Flipper said:
...........Here's my theory (and I'm a mechanical engineer who's done a little work with this kind of thing). Mazda probably designed the bars to be smooth end front and spoiler in the back. The bumps in the back would serve as vortex breakers. Vortices are the little eddies (like off of a rowboat oar) that shed off of a trailing edge of a surface. Depending on the speed of the air (or the car in our case) these vortices can form at a frequency that causes lots of noise........
Similar to what I had posted on the second post - I majored in ME for two years, but finished with a Industrial Engineering degree, but I still maintain my knowledge-base...
Aerodynamic reasons would tell one that you would create more turbulence (noise) with a reverse teardrop profile (airfoil). You would effectively need a profile that was "streamlined" - pointy to pointy, but even that would cause noise as the leading edge cutting through the air causing sick amounts of eddy currents to cary over the top surface and "collide" with currents running off the trailing edge.
Flipper said:
...........(Side note: Look at some tall smoke stacks in some industrial areas and you'll see plates just sticking out of the side on the top half of the stack. These are to break up vortices to avoid failure similar to that famous bridge collapse.)
Actually, the "fins" are there to
create vortices - 1/2 of one dozen, 6 of another

same thing....; shearing forces of the wind against the face of the stack are what cause structural damage over time. The fins break up the wind and create eddy currents that will dissipate the dynamic stresses put on it by the wind. Stacks have the fins because their shape inherently causes unwanted forces; eddy currents around a sphere/round object on the "trailing edge". Laminar flow ends up increasing in velocity around "rounder" objects.
Stacks are usually placed in series so currents flowing off of one stack would (assuming the conditions are right) would flow into the others, causing undue stresses. Contrast this to F1 or other race cars when they trail in the currents of cars in front of them. The cars are designed such that their trailing flow will create a "vacuum" for a given distance (often why if you've ever been to an F1 or Indy race up close, you don't "feel" the car until a second or so after it passes; similar to the dopplar effect, but via air currents)
Here's a good thesis on the topic:
http://scholar.lib.vt.edu/theses/available/etd-5024212489622671/unrestricted/thesis.pdf
Flipper said:
...........Dimitrios mentioned the proximity to the roof. Maybe the designers had a great cross bar design that they tested independent of the car. Maybe it didn't work so hot on the car and reversing them was actually an improvement, a cheap workaround. I won't open my sunroof until the spring now, but I don't think the wind noise is so bad. At 70 mph, I'm hearing the engine and wind around the windshield as much as noise in the roof area. Not saying the car is quiet. I just don't think the rack is the worst culprit.
It may be an issue. I feel that design itself is flawed - everything from removing the bars (if wanted) to the very construction (I , as well as others, get "creaking" when taking turns when the rack is loaded with a bike, etc). Mazda needs to redesign a bar system as "aftermarket" that is similar to the Audi/VW or Subaru setups (honestly, I feel highly about those systems). Has anyone noticed how the Mazda3 hatchback roof-rack is setup? Any improvements?