Miami Pedestrian Bridge, Part XIII 81

[JAE]

A continuation of our discussion of this failure. Best to read the other threads first to avoid rehashing things already discussed.

Part I
thread815-436595

Part II
thread815-436699

Part III
thread815-436802

Part IV
thread815-436924

Part V
thread815-437029

Part VI
thread815-438451

Part VII
thread815-438966

Part VIII
thread815-440072

Part IX
thread815-451175

Part X
thread815-454618

Part XI
thread815-454998

Part XII
thread815-455746


Check out Eng-Tips Forum's Policies here:
faq731-376

 

[Sym P. le]

I do not see what would cause a lateral torsion (east-west)

I'm not indicating lateral torsion, it is longitudinal (the axis of rotation is lateral through the lower diaphragm, centered on the 8 heavy rebar). Consider 12 extended to the base of the diaphragm with 11 tied into it. Both are free of the deck. You can push/pull the top of 12 longitudinally and torque the diaphragm. This is the freedom of movement when the weak surfaces let go.

I'm saying the the EOR erred in his design process and in his evaluation of the message that the cracks were giving. I have not expected it to be popular but eventually people might wake up.

Regarding the reversed images, I took the images from the posted and linked slide and applied them to this situation. The torque from 12 will impart opposite shear patterns to the left and right connections with the diaphragm. These are mistaken as punch out, shear stress, vertical load etc. but I believe my presentation is compelling. All things considered, this structure was very tough to bring down.

This is not to disregard the plethora of stress loads that the node and surrounds experienced but the signature torque failure pattern has not been addressed.

My previous posts (amoungst others) on this page, 23 Sep 19 00:09, and on page XII, 25 Jul 19 02:18, provide more development on this theory, including the lower PT rod as a control mechanism.

 

[hpl575]

Retensioning the rods, where do I begin, several threads ago I posted a comment to similar to, "I hope FIGG's decision to retention the PT rods in Member 11 went beyond, 'Seems like it was performing better before the rods were detensioned, lets put it back on'". Reading the FIGG employees' post incident NTSB depositions, it would appear not.

Pate, "..., it seemed prudent to try to get back to that state of stress that we had when things were good."

First, the extent of the cracks had changed significantly since the bridge was setting on piers with rods tensioned.

Second, the state of stress changed significantly after the rods where detensioned, restoring the state stress would seem a long shot at best.

Third, observation of the cracks at the end the deck would indicate that the top bar, or north bar in some reports, could not be reliably restored, pulling on the opposite side of the cracks would only further fragment the concrete.

Last, at the angle of Member 11, more shear is created than friction resistance, even at shear friction coefficient of 1 for roughened concrete.

 

[Vance Wiley]

Last, at the angle of Member 11, more shear is created than friction resistance, even at shear friction coefficient of 1 for roughened concrete.

Great point. Not only does the angle result in more sliding than normal force from tightening the PT rods, the normal force is only 60% effective or 1/0.6=1.66 times worse than the angle of 31.8 degrees. Using the tan of 31.8 degrees as 0.620 and multiply by 0.6 = 0.37 vert effective (to use coeff of 1.0)with a horizontal component of 1.0 horiz sliding created. That looks like a 20.3 degree angle. So one PT rod at 280 kips X cos 20.3 = 262 kips effective sliding. Two PT rods - wow.

Of course at the time the EOR thought he had a coeff of friction of 1.0 to work with.
Great post.

 

[Earth314159]

The diaphragm really isn't a torsional member. What you are seeing there is a punching shear. However, what it comes down to is compression struts and tension loads on the rebar. There are only compression and tension stresses. Shear stresses are tensor components of the principle compression and tension. Torsional is a compression strut twisting around the outside of the torsional member and rebar resist the tension components (I am taking some liberties and simplifying here to a certain degree).

You can argue that there is a bit of an eccentricity between the tension load of the cables/rebar and the compression strut of #11 but this doesn't amount to a torsional member as we think of them. You can resolve the forces into compression struts and there is no need for torsion. A cracked torsional member in concrete tends to be far more flexible than the assumed elastic member. Once you have a torsional crack and there are other stiffer load paths, all the torsion is relieved. It is like wood-armer moments in a slab. You can ignore the torsion disappears and redistribute the torsion bending into bending moments.

Also, if this was a torsional moment, you need a resisting torsional force at the other end of the torsional member.

 

[Earth314159]

I don't think it is a good thing for interface shear (AKA shear friction) to be ignored in the Australian code. Interface shear is a real think in any concrete structure. You can't wish away. Even in a monolithic pour there is interface shear (the mu values and cohesion just have higher values). I don't think it is correct to say the Australian code won't allow it. It doesn't account for it which is worse. Other countries don't ignore it. It is a mandatory check and there are decades of experience using the code equations.

 

[Vance Wiley]

I see. And to an extent I agree. Here is a quote from earlier:
That reads a lot like your thoughts.
I posted a spreadsheet study of the shortening of the canopy due to dropping at node 10/11. The top of 12 definitely gets pulled south as the collapse proceeds.

 

[hokie66]

We have had many discussions about shear friction theory on this site. The early proponents of the theory acknowledged that the values obtained in tests were a combination of shear and dowel action, but were not able to separate the contributions of each. Shear friction is said to result from passive clamping forces, and that is where the logic falls short. For a clamping force to develop, there must be elongation in the reinforcement normal to the joint, and that elongation must be accompanied by opening of the joint.

An anomaly is that the clamping force is based, I think, on all the developed steel crossing a joint at right angles, but surely only the steel in tension can contribute.

Australia has its share of building issues, but I don’t believe any have been caused by the intentional omission of shear friction as an accepted method.

 

[ldeem]

hokie66

ACI accounts for rebar at an angle to the shear plane by resolving for the perpendicular force. Which brings up an interesting question as to how many different shear planes should be considered in design. Typically it would just be the cold joint which is easy to locate. However in something like this truss where it appears the failure plane was partly in the cold joint and partly through the monolithic pour the critical shear plane is much less obvious. When you ad in the drain pipe, conduits and all the other pieces and parts in the joint the complexity jumps up to whole new level. Looking at the picture earlier in this tread showing the joint with all the rebar, conduits, etc. in the forms makes we wonder if in design they ever had a good sense of how crowded this area would be. I recall the engineer considered the 4" conduits and drain pipe but I wonder if all the other conduits and obstructions were known at the time of the design.

Hindsight is 20-20 but looking at the picture of the actual joint I wonder if anyone on the engineering team thought if all that congestion was appropriately accounted for.

 

[LittleInch]

I've been following this since the start, though I'm no structural designer.

My thoughts are that when you get all this level of detail it can be easy to miss the big picture.

I haven't read all the recent docs in great detail, but the key issues which seem to be in danger of getting lost.

This bridge was a one off strange design, but seems to have been designed using "standard" methods and analysis.
It has different sized top and bottom chords.
It's a concrete truss.
The truss design is asymmetrical and appears to lead to out of balance forces which no one other than me seems to care about. I think they might add to the compressive force seen by member 11 and possibly also add to a bending moment on members 1 and 12.
The design of the temporary construction mode does not seem to have received the full level of attention that it should have done.

How the bridge was transported gave rise to potential damage - Never clear how exactly they worked out 0.5 degrees twist was ok but 0.55 wasn't. This was a large rigid, brittle beast of a thing to move across an undulating surface.

The presence and movement during the design of a whole bunch of tubes, large and small into the node area may or may not have been included in the final calculations and analyses - some earlier drawings have the two larger vertical tubes some distance away from member 12, but ended up right next to it. Whatever they appear to be a key factor in reducing strength, increasing congestion and complexity and their impact should be highlighted.

Not withstanding any of that and the construction joint issue, the bridge survived for three days slowly failing. For the PE and the design firm to witness this and not be man enough to say something's wrong, we need to stop is nothing short of criminal (IMHO). The other parties will look to the designer for guidance and re assurance. To provide that reassurance on what seems to be a wing and a prayer about "well why don't we try re-tensioning it and see what happens" seems to be being swept under the table and I'm not surprised it doesn't seem to figure much in the Figg document. It should.

There are many many lessons to be learnt from this collapse but it can be lost if some of the parties are trying to focus in on one particular aspect ( this construction joint / friction factor) and not the big picture. It may well have been a contributing factor, but as other have pointed out - this was a very congested area and as a designer you always need to consider exactly how you're going the build the bloody thing safely and effectively. Then when you see it not working in practice, stop and re consider, don't just keep going.

The NTSB final report awaits.

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[Vance Wiley]

Your entire post is quite good. So many points are so easy to agree with.
While the truss is asymmetrical and that causes internal loads not immediately evident, there do not appear to be any issues there that are beyond current engineering knowledge as far as analysis goes. The flatter angle of member 2 compared to member 11 causes thrust into the canopy and tension in the deck that overwhelms or strongly influences expected connection loads and member loads over the southmost 3/4 of the span. The reverse angle of member 3 causes unusual loads in the members and joints in its zone of influence. But these can be predicted and supposedly dealt with. As you correctly note, this is a one-off design, and these geometry issues complicate the design, as does the choice of materials used, the need to correctly deal with connections in concrete having sharp angles between web members, the mixture of PT and normal reinforcing in adjacent members, and strain compatibility. These combine to become major challenges in this structure.
It should be noted the truss geometry is dictated by the visual design intent.
FIGG has proposed some visibly beautiful towers to support a proposed cable stayed skyway structure over congested city streets in another Florida city. While beautiful, with twists and partial curls and tapers, the towers appear at a glance to create unnecessary engineering challenges. This pedestrian bridge also created unnecessary engineering challenges. Apparently many of those were not recognized.

 

[Eufalconimorph]

There are three important questions.
[ol]
[li]Why did the bridge collapse?[/li]
[li]Why were workers killed in the collapse?[/li]
[li]Why were uninvolved people killed in the collapse?[/li]
[/ol]

Question 3 is (in my mind) the most important. Why wasn't the road closed? Why were there cars under a clearly failing bridge? This is the question that should lead to criminal charges (for negligence at least AFAICT).

Question 2 is a matter for OSHA. What training was missed for the workers not clipped in? What oversight by the foremen was skipped? Etc, etc. Also possible criminal charges here, but less importantly. It's a dangerous occupation, the PPE was provided, some of the fault is the workers'. (Some of the workers injured were in situations where the PPE wouldn't help, that's a separate issue.)

Question 1 is the least important, but the most interesting. Why did the design fail? Could the architectural design have been implemented safely? Etc, etc. Most of the discussion here focuses on this question, which is appropriate since this is an engineering forum and it's the engineering question.

 

[RVAmeche]

The simple answer to #2 & #3 appears to be because FIGG didn't believe the bridge would collapse.

 

[hpaircraft]

The simple answer to #2 & #3 appears to be because FIGG didn't believe the bridge would collapse.

I would phrase it differently: Instead of "FIGG didn't believe the bridge would collapse," I'd say they did believe that the bridge would not collapse. There's a subtle but important semantic difference; the former features a passive lack of belief and the latter has the active presence of belief. A belief that was not founded in an accurate assessment of the status of their structure.

 

[epoxybot]

hpaircraft (Aeronautics) I agree, there seems to have been a profound ignorance as to what the cracks were telling those who actually examined them on site. Even the photos display a lack of experience and little in the way of increased skill or thought given to better documenting the events.

 

[RVAmeche]

That's a really interesting tweak. I imagine lawyers would prefer that phrasing

 

[FortyYearsExperience]

there seems to have been a profound ignorance as to what the cracks were telling those who actually examined them on site. Even the photos display a lack of experience and little in the way of increased skill or thought given to better documenting the events.

This is a profoundly insightful comment. And of course, it is precisely what Denney Pate expressly stated in the voicemail he left behind for the FDOT.

 

[Sym P. le]

Thank-you Vance and Earth for your replies.


Earth, I see your response as a blinding example of tunnel vision. In a conventional situation, exactly as you say, but in this deconstruction, my lack of academic rigor may leave me more room to explore.

... need a resisting torsional force at the other end of the torsional member

The deck/diaphragm duo compete with the 11/12 duo. Once the 11/12 node is free of the pocket, the structure becomes indeterminate and there is a tremendous amount of torsion as 11 pushes 12 north while the slab rotates down.

... how did it "root" out the section below the top of the deck?

I've questioned this also but I now believe that the 11/12 node became detached as a whole thus rendering this point moot.

The top of 12 definitely gets pulled south as the collapse proceeds.

FWIW, the top of 12 has to move north approx. 4" at the initiation of the collapse (Part XII Sym P. le (Mechanical) 21 Jul 19 02:54)

So to bring my thinking around, 11/12 node is cracked out of the pocket by the lower PT rod, torsion takes over and blows apart the whole bottom end of 11 and 12 ???? 11 slides down the PT rods in a controlled fashion while 12 drops vertically.

 

[Earth314159]

The deck/diaphragm duo compete with the 11/12 duo. Once the 11/12 node is free of the pocket, the structure becomes indeterminate and there is a tremendous amount of torsion as 11 pushes 12 north while the slab rotates down.

A structure does not become indeterminant or more redundant as a load path is lost. It becomes less redundant and closer to a determinant structure.

The torsion is not large. There is an equal and opposite force from the deck and PT cables in the same line that resist the horizontal component of the strut force. Once the #11 and #12 lest loose from the pocket, you have no more load paths. It is a failure. To create torsion, you need a couple and there is no couple.

Earth, I see your response as a blinding example of tunnel vision. In a conventional situation, exactly as you say, but in this deconstruction, my lack of academic rigor may leave me more room to explore.

Sorry, I guess I let facts and evidence get in the way of my analysis.

 

[Vance Wiley]

Reinforcing in Member 11
We see photos of pre-collapse cracking in member 11. WJE reports member 11 triggered the collapse by failing near node 11/12 after node 11/12 slipped on the deck.
On sheet B-39 of FIGG's RFC drawings Detail A-A shows a 21 X 24 section with 10 - #7 longitudinal bars and a note "Members with no PT bars". So the interior web members which are in compression have 10 - #7 longitudinal bars. Interior members in tension have PT rods to resist the tension and nominal reinforcing to anchor and retain the concrete.
On sheet B-40 we see Detail B-B as a detail of a section 21 X 24 with 8 - #7 longitudinal reinforcing bars and what I assume are PT rods in ducts, and the note "Typ all members with PT bars".
So when PT bars were added in member 11 for transport, how did the reinforcing detailers interpret the drawings for member 11? Member 11 has PT rods - which is it? Detail A-A or B-B?
It may not make much difference in capacity, but I find it interesting that the 21 X 24 section with the greatest load by far has 8 bars and the far less loaded compression diagonals in the interior have 10 bars of equal size.
The fact is FIGG drawings, ELEVATION on B-40 clearly notes for member 11 " 7S11 (E F)" and there are two arrowheads so I guess that means 2 bars each face.
That would be a total of 8 bars - as Detail B-B requires. So FIGG also fell into their own trap - the highest loaded compression member of 21 X 24 dimension has less reinforcing than all others because PT bars were placed there to resist forces from transporting, and the PT was not really PT because it was intended to be released by detensioning.
In looking at member 4 on the ELEVATION on Drawing B-39 we see Detail A-A cut and the note "7s04 (E F)" and two arrowheads and 4 lines which indicate reinforcing in other locations on the drawings. Are there supposed to be 3 bars in the 24" side as shown on A-A or 2 bars if we count the arrowheads or 4 bars if we count the lines?
The coordination and review exhibited in the drawings is lacking in this case and it seems to have contributed heavily to the failure of member 11.

 

[TheGreenLama]

Or how about just the base of member 11 failing first, just exploding. This is nothing new here. It has been posited numerous times earlier. Even FIGG's report on the failure alludes to it. In the most basic terms: Base of 11 slides > causing distress and deterioration of base of 11 > causing base of 11 to finally explode. After that everything else is secondary.