The first sign something was terribly wrong with Florida International University’s ambitious, unfinished pedestrian bridge over Tamiami Trail came nearly three weeks before it collapsed onto idling cars and killed six people.
On Saturday, Feb. 24, 2018, the 174-foot span, manufactured on site at the side of the trail, was still resting on temporary supports just off the ground. As crews stripped off the framing pieces used to shape and support the concrete structure, a loud cracking sound rang out — so sharp that startled workers ran out from under the span.
“As we were stripping it, I looked out of the office and everybody was out from underneath the bridge. I said, ‘Oh, shit,’ “ John Jackson, a supervisor with subcontractor Structural Group of South Florida, later told an investigator from the National Transportation Safety Board, which is expected to rule early next week on what caused the collapse. “So, I ran out there and everybody said, ‘Aw man, aw man, the bridge went whoop like that.’ ... I said, ‘Well, you guys stand down. I’ll go up on top of the bridge to find out what’s up.’ So, I get up there and both trusses on each end [had] cracked at the bottom where it meets the bridge.”
Jackson told a supervisor for the project’s general contractor, Munilla Construction Management, about the alarming incident — but FIGG Bridge Engineers, the Tallahassee-based firm that designed the bridge and was ultimately responsible for ensuring its structural soundness, gave the green light to keep working.
“I thought it was very unusual and we were very leery,” Jackson said. “[The cracks] were significant.”
The cracks were indeed significant, a symptom of a hidden and catastrophic weakness in a key structural connection at the bridge’s north end that would end in the March 15, 2018, collapse, according to evidence gathered by federal investigators.
But over the next two weeks, as cracks spread and expanded and a chunk of concrete split off a support column, no key project participant acted to exercise independent judgment or oversight, more than 6,000 pages of records released by the National Transportation Safety Board last week show. Not FIGG, the designers. Not MCM, the contractor. Not the Florida Department of Transportation, which supervised the project, or FIU, which conceived it.
In interviews with contractors and crew members, as well as construction records, analytic reports and photographs, the new trove of documents paints a detailed picture of organizational dysfunction.
Even as inspectors and subcontractors dutifully documented cracks that grew to proportions that the Federal Highway Administration labels as “abnormal” in a newly released document, project leaders hewed to contractually and strictly defined roles. The NTSB documents show no project leader discussed suspending work, even thought to ask for an independent evaluation or raised red flags that might have prevented the collapse or at least shut down the busy road under the bridge.
The lack of urgency reflects two major factors that the NTSB will likely focus on as contributing to the disaster: a faulty design by FIGG that left the bridge critically weakened at that crucial point, and a paralyzing culture of group-think that prevented the team behind the project from recognizing that their signature bridge had become a 950-ton deathtrap.
In the end, the records and interviews show, no one ever raised the possibility of shutting down Tamiami Trail as the team embarked on an ill-considered attempt to repair the cracks that prompted the bridge to crumble and crush cars below, killing five motorists and one bridge worker and injuring 10 others.
David Beck, a veteran New Hampshire engineer who independently analyzed FIGG’s bridge plans for the Miami Herald and reviewed the federal docket, said no one on the project team seemed to have a comprehensive overview of how the bridge was supposed to work.
“No one designer had a full understanding and appreciation of the structure and mechanics of the main span,” Beck said. “No one from FIGG could see what was going wrong. And there was no follow-through on the quality control. No one was studying the cracks in a serious way. No one had the knowledge or took the initiative to understand the danger these cracks presented.
“Anytime you want to mess something up in engineering, you get a whole bunch of people involved a little bit. This bridge is a prime example.”
The systemic breakdown in oversight became abundantly clear last week, when the NTSB released the full docket from its 19-month investigation. The Miami Herald had sued — winning a decision in state court but losing at the federal level — to force the release of some of the records last year.
On Tuesday, at a hearing in Washington, D.C., the NTSB is expected to issue its final determination on what caused the bridge to collapse. It will also make safety recommendations to prevent similar accidents from happening again.
FIGG disputes the evidence collected by federal agencies, including the contention that its design was flawed. In a press release, issued after the NTSB records were released last week, it called the accident the result of “a failure by contractors.”
At the same time, the Florida Department of Transportation acknowledged for the first time that the road should have been closed. The state agency, which had consistently distanced itself from the bridge despite significant involvement, also announced an overhaul of how it supervises such projects that it says will bring more oversight and accountability.
Web of dysfunction
The pattern Jackson described was repeated time and again as the $14.2 million federally funded bridge began to fail in plain sight.
Inspectors and contractors on site relayed reports and photos by email of spreading cracks to FIGG, only to receive reassurances that they were within expected parameters and posed no safety threat — conclusions no one questioned or challenged.
What the cracks should have told project team members is that the bridge span was in serious structural distress, engineers from three separate federal agencies have now concluded. New assessments from the Federal Highway Administration echo a preliminary report by the NTSB issued earlier and a separate report by the Occupational Safety and Health Administration agree that the distress stemmed principally from a series of errors by FIGG engineers in the design and review of the bridge structure.
Those errors caused the engineers to “significantly” underestimate the forces that the bridge’s dead weight put on the connection at its north end while underestimating the strength of the joint to withstand those forces. Fatally, they also failed to understand what was producing the cracking and what it portended, documents included in the federal docket show: namely, that the structure’s north end was coming apart as horizontal support columns split from the bridge deck.
Those errors were compounded when FIGG ordered crews on March 15 to tighten internal steel support rods in a key diagonal strut at the north end in a poorly conceived attempt to close up the cracks, documents show. The force of that re-stressing operation caused the damaged, badly overtaxed connection to fail abruptly, sending the span crashing down to the roadway below, where motorists were stopped at a red light.
The cracks by then had grown to such an extent that one inspector sent by a subcontractor to observe the repair work the day of the collapse called them “scary” in a subsequent interview with federal investigators.
Federal investigators say that, by the time of the collapse, the size and depths of the splitting in the concrete at the span’s north end were far larger than cracking that is considered normal and acceptable in concrete construction — and by a jaw-dropping margin.
“The structural cracks in this bridge, located in critical regions of a non-redundant single plane truss, were more than forty times larger” than normal, a highway administration assessment concludes. “This scale of cracking in this structure was a clear indication that the intended load resisting mechanisms were failing. ... As those unanticipated load paths were pushed beyond their limits, the structure exhibited ever more distress until failure occurred.”
But FIGG, confident in its own reassurances and analysis, did not ask that the road be closed while the work was going on. Neither did anyone else — even after FIGG’s engineers acknowledged they did not know what was causing the cracks, the documents show.
The doomed bridge was the linchpin of a broader “prosperity” project devised by FIU meant to help the working-class town of Sweetwater across the Trail improve its tiny downtown by tying it into the school’s sprawling main campus, while making it safer for students and others to get across the busy, multi-lane roadway on foot to new apartments for students and faculty on the town side.
But the university also wanted an iconic design to serve as a new campus gateway.
For aesthetics, ease of maintenance and durability, the proposal from MCM that FIU embraced called for a bridge made of steel-reinforced concrete instead of a more common and simpler to erect steel span. MCM hired FIGG, a renowned bridge design and engineering firm. FIGG came up with a one-of-a-kind scheme for an expansive concrete deck topped by a broad concrete canopy and connected by a single line of dramatic diagonal concrete trusses.
The bridge was designed to use a relatively novel method of bridge construction that FIU’s engineering school champions. In so-called Accelerated Bridge Construction, a bridge span is manufactured on site, then lifted into place over pylons, largely eliminating the need to close a roadway for an extended period of time.
That meant the bridge would be under varying stresses during different phases of construction, first while lying supported on ground level, then while being lifted into position, and finally when set to rest in place.
While it was still on the ground, steel cables and rods inside the bridge concrete were tightened, or “stressed,” to provide internal strength, and formwork supports were then removed — the work that was going on when the FIU bridge first cracked significantly, sending workers scattering.
The span was then lifted into place on March 10 by a pair of massive transporters, creating different forces and points of stress on the bridge as it was carried aloft — all monitored carefully by instruments placed along the structure. Those forces shifted again as the bridge was placed on its support pylons.
That structure was actually only the bridge’s main span. Plans called for the bridge to be completed by construction of a shorter back span connecting over a canal that runs along the Trail to Sweetwater. That back span would provide additional support and strength to the entire bridge, uniting separate components into one self-supporting structure that required no columns or supports in the roadway, only at the ends.
The plan also required one unusual element: steel rods placed inside the last two horizontal support trusses at either end. Those were inserted to support the bridge only while the span was being transported. The rods were tightened before the span was lifted in the air to prevent sagging at each end. Once the bridge span was resting on the pylons and supported by its own weight, the rods were “de-stressed” because their support was no longer needed.
That element would prove key in design errors and the bridge’s failure, federal documents in the investigative docket show.
Unclear and blurred lines of responsibility among project contractors hindered design reviews, oversight and the response to the cracking, the new records show.
To run the project, which was federally funded but locally administered, federal rules set up a complex structure of management. The bridge was built not directly by FDOT, the agency in Florida that manages federal transportation funding, but outsourced to the university. FIU hired MCM, which hired FIGG and other subcontractors. To conduct inspections and make sure the project was on track, FIU also hired local engineering firm Bolton Perez & Associates.
As a required backstop on its design, FIGG hired Louis Berger, a global engineering firm based in New Jersey. Berger’s role was to review FIGG’s design.
FDOT, meanwhile, was represented on the project team by yet another consultant, Alfredo Reyna of the engineering firm Keith & Schnars (now KCI Technologies). Agency officials have contended they took mostly a back seat, although its engineers did review and comment and sign off on FIGG’s design, which was directed by its chief engineer, W. Denney Pate.
But the federal investigation revealed numerous technical flaws in FIGG’s design and the plan reviews. The newly released highway agency assessment concluded that FIGG’s engineers misapplied certain standards, misused computer models and misinterpreted the data they produced. When the agency ran its own checks on the design, it found FIGG had dramatically miscalculated the forces working on the north end at a critical juncture of support elements Number 11 and 12 that bore a disproportionate percentage of the bridge’s weight.
FIGG also overestimated the strength of the structural support elements provided in its plans at the connection, the report says.
The firm’s engineers also appear to have failed to fully account for the shifting forces on the bridge in each stage of construction, the assessment concluded, though investigators could not say for sure because FIGG’s plans and calculations lacked sufficient detail to assess.
FDOT’s reviews, while making some tweaks to the design, failed to catch the critical errors. Berger, the firm hired by FIGG to review its plans to ensure the design calculations were correct, also failed to catch any significant mistakes.
That may be because its contract with FIGG, which agreed to pay Berger a maximum of $61,000, provided only for a limited scope and time frame for the work. Berger analyzed the plans for only one of the phases of construction — when the bridge was completed. That meant Berger’s engineer, Ayman Shama, did not examine the effect of the forces on the main span while it lay on the ground, while it was being lifted, or when it was set down on the pylons without the back span.
Shama told the NTSB he was also not asked to examine another critical point: whether the bridge was sufficiently “redundant” — that is, whether it was designed so that failure of a single structural element, such as the weak “node” or connection at the north end that caused the FIU bridge failure, would not bring it down.
In an interview, Shama, who has a doctorate in engineering, told federal investigators that such an extensive check would have exceeded “the budget and time agreed about with the designer.” Shama, who has since left Berger, told the NTSB his work lasted about seven weeks. Berger submitted its letter to FIU certifying FIGG’s design for the bridge’s superstructure on Feb. 10, 2017 — a full year before the first significant cracks appeared.
“In the beginning, I suggested to do this kind of analysis, to analyze the connections,” Shama said. “I’m talking about the nodes, or the joints to analyze the connections. However, the budget and time to do this actually was not agreed upon with the designer.”
It’s unclear whether Berger even should have accepted the job. The firm was not rated as “qualified” under state rules to review a project like the FIU bridge, which was categorized by FDOT as a “complex bridge design - concrete.” Berger had not been qualified to perform such work since at least 2013, according to a factual report by NTSB released as part of the docket.
That didn’t stop a Berger employee from telling FIGG in an email that it was qualified, according to the NTSB report. Compounding matters, a state website wrongly listed Berger as qualified. The glitch was due to a technical error by FDOT, the NTSB found, though investigators noted that Berger should have known it was not certified by the state and FIGG could have discovered that fact through simple “due diligence.”
The documents, however, do not tie Berger’s lack of certification to the bridge collapse or raise issues with Shama’s work.
‘It cracked like hell’
There were already hairline cracks visible in the concrete before the ominous episode on Feb. 24 that sent workers running.
In mid-February, after the internal support rods at either end were tightened but before the supporting formwork was removed, inspectors reported and documented hairline cracks. Small cracking in concrete structures is common, and those were expected and well within norms, inspection reports show. Bolton Perez sent a report detailing “several” cracks to MCM, which relayed that to FIGG. FIGG engineer Manuel Feliciano said the cracks were “temporary.”
Feliciano also asked inspectors not to mark cracks with Sharpie because that would “discolor” the concrete.
On Feb. 28, though, as the cracks appeared to worsen, inspectors and MCM again asked FIGG for an evaluation. FIGG did not respond until more than a week later.
“Having not received a response from FIGG for more than a week, MCM had to follow up again on March 7, 2018. FIGG then advised, among other things, that the cracking was ‘not .... a structural concern, and that FIGG was ‘not concerned about these very small cracks,’ “ a federal highway assessment said.
FIGG said the cracking Number 11 strut could be repaired by filling in the separations once the bridge span was in place over the roadway.
But photos taken on March 8 by the transporter contractor, Barnhart Crane & Rigging, in preparation for the move were in hindsight a troubling sign, investigators say. The cracks were over a critical “node” at juncture of that diagonal Number 11 where it met the deck and the back upright truss, Number 12 — even before the stresses of the move.
Two days later, on March 10, the bridge was lifted into place, an operation witnessed by most of the lead team members, FIU administrators, politicians and other public figures to significant fanfare. FIU trumpeted what initially seemed a successful operation in videos, news releases and social media.
FIGG’s chief engineer W. Denney Pate boarded a flight out of South Florida shortly afterward.
After the transporters were removed, Barnhart reported that all had gone well during the move despite minor glitches, such as the Wi-Fi connection to instruments going out several times, and that the span was positioned as planned: “All global deformations such as rotation, twist, and deflection indicated the condition of the span after the move was nearly identical to its initial state.”
Sometime after 12:30 p.m., when the span was set down and the transporters removed, an inspector and a junior FIGG engineer walked along it. Inspector Alexis Molina, working for Bolton Perez subcontractor Corradino Group, noted what he thought to be newer and bigger cracks and other damage at the north end, which he showed to project manager Rodrigo Isaza of MCM.
But real signs of trouble began showing up after the internal support rods at each end were de-stressed that afternoon, as planned. After that operation ended around 6:30 p.m., documents show, more and alarming cracking was discovered at the Number 11 strut connection.
“It cracked like hell,” Kevin Hanson, a crew chief with the contractor doing the de-stressing, said in a text message to a supervisor. (Hanson was badly injured in the collapse.)
In its own report to the NTSB, MCM contends it did not find out about the additional cracking until Monday, March 12, when Isaza sent an urgent email and 16 photos of the cracks at the Number 11 and Number 12 connection to FIGG engineer Dwight Dempsey. FIGG contends in documents and interviews that this was the first time it learned of cracking after the move.
“It is our opinion that some of these cracks are rather large and/or of concern; therefore, please review and comment as promptly as possible and advise if there is a required course of action to remedy or address these right away. Your immediate attention and response is required,” Isaza wrote to FIGG.
FIGG replied it was working on a fix and performing calculations to ensure the connection was strong enough to hold up, but that assessment was based only on photographs, without an inspection of the cracks in person. The federal assessment concludes that the same errors and miscalculations they made in the design caused FIGG engineers to significantly misdiagnose the problem.
According to federal highway engineers’ assessment report, what was happening was this: The node or connection was detaching from the deck as it slid off its anchors towards the end of the bridge. The cracks had split the concrete at that spot to such an extent that steel rebar installed to provide strength at that juncture was no longer anchored properly and provided little reinforcement.
In its review, the highway agency found no other deviations from construction plans, and analysis of construction materials found no issues with quality of steel or concrete.
The re-stressing of the rods at the weakened connection point, it concludes, was clearly what prompted the collapse.
“The bridge collapsed when the demand placed on the truss member 11 and 12 nodal region exceeded the resistance that the structure was able to provide,” its assessment reads.
In its submission to the NTSB, FIGG denied responsibility for the accident. Its report concluded that a construction joint at the north end of the main span between the truss members and the bridge deck was not properly “roughened” to provide resistance through friction, and that this construction error, not a design flaw, doomed the bridge.
“Full-scale tests show that if the construction joints below Members 11 and 12 were roughened as required by the FDOT Standard sophistication, the collapse would not have occurred,” FIGG’s submission states.
FIGG’s review was conducted by prominent forensic engineers Wiss, Janney, Elstner Associates. Wiss Janney found FIGG’s work was not flawed, saying that the “design of the truss member 11 connection to the bridge deck ... was in compliance.”
The report listed other factors that Wiss Janney believes played into the collapse, including: damage from twisting when the bridge was lifted into place; “inadequate, inaccurate and untimely information provided by MCM and others”; the failure of MCM, Bolton Perez, FIU and FDOT to close the road to traffic; and the failure of Structural Group, MCM and Bolton Perez to monitor the cracks.
The highway agency assessment says the connection ends were likely not properly roughened, but notes that FIGG’s instructions for that in its plans were “inconsistent” and did not clearly specify it. The agency does not conclude lack of roughening played a major role in the bridge’s failure.
FIGG’s temporary fix to the cracking observed in March was to re-stress the internal rods in the Number 11 truss. The thinking was that would return the span to the status or condition it was in before the bars were de-stressed and the cracks worsened — entirely missing the fact that the connection in fact was far too damaged for the cracking to be simply reversed, the assessment concludes.
In fact, it only made things worse.
“Retensioning member 11 post-tensioning rods increased the force effect on the node and exacerbated the damage,” the highway engineers conclude.
The cracks at this point were disturbing, the documents show. Photos in the docket display cracks ranging from three inches to six inches deep. FDOT standards classify cracks as structural if they are deeper than half an inch. But no one appears to have raised that issue.
Carrying on with the plan
Instead, FIGG carried on with its plan. The next day, Tuesday, FIGG again said the fissures posed no safety concerns. Dempsey told Isaza to install plastic shims under the centerline of the bridge span at the north end to better distribute the load — a measure federal investigators say did nothing to address the structural problems.
“FIGG is evaluating this situation as a top priority and will be making recommendations as a result of this evaluation. As of right now, we do not see this as a safety issue,” Dempsey told Isaza in the email.
Isaza responded: “As just discussed, we are glad to hear that upon further evaluation by your team, this matter does not pose a safety issue and/or concern. We are also proceeding to install the temporary shims (plastic/metal), as recommended, later today and will provide you with the additional photos requested. Moreover, we will be monitoring the cracks to ensure these do not develop further.”
In an email to Pate, his boss, Dempsey appears to downplay the cracks: “Rodrigo said that the cracks were observed prior to destressing then grew slightly once [the post-tensioning bars] were destressed.”
FIGG’s Pate also left a voicemail that Tuesday “as a courtesy” with Tom Andres, an FDOT engineer who had reviewed the bridge plans. Andres, who was out of the office on an assignment, did not hear the message until the day after the collapse. But Pate expressed no concern about the cracks.
“[O]bviously some repairs or whatever will have to be done but from a safety perspective we don’t see that there’s any issue there,” Pate said.
According to the NTSB documents, Pate did not see the cracks in person until March 15, early on the morning of the collapse. He was taken up in a lift for a closer look, but was apparently still untroubled by what he saw. In a 9 a.m. meeting in the MCM construction office at the site, Pate ran through a technical slide presentation that concluded the span was safe, and discussed other repair work to bolster the strength of the cracking connection later. But he displayed no sense of urgency, saying he would have something further for the team by Saturday.
“Based on the available information at the time, we did not see this as a safety concern,” Dempsey told NTSB investigators about the days leading up to the collapse. “From what we saw based on the pictures that we saw, that ... it still had reserve strength and the capacity that would be sufficient to meet the design codes.”
Dempsey said he was not informed that the cracks had grown as wide as they had. Investigators concluded FIGG had been told.
Others who saw the cracks that morning were far more concerned. Bolton Perez project administrator Rafael Urdaneta asked FIGG during the meeting whether the bridge should be shored up, firm inspector Carlos Chapman told investigators in an interview.
“He saw those cracks. They were scary cracks,” Chapman said.
But FIGG said shoring wasn’t necessary, Chapman said.
FIGG could provide no explanation for the cracks, several attendees noted in interviews. But no one appeared especially concerned. No one requested an independent, third-party review of FIGG’s intended fix. Deviations from construction plans require such an objective “peer review,” but FIGG engineers insisted in interviews with federal investigators that they did not regard the re-stressing of the rods as a departure, since they were only trying to return the span to the status before the supports were loosened.
No one even brought up closing the roadway, the reports and interviews show.
“In the meetings I had and in the conversations I was a part of, both here on-site and elsewhere, that was never discussed,” Pate said in an interview with investigators. “None of the — when we were all here together, you know, FIU didn’t discuss it; FDOT didn’t discuss it; MCM, FIGG, no — it just was not a discussion.“
MCM, in its submission to the NTSB, notes an independent review might have made all the difference. But the general contractor also sought to distance itself from the decision not to pursue a review or press for a closing of Tamiami Trail, also known as Southwest Eighth Street, saying neither of these fell within its purview. MCM noted its representatives were the only ones in the room who were not qualified engineers or a party with a say in closing the roadway — something that fell to FDOT, Bolton Perez and FIU.
“It appears that FIGG’s restressing was not rooted in science and was not independently peer reviewed, but was based on a misinterpretation of the sequence of events and timing of cracking,” MCM wrote. “Further, Engineers representing FDOT and FIU were present at the meeting and did not take issue with any of FIGG’s technical analysis methods or findings. FDOT and FIU engineer representatives also did not object to proceeding with the repair (restressing) immediately after the March 15, 2018 meeting, despite evidence that FIGG did not truly understand the root cause of the observed distress.”
FIGG’s Pate again left quickly, boarding a flight to Tallahassee before subcontractors began the re-stressing work. He found out about the collapse, interviews show, only after landing.