Diffuser Profile - Convex or Concave?

[DRCorsa]

As DRCorsa summarized and demonstrated:

"For the concave, downforce and drag levels are lower, there is the suction peak at the kink line and there is notably less downforce and stall after the kink line."

So why would you use what on the surface looks like an inferior roof (floor) shape?

I would be "restricted" to use it if the regulations "forced" me to do it.
I mean, if the regulations call for a very high roof angle when using a convex shape (small length and large height of throat), i would use the concave if this proved to be "less bad" in terms of stalling.
The convex throat in my analysis is really smooth because i didn't have to follow any rules. If the convex shape would have been much steeper, maybe thing would be reversed.
And if i could use the exhaust to reenergize the boundary layer and limit the negative effects of this stall then i would be happy at the end of the day.
Just have a look at how much lap times has been improved the recent years. In many circuits, the best lap time is similar to those during 2004 with those 3-litre super-powerful engines. And i really believe that the largest stake of improvement has been made due to aerodynamic developments rather that other factors (tyres, suspension etc).

[ringo]

DR corsa, sorry to bother you, but could you do a straight diffuser?
This will add perspective.

[hardingfv32]

DRCorsa

If the bubble you show is generated by the simulation, then I can safely say that none of the separation bubbles I saw in the studies I found on straight floors were even remotely as bad as your concave simulation.

Obviously the rules are the cause this design in a VERY broad sense, but in a very narrow point of view, the rules would not seem to be the issue relative to getting a good diffuser shape that functions well as just a diffuser.

We MUST keep the exhaust out of this, as that would blow us out of the water with more unknowns. Again, the concave shape has been used with the exhaust located in many locations.

This is a good challenge...

Brian

[DRCorsa]

We are not discussing how "large" the "bubble" is but whether it's always there or not. Do you have any study of concave diffuser without a bubble at all?
Mr. Newey maybe has one..

[hardingfv32]

I yet to see a concave shape ANYWHERE in nature or industry in the context we are speaking. That is a bold statement I hope someone will take the time to disprove.

I did find a paper that could change the size of the bubble on a straight floor with BIG changes in Reynolds numbers. I did not think that applied to our situation. Am I wrong?

Here is one of the better papers I found, but still not on point with out discussion. It shows the use of an Ahmed vehicle model which is very common in most of the studies I found.

CFD Study on the Diffuser of a Formula 3 Racecar

http://escholarship.usyd.edu.au/journal ... e/193/1255

Brian

[DRCorsa]

Thanks for the paper, pal!
I will look into it.

[ringo]

DR corsa, sorry to bother you, but could you do a straight diffuser?
This will add perspective.

corsa while you are at it. Show us the static pressure for the convex and concave designs.
Also the shear stress at the diffuser walls. I.e. if you have included surface roughness.

[slimjim8201]

A shameless self-reposting from the thread "Diffuser Confusion"

I've run a simple but informative study to examine the effects of both diffuser angle and type (straight, convex, concave). I've included data for three different angles (5, 10, and 15 degrees) and for each of the three configurations. There is also a baseline case in which there is no diffuser.

Study Details:

- Generic 2D body
- Velocity = 25 m/s

Enjoy...

[ringo]

Your results are contradicting Rcorsa's.

I can see where the underfloor pressure is lower on average for the 10 degree concave design than it's competitors.

Why do you think your results are different?

My opinion is that the diffuser needs to have the same contraints as what is in the F1 technical regulations. As this will determine which shape is optimum. The optimum shifts based on the constraints.
The goal of the diffuser is to have maximum pressure recovery which means the shear stress along the walls needs to be zero or a close to zero as possible.

Chances are the optimum is concave up to a certain distance from the throat, then linear onward. That must be determined from the restrictions and boundary conditions.
But the goal is always to reduce the shear stress at the diffuser walls.

[Mystery Steve]

slimjim8201 and DRCorsa,

Out of curiousity, do either of your models include a velocity vector boundary condition to simulate the ground moving relative to the floor? At the very least, that will definitely effect the way the boundary layer develops, and it may even change the way the flow stays attached depending on how turbulence and vorticity are modeled. While a typical boundary layer on a bluff body start out laminar and progresses to turbulent, the boundary layer between a moving and a stationary object can develop into a different flow pattern called a Couette flow where the boundary layers of the floor and the ground interact:




As for the separation bubble in the concave diffuser simulated by DRCorsa, I'd be curious to see how the velocity streamlines change if the lower beam wing is added to the model. Not to trouble you too much, but would it be difficult to add that to the simulation? Maybe just a high-camber NACA wing, and make a rough ballpark guess of the location based on pictures of current cars?

Another thing to consider is that air will be pulled under the floor from along the side pods, and if it is managed effectively, it can "energize" the flow into the diffuser and reduce separation.

I don't mean to completely dismiss your results thus far, but as others have mentioned, the diffuser flow is influenced by the other aerodynamic elements around it. That being said, the approach of starting with a dead simple model and methodically increasing the complexity will yield a greater understanding at the end of the day, as opposed to just starting with a model of a hypothetical car.

[hardingfv32]

1) I might have a study with the bluff body tested with a rolling-road. I did not take much note of that at the time I read it.

2) What do you expect the lower beam to add to the picture? It might be easier to just adjust the flow over the top of the diffuser to simulate what you think the influence of the beam wing is going to be.

3) Exactly what is "energize" in this situation? Increased velocity or more vortex activity?

Brian

[Mystery Steve]

Depending on the proximity of the beam wing to the diffuser and its overall size and shape, the pressure field around the wing, and the upwash it produces, can greatly affect the overall downforce produced by the underbody. Joseph Katz briefly discusses this in his book, Race Car Aerodynamics: Designing for Speed, and he has also published an SAE paper on the subject, but I don't personally have a copy of the full article.

Obviously, there are several parameters related to size, shape, and placement of this wing that can affect the overall flow. This is why I suggested trying an established, relatively high camber wing and approximating its location based on pictures of current cars. At the very least, we can interpret the results by comparing and contrasting the results obtained with and without the beam wing. (And if I may go on another slight tangent for a moment, it would be interesting to model the downforce and drag of the beam wing in isolation, to see if the overall downforce is "greater than the sum of the parts" like Katz describes in the abstract of his SAE paper. This would be more out of curiosity than practicality, though.) I'm not sure how you might "adjust the flow" over the top of the diffuser to simulate this effect, at least in a way that would make the simulation process easier.

Also, by "energize" I was referring to the vortex activity which can be used to help keep the flow attached in the diffuser. On that note, have these simulations been 3-dimensional or 2-dimensional models? If you aren't already, it might be prudent to run a simpler 2-dimensional simulation initially to investigate the elements of the problem that don't rely on 3-dimensional flow, namely vorticity. However, three dimensions is still better than nothing. I don't want to ask too much, of course.

[hardingfv32]

Mystery Steve

I have thank you for hanging in. I toke the Katz paper and went looking for a free reference. Sometimes you luck out, but not is this case. But, as I am searching I check out the other results to see if there is anything of interest and we lucked out.

This paper shows CFD of a concave diffuser with and without the rear wing assemblies. The wings seem to solve the separation problem.

The name doesn't sound on point, but they seldom do:

Investigation of Turbulence Created by Formula One™ Cars with the Aid of Numerical Fluid
Dynamics and Optimization of Overtaking Potential

http://www.f1-forecast.com/pdf/F1-Forecast-Tech-11.pdf

Brian

[shelly]

In 3d the same floor shape will suffer less from separation.

However, as an aside issue, what is nice is that che convex floor example form drcorsa, which is not stalled, seems to show a nice couple of footplate vortices, which are the cause for the blue streaks on the footplate which are the lower pressure regions of that setup.

@drcorsa: could you please post a picture of pressure on a x-plane across the diffuser? We should see two blue almost circular patches under the footplate: the tip vortices which, acting on a downward facing surface, genrate downforce.

In a real car, those are the vortices the ebd interacts with.

[DRCorsa]

@shelly: