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

 

[graybeach]

Here in NYC a 3 year old, 450 foot long pedestrian bridge in Brooklyn has to be replaced because it is too "bouncy" and has some other problems. Supposedly the bounciness was by design or at least this was the bill of goods sold to the owner by a certain genius grant "star-gineer". I bring this up because engineers who become "famous" as was somewhat the case with Mr. Pate have a huge responsibility to question themselves at all times and be ready make corrections and in fact rejoice when mistakes are caught before any loss of life or major economic loss. Very dangerous to be in a situation where star power could tamp down legitimate concerns, even unintentionally.

 

[samwise753]

jrs_87, can you point me to what you are talking about?

 

[jrs_87]

Sorry to disappoint, but that's the point. There is no such thing.

 

[Vance Wiley]

Dave - Thanks for the questions.
Somewhere in there I used the term "if undamaged". I was trying to find a capacity/demand condition to define the adequacy of the FIGG design at the time of the collapse, with no other factors involved. No cracking from shrinkage or stressing, no twisting from transport - just the designed and detailed structure.
I did use a shear area of 21 X 40, a bit more than you suggest.
What those numbers indicate is there was some minimal buffer against failing, but very little. The design certainly did not meet an AASHTO LRFD design which would include 42 kips reaction for construction load and not the 3 kips which I used.
AASHTO for Dead and Construction load:
DL 810 K /tan 31.8 deg = 1306 kips slide X 1.25 = 1632 kips
LL 42 k /tan 31.8 = 70 kips slide X 1.75 = 122 kips
Demand + 1632 + 122 = 1755 kips Game over.
Then factor capacity of 1548 kips X 0.9 = 1393 kips gives Design capacity demand of 1755 kips and Capacity of 1393 kips and the design is seriously deficient under ASSHTO code for this load case. EDITED

Back to the actual load at the time of failure - 1311 kips from the earlier post - and comparing that to the factored capacity of 0.9 X 1548 + 1393 kips, there is a margin of 6%, if completely undamaged and joints were prepared to 1/4" amplitude roughness. From a previous post, it appears the AASHTO factor of 0.9 provides a possibility of failure of ONE in 250 ( 0.4 % ) if the Capacity and the Demand are equal.

Clearly any damage such as you describe or construction process such as lack of preparation of the joint surfaces which compromised the as designed capacity in any way would preclude any chance of the intended design to support itself until other construction could provide added capacity.
Added capacity was necessary and quickly. It was too late by the time the condition was recognized.
And with the joint at node 11/12 un-roughened and damage already evident it could have fallen the instant the transporters placed it on the final supports.

 

[jrs_87]

"Accelerated Bridge Construction Risk, A Contractor’s Perspective" March 17, 2016

https://abc-utc.fiu.edu/mc-events/a...ion-risk-a-contractors-perspective/?mc_id=100

 

[Vance Wiley]

Thank you SAM! You are absolutely on track!
IF the last PT was stressed BEFORE falsework was removed under the back span, that could provide some lift effect and add to the negative moment over the pylon. But the simple span weight of the main span would have to be lifted to a level condition before it could contribute. And your numbers will likely tell how much the PT can help.
This is the PT that was intended to provide some clamping force to assist node 11/12 and prevent the collapse. I think in order to provide clamping of 11/12 they should have left a gap in the canopy while stressing and grouted it later. Or maybe never.
EDIT
I think I recall a reference to continuity over a support as providing redundancy. So this was their token effort at redundancy.

 

[samwise753]

*facepalm* Well played jrs_87.

 

[The Mad Spaniard]

Vance Wiley (Structural)1 Nov 19 07:24
Some points from NTSB Meeting Miami and further down, comments about loads and capacities at failure.

Very informative post.

But, do not forget that the equations have some limits:

V_u =K₂*A_cv

In this case (assuming slab)

V_u = 1.8x21x40 = 1512 kips

If we have a resistance factor of 0.9 then

ΦV_u = 0.9x1512 = 1360 kips

If instead of 1.8 we use 1.5, then the whole thing changes to 1260 kips

 

[hpl575]

I just keep coming back to; Looks like a duck, quacks like a duck, walks like a duck, must be a goat, mentality that appears to have been used in the design and analysis of this bridge. The design codes dictate deep beam provisions to account for the effects of struts and ties forming and consequences thereof even though it may not be apparent for a beam. FIGG presented a deep beam analysis for the end diaphragm during their presentation the morning of the collapse. Why strut and methods and provisions were not followed for the design of the truss, the very picture of a strut and tie system, somewhat baffles my mind, of course hindsight being 20/20.

I was hoping the NTSB investigation and contributors thereof would at least explore the significance of not resolving the forces and reliably delivering the forces to the nodes as is required the strut and tie method, but it appears from what I have found so far in the documents that everyone is focused on shear friction. The better solution to me would have been not to need shear friction in the first place at the truss joints.

 

[3DDave]

Looking at alternate analysis is worthwhile, but with that cold-joint sitting there and it's place as a primary contributor to the failure it's natural that the focus is on shear friction. The NTSB is more interested in what went wrong rather than creating alternate designs that would have worked. I know they often make recommendations, but I cannot fault them if they didn't want to create a do-it-yourself concrete strut-and-tie analysis kit. Let sellers of concrete work out the best solution to this problem and how to implement it.

 

[hpl575]

FIGG was criticized in the course of the NTSB investigation for not applying a 1.05 factor for a non-reductant member(s)/structure. I wish I knew actual loads within 5 percent, other than fresh water, for any structure I have ever designed. In my mind increasing the load by 1.05 does very little to increase the actual reliability of non-redundant member other than account for an additional 5% overload. A better means of increasing reliability is to increase ductility and resistance after cracking has occurred as is required for high seismic risk locations for load resisting members.

In a statement in the WJE report prepared for FIGG something similar to, even the first beam placed in a steel bridge is non-redundant, is a made for lawyers and trials conclusion. The argument goes something like this: "Aren't some parts of most structures non-redundant at some point during construction... wasn't this structure currently still under construction."

But, anyone who would infer that since there is more than one layer of rebar in member 11, therefore it is redundant, don't even know how to respond to that.

 

[Earth314159]

Samwise,

That is 12 cables in each conduit. A factored resistance of about 2500Kips.

You don't have to take full continuity for dead load. You set the amount of hogging moment with the amount of PT.

 

[hpl575]

Thanks 3DDave, and I do mean that sincerely, I guess I was hoping after waiting so long for the NTSB findings to be made public that some of the other facets of the design and methodology would be explored. Maybe such recommendations are best left to the Code Committees and researchers.

 

[samwise753]

Ah. I saw the number 12 (and others depending on the strand designation) in front of the strand, but didn't connect it as the number of strands per conduit. Not very familiar with post-tensioning; looked up something and thought the number designation for couplers was for size of coupler. It is just the number of strands the conduit can handle. I need to research hogging moment. If two separate spans already become compromised, then will hogging moment help the situation? It sounds like it could have even if the concrete cracked within normal bounds but remained sound.

 

[jrs_87]

I'm wondering, given time, what is the possibility FIGG is going to provide an explanation that will surprise us?

 

[Vance Wiley]

OUCH !! I did not check that. Thank you.
So for this span to have supported construction live load also the factored demand would have been 1755 kips. That would require an area of 21" X { 1785/(1.8 X 21) = 47". The area under the projection of member 11 is 24/sin31.8deg = 45 inches and its northern projection seems to coincide with the edge of the deck and 10-1/2" from the north face of 12. The placement of one of the # 7 hoops mobilized the fillet under 11 but that capacity was not tied back to 11 and was one of the first things to crack.
The result is my statement that it had some headroom in terms of actual (unfactored) demand and resistance must be modified and look like this:
Expected Capacity at failure on March 15 (if undamaged) is 1548 kips but is limited by code defined Vu to 1512 kips. That capacity is based on undamaged conditions. Actual demand as I calculated it was 1311 kips, and with only the actual live load at the moment of collapse estimated to be 3 kips.
The structure had a limited amount of excess capacity and should not have failed on March 15 had it been undamaged and had the construction joints been roughened to 1/4" amplitude.
But it was seriously under designed for Dead Load plus Construction Load and far more deficient when prescribed pedestrian Live Loads are considered. But before being opened for pedestrian traffic, the north span and pylon would have been completed, making that load condition far less critical in the big picture.

Your comment brings to mind the results of testing by WJE. In those tests they cast a 32 degree joint in a 21 X 24 section and tested that capacity in axial load, creating shear across a 32 degree plane in the 24 inch dimension of the section.
The test results reported an average peak resistance of 2594 kips for the intentionally roughened surfaces. That calcs to be 2594K/(21 X 45" ) = 2.74 ksi. Apparently the max value of 1.8 ksi used to set max Vu is based on a factor of 0.67 for shear when compared to the limited test results, and that seems reasonable.

 

[Vance Wiley]

I totally agree.
As pointed out by the great Takes me back to the words of a wise Architect - "Never be the first to try something and never be the last guy to use something."

 

[Tomfh]

All structures with shear loads rely upon so called “shear friction”. If you have a monolithic pour it’s still theoretically shear friction. It’s just that the frictional force is vastly better in monolithic concrete.

As for not having a cold joint heavily inclined with respect to the load path, yeah, that would have been best avoided.

 

[hokie66]

I keep coming back to the issue of longitudinal shortening of the chords. A combination of drying shrinkage and PT would have shortened the structure by in the order of 1.5 to 2.5 inches. This amount of shortening creates irresistable forces if there are restraining elements. The top chord had little restraint near the end, but the bottom chord was restrained by the diagonals, referred to as members 2 and 11. So the end diagonals and deck were already moving in opposite directions before the external load was applied.

 

[Vance Wiley]

I can certainly support that. Members like the web diagonals should have a lot of confinement reinforcing. A lot.

I would like to present some thoughts about redundancy. Just for discussion, mind you. I am not sure I will even agree with myself tomorrow - but it was only a half glass of wine, so the probability is increased.
Where does redundancy start? Can a bridge be redundant - or would you build a second bridge beside it to provide the function if the first one fails? From the viewpoint of traffic flow, that looks good. From the viewpoint of those who went into the river, not so much.
I have more than one 9/16" combination wrench - now that is redundancy. One breaks, I can keep on going.
So it would appear redundancy can be defined by one's perspective.
In the case of this structure, there is only one deck - how do you make that redundant? There are dozens of PT strands in the deck, and if one fails the structure will not fall, so is the PT considered as having redundancy? Does that then extend to the deck? It does not make a lot of sense to construct a second deck - .
The idea of providing capacity beyond that needed and therefore creating redundancy is one way and in many cases may be the only way to provide redundancy. What level of increase? 1.05 is not much. Maybe define different factors for different conditions. So should a 4 bar column get a 1.2 increase factor? An 8 bar column a 1.10 increase factor? If you choose bars having 20% greater area is it more redundant? Or should there be more bars? With 4 corners which corner would get the extra bar? Would it do any good to place a #4 bar (0.20 in^2) beside each #9 bar (1.0 in^2)? That would be a 20% increase.
In a small beam that might use 2 - #9 bars, adding 2 # 4 bars would do little if a #9 failed.

And lastly, what good does it do to to provide 150% of demand strength everywhere if the designer screws up and only provides half enough capacity in a critical location?
i just realized that is all questions and no answers.
Maybe expecting a solution is just a "half pipe" dream.