Possible design flaw found in collapsed U.S. bridge

http://news.scotsman.com/latest.cfm?id=1247802007 ^

[traumer]

MINNEAPOLIS (Reuters) - A week after a deadly bridge collapse, U.S. Navy divers cut through tangled debris with underwater torches and saws on Wednesday in the search for victims while investigators identified a possible flaw in the 40-year-old span's design.

The August 1 rush-hour collapse of the Interstate 35W bridge hurled vehicles into the Mississippi River 65 feet (20 meters) below, with many tumbling onto the bridge's crumpled concrete deck.

Reacting to the disaster, officials demanded inspections of potentially suspect bridges across the United States amid renewed calls to shore up the country's aging infrastructure.

Five people were killed in the bridge collapse, a death toll that was confirmed within a day of incident. Eight other probable victims are listed as missing.

The recovery process has been slowed by huge slabs of steel-reinforced concrete and dangerous chunks of debris submerged in the river's swift, turbid waters. In some cases, divers had to use their fingertips to read license plates.

"This is going to be a process of having to, most likely, pull these vehicles out and do a long-term extraction, taking apart the vehicle to recover evidence, (and) any (human) remains," Minneapolis Police Capt. Mike Martin told reporters.

National Transportation Safety Board investigators said they had found a potential design problem with gusset plates, or steel plates that tie together angled steel beams of the bridge's frame.

Investigators are trying to verify loads and stresses on these plates at specific locations as well as the materials used to construct them.

Officials stressed the finding is preliminary and would not say exactly where the plates were located or whether failure would have caused the collapse.

"We are continuing to make progress on this investigation, and each area of inquiry gets us closer to ultimately determining the cause of this tragedy," National Transportation Safety Board Chairman Mark Rosenker said in a statement.

Out of some 100 people injured, only a handful remained in hospitals with one in critical condition.

City officials have called the large number of survivors miraculous.

For the families of the missing it has been an agonizing wait. Members of Minneapolis' large Somali immigrant community are grieving over the presumed death of a 23-year-old nursing student, who was pregnant, and her 2-year-old daughter.

Minnesota officials were quickly laying the groundwork for replacing the vital eight-lane bridge, which had been the state's busiest with 140,000 vehicles crossing it each day.

Construction bids were due on Wednesday and officials hoped to choose a contractor within weeks to build a new bridge by the end of 2008, with the help of $250 million (123 million pounds) promised by the federal government. One proposal called for two spans of five traffic lanes each, with room for light rail or buses.

It was unlikely a new bridge could be completed before the Republican Party convention in September 2008, to be held in neighbouring St. Paul.

[DoughtyOne]

Thanks for the comment.

To my way of thinking the vibrations resulting from the repairs may have contributed, but then it would seem your average traffic would cause vibrations also.

Along those lines, harmonics is an interesting field. Perhaps the jackhammer or whatever hit just the pace to cause the worst damage for the failed part.

Take care.

[Lancey Howard]

Construction bids were due on Wednesday and officials hoped to choose a contractor within weeks to build a new bridge by the end of 2008, with the help of $250 million (123 million pounds) promised by the federal government. One proposal called for two spans of five traffic lanes each, with room for light rail or buses.

Wait until the EPA gets involved. If you think that's a joke, ask homeowners in Ocean City, NJ about their bridge rebuilding project.

[DoughtyOne]

I get pretty angry with Bush, but I do find a lot of humor in the stupidity of the left, that is just as likely as not to pick up on your post and adopt the talking points.

Thanks Mad_Tom...

[KAUAIBOUND]

I am a structural engineer with 28 years experience and you are 100% correct. The contractor overloaded the bridge with construction material. The NYT slant attempting to create fear over the use of gusset plates is IMO a deliberate attempt to create fear in the general (and generally uninformed) public and cause an outcry about spending money on the WOT. It’s all about Bush.

I haven’t seen any drawings or diagrams of the bridge, but I suspect that the reason that failure of one part of the bridge caused the failure of the entire bridge was because the bridge was designed as a continuous span structure, unlike the Oakland bridge that collapsed in the 1989 earthquake, which was a single span bridge. Continuous span bridges permit longer spans and more economical used of material, but are susceptible to progressive failures such as occured.

[jeffers]

That gusset appears to have failed due to a moment. The far end of the member attached at that point was rotated about that gusset connection point, popped the bolts and pulled free.

A "possible gusset design flaw" is not much to go on. Much more useful would be the location of that gusset or gussets, as we still don't have the major components of the failure sequence in proper order yet.

Still focusing on the north end of span 5/pier 5, and/or pier six and the superstructure above pier 6 here. Both still look viable. Initial failure could have been anywhere from the midspan of span 5 through the south half of span 7 (center span), probably nearer the east side than the west.

The most interesting point on the bridge for me right now is just north of pier 6, east truss. Road deck south of pier 6, and presumably the span 6 east truss look prety good comparatively, but the span 7 east truss got hammered hard just north of pier 6.

I've isolated the SE kingpost (buckled under load) and the top chord, (folded down under moment at north end of span 6 road deck) and one other member, either a tension brace or the bottom chord, but that panel is so scattered that it's hard to piece it back together.



Blue arrow 1 indicates the SE kingpost, blue 2 is probably the tension brace angling north and down from the top of the SE kingpost, blue 3 is probably the east truss span 7 top chord, and blue 4 is probably the span 7 east truss bottom chord, although it may also be the diagonal tension brace if blue 2 is wrong.

Red arrows 1 and 2 seem to indicate that the top of the southwest kingpost rotated south, both bottom chords are bent consistent with this theory. So is the top chord of the pier 7 west truss, just slightly off the end of blue arrow 4.

Red arrow three indicates similar behavior of the span 7 east truss top chord.

If the center span broke free early on, its counterweight would be deprived to span 6 just south of pier 6, and would explain why the road deck was so compromised just north of pier 6.

However, the SE kingpost or other members closely associated with that panel of the east truss could easily have failed first, dropped the south end of the span 7 east truss, overloaded the south end of the span 7 west truss, and accomplished the same result.

I'm starting to consider a failure around the second or third panel of the east span seven truss, just north of the SE kingpost. That would be at the right end of the member marked with blue arrow 3. Supporting this is the position of the east truss span 7 bottom chord (blue arrow 4.), which does not appear to have leaned over eastward with the rest of the eastern truss at pier 6.

Initiating trigger or not, bad things happened there.

[jeffers]

Yes, the more i think about this, the more sense it makes. It resolves an issue that’s been bothering me since I first realized the trusses leaned east at pier 6.

THE SPAN 7 ROAD DECK DID NOT LEAN EAST WITH THE PIER 6 TRUSS PANELS. IT DROPPED STRAIGHT DOWN.

Since the trusses above pier 6 clearly leaned east, and since they did so without taking the south end of span 7 with them, span 7 MUST have seperated from BOTH main truss kingpost panels, PRIOR TO those panels leaning east.

It explains why the SW kingpost rotated, and why span 6 failed, it lost its center span cantilever counterweight.

The critical evidence is in the river, close to the south shore.

[jeffers]

Still doesn’t resolve the basic question though, whether span five dropped first, or something near pier 6 failed first.

With the cantilevers in play, I can build a good case either way.

Still, another brick in the pile of understanding.

I can now be reasonably certain that span 7 seperated from the pier 6 kingposts before those kingposts started leaning east.

Not the whole answer, but better than a poke in the eye.

[metmom]

Along those lines, harmonics is an interesting field.

That's an interesting thought.

I had a glass statue that my sister slid along the surface of a hutch to put it back in position. About 15 seconds after she took her hand off it, we heard a loud *PING* and the thing was cracked right in two.

[Last Dakotan]

To my way of thinking the vibrations resulting from the repairs may have contributed...

Fatigue is from cyclic loading, heavy truck rolls on, heavy truck rolls off, heavy truck rolls on... etc.

You get the picture.The crack is opened and closed repeatably and works it's way through the member.

[ninonitti]

“If it’s not Scot.....it’s crap”

[DoughtyOne]

How about that. That’s a strange one.

I’ve been told that military members never cross bridges while marching in step. That may no longer be true or it may have been an old wives tale. The idea was to avoid rhythmic vibrations.

Harmonics as I understand it, is the study of waves or vibrations. A wave or vibration of one length or cycle would cause no problem, but if adjusted could cause a buildup of tension in an object to the point it could weaken or break.

I’m sure you’ve seen examples of the singer who hits a certain note and the goblet shatters. It’s the same thing. Hit just the right harmonic wave and poof. Adjust the wave and no problem at all.

Harmonics. I think a snooze is in order. Good night. Oh, it’s morning.

Bye.

[DoughtyOne]

Thank you.

[Jonah Hex]

One other aspect to consider. A bridge is usually designed to evenly distribute a load both longitudinally and laterally along the bridge structure.

Considering two lanes were closed down for maintenance resulting in channeling rush hour traffic into a limited space, is it possible an unevenly distributed spanwise (i.e. laterally) heavy load could have contributed to the structural failure?

[supercat]

Of course there’s a design flaw. While we don’t know what initiated the collapse, we do know that because of the design, the whole bridge came down. A proper design wouldn’t have the whole thing coming down when just one section has failed.

Designing redundancy into structures like bridges is not nearly as simple as it might sound. Among the issues to be dealt with:

1. Failure of certain structural members will often cause forces on certain other members to increase beyond normal values by one or more orders of magnitude. For example, in a typical suspension bridge, the cables on both sides of each tower pull roughly equally. The tower must resist a very substantial downward force, but a much smaller lateral force. If the cable on one side of the tower fails, the lateral force on the other side will have nothing to counterbalance it and may be even greater than the downward force. There would be no practical way to make the tower strong enough to resist such force.
2. Redundant structures must include roller bearings or other such features. In many cases, bearing failure may subject multiple parts of the structure to forces that are orders of magnitude beyond normal; this can become a failure mode in and of itself. Such joints are often required, and can cause problems, even in non-redundant structures, but in redundant structures they offer even more opportunities for failure.

A properly-designed system of breakaways could in some cases limit the structural consequences of a localized failure. Of course, one would have to hope the breakaways themselves don't cause problems themselves.

[TXnMA]

Thanks or posting that great overhead view of the south bank piers and associaated damage -- it is just what I was looking for.

To aid in visualization, I have overtinted the piers & footing in yellow, and have tinted the following elements in orange:

A. What I interpret as the western kingpost and its associated "triangle" -- with an arrow showing how it rotated clockwise (in the plane of the view) off of its shoe. (The shoe [possibly damaged] still remains atop the pier.)

B. The western edge of the cantilevered deck "tucked back in under" the remains of the deck where the top of the kingpost went under the deck. (Does this indicate that the western kingpost collapse was early in the collapse sequence?)

C. A secction of ?beam? bent into a curious arc shape. What does this say about the various moments in play here?

Also, this view confirms that the decking riverward of the piers is missing (and, presumably in the river). What does this say about the forces imposed on this system?

[jeffers]

Re your question B:

“Does this indicate that the western kingpost collapse was early in the collapse sequence?”

Yes and no. It means the road deck didn’t fall and bring the truss with it as a result, but we’ve never looked at that possibility anyway. “What’s under what else” is usually only a good indicator as to sequence when both pieces started out at the same elevation level. The kingpost began life under the road deck, it ended life under the road deck, it’s hard to draw useful conclusions from this.

Re question C:

“What does this say about the various moments in play here?”

Indeterminate. Not sure what that member is, therefore speculation about it’s failure mode is difficult and inconclusive. From size and placement, it could be part of the sway bracing between the east and west pier 6 kingpost, or it could be part of a deck truss, or it could be a deck stringer.

The image shows it subjected to either bending moment or compressive overload. If it’s part of the sway brace assembly, it probably indicates that the kingposts did not fail in unison. The truss tops could have leaned towards each other, or one could have buckled first, either mechanism could produce the deflection show in the image.

If it’s a chord from a deck truss, it might have folded during failure when something else hit it on the way down, on final impact, or, it could indicate that the deck truss itself failed, allowing deflection between the main trusses.

If it’s a deck stringer, that span of deck, between deck trusses, could have failed from point loading, (unlikely in my opinion) or impacts during the collapse could have affected it.

Speculation, it’s a sway brace crossmember, and was all or part of the trigger event in the failure, or else it merely indicates that the south kingpost buckledor leaned east before the west kingpost rotated off the pier.

Best I have, sorry it isn’t more.

Interesting note...the member you indicate at yellow arrow B is almost certainly the top chord from the west main truss, from just south of pier 6. Your dashed lines are either improperly drawn, or else something significant happened at its connection with the west pier 6 kingpost. I suspect the former.

There is visible deflection just above and left of the yellow arrowhead. I suspect this trend continues underneath the road deck, and that this member remains attached to the top of the SW pier 6 kingpost, further “down” in the image.

This would indicate even more deflection that what is visible, further supporting my contention that the west pier 6 kingpost rotated, base northward, at or during main collapse, as previously noted by red arrows 1 and 2.

Could have happened two ways:

1. South end of span 7 shears off and drops, relieving cantilever counterbalance weight to the main trusses just south of pier 6, such that span six deflected in the center, rotating one or both main truss assemblies at pier 6, just as we see on the north side of the river at pier 7.

2. Crossbeam/endbeam/rocker bearing failure at pier 5 relieves span six of its southern cantilever counterbalance, allowing span 6 to deflect midspan, rotating one or both main truss assemblies at pier 6, and severing same from the southern end of span 7.

There are probably other sequences to account for what we see, but few that also incorporate the bridge’s significant history, namely chronic and severe problems at the expansion bearing assemblies.

Of the two, I prefer sequence one, though I have no strong evidence to substantiate that preference. That evidence would be underwater. However, the main span would necessarily be subject to greater stresses, and in my opinion, subjected to greater chance of member failure due to increases in loading brought about by frozen bearing assemblies.

Important to note, either span 7 truss, east or west, could have failed first. If the east truss let go at river’s edge, most of span 7’s southern gravity loading would have applied to the west truss and could have severed it. Same applies if the SE kingpost buckled first.

On the other hand, if the west truss sheared off first at revier’s edge, the span seven loads would be applied to the east truss, possibly failing it, or concentrating load on the east pier 6 kingpost buckling it.

My preference has the failure on the east side first. Failure of the west truss would have tended to rotate the east truss kingpost to the west, not the east as we see in the imagery. East truss severed, one or two struts north of pier 6, east pier 6 kingpost buckled, or sway bracking between east and west kingposts let go, or a combination of all three was the first major components to fail, in my opinion.

Something small could have failed elsewhere, concentrasting loads in this area, but this was the first major piece of the bridge to fail, in my opinion.

If not, look back around pier 5, but that area appears to be a distant second place to me.

Main failure, east truss or sway bracing, at pier 6 or within two panels north of pier 6.

Best I have for now, I’ll stay with that educated guess till I see something new to make it unlikely.

[TXnMA]

jeffers,

Thanks very much for your efforts in posting these analysis details for me -- and for all FR posterity!

My interpretation of "B" was that it was the edge of the western deck. I am going back to Canvas to annotate what I believe to be the probable resting position of the top chord on that side.

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

Meanwhile...I don't know if you are following all the (four -- at least) FR threads where failure analysis comments are being posted, but in the thread,

Enginers puzzle over bridge collapse

Ronzo has posted some very significant photos of the rollerplate mechanisms atop the piers that I think you will find to be very informative:

1) http://www.freerepublic.com/focus/f-news/1876107/posts?page=105#105 has a superb "before" closeup of the western rollerplate assembly and kingpost base -- from ca 2002 -- showing significant corrosion there. The detail on that assembly atop the northwestern (riverside) pier is outstanding -- and informative!

2) In http://www.freerepublic.com/focus/f-news/1876107/posts?page=128#128", Ronzo has identified what appears to be the "shoe"/"rockerplate"/rollerplate" (missing from the northeastern pier top) -- lying on the ground!!.

3) In http://www.freerepublic.com/focus/f-news/1876107/posts?page=131#131, Ronzo posted a 'before" photo of the northeastern piertop and rollerplate assembly that appears to show significant corrosion there -- and the likely cause. He shows a drainpipe coming down that pier, and, just above the rollerplate, the drainpipe is disconnected and offset, allowing deck drainage to fall directly on that piertop assembly!!

By Jove, Ronzo I think you may be onto something!!! ;-)

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

I'm actually on a trip (with my wife in El Dorado, AR -- doing genealogy research) and am sending this from our motel room. Next time we "land" somewhere, I'll FTP up to my webspace some more annotated photos and comments on what else I see in this critical north shore area of collapse.

jeffers -- I have done some more differential video frame analysis on the south pier collapse, and will have graphics to send soon. do either of you have a source URL for that (by the locks) surveillance video that includes frames of the bridge before the collapse began? The only ones I've found start with the collapse already in progress... :-(

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

Again, thanks for all the great engineering analysis -- and sharp eyes! I'm convinced that FR folks will have the likely cause for failure posted in the FR archives -- well before anyone else reaches the same conclusion!!