using a diffuser without a flat bottom, the odds are not in your favor. The air under the car is likely turbulent to begin with, and mother nature herself is actively working against you. You have to protect the area of suction as best you can, and if everything goes 100% to plan, you’re still only making downforce over just the rear wheels. When all is said and done, rear downforce is usually gained much more efficiently using a wing.

But I have CFD proof!

But wait, you say, I’ve seen CFD that proves that a diffuser works without a flat bottom!

In the Verus Engineering blog they did a neat CFD study called Is a Flat Underbody Necessary for a Rear Diffuser to Function? The data shows that, compared to a car with a dirty bottom, a diffuser without a flat bottom reduced drag by 26.2 lbs and made 10.2 lbs of downforce. Anytime you can reduce drag and gain downforce, you take it, so this looks like a clear win.

But you could also read their data in a completely different way. Note that the flat bottom alone (without a diffuser) made 23% more downforce than the flat bottom with a diffuser, and only gained 3.5% drag in doing so. So based on this CFD data, one could conclude that this diffuser reduced the effectiveness of a flat floor.

Now that’s a pretty strange conclusion, because I would imagine that any diffuser would help a flat bottom work better. It makes me wonder if the CFD model is too simplistic. Let me not be too critical, because Verus and everyone else publishing CFD is doing us all a favor showing us this data. Computers are simply tools, and with a refinement of those tools, we’ll get better and better data. Let’s just keep moving ahead.

Next, I’ll take a look at Kyle Forsters videos. He has two CFD videos examining flat floors and diffusers, using a NC Miata:

Cut bumper vs Diffuser – Kyle’s first video is just a cut bumper vs diffuser. He didn’t test a flat floor, and the muffler got in the way a bit.
Flat Floor vs Diffuser – In the second video, he uses a flat floor and gets different results.

I watched the videos, and made the following notes. (The downforce and drag values are at 180 kph, or about 112 mph.)

A cut bumper added 6 kg downforce and reduced drag by 1 kg.
A diffuser with the muffler in the way made 4 kg downforce and 4kg more drag. This is not as good as the cut bumper.
Improving the diffuser by removing the muffler added 11 kg downforce, but drag remained the same. Based on this, the cut bumper is still better than the diffuser without a flat bottom.
A flat floor with a cut bumper made 49 kg of downforce and reduced drag by 15 kg.
A flat floor with a diffuser made 50 kg of downforce and reduced drag by 18 kg.

Gleaning this data was a bit difficult, because the comparative data is split across two videos. There are inconsistencies as well; In one video he says the diffuser is 3x more effective at creating downforce than a cut bumper (both with flat floors), and in another video the cut bumper data is virtually the same as the diffuser. So, just like with the previous CFD, there are inconsistencies with the tool, or the operator, or the person interpreting it (mea culpa).

I’ve taken Kyle Forster’s course on aerodynamics, and there’s a lot of CFD on underbody aero, but not much without a flat bottom. However, I’ve had some private consultations with Kyle, and in one of those he showed me a hush-hush diffuser from overseas that was designed to work without a flat bottom. He wouldn’t go into details on the design, and when I asked him how many points that was worth, he shut me down quickly, saying that was private information. But know this, it is possible for a car manufacturer and surely an F1 aerodynamicist to make a diffuser work without a flat bottom. The question is, can you or I modify our cars to do that?

My DIY diffuser

I wanted to know if a diffuser without a flat floor was worth it, so I built one and tested it in the A2 wind tunnel. It’s actually quite neat looking and follows basic aerodynamic principles of vertical and lateral diffusion. With it mounted on the car I was like… damn, that’s cool looking!

Obviously the results in this section apply only to this diffuser on this car, and your results will certainly differ. If you are building your own diffuser, you might take some of my build details and do them differently. (Or better yet, do something else with your time.)

The diffuser is as wide as I could make it between the wheels and extends to the rear axle. It has about 3″ of ground clearance at the front, and the leading edge angles up slightly. My thinking here was to make the area of least restriction (where the downforce is located) within the diffuser itself. I thought this was pretty clever, but this might have been a mistake.

Test fitting my diffuser and setting the height.

I added strakes that diverge from the centerline. These are supposed to spin a vortex off the trailing edge, which should help defeat some tire squirt, by sealing off the center compartment. Thus, the outer chambers of the diffuser are mostly sacrificial in nature, allowing the center tunnel to do most of the work.

The strakes are 3″ off the ground, because at least two of the rulebooks I was looking at state that the minimum ground clearance was 3″. Perhaps the diffuser would have worked better if the strakes were closer to the ground, but rules be rules.

Another detail that didn’t work in its favor is that I made a quick-release diffuser that attached to the car via the trailer hitch and two zip ties. On the upside, I was able to remove the diffuser in one minute. On the downside, there’s a longitudinal cross member behind the muffler that blocks air moving rearward above the diffuser. See the image below.

Quick release splitter mounts to the trailer hitch.

Is this significant? I don’t know. I was mostly concerned with air going under the diffuser, and reckoned anything that snuck between the muffler and top of the diffuser wouldn’t be important. If I had to guess, this probably added some drag, but might have also kicked up some air like a spoiler or big Gurney flap. I’ll have to retest this on my Miata in the future, with topside designed so that air can travel cleanly on top and below.

So how did my DIY diffuser work? If you buy my wind tunnel report you’ll get the full story, but the abridged version is the diffuser made about 15 lbs of rear downforce at a cost of about 3.5 lbs of drag (at 100 mph). That’s not a lot of downforce, but a favorable lift/drag ratio. Except that isn’t the whole truth.

The front of the car lost 14.5 lbs of downforce (when you push down on one end of the car, it goes up on the other), and so the net gain was only .5 lbs of downforce for 3.5 lbs of drag. The data shows that the diffuser would make the car slower. Yuck.

The loss of front downforce is normal, but for whatever reason, the ratio is really bad. I tried 16 different rear aero options in the wind tunnel (5 wings, 3 Gurney flaps, 4 end plates, 2 spoilers) and they all added rear downforce at the expense of front downforce. But all of them did so much more favorably than the diffuser. (In fact one spoiler added both front and rear downforce, but that’s a story for another day.) I don’t know why the diffuser lost an equal amount in front, it’s downright puzzling.

I’m an amateur aerodynamicist, but I’m well studied, and I have good DIY fabrication skills. I generally feel like I know what I’m doing, but I made a diffuser that made my car slower. But this was my first stab at a diffuser, and I can only improve on this. I also didn’t throw all the tricks at it, and I’m sure that side skirts could have helped. But note that the A2 wind tunnel doesn’t have a rolling road, and so the negative effects of tire squirt aren’t quantified here.

But even if I could have made the diffuser work better, I’m still only making rear downforce, and losing front downforce in the trade. In my mind, the entire game of budget touring car aerodynamics is getting more front downforce, that’s the real limiting factor in an aero package. From all of this I can conclude one thing, which is that a diffuser without a flat bottom is fuggin useless to me; it’s much easier for me to gain rear downforce by adding more wing angle, a larger Gurney flap, or using a larger wing.

Rules and reasons

Given this, why do the people who write racing rules allow a diffuser, but ban flat bottoms? I can think of three reasons:

Cost control – Rule makers are trying to reduce the cost of a full-scale aerodynamics war. Enh, I guess that’s a concern, but compared to engine mods and tires, aero is the cheapest way to get performance, and a flat bottom is dead simple. I mean, it’s flat.
Give the people what they want – Rule makers don’t want to restrict people from bolting on parts that look cool. This one I sort of understand. If someone likes the look of a diffuser, they should be allowed to use one. Even if it makes the car slower. But what if someone likes the look of a flat bottom? I mean, give the people what they want, right?
Copy/paste – Rule makers are lazy or don’t know any better, so they copy and paste someone else’s rules. We’ve all been there.

Given those reasons, it’s understandable that some racing rules allow a diffuser, but not a flat bottom. Which rule sets are we talking about? I’m sure there are many more, but here’s a quick look:

Grid Life – Street Modified
NASA Time Trials – TT4
Ontario Time Attack – All classes
SCCA Time Trials Nationals – Max classes

But… not all cars are created equal. Some manufacturers have done a great job fairing and flattening the underside, and a diffuser will work well on those cars. Whereas on many cars, a diffuser without a flat bottom is going to be totally useless. For parity across different cars, I feel that any rules that allow diffusers should allow a flat bottom. Full stop.

Next stop… flat bottom

Despite failing miserably in the wind tunnel, I think my diffuser design is pretty good, and so obviously the next thing to test is how the diffuser works with a flat bottom. But this time I’ll test it on a proper racecar, not my Veloster. In the Spring, I’ll trailer Falconet down to A2 and test every combination of underbody aero including flat bottom, diffuser, side skirt, etc. And maybe I’ll acquire some diffusers manufactured by various companies and test those. Finally, I’m really curious about side skirts with no other aero, that should be an interesting one to throw into the mix.

To stay in the loop, subscribe to my blog or follow my Facebook page. If you value this kind of content, and want to support future aerodynamic tests and articles, please buy me a coffee.

Miata Aero Testing Results

This article was originally spread out over several different pages. I’m not sure what I was thinking at the time, but I’ve reorganized this as a single (rather lengthy) article now.

For the full story on how I performed these tests, see Testing Miata aerodynamics with $20K worth of computers. This article is essentially Part 2 of that one, so I can deep dive on the test results of the various aero options.

While datapoints like pounds of downforce at 100 mph, or horsepower consumed, are things we can wrap our heads around, it’s difficult to translate that into the only thing that matters: lap time. Therefore, in the following sections, I also include lap time simulations using OptimumLap.

Testing Miata tops

The first thing I wanted to test was the largest knowledge gap, roofline shape. This meant I had to have different options, that would come on and off quickly, using the same brackets. The four options were an open top, an OEM hard top, a Treasure Coast Chop Top (which should approximate a hard top with the window removed), and a fastback of my own design.

I built the fastback before I had the notion to do this test, or I would have built it differently. The main problem is that I made long brackets along the bottom edge, and this required removing the trunk lid. This meant I didn’t get to test any of the other tops with an OEM trunk lid. Instead, I bolted a plywood cover over the trunk cavity. This new trunk lid is about 3.5” taller at the back than a stock trunk. It’s hard to say exactly what the effect of this was, but it’s likely a reduction in drag and lift, akin to adding a spoiler. So when you look at the data later, note that none of the tops used an OEM trunk.

Open top results

Miatas are meant to go topless, let’s start there and address some burning questions:

What happens when you use a wing with an open top?
How much does an open top affect a wing’s performance?
At autocross speeds, is it better to remove the top or leave it on?

Take a look at the following table, and you can see that an open-top Miata generates about 40% of the downforce as one with an OEM hard top (Total Cl field). Of all the options, this was the worst at creating downforce.

It might be a little confusing that the coefficient of drag (Cd) is better with an open top than with a hard top. This is likely the result of running the tests with the windows open, which turns the hard top cabin into a parachute.

Top	Front Cl	Rear Cl	Total Cl	Cd	L/D %	HP @ 100mph
Open top	-0.20	-0.23	-0.43	0.43	1.01	47.16
Hard top	-0.17	-0.84	-1.01	0.48	2.11	52.78

Let’s plug these numbers into OptimumLap and see what happens. I’ll use three different tracks to represent a range of speeds. These tracks are already in OptimumLap.

Top	Watkins Glen	Waterford Hills	2010 SCCA Nationals
Open top	2:24.02	1:20.97	1:03.88
Hard top	2:22.96	1:19.95	1:03.34

The hard top is worth about one second at both Watkins Glen and Waterford Hills, and just over a half second on the autocross course.

OEM hard top with plywood trunk lid, a concession to the fastback.

For these simulations, the car weight was kept the same. Someone will point out that the top weighs 45 pounds, and that OptimumLap doesn’t factor in the change in center of gravity. Both true. But I can calculate how much weight you’d have to remove from the open top car to match the autocross time of the hard top, and it’s 210 pounds. I’m not sure how high you’d have to place 45 pounds above the car to equal 210 pounds, but it’s probably pretty far up there!

But running an open top car with a wing has two advantages. One, it looks cool. Two, an open top car with this wing beats any top without a wing, every time. That’s kind of jumping ahead in the data, but it’s worth noting.

Let’s get back to the real world and the test at Watkins Glen. Alyssa reported that the car was more difficult to drive with the open top. She had to brake before entering Turn 10, and then had to manage a car that was oversteering badly. With a hard top, she could mash the throttle from the exit of Turn 9 to the exit of Turn 10. That kind of confidence over an 8-9 hour race can mean a lot more than a second per lap.

Chop Top results

Treasure Coast Miata sells their “Chop Top” for budget endurance racing. It’s an economical and lightweight top that does the job of enclosing the roof. This has two benefits: better aero, and you don’t have to wear arm restraints when racing (the car is no longer considered a convertible). I fabricated mounts that attach to the hard top brackets, and with those the total weight of the Chop Top was a scant 7 pounds.

There is a persistent myth in Miatadom, that removing the rear window from a hard top is aerodynamically better. So I put two small Lexan covers on the sides of the Chop Top, closing in the sides. This made the chop very similar to a hard top without a rear window. Let’s add this data to the open top and hard top.

Top	Front cL	Rear cL	Total cL	cD	L/D %	HP @ 100mph
Open top	-0.20	-0.23	-0.43	0.43	1.01	47.16
Chop top	-0.20	-0.33	-0.53	0.45	1.19	49.40
Hard top	-0.17	-0.84	-1.01	0.48	2.11	52.78

As you can see, the chop top allows the wing to work a bit better than an open top, with an increase in downforce. But it’s not as much as you’d think.

However, once you add a wing, the Chop Top performs barely better than an open top. This is interesting, because you’d think airflow over the roof is considerably smoother than an open top. However, it’s what’s happening on the underside of the wing that’s more important, and the Chop Top roof can’t defeat the turbulence coming from the open sides of the cockpit and going beneath the wing.

Chop Top with plywood trunk cover. Note clear lexan and clear gas line, so we can see if the gas is about to overflow.

Next I’ll do the same track simulations, and what I find interesting here is that the Chop Top isn’t really that much different than an open top at any of the tracks. Not enough to really make a difference.

Top	Watkins Glen	Waterford Hills	2010 SCCA Nationals
Open top	2:24.02	1:20.97	1:03.88
Chop top	2:24.04	1:20.82	1:03.79
Hard top	2:22.96	1:19.95	1:03.34

Nevertheless, for those racing with an open top and a wing, the Chop Top is worth a look for a bit of weather protection and not using arm restraints. In addition, we’ve finally dispelled the myth that removing the rear window is more effective. It isn’t. At least when used in conjunction with a wing.

OEM hard top

All along I’ve been citing the data for the OEM hard top without really discussing it. It’s the status quo in racing, looks great, and performs its duty.

In the data, the OEM hard top generated more drag and lift than what I expected from published data. This is likely due to the open windows and wide canopy, which turns the cabin into a parachute. The drag is supposed to be around .38 with closed windows, but we measured over .5. Lift is also supposed to be the high .30-somethings, and we measured .55 (with vortex generators, I don’t have the raw data without).

The hard top with airdam, splitter, and wing made a killer combo: .48 Cd and 1.01 Cl. Those are good numbers. Racing numbers.

Fastback results

The front of my fastback uses the Treasure Coast Chop Top, and the rear fastback section bolts on and slopes back at about 14 degrees. So essentially the roofline is the same as OEM to about the rear window. Starting with the Chop Top made building the fastback fairly easy, and it was also easy to add and remove for this test. (You can see construction photos this and other tops I’ve built at the end of this article.)

An older photo, but the top is the same.

The Chop Top plus fastback weighed 17 pounds less than the stock hard top, and to equalize the two I bolted 8-pound lead weights to the top of the seat belt towers. This was the only time I made adjustments to the weight of the car, and so the open top and Chop Top configurations were a bit lighter.

The fastback significantly reduced drag, and helped the wing create more downforce. Compared to the OEM hardtop, downforce increased 129.7%. Another way of thinking of that is that the fastback turned a 60″ wing into a 78″ wing. Or you could say that the OEM hardtop is so bad that it made a 60” wing behave as a 48” wing….

The large gain in rear downforce was offset by a small loss in front downforce. Essentially, the wing was so effective with the fastback that the front end lifted, changing the height and angle of the splitter, reducing its effectiveness.

Top	Front cL	Rear cL	Total cL	cD	L/D %	HP @ 100mph
Open top	-0.20	-0.23	-0.43	0.43	1.01	47.16
Chop top	-0.20	-0.33	-0.53	0.45	1.19	49.40
Hard top	-0.17	-0.84	-1.01	0.48	2.11	52.78
Fastback	-0.12	-1.09	-1.20	0.41	2.97	44.81

In addition, the fastback reduced drag by 15%. This in itself is pretty surprising, and not only helps top speed, but fuel economy. Combined, the downforce and drag created a lift/drag ratio that was 50% better than the OEM hard top with a wing. Astounding.

Note that the .41 coefficient of drag is actually quite good when you consider that the wing made the most downforce in this configuration, and just as in all of the tests, the windows were open.

But all was not rosy with this setup. Both Anthony and Alyssa commented that the car understeered badly in this configuration, and was boring as shit to drive. Given time, Jeremiah would have changed the mechanical grip by adjusting the front roll couple, by means of spring and/or stabilizer bar. This would have helped balance the vehicle at speed.

Let’s do another simulation in OptimumLap. The fastback gains 1.9 seconds at WGI, and about half that at Waterford, which is pretty spectacular.

Top	Watkins Glen	Waterford Hills	2010 SCCA Nationals
Open top	2:24.02	1:20.97	1:03.88
Chop top	2:24.04	1:20.82	1:03.79
Hard top	2:22.96	1:19.95	1:03.34
Fastback	2:21.06	1:19.02	1:03.12

Vortex generators on an OEM hard top

The shape of the Miata’s canopy is abrupt, and if you look at wind tunnel tests, you can see smoke trails that are turbulent, and then separate, as air moves over the top. Vortex generators (VGs) create a thicker turbulent layer of air, which keeps air from separating completely. This should result in less drag, and may also help interaction with a wing.

Most vortex generators you see are cosmetic fakery and don’t create vortices. I bought the real deal from AirTab. Made of thin plastic, they go on with double-sided tape, just peel and stick. The manufacturer says they should be mounted no closer than 4” apart. I set them at 5” on center, and so that made 9 for the roof.

If you do some research on VGs, there’s good data that they work. They’ve been used on semi trucks, RVs, the underside of race car wings, and many places where flow separation can occur. For cars, take a look at the four-part series on Autospeed, where they tested VGs on a Prius and Insight. Even better, check out Hi-kick Racing’s blog on adding VGs to a Miata. VGs decreased his lap time from 1:02.8 to 1:02.1. Here’s a photo from his site.

Nevertheless, my expectations were low. If vortex generators are the cat’s meow, then every cat would have them, right? As you can see in the table below, VGs made things worse. Total downforce decreased by about 20%, and drag increased substantially. Take a look at HP consumed at 100 mph and you’ll see you have five less ponies at that speed.

Configuration	Front cL	Rear cL	Total cL	cD	L/D %	HP @ 100mph
Hard top, splitter, wing	-0.17	-0.84	-1.01	0.48	2.11	52.78
VGs, splitter, wing	-0.20	-0.61	-0.82	0.52	1.57	57.58

And this is how those values affect lap time in OptimumLap.

Vortex Generators	Watkins Glen	Waterford Hills	2010 SCCA Nationals
Hard top, splitter, wing	2:22.96	1:19.95	1:03.34
VGs, splitter, wing	2:24.32	1:20.42	1:03.54

We placed the VGs at the trailing edge of the hard top, but it’s possible that moving them forward may have helped some. Or perhaps we used too many? I followed the instructions and they were supposed to work.

The double-sided tape was difficult to remove, and so experimentation with the number and location of VGs wasn’t possible. The only lasting impression the VGs made was the adhesive, it was a bitch to remove and so all further tests would use the OEM hard top and VGs combined.

Miata top conclusions

Different tops change airflow over the roof, and this affects how a wing works. It’s unclear how much of this is is based on turbulence, or because of a change in downwash angle as air hits the wing. It’s likely a combination of both, but we didn’t have time to experiment with wing angle and this mystery remains.

In addition, our data revealed a rolling rake angle that changes ~ ½ degree depending on rear wing configuration, and this impacted front downforce and distribution more than expected. By adjusting chassis rake and splitter angle, it’s likely the total downforce and Lift/Drag efficiency would have been higher, and this may have reduced rolling-rake changes as well.

We can make the following general conclusions about using a wing with different tops.

An open top reduces a wing’s effectiveness by about 2.5x. Or, if you thought you were getting 200 lbs of downforce, you’re getting 80.
A Chop Top performs marginally better than an open top.
The OEM hard top is actually quite good with a wing. But don’t remove the rear window. And don’t use vortex generators.
A fastback allows a wing to perform the best, increasing downforce, while also decreasing drag.

9 Lives Racing Big Wang vs cheap dual wing

The low price and availability of aerodynamic car wings are making them more common in crap-can endurance racing. You can buy a cheap extruded aluminum wing on eBay, Amazon, or other online retailers for $50, but are they good for anything?

I broke the piggy bank and purchased a 53” double-decker wing for $75, shipped. This wing is sold under a variety of brand names like BestEquip, Mophorn, Neverland, etc, and I’ll refer to this as the eBay wing, because that’s where I got it.

Immediately upon unboxing, I knew I’d have to make some modifications. Like most budget items from China, it came with trunk mounts too low to allow airflow to go under the wing. I trusted the supplied mystery-metal hardware as far as I could throw them, which was directly in the trash.

The main wing felt light, yet surprisingly rigid. The stiffness is partly from the dual horizontal mounting rails, which allow the wing to be mounted at just about any width. However, the exposed slots and flat underside of the wing can’t be great for keeping airflow attached. In the future, I may add some curvature here.

The upper wing felt like a noodle, and I feared it would vibrate and hit the lower wing at speed. So I riveted on a Gurney flap of 1/4” aluminum angle to stiffen it up. I also added a small stop in the middle of the lower wing to limit downward movement of the upper wing.

The end plates had to go, not only because they were too small, but they didn’t allow the upper wing to pivot into the correct position. According to McBeath in Competition Car Aerodynamics, the upper wing should overlap the lower wing at about 4% of the chord (.3” in this case). In order to accelerate the air, the gap must be larger at the front than at the rear. With those two factors set, the slots in the supplied end plates, which support and locate the upper wing, wouldn’t allow us to pivot the upper wing into a useful angle.

Since I couldn’t get the correct spacing and wing angle with the supplied end plates, I made my own from 12”x10” sheet metal, and drilled new upper wing mounting slots. All done I paid a little over $110 for the wing and modifications, and a couple hours figuring all that out.

The car was set up with -1 degree of negative rake, meaning the back was lower than the front. Ideally it should be the other way around, but we got lost debugging what we thought was a clearance issue with a front sensor and didn’t correct the angle.

I set the lower wing angle to 3 degrees, but because of the chassis rake, the main wing was closer to 2.5 degrees. I set the upper wing to 12 degrees. Measured over the entire chord, this created a total camber of about 14 degrees, which is right in the middle of the values that McBeath cites in Competition Car Aerodynamics. Given more time, I would have experimented with the angle of attack of the upper element, as well as the entire wing. However, not knowing how the wing would perform, I chose middle-of-the-road values from a published text, and figured it was more important to avoid a stall condition than maximize downforce. The wing weighed 7.6 pounds altogether, which is very light.

60″ 9 Lives Big Wang vs 53″ eBay double wing.

The wing I used through all the other testing on at WGI is a 60” 9 Lives Racing “Big Wang”. The standard Miata wing is 64″, but my eventual plan was to end-plate mount this, like a Ferrari F40. I never did get around to that, so the wing is a bit smaller than you’d see on most Miatas. (But as you already saw above, the fastback made it behave as a much larger wing.)

When I took the wing out of the package I was immediately impressed by the sturdy construction. When there are only cockroaches left in the world, there will also be 9LR wings. I made my own 12” x 12” end plates and mounting brackets, and had a local shop weld the mounts underneath. The entire setup was about $500, and weighed in at double the double wing, at 14.4 pounds.

I set the wing angle at 5 degrees, but with chassis rake the actual wing angle measured 4.6 degrees. If you look at the 9 Lives CFD open air data, this is right in the middle of the wing’s working range. I knew the downwash angle would change with every top, and that this would change the effective wing angle. But due to intermittent track closures, I didn’t get a chance to sweep the wing angles, and left it where it was.

For both wings I used the same wing stands, which I cut from aluminum plate. Many racing series limit wings to roof height, so I made sure the highest part of the wing was level with the roof. I bolted the base of the wing stands through the sides of the trunk gutter, and while this seemed strong enough, I added another L-bracket on top of the rear fenders, and this stiffened things up considerably.

Wings compared

So let’s see how these wings compared. In this first chart, we’re looking at front and rear downforce using GPS speed alone. It looks like the eBay wing (red) creates more rear downforce as the 9 Lives Racing wing (blue), and when combined with the front, the total downforce is pretty close.

However, the weather had changed during this run, and we had an 11 mph headwind. After correcting the wind speed detected by the pitot tube, a clearer picture developed. See the bar graph below. Here we can see that the eBay wing generated less downforce than the GPS speed would have us believe. This is why you can’t trust testing using GPS speed alone, and why you hire a guy like Jeremiah.

When you add downforce on one end of the car, you can expect to lose downforce on the other end. This is the natural see-saw effect of pushing down on one end. What’s interesting here is that the 9 Lives Racing wing not only made more rear downforce, but it also had more front downforce. How can this be?

The most likely reason is drag. The eBay wing creates more drag, and this rear-biased force lifts the front end. Whatever the case, front downforce, and thus total downforce, is a lot less with the eBay wing.

Take a look at the following table and you can see the corresponding values for coefficient of lift (which we’ve been familiarizing as “downforce”), and drag. We already saw that the 9 Lives Racing wing practically doubled the total downforce, and here you can see it did that while creating 15% less drag. If you look at the final column in the table, you’ll see that the drag reduction alone equals an extra 8 hp at 100 mph.

Wing	Front cL	Rear cL	Total cL	cD	L/D%	Front Load	HP @ 100mph
9LR	-0.17	-0.84	-1.01	0.48	2.11	16.76%	52.78
eBay	0.03	-0.58	-0.56	0.55	0.90	4.43%	60.79

If we divide the total coefficient of lift by the coefficient of drag, you get the L/D ratio, which tells you how efficient the entire aero package is. Here you can see the 9 Lives Racing wing contributes to a setup that is over 230% more efficient at creating downforce.

The data from testing other tops shows that the 9LR wing changes the coefficient of drag by about .03 across all tops, from open top to fastback. By contrast, the double wing changes the Cd by .10. Yowza, that’s a lot.

One final calculation is the front aero load distribution percentage, which gives you an idea of how much the car will understeer (a low percentage) or oversteer (a high percentage). The low values here indicate that with either wing, the car would understeer badly. This is partially due to the negative rake of the chassis and negative splitter angle (both setup mistakes that should have been corrected before testing). However, even with these setup details corrected, the eBay wing would produce a car that understeers more.

As usual, let’s see what happens in OptimumLap. To spice things up, I’ll also add the data from the 9LR wing with an open top.

9LR vs eBay	Watkins Glen	Waterford Hills	2010 SCCA Nationals
Hard top, 9LR	2:22.96	1:19.95	1:03.34
Hard top, eBay	2:25.71	1:20.99	1:03.81
Open top, 9LR	2:24.02	1:20.97	1:03.88

The 9 Lives Big Wang outperforms the cheap eBay wing in every way. In fact, the 9LR wing with an open top out performs the eBay wing with an OEM hard top on any track that isn’t autocross. 9 Lives Racing is a small, made-in-the-USA business with employees who race cars, and you can feel good about supporting them.

But if you’re racing in 24 Hours of Lemons on a $500 budget, you might find that a cheap wing suits your janky crap-can just fine. The wing could have performed better with more fine tuning, but it’s clearly a case of “you get what you pay for.” If you purchase this wing, you might want to take similar steps that I did to limit movement of the upper wing, optimize the convergent gap and wing angles, and get the wing about 6 feet above your roofline. It is Lemons, after all.

Wing vs no wing

For most of the test I used a 60″ 9 Lives Racing wing, but I wanted to see what would happen without it. The coefficient of drag went down by .03 for all configurations. This is rather interesting, because usually when downforce goes up, so does drag. But this wing had the same drag in all configurations.

When I removed the wing, total downforce took a nosedive, and for the first time in the test, the car generated lift instead of downforce. This was an appropriate time to test the hardtops and see how they did without a wing.

First we threw the Chop Top back on and did a run. Then we attached the rear section, making it into a fastback again. And finally we tested the OEM hardtop. Unfortunately the OEM hard top still had the vortex generators attached, so I’ve made an educated guess on the Total Cl and Cd values below (these are in italics in the table below).

Configuration	Front Cl	Rear Cl	Total Cl	Cd	L/D %	HP @ 100mph
Hard top, VGs, no splitter, with wing	0.18	-0.64	-0.47	0.53	0.88	58.75
Hard top, VGs, no splitter, no wing	0.22	0.38	0.59	0.49	-1.20	54.53
Chop top, no splitter, no wing	0.11	0.20	0.31	0.48	-0.64	52.93
Fastback, no splitter, no wing	0.16	0.18	0.35	0.38	-0.80	42.00
OEM hard top, no splitter, no wing			.50	.45

What’s surprising here is that without a wing, the chop top has the best L/D ratio. This is largely because it creates the least lift. Remember that negative lift values are what we’re looking for (downforce), and the chop top’s .31 Cl has the least lift. The fastback creates more lift than the chop top, but it does so with less drag, and in the end, this is makes a faster car.

Unfortunately we didn’t get data for a bare OEM hardtop (without VGs, wing, or splitter), so we don’t know if it’s the shape of the OEM roof, or the VGs that create so much lift. But the total Cl value of 0.59 is quite a bit worse than either the chop top or fastback. Based on the data we obtained doing the wing tests, a bare OEM hard top should have a CD of about .45. It’s hard to imagine the vortex generators adding more than 10% lift, and that would put the total lift around .50.

Let’s see what happens in OptimumLap when we remove the wing.

Wing or no	Watkins Glen	Waterford Hills	2010 SCCA Nationals
VGs, no splitter, 9LR wing	2:25.65	1:21.12	1:03.89
Hard top, VGs, no splitter, no wing	2:29.14	1:23.14	1:04.91

Yikes, that sucks! The wing is worth 3.5 seconds at WGI and even 2 seconds at a short track like Wateford? Amazing.

Now let’s just compare the different tops without wings.

Tops without wings	Watkins Glen	Waterford Hills	2010 SCCA Nationals
Hard top, VGs, no splitter, no wing	2:29.14	1:23.14	1:04.91
Chop top, no splitter, no wing	2:27.74	1:22.85	1:04.63
Fastback, no splitter, no wing	2:26.51	1:22.42	1:04.63
OEM hard top, no splitter, *no wing*	2:28.17	1:22.87	1:04.80

Here we can see that the fastback is still the fastest configuration, but it’s not a huge difference unless you’re at a high-speed track like WGI. Without a wing, I’d be happy to use a Chop Top at most tracks for the light weight and convenience of strapping in the driver and accessing things like a cool suit, radios, cameras, etc., in the cockpit. And on performance, the Chop Top beats the OEM hardtop, even without factoring in the 38 pound weight difference.

One thing that’s conclusive here is that if the rules allow it, use a wing. This probably even applies to classes like NASA ST6/TT6 that carry a substantial penalty for running a wing.

Airdam and splitter

For most of the test we used a 4″ splitter. This was bolted to a flat undertray, flush with the airdam. I wanted to see what removing the 4″ splitter extension would do. We expected a loss in downforce, but I wasn’t sure if drag would go up or down. You see it both ways online, with CFD data showing that a splitter reduces drag, and the occasional internet expert claiming that drag goes up.

Splitters	Front Cl	Rear Cl	Total Cl	Cd	L/D %	HP @ 100mph
VGs, splitter, wing	-0.20	-0.61	-0.82	0.52	1.57	57.58
VGs, no splitter, wing	0.18	-0.64	-0.47	0.53	0.88	58.75

Score one for the CFD team, the splitter reduced drag slightly. When I removed it, the drag went up from .52. To .53. More importantly, we lost a lot of front-end downforce. Our raw data showed a loss of 69 lbs on the back straight, which calculated to a .38 delta in front coefficient of lift. Let’s see what that’s like in OptimumLap.

Splitters	Watkins Glen	Waterford Hills	2010 SCCA Nationals
VGs, splitter, wing	2:24.32	1:20.42	1:03.54
VGs, no splitter, wing	2:25.65	1:21.12	1:03.89

Obviously, if you’re running just an airdam, and the rules allow it, add the splitter. It’s significant. It’s also worth noting that this was just a plain splitter with no rear curvature or splitter diffusers. Knowing what I know today, the splitter would be twice as effective.

Endurance racing simulations

For endurance racing it’s important to know the amount of energy used per lap, because that determines how far you can go on a tank. You can get this data from OptimumLap simulations. From the energy used, you can determine the length of each driver’s stint, as well as how many laps they can complete in each stint. This can be very important for pit strategy, especially in longer races. Note that simulations are exactly that, and aren’t intended to be exact. But they are useful for making direct comparisons.

The configurations I used for the simulations follow this key. Note that the “B” group has less aero: it uses an airdam, but not a splitter, and no wing.

1a – Open top, wing, airdam, splitter
2a – Chop top, wing, airdam, splitter
2b – Chop Top, no wing, airdam, no splitter
3a – OEM hard top, wing, airdam, splitter
3b – OEM hard top, no wing, airdam, no splitter
4a – Fastback, wing, airdam, splitter
4b – Fastback, no wing, airdam, no splitter

In the table below, the Energy value in the table comes straight from Optimum Lap, and I’ve simply taken a value of 16,500 energy units divided by the energy used to get the number of hours per tank. The Miatas I’ve raced have burned about 7 gallons per hour, so these values aren’t far off.

Config	Cl	Cd	WGI Lap	Energy	Hours per tank	Pit stops	Laps in 8 hours
1a	-0.43	0.43	144.02	9087.53	1.82	4	174.98
2a	-0.53	0.45	144.04	9141.08	1.81	4	174.95
2b	0.31	0.48	147.74	9183.12	1.80	4	170.57
3a	-1.01	0.48	142.96	9223.37	1.79	4	176.27
3b	0.5	0.45	148.17	9078.14	1.82	4	170.07
4a	-1.2	0.41	141.06	9031.23	1.83	4	178.65
4b	0.35	0.38	146.51	8911.28	1.85	3	174.05

Take a look at the first two, this is the open top vs Chop Top, and it’s interesting that they turn almost exactly the same number of laps. The OEM hard top beats those by a lap and change.

But the fastback configuration 4a is the clear winner here, turning 178 laps, two more than the OEM hardtop with the same aero. Make this into a 24 hour race and the fastback wins by seven laps. Notice the Energy column, the fastback with wing is not only turning the fastest laps, but it’s using less gas than any other configuration, save the fastback without the wing.

In the non-aero B-group, the fastback (4b) wins by three laps over its wingless brothers. This is partially because the fastback can take one less pit stop. If I re-run the data with a larger gas tank so that all configurations have the same number of pit stops, then the fastback wins by only two laps.

But once you add a wing, the race is over. In fact the worst combination with a wing (open top) beats the best performing top without a wing (fastback) by a full lap, even though the fastback does one less pit stop.

If you want to check my math, I have a spreadsheet with these values, and for simplicity, I’ve removed a lot of the variables in this table. You have to keep track of things like yellow flags and time taken for each pit stop, which determines the actual driving time per race. The 8-hour race I’ve simulated uses 420 minutes of racing time instead of 480 minutes. Yellow flags at Watkins Glen take longer, and I’ve also subtracted 5 minutes per pit stop.

For shits and grins, the eBay wing again

Let’s see what happens when we use the airdam, splitter, OEM hardtop (without the vortex generators) and the cheap eBay wing. This is configuration 3c in the table below. Look above and compare.

Config	Cl	Cd	WGI Lap	Energy	Hours per tank	Pit stops	Laps in 8 hours
3c	.56	.55	145.71	9393.39	1.76	4	172.95

The eBay wing (3c) loses to every configuration that uses the 9 Lives Racing wing. Yup, even the open top car with a 9LR wing is going to be the hardtop with a cheap wing. However, the eBay wing beats OEM hardtop without a wing (3b) in both a sprint race or an endurance race. So it’s worthwhile running a cheap wing if you have nothing at all.

The biggest difference is the Energy field. Compared to the lowest drag version (4b), the eBay wing uses 5% more gas on every lap. That may or may not be a consequence, depending on how long you’re in the car.

It wouldn’t matter in a sprint race; the eBay wing (3c) would win against the fastback without a wing (4b). But in an endurance race, it would depend on the tank size and driving stint time. If I change the data so that they all take the same number of pit stops, then the eBay wing wins. If I leave the gas tank size as it is, then the fastback without a wing wins.

Conclusions

I wrote an article called The Dunning-Kruger of Car Aerodynamics, wherein I examine the steps most people go through on their aerodynamic journey. It was lucky that I met Jeremiah online and was able to do these real-world tests, and jump right past the pit of despair.

I’m sort of backsliding on the D-K chart by doing wind tunnel testing now, and who knows, I may slide further backwards towards CFD. But in the future, my ‘Busa swapped fastback Miata, Falconet, will have a full onboard system with strain gauges, pitot tube, and all the works so that I can do the same kind of testing I started with.

In 2024 I’ll do a full writeup of how to install all the parts and hook it up to an Aim dash (ahem, with a lot of help from Matt Romanowski), but this should bring me full circle back to a system akin to what I started with. With those tech gizmos I’ll be able to get real-world data on downforce and drag whenever I want, essentially making any race track into my own private wind tunnel. Fuck yeah.

Testing Wing End Plates in a Wind Tunnel

The following article is made up of excepts from my wind tunnel report. You’ll get a more cohesive story, and a lot more data on many more aerodynamic parts, if you buy the report and read it end to end.

End plates on wings are necessary; they separate the low-pressure region under the wing from the high-pressure area on top of the wing. The suction side of the wing is what does most of the work, so by keeping the high pressure side from bleeding into the suction side, the wing makes more downforce.

The shape of the low-pressure region under the wing is different for every airfoil. However, for most wings designed for motorsports, you’ll find that the low-pressure region is at the front of the wing and often extends ahead of the wing. The low-pressure area extends about a chord’s distance below the wing as well.

*The shape of the low-pressure region below the wing depends on the airfoil. I inverted the images so that it relates to car wings. This image is from Race Car Aerodynamics, and I highly suggest you* buy the book.

Given that information, and after looking at the preceding image, you might conclude that a good endplate should be shaped to exactly cover the high and low pressure regions of the wing. And from that, you might surmise that a good endplate for a 10” chord motorsports wing should extend 10” below the wing, and should have a lot of surface area concentrated at the front. And that’s how I see it as well. However, some end plates have most of the area towards the rear of the wing, and I can’t say I understand that. But aerodynamics is full of weird contradictions, and perhaps some of those end plates work.

With the amount of companies selling improved end plates with different shapes and sizes, you’d assume there was something to be gained over a plain rectangular end plate. And because some of these fancy end plates cost a couple hundred dollars, and boast CFD-designed pedigrees, they must be doing something useful, right?

CFD and wing efficiency

Some of my wind tunnel data conflicts with published CFD (computational fluid dynamics) data. This isn’t surprising, as CFD is just a computer calculation, and not real-world data. Manufacturers typically test wings in free-stream CFD, meaning that the wing is suspended in mid-air, as only a computer simulation can do. This is the best way to calculate what happens when you change wing angle, add Gurney flaps, or change the shape and size of end plates. Free stream CFD is essential, because it eliminates everything in front of the wing. This is really the only way to compare one thing to another.

But when you put a wing on a car, everything in front of the wing affects its performance. Wind speed and direction, cars in front of you, and open windows can make a huge difference. Plus there’s the shape of your car, the angle of the windshield, aerodynamic devices on your car, like splitter, canards, hood vents, vortex generators, GPS antenna, wing stands, … you name it, every single thing that’s in front of your wing changes how it performs. You’ll never get the same amount of downforce from your wing as the free stream CFD data shows. Not even close.

You can research wings on Airfoil Tools or Bigfoil, or use tools like Javafoil and CFD, and you’ll find wings that have a 14:1 L/D ratio, or better. But when you put the wing on the car and adjust the angle of attack, you’ll be stoked when your wing has half of that.

For this reason, anything you do to improve the efficiency of the wing in free-stream CFD is meaningless until you put it on the car. For example, modifications to wing end plates can reduce drag, and this shows up in CFD as a gain in wing efficiency. But when you put the wing on the car, the drag of the wing is inconsequential to the total vehicle drag. Touring cars are essentially huge rounded bricks, and wings are tiny streamlined objects by comparison, and so you can understand that the drag from the wing is essentially nothing compared to the drag of the vehicle.

In reality, the only thing that matters is the aerodynamic efficiency of the entire vehicle, and you typically get that by going after as much wing downforce as possible. Modifications to the end plate that reduce drag might increase free stream wing efficiency, but they do that by reducing wing downforce. And this makes the L/D ratio of the car worse, and the car goes slower. Ergo, it’s utterly worthless to optimize wing efficiency in free stream CFD by reducing drag. If you use CFD for anything, it should be for optimizing the wing for maximum downforce.

Testing end plates in a wind tunnel

Before I get to the testing data, let me tell you exactly how shit stupid I am. My aerodynamics sensei Kyle Forster had these things to say about end plates:

Use a rectangular shape. Adding vents, cuts, and other tricks are more likely to reduce the performance of the wing than improve it.
Optimizing the performance of the wing end plates is the least important part of the entire vehicle’s aero package.

You might think I would take Kyle at his word. He worked for the Mercedes Formula 1 team as an aerodynamics engineer during the manufacturer’s most dominant years. But I’m also a stubborn, pig-headed ass who believes in getting his own data. So I took three end plates (four if you count that I turned one backwards) to the A2 wind tunnel and spent my hard-earned money to see if he was right.

I tested four wings in the wind tunnel, but when I got around to testing the end plates, they were all swapped onto a 55″ (1397 mm) 9 Lives Racing wing. This is the benchmark motorsports wing for many good reasons, so I figured why not go with the industry standard.

I first tested the basic rectangular end plate, which is made from an aluminum street sign that I simply cut in half, rounded the corners, and called it done.

Sorry about the image quality, these are stills from the video monitors in the wind tunnel.

I then swapped those for a popular CFD-designed end plate. I’ve always found this design to fly in the face of reason – why is there a big cut out right where the low-pressure region is?

CFD end plate with a pressure relief cut on the top, and a large radius cut into the leading edge.

I then turned the end plate backwards and tested that. From my point of view, it seems like the end plate might perform better with more surface area facing forward and less at the rear. It wasn’t a great fit, though.

The CFD end plate didn’t fit very well when I flipped it around backwards. I got two of the holes to match up and called it good enough.

Finally I tested an end plate of my own design. It’s in some ways the opposite of the CFD end plate, having a lot of surface area forward and tapering towards the rear. There’s a very small relief cut on the upper back corner that’s supposed to reduce a vortex there (er… so I’ve read). But more significantly, this end plate has a very small wicker on the trailing edge.

My Occam’s Racer end plate with more area forward and Gurney flap. This is the same end plate in the cover image.

Let’s see how the end plates performed:

End plate	cD	cL	Vehicle L/D
Rectangular	.467	-.382	.82
CFD	.475	-.386	.81
CFD backwards	.474	-.385	.81
Occam’s Racer	.480	-.398	.83

Coefficients of drag and lift with various end plates. cL is a negative number because it’s showing downforce; more negative is more better.

Wind tunnel data

So let’s unpack the coefficient data and translate that into more common figures, like pounds of downforce and horsepower consumed.

Rectangular – Vehicle L/D ratio .82
- Baseline to compare with other end plates
CFD – Vehicle L/D ratio .81
- +1.8 lbs total downforce
- 4.1 lbs drag = 0.4:1 L/D ratio for end plate
- +1.1 hp used from drag
CFD backwards – Vehicle L/D ratio .81
- +1.5 lbs total downforce
- 3.5 lbs drag = 0.4:1 L/D ratio for end plate
- +.9 hp used from drag
Occam’s Racer – Vehicle L/D ratio .83
- +8.8 lbs total downforce
- 6.9 lbs drag = 1.3:1 L/D ratio for end plate
- +1.8 hp used from drag

The first thing you’ll notice is that the CFD end plate increased downforce by almost 2 lbs at 100 mph. That’s more than the rectangular end plate, but not much. As a consequence of that additional downforce, there’s a bit more drag.

Turning the CFD end plate backwards resulted in less downforce, but also less drag. Overall, the CFD end plate performed the same forwards as backwards. Surprising.

Another surprise was that my Occam’s Racer end plates gained 8.8 lbs of downforce over the rectangular plate. Not surprising, this also resulted in more drag. However, these end plates resulted in the best vehicle L/D ratio.

Racing simulations

Those numbers are all very close, and you might be wondering how they affect the only thing that matters: lap times!

To find out, I put the coefficient of lift and drag values into OptimumLap and ran them around two race tracks, the autocross course from 2010 SCCA Solo Nationals, and Lime Rock Park. I typically use these two tracks because they are close in lap time, but are completely different with respect to speed. I’ve also included my local track Watkins Glen, because it’s very high speed and should spread the results out more. (My wind tunnel report shows lap time comparisons for every part that I test, as well as some useful combinations.)

End plate	cD	cL	Autocross	Lime Rock	WGI
Rectangular	.467	.382	61.79	61.03	134.74
CFD	.475	.386	61.79	61.05	134.81
CFD backwards	.474	.385	61.79	61.05	134.81
Occam’s Racer	.480	.398	61.78	61.04	134.81

Lap times

On the autocross track, my end plates won by a whopping .01 seconds. At Lime Rock, the rectangular street signs won by the same insignificant margin. At Watkins Glen, the “No Skateboarding” sign went .07 seconds faster than either of the fancy end plates.

Discussion

The CFD-designed end plates were a disappointment, and put an exclamation point on my rant at the beginning of this post about using free stream CFD. Look, I’m not at all doubting that these end plates worked better in CFD and returned exactly what they calculated. But overall performance didn’t change facing either way, and shows how useless free stream CFD can be in the real world.

The custom end plates I designed are based on a hunch that I should put most of the area low and forward. More significantly, these end plates have a small Gurney flap on the outer edge, and it’s likely that the shape of the end plate was no better than the others, and it was simply the addition of the wicker that gave this end plate the most downforce. Most downforce doesn’t mean best, as it wasn’t terribly efficient and only returned a 1.3:1 L/D ratio above and beyond the rectangular plate. Using these end plates would make the car faster on a tight track, but slower on most race tracks.

In the end, there really is nothing wrong with a rectangular end plate. If you don’t have the means of CFD testing your entire vehicle and optimizing the end plates to your entire car, then a rectangular end plate is your best bet. I make mine from street signs that I buy from my local metal recycler for $1 per pound. And so both end plates are less than $2 and come with amusing graphics. It astonishes me that people will pay much more for something that performs worse.

But let’s face it, no matter what you do to the end plate, it’s not going to make it significantly worse, either. End plates with cuts and vents are a great place for personalization, they are a source of many silly conversations, and they throw competitors off the scent. In the end, I say do whatever you want, it won’t matter much anyway.

But if you want to hedge your bets and do the least work, just stick to a rectangular end plate. Learning this lesson cost me $300 plus a lot of time and effort, and it proves I should just have just believed Kyle Forster.

But it wasn’t a total waste of time and money if that keeps other people from making the same mistake. So if this article saves you money, please consider buying me a coffee, or if you want 50+ pages of the same kind of data, then buy my wind tunnel report. It’s only through contributions that I can afford to do wind tunnel testing and continue the lord’s work. Thanks!

Postscript

After posting this article, I got a bunch of great comments on the Professional Awesome Technical Forum on Facebook. This group has a wealth of knowledge that surpasses my own, so please see this post, scroll through the comments, and benefit from the global knowledge base.

How Much Downforce Do Canards Make?

Canards, also sometimes called dive planes or dive plates, are little winglets on the front bumper fascia. They are mysterious pieces of aero, because the shape, size, curvature, and mounting locations are all over the map. There’s no standardization at all. What they do depends on a combination of factors, and also who you ask.

I really like the explanation of dive planes on the Verus Engineering site: “Dive planes, also referred to as canards, allow you to shift the aero balance forward, possibly aiding you and your setup in balancing out a large rear wing or diffuser. Dive planes also help seal the sides of the car and help evacuate air from the wheel well, further reducing lift and drag in some cases.”

I love how much conjecture is in that description, and if you read between the lines, you’ll see that they might not help your car at all. I would sum it up like this: “Sometimes canards will make your car faster.”

The reason canards might make your car slower is because they are at the leading edge of the car, so they affect everything behind them. Airflow down the sides of the car, underbody, wake, and especially the rear wing could all be negatively affected.

Canards don’t have an airfoil shape, and so they create downforce through pressure, rather than suction. In Race Car Aerodynamics, Katz examines dive plates, and gives them a Cl of .03 and Cd of .01. That 3:1 lift/drag ratio is like a spoiler rather than a wing, which makes total sense all things considered.

You might use canards under the following circumstances.

If you have an older car that has the front tires exposed to airflow (Miata, etc), then canards can help deflect air away from the tires.
If you have a car with a flat bottom, canards may help seal the sides of the car via a vortex, which may help the underbody creates more downforce.
If you have a splitter and aren’t using spats or other tricks to extract air sideways, then canards can be beneficial when placed low on the bumper fascia. I’m going to dig into this one in more detail.

DIY canards

The Velsoter N bodywork has a lot of vents, swoops, and hard angles, so there aren’t a lot of options on the size of the canards, their shape, or where to mount them. Effectively, the bodywork itself dictates the dimension and locations of the canards.

The vents and bumper fascia elements make it difficult to mount canards. There were only three logical places to put them.

I made my canards out of aluminum street signs, which I get at the local recycler for cheap. I shaped them to fit the bumper fascia and curved them so they would fit where they would. All of the canards are identical in shape and curvature.

The top canard mounts just below the headlight. The headlight has a flat spot below and in front of it, and by putting a canard below that, I essentially increased the amount of planform area for free. (More area means more area for high pressure to form.)

Upper canard alone fits below the headlight and utilizes the extra planform area of the bodywork.

I placed another canard 5” lower than the top canard. It sits on top of the air curtain duct on each side of the car, so there’s no way to mount it any lower. I’ve several aero companies put their canards in this general location, so I figured it should work there.

The canards are duct taped on, because that works fine for wind tunnel testing.

I also placed one canard down on the bumper fascia as far as it would go. This was an audible call made by AJ Hartman during the wind tunnel test. I hadn’t planned to put a canard there, but I’m glad I did, the results were astounding. This is why you bring a professional with you to the wind tunnel!

Middle canard moved to the bottom position.

Test data

Recall that all of the canards are identical in size and shape, and so the primary thing I was testing was the location of the canards: top, middle, and bottom. It’s important to note that the car has a splitter and a wing, because canards will interact with both.

In the table below, front downforce is listed as a positive value, while rear downforce is negative. This is the normal see-saw effect of pushing down on one side of the car, the other goes up. But also, there may be some loss of rear wing downforce via turbulence from the canards. The Total downforce is obviously the two added together, and this is important when we look at Lbs of drag.

The next columns are drag, and if you divide the total downforce by the drag lbs, you get the L/D ratio, or how efficient the canard is. The top canard is the most efficient at 3.91:1. Finally the last column is HP, which is how much power is consumed by the canards at 100 mph.

	Downforce @ 100 mph			Drag
	Front	Rear	Total	Lbs	L/D	HP
No canards	0.01	0.05	-0.04	-0.04	0.89	0.04
Top only	15.18	-2.17	13.11	3.35	3.91	0.95
Mid only	11.13	-1.91	9.21	3.30	2.80	0.88
Top and mid	26.30	-4.08	22.32	6.64	3.36	1.82
Bottom only	85.61	-17.76	67.95	28.98	2.34	7.78
Top and bottom	100.79	-19.94	81.06	32.33	2.51	8.73

The top canard was the most efficient, probably because it’s mounted below the headlight where the extra planform area effectively adds to the canard’s surface area.

The middle canard has exactly the same drag as the top canard, but because it didn’t have the extra bodywork area to work with, it made less total downforce. The 2.8:1 L/D ratio is quite close to what Katz cites. When I look at canards from various manufacturers, they usually put them in about this location, but you can see it’s the least effective of all.

Using both the top and middle canard together was simply additive. Meaning, the data doesn’t show any effect of the canards working together to create more downforce than either one did individually. It makes me wonder why canards are usually mounted in pairs.

Take a look at that bottom canard, it made over 700% more downforce than the canard that was just 8″ above it. Drag also went up by a lot, but that’s an acceptable tradeoff for this much downforce.

Finally, I added the top and bottom canards together to combine the better L/D ratio of the top canard with the high downforce from the bottom. This gives just over 100 lbs of front downforce, and when matched with more rear downforce, would provide a lot more grip.

Discussion

I didn’t expect much out of canards, and I’ve often slagged them off as poseur junk. I was wrong about that. If AJ hadn’t prodded me to try the canard in the bottom position, I’d have missed some very useful information.

Canards aren’t an airfoil shape, and don’t create suction underneath like a wing does. The top surface is theoretically limited in the amount of downforce it can create via pressure to less than cL 1.0. But since the surface area of the canard and the angle of attack was identical in each position, then the canard itself can’t be making more downforce.

Suction is really the only way to explain a gain of this magnitude, and it’s not the canard that’s doing the heavy lifting, it’s the splitter. The canard must be extracting air from the wheel wells, and/or from the sides of the splitter, which is creating more suction under the splitter.

As a consequence of that, the bottom canard also makes so much more drag. Drag is a normal byproduct of downforce, but the canards in either of the other locations were much more efficient than the bottom.

One possible explanation is that when placed this low, the canard creates a much stronger vortex. Vortices take a lot of energy to spin up, and a bigger vortex makes more drag. Or perhaps it’s simply that canards are an inefficient way to create downforce, and drag increases at a higher rate than downforce? In any case, the location is obviously a very important factor in canard placement, and I’ve only played with one variable at this point, height.

Sometimes you see a canard with end plates, or a vertical outer edge. That kind of thing might help a canard create more local downforce, because it should hold more pressure on top of the blade. But this may not be ideal for the lowest canard, as the whole point of that one is to spin a vortex off the outer edge and suck air out from below the car. Or it might work better. Further testing is required.

It’s worth noting that you shouldn’t apply these downforce numbers directly to your car, unless you have a Veloster N with the same splitter and wing. If I’ve learned anything from this test, it’s that canards are finicky. About the only thing you can conclude from this test data is that height matters. And that’s only one variable that’s been isolated, with angle of attack, size, shape, fore-aft location, and wicker edge yet to be determined.

Before I sign off on this one, I’d like to go back to my statement that “Sometimes canards will make your car faster.” AJ recently stated in the Professional Awesome Technical Forum that canards have improved the L/D ratio of every car he’s tested in the wind tunnel. Well this is because he’s a professional aerodynamicist! He knows where to place canards, and what angle to run them at. The average aero enthusiasts buying appearance-grade canards and placing them where they look cool is a performance crap shoot. Good luck with that.

If you enjoyed reading this article, check out my wind tunnel report. It’s over 50 pages of similar data, but goes over many more pieces of aero, and to a much greater depth.

Car Wing Comparisons

Updated 8/2023 to add Benzing wings and correct an assumption about the Douglas LA203A from Wingmen Aerodynamics.

Airfoil Tools is an amazing website, and was once the primary way I researched wings. When I look at wings, the most important factor to me is the coefficient of lift (Cl); I’m only interested in airfoils that have a Cl 1.5 or higher.

I also look at the efficiency of the wing, which is the coefficient of lift divided by the coefficient of drag (Cl/Cd). Wing efficiency is less important than downforce, because the only thing that matters is the efficiency of the vehicle, and that is typically achieved by using a wing with the greatest downforce. Even if wing efficiency isn’t that important, I wouldn’t use a really draggy wing, so I look for a Cl/Cd of 100 or higher.

So, lift should be at least 1.5, and efficiency should be at least 100. Those are nice round numbers, and easy to remember. I look at these two values at a Reynolds number (Re) of 500,000, because that represents a normal-sized wing at a realistic car speed. For a 9″ wing, 500k Re is 73 mph. For low speeds (or small wings) I’ll look at Re 200k, and for huge wings or really fast speeds, I’ll look at 1 million. But there’s really no reason to look outside of the 200k-1M Reynolds numbers and 500k is a happy medium.

I also set the turbulence value to Ncrit=5 because Airfoil Tools doesn’t allow me to set it any lower (in some cases there’s data for lower Ncrit numbers, but it’s rare.) The default setting in Airfoil Tools is Ncrit=9, and that replicates what the wing would experience in a wind tunnel, and doesn’t represent the turbulent nature of air flowing over a car, nor what’s happening in a race, behind other cars, with cross winds, open windows, and other factors.

The next important factor I look at is how the wing deals with stalling at high angles of attack. I look at this for two reasons. 1) Most cars will go faster with the wing set to maximum downforce rather than maximum efficiency, and 2) car roofs are often cambered and this means air hits the wing at different angles across its length.

For example, on my Miata, if I mount the wing at roof height, the ends of the wing are at 0 degrees, but the angle changes from -5 to -7 degrees along the roofline. If I mount the wing closer to the trunk, that angle increases to nearly 15 degrees. You can read about that and see a video Visualizing Airflow.

Given that air hits a wing at different angles of attack across the wing, at some point along its length, a wing may be stalling. When wings stall, they lose downforce and gain drag. Some wings do this gracefully, with a very gradual falloff in downforce, and others stall dramatically, with downforce crashing and drag spiking way up.

An obvious way around this is to use a wing with a 3D profile, so that air hits the wing at the same angle across the entire wing. You can read about that in my article on 3D wings.

Airfoil Comparisons

Now that you know my criteria for looking at wings, in the rest of this post I examine different wing profiles and give my thoughts on them. I’ve ordered these by Cl, which is how much downforce they make, At the end of this post I’ve included a table with summary values and some parting thoughts.

With all of that front matter and grey matter out of the way, let’s check out some wings!

Clark Y

The Clark Y airfoil (airfoil tools, wikipedia) is distinguished by a flat bottom, or when used upside down on a car, a flat top. Most wings are cambered on both sides, but the flat surface can make it easier to manufacture, and for an airplane, it’s good for training because it has gentle stall characteristics. But as a car wing, the flat top means more drag and less downforce than a wing with a cambered topside. As such, the Clark Y doesn’t quite meet my downforce threshold of 1.5 Cl at 500k, and the efficiency (Cl/Cd) is below 100 as well. Personally, I wouldn’t use this wing, but it would be easy to build.

NACA 6412

MacBeath often cites the NACA profiles as examples in his books, and for good reason, they are easy to understand. The first number in 6412 means percent of camber relative to chord (6%), the second number is where the camber occurs (4 means 40% of the chord), and the third number is thickness (12%). NACA 6412 meets my criteria for a good car wing with a Cl over 1.5 and a Cl/Cd over 100.

I like NACA profiles for another reason: they allow me to change the variables and see what happens to lift, drag, and efficiency at different Reynolds numbers and angles of attack. For example, I’ve read that maximum lift on single-element wings occurs at 12% thickness, and after experimenting with different NACA profiles that are identical in other respects, I know this to be true.

You can also use the NACA 4-digit generator to create your own wing profiles. For example, this is a NACA 9512. I’ve maxed the camber allowed in the tool (9.5%), set the camber further rearward (50%) to increase lift, and used the max lift thickness of 12%. I’m certain this would be a good car wing.

Cambered Plates

Probably the easiest way to make a wing is to cut a metal pipe lengthwise into strips, and then lay two of the curved pieces on top of each other. Put a semi-circular nose on it and weld the three pieces together. This is a cheap and easy way to make blades for small wind turbines, I don’t see why you couldn’t do the same for a car wing. It’s so DIY I want to make one for 24 Hours of Lemons.

You can make these wings in different thicknesses, and at 12%, it has a high Cl of 1.7. However, this wing has a lot of drag and so the efficiency is quite miserable, less than half of my threshold value of 100. Still, for a low-speed wing where drag is inconsequential (autocross), this would totally work. And for 24 Hours of Lemons, it’s better than a snowboard, skateboard, angled plywood, etc.

LNV109A – NASCAR COT Wing

NASCAR flirted briefly with car wings, but after 93 races went back to spoilers.

The airfoil they chose was the Douglas/Liebeck LNV109A high lift airfoil. I don’t know what decisions went into that choice, but the numbers don’t excite me. I mean, it would be cool to have one for historical reasons, but I’m not actively searching eBay for one.

If you look at the Cl vs Alpha chart (which you can think of as how much downforce the wing makes at different angles of attack), you can see the wing has a max Cl of 1.7. Downforce peaks at 12 degrees and then falls off drastically. Compare this to the Clark Y above to see what I mean.

Next, take a look at the efficiency of the wing. Most wings have a bell-shaped curve with maximum efficiency in the middle of the range (say 5 degrees). What’s interesting here is that max efficiency occurs at 10 degrees, with a Cl/Cd is 117. Maybe that’s why NASCAR chose this airfoil shape? If you want a wing that creates close racing at high speed, this would be a good choice. But if your agenda is low lap times, I think there are better choices.

FX 72-MS-150A

I have three different made-in-China wings, one came as a double wing, the other two are single wings. They are fun for experimenting with, cheap, and disposable. Whoever designed them chose a similar profile for all of them, which is akin to the Wortmann FX 72-MS-150A.

Some of these wings are sold as “universal” and so they are flat on the bottom with two mounting rails underneath. I modify these by adding wood to the bottom and rounding it.

By the numbers, this is a decent airfoil for a car, it makes a lot of lift (1.8) and is very efficient (121). The only drawback is when this wing stalls, it falls out of the sky. This isn’t a wing that you want to set for maximum downforce. Make sure you take into account the downwash angle on the roof.

Douglas LA203A – Wingmen Aerodynamics

Wingmen Aerodynamics makes a wing based on this profile. I originally thought it was a Clark Y, but was corrected in the comments section. I’ve seen this wing at a couple different races, and I was immediately impressed with the build quality. It looks very light and feels stiff and strong.

With a max thickness of 15.7%, the wing is on the thicker side, but oddly, it also has less topside camber than most motorsports wings. At around 1.75 Cl, the LA203A, doesn’t make the most downforce, but it’s the second most efficient wing in this article. It also has a very gentle stall characteristic, and should work well behind a cambered roof line, or lower on the trunk lid where the center and ends of the wing are at very different angles of attack. Nice choice, Wingmen.

GOE 464

This is a very thin airfoil, almost potato chip in profile. The only reason I find this aIrfoil interesting is because APR makes a carbon fiber wing using a similar profile. APR’s GTC-300 wing has more camber, but the GOE 464 is close enough to look at some numbers.

The GOE 464 has a max Cl of 1.85, which is a lot of downforce, and the efficiency at 500k almost reaches 100. It’s an interesting wing, but the numbers aren’t blowing me away. Also, it’s so thin that it would be difficult for me to build and make rigid enough, and I feel there are better choices.

Better airfoils

The previous airfoils were all interesting in one way or another, and some can be bought or made pretty cheaply. But all of the following airfoils have superior numbers, and I feel would be better choices for a car wing. Choosing between them is difficult, as there’s always a trade off between lift, drag, and stall. Each airfoil has a niche where it outperforms the others, and I’d be happy with any of them.

Church Hollinger CH10

Any wing that makes around Cl 2.0 is in the category of ultra high lift. The most efficient of these is the Church Hollinger CH10. At 500k, this wing has a Cl/Cd of 132, which blows away the others.

*Get thee to church*. Church Hollinger, that is.

If you look at the data, peak downforce is around 10 degrees. Max efficiency (Cl/Cd) occurs at around 3 degrees, which is where I’d set the wing on a Miata. This would put the angle of attack at 8-10 degrees over the roof, and 3 degrees at the wing ends. Perfect.

I believe that the Wing Logic extruded aluminum wing is a CH10, and if you think it looks a lot like the 9 Lives Racing wing, it’s because they are close cousins in shape. Both airfoils have their maximum thickness at 30% of chord, and their max camber at 50% of chord (measured from the nose of the wing). The difference between them is the 9LR wing has two degrees more camber and 2% more thickness.

GOE 652

The first thing you notice about this wing is the thickness of the leading edge. It carries that thickness over much of the wing, and the 17% chord thickness is unusually phat. The purpose of that (I think) is to keep air attached at steep angles of attack.

The 652 has a very gradual stall, and should tolerate being set at too steep of an angle. For this reason, I believe this wing would be a good for a car with a highly cambered roofline, or where you have to mount the wing closer to the trunk. In both cases, there are large changes in apparent wind angle across the wing, and this wing won’t care as much as other wings.

This isn’t a particularly efficient wing, and even though it exceeds my threshold value of 100, it does that at a very low angle of attack. However, the high lift of Cl 2.0 puts this wing into elite company. I’d wager this would make a good upper element for a dual-element wing, not just the because of the shape, but because the added thickness would make it stiffer in a smaller chord. I’m definitely making one of these someday.

Eppler 420

The Eppler 420 isn’t as efficient as the CH10, but has slightly more downforce and a gradual stall. It’s a good all-purpose shape, and because it’s thicker, would be a strong contender for either element in a dual-element wing. As an all-purpose wing, it’s hard to choose between the CH10 and E420. The former is more efficient, the latter makes more downforce.

It’s also a pretty good wing for low Reynolds (low speed or small wing). The Porsche Cayman R has a tiny rear wing, and it’s probably not a coincidence that the profile looks a lot like the Eppler 420.

Porshe Cayman R wing looks like an Eppler 420 at 5 degrees.

Wortmann FX 74-CL5-140

This airfoil wasn’t on my radar, but in writing this article I looked at every airfoil on the Airfoil Tools site. Glad I did, this one is a keeper! With a Cl that’s nearly the same as the Selig wings, and an efficiency closer to the CH10, this wing sits in rarefied air.

You don’t get something for nothing, and the tradeoff is a steep drop when it stalls. If you want to go after maximum lift with this airfoil, mount it high where the angle of wind doesn’t change much. I’m not sure what’s happening around zero degrees, but I wouldn’t set it there anyway.

I have more to say about this wing after reviewing the next wings. I’m tempted to build one, so stay tuned on that.

Selig 1223 and 1223 RTL

The Selig 1223 and the Selig 1223 RTL are Downforce Royalty. The RTL version is slightly thicker, which results in higher lift and drag. The RTL can be set to 15+ degrees and approaches a Cl of 2.5. That’s huge.

Both airfoils make a lot of downforce, but also a lot of drag, and their Cl/Cd efficiency is less than 100 at all angles. Ergo, I would use this airfoil for low speed or for a car with a lot of power. Those are also usecases for a dual element wing, which might be a better choice if your racing rules allow that.

AeroDesign wing from Australia appears to be Selig-ish.

Let’s compare the two Selig wings to the FX74. These graphs are from Re 1M because I plan to use these for a larger chord wing. In the comparisons you can see how the S1223 wings are clear winners in downforce, but the FX74 is far more efficient at Cl 2.25 and below. I’ve drawn a blue dashed line at 2.25 Cl, and you can see that the FX74 has a sweet spot where’s it’s making a lot of downforce without much drag.

This is how five of the airfoils stack up at 500k Re. I didn’t include the RTL because that’s in the previous charts. I’ve pointed out a few areas that differentiate one wing from another.

Motorsports wings

All of the previous airfoils are aviation wings that have been adapted to motorsports use. Cars don’t (or shouldn’t) fly, and they have much lower Reynolds numbers, and so it makes sense that car-specific wings would be shaped differently.

Indeed, Enrico Benzing did a bunch of research and came up with a lot of different designs specifically for motorsports use. None of these would be particularly good at flying, but they are better choices on cars.

Similar to the NACA wings, the Be-series of numbers tell you about the shape:

The first two digits are the thickness of the wing, relative to the chord. Aviation wings make maximum downforce around 12% thickness, but cars wings are often thicker.
The next number is the location of the thickness. Many of Benzings wings have the maximum thickness at 20-30% of the chord. This is supposed to add more downforce.
The first two numbers after the dash are how much camber the wing has. If you use the Airfoil Tools NACA generator to make your own airfoil shapes, you can’t enter anything higher than 9.5%, and Benzing’s wings are all more than that.
The final number is the position of the maximum camber, in tenths. More rearward camber generally means more downforce.

Be 183-176 – TCR Wing

As I stated in the NACA wing details, downforce typically increases as you increase thickness, up to about 12%, at which point that trend reverses. But that’s not true for motorsports wings, but even so, this is a very thick wing, the chubbiest in the Be series. This is also extreme camber, I haven’t seen an airfoil with this much.

The TCR regulations state this wing must be 1380mm x 250mm (54.3″ x 9.8″) and you’re not allowed to use a Gurney flap. From the wind tunnel testing I’ve done, I can tell you that a wicker would absolutely help this wing, maybe to the point where it works.

Make your own wing using these coordinates.

Knowing what I know about wings, I wouldn’t use this one unless I was forced to. Unfortunately, like the LNV109A was with NASCAR, this wing is a homologated part, and all the TCR cars are forced to use it. Worse still, you need to mount it (including brackets and endplates) below the roofline. On a hatchback.

My buddy Josh and I inspected one of the hatchbacks at a wet Grid Life race, and the water left streaks that were like flow-vis paint. You could clearly see where the air was separating on the back third the wing. In my opinion, this airfoil is too thick, it has too much camber, and air can’t stay attached along the entire surface, causing a loss of downforce and increased drag. I could be totally wrong about the TCR wing, and may have to eat my words. One day I’ll test one in a wind tunnel to be sure.

I wrote some additional comments about this wing in my Grid Life recap, and I’ll copy the most salient one here. “Leave it to the people that fucked our asses with diesel-gate to bugger a racing series with a wing that stinks of shit.”

Be 123-125 – 9 Lives Racing Big Wang

Probably the most popular motorsports wing is the 9 Lives “Big Wang”. The shape is essentially a Be 123-125, but with a modification to the rear geometry to support a Gurney flap. The shape is conservative compared to other Benzing wings, and measures in with numbers closer to aviation wings. Never mind that, because it’s a fantastic motorsports wing.

Big Wang with Gurney flap slot cut off for experimentation with chord.

In my wind tunnel testing the Big Wang was second best to the MSHD, and outperformed my S1223. This surprised me considering the amount of research I’d read on low-Reynolds high-lift airfoils; I predicted the S1223 would be a clear winner. Nostradamus I ain’t.

In a research paper on multi-element wings, this shape (or rather a very close cousin of this shape, the 122-125 with thickness 10% further forward) was the best performing main element. So if you are looking at a dual-element wing, this is the best off-the-shelf choice. Or even if you’re looking for a single element wing, it’s pretty hard to beat this on performance.

Be 183-155 – Procar Innvoations

Procar Innovations makes a wing that looks like the TCR, but it has less chord and is less extreme. I measure it up at 183-155, meaning it’s just as thick as the TCR wing, but it has less camber and the location of the camber is 10% further forward.

I tested this wing in the wind tunnel and it was really good, on par with the 9 Lives Racing wing. The end plates fasten with two 8mm bolts instead of four 5mm bolts, and could easily support a double element. The PCI wing doesn’t have a slot for a Gurney flap, but I would rivet on a wicker. My wind tunnel testing confirms that wings with a lot of camber really like a wicker, and this one gained 165% downforce with a 1/2″ piece of angle aluminum taped on top.

MSHD

I didn’t know about this airfoil until very recently, as it doesn’t show up in Airfoil Tools, Bigfoil, Enrico Benzing’s Wings, or any other text I own. Because of that, I can’t display the same charts, but I can describe some of the advantages.

There’s a scientific paper on this airfoil making a direct comparison to the S1223 for use in low-speed motorsports applications. The main advantages of the MSHD over the S1223 is an even higher Cl value, softer stall characteristics, and a large range of AoA. To my present knowledge, this airfoil makes the most downforce of any wing listed here. If the Selig wings are the royalty in a deck of cars, then the MSHD is the ace of spades.

Shape-wise, the airfoils are not that much different, but the MSHD has more camber. It almost looks like someone took the S1223 and kept pushing the middle downward until they got flow separation. To plot the wing you can use the values in the spreadsheet.

In my wind tunnel testing this wing produced the best numbers. It also did very well with a Gurney flap. I don’t know of anyone making these for sale, but it was fairly easy to build in foam and fiberglas. I did a DIY writeup on that in another article.

Summary Data

Here are all the wings sorted by cL. For each wing, I’ve also listed the max efficiency, and the angle where that occurs. Note that the fastest way around the track is often right around max downforce, not max efficiency. But efficiency is somehow important for marketing, even if that doesn’t really matter.

Wing	Max cL	Max cL/cD	Notes
Clark Y	1.45	90.2 at α=4°	Clark why?
NACA 6412	1.6	111.3 @ 6.25°	Good reference
Cambered plate	1.7	42.5 @ 5.75°	24 hrs of Lemons
NASCAR COT	1.75	113.7 @ 10°	Wing of yesterday
Douglas LA203A	1.75	127.7 at α=6.5°	Wingmen of today
MIC FX 72	1.8	121.1 at α=6.75°	MIC-Wing
GOE 464	1.85	97 at α=7.5°	Potato chip
CH10	1.95	132.4 at α=3.25°	Effin efficient
GOE 652	2.05	102.8 at α=2.25°	She thicc
Eppler 420	2.1	106.1 @ 6.25°	Dude, 420
Wortmann FX 74	2.25	115.5 at α=6.75°	Warts and all
s1223	2.3	97.4 at α=5.75°	DF Queen
s1223 RTL	2.35	85.4 at α=5.5°	DF King
BE 183-176	?	TBD	TCR shite
Be 123-125	>2.35	TBD	9LR
Be 183-155	>2.35	TBD	PCI
MSHD	>2.35	TBD	Ace in the hole

cL and cD at Re 500k, Ncrit=5

How Downforce Affects Tires

My twin brother wrote an article on You Suck at Racing about tire grip, and I’m going to steal some of that content to explore how tire grip and aero are related.

Braking, cornering, accelerating: everything depends on grip. Understanding how rubber tires create and lose grip is therefore fundamental. Let’s start with some theoretical laws of friction.

Amonton’s First Law: The force of friction is directly proportional to the applied load. If this law is true, then a 4000 lb car should stop in the same distance as a 2000 lb car. It weighs twice as much, but it also experiences twice as much friction, so theoretically, the weight of the car doesn’t matter.
Amonton’s Second Law: The force of friction is independent of the area of contact. This means that it doesn’t matter how wide your tires are. Skinny or fat, they have the same amount of grip. And grooves wouldn’t matter either.
Coulomb’s Law of Friction: Friction is independent of velocity. Which means you should have the same amount of grip at all speeds.
Finally, static friction is always greater than kinetic friction.

You might not believe these laws, because you’ve experienced that tires don’t really follow these laws of friction.

If weight doesn’t matter, then why do lightweight cars like Miatas out-handle bigger cars?
If tire width doesn’t matter, then why are wider tires faster? And by the same logic, given the same amount of rubber area, why are slicks faster than a tire with grooves?
If static friction is always greater than sliding friction, why is it faster to have slip angle through a corner?

Four important graphs

In order to understand how tires work, you have to understand the following four graphs. Introductory physics assumes that the coefficient of friction, Mu (μ), is a constant, and that may be true for a block sliding against a table top, but when it comes to tires, μ is not a constant.

Tires generate grip from molecular adhesion, mechanical keying, and hysteresis, and those factors are based on a combination of variables. In each of the graphs below, the coefficient of friction, Mu (μ), changes due to load, temperature, speed, and slip angle:

Graph A shows μ as a function of load (weight). If doubling weight doubled grip, then the line would be flat. But when you double the amount of weight on a tire, there are diminishing returns. When cornering heavily, the outside tires experience more load, and because of that, heavier cars lose more grip than light cars.

Graph B shows μ as a function of temperature. Every tire has an optimum temperature. Both cold and hot tires have less grip than one in the optimal range. If your tires are too wide, they may never get up to optimal temperature, and a narrower tire may heat up more favorably. For this reason, wider isn’t always better.

Tire Rack did a great test where they tested a bunch of wheel and tire widths. The fastest tire wasn’t the widest. And when they went to a wet track, the fastest lap was the narrowest tire.

One thing that contributes to heat is grooves. Squirming tread blocks are a major source of heat. As a result, grooved tires heat up more quickly than slicks. One reason for using slick tires is to spread the load better, but an even more important one is to prevent the rubber from overheating.

Graph C shows μ as a function of speed. The faster the car goes, the less time there is for keying; the ability for rubber to change shape and interact with the road. Under wet conditions, where adhesion no longer applies, grip is highly affected by speed.

Graph D shows μ as a function of slip angle. Every tire has an optimal slip angle. When a tire is twisted, which it always is to some degree, some parts of the contact patch are experiencing static friction while others are kinetic. This mixed state isn’t really addressed by any of the laws of friction, but it doesn’t make this any less real.

Ian made a drawing of what is happening between the road and the surface of your tire, which can help you further understand how tires grip.

Visualizing grip

The following image shows the surface of the road as a jagged green line on the bottom. (If the road surface was perfectly smooth, then the line would be horizontal. But because asphalt has imperfections with peaks and valleys, the road surface is represented as a jagged line.)

Panel A represents a tire (squiggly line) at rest, pressing into the surface of the road (jagged line).

Panel B is what happens when you add load: the rubber goes deeper into the surface, creating more grip. But there’s only so far you can push the rubber in. This is why doubling the load on a tire doesn’t double its grip. Panel B could also be softer rubber or hotter rubber. In both cases, the rubber conforms more easily to the surface, and with more contact, you get more grip.

Panel C shows what happens at high speed. The rubber doesn’t have as much time to change shape, so it doesn’t deliver as much grip from keying.

Panel D shows what happens when a tire overheats. The rubber comes apart, providing less contact with the surface. If the rubber gets hot enough, it may liquify or sublimate, creating a slippery layer between the surfaces.

This visual model isn’t perfect, as it doesn’t give why slip angle matters so much. But hopefully it helps you visualize the interface between your tires and the road, and why some factors add grip, and other factors take it away.

Aero and tires

So that’s how rubber grips the road, but there are other dynamic factors at play here, namely the aerodynamics of the vehicle.

Most cars without aero lift at speed, because the curved surface of the top of the car is longer than the bottom. In other words, cars are shaped like airplane wings, and like wings, they generate lift. The higher the air speed, the less traction there is.

Most cars have a coefficient of lift of around 0.3. Cars with a lot of curvature, like a fastback, have more lift than a three-box sedan or hatchback. By nature of their shape, most cars generate more lift over the rear tires than the front tires.

Cars also generate lift when in yaw, so if your car is pitched slightly sideways in a corner, it has even less traction. Nissan did some tests on this and found there is a fairly linear relationship between yaw angle and lift, and so the more sideways you get, the more the car lifts.

This means as you corner faster and faster, your rear tires have less and less rear grip. You already saw in graph C and panel C that tires have less grip at speed because they have less time for keying. So if you combine the keying losses with the lift and yaw losses, you get a car that’s lost a lot of grip on the rear tires.

And this is why it’s so important for race cars to have spoilers and/or wings.

Aero and lap times

To put some numbers on it, let’s take a look at a few NASA classes at Watkins Glen.

NASA Spec3 is a class for stock-bodied (no aero) E36 BMWs on Toyo RR tires measuring around 14.5 lbs/hp. The Spec3 lap record is 2:13.6.

NASA also has the ST5, class, which is a similar lbs/hp ratio to Spec3, but allows a splitter and wing. The ST5 record is 4.3 seconds faster: 2:09.27.

NASA also has a time trial class, TT5. TT5 and ST5 are the same formula, but the ST5 cars are racing wheel to wheel, whereas the TT5 cars are in a time trial situation with less traffic. Therefore, the TT5 laps are usually faster, but in this case, a surprising two seconds faster: 2:07.202.

If you compare the Spec3 lap record to ST5/TT5, can see that aero (splitter and wing) are worth about 4-6 seconds at Watkins Glen. Let’s call it 5 seconds for simplicity. This isn’t a difference in tire grip, as most cars are on Toyo RRs, but some cars are on Maxxis RC1 for the same lbs/hp (or Hoosiers at a significant penalty to lbs/hp).

Let’s take a look at where the difference is. On the front and back straights, there isn’t a huge difference in top speed, so the cars are pretty similar in lbs/hp. In Turn 7, where aero doesn’t make much difference, the minimum corner speeds (vMin) are pretty similar, and so we’re looking at cars with equal grip, as well. But take a look at Turn 10, the aero cars are going about 10 mph faster!

	Lap	Front	Back	T7	T10
Spec3	2:13.6	121	126	62	87
ST5	2:09.3	118	1277	63	96
TT5	2:07.2	123	128	62	98

Now I’m making some pretty big assumptions on driver skill and track conditions being equal. So I’ll do some simulations in OptimumLap, and see if the computer world agrees with the real world.

I’ll start with the Spec3 car using drag and lift values of .44 and 0.3 which is probably in the right area. With these values I get a lap time of 2:13.82 which is pretty close to the Spec3 lap record. I’ll then add a splitter and wing to bring the Cd to .47 and Cl to -0.8 which are pretty fair values for the added aero. Doing only these aero mods on exactly the same car, I get a 2:08.84 lap, which is right in the middle of the 5-second delta we saw in the real world.

So that’s a pretty good verification of aero being worth about 5 seconds. So next I want to take a look at Turn 10 and see how much aero helps here, and if there’s really a 10 mph delta.

Turn 10 Watkins Glen

For the OptimumLap simulations, I’ll use four cars instead of two cars to get more granularity. I’ve given them the same tires with 1.2g of lateral grip, but different aero packages.

No aero – This is the Spec3 car with a coefficient of lift of 0.30, and is represented by the red line.
Zero lift – This car has some minor aero like a spoiler, which cancels out lift, and so the Cl is zero. This is represented by the orange line.
Mild aero – This is the kind of lift and drag you’d see from a spoiler and airdam done professionally (NASCAR level). This is represented by the blue line.
Good aero – This is a car with a splitter and wing, and is represented by the green line.

I’ll examine the grip in the middle of Turn 10, at the 16000′ mark from the start/finish line. This is not quite the minimum speed in the corner, but shows a high lateral load and is as good a place as any to look at G forces. There are a lot of spikes in the graph (like you’d see in Aim Solo data), so imagine it’s more of a smooth arc.

Starting at the bottom of the graph, notice that the car represented by the red line is pulling only 1.14 Gs. Recall that I gave all the tires the same 1.2g of grip, but because of aerodynamic lift, they are losing traction at speed. This is a normal situation for a street car or spec racer with no aero. What the simulation doesn’t show is that most of the grip is lost on the rear tires.
The orange line is a car with a spoiler, which mostly cancels out lift. Lateral Gs are very close to the the static 1.2g value.
The blue line has more than 1.29g grip because tire load is increased with downforce.
The green line is even more dramatic, with 1.4G grip. This is significantly more grip than the car with no aero.

You might be wondering how an increase in lateral Gs plays out in speed through the corner, which is the next graph. I’ve chosen the same 16000′ spot on track to measure the speed, and you can see it’s a difference of almost 10 mph between fastest and slowest.

Turn 10 is a very fast corner, and you’d see a smaller delta on a slower corner, but this is still pretty remarkable. By increasing the load on the tires, tire grip went from 1.14 to 1.4 Gs, and the car with good aero went about 10 mph faster in the middle of the corner.

Downforce and tire wear

Most people imagine that aerodynamic downforce will make your tires wear out faster. More grip = more wear, right? No. Oddly, downforce makes tires last longer.

Tires wear by abrasion; from sliding or spinning. Have you ever flat spotted a tire? Then you know that sliding a tire can wear it out in a couple seconds! Aero increases the load on the tire, giving you more grip, which makes it less likely to spin or lock up.

Aerodynamic downforce also loads the tires more equally. When cornering, the outside tires get loaded more, as a normal byproduct of mechanical grip. However, aero loads are based only on air, and is balanced across the car, left to right, helping to balance the car better.

One could even imagine an active aero device that would split the wing in half and only load the inside tires. Or rudders or vanes that help the car turn using air alone, and use the tires even less.

But let’s jump back from fantasy land… in reality, you get more aerodynamic downforce from rear aero than front aero, and this helps a lot in braking zones, increasing rear load and rear grip. The same is true in acceleration, and more rear grip reduces tire wear (on rear wheel drive cars).

Another way aero increases tire life is when you drive under the limit of the tire. For example, take Turn 10 at Watkins Glen again. Miatas can usually go through at full speed or with a slight lift on entry, and that’s because there isn’t much of a straight between Turn 9 and Turn 10 and Miatas are dog slow. If you’re going through at 85 mph without aero and on the limit of traction, you’ll go through it with aero at the same speed, but well under the limit, and you’ll wear the tires less.

The uphill esses are another place where a Miata is flat out while cornering. You can’t ever reach the limit of lateral grip because the car can’t accelerate fast enough to get there. So you slip less and use less tire.

Now this isn’t going to be true at every race track, most of the time when you have a higher limit, you fuggin use all of it. But sometimes there’s a corner or two where aero now puts you under the limit, and in a long endurance race, this can be the difference between changing tires mid race, or simply saving money. The point being, aerodynamic downforce can make your tires last longer if it keeps your car from sliding.

A good real-world example is the 2000-2002 Corvette SCCA cars, which went from a 315 rear tire in 2000 to a 275 tire in 2001. To increase grip, a rear wing was added in 2001, but it wasn’t enough and tires would only last about 4 laps before starting to go off. After optimizing the aero in a wind tunnel to create more downforce, the same tires in 2002 would last an entire race.

Aerodynamic balance

At this point I’ve only looked at how aerodynamic downforce affects grip and longevity due to increased tire load. But there are other aero factors at play that are important.

Earlier in this way too long blog post, I by mentioned that cars without aero lose rear grip from lift and yaw. In truth, the car loses both front and rear grip, but it loses more rear grip. As a consequence, as the car goes faster, it transitions more and more to oversteer. Most people find this an unsettling situation.

Personally, I don’t mind if a car oversteers at low speed, in fact, I like it. But if it does that at high speed, it scares the shit out of me. Ideally, I like a car that rotates easily at low speed and then transitions to understeer at high speed. This is ridiculously easy to do: add a big wing, and then tune the amount of understeer by adding or removing wing angle.

This is the magic of dynamically balancing the load on your tires using downforce. It’s so easy, and it’s so tunable. However, rear wings and end plates also increase stability by increasing the static margin.

Static margin is the distance between the center of gravity and the center of pressure. Anything that adds rear drag increases the static margin, sort of like streamers on a kite tail. In addition, horizontal areas on the back off your car, like big end plates, shark fins, or even bodywork like a hatchback or station wagon, increases the static margin through sideways resistance to air.

A greater static margin makes the car harder to turn, but also makes the car more stable. When a car goes over the limit of grip, the driver must make steering corrections. Cars that have a higher static margin require fewer steering corrections to bring the car back into line, which is easier on the driver. This also ends up being easier on the tires, and can make the car faster, if a bit less exciting to drive.

Conclusions

Tire grip is arguably the most important factor on a race car. Understanding how tires make grip is therefore one of the most important things a racing driver needs to know. Aerodynamic downforce can greatly influence the balance and grip of the tires at different speeds, and can be used for tuning the car’s handling and ultimately make the car turn faster lap times, and/or stay on track for a longer duration.