Bicycle fairing design in wind tunnel 1

[GeezerGeek]

After searching the net for help designing recumbent bicycle fairings, I found good technical information about fully faired bikes used to break land speed records but there was very little technical information about partial fairings (windscreens). It seems that most people who post have designed their partial fairings by trial and error and test them by timing themselves on long rides – very imprecise.

A good partial fairing should have a low Cd from the front, minimal side area, be light weight, and get lift from side winds. Some of these are conflicting requirements so I want to make a design decision after testing wind loads from different angles. My goal is to develop an easy and inexpensive tool that bicycling hobbyists can use to test different designs. To do so, I built a wind tunnel out of cardboard boxes and a window fan.

The Reynolds number for the bike should be in the 300K range and my models with this wind tunnel have a Reynolds number of about 50K. Spherical designs will have poor correlation because of the different Reynolds numbers but I am hoping for the best with a shape similar to a thin slice off the side of a cylinder.

To test the wind tunnel, I have tested different sized disks. They should have a Cd of 1.17 but I am getting numbers from 1.1 to 2.0 depending on size. The larger the disk the larger the tested Cd. The test section of the wind tunnel is 12 inches X 15 inches ( 30.5 cm X by 38 cm) and the largest disk has an area 10% that of the wind tunnel or 5 inches (12.7 cm).

Are the errors in Cd because of boundary layer effects?
If so, how far from a wall does the disk have to be or conversely, how big does the wind tunnel have to be to reduce boundary layer errors to less than 20%?
Does anybody know of any good info on the web for designing bicycle wind screens for reumbents besides those designed by trial and error on full scale models?

 

[IRstuff]

Are these disks flat on the back?

If so, wouldn't you expect some drag effects that are proportional to circumference?

Even the best frontal aeroshapes are significantly enhanced by using a boattail shape in the back

TTFN

 

[GeezerGeek]

Yes, the disks are flat on the back. I am not going to use them for a fairing just to test the wind tunnel and sensors. From what I read, the Cd of a flat disk is nearly constant from Reynolds numbers of 10^3 to 10^6 and they are easy to make.

 

[btrueblood]

GG,

Yes, the tunnel has an effect on your test model, and vice versa. The simplest interaction to account for is the blockage effect - if your model blocks 10% of the flow area, the true mean flow speed near the model is 10% higher than you would otherwise calculate. There are lift and moment coefficient interactions you can account for by theory as well. Somebody on this site should be able to give you a good recent textbook for experimental aerodynamics and wind tunnel testing techniques (or, if not, I'll dig back through my notes for one).

 

[Sparweb]

Do you have something like a pitot tube to check the air speed in the tunnel? Placing it in different positions relative to the walls will show any gradients across the test section. Also, it would be intresting to put it in plane with the disks to directly measure the effect of disk size.
Did you build in a "settling chamber"? Is the fan blowing or sucking through the tunnel?


Steven Fahey, CET
"Simplicate, and add more lightness" - Bill Stout

 

[GeezerGeek]

Here is some more information to answer the above questions and to answer questions which have not yet been asked.

I did not build a settling chamber. To minimize jets, I placed the wind tunnel on the upstream side of the fan and built a star straightener at the inlet of the fan to prevent a vortex from backing upstream. On the inlet of the tunnel I have a large bell, then a converging section with a 7 degree slope that reduces the area by a factor of 3, and then I have one equivalent diameter of straight duct. To measure flows, I have a pitot tube and a 0 to 0.25 inch magnehelic.

To qualify the empty tunnel, I checked 9 equally spaced points in a 3 X 3 grid. The velocity pressures varied from 0.120 inches to 0.125 inches. There is a lot of damping in the magnehelic and so I am only measuring average VPs. While running the tests with models, I left the pitot tube in a corner and found that the velocity pressure did not change a measurable amount. Remember, this tunnel uses a window fan that is basically a thrust fan. I suspect that I am running very close to the peak of the fan curve. By adding the extra resistance, the fan produces less air and the velocity ends up to be about the same. To verify this hypothesis, I added a small additional resistance across the fan and it started to surge.

To measure the force applied to the models I am using a modified joystick. I took out the the springs because of the nonlinearity of the joystick's responses. Next, I replaced the thick handle with a small diameter threaded rod and attached 4 apposing springs to center the threaded rod. On the inside I feed 14v DC across the two pots to get a high level voltage signal. This provides a two plane load cell with a sensitivity of about 6.25 grams per volt. The joystick is recessed into the bottom of the tunnel with the threaded rod sticking out of a hole. The only problem with this is that the models do not stay vertical. Fortunately, the weight of the models bend the stick downwards and then the force of the air straightens the stick so it is close to vertical. This was just plain luck.

So far I have only used materials that I had laying around the house and some cardboard from a dumpster. Except for the pitot tube and magnehelic that I picked up at a garage sale 10 years ago, any bicycle hobbyist should be able to find the materials for next to nothing and recreate the tunnel for their own tests.

 

[Sparweb]

Wind tunnels can be calibrated with shperes instead of disks. I have a textbook called "Airplane Aerodynamics" by Dommasch, Sherby and Connoly. Though far from modern, it contains a description of the method of measuring the Transition Factor for your wind tunnel using spheres.
I can't think of an easy way to vary the air speed in your tunnel without causing turbulence, but different sizes of sphere may get you the results you need. There may be other ways of finding the transition factor if that doesn't work. A good place to look is old NACA reports.

http://naca.larc.nasa.gov/

You might also want to track down Hoerner's Fluid Dynamic Drag (see my profile), because it has drag data, too. Come to think of it, it may have data related to bicycles.


Steven Fahey, CET
"Simplicate, and add more lightness" - Bill Stout