Bridge Faults Big and Small

The Interstate 35W bridge over the Mississippi River in Minneapolis, Minnesota that collapsed yesterday evening was an arched truss bridge; you can see photos and details here. Let’s discuss a little terminology, first, then look at the bridge survey performed in 2001:

The road surface – well, the load or traffic area, to be technical – of a bridge is called the “deck”. On a bridge like this, the part of the box truss that supports the deck can be, obviously enough, called the “deck truss”, though it’s not entirely obvious. The other portions are referred to as the “main truss” and the “floor truss”. Put in layman’s terms, the “main trusses” were the top parts of the span, the floor trusses the bottom of the span, and the deck trusses the parts spanning the main trusses and supporting the deck.

Clear? Good.

Now, according to the survey (big PDF), the bridge had problems:

Although fatigue cracking has not occurred in the deck truss, it has many poor fatigue details on the main truss and floor truss systems.

But they weren’t too bad:

The detailed fatigue assessment in this report shows that fatigue cracking of the deck truss is not likely. Therefore, replacement of this bridge, and the associated very high cost, may be deferred.

Oh, well, no worries, then. But let’s look at that a little closer:

Although fatigue cracking has not occurred in the deck truss, it has many poor fatigue details on the main truss and the floor truss system.

Wait, the main and floor trusses are in bad shape, but the deck truss is okay, so nothing needs to be done, right? Well, yeah. That’s what it says. However, it also mentions this:

Concern about fatigue cracking in the deck truss is heightened by a lack of redundancy in the main truss system. Only two planes of the main trusses support eight lanes of traffic. The truss is determinate and the joints are theoretically pinned. Therefore, if one member were severed by a fatigue crack, the plane of the main truss would, theoretically collapse.

However, it is possible that collapse may not occur if this happened. Loads may be redistributed and joints may resist rotation and develop bending moments. If the fractured main truss deflected significantly the slab could prevent the complete collapse through catenary action.

Okay – the main trusses had little if any redundancy, and were in poor shape, and the failure of one would – theoretically – lead to collapse. However, joints could resist rotation and “develop bending moments” (not a good thing). As a last wave of defense against catastrophic failure, the strength of the deck slab could have mitigated the deflection of a fractured main truss. Sounds good… except nobody considered what would happen if the deck was compromised by repairs while serving as not only a structural member, but the only thing holding the bridge up.

Part of the bridge appears to have experienced a severe “bending moment”, consistent with the predicted results of a main truss failure. Food for though, no?

Reportedly, the bridge was last inspected in 2004, and no significant problems were discovered. Is it possible, I wonder, that a main truss failed, a result of the problems identified in 2001, between 2004 and now? Is it possible that the deck slab did, indeed, maintain the structural integrity of the bridge? Is it possible, then, that this led to the cracking and other damage that was being repaired this summer? And is it possible that those repairs weakened the deck to the point that it was no longer able to resist the deflection of the main truss, with catastrophic results? I think so. Whether this proves to be the case, only time will tell.

In closing, I’ll leave you with this quote, from the report:

…the bridge could most likely tolerate the loss of a floor truss without collapse, whereas the failure of one of the two main trusses would be more critical.

Update 2 Aug 1340 CDT: Video footage of the collapse suggests the failure began on the south end of the bridge, possibly with the failure of the floor trusses at the support column there. From there, the bridge fails in two more places as the main and deck trusses are unable to handle stresses they were never designed for, ultimately separating cleanly near the middle of the main span; the north end of the span gives way moments later, though it’s largely clouded from view by the splash from the first section to hit the river.

Published in: General | on August 2nd, 2007| 6 Comments »

Both comments and pings are currently closed.


  1. On 8/2/2007 at 12:09 pm William Jolitz Said:

    Thanks for a deft analysis. One other item – could the freight train have been a trigger for the final roadway break up of a compromised structure, with the vibrations not unlike the wind with the Tacoma Narrows Bridge, perpendicular to the plane of main trusses (e.g. up/down shearing)?

  2. On 8/2/2007 at 12:40 pm Nemo Said:

    It’s possible, but I would guess only as a secondary or even tertiary factor. The train tracks ran under the north end of the bridge, which is the end structural engineers, according to reports, had concerns about the soil at in the late 1980s when the last two traffic lanes were opened. It’s worth noting the railroad tracks predate the bridge; indeed, predate the interstate, and are used much less now than they would have been in the past. On the other hand, the Twin Cities are in one of the worst droughts for a couple decades, but I’m not positive how much of an effect this would have on soil strength or transmissiveness.

    The train that got crushed by the bridge wasn’t moving when the collapse took place; I’m not sure a locomotive was even present, and that it wasn’t just a string of grain cars waiting to be moved.

    Updated: With the failure now known to have first manifested at the south end of the bridge, I’m even less inclined to consider train traffic as a contributing factor, though it probably can’t be ruled out completely.

  3. On 8/2/2007 at 3:26 pm B Roberts Said:

    Looking at the footage, and after effect photos, the road bed of the southern approach bridge to the main span has fallen to the east side of the support column. While photos of the north column show it leaning south towards the river. Also the remains of the steel arch in the viscinity of the southern support appear to be twisted the most, suggesting the most torque was applied there.

    Is it possible that approach bridge collapsed at the southern support column? If so, with the southern approach bridge suddenly collapsing is it then possible that the main span shifted south by virtue of it’s mass alone, ripping it from it’s northern column cleanly and making it drop like the video shows?

  4. On 8/2/2007 at 3:49 pm Nemo Said:

    B – I wouldn’t think so, but everything is so interconnected that it’s hard to say. The north column does seem to be shifted towards the river, but the floor truss assembly that sat on it fell to the north, i.e. away from the river, and on top of it; I’m inclined to believe right now that the lateral forces of that weight are responsible for the shift that seems to be visible.

    It’s very possible that the approach bridge on the south end collapsed at the support column at the periphery of the truss bridge, but I don’t completely agree that the pattern of damage visible is consistent. The south supports of the truss structure appear to have shifted south of the pillars they were on (so that, in essence, the two “ends” of the truss bridge fell outward, away from the middle of the river), and my first inclination is to consider the collapse of the approach spans at the south end as secondary failures caused by the collapse of the truss, rather than vice versa.

    I believe that the preliminary information supports a strong argument that the truss bridge failed first; being arched, as it failed, there were incredible horizontal forces produced along the axis of the bridge, as the arch attempted to flatten, for lack of a better word. This severed the bridge deck and the truss structure near the middle, as metal and concrete were asked to bend, flex, and stretch in ways far beyond their capabilities. The south end, perhaps being the weakest (the origin of the failure?) then fell, moving south and east, followed moments later by the north end, which fell straight down but to the north. The decking and possible even supports of the approach bridges were in the process impacted by the “down and out” force of the truss collapsing, and themselves then failed in a sort of domino effect.

    On the other hand, it could have happened like you describe, but both ends in the process pivoted vertically on their supports and slid away from the river, off the support columns. I’d expect both supports to have fallen in the same direction, to the north, in those circumstances, though…

  5. On 8/3/2007 at 1:05 am Bob Denney Said:

    Can the curve af the bottom of the truss be called an ‘arch’?
    The video of the bridge collapsing seems to show the ‘arch’ becoming straight. Which would not seem to be very strong.

    Seems to be a very weak design.

  6. On 8/3/2007 at 2:35 am Nemo Said:

    Well, I’d call the bottom “arched”, because it technically is; a regular, non-arched truss has parallel upper and lower surfaces. While it does kind of look like it becomes straight during the collapse, I think it might be a trick of perspective. If it did, how much was gross distortion of the structure, and how much was subsequent sections of the trusswork separating from each other, I don’t really know. I’d bet on distortion, probably in two planes… but that’s really just a guess.

    Either way, it was definately a weak design; accoring to local television (KARE), it was one of just four extant non-redundant arched truss bridges in the state.