Importance of the diffuser?

[xpensive]

When I opened this thread, I would like to take the opportunity to re-iterate a draft calculation I made based on standard Bernoulli calculations with different air-speeds on the two sides of a plane:

Resulting vertical force will be: Area * Density * (Speed2^2 - Speed1^2)/2.

With a realistic scenario, let's say a car with an underbody area of 2 m^2,
is moving at 144 km/h(40 m/s) through a corner with an air-speed underneath of 70 m/s, resulting downforce is 4000N.

When a 7% higher speed of 75m/s would increase downforce with 20% to 4800 N.
With virtually no drag-penalty to carry for the coming straight, unbeatable.

No wonder the teams are bickering over the interpretation of diffuser-rules.

[speedsense]

Speedsense,
I, and I think many others here, would disagree with you. I don't like referring to the diffuser itself as a downforce making device, but the underbody as a system sure is a downforce making device. The diffuser is just the most visible part of the underbody so it gets a fair amount of attention.

Whether you call it anti-lift or downforce is just semantics. It's the same thing. Any force in the downward direction, even if it only partially cancels a net upwards force, is downforce. Underbodys can generate significant amount of net downforce both with and without wings. However a diffuser/underbody designed to work with a rear wing should not be the same as a diffuser/underbody designed to work with no wings. Just removing a wing from an underbody setup built to work with a wing isn't proving much of anything.

My point is, with and without wings the underbody is a giant source of net downforce. With a winged car the wings create yet more downforce individually, further help drive the diffusers effectiveness, and most importantly tune aero balance.

The word "downforce" came about in creation of the term, when wings were used on cars instead of the using the term "negative lift" to define the force. It became convention to use downforce as the term. Automotive windtunnel engineers (ones that I've met) use the term downforce only in description of a device that can operate in both conditions-negative lift and lift. Otherwise they term as having "down ward forces" when there are high pressure areas present on both sides of the device. Or in otherwords a device that operates by removing/lowering the percentage of high pressure below it so that the above "higher" high pressure areas can act on it.

Seems to be the same thing, but fundamentally it isn't, especially if you reference a wing with the same properties. Yes, it's only semantics or a type of "short hand" in speech. But the devices operate in totally different manners and can't be explained both in theory or math in the same manner. Labeling both a wing and a diffuser, as a producer of downforce, isn't exactly correct either as a wing does provide production and a diffuser aids in the production of it.
So in a correct use, downforce=negative lift and is the exact opposite of lift and the device should be capable of producing both.
I too thought there wasn't a difference, until I worked in a few windtunnels with some very experienced Automotive Aero engineers (with race cars), that pointed out the differences between downforce, down-ward forces and the defining of an anti-lift device. Though they share similar properties, their use is far different in the theories/math that are applied to them and the enviroment that makes them work.
One interesting point to this, is when it came to testing full blown ground effect tunnels (Formula Atlantic) blurring the above statements further though once it is seen what happens directly under the floor of a car with GE tunnels (the tunnels are open to it) and what happens when you close the floor (inboard) to the ground. The loss of force production is tremendous, vs open, here too, a tunnel is operating with antilift properties, though it can still produce some amounts of downforce when sealed off from the underbody, though it's effectiveness is greatly reduced. Tunnels are more antilift than downforce producers.
When a race car is "approached" with this line of thinking, the use of these devices and gaining higher effective use from them becomes possible and as is applying the correct adjustments.
All I know, is this line of thinking has won a lot of races for me and made understanding what the pressure data under a car means and how to change it for the better.
Until someone tells me it's wrong and why, I ain't fixing it, until it's broken
IMHO.

[PlatinumZealot]

Yup, That is how I saw it too. I did not do aerodynamics but I learnt of Diffusers that are used in water pipes and turbine exhausts. And they don't make downforce in those applications.

The purpose of is to keep a smooth pressure transition from one point to another.

[RideRate]

Otherwise they term as having "down ward forces" when there are high pressure areas present on both sides of the device. Or in otherwords a device that operates by removing/lowering the percentage of high pressure below it so that the above "higher" high pressure areas can act on it.

This is exactly how both a wing AND underbody work, by manipulation of high and low pressure areas. They may not work by the exact same methods, but they attempt to create the same results by manipulating flow and thus pressures. I have no qualms with how you want to think about it. That is well and good. I guess my point is flinging around the term downforce to describe both what wings and underbodies do is in no way implying they operate via the exact same methods or that they require the exact same type of analysis.

When I design an underbody I set out to achieve some amount of downforce. I analyze that underbody with the math, mindset, and tools that govern underbody behavior. In the end I want the extra grip that comes with the increase in normal load. Whether I call in anti-lift or downforce has no correlation on the method I used to determine its effectiveness.

[xpensive]

I think our new friends "speedsense" and "riderate" are taking us for a good ride in the garden of semantics!

Have you read the above postings kilcoo?

But in either case, having a little Bernoulli for breakfast has never hurt anyone I think.

[kilcoo316]

Tunnels are more antilift than downforce producers.

I'm finding that statement very interesting.

If an isolated ground effect tunnel does not generate much downforce in itself, and instead reduces the losses from conventional front/rear wings...

Why was the Lotus 78 so effective?

Its not like the front or rear wings were mighty aggressive... or the rear wing was even in the wake of the tunnel to influence it (although you could make an argument for the wingtip vortices).

[speedsense]

Tunnels are more antilift than downforce producers.

I'm finding that statement very interesting.

If an isolated ground effect tunnel does not generate much downforce in itself, and instead reduces the losses from conventional front/rear wings...

Why was the Lotus 78 so effective?

Its not like the front or rear wings were mighty aggressive... or the rear wing was even in the wake of the tunnel to influence it (although you could make an argument for the wingtip vortices).

I can talk about this test freely, as my client is no longer involved in racing and the head engineer is retired (I was a junior engineer at this time). Though I cannot give you actual downforce/ride change numbers (I don't have them).
This test primarily was geared at understanding tunnels and how to increase efficency in them and we did things outside of the rules for reasons of discovery on a full size car, with an non moving ground plane, though the wheels were spun. 150 mph wind.

We sealed the tunnels on the outside to the ground (ALA the Lotus,though not moveable as it wasn't a moving ground plane), turned out this was the largest increase in the downforce package. and we added strakes inside the tunnels, seperating the tunnel into channels. First into two channels, then three. Adding four, five...created no improvement, though one and two added measurable gains.

Lastly in this portion of the test, we removed the strakes and added an inboard sealing plane to the ground, sealing the tunnel from the chassis floor, rendering the tunnel on it's own merit. This changed the ride numbers by a large amount (lessening the downward forces on the whole car by just over half) and increased the amount of high pressure areas (literally 10 times the number of HPressure pockets) under the chassis floor.

We then shortened the inboard "sealer", half way from ground level though still below the floor. This only provided a small change and increase to efficency. The high pressure areas (under chassis) were still numerous, though were lessened in pressure readings, though still high pressure; it was only when this "sealer" was put at chassis level, that "most" of the numbers returned. Removal of the inboard sealer, allowing the tunnels full access to the underbody netted most of the high pressure under the chassis disappearing and the downward forces at full strength.

Our discovery was that the dispelling of the high pressure areas (coming from other parts of the car and entering under the car), rendered the tunnel most effective. The removal of the high pressure pockets that exist undercar gave a highly effective downforce package and reduction in drag.
One other interesting point, is that the tunnels had very little pitch senstivity when sealed from both sides of it, and the senstivity returned on removal/shortening of the inboard sealer.

Removal of the outboard sealer, greatly dropped all the numbers and literally doubled all the pitch sensivity numbers. The center of pressure movement inside the tunnel became critical to force distribution and appeared mostly effected by pitch, significantly less in roll only. Seemed that when pitch became involved at any point the CP numbers greatly effected efficency in long movement rather than lateral. It was summarized/theorized that the bottom of the chassis rake movements "allowed" higher numbers of high pressure areas to accumlate, and only the areas where the tunnels were closest to the ground, were they being removed rather than added.

Lastly, the removal of the front wings or flattening of them, greatly reduced the number of high pressure area accumulation (wake and unintended vortices), in turn rendering the tunnel more effective in removal and reducing front pitch sensitivity and 50% reduction in the front wing drag numbers (wing removal) Though front wheel turbulance intervention increased. Though over all the downforce package suffered when the wings were removed instead of flattened, also by rather large numbers.

It not so much what happens inside the tunnel that is so high in importance. It's how the surrounding areas (front and sides of the tunnels) are effected in giving strength to the aero package. Basically removal or scavenging of high pressure areas created by other parts/compartments of the car that the gains are made in efficency and the downforce is created, not the device itself creating it.

Isn't is anti-lift production and not a product of downforce, but rather allowing the downforce producing elements (namely wings) or the higher pressure upper body panels parts (down ward forces) to have further strength. no?

[kilcoo316]

Another thought.



We are dealing with pretty much incompressible subsonics here... so raising the dynamic pressure will reduce the actual pressure acting on a tangential surface. Its a simple linear relationship.


Now... since the floor is that close to the ground, no significant vertical velocities can exist between floor and ground plane - so a pressure head need not be considered. Therefore the problem reduces to effectively a 1-D (longitudinal) issue (or 2-D if you consider tranverse flow)...

Thus, the only significant sources of high pressure areas can be low dynamic pressure areas... or areas where there is flow stagnation... which requires a significant deceleration on the ambient flow... on a flat floor, I'd be interested to know how that comes about.



You can make a case for the wakes of the front wheels... and the stagnated zone around the rear wheels... thats about it.

Since both are located towards the lateral extremeties of the floor, their effects on downforce production will be small due to an effective Kutta condition existing around the floor and sidepod of the car.

(Unless the front wheel wake is entrained under the floor of the car in significant quantities)

[RideRate]

Speedsense,
Nice post. I am certain now we are arguing nothing but semantics. Most of what you state are things we know about underbody function. We know if you seal it you can reap greater gains along with reducing pitch and roll sensitivity. We know that the distance between the ground and underbody is key to underbody function. We agree the whole purpose of the underbody is to create low pressure under the car while minimizing/removing high pressure areas. IMO, and I know not yours, removing high pressure under the car is a downforce making device. Sure the new net downwards force is really the higher relative pressure now acting all over the top of the car, but it's the clever underbody that allows this to take place. Whatever you term it is irrelevant as long as you understand how it is that underbodies and wings efficiently do their work. And I think we got that.

Here's the only final thing I don't 100% agree on. When you add underbody you are not gaining that much more efficiency from wings. The wings make downforce by the relative pressures acting on the top and bottom of the wings themselves so if this wing is up in free-stream behind the car, it makes no more net downforce itself even when you lower underbody air pressure. An underbody producing low pressure under the car is so effective because it turns the whole car into a 'virtual' wing. The atmospheric (or relatively higher pressure air on top) acts over the entire upper surface area of the car and it is this large area that reaps great rewards from the small pressure gradients generated. The force acting on your car due to the wing doesn't change so the wing isn't really any more effective.

Thank you for the discussion!
-B

[speedsense]

Another thought.



We are dealing with pretty much incompressible subsonics here... so raising the dynamic pressure will reduce the actual pressure acting on a tangential surface. Its a simple linear relationship.


Now... since the floor is that close to the ground, no significant vertical velocities can exist between floor and ground plane - so a pressure head need not be considered. Therefore the problem reduces to effectively a 1-D (longitudinal) issue (or 2-D if you consider tranverse flow)...

Thus, the only significant sources of high pressure areas can be low dynamic pressure areas... or areas where there is flow stagnation... which requires a significant deceleration on the ambient flow... on a flat floor, I'd be interested to know how that comes about.



You can make a case for the wakes of the front wheels... and the stagnated zone around the rear wheels... thats about it.

Since both are located towards the lateral extremeties of the floor, their effects on downforce production will be small due to an effective Kutta condition existing around the floor and sidepod of the car.

(Unless the front wheel wake is entrained under the floor of the car in significant quantities)

Kilco, please be gentle , I'm not a "schooled" engineer and my teachings are backwards in my learning process, I understood data acquisition and analyzed data to a professional degree before developing a better understanding of chassis dynamics or aero. It was the analysis that taught me these things and being around the many engineers (chassis design, aero and setup) to explain the data I would find. Any education or reading of theory or math came later in life or as I needed it.
I have somewhat an understanding of Bernoulli and fluid forces though you will lose me in a hurry quoting "Kutta condition" or what a pressure head is (leading edge of a high or higher pressure area?) I can make assumptions as to what you are referring to, but we all know where assumptions lead to.
I do have a significant library of books on aero and chassis, of SAE strength, and can't find a Kutta reference anywhere.

Now... since the floor is that close to the ground, no significant vertical velocities can exist between floor and ground plane - so a pressure head need not be considered. Therefore the problem reduces to effectively a 1-D (longitudinal) issue (or 2-D if you consider tranverse flow)...

Have you considered that ashpalt and concrete are imperfect in creating a flat surface and that a car doesn't maintain an attitude that maintains anything other than a chaotic movement to the floor gap? Most of the movement can be predicted, though yaw,heave,warp,etc. leaves a chaotic enviroment. Unlike a windtunnel floor or CFD program(though these are starting to be included in some sim and CFD programs) or a static measurement.
All of the tests in the previous post, were done in mostly static conditions, though with turning wheels and actuators replacing the shocks, with actual (suspension) track data running through them or controlled movements, roll pitch for instance. (this was before access to a 4/7 poster was possible)

[kilcoo316]

Kutta condition just means there can be no discontinuities in the flow* (be that pressures/velocities), it really applies to trailing edges of wings, but also applies along the side of the floor.



*Kutta does not apply to transonic flow where shocks can exist, these by their nature are "discontinuities" of sorts.



As for the thought on imperfect road surfaces, good point. What would the typical suspension travel on an F1 car be, and what would the frequency of a typical F1 damper be?

(To get a feel for the speed of vertical movement of the car suspension)

[speedsense]

Kutta condition just means there can be no discontinuities in the flow* (be that pressures/velocities), it really applies to trailing edges of wings, but also applies along the side of the floor.

I don't see how a gap between the ground and a floor gap, and a wingtip would have similar properties. Though floors and sidepod "floor" treatments tend to have pretruding "edges", that seperate air flow from upper body and lower. (simulating very long wing tips, I suppose)
Though this gap is bombarded with turbulent air flow in an open wheel car and somewhat in closed wheel. Not only from the inside of the wheels but also from the outsides of a wheel. The wheels act as an air pump, pumping in both directions a "cyclone" of turburlent air. This "cyclone" turns in it's path and continues down stream. Not only does the head wind cause this turning motion (path of the cyclone) but intended and unintended vortices also cause the turning motion. As the travel downstream continues, gaining in size, and slowing in circle velocity, it bombards not only the opening for the side pod (reason barge boards are so effective) but also strikes the side pod and enters underneath the sides of the side pods.
By way of Kutta princple, are you saying that this can't happen, because it does and these high pressure areas end up going underneath disrupting the stream underneath and creating lift. Maybe I just don't understand what you mean (most likely). BTW, this is only one example of many turbulent streams acting of the floor.
I can't find a great picture, of a F1 car in the rain, and streams of water launched into the air, arcing and headed to ground level and streaming into the side pod ground gap, as though it is being sucked,forced into it.

*Kutta does not apply to transonic flow where shocks can exist, these by their nature are "discontinuities" of sorts.

Over my head and don't understand. Does this mean because there are shocks on a car that Kutta cannot apply. Don't wingtips connected to wings that have flex or "shock" princples applied to them, negate this princple also? If I added a shock to a wing would that also fall into this?
Or do you mean shocks to the wing tip? Doesn't this happened with crosswinds and turbulent air...?

As for the thought on imperfect road surfaces, good point. What would the typical suspension travel on an F1 car be, and what would the frequency of a typical F1 damper be?

I've never worked with F1 car, though can relate what I've read. That the movements are in the .3", 8mm.. range. The ground effect cars that I've worked with operated in the .4 to .8", range of movement(depending on spring rate).
Spring Frequency (also inclusive of damper velocity) (again on ground effect cars) is in the 2 to 4 hz range, mostly + -3" at chassis readings and 4 hz is high end frequency. Don't know if it's a slower range or higher on an F1 car. Ranges out at the wheels tend to be higher values..

[Ciro Pabón]

I'm no aerodynamicist but I remember well my hidrodynamics course. For those in my position, here is a (repeated but more confusing ) explanation of kilcoo316 argument.

As usual, by showing my limitations, I'm sure kilcoo will teach me one thing or two.

The general idea of Kutta is that given the equations of flow, you can have multiple solutions. For example, given the equations to a wing, here you have two possible solutions:



Clearly the solution on the left is the correct one, isn't it?


So, what Mr. Kutta said is: at the trailing edge, given those two solutions, you could have these two different flows:



You can see that the flow around the trailing edge, in the second, lower image, turns around the trailing edge sharply: the speed of flow must be infinite there for the air to turn around "instantly".

Thus, to "forbid" that kind of (mathematically valid but physically impossible) solutions, he said:

In a physical flow (i.e. having viscous effects), the flow will smoothly leave a
sharp trailing edge. -Kutta Condition-

Thus, the left flow in the first image is the correct one. My teacher used to call this "boundary conditions". They are restrictions you don't find in the equations, but in the logic of the real world.

We, alumni, used to call it the Law of the Machete (you know, the tropical sword or knife used in the jungle): if equations don't fit, then you "cut" through them. It's a very useful law in engineering!

Kutta has a corollary: you cannot have two different vectors of velocity, like this:



Thus, Vupper and Vlower must be one, they cannot be two different vectors: the air moves in a flow that is (in that image) horizontal, like this:



--End of Kutta condition explanation--

Now, if I understand Kilcoo argument (which I doubt) what he says is that in the leading edge of the sidepod you cannot have air from the nose "crossing" toward the tunnel.

In the last image, "reverse" the speed vectors, so they point to the left. Ready?

Now, T.E. is not the trailing edge: it's the leading edge of the sidepod, seen from above the car. Vupper is the air coming from the nose. Vlower is the air coming from the wheels. So, they cannot be different vectors, they must be one.

I'm not sure at all if what kilcoo means is that air coming "from the wheels" is kind of a barrier that doesn't allow the air coming "from the nose" to enter the tunnel.

I know I got something wrong, but if I don't explain it, I'll never be corrected.

By the way, welcome, speedsense and Riderate. I found your explanations fascinating. Please, keep the posting. I believe you more than I believe kilcoo! (sorry, pal, but they seem to have hands on experience: the "Law of the Machete" applies! ).

[speedsense]

I'm no aerodynamicist but I remember well my hidrodynamics course. For those in my position, here is a (repeated but more confusing ) explanation of kilcoo316 argument.

As usual, by showing my limitations, I'm sure kilcoo will teach me one thing or two.

The general idea of Kutta is that given the equations of flow, you can have multiple solutions. For example, given the equations to a wing, here you have two possible solutions:



Clearly the solution on the left is the correct one, isn't it?


So, what Mr. Kutta said is: at the trailing edge, given those two solutions, you could have these two different flows:



You can see that the flow around the trailing edge, in the second, lower image, turns around the trailing edge sharply: the speed of flow must be infinite there for the air to turn around "instantly".

Thus, to "forbid" that kind of (mathematically valid but physically impossible) solutions, he said:

In a physical flow (i.e. having viscous effects), the flow will smoothly leave a
sharp trailing edge. -Kutta Condition-

Thus, the left flow in the first image is the correct one. My teacher used to call this "boundary conditions". They are restrictions you don't find in the equations, but in the logic of the real world.

We, alumni, used to call it the Law of the Machete (you know, the tropical sword or knife used in the jungle): if equations don't fit, then you "cut" through them. It's a very useful law in engineering!

Kutta has a corollary: you cannot have two different vectors of velocity, like this:



Thus, Vupper and Vlower must be one, they cannot be two different vectors: the air moves in a flow that is (in that image) horizontal, like this:



--End of Kutta condition explanation--

Now, if I understand Kilcoo argument (which I doubt) what he says is that in the leading edge of the sidepod you cannot have air from the nose "crossing" toward the tunnel.

In the last image, "reverse" the speed vectors, so they point to the left. Ready?

Now, T.E. is not the trailing edge: it's the leading edge of the sidepod, seen from above the car. Vupper is the air coming from the nose. Vlower is the air coming from the wheels. So, they cannot be different vectors, they must be one.

I'm not sure at all if what kilcoo means is that air coming "from the wheels" is kind of a barrier that doesn't allow the air coming "from the nose" to enter the tunnel.

I know I got something wrong, but if I don't explain it, I'll never be corrected.

By the way, welcome, speedsense and Riderate. I found your explanations fascinating. Please, keep the posting. I believe you more than I believe kilcoo! (sorry, pal, but they seem to have hands on experience: the "Law of the Machete" applies! ).

Thank you Ciro, the explaination helps me realize what may be what Kilcoo is speaking of (or not speaking of)...
though still not sure of the "shock" statements, except that maybe that sharper angle or converging of the steam would be quite different in the case of a shock to the tip or leading edge.

[riff_raff]

The underbody of an F1 car is more efficient at producing downforce (ie. it generally has a more favorable L/D coefficient) than the front and rear wings. But the front and rear wing elements are still useful, in spite of their less favorable L/D characteristics, because they can be adjusted to balance the front-to-rear aero grip of the chassis, and the underwing cannot.

All aerodynamicists strive to produce the most lift (or downforce) for the least drag penalty. And since the power required to overcome drag increases exponentially with speed, that is why F1 teams spend so much money on wind tunnel work. It gives the most bang for the buck.

The reason the underbody is more efficient at producing downforce than the front or rear wing is that the basic laws of physics dictate that accelerating a large mass of air at a small rate is more efficient than accelerating a smaller mass of air at a higher rate. Thus the larger mass flow rate at a lower delta V across the under body will produce a given downforce with less drag penalty, than the smaller mass flow rate with a greater delta V across the front or rear wing airfoils.

If you look at some of the unrestricted ground effect F1 cars of the 70's, you'll note that they eliminated the front wing completely.



Regards,
Terry