Any engineers who can analyze this report on fatigue cracking I35W Bridge?

[RamboLiberal]

http://media.startribune.com/smedia/2007/08/01/21/bridges200110.source.prod_affiliate.2.pdf

[frogcycle]

1. yes, but what are you looking for?

[RamboLiberal]

4. Your viewpoint if anything in this report could point to reason

for collapse.

[frogcycle]

9. The analysis looks sound enough

They made a somewhat simplistic finite element model, made some assumptions that could be argued.

But the numbers they got support the conclusions. And they recommended periodic inspections that should have been appropriate.

In a highly-redundant structure such as this truss, some fatigue cracks growing around rivets here and there are not going to cause catastrophic failure like this.

The analysis did not address whatever the cause was.

Back when I was designing railcars and using finite element analysis I used to use the analogy that I had a sound structure from static analysis, and with the assumed dynamic loading it should last forever from a fatigue standpoint. So the ones coming into the shop all falling apart obviously were not following the rules re dynamic loading. They were clearly sneaking off when nobody was looking and banging into each other!

This sort of catastrophic failure, in my opinion, came about either because somebody made a gross oversight in the fundamental design - that is, the finite element model did not represent the real structure (very, very unlikely) - or there was loading, probably repeatedly, far outside the specs.
A whole lot of grossly overloaded trucks could do that over the years. But you'd still be likely to see the problems appearing as small cracks, long before a catastrophic failure. Unless the report is incorrect in stating that the critical members are readily visible. that , again , would mean somebody didn't know a damned thing about truss design (very, very unlikely).

My opinion is the pile driver that was pounding away next to the bridge probably is what tipped the scales. Possibly by aggravating existing incipient fatigue damage that would otherwise have been found on inspection and not been of major concern ( a redundant structure tends to "compensate" for localized weakening - like a sagging barn roof that remains standing for years) and causing a sudden failure of a critical component that then 'shocked" the neighboring components. It could possibly turn out that the construction crew screwed with load bearing members thinking they were just guardrails or something. Also not likely.

More likely, and this is purely wild speculation by me - the pile driver hit something subterranean that shifted the nearby footings, caused the roller bearing support to move and a panel slipped off its support. Then it came down like a house of cards. May sound farfetched, but suppose there was erosion around the footings, suppose currents had cut a channel right by the base. I have no idea whether there are regular inspections of the underwater parts of bridges. Suppose there was a subterranean cave that was not flooded and they punched through. The whole riverbed could then have collapsed.


I guess what i am trying to say is the failure suggests to me more that it had its legs kicked out from under it, as in an earthquake, then that it crumbled from deterioration. Once one panel went, the jolt to the supporting structure was enough to jerk the legs out from under the next panel, and so on.

[malaise]

14. Please explain pile driving for me n/t

[dipsydoodle]

16. If you look at the columns,

two each side of the river, they go down deep into the ground. They were probably created by driving steel deep into the ground and then surrounding them with concrete. It's a pile driver which hammers the steel into the ground by a slow percussive action.

Something like this :

[malaise]

18. Thanks a lot

I learn something new on DU every day. :D

[frogcycle]

20. I have heard speculative comments about the jackhammers and passing train - that is bs

if that thing was so close to shattering that the high frequency chatter of a jackhammer against concrete was sufficient to make it go, then a sneeze would have too. That would mean that any half-assed inspection in the last six months should have called for immediate closure. And trains don't shake the ground all that much. They rattle a lot, but the wheel-rail interface is not like the bouncing you get with trucks on the highway.

People also talking about the fact that it was bumper-to-bumper and wondering if that overloaded it. That is also crap. Design specs for things like this would call for it to take as many heavily-loaded trucks as could squeeze on it, with a big safety factor.

The pile driver, on the other hand, generates very-high energy low frequency shocks in the ground. BOOM!.....BOOM!.... Remember the scene in Juraissic Park where the glass of water is vibrating from the approaching dinosaur? The waves propagated from that would travel into the supports and up the structure. Every jolt could be applying loads to joints that they weren't designed for. Picture a movie where a submarine has gone down too deep, and seams are starting to pop. If there were uninspected corroded/cracked joints and that had been going on for a few days...

[ProfessorGAC]

17. Excellent Post, Frog

Technically very sound and learned, and explained very well. Nice job. I'm a scientist/mathematician, not an engineer, so i don't know much about bridge loading. But, i know more now than i did. Thanks.
The Professor

[frogcycle]

21. thanks - but I should repeat my disclaimer

I am mostly guessing from sketchy incomplete data.

I have heard that there were comments in the more-recent reports about a lack of redundancy in the design. If that is true, then the problem goes back to 1967, and somebody is a flaming idiot.

A fundamental in designing something like this is to look at failure modes. You design it not to fail, but then ask "what if?" You have secondary and tertiary plans - whether its a structural design, or a system design. For those components that are "single point of failure" you have a strategy to shut down, or abort the mission, whatever, withouf catastrophic failure.

Katrina would be an example of NOT doing that. You build levees, install pumps, etc, and theoretically have a fail-safe design, then you say "but what if the storm is THIS big and the levees are breached?" that's why submarines have watertight compartments, and why NOLA should too (see my plan on that topic: http://www.dbc3.com/NOLAPlan). But I digress.

At some point you have to make a trade-off. NASA says "ok, we have done everything we can, tried to anticipate every eventuality, have backup systems, backup plans, and backups for the backups. We still expect to lose a shuttle every x number of launches."

For space flight you have to do that. For NOLA, lacking a complete rethinking and redesign (again, see my website), you also have to do that. For the Space Station they have escape modules. For the shuttle, they know they'll lose the crew. For something like this bridge, you really don't have to do that. A proper inspection and honest appraisal of its condition, should, in my opinion, have found it to be on the verge of disaster. Or the inputs were so unexpected, so out of spec, that it is a freak incident. In which case it should have had sufficient redundancy that if, say, one major support beam broke loose from the ground support at one end, then that portion of roadway would have failed, there would have been a minor disaster, but the span would have stayed up = like an airliner flying on one engine with two out of commission. It would have had to be condemned, to be sure, but the whole damned thing would not have dropped like a rock. Either grossly risky design, or gross lack of inspection, or gross bad judgement on the data.

I expect all three, with the latter being driven by some twenty-something Liberty University grad.

[karlrschneider]

2. I took a cursory look at this, and it started out fairly damning.

Seems to be saying (I will read it more carefully when I have some time) it was a somewhat questionable design and not put together very well.

[frogcycle]

12. it is a somewhat unusual design

using steel beams as pillars

can't say that I've noticed a lot like that.

But with sufficient redundancy and lateral stabilzation of long beams, there is nothing to say that can't work.

The beams aren't the issue, though. The connections are where the stresas concentrations and localized reversals occur due to cyclic load. I have a real hard time imagining that the connections were so damaged across the full span as to just "let loose" like that though. With ANY periodic inspection, there should have been observation of "gee, this rivet hole is elongated", and "there are "hairline cracks at the tips of these welds," etc.

I think something really bad was going on out of sight - probably under water.

[madokie]

13. didn't I read there was no supports in the water?

maybe at the edge of the water,

[frogcycle]

19. yes, I just heard that

the schematic in the analysis didn not make that clear and i just assumed...

[frogcycle]

22. from comcast homepage:

"It was built with a single 458-foot-long steel arch to avoid the need for piers that might interfere with river navigation."

Funny, the bridge right next to it has a piling in the middle of the river - I assumed this one did too - but what is the logic? The other one interferes with river traffic as much as this would have. I wonder which was built first?

I just looked at the picture in the doc in OP again - and now see that at a glance the piling for the OTHER bridge looks like it is for this bridge.

[bleedingheart]

15. It could have been the sum total of some other factors as well

as you suspected there could have been a problem with the structural support of the bridge in the water.
Rivers and erosion go hand in hand. Any number of things could have weakened the concrete bridge supports in the water.

That could have been the starting point the video I saw on CNN seems to lend itself to that a bit.

Once a main support gives...then there really isn't much that is going to stop it from just collapsing as it did.

The bridge design is one I have seen before (here in Pittsburgh..where we are called The City of Bridges)...

[tularetom]

3. Wow it's a pretty long report

I just read the executive summary. I'll read the rest tomorrow. A few thoughts however:

fatigue is another term for stress reversal. Take a short piece of wire and pull on it. You can't pull it apart. Now bend it back and forth four or five times. It snaps. You have reversed the stresses in the wire by bending it and returning it to its original shape. This is what happens to the support members in a structure like a bridge. As a heavy truck passes over a particular segment of the bridge it puts stress on the members supporting that segment. And those members flex slightly. After the truck has passed, the member relaxes and returns to its original shape. Multiply this by millions of trucks and the steel that makes up the structural member begins to lose strength. There is a finite number of these stress reversals that any steel member can stand before it fails.

Usually there are enough members making up a structure (especially in a truss bridge like this one) that one could fail without an overall structure collapse. But I noticed in the summary a comment about the lack of redudancy of the truss members. In the worst case this might mean that there is literally no backup for a failed truss member and one overstressed member could cause the collapse of the whole bridge.

And this makes me want to read the rest of the report.

[karlrschneider]

6. The truss configuration is what got my attention as I mentioned in another post

there doesn't seem to be many structural members that utilize tensile strength...almost all of them are in compression. Not a great design. Oh, the phenomenon you alluded to is commonly called 'work hardening' which is something non-engineering types might recognize. ;-)

[Bolo Boffin]

5. Ouch.

In March 2001, the report said that "the bridge should not have any problems with fatigue cracking in the foreseeable future."

That's plain enough. However the report did recommend an inspection regime, and that is what likely caught the problems they've been trying to fix over the summer.

One conclusion stands out to me: "Live load stress ranges greater than the fatigue threshold can be calculated if the AASHTO lane loads are assumed. The actual measured stress ranges are far less primarily because the loading does not frequently approach this magnitude. While the lane loads are appropriate for a strength limit state (the loading could approach this magnitude a few times during the life of the bridge), only loads that occur more frequently than 0.01% of occurrences are relevant for fatigue. For this bridge with 15,000 trucks per day in each direction, only loads that occur on a daily basis are important for fatigue."

The first page of the report (pdf 11) says the ADT is 15,000 per day with ten percent trucks. A mistake in this paragraph really doesn't bode well. Actually this mistake is on page 2 of the report as well (pdf 12), where the 15,000 trucks in each direction are ridiculed (the "details should have cracked open soon after opening if the stress ranges were really this high"). That looks really nasty.

There are two standards in play for this bridge. One is AASHO, the one the bridge was built under (called non-conservative by the report). The second is AASHTO, which were developed in the 1970s. Perhaps the AASHTO standard is to calculate stress loads as if all expected traffic were trucks, and that's why they're citing 15,000 trucks each way.

Yikes, this looks awful to this layman's eyes.

[bridgit]

7. those were the sorts of things hubby had mentioned; what the report refers to as:

'unanticipated out-of-plane distortion of the girders' keying upon the general condition of the deck they do go further: 'Concern about fatigue cracking in the deck truss is heightened by a lack of redundancy in the main truss system' concluding rather flatly under the circumstances it would seem 'Therefore, replacement of this bridge, and the associated very high cost, may be deferred.'

hm...deferred to when? Cause the time to retro-fit a system of supports to the truss system seems to have come & gone...a disaster

[IDemo]

8. self-delete

[ShaneGR]

10. Well, that professor has some explaining to do...

They said the deck truss most likely wouldn't crack, even though the girders had exhibited poor loading capacity. But according to what I watched tonight on the news, cracks had been reported since that report was done in 2001. So, that whole reports conclusion was evidently terribly wrong.

[annabanana]

11. kickin' in the am. . . . . . . n/t