Bridge Collapse Caused By Design Flaw, Not Maintenance

Captain's Quarters ^ | Jan. 15, 2008 | Ed Morrissey

[jeffers]

This image shows the 2004, 2005, and 2006 relevent sections of the bridge reports side by side for comparison. As you can see, these excerpts are identical. You can make further comparisons on your own by downloading the relevent documents from here:

http://www.dot.state.mn.us/i35wbridge/history.html







That is NOT to say each report is fully identical with all the others. Some have pictures, and there are sections in most or all reports which DO appear to have been updated from previous years. However, there are just as obviously cut and pasted sections, and since some of these concern the critical mainspan failure point, it is something that needs to be investigated.

Also, from the 2003 report, I found reference to graffiti which had been painted on the U10 East truss gusset plate. There is an access ladder attached to L10-U10, as you can see from previous images, and there are reports that un-authorized persons were climbing up into the superstructure from a different section of the bridge.

It is possible, perhaps even probable, that someone would have painted over an obscene word or phrase, perhaps not taking the time to match the paint color.

For the outlines of paint which might have been used to cover graffiti, however, to match the fracture lines which dropped the mainspan into the river, stretches credibility further than I am willing to just gloss over.

There are explanations I might accept for some of this coincidence, centered around East Truss panel point U10. Obviously this is a critical point in the mainspan, along with three other symmetrical and similarly located points, where the effective cantilever changes over into a simple truss. The top chord member immediately south of U10 is in tension, whereas the top chord member immediately north of U10 is in compression.

This is an important consideration. The weight of the bridge is pulling the top chord towards midspan for the first two panels of the truss riverward from the piers. Between those two points, beginning one panel riverward from U10, the bottom chord is in tension, carrying a large percentage of the weight of the rest of the mainspan. That tension is transferred up the L11-U10 diagonal to the U10 gusset, and thence along the top chord back to the piers and the counterbalancing opposite sidespan.

Clearly, this is why U of M placed strain sensors there, and at the matching symmetrical structural points around the bridge. This MAY be why there is an access ladder there. It is distinctly POSSIBLE that a graffiti artist chose not to reach further than from the top of this ladder, for fear of falling or just plain laziness.

But unless it can be shown that the repair paint was corrosive to some degree, I find it difficult to explain why the fracture lines so closely match the discoloration perimeter.

Sheer coincidence?

Perhaps, but people died as a direct result of this "coincidence". The damages total in the billion dollar range, or will by the time the survivors are paid off and the bridge is rebuilt. The East Truss U10 gusset was one of the very first sections of steel to be cut free, barged, and hauled away after clean-up and analysis began.

Questions like these, coincidences like these, when they involve the loss of human life and hundreds of millions of dollars, should not, in my opinion, be swept under the rug, they need to be examined VERY closely, preferably on the record, in a sworn statement.

[spunkets]

Thanks for the ping.

Just a note about the paint... The black color where the plate was torn away is an oxygen deficient Fe corrosion product. It's between FeO and Fe₃O₄. It's stable in that crevass, because of the pressure and lack of O₂. The normal red oxide is Fe₂O₃. I don't think there's any areas that weren't covered with paint. The stains are on the surface of the paint, except in a few small areas where the paint lifted. Fe corrodes by first producing Fe²⁺, which is water soluble. The Fe²⁺ remains in solution until it's oxidized by O₂ from the atmosphere.

It's clear the drain dumped water on that gusset area and kept it wet. The corrosion occurred around and within the rivet crevasses. Fe²⁺ would wander out as a corrosion product and precipitate as the the water flowed down the struts, after being oxidized by atmospheric O₂ to Fe³⁺. That's how the red rust color on the paint got there.

[B4Ranch]

I don’t know where to locate the pictures you mention. Your theory does make sense to me.

[TXnMA]

"My gut instinct tells me..."

Still chasing the "scour" ghost, eh? I thought we put that theory to rest shortly after the failure... :-(

IM(NS)HO, the tilt of Pier 6 was what we in failure analysis call a "verified ugly" -- an obvious artifact that distracts those who are prone to jump to conclusions from the real point and cause of failure. Again, IM(NS)HO, Jeffers has it "nailed".

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

In the period immediately after the collapse, I did extensive, frame-by-frame time-lapse analysis of the security cam video taken of the pier 6 end of the bridge -- looking across the river from the locks area at the pier 6 end of the main span. And, IIRC, I copied you on those findings...

Although the splash/spray (after the main span hit the water, of course) obscured much of the pier itself, I detected no riverward shift of the top of pier 6 until after the mainspan had collapsed, and after the splash had subsided and after failure of the side spans on the pier 6 side was well underway.

IOW, the tilt of pier 6 was one of the very last things to occur -- not the first.

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

In failure analysis, "gut instinct" can lead you into never-never land. Try presenting "gut instinct" as your data some time at the Reliability Physics Symposium -- but put your earplugs in first... '-)

[1COUNTER-MORTER-68]

BTTT...

[TXnMA]

I don’t know where to locate the pictures you mention. Your theory does make sense to me.

That is because exit82 got the pier numbers (6 & 7) reversed.

And, like a dummy, I didn't double check his nomenclature before I replied to him in #64!! :-(

In my post #64, please replace all incidences of the numeral, "6" with the numeral, "7".

Thank you!

[exit82]

I’m not a structural engineer.

But why did Pier 6 move at all? It is not only a tilt, but a dislocation. If it was anchored properly, it should not have moved the way it did, but rather collapsed in some way. The center span collapsing most likely would have separated at the expansion joint at the approach span, at both ends.

The longer part of the collapsed center section started from Pier 6. The shorter section of the span came from Pier 7. Pier 6 has its rocker bearings and plates intact atop the pier. Pier 7 has the piertops shorn off where the bearing plates were.

The longer section of the bridge fell pretty much in line vertically. The shorter section is off to one side of Pier 7 by a fair degree.

What does that indicate? To me it indicated the presence of a tremendous horizontal force,possibility building over time, not just a failure where gravity then took over.

Could the bridge after failing at U10 pushed toward Pier 7 as it collapsed? Sure it could.

And the moment force at Pier 6 could have been sufficient to dislodge the entire pier structure? Possibly.

The bridge was inspected for scour. But the visibility by Pier 6 was always almost zero due to turbidity caused by the extreme current of the river. Even post collapse recovery efforts were hampered by the extremely low visibility in the water.

I am having trouble with the article’s explanation that the gusset plates were supposed to be twice the thickness. The bridge should have collapsed as soon as the number of lanes was doubled from four to eight years agoif that was the case.

The bridge was designed in the 1960s when design was very conservative, so adequate safety factors should have been involved.

Jeffers submits that U10 failed due to corrosion—that would make more sense than this article.

But until the subsurface condition of Pier 6 is investigated, I cannot rule out the possibility of its role.

That’s just my opinion. Sometimes Occam’s Razor is right and sometimes it is not.

[exit82]

I thought I got it right from B4’s post 21’s pictures.

But I didn’t. I reversed the numbers. Same in preceding post.

Thanks for pointing that out TXnMA.

[B4Ranch]

>In failure analysis, “gut instinct” can lead you into never-never land. Try presenting “gut instinct” as your data some time at the Reliability Physics Symposium — but put your earplugs in first... ‘-)<

‘-{

[toddlintown]

One day Bush drove by the civil engineering firm that designed the bridge; ergo, it’s Bush’s fault.

[B4Ranch]

While I am not sure who would be responsible or interested in your analysis, I do think that you should send it to the appropriate people for further study.

Thank you for your efforts

[TXnMA]

"But why did Pier 6 move at all?"

~~~~~~~~~~

I know you know this -- but it was pier 7 that moved... '-)

When I first saw the tilt of Pier pair #7 -- and the (intact) condition of its "shoes" -- and the position of the deck leaning on Pier 7, I had the exact, same concern you did.

However, I took the time and effort to do a lot of analysis -- to prove my own first impression to be wrong...

The collapse of the deck landward of pier 7 was very sudden and very rapid. That appeared to me to be the force that shoved the top(s) of pier 7 toward the river. I don't have the video on this machine, but, IIRC, that event occurred some fifteen seconds after the bridge started collapsing.

You are probably correct that the proximity of the piers' base to the river allowed the pier-base displacement to occur -- and that scour may have played some role in that weakness.

However, I see no evidence that said weakness contributed at all to the initiation of the failure.

[UCANSEE2]

Let’s say you have a truss that has rusted over the years,
and shows a stain.

Other trusses may be rusted as well.

Now add extra weight (more lanes)
Now add extra weight (construction equipment)

Now....add.... JACKHAMMERS.

I have seen SEVERE VIBRATORY OSCILLATIONS do incredible damage to structures.

Why not here? (they were using jackhammers, but that didn’t seem to be mentioned)

[jeffers]

This is complicated, but here goes. The earliest failure I can detect took place at U 10, east side.

Why?

Because it’s at the bottom of the pile. There’s one other piece of bridge, at the bottom of the pile too, in a different place, a short section of sidespan quite a bit south of the river, specifically the north end of the road deck that came to rest vertically, that the southbound semi crashed into, and that the southbound school bus almost crashed into.

Either could have been the initial trigger for the event. I prefer the mainspan trigger, because more bridge is involved, but from this distance, given the imagery available, I can’t say either way with any certainty.

For now, I consider the original trigger, one of the two specified above, to be moot.

The key to understanding the failure sequence, once U10 was not delivering in function, after it tore, is the understanding of the nature of the bridge itself, as designed.

At the highest level of simplicity, there were three bridges, from south to north, a standard beam and post highway overpass type bridge, a trussed span over the river, and another beam and post highway bridge at the north end.

At the next, more complex level, we look only at the trussed span. Strictly speaking, there were three trussed spans, supported by four double piers, from south to north, pier 5, 6, 7, and 8, but even this is misleading. The real meat of the nature of the bridge design comes now.

Instead of three trussed spans and four pairs of piers, think of four, T shaped piers, with very short sections of road deck connecting the horizontal ends of each T shaped pier.

The superstructure of the bridge was designed as a complex combination of a cantilever and a simple truss, such that the first few panels of truss, on either side of any given pier, are in balance with the oppossing first few panels on the opposite side of the same pier. In a perfect world, you could remove all but three panels of truss on either side of one pair of piers, and they would stand there in perfect balance, because that’s precisely what they were designed to do.

If, however, you remove one bolt from either side, the structure becomes unbalanced about the pier and will tilt off to the other side like an unbalanced teeter-totter on the playground.

With this in mind, we can now see the failure sequence.

1. U10 east fails, creating two significant problems for the entire east truss, basically everything south of the failure, and everything north of the failure. U10 is at the north end of the pier 6 “Tee”.

2. The south end of the east truss atop pier six now carries itself plus the short section of southern road deck, while the north end, separated by the U10 fracture, only carries itself. Like the unbalanced teeter-totter, the north end begins to rise, while the heavier south end drops, the whole rotating about the top of pier six.

3. The south end of the short section of road deck still attached to the pier seven “Tee”, is not designed to be self supporting, nor is it carried by pier 7, so it also begins to fall. Remember, so far, we are only looking at the east truss.

4. While the superstructure above pier 6 (east side) is rotating about pier 6 due to the unbalanced forces around pier 6, the short section of road deck still hanging from the pier 7 “Tee”, (actually a pretty long section of road deck, the full length of the mainspan minus 4 panels) sags enough to put most or all of its load on the west truss, centered at pier 6.

5. The west truss around pier six cannot sustain the additional load and begins to distort. The connections between the short section of mainspan road deck and pier 7 cannot hold the mainspan up either and shear off. Things are happening very quickly, and even a supercomputer might not get the sequence here exactly right.

6. The center span road deck is now in freefall two panels south of pier 7, in freefall on its east side two panels north of pier 6, and all or in great part, loading the west half of pier 6. The entire superstructure, the road deck and both trusses, begins to lean to the east, but cannot move laterally to the east, because it is still connected to pier 6 via the west truss. It twists instead.

7. This twist imposes a rotation onto the “tee” supported by pier 6, including both the east and west truss. The lateral crossbracing between the two trusses fails, and almost, but not quite simultaneausly, the U10 and L10 connections on the west truss let go.

8. The mainspan, now in complete freefall, drops straight down into the water, coming to rest where we see it in the post collapse imagery. The superstructure above pier 6, continues its sideways rotation, to collapse with the west truss atop the east truss, and it takes the north end of the pier 5-pier 6 span eastward with it.

9. The superstructure atop pier 7 doesn’t know it’s dead yet and remains standing for several seconds. But it too is unbalanced. Its south end carries nothing, while its north end carries the south end of the pier 7-pier 8 span. Slowly, it begins to rotate, both trusses in unison, river end up, shore end down, another unbalanced teeter-totter.

10. But there is a critical difference between the pier 6 assembly and the pier 7 assembly. The pier 6 superstructure rests atop a roller bearing nest, basically some very large steel rolling pins. This is done so the bridge can expand and contract as seasons change, sliding north and south atop pier 6. Pier 7 however, is fixed, pinned in place, the bearing plates are fixed, with no rollers, directly atop the concrete of pier 7.

11. Had this been simple beam and post construction, with a span depth miniscule compared to span distance, the pier 7-pier 8 span would have broken and fell, without greatly disturbing pier 7, even though there were no roller bearings at pier 7. But it wasn’t. The superstructure thickened atop the piers, and simple geometry, as the span sagged, pushed pier 7 riverward as if the earth holding it in place didn’t exist.

12. At some point in the sag of the north sidespan (Pier 7-pier 8), and the rotation of the top of pier 7, the pier 7 bearing plates disconnected. At that point, pier 7 itself stopped moving, and the superstructure slid atop pier 7, coming to rest as seen in the post collapse imagery.

The above is as close to a non-engineering failure sequence as I can manage. Jim Trent posted a must read opinion, from the point of view of an experienced inspector/designer, here:

http://www.freerepublic.com/focus/news/1879030/posts?q=1&;page=1#1

Since I’m not a professional engineer, and enjoy the professional freedom that allows without retribution, or penalty for error, I posted some deeper analysis and speculation, from the point of view of an engineering trained, longtime construction field supervisor, here:

http://www.freerepublic.com/focus/news/1884493/posts?page=1

The latter has imagery depicting the failure sequence, though the description may be more technical in nature.

If you still have questions, you can look deeper at the sources above, or ask them here and I will try to answer them.

[exit82]

Thank you for your very civil reply, and for your analysis.

Having done some legal forensic work, I tend to question assumptions only in the hope of arriving at the truth.

Many times the obvious conclusion is not the correct conclusion. Questioning sometimes leads to a different path that would not be pursued initially.

And then sometimes, the obvious is just that, the obvious.

Thank you for advancing the discussion, and for correcting my reversing of the pier numbers.

[jeffers]

You are right, of course, about the anti-corrosive coating. I’ve never seen bare steel raised in an exposed design. It’s always covered with something.

In saying the U10 gusset “wasn’t painted”, my meaning was “wasn’t painted green, after assembly”, as opposed to “not painted, therefore bare steel”.

I appreciate you taking the time to add the oxidation resistance information, which I was just too otherwise directed to do.

Not sure if you agree or disagree, but I do NOT see rust on U10, before or after the collapse. I do NOT see dampness there either. It’s something else, because dampness would have gone away after the collapse, or spread considerably, This discoloration kept the same shape before, during and days after the collapse.

At the risk of being repetitive, the discoloration very closely follows the fracture lines too.

[Kickass Conservative]

If a pigeon poops, raise taxes.

[exit82]

Jeffers, your anaysis is very complete, and I thank you for providing it, as well as those which you have on the other threads last year.

I am very impressed by your ability the identify parts of the wreckage, and to sequence events in almost real time in a way that is understandable.

I know that ASCE is very interested in hearing input from their members on this bridge collapse in response to this very article.

Won’t you consider emailing your research to them? I think they would find it very helpful, as they are big into infrastructure improvements nationwide, working with Congress, etc. and this collapse is certainly the hot topic now as the poster child for all that ails the declining condition of the nations highways and bridges.

Your practical experience as a construction supervisor yields a viewpoint many engineers are lacking in, bridging(no pun intended) design and field erection. Sometimes the two don’t resemble each other.

[jeffers]

I completely agree.

This report is about like saying the bridge failed, because it fell down. It explains nothing, and it conceals much.

The gussets, as designed, held fast for decades. They held four extra lanes of additional traffic, after the expansion project. They held concentrated point loads of construction traffic, a hopper truck and apparantly a concrete applicator truck, just north of the failure point.

The gussets, as originally designed, were re-analyzed by the U of M study, and the original design criteria were used in comparison with the actual strain measurements gathered under that study, and the traffic loads DID NOT APPROACH DESIGN LIMITS.

Like you, I am appalled at the visual condition of that bridge, from end to end, as shown in the inspection report photos.

That bridge was neglected, for decades. serious problems were neglected, for decades. In spite of that, the bridge held four additional lanes of extra traffic, and held additional construction loads too. The I-35 bridge did not fail Minnesota. MnDOT and Minnesot politicians abdicated their responsibility, and failed the bridge, along with the people who died using it.

What we see here, in my opinion, is a perfect example of the wolf guarding the sheep. I smell a fix, I do not trust or believe this report’s conclusions, I believe it is an attempt to sweep the true failings under the rug, to avoid responsibility for professional and political neglect, and believe further, that many, and quite possibly, ALL OTHER BRIDGES IN MINNESOTA ARE SUSPECT.

Minnesota has a HUGE problem right now. The people paid to watch and take care of your bridges, are not.

Instead of fixing problems, they pretend there are no problems, even now, today, after the collapse.

Minnesotans, you need to speak up, and don’t stop until you get action.

Your lives are at risk.

[MaxMax]

The Rats owe Bush an apology.