Lets say that one of the tension members in the center span (one on the bottom) had a fatigue crack that was ignored or missed. The bumper to bumper traffic and the construction equipment were too heavy for the remaining metal.
The tension member would snap. That side would start to fall straight down. It would pull the other side out of vertical and that would fail before it got very far off vertical. The center section would then fall straight down (although tipped slightly since one side got a head start), pulling the supports on each bank towards the water. That is what it looks like in your photo.
Once those supports were gone, the bridge approach structures on both sides would fall, starting from the missing supports and working in both directions until it reached solid ground (or very close to it).
BTW, the researchers will be looking for a piece that has a crystalline (brittle) crack for about half its width. That part will be rusty because it took YEARS to develop. The rest of the break will be ductile and have little rust (only that which forms from the yesterday to when they pull it out of the water). That will be the initiation point for failure.
Anyway, that is what my money is on.
I wonder what would happen if it was found to be cutting torch marks at the critical stress points, anybody can buy a portable cutting torch.
When the center span came down it was tipped slightly side to side, as you suggest, but it was also tipped end to end, south end first, by the video. As far as the main span goes, it’s pretty clear that whatever failed affected the south set of cantilevers first. You can see the north cantilevers flex and then snap, just south of the north piers, and the northern approach span sits there several seconds before rocking back shoreward and collapsing.
However, it’s hard to see in the photo, but the southbound semi truck that burned did so because he went head on into a near vertical section of road deck, supported on one end, tilting down to ground on the river end. You can see it in the photo I posted, right next to the school bus.
From a different angle, it’s clear that the north end of that vertical slab sits under the approach span that the bus and semi sit on. The flat slab the semi is on rests on the hood of an SUV which was northbound, and another northbound vehicle smashed head on into the end of the slab the semi and bus rest on.
Probability says that the triggering failure occurred in the first span to fall, in other words, the bottom of the debris stack.
There are two main bridge components at this level. One is the southeast cantilever superstructure, the other is the north end of the approach span referred to above. The failure of a single member in either of these assemblies could have initiated the collapse progression.
However, if the trigger had been the approach span, it left the intervening approach span suspended in place while the south end of the center span beat it to the ground. The same is true if the southeast cantilever of the center span failed first. The “bottom of the stack” sits both north and south of a span which is not at the bottom of the stack. The slab the semi and bus are on probably failed laterin the sequence.
Given this, it makes sense for the larger structure to fail first, because it would send bigger ripples across the intervening approach span to drop the next adjacent span than if the sequence occurred in the opposite direction. If the approach span failed first, the tail would be wagging the dog. Further still, half the tail would be wagging a very large dog, because the south end of the near vertical span appears to rest on its original supports. Obviously this is not conclusive, just playing the odds here. The failure of a big section of bridge is more likely to shake loose a much much smaller section of bridge, after jumping over a section which failed later in the sequence, than the other way around.
From that point, just about any member of the southeast cantilever could have initiated the collapse, but very rapidly, the failure transferred considerable load to the kingpost of the southeast cantilever, which was unable to maintain integrity and buckled. That’s what initiated the major collapse sequence, because as you have noted earlier, there are considerable point loads at each of those four main piers, and of the four, the imagery suggests that the southeast kingpost failed first. Because the southwest cantilever kingpost is largely intact and visually straight, the lateral stringers and diagonal braces between the two had to let go early on to allow the southeast kingpost to crumple.
Because the center span fell without significant deflection, no major curves in the road surface visible in the video, it is safe to assume that whatever failed was out of sight to the right of the camera’s view, in other words, towards the middle or the southern end of the southeast cantilever structure. I mean, it’s possible the connecting pin between the two eastern cantilevers failed, or any other point on the east side, and the north kingpost was better able to resist the additional loading that its southern counterpart, but much less likely since the load transfers would have been progressive and as yoi p[osted, there wasn’t redundancy in the structure. If a tension member had failed out in the mainspan, I’d expect to see it, then adjacent panels fail progressively towards both ends, but the video doesn’t support this.
Putting it all together reduces the possibilities to two main groups.
1. One or more members in the approach span two south of the main south piers failed, triggering collapse of the north end of that span, inducing through shock, vibration or load transfer, greater than design loads on the southeast cantilever, which failed, bringing down most of the rest of the bridge.
2. One or more members failed very close to the southeast pier, either in the main cantilever assembly, or in the lateral bracing system. The video indicates that the triggering failure propbably wasn’t very close to midspan, That leaves the river side close to the southeast pier, and the entirety of the counterbalancing shore side of the cantilever assembly.
Of these two, I favor the second, because of the “wag the dog by remote control” discussion above. Naturally, any part of this chain of observations could be wrong. Sometimes things bounce when they hit the ground, and end up on top of other things which fell later in the sequence. But the odds and imagery generally support this conclusion, and until they fine tooth comb every piece of steel from the debris, that’s about all we have to work with.