Tubulence - Friend or Foe? | Here's a lot of Aero info, not spread in several threads

[trebuchet03]

Rather than sporadically give information in posts... Let me take a moment to explain a few things with respect to aerodynamics, cD and a few terms that go with that.... Basically, I'm about to regurgitate some of the concepts you might learn in a fluid mechanics course (but minimizing the math component)

First, some definitions

Reynolds number: a ratio between the inertial forces (momentum of the working fluid) to viscous forces (property of resistance to shearing) | dependent on: body length, fluid velocity, fluid density, dynamic fluid viscosity, kinematic fluid viscosity.

Frictional Drag: this is the drag between the moving fluid and relativity stationary surface. May be referred to as viscous drag - I'll switch back of forth most likely. This value is dependent/sensitive on the Reynolds number.

Pressure Drag: this is the drag as a result of a pressure drop - wake. Think of that big blast of wind from a semi truck, or the waves generated by a boat. Pressure drag is mildly dependent on the Reynolds number, but generally not sensitive to it. Pressure drag is highly sensitive to body shape.

For both Frictional and Pressure drag.... It's important to note that they are both VERY dependent on fluid viscosity. That is, in inviscid (viscosity = 0) flow, both of drop to 0.

AoA: Angle of attack - draw a line in parallel to fluid flow. Now draw a line from the leading edge of a body to the trailing edge. Measure the angle between these two lines.

Streamlined Body: Drag losses are primarily due to viscous losses. Streamlined bodies of the same thickness are significantly more aerodynamic than a bluff body. Note that a streamlined shape doesn't necessarily mean a streamlined body - at a high AoA, flow separation causes great pressure drops.
Bluff Body: Drag losses at high Reynolds numbers are primarily due to momentum losses - meaning, pressure drop as a result of a wake.

Free Stream: This is the flow that is not impeded by a body.


Boundary Layer: I can spend a lot of time on this, but I'll abridge it... The boundary layer is the fluid directly beside the body. The general profile of the boundary looks something like this:
http://history.nasa.gov/SP-4103/p529.jpg
where the length of the lines represent a velocity vector - the longer the line, the faster the flow. At the point where the fluid meets the body, there is a zero slip condition. That is, at this point the fluid does not move. There are some important implications of this as it applies to differential calculus - but don't worry about that, just understand this concept ;) Below is the shape of the boundary layer (note that the y scale is increased for easier reading):
http://www.cartage.org.lb/en/themes/Sciences/Physics/Mechanics/FluidMechanics/RealFluids/BoundaryLayers/img00032.gif
Note the several phases of the boundary layer:

Laminar
Transitional
Turbulent

And just for fun - here's a real world example:
http://history.nasa.gov/SP-4103/p530.jpg

Laminar Flow: Fluid "layers" (streamlines) flow over each other in a parallel fashion smoothly over each other.
Transitional Flow: This is more of a mathematical zone as laminar turns into turbulent. I call it mathematical because there are specialized equations that apply to transitional flow only.
Turbulent Flow: The streamlines, instead of slipping over each other, diverge in chaotic eddies. Note that chaotic does not mean random, it "simply" means that it is highly dependent on initial flow conditions - small changes in these conditions create great differences in output.

Please note, there is no critical distinction between laminar, transitional and turbulent flow. The mathematical distinction is almost arbitrary - it's just a number that the community has agreed to use :thumbup:

Okay, I think that just about covers some minimal basics - if I make any knowledge assumptions, please let me know and I'll elaborate :thumbup: :thumbup:

I'm going to talk about bluff bodies for the most part as that's what our cars are (with exceptions - basjoos!).

So lets look at some simple objects and what happens.... A cylinder is a classic example:
http://www.flometrics.com/services/cylinder/cylslo.jpg
So, the flow in the front is generally nice, even and laminar. But downstream, there's some nasty eddies. The large vorticies off of a cylinder have their own natural frequency dependent on fluid velocity and cylinder diameter - coming at regular intervals. These remove energy from the body. The wake of a body typically has this same eddying motion, but not necessarily at a consistent frequency.

This is actually a VERY important phenomena as if this frequency matches the natural frequency of the cylinder (think a street lamp pole), the cylinder will vibrate divergently and without control, eventually self destruct. This is why many light poles, especially in wind prone areas, are tapered. Additionally, power lines may have spoilers and damping mechanisms attached. Even large sky scrapers have vibration control for wind induced motion - one method uses VERY heavy weights (think tons) on the top floor of the building; given a certain wind speed, oil is pumped under the weights so they can move (constrained and attached to the structure by springs)...

How about a sphere?

http://www.princeton.edu/~asmits/Bicycle_web/pictures/R_combined.GIF

The sphere on the left (a) shows what happens with laminar flow. As soon as the pressure drop becomes significant you get separation, bugger.

So what's the deal with the picture on the right? How did we delay separation? That sphere has a trip wire. What does a trip wire do? It perturbs the boundary layer, a lot. Enough so that nearly the entire boundary layer in contact with the sphere is turbulent.

So how does this change cD? Here's some graphics:
http://www.princeton.edu/~asmits/Bicycle_web/pictures/drag_coefficient.GIF
cD versus Reynolds #

http://www.princeton.edu/~asmits/Bicycle_web/pictures/flow_patterns.GIF
Points on the graph above correspond to the letters below each case.

"Whoa Whoa Whoa!! Are you telling me cD changes with velocity?"
Yes, yes I am - but don't get too excited as this mainly applies to small objects. cD drops significantly as the Re# increases. BUT, it remains nearly constant for large Re#'s in the laminar region and approaches a constant value after the transitional region. As a car is very wide compared to something like a tennis ball, the Reynolds number is high - very high in fact (on order of 10^6 @ 55mph and 10^5 at ~3mph).


Balls!
Sports balls, believe it or not, have some great design... Smooth balls are not very efficient aerodynamically. This is why golf balls have dimples, tennis balls have fuzz, cricket balls have a stubby seam, baseballs have stitches, American footballs are textured, soccer balls have seams (Everywhere). Okay, some of this is traditional - but it still plays a role.
http://www.princeton.edu/~asmits/Bicycle_web/pictures/roughness.GIF
Roughness and cD... Having a bit of roughness induces turbulence. Look at the photo above of the sphere with the trip wire - surface roughness is very similar, in effect, to a trip wire.

This results in a reduction of cD as shown in the above graph (which just so happens to also be in my fluid mechanics book - Fundamentals of Fluid Mechanics produced by John Wiley & Sons :p). Relative roughness is t/D, t being the thickness of "rough" and D being the diameter of a sphere/cylinder. The dashed line on the left is a golf ball.

And on the subject of golf balls... why? The dimples! Why? The dimples act, in effect, like a trip wire. They promote a transition to turbulent flow. By doing so, we get higher skin friction (yet another name for viscous losses). BUT, remember the definition of a bluff body? A golf ball is a sphere - and spheres are bluff bodies. Most of their losses come from the wake - so if we can move the wake further back, we can reduce the wake size and reduce the losses from the wake. It just so happens that by increasing the turbulence, you decrease your drag by reducing where most of the drag is coming from - flow separation (wake).

------
Okay, so what's all of this talk about the benefits of turbulent flow? Turbulence is the most misunderstood concept with regards to aerodynamics, in my opinion. Nature has known for millions of years of the benefits of turbulence. This is why dermal denticles on sharks help lower their cD. Likewise the bumpy tubercles of a whale fin. etc.
http://www.whalepower.com/drupal/files/userfiles/image/humpback_fin.jpg

The key difference between induced turbulence and the turbulence as a result of a shape is the size. Larger eddies, swirls, etc. contain more energy than smaller ones. So if we induce smaller eddies, with a small amount of energy - we move into the turbulent regime without a whole lot of loss that you'd typically associate with the word "turbulence". This will typically be referred to as "energizing the boundary layer." This can happen on a very VERY small scale - the smallest of which is the Kolmogorov micro scale (http://en.wikipedia.org/wiki/Kolmogorov_microscales). <-- keep in mind that the mathematic description of turbulence is not solved - we just can't do it, yet :p

Some very notable physicist when asked what they would like to ask God have replied:
Werner Heisenberg: "When I meet God, I am going to ask him two questions: Why relativity? And why turbulence? I really believe he will have an answer for the first."
Horace Lamb: "I am an old man now, and when I die and go to heaven there are two matters on which I hope for enlightenment. One is quantum electrodynamics, and the other is the turbulent motion of fluids. And about the former I am rather optimistic."
^^both of these were mentioned in my lectures :p

--------
Vortex Generators.....
Okay, so I've been building up to this as it has come up a lot recently - trying to give some background before I get to it.

Here's the answer to some straightforward (I hope) questions:
1. Does it increase turbulence?
Yes - the idea is, move the wake back with turbulent flow
2. Does it reduce cD?
The devil's in the details. Properly applied, absolutely yes. Nature has proven this through millions of years of evolutionary change. CFD and tunnel testing too :p
3. wtf counts as a vortex generator?
Almost anything that perturbs flow - from commercially sold tabs to dimples to stiff fuzz (think tennis ball). Vortex Generator does not necessarily refer to one particular product (nor do I endorse any such product).
4. How much improvement?
See #2 - Mitsubishi has published a paper reporting a cD reduction of 6 points (.006) on a Lancer which can be considered significant in the cD world. They used delta wing type vortex generators placed 100mm in front of the predicted separation point:
http://www.primitiveengineering.com/blog/img/delta_effect_vg.jpg
http://www.primitiveengineering.com/blog/img/vortex_top.jpg

Note the smaller wake size and how the separation point is further down the rear glass:
http://www.primitiveengineering.com/blog/img/vortex_gen_cfd.jpg


Whew, freeking long.... Questions/additions? Let me know if I missed a typo too :)

[TomO]

Good read, trebuchet! Thanks for the "scoop" on aerodynamics, lol.

[CO ZX2]

Tubulence - Friend or Foe?

Tubulence. Is this effect brought about by sitting down in the bathtub too fast?

[Silveredwings]

Tubulence - Friend or Foe?

Tubulence. Is this effect brought about by sitting down in the bathtub too fast?

I think it's a term for the bubbles you create while sitting in the bathtub. ;)

[rh77]

Excellent write-up!

I vote to stickify it...

RH77

[trebuchet03]

Tubulence - Friend or Foe?

Tubulence. Is this effect brought about by sitting down in the bathtub too fast?

More the horizontal growth as a result of middle aged spread....

But how embarrassing, a critical spelling error in the title :o

[Mighty Mira]

One thing I have to wonder... if vortex generators are so great, why aren't they on the back of the insight, the prius, the UFE-III, the VW 1 litre car, etc. etc? I know length is a constraint, but given the same length, why wouldn't they attempt to drop the rear at a steeper angle and add some VGs, that way they could decrease the size of the wake still further, if it were possible.

[DRW]

Good stuff!
I think I'm seeing a Karmann vortex that's formed at the back of the car in the graphic that's labeled '(a)With VG' while the car '(b)without VG' doesn't show signs of a Karman vortex. Do you see it too? Does it mean anything? Is that a good thing or not?

[trebuchet03]

One thing I have to wonder... if vortex generators are so great, why aren't they on the back of the insight, the prius, the UFE-III, the VW 1 litre car, etc. etc?

It would be great if someone had some tuft photos of a prius' rear glass at speed. The Prius hatch is closer to a streamlined shape (as is the 1L and UFE-III). So if the separation point is really far down the glass, or not until a sudden change in body shape (rear cut off), there's more benefit of keeping laminar flow and dealing with skin friction.

I dug around autospeed a little more, and found some more information :p http://www.autospeed.com/cms/A_3061/article.html. They specifically tested the air tab product and they did have promising results - similar, in effect to my tuft testing while behind a semi.

I know length is a constraint, but given the same length, why wouldn't they attempt to drop the rear at a steeper angle and add some VGs, that way they could decrease the size of the wake still further, if it were possible.

I'm not sure I follow that? Why would that make the wake size smaller? Intuitively, I'm thinking of a steep angled rear end compared to a gently sloping rear end - the steep one has a pretty bad wake issue... Maybe I'm just not understanding what you said?


Good stuff!
I think I'm seeing a Karmann vortex that's formed at the back of the car in the graphic that's labeled '(a)With VG' while the car '(b)without VG' doesn't show signs of a Karman vortex. Do you see it too? Does it mean anything? Is that a good thing or not?

I do see an obvious downward motion there - but I wouldn't call it a Karman Vortex - a requirement for that would be periodic oscillating eddies - they could be there, it's just hard to tell from one image:
http://www.princeton.edu/~asmits/Bicycle_web/pictures/flow_patterns.GIF
^center image (c)

I think that little downward tail has something to do with the upward curve (compared to photo (b)) and the faster moving flow above it (in yellow - compared to the slower green in (b)). It would be nice to see a few frames to get an idea of the bigger picture there :)

[Gary Palmer]

Very nice writeup.

[Mighty Mira]

I'm not sure I follow that? Why would that make the wake size smaller? Intuitively, I'm thinking of a steep angled rear end compared to a gently sloping rear end - the steep one has a pretty bad wake issue... Maybe I'm just not understanding what you said?

I'm referring to the area of suction behind (the area below boundary layer separation) as "wake". Probably the wrong choice of words.

My understanding is that with the sphere and the trip wire, the trip wire trips the boundary layer into turbulence (energizes it), and as a result, the boundary layer can sustain a steeper angle than the prius-like 17 degrees or so before separating. For the same length car, you might have a more effective boattail that way, since the area of suction would be smaller. i.e. On one hand (prius style), you have a larger area of suction behind the vehicle, and on the more extreme rake + VGs or turbulator tape or whatever you use to trip the boundary layer, you have less area of suction but more energy lost to trip the boundary layer.

Do you get what I'm saying? Think VW beetle but with boundary layer tripping somewhere on top, and a small wing near the very bottom, to get the air leaving in a horizontal direction.

[trebuchet03]

Mighty, I had an interesting conversation with the South East ASME district leader (at least, afterwards - I was told that was his position) - and in speaking with him with respect to HPV fairings, it was recommended (among other things) to explore completely turbulent flow - tripping with some sort of speed bump on the leading edge...

In the next couple months, we'll be modeling that to see what happens - as turbulence is chaotic, there's good chances that it will help, hurt and/or do nothing :p But for us, we're not sure how much of an effect it will have at our relative low velocities :p

On the polar opposite side of things, it was also recommended that we check out deHavilland slots to maintain laminar flow (and provide necessary ventilation). deHavilland slots basically suck air into the foil just before the transition point - I believe this basically "resets" the BL, or at least decreases the thickness.


-----
Any good ideas on how to trip the BL? I was thinking of just experimenting with a triangular shape -like a FedEx Triangle mailer sort of shape. Any suggestions?

[Mighty Mira]

Any good ideas on how to trip the BL? I was thinking of just experimenting with a triangular shape -like a FedEx Triangle mailer sort of shape. Any suggestions?
Go with some sort of thick tape, and cut it in a zigzag pattern as per here. (http://pagesperso-orange.fr/scherrer/matthieu/aero/nimbus4e.html) I'm guessing that a material store will have these sorts of scissors. As to where to find 1mm thick tape, not sure exactly.

I suspect that a lot of hatchbacks where the hatch is roughly 30 degrees or so might conceivably have good airflow with the addition of such turbulator tape. However, I have to wonder why the front windshield wipers aren't tripping the boundary layer. Or maybe the energy they add is too diffuse by the time the air gets to the rear of the car. Or the top of the windshield for that matter.

[Jim Dunlop]

Thank you for the nice write-up. It's interesting for me to see how N(Re), viscosity, turbulence, etc. apply to aerodynamics. In my field, we mostly think of turbulence in terms of how it affects mixing in a pipe (i.e. flushing out "dead legs") and heat transfer. Also we only care about Newtonian fluids for the most part.

[DRW]

To summarise, it seems like turbulators help when there is unwanted airflow separation. A successfull turbulator is placed a certain distance in front of the point of separation.

[trebuchet03]

To summarize, it seems like turbulators help when there is unwanted airflow separation. A successful turbulator is placed a certain distance in front of the point of separation.

Technically - they should be placed just before the transition. So it's correct to say before separation, but it's the zone before separation you're attempting to fix (and thus prevent separation).

[Snax]

On the polar opposite side of things, it was also recommended that we check out deHavilland slots to maintain laminar flow (and provide necessary ventilation). deHavilland slots basically suck air into the foil just before the transition point - I believe this basically "resets" the BL, or at least decreases the thickness.
This sort of thing can be easily seen in commercial airliner flaps. At full extension, they essentially become multiple airfoils working together as one virtual airfoil. The same thing is seen in rear wings of racecars like F1.

As for a boundary layer trip, I think any shape that does not promote laminar flow over it would be effective. I think the more significant question is the height of the trip from the surface. How thick does the trip layer need to be for best efficiency?

There is obviously a sweet spot where too much depth unnecessarily accelerates more than just boundary layer air. It will remain attached further back, but it ends up being wasted energy. Likewise, too little depth won't accelerate the boundary layer enough and it will detach before the trailing edge.

I suspect there is a formula that can explain this relationship fairly accurately - or at least ballpark a decent starting point based on some general assumptions. Intuitively, I 'get it'. But mathmatically, er....

[Mighty Mira]

To summarise, it seems like turbulators help when there is unwanted airflow separation. A successfull turbulator is placed a certain distance in front of the point of separation.
From what I can see (from the picture of the sphere with boundary layer tripping), there is still a minimum angle that must be maintained or the flow will still separate. i.e., no turbulator is going to help the back of a semi (or similar shape). The boundary layer on the sphere separated at 45 degrees. Note that on the back of the lancer the flow still appears to be separating. (Although that's trusting that the CFD is a true picture, which it isn't necessarily.)

I suspect that this angle will vary depending on how much the boundary layer is energized (turbulator height) and how far back the boundary layer was energized.

[trebuchet03]

As for a boundary layer trip, I think any shape that does not promote laminar flow over it would be effective. I think the more significant question is the height of the trip from the surface. How thick does the trip layer need to be for best efficiency?

All good questions - we intend to investigate :D


I suspect there is a formula that can explain this relationship fairly accurately - or at least ballpark a decent starting point based on some general assumptions. Intuitively, I 'get it'. But mathmatically, er....

I am not aware of equations for such a thing - there very well could be though. The reason, Turbulent flow isn't fully understood (especially in a mathematical sense) - Turbulence is a chaotic system. Complicating things more - we're dealing with curved surfaces :D There are, however, relationships for laminar flow, and fairly accurate equations in the transition zone. But once you get to high Reynolds numbers, small changes drastically change the outcome.



no turbulator is going to help the back of a semi (or similar shape).

Absolutely right :thumbup: At least, there no research that supports this. Really, a vortex generator fixes an aerodynamic flaw in the design. Think of it as a band aid - the cure being a costly redesign or potentially difficult modification that many/most are not willing to do.


Thank you for the nice write-up. It's interesting for me to see how N(Re), viscosity, turbulence, etc. apply to aerodynamics. In my field, we mostly think of turbulence in terms of how it affects mixing in a pipe (i.e. flushing out "dead legs") and heat transfer. Also we only care about Newtonian fluids for the most part.

Awesome :thumbup: Are you involved with heat exchanger design? Regurgitating class lectures - turbulence enhances heat transfer (higher Re# = better) - I'm sure you knew that already. As for the "N" - is that a reference to the Nusselt number (I imagine that, and perhaps the prandtl number are used often in your field :) )?

[Mighty Mira]

Absolutely right :thumbup: At least, there no research that supports this. Really, a vortex generator fixes an aerodynamic flaw in the design. Think of it as a band aid - the cure being a costly redesign or potentially difficult modification that many/most are not willing to do.

Thinking about it some more... aren't VGs and turbulators separate things? i.e. you'd use VGs if you are generating large vortexes at the corners of the car (e.g. back of a BMW). In something a bit rounder but with a rake of 20 to 45 degrees back, you'd want to trip the boundary layer with a turbulator so as to decrease suction at the back.

And with something like the Lancer, they could have achieved the same thing with a small ridge on the roof of the car somewhere.

[Snax]

Or perhaps a wing like Subaru did with the STi.

[ZugyNA]

I think it's a term for the bubbles you create while sitting in the bathtub. ;)

That is flatulence. The French use wine corks for this problem.:cool:

[ZugyNA]

It would be great if someone had some tuft photos of a prius' rear glass at speed. The Prius hatch is closer to a streamlined shape (as is the 1L and UFE-III). So if the separation point is really far down the glass, or not until a sudden change in body shape (rear cut off), there's more benefit of keeping laminar flow and dealing with skin friction.

I dug around autospeed a little more, and found some more information :p http://www.autospeed.com/cms/A_3061/article.html. They specifically tested the air tab product and they did have promising results - similar, in effect to my tuft testing while behind a semi.

There are four articles there...good reading. But with their last test on the Honda Insight...why weren't the airtabs placed up above the rear glass as with the Toyota Prius right above? Separation not expected? Not a fair test...who would expect improvements on an aero tweaked shape?

They need to test something like the green Honda SUV.

http://us1.webpublications.com.au/static/images/articles/i30/3060_13lo.jpg

[Snax]

I think it would have made sense for them to evaluate the Insight at significantly higher speeds. But as it was, the whole exercise appeared to be pointless on that car.

[trebuchet03]

There are four articles there...good reading. But with their last test on the Honda Insight...why weren't the airtabs placed up above the rear glass as with the Toyota Prius right above? Separation not expected? Not a fair test...who would expect improvements on an aero tweaked shape?

They need to test something like the green Honda SUV.

http://us1.webpublications.com.au/static/images/articles/i30/3060_13lo.jpg

As far as "fairness" is concerned... They weren't comparing vehicle to vehicle. They were comparing a concept to another concept. The Prius has a much steeper angle compared to the insight (separation is not expected - as can be seen while driving in the rain). Doing the "wake fill" test on the Prius wouldn't really help if flow has already separated.

As far as the whole "wake fill" idea....

But what about the idea that vortex generators can pull extra air into the low pressure wake, increasing its pressure and so decreasing drag? As far as we’re aware, there is no independent scientific evidence for this idea (as opposed to the energizing of boundary layers, where there is 50+ years of experience on aircraft!).

[Erik]

OK, I am sold on the idea of using a turbulator to trip the boundary layer ahead of my hatchback so I don't get held back by suction at highway speeds.

I am going to put a thin zig zagged layer of clear silicone caulk across the roof of my civic hatchback. (I can always scrape 99% of it off later with a razor blade)

The question is, how far should I put the caulk "trip" strip ahead of the hatchback? 100mm like Mitsubishi did?

[DRW]

OK, I am sold on the idea of using a turbulator to trip the boundary layer ahead of my hatchback so I don't get held back by suction at highway speeds.

I am going to put a thin zig zagged layer of clear silicone caulk across the roof of my civic hatchback. (I can always scrape 99% of it off later with a razor blade)

The question is, how far should I put the caulk "trip" strip ahead of the hatchback? 100mm like Mitsubishi did?

In Mitsu's test, the VG were placed 100mm ahead of where the airflow separated from the rear glass. Similarly, in the link to the glider modification, the zigzag tape on the wing was placed ahead of the point of separation. The first step is to find where airflow separates from the hatch. I think a good way to determine where flow separates on your car is to do some tuft testing at various speeds.

I did some tuft testing on my car. There's side view pic of my car in my profile for reference. At 60mph flow started to separate at about 2/3 of the way down the rear hatch, or about 6 inches away from the rear deck. The angle of the glass at the point of separation is about 21 degrees. The flow wasn't completely turbulent, the tufts were still pointed downstream, but they wiggled noticeably more than the tufts just upstream.

Now you've got me thinking of using clear silicone caulking to make a zigzag strip!

Edit: oh wait, I just saw your car pic in your profile. It's pretty clear where the air would separate.

[Erik]

DRW- Thanks for the info/help, but my hatch glass is pretty close to vertical. Maybe a 15 degree slope downward (a station wagon would be 0 degrees). So it seems like I need to put the caulk strip on the roof- right?

Its gonna be hard for me to do tuft testing with tufts on the roof unless I can find a volunteer to hang out the window and watch them at 55 mph :D

[DRW]

Ah! I edited my post at the same time you wrote yours!

Erik, the trailing edge of the roof of your car is similar to the Insight. They both have a gently tapered rooflines terminating with a sharp edge. This is done to give the air a definite point to separate from the body. See part one of the vortex article here www.autospeed.com/A_3058/cms/article.html
where they state:

" Cars with a two-box shape (eg hatchbacks and wagons) are always stuck with flow separation at the end of the roof, so creating a larger wake. In those cars, and also at the trailing edge of the boot in three-box cars, the separation should be clean – ie the flow shouldn’t wrap around the end of the roof or the boot lid.


On hatchbacks and wagons, roof extension spoilers achieve this clean separation, as do sedans by means of the sudden transition from horizontal to vertical at the trailing edges of the boot lid."

My interpretation of what that article is saying is that your car already has a nice aero shape at the top of the roof/hatch junction. Sure, it's a large wake, but it has clean separation. VG's might reduce the wake on your car, but it would also make the wake more turbulent.

[Erik]

DRW- Thanks, that was a great article. I am going to try it just to see if ther eis any change in my mileage, but it looks like a Kamm back is the only aero option to help the back end of my car

[GeneW]

Have to see the actual data but do those delta wing doodads generate vortices or generate laminar flow?

Gene

[DRW]

Have to see the actual data but do those delta wing doodads generate vortices or generate laminar flow?

Gene

Those delta wing doodads are called vortex generators (VG's).

[mji988]

Thanks for the fluids review. Besides vortex generators, will adding heat help trip the boundary layer? If so, do you think exhaust heat could be used to help accomplish this?

[RoadWarrior]

Thanks for the fluids review. Besides vortex generators, will adding heat help trip the boundary layer? If so, do you think exhaust heat could be used to help accomplish this?

You can probably use exhaust flow to do it, in a similar manner to "blown flaps".

It's interesting actually how many vehicles now have a large transverse muffler between the bumper and the gas tank, yes the oval shape probably helps fill the hole, stopping the bumper becoming a drag parachute and cleaning up underbody aerodynamics. There is however potentially a small benefit from them expanding air as it passes over it, when hot, and providing a minute amount of thrust.

By exploiting the Coanda effect, exhaust flow could be made to "drive" a Coanda surface at the underside rear of the vehicle. This would have the benefit of helping clean up rear airflow, while also "sucking" the vehicle to the road somewhat, and should also provide around 15lb of thrust at highway speed.

[green swift]

Tubulence - Friend or Foe?

Tubulence. Is this effect brought about by sitting down in the bathtub too fast?

When you installed your air dam did you see any increase on the scan gauge?
IF YOU DID HOW much?

green swift

[sonyhome]

I had done a mini survey on VGs in cleanmpg.com,here (http://www.cleanmpg.com/forums/showpost.php?p=34890&postcount=6), which I'll recap and complement:

Vortex Generators - VGs:

Mitsubishi EVO MR (http://lacar.com/modules.php?name=Ne...rticle&sid=479) $300
Air Tabs (http://www.airtab.com) $2.50/ea (+10% MPG CR-V gen-1)
AirFlow Systems (http://www.prfprod.com) $10/25ft (I contacted them on FE and they honestly replied (http://www.gassavers.org/showpost.php?p=103944&postcount=23))
Vortekz (http://www.manufacturersdepot.com/ProductDetail.jsp?LISTID=800000A6-1180622063) $20/10pc
Fuel Savers (http://www.fuelsavers.com.au) AU$110/9pc (up to 11% savings)

Boundary Layer Energizers etc., not quite VGs:

UC Davis BLEs (http://flight.engr.ucdavis.edu/~hjshiu/research/irflowviz.html)

Sinha Deturbulator tape (http://www.sinhatech.com) (This claims up to 30% MPG increase, still in R&D)
Found in Bman83GL Deturbulator tape (http://www.gassavers.org/showthread.php?t=5638) topic.

Zig-zag (http://pagesperso-orange.fr/scherrer/matthieu/aero/nimbus4e.html) and dimple tape(2 (http://ecomodder.com/forum/showthread.php/showpost.php?p=30203&postcount=39),3 (http://www.wingsandwheels.com/page29.htm))

Spoilers
Spoilers (not wings) are used to detach the flow of air from the rear of the car to create a pocket of non-turbulent stagnant air behind the car to finish the air flow like a teardrop (makes a virtual boat tail) and reduce Cd caused by drag from poor rear aerodynamics. Used by modern wedge shaped hatchback cars.

(need references) see Citroen Cx (http://en.wikipedia.org/wiki/Citro%C3%ABn_CX) or Toyota Prius.

Other stuff

Aerody.org articles (http://aerodyn.org/Drag/)
Wolf-Heinrich Hucho - Aerodynamics of Road Vehicles From Fluid Mechanics to Vehicle Engineering (http://www.normas.com/SAE/pages/R-177.html)
Recumbents.com (http://www.recumbents.com/car_aerodynamics/)
Aerohead's post ecomodder (http://ecomodder.com/forum/26356-post23.html)
Car Aerodynamics (http://en.wikipedia.org/wiki/Automotive_aerodynamics)


Diffuser (http://en.wikipedia.org/wiki/Diffuser_%28automotive%29) (read for front/rear impact on underbody drag) and ground effect (http://en.wikipedia.org/wiki/Ground_effect_in_cars)

Wing (http://en.wikipedia.org/wiki/Wing_%28automotive%29) and Aerofoil (http://en.wikipedia.org/wiki/Airfoil)

Maxing out tire pressure (http://www.officer.com/article/article.jsp?siteSection=19&id=27281)

Drag Coefficient - Cd/Cx

Wikipedia listing (http://en.wikipedia.org/wiki/Automobile_drag_coefficients)
Daox CdA listing (http://www.tercelreference.com/Downloads/CdA_List.htm) from ecomodder (http://ecomodder.com/forum/showthread.php/cda-spreadsheet-1935.html)

Not aero

Driving tips (http://ecomodder.com/forum/EM-hypermiling-driving-tips-ecodriving.php) (some usefull A/T tips)
Pulse and glide explained with MPG results (http://www.metrompg.com/posts/pulse-and-glide.htm)

Autospeed articles

As for Autospeed aero articles, I had read a few cool ones (VG experiments, wool tuft testing, airflow test):

Vortex:
http://autospeed.drive.com.au/cms/A_3061/article.html
http://autospeed.drive.com.au/A_3060/cms/article.html
http://www.autospeed.com/A_3059/cms/article.html
http://www.autospeed.com/A_3058/cms/article.html

Aero testing:
http://autospeed.drive.com.au/cms/A_108676/article.html
http://autospeed.drive.com.au/A_108675/cms/article.html
http://autospeed.drive.com.au/A_108674/cms/article.html
http://autospeed.drive.com.au/A_108656/cms/article.html

Aerodynamics:
http://autospeed.drive.com.au/cms/A_107773/article.html
http://www.recumbents.com/car_aerodynamics
http://www.edmunds.com/advice/specia...4/article.html
http://www.ibsinger.com/gallery.htm?...&galleryId=768

Under car airflow (Prius +5MPG):
http://autospeed.drive.com.au/cms/A_2456/article.html
http://autospeed.drive.com.au/cms/A_2455/article.html

Undertrays spoilers and bonnet vents
http://autospeed.drive.com.au/cms/A_2162/article.html
http://autospeed.drive.com.au/cms/A_2160/article.html
http://autospeed.drive.com.au/cms/A_2159/article.html

Wiper blade deflector for noise
http://autospeed.drive.com.au/cms/A_2445/article.html

[dosco]

You can probably use exhaust flow to do it, in a similar manner to "blown flaps".

In F1 cars this method was used in the diffuser. Designers found that the change in gas flow due to changes in engine RPM created balance problems. It was eventually scrapped.

In a regular roadgoing car, I'm not sure there would be enough energy left in the flow (after the muffler) to make much of a difference. Get rid of the muffler, then maybe you might have something (in addition to a headache from the noise).

[RoadWarrior]

Yeah, I can see that being a problem, great downforce through a long curve at 10K rpm, but it's time to shift.... annnnnnnnd the back end goes loose.

[RoadWarrior]

Sinha Deturbulator tape (http://www.sinhatech.com) (This claims up to 30% MPG increase, still in R&D)
Found in Bman83GL Deturbulator tape (http://www.gassavers.org/showthread.php?t=5638) topic.


That's interesting, I mentioned on mpgresearch a while back that it might be possible to use closed cell foam "draft insulating" tape strip to get the same effect. I knew about the effect from a 1970s aeromodeller annual article on use of turbulation strips on model sailplanes. I have 2 points I am going to apply it on Marvin when I get off my butt and get into the aero mods. i) near the back edge of the rear ventilator windows, such that when they are cracked more air will tuck in behind them, and ii) in front of the back light clusters, because modding rear shape there is difficult.

[DRW]

Dam, that's a lot of good info! I'll be reading for weeks.

[sonyhome]

That's interesting, I mentioned on mpgresearch a while back that it might be possible to use closed cell foam "draft insulating" tape strip to get the same effect.

If you have it handy, link it... We gotta be careful planes use those to attach flow and avoid bubbles that disrupt plane stability, and not to improve MPG. I dunno if one implies the other? BTW, added http://aerodyn.org/Drag link to my previous post.

[sonyhome]

Diffuser:

In one of the links, it mentions for diffusers the use of a hollow space under the rear of the car to allow airflow to expand and slow down before exiting the underbody, to avoid further turbulences in the wake. If anything then the holes behind the rear bumper should not ne covered up. Maybe poke holes in the bumper at best. Maybe just put vertical separators to direct the flow and reduce turbulences in that gap?

[RoadWarrior]

If you have it handy, link it... We gotta be careful planes use those to attach flow and avoid bubbles that disrupt plane stability, and not to improve MPG. I dunno if one implies the other? BTW, added http://aerodyn.org/Drag link to my previous post.

Yes, sometimes you've gotta think upside down and backwards on what applies to planes to apply it to cars, because you don't really want high lift.

Unfortunately what I thought I posted over there seems to have been eaten by one of the hacks or server crashes.

Anyway, I think it was the mentioning of use of turbulation devices to pre-empt flow detachment and formation of large vortices, and to cause small vortices that would entrain the bulk airflow round sharper angles than it would normally follow without separation. I determined that a minimum 1/4 inch feature was needed at the typical Reynolds numbers we see at highway speed in a typical size car, for it to have significant effect on airflow. The ideal spot for such turbulation devices is supposed to be in the rear 3rd of a body. I did some rough back of envelope figuring that seemed to suggest that about 2 inches of surface is the minimum that flow will reattach to at "our" reynolds numbers, so placing of turbulation strips should be a minimum two inches from changes in curvature, unless you're using them to separate flow.

Bear in mind that trucks by virtue of length are in a different calss of Reynolds numbers than cars, so that devices which work on trucks may need to be scaled differently to be effective on cars and vice verse i.e. quarter inch strips won't do crap on a semi-trailer, they may be getting towards marginal on full size extended vans. Also they would have to be placed further than 2 inch from the trailing edge.

The most effect is seen apparently with groups of 3 strips spaced their own height apart.

However, I have come to consider a slightly different application approach more theoretically desirable. Because the angle at which the air departs the vehicle body determines the "angle of attack" as it were of the body shape as an airfoil, you want to get that angle as shallow as possible, near horizontal. There is of course though a balance between doing this and adding too much base drag. One needs to keep the angle of attack as neutral or negative as possible otherwise the lift forces are actually pulling backwards on the body, giving induced drag. Keeping the "trailing edge" higher than the "leading edge" is one reason why it's beneficial to have a clean edge separation off a trunk or stub-trunk at about half the height of the vehicle at the rear. However, this doesn't fix everything because of a outflow from the roof centerline that will happen due to air at higher pressure trying to take the easiest escape route rather than going all the way over the top. This leads to similar phenomena as tip vortices on aircraft wings. This flow will attempt to wrap over the sides of the vehicle, and will cause a net downflow effect on air departing the vehicle at the rear. This is actually not good, Kammbacks get talked up, but to be truly efficient they should kick the air back up before separation. There is a balance however between where base drag reduction is good and where induced drag becomes a problem.

Anyway..... turbulation strips could be doing 3 things at once... Placed at a shallow angle to the airflow, 15-30 degrees in groups of 3 they would i) act to keep flow attached as turbulators, ii) enhanced lower drag action by virtue of delta wing swirl effect exploited by other turbulation devices, and iii) direct airflow away from the vehicle with an upward component at the rear, helping to kill downwash from the "tip vortex" effect and reduce induced drag. These would be most effective in this manner on the lower half of the vehicle on the sides. With vehicles like vans I think you'd be better off having them straight on the top half of the side and angled towards the centerline on the roof.

[dosco]

Diffuser:

In one of the links, it mentions for diffusers the use of a hollow space under the rear of the car to allow airflow to expand and slow down before exiting the underbody, to avoid further turbulences in the wake. If anything then the holes behind the rear bumper should not ne covered up. Maybe poke holes in the bumper at best. Maybe just put vertical separators to direct the flow and reduce turbulences in that gap?

I wouldn't spend too much time worrying about it. A diffuser is mostly applicable to race cars for generating downforce. They were developed to recover underbody-generated downforce that was lost when cars moved from "airfoil tunnel" to "flat-bottom" underbodies.

For a normally used road car, just putting something on the bottom of the car to "clean it up" will help reduce drag, thereby benifitting FE.

[RoadWarrior]

However, killing "upforce" would be beneficial or at least translating it from up and 15 degrees backwards, to up and 5 degrees forward.

[sonyhome]

RoadWarrior,

i'm trying to decypher your last message in my brain into something applicable to my CR-V... It's a 2-box style SUV, with garbage sticking out in the back (spare).

I still think I should want a clean airflow in the back.

From what you say, it'd be best to detach air like kammbacks (read like honda CRX) in the middle height of the vehicle, with a spoiler kicking air back up 15 degrees to clean the virtual boattail airflow.

For my CR-V, I'd want a clean break.

From what you say, on the roof I could use 3 rows of V-strips (or VGs since you seem to say they do the same... maybe at a larger dimension), say 4~6 inches away from the rear hatch door (farther than for cars) and angled in a V shape with the point towards the front of the car. Also I don't have to worry about the roof airflow spilling out to the sides, as i have a built-in raised roof rails that helps direct the flow.

If I do this, what I understand is the air would reattach and go around the rear hatch bend better, instead of detaching into turbulences.

However doesn't that mean I'd still need a spoiler to get the air kicked out clean of the roof to form a cleaner virtual boattail to eliminate turbulences?

Or is it better to have the air attach the whole length of the trunk to eliminate the dead air bubble in the back, at the risk of creating turbulences from the roof airflow crashing with the underbody airflow?

I noticed that many 2-box suvs have a rear spoiler with a hole. I assume some of the air is accelerated to attach to the rear window and some of it is kicked out and cleanly separated from the car. Maybe that helps fill the stagnant air pocket with some low pressure air turbulences.

A few questions:

- What's the reynold number for highway speeds (car/small SUV)? I suppose it's small.

- Wouldn't the best spot for the V-tape be above the windshield, to reattach cleanly the roof flow, and maybe influence how the flow happens above the hood?

- How about the leading edge of the hood itself?

... or maybe that would just make the flow more uniform and attached, aka more stable ride, but increase surface drag?

[RoadWarrior]

I guess I was unclear, the V should point rearwards when on a topside surface, pointing it forwards enhances the downwash vortices. If using angled strips on the lower half of the side, they should be low in the front high at the back.

Looking at the best photos I could find of an '02 CRV, it looks like you've got a semi-sharp lip on the back of the roof. Since the CRV is fairly tall for it's length, causing too much air to come down the back probably makes for a large increase in induced drag. So I'd say to avoid using VGs on top. You might try something to crispen the lip up a bit and allow cleaner separation. Just a strip of something protruding half an inch over the edge might do it. The downwash vortices look like they will be taken care of with the roof rails. Also around the rear corners it looks like things have been taken care of with the shapes in the bumper and ahead of the light clusters, the way the air trips and falls into those kind of scallop shapes should attach it further round the back. If anything the tire carrier helps a little, allowing small vortex cushions to form either side of it and round the back of the vehicle. The only place I see where you might consider turbulators on the back, is in the area of the rear side window. If that has a black border on it, you might consider just two strips of 1/4 high closed cell insulating strip, top to bottom there.

Now, any turbulation device can't magically pull down flow in an area that is already turbulent, so to have any effect on the windshield area, it would actually have to be on the windshield before the flow separates. If you've got a darkened windshield at the top and your wipers don't go up that far, then you might get two or three straight across strips there, spaced their own height apart. Likewise on the front, they would have to be on a surface that had attached flow. There doesn't look like anywhere where you could put any, apart from maybe curving a single strip around between the headlamp area and the signal light, making sure you don't block either. If you did an upper grille block then maybe just below the lip of the hood could take a set. Mirrors, standing out from the body somewhat, might be acting in a different class of reynolds number, and the 1/4 inch strip would be too big. If you could find something sticky and 1/8th of an inch high, then putting those in sets at 1/3 and 2/3 of the mirror height might marginally improve it, however, these might also tend to get the mirror dirtier with spray.

With newer vehicles there's not a heck of a lot to fix, the major comprimises are in the general body shape, and all the details to work around that have usually been taken care of.

Reynolds number calculation for a typical car @ 55mph...
http://galileo.phys.virginia.edu/classes/311/notes/fluids2/node4.html

Wikipedia Reynolds number...
http://en.wikipedia.org/wiki/Reynold%27s_number
NOTE, this mentions the flow transition at about 3x10^6 for a cylinder resulting in a DROP in Cd... with cars being in the region of 2.4 @ 55mph, transitioning to a lower drag regime by going faster is possible. It's probable that the laminar to turbulent transition for a boxy 1970s car didn't gain anything over 55mph, but with cleaner aero in later cars it's probable that immediately after transition, DRAG IS LOWER, then climbs again. Hence 55mph is probably not such a "saver" for anything with a Cd under .45 or so, that sheds turbulence with any semblance of aero design competence.

Pretty pictures...
http://www.featflow.de/album/catalog/asmo_low_2d/data.html

[dosco]

However, killing "upforce" would be beneficial or at least translating it from up and 15 degrees backwards, to up and 5 degrees forward.

Sure, but how much lift are we talking about? For most of us, the answer is "not much."

Out of curiousity, I've seen you mention the "15 degrees backwards" idea a few times. Where did you come up with that?

[RoadWarrior]

Actually it's the combination of the lift and drag vectors, due to free body diagrams of aeroplane aerodynamics being rigidly fixed in the up/down left/right frame of reference rather than in terms of direction of incident airflow. Soooo aerodynamics texts will talk of lift and drag separately there, when the actual physical force of lift might be pulling at an angle some degrees off from the vertical. So if a wing makes 100lb of lift going forwards, at a certain airspeed, if it can maintain that airspeed through thrust, flying straight up, it's still got 100lb of "lift" pulling at right angles to the plane of the wing, but in the way that aero textbooks describe it, it's now 100lb of "drag" because up is always up...

So, what I'm meaning is that in considering the average angle of airflow over the vehicle, and the angle at which it meets the "leading edge" and departs the "trailing edge" you might get "lift" pulling forwards if the angle is inclined forwards, or you might get "lift" pulling you backwards and thus becoming drag, what's known as induced drag if the angle is inclined rearwards.

By expanding air out from under the car, you effectively kick up the angle of the air departing the car inclining it forwards, thus altering the average angle of airflow over the body for the better.

Now, all this seems a bit odd, because we'd think that the average angle of airflow would be horizontal at all times. Well it would be if the car was flying in free air, 1000 ft above the earth, but it's not, the air is prevented from fully flowing underneath it and compensating. It all sorts itself out a hundred feet behind you, but by that time it's not having dynamic effects on the vehicle.

Best way to think of it, is imagine you've got a flat block moving through the air, average angle over that is horizontal, now a wedge moving point first, average angle of flow over that is going to be inclined forward from horizontal, now the wedge going backwards, inclined backwards from horizontal, Stick a flat bottom airfoil on the block, lift is straight up (for an airfoil designed to have lift at zero incidence) stick it on the wedge going forward, the lift force is inclined forward from the vertical, stick it on the edge going backwards, it's inclined backwards from the vertical.

So, if you took a chassis cab, and gave it a smooth sweeping kammback down to the rear bumper, you'd be approximating an airfoil at a high angle of attack, which in conventional terms would make little lift and lots of drag, but the actual force of lift is a component of the drag, because it's pulling the body backwards. This is analogous to our backwards wedge. However, if there were more expanding air coming out from under the back bumper then the air would not depart the body at such a downward angle, it would effectivly "kick up" the back edge of the airflow, such that it moved the lift force vector forwards WRT to the body and stopped the lift making induced drag. In other words you'd tilt the average airflow angle back to the horizontal.

Hope that's not as muddy.... it's clear in my mind, but it's hard to get around the "broken" method of describing forces on a body that's favored by aircraft aerodynamicists.

[sonyhome]

There's 3 generations of CRVs, you may have looked at the most recent, with better aero and no outside spare?

Mine doesn't have a lip on top.
http://membres.lycos.fr/sonyhome/Honda_CRV/Vanity/HPIM0016.jpg

Unless you're talking about this aftermarket air deflector
http://www.drivah.com/marketplace/images/oem_crv_base_thumb.jpg http://www.drivah.com/marketplace/images/oem_crv_rearair_thumb.jpg

which people seem to say is just decoration... I don't recall seeing anyone claiming gains from it.

You also mention a change of Cd. Could that explain that -I believe- I get same or better MPG at 75 than at 55?

So from reading you, I also have the feel I should not try to fill the void behind the rear wheel to flatten the underbody. What I did so far is tape holes under there to closed internal voids between metal and plastic parts like the rear bumper and side runners.

The airflow seems from what you say to detach way before a curvature point... So a VG shouls always be a couple of inches before it.
Note: I get lots of bugs and some pebbles impacts on the windshied. It must indicate rough air direction changes that could benefit cleaning?

[RoadWarrior]

Yes it doesn't look so sharp on your pic as it did on the one I found labelled as an '02. However, it's also not well curved enough to indicate that they intended the air to wrap round the back any. If the rear air deflectors aren't known to do anything, then sharpening that edge up sound like the better plan.

The void between the axle/tank and the rear bumper could be handled in either of two ways, either by boxing it in, or by taking advantage of the space there to make something of a diffuser. Boxing it in helps because it kills a lot of drag, making a diffuser helps more because it kills that drag and lifts air behind the car, killing some induced drag. The decider would be in how easy it is to mod the rear bumper, if the bottom half looks "useless" then making big holes in it, or cutting it away entirely is doable for the diffuser plan, but if it looks solider, has a metal box section behind it etc, that is part of the structural integrity or crash protection, then you should just settle for boxing it in.

Airflow detaches after about 11 degrees deflection, but there's also a minimum surface area it will attach to, at our typical speeds this seems to work out that it will stay attached to a curvature of around 3-4 inches radius, so curves of that radius it will tuck quite far round, curves of lesser radius it will detach messily at 11* deflection and greater. But... what you're doing in turbulation is pre-empting the "messy" formulation of large vortices, by forming small controlled vortices, that stay semi-attached and roll along the surface, rather than being large and shedding unsteadily making chaotic flow above them or stagnating in place and causing airflow deviation as bad as if they were a chunk of protruding bodywork. Behind the vehicle instead of shedding smoothly they may also suck back at the rear edge causing vacuum drag. So, what you're doing is kind of making loads of little energetic "sticky balls" instead, that because they're rolling can sort of disobey the 11* rule and manage to follow a sharper curve... however, you have to get them attached to the surface before it deviates, otherwise they're just jumping off into space... shoving an air stream upwards that will have a large vortex form behind it that does the same as what you were trying to avoid in the first place. So, you need to make these sticky balls just before the surface deviates, with enough surface left for them to "cling" to to roll themselves round the edge more tightly, to help the bulk flow take that path instead of getting pushed higher by the larger vortices.

The change of Cd thing due to flow transition, it means that your aerodynamic drag going up with the square of speed type graph is still starting to head sharply upwards at 55, but when the flow transitions if it was plotted as actual drag force against speed, you'd see a sudden dip in the graph where the flow transitioned from one regime to the other, and then it would take off straight up again. I think that probably older cars with very little aerodynamic refinement maybe get nothing at all when the flow transitions, and that newer cars probably find that point quite high. So it probably doesn't change the "in general, stay at 55" advice, since some cars might need to go to illegal speeds to find the dip, and draggier cars might not have a dip to find.

[sonyhome]

RoadWarrior, thanks.

It looks like my avenues are pretty clear: Test a lip and fill the undercarriage somehow (there is metal structure behind the plastic bumper).

For the lip, should I try to just angle it down or flat, just following the roofline
http://i9.ebayimg.com/04/i/000/93/14/048c_1.JPG
http://z.about.com/d/suvs/1/0/T/1/-/-/24_VueHybrid.jpg

Or just stick a wedge to kick the airflow upwards a little
http://thepassionatepursuit.com/images/weblog/07-09-01-lexus-es-350-spoiler.jpg
http://www.cardata.com/spoilers/images/DAR_Spoilers/2008_Chevy_Tahoe_FG-120.jpg

And why VGs would not achieve the same?

I'd expect the vortices to provide support for the airflow above it, and shape a virtual spoiler, creating a clean cut of the air away from the body instead of a messy decompression near the curve of the rear hatchdoor when the air cannot follow the curve > 11%.

http://www.airtab.com/Images/gallery/CRV/DSC05730.JPG

[RoadWarrior]

VGs on top risk turning the airflow through too great an angle and creating more induced drag (due to shape and lift) than the base (Pressure/vacuum) drag they reduce. I'd guesstimate that you want the airflow coming off the top of there at no more than about a 20 degree angle from the horizontal. Therefore picture number 1 would be my choice, maybe with the angle tweaked downward just a touch, or made adjustable so you can find the sweet spot (Shim washers under the fittings???)

The other way to do it, would be to bring air completely down the back screen then seperate it from the body at about halfway down the back. For this you might need a large "whale tail" hatch spoiler at the bottom of the window, or to make a "trunkette" type shape. However, the angle the roof lip turns and the angle of the back screen don't look particularly helpful for this, VGs or turbulators might manage about a 45 degree deflection, but to "catch" that and turn it horizontal, you'd have to stick something way out of the back there, like a 2 foot wide plank strapped on top of your tire carrier. So then you'd be left with the rear screen air deflector method, but again need to turn the air. However, since this should be bringing the air round a sharper turn than the VGs will, you don't need a large surface to turn it horizontal, maybe just a 3 or 4 inch "shelf" type spoiler at the base of the rear screen.

A third option might be to put a large air dam on the front, and use the VGs on top anyway. This may also get gains in conjunction with 1 on it's own or the deflector on it's own. The reason would be that the airdam will lower the point that the airstream separates at the leading edge of the vehicle, thus effectively "tilting the airfoil" forwards, such that greater departure angles of air from the rear of the vehicle will not cause extra induced drag, but you get the gain from the reduced base drag. Adding the airdam should improve things anyway. but say you got 1 mpg from the airdam on it's own and 2 mpg from the "better" options on the back on their own, and 0 mpg for the the others. What should happen is that they are complementary, such that you might see 5mpg with airdam and better options, and 2 or 3 mpg with the airdam and the other options.

[sonyhome]

I don't plan to do monster aero mods to the 'V (I'm not hateful LOL!) so I'll start by the spoiler.

- I'll probably make a proto with an L corner to shove it in the door, and use spacers to manage angles. SInce it has to be 20ish degrees downwards, it's going to be interresting to keep it in place there.

- As for the air dam, I'm not sure it would help: The 'V is high up so that's quite a lot of frontal surface added... I'll try to make sketches to see if I understand what you say about the dam on the final airflow.

[sonyhome]

Ok, with some ABS plastic, that was a quick prototype... I did a quick drive... I seem to be at 30MPG at 72MPH with cruise control over a short drive, that's pretty good, the # looks higher rather than lower than normal (no A-B-A test).EDIT Did another 2x 25miles trip and got 27MPG this time at 73MPH cruise ctrl... That's not very good.

I put some pics here (http://tinyurl.com/6xg6ez) of the spoiler and shots of the geometry/shape of the car.

Aero

Here's 3 figures hand drawn. I suspect I probably did not quite understand all the car aero so maybe my description is wrong and so plz correct if you see stupidities.

1- Ideal shape, the car pushes the air out evenly and it regroups behind cleanly. Symmetry avoids lift drag.

2- CR-V SUV shape. The desing creates lift, the air travels more on top than bottom. Increasing lift also increases drag (?). Furthermore sharp tailgate angle forces undercarriage and roof air to rejoin at sharp angle, colliding while decompressing, causing turbulence drag.

3- Spoiler extends roof, detaches air cleanly forming a virtual boat tail with less angle so less lift and less associated drag. Undercarriage and roof airflows meet without colliding as much, less turbulent flow means less energy spent in eddies, less drag. This does create an air hole depression behind the SUV.

http://tinyurl.com/4fkndh

Airdam
If I were to make one, I'd increase the surface area by 15% or more:
Ground to bumper 15", ground to bottom of car 9", width of car 67" (tires 7" wide each), height 60". That's about 3100sqin frontal surface vs3600sqin with a full air dam.

Spoiler
I made the spoiler to follow the roofline, slopes down a little, and the ABS is flexible so it should sump by 1" or so with wind pressure. Is it still riding too high?
http://tinyurl.com/59sbw7
http://tinyurl.com/6mv92m

underbody diffuser
Behind the rear left wheel I've got a air hole, that should be fixed. Notice the tape job I did to seal holes around the bumper. Tape joins metal parts to plastic parts. Near the mud flap I didn't finish the tape job.
http://tinyurl.com/624mag
Incidentaly a friend took me to the BMW dealer and I looked at SUV undertrays. It's all covered & flat, surfaces just have some ridges parallele to airflow.

[dosco]

Actually it's the combination of the lift and drag vectors, due to free body diagrams of aeroplane aerodynamics being rigidly fixed in the up/down left/right frame of reference rather than in terms of direction of incident airflow.

What text is that? I studied this stuff in school and lift and drag are defined by the direction of the freestream velocity. Lift is always perpendicular to the freestream and drag is always in the direction of the freestream. What doesn't change is the direction of gravity and the weight vector.

So if a wing makes 100lb of lift going forwards, at a certain airspeed, if it can maintain that airspeed through thrust, flying straight up, it's still got 100lb of "lift" pulling at right angles to the plane of the wing, but in the way that aero textbooks describe it, it's now 100lb of "drag" because up is always up...

The lift you describe is still lift, but what matters is resolving it when compared to gravity.

The amount of lift generated is a function of the angle of attack, the angle between the zero lift line of the body and the freestram velocity.

Now, all this seems a bit odd, because we'd think that the average angle of airflow would be horizontal at all times.

Here is where I agree. It would be nice to know the zero lift line of a car. Perhaps some manufacturers or grad students might have that info.

the air is prevented from fully flowing underneath it and compensating.

The air is flowing under the car, it is just a mess. The other complicating factor is the boundary layer under the car. Recall that the car is moving over the pavement and through the air. The air is stationary at the ground and the car is dragging the air as it moves. You may notice this when driving in the rain and looking at the car ahead of you. You may also see this during auto races in the wet. This is an issue with wind tunnel testing, the most accurate results are generated in tunnels with floors that move at the same speed as the freestream.

Not sure what you mean by "compensating."