I-35W Minnesota Bridge Inspection Report [opinion]

I just received a copy of the last bridge inspection report on the bridge that collapsed in Minnesota last week. While there is no smoking gun, it points to MANY possible failure points. Where I am coming from is this: I am a Certified FHWA bridge inspector and have additional training in fracture critical bridges (which this bridge was). I am mainly concentrating on the center section, since that is where the failure started.

The report was dated June 2006. It is 50 pages long. Interestingly, it was NOT done by a private engineering firm (like mine) while under contract to the MNDOT. MNDOT did their own inspection for their own people to review. The bridge had been inspected yearly back to 1996 and every two years before that to 1988. There was no Federal requirement for bridge inspections before that.

Although I have read elsewhere that the engineers supposedly had used “exclamation points” in their report to emphasize the importance of what they were saying, I found none in this report. It recommended yearly inspections, some small repairs, and nothing else. At most, there was a weak recommendation that “the eventual replacement of the entire structure would be preferable” (to the repairs listed). The word “eventual” does not denote any particular urgency to me.

It does list some things that should have alerted engineers to the problems, but nothing was evidently passed on higher (to the politicians that control the purse strings). For example, it says about the Main Truss Members, “The truss members have numerous poor weld details.” Then it lists numerous cracks at the ends of tack welds, at internal diaphragms that did not have outer stiffeners along the web, welding tabs left in place, plug welds, etc. These are all VERY bad when it comes to fatigue cracking.

But the worst problem was rust. There are about 20 pages of color photos, mostly of badly corroded details. There were some places that there were actually holes rusted through the metal. A combination of fatigue cracks and corrosion is death for any structure. Some of the statements are as follows: “Pack rust is forming between the connection plates.” “The floor beam trusses below stringer joints have section loss, severe flaking rust.” “Truss bottom chord gusset plate has section loss, flaking & pack rust.” “Sway bracing has severe pitting and a 3” x 8” hole due to rust.“ “Some areas (of the trusses) have section loss with holes due to rust.“ No use repeating any more. From the pictures, this is worse than any bridge I have personally inspected. The deicing system was installed in 1999, which could have only made the situation worse. Some of the floor drains dropped directly onto the truss and the corrosion is even worse there.

In addition, there was vertical and horizontal movement at several support locations. Some of the gusset plates were bent. Some had shed bolts (there were empty holes where bolts had been originally), probably from a combination of rust and force from shifting. Just a few inches shift, but that can induce large, unplanned forces into the bridge before a single vehicle drives over it. About half of the expansion joints were non-functional, too. This alone would not cause failure, but it cannot help.

They say that they remove the “plastic pigeon screens” every other year to check the inside of the trusses. They were put on because of the buildup of bird crap inside the box trusses several years ago. There was nothing said about cleaning it out so a thorough inspection could be made. A quick look-see into an uncleaned box could hide a lot. Also, this means that the yearly reports could not be as thorough as they should have been, considering the condition of the bridge.

In the back of the report are several drawings of the truss with the type of stress in each member. About 1/3 of the lower chords were always in compression. About 1/3 were in tension all the time. And about 1/3 reversed stress (went from compression to tension as a vehicle traveled over the bridge). At least that part of the bridge was well designed. The top chords were about 1/4 in compression. About 1/2 in tension. The remaining 1/4 reversed stress. The members between the top and bottom chords were alternately compression and tension.

My guess is that the failure was in probably in a member that reversed stress. That could be either top or bottom chord, but I am guessing bottom. It could have also been in a tension member. That does not narrow it down much. However, it looks like this bridge was an accident waiting to happen. If it did not fail in the spot that they finally decide it failed, it would have failed somewhere else -- and soon.

The fault was not totally with the inspectors. They accurately portrayed the bridge as a piece of crap (although I think they downplayed urgency more than they should have). I believe the fault is the people within MNDOT who got the report and sat on their hands.

BTW, the bridges built when this one was built had a combination of bad factors that made them “wear out” much sooner than planned. Three things came into being that all made fatigue a problem -- something that bridge designers never had to deal with before. One was the introduction of computers and hand calculators, which allowed more loads to be checked and the use of thinner material. Higher strength steel became widely available. A7 (30ksi) and A36 (36ksi) steel were used before that -- very ductile and low strength (thick), so that rust would not affect it as badly. 50ksi to 100ksi steel became readily available at about that time. That meant thinner material, again, more susceptible to rust. In addition, welding substantially replaced bolts and rivets. Along with bad welding details, fatigue cracks were inevitable. Although it came along a few years later, the adoption of deicing (either on trucks going over the bridge or mounted directly on the bridge) was also bad. “Stress-corrosion” cracking is what did this bridge in.

I’d be willing to lay the imagery out for you here, (have already done so in various other threads) in support for the contentious items on the list, but I understand your position in the engineering community and respect your professionalism and sense of responsibility in abiding by the conventions of same.

I’m not restricted by any of that, but I do hope I’ve been able to convey that my speculations are just that, best guesses, based on the admittedly limited pool of available data, not conclusive by any means.

Although I’m reasonably confident your posted material on the political claims that we won’t see a conclusive report was done so tongue in cheek, it wouldn’t be the first time a final report missed some aspect of a catastrophe. This has little to do with politics, it’s the nature of the situation. The investigating authorities do not have carte blanche, and are not freed from other duties and projects to concentrate solely on this one accident. Inspection time, analysis time, even processing time on large computers in simulation are all finite and limited resources.

Both the FEMA and NIST reports on the WTC collpase were less than perfect and missed some significant connections that were available. In plotting the available O2 budget for the fires, they failed to consider supplies located on floors away from the impact zone, even though burned victims in those areas make it clear that the initial fireball opened those areas to venting which could and probably did affect the overall thermal loads.

A significant collapse of several stories worth of floor systems on the northwest corner of Tower One several minuites before the onset of progressive collapse was not thoroughly analyzed in the dynamics of gravity load transfers which, along with the buckling of the Vierendiel truss structure of the south perimeter wall, led to the upper section settling almost wholly onto a very few core columns and initiating the pancake sequence.

The ASCE reports on the Louisiana Hurricane Protection levees spectacularly omitted the direct role subsidence played in the breaching of the St. Bernard Primary Protection systems.

These elements were not omitted (or concealed) for political reasons, they necessarily would have been speculative to some extent, and there was simply a limited budget that did not permit unlimited, open-spectrum analysis.

If the final report on the I35W bridge collpase does not fully answer all of our questions, it will be for similar reasons, and I fully expect that there will be enough data included in the report for us to satisfy all of our curiosities.

As to whether or not specific nodes of failure within the systems designed to prevent such disasters are identified and successfully addressed, I will await further developments and just see what happenes.

Thanks for sharing your well situated point of view in the alanysis we have cinducted here, and for some stimulating discussion, and all the best to you in the future.

Next time we cross paths, hoopefully, it will be under better circumstances.

> “A lot of your report indicates that the bridge was highly susceptible to failure by cyclic fatigue. What makes you conclude that failure was by SCC? Was failure in the supports in tension or in the compression-tension supports?”

The reason I believe it was done in by fatigue is because other bridges have failed or had to be replaced early for that reason. Most of them were the same vintage that this bridge was.

Going to thinner material and using bad welding details made the structure much more prone to fatigue failure than the bridges that came before or after. In fact, if you look at the fatigue charts for different welding details, you will see that the ones identified in this report are right at the bottom of the list. I don’t have the charts handy, but I seem to remember that the kind of welding details mentioned in the report reduced the life by 95% over a good welding detail (or, alternately, reduced the allowable stress to a tiny fraction of what was actually used for design purposes.

That is the major problem. Each of the members that make up the structure were designed for a specific load. A certain shape and area member was designed to handle that load. In most areas of that member, it is capable of taking that load (assuming the engineers who originally designed it were competent). However, at the bad welding details, the allowable stress was reduced to below the actual stress in the bridge. That gave rise to the fatigue cracks that were detailed throughout the report. Unfortunately, the average bridge engineer did not know very much about fatigue when the bridge was designed.

I cannot say if it failed first in the reverse-stress (compression-tension) members or the tension only members. That depends on the stress levels in each one, the welding details, and the existing cracks. It is impossible to tell all that from the information available to the general public right now.

Generally, reverse-stress members fail first provided they and the ones they are compared to have similar stresses. However, if the reverse-stress member had much less stress than the tension-only member, the tension-only member would fail first. Although, in theory, it is possible to have a compression fatigue failure, I have never actually seen one.

That is why competent failure reports take so long to investigate and write. Anyone can spout off in a forum (including me), but if an engineer is going on record, they had better have ALL the information available and be able to back up any and every statement made.

The deck trusses were cantilevered out past the outer edges of the twin truss assembly, and many of these cantilever sections failed on impact. Because of this, it’s going to be difficult for us here to tell whether any of those failed early or whether such failures may have been a trigger for collapse.

The deck trusses wouldn’t even have to fail to impose the kind of point loads you discuss. However...

I’m uncomfortable with the contention that a few piles of gravel or sand, and even ten or 20 cement trucks overloaded the bridge, in general distributed terms, because even though that seems like a lot of weight to a layman, it’s nothing compared to heavy bumper to bumper traffic, times six lanes, times 1500 or so odd feet of the three trussed spans.

In addition to that, the sum total of all that traffic is but a fraction of the dead load that the bridge had to carry, resisting its own weight in order to span the river.

In broad terms, we’re talking construction equipment loads a hundred or more times less than the bridge’s rated load carrying ability.

But you aren’t taking about general, distributed construction loads, you’re talking about point loads, maybe even a single point load.

At first glance, the 27 inch deep cold rolled deck stringers would tend to spread the weight of a cement truck or big pile of gravel out, so that no one deck stringer carried the load. However, each of those rested on a set of floor trusses, and the kind of weight you envision could have at most been distributed to three of the floor trusses, more likely one or two.

While the deck stringers are deep enough to carry deck loads in between floor trusses, I do not think they were deep enough, or designed to carry deck loads to adjacent floor trusses in the event one floor truss deflected out of plane, buckled, or failed.

Further, as you note, the floor trusses were cantilevered in design. They extended out well to either side of the main trusses.

Three general cases:

1. Normal traffic loads greatly exceed total construction traffic loads. Southbound traffic in the far west lane concentrates on the west main truss, and through cantilever action in the floor trusses, actually imposes a lifting force on the east truss. Northbound traffic in the far west lanes imposes loads on both main trusses. Net result, significantly greater load on the west truss than the east truss.

2. Traffic loads roughly equal construction traffic loads. In this case, the live loads are equally distributed on the east and west main truss.

3. Opposite of case 1, construction loads greatly exceed normal traffic loads. Now gravity loads on the east truss exceed those on the west truss.

Note, these are general cases, distributed loads along the full axial length of the main trussed spans.

I have a hard time believing the total weight of construction traffic on the bridge exceeded the heavy normal traffic on the bridge at collapse time, and an even harder time believing that construction loads even approached design limits of the bridge, which include dead loads (weight of the bridge itself) plus live loads, (traffic only).

However...

Couple of big piles of gravel close together, one dump truck dumping more yet, a couple more close by, waiting to dump, add in a cement mixer or 5000 gallon water tank truck and maybe, just maybe, you recreate case three on a single floor truss.

From there, it is possible that floor truss loads a rusted, fatigued, cracked main truss member, held together by cracked welds of poor welding technique, past what it can carry, and if this in one of Jim’s 57 critical members, then yes, you fail the bridge. It could all come down from there.

But...

(getting real deep in the soup here, and pushing me even further out on a limb)

I like the west main truss failing first.

Maybe.

Somehow, something vaporized the pier 6 east main truss panel, and the two main truss panels adjacent to that at the north end of span 6 and the south end of span 7. If what we see in the images atop (barely) the east side of pier 6 is indeed the remnants of the southeast kingpost, then the largest and strongest designed piece of steel on the bridge folded up like wet spaghetti.

If that’s not the kingpost we see in the images, then something up and just threw the largest and strongest piece of steel on the bridge completely out of sight. Either way, bad things happened there.

The simplest way I know of to disappear the SE kingpost is to sever the west main truss somewhere out in span 7. Now you have fully half of all the span 7 load concentrated on the SE king.

And that’s not all.

Now the west main truss can no longer act as a cantilever for span 6, balanced over pier 6, because the counterbalance of span 7 has been removed. So the west truss on span 6 begins to sag, transferring the additional load of a significant fraction of span 6 to the SE kingpost as well.

I think such a speculative sequence as this could easily make the SE king and adjacent main truss vaporize, which is what we see in the pictures, but...

Soon as you sever the west truss mid span 7, you also induce a rotational moment on the SE king, to the WEST, and we all know that the road decks fell to the EAST at pier 6.

Sooooo....the best sequence for vanishing the SE kingpost is to sever the west main truss, but doing so makes the bridge want to fall west, and the pics show it fell east.

Getting around this is complex.

Let’s say you fail one member of the east truss out in span 7, not far north of pier 6. Just for a point of reference, say it is a diagonal main truss member in tension, running down towards the base of the vertical strut under the floor truss and construction traffic point load we discussed earlier.

Pop, it’s no longer doing its job. The vertical strut it was carrying begins to sag. In a simple truss, the bottom chord nearby goes into tension, top chord goes into compression, but this was not a simple truss for its full span. If close enough to pier 6, it acts as a cantilever, top chord in tension, bottom chord in compression. Either way, you could easily envision significant axial loading on the SE kingpost, differiential axial loading comparing top to bottom, and dynamic axial loading as different parts of the span 7 east truss let go, especially if either the top or bottom chord suddenly reversed stress or was subject to a sudden jerk.

Now, you have a mechanism which might deflect or even buckle the SE king. Once it’s no longer dead vertical, yes, bad things happen. To do their jobs, compression members like to be straight. Even if you don’t buckle the SE king by jerking span 7 east truss loose, you have a severed east main truss, or it’s still in the process of severing itself, which is affecting the ability of the east truss to cantilever the north end of the span 6 east truss, again concentrating loads on the SE kingpost.

Okay, nine miles out on a half inch limb here, but bear with me, because now, you have a mechanism to rotate the pier 6 decks to the east. Not the best mechanism, but a viable mechanism.

When the SE kingpost buckles, the sway bracing between it and the SW king goes into tension and pulls the SW king to the...wait for it....east.

So, where does that leave us?

If you followed me through this rambling, highly speculative narrative so far, you can see:

1. A concentrated gravity load imposed by a LOT of construction materials and equipment, spacially localized, could have transferred to a single floor truss and thereby, could have overloaded a single, rusted, cracked, poorly welded, corroded main truss member past its tensile limits. Same goes for a gusset at an adjacent connection.

2. Severing the west main truss, midspan span 7 gives us an easy mechanism to explain why the SE king ceased to exist as a viable (and visible) entity.

3. Severing the east main truss, midspan span 7, gives us a handy mechanism to explain why the pier 6 superstructure rotated east.

Why is this important, and why did I lead you into this dark and lonely forest?

Because severing the west truss on span 7 makes the most likely case for obliterating the SE kingpost, AND gives us a viable (though still not my favorite) explanation for the east rotation of the pier 6 superstructure.

Yes, the loads of span 6 and 7 tend to rotate the SE king west if the span 7 west truss is severed, but if the SE king buckles before it leans west, the sway bracing pulls the SW king to the...east.

And that brings us full circle back to your question.

If construction loads greatly exceeded normal traffic, at any given point on the trussed span, the load imbalance would affect the east truss more than the west truss. The cantilevered floor trusses amplify this effect.

If we like the west truss severing first, then point construction loads do not help us buckle the SE king, but point construction loads do give us the best chance to rotate the pier 6 superstructure to the east.

Now...before you take this to any banks, remember, we’re talking about VERY slight POSSIBLE differences in probabilities here, in a foolhardy attempt to explain something we didn’t see, and haven’t got to look at yet in detail, in an event where bad things happened so quickly and successively that critical members were probably still failing in the span 7 truss assembly, even after it broke loose and was in free fall.

The point of all this is not to say what happened, or even what might have happened.

The point is to show you what any answer to the question you raise will entail.

The point is to show you some basic structural dynamics processes, in the hope that your gut level “feel” for how bad things happen will increase.

If somebody else posted this answer to you, I could poke a million holes in it without even going for a second cup of coffee.

But you asked, and I think it was a good question.

I like the idea of point construction loads failing the bridge.

I like the idea of frozen bearings failing the bridge.

I like the idea of bad welds, fatigue cracking, broken bolts, and lingering damage failing the bridge.

The differences in probabilities related to these possible causes are insignificant in my opinion. I don’t have enough information to say yes or no.

I won’t till the final report comes out, and even then...

But if it makes you feel any better, I don’t like resonant vibration, as a mechanism for failing the bridge.

I don’t like construction traffic overloading the bridge.

I flat can’t stand scouring as mechanisms causative in this bridge collapse.

Construction traffic point loads? Exacerbated by the cantilever design of the floor trusses, and possibly the west lane construction traffic, east lane normal traffic scheme?

Yeah, maybe.

You’re still in the running, good question.

Jim, you’ve got to shake it off, man.

I maxed a 50 ton crane, setting a lam truss 122 feet from its center pin by rocking the ball and cable two feet out of plumb, because we only had 120 feet of boom and jib.

Shattered two four by mats under the near corner outrigger, drove the splinters 18 inches into grade, and the operator pulled me off afterwards to tell me he went light on the back end during the lift.

Light, bit not over.

Broke three vertebra falling 14 feet onto green crete a week before my last heavy job ended 10 years ago. It only costs me now when I pick up my kids, and they’re too old to lift anyway.

Set ridges in wind strong enough to take the Marlboro out of clenched teeth, and at minus 47 windchill, and at minus 20 actual.

Once set a 2400 pound beam 12 feet above grade with just me and one other guy.

That which does not kill us, makes us rheumatic old fogies before our time, but it gets them up in the air.

Yep, this bridge tried to kill you, but you whupped her. That’s yesterday.

New jobs today, more yet tomorrow.

Back up on that horse and ride, or else it’s “sidewalk superintendant” status for you.

(You know I’m yanking your chain, right?)

One good thing about liberals.

Yes, they either lie or perish, (or become conservative) but they lie very, very badly.

Liars don’t last long in construction because they either take one step too many out on something they claimed they did right, and find out the hard way that they didn’t, or else somebody else does and then...well, you know what happens to liars and Nancies who get someone hurt in the field. It’s not pretty and I’ve yet to see any ‘victims’ come back for more.

So when liberals try to lie about construction, they can’t even field GOOD lies. They’re clueless.

If they try, we’ll be all over them like white on rice and they’ll end up wishing back for the good old days when all they had to worry about was Buckwheat ripping them new ones over 1990’s typefonts on documents from 1965.

They can try, but it’s not going to fly.

Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.