Diffuser Profile - Convex or Concave?

[marekk]

Brian

As for the question about engine diffuser, there is no contradiction. Along the diffuser, except for fins effects and voritces, we see a pressure increase due to the increasing cross section. In a f1 diffuser downforce is produced by air running along a curved path across the kinkline and running fast along the flat floor

I have less comfort with this answer. I have to rethink in the context you have provided. Maybe more question coming.

Brian

Pressure gain in engine inlet diffuser is quite obvious - even at full 18,000rpm, the flow through the 2.4l engine is about 0.4 m^3, and for 200mm diameter inlet at 50m/s, if not for spilling, inlet volume would be 1,57 m^3. As shelly said, with correct geometry incoming air is slowed down due to increasing cross section and compress to some extend. The trick described in the linked study is to maximise the effect.

I like to think about F1 diffuser as a pump driven by low pressure area behind the car - for me it's easier to understand and visualize pressure profiles this way. Of course it's just the reference frame positioning, but works for me.

Thanks for great post @reca.

[Pierce89]

I've often used the same description as Marrek to help people understand and vizualize the underfloor flows.

[Pierce89]

With a concave profile the aim is to obtain a more abrupt deceleration in the very first part, near the throat/leading edge, getting closer to the maximum ideal pressure recovery, coherently with the energy of the boundary layer, which is in fact achieved when the b.l. itself is constantly kept at limit of separation.
The same theory is valid not only for diffusers, but for the deceleration area of airfoils too (obviously, the aim is the same...), google for Liebeck airfoils to see a practical application of the concept.

The problem is that, exactly because the maximum pressure recovery demands pushing the b.l. to limit of separation, it's a potentially more critical design that operates optimally in a rather limited range of conditions and dramatically loses efficacy at the minimal variation, and that makes often preferable a more conservative convex profile, that induces a more limited peak deceleration the b.l. is more easily able to cope with, sacrificing pressure recovery in favor of versatility.

That presumably explains the difference we see in F1 diffusers, with the central, narrow, part typically convex, while lateral channels are concave.

In the middle the floor gets quite close to the ground, so the area at the throat is small and changes greatly with small variations of ride height, and geometry is not particularly favorable being narrow and open at the sides, things making more difficult to control the flow thus increasing risk of separation; a smoother transition with a gradual change of slope, imposing a smoother deceleration and keeping the b.l. more distant from limit of separation, is preferable in order to get a more stable behavior for all the working conditions (obviously any trick to improve the situation, like blowing thru the starter hole, energizing the flow, can allow to be more "aggressive" and move a bit towards the ideal)

For the lateral channels, more distant from ground at the throat, thus less sensitive to ride height variations, that problem is lot less evident and they can afford using a more effective profile, concave, closer to the ideal shape for maximum pressure recovery.

BTW, concave profile is nothing new, as you can see in this WT model of Ferrari's F310B (1997) floor, that was on display at Galleria Ferrari in 2005 showing pressure distribution, with distinctly in evidence the peak of pressure localized at the kink:

Sorry to qoute such huge post but I just want evryone to see it. This is quite possibly the best written most informative post in the history of F1techinal.
Truly Excellent Job, Reca!! Well Done!

[DRCorsa]

Reca, thanks for the wonderful info!
Do you recognise a "blend" of conxave and convex diffuser here? I think that this one is concave at the front and convex at its end? BTW, it's from last year'a HRT.

[hardingfv32]

While I appreciate the possible reasons for the shapes of the inner and outer diffuser roof shapes, something seems odd about the F310B diffuser. There is no pressure reduction at the center section. Is it being implied this is the best they could do without suffering separation?

Could the shape be dictated by the transaxle or the rear crash structure mounting requirements?

Brian

[Reca]

Well, first of all a big thank you to everybody showing appreciation for previous post, you are being way too generous...

Reca, thanks for the wonderful info!
Do you recognise a "blend" of conxave and convex diffuser here? I think that this one is concave at the front and convex at its end? BTW, it's from last year'a HRT.

Yeah, looks like that, albeit with images from behind it's always difficult to tell the curvature. If that's indeed the case it could be that with the maximum slope they managed to achieve, without separation, in the first part, they couldn't exploit the whole vertical height of the diffuser allowed by rules; getting closer to the trailing edge then presumably the b.l. stabilized a bit, helped by increasing influence of pressure field behind the car, so they could afford demanding a further bit of deceleration/pressure recovery using the remaining height.

While I appreciate the possible reasons for the shapes of the inner and outer diffuser roof shapes, something seems odd about the F310B diffuser. There is no pressure reduction at the center section. Is it being implied this is the best they could do without suffering separation?

Could the shape be dictated by the transaxle or the rear crash structure mounting requirements?

Brian

On the F310B certainly the integration with mechanical parts in rear end played a role on defining the geometry, that image though was meant to show the concave profile isn't unusual, but was adopted already years ago, and also help visualize the localized peak of pressure at kink, not specifically to show the difference between central part and sides; there were quite different rules back then for bodywork at the rear and the interaction of various areas was quite a bit different (also worth mentioning in that car exhausts were blowing just above diffuser's edge, adjacent to the central channel, but remains to be seen if it was tested in the WT).

With current cars that's not an issue, a more aggressive curvature, especially at the kink, of the central part wouldn't need much room, if they could afford finding the volume for an entire additional deck, certainly they could find that little bit for a different curvature; packaging limitation doesn't look to be, nowadays, what suggests not to use a more aggressive shape for slope of central part, distance from ground and other details of the geometry most likely are.

[BreezyRacer]

This is a very interesting thread but please consider this in your CFD testing .. drawing any conclusions about diffuser design by running a diffuser box in isolation, without the rest of the car and tire layouts is beyond a waste .. it's very misleading!

You are wasting your time because what you are "learning" is wrong. This is largely because of tire turbulence, airflow and airspeeds above the diffuser box, splitter/leading edge profiling, and other outside influences such as wings and wing proximation to the diffuser exit.

Sorry ..

[hardingfv32]

This is a very interesting thread but please consider this in your CFD testing .. drawing any conclusions about diffuser design by running a diffuser box in isolation, without the rest of the car and tire layouts is beyond a waste .. it's very misleading!

You are wasting your time because what you are "learning" is wrong. This is largely because of tire turbulence, airflow and airspeeds above the diffuser box, splitter/leading edge profiling, and other outside influences such as wings and wing proximation to the diffuser exit.

Sorry ..

This is never going to happen on this forum. I have studies of very simple diffusers testing expansion ratios, roof angle, ground clearance, etc. and they take 4 days to process. This is using a research accepted, VERY simple bluff body (block) shape for the model.

Bring the rest of car and tires in to the test adds variables that we have little knowledge of. Generally adding more unknowns does not help solve a problem.

The question asked was very narrow, convex or concave. This can be answered if we can fine some vary narrow research on venturi or throat shapes. We need to solve for the individual parts of the diffuser puzzle before assembling the complete puzzle. Unfortunately, most people on this forum find this boring and want to jump to the "big picture" immediately.

Brian

[shelly]

if cfd says so, it must be correct! (ok, I'm joking)

Takingg cfd too literally is always a risk. I think that till now in this thread we are managing to go along better then in previous threads, like "diffuser function".

Examples,and posts like reca's are good material.
I think theta we have already got some good points that in other threads have been refused even if they are correct, like the concept of kink line.

I think it could be a good idea to open a separate thread about non linear downforce and vortices.

@hollus: I think that an accelerated vortex gives more downfoce for 3 reasons:
-it is narrower but it's longer (for continuity the volume is the same, so the projected area is equal or bigger)
-for helmholtz it rotates more quickly, so the inside pressure is lower
-being the inside pressure lower, even if the gradient is aggressive, the area around the vortex where the pressure is low will be slightly bigger.

[hardingfv32]

Any suggestions for descriptors to use for a search on the convex vs concave subject? Something without the word venturi. Various adjectives with the word venturi got me nothing.

Brian

[Tozza Mazza]

I think the point of the thread is to find the best diffuser profile, see if it's convex and concave, neither, or a mix of both, and give an answer to why. We seem to have found reasons to support different shapes, and have had some nice evidence given, with some very informative posts.

All we lack is clear cut evidence using a whole car as a model now!

If we could get some CFD run (I know we can't take CFD results literally), to support our findings, then the thread may develop into new areas, including how things such as the exhaust, and tyre wake, affect diffuser performance, and in what way.

Would be cool if that happened!

[hardingfv32]

I think the point of the thread is to find the best diffuser profile, see if it's convex and concave, neither, or a mix of both, and give an answer to why. We seem to have found reasons to support different shapes, and have had some nice evidence given, with some very informative posts.

We have not answered which is best for a diffuser. We can't make such a determination without understanding the the function of a concave venturi or nozzle exit wall (cone).

I can find no other mechanism or system that uses such a shape. At a minimum this wall design is not a simple commonly used solution in industry. Why it has value in a F1 diffuser is still undetermined.

Is this another case of moving on without solving for one small piece of the F1 aero puzzle?

Brian

[DRCorsa]

This is a very interesting thread but please consider this in your CFD testing .. drawing any conclusions about diffuser design by running a diffuser box in isolation, without the rest of the car and tire layouts is beyond a waste .. it's very misleading!

You are wasting your time because what you are "learning" is wrong. This is largely because of tire turbulence, airflow and airspeeds above the diffuser box, splitter/leading edge profiling, and other outside influences such as wings and wing proximation to the diffuser exit.

Sorry ..

Nobody here claimed that a simple CFD analysis of a diffuser in isolation will provide the full picture of the aerodynamic behaviour of an F1 car.
I will agree with Brian. To solve a large problem you should make a research on a narrow field first.
I know that a diffuser of an F1 car is not an isolated part and its performance will be affected by numerous other interconnecting factors. I designed the first convex diffuser to actually build it and test it on a car much different from an F1 car, where there is no flow on the upper surface, no rear wings etc. I don't have the time (and expertise i am afraid) to design a whole F1 car to show the people how it works in real life.
We are not here to present absolute facts simply because we can only speculate given that we are not working for the F1 teams. We are here (i suppose) to make contributions towards a given question and everyone contributed to the thread so far.

[hardingfv32]

DRCorsa wrote;

"I designed the first convex diffuser to actually build it and test it on a car much different from an F1 car, where there is no flow on the upper surface, no rear wings etc."

Why would you consider concave? What in nature (industry) provided you with this thought/idea?

I need some clue to further my search.

Brian

[BreezyRacer]

I wasn't trying to poo poo what has gone on here but I went thru similar exercise a few years ago .. designing a diffuser in isolation. It worked very well by itself after maybe 10 revisions. When I attached it to a body model it all feel apart. Many revisions later I got back where I was, or close, but many of the changes I had to make related to the body, etc. I learned a lot about lots of critical interactions, as I listed in my previous post. When i saw some of the flows of diffuser boxes in isolation I just wanted to mention my experience.

I have to clarify that I did NOT try the concave design. My understanding is/was that it's beneficial when you can a strong exit interaction with wings, and of course when the rules require a very short throat, as in F1. But I have to say that I haven't tested it.

You cannot test all these concepts .. there are just too many possibilities. Strakes in diffuser boxes are a particular interest of mine and there's not much hard data on them, though a well written paper was released a year or two ago.