Engineers puzzle over bridge collapse

http://news.bbc.co.uk/2/hi/americas/6927526.stm ^

[TXnMA]

Yes, I am saying that:

1) Initially, the lateral thrust (N-S -- along the axis of the roadway) on the north-bank piers was balanced via the combined thrusts of the fullspan arch over the river opposing that of the half-span arch to the north shore.

2) Collapse of the fullspan arch removed its side-thrust -- leaving the piers subject only to the side-thrust of the landward half-span.

3) ?(The splash wave [or impact from the falling deck] may have further affected the stability of the piers.)?

4) The unbalanced thrust from the half-span shoved (tilted) the tops of the piers river-ward. (This slight shift is visible in the differential image of the single frame prior to collapse.)

5) Riverward movement of the structure -- following the piers as they tilted -- pulled the shoreward end of the half-span off of its bent.

6) Collapse of the half-span section dragged (rotated) the top of the kingpost northward.

7) The kingposts finally detached from the shoes -- leaving them relatively intact.

~~~~~~~~~~~

The second and fourth photos in Danette's #1662 on the live thread are instructive. That fourth photo

illustrates how the half-span "pancaked" after it separated from its north bent. ( I find it to be remarkable that the entire truss structure was "squashed" down to virtually zero height by that collapse...)

There is a better view online of the above scene [with no obscuring TV banner] (taken by Josh Fisher) that shows both ends of the collapsed half-span -- but I captured only the image and not the URL. :-(

Also, there is a nice detail of the tops of the tilted piers showing the relatively undamaged shoes and the kingpost bases (by Tony Webster) but -- same story -- I have no URL... :-(

~~~~~~~~~~~~~~~

Of course, the really interesting action took place on the south shore!

(Have you read the comments by the cement-truck driver who cited examples of how his unbalanced, rotating load put a similar bridge into severe sympathetic vibration?)

[Polybius]

... but the bridge was last inspected in 2006 and no significant structural problems were found.

In reality, the bridge was only one levl above "basically intolerable".

Few high-traffic bridges in the United States scored worse than the Minneapolis bridge performed on its 2005 inspection, according to an analysis of federal records by MSNBC.com. ...........Taking into account many factors in a complex formula, the sufficiency rating ranges from 100 (the highest score) down to 0. A rating of 80 indicates that some rehabilitation may be needed. A score of 50 or less indicates that replacement may be in order. The I-35W bridge in Minneapolis was given a score of 50.0 in the 2005 inspection. ..........Overall, the structural evaluation of the bridge was listed as "meets minimum tolerable limits to be left in place as is." The next lowest rating is "basically intolerable."

[TXnMA]

I'm "into" technical visualization graphics, and, I must say that the one you posted is one of the most vivid that I have seen in quite a while. That combination of a color-coded side-scan sonar mosaic -- georeferenced to a gray-scale aerial photo -- is spectacularly good!

In contrast to the severe scouring around the pilings on the left, look at how effectively the jetty wall on the left has collected silt -- and, has, apparently, done its job of stopping the in-cutting of the bank just downstream. That makes one wonder if analogous structures placed just upstream of the pilings would have protected them against scouring.

If I were responsible for maintenance on that bridge, I'd be considering replacing the whole thing -- or, at least, dumping many barge-loads of rock rip-rap around those piers...

I'd say that your graphic is worth more than a thousand words. Even a politician who wants to spend money on pork rather than maintenance should be able to understand the risk it depicts!

~~~~~~~~~~~~

OTOH, if you read my other posts here, you will see that (at the present level of available info) I perceive that scouring played a (relatively) minor role in this particular collapse.

[null and void]

I would be very surprised if the engineers that oversaw the installation of the de-icing system didn't consider the corrosive affect it would have on the bridge. My guess would be that it was minimal.

Good point. OTOH, I'd also be willing bet the formulation of the deicing fluid has changed over the last 50 years.

[Gandalf_The_Gray]

My best guess to the cause was 4 decades of harsh winters (salt, chemicals, etc for the snow)

Although that sounds reasonable as a guess I rather doubt it.

I worked for a SE Wisconsin based company which, through corporate acquisitions, became a sister manufacturing plant with a former competitor in Minneapolis. The engineering personnel made numerous trips from Milwaukee to Minneapolis (always in midwinter it seemed!) My first trip up to "Minni-no-place" was more then twenty years ago. The city had banned studded tires and all road chemicals with the exception of sand.

Sand was only used at intersections and about fifty feet or so back from a stop sign. Most side roads were not plowed and bore a singular resemblance to a bob sled run. I will never forget that trip as the temperatures were in the -20s and my boss assigned me the task of starting the car before breakfast so it would be somewhat warmed up by the time we left the motel. It was not unusual to see empty cars sitting at the curb with the engine running all over town.

I grew up in Wisconsin and have lived here all my life and I can say I have never been so cold as those midwinter trips to Minnesota! At least we get some moderation from Lake Michigan. Minnesota gets their "Canadian air" straight from the frigid arctic!

Regards,
GtG

[Doe Eyes]

I'd also be willing bet the formulation of the deicing fluid has changed over the last 50 years.

The antifreeze in your car is less corrosive than water.

[null and void]

Yes but it is also highly poisonous, that’s why roads are deiced with mixtures of various salts.

[Doe Eyes]

that’s why roads are deiced with mixtures of various salts.

And they use ethylene glycol on aircraft because salt is to corrosive.

[null and void]

Yes, ethylene glycol, and its proper disposal are expensive, worth it for a multimillion dollar aircraft deiced in a controlled area. It's prohibitively expensive over thousands of miles of roads and bridges. Salts, whether sodium, potassium or calcium based, are much cheaper and relatively benign.

Environmentally, pouring ethylene glycol onto a bridge and allowing it to run off untreated into a river would be a non-starter.

(Although 50 years ago, they might have thought that a flowing river would dilute it to safe levels.)

[VOA]

I’m no engineer or construction-type person.

But I can recommend an expert source!
All I really know about bridges I picked up in tidbits from the columns
Henry Petroski (of Duke U.) writes (or at least used to write) for
the journal “American Scientist”.
He does an elegant job of writing about the history, business,
politics, and personalities of bridge design and contruction.
He gets all sorts of interesting tibits in his columns, e.g.,
on the opening day of a new bridge in Russia/USSR, the chief engineer
and his family had to stand under the bridge for the first day
of operation.

Petroski has penned some books as well (one linked below)

Pushing the Limits: New Adventures in Engineering
by Henry Petroski
http://www.amazon.com/Pushing-Limits-New-Adventures-Engineering/dp/product-description/1400032946

[RayChuang88]

If you look at that picture, note the following potential problems:

1) The spindly structure of the steel holding up the wide, heavy roadbed deck.
2) Only four relatively small concrete posts holding up the bridge over the Mississippi River.
3) The obvious rust of the steel just over the concrete posts.
4) The large amount of exposed steel subject to effects of the wide temperature variations of Minneapolis weather (remember the city can get under 0° F. easily during the winter and can reach around 100° F. in the hottest summer days).

I'm surprised the bridge didn't collapse YEARS earlier.

[jeffers]

Ok, if you’re still following the live thread, you know there’s dissent over the state of the bottom chord of the arched trusses, tension versus compression.

I’m lobbying compression, Spunkets is arguing tension, and Supercat seems to be arguing with me (?) for compression, not realizing that I’m already there.

If the bottom chord is indeed under compression at equilibrium, then yes, removal of the center span will permit (key word there) unbalanced axial forces to act on the tops of the piers. Simply put, as the arched bottom chord of the sidespan sags, it will tend to straighten out, getting “longer” in the horizontal direction.

This raises two problems with your analysis, however.

One, there is no axial force applied to the piers UNTIL the truss begins to deflect.

Two, straightening the arch would imply an equal and opposite reaction that will be applied at BOTH ends of the truss, resisting any tendency to drag the northern end of the sidespan off its supporting piers.

What I think I’m hearing from your analysis is that the tops of the piers moved south PRIOR to any deflection of the sidespan road deck. That would imply that the piers rotated about their base INDEPENDENTLY from the truss, due to forces unknown, and of magnitude I’m going to have a hard time accounting for.

I can see some issues raised by the stress imposed on the entire structure as the south end let go and dropped, ripping the mainspan loose from the truss system just south of the north piers. Problem there is that any yank that would have pulled the north piers out of plumb would have had to either rip the sidespan truss apart, or pulled it off its suppoert then. It should have failed immediately under thos circumstances, but it didn’t, it remained in a state of semi-equilibrium, redistributing loads, for several seconds, until it could no longer do so and began to accelerate downward under gravity.

Another potential solution would involve hydraulic effects, or impact effects of the mainspan hitting the water and river bed, and causing problems with the pier foundations which took a few seconds to manifest, but again, the energy budget seems lacking.

I hate to badger a point, but from what I hear you say, the sidespan stayed in equilibrium, and moved as a unit towards centerspan.

I have problems with that order of events. I could see the deck sagging, even imperceptibly, rocking the pier tops riverward, and if this was the case, the first visible sign of failure, depending on view angle and image resolution, might have been pier rotation.

I could stretch and see the mainspan collapse dragging pier tops and the sidespan truss to the edge of the sidespan’s northern supports, which took a few second to crumble, and then multiple sections of sidespan dropped, the end effect of which was to push the pier tops further yet out of plumb, and eventually rip the bridge shoes loose from the tops of the piers.

But the idea of that sidespan remaining stable, and then pier and sidespan trusses together moving towards the river as a unit, is tough to find a mechanism for.

The idea that the piers rotated out from under the trusses, while the trusses stayed still, is even harder to account.

If the bottom chord of the truss was under tension, as Spunkets wants to think, then I’m pretty much at a standstill.

Do you still have your subtractive imagery? Can you post the key image that shows the pier out of plumb without any truss deflection? If not, can you look at it again for even slight signs of sag on the truss and/or roaddeck coincident with pier rotation? It woulkdn’t take but a very small amount of truss defelction to straighten that arch out enough to push the piers over.

If you could find that, it would really make this easier to understand.

[TXnMA]

But the idea of that sidespan remaining stable, and then pier and sidespan trusses together moving towards the river as a unit, is tough to find a mechanism for.

I agree. I am just trying to explain why that sidespan remained (apparently) stable for 15 seconds or so after the collapse began, and then ended in a state with the piers clearly shoved out of plumb toward the river. It's almost as if it were a separate, independent failure event...

I'm going to have to download the video again (over a s-l-o-w dialup connection) but, then I will re-do the differential analysis and add some vector graphic reference lines with Canvas and upload them to my webspace for viewing here. I also may add in a side-view sketch of that side-span action -- as it appears to me.

I'm not convinced I am right; but (unbalanced) forces from the span chords being in (opposing) compression were the only thing I could come up with,

Big questions:

1. Why did that sidespan failure have such a "long fuse"?

2. What finally triggered its collapse (after the main collapse was, essentially, over)?

3. What causative role (if any) did the tilting of the water-side piers play?

4. Why (aside from puzzlement) are we struggling over something that happened a quarter-minute after the first failure occurred? '-)

[Lonesome in Massachussets]

Actually, Keith Eaton notwithstanding, most bridge collapses are the result of allision, a ship hitting a support.

[Lonesome in Massachussets]

I think it’s more likely there were workers removing something they were not authorized to remove. Or there was a bridge engineer not quite up to speed in old fashioned bridge design and authorized something he shouldn’t have.

My suspicion focuses on the maintenance crew as well. In addition, the traffic was routed to one lane making an asymmetric load that may have aggravated whatever other deficiencies were present. Time will tell.

[Lonesome in Massachussets]

A truss is an exercise in minimalistic design.

It's also not prone to buckling. If you add extra members, "redundancy" to a truss it's prone to buckle with changes in temprature or load.

[Vaduz]

But engineers don’t make mistakes do they?.

[ProtectOurFreedom]

“Allision” — The act if striking or collision of a moving vessel against a stationary object. Never heard that word before. Thanks for pointing that out.

[jeffers]

You have an excellent grasp of engineering reality, and space-time mechanics. Your questions make this obvious.

The answer to your question is not so simple.

It probably won’t hurt to read my latest three or four posts on the live thread, but I’ll try to sum up here.

Structural failures can progress rapidly, if design loads are exceeded rapidly and by orders of magnitude. On the other hand, structural failures can proceed slowly if the loads imposed are only at or near design limits.

The trusses used in this bridge were designed with both cantilever and simple truss load distribution mechanisms. Simply put, the trusses acted as balanced cantilevers near the mainspan piers, and as simple trusses near midspan on the approaches and center span.

When the mainspan failed, the counter-balancing weight for the northern cantilever assemblies was removed. Had this been a full and true cantilever design, the northern sidespan would probably have failed much more quickly.

But because it was not a full and true cantilever design, the unbalancing forces were comparatively small compared to the case of a true cantilever.

With the counterbalancing weight of the mainspan removed, the sidespan went into (terminal, but we didn’t know that for sure then) instability. Because the counterbalance weight was not the only thing holding up the sidespan, the forces were not in excess of design limits by orders of magnitude. Design limits were exceeded, but only by relatively small increments.

Instead of catastrophic failure, the structure began an accelerating series of load redistributions. In essence, the simple truss portion of the design scheme began trying to take the additional loads imposed by the loss of the counterweight.

Because the simple truss mechanism was still structurally viable, even with the counterweight removed, it was able to successfully redistribute greater than design loads for an appreciable time after the mainspan failure sequence terminated.

Individual truss member components, instead of tearing, shearing off, or buckling instantly, slowly elongated (tension members) or distorted (compression members) and a whole series of load re-distributions took place, back and forth, and members changed shape and began to shed loads they could no longer accomodate.

Eventually point loads accumulated on individual members that they could not, by design (and in implementation during the throes of failure), shed, or bear any longer and these individual members then underwent catastrophic failure.

Due to the non-redundant design of the bridge, it probably didn’t take many single point failures before big changes began to take place, and at this point, full failure was in progress. That would be when the video would begin to show visible results.

An important part of the failure sequence is missing from the video. There is no audio signal to go with it.

Between mainspan failure termination and the onset of catastrophic sidespan collapse, I’m reasonably confident that any human observers witnessing the event were fully aware of what was taking place. That steel would have been emitting a cacaphony of noise, screeches and groans, popping sounds, and in the late stages, the reports of tearing steel. The structure was dying and in spite of its best efforts to redistribute loads to survive, the loss of the counterbalancing weight of the mainspan proved to be a mortal injury.

We can’t hear that on the silent video, but anyone there would have. It’s probably one of the saddest (and most terrifying) sounds they ever heard.

[twntaipan]

How close are the railroad tracks to those support columns for the river span of the bridge?

Has any thought been given to the possibility of a train derailment at that point, with that impact causing the bridge collapse?