Diffuser / under body questions

[Apex]

Mickey_s hope this helps

http://flowlab.fluent.com/collaboration ... ffuser.gif

[Gecko]

Mikey_s,

while you are absolutely correct that the molecules only know whatever is happening in their very nearest neighbourhood, the theory on this basic level has very little predictive power. As others have pointed out, any disturbance in the flow field is transmitted across the system with the speed of sound, which for the types of flows we are interested in here is effectively instantaneous. This is also reflected in the equations governing the flow; while all the terms in those equations indeed take only local information into account, the pressure term is present in those equations in such a way that it needs to be determined globally at all times for the flow to be self-consistent, simply because the pressure changes are propagated (almost) instantly over the system.

To sum it up, indeed each molecule only feels the pressure differences in its neighbourhood, but the whole pressure distribution in the system is a sort of a self-organising phenomenon that occurs almost instantly.

The whole faster velocity/lower pressure mantra can only be understood after taking this larger picture into account. It is a theorem coming from the stationary inviscid incompressible flow theory that, if you follow a particle along its trajectory, the sum of its local energy density and its pressure is constant. It's a very nontrivial result that emerges only after you take collective phenomena of many molecules into account. In a sense, the fluid has emergent properties that are more than just the sum of its parts (molecule dynamics).

Also, about the diffuser, while the pressure differences in the air flow around the car indeed produce the density variations, those can be neglected for the most part and air can be considered as an incompressible fluid for most automotive purposes. Certainly the density differences are far, far less than the cross section differences between the diffuser and the underbody. To understand the diffuser, one really needs to think of an incompressible flow and the changes in cross section, with the resulting velocity and hence pressure differences. By imagining that there occurs an actual expansion of air, one obtains exactly the opposite (and wrong) results for the pressure distribution, which should (neglecting vortices) show higher pressure in the region of the diffuser and lower at the underbody.

[AeroGT3]

I disagree about compressibility only happening at extremely high speeds, though I will say that undertray flow should be largely incompressible if you're doing things correctly. I don't know how to get it up here, but I can show you a wing at a freestream Mach number of 0.1 that gets compressible. Compressible flow happens way before transonic regimes - more like less than Mach 0.5!

Also, the conservation of mass statement (in aerodynamics it translated into the continuity equation), is NOT mass in equals mass out. At steady state, that is true. But there are transient effects - mass accumulation within a set volume due to changes in density, or changes in time derivates.

A better statement would be "the difference between mass in and mass out is equal to the time rate of change of mass contained within a specified volume."

To the original poster: which school are you at? Does your team typically do Aero? As a first year Aero team, I would recommend against using an airfoil shaped undertray as the analysis will be over your head in my opinion unless you're a truly talented graduate student with access to very fast computers or a rolling ground tunnel.

As far as the tray, you have to stop looking at it in terms of velocities or shapes or flow directions, and look at it purely in terms of pressures. You want low pressure in back of the car, high pressure in front, high above the tray and low underneath. The flow is pressure driven, and like someone already said, the pressure at the outlet of your diffuser will specify your flow rate. The ONLY reason trays work is because the pressure at the diffuser exit is LOWER than that at the inlet. Start at the back of the car and carefully work your way forwards. Data aquisition for measuring these backpressures would be ideal, but unlikely for FSAE or Formula student.

[Mikey_s]

Firstly I (we?) should apologise to FF as this thread has crept a little.

Secondly, I am perfectly happy to accept that I am wrong, with the caveat that I am (as yet) not convinced that we are necessarily in disagreement. Please bear with me and, when you have considered my argument(s) deliver your verdict(s)

Reca, my comment about FF's original design with the venturi at the front and rear of the floor (btw as I write this AEROGT3 has just made some of the points I was about to make!).

I do understand Bernoulli's theory (it's not that complex ), but those assumptions hold for systems where the flow rate remains constant. I have no clue whether this occurs in practice, but my sense in the original plan of FF is that if you have a trumpet at the front of the undertray and move the apparatus through the air the cross sectional area will decrease and in order to maintain the flow rate the pressure would need to decrease. However, in practice I believe that this would not happen; all of my senses tell me that the flow rate would decrease because of boundary friction and a reduction in the cross sectional area. This would, I believe, lead to a pressure increase at the constriction leading to a decreased flow. Your argument holds only where the mass flow is forced to remain constant. In any case this is not how the undertray of an F1 vehicle is constructed - an F1 vehicle has a splitter to separate flow under the car from that which is then directed into the sidepods and around the side of the car.

My second argument refers to the diffuser itself and the equation Reca posted with rho*area*v=constant;

If the regular laws of physics are to be considered as binding (for the purpose of this discussion we should assume that they are) then an increase in cross-sectional area must lead to a decrease in either velocity and/or density. on the basis that density = mass per unit volume and the mass is fixed then the density of the air in the diffuser must decrease. It is also clear that in principle the flow velocity should also decrease (there is an english saying that still waters run deep - so deepening the area leads to a reduction in velocity), but the missing term in this discussion is time - there are no kintetics given and this is important to understand what is happening in respect of driving (no pun intended) the mechanism of action of the diffuser.

I believe it is a gross oversimplification to say that air should be considered incompressible, even at low speed, also that the ideal gas law is irrelevant to open systems. Anyone who has seen a diffuser in real life will have seen the vertical flutes to try and 'close' the system and furthermore, in the good old, bad old days the skirts aimed to do the same thing. Of course the system is not totally closed, but they aim to close it as far as the regulations permit. This obvioulsly will reduce the extent to which the law holds, but directionally it still applies.

Notwithstanding the above, there are some assumptions made (on both sides) that heavily influence the answer obtained in the models. I still maintain that what I have stated is not necessarily at odds with what Reca, kilkoo and gecko have said. What influences the answer is whethe the flow is limited, or constant. However, the main issue is that the diffuser is what drives the flow under the car. I maintain that the flow is resultant from the pressure differences generated by the movement of the vehicle.

Finally, to answer the comment of Reca about the work of the aerodynamicists; If the models you and others use in your cfd computations were perfect there would be no need for a wind tunnel. They are not, and the important thing about a model is knowing where it works and what the range of validity is. In the real world flow is complex and the models, however complex, must be tested in a tunnel to detemine whether they give the right answer - and even then they don't always do that... ask Toyota!! The aero guys working in the wind tunnel perform a highly valuable task - they make sure that models generated by the CFD guys work!

[Gecko]

The point about compressibility being visible at speeds we are considering is of course true, I never wanted to argue against that. The point, however, is that, to understand the functioning of the diffuser, the compressibility does not need to be taken into account. To illustrate the point, the compressibility effects scale with the pressure differences, and these differences are roughly proportional to the square of the car velocity. That way one can always slow the car down to a speed where the compressibility effects are not important anymore (though they may be at speeds we are interested in, I am not trying to argue otherwise!), yet the diffuser will still work (although the force generated by it will naturally decrease, but in a well understood way). In an ideal incompressible fluid (no viscosity) the flow picture should stay the same regardless of speed, just that the flow velocities will scale overall, and the flow can be slowed down to the point where compressibility can be neglected, as long as the boundary layers don't thicken too much with respect to the dimensions of the system.

One can of course argue that compressibility and viscosity will change the behaviour and that they are always present, but this is obfuscation through introducing too many variables, especially since those are not the dominant terms in this application and can be thought of as secondary effects. The main principle behind the action of the diffuser is obtained from a simple ideal incompressible fluid theory which can be shown to be valid to a reasonable degree in such an application. In a real optimized design one should indeed take things like compressibility and viscosity into account, but the basic principle will not change because of those effects.

[Reca]

Mike, just a question.

If I take a long tunnel, then I fill it up with water, then I put in it a real size model of a f1 car with and then I drag it, for example via a long wire from the end of the tunnel, at a speed of say 5 m/s, is the car going to generate downforce ?

[BreezyRacer]

I'm a newbie here and this is my first post.

Where do you draw the line between high pressure and compressed air? Is a high pressure area, such as that created around a vortex, compressed air or just air at a higher pressure than the air around it? Maybe this is just a matter of semantics?

I'm a bit of a newbie at studying aero and my focus has been on diffusers so I've been following this post as best as I can. A couple of Q's ..

1. Wouldn't some inlets at the front help to guarantee a supply of air to make the diffuser work at ultra low ride heights? I say this because it's well known that if the front of the car gets sealed off (as in bottoming under braking) there is only air from the sides of the car to make the diffuser work and most downforce is lost. Of course current F1 cars don't have this problem as they have stepped bottoms regulated into the design.

2. What is the effect of velocity on aero design in general? Does the near optimal design for 100kpm look the same/similar as the design for 200kpm? All the test data I've been able to see happens at the static velocity, and is often not even mentioned in the CFD's ,etc that are published. I would think that airflow at even 50kpm is a great force and should be worthy of harnessing. Just stick your hand out the window .. Also I never see/hear of new trays for Monaco to try to maximize DF at lower speeds, so that makes me wonder about the role of velocity on design.


3. Most books I've read indicate that the roof of a diffuser as a flat panel between 5-15%. It seems though, in real life many of the roof designs are curved (some concave at the entrance and convex at the exit). How do you determine a diffuser angle for that type of floor? Just curious ... I guess very curious.

It seems to me that an F1 car is a partularly bad example to use for an ideal diffuser design, though a captivating study because it shows all the ways to get around space and rule limitations. Certainly not that I understand it all (I did get the part about the center strakes producing DF through creation of votexes though so I'm somewhat understanding)

I've been trying to study diffuser designs from 1994 (the first year without active suspension, and planks and diffuser limitations were added only after Imola and the loss of our greatest ever driver). So far the Williams FW16 is all I've been able to get my hands on and that diffuser design was not much more than a smoothly radiused blend with some pretty straight strakes. It didn't look to be aligned with a lower wing element either.

[Mikey_s]

Reca,

I would guess that the answer is yes, although what the relative contribution between wings and underbody is - I have no clue. What is your point?

I'm not sure that water is a particularly good example in any case, it is 9in common with other liquids) quite incompressible, and has a tendency to cavitate in areas of low pressure generating highly non-linear results.

[Reca]

Actually water is the perfect example exactly because it’s used for a quantity of tests to simulate behaviour of cars, airplanes, wings in ground effect etc. etc etc. at low speed and gives the same results the real case, in air, would give. (granted that, obviously, you respect the requirements of fluid dynamics similitude, but that’s true also for any wind tunnel simulation).
Water tunnels (closed loop) and tow tanks (basically the system I described in the previous post, although the towing mechanism is usually different) are quite effective tools, for some things, as for example visualizations with dye, they are even superior to the wind tunnel.
The fact that water is incompressible is not a problem, exactly because in that range of speed also air is, even if it’s something you apparently will never accept.
On the contrary that air is de-facto incompressible at low speed creates problems with simulations because to simulate in wind tunnel, the wind speed necessary is the real case speed divided by model size. So to simulate a real case speed of say 250 km/h in a wind tunnel, with, for example, 40-50% model (that is feasible for cars but is definitively too big, hence infeasible for airplanes), you need speed in excess of 500 km/h, and, also neglecting the power required to generate that flow, the results would be useless because the test would be affected by compressibility effects that in the real case are negligible.
Thanks to the fact that kinematic viscosity of water is at 20°C about 15 times that of air (and the ratio increases relatively fast with temp), water allows to simulate with small models and low speeds real cases that with the same scale models on air would be either impossible or very difficult, and very very costly, to simulate since would require to modify temp and/or pressure of a huge mass flow.

[Mikey_s]

Reca,

I, like most others, visit this site to learn. It is clear that you have a great understanding of CFD, and I respect your expertise. BUT, I do have a scientific background and, whilst I am demonstrably not a CFD, or aero guy, I still have some difficulty in accepting that you and I are on different pages. Please don't get frustrated with another scientist that takes a different (and, at least in my view, not opposed) view of the world to yourself.

Your own equation (rho*area*v=constant) has three variables in it. We agree that in the diffuser the cross sectional area is changing (only two more left), but you seem to suggest that the only other variable is velocity of the flow, whereas I take a somewhat different view. My view is that the flow increases for a reason, that reason is a pressure gradient. The flow speeds up because it must (in other words it is resultant), can you be certain that the density of the airstream in the diffuser is unchanged? I have no data, just a feeling, based on a number of pressure experiments that I have personally carried out in the (now distant) past.

I would not say that I refuse to accept the incompressibilty of air - but practical experience tells me that it does happen - anyone who has looked out of an airplane window at take-off, or landing on a humid morning will have seen the vapour cloud forming/disappearing instantaneously above the wing resulting from aerosol formation in the low pressure zone above the wing - it does not persist and therefore arises from the passage over the wing. The phase change happens for a reason - lower pressure, lower air density (hell for all I know the temperature might also be changing, pV=nRT anyone? - LOL). Squeezing my water bottle also demonstrates how compressible air is relative to a liquid.

I think our main area of difference Reca is mainly on the issue of what is driving the process - and I think that we are nearly in violent agreement (you may disagree on this point ) - the issue is which is derivative and which is driving the process. The nett result is clear, downforce is generated by the undertray, the flow in the undertray is driven by the diffuser. Ultimately the downforce from the undertray can only result from a difference in pressure above the car relative to below the car. The energy to drive the process comes from the motion of the car.

Just to tweak your tail one time Reca; if your CFD models are really the answer, then to use your own phrase, do you consider that you believe that thousands people working 24/7 on wind tunnels all over the world are just wasting time? (should the cars just be designed on the computer and raced without testing?) no need to answer that

I hope there are no hard feelings Reca, that is not intended and not desired - the only solution to science it to keep asking questions.

[AeroGT3]

Guys, at any kind of speeds a Formula SAE car will see, the density WILL be constant except on EXTREME high lift surfaces. The effects of air compressibility will be incredibly small. You're looking at the ideal gas law and saying well, if pressure changes density will too! You're totally forgetting that there's a big, huge constant in front of one side that isn't present on the other, and that Temperature variable is being totally ignored. Believe me, Formula SAE trays are incompressible.

[AeroGT3]

I have no clue whether this occurs in practice, but my sense in the original plan of FF is that if you have a trumpet at the front of the undertray and move the apparatus through the air the cross sectional area will decrease and in order to maintain the flow rate the pressure would need to decrease. However, in practice I believe that this would not happen; all of my senses tell me that the flow rate would decrease because of boundary friction and a reduction in the cross sectional area. This would, I believe, lead to a pressure increase at the constriction leading to a decreased flow. Your argument holds only where the mass flow is forced to remain constant.

There is no "nice" way to say this, so I'll just say it. You are 100%, totally, completely, and undoubtedly wrong.

A reduction in area decreases pressure. Boundary friction are small. Go do some calculations on boundary layer thicknesses -they are very thin.

The flow rate is NOT controlled by area, and you can decrease it all you want. What controls flowrate is the pressuire at the exit and inlet of the tray AND THAT'S IT. The only exceptions to that are when you reduce the area so much that the flow becomes sonic (speed of sound), or when you reduce it so it's so thin the boundary layer is the same thickness as your rideheight. This never happens with cars, but if it did, it would be the only scenario where the flow rate is constricted by area.

[Reca]

The flow speeds up because it must (in other words it is resultant), can you be certain that the density of the airstream in the diffuser is unchanged? I have no data, just a feeling, based on a number of pressure experiments that I have personally carried out in the (now distant) past.

I am sure that a car model with a diffuser to generate downforce works even if the fluid is water. Isn’t that enough to demonstrate that you can have change of pressure without change of density ? If that wasn’t the case, in an incompressible medium like water, the underfloor (or even a wing) wouldn’t work while not only it works, but, if fluid dynamic similitude requirements are respected, it gives the same qualitative and quantitative results it would give with air.

I can also be sure that if you take schlieren or shadowgraph images (visualizations methods based on deviation of light caused exactly by change of density) of a given airfoil at different velocities, you are going to see an uniform bright field until the airfoil reaches at minimum a freestream Mach between 0.3-0.4, at that point you’ll start to see small areas of compressibility in the high velocity regions, but still we are talking about only about small and limited regions, in the remaining part of the field the density is constant. As speed increases these regions of compressibility will become larger and larger.

I would not say that I refuse to accept the incompressibilty of air - but practical experience tells me that it does happen - anyone who has looked out of an airplane window at take-off, or landing on a humid morning will have seen the vapour cloud forming/disappearing instantaneously above the wing resulting from aerosol formation in the low pressure zone above the wing - it does not persist and therefore arises from the passage over the wing. The phase change happens for a reason - lower pressure, lower air density (hell for all I know the temperature might also be changing, pV=nRT anyone? - LOL). Squeezing my water bottle also demonstrates how compressible air is relative to a liquid.

Where pressure drops, due to acceleration of air at the leading edge (and particularly in takeoff/landing due to the higher AoA hence larger speed increment), the tiny droplets of water present in the humid air aggregate and become big enough to cause refraction of light, that is what you see. Then velocity drops and pressure consequently increases again, and in just few % of chord after the leading edge the big droplets are separated in smaller and again invisible ones.
Similarly the wingtip vortices that are visible in an humid day (also on F1 cars) are due to the accumulation of the water droplets caused by increment of vorticity. (since water is denser than air the droplets follow a different trajectory, that’s also the reason why visualizations, particularly of vortices, are way more effective in water tunnels, dye density is basically same as water so the colour follows the same trajectory of the fluid, while in air isn’t the case.)

The squeezed bottle then is a bad example. First, because while doing it you applying a pressure lot higher than the change of pressure caused by an airfoil. In fact even with very high peaks of velocity on the nose, we are talking about a variation of pressure in the order of few % of atmospheric one. While squeezing the bottle you are applying a pressure that is easily a few times the atm pressure.
Second because squeezing a closed bottle (I assume you are talking about a closed one, right, otherwise the example would be even more pointless) you are working with a closed system, excluding the minimal leakage thru the cap the air has no way to go and consequently has to compress, no choice. In open atmosphere situation is totally different, air molecules have plenty of space to go.

BTW, the only reason I asked you if the guys working in wind tunnels are wasting time is because you said that a common misconception is to assume that air is moving over objects while on the contrary are the objects that move against stationary air.
If that leads to a difference in the distribution of pressure on the underfloor (and your description of pressure on the underfloor is what led us here), then not only the guys on wind tunnels would waste time, but I did waste a lot of time too, because contrarily to your assumption that I’m a CFD guy living in a mathematical world unrelated with reality, I had my good share of “dirt” work too with oils, dyes, scanivalve, HWA, PIV etc etc...

[Mikey_s]

I wave a white flag and surrender... and apologise to FF once again for having hijacked his post... at least I learned a lot from this thread.

Reca and AeroGT3, thanks for your patience in educating me.

To try and salvage an answer to FF's question at the begining of the thread though; is the best shape for his undertray then a trumpet entry and a trumpet exit with a flat underfloor? (still having conceptual difficulty with that one on the basis that the front end ought to work in the opposite sense to the back end, thereby cancelling the effect out )... or is it to have splitter?

[flynfrog]

reca you mentnion oil do you have a recipe for areo oil just disel or somehting else

the cfd is running now on a few of those curves and im working on the radatior ducting as we seak all seems to be coming together