TL;DR Go get a Wing Logic MSHD wing right now. The wing is over 2.5 lbs lighter than the previous version, makes more downforce, and is just $350 for a 65” wing!
In my article on Car Wing Comparisons, I review many of the airfoils used in motorsports. The wings I researched for that article were mostly from the Airfoil Tools website, and designed as airplane wings. Of those, the best performing wing was the low-Reynolds high-lift Selig S1223.
But cars don’t need to get airborne, and so motorsports wings can be designed differently than airplane wings. In Enrico Benzing’s book Wings, he examines the various factors that make car wings different than airplane wings. I’ll do Benzing a great disservice and sum up his entire book in a sentence: car wings can have more camber and thickness, and the location of the maximum thickness is important.
Benzing’s research was groundbreaking 13 years ago when it was published. Since then, car wings have continued to evolve from high-lift airplane wings, to motorsports-specific designs. I built a S1223 wing and tested it in the wind tunnel vs two wings in Benzing’s book, and the S1223 airplane wing did not perform as well as either of the car-specific wings. However, the best performing wing I tested isn’t in Benzing’s book.
The wing that won my wind tunnel comparison test was a new airfoil called the MSHD. The 144-page thesis paper by Sriram Saranathy Pakkam goes into great detail on the airfoil’s design, and how it compares with other popular wings.
At 10-12 degrees AoA, which is where you’d set most wings for maximum downforce, there isn’t a lot of difference between MSHD and S1223. But at low angles or very high angles, the MSHD is in a class by itself.
It seemed too good to be true, so I had to build one, try it on my car, and test it in the wind tunnel. It outperformed my homemade S1223, a Procar Innovations wing (Benzing 183-185), and a 9 Lives Racing wing (Benzing 123-125). These results are in my first wind tunnel report.
The shape of the MSHD looks a lot like a Selig S1223, but with more camber, and an absolutely wicked trailing edge. The trailing edge was difficult for me to build in foam and fiberglass, as it’s nearly impossible to get the edge thin enough and still retain some strength. So I’m pretty sure the performance of my DIY wing could have been improved if I made it in some other material.
MSHD compared to some other wings, from Pakkam’s paper.
Carbon fiber is the best material, but is banned in series like Champcar, and can be prohibitively expensive. A better way to build this wing for the average club racer would be to extrude it out of aluminum. This would allow a very thin and strong trailing edge that closely follows the ideal geometry. And if mass produced, it could drive the price down to where even cheap bastards like myself would buy one, rather than DIY an inaccurate copy.
Wing Logic
So now that you know the backstory of why the MSHD is the motorsports airfoil of choice, let’s move onto the part of the story of how this wing gets mass produced. It starts with Michael Jui, an entrepreneur who was already doing stuff in motorsports. If you’ve ever shopped for a new helmet and sorted on lowest price, you’ll see Conquer helmets at the top of the list. Conquer is one of Michael’s companies, and embodies his philosophy: high quality at a low price.
Michael contacted me and asked if I could test his extruded aluminum Wing Logic wing in the wind tunnel. That information is here on my site in the article Wing Logic vs 9 Lives Racing, and also more in depth in my second wind tunnel report. In short, the Wing Logic wing did pretty well for being based on an airplane wing. But the 9 Lives Racing wing made the same downforce for less area, meaning it was more efficient and also weighed less. In the end I opined that the 9 Lives was their better choice if price was no object, but that Wing Logic was a fine wing for the budget conscious.
After that test, Michael and I got talking about a new car wing, and I suggested he build one using the MSHD shape. I advised that he should make 10” and 12” chord sizes. Alyssa Merrill, who does all my computer designing and 3D printing, sent Michael the CAD file we used to 3D-print Steve Leo’s foam wing. And with that, Michael’s engineer got started on the two wing designs.
Weighty decisions
Michael sent me the engineer drawings for both the 250mm and 300mm wings. The larger wing was going to be almost 150% heavier, and I’m not sure how many people would want over 20 lbs high up above their car. The end result is that Michael has only the 250mm available at this time, and I think that was a good call.
Initial design of the 250mm (9.85”) chord.
Weight is an issue for aluminum wings, and so I asked Michael if he could have his engineer lighten it up and make it thinner? Not possible. The wing was already at the minimum thickness for making an aluminum extrusion. So I suggested reducing the number of end-plate bosses from four to three, and moving one of them to the rear of the wing. I also suggested removing one of the spars. This resulted in a minor weight savings of .166 kg per meter, or 1.785 ounces per foot.
On a typical 72” wing this is only a savings of 11 ounces. But as the saying goes, if you concentrate on the ounces, the pounds take care of themselves. Also, the location of that weight is important, as wings are typically the furthest distance from the center of gravity.
My tweaks included removing two bosses and one spar. The end result is less weight and wider support for the end plates.
I also felt the endplates on his original wing were too heavy. At 10” x 5.5”, there’s very little flex to them, and so there’s no need for such a heavy endplate. Thickness was originally around 9 gauge, and is now about 13 gauge, which is a savings of 7.3 ounces.
The Wing Logic endplates are rather small compared to what you see on most wings, and you might wonder how they perform. In my 2nd wind tunnel report I tested the standard Wing Logic endplate versus larger endplates, and the results were about even. There was slightly more downforce and drag for the larger endplates, and slightly less from the smaller ones. But the overall efficiency of the car remained about the same with both, and so small endplates are just fine.
Another decision not to be taken lightly is whether or not to put a Gurney flap on the wing. 9 Lives Racing has a slot in the trailing edge of the wing for a Gurney flap, and I advised specifically against this. The problem is that this adds weight, but more significantly, it changes the rear geometry of the airfoil. If you compare a Be 123-125 airfoil with a 9 Lives, they are identical… except for the trailing edge, which is far too thick on the 9 Lives.
How much does this change the performance? I don’t know. But if the wing performed better with that thicker rear section, we’d see wings designed like that from the get-go, and we simply don’t.
The other way to do a Gurney flap is to mold it in. This is what Michael did on his original Wing Logic, adding a 1/4” Gurney flap as part of the extrusion. While this is economically efficient for the customer, I advised against doing that on the new wing.
First, because it’s just so easy to tape on a Gurney flap. The flap is on the pressure side of the wing, and so simply using duct tape to fasten it down makes for easy application of any size wicker. Some people may want to use double-sided 3M tape or rivets for peace of mind, and I can’t argue with that. I use 1/2” aluminum angle for a Gurney flap, and don’t mess with different heights. Changing wing angle has about the same effect as changing Gurney flap height, and so nobody really needs multiple sizes; a slot for a custom-sized Gurney flap is just more gimmicky and expensive shit I can do without.
The other reason not to mold in a Gurney flap is it can be trickier to set up it up as a dual element. Gurney flaps can absolutely be used in a multi-wing setup, but the wing overlap and gap distance become crucial and finicky variables. You can read about that elsewhere on this site, but rather than having a customer suffer those headaches, I suggested to leave the trailing edge sharp.
So you might be wondering, how does the MSHD behave as a dual element? The maximum coefficient of lift goes from 2.3-ish as a single element to almost 4 as a dual element. Yowza. Also, there’s a very soft or non-existent stall. So as good as this is as a single wing, it might be even more remarkable as a dual. (I have one set up for a dual element, but more on that in a future article.)
Dual element CL at 300k and 400k Reynolds. For a 250mm chord wing, this is about 38 and 50 mph, which is very slow; wings generally improve with higher speed, so this looks great.
Unboxing
It took a year from concept to reality which is probably a short development time, but it seemed like forever. The wing came really nicely packaged, with protective foam all around. My first thought upon unboxing it was “holy shit that’s a lot of camber.” My next thought was “good lord that’s a wicked trailing edge!” And then I thought, “man, this is pretty light.”
I haven’t been this excited since Xmas in my jammies.
The chord measures 250mm, or about 9-7/8”. Available lengths are 60”, 65”, 71”, 72”, and 74”. Most racing series limit wings to body width, but if you are unbound by rules, wider is better. If you race in a series that has a 701 square-inch limitation, you’ll want the 71” wing.
250mm chord. Or 9 7/8” ish.
End plates are included in the box, and as stated earlier, they are the same size and shape as the old endplates, just thinner. They aren’t pre-drilled, but there’s an included template for that. The M6 bolts are stainless button-head, and thread quite deeply into the end of the wings, which I appreciate. (For reference, 9 Lives uses four M5 bolts, PCI uses two M8.)
Bottom mounts are included, but the holes are oddly large. I typically use M6 hardware for wing mounts, but these are much larger. Which is fine I suppose, just different. But since I’m going to do some back-to-back testing, I made up my own mounts using a 9 Lives pattern.
Original undrilled Wing Logic mounts (top), new pre-drilled MSHD mounts, and DIY 9 Lives pattern mounts (bottom).
I’m sure someone wants to know what the Wing Logic MSHD looks like next to the industry standard 9 Lives wing. MSHD has more camber, and a much finer trailing edge.
9LR top, WL bottom.
I weighed the 65” Wing Logic MSHD with endplates and wing mounts and it came in at an even 14 lbs. The original 65” Wing Logic was 16.6 lbs, so this is a significant savings over the old wing. Incredibly, the 65” x 9.8” MSHD weighs even less than a 9 Lives Racing 64” x 9.2” wing, which came in at 14.6 lbs with the same mounts and endplates.
Now this is the older 9 Lives wing that I weighed, and what is now sold as the “Express” wing. 9 Lives Racing has a newer wing, which is the same shape but has been lighten somewhat. I don’t have one, and so I can’t compare the weight on that.
65” wing with endplates and wing mounts (before welding) is an even 14 lbs.
I have to say, it’s really satisfying to have the Wing Logic MSHD in my hands. My role in helping bring this to market was very minor, but I’m chuffed to have had any input in this, and to see it IRL
It’s bigger, lighter, and has more downforce than any other aluminum wing on the market. And it’s currently on sale for $350 in a 65” wing!
CFD
A lot of wings get developed, sold and evaluated based on CFD alone, which is a highly suspect practice. CFD is a tool to get you part of the way there, and you to need to follow up with a wind tunnel or track testing. That’s how the F1 teams do it.
Unverified CFD results in ridiculous claims of downforce and efficiency, and optimizations that show up in the computer world, but not in the real world. I’ll give you an example: I tested some CFD-designed endplates that were supposed to increase downforce and reduce drag over a standard rectangular endplate. But in the wind tunnel, the CFD designed ones were actually worse than a plain rectangle I cut out of a street sign. (This data is in my first wind tunnel report.) Moreover, I flipped the CFD endplates backwards to see what would happen; the data showed very small changes to drag and downforce, but the same overall performance. So whenever I see something that is developed and marketed completely from CFD, I distrust it.
However, I do trust the CFD from Kyle Forster, since his CFD models are much more accurate, and he puts the wing on a virtual car, rather than relying only on free stream simulations. I’m not going to delve into the computer stuff here, but if you’re wondering how the old vs new wings compare, Wing Logic published the full CFD comparison of the original Wing Logic versus the new MSHD.
Yes, this is a lot of downforce. No, I don’t know why 4 decimal places are required.
Note that the MSHD CFD data stops at 10 degrees AoA, and as you saw from the original thesis paper, the MSHD can be run at even higher angles. And so while these numbers look great on paper, I believe they can get even better.
Based on the CFD, and more importantly, the testing I’ve done on my own DIY MSHD, I fully expect the new Wing Logic to outperform every other wing I’ve tested, and relocate all my other wings to the recycling bin.
Track testing and wind tunnel
I’ll schedule wind tunnel time later this year, and will test the MSHD as a single and dual element. I have some different endplate ideas, and a few secrets I’m not sharing yet that could be game changers. Subscribe to the blog, you don’t want to miss that.
I’ll also be track testing the wing on a few occasions. I’m racing with Lucky Dog at PittRace and will be able to A/B test the MSHD vs 9 Lives wings on the Miata I took to the wind tunnel. So that will be great to get real world data that I can use to compare with the wind tunnel data. And I have some other track tests scheduled for Watkins Glen and Pineview, before going back to PittRace one more time (the final time, ever).
In a previous article I examined Veloster N Cooling and Hood Vents. I bought a fiberglass hood with an extractor vent, modified the vent to be larger, and then tested that on track and in a wind tunnel. The results were great, with up to a 10-degree drop in water and oil temperature, reduced aero drag, and increased downforce.
That car was totaled, and now my new car needs hood vents. But this time I decided to install Race Louvers in the original hood, rather than use a fiberglass hood. The reasons for that are two-fold: I always felt like the black hood was a bit too boy-racer (boi ricer?) for me; Race Louvers makes the best performing hood vents in the industry.
I know that because Race Louvers publishes their test data publicly, and they’ve tried just about everything. I’ve also tested my own vents and others in the same wind tunnel, and Race Louvers make more downforce every time.
So I contacted Al Watson from Race Louvers and begged him for a deal. “I’ll put stickers on my car,” I whined. “I’ll write an article on my site,” I pleaded. His reply was something to the effect of, “Enh, don’t worry about it, I saw that your car got wrecked, I’ll send you whatever you want.”
No fucking way. Thanks Al.
We discussed which Veloster louvers on his site would work best on my car, and I confess I didn’t realize there was more to it than I thought. We decided on a center vent alone, because using the center and sides, or the new fancy CF Extreme louvers might actually be too much area. Two days later I had the louvers in hands.
Installing Race Louvers
The first thing you need to do is measure the distance from the front of the hood to the louvre opening. This should be directly behind the radiator. In the case of the Veloster, this is just about 10” to the back of the fan shroud.
Measure from the front of the hood to the rear of the radiator.
Next I used blue painters tape and a sharpie to draw two lines 10” back from each side. Because the hood is contoured, I used a thin piece of wood as a straight edge.
Next I connected the two sides and drew a line between them. This would be the front of the vent.
Center line and front edge scribed on blue tape.
Next I used the full-sized template to draw cut lines on the hood.
I then masked off the area some more to protect the hood when I cut it. Now I was ready to make the first cut. I used a grinder with a cutoff wheel to make the initial rectangle, but a jig saw was necessary to make the final cuts in the outside corners.
A cutting wheel makes fast work of this job.
Next I test-fitted the louvers and found some unsightly gaps in the hood. If you look from the side, you can see the support webbing layers. On a race car I wouldn’t care, but on a dual duty track car, I felt this looked sloppy.
Notice the bare sheet metal below my fingers. Yuck.
I figured if I put some aluminum flashing here, with a 90-degree bend on it, this would cover the side gaps. I don’t think there’s any performance gained here, this was just an aesthetic thing.
Aluminum flashing test fit, before trimming.
I made four pieces.
With that done, I could finally rivet the louvers into place.
Louvers installed.
From the side, you can see the aluminum flashing covers up those side gaps nicely.
Hey down there!
I did the whole operation without removing the insulation under the hood.
Hey up there!
I haven’t tested the louvers yet, but I’m betting they work slightly better than my FRP vent did. The total area on the Race Louvers is 158 square inches, which is 17% larger than what I tested in the wind tunnel. In addition, the louvers should add some differential pressure that should aid in both front downforce and heat extraction.
From start to finish this is about a 2-hour job, and was a fine way to spend a Saturday morning. Next I’ll make a custom cover for it, to keep leaves and water out of the engine bay. But autumn is a month away yet, and there’s a Hayabusa swap calling my name.
Last weekend I instructed on track for the first time since my accident, and while my neck isn’t 100%, I can right-seat coach and drive a car on track again. I’m definitely concerned about things like sudden and unexpected stops (tire walls, Armco, and other barriers), but the general motion of a car on track is no longer a problem.
My student for this recent track event with Mass Tuning was a 16-year old girl who got her driver’s license two months ago. She was driving her dad’s clappped out Chevy Cruze, which I fully support (except the seats, which lack any support). But we found out quickly the budget Accelera Phi tires suck in the wet. Naturally, it rained the whole day!
Pineview Run is a busy and technical track, not the easiest place for your first track day. But she was a good student, listening to instruction, implementing changes, and without ego. So pretty soon she was safely navigating the track with pace and giving point bys. The next session we were quickly on pace with other cars, and right in the swim of things, cutting wakes through the wet track. By the last session we had a few dry spots, and she started throwing the car into turns and squealing the tires. With that she earned a few point bys, and even dropped her 18-year old brother (in a track-prepped BMW) in her rear-view mirror.
That was a good day with Mass Tuning, and a nice way to ease back into coaching again. This was my first event since coaching with Ross Bentley, Peter Krause, Sam Smith, Josh Herbert, and my brother Ian. We called it the Speed Secrets Academy, and it was a smashing success. I haven’t written about that event yet, but it went really well, far exceeding the high expectations the students or coaches imagined. Ian wraps up some of that in his blog about the Speed Secrets Academy, so I won’t repeat that here.
There have also been a few podcasts on that event:
Ian and I were guests on Garage Heroes in Training, reviewing how the program went, and how it was different from regular HPDE. .
People keep asking if we’ll do another Speed Secrets Academy, when, and where? Yes, and I’m not sure. We had always envisioned this as a traveling road show, and we’d like to hit tracks in the Northwest, Southwest, and Southeast. But a lot of this hinges on the availability of Ross and Peter, so I don’t have even tentative dates yet.
In the meantime, I’m available for private coaching at Pineview Run. I prefer to coach here as we can do drills, data, and run through many of the program elements that we did in the Speed Secrets Academy. Alternatively, I can also meet you at Watkins Glen, but you’d get less out of a coaching session there than at Pineview.
If you are a Pineview member or have an annual track pass, you can contact Pineview via the website. In the notes, mention that you want me as your coach.
If you aren’t a member, contact me and we can schedule it. Pricing will depend on if you need a guest pass, and what day, but figure around $400-800.
The best track layout for coaching sessions is the short track, as you can get in more laps in than on the long track. This is especially important for doing drills, and we will be doing mostly drills.
Pineview’s track schedule goes like this:
On Wednesday and Thursday, Pineview uses the short track, but it’s a shorter day that goes from 12:00 to 5:00. Still, that’s plenty of time and you’ll be exhausted.
On Fridays they use the short track starting at 10am, and switch to the long track in the afternoon.
On weekends, they sometimes use the Friday schedule, and sometimes use the long track all day. This layout is fast and flowy, but not quite as good for coaching sessions.
I’m working with Pineview to offer other coaching programs, and with Ross on the next Speed Secrets Academy. But until we get that figured out, hit me up for intensive drilling and 1:1 coaching.
I was able to find another Veloster N with a 6-speed manual and a factory warranty! Unfortunately it has almost twice the miles and it’s a year older, but there were only two in the country to choose from. I would have taken any color, but it’s red, same as the last one. Insurance almost covered the cost of replacing it, but covered none of the performance or aero parts.
AAA sucks. I documented 14 pages of parts that I’d added, with receipts and links to articles here on this site, but many were DIY things I designed and developed and tested in the wind tunnel, and they didn’t give a hoot about any of it.
They also said the upgraded suspension, FRP hood, Konig wheels, RS4 tires, Pagid brake pads, camber bolts, alignment, and all the other stuff are “preference items” and don’t increase the value of the car. Also, anything over a year old is not insured. They were willing to consider the tow hitch (which is a custom job I had to weld up), but in the end they said the best they’ll give me is permission to drive to the junk yard 90 miles away and remove it. Thanks man, it’s the thought that counts.
AAA says I can take the tow hitch. Uh… might be a little bent?
If AAA sucks, maybe New York laws suck even more? There’s something called a No Fault law that provides $50k reimbursement for medical, my car, and loss of work. But I can’t go after the person who did this unless I’ve lost a limb or been permanently disfigured.
And because the fire and rescue driver was responding to an emergency, I’d have to prove he was grossly negligent and driving recklessly. Which is really the lawyer’s job, but they aren’t interested because of the No Fault law.
No fault. Indeed. Your job is to rescue people, not put them in the hospital!
I’m going to be a better person at the end of this, I swear. But right now, I just want to swear.
My wife has had enough of the swearing and moping around and set up a Go Fund Me. I feel sheepish putting this on here, as I feel like these things should be for people truly in need. So please feel no obligation to contribute. [Update: Go Fund Me is now closed.]
One of the things I’ve learned about this car is that it’s more fun on crappy tires. So rather than doing a lot of aero or modifying it for setting a fast lap time, I’m going in the opposite direction: less aero and slower lap times. No more fussing about weight, or power, or sticky tires. In fact, I’ll probably track it on all-season tires most of the time.
[Update #2] Here’s the new car, still waiting on the plates.
Going lighter on aero this time around and making it more for sliding around and having fun.
Instead of lowering springs I got a Mando ECS10 to make the suspension softer. I’m going to figure out how to lift it. And then maybe I’ll do a rallycross or something? Anyway, a new adventure beginning, and more time will be spent on Falconet for track use.
Thanks to everyone who contributed to the Go Fund Me, I will send you out wind tunnel reports and maybe I can run some OptimumLap simulations or do 1:1 aero consultations with you.
I was data coaching at Watkins Glen today. I went back into town to get a few things and on the way back to the track, I heard a siren behind me. I pulled to the side of the road and stopped, as one does to let them pass.
The emergency vehicle’s driver was distracted or didn’t see me, and plowed into the back of my car at full speed. My car is totaled.
Hit from behind and then pushed into a fence.
It was a fire and rescue truck, perhaps a Chevy Suburban. I saw it coming the whole time. In disbelief I watched the distance close. And then the impact. It was loud. Jarring. Painful. The airbag didn’t go off, and I got a face full of my spectacles on the steering wheel.
Some would say, “an improvement.”
This comes at such an inconvenient time; I’m coaching with Ross Bentley and Peter Krause next week, doing a new program we’ve developed. Luckily my brother Ian and co-coach Josh Herbert are helping with last minute stuff. But, man, super unhappy about this turn of events, and the timing!
The front got munched on a metal fence post.
I’ve put so much work into this car. It’s truly gutting to see this mess. And to know it’s not coming back.
That’s a huge impact. Hyundai makes a solid accordion car!
There’s a 90% chance I get another Veloster N and just hit the reset button with everything I’ve learned. I probably won’t be able to get another 10-year track warranty, though.
World’s smallest splitter has become one big splinter.
I used to track a ’93 Miata, and it occurs to me that if I was in that car, you might be reading my obituary. The height of the bumpers on the truck that hit me are considerably higher than standard bumper height, and I probably would have been steamrolled. Now that I’ve crash tested the Veloster, I’m more convinced this is the right car for me.
I feel like an airbag should have gone off.
I’ve been to the ER twice, and passed the CAT scans and X-rays tests. There’s no glass in my face, even though it feels like it. My back and neck are pretty sore, and my left shoulder is lower than my right. This is going to some time to come back from, both mentally and physically. But I’m also thankful; it could have been a lot worse.
Tell your loved ones you love them. Appreciate the time you have. We are lucky to be here.
Update 7/17: I think I found my replacement car in Florida. Same color, 6-speed manual, Hyundai certified (100k powertrain warranty). It’s a 2021 not a ’22, and it has almost double the miles, but I don’t want a DCT, and there just aren’t many to choose from.
In this post I show you a small splitter I made for my Veloster N. It’s about the size of your typical aftermarket front lip, but it’s much better performing, because it leverages all of the wind tunnel work I’ve done.
If you’ve read my wind tunnel reports, you know that 9 Lives Racing splitter diffusers on a Miata made only made about 8 lbs of downforce at 100 mph. This is pretty underwhelming, but they also don’t have a lot of area. Meanwhile, curving the entire trailing edge of the splitter upwards (essentially making a splitter diffuser across the full width of the car) made 150% more downforce (added 60+ lbs) on my Veloster. And so those are the numbers behind why I have a 36” wide central diffuser in my undertray, and not skinny little ramps in the wheel wells.
My aluminum undertray compared to OE plastic. I tape over the oil filter hole, of course.
I chose to make the splitter quite small, it measures just 36” wide, so that it fits between the two big black chunks of plastic on either side. The splitter is also very short, only 2” lip maximum, and the sides taper back to follow the lines of the body. Area-wise, this is definitely the world’s smallest splitter.
36” x 2” is the smallest splitter I’ve made, by a long shot.
You might wonder why I chose such a small splitter lip. My recent experiments with rear aero have shown that my car goes faster and faster with more rear downforce. So if I were to use a big splitter, I might need a 70” wing to balance that out. Maybe a double wing.
I don’t want to do that because this is a dual duty car, primarily for coaching and DEs. I don’t use the car for any kind of competitions, and so it’s more important that I don’t ground out the lip on driveways, or hit logs at my mom’s house! I also want the car to look cool, and I don’t like huge wings on a street car, but that’s just an aesthetic thing.
The splitter is bolted to the undertray and fits between the undertray and front fascia. From this angle, you can see how much diffusion the undertray has. This allows air to expand and lowers the pressure in front of it.
The splitter fits in between the undertray and the middle fascia. There’s quite a gap here, and so I tripled the thickness of the leading edge to 3/4”.
Thickening the lip to 3/4” allows a very large roundover, which helps keep air attached.
It’s easy to add extra thickness to a splitter and round it over, and with that you get better flow attachment. I don’t know why I don’t see more people doing this. Also worth noting, the big roundover made the splitter about a pound lighter!
Front lip rounded over and strakes riveted on. It’s ready to mount on the car.
I also decided to add some vortex strakes to send air sideways. Typically these go directly in front of the tire, which reduces tire squirt. But because my undertray is narrower than a full-width splitter, only the outermost strake lines up with a tire. The other two strakes should direct more air to the brakes.
Strakes direct some air to the brakes.
To mount the splitter and undertray I use speed clips through various parts of the fascia, long bolts to the radiator bottom (where the OE undertray mounts to), and splitter rods in front. The splitter rods mount to the bumper beam using U-bolts and other hardware from the Lowe’s racing department. All of the metal parts to mount the splitter are less than $50.
Splitter rods are all-thread, mated to a turnbuckle at the top and a T-bolt underneath. Speed clips replace all of the pop fastners around the outside.
The splitter is meant to be set for maximum ground clearance, so that it’s flush with the underbody. But the turnbuckles allow me to add some rake, and I could angle it down a few degrees if I want to. But like I said, this is mostly a street car and ground clearance is important, so I’ll keep it flush most of the time.
All said and done it weighs 11 lbs, and is built way stronger than necessary. But it makes the whole front end more rigid, and ties everything together nicely. The front fascia used to flex quite a bit if I pushed in on it with my foot, and now it’s much more solid.
Bottom view shows T-bolt, strakes, and flush mounted height.
Which brings me to the final step, which has nothing to do with fixing my mishap, but something I’ve been meaning to do for a while: add canards. I’ve tested canards in the wind tunnel on a couple occasions, and while I liked the results, the front fascia felt a little too flexible for my tastes (the bottom canard would flex downward at speed) and so I never used them on track.
I have two sets of canards, some commercial productsfrom Verus and DIY ones I made. In the wind tunnel, the CFD-designed Verus canards made only 15 lbs of front downforce at 100 mph, compared to the 85 lbs my canards made. However, my canards were pretty draggy and the Verus ones had zero effect on drag or rear downforce. So I figured I’ll mount the Verus ones permanently, and make mine removable, and use them only when I’m at the track.
Note that the reason my canards made more downforce (and drag) is simply the location. In all of the wind tunnel data I’ve accrued, the bottom canard does the work, and the upper canard helps reduce the drag in the entire system. It’s important to get the bottom canard as low down as possible; mine go on the ridge of red plastic, at the bottom of the bumper fascia. (See more in my blog post on canards for more details.)
How much downforce will these changes make? I’ve optimized the splitter in several ways, but it’s too high for there to be much ground effect, and the splitter lip itself is rather short. Let’s call it 50-60 lbs of downforce at 100 mph, plus maybe 110-120 lbs for the canards and hood vent. Both of those items help with splitter extraction, so they go hand-in-hand.
Around 170 lbs of downforce at 100 mph is about the same as my 12-degree full-curve splitter made alone (without canard or hood vent), which I felt was pretty decent. However, the bottom canards are doing half the work, and they are draggy sumbitches. At a slow track like Pineview, I’ll surely use the lower canards, but I’ll have to A/B test them at Watkins Glen before knowing if they are beneficial at a fast track.
This was a good stopping place for the day, and I think I earned myself a cold one (Best Day Brewing N/A West Coast IPA.)
Last year I borrowed Phil Sproger’s NA Miata so that I could test various parts in the A2 wind tunnel. Phil’s car has the full catalog of parts from 9 Lives Racing, and so we were able to set up the car in real life exactly as it’s been tested in CFD. But we didn’t stop there, we also tried various combinations of parts, including the following:
Splitter diffusers, spill boards, and tire spats.
Canards in various locations and combinations.
Closed windows versus open, plus modifications to reduce drag and turbulence from the open windows, including wickers, mirrors, and venting the rear window in two different locations.
Singular hood vents fender vents.
Brake ducts, NACA ducts.
OEM hardtop with and without a rear window spoiler, versus a CCP fastback.
Blackbird Fabworx spoiler at different angles/heights.
Wings from 9 Lives Racing, Wing Logic, and a couple prototypes.
The report is over 50 pages long and includes all of the wind tunnel data, plus explanations on how each part contributes to drag and lift. I then assemble the parts into logical builds, so you can see the lift-drag ratio and aerodynamic balance of the entire car. And I give you the details on how to do that, so you can build your own custom Miata and know its values. The report concludes with racing simulations using Miatas of different power levels on different race tracks, so you know the most important thing: how different aero parts affect lap time.
The Miata wind tunnel report costs $35. After paying, you’ll get a download link.
I talk about the wind tunnel test and go over some of the results in an episode on The Blind Apex podcast. If you listen to that, you might wonder “where’s Johnny?” Well, we invited Mister 9 Lives Racing, but he was a no-show! Which was surprising because he was a guest of mine at the wind tunnel, and could have provided some color commentary. I also had some follow up questions on how we can use my data as a feedback loop for better CFD, and move the science of Miata aerodynamics forward.
I share some of the data from the wind tunnel in the podcast, but please don’t take that as a sign that you should do the same. It was expensive and a lot of work testing parts and putting the data together in a whitepaper. Thanks for caring and not sharing, and supporting this website and future wind tunnel tests.
It’s been a while since my last article and you might be wondering what’s going on over here. Mostly I’ve been doing non-aero experiments based on driver mod, rather than car mod. I’ll report back on that when I have some reliable data, but the spring weather (rain) has made data gathering difficult.
As part of this driver mod theme, the last couple weeks I’ve volunteered as a mentor to do instructor training with NASA and Hooked on Driving. The programs are a little different, but the mentoring part is quite similar, and quite a lot of fun.
National Auto Sport Association Northeast
Last week Ed Cangialosi of NASA Northeast sent out a plea for additional mentors to help with their instructor training event at NJMP Lightning. My wife was out of town, so I heeded the call. This wasn’t entirely out of generosity, as I always learn some nugget from my student or other coaches.
The trip down to Millville is 275 miles, and I averaged 32 mpg in my Veloster N. It occurred to me that I burn more gas in two 20-minute track sessions at Watkins Glen than I do driving all the way to NJMP. Wild.
The NJMP instructor training is designed to get new instructors into the game. I don’t believe any of the students have instructed before, and so this is to simply gain provisional instructor status. And NASA needs a lot of instructors, partly because of Toyota’s GR program. If you buy a new GR86, GR Corolla, or GR Supra, you get a free NASA track day. This puts a lot of pressure on the already oversold novice group, and so many more instructors are needed.
I applaud Toyota for having a program like this, but “Oh What a Feeling!” it would be if they could build some track-ready cars! The Subaru engines in the GR86s are well known for their oil pan problems, and Toyota is well known for not warrantying them. So why do they give away a track day for a car that can un-warranty itself on track? The GR Corolla has a more reliable engine, but it’s also known to overheat its rear diff, thereby becoming a front-wheel drive car after a few hot laps. I guess the GR Supra is the car to have then, but the one GR Supra student I instructed had already done a 550 hp Stage 2 tune before his first track day, and while I’m not going to paint all the Supra owners with the same brush, I have seen more than a few novices with modified Supras.
But to get back to the point of this post, to become a NASA instructor, there’s some pre-work you must do, and then take the in-person class. During the class the chief instructor hits the high points in a lecture format, but most of the education is hands-on with mentors.
As a mentor, my role is to drive the car as a student would, play-acting three types of student. The instructor candidate (my student, as it were) sits in the right seat and coaches me through some typical personality types.
Novice, timid – The first type of novice is what we refer to as a never-never, as they’ve never been on track before. They don’t know the limits of their car, they don’t know the jargon, and may not even know how to buckle their helmet.
Novice, aggressive – The next type of novice has more confidence than skill. They have strong inputs that engage the car’s stability and traction control systems, and they don’t listen to you. This is a deadly combination the instructor needs to control.
Novice, skilled – This driver has been to a couple track days and has the skills and jargon down. They need more refined instruction, and might be ready to advance to the solo group if they have awareness and play well with others.
Hooked on Driving MSF Level 2
The Motorsport Safety Foundation is a national organization that seeks to formalize the requirements for different levels of instructors, by establishing a path of continuing education, and providing a formal record of achievement for HPDE instructors.
Unlike the NASA training, MSF Level 2 assumes the student has done some instructing before, and passing this course means you earn more than a provisional status; MSF2 means you can instruct with any group that recognizes these credentials. And that is a lot of HPDE groups nationwide, and so earning your MSF2 is a pretty big deal.
For that reason, the training is more intense. There’s a full day of classroom, with slides, discussion, and poignant stories from the trenches. Added to that, you also have six sessions on track with a mentor, who now plays six different roles, not just the three with NASA.
Hooked on Driving did their training at Pocono North, which if I’m being brutally honest, isn’t a great track. The transitions from the banking to the infield are abrupt, and lowered track cars with splitters will absolutely hit bottom here. The banking is fast, the infield is slow, and there’s not enough variation to get anything more than a low grade from this sourpuss.
Anyway, I wasn’t there to drive the track, I was there to instruct two students. I started the day with one student, and then half way through the day, switched students with another mentor. This is something HoD does, and I love it. Not only does the instructor candidate get two different coaches, but there’s less chance of favoritism when it comes time to pass or fail the student.
Not every student passes, and it’s difficult to be the mentor that fails their student. In fact I didn’t pass mine, I asked that he work with HoD a few more times and get more mentoring before he gets his MSF2 badge. I know he’s going to be a good coach, but that level of certification has to mean something, and my guy was close, but not quite over the bar.
Play acting
I come from a family of actors. My dad was an amateur actor his whole life, and my sister Mia Korf (IMDB) was on soap operas and other stuff. So maybe it comes naturally to me to act out the characters, and so I do some pretty funny shit in the car. I’m still laughing about it now.
When the instructor says “open the wheel” I take both hands off the wheel.
At times I won’t listen to anything the instructor says. I’ll shut down and say nothing when they ask me questions.
I’ll shift to 2nd and 3rd and back to 2nd in the middle of a corner.
I leave my hand on the shifter the whole time.
I steer with one hand on the wheel. Or I’ll shuffle steer, and then justify it by saying that’s what Randy Pobst does.
I hide my phone under my left leg and time my laps.
I leave things in the car that rattle around, usually under the passenger seat.
I intentionally drive off the track and into the grass at least once. Sometimes twice.
There are other things I do, but they don’t come out of left field. I make sure I’m in character during the pre-drive interview, and even while waiting on the grid. “Oh, that’s Mikey ahead of us, he says he’s going to do a 1:20, so we’re going to do a 1:19.” If you hear that shit, we’re going flat out.
Advice
Get your MSF Level 1 or sign up with NASA and become a provisional instructor. I started my journey with Chin Track Days, but there are many organizations that will let you instruct with them, and then mentor you along the way.
As an instructor, you’ll start seeing patterns and learn to anticipate your student. The most common safety issue I see in novice drivers is that they don’t brake hard enough. Many are afraid to engage the ABS, and so they carry 30% brake pressure all the way into the corner, and beyond. If you’re a new instructor, get the students to engage the ABS as early in the day as you can. Once the student realizes they can’t break the car by braking, they’ll achieve a necessary level of confidence that both of you will benefit from.
Another thing I do with most students is to stop them from shifting altogether. If it’s possible, run the entire track in one gear, as tall as possible. This one change can keep a hot-head under control, and can later be used with intermediate drivers to make them corner faster.
The one thing I disagree with in HPDE curriculums is to “drive the racing line.” But because I’m there to instruct under their umbrella, I abide by their rules. Mostly. The one subtle rebellion I have is that I don’t make students steer to the track-out cone. I just don’t see the purpose of adding distance for no reason. Instead, I get them to gradually add more and more throttle, and eventually we find the corner exit as a natural course of action. Now the racing line becomes a result of inputs, not a goal in and of itself.
The most common problem I see with new instructors is they provide too much information. For example a recent instructor candidate told me to “get on the brakes early to load up the front tires so that you gain more traction for turning in.” I mean, yes that’s true, but that’s a conversation for the parking lot. Instead, tell the driver to make an input based on a reference. My version of his directions was “Brake at the bridge.”
Becoming an instructor has many benefits, not the least of which is free track time. But you’ll also find the rare student that suddenly “gets it,” and the pride is felt equally in both seats when that happens. Believe me, this is as rewarding as setting a personal best lap. Through this journey you’ll also become a better driver yourself. Because, as the saying goes, “You never really know something until you teach it.”
I plan on doing a few tests at Watkins Glen International this year, and this is the first of (I hope) several reports. WGI is 25 miles from my house, and I have a lot of opportunities for free track time with groups I instruct for (Chin, HoD, MT, NASA, PCA, SCDA). The Glen is convenient for testing, and the corner speeds are high, making it a good place to test aerodynamic downforce. But the straights are also quite long, and so drag matters a lot here.
However, as a testing venue, WGI is only so-so because it’s a popular track, and so there might be 40+ cars in your run group. It’s not just the traffic that’s the problem, but highly variable weather, and virtually no runoff. So it’s not uncommon to lose a session due to rain, fog, or because someone hits a wall. For this reason, I may need to use sector times and predicted laps in order to get good data.
But on April 13th we had decent weather and lots of space on track with Mass Tuning, and so I was able to test different things and get clean laps. But because I made aero changes each session, and was also playing taxi (guest rides), I wasn’t able to string together many hot laps, and so there are only three decent laps from each aero configuration.
Another reason for the lack of hot laps is it takes a full three laps to get my tires up to temperature. I start them at 27 front and 30 rear, and at the end of the session they are properly at 37 F and 36 R. But I lose a lot of time early in the session, braking and accelerating hard to get the temps up, while keeping corner speeds low. If I’m impatient and corner hard at less than 32 psi, the tires roll over and I cord the outer edge. I learned this the hard way last year, getting only about 1/3 of the life out of my tires.
In the future, I may start the tires at 30 psi, which should put the tires out of the danger zone after one lap. This would result in the front tires being 40-42 psi for most of the session, and while this won’t return the best lap times, I should be able to normalize the data better.
The long warm-up times are probably the result of not enough negative camber. In fact I never hear the tires working hard, they just make a low-frequency moaning sound, not the screeching howl I get out of my Miata on the same tires. The Veloster sits on lowering springs and has one set of camber bolts, and this maxes out the camber at -1.8 degrees. With offset bushings (or slotting) it’s possible to get -2.5 degrees, but any more than that and I risk blowing the OE struts. This is still a street car, and I’m not going to ruin it by turning it into a track car with coilovers, stiff sway bars, and solid bushings. Well, not yet anyway.
OE N wing 2:17.4
In the wind tunnel, the OE wing (or it could equally be called a spoiler) made 30.8 lbs of downforce at 100 mph. This cancelled out the rear lift on the base model hatchback, giving the car a slightly negative coefficient of lift. This is very rare in a street car. In my article Thinking in Aerodynamic Coefficients, I show that most cars have positive lift of around Cl 0.1-0.2. Cars that have zero lift or even some downforce are rare, and usually an exotic sportscar. But the Veloster N is one of those unicorns that has downforce straight from the factory.
The OE wing does a good job of stabilizing the rear of the car, and I felt none of the rear instability issues I felt when testing the car with the base model roof extension (meaning no wing). That was a test I ran at NYST, and the rear lift made the car twitchy under braking going into T1 and T5. But at WGI where I was 30 mph faster, the N wing felt stable under braking, and so the OE wing is definitely an improvement on the base model’s featureless roof extension.
The N wing also had the least drag, and posted the fastest top speed of 132 mph on the back straight. But the OEM wing also had the least downforce, and you can see this on the speed trace below, where I compare the OEM wing (red) to the same wing with a Gurney flap (blue). Note the difference in speed through the esses, it’s huge. But also note the vMins in each corner that I’ve circled. Not only are the vMins higher, they are shifted to the left, indicating that I’m backing up the corner better when the car has more rear downforce.
Red is OEM wing; Blue is the same wing with a Gurney flap.
Wicker kicker 2:16.4
I put a 1” tall Gurney flap (wicker) on the trailing edge of the N wing, and in the wind tunnel this made 114.7 lbs of rear downforce and 8.9 lbs of front downforce. The fact that this made front downforce means the wing is behaving more like a spoiler than a wing, and the combined 123.6 lbs of downforce is a lot more than I would have thought from adding a simple Gurney flap. When I add a Gurney flap to a wing I usually get 150-170% more downforce, not over 400% more! With an increase in downforce, there’s more drag, and 8.3 hp is lost at 100 mph.
Wicker-kicker Gurney flap thingy.
But the extra grip through the esses is worth it, and by the time I get on the brakes for the bus stop, the max speeds are identical with or without the wicker. Around the rest of the lap, the modified OE wing has more grip, and the lap times are 1.1 seconds faster, lap after lap.
I uploaded a video of three consecutive laps, and I apologize in advance for the shitty audio. I use an original SmartyCam with an internal mic, and I can’t figure out how to make it less terrible. Anyway, I do a brace of identical 2:17.031 laps (down to the thousandth of a second!), on either side of a 2:16.415.
Three laps, turn the audio off.
S1223 54×11 wing 2:16.8
In the wind tunnel, my DIY Selig S1223 wing made 179.5 lbs of rear downforce at 100 mph, and 31.6 lbs of front lift. Wings are located higher and further rearward than spoilers, and so it’s normal for wings to lift the front end through leverage. As a practical matter, this is why people typically use a splitter and a wing together, but a spoiler can often be combined with a simple airdam, or just used on its own.
The wing’s total downforce of 147.9 lbs, divided by 45.3 lbs of drag force, returns a 3.27:1 L/D ratio, which is about the same as the OE wing, but not as good as with the Gurney flap. The drag amounts to 12 hp at 100 mph, and on the back straight of Watkins Glen, the wing is 2 mph slower than the other options.
One of the reasons the wing isn’t very efficient is because it has a span of only 54”. This is problematic for two reasons: first, wing-tip vortices have a detrimental effect on the wing’s overall performance, and so the greater the aspect ratio, the better the wing performs; second, most of the wing is in the silhouette of the hatchback roofline, and so very little of the wing is in clean air.
The reason the wing is 54” is because I wanted a cheap way to test a dual element wing, and bought a $35 extruded wing as the upper wing. The dual wing worked well at Pineview Run and NYST, setting the fastest laps (.8 seconds and 1.0 seconds faster than the single wing, respectively). But after seeing the initial results from the single wing at Watkins Glen, I didn’t even bother attaching the upper wing, as I’m sure the dual wing would have been the slowest.
Single wing has rather tall end plates, as they are used to hold a second wing element. I didn’t try the dual wing option at WGI.
Under most racing rules, wings are allowed to be body width, which in the case of the Veloster N would be 71.7”. I tested a 70” Wing Logic wing in the wind tunnel, and it made the same amount of total downforce as my 54” DIY wing, but with 44% of the drag. The resulting 7.4:1 L/D ratio shows how important it is to get the wing tips as far apart from each other as possible, into clean air, and away from the hatchback roofline. I may test this on track in the future.
Conclusions
Based on testing the single wing at Pineview Run (.7 seconds faster) and NYST (1.5 seconds), I felt for sure the S1223 single element wing was going to be worth 3 seconds at WGI. The fact that it was only worth about 1 second has me tail spinning into the D-K pit of despair. Let me think about why that may be….
One reason for the lack of performance may be the aero balance. As mentioned, the wing adds rear downforce with a lot of leverage, and so it was the only one that lifted the front. The result is the car may have too much rear aero bias, and not enough grip for turning in. Perhaps when I put a splitter on the car, I’ll see better results.
But note also that the aero balance was even worse at NYST and Pineview. The fastest rear aero on those tracks was the double wing, which I didn’t measure in the wind tunnel, but certainly has the most downforce and drag.
Another reason the wing underperformed could be the driver underperformed. I don’t believe I was cornering hard enough, and just barely edging into the performance envelope where aero is adding to what the tires alone can give. With more laps and coaching, we may see the driver perform better, and with that, the results may change.
Of course drag is a factor at Watkins Glen, and the 54” wing had the most drag. If I swapped to a 70” Wing Logic wing, I’d gain the equivalent of 6.7 hp (at 100 mph).
But as it sits now for both car and driver, the easiest and cheapest way to go faster in a Veloster N is to put a 1” Gurney flap on the OE wing. The performance of this modification was one of the biggest surprises in the wind tunnel, and it’s nice to see that reflected in real life, as well.
Three hot laps from each configuration went like this:
Config
Best lap
Average
Predicted
OEM wing
2:17.434
2:17.91
2:16.841
OE + wicker
2:16.415
2:16.83
2:15.783
54″ wing
2:16.795
2:17.02
2:16.284
While I have some work to do to get faster, I’m a consistent driver. Watkins Glen is a long 3.38 miles, but my lap times are usually within a couple-three tenths of the previous lap. That’s around a 0.3% difference, and probably similar to the noise you’d see in other variables that change throughout the day, such as air and track temperature, wind speed and direction, etc.
<brag>On a track that I have more laps on, like Pineview Run, I’m a metronome. Here’s six laps in a row with 0.372 seconds between all of them. If you throw out the fastest and slowest, I do four laps separated by 0.076 seconds.</brag>
Consistent laps matter for testing.
I mention this not just so that I can thump my chest, but to throw some validity on track testing aero components. There are a lot of variables that change throughout the day, or even within a single track session. As those tolerances stack up, lap times can vary a lot. However, my driving is probably less of a factor than you might think, and I just want to point that out so I can deflect some of the “you’re driving like shit” comments. (Which are true, but at least I’m consistently driving like shit.)
Future tests
For Watkins Glen to be useful as an aero testing venue, I’ll need to fix my attitude, and turn my frown upside down. I’ve never gotten along well with this track, and I have to get to the point where I enjoy driving here. With some help from Gregg Vandivert (Omega 13 Coaching) and many more visits to the track, perhaps I can flip this script.
[sigh emoji] I hesitate to list the tests I want to do in the future, because the best laid plans of mice and men usually amount to the same dung heap of disappointment. But if things go as planned, I’ll test the following:
Driver mod – I’m curious to see how much time I can lose from professional coaching alone. In order to A/B test myself, I’ll need to use the same baseline setup and pray for days with similar weather.
Splitter – I didn’t use a splitter in any of these tests because I was matching the same setup I already tested at Pineview and NYST. But now that I’ve done those tests, I can put a splitter back on my car and see how this compares to the wind tunnel, and how getting more front aero load helps the overall balance. Naturally I’ll need to test the splitter at various heights and angles of attack, and so this could be a whole day of testing on its own.
Canards – I tested canards in the wind tunnel on two different occasions, and now it’s time to take the best results and see how they do in the real world.
OE wingmodifications – The 1” wicker worked great on the OE wing, but I’ve only tried the one size. Next I should try 1/2” and 1.5” tall, and see what happens. Also, I believe that getting the OE wing a little higher may increase performance, and with that, I could also add some angle of attack. This should be as simple as installing a few shims beneath the wing.
Ducktail spoiler – I tested a DIY spoiler at Pineview and the results were similar to the single wing. But WGI has already proven to be quite different than other tracks, and so I should test this one.
Bigger wings – The wind tunnel already showed me how important wing span is, and so a 71.7” wing will be a lot more efficient. I’m building a big wing now, and shall test this for sure.
Active aero – I wrote an article on active aero, in which I did racing simulations at Watkins Glen. Now it’s time to put my money where my mouth is and do the same experiments in real life. The easiest active aero to fabricate would be a DRS spoiler, so I may start with that rather than jumping straight to a dual wing.
Diffuser – I tested a diffuser in the wind tunnel and it was pretty lousy, losing about as much front downforce as it made in the rear. But the A2 wind tunnel doesn’t have rollers for the wheels and so the effect of the underbody can’t be 100% trusted. So I should probably test the diffuser IRL and shoot myself up with another dose of disappointment.
vMins and driver performance
If you saw my previous post on vMins, you may wonder how well I performed on the vMin table. I put the OE wing on the Street table, and the modified wings on the Track side.
The red circles indicate my vMins with the OE wing, and you can see that I’m OK, but not great. My T7 may be a bit high, or more likely, all of my vMins can come closer to that level. However, when I add rear aero, I park it in the bus stop! I’m probably using the same braking marker, and then just over-slowing the car with the extra downforce and drag. I expected my vMins to be a lot worse than this, so I’m pleasantly surprised that I’m not driving like shit. And at least I know what to work on next time.
Gran Turismo nerds
I prepared for my Watkins Glen weekend by doing some laps in Gran Turismo 7. GT7 doesn’t have a Hyundai Veloster N in the game, so I’ve been using a Scirocco R. Now that I have data to refer to, I can modify the car to be as close as possible to the real world.
In the game I set the car weight to 3000 lbs and the horsepower to 244 (buy the lightweight modifier and add ballast weight and a power restrictor). This approximates the dry weight of my car (I have lighter wheels and the rear seats removed) and what my car puts out on the dyno (my engine is bone stock, not even a cold air intake). More importantly, this also gives me a realistic speed of 132 mph on the back straight. (I’m sure you could use a heavier car with more power, but this is just what I landed on.)
I also fitted the non-adjustable Sport suspension because I have lowering springs, and added the Sport brake pads (which do nothing in the game, but I’m trying to match the car IRL). I added a rear wing so that I can adjust the rear downforce from low (OE N spoiler) to high (single wing). GT7 doesn’t simulate aero accurately anyway, but I added no other downforce because I’m not using any. I use Comfort Soft tires, which puts the corner speeds in the right window for Hankook RS4s with painfully not enough negative camber.
If you have GT7, make a Scirocco R like this and play along. The PP value should be 492 with the wing maxed out. You can drop a comment here or contact me if you want to compare notes.
With the Scirocco R set up like this, I can click off low 2:14 laps regularly, and get the occasional 2:13 at WGI. That’s 3 seconds faster than I’m doing in the real world, which seems about right seeing as I can drive much harder in the game than IRL. As I get better in the real world, I expect the lap times to get closer to GT7. We shall see.
If you made it this far, thanks! If you’d like to support more content like this, hit the Buy Me a Coffee link. If you’d like the Veloster Wind Tunnel Report, it’s just $25 and goes through a ton of stuff you can do to make your car faster.
TL;DR: A long story on how I got started with data, became frustrated with data coaching, and developed a simple tool for self coaching. You can skip this epic tale and print the PDF of the Watkins Glen vMin table.
I’ve been a data coach for a few years, and everyone I’ve coached has made massive improvements. Most of the time data coaching takes just one or two sessions, and on average, I’d say people go about a second faster per mile.
On a short track like Pineview, that’s only one second, but at Watkins Glen, that’s three seconds. Three seconds is like going from a track tire to a slick. Or adding aero. Or buying a lot of horsepower. Any of those things could cost $2000 or more, and so getting three seconds for free is phenomenal. Yes, free.
For a couple years I was the lead data coach for the Niagara Region Porsche Club of America (NRPCA), and we offered free data coaching. While PCA clientelle can certainly afford paying for it, we were trying to lower the bar and get more students interested. Nevertheless, I’d have only one or two students per day.
This became a discouraging waste of time, and so last year I took a hiatus from data coaching. I’ll get back to that again, but in the meantime, I want to share something I invented, which is a vMin table for self-coaching. It looks very simple, but there’s a lot of research and effort that went into it.
It’s a long sorry, hence the TL;DR at the top. I won’t blame anybody for simply clicking the link and moving onto using the tool. But for those that want the backstory, it begins like this:
I never wanted to use data
14-odd years ago I did one or two HPDEs with Hooked on Driving in California, and after seeing a shit-heap of a car on its way to race at Chuckwalla, I convinced my brother and a couple friends to buy a first-gen MR2, cage it, and race it in the 24 Hours of Lemons. For several years, everything I knew about performance driving I learned with the bit between my teeth, going wheel to wheel in an underpowered car, trying to catch faster cars in the corners.
I was good, always setting the fastest times on my team. When we acquired Ben Dawson (Dominating with Dawson, on Garage Heroes in Training), I was able to match his times, but not better them. Then my identical twin brother became a real student of the game, reading Speed Secrets, getting into data, sim racing, instructing… and he left me in the dust.
After being the fastest on the team for a couple years, I was now a few seconds slower per lap. Ian encouraged me to follow in his footsteps, but I countered that I wanted my skills to represent the everyman driver. It could maybe turn into a twin study, where one brother used data, the other didn’t, and look where they ended up years later.
That’s how much I didn’t want to use data; I was willing to let my identical twin trounce me on track, handicap our team’s position in the final standings, pretend to do a bogus twin study, and skill-shame myself with a milquetoast modifier like everyman.
(When reading this article to my wife, she said if people only knew how competitive I am, the fact that I would let my twin brother beat me is incomprehensible. This is the exclamation point on how much I didn’t want to get into data.)
The transition to using data happened sometime after leaving California and starting a new team in New York. Ian flew out a couple times to race my new Miata, and he’d annoyingly show up to some unseen track, and destroy the team’s best times in his first few sighting laps.
It took a while, but the everyman got tired of his everyday beatings, learned how to use data, and then began to coach others through the same process. The track to hell is paved with good intentions.
Data coaching barriers
Data coaching hasn’t taken off for me like I had imagined it would. The reason for that isn’t a hardware problem, since I bring six Aim Solos and assorted brackets with me to all events. Neither was it a financial hardship, because I offered it for free. And nobody ever lost a driving session, since I’d schedule their coaching sessions around that. There were literally no barriers to data coaching, and so the reluctance was puzzling.
Part of the reason data coaching didn’t catch on was due to PCA itself being different from other clubs: we weren’t supposed to talk about lap times, none of the senior instructors were using data or pushing it into the curriculum, much of the clientele were older and skeptical of things new, and in any national organization, change moves at a snail’s pace.
There’s also the “I’m just here to have fun” contingent that comprises most of the people that sign up for a HPDE, and I totally get this group. If I went to a pickup basketball game and some “coach” was on the sidelines saying he could run me through drills between games, I’d walk the other way. The fact is, most of us sign up for track days to have fun with friends, and while the E in HPDE stands for Education, we are really there for the E in Enjoyment.
However, what I’ve found through years of trying to get people to dip their toe into data, the largest barrier to data coaching is this: nobody wants to know they are slow. And they don’t want anyone else to know that either. Pride is the wall that stands between having fun and and going faster.
The real barrier to data coaching.
The thing is, I haven’t met any slow drivers. Everyone has at least one corner where they are fast, and we all have one or two corners where we are slow. You use the former to build on the latter. And that’s what data coaching does – it finds your strengths and creates an individualized strategy to address your weakness. People call it driver mod and it’s a good term, because it’s something you take to every car you drive.
But because pride is more delicate than money is valuable, people will spend thousands on car mods and won’t accept a free driver mod. This was doubly frustrating because NRPCA HPDEs are mostly on weekdays, and so I was burning vacation time to help one or two students per day. So I pondered the incredible value of data coaching, against the struggle of getting people to try it, and I realized this: people will only try data coaching if they can do it in private.
Technological barriers
I wasn’t the only one who figured this out, Garmin did, too. Their Catalyst product combines AI coaching with data privacy in a compact package that doesn’t require a laptop or a professional to review the data.
I read the Garmin Catalyst review by Mark Petronis at AMT Motorsport, and that should have deterred me from buying one, because I’m already an advanced data user. But a virulent case of FOMO swayed me into buying one and experimenting with it for a couple months. I found the following shortcomings.
The Catalyst doesn’t actually know where you should brake, turn in, accelerate, or track out. It only knows what you’ve already done. so you’ll never know the fastest way around the track, just the least-worst that you’ve done so far.
If you drive consistently, the Catalyst won’t help; without variation, it returns nothing. You need to make errors in your driving that end up being beneficial mistakes.
The Catalyst uses only your recent driving. There’s no way to save and then upload best laps from yourself or other drivers to compare with. And so if you live for comparative data (driver mod, tire testing, aero testing, etc), you can’t A/B test jack shit versus shit jack.
You also can’t download your Catalyst data and then compare it with other people using different devices. This is infuriating, because the Catalyst is gathering GPS data. All I need is a CSV file so I can dump that into Race Studio, but Garmin refuses to play well with others.
The coaching “opportunities” were mostly good, but occasionally had advice I disagreed with, or missed some obvious wins. For examples of that, see Gregg Vandivert’s article When Garmin Catalyst leaves time on the table.
The audio hints are fun and come at a good cadence, but the advice has nothing to do with skill, just to drive with more gusto. As useless as the advice was, some of it remains memorable; I can still hear her say “keep pushing,” but it’s mostly her accent, and no longer in the context of driving.
The Catalyst isn’t plugged into the OBD2 port, so it doesn’t know the most basic information, like whether you had pulled your foot off the gas, braked lightly, or balked a shift. It doesn’t know what gear you’re in, so it can’t whisper “hey there big guy, don’t downshift before the next left, hold 4th gear.”
There are numerous other quirks or strategic decisions that are baked into the firmware or blocked from export, and I simply don’t agree with a closed system that reeks of rotten Apple products.
In the end, I discovered what the Catalyst is: an excellent delta timer and video camera. [sigh] Perhaps there have been updates that address these issues, and maybe I wasn’t using mine in a way that got the best use out of it. In any case, I sold mine and haven’t looked back. But if Garmin could export the optimal lap video as 10hz GPS data, I’d buy it back at twice the cost.
My requirements must not represent the majority, as I see a lot of Catalysts at the track. However, it appears that most of them are being used as a lap timer and video camera. I rarely see people removing the devices from their cars in between sessions, and so I doubt they are using the coaching opportunities during the day. I guess this is like buying a gym membership and then only using the sauna, but least they are in the right area for self betterment. I understand many people really like their Catalyst, but those people probably never had real data coaching, and accept the shortcomings as a tradeoff for a great UX and data privacy.
On the technological flip side you have Aim products, which sets the gold standard for motorsports data loggers. Unlike the Catalyst, you can plug an Aim SoloDL into your car’s OBD2 port and/or jump into the CAN bus and get brake pressure, throttle position, steering angle, and anything else the car reports on. The amount of data is staggering, and Aim’s decision to display everything by default (including useless data like the device’s internal battery voltage), is complicated and off-putting.
But the true barrier to using Aim products is the software. I’ve worked in software development for 25 years, for companies like Oracle, Salesforce, and Google. I know what good usability looks like, and the amount of effort it takes to take a very powerful tool and make it easy to use.
The Race Studio 3 user interface is a cornucopia of poor decisions, many of which could have been fixed using setup wizards to create custom profiles. Or hire a technical writer like myself to explain how to use three lefts to make a right. But there’s nothing intuitive, streamlined, or adequately documented in Race Studio 3, and so I use Race Studio 2 most of the time, which also isn’t great, but sucks less.
It’s also worth mentioning that unless you put a password on your Solo2, your data isn’t private. When I turn on my laptop in the Watkins Glen garage, I can walk around and pick up everyone’s Solo2 on the wireless network. Because the Solo2 is so damn confusing, most people are using it only as a lap timer, and so they haven’t set a password, and everyone within earshot can access that data.
At this point of the story we come to the realization that technology sucks, and that the best tool for introducing people to data doesn’t exist. If I wanted to get people to dip their toe into the waters of data coaching, I’d need to invent something with the following qualifications: free, private, and devoid of all technology.
[vMin table entered the chat.]
vMin data coaching
The intro to data coaching tool I came up with is simply a table for minimum corner speed, or as data coaches call it, vMin. Every track has thousands of laps of historical data, and my approach takes advantage of that knowledge, using the best drivers. The primary use of this table is for identifying which corners you need to work on. I’ll show you how to use the table in a minute, but first I want to explain why vMin.
vMin is only a single piece of data, so is it really that important? Well, after seven-and-a-half years of weekly email tips and tricks, Ross Bentley thought so. The final Speed Secrets Weekly #400 could have ended on any subject, but there’s a reason he chose vMin; it’s the thing that separates average drivers from great drivers.
Unlike lap times, which can be bought, setting an ideal vMin is 100% skill. It’s the one variable you can look at that separates the wheat from the chaff. It’s like a unique identifier for expertise, or a genetic marker for driving greatness. Here’s why:
Average drivers throw away entry speed without a thought, and that’s because their thoughts are on two things they learned in the novice group: late braking and early acceleration. When you’re adept at those skills, and combine that with a modern car’s nannies and horsepower, you can pass everyone in your DE group. Advancing to the fastest run group and passing the instructors is simply a matter of more power and better tires.
On the other hand, advanced drivers hoard entry speed like a dragon with its gold; they save it and defend it. Having already mastered late braking and early acceleration, they can instead concentrate 90% of their attention on the last 10% of corner entry. In that very small and compressed zone just before the apex, they make micro adjustments in a delicate dance that maximizes traction, minimizes steering angle, and positions the car for the optimal exit. And they do all of that while retaining the highest minimum corner speed. Herein lies the real art of driving, and if you want just one metric that separates the artistes from the poseurs, you can see it in a single number: vMin.
Another great thing about vMin analysis is that you can still get data in a DE session that’s full of traffic. Unlike looking at lap times, which can be spoiled by a single slow driver, you only need to hit each corner once perfectly, and you’ll have representative vMin data. Of course the more laps you do the more accurate your data is, but the point is, you don’t need to be enraged by backmarkers spoiling a lap, you can still get good data on a crowded day.
Data coaches may add the fact that while vMin is an important metric, where the car is at the point of vMin, and what direction it’s pointing, are actually more important. Yes, I agree wholeheartedly. But if you are able to get that kind of information, then you’re already using data, and this vMin table was not invented nor intended for you! But I will also say that that the where and what direction are already subtly baked into those numbers.
But let’s get back to vMin simply as a number, and not as a position or angle. By raising your vMin speed, you have an advantage until the next corner. For example, at Watkins Glen, let’s say you typically go through Turn 1 at 64 mph. If you can roll another 1 mph through the corner using a different line or technique, that might be worth a full second by the time you reach the bus stop.
Is a higher vMin always better? No. If going through the corner at 66 mph means you’re later to full throttle, then you might be a full second slower by the time you get to the bus stop. So the point isn’t to get the highest vMin in each corner, but the appropriate vMin. If your vMin is at the appropriate speed, and if you’re also at the correct location and angle, and you do this through every corner, this shows up as horizontal line in the vMin table.
Before I explain how to use the vMin table, you should know it’s a coarse measurement, and it’s not perfect. Necessarily so. Later in this article I go into how I created this table, and some decisions I made for ease of use over accuracy. The purpose of the vMin table is to give the everyman a free, private tool they can use to data coach themselves with nothing more than pencil and paper.
Using the vMin table
The vMin table (download PDF) shows the ideal minimum corner speeds at Watkins Glen International. It is divided into three columns, based on how much aero your car has. Circle your vMins in each corner, and they should (ideally) form a horizontal line across one or two rows. Speeds that are above that line (slower) are corners where you can increase your vMin. Speeds that are below (faster), indicate that you are either late to full throttle, or can bring several vMins higher.
In the previous image, there’s a horizontal-ish line formed by the inner and outer loop, and turns 7, 8, 9, and 10. But T6 and especially T11 are slower, and are areas for improvement. T1 is faster than other corners, meaning the driver is either late to full throttle, or all corners could be faster.
And that’s basically it. Draw circles around your vMins, look at the ones above and below the line, and try different strategies so that you end up with a horizontal line.
You probably noticed that there are no vMins for Turns 2-3 (the esses). That’s because low-powered cars can’t reach a high enough speed before these corners. If you’re flat footed before, during, and after the corner, it’s really just a straight, isn’t it? For the same reason, if you’re in a slow car, your vMin data for T10 could be off, because you may not reach the T10 vMin on the straight between T9 and T10.
Those caveats aside, one of the great things about vMin analysis is that it doesn’t matter if it’s raining out. All your vMins will come down by the same amount. For that matter, tires don’t matter. If you’re on all-season tires or racing slicks, your vMins should still form a horizontal line.
The vMin table can also be used as a comparative metric with other cars or drivers. If you and a buddy have similar aero (more on this later), you can compare vMins. You may find one of you has a corner you need to work on, and the other has the answer. No data coaches needed, just share driving notes with friends.
Improving your vMins
You’ve circled your vMins on the table, and it’s time to address the outliers. Most of the time you’ll be working on raising your vMins, and so I’ve included various strategies for that below. I can’t take credit for these, it’s things I’ve learned standing on the shoulders of giants. Certainly much comes from Ross Bentley, and his excellent Masterclass online webinars and Speed Secrets books, emails, and other content. I also compare notes with other coaches, and so if you have a method for raising or refining vMin, please contact me and I’ll update the list.
What I’ve found is the best way to raise vMin is deprogramming. We need to challenge our beliefs, break old habits, and even unlearn things we were taught as novices. These are things that we rely on for speed and safety, and they are deeply ingrained. But they are also holding us back. Deprogramming requires a change of focus and a deliberate (sometimes diametrical) change in our actions. By completely changing what you were concentrating on and performing through habit, and intentionally moving that focus and doing something else, you’ll begin to deprogram yourself.
Try to change your focus and actions in the following areas, and see how it changes your vMin.
Existing focus
Focus on this to raise vMin
Beginning of braking point
End of braking point
Late braking
Brake earlier/lighter/longer
Threshold braking
Release longer, softer
Downshifting before the corner
Hold a taller gear
Driving a late apex line
Take a late apex as early as you can
Always on one pedal
Coast/pause mid-corner
Early acceleration
Maintain momentum
Setting a fast lap
Learning
Driving
Drilling
Note that different corners will require different strategies, and not all of them will work out. People are different, too. You may find that one strategy works for you, while someone else may have a very difficult time with it. For example, I find that the best way to improve my vMins is doing the no-brakes drill, and that may not be practical for most people who are sharing a crowded track with others.
I’ll explain each of the focus areas, so you understand why they can increase your vMin.
BoB vs EoB – Most of us have a solid reference for the beginning of braking point (BoB). At Watkins Glen, it’s usually something very obvious, like the 400 board in T1. I’m not suggesting you stop using that, but add a second reference point for the end of braking (EoB) point. It’s deeper than you think, near the apex. By shifting your focus to EoB, you’ll spend less attention on late braking and more on that crucial zone, where the wild things are.
Brake lighter – For someone who is really good at late braking, it may seem that breaking earlier, lighter, and longer will result in a slower lap time. However, some combination of earlier, lighter, longer will set you up for a higher vMin, and with that, your lap time will come down. You may eventually get back to later, stronger, shorter, but only after you understand the true vMin in that corner, and position your car appropriately.
Release softer – It’s easy and fun to pass people on the brakes. Drop anchor a fraction of a second later, brake hard in a straight line, and turn in at the last possible moment and you’ve made a great pass. But once the pass is complete, it requires a lot of effort to turn the car sharply at a low speed. If you release the brakes softer (earlier and longer), you can add a little bit of steering into the final part of the braking zone, making the car a wee bit less stable. This will allow you to pivot the car using yaw instead of steering angle, which results in earlier and acceleration and less tire scrub.
Hold a taller gear -Because the engine has better acceleration at higher revs, most people downshift before the corner. Instead, try holding a taller gear. By not downshifting, you can use all of you concentration on corner entry. By using a taller gear, you can apply full throttle earlier; it’s like a passive traction control system. If there’s just one exercise you try, I hope it’s this one, you may be surprised by the results. If this doesn’t work, then your car may have large gaps in the gearing, and so try downshifting after the corner. This will still free up your concentration on corner entry, while providing more oomph at corner exit. My twin brother has been trying go under 1:02 sim racing a Miata at Brands Hatch Indy, and was stuck there for years. He changed his shifting to after the corner, and can now consistently click off high 1:01 laps.
Apex earlier – You were taught a late apex line for good reasons, it allows a larger corner radius, earlier acceleration, and it’s safe. Do what you’re doing now, just do it earlier. We call this backing up the corner, and it’s one of the things I look for immediately when data coaching. Whenever I compare two or more drivers, whoever has their vMin further to the left will be the fastest, because they broke earlier, turned the car earlier, and got on the gas earlier. Another reason you should apex earlier is because not all corners require a late apex. But most of us learned the late apex line early in our driving careers, and because it’s so effective, we’ve programmed ourselves to use that strategy in all corners. This is the most common mistake I see as a data coach, but unlearning the school line requires a deliberate change of focus.
Coast – Whether or not they’ve heard “always be on one pedal,” many people are. They transition immediately from hard on the brakes to hard on the gas. The term is pedal mashing; the definition is slow. You’ve probably seen a friction circle, and understand that if you use 50% of the tire’s traction for braking or acceleration, you only have 50% left for cornering. The result of immediately jumping from the brakes to the gas means the car never experiences 100% of the tire’s grip for cornering. However, if you allow the car to settle in the middle of the corner, then it can use all of the the available traction for lateral grip. We aren’t talking a lot of time spent coasting, it could be just a couple heartbeats between the pedals, but it can pay huge dividends in vMin. If you’re a pedal masher, it may take some effort to delay your transition from brakes to throttle, but try coasting for a full second mid corner, watch your vMins go up, and your lap times come down.
Conserve momentum – Drive a Miata. No, seriously! Every car is a momentum car, and if you can’t go fast with 90 hp, 900 hp isn’t going to help you. The only way to go fast in a slow car is to conserve momentum, and the best way to learn that is in a Miata. If you can’t borrow or rent one, drive your car two gears taller than normal. It’s probably still faster than a Miata, but you’ll get the point of the exercise.
Focus on learning – Lap times are an important measurement of self worth, and at some point during a weekend most of us are focused on setting a PB lap. But you only need one, and it’s usually set around 10-11am. So for the afternoon sessions, change your focus to learning, and experiment with things that can raise vMin. You will go slower than normal, but your next morning session will thank you. Also, if you drive a modern car with traction- and stability-control systems, turn off all the nannies, even if it’s raining. Heck, especially if it’s raining. Those things may help you set a fast lap right now, but they are seriously holding you back in the future. In my car, the difference between letting the car’s computer think for me and me thinking for myself is about 1.5 seconds per lap, and that’s on a short 1-mile course.
Drills – It’s difficult to learn how to play a musical instrument without doing scales or other repetitive exercises. And when you want to learn how to play a song, you need to break it down into multiple pieces and do each piece separately for a while. That’s drilling and it’s an essential part of mastering anything. How much drilling do we do at the track? None. Part of that is because drilling isn’t fun, but it’s also both unsafe and rude to do drills when sharing the track with people who are mostly trying to set PBs. The answer is to use your warm up and cool down laps for drills. But if you want real mastery, you should spend more time drilling than driving, and that requires a mostly vacant track. For this I suggest either becoming a member at a private club like Pineview Run or Circuit Florida, or hitting a less popular track mid-week. I won’t go heavy into which drills you should do, but the no-brakes drill, followed by no shifting, is a good one-two punch that will raise your vMin and drop seconds per lap.
Now that I’ve addressed your low vMins, what about those corners that have a vMin that’s faster than others? These are both rarer and easier to solve. Like in golf, most of us are trying to correct a slice (low vMin), but the odd duck has a hook (high vMin).
If you have one vMin that’s too high, you might be a novice that’s under-driving the whole track. That’s OK, this gives us something to work with! The confidence and skill you’re experiencing in that one faster corner can be applied to every other corner, and all your vMins can come up. Ask you instructor to help you with this, you’re about to drop 5 seconds per lap!
If you’re an intermediate or better driver with one vMin that’s too high, then you’re getting on the gas too late. Review your data in this corner and you’ll see you are either late to full throttle, or rolling off after an initial throttle application, and then reapplying throttle at corner exit. On some corners, this is an appropriate strategy, but Watkins Glen doesn’t have a turn like that.
The reason for your high vMin is likely one of two things. Either you started tracking in a low powered car, in which case your natural default is to conserve momentum, or you simply charged in too fast. Good for you, because too much is easier to fix than not enough! Change your focus to braking and turning earlier (backing up the corner), and get to full throttle before the apex. You may eventually move your acceleration point a bit later, and dial in some maintenance throttle, but to break old habits you need to do something extreme, and changing your inputs such that you can apply full throttle before the apex should do it.
Using the vMin table as a shortcut in data coaching
You’ll recall that the purpose of the vMin table was to allow people to keep their data private and coach themselves. What I didn’t expect was that I’d also use this table with every single student I coached at Watkins Glen.
Before I had the vMin table, my normal process was to send a student out for two sessions with the Aim Solo. The first session was mostly to make sure the unit was working and to warm up the driver, but sometimes there are good corners or sectors that are worth saving. The second session is usually the fastest of the day, and I bring them in after that and download the Aim data into Race Studio.
I’d load up their best three or four laps, and look at the the shape of the speed trace. From that I can get just about every input they are doing, and so I don’t need things like steering angle, brake pressure, throttle position, etc. I’ll also look at the time slip on the bottom to look for any large gains due to beneficial mistakes. Next I’ll do a sector time analysis to calculate a realistic theoretical best lap, and expand the histograms to show them how consistent they are. I may then open the friction circle to show the student how they are blending inputs (there’s often a discrepancy in left and right turns), although I could already suss that out myself from looking at the speed trace.
Now that is a pretty long and complicated process, but it’s also very accurate and illuminating. But what I found was that it wasn’t necessary with the average student. Most people have the same things to work on, and it’s easy to grab the low hanging fruit. The largest, lowest, and juiciest fruit is vMin.
These days my process has changed, and after downloading their data into Race Studio, I load up all the laps (not just the fast ones), take a note of the vMins in each corner, and circle them on the vMin table. This allows me to immediately assess the general skill level of the student, and which corners they need to work on.
For example, if I see that T7 and T8 are off by say 2-3 mph, I know the student has some work to do in general. But if those corners are the same, then I can assume a higher level of skill. There are other shortcuts you may discover for yourself, so if you’re a data coach at Watkins Glen, try my vMin table and see if it speeds up your data coaching process as well.
How I created the vMin table
You might be wondering where the data comes from, and how accurate the vMins are. I started by analyzing hundreds of laps at WGI, from rookie drivers to pro racers. I found a lot of 10hz GPS data online, and so it was easy to import a .drk file or convert a CSV from some other system, and then analyze the data in Race Studio.
But when I didn’t find enough pro-level data, I found another way, which was to watch in-car videos on YouTube, and simply jot down the vMins in each corner. It’s definitely a longer and less accurate process, as I needed to watch multiple laps and get the highest vMins in each corner. Data was useless at less than 10 hz, but I found some quality videos online, and it’s worth mentioning the Catalyst videos were excellent. (However, most of the Catalyst users aren’t at a level where their data was super useful.) By combining GPS data with videos I was able to get an expert-level dataset large enough to work with.
After acquiring that mountain of data, I noticed that most advanced drivers go through turns 7 and 8 at the same speed. Drivers of less skill might go through T7 a couple mph faster, or it could be T8, but the best drivers were just about the same speed through both corners.
That got me to thinking of these two corners as a base corner speed, and that every other corner was simply some multiple of the average of T7 and T8. But after some pondering, I realized it wasn’t the average of the corners that mattered, it was the higher of the two. Because the best drivers went through both corners at the same speed, everyone else should be able to as well.
I then looked at my consistent drivers, meaning those who drove T7 and T8 the same-ish speed, and jotted down what their vMin was in every other corner. Then I made that into a percentage of the base corner speed. For example, I found that an expert driver usually goes through T1 at 108% of the base corner speed (the higher of T7 and T8).
Now that I had multipliers on the base speed for every corner on the track, I could make a table for every 1 mph interval of the base corner speed. If you look at the table, you’ll notice turns 7 and 8 are always whole numbers, while every other turn is a decimal value. Now you know why.
At this point I had a workable vMin table, except for one thing: aero. Cars without aerodynamic downforce lift at speed, which means they lose grip the faster they go. Conversely, cars with aero downforce gain grip the faster they go (well, in relation to cars without downforce). Because WGI has so many fast corners and sees everything from IMSA racers to Spec Miatas, my data was not actually correct, since some cars were lifting off the pavement, and others were pushing into it. For an in depth look at the interplay between tires and aero, see my article on How Downforce Affects Tires, which shows the cornering speed of cars with varying levels of downforce, and their speed through Watkins Glen T10.
To get accurate vMin information, I’d need to know the frontal area and the coefficient of drag and lift from every car in my dataset. This is impossible, especially at the pro level where these secrets are guarded. In addition, pro teams may optimize their mechanical grip for just a couple corners, and make the driver earn their pay in the rest of them. So even if I could get specific aero data on the car, I’m not sure I could use those values for every corner equally.
OK, so instead of 100% accurate data, I figured I could estimate the amount of downforce on every car and log it next to the vMin speeds. After doing that for every car in my dataset, I would then be able to create each car in OptimumLap, run a simulation to see what the difference was in speed and lateral grip, and factor that into the final vMin value for each corner.
That’s a lot of work, but it’s the kind of thing I’m good at. I’ve done a lot of aero research, and combined with my wind tunnel testing, I can make an educated guess on a car’s drag and lift values, and land within a few percent.
After assigning values to every car and running test simulations, I now had an aero factor for each corner. To get accurate vMins, I just had to classify the car’s aero on a scale from 0 (none) to 1 (lots), factor that into my formula, and then I’d know the approximate vMin.
And here I ran into my first usability problem. My initial goal was to create a mathematical formula to say, “if the car has this coefficient of lift, it should go this much faster through the corner.” But this would require the user to know the coefficient of lift on their car, and then apply a mathematical formula to find each vMin value in the table. Nobody is going to do that.
So what I ended up with are three columns to represent aero: none, medium and big. It’s still up to the user to determine which column to use, and for that I’ll provide some quick guidance here, but also you should look at my article on Thinking in Aerodynamic Coefficients.
No aero – This represents most cars without aero, but also a lot of cars with factory wings, front lips, side skirts, and body kits. None of that OE stuff does much for downforce. Coefficient of lift ranges from +0.25 to -0.1.
Medium aero – This represents most track cars with wings and splitters or pro-level cars with just airdams and spoilers. Coefficient of lift is all negative (downforce), in the range of -0.3 to -0.8 or so.
Big aero – This represents professional aero done correctly, and some amateur-level time attack aero. It’s rare to see this level of aero at the track. Anyway, coefficient of lift is -1.0 and better.
As I’m writing this, it occurs to me that you could average the vMins between two adjacent tables, and in that way get five values for aero. That level of accuracy shouldn’t be necessary for most tracks (corner speeds are lower), but that’s kinda neat.
One more thing to note about aero (and it relates to the dataset in general) is that if you compare the three aero columns, the speeds are the same in Turn 1. This isn’t a typo or error on my part. T1 is certainly fast enough for aero to make a difference, but for whatever reason, aero cars and non-aero cars (driven by pro and expert drivers) go through T1 at virtually the same speed, respective to the base cornering speed.
This is why it’s not possible to assemble an accurate vMin table by measuring corner radius, camber, lateral grip, and coefficient of lift, and then running that through a number crunching algorithm. The drivers are the real computers here, and so it’s necessary to see what they are doing in each corner, not what is theoretically (mathematically) possible.
To find out who the best drivers are, it was necessary to quantify driver skill. This required estimating tire grip for each car and factoring that in. Essentially I was asking this: given this tire and aero, how fast should this car be able to go through each turn? This wasn’t terribly difficult because I have a lot of tire data, and I know how to tweak the values in OptimumLap to return realistic values. This got me close, and in some cases I had to slightly correct the values based on what the best drivers were doing.
With all of this information on tires and aero, I could assign every driver a cornering score. What you see in the vMin table is the best of the best. Some of these are pro drivers, but there are quite a few drivers that have pro-level vMins, with only a single corner and/or some consistency they need to work on.
vMins at other tracks?
You might be wondering if I’ve created vMin tables for other tracks, and of course I have. The ADHD part of me loves the dope rush of starting a new project, but then the realization of actually having to finish it sets in. This ends up with me starting another vMin analysis at some new track and not finishing that one either.
But I do intend on releasing more vMin tables in the future. The next tracks are likely Lime Rock, NYST, PittRace, Thompson, and Mosport, all of which I started and then backburnered.
It’s a time consuming process to find all the data, cull that into an expert driver dataset, log all the data in a spreadsheet, calculate a base corner speed, and then figure out the multipliers for each corner. But the endgame where I have to factor in the aero is particularly lengthy and difficult to get right. Now you know why I move on to the next before finishing.
There’s also the fact that not every race track has corners that are useful for vMin analysis. I mentioned previously that T2 and T3 at Watkins Glen don’t have vMins, and the climbing esses at ViR would be similarly problematic in a slower car. Also very long corners, like the Octopus at NJMP Thunderbolt, or the T8-9 complex at Brainerd, just don’t lend themselves well to vMin analysis, since vMin is then very location dependent.
If you have a track that you’d like me to create a vMin table for, contact me and perhaps I can be compelled to start a new one. (I mean, I can always be compelled to start a new one; I really need some motivation to finish one.) But know this: the accuracy of a vMin table depends on having pro- and expert-level driving data from that track to create a dataset. Without that, it’s a non-starter.
Find a data coach
I hope you get something out of the vMin table and try some of the exercises on the reverse side. Most of you will see a significant drop in lap time, but you’ll eventually hit another plateau, because where your car is at vMin, and where the car is pointing, and how long it’s at vMin (thanks Ross) are more important than the actual vMin number. To work on those, you’ll need a data coach.
I suggest hiring a data coach at the track. While there are coaches and services that will analyze your data remotely, having a data coach at the track with you is much more effective. You’ll shorten the feedback loop and get results that day.
If you’re well connected and monied, absolutely go straight to Ross Bentley or Peter Krause. For people coming to Watkins Glen, I suggest the following.
My data coaching partner Chris White is still with Niagara PCA, and also data coaches for the WGI-based Trackmasters group. So if you make it to Watkins Glen for those events, book time with him, I believe this service is still free. Also, Chris tells me that the vMin table is now being used in the PCA classroom, and so perhaps the worm has turned?
Gregg Vandivert runs Omega 13 Coaching and is available at several tracks in the northeast. He uses both Aim and Garmin devices at the same time, and combines that with right-seat coaching. I don’t believe you’ll find a better full-service coaching experience. Gregg told me a great story where he got someone six seconds at WGI, and it wasn’t even his student!
Patroon BMW visits WGI at least once per year and has a great crew of data coaches. They are headquartered in the Albany area, and so you’re more likely to see them at New England tracks like Lime Rock, but look them up if you’re in their area.
I’m available for private data coaching at Watkins Glen and other tracks in the Northeast, but I prefer to do this at Pineview Run. I have a lot of comparative data there, the track is usually pretty empty, and the lodge is air-conditioned with big tables where we can sit and review data. Pineview is a highly technical track, and so it can be even more useful if you’ve never been there before, as you can test your assumptions on driving strategy versus a mountain of data from other drivers. I have a sliding scaled based on dates (was I planning on being there already) and whether or not you own a Porsche.
If you’d like to get drill-based instruction and data coaching from me and Ross Bentley at Pineview, sign up for the Ross Bentley Driver Development Program. The 2-day class doesn’t include data, but the 3-day class does, and is offered only on July 25-27. The cost is $4000, but the driver mod will be worth it, many times over.
Occam's Racer 