Blog

It’s been quiet on the Occam’s Racer site for a couple months. Most of that is the aftermath of knee surgery. Meds, physical therapy, lack of sleep, and persistent pain, have kept me away from my computer entirely. On top of that I tore my rotator cuff (downhill mountain bike injury, just before knee surgery), and I am going under the knife again in December. This has also compounded my inability to sleep, which makes writing or doing anything quite difficult.

Boo hoo, so what else is going on? Falconet.

The Hayabusa is a Japanese falcon, and many people know that name from the legendary motorcycle that was inspired by it. The smallest bird of prey is called a falconet, and that seemed like an apt moniker for a Miata with a Hayabusa motor. That’s right, I’m not getting out of Miatas after all, I’m putting a motorcycle engine in my race car and going after track records.

Mike Smith of Spec13, who is the engineer responsible for Hayabusa swapped Miatas (and soon for other cars), is the guilty party here. I was completely ready to sell everything Miata related and live the easy life with my Veloster. But after being a passenger in his test mule (1600 lb Miata with Busa swap), and doing a little napkin math, I’m back in.

The Hayabusa engine and transmission are about 200 lbs total, and this should get my race car below 1900 lbs. More importantly, there’s a lot of empty space in the engine compartment, which will allow me to do some unique things with front aero.

I’m fairly confident that Falconet will make more front downforce than any Miata in existence, and that means I can also add a shit ton of rear wing. Downforce in a lighter car means more grip, because grip gains are relative to the weight of the car.

Let me explain. Let’s say you have a modern track car (BMW, Camaro, Mustang) that weighs 4000 lbs with the driver, and you add 400 lbs of downforce. This will get you nearly 10% more grip, because grip goes up fairly linearly with weight (not exactly, though, see How Downforce Affects Tires). Now add the same amount of downforce to a car that weighs 2000 lbs, and you get almost 20% more grip. Lightweight cars benefit more from aero than heavy cars, and that’s where Falconet will shine.

It won’t be the fastest car at horsepower tracks, but at places like Lime Rock and Nelson Ledges, where there are long, high-speed corners, I expect Falconet to swoop in on some lap records. And on low speed tracks, where it’s difficult for big V8s to get all the power down, I imagine Falconet can prey on heavier cars as well.

Personally, I think I can do a record lap at Pineview, but probably not any other track. Not only do I know Pineview very well, but I’ve kind of lost my nerve on big tracks. But Mike Smith is a very fast Spec Miata driver, and Occam’s Racer teammate Alyssa Merrill has uncanny speed herself, and being all of 120 lbs, she’s probably the one who will break hearts and records.

Alyssa’s job also includes adding BMW-sourced antilock brakes, an onboard telemetry system, and general chassis setup. I’ve driven her car, she knows what she’s doing far better than I do. She also has mad skills in 3D design and will be making custom aero pieces like canards, wings, and vortex generators (kidding!).

Mike’s job is, of course, building a Busa engine and fitting it to the car. But not just any engine, it’ll be a stroker. He’ll also install paddle shifters, reverse gear, a taller drive ratio, and anything else required in the drivetrain.

My job is providing the car and doing all the aero. The top will be my fastback version 2, which is a single piece that connects from the front windshield frame to the trunk. The previous version of my fastback reduced drag by 20% and added 30% more rear downforce. Some of this can be attributed to improved backlight angle, but even more might be the width at the B pillars.

The extra width of the factory hardtop, or any street-based hardtop that has to seal the side windows, is wider than the windshield frame. Therefore, air gets shoved into the cabin, and causes downstream turbulence and drag. This effect is compounded when the car is cornering. Unfortunately every fastback I’ve reviewed, from 3D printed to curvaceous fiberglass, has the same problem at the B pillar, and will never perform as well as mine.

One of the things I discovered in my wind tunnel test was how to reduce the effect of open windows, and I expect fastback v2 to be even better, as it has some new tricks to further reduce drag and turbulence. First is a reduced window opening at the top edge, which is where most of the turbulence is. I also added a drip edge here, but the purpose is really to keep air from curling inside. Finally, the B pillar is well rounded for smooth extraction.

I don’t know exactly what the front end will look like yet, but I’m taking a lot of inspiration from the Fittipaldi EF7, with it’s front wing molded into the front bumper, and the extractor hood.

This wouldn’t be possible on a normal Miata, because the engine and radiator take up too much space. But look at the Busa engine and how much room there is in front for an extractor radiator and some kind of wing arrangement below that.

You’ll notice how far back the engine sits, and this, coupled with the lack of engine weight, means the car will have more weight on the rear wheels than the front. I’m not sure how this will affect the handling, and this might require non-standard spring rates. There will be a lot of experimentation, but that’s not a problem because we have unlimited testing days at Pineview Run.

But don’t expect rapid progress on this project from my side, I’ll be in a sling for 6 weeks come December, and won’t be able to life more than 5 lbs until next summer. But I have a couple partners in this project, and they can make a lot of progress while I’m on the mend.

So that’s what’s going on over here, big plans to smash some lap records. Falconet FTW.

Grid Life Touring Cup (GLTC) has rules for three different size wings:

• Small – You can use a small wing or spoiler that measures under 250 square inches, and there’s no lbs/hp penalty to using one. You can buy an extruded 135cm wing on eBay for $60 or make yourself a spoiler. I tested both in the wind tunnel, and it’s definitely an advantage over using no rear aero at all.
• Medium – If your wing measures under 500 square inches, your car takes only a 1% penalty to lbs/hp ratio. The easiest solutions are a 54” 9 Lives Racing wing, or 55” Procar Innovations. I tested both wings in the wind tunnel, and they are both solid choices.
• Large – The maximum size allowed in the GLTC rules is 701 square inches, and incurs a 3% penalty to lbs/hp ratio (or 4% when used with a splitter). There are a number of quality 2D and 3D wings to choose from that are class legal.

The focus of this article is only the mid-sized 500 square inch wings. There are already a couple good aftermarket wings to choose from, but I wanted to create a custom wing that makes more downforce than anything currently available. The way I’ll do that is the following:

• By maximizing the wingspan to chord ratio, for a particular car.
• By choosing an airfoil that has the most downforce.
• By creating a 3D shape that is customized to the car’s roofline shape, at the ideal wing height and setback distance.
• By making it very stiff and light. I made this from foam and fiberglass, and the end result is half the weight of an aluminum wing, and only slightly heavier than carbon.

Dimensions

Since this wing is completely custom, I can make it any size I want. When designing a wing for a car, the general rule is to go as wide as the rules allow, which reduces relative losses from wing tip vortices. Many people have heard the concept of a high aspect ratio wing, and making the wingspan as wide as the rules allow, you get the highest aspect ratio.

This particular wing is designed for a Miata, which puts the wingspan at 64″ max. For a 500 square inch wing, this defines the chord at 7.8″. (For a 700 square inch wing, I’d use the same wingspan and make the chord 11″.)

Airfoil

The aviation airfoil with the most lift at low Reynolds numbers (low speed or small chord – same thing) is the Selig S1223 and S1223 RTL. There are several scientific papers written on the Selig S1223, and you can chase those down if you want confirmation. These scientific tests were all done on aircraft, and cars are different.

A proper motorsports wing can have even more downforce, because it doesn’t have to fly. A bunch of research was done on this by Enrico Benzing, and I refer you to my article on Car Wing Comparisons for those conclusions. There’s been additional research into car wings, and one that is particularly good is called the MSDH.

The MSHD airfoil is a lot like the Selig S1223 in shape, but has more camber. I traced the two airfoils onto a piece of paper, and the MSHD looks like someone just pushed down on a S1223 and increased the camber. The airfoil shapes are otherwise very similar.

3D shape

When you extrude a wing from aluminum, it comes out as a 2D shape. When making a custom wing from composite materials, there’s no reason to build a 2D wing.

Cars have cambered rooflines, and so air comes over the center of the roof at a different angle that at the sides. For example, on a Miata, air going over the middle of the roof comes down at a 5-degree angle. Towards the sides of the roof the angle is 7 degrees. And the area outside the wing stands is at zero degrees.

If you set a 2D wing to 10 degrees angle of attack (AOA), that means there are parts of the wing that are at 15 and 17 degrees, which on most airfoils would stall. Stall means air separates which results in more drag and less downforce. When airplane wings stall, they fall out of the sky.

If you set a 2D wing at say 3 degrees AOA, the middle of the wing is at 8 to 10 degrees, which is ideal. But you do lose a little bit of downforce at the ends of the wing, which are only at 3 degrees. So ideally the ends of the wing should be twisted downward so they are in the same 10-degree angle.

The angle that air comes down the roof different at different heights, and setback distances. I made a tool for measuring that, and so I know the exact shape of air as it hits the wing. I turn, this allows me to design a 3D shape that makes the most downforce over the entire wing.

On paper, a 3D wing calculates to about 5% more downforce than a 2D wing, which results in a 14% better lift/drag ratio in free stream air. (Free stream data is worthless, I only include that here because everyone else does.) 5% more downforce is not a huge difference, but if you’re limited by 500 square inches, you need to optimize every inch of it.

Construction

The wing is constructed of foam and fiberglass. Carbon fiber would be better, but for a prototype, the increase in cost and complexity isn’t worth it. I used biaxial fiberglass and silica microspheres to create a strong and lightweight composite.

This is a essentially surfboard construction, and weighs just 5.5 lbs. It supports my full body weight, no problem. It has some flex to it, but is way stronger than I’d anticipated, and I certainly overbuilt it.

Gurney flap

Most 3D wings have expensive Gurney flaps made out of carbon fiber. This is necessary because the rear edge of the wing is also a 3D shape, and so you can’t just slap a piece of angle aluminum on there. But I spec’d this wing to have a straight line across the rear of the wing, so that it would be easy to add a wicker (or even a second element).

The general rule in aviation is that Gurney flaps are 1-3% of the chord. Smaller is usually more efficient for airplane wings, but we are talking cars here which go slower and have terrible aerodynamics. Using a larger gurney flap makes the wing itself less efficient, but makes more downforce, which in turn makes the car as a whole more efficient.

For cars, 1/2″ Gurney flaps are the most common. This works out to 6.4% of the chord on this wing, which is a little large, and if I could find 3/8″ locally I would have used that instead, as it would result in a 5% chord wicker.

I tested the wing with and without a Gurney flap, and while the wing was 11% more efficient without the wicker, the L/D ratio of the entire vehicle was 12% better with it on. In this restricted 500 square inch size, you definitely want the wicker.

Wind tunnel testing

I tested three 500 square inch wings in the wind tunnel, and the MSHD was the clear winner. It’s difficult to say how much of the performance is from the advantages of the MSHD airfoil, the higher aspect ratio, or the 3D shape. (This was tested on a Veloster, not a Miata, but the roofline is similar enough that the 3D shape conferred some benefit.)

By fully optimizing the MSHD wing with a Gurney flap and the best performing end plate I tested, it’s possible to outperform a standard 701 square inch wing. I was surprised by this result, and it shows how important it is to customize aero to a particular car and rule set.

To recap the benefits of this wing.

• The MSHD airfoil is designed for motorsports applications and has the most downforce of any airfoil I’m aware of. This results in the greatest lift/drag ratio of the vehicle.
• The wing can be built custom to any car or rules specification. By using a 3D shape that conforms to the shape of the roofline, the wing can make the maximum amount of downforce across the entire wingspan at any height or setback distance.
• The wing is composite structure that is both strong and light, reducing weight and polar moment of inertia.

Am I selling these? No. Or at least not yet. I might sell prototypes that I’m no longer using for testing, or make a one-off for a friend. But I have zero interest in dealing with material shortages, shipping problems, irate customers, and all the other crap that goes along with being a manufacturer. Hopefully I’ve provided enough information for DIY wing builders to go on. So get on with it.

I’m having my right knee replaced tomorrow. My previous operation was a temporary measure to correct for 9.9 degrees of negative camber, and I knew that eventually the arthritis and pain would get to be too much. I just felt like it was a couple years away, so I was a little surprised by how immediately necessary it became.

The reason I’m having it done right now has everything to do with aerodynamics: Car aerodynamics and tree aerodynamics.

My wife went on an African safari for a couple weeks in July, and so I had some time on my hands. I didn’t have anything planned, but I’m also not one to sit around.

So I suddenly decided it was necessary to rent time at the A2 wind tunnel and test aero parts. But my Miatas are out of commission and I didn’t have that much to test on the Veloster, and so I spent the next week building aero parts.

My race barn has a downstairs area dedicated for cars and bikes. The tools, welder, hardware, etc., are all down there. The upstairs is my wood shop, where I build most parts, do fiberglassing and painting, etc. I went up and down those stairs more than 100 times per day building and fitting and rebuilding parts for the aero test. That put a lot more strain on my knees than normal.

And then, while I was testing my car in the wind tunnel in North Carolina, a wind storm 650 miles north knocked down a bunch of trees. A big willow went down, a couple pines and apple trees, and a silver maple put two holes in my roof. Wind storm. Ironic, innit?

So I spent a couple days cutting and hauling wood and cleaning up the yard. After which I had two days of coaching at Watkins Glen with PCA. As a result of all that activity, I had no time to rest my knees and I woke up on Friday morning and couldn’t walk.

I literally couldn’t put even a pound of weight on my foot, the pain was excruciating. I had a friend fetch my crutches while I called around to get x-rays and whatnot. Since it had been a few years since my last visit, I couldn’t even get a orthopedic appointment inside of 6 weeks. Thankfully, motorsports came to the rescue.

Warren Wulf is a fellow member at Pineview (and also a classmate from Cornell), and races a BMW. I reached out to him and he got me an appointment that week with Dr Max Greenky. Max does something like 500 knee replacements per year, and after seeing my knee, said that I could get a new one at any time. So I booked an op as soon as possible, and that’s tomorrow.

This will be my third surgery on my right knee, I’ve previously had a lateral release and osteotomy (plates and screws), so I know what to expect regarding recovery, physical therapy, and such. The osteotomy was not my first Rodeo. For whatever reason, my left knee, which had never given me any problems in my entire life, developed the same problem as my right knee. And so Dr Scott Rodeo has done osteotomies on both knees. I’ll need a knee replacement on the left one sometime in the future as well. Gad’s, I think that’ll be my seventh sports-related surgery. I’m falling apart.

FFS Mario, how old are you? I’m 56. I played lacrosse in high school, college, and in a club league afterwards. Probably worse, I skateboarded and did other impact sports my entire life, and have abused the shit out of my body. I don’t count the broken bones. I also have Ehlers Danlos syndrome, and so my joints were predisposed to betray me since birth.

I’m not sure how long the recovery will be. The plate and screws need to come out before the knee comes off, so it’s not a straight replacement. The literature says I’m not supposed to drive for 3 weeks, but you probably guessed that I’m a competitive person, and that just means I’ll be driving at 20 days or sooner. I’m hoping to get back on track in late October.

This gives me some downtime to write, and so I’ll go back through previous articles and inject some wind tunnel test data. I’ll also have to correct a few things I was wrong about (canards), and put an exclamation point on things I was right about (end plates).

This downtime is a good time to Contact Me and ask me aero questions, or just message me in the middle of the day and say yo. I’m also scheduled to be a guest on the RacersHQ podcast, and maybe the Garage Heroes will invite me back.

Once I’m able to drive in anger again, I’ll go to Pineview and get more Aim Solo data on the new track. I’m not sure if I’ll update the track guide, but I want to put something up on this site in the meantime.

My brother was supposed to visit, but because of injuries of his own (we are identical twins after all), I have four non-transferrable guest passes I need to use this year. If you’ve never been to Pineview and want to check it out, hit me up.

OK, then. New parts tomorrow!

If you follow my website, you know I’ve been building aerodynamic parts and testing them for a few years. Much of that time I was focused on Miatas, but the development, testing, and conclusions apply generally to all road racing vehicles. I’ve recently started testing in a wind tunnel, and wrote a report which I’m selling for a modest fee.

The report is over 50 pages and contains numerous pictures and tables, racing simulations, and analysis of the results. The report is aimed at the average racer who has no patience for mathematical formulas and just wants to know what works.

All of the data is presented as coefficients of lift and drag, which is useful to aero nerds and anyone who uses a racing simulation program. But all of the data is also translated into pounds of downforce and drag, and gains or losses in horsepower.

For each aerodynamic part I also do race track simulations on an autocross course and a road course, so that you’ll know the most important thing: how much faster you’ll go. At the end of the report I put various parts together into logical “builds” that comply with the rules and classes of different racing organizations, like Grid Life, NASA, and SCCA. And I also run simulations at more race tracks.

Here’s what you get in the report:

• Aerodynamic principles as succinctly as I can put them, and only as they relate to the data in this report.
• Three splitters with identical front size and shape, but very different undersides resulting in huge gains in downforce and drag reduction. I’ve never seen anyone test splitter curvature, so this is all new.
• Canards in three different locations, with results that vary from 11 pounds of front downforce to over 85!
• Hood vents, showing the drag and lift on cars with and without aero.
• Open vs closed windows, and how that affects a car with and without aero. And how to easily reduce the negative effect of open windows by half.
• Two different spoilers compared to a bare roof without a spoiler. Drag, lift, and lap times as normal, but also MPG.
• 250 square inch wing vs spoiler, for cars that race in GLTC.
• Three different 500 square inch wings, with and without Gurney flaps. This shows the effectiveness of different airfoils, and how each changes with the addition of a wicker. The wing size is especially useful for cars that race in GLTC.
• How does a low-Reynolds high-lift aviation airfoil compare to a motorsports wing? I test a Selig S1223 vs 9 Lives, PCI, and MSHD to find out.
• End plate shape and its effect on wing vehicle performance. The results don’t correlate to CFD, and they aren’t what you’d expect.
• Diffusers work great if your car has a smooth, flat underbody, but how do they work on a regular road car? I build a diffuser designed specifically for use without a flat bottom, and give you the dimension of it, and the results.
• Base car performance, hatchback aerodynamics, drag and lift, and race simulations on different tires, and more.

Fill out the form and click then click the link to download the report. If you’ve bought me a coffee, use the password for a $5 discount.

If you close the browser tab before you click the link, you won’t get the report. In that case, contact me and I’ll send you the link.

Please don’t share the report, it’s a lot of money and effort to build and test the parts, and a lot of analysis and writing to present the information in a meaningful way. I firmly believe you won’t find a cheaper way to go faster than by using the aerodynamic tricks and tips within this report.

If you feel this information was worth the money, I’d appreciate it if you leave a comment. If you don’t feel that way, tell me why and I’ll refund your money.

I wanted a proper wing for my Veloster, that makes enough downforce to match my splitter. The OEM wing on a Veloster N isn’t really a wing, because it doesn’t have an airfoil shape. It’s not exactly a spoiler either, because air goes both on top and underneath it. What the wing/spoiler does is straightens out airflow and cancels some of the lift created by the downward-sloping roof. It makes downforce, but not enough to counteract a decent splitter.

There are a lot of good off the shelf wings, but I wanted to make my own (again). This time I opted for a Selig S1223 airfoil. There’s a higher lift version called the S1223 RTL, but it really only has an advantage at the steepest angle of attack, and otherwise has less lift.

Designing aero for FWD hatchbacks is new to me, and there’s definitely some guesswork involved. As a rule of thumb, you want the widest wingspan allowed in your racing rules, because this lessens the detrimental effect of wing-tip vortices. Most rules limit wings to body width, which on a Veloster N is 71.3″ (181cm). That’s a large wing, but because front-wheel drive cars have proportionately less weight on the rear of the car, they don’t need as much rear downforce.

I could have made a high aspect ratio wing the full width of the car, but this would result in a very small chord. At Reynolds numbers under 2 million, wings are actually more efficient with more chord, not less. So this is an area where I’m totally stumped: is a narrow wingspan with more chord better or worse than a high aspect ratio? I’ll have to test this one day.

One thing is for sure, a longer wingspan is a pain in the ass (and often a pain in the head) when you work on the car. The number of times I’ve bonked my head or run into an end plate…

So in the end, I decided on a wing with a 53.3″ (135cm) wingspan and 11″ chord, with a total area of 583 square inches. That’s the same total area as a 64″ 9 Lives Racing wing that you’d see on a Miata, which kinda makes sense, as a Miata and Veloster have the same amount of weight on the rear wheels.

I traced the S1223 airfoil shape onto Meranti Hydrotek (BS6566) plywood scraps I had lying around from splitter projects, and then cut them out on my band saw.

Wing construction

I began by gluing up the ribs to a central spar. You can see in the pic that the spar is notched into the wing mounts and end plates. I then glued a wood dowel to the front and a thin spar towards the trailing edge. This skeleton provides the basic structure and shape, so that thin plywood panels can be glued on top and bottom.

The plywood itself is very thin and offers little strength, but the whole outside will be covered in fiberglass, which is where the strength comes from. This construction style is called a torsion box, and is used on everything from airplane wings to the doors in your house.

The wing mounts are on the bottom, and an integral part of the structure, so I had to laminate the bottom of the wing in three pieces. I did one at a time.

After gluing and filleting the bottom panels using epoxy and microspheres (filler), I flipped the wing over and glued the top of the wing. You might be wondering about the plywood I use for the upper and lower skins. Big box lumber stores like Home Despot and Lowe’s get pallets of wood that are strapped down with a very thin plywood sheet over the top to protect the wood below. This thin 3-ply plywood is so flexible you can almost bend it in a circle. There’s no SKU for it, and they’ll give it away if you ask nicely.

I can make a couple wings from a sheet of this plywood. But while it’s free, it isn’t easy to find, because it doesn’t come on every pallet of wood, and they typically throw it away. So make sure you check back often or make friends with the store manager, this shit is gold.

After gluing on the plywood panels I wrapped the whole thing in fiberglass fabric. After gluing the wing up, I noticed the rear edge of the wing was a little wavy. Damnit. I don’t think this is a big deal for a single element wing, and I’ve even seen some wings that have a wavy rear edge on purpose. But I’m going to add a double element in thr future, and it needs a straight trailing edge for that.

The solution was to clamp and glue a 1/2” (13mm) Gurney flap to help flatten it out. I wasn’t sure I was going to add a wicker on this wing, but it was the obvious solution to this problem and now it’s not coming off!

The wing weighs 6.6 lbs (3 kg) which is pretty light considering how stiff and solid it feels, and this includes the gurney flap, end plates, and bottom mounts. That’s less than half the weight of an aluminum wing of the same size, and probably a pound heavier than a $3000 carbon wing. The bigger difference is my material cost is less than $100.

Wing mounts

While building the wing I spent a lot of time thinking about wing mounts. Many racing rules limit the height of a wing to roof height, but they usually give hatchbacks a bit more room, and 5-8″ above the roof is common. That’s plenty of height considering how much the roof slopes back in the rear.

On a sedan, you typically mount a wing with about 1/4 of the wing overlapping the rear of the trunk. On a hatchback, you need to get the wing a bit further back. The wing needs area below the wing for the low-pressure region to form, and the rear window glass is in the way.

Because I don’t know a lot about hatchback aerodynamics, I want to be able to adjust the wing not only for angle, but for height and setback distance. Because of the high roofline, being able to make adjustments here could be critical.

For this reason, I built a four-point mount with a long, flat top. This allows me to move the wing fore and aft 6”. I also put a 8-degree fixed angle into the top of the cradle mount, and so moving the wing forward or backward changes the height 1”. The wing itself has built-in uprights with an additional 2″ of height adjustment, so altogether I can move the wing forward or backward 6″ and up or down 3″. For the lowest drag settings, I zero out the wing and move it low and forward. For the highest downforce configuration, I’ll move the wing rearward and up. Makes sense to me, but let’s see how it works.

Airfoil data

The following images come from Airfloil Tools and show the S1223 data at 500k Reynolds, which is 57 mph (92 kph) for an 11″ chord wing.

The first graph is the coefficient of drag (Cd) at different angles of attack. The lowest drag setting is right at zero degrees, but anywhere between -2 to +2 degrees is going to be indistinguishable. From 3 degrees to 12 degrees drag increases linearly, but then really spikes upwards at 13 degrees as flow separations occur.

In the next graph you can see the best lift/drag ratio of the wing is around 5 degrees. This isn’t hugely important, since it’s the lift/drag ratio of the vehicle that matters, not the wing. Anyway, it’s an interesting data point.

The most important graph is the coefficient of lift vs angle, because you can use that to estimate the amount of downforce at different angles of attack. This wing ranges between .5 Cl at -2.5 degrees to 2.4 Cl at 13 degrees

Therefore, I’m going to spec a working range of 0 to 12 degrees, and use a baseline of 8 degrees. Now, because the roof also slopes downward at around 8 degrees over the roof extension, that means I’m actually setting the wing to zero degrees in relation to the ground.

Given all this data, it would be useful to have five settings for wing angle: 0, 3, 5, 8, and 12 degrees. In terms of pounds of downforce, if the wing is making 120 lbs of downforce at zero degrees, then each setting would be about +25 lbs of downforce. I don’t think settings in between those will be super useful, as I don’t think I could feel a difference of having 12 pounds more or less in the trunk.

Wind tunnel results

I tested this wing in the wind tunnel and it did very well. It made about 180 lbs of rear downforce at 100 mph, and lost about 30 lbs on the front. The loss of front downforce is the natural effect of pushing down on the rear of the car, it goes up on the other.

If 180 lbs seems low, it’s because it is. See my previous comments about height, angle, and setback distance. With those variables optimized, I expect to almost double the downforce.

When compared to the other wings I tested, the Selig wing was very efficient. For some reason, this wing lost less front downforce than the other wings I tested, and as a result, the S1223 wing had the highest L/D ratio of all the wings. However, it didn’t make quite as much downforce as the others, so the L/D ratio of the vehicle was slightly worse with this wing than the other wings.

This is exactly why shopping for a wing looking at its efficiency is utterly worthless, because the most important aspect of a wing is how much downforce it makes. If two wings make the same downforce and one has less drag, of course use the one with less drag. If one wing makes more downforce than the other, but it also makes more drag, 99 times out of 100 you want the wing with downforce, as that will make the vehicle the most efficient.

As an airfoil, the S1223 is still a solid choice for a car wing. I’ve recently shifted my focus from low-Reynolds high-lift aviation airfoils to Enrico Benzing’s designs, and the MSHD airfoil. The latter I recently built and tested in the wind tunnel, and I’ll report on that soon. Subscribe to my blog and you won’t miss that. Believe me, you don’t want to miss that.

Engineers design cars with the sound belief that people will drive them around with the windows closed. But in high performance driving, we sometimes have to roll then down.

In Europe, they keep the windows closed for racing and track driving because it’s safer. The reasoning is, that in case of a rollover, the car is stronger with the windows closed. And the closed windows also keeps insects, tire debris, and other shit from hitting the driver.

In the USA, we race and do track days with the windows down because it’s safer. The reason being that after a crash, we may need to extract someone through a window if every single door on the car no longer works. It isn’t exactly logical, but that’s our time-worn tradition.

In the USA, we have another time-worn tradition, which is to let people who have a lot of money bend the rules and do whatever the fuck they want. And so if you buy a Porsche 911 GT3 Cup, which has Lexan windows that are fixed in place and don’t roll down, that’s OK, you can keep the windows closed and do point-bys with your turn singals. Nevermind that anyone who can afford a Cup car can afford a spare set of doors with no windows….

And so I’ve wondered about the aerodynamic effects of open and closed windows, and how much of an advantage these entitled rule dodgers have when they bring a Cup car to a track day.

Calculator results

Everyone who has experienced buffeting when they open their car window at speed understands that open windows causes turbulence. From this you can conclude that there must also be an increase in drag, and if your car has a rear wing, a loss of rear downforce. But how much?

At first I tried to figure this out with an online calculator, and then later tested open vs closed windows in the A2 wind tunnel. If you skip ahead to the wind tunnel results, the only thing you’ll miss is that the Veloster N seems to be better than most cars, and even most race cars, with respect to open windows. How did I figure that out?

I used the Drag and lift calculator from HP Wizard. This aerodynamic estimation tool calculates the drag and lift of your vehicle based on its body shape. I’ve gotten very close to real-world values using this tool, it’s pretty dope.

To use the tool you click on the little pictures that describe the various shapes of your car. When you’re done, you’ll have a close approximation of drag and lift. You can also use this tool to do things like change your sedan into a fastback and see how much that reduces drag. Or in this case, open the windows and see what happens.

I built my Veloster N in HP Wizard and the tool says it should have a drag (cD) of 0.357 and positive lift (cL) of 0.15. This is a fair approximation for a hatchback with a similar shape. However, in the wind tunnel, my Veloster N had a drag coefficient much higher than that, at .421. What the fuck?

Most production cars generate lift, because they are longer on the top than on the bottom. That shape is akin to an airplane wing, and so cars typically lift at speed. Cars with more streamlined shapes, like coupes and fastbacks, generally have even more lift.

To counteract lift you add devices like airdams, splitters, spoilers, and wings that create downforce. The byproduct of downforce is drag. And guess what? The Veloster N creates downforce.

Yep, believe it or not, the Veloster N has a cL of -.027, meaning it makes downforce, straight from the factory. This is normal for a Ferrari or Porsche, but very rare for a production car. And for a hatchback? It’s the first I’ve ever heard of it.

When you create downforce via suction you can return a high lift to drag (L/D) ratio, but when you work only from the pressure side, it’s much less efficient. The Veloster doesn’t have a proper rear wing or splitter, and generates downforce primarily through areas that hold positive pressure. All of those angled surfaces on the front bumper fascia plus that funky spoiler combine to make downforce. But this also results in a lot of drag.

To go back to HP Wizard for a minute, it uses a default value of .15 for lift (cL) for all cars. If we give the downforce-generating bodywork an inefficient but fair 2:1 L/D ratio, then we can add .059 to the .357 drag that HP Wizard estimated, and we get a .416 cD. So that’s pretty close to the Veloster N’s measured .421. That’s with the windows closed, of course.

So now that we have a fair representation of drag, let’s see what happens when we open the windows. HP Wizard has two options for that: race car and production car. The race car version is more streamlined, and is less affected by the open windows.

Keep those values in your head: An additional cD of.027 (or 2.7 points) is the additional drag on a race car, and .047 is the amount of drag on a production car. The Veloster N is better than that, but we’ll geto that in a moment, first we must also discuss downforce.

The problem with opening your car windows isn’t just that it adds drag, it also reduces rear downforce. The net effect is a worse L/D ratio. But how much worse?

Well if we look at HP Wizard, it says there’s no change in downforce when you open the windows, the cL stays a static .15 (meaning positive lift) no matter if the windows are open or closed. Now HP Wizard is a simple calculator, and perhaps the values are too variable to be meaningful. So I’ll give them a break. Luckily, I got this information from the wind tunnel.

Wind tunnel results

In the wind tunnel, the data shows that opening the front windows results in a significant 15% loss in rear downforce. What’s interesting is that the percentage of loss is the same with either a spoiler or a wing. However, because a rear wing might generate on the order of 400 lbs of downforce, a good 60 pounds of that is sucked out when you open windows.

Whether using a spoiler, wing, or OEM bodywork, the result of this 15% rear downforce loss is that that the aero balance shifts forward. For example, on my stock-bodied Veloster N, opening the windows at 100 mph results in a loss of 5.6 lbs of rear downforce, and with that, a gain of 4.8 lbs of front downforce. (This is normal: gains at one end of the car often result in losses on the other end.) This isn’t a huge change, but the car will now start to shift to oversteer slightly at high speed. Not so good.

Now let’s talk about what happens to drag. Recall that HP Wizard predicted that the race car would gain 2.7 points of drag (.027 cD), and the production car would gain 4.7 points of drag, with open windows. The Veloster gains only 2.3 points with open windows!

This is surprising, because HP Wizard is a fairly accurate calculator. So if we go along with that, according to the data, the average production car will lose about 7 hp at 100 mph when it opens the windows, while the Veloster N loses only 3.4 hp. (Side note – the drag loss was the same with OEM spoiler or a wing.)

At this point you might be imagining how your car will react with open windows. I’m sorry to say, it’s almost certainly going to be worse than mine. 1) Because the Veloster doesn’t lose a lot of drag with open windows, we can conclude it’s loss of downforce is also less than other cars, 2) The hatchback body style is likely to be less affected by the open window turbulence, 3) the short 135cmm wing doesn’t stick out into airflow as much as other wings, and will be less affected by open window turbulence.

On a car with a less efficient body shape, like a hardtop Miata, I’d guess that the wing loses 25% of its downforce between the open windows and the compromised shape of the canopy, and probably a great deal more drag as well.

And these results come from linear airflow in a wind tunnel. When the car is cornering and in yaw, the outside window will have even more air shoved inside, and I’m not sure what that will do for drag and turbulence, but it could be a lot worse.

Open window improvements

Given that open windows reduce performance, it would be useful to minimize the amount of drag and turbulence from them. There are several ways you might do that: by addressing air at the front of the A-pillar, helping airflow re-attach at the B-pillar, reducing the size of the window openings, or blocking the air as much as possible using a window net.

A good source for ideas is NASCAR, because they race with the windows open and at such high speed they need to employ all the tricks. For example, their window nets have more fabric than holes. Do you think this is for safety or for aero?

A pillar

The relationship between the width of the A and B pillars is important. If the A pillar is wider than the B pillar, less air will go in the cockpit, as air can more easily jump the window gap and reattach at the B pillar. On older cars, the A and B pillars are often the same width, and you’d guess that open windows are slightly worse.

On convertibles with hardtops (Miata, S2000, etc), the B pillar is wider still, because it has to cover the gap where the convertible top stows away. As such, the B pillar sticks out into airflow and turns the cabin into a parachute. The result is lot of turbulence, drag, and loss of downforce.

To help air jump the window gap, you could do something to widen the A pillar, or add a wicker on the quarter window frame. This would move the stagnation point outward, and that should help air pass by the gap. This is a fairly easy modification, and I’ll have to wind tunnel test that in the future.

Another idea I’ve seen is to add vortex generators in the same location. VGs are fairly draggy themselves, but they can reduce drag overall if they delay flow separation.

In my testing, I found that vortex generators on the roof were particularly bad; they caused a loss of 5 hp (at 100 mph) due to drag and reduced the rear wing’s downforce by 28%! I haven’t tested VGs on the quarter window frame, but my guess is they are going to have a lot of downstream losses, and won’t work as well as a simple wicker. But this would be an easy one to test in the wind tunnel, and put to rest a lot of silly speculation.

I asked my aero sensei Kyle Forster if there was anything I could do to mitigate losses from open windows, and he said there’s more to be gained by smoothing the air at the B pillar than doing anything with the A pillar. So there you have it from a F1 aero engineer, maybe don’t waste too much time on the quarter window.

B pillar

If you look at any NASCAR stock car you’ll see a lot of rounding on the B-pillar, which extends inside the cockpit slightly, behind the driver’s head. I did the same thing with my Miata fastback, trying to scoop air that was in the cockpit and coax it out the back of the window.

I’ve written about my fastback Miata a few times here; it reduced drag by 15% and increased rear downforce by 130% compared to the OEM hardtop. Those are crazy good numbers for just a top.

In the past I had attributed most of that to the improved backlight angle, and opined that the narrower B pillar was potentially part of the performance. But knowing what I know now, I believe that the narrower B pillar has more to do with it than I thought. It’s the only way to explain how the L/D ratio went from 2.11 to 2.97. That’s an insane improvement from simply changing the shape of the roof.

Roofline junction

The one thing I tried in the wind tunnel that would affect open windows was the addition of WellVizor wind visors. These reduce the top edge of the window opening by 1″ at the front, gradually expanding to about 2″ at the rear. That’s not a lot of area, but it turns out to be a very significant area for turbulence.

The visors attach nearly flush with the body, and don’t bulge out like some I’ve seen, nor do they cover much of the window. So imagine our surprise when the two little pieces of smoked plastic reduced the effect of open windows by half! In practical terms of 100 mph, that means a gain of about 1.5 hp due to drag reduction, while also gaining 7% more rear downforce.

Given that these simple visors reduced open window losses by that much, I can conclude that most of the turbulence comes from the top of the window. Ergo, you can do whatever you want to the A pillar or B pillar, and it won’t amount to what you’ll get by smoothing airflow and/or reducing window size at the roofline junction. At least on this car.

The only downside to the window visors is you may hit your arm on it when you signal a point-by over the roof. However, they are flexible and bend out of the way. This is especially important if you had to exit the car window with your helmet on.

I can’t fit window visors to my Miata because there’s no window frame, so I modified my fastback roofline to have a lower opening, and added a drip edge hoping this will reduce air going in the cockpit.

Racing simulations

Now that I have all the data, I can get back to the question of how much of an advantage the Cup Cars with their fixed Lexan windows have. To find out, I’ll run a few simulations in OptimumLap, comparing cars with open and closed windows. I’ll do this for the base model Veloster, and also one with a wing, because that will have larger losses. I’ll also throw the window visors in the mix to see what they do.

I’ll run the simulations at two different venues, the autocross track from 2010 Solo Nationals and Lime Rock. I use these two tracks because they represent opposite sides of the aero spectrum, and have lap times that are quite similar to each other. So this is a good way to see the relative benefit of aero on a slow and fast track.

For the simulations I’ll use the following values for the Veloster: 3250 lbs, 234 hp, long-G .95, lat-G 1.2, and simply change the cD and cL to match the different builds. The aero car is just a splitter and wing, and represents a low effort build, so don’t get overly excited about it’s lackluster performance.

The results show that whether you have open or closed windows on an autocross course, it won’t make a difference. It doesn’t even matter if the car has aero.

On a proper race track like Lime Rock, a Veloster with the windows closed would go .09 seconds faster per lap than one with the windows open. On the mildly aero’d car, that would double to .18 seconds per lap. Lime Rock is a short track, and so on longer track with a 2-minute lap, a car with closed windows are gaining .36 seconds on the rest of us. That’s not as much as I thought it would be, but it’s still significant.

The budget solution is to put window visors on your car, and these resulted in a 1/10 of a second advantage at Lime Rock, on the aero car. If this seems like a very small number, think about this: in a two-hour endurance race stint, the rain visors would gain almost 12 seconds over a car with open windows, while simultaneously reducing fuel consumption.

Conclusions

Open windows are a drag, but you can mitigate losses by using simple tricks like modifications to the A and B pillar, or addressing the roofline junction window visors. WellVizors are just one of several different manufacturers that make rain/wind guards, and it might be that the more bulbous versions work better that the streamlined ones. There’s more to experiment with here, and the results could vary widely, depending on the car. (Note that you can contact WellVizors directly and order just the fronts at a great discount.)

The Veloster didn’t suffer a lot from open windows, and it’s just one more way that I’m constantly surprised by its design and performance. Most cars will certainly perform worse with the windows down.

Convertible cars with their shitty hardtops are particularly screwed by open windows by an exaggerated parachute effect at the B pillar. Furthermore, they don’t have window frames, and so you can’t even attach window visors! But there are ways around this if you put some creativity into it.

As for the moneybags bringing their 911 Cup cars to track days, the closed (fixed) windows aren’t actually that big of a deal. Most of the P-car owners can’t drive worth a shit anyway, and they need every advantage they can get.