Thanks for the comments, Jeffers.
Here’s more food for thought. I haven’t seen these issues discussed.
Look at the big photo at http://fullyarticulated.typepad.com/sprawledout/2007/08/article-are-the.html (Sorry about the political content, but it’s a great photo) and I notice 3 things.
1. South landside span (the one that twisted), the 2 southbound lanes are folded in so as to be almost vertical. A tighter fold than anywhere else in the main bridge system. Why did they fold so vigorously, especially when the main truss that they were on top of seems to have become completely separated? If it had folded down on both sides of the main truss, it would have pinched it, and the main truss wouldn’t be lying down on the riverbank all casual-like.
2. No cars on the SB lanes or on the ground on the twisted south landside span. You’d think if there were cars on that span when it collapsed that you’d see them on the ground. And there was the schoolbus and semi truck just ahead of them, which you’d normally get cars queing behind. If the main span fell first, thes cars would have gotten off without any additional cars coming on. No cars visible in http://www.flickr.com/photos/thecurseofbrian/981931600/in/set-72157601159020144/ either. Maybe there just weren’t any cars, the traffic cam video should show this.
3. The west truss is lying horizontally on the pier, and it’s in pretty good shape. It’s still pretty much all in plane, it has a couple web members all intact, in fact, it’s the only non-mangled piece of the whole supertructure.
Another shot of the relative good condition of the west truss over the south pier http://www.flickr.com/photos/s4xton/980394803/in/set-72157601157770382/
and 2 other things:
* South approach girder spans failed due to compression—they got shoved back away from the river. South side http://www.flickr.com/photos/s4xton/980437381/in/set-72157601157770382/ and north side was plainly tugged, as one span even got tugged off its foundations a couple spans back from the RR tracks. (Though it looks like the south spans slid backwards after it kinked on the pier, so it could be a post-collapse feature (slipped on the way down). In any case, the south side girder span that fell on the Tastee truck cab seems to have fallen in a different sort of way than the north side approach girder span that fell on the train cars.
* High survivorship is really remarkable. Lots of folks dropped 105’ into the river and walked away from it. 3 of the folks that died did so after they fell—one drowned rescuing people, one had a lampost hit them, one had a sign land on their car. And the truck driver had the fire to contend with.
http://www.flickr.com/photos/s4xton/980407219/in/set-72157601157770382/ this pic shows the plastic pylons still rightside up...
*************
So, seems to me the south approach cantilever arm (from pier 5 to 6, I think) with the twisted roadway almost has to be where it started. Everything else failed symetrically across a cross-section of the deck. The cars are all where they were on the road, so the main span didn’t twist on its way down to the river.
And it was just good luck that there weren’t cars going south at that instant in time, or they’re hidden under the bridge in the one photo. But with a sideways fall seems like you’d have lots of fatalities.
The intact west-side truss suggests that it simply fell down, didn’t fail at all, wasn’t crushed.
So somehow the main deck over the river didnt get twisted, even though the underlying truss did...
Anyhow, more stuff for everyone to think about, piece together.
Kwuntongchai
Those are some good pictures. Thanks for posting the link. More on that later.
In answer to your questions:
1. The cantilever arms of the floor trusses (supporting the outer edges of the road deck) seem to have failed near universally across the length of the trussed spans.
This suggests to me that there was either less margin of safety in the design there, or else the bridge was subject to a powerful jerk in the vertical plane at some point in the failure process.
We can account for such a jerk when the main trusses, largely intact, hit the ground falling straight down. The cantilevered outer deck lanes just kept falling even after the fall of the main trusses was arrested by impact with the terrain underneath.
I don’t thing that applies to the area you describe, since the west truss fell over sideways, nearly intact. The folded roadway edges there suggests a powerful shock in the vertical plane, nefore the lean became pronounced, but accounting for this heads quickly into gray territory.
As always, “gray territory” in this bridge failure, for me, anyway, always comes down to a single question: “Initial main truss failure at the south end of span 7, or at the north end of what we’ll call, for lack of a better term, span 5.”
More on this later too.
2. Implied question, where are the cars? Answer, in the woods west of the bridge, between the northbound and southbound lanes, on the northbound lanes heading south, under the northbound lanes, or there weren’t any.
Or...
...hmmm, did you just imply a 64 dollar question?
If the center span, span 7 over the river, severed from span 6 very early in the failure, and span 6 survived for a few more seconds, the truck and bus may have been the last southbound cars that made it onto span 6 before span 7 went in the drink.
Interesting.
3. The pier 6 panel of the west truss is in pretty good shape. Members still attached in the panels immediately north and south suggest one of two things:
A. The pier 6 west king post rotated clockwise, as seen looking east from west of pier 6, prior to collapse to the east. Both the top chord and bottom chord of the panel just south of the pier 6 panel, of the west truss are bent “up” in relation to the trusses design oriantation. Both the top chord and bottom chord of the first panel north of the west trusses pier 6 panel are bent “down” in relation to the truss’s design orientation. The bottom chord is ripped half out of the connection gusset, which itself is mangled. When I say “ripped”, I mean the four sides of the box beam are no longer joined at the corners of the original box. Powerful stresses were imposed there.
The pier 6 west truss panel’s post collapse orientation also supports early rotation in the plane of the standing truss. The base of the kingpost now sits riverward of pier 6, at least one rocker plate from pier six now sits riverward of pier six, and the top of the SW kingpost sits shoreward of the bottom of the SW kingpost.
Such rotation implies a failure sequence similar to that seen on the north end of the bridge, at span 8. In fact, there are very many similarities between the post collapse visual appearance of the east truss at pier 7, and the west truss at pier 6, even though both trusses at pier 8 are still vertical. The pier 8 east truss has both top and bottom chord bent “up” at the next shoreward panel, and both remnants of the top and bottom chords bent “down” at the next riverward panel, just like the west truss at pier six, except now that truss is lying on it’s side.
This suggests the possibility that span 6 failed in a manner similar to span 8. We are reasonably confident in understanding the mechanics of the span 8 failure. From the video we see span 7 drop close to the camera, sever at the far end of span sevem, and see span 8 remain standing for several seconds before collapsing.
With the counterweight represented by span 7 missing from the cantilever calculation at pier 7, the center of span 8 sagged, drawing the kingpost tops at pier 7 and 8 together. Eventually, the midspan sag of span 8 was so great that both ends were dragged off their piers.
Did this happen on the south side too?
We see similar rotation at pier 6, but we also see that the centroid of mass and structure of the pier 6 west truss seems to have come to rest riverward of pier 6, while the span 8 mass ended up shoreward of pier 8.
Let’s hold there for a minute and look at the other possible reasons why the top and bottom chords adjacent to the pier 6 west truss panel might be bent the way they are.
B. If span 7 ripped apart just north of pier 6, it would begin falling immediately thereafter. Nothing left to support it and the pier 8 cantilever could never reach that far south. Members still partially attached at the sever point would be pulled downward by the falling portion of span 7.
That can account for the bending we see in the west truss, one panel north of pier 6.
Again, with the span 7 counterweight removed, the shoreward cantilever is disturbed, this time at pier 6, and affecting span 5. Span 5 sags in the middle similar to span 8, but in this case, the riverward ends of the trusses also fall over to the side. Why?
Clearly, because their transverse integrity, their ability to resist lateral forces, was compromised. The sway bracing between kingposts, and between vertical struts at every panel connection HAD TO fail before the trusses could fall over sideways.
How does this happen?
Two ways.
The SW kingpost buckles and rips loose from the sway bracing.
The SW kingpost topples, still straight along its length, ripping loose the sway bracing.
So....we have two mechanisms to account for what we see, both near equally plausible.
One, span 7 severs near pier 6, one truss, east or west, before the other, which follows quickly, but not instantaneausly. The removal of the span 7 counterweight causes span 6 to sag and then collapse.
Two, span 6 sags and collapses, removing the counterweight for span 7, which sags, severs, and as the pier 6 truss assemblied fail assymetrically, they lean east.
Wait a minute, who’s talking about a span 6 trigger?
Well, MnDOT, for one, indirectly anyway. The worst problem the bridge ever had was when the SE rocker bearing at the endbeam/croissbeam assembly froze, necessitating closure of the whole bridge, and jacking it up to repair it, back in 1986.
What is a crossbeam/endbeam/rocker bearing, anyway?
It’s another cantilever.
At the south end of span 6. (and the north end of span 8, but that’s not important now).
At the south end of span 6, the big steel trusses end, and the road deck is supported very differently. Instead of two large steel trusses spanning pier to pier, supporting crosswise (transverse) floor trusses, which in turn support longways (axial) steel beams which in turn support the road deck, we change over to concrete piers each supporting a wide array of axial steel beams and then the road deck, just like 95% of all the highway overpasses most drivers are very familiar with.
But the span5/span6 junction is special. There is a cantilever in play there. I’m going to simplify this a little bit to get the point across without drawings and three D models.
Assume span 6, the southernmost trussed span, extends well past its southern piers. It’s hanging out into space, south of the piers. Assume the weight of the south end of both trusses rest of a massive steel beam made up of cold rolled (not welded) pieces all bolted together, and that this beam rests on the span 6 piers, pier 5. Concrete piers holding up a huge beam, which hold up the trusses which then reach further south, just hanging in space.
With me so far?
Good, that massive steel beam is called the crossbeam.
Now, attach another massive steel beam to the ends of those trusses hanging out in space. This is called the endbeam. Now attach the north end of a normal highway overpass to the endbeam.
That’s the idea. If the span 6 trusses fail, so does the north end of span 5.
But it’s more complicated than that. You have the concept, but in reality, the crossbeam directly supports the main trusses, AND it directly supports the endbeam. There are axial beams running THROUGH there, and there are rocker bearings mounted on the crossbeam which actually support the endbeam.
Not necessary to understand exactly how, just the concept. Even I don’t know exactly how this all fits together, I haven’t pored over the engineering drawings. But the concept is easy, there is a cantilever over pier 5.
The north end of span 5 helps support the south end of span six, the first big truss span as you move north. The south end of sopan 6 helps support the north end of span 5.
They are co-dependant.
Fail one, the other soon follows.
Just like we saw onb the north end of the river. the main span failed, then span 8 failed, north of the river, a few seconds later.
Why is all this possibly critical?
Look at the pictures.
You WILL NOT SEE SPAN 5.
The southbound semi crashed into it, head on.
Span five came to rest near vertical, north end down, and the semi ran off the south end of span six, ran into the vertical concrete wall that was then span 5, and burst into flame.
More important yet, the north end of span 5 is UNDERNEATH the south end of span 6.
There is one photo I know of, showing two northbound SUVs resting on span 5, with their hoods crushed under span 6 and on fire. In the distant background, you can see the light from the fire caused by the southbound semi.
If the bridge failed at the south crossbeam/endbeam, dropping the north end of span 5, span 6 would have lost it’s cantilever. Span 6 would then sag in the middle, just like span 8 did. The north end of span 6 would rotate, top towards the shore, just like span 8 did.
If only one side of the crossbeam/endbeam failed early on, then only one of the span 6 main trusses would have lost it’s counterweight. If the east end of the crossbeam/endbeam failed first then the east truss of span 6 would sag earlier than the west truss of span six.
Both would sag quickly, because one truss was nowhere near strong enough to carry the whole bridge by itself, but when one sags even a little quicker than the other, now you have a transverse moment.
Now you have “lean”.
The lean we see in all the photos.
So, the trigger point of failure could have been...well...probability suggests one of three main places.
1. The SE kingpost crumples, either ripping out the swaybraces attaching it to the SW king, or because the sway braces failed first.
Mental exercise: stand two wooden children’s toy blocks, oh, a foot long, on end, six inches apart. Place two smaller blocks between the two horizontally, one midway up the tall blocks, the other between the tops of the two standing blocks. Wrap six feet of duct tape, horizontally around the horizontals and verticals, tying them to the uprights.
Now stand on it. Try to either crush one of the uprights, or make one of the uprights rip the duct tape and fall over sideways.
Hold that thought.
2. One side or the other of the south crossbeam/endbeam fails, dropping first the north end of span 5. Span 6, relieved of its cantilever counterbalance, sags midspan, rotating the pier 6 kingposts towards the shore at their tops. Span 7 severs just north of pier 6. Span 8, relieved of its counterweight, sags and fails as described earlier.
Mental exercise: stand four children’s blocks on end. Lay three blocks on top of these four “piers” to create three “spans” of bridge. Run one strip of duct tape end to end atop the spans.
Now, jump on the middle of one of the end spans, breaking it between piers.
The goal is to end up with the broken span underneath the middle span, and the far end span underneath the middle span. The middle connecting span has to fall LAST.
Hold that thought.
3. Span 7 fails just north of pier 6, one truss, east or west slightly before the other truss. Span 7 severs completely, both trusses, just north of pier 6. Both trusses at pier 6 are yanked northward, and span 6 loses its counterweight as soon as span seven breaks free. Span 6 sags, disturbing the crossbeam/endbeam assembly, actually pulling it apart, remember there’s only rocker bearings connecting the two, span five’s north end just rests atop span 6’s south end. The north end of span 5, along with the endbeam, drops, and span 6, still sagging, eventually fails like span 8 later did, as described earlier.
Mental exercise: create the same structure as in exercise 2, four piers, three spans, but this time, only duct tape two of the spans on top. Jump on the middle of the end span which IS connected to the middle span with duct tape, breaking it.
The objective is to break the end span and pull the middle span sideways towards the break, jerking the far middle span pier out from under the farthest endspan.
That’s easy to do.
The earlier two are harder to accomplish. The first one is near impossible. Crushing the kingpost, failing it in compression, is VERY hard to do. Your own intuition supports this, and engineering experience suggests very very few members fail in compression.
That’s why I favor initial failure in span 7, just north of pier 6. One truss or the other first, not simultaneausly.
That’s where I’ve been since 24 to 48 hours after the collapse, what, two weeks ago?
But it’s not conclusive.
Because option two, failure at the crossbeam/endbeam, slightly more difficult than option three, severing span 7 just north of pier 6, while less likely, is still plausible and because the north end of span 5 is UNDER the south end of span 6, at “the bottom of the pile”, meaning it failed early on.
For two weeks, I’ve been trying to decide, span 5 falling first, or span 7 falling first? A week ago, I decided there wasn’t enough visible in photos to be sure, but that span 7 severing was more likely than span 5. Span 5 falling first would be the tail wagging the dog.
It happens, but it’s usually the other way around.
Now, to wrap up, back to your new pictures.
I found a new member.
It’s a very large piece of steel.
Very large.
It could be the SE kingpost.
It’s big enough.
It’s in a strange place, and bad, bad things happened to it.
It could also be a top or bottom chord.
I’ve torn the images apart, downloaded tens of them at full res and those are SLOW on dialup.
I can’t tell.
It might be the SE kingpost.
It’s draped on the concrete wall at the river’s edge.
It’s a very large cross section box beam, more rectangular in cross section (like the SW kingpost), less square in cross section (like the top and bottom chords).
The key is the holes. Big ovals mean chords. Two slits mean kingposts or struts. Diagonal’s have two slits, but they are much smaller in cross section.
I can’t see the holes well enough to tell.
If it’s the SE kingpost, it’s a loooooong way from where it was before the bridge fell.
No matter what it is, it took a beating. One end of it is wrapped in a corkscrew spiral. The other end used to be in a gusset connection, you can see where it didn’t get painted last time around. The gusset was forcibly ripped off and is not visible.
It would take enormous force to twist a piece of steel that big like that. Twisting force. Rotational force. A moment. The only way I can think of to do that would be to fix one end, attach the other end to a huge bridge, and push the huge bridge over sideways.
Hmmmm....
That’s where I’m at right now.