Diffuser Profile - Convex or Concave?

[hardingfv32]

Yes, you are right about the wing placement. Sadly not perfect, but the best I have found.

But... he did make the diffuser work, why?

What is the clue being revealed by these images? A need for higher levels of low pressure or more velocity at the exit of the diffuser?

Do his statements help solve the mystery?

"shows clearly the breaking at the diffuser and the turbulences at the backlash"
"steadying current setting back down"?

I need help understanding what he is talking about.

Brian

[DRCorsa]

I don't understand the author's comments on the images, but this could be normal as English is not my mother language anyway.
The beam wing, where he positioned it, is clearly working to energize the diffuser's boundary layer. I would say that the high velocity/low pressure of the wing's lower surface helps to "pull" air from the diffuser, helping the flow adjacent to the concave surface to reattach.
While this is (i think) clear enough, i can't see how this concept could work -at this level at least- with the normal placement of the beam wing on an actual F1 car.

[hardingfv32]

So at a minimum we have learned that we need much more flow after the throat of a concave diffuser to maintain flow attachment. Is this correct? Does this agree with general aerodynamic principles as you know them?

Brian

[ringo]

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.

Brian

Well, think this one is not that uncommon:




If you imagine ground effect as mirrored (second half under the ground) picture of the aero shape - which is basically correct, it would be quite similiar.

It's the best shape for pressure recovery efficiency in supersonic flows, but works for lower speeds to.

The basic idea behind is the same in both cases - optimal expansion from higher (at floor's leading edge or rockets combustion chamber) to lower (behind the nozzle/diffuser) pressure.

The difference between both cases is rocket nozzle is driven by high pressure in combustion chamber, diffuser by low pressure behind the car, but physics works almost always in symmetrical way, so no big deal at all.

I'm quite sure there will be separation bubble just behind kink line with such extreme design (air molecules do have a mass and just can't change the direction 90 degree at the sudden), but the flow has a chance to re-attach later.

As far as strakes goes, my very uneducated guess is that beside obvious roles of separation of regions with different pressure, they are slightly cambered, vertical airfoils, and as any lift generating airfoil they produce tip vortices, one being shedded near the roof of the diffuser (which helps to energize boundary layers), and the other on the ground facing side, probably co-rotating and merging with vorticies entering the diffuser. Nothing is free, so they add some drag (both form and induced) to the system.

You are not quite correct about that shape.
I think the discussion is getting carried away. Too much side tracking and confabulations on wings and beam wings. Not sure if any reader would reach a good understanding of the topic in the thread.

The truth of the matter is a diffuser isn't really sensitive to geometry once you pass a certain distance from the throat. There are many shapes that can give the same result. Generally a bell shape with concave shape near the throat is optimum, but bells have many shapes, and many of these can give almost the exact same result. To give some perspective, optimization is really like 1% improvement when we are talking about diffusers.

[hardingfv32]

The truth of the matter is a diffuser isn't really sensitive to geometry once you pass a certain distance from the throat. There are many shapes that can give the same result. Generally a bell shape with concave shape near the throat is optimum, but bells have many shapes, and many of these can give almost the exact same result. To give some perspective, optimization is really like 1% improvement when we are talking about diffusers.

What do you base this statement on? I can not find any studies on concave diffusers. I find it unlikely you have done the necessary simulations.

There is not much height available to do much with the roof (floor) after implementing an aggressive concave floor. It looks like the slope of the concave section at the front of the diffuser floor on the RB7 at Monza is about 30 degrees or more.

So what is the secret to keeping the flow attached to a very aggressive concave roof (floor)?

Brian

[ringo]

It shouldn't be aggressively concave, and i think we are looking at the diffuser incorrectly, ie more aggressive than it really is. It only needs to be concave near the throat with attached flow. The ultimate goal as i said is to have zero shear stress on the wall. Whatever shape achieves that is best for a given height, opening and length.
It can be concave for the first 5%, then convex for 30% and so on for example. It wont affect it much if it's mixed curvature, as long as the flow is attached.
Detached flow shouln't be on it, unless the height to length ratio is such that separation is unavoidable in a very very small region.

believe it or not this can be an optimized diffuser shape:


There are many variations, it's generally bell shaped and depends on the surface roughness, the dimensions of the inlet and outlet etc.
These things are computer calculated, but the whole idea is to have zero shear stress at the wall.
You can manually "feel around" with the cfd if you observe the friction at the surfaces. Again, the ideal is a bell shape for asymmetric diffusers based on research, not my experimeting.
At the end of the day we are talking about 1% is the starting design is fairly good. Nothing to beat up oneself over i guess.

[MIKEY_!]

How about someone run increasingly less aggressive concave diffusers through CFD. Just to see how much less aggressive it needs to be to prevent flow separation.

[ringo]

this is actually gentler than we were proposing in the cfd, and this also has multipe curvature.

[hardingfv32]

Ringo

The RB7 at Monza was 35% concave, 50% straight, and 15% convex.

You have run on about bell shape, surface roughness, dimensions, shear stress at the wall and something about 1%. It has been two posts now. If you actually have any idea what you are talking about, please relate it to the images that have been posted. Just pick one or two of the many variations you say are possible and explain how they apply. Demonstrate that anything you have mentioned is relevant to the discussion.

What are they doing to presumable keep the flow attached to the concave diffuser surface?

Brian

[Tozza Mazza]

Those percentage figures you have Brian seem quite precise.

Where did you get them from?

Can you draw out this profile inside a box 350mm long and 125mm (75mm from step plane) high?

I think Ringo is probably right, it's not one curvature, it's a mix to beat the flow seperation.

Now for some CFD evidence.

[DRCorsa]

Nice shot Brian!

Here's a pic of the Benetton B196 diffuser froms scarbs website. It shows (in the background, yellow arrow) clearly that it's a pure concave shape, similar to that used for the CFD analysis.

[DRCorsa]

After studying Brian's latest picture of the RB7 floor, i came up with this design which follows the F1 rules (125mm height, 350mm length) so the downforce-drag levels are not comparable with the previous analyses.
Very smooth concave at the start and convex shape at the end. The flow is almost perfectly attached, the suction peak at the kink line is unchanged, plus we have a second (lower intensity) suction peak at the transition between concave and convex.
With a bit of design fine-tuning we could come up with a stronger suction peak at the second transition and also with an improvement of the pressure at the rearmost edge of the throat.





I think we are quite close to the colour plot of the Ferrari model, posted by Reca a few pages back.




And here is a pressure plot on a plane closer to the diffuser lateral wall, where the second suction peak is more evident.




And lastly, a velocity streamline on this same plane:

[DumHed]

I imagine that concentrating the suction peak at the kink line and transition keeps the low pressure zone more vertically below the diffuser, rather than "behind" it - which will give less drag penalty.

Imagine a 45° flat diffuser with a constant low pressure under it. The force will be acting 50% down and 50% back (drag).

If the shape can be optimised to keep the low pressure zones as close as possible to surfaces parallel to the ground, most of the force will be downward.

[DRCorsa]

So at a minimum we have learned that we need much more flow after the throat of a concave diffuser to maintain flow attachment. Is this correct? Does this agree with general aerodynamic principles as you know them?

Brian

I think yes.
When a surface is stalled, it means that the flow velocity close to the surface is very low and you need to make something to energize the boundary layer.
A similar concept was used with the F-104 Starfighter aircraft where engine exhaust were blown to the trailing edge flap when its angle extended beyond 15deg. This energized the b.l. making the aircraft capable to land at lower speeds, something that in normal conditions would be impossible due to its tiny wings.

But, as we have seen, in F1 such a system is not needed because they can control the flow on the diffuser by making a blend between convex and concave.
And, as shelly has pointed out, exhaust blown diffusers are probably more about vortex generation/enhancing, than b.l. energizing.

[ringo]

You see, it's a mixture of curves. You guys biting off more than you can chew with beam wings and what not.
Just take it easy and focus on the diffuser alone.

So brian, what were saying about evidence?
The trend on F1T nowadays is to speculate isn't it? Why have a different standard for me?

Any way it's more transparent if i give the tips and let corsa do the tests.

Bell shaped is best, simple. Depends on the situation, it can be optimized from there and give many different bells.