Florida International Bridge Collapse

DaveC426913 · Mar 21, 2018

And, I presume, insufficient testing before installation.
Would this have failed if it had been set up at the assembly yard and left for a week?

Peter Dow · Mar 21, 2018

DaveC426913 said:
And, I presume, insufficient testing before installation.
Would this have failed if it had been set up at the assembly yard and left for a week?

The problem with a house of cards is not "insufficient testing" because it fails whenever it is tested in a sufficiently rigorous way every time.

If the mainspan was "tested" in the assembly yard, on in situ when it failed, or in situ weeks later, by doing to it, what I suspect was done to it - over-jacking prestressing strands to failure - then it would have failed each and every time regardless of where it was, or when that was done to it.

The point to note is that over-jacking prestressing strands to failure would never happen in normal use, so doing that would not be a fair test of the structure. Unless you were testing for vulnerability to deliberate sabotage. I suppose some terrorist or wanna-be school-shooter might turn up with a jack to over-jack prestressing strands maybe. So if you were testing for malicious damage then an over-jacking test might be in order.

You would test the mainspan normally and fairly by loading it with lots of weights and, whilst I am the last one to vouch for a failed design, the mainspan would likely have had a much better chance of passing such normal and fair tests.

hardalee · Mar 21, 2018

DaveC426913 said:
Not an engineer. Use small words plz.

Was this a single point of failure that cascaded? Was there no redundancy? That is correct.

Surely there must be multiple design flaws and insufficient redundancy for it to fail so rapidly and catastrophically?

Unknown yet but usually more than one design or construction flaw for a catastrophic collapse.

The tendons were inside 3" diameter steel tubes called ducts. The failure of the tendon or tendons would not have broken he concrete in compression, though a torsion could have been applied. Should not have been a problem if the beam was properly designed with mild steel and steel ties.

My current bet is a void in the concrete diagonal, not seen, or some unspecified cracking.

Ground penetrating radar could have been used to search out voids, maybe it was, don't know.

The lack of redundancy in truss bridges is well known, but usually there are two trusses, not one in the center. This comment is all over the news this morning.

I'm still waiting to hear about the cracking and stressing.

Peter Dow · Mar 21, 2018

Peter Dow said:
There's an article in New Civil Engineer

Experts cite explosive joint failure as cause of Florida bridge collapse

with some nice diagrams.

Experts cite explosive joint failure as cause of Florida bridge collapse
NEW CIVIL ENGINEER 20 MARCH, 2018 BY KATHERINE SMALE
"In its temporary condition, diagonal member 10 (see diagram), was carrying half the dead load of the bridge. The joint at the bottom end of the member was critical as it had to transfer all of the load into a horizontal force in the bottom chord of the truss and vertical force in the supporting pier."

Actually, from the appropriate force vector diagram, we can estimate that the compressive force carried by the diagonal truss member which seems to have failed (number 10 in the NCE diagram = number 11 in the MCM-FIGG proposal page 115 and in my diagram below) is better estimated at about 85% of the dead load of the bridge.

Peter Dow · Mar 21, 2018

hardalee said:
Unknown yet but usually more than one design or construction flaw for a catastrophic collapse.

The tendons were inside 3" diameter steel tubes called ducts. The failure of the tendon or tendons

The "tendon" which failed, I presume by over-jacking, is pictured here.

3104086_collapsedendoffloridabridgecloseupofbluejackpa.jpg

That's the strand / tendon / bar which you reckon is 0.6" diameter, presumably?

It is not possible to see from that photograph whether that failed strand / tendon / bar is in a duct or if so, what size of duct it is.

The only "ducts" that I can see in the proposal pdf illustrated for truss members are the ducts shown surrounding the 1.75" P.T. bars in Sheet B-17 on page 115.

The size of duct illustrated appears to be about 2" inner diameter or maybe slightly bigger.

Admittedly, the deck transverse post-tensioning layout on Sheet B-13 on page 111 seems to show a duct for 0.6" tendons. So there are ducts for 0.6" tendons certainly.

However, I don't see a drawing illustrating either ducts for 0.6" prestressing strands or 0.6" tendons for truss members anywhere, did you and if so where?

Sheet B-2, page 100 mentions "PLASTIC DUCT" in relation to prestressing strands to specify a "FRICTION COEFFICIENT" but not duct size.

hardalee said:
The failure of the tendon or tendons would not have broken he concrete in compression,

The broken end of the strand / tendon / bar (the end not attached to the jack but still attached to the truss joint) would whip back into the joint at high speed smashing into the joint and very likely breaking it.

hardalee · Mar 21, 2018

To help get back on track. The failure of one pt strand in a 3" diameter duct would not have caused failure as is easy to see by calculation. (It carries about 33,000 lbs which is a small fraction of the load. Loading above described as 1.7 times some factor of the dead load is OK) . Something else must have been wrong and tendon may have been the straw that broke the camels back.

The bridge was assembled beside the road in the south side of southwest 8th street and rotated in place. The pick up points had to be adjusted from the original proposal due to obstruction in the road way to the north. This moved the big trucks south and caused a portion of the bridge to cantilever during the assembly which would have necessitated the addition of tendons in the compression member that failed.

We are still waiting for the real smoking gun, the flaw that actually caused the collapse. If it does not show up soon, I intend to do a finite element analysis of the bridge to check what it should have had in it, but without the real reinforcing, it may not show much.

If it had not failed when it did, it might have held until it was full of students and failed then. Not a pretty thought.

Peter Dow · Mar 21, 2018

hardalee said:
To help get back on track. The failure of one pt strand in a 3" diameter duct

Why 3" diameter duct?
Where did you get 3" diameter from?
There is NO mention of a 3" diameter duct anywhere in the proposal pdf that I can see!
Show me, please.

hardalee said:
would not have caused failure as is easy to see by calculation.

The cause and effect of a strand / tendon / bar breaking has got nothing to do the diameter of the duct it is in and everything to do with the diameter of the strand / tendon / bar and how much force (and stored energy) it is carrying when it breaks.

hardalee said:
(It carries about 33,000 lbs which is a small fraction of the load.

On page B-13, page 111 of the proposal pdf, states for a 0.6 inch strand

"STRESSING FORCE SHALL BE 46 KIPS PER STRAND" or 46,000 lbs"

- which is possible with grade 270 KSI.

However, what makes you so sure that the bar which failed was a 0.6" tendon?

Take a closer look in this picture.

That looks much more like a 1.75 inch diameter P.T. BAR than a 0.6 inch diameter tendon to me.

In which case the proposal pdf, on Sheet B-17, page 115, recommends a P.T FORCE/BAR between 200 KIPS and 320 KIPS depending on the truss member concerned.

There is no specification for P.T FORCE/BAR given in the proposal pdf for member #11, which specifies ZERO P.T. bars, but the above photograph shows a bar nevertheless.

I understand that the construction plan intended to DE-STRESS truss member #11 bar after the mainspan was placed in situ.

I don't think that simply destressing the bar would have caused a problem.

Whereas if a 1.75 inch diameter bar was stressed by being over-jacked to failure, it might have failed at a force of something like

270 KSI x 2.4 square inches = 649 KSI

hardalee said:
Loading above described as 1.7 times some factor of the dead load is OK) .

Sheet B-9, page 107 gives the approximate lifting weight of the mainspan as "915 tons"
1 ton = 2000 lbs, So
915 tons = 915 x 2000 = 1,830,000 lbs = 1,830 KIPS
Half weight of the mainspan =
0.5 x 1,830 KIPS = 915 KIPS
1.7 x 915 = 0.85 x 2 x 915 = 0.85 x 1,830 = 1555 KIPS
So the compressive load in member #11 was 1555 KIPS

hardalee said:
Something else must have been wrong and tendon may have been the straw that broke the camels back.

I suggest to you that a 1.75" P.T. BAR over-jacked to failure would unleash an amount of energy rather more destructive than a "straw".

I think it was reported that people heard a loud CRACK and the New Civil Engineer article suggests an explosive failure of the bottom joint so maybe that's what did it?

hardalee said:
The bridge was assembled beside the road in the south side of southwest 8th street and rotated in place. The pick up points had to be adjusted from the original proposal due to obstruction in the road way to the north. This moved the big trucks south and caused a portion of the bridge to cantilever during the assembly which would have necessitated the addition of tendons in the compression member that failed.

We can see one tendon / bar in the photograph attached to the jack.

I am not sure if that is another tendon / bar there, at the upper right of the picture? You can see something that looks like a bar sticking out of a tube, so maybe?

hardalee said:
We are still waiting for the real smoking gun, the flaw that actually caused the collapse. If it does not show up soon, I intend to do a finite element analysis of the bridge to check what it should have had in it, but without the real reinforcing, it may not show much.

You don't think that the jack with the bar sticking out of the truss member shown in the picture is the "smoking gun"?

hardalee said:
If it had not failed when it did, it might have held until it was full of students and failed then. Not a pretty thought.

Indeed.

hardalee · Mar 21, 2018

The bar may have been a solid bar in the photo, not a strand as I assumed from information provided to me on the other pt in the job. I was not informed about that area.The 3" duct was quoted to me by the same person that was onsite for the demolition of the standing section to the south last Friday. I will re-visit with him and see if I can clarify the situation.

After stressing there is a loss due to anchor seating. Then further losses due to elastic deformation, creep and shrinkage. The final effect force I have been using is 33 kips for a .6 tendon approximate final effective force.

No, I don't think the bar or tendon shown is the smoking gun. It is a good observation but not conclusive. It may be the result of the collapse, or something that initiated but not caused the collapse. If the structure were that tinder, it would have been a terrible design error.

Good observation. Yes, that is another bar in the upper right. From the look of the anchor, it looks like they did use bars in this section instead of strands.I will do a few calculations when I have time.

BTW,if bars were used, it is very doubtful bars were 270 ksi as strands are. More than likely 150 ksi threaded dywidag bars.

We will see.

Peter Dow · Mar 21, 2018

Peter Dow said:
Whereas if a 1.75 inch diameter bar was stressed by being over-jacked to failure, it might have failed at a force of something like

270 KSI x 2.4 square inches = 649 KSI

Correction -

270 KSI x 2.4 square inches = 649 KIPS

Peter Dow · Mar 21, 2018

hardalee said:
The bar may have been a solid bar in the photo, not a strand as I assumed from information provided to me on the other pt in the job. I was not informed about that area.The 3" duct was quoted to me by the same person that was onsite for the demolition of the standing section to the south last Friday. I will re-visit with him and see if I can clarify the situation.

OK but I was trying to inform you about the 1.75" diameter P.T. Bars specified for truss members and showed you images from the proposal pdf Sheet B-17, page 115.

Hopefully we are now on the same page.

hardalee said:
No, I don't think the bar or tendon shown is the smoking gun. It is a good observation but not conclusive. It may be the result of the collapse, or something that initiated but not caused the collapse.

Check out this
A Foulmouthed Canadian YouTuber Might Have Solved the FIU Bridge Collapse

hardalee said:
If the structure were that tinder, it would have been a terrible design error.

Whatever the design was, it wasn't idiot-proof therefore it was a terrible design, right?

hardalee said:
Good observation. Yes, that is another bar in the upper right.

Thanks! It looks like it is at a different angle, perhaps from a neighbouring truss member, #10 maybe but that is supposed to have 4 P.T. Bars in it so maybe not, maybe it is a second P.T. bar in member #11?

hardalee said:
From the look of the anchor, it looks like they did use bars in this section instead of strands.I will do a few calculations when I have time.

OK.

hardalee said:
BTW,if bars were used, it is very doubtful bars were 270 ksi as strands are. More than likely 150 ksi threaded dywidag bars.

Sure, in which case if a 150 KSI 1.75 inch diameter bar was stressed by being over-jacked to failure, it might have failed at a force of something like -

2.4 square inches x 150 KSI = 360 KIPS.

hardalee · Mar 21, 2018

Peter Dow said:
OK but I was trying to inform you about the 1.75" diameter P.T. Bars specified for truss members and showed you images from the proposal pdf Sheet B-17, page 115.

Hopefully we are now on the same page.

Check out this
A Foulmouthed Canadian YouTuber Might Have Solved the FIU Bridge Collapse

Whatever the design was, it wasn't idiot-proof therefore it was a terrible design, right?

Thanks! It looks like it is at a different angle, perhaps from a neighbouring truss member, #10 maybe but that is supposed to have 4 P.T. Bars in it so maybe not, maybe it is a second P.T. bar in member #11?

OK.

Sure, in which case if a 150 KSI 1.75 inch diameter bar was stressed by being over-jacked to failure, it might have failed at a force of something like -

2.4 square inches x 150 KSI = 360 KIPS.

hardalee · Mar 21, 2018

"Whatever the design was, it wasn't idiot-proof therefore it was a terrible design, right?"

I never said it was a terrible design. I said that if it failed from one pt break, it was a very tinder structure. Quotes out of place do not express my current opinions.

I am not willing to assign fault until all the data is in. At which time I will point the skinny finger of guilt at those responsible.

hardalee · Mar 22, 2018

Peter Dow said:
OK but I was trying to inform you about the 1.75" diameter P.T. Bars specified for truss members and showed you images from the proposal pdf Sheet B-17, page 115.

Hopefully we are now on the same page.

Check out this
A Foulmouthed Canadian YouTuber Might Have Solved the FIU Bridge Collapse

Whatever the design was, it wasn't idiot-proof therefore it was a terrible design, right?

Thanks! It looks like it is at a different angle, perhaps from a neighbouring truss member, #10 maybe but that is supposed to have 4 P.T. Bars in it so maybe not, maybe it is a second P.T. bar in member #11?

OK.

Sure, in which case if a 150 KSI 1.75 inch diameter bar was stressed by being over-jacked to failure, it might have failed at a force of something like -

2.4 square inches x 150 KSI = 360 KIPS.

DaveC426913 · Mar 22, 2018

Peter Dow said:
The problem with a house of cards is not "insufficient testing" because it fails whenever it is tested in a sufficiently rigorous way every time.

If the mainspan was "tested" in the assembly yard, on in situ when it failed, or in situ weeks later, by doing to it, what I suspect was done to it - over-jacking prestressing strands to failure - then it would have failed each and every time regardless of where it was, or when that was done to it.

The point to note is that over-jacking prestressing strands to failure would never happen in normal use, so doing that would not be a fair test of the structure. Unless you were testing for vulnerability to deliberate sabotage. I suppose some terrorist or wanna-be school-shooter might turn up with a jack to over-jack prestressing strands maybe. So if you were testing for malicious damage then an over-jacking test might be in order.

I see your point. It didn't fail because of a design flaw; it failed because of an implementation flaw*. Which you can't really test for. That's what inspections are for.

*at least, that's how it is beginning to appear

sculptor · Mar 22, 2018

Damage while moving the span into position?

Peter Dow · Mar 22, 2018

There are some interesting engineering tips to be read here at this link, including this interesting slow-motion video.

Peter Dow · Mar 22, 2018

Quotes from Ingenuity on eng-tips.com

"Here is a photo of a closer view of the north-end failure, from NTSB video:"

"And another showing spalling to underside of member #11 - the same side that the PT bar that was being stressed/de-stressed was located:"

If a P.T. Bar fails under high stress the two ends fly apart like the whip from hell. All that energy has to go somewhere, hence the spalling that you see on the underside of member #11.

The question now is - was the failure of P.T. Bar the first event that caused the failure of the bottom joint of member #11, "the critical joint", or conceivably a secondary event that happened subsequent to a prior failure of another component?

Peter Dow · Mar 22, 2018

DaveC426913 said:
I see your point. It didn't fail because of a design flaw; it failed because of an implementation flaw*. Which you can't really test for. That's what inspections are for.

*at least, that's how it is beginning to appear

There is no satisfactory way to "implement" a house of cards. It is an intrinsically precarious structure.

Maybe somewhere there is a house of cards which has stood the test of time, but it is generally understood that the metaphorical reference to a "house of cards" is to compare it with something that is precarious, unstable and prone to failure - in this case the FIU pedestrian bridge.

So I am saying that it was a bad design and so you have missed my point.

If, as it seems the evidence may be pointing to, the bridge failed because of what one worker did in a minute dangling from crane with a jack to a P.T Bar then that proves that the bridge was precarious and so it had a bad design.

A good design should exclude the possibility for one worker doing something inept, whether under orders to do that something inept or not, which causes the collapse of the whole structure.

A good design would build in redundancy so that if one component failed - like a P.T. bar or a truss member or a truss joint - then other P.T. bars or truss members or joints would save the bridge.

Or a good design would use a truss made from rigid metal-only members (tubes or girders) and metal-only joints and avoided the problems of trusses made from prestressed or post-tensioned concrete on such a critical component of a bridge.

Peter Dow · Mar 22, 2018

Damage to the side of truss member #11 is spalling caused by the explosive release of elastic energy which was stored in the highly stressed post-tensioning bar within when it snapped.

This, along with the picture of the jack still attached to the P.T. bar, is the smoking gun.

LaurieAG · Mar 22, 2018

30 years ago I saw my first post tensioned slab and there was just as much reinforcing as a normal slab. 2 years ago I saw one that only had light reinforcement around the pillar caps. 2 weeks ago on cable I saw slabs like this poured and cured in one day before being transported.

I worked as a concrete tester nearly 40 years ago and a couple of years more recently and have seen some jobs that use low grade concrete (20 mpa) and so much styrofoam that you can punch through your slab if you try to jack up a four wheel drive in the garage!

There should be some sort of happy medium trade off between light weight design/fault tolerance because this bridge had a ratio of infinity (divide by zero error).