WHY DID THE FIU BRIDGE COLLAPSE

The FIU bridge tragically collapsed in 2018 killing 4 motorists and 1 construction worker. Since the event many have asked why did the FIU Bridge collapse? what were the causes and who was responsible?

In summary, the National Transportation Safety Board report indicated that the FIU bridge collapsed because of design calculation errors made by the Structural Bridge design consultant Figg Bridge Engineers.

As a Structural Engineer, this explanation only scratches the surface. To ensure that Structural Engineering continues to advance and keep our society safe, its important for our profession to learn from histories tragedies.

In this article I am going to explore exactly what errors were made. I am also going to pinpoint the exact failure mechanism and location on the bridge which caused the eventual collapse. Also we will look at some measures that could have been put into place to prevent the eventual collapse.

For those less patient, (spoiler alert), the Structural reason for the FIU bridge collapse is…

The first Structural reason why the FIU Pedestrian Bridge collapsed was shear failure at the junction where the diagonal truss member joined the bottom deck of the bridge at the northern support node. In-sufficient reinforcing was provided at the construction joint where these two elements met which ultimately caused its collapse.

The second Structural reason why the FIU Pedestrian Bridge collapsed was shear lag at the support node which prevented the post-tensioning cables in the trusses bottom deck from being fully effective and developed to support the tension load demands in the bottom deck.

If you find this article interesting and like to learn more Structural Engineering from forensic examples… Take a look at THIS article which covers a detailed look at the Structural reasons for the collapse of the West Gate Bridge, a significant steel box girder bridge in Melbourne Australia.

To see how we can determine this ourselves using simple hand calculations, grab a coffee and your thinking cap, we are going to be using maths and science to perform a deep dive into the reason the FIU bridge collapsed. But first, we need to understand, the background, how it was built and how it was detailed and the events leading up to the collapse…

FIU Pedestrian Bridge Summary

The Florida International University (FIU) pedestrian bridge collapsed on March 15, 2018 at approximately 1:45pm.

The pedestrian bridge was designed to connect the town of Sweetwater (to the north) to the Florida International University campus in University Park (to the south).

The connection between the Florida International University Campus and the Sweetwater precinct required the bridge to span across 8 lanes of highway traffic as well as the Tamiami Canal.

At the time of the collapse, the longest span above the 8 lane highway SW 8th Street had been constructed and rectification works were being conducted at the northern end. The bridge element was a concrete truss spanning approximately 174 feet (53m) and weighed approximately 930 tons. A team of workers standing on the roof deck of the bridge were re-tensioning PT bars within one of the members (this will later become a critical component in the bridges collapse as discussed later on). The highway was open to public traffic at this time…

A team of construction workers were re-tensioning tension rods within one of the members in the FIU pedestrian bridge immediately before its collapse
A team of construction workers were re-tensioning tension rods within one of the members in the FIU pedestrian bridge immediately before its collapse

The parties involved with the bridges construction and design are as follows:

PartyOrganisation
ClientFlorida International University
BuilderMagnum Construction Management, MCM (previously Munilla Construction Management)
Design EngineerFigg Bridge Engineers
Proof Engineer (Review/Checker)Louis Berger, since Acquired by WSP
Construction Engineer and insepction/monitoringBolton Perez, now Colliers Engineering and Design

The incident received widespread news coverage at the time both locally in Florida and world-wide. The occupational Safety and Health Administration (OSHA) sent its own officials and forensic engineers to determine the cause of the collapse and whether OSHA standards had been breached. During the first week following the collapse, the site was monitored continuously.

Equally, the National Transportation Safety Board (NTSB) sent their own representatives to inspect the remains of the bridge and commence a detailed investigation.

How Many Died in the FIU Bridge Collapse

The tragic collapse of the FIU Pedestrian bridge resulted in six deaths. Of the deceased was a site worker and five were motorists. There were ten injuries (6 serious injuries and 4 minor).

This article is written in dedication to those who were lost or injured in this event as well as their families, friends and loved ones. We pray that the events which caused this tragedy serve as lessons for the engineering and construction industries to ensure such events don’t occur in the future.

Structural form of the FIU Pedestrian Bridge

The FIU bridge was intended to be a two-span continuous bridge structure spanning across 8 lanes of high-way and an adjacent canal.

The structural form of the bridges main spanning elements was a unique arrangement never before used in the region. In elevation view, the form of the spanning members resembled a truss. In cross section view the spanning members resembled an “I” section.

Elevation view of the FIU bridge with structural form resembling a truss element.
Elevation view of the FIU bridge with structural form resembling a truss element.
The cross section view of the main bridge elements resembled that of an "I" beam
The cross section view of the main bridge elements resembled that of an “I” beam

Although it resembled a cable stay bridge, the structural detailing and indeed the structural analysis did not rely upon the cable elements for its structural support. The cables were however intended to contribute to the vibration performance of the bridge in the final built condition.

The apparent “redundancy” of these cable elements can be verified upon review of the structural documentation. The connection between the cables and the roof deck of bridge was a nominal post fixed bolted connection.

Structural Detailing and Construction Methodology of FIU Pedestrian Bridge

The foundation system adopted for the FIU bridge varied between the external support piers and the interior pylon.

The main interior pylon support foundation comprised several precast concrete driven piles tied together with a 5’6″ (1.67m) deep concrete pile cap.

The foundation supporting the southern pier was a series of discrete shallow pad foundations 3 feet (0.91m) thick.

The central main support pylon was a cast in place reinforced concrete element shaped as a trapezoid. The section was widest at the top where it met the underside of the bridge deck and thinnest at its base where it met the top of pile cap…

The main spanning truss elements were comprised of concrete construction using an innovative type of precast construction. Rather than being constructed in a factory and trucked to site, the truss elements were constructed on the side of the road then wheeled into position…

The diagonal members of the truss contained post tensioned rods within them. Not all diagonal members contained tension rods but rather those diagonal members which contained increased tension forces. The post tensioned rods were jacked from the trusses roof deck. To allow development of the rods as well as allow space for the spiral and end plate, additional concrete was provided at each top node where the diagonals and the roof deck met. These are referred to as “blisters” in the structural documentation.

Additional concrete in the form of "blisters" was provided at each top node location where the diagonal members med the roof deck of the FIU truss bridge.
Additional concrete in the form of “blisters” was provided at each top node location where the diagonal members med the roof deck of the FIU truss bridge.

The support condition of the main truss differed between temporary construction/transportation and the final built case. The locations of the dolly wheels which pulled the truss into position were positioned directly beneath the first internal bottom nodes of the truss. Whereas the support condition in the final built case was the bottom nodes at the extreme ends of the truss. For this reason, tension rods were also provided within the last diagonal members of the truss…

Illustration of support condition and force within last diagonal members of the FIU bridge truss, during construction (top) and in the final built case (bottom)
Illustration of support condition and force within last diagonal members of the FIU bridge truss, during construction (top) and in the final built case (bottom)

Because the post tension rods were only require during temporary transportation of the truss section, they were scheduled to be de-stressed after the truss elemental was placed at its final location.

Elevation view of post tension rod arrangement for diagonal members 11 and 10, not all diagonal members were provided with PT rods (notably member 9 illustrated here), the rods within member 11 were to be de-stressed once the truss was positioned at its final location.
Elevation view of post tension rod arrangement for diagonal members 11 and 10, not all diagonal members were provided with PT rods (notably member 9 illustrated here), the rods within member 11 were to be de-stressed once the truss was positioned at its final location.

The trusses roof deck and bottom deck were both post-tensioned concrete. The roof deck contained post-tensioning cables in the longitudinal direction only, while the bottom deck contained PT cables in the longitudinal and transverse directions. For a full explanation on what post-tensioning is and how it works, see the article located via THIS link.

Cross section view of the roof deck of the FIU bridge  truss with post tensioning cables highlighted in green, in the longitudinal direction only.
Cross section view of the roof deck of the FIU bridge truss with post tensioning cables highlighted in green, in the longitudinal direction only.
Cross section view of the bottom deck of the FIU bridge truss with post tensioning cables highlighted in green, in the longitudinal and transverse directions.
Cross section view of the bottom deck of the FIU bridge truss with post tensioning cables highlighted in green, in the longitudinal and transverse directions.

Due to the form of the truss and the nature of concrete construction, the truss was required to be constructed in three separate pours:

  1. Bottom Deck
  2. Diagonal Members
  3. Roof Deck

As a result, this meant that construction joints were inherently required between each staged portion.

For more on construction joints and how to check different construction joint types, take a look at THIS article.

Image showing the main truss span of the FIU bridge and its formwork on the side of the roadway near the construction site
Image showing the main truss span of the FIU bridge and its formwork on the side of the roadway near the construction site
Image of the FIU bridge main span truss during construction showing reinforcing, PT rods and PT cables.
Image of the FIU bridge main span truss during construction showing reinforcing, PT rods and PT cables.
Elevation view of the northern support node of the main span truss of the FIU bridge indicating construction joint between the bottom deck and vertical/diagonal members.
Elevation view of the northern support node of the main span truss of the FIU bridge indicating construction joint between the bottom deck and vertical/diagonal members.

Events Leading up to the FIU Bridge Collapse and the Collapse itself.

The following timeline outlines the events which occurred to the leadup of the 15th of March 2018 collapse of the FIU Pedestrian bridge.

February 26, 2018: After the construction of the truss element on the side of the road, the temporary shoring was removed. Soon after the shoring was removed, cracking was evident at several locations on the truss. Bolton Perez and Associates (BPA, now Colliers Engineering and Design), monitoring and inspection engineer identified the cracks and took photos.

February 28, 2018: The photos of the cracks were issued to the builder MCM for action and attention by the Engineer on Record (EOR) FIGG engineers. On the same day MCM forwarded the photos to FIGG.

PHOTO 1 - Evidence of cracking at the FIU Pedestrian bridge after shoring was removed (refer later for location identified on elevation)
PHOTO 1 – Evidence of cracking at the FIU Pedestrian bridge after shoring was removed (refer later for location identified on elevation)
PHOTO 2 - Evidence of cracking at the FIU Pedestrian bridge after shoring was removed (refer later for location identified on elevation)
PHOTO 2 – Evidence of cracking at the FIU Pedestrian bridge after shoring was removed (refer later for location identified on elevation)
PHOTO 3 - Evidence of cracking at the FIU Pedestrian bridge after shoring was removed (refer later for location identified on elevation)
PHOTO 3 – Evidence of cracking at the FIU Pedestrian bridge after shoring was removed (refer later for location identified on elevation)
PHOTO 4 - Evidence of cracking at the FIU Pedestrian bridge after shoring was removed (refer later for location identified on elevation)
PHOTO 4 – Evidence of cracking at the FIU Pedestrian bridge after shoring was removed (refer later for location identified on elevation)
PHOTO 5- Evidence of cracking at the FIU Pedestrian bridge after shoring was removed (refer later for location identified on elevation)
PHOTO 5- Evidence of cracking at the FIU Pedestrian bridge after shoring was removed (refer later for location identified on elevation)

March 7, 2018: FIGG Responded to the photos indicating the cracking. The following outlines a summary on their assessment of the photos.

  • Photo 1: No Structural concern
  • Photo 2: No concern
  • Photo 3, 4 and 5: These were the cracks evident at the bottom of the diagonal member 11. No concern was expressed by FIGG however FIGG indicated that the contractor MCM will seal them in accordance with FDOT standard specifications.

March 10, 2018: The truss was transported (wheeled) into its final position and placed on the pier and pylon supports. At this time, no other cracks were evident other than those outlined previously on the 7th of March 2018.

VSL then sent representatives of their post-tensioning team to de-stress the bars in diagonal members 2 and 11. (as outlined previously, the tension force in these members were only required temporarily during transportation of the truss segment). First the PT bars in member 2 were de-stressed, then the crew began de-stressing the bars within member 11.

As they began the de-stressing process, cracks began to appear at multiple locations around the node point where the bottom of member 11 met the bottom deck of the truss. Representatives from VSL took photos of the cracks then issued them to their supervisor. The photos were included with a comment from a VSL site representative that during de-stressing “it cracked like hell”.

March 12, 2018: MCM employees took their own pictures of the cracks. The photos are sent to FIGG engineers for comment. In the email, the MCM representative commented that “some of these cracks are rather large and/or of concern” and “your immediate attention and response is required”.

Image taken vertical view at the FIU bridge truss bottom deck where it meets member 11
Image taken vertical view at the FIU bridge truss bottom deck where it meets member 11
Image taken vertical view at the FIU bridge truss bottom deck where it meets member 11 (opposing side of vertical member compared to previous image)
Image taken vertical view at the FIU bridge truss bottom deck where it meets member 11 (opposing side of vertical member compared to previous image)
Image taken at the northern face of the FIU bridge truss bottom deck where the bottom deck meets member 11
Image taken at the northern face of the FIU bridge truss bottom deck where the bottom deck meets member 11

March 13, 2018: After reviewing the photos, FIGG instructed MCM at around 9:45am to put an additional plastic shim similar to the existing shims on the pylon directly under the bottom deck of the truss.

At 5:18pm, FIGG confirmed again that they had evaluated the cracks and no safety issue was present.

FIGG gave additional instruction to MCM to re-tension the two PT bars in diagonal member 11 which had previously been de-stressed.

FIGG further instructed that the bottom deck of the truss should be closely monitored during the PT bars being stressed to ensure the crack size does not increase. It also noted that while no safety issue was evident, the re-stressing works should be undertaken as soon as possible. By this stage, the already evident cracking had appeared to become worse (both in terms of width and extent).

Additional photos were taken of the cracks and issued to FIGG via email.

Image taken from vertical view of the truss bottom deck three days after the PT rods in member 11 were de-stressed
Image taken from vertical view of the truss bottom deck three days after the PT rods in member 11 were de-stressed
Image taken from vertical view of the truss bottom deck three days after the PT rods in member 11 were de-stressed (opposing side of vertical member compared to previous image)
Image taken from vertical view of the truss bottom deck three days after the PT rods in member 11 were de-stressed (opposing side of vertical member compared to previous image)
Close up view of the cracking evident on the truss bottom deck showing severity and width f crack.
Close up view of the cracking evident on the truss bottom deck showing severity and width f crack.
Close up Elevation view of diagonal member 11 showing crack where the member meets the truss bottom deck of the FIU bridge.
Close up Elevation view of diagonal member 11 showing crack where the member meets the truss bottom deck of the FIU bridge.
Elevation view of diagonal member 11 showing spalled concrete on its face.
Elevation view of diagonal member 11 showing spalled concrete on its face.

March 14, 2018: FIGG spent the day at its office analysing the cracks as they appeared on the latest photos sent from MCM.

MCM placed the recommended shim (steel instead of plastic however) under the bottom deck of the truss as previously instructed by FIGG.

VSL who was concerned about the cracks wanted FIGG to analyse them to determine if an epoxy could be used as a measure of repair and asked MCM if the cracks would be repaired before re-tensioning the PT rods in member 11. MCM responded that the cracks would not be repaired prior as under FIGG’s assurance, the cracks did not present any safety issues.

MCM instructed VSL to proceed with the preparation to re-tension the PT bars in diagonal member 11. MCM organised a crane to be present on March 15, 2018 to facilitate the works.

March 15, 2018: At approximately 7:45am, two structural engineers from FIGG attended the site to inspect the bridge and the cracks. The inspection was performed both on foot on the bridge deck and from an elevated platform from underneath the bridge deck. Present at the time were also a representative from MCM and BPA.

It was reported that one of the FIGG engineers commented that the cracks appeared worse than in the photos.

At approximately 9am on site, a meeting was held with attendees from FIGG, FIU, MCM, FDOT and BPA being present. FIGG was to present its findings and recommendations to the group regarding the cracks.

During the meeting, FIGG presented their opinion on the cracking which was witnessed on the bridge deck which included:

  1. The cracks do not present any safety concern as per FIGG’s evaluation
  2. This is a temporary loading condition during construction until the intermediate pylon and back span truss are completed.
  3. MCM must expedite pouring of intermediate pylon and construction of back span truss.
  4. After the construction of the back span, the structural behaviour of the main span would change the forces in diagonal member 11 and the node at 11/12 will be reduced. The intermediate pylon and back span will be integrated with the existing diaphragm and the main span and the reserve strength will be increased.
  5. Spalled areas have not been replicated by the engineering analysis
  6. The spalled areas are minor

Later that day at approximately 1:45pm on the 15th of March 2018 the main truss span collapsed while the VSL workers were completing the re-tensioning of member 11.

The collapse was sudden and dashcam footage indicates the collapse mechanism beginning at the bottom node of member 11.

At the time of the collapse the roadway beneath was open to public traffic. Multiple eastbound vehicles were stationary under the span while waiting at traffic lights. Eight cars were crushed under the wreckage.

A driver who survive the collapse reported that small rocks fell onto her car just before the front of her car was crushed by the span.

Tragically six people were killed in the collapse and ten were injured. One of whom was a VSL worker who was part of the crew that were re-tensioning the PT rods from the roof deck of the truss.

In the months following the event, a detailed investigation was undertaken by the national Transportation Safety Board (NTSB).

The inquiry and subsequent report found that while different parties each shared varying levels of responsibility for the collapse of the FIU pedestrian bridge, the main of which was the bridge designer FIGG engineers; every company, institution and agency involved in the project was partly to blame for the bridges collapse.

Engineering Analysis of the FIU Pedestrian Bridge Collapse.

We will now perform our own analysis on the strength and adequacy of the FIU bridge at the time of its collapse. We will be focusing our check on the bottom node of member 11 as this is where the cracking had been reported and where the initial stages of the collapse was evident on the dashcam footage.

When checking collapse mechanisms as apposed to code compliance it is best practice to not apply safety factors to the analysis. This is because we are after the real world behaviour of the bridge as close as possible to what the realities were at the time of collapse. Inclusion of safety factors can give misleading causes to a collapse depending on the conservatism of the safety factor.

For this reason the calculated loading on the FIU bridge at the time of the collapse will be assessed as the un-factored working load. Equally, the concrete strength and reinforcing strength capacities will not be reduced but rather their un-factored values will be adopted.

The bridge form is a simple truss structure. While we will not go into the details on truss theory in this article, you can refer to an in-depth first principles approach to truss design by following THIS link.

Loading up the FIU Pedestrian Bridge.

At the time of the collapse, we know that the main truss span was relatively un-loaded. The only apparent loading that the truss was supporting included:

  • Self-weight of the truss itself (concrete dead load)
  • Nominal safety handrail loading around the perimeter
  • Circa 3-4 construction workers operating at the top of member 11.

From images taken from the roof deck of the truss while the VSL workers were re-stressing the rods, the weight of the jack was being supported by the adjacent mobile crane next to the bridge span.

Compared to the self-weight component of the truss span, the handrail loading and construction worker loading are negligible in magnitude and therefore will not be considered as part of our loading assessment.

Using the structural documentation we can determine the cross-sectional area of the bottom deck and roof deck of the truss segment. For accuracy, I have scaled off the structural documentation using Bluebeam Revu’s area calculation function…

Cross section view of the FIU bridge truss member,  top roof deck and bottom deck cross sectional areas indicated
Cross section view of the FIU bridge truss member, top roof deck and bottom deck cross sectional areas indicated

Also from the structural documentation we can see that the clear span of this truss segment is circa 53.3m (174.9 ft)…

Temporary construction view of the FIU main truss span, span length indicated.
Temporary construction view of the FIU main truss span, span length indicated.

Multiplying the cross section areas of the roof deck and bottom deck of the truss with the trusses span gives us the total volume of these components. The density of concrete I have chosen is 25kN/m3 (159.1 lbs/ft3). Multiplying the volume by the density gives us the total weight of the roof and bottom deck…

Next we’ll move onto the self-weight of the diagonal members of the truss. Half the lineal length of the diagonal members contribute to the reaction force at the northern pylon support. The length of the diagonals can be scaled off from the drawings. The cross-sectional area is also indicated in the details within the drawings. Multiplying the tributary volume of the diagonal members again by the density of the concrete gives us the total weight of these elements…

Elevation view of the FIU pedestrian bridge, tributary length of diagonal members which contribute to the northern reaction force highlighted green
Elevation view of the FIU pedestrian bridge, tributary length of diagonal members which contribute to the northern reaction force highlighted green
Cross section view of diagonal member with dimensions indicated (taken from structural documentation of the FIU Bridge)
Cross section view of diagonal member with dimensions indicated (taken from structural documentation of the FIU Bridge)

The final concrete mass within the truss are the concrete blisters which provided clearance and development to the PT rods at the roof deck of the truss. Again, we can determine that around half of the blisters contribute to the reaction force on the northern pylon. The dimensions for the blister elements can be taken from the structural documentation.

Elevation view of the FIU pedestrian bridge, blister elements which contribute to the reaction force at the northern end highlighted green
Elevation view of the FIU pedestrian bridge, blister elements which contribute to the reaction force at the northern end highlighted green
Detail view of blister element (highlighted green) taken from structural documentation of the FIU bridge
Detail view of blister element (highlighted green) taken from structural documentation of the FIU bridge

Adding all the concrete components together we can now determine the total reaction force at the northern pylon due to the self-weight of the spanning truss in the FIU pedestrian bridge…

Overall colour coded view of the FIU pedestrian bridge dividing each concrete component in groups (top/bottom deck, diagonals and blisters)
Overall colour coded view of the FIU pedestrian bridge dividing each concrete component in groups (top/bottom deck, diagonals and blisters)

Determining the member Forces within the FIU Pedestrian Bridge.

We can now focus directly on the northern support where the bottom node of member 11 is located. Resolving the reaction force to its horizontal component gives us the tension load which is placed on the bottom deck of the truss…

Bottom node of member 11 idealised as a series of vectors meeting at a single node for the support, bottom deck and diagonal member.
Bottom node of member 11 idealised as a series of vectors meeting at a single node for the support, bottom deck and diagonal member.

The tension force within the bottom deck of the truss is a direct reaction of the compression force applied by member 11 to the support node. The coupling of the tension in the bottom deck and compression in member 11 is what keeps the support node stable and in abeyance with the laws of equilibrium (or statics).

For equilibrium to be satisfied, the compression load from member 11 needs to “enter” the support node. Before it can enter the support node, it first must cross the construction joint which lies at the interface of the bottom deck and member 11.

You will recall from previous sections that the FIU pedestrian bridge was constructed in stages which required introduction of these construction joints (also referred to as a cold-joint within the concrete)…

Cross section view of member 11 node at support indicating construction joint, node location and generated shear plane.
Cross section view of member 11 node at support indicating construction joint, node location and generated shear plane.

Due to this detail which is unique to a bridge truss of this form and material, a shear interface check needs to be performed immediately above the support node. The majority of truss theory involves checking members for simple tension or compression, a non-experienced structural engineer may easily overlook this requirement. Lets now look at the capacity of this interface to support the applied shear loads.

Calculating the shear Capacity of the Construction Joint Between member 11 and The bottom Deck of the Truss

The elements resisting the shear force at the construction joint are:

  • The friction of the concrete-on-concrete interface where member 11 and the bottom deck meet, the overlapped area is the effective friction surface.
  • The shear capacity of any reinforcing bars in dowel action which pass through the construction joint.

Both the US and Australian concrete design codes provide guidance on how to check the capacity of such a joint.

In the US code (ACI 318-08) the following equation is given in chapter 11.6…

In the Australian Concrete Bridge Code (AS5100.5) the following equation is given…

While these formulas look quite different, they are approaching the capacity check in a similar manner. The main difference between the Australian code and the US code is that the Australian code is determining the shear stress of the interface, whereas the US code calculates the shear force (stress simply being force/shear surface area).

Considering that the FIU pedestrian bridge is located in Florida US, we will use the US concrete code ACI 318-08.

Chapter 11.6 is where the previous equation is found. Lets unpack the different coefficients within this equation…

The equation above caters for reinforcing bar passing through the joint at a perpendicular angle. However there were also reinforcing bars arriving at the node at an angle equal to that of the angle of inclination of member 11. ACI 318-08 also caters for this scenario, the equation is very similar however with an angular modification (all other coefficients remain the same as the previous equation)…

Lets now look at the quantity of reinforcing (both perpendicular and at 32 degree angle) which is passing through the construction joint. In terms of vertical reinforcement, the end vertical element of the truss acted as a column and was provided with its own reinforcing bars.

Locally at the joint, it appears the engineer provided joint reinforcing bars in addition to the columns own longitudinal bars. The US concrete code however requires that all effective reinforcement needs to be fully developed with enough embedment either side of the joint. This unfortunately means that the additional joint reinforcing bars cannot be considered as contributing to the joints shear capacity…

Elevation view and plan view of member 11 base node indicating effective reinforcement which passes through the construction joint with the truss bottom deck.  Bars indicated with purple dash line not effective in reinforcing the joint for shear.
Elevation view and plan view of member 11 base node indicating effective reinforcement which passes through the construction joint with the truss bottom deck. Bars indicated with purple dash line not effective in reinforcing the joint for shear.

The joint reinforcing bars are indicated as a dashed purple line in the image above. As you can see, they do not continue above the construction joint very far at all. The effectiveness of these bars (or lack thereof), is evident when examining the crack pattern as indicated in the photo below.

There are a total of 9 reinforcing bars of size #7 passing through the joint as well as 3 reinforcing bars of size #11.

Now lets look at the additional reinforcement passing through the joint from member 11…

Elevation view (top) of base node at member 11 and cross section view (bottom) of member 11, indicating reinforcing both within member 11 and at the construction joint
Elevation view (top) of base node at member 11 and cross section view (bottom) of member 11, indicating reinforcing both within member 11 and at the construction joint

There are two 1.75 inch stressing rods passing through the construction joint. Although they don’t continue far below the construction joint, they are provided with an imbedded anchor plate which would make them at least partially effective.

there are also 8x #7 reinforcing bars within member 11 which pass through the construction joint. However these are not fully developed with adequate embedment below the joint and cannot be considered to contribute to the joints shear capacity.

Here is a summary of the total effective reinforcing passing through the construction joint…

Bar QuantitySizeAreaTotal Area
Column Longitudinal9#70.6 inch27.2 inch2
Column Longitudinal3#111.56 inch24.68 inch2
Member 11 Stressing Rod21.75 inch2.58inch25.16 inch2
Total 90 Degree Bars11.88 inch2
Total 32 Degree Bars9.96 inch2

The yield strength of the conventional reinforcing bar was 60,000 psi while the tension rods were approximately 153,740 psi. The contact area of the joint can be readily obtained through the measurements on the drawing…

Elevation view of the member 11 base node, contact area of construction joint indicated.
Elevation view of the member 11 base node, contact area of construction joint indicated.

We now have all the value we need to check the joints overall capacity, lets start with the perpendicular reinforcement component…

In addition to this, we add the capacity from the stressing rods from member 11…

The code also allows for the assisting effects of compression forces acting perpendicular to the joint. This can greatly increase the joints overall shear capacity. The compression force (or clamping force) is equal to that of the vertical reaction at the support node. This needs to be multiplied by the coefficient of friction (which is 0.6 due to no deliberate roughening being undertaken to the concrete surfaces). Therefore, adding the contribution of the vertical reinforcement and the angled reinforcement as well as the assisting effects of the compression force, we can determine the shear capacity of the joint…

We need to also consider the effects that the force within the PT rods had on the joint. The behaviour of the PT rods when stressed would have acted in two ways:

  • The horizontal component of the stressing load would have increased the shear demand across the construction joint
  • The vertical component of the stressing load would have increased the shear capacity of the joint through clamping effect.

So we can see that the stressing rods would have both “helped” and “hurt” the overall capacity of the joint. Lets account for both components in our analysis to obtain our final utilisation.

Each stressing rod was tensioned to 280 kips or 280,000lbs. We take the horizontal stressing load component as follows…

Now we take teh vertical component of this load as follows…

Multiplying the vertical component by the coefficient of friction we get 148,377 x 0.6 = 89,026 lbs of additional assistance to the joint. However the demand on the joint is increased by 237,453 lbs.

Joint Shear CapacityShear Load On JointPercentage Utilisation
2,658,581 lbs (11,825 kN)1,780,415 lbs (7,919 kN)66%

Note that this is with no safety factors applied at all. This would explain why the first signs of cracking on February 28, 2018 did not appear significant or of concern. The curing and stressing sequence of the different truss elements is un-clear. The initial cracks which were evident may have been a result of shrinkage of the concrete elements during construction.

The most significant cracking occurred after the stressing rods were de-stressed and the bridge was spanning across the road-way.

A simple modification to our calculated joint capacity now tells a different sotry…

Now we compare the joint shear capacity against the shear load on the joint (without the clamping load from the stressing rods or the increased shear demand caused by the stressing rods)…

Joint Shear CapacityShear Load On JointPercentage Utilisation
1,644,570 lbs (7,315 kN)1,542,962 lbs (6,850 kN)94%

Noting that no safety factors are used in calculating the utilisation of 94%, this not only indicates that the construction joint design was non code compliant but also quite likely the cause of the bridges collapse. The inclusion of safety factors in Structural Engineering deign caters for unknowns and potential things that may not go according to plan when comparing the theoretical design with the as-built condition. With safety factors applied, the percentage “real life” utilisation should be somewhere between 50%-60% (i.e. the design joint capacity is twice that of the load demand, or a safety factor of 2)

Lets take a look at some things which may influence the utilisation of the joint that aren’t accounted for in this analysis..

  • The PT rods are housed within hollow ducts. This allows the rods to be stressed and de-stressed. The hollow ducts would have reduced the concrete-on-concrete contact area of the joint, therefore reducing its capacity. To make maters worse, because the rods enter the construction joint at an angle, the hollowed out zone is elongated further reducing the contact area.
  • The shear load demand on the joint assumes that the truss was constructed with perfect geometry in accordance with the drawings and no overpour of concrete occurred.
  • There were cast-in pipes and other services within the bridge, if any one of these passed through the construction joint, this would have reduced the effective contact area of the joint.
  • All material strengths including reinforcing bar and concrete would need to be at the required design strength.
  • Any defects in the concrete near the joint would have effected its capacity through inadequate vibration/compaction etc.

The numbers we have ran so far is in keeping wit the behaviour of the truss and the cracks which were evident both before and after the PT rods were de-stressed.

Why did the FIU Bridge Collapse when the PT Rods were Re-Stressed?

The next logical question would then be, if the PT rods were originally providing significant contribution to the joints capacity, why did re-stressing them actually cause the bridge to collapse?

To answer this, we need to conduct a thought experiment on what would have been occurring at the construction joint when the rods were de-stressed and then re-stressed again.

In the first instance, the rods were stressed and from our previous calculations the joint was adequate to support the load demand…

Elevation view of the support node on the FIU bridge, construction joint indicated as well as effective reinforcing bar passing through the joint
Elevation view of the support node on the FIU bridge, construction joint indicated as well as effective reinforcing bar passing through the joint

When the rods were de-stressed, our calculations indicate that the joint had essentially failed. This would have meant loss of effective contact area between concrete and yielding of the main effective bars which passed through the joint (and in some cases some of the reinforcement may have failed altogether).

Elevation view of the support node on the FIU bridge after PT rods are de-stressed, slippage of the joint likely occurred causing the concrete-on-concrete interface to fail as well as yielding or failure of the reinforcing bars.
Elevation view of the support node on the FIU bridge after PT rods are de-stressed, slippage of the joint likely occurred causing the concrete-on-concrete interface to fail as well as yielding or failure of the reinforcing bars.

From our previous calculations, we determined that the force within the rods when they are stressed applies more load demand on the joint than assistance. This process may have contributed to further failure of the concrete contact area as well as the vertical reinforcing bars. Therefore even though the PT rods are now providing assistance to the joint, the remaining elements such as the concrete interface and the conventional reinforcing bars may not have been effective. At this point, we can determine a rough percentage utilisation of the joint…

Joint Shear CapacityShear Load On JointPercentage Utilisation
1,503,475 lbs (6,687 kN)1,780,415 lbs (7,919 kN)118%

The percentage utilisation leaves little doubt that the joint was inadequate following the re-stressing procedure of the PT rods.

Further Inadequacies in the Design.

Lets say that the engineer had either not specified for the PT rods in member 11 to be de-stressed, or provided enough properly developed reinforcing bar through the joint.

The load would from member 11 then have effectively “entered” the support node. However the corresponding tension load in the bottom deck of the truss needs to adequately “exit” the node for everything to remain stable. Unfortunately there was an inadequate load path to allow the PT tendons within the bottom deck of the truss to be fully engaged and support this tension load close to the node…

Shear Lag at the Member 11 Node

The width of the diagonal member 11, and therefore the node, was only 19 inches. The width of the truss bottom deck was over 31 ft.

In truss theory, all the tension reinforcement needs to be developed and anchored into the footprint of the node. Post tensioning cables were provided in the bottom deck of the truss. However they were evenly spread across its width and no cables were provided directly in the middle of the deck (where the truss nodes are located).

Elevation view of the bottom deck of the truss for the FIU bridge.  Detailing indicates that the PT cables were not placed within the node region but evenly spread across the decks width.
Elevation view of the bottom deck of the truss for the FIU bridge. Detailing indicates that the PT cables were not placed within the node region but evenly spread across the decks width.

This prevented the PT cables from being properly engaged to support the tension load within the bottom deck of the truss. A pull-out failure mechanism is then produced (or cone failure) as the horizontal force component of member 11 tries to thrust the node out of the truss…

Isometric view of the support node of the FIU bridge.  Without adequate PT cables within the node region, a pull-out of cone failure begins to result, this image indicates the crack formation which occurs as a results.  PT cables lying outside of this cone failure zone are not effective in supporting the required tension load caused by diagonal member 11.
Isometric view of the support node of the FIU bridge. Without adequate PT cables within the node region, a pull-out of cone failure begins to result, this image indicates the crack formation which occurs as a results. PT cables lying outside of this cone failure zone are not effective in supporting the required tension load caused by diagonal member 11.

The PT cables which lie within the cone failure zone are effective in supporting the tension load in the deck produced at the node. Based on the geometry of the node and the truss, only the first two PT cables are likely to have been able to contribute to supporting this thrust load. The capacity of these cables were in-sufficient on their own without the assistance from the adjacent cables. The effectiveness of these two cables is also questionable as they are required to be fully developed almost immediately from the end face of the deck…

Isometric view of support node in fully failed condition indicating extent of cone failure (pull-out failure) on the FIU pedestrian bridge.
Isometric view of support node in fully failed condition indicating extent of cone failure (pull-out failure) on the FIU pedestrian bridge.

Some of the more serious cracks which were evident after the PT rods were de-stressed appear to resemble this cone type pull out failure mechanism.

Actual image of the bottom deck at the support node which appears to indicate evidence of cone failure on the FIU pedestrian bridge.

It would appear that the failure mechanisms were localised around the northern support node at the base of diagonal member 11. Failure was likely to have been brought about through a combination of pull-out cone failure due to inadequate tension reinforcement at the node in the bottom deck as well as inadequate developed reinforcement crossing the construction joint between the bottom deck and diagonal member 11.

Will the FIU Bridge be re-built again?

On May 6, 2020, FDOT announced plans to design and rebuild the FIU bridge with guidance from the NTSB. The design stage is scheduled to begun in 2021 an last for two yeas, with further two years estimated to construct the bdridge.

Published by

Quentin Suckling is a full time practicing Structural Engineer based in Melbourne Australia. He has been practicing in the local market at tier 1 engineering consulting firms over the last 16 years.

8 thoughts on “WHY DID THE FIU BRIDGE COLLAPSE

  1. the engineers were incompetent and should be tried for manslaughter. expect much more of this in the coming decades. diversity is not a bridge’s strength.

  2. Thanks for sharing Quentin. Note that the stress bars in member 11 were in ungrouted ducts and as such i’m not so sure you could rely on their shear capacity to strengthen the joint in dowel action.

    My guess is the top and bottom chord provided some minor capacity in flexure to transfer some of the load back to the support (although not designed for this load path) which is why it didn’t fail when the cracking propagated initially. Once the bars were reloaded and stress increased on the node it finally gave up.

Leave a Comment

%d bloggers like this: