What I was looking for was someone to talk about the temperature distribution of the exhaust gas across a cross section of a given tube. The temp distribution is shaped like a bell curve. The centerline of the tube has the hottest gases while along the wall you will find the "coolest" gases.
You want to maintain the highest average gas temperature to promote flow. Thus the reason for ceramic coatings and stainless steel.
With ceramic coating you want the inside and outside (especially for mild steel tubing) of the tubing coated. Having the inside coated will locate the "insulation wall" closer to the core and increase the average exhaust gas temperature in the tube. Stainless steel is better because its heat conduction properties are poor, therefore it insulates better.
Now think about the bell curve temperature distribtion across the tubing at a given point from the exhaust port. Let's assume we run a 30" long primary that is perfectly straight. As this column of exhaust gases travels down the tube the bell cuve distribution starts off with a higher average temp and a large peak. As it moves down the tube the peak begins to soften and the average temp falls. The rate of drop in the average temperature is based on the tubing material, wall thickness and coating (if any). So, the temperature of the exiting gases is less at the 30" location then at the 1" location.
In the above case we used straight tubing, which can be thought of a "bend" with an infinitely large radius. Since it would be difficult to run 4 straight 30" tubes from a front transverse mounted engine we really need to add some bends for packaging. We have just added something that will impede exhaust flow. Why? because the flow is changing direction and changing at different rates depending on the radius of the bend.
I think everyone agrees that mandrel bends provide better flow chacteristics than crush bend since the diameter of the tube is larger. However, even in a mandrel bend you will get some reduction of tubing diameter. The outside of the bend looses wall thickness while the inside may gain a little and therefore it can't hold the perfect circular cross section as found in a piece of straight tubing.
You'll loose a little bit in the diameter of the tubing each time you go tighter with smaller bend radii. Yes these amounts are very small but they do exist.
Now what about the effect of a radius change on the temperature distribution. In comparison, there is little mixing of the exhaust gases in straight tubing compared to a bend. So when the exhaust gases move through a bend they mix because they have to change direction. The sharper you make the bend the greater the extent of the mixing. What happens to the temperature distribution and average in the tube? If looking at it near start of the bend the average temperature will be just a little lower (since the gases are moving away from the exhaust port) but the temperture distribution will be flatter since some of the hottest core gases have now mixed with the cooler boundary layer gases. This has increased the temperature of the boundary layer gases and this causes an increase in the heat transfer rate. The heat transfer rate increases since there is now a greater temperature delta across the wall tubing.
As the exhaust gases moves through the bend it's loosing more heat because of the hotter boundary layer gases and the temperature distribution returns to the classical bell shaped curve. But this has caused a drop in the average exhuast gas temperature.
For example a 90 degree bend on a 4" radius is 6.28" long. With all other things remaining constant the average exhaust gas temperature exiting that 90 degree bend will be less then if it had gone through 6.28" of straight tubing. As the bend radius gets tighter the mixing will be "better" and happen quicker. So each and every bend causes mixing and thus an accelerated reduction in the average temperature of the exhaust gases.
So what will work best? Minimal bands on larger radii.
What other parts of a header cause mixing? The merge collectors.
A good way to verify the above is to look at the color of the tubes of a stainless steel header, and that's much easier to see when the header is new. Hotter parts will be closer to a purple color while the cooler areas are more golden. You'll see these color changes through out the bends and near the collector outlets.