CO2 Dragster Wind Tunnel
Carve a body, watch the air, and find out what would actually make it faster. Nothing is uploaded — it all runs in your browser.
Everything on this page — the smoke tunnel, the wheels, the race — belongs to the car selected here. Build two and the race below runs them against each other.
This shows WHERE the air separates from your body — at a slower, thicker-fluid scale than the real race. It is not measuring your car. This simulation runs at a Reynolds number of about —. Your real car on the track is around 340,000 — roughly a thousand times higher. No simulation that fits in a browser tab closes that gap, so the drag numbers come from the panel below, not from this picture.
Race the two cars
A simulated 20 m run: the cartridge fires, and each car carries its own shape, wheels and weight down the track. Everything below is a prediction from the numbers — it is not a measurement of a real car, and it is not the smoke picture above.
Race it
A simulated run down a 20 m track — cartridge fired at the line, your carved body, your wheels. This is a prediction from the numbers, not a measurement of a real car.
What would actually save you time?
Each row re-runs the race with one thing changed on your car.
- Mass10 g lighter−125 ms
- Build quality0.1 N less bearing friction−119 ms
- Wheelsspoked wheels instead of solid−33 ms
- Aerodynamics10% less drag area−6 ms
This is the finding worth arguing with. On a 20 m run, mass and how freely the wheels spin move the clock far more than the shape of the body does — which is why a wind tunnel alone would point you at the wrong end of the car.
Where your drag comes from
Drag area (Cd·A) = 8.0 × 10⁻⁴ m², including 15% for parts interfering with each other.
- Body1428 mm² frontal · Cd 0.23
- Wheels (×4)30×5 front, 30×5 rear · Cd 0.55
- Tether guides (×2)screw eyes · Cd 1
How these three are worked out
Nose and tail drag come from the profile's heights, and skin drag from a rectangular section wrapped round the car. Upload a model and both are measured off the real surface instead.
Where the energy goes
A charged 8 g cartridge holds about 1700 J. Almost all of it leaves through the nozzle as noise, cold gas and turbulence before it ever pushes the car.
- Out of the nozzle 1688 J
- Into the car 12 J
Of the 12 J that does reach the car:
- Speed at the line 6.9 J
- Bearings & tether 3.6 J
- Air resistance 1.0 J
- Escaping gas 0.18 J
Mass
- Carved body
- 107 g
- Wheels (×4)
- 12.8 g
- Axles, eyes, glue
- 6 g
- Cartridge (full)
- 33 g
- On the start line
- 158 g
- Wheels' spin
- +8.0 g
Newton's second law actually sees 166 g.
Where that weight comes from
Body weight is estimated from the profile, which treats each slice as a rectangle — so a rounded shape comes out heavier here than it would be in balsa.
Compare designs, don't trust absolutes. The published spread on dragster drag is about ±30%, and the two available measurements of an 8 g cartridge disagree by 2× on peak thrust. The thrust curve here is one published fit; measuring a real cartridge with a force sensor would put this model on firmer ground than anything in the literature.
How this works, and what it can't tell you
The picture is a lattice-Boltzmann simulation running live in your browser. It is honest about flow structure — where the air separates, how big the wake is — but it runs at a Reynolds number around 500, and your real car is around 340,000. Closing that gap would need roughly 9,000 cells across the body: about 29 GB of simulation. So the picture never reports a force.
The numbers come from a component drag buildup (body + four wheels + two tether guides, plus 15% for interference) feeding a step-by-step integration of Newton's second law over 20 m, with the cartridge's thrust curve, its gas depleting as it fires, the wheels' rotational inertia, and bearing friction. Calibrated against NACA Report 614 for body shape and against published whole-car measurements of 4.3–7.1 × 10⁻⁴ m² of drag area.
What it can't tell you: an absolute race time you should trust. Published dragster drag has a ±30% spread; one study that measured its own model in a real wind tunnel missed its own CFD by 24%. And the thrust curve rests on a single published fit — the only measured curve anyone has published sits on a web page that can no longer be reached. Compare designs against each other, not against the clock.
The most useful thing a class could do with this: measure a real cartridge with a force sensor, and race the cars you actually built. A desk fan, some incense smoke and a phone at 240 fps will also show you more about your own car than any simulation.