I didn't understand the scale of this problem until I started to monitor the behaviour of the bearings on Cleddau bridge some years ago. I noticed that, as well as moving backwards and forwards as they were intended to do, one bearing went ahead of the other at different times of day and the difference was large. about 0.5mm in the width of a 440mm wide roller. Rollers are really not capable of taking that so we may reasonably assume that the engineers concerned hadn't thought of it.
The Cleddau was started about the time FRB was opened and was designed by the same people, so we might reasonably suppose that they didn't think plan rotation was an issue there either.
However, Suspension bridges are flexible in plan, as well as in elevation. The flexibility in elevation was amply demonstrated by the load test. When the bridge was just finished there was a very big gale and the engineer reckoned the bridge bent sideways so that at mid span, the central reserve was roughly in line with the tower legs, a side sway of 6m. (I worked n big bridges in my boyhood and knew some of the people who were then my bosses).
Let's just put that on hold for a moment and consider a much smaller sway of say 2.5m. The bridge will bend into something resembling a sine wave in which the end rotation is roughly twice the deflection in half the span. so say 5m in 500 or 1in100. (You see why I chose 2.5m deflection, this is an engineer at work).
In the sketch above I have shown a 3om deflection to indicate the shape clearly and also a 3m deflection with the tangent line extended to show a true shape.
The top of the links are fairly stiffly attached to the tower.
The bottom is equally stiffly attached to the main truss.
The bottom pin is therefore trying to twist its link by 1 in 100 and that will develop a bigger stress than imagined by the designers, concentrating at the point where the link becomes narrowest.
It is the scale of these rotations that I believe catches people out. Engineers live by picking things that can be discounted as too small to cause trouble and without some serious thought (and possibly actual measurement of the effect) I certainly would not have considered this to be a problem. Even with modern computers, the stress pattern would be extremely difficult to calculate, and the original design was done in the days of slide rules.
So, I would defend, quite robustly, anyone who had failed to spot that as a danger. The stresses would not have to be enormous to create fatigue and if you haven't been warned to look for it, you would very easily miss the initial very small cracks. Only when the crack got to a critical length and went bang would the trouble become evident. And if you happened not to look in the few days before the bang, you would never spot the trouble before failure.
Having said all that, there was no really serious danger. The link supports the end of the truss and keeps it aligned. If it had broken completely it might have cause a steep ramp to appear in the expansion joint but the joint is a good metre long (a sort of mini bridge in its own right) so it would be a ramp, not a step.
On the other hand, it would have been a bang if the other link went. Repair would have become very much more difficult. Closing the bridge was definitely the right thing to do.
And finally: What I am doing here is mapping my own observations on to reported damage. My ideas could only be checked by close examination and measurement. I may be completely wrong.