Go Back   EcoModder Forum > EcoModding > Aerodynamics
Register Now
 Register Now
 

Reply  Post New Thread
 
Submit Tools LinkBack Thread Tools
Old 05-03-2020, 04:21 AM   #1 (permalink)
Master EcoModder
 
Join Date: Nov 2017
Location: Australia
Posts: 815
Thanks: 54
Thanked 755 Times in 475 Posts
Aerodynamic testing

I think that on-road testing beats any theorizing, hands-down. With car aero there are just so many inter-relating variables, that predicting what will happen in any given aerodynamic situation is very hard. (It's for this reason that I think many 'rules of thumb' are useless.) Testing shows you what is really happening, not just theory or speculation.



I will skip past tuft testing (that I think is fantastic) because most people already know about it. But tuft testing is still one of the best aero testing techniques, ever.

But what about the others?



I think that surface panel pressure testing is critical. You'll need a Magnehelic gauge of suitable range (cheap on eBay), some tube, a surface pressure sensor (that you can make yourself) and a few other bits and pieces. With this equipment, you can then measure average pressures in the wake, and on any panel you wish to - both above and below the car. Suddenly you can see what the airflow is doing in terms of pressures - and it's pressures that create drag and thrust, lift and downforce. We've all seen the CFD images showing pressure distributions on a computer-modeled car, but you can easily measure them yourself on the road. Cost? About US$100.



Lift and downforce are also easily measured. It's admittedly not as easy as measuring panel pressures, but directly measuring lift and downforce takes into account all the vectors (directions) that those panel pressures have. Any car having overall lift (basically all cars) has what is called 'induced drag' ie drag resulting from the fact the car body is developing those vertical forces. So reducing lift is very important if you are after low drag. Cost to measure lift? About US$75 - but some fiddly work in fitting sensors.



Local airspeeds can also be easily measured. That is, see where airflow speeds are above or below the 'free-stream'. (Free-stream equals forward speed of car on a day without wind.) The easiest way is to use the trusty Magnehelic gauge and a miniature pitot tube, as sold cheaply for use on model aircraft. The pitot tube can also be used to accurate detect airflow angles, eg off the trailing edge of a roof on a sedan. Cost? About US$10 if you already have the gauge.



Drag is the really hard one to measure. I think if you have sufficient time and patience, nothing beats mileage - but it needs to be over many thousands of miles for cast-iron results. And, not many people have the patience to make only one aero change over such testing regimes. In the absence of that, I think throttle-stop top speed testing looks very promising. At least it is consistent, quick and the first testing matches published data (eg windows up / windows down).

  Reply With Quote
The Following User Says Thank You to JulianEdgar For This Useful Post:
Gasoline Fumes (05-03-2020)
Alt Today
Popular topics

Other popular topics in this forum...

   
Old 05-03-2020, 12:58 PM   #2 (permalink)
Tyrant-at-large
 
Vman455's Avatar
 
Join Date: Feb 2012
Location: Champaign, IL
Posts: 1,648

Little Red - retired - '05 Honda Civic EX
90 day: 49.03 mpg (US)

Pope Pious the Prius - '13 Toyota Prius Two
Team Toyota
SUV
90 day: 59.03 mpg (US)
Thanks: 176
Thanked 1,438 Times in 757 Posts
This is great! I already have a Magnahelic that I've used to test pressure above and below the hood, but I think this summer I'll play around with it some more.

Quote:
But tuft testing is still one of the best aero testing techniques, ever.
I would be careful with tuft testing, however--it doesn't always tell you what the lowest-drag solution is. I was reading some papers on Hucho's website and came across this one on fastback shapes:

"This pseudo-tail [used on the Tatra 87, VW Beetle, Volvo PV544, etc.] wasn't that fast [i.e. low drag]! On the contrary, it was characterized by a particularly high air resistance. And that met with incredulous amazement. Because if you looked at the flow around these cars, be it with woolen threads on the street or with the smoke probe in the wind tunnel, it looked as if the flow remained close to the back. How should there be a high resistance? This secret could only be revealed much later, during the development of the VW Golf I." (emphasis and notes added)

As he goes on to explain, the flow remained attached because a pair of inward-rotating vortices "pulled" the flow down to the surface but at the same time, the low pressure of the vortices induced lift and overall drag was increased--despite the tufts showing attached flow.

Tuft testing might be the best technique in conjunction with something like pressure testing, then, just to make sure one is getting a complete picture.
__________________

  Reply With Quote
Old 05-03-2020, 01:37 PM   #3 (permalink)
Master EcoModder
 
freebeard's Avatar
 
Join Date: Aug 2012
Location: northwest of normal
Posts: 16,708
Thanks: 4,427
Thanked 5,615 Times in 4,476 Posts
Quote:
"This pseudo-tail [used on the Tatra 87, VW Beetle, Volvo PV544, etc.] wasn't that fast [i.e. low drag]! On the contrary, it was characterized by a particularly high air resistance. And that met with incredulous amazement.
Are you sure? Compare the Beetle and the Tatra. The Tatra has elephant ear-sized scoops on the rear quarters to capture engine-cooling air. The Beetle has surface vents below the rear window.

Somebody knew about those vortexii.
__________________
.

Change my, you know, the thing.

_________________
  Reply With Quote
Old 05-03-2020, 05:51 PM   #4 (permalink)
Master EcoModder
 
Join Date: Nov 2017
Location: Australia
Posts: 815
Thanks: 54
Thanked 755 Times in 475 Posts
Quote:
Originally Posted by Vman455 View Post
This is great! I already have a Magnahelic that I've used to test pressure above and below the hood, but I think this summer I'll play around with it some more.



I would be careful with tuft testing, however--it doesn't always tell you what the lowest-drag solution is. I was reading some papers on Hucho's website and came across this one on fastback shapes:

"This pseudo-tail [used on the Tatra 87, VW Beetle, Volvo PV544, etc.] wasn't that fast [i.e. low drag]! On the contrary, it was characterized by a particularly high air resistance. And that met with incredulous amazement. Because if you looked at the flow around these cars, be it with woolen threads on the street or with the smoke probe in the wind tunnel, it looked as if the flow remained close to the back. How should there be a high resistance? This secret could only be revealed much later, during the development of the VW Golf I." (emphasis and notes added)

As he goes on to explain, the flow remained attached because a pair of inward-rotating vortices "pulled" the flow down to the surface but at the same time, the low pressure of the vortices induced lift and overall drag was increased--despite the tufts showing attached flow.

Tuft testing might be the best technique in conjunction with something like pressure testing, then, just to make sure one is getting a complete picture.
None of that casts a negative light on tuft testing - you just need to know enough to make a judgement on what you are seeing. For those who have my Veloce aero book, attached flow on the rear part of the car being associated with higher drag and lift is shown on page 65 (new Beetle), page 134 (rear roof extension angle on the Roomster) and page 195 (Porsche with ducktail).
  Reply With Quote
Old 05-03-2020, 07:38 PM   #5 (permalink)
Ecomodding amateur
 
M_a_t_t's Avatar
 
Join Date: Nov 2015
Location: Indiana
Posts: 413

The Van - '95 Chevy Astro Cl
Team Chevy
90 day: 16.39 mpg (US)

The new bike - '17 Kawasaki Versys X 300 abs
Motorcycle
90 day: 64.78 mpg (US)

The Mercury - '95 Mercury Tracer Trio
Team Ford
90 day: 33.66 mpg (US)
Thanks: 61
Thanked 140 Times in 103 Posts
Most of these methods are new to me, hopefully I have time to do some reading later. What method (or combination) would you suggest aeromods be tested with? Especially the ones that take awhile to make, such as boattails and the like. Currently the most common method for checking is based on coastdown testing. This provides some actual numbers for comparison, even if they might be skewed, to see if a mod works.

If tuft testing is not indicative of lower Cd, and coastdown testing not reliable* both of which are the least cost to do some testing what would be the next best bang for the buck?

*In one of your other threads you suggested you never had any luck with coast down testing. Could you expand on why you feel that way? I was just assuming unreliability in the data.
Quote:
Originally Posted by JulianEdgar View Post
Coastdowns should never be done coming to a standstill, in my opinion. Why? Because you are then spending too long at speeds where aero drag is basically not relevant - but it potentially adds to the error.

But I have never had any luck with coast-downs - of any sort. And Rob Palin (ex Tesla aerodynamicist) was scathing about them in a recent email to me.
__________________
My herd includes:
1973 Fiat 124 Special
1975 Honda Civic CVCC 4spd
1981 Kawasaki KZ750E
1981 Kawasaki KZ650 CSR
1983 Kawasaki KZ1100-A3
1986 Nissan 300zx Turbo 5 spd
1995 Chevy Astro RWD
1995 Mercury Tracer
2004 Chevy Astro AWD
2017 Kawasaki VersysX 300

https://www.youtube.com/channel/UC6s...LulDUQ8HMj5VKA
  Reply With Quote
Old 05-03-2020, 08:06 PM   #6 (permalink)
Master EcoModder
 
Join Date: Nov 2017
Location: Australia
Posts: 815
Thanks: 54
Thanked 755 Times in 475 Posts
Quote:
Originally Posted by M_a_t_t View Post
Most of these methods are new to me, hopefully I have time to do some reading later. What method (or combination) would you suggest aeromods be tested with? Especially the ones that take awhile to make, such as boattails and the like. Currently the most common method for checking is based on coastdown testing. This provides some actual numbers for comparison, even if they might be skewed, to see if a mod works.

If tuft testing is not indicative of lower Cd, and coastdown testing not reliable* both of which are the least cost to do some testing what would be the next best bang for the buck?

*In one of your other threads you suggested you never had any luck with coast down testing. Could you expand on why you feel that way? I was just assuming unreliability in the data.
In my experience of coastdown testing, the data tells you nothing. For example, I couldn't measure a statistical difference in windows up/down testing. (That testing was done over many runs.) I did this a few years ago when I was preparing material for my book. That followed coastdown testing I did in the late 1980s when I found that the results were all over the place - such a contrast to (say) 0-60 mph testing that was consistent within tenths of a second.

When writing the book, I found two SAE papers on coastdown testing and looked carefully at what they had done to get good results. It was all way beyond the technology available to an amateur eg specially calibrated anemometers on probes out the front of the car constantly logging data on things like yaw angles of the oncoming air, and sophisticated computer modeling of driveline inertia, etc, etc.

I have seen people using coastdown testing to purportedly find tiny changes in aero drag and, frankly, I don't believe it. If you were doing coastdowns from (say) 100 mph down to 80 mph, it may start developing useful data - but otherwise, I can't see it. If you are attempting coastdowns, you should at minimum test the results from something like windows up / windows down and expect to see a calculated 10-12 per cent (or similar) change in drag.

All testing techniques - tufts, pressures, downforce/lift, pitot tube, mileage, throttle stop - can give you extremely useful information. But you need to:
  • ignore changes that are tiny (because they're probably statistically inconsequential)
  • do the testing with rigorous methodology
  • understand the significance of what you are actually seeing (tuft testing a good example)

Frankly, I believe that most of the aerodynamic modifications that I see people undertaking do very little, especially those that attempt to decrease drag. However, I can see some major modifications (full length belly pans / undertrays, large boat tail extensions) as giving results above noise.

In my aerodynamic modification, I first do a mock-up out of cheap and easy materials, and if I cannot get a measurable positive change in behavior, I take it off. If I get a positive change, I make a proper one. Never invest time and energy in the 'proper' version when you can't be sure if it works. Cardboard, coroplast and tape first.

For most accuracy in measuring changes in drag, long distance (thousands of miles) mileage tests are best. I am hoping that the throttle-stop measurement technique I have developed will be next best.
  Reply With Quote
The Following User Says Thank You to JulianEdgar For This Useful Post:
MeteorGray (05-05-2020)
Old 05-03-2020, 09:07 PM   #7 (permalink)
Master EcoModder
 
Join Date: Nov 2017
Location: Australia
Posts: 815
Thanks: 54
Thanked 755 Times in 475 Posts
Quote:
Originally Posted by M_a_t_t View Post
Most of these methods are new to me, hopefully I have time to do some reading later.
I did have this in the pipeline but I brought it forward as it answers some of your questions.

  Reply With Quote
The Following User Says Thank You to JulianEdgar For This Useful Post:
M_a_t_t (05-04-2020)
Old 05-04-2020, 04:34 PM   #8 (permalink)
Ecomodding amateur
 
M_a_t_t's Avatar
 
Join Date: Nov 2015
Location: Indiana
Posts: 413

The Van - '95 Chevy Astro Cl
Team Chevy
90 day: 16.39 mpg (US)

The new bike - '17 Kawasaki Versys X 300 abs
Motorcycle
90 day: 64.78 mpg (US)

The Mercury - '95 Mercury Tracer Trio
Team Ford
90 day: 33.66 mpg (US)
Thanks: 61
Thanked 140 Times in 103 Posts
Quote:
Originally Posted by JulianEdgar View Post
When writing the book, I found two SAE papers on coastdown testing and looked carefully at what they had done to get good results. It was all way beyond the technology available to an amateur eg specially calibrated anemometers on probes out the front of the car constantly logging data on things like yaw angles of the oncoming air, and sophisticated computer modeling of driveline inertia, etc, etc.

Frankly, I believe that most of the aerodynamic modifications that I see people undertaking do very little, especially those that attempt to decrease drag. However, I can see some major modifications (full length belly pans / undertrays, large boat tail extensions) as giving results above noise.
When replying I wrote a paragraph filled with questions. I think it would be easier to list them.

1.You said "most of the aerodynamic modifications that I see people undertaking do very little, especially those that attempt to decrease drag." Are you suggesting the mods themselves are ineffective or that our current testing system is not good enough to show small improvements and therefore unable to tell if they are actually helping?

2. Do you think these smaller mods would show up using the throttle stop method and could you give some insight into when you think mods start getting too small to make a measurable difference?

3. Would a higher test speed change the threshold of mods too small to test? E.g your experiments have been around 105 kph (65 mph) if we tested at ~125 kph (78 mph) would smaller changes be more evident?

Can you link those papers? I have been trying to get a small collection of reading to do this summer on aerodynamics (when I have the extra cash to purchase some of the texts anyway). If it seems I am avoiding the other methods of testing it is because I haven't completely grasped how to interpret results from them. This is not a direct request for you, I just haven't done my part to sit down and read up on anything yet to be able to ask informed questions about them.

I do have 3 questions for the magnehelic gauge. The "ideal" car would have very little pressure change over the surface of the body correct? So with the magnehelic gauge you check for high and low pressure zones and try to reduce the size of these zones. I think thats how it works right? Is there a limit to effective hose length? For example, if I have the gauge towards the front of my van and ran a hose (~15-20 ft or 5 meters) to the top edge of the rear would the hose act as a restriction to the changes or is that effect really tiny and therefore wont matter? The hose running along the body line wouldn't be enough to change the results of the test correct? Like if I had a hole in the roof above the driver seat and ran the hose down the length of the roof rather than just poking out the top.
__________________
My herd includes:
1973 Fiat 124 Special
1975 Honda Civic CVCC 4spd
1981 Kawasaki KZ750E
1981 Kawasaki KZ650 CSR
1983 Kawasaki KZ1100-A3
1986 Nissan 300zx Turbo 5 spd
1995 Chevy Astro RWD
1995 Mercury Tracer
2004 Chevy Astro AWD
2017 Kawasaki VersysX 300

https://www.youtube.com/channel/UC6s...LulDUQ8HMj5VKA

Last edited by M_a_t_t; 05-04-2020 at 04:50 PM..
  Reply With Quote
Old 05-04-2020, 06:05 PM   #9 (permalink)
Master EcoModder
 
Join Date: Nov 2017
Location: Australia
Posts: 815
Thanks: 54
Thanked 755 Times in 475 Posts
Quote:
Originally Posted by M_a_t_t View Post
When replying I wrote a paragraph filled with questions. I think it would be easier to list them.

1.You said "most of the aerodynamic modifications that I see people undertaking do very little, especially those that attempt to decrease drag." Are you suggesting the mods themselves are ineffective or that our current testing system is not good enough to show small improvements and therefore unable to tell if they are actually helping?

2. Do you think these smaller mods would show up using the throttle stop method and could you give some insight into when you think mods start getting too small to make a measurable difference?

3. Would a higher test speed change the threshold of mods too small to test? E.g your experiments have been around 105 kph (65 mph) if we tested at ~125 kph (78 mph) would smaller changes be more evident?

4. Can you link those papers? I have been trying to get a small collection of reading to do this summer on aerodynamics (when I have the extra cash to purchase some of the texts anyway). If it seems I am avoiding the other methods of testing it is because I haven't completely grasped how to interpret results from them. This is not a direct request for you, I just haven't done my part to sit down and read up on anything yet to be able to ask informed questions about them.

5. I do have 3 questions for the magnehelic gauge. The "ideal" car would have very little pressure change over the surface of the body correct? So with the magnehelic gauge you check for high and low pressure zones and try to reduce the size of these zones. I think thats how it works right? Is there a limit to effective hose length? For example, if I have the gauge towards the front of my van and ran a hose (~15-20 ft or 5 meters) to the top edge of the rear would the hose act as a restriction to the changes or is that effect really tiny and therefore wont matter? The hose running along the body line wouldn't be enough to change the results of the test correct? Like if I had a hole in the roof above the driver seat and ran the hose down the length of the roof rather than just poking out the top.
1. Probably a bit of both. For example, moving from a good standard wheel to one with a full cover I doubt does anything at all, and if it did in fact do anything, it would be so small as to be next to unmeasurable. I also see lots of modifications that I am sure do nothing. (Unfortunately what I have found with discussion groups is that if you attempt to call these out, people get very upset. Notably, there is never any evidence that these mods work. I've stopped looking through build threads because I find it so frustrating.) To keep it simple, I think the physical size of the modification usually correlates pretty well with its likely impact - positive or negative.

2. I am still exploring the throttle stop method but I think that mods that makes less than a calculated 1-2 per cent difference in drag won't be measurable. My latest test session showed that pretty well I think. Taking the rear spats off the Honda didn't give any real measured change (only within margin of error), whereas opening the windows is always measurable, with a greater difference occurring in gusty crosswinds. To put that another way, if I can measure a change that is anything like as great as that caused by opening the windows, I know I am measuring something real. (That was the case with the Edgarwits, so they are very exciting to me.)

I have been measuring coarse body aero pressures since 2000, body panel pressures since 2018 and lift/downforce also since 2018. I think I have a good feel for these measurements in that I can quickly see if things are going in the right direction or not - as the case may be! What I have found with these measurements, and it looks the same with throttle stop method, is that it's best to make large changes and see what happens. If they appear to work, you then finesse them. Fiddling around with initially tiny changes often means you spend a lot of time and get nowhere.

3. Yes, as I have said in the videos, test at the highest legal speed. I have done some testing at 140 km/h and I did notice that the throttle-stop top speed wasn't held as consistently as at lower speeds - there was say a 1-2 km/h variation around a mean. But the mean was still obvious to the eye. Of course, from an aero perspective, top speed without a throttle stop is best of all, but then there is a much greater likelihood of the engine's power output varying as the engine will be developing full power eg on a turbo car intake air temperature rising. (I did do just one top speed run, and noticed that the speed at the end of my straight (still rising, but I don't have a longer straight) was 190 km/h, down 10 over what I achieved with the rear wing set to provide less downforce. That's when I started thinking about the throttle stop method... so I wouldn't be arrested.)

4. For coastdown testing: SAE 950626 - ABCD ľAn Improved Coast Down Test and Analysis Method and SAE 940420 - A Detailed Drag Study Using the Coastdown Method. I think after you read these papers you will probably discount doing it without their equipment and computer modeling. Measuring of body pressures with the surface measurement sensor, and measuring downforce, are as far as I know covered only in my book - I've not read coverage anywhere else on how an amateur can do these.

5. Actually you want to maximise low pressures on the front of the car (creating thrust) and decrease low pressures on the rear of the car (reducing drag). But even that isn't accurate, because you need to take into account the vector (direction) those forces are acting in. (Page 15 of my book.) But in fact you don't even use a Magnehelic gauge like that much anyway! There are basically two uses of the Magnehelic gauge:

(a) Fairly coarse measurements eg finding the right place for hood vents, measuring pressure drop across a radiator core (eg to see if an undertray is reducing radiator flow), making cooling ducts that work, etc. These measurements are easy and don't require a reference pressure reservoir or any sensor.

(b) Measuring very small pressures eg on body panels above and below the car. These require the use of a surface pressure puck (I made my own) and a reference pressure reservoir. You need to be quick and careful when doing these measurements. You can use these when assessing if an undertray is working at developing a low pressure (reducing lift), see if a rear spoiler is increasing pressures on rear panels, see if average wake pressures are being changed, etc.

It doesn't matter in terms of your pressure readings how long the hose is. As long as it has no leaks, it can be (say) 5 metres long without issues. You may find the reading is more damped (changes more slowly) - that's all. When doing coarse readings, the presence of the hose doesn't make any difference. In fine readings, I always use a very small diameter hose near the surface pressure measuring puck, and try to position it directly upstream or downstream of the puck so that the blockage factor is kept as low as possible.
  Reply With Quote
Old 05-05-2020, 10:06 PM   #10 (permalink)
Ecomodding amateur
 
M_a_t_t's Avatar
 
Join Date: Nov 2015
Location: Indiana
Posts: 413

The Van - '95 Chevy Astro Cl
Team Chevy
90 day: 16.39 mpg (US)

The new bike - '17 Kawasaki Versys X 300 abs
Motorcycle
90 day: 64.78 mpg (US)

The Mercury - '95 Mercury Tracer Trio
Team Ford
90 day: 33.66 mpg (US)
Thanks: 61
Thanked 140 Times in 103 Posts
Quote:
Originally Posted by JulianEdgar View Post
1. Probably a bit of both. For example, moving from a good standard wheel to one with a full cover I doubt does anything at all, and if it did in fact do anything, it would be so small as to be next to unmeasurable. I also see lots of modifications that I am sure do nothing. (Unfortunately what I have found with discussion groups is that if you attempt to call these out, people get very upset. Notably, there is never any evidence that these mods work. I've stopped looking through build threads because I find it so frustrating.) To keep it simple, I think the physical size of the modification usually correlates pretty well with its likely impact - positive or negative.
In one of your videos (Porsche on low drag wheel design) you say "Ventilation drag is...primarily through air passing through the openings in the wheels. Ventilation drag alone contributes up to 8% of total vehicle drag. Any difference in ventilation drag is going to be significant in terms of overall drag of the vehicle." (this isn't quite word for word, but very close) @~1:10

Then: ventilation drag is 50 percent tires and 50 percent wheels. This would suggest that 4% can be saved by putting full wheel covers on. This is double checked in the wheel graph you show later (@2:50) by a fully covered rim giving a -.015 Cd. On a car whose Cd is .30 this is a 5% decrease in drag.

In the qoute above you say you don't think it makes much of a difference. It sounds you are contradicting yourself. Maybe what you mean by "much of a difference" is different than my version? I know if Porsche went through the effort to find these results and I would assume that Honda did for the insight then full wheel covers would be worth their effort on our own cars (rather than buying a new set of rims).

I would like to thank you for the compliment in wheels (again in the video) as I've put insight wheels on my mercury. What tires are you running on yours? Some have said some of the newer FE tires have worked better than the original spec potenza re92's.

On longitudinal drag: If you put wheel skirts (or spats) on the outside and enclosed the wheel well as best as you could while still allowing the suspension to articulate do you think this would mitigate most of the longitudinal drag and ventilation drag? Possibly worth 6 or 7% in the case you were making in the video?

Edit: I just watched your video on "deceptive rules of thumb in car aerodynamics" and could see a point to be made that not every car will see these results. None of the evidence I've seen has said you would get worse Cd by adding them though so they should always show some benefit. Correct?

__________________
My herd includes:
1973 Fiat 124 Special
1975 Honda Civic CVCC 4spd
1981 Kawasaki KZ750E
1981 Kawasaki KZ650 CSR
1983 Kawasaki KZ1100-A3
1986 Nissan 300zx Turbo 5 spd
1995 Chevy Astro RWD
1995 Mercury Tracer
2004 Chevy Astro AWD
2017 Kawasaki VersysX 300

https://www.youtube.com/channel/UC6s...LulDUQ8HMj5VKA

Last edited by M_a_t_t; 05-05-2020 at 10:33 PM..
  Reply With Quote
Reply  Post New Thread


Thread Tools




Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, vBulletin Solutions Inc.
Content Relevant URLs by vBSEO 3.5.2
All content copyright EcoModder.com