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

 

[SFCharlie]

The Feds ran 4 runs and got different answers each time - by maybe 10% - maybe more.
Strikes confidence in me.

I got the impression that the different runs were for different conditions of support; un-tensioned, tensioned, transported, etc.

SF Charlie
Eng-Tips.com Forum Policies

 

[charliealphabravo]

How is a forensic report as massive as the Figg "submission" not signed and the author(s) identified? It's almost like the FL engineering rules mean nothing.

 

[epoxybot]

How is a forensic report as massive as the Figg "submission" not signed and the author(s) identified?

Maybe when the language used in the report, is the result of a negotiation between the forensic investigators & the client.
The report puts a lot of focus on the roughening of the joint between the deck and #11 and rightly so but the in the few instances where surface prep was called out, I do not believe they ever added (TYP.). It would have made a world of difference.

Concrete finishers love their hand tool. I wonder if a small hand cultivator would be an acceptable means of roughening the concrete before it completely hardens.

 

[Earth314159]

My favorite words are Typ. U.N.O.

 

[hpaircraft]

...Maybe when the language used in the report, is the result of a negotiation between the forensic investigators & the client.
The report puts a lot of focus on the roughening of the joint between the deck and #11 and rightly so but the in the few instances where surface prep was called out, I do not believe they ever added (TYP.). It would have made a world of difference...

I can't argue that the joints didn't get the prep they ought to have. But I think that if you're in the realm where the absence of 1/4" surface roughening in a construction joint causes your bridge to fall down under its own weight, your design embodies way too little margin of safety.

 

[Vance Wiley]

Nodal Shears FHWA analysis of FIGG design - Bridge Factor Attachment 73 - pg 77 and 78 of 90

Bridge Factors Attachment 73 – FHWA Assessment of Bridge Design and Performance

https://dms.ntsb.gov/pubdms/search/document.cfm?do...

FHWA finds FIGG underestimated nodal shear demand at 11/12 and deck. FHWA represents FIGG calculations as showing shear at 11/12 to deck to be twice as great (approx) under full two span condition than when under single main span with full fixity at the pylon/pier (north end of main span), and the greater value should have been used for design and was not. Both conditions are noted to be Stage 4.
Note the Fixed Pylon (Stage 4) is an intermediate construction stage subject to construction loads only. Failure prevented this condition from being encountered.
Note also that the Longitudinal Model (Stage 4) is also an intermediate condition subject to construction loads only.
I do not see how it is possible for moments at the north end of the main span to be greater than when full fixity is assumed at that end. Unless one condition is unfactored DL only and one is full factored DL and LL.
First, DL is about 11 kips/ft and LL is 90#x30 feet = 2.7 kips/ft. Factored that is 11k X 1.25 = 16.75 kips/ft DL and 2.7 k X 1.75 = 4.37 kips/foot factored LL for a total of 18.5 kips/ft factored load. The factored TL load is 1.67 times unfactored DL. So the 2X increase was not from considering unfactored DL vs factored TL. Factors are FHWA representation of those used by FIGG.
The only way I can see the shear demand in node 11/12 to deck increasing beyond that from full fixity at the pier is if enough negative moment is created over the pylon by the continuity PT in the canopy to draw even more load from the main span to the pier. But that condition is Stage 5, after the north span has cured.
So I am confused as to how FIGG could have had such a discrepancy between shear demand at node 11/12 to deck under Fixed Pylon (Stage 4) vs Longitudinal Model (Stage 4) conditions.
As I recall prior discussion here the unfactored DL shear at node 11/12 to deck was in the order of 1300 kips. Or was that the axial load in 11? Adding LL and factoring the loads by the 1.67 from above (DL to Factored TL ratio) yields 2170 kips - pretty much what the graph by FHWA shows for 11/12.
Perhaps FHWA is misinterpreting something also?
Whatever the answer, this points out the complexity of this design and the various support conditions which existed - full form support, full span at casting yard, intermediate span during transport and reversal of loads in many members, and final support in place. All happening in a brittle structure.
And if the non-redundant factor is only 1.05, that would not have saved this structure at this phase.

 

[Vance Wiley]

Table 2 Pg 81 of Attachment 73 has 4 runs - you are correct, two are with and without PT rods in 11 and 2 and there is a 2D and 3D run to compare.
I should have been more specific. Thanks.

 

[Vance Wiley]

Easy there, guys. This is a family hour forum.
The tool you recommend cannot be approved. First, it requires a seat, a safety belt, and special gloves.
Seriously - I think special provisions for joint preparation should be made at joints as critical as node 11/12 to deck. The demand and unit stresses and performance required are so much greater than in any construction joint in a pier or wall or at the top of a girder to a composite slab. So I think special forming is required. I suggest formed indentations in a 45 degree shape with amplitudes greater than the largest aggregate size.
Forget cohesion - the slightest slip has destroyed that. Forget raking it - heck, I have seen bridge decks with amplitudes greater than 1/4 inch. So often raking results in dingleberries which are easily dislodged by structural loads but are not loose so washing does not remove them. And yet raking is one of the predominant method of preparation for a joint.
The sawtooth configuration will develop the tension in reinforcing across the interface while the aggregate can move into the "teeth" and those areas will act like concrete instead of paste, dingleberries, and ball bearings. If reinforcing creates access problems great - make the shear plane bigger. If this project demonstrates anything it is "get enough".

 

[epoxybot]

Vance Wiley - Yes a more "keyed" approach would not have been difficult to form. It is sad that FIGG was able to recognize in the design phase that the area between and north of the plastic pipes penetrations were not a load path but when inspected in the field, failed to appreciate that the loss of bond at the deck/node interface & the loss of the southern most #7 in the chamfer, was so disastrous.

 

[hokie66]

I doubt that shear friction theory was ever intended by its inventors to be used in a situation like this. Hopefully, it never will be again.

Shear friction is not mentioned in the Australian concrete standard AS3600, but I know that some Australian engineers use it at times under pressure of contractors who don't want to provide decent bearing at joints.

 

[Vance Wiley]

I am quoting myself because i may have an answer to my own question.
I have not ran the numbers - that took FHWA 18 months - but in concept the increase in load at node 11/12 to deck may be due to the details of staging the construction.
Calculating for fixity at the pylon end (north) of the main span under dead load is unrealistic - that could only happen if the span were again supported or lifted along its length and then fixed. So calculating fixity at the north end of the main span should only apply to added loads after it is connected to the pylon and north span.
Thus shears at node 11/12 to deck would be from full DL of the main span as simply supported PLUS added load on the main span under finished condition PLUS load drawn from the main span due to negative moment produced over the pylon from removing shoring under the north span and from the continuity PT force added in the canopy under Phase 5.
In concept it appears the loads on node 11/12 to deck could increase. I do not see how they could double - but I have not ran the numbers.
So I get the following approximate added load (vertical) to contribute thru whatever angle to cause shear in 11/12 to deck. All calcs by "back of envelope" method and to that accuracy. Remember - this is the increase at the end reaction of the main span and must be adjusted to show actual shear in the node. So compare only the loads causing end shear.
First, LL from the main span will add about 300 kips shear in the main span at the pylon and is the major thing that will influence the node shear. But this does not allow us to compare the effect of releasing the north span. I propose to do LL and released DL on the north span to see maximum influence on node 11/12 to deck from that action.
Distributing the released DL from the north span to both spans with spans fixed to pylon adds 14 kips shear to end reaction of main span at pylon. Loading LL onto north span only increases shear at the north end of main span by 4 kips. General concept is Unbalanced Moment X stiffness factor of main span / stiffness factor of both spans and pylon.
The pylon, if fixed to superstructure trusses is rather stiff becaus it is short. Pinning the top of the pylon increases the draw of shear to north end of main span to a total of 34 kips. At 32 degrees, that is about -what - 60 kips in a connection that has 1300 or more? Insignificant. Remember - all numbers unfactored. The fixed end moments of DL and LL from the short span remain mostly in the short span because the long span is less stiff, and the shear draw is M/L.
Would appreciate someone checking me out - for a pinned top of pylon the distribution is simply L/(L1+L2) and with same bridge section on both spans it is rather easy.
I find it interesting today because if I do it right it keeps me fresh. If I screw up, I gotta make adjustments to my thinking. So feedback is welcome.
Anyway, I still cannot see the shear at node 11/12 to deck doubling from the release of the north span.

 

[PA PE]

The interface area was extremely congested. It would have been difficult to put a hand inside that mess of steel, let alone get a tool in there and maneuver it around to roughen the surface of the cold joint. In the Figg rebuttal there is a photograph of of this area showing forms and steel in place before the deck concrete was placed. There are electric conduits going through this area as well, further reducing the strength of the members.

At the risk of being accused of Monday Morning Quarterbacking, I put together a quick sketch showing an alternate way to create the interface area. This is based on Figure 56, on Page 60 of 90 in the Bridge Factors Factual Report Attachment 73.

The sketch shows a buttress above the deck surface that would be placed with the floor, making it monolithic with the floor. The shear plane would not go through a cold joint. The Interface bars would be angled, The resulting cold joint would be in pure compression.

Thoughts?

 

[PA PE]

I understand that the requirement to roughen the surface was not explicitly stated on the Drawings. Rather, this requirement is said to be included in a referenced standard: A Florida DOT publication.

Since this is such a critical requirement, would it make sense to restate this requirement on the Drawings so it's right there in front of the Contractor? Or does that create a slippery slope that would end with repeating the entire referenced document?

 

[Vance Wiley]

a quick sketch Thoughts?

I think it is a step in the right direction. The numbers will likely show you will need a greater length and/or width to get enough reinforcing across the plane at the top of the deck. The monolithic casting will improve the friction factor from 0.7 (smooth) to 1.3 (monolithic (if I remember). The sloping joint has been criticized previously. Your illustration is a great representation of the joint I would support. (oops-no pun intended). The fixity of the monolithic nodes (or those with proper joints) has created some moments in the web members with corresponding non critical shears so a formed key would be a good idea in the square joint face.
And to transfer the tear out force to a point where the PT in the deck can handle the load it will need longitudinal reinforcing well anchored under the joint and extending far enough south to transfer the load. Had these ideas been at work on March 15 the structure would be standing today, IMO.
Absolutely. My office issued many many drawings over 40+ years and in every case in which construction joints were needed we included a note in the general form of "Construction joints shall have surfaces removed to expose aggregate solidly embedded and remove all laitance ---" etc. And that was for tops of foundations under walls, joints in walls, - - not in something with the demand and critical nature of node 11/12 and the deck.
Which brings me to a comment about aggregates. In many photos of the damaged structure I see what appears to be fractured aggregates embedded in the cement matrix - and though the photo is a 2D image it apppears the aggregates fractured in the same plane as the matrix. In the use of light weight aggregates there is a requirement for "splitting tensile strength". I wonder if the aggregates used are appropriate for 8500 psi concrete? It seems to me if you want to make 8500 psi concrete you need 8500 psi aggregates with suitable shear strength. I am aware that FHWA found the concrete to be suitable for the design requirements by FIGG.

 

[PA PE]

Referenced standards are often seen as boilerplate and for the purpose of CYA. I like to put plenty of notes on the drawings. Referenced standards are supposed to be available on site. How often does that happen, who takes the time to read them, and who enforces that requirement? At the very least, critical requirements should be highlighted during the preconstruction conference, or at a pre-installation meeting.

Figg made a valid point that their involvement during construction was limited. My project managers don't like to send me to the sites either because of my billing rate, and it becomes very annoying when the inspector calls in with questions that should have been addressed before they become a problem in the field. The person who does the design is in the best position to see things before they become a problem, and even allow minor deviations that accommodate a contractor's concerns without compromising the work. I wonder if Figg was tempted to take the position that if they can't be on site then it's not their problem. This could point to an issue in the construction process and the contractual arrangements between the various entities.

I wonder who is paying for all the forensic work?

 

[Vance Wiley]

I wonder who is paying for all the forensic work?

Which has cost probably 10 times the design fee.
The design process can involve maybe hundreds of decisions each day. The litigation process can spend maybe hundreds of hours on just one of those decisions. It does seem that, for the most part, after a year and more of posturing, most of the parties in this event have decided to avoid litigation and settle.
I applaud that.

 

[SFCharlie]

It seems to me if you want to make 8500 psi concrete you need 8500 psi aggregates with suitable shear strength. I am aware that FHWA found the concrete to be suitable for the design requirements by FIGG.

I think you're on the right track. The concrete for the new locks in the panama canal was failing specs because the basalt aggregate wasn't strong enough.

SF Charlie
Eng-Tips.com Forum Policies

 

[Earth314159]

PA PE

Your concept is theoretically acceptable but is more difficult to build/form. There are also other joints that were an issue which would also have to be altered. There is actually more shear at the #10/#11 joint than the #11/#12 joint but it was not an issue since the PT from #10 came through the canopy and up into the blister to form a shear plan on the upper side of the canopy (the PT also clamps the joint). Although the shear stress across the pour joint was still very high.

The steel was actually not that congested (at the slip plan directly below #11). I would increase the base as you have shown, increased the steel (as you have shown) and roughened the surface. For ease of construction, I would keep the pour joint at the top of the deck but make sure my stresses were low enough to be adequate. The steel should all be ties for good anchorage for shear friction.

I would also say, under similar circumstances, I have had hassles from rebar suppliers. They don't like to make each tie a different height. It is more labour and difficult for them to organize. Although I would say that the extra complication is worth in this case.

The other option would be to organize a continuous pour of the deck and diagonals. You could form the a square end at the top of the diagonals and then pour the canopy. You don't want to pour the canopy with the diagonals because of the plastic creep issues. You can get a gap between the diagonal and the canopy.

 

[Vance Wiley]

The other option would be to organize a continuous pour of the deck and diagonals

This seems difficult on first glance. Could you explain how they would prevent the pressures from concrete in the webs from 'boiling out' at the deck surface? I like the idea of reconsolidating the concrete in the deck and diagonal joint to create a monolithic joint. That increases the coeff of friction from 0.7 or 1.0 to 1.3 {or is it 1.4?).
As for the canopy, I have thought that a continuous concrete beam should have been cast full length under the canopy, providing rotational resistance to those joints and plenty of area for shear transfer - something like a bridge deck on a bridge girder. That might leave the plastic creep issues you mention, however. What are your thoughts about creep and elastic shortening of the canopy from PT forces relative to its effect on the already cast diagonals?
It is interesting how many opportunities there are for 'improvements' to this structure. Points out the complexity of the idea.

 

[Earth314159]

Vance Wiley

Doing a continuous pour takes some experience. We sometimes do it for footings and walls/columns. You have to go back and forth on the pour. This also does not work with high slump concrete but it is done. In my area, it is actually common for house strip footings and foundation walls.

I don't have my code at home but I think it is 1.4 (I can check later). It may also be slightly different in the US. A continuous pour makes the failure plan no different than a failure plane in a typical column. If you look at Figg's report from the NTSB docket, they actually made columns will deliberate cold joints to simulate the failure plane of the bridge.

I though of a continuous beam on the canopy as well but that would have to be poured with the diagonals as well to get an advantage. You might as well pour the canopy with the diagonals.

If you have a true determinant truss, the PT does not affect the member forces in the diagonals or end verticals regardless of the creep. However, you do not have a true truss (nor do you ever have a true truss in the practical world). The bending stiffness of the canopy, diagonals, verticals are affected to some degree by the PT. You can argue that if you have enough ductility demand, it really doesn't mater in terms of safety. Elements like #1 would have double flexure due to the PT. The PT would cause visible damage but not critical damage as the axial stresses due to gravity load on #1 are low. The only way to know for certain what the effect of the PT would be is to model it. Even a simple 2D stick frame could be used to do this. To cause high shear forces at the joints with the PT would require the deck and/or canopy have a relatively high bending stiffness compared to the truss as a whole. In general, creep reduces the level PT force as the structure essentially softens with response to long term loads. Any deleterious effect from the PT would also likely diminish with creep. When I have had a complex PT structure, I model concrete both long term and short term. I reduce the E for concrete but keep the same temperature difference on the PT to determine the relaxation due to creep.