A Miata and a Porsche 914 are somewhat similar in size and shape, and there’s some interesting studies and CFD data on the 914 that could help fill some knowledge gaps that exist in the Miataverse.
They have similar frontal areas, but the 914 has a better coefficient of drag (.363 vs .38). I suspect this is partly the front end, which covers the tires completely from airflow. The 914 canopy might be better as well, since a vertical rear window is better than one that slopes back at 25-35 degrees, which is the worst possible angle, and about what a Miata hard top has. But in many ways, these cars are not so different, so let’s take a look at the 914 in more detail.
Published drag data
I’ve been digging around and found a few pieces of interesting data on a Porsche 914. In the May 1978 Porsche Panorama magazine, there are drag figures for a Porsche 914 in various configurations.
|Configuration||Cd||Change from stock|
|No roof, windows open||.447||-.84|
|No roof, windows open, headlights up||.464||-1.01|
- One of the things I didn’t test at Watkins Glen was the effect of pop-up headlights (because I removed them from the race car). Miatas have slightly larger headlights, but at least this is a ballpark figure I can use for simulations in the future.
- I also wasn’t able to test the difference between windows open and closed. Again, because race car. However, the windows open drag figure published here is very good compared to a Miata. I think it’s the Miata’s wider rear canopy that’s the culprit. The 914 canopy doesn’t stick out into airflow the way the Miata does, and it’s probably a significant drag factor when the windows are open. The 914 also has turn signals mounted on fender bulges, which redirects airflow along the car. But more on that later.
Scientific experiments on a 914
Chris Cassidy works at UCSD, in the Aerospace and Mechanical Engineering department. He also raced a Porsche 914 in autocross, and sponsored a project where his students used his car to do aerodynamic experiments to reduce drag, which would make his car faster in autocross. Much of this blog focuses on his work, thank you Chris and students!
Chris had several groups of students over a few years, each group doing a slightly different project, from preparing a scale model, to validating the model, and then doing computer analysis. His second group of students did the CFD vs water tank experiments, summarized in the table below.
From the scale model and flow visualizations, they concluded that the CFD model was accurate enough. And from this, they could test a few different configurations, and see how drag affects HP at 50 mph.
Now I’m grateful to Chris for allowing me to use his data, but I think the primary supposition is wrong, which is that drag reduction equals speed in autocross. The students concluded that the standard car with no airdam or spoiler would be the fastest, because it had a 2.1 horsepower advantage due to drag reduction.
However, I’ve done plenty of computer simulations, and I can tell you that at low speed, drag reduction isn’t that important, certainly nowhere near as important as creating downforce. If Chris wanted to go faster, he should have had his students work primarily on downforce. But let’s not skip ahead, let’s check out what they did.
CFD studies on a 914
Chris’s third group of students put 40 different configurations into CFD analysis, totaling over 400 hours of computer simulations. You can check out his website for all the results, but I’ll hit some of the higher points and relate this to a Miata.
First, take a look at the following image. The body styles are listed in order by the amount of HP gained in each configuration, which roughly (but not exactly) corresponds to the least drag. Not surprisingly, the most streamlined version, #15, gains the most HP. Being wing shaped, it also has the most lift. In the downforce column, the lower the number the better (a positive number is lift, negative number is downforce). The last two configurations are missing from this image, but are on the poster. All of this makes me wonder which one would be fastest on a race track….
Before I get to finding out that burning question, I’ll need to figure out the coefficients of drag and lift from Newtons.
Figuring out drag and lift
OptimumLap uses coefficient of drag and lift, and Chris’s students used drag force and downforce in Newtons. How am I going to convert this? I wish I paid more attention in math class, but here’s a formula I can almost get my head around. Solving for coefficient of drag Cd, Fd is drag force, p is the density of fluid, V is velocity, and A is surface area.
Cd = (2*Fd)/p*V^2*A)
For each configuration, the velocity is the same (50 mph), as are the fluid density (air) and frontal area. (17.8 sq ft). So it looks like the Cd is directly based on the drag force. This means I can set up a simple ratio, and get the relative Cd.
Likewise, I need to convert downforce in Newtons to a Cl value. I’m sure there’s a formula for this, but if I call the stock body .30 lift (a reasonable guess), I can figure out the lift of other body styles by a straight ratio using the downforce value.
Let’s see which of these is fastest on an autocross track using a 914 with stock weight and power (94 whp) and tires at 1.1g grip. I’ll also throw in Watkins Glen to see if anything different happens at high speed. Some of the configurations in the big image above make no sense (like the car resting on the ground), and so I’ll use only some of them. I’ve reordered these from mild to wild:
|#33 – Stock||.363||0.30||65.08||150.17|
|#5 – Rear spoiler||.385||-0.37||64.44||148.30|
|#4 – Front spoiler||.413||-0.05||64.77||149.77|
|#29 – Slotted pillar||.320||0.30||65.06||149.34|
|#23 – Big wing||.405||-0.23||64.59||149.20|
|#34 – Racing open||.303||0.53||65.28||150.10|
|#22 – No glass||.429||-.48||64.36||148.86|
|#10 – Slant back||.319||0.73||65.48||151.30|
|#14 – Scoop front, slant back||.299||0.43||65.18||149.57|
|#15 – Full streamliner||.270||1.10||65.82||152.42|
The students theorized that the stock configuration, with the least drag, would be fastest, but it doesn’t come out that way in OptimumLap. Configuration #22, which had the most drag, won on the autocross course. It also placed second at Watkins Glen, but got beat by the spoiler, Configuration #5, which has less drag.
A conclusion you can draw from this is that at low speed, ignore drag and go after as much downforce as possible. At a high-speed venue, go after the best L/D ratio. Most tracks are somewhere in between these two, and in this case, I would maximize downforce as long as it doesn’t require an extra stop in an endurance race.
There are some strange discrepancies in the data, like why is an airdam and a spoiler worse than just a spoiler? And why didn’t the wing perform very well? I don’t have the answers, and this is just CFD, so we have to take all of this with crack-rock sized grain of salt.
What can we apply to a Miata?
Well that was fun to see what happens to a modified 914, but what can we apply to a Miata?
The CFD study here shows that a spoiler is very effective on a 914. I’ve written about Miata spoilers before, and mine works really well. There’s really no downside.
Directing airflow below the window
Although the 914 front end directs air under the car just as a Miata does, it covers the front wheels nicely. But that’s not the part of the front end that’s the most interesting, take a look at the turn signal bulges. They look like a styling feature, but in fact they work to pull air down the side of the car, and away from the windows. These days you see this done with canards, but the 914 does this with a very simple forward bulge for mounting turn signals.
Most of us race with open windows, and keeping air out of the cockpit will reduce both drag and lift. For top-up-windows-down driving, it should also mean less buffeting inside the car. I may have to build some of theses bulge-thingies and yarn test it, this could be a great way to reduce the amount of air going into the open window.
I’ve always wondered if the Miata roof spoilers were anything more than cosmetic.
If you look at configuration #24, they added a small roof extension to the 914. It reduced lift but increased drag, both by a tiny bit. Performance-wise, it was a wash. Maybe I spent too many years in California, but I like the way they look, and I certainly wouldn’t fault anyone for using one.
Miscellaneous drag reduction
In the CFD study there were a number of small things that reduced drag. If you did all of them, it could prove beneficial for endurance racing.
- Removing the 914’s mirrors (configuration #19) reduced drag quite a bit. If I use the RSR calculator to calculate the power used at 50 mph, then the mirrors contribute about .035 to drag, which is a lot. Note that 914 mirrors don’t have the rounded aerodynamic shape that Miata’s have, and so it’s not apples to apples. (For more on mirror aerodynamics, you have to check out this lesson in Flow Illustrator.) A more accurate figure is probably half, or around .018.
- Covering the rear wheels with flush covers (configuration #35) reduced drag by .02. If you use the HP Wizard tool, you get a similar value (.022). This is something I’ve been meaning to do to my brother’s Yaris, because the rear wheels don’t do much except keep the trunk off the ground. We can run skinny wheels with less offset, and get them flush underneath a side cover. However, I don’t know if it’s as practical to do rear wheel covers on a Miata, since the wheels are wider than the bodywork. Rear wheel covers that aren’t flush with the body would be configuration #17, and you can see that did nothing for drag reduction.
- Slotting the pillar (#29) reduced drag by .043, which is a very significant amount. This directs air into the vacant area behind the rear window. We don’t see this much on street cars, I guess because they are ugly. But slots like this or guide vanes could be useful when you have an abrupt back end and turbulent wake. On a Miata, I’d figure out how to do this with air rushing past the window, and duct that internally to fill in the wake behind the hardtop, or behind the bumper.