The problem is in how the LCAs mount. Instead of attaching to the axle and chassis (or linked directly to them via stable
suspension components such as in use in formula cars) like a good setup should, they mount to diverging
control arms that are linked to a laterally unstable axle. Hmmm, allright, how about in English?
The LCAs are not parallel with each other. So in a turn, the distance between the
control arms is trying to change and that places stress on the sway bar. When you couple this with an axle that has been known to have up to 2" of lateral movement, even more stress is introduced as the angles between the LCAs is becoming even more extreme. A lateral control device (such as a
panhard bar or watts link) will help quite abit, but you're still left with the problem of diverging
control arms.
If you'll notice, these stock sway bars always break at one of the rear bolt holes to the LCAs (I've never seen a failure at another place and wrecks don't count
). This is where the majority of stress is placed as that's where the most leverage is placed on the sway bar.
One of these days, I need to educate myself on computer animation and modeling. A view of all of this in action would make it infinitely easier to understand. If there's still some confusion about it, I'll make up some sketches in paint that better illustrate the issues.