Test Drive
Citroën C4 (B5)
An instrumented readout computed from this car's published specifications.Figures are for the 2.0 16v HDi.
0–100 km/h
modelled
published
Model is 13% higher
Top speed
modelled
published
Model is 4% higher
Braking 100–0
modelled
Tyre-limited ideal. Ignores fade and assumes peak grip throughout. A floor, not a test result.
Power to weight
at kerb weight
Traction-limited to 33 km/h: grip, not power, sets the launch.
Acceleration
Speed against time, integrated forward: traction-limited off the line, then power-limited.
Power required against speed
Where the curve meets the flat line is the top speed. Drag rises with the cube of speed, which is why the last 20 km/h cost more than the first 100.
Dashed line: 90 kW at the wheels, after drivetrain losses. Horizontal axis in km/h.
Holding a steady speed
- Power at the wheels
- 19.3 kW
- Spent on drag
- 72%
- Steady-state use
- 5.3 L/100 km
The published figure of 5.4 L/100 km is a drive cycle, with stops, acceleration and hills. This is a constant speed on level ground, so the two are not comparable and neither is wrong.
Below about 75 km/h rolling resistance dominates; above it, drag does.
Aerodynamic drag
Cd 0.28 × 2.19 m² = CdA 0.614 m². Frontal area estimated.
Newtons against km/h.
Estimated as 0.85 × width × height = 0.85 × 1.769 m × 1.458 m ≈ 2.19 m². Not a measured figure.
Every figure on this page is modelled, not measured. The formulae and every default constant are published on the methodology page, including where the model is weakest: it knows nothing about gearing, so a car that runs out of gears before it runs out of power will model faster than it really is.
A lattice-Boltzmann solver, running on your GPU. It starts only when you ask it to.
Published Cd
0.28
The manufacturer's figure for the real car. This stays authoritative.
This solver, on a fitted shape
—
Measured on a representative shape, not on Citroën C4 (B5). Not comparable with the figure on the left.
What you are looking at, stated plainly. A real lattice-Boltzmann solver running on your GPU. The smoke is carried by the velocity field it computes, not a decorative particle effect, and turning the car re-rasterises the body the fluid sees. But the body is a representative shape stretched onto Citroën C4 (B5)'s published dimensions, not a scan of a car. At — mm per cell it resolves large-scale separation and the shape of the wake; it does not resolve the near-wall boundary layer, the gap between tyre and arch, or absolute drag to engineering tolerance. Treat the number above as a property of the shape at this scale, not of a car.