How to improve aerodynamics with a virtual wind tunnel

Tom
By Tom Suddard
Oct 23, 2024 | Aerodynamics, CFD | Posted in Features | From the June 2022 issue | Never miss an article

Image Courtesy Morlind Engineering

[GRM+ members read this article first. Subscribe and gain access to more exclusive content for only $3/month.]

Wind tunnel? What wind tunnel?

We successfully scanned our LS-swapped 350Z using chalk dust and an old camera so the wizards at Morlind Engineering could meticulously model its aero properties, eventually producing a perfectly accurate, fluid-tight CAD model that was ready for CFD–industry-speak for computational fluid dynamics. 

The next step? It’s time to put our virtual car into Morlind’s virtual wind tunnel.

Loading the Virtual Trailer

Prepping your race car for a visit to a normal wind tunnel–or a test day designed to measure aerodynamic effectiveness–requires a lot of work. Why? Well, besides preparing your car to physically be there, you have to prepare every single aerodynamic device to be tested as well. 

Why not just build the aero you know will be effective? Because, sadly, it doesn’t quite work like that. 

The field of aerodynamics is complicated, and 2+2 doesn’t always equal 4 the way you’d expect. Think of aerodynamic testing like developing a new drug: Sure, there are best practices and other drugs you can use as a starting point, but you’ll never know if your wonder pill cures cancer without first testing it in a clinical trial.

So while there’s a strong baseline of knowledge to use as a starting point, you actually have to build every assumption in reality in order to test them. Sure, it’s common knowledge that splitters make race cars faster–but how much splitter will balance the rear wing? 

And when doing that math, did you take into account the additional airflow over the car that a bigger splitter will create? How will that airflow affect the wing? And how does that affect how much splitter you’ll need? Aerodynamics is full of compounding variables that make assuming anything perilous at best and frivolous at worse. 

So let’s build some parts to test. Because we’re using CFD, building parts actually stops after the first step: Draw it in a computer. And because there’s no ticking clock on a virtual wind tunnel, we were able to run the 350Z without any aero modifications, then pause to study the results before we got to work designing parts. 

Making That Virtual Baseline Run

What exactly does a run look like? Honestly, it looks like a big space heater. Morlind loads the model into its software, clicks a few buttons, and then code starts scrolling across the screen. 

For the next 12 hours or so, a big stack of servers turns electricity into heat and data, doing the math until the run is complete. If you’ve ever seen a crypto mining farm, you know exactly what a virtual wind tunnel looks like. 

After the calculations are complete, Morlind does some post processing on yet another powerful computer, then finally presents the data: numbers showing downforce and drag, along with visualizations showing how and where the air flows around the car. 

No aero aids

Before we could test the various splitters and wings, we needed a baseline for our Nissan 350Z. The CFD modeling showed what our butt felt: The car got light at speed due to lift.

What did the baseline data say about our 350Z? It said, well, exactly what we expected: Our car made lift, totaling 188 pounds at 150 mph, and 62.1% of it was over the rear axle. That confirmed what we’d noticed on track: The rear of our car seems to get light at speed. Overall, though, Morlind didn’t see anything too troubling in the data. Our 350Z behaved more or less like a fastback coupe should. 

Baseline data in hand, it was time for the fun part: adding downforce. 

Building Virtual Parts

Now Morlind Engineering could start designing alternate CAD models. The goal was to test as many combinations as possible, then pick the one that produced the most downforce with the least drag. 

Morlind used three different approaches when designing test parts for our 350Z: virtually “installing” parts from its manufacturing partner, Nine Lives Racing; repurposing existing parts, like splitter ramps; and designing new parts from scratch, including a set of bespoke digital hood vents for our digital car.

During this design process, Morlind continually referred to the CFD data. The shape of the splitter and dive planes, for example, was influenced by slicing cross-sections of the car to see exactly how the air flowed around the model. 

Another cool thing about CFD is that the annoying details of real life–like how to run splitter support cables or whether you need to trim a hood brace to fit those vents–don’t get in the way of testing parts. Mounting a splitter was as simple as drawing an appropriate outline and dragging it over to the car with a mouse, while installing a Nine Lives Racing rear wing was a copy-and-paste job.

Once Morlind had worked out a pile of virtual parts for our virtual car, it was time to put everything back into the virtual wind tunnel. That produced what we’ll affectionately call our aerodynamic cookbook, which summarizes each combination’s performance and compares them against each other. Every metric was calculated at a speed of 150 mph, and for those keeping track at home, each trip to the server farm costs roughly $300 in computing time. 

The Basic Recipe: Wing and Splitter

Adding a wing and a splitter to a race car is a well-worn path to speed, and that’s where Morlind suggested we start. Adding a Nine Lives Racing Big Wang kit was our first huge improvement, turning 188 pounds of net lift into 267 pounds of net downforce at a zero-degree angle of attack–aka AOA–and 323 pounds at a 5-degree AOA. 

One problem: This downforce was all on the rear of the car, overcorrecting our rear lift to the point that our wing was using the car as a giant lever, badly lifting the front at speed.

What about drag? Adding parts did increase the amount of horsepower we’d be wasting pushing air molecules around. 

In stock form, our 350Z made 452 pounds of drag, which correlates to a lift-to-drag ratio (a measure of how efficiently it’s making downforce) of 0.42. That’s a great L/D for fuel mileage but bad for racing. We want a negative number, which means the car is making downforce in exchange for those pounds of drag. 

The Big Wang set to an AOA of 5 degrees increased drag to 547 pounds, a 21% bump from stock. More importantly, that wing changed the L/D ratio to -0.59, a huge improvement. 

If you’ve ever wondered why wings are preferable to spoilers, this is why: They’re an extremely efficient way to make downforce. And the news only got better. Pairing a splitter with the wing actually reduced drag, down to 531 pounds, and made far more downforce. Our L/D ratio dropped to -1.19.


With wing (0° AOA)


With wing (5° AOA)

  
With wing (5° AOA) and splitter

Now we could start adding aero equipment–first a wing and then also a splitter. The CFD modeling showed that the wing added downforce without creating too much turbulence, while the splitter helped balance the nose of the car with additional downforce. We’re making progress. 

Huh, so that’s why race cars have splitters. Adding one to the 350Z immediately brought things back under control, netting 630 pounds of downforce, 32.5% of it on the front wheels. 

Not bad, but not quite what we were after. We wanted our downforce distribution to match the car’s static weight distribution, with the balance remaining unchanged as speeds increase. 

From this simple baseline, Morlind started iterating, with the goal of increasing front downforce and investigating ways to reduce drag along the way. This work followed two distinct paths in 12-hour chunks as each CFD run was completed. Here’s what we learned from the results.

Path No. 1: Hood Vents

Our 350Z has a sealed engine bay, meaning underhood air pressures are extremely high. It was obvious early on that getting that air out from under the hood would be beneficial for cooling and downforce, even if it might come at the expense of additional drag. Air flowing through heat exchangers and engine bay stuff produces more drag than air flowing against smooth body panels, but more air through the radiator isn’t something most racers would complain about. 

So Morlind built some virtual hood vents, added them to the car and ran the CFD. The results, pardon the pun, sucked. Adding vents to our 350Z reduced drag by 10 pounds, but also reduced front downforce by 41 pounds. The vents were indeed evacuating air from under the hood, but reducing our splitter’s effectiveness in the process. Every aerodynamic device does indeed affect every other device, and we proved it with disappointing data. 


With wing (5° AOA), splitter, hood vents


With wing (5° AOA), splitter, hood vents, dive planes


With wing (5° AOA), splitter, dive planes


With wing (5° AOA), splitter, dive planes, ramps

The CFD modeling allowed us to test different setups without getting our hands dirty. For example, the hood vents lessened drag but also reduced the splitter’s effectiveness. Adding dive planes to the front corners of the body, however, restored that lost downforce. Tacking on some splitter ramps added downforce while increasing drag.

But that rule works both ways. The vents had increased cooling and reduced drag, which are both good things in a race car. So Morlind added dive planes to each corner of the 350Z’s front bumper, then ran the CFD again. Success! Drag remained low–just 9 pounds above our wing and splitter baseline at 540 pounds–but the car was now making 673 pounds of downforce, with 36.8% of that on the front of the car and a L/D ratio of -1.25. 

Morlind tested the car with dive planes but no hood vents and confirmed the hypothesis: Each device on its own wasn’t helpful, but together they produced a fantastic result. Morlind also tested various combinations of splitter ramps and splitter end plates, but none beat the teaming of dive planes and vents. 

Path No. 2: No Hood Vents

Morlind had developed an effective aerodynamic package with hood vents, but what about running without them? Could there be more front downforce on the table?

Starting with the assumption of a sealed hood, Morlind went to work. Adding splitter ramps to our wing and splitter baseline actually increased drag and reduced downforce slightly, pushing the L/D ratio down to -1.06. And adding dive planes to the car without ramps produced a similarly disappointing result. 

Pairing them together, though, was a promising option. With splitter ramps and dive planes, drag increased to 553 pounds, but downforce increased right along with it. The 350Z netted 658 pounds of downforce and a -1.19 L/D ratio with this combination. Even better: 37.5% of that downforce was on the front wheels, making this our best-balanced combination yet. 

Everything at Once?

Morlind had found two good paths to making downforce, but what about combining them? The 350Z went back to the virtual tunnel one more time, digitally festooned with a wing, splitter, ramps, dive planes and hood vents. 


With wing (5° AOA), splitter, hood vents, dive planes, ramps

What about running everything at once? The data showed that we weren’t quite there yet, but at least we now had some paths to follow. 

The result: 541 pounds of drag, 632 pounds of downforce, and a L/D ratio of 1.17. Sure, front downforce was great at 38.8% of the total, but this result wasn’t what we hoped for. The combination was less than the sum of its parts, meaning we needed to choose a path before proceeding.

Join Free Join our community to easily find more Aerodynamics and CFD articles.
Comments
Colin Wood
Colin Wood Associate Editor
5/4/22 2:30 p.m.

Sure, it has a price tag, but I think it's pretty wild that this sort of service is available to pretty much anyone.

I'm not looking to improve the aero setup on any car, but it's good to know that I don't have to reserve time in a physical wind tunnel if I change my mind.

Johnny_at_NineLives
Johnny_at_NineLives GRM+ Memberand Reader
6/30/22 3:40 p.m.

In reply to Colin Wood :

best part it's free. we do the test and build the parts. the parts are a normal price. 

fidelity101
fidelity101 UberDork
10/11/22 9:28 a.m.

the OEMs print small scale models out of SLA (process) and wind tunnel those, it scales well to production size vehicles with simple math rather than print a whole 1:1 car. 

corsepervita
corsepervita HalfDork
5/1/23 12:13 a.m.

I've always wanted to do CFD testing but the software you can get for it is either free and insanely complex to even run, or very incredibly expensive and basically out of reach of most people who are doing this for hobby.

j_tso
j_tso Dork
5/1/23 12:37 a.m.

DavyZ
DavyZ New Reader
9/11/23 2:36 p.m.

It's also "free" for those of us who look at what others have successfully done and then copy their modifications as closely as possible.  Sometimes "monkey see, monkey do" does work beneficially.  

Seriously, this article was very informative and shows what can be done without actually doing anything at all, so to speak.  (no wind tunnel)

 

Datsun310Guy
Datsun310Guy MegaDork
9/11/23 2:50 p.m.

Hey, I'm in charge of the Bob Seger references.  

footinmouth
footinmouth GRM+ Memberand Reader
9/13/23 7:23 a.m.

That testing would cost more than my vehicle is worth 

 

TiviamTonal
TiviamTonal New Reader
1/9/24 12:15 a.m.

Something that was surprising to me was I recently watched a video about drag coefficients that made it seem like it is not a super exact science. The choice of wind tunnels can apparently have a fairly significant effect on the results from what that video said. We aren't talking totally different, but enough that differences like in this article of .30 vs .32 kind of just happen. Then again with what they say in the article maybe part of that is just that different vehicles of the same model can vary a fair bit based on tiny differences in alignment of panels and things.

stuart in mn
stuart in mn MegaDork
1/9/24 7:23 a.m.
fidelity101 said:

the OEMs print small scale models out of SLA (process) and wind tunnel those, it scales well to production size vehicles with simple math rather than print a whole 1:1 car. 

Many years ago I worked for a company that designs and builds wind tunnels.  Testing is mainly done on scale models simply due to economics, since building a wind tunnel large enough to hold a full size car (or airplane or whatever) would cost a lot more money.  NASA has a couple tunnels at their Ames facility in California that are large enough for full size testing; those are the only ones I can think of offhand, but there may be others.

You'll need to log in to post.

Our Preferred Partners
wLKVKySUQFVrao0su5Urbozi6dSD17mLpsSLjSsaMot0EMvtN3OFzjRIw6VA2w51