Drag Coefficient - Is it really constant

[pascal28]

In the equation of drag the drag coefficient (CD) is assumed not to vary and hence indicates the aerodynamic efficiency of an objects shape.

The drag is then a product of this efficiency multiplied by the frontal area, 0.5 density of fluid traveling through and the velocity squared i.e.

Drag = CD * A* 0.5 * rho * V^2

Manufacturers quote a CD, which as far as I am aware can be derived from from testing in a wind tunnel (& possibly from CFD).

Questions:
1) Is drag coefficient really a constant with speed?
2) If it isn't what would be the typical variation with normal road speeds (say 30 to 70 mph)?
3) Is there an agreed speed that manufacturers quote CD at?

I would like to know this to correctly calibrate a CFD model of the Honda Insight that has a quoted CD of 0.25.

[IamIan]

Originally Posted by pascal28

In the equation of drag the drag coefficient (CD) is assumed not to vary and hence indicates the aerodynamic efficiency of an objects shape.

In the generalized / simplified version ... yes ... not in underlining science it is based on ... ie Fluid Dynamics , and then bellow that the science fluid dynamics' generalizations are based on allows for even more potential variation.

Originally Posted by pascal28

Questions:
1) Is drag coefficient really a constant with speed?
2) If it isn't what would be the typical variation with normal road speeds (say 30 to 70 mph)?
3) Is there an agreed speed that manufacturers quote CD at?

I would like to know this to correctly calibrate a CFD model of the Honda Insight that has a quoted CD of 0.25.

1> No... see above.

2> the amount it varies depends on more than the speed ... if you want generalized / 'typical' , just use the Cd... if you are interested enough to explore a degree in fluid dynamics, enjoy.... kind of like F=MA or V=IR are not 100% accurate, but are a fairly good generalizations for most people.

3> No... it is either computer estimated from computer models of the shape ... or they put it in a wind tunnel measure the force at some air speed and run the calculations backwards... no global governmental or scientific standards are imposed on the methods used... it's on the 'honor system' ... if you don't like it, your welcome to pay for your own wind tunnel tests ... although over 99% of people can't afford to do that, so they just use the published number.

[NeilBlanchard]

If you are referring to Falcon, I don't think that you *can* calibrate it.

[MetroMPG]

I recall reading a comment posted by an aerodynamics expert working for an automaker. I'll see if I can find it.

It boiled down to: while Cd does vary with airspeed, the variation is small enough in the range of speeds a road vehicle typically drives at that it's not worth worrying about. (I seem to remember -- but will check for it -- that he was talking in terms of thousandths - eg. 0.00x).

[aerohead]

*Aerodynamic coefficients of drag depend upon the type of boundary layer/Reynolds number.
*For road vehicles,due to ambient atmospheric turbulence,the boundary layer will transition from laminar,to turbulent at around 20-mph.
*From 20-mph,to about 250-mph,compressibility effects are non-existent and the drag coefficient is constant and stable within this velocity regime.
*So,for 'normal' driving velocities,automotive coefficients of aerodynamic drag can be considered constant.
*During SAE coastdown testing,the initial and terminal velocities are held within this turbulent boundary layer spectrum.
*Model wind tunnels have turbulence screens to introduce turbulence into the airstream,as would be found in the 'real world.'
*Full-scale wind tunnels will generate normal turbulence as a consequence of viscous shearing of the infinite strata of air layers as they move away from tunnel boundaries.

[MetroMPG]

For further reading, an excerpt of a discussion found here:

Drag CRISIS! The Coefficient of Drag is NOT a Constant! « Realworldnumbers

(Phil - are you the "Phil" in the discussion on that page too?)

[Re:] the variability of Cd – we all know that it does: it just doesn’t generally vary that much in the range of vehicle speeds that we are interested in. I’d like to add in a few comments:

1. [...] When we test cars (I work with Saloons and SUVs mostly) we see very little variation between 50 km/h and 100 km/h (our usual test speed. At this speed Re will be in the range 106 to 107). Variations are of the order of 0.005, or less, in Cd.

2. At practical road speeds we have very little (perhaps even no) laminar boundary layer -so boundary layer transition is of little relevance. Many (depending on the vehicle) of our separation lines are geometrically fixed; also gaps, seals, finishers etc act to ‘trip’ the boundary layer. Aside from maybe a small region around the bonnet (hood) leading edge radius we don’t really see laminar oundary layers (you do see some re-laminarisation on door mirror casings sometimes). The upshot of this is that we see relatively little change in flow topology in the speed range of interest. Hence Cd doesn’t vary much, as I’ve said.


3. Cd should fall slightly with speed as, for cars, the total drag force that is being non-dimensionalised by area and dynamic head comprises pressure drag, skin friction drag and a small amount of (vortex) induced drag (perhaps). Whilst the pressure drag force scales with the square of velocity the skin friction drag does not: it increases with velocity raised to a power of less than unity. Thus it is being ‘over corrected’, for want of a better term, by the velocity squared term in the denominator of the drag equation. Hence, although the force increases with velocity, its contribution to the total coefficient diminishes. However this variation in Cd will be small as skin friction drag is worth something around 5% of the total for contemporary cars; so a variation in 5% of the total will be small!


4. On the road the car will be subject to relatively large levels of atmospheric turbulence, plus turbulence generated by vehicle wakes and road-side objects. This will also ensure that the flow is post-critical and Cd will not tend to vary much during typical high-speed or constant-speed cruising.


[...]

[aerohead]

Darin,that's not me.
I'm deferring to Hucho with respect to a constant Cd,and velocity at which it would occur.Any fluctuation would seem statistically insignificant if back-to-back testing were conducted with the same vehicle at equal velocity.
Sure,we'd want to track any meteorological differences which could affect results,but for lay people,lacking thousands of dollars per hour in a world-class full-scale tunnel,we might allow the luxury of presuming a constant Cd for the subject vehicle.
I'd go nuts otherwise!
PS,thanks for sharing the link,it's a good discussion.

[Varn]

We went to the Nissan test track near Casa Grande Arizona a number of years ago.
They had the 5.5 mile track set up with the straight aways set on a small slope. Uphill on one side, down hill on the other. The track manager said it was to measure drag.