Did it crack when it was lifted into place? Was it a snapping steel support that did it in? Or was it a fatally flawed design from the start? And what was the doomed construction crew member doing when the bridge buckled under him?
In the two weeks since the deadly collapse of a pedestrian bridge under construction at Florida International University, engineers the world over have obsessively dissected photos, scrutinized video clips and picked over preliminary plans for the singular structure in search of the mysterious and elusive breaking point.
Never mind that how and where precisely the bridge broke apart likely won’t be known for months, until the National Transportation Safety Board issues an official finding: On engineering online discussion boards, around office water coolers and computer monitors, divining the cause of failure for a structure by one of the country’s best-known bridge designers has become the professional cause du jour.
The upshot so far: several compelling hypotheses and lots of sharp assessments, some loopy detours and a sharpening focus on a pair of support trusses that appeared to shear and shatter just as the unfinished concrete bridge suddenly crashed to the roadway bed below, but nothing close to a conclusive explanation.
Still, after clearing away some early misconceptions — confirming, for instance, that the FIU overpass was not a suspension bridge but a modern take on an old-fashioned truss design — outside expert speculation has moved beyond just potential construction error to focus increasingly on possible design flaws that might have made the bridge structurally vulnerable.
“I would say they were pushing the envelop,” said Neil Hawkins, an emeritus engineering professor at the University of Illinois who specializes in concrete construction. “The question is, to what extent had the new concepts in this been validated through testing prior to actually putting it in place?”
The admittedly speculative consensus has coalesced around a view of the FIU project as a novel design with a high “wow’ factor, but also hidden structural risks and vulnerabilities that were possibly compounded by a last-minute design change, all brought to a catastrophic conclusion by engineering and construction miscalculations.
It’s the sort of baffling accident that makes structural engineers break out in sweats — a product of the fear of some overlooked fatal flaw in their own work and the ethical obligation to ensure the same thing doesn’t happen again, engineers say.
Experts note that it’s impossible for them to narrow down possible explanations for the bridge failure without critical documents and information that so far are not publicly known. One highly regarded engineers-only forum, Eng-tips.com, “paused” its discussion of the bridge after 800 comments and five threads because speculation and analysis based on limited evidence hit a dead end.
On Wednesday, FIU officials refused to provide the Herald construction documents that might shed light on the collapse, citing a federal regulation they say prevents them from releasing documents that are part of an ongoing NTSB investigation. But an attorney for the Reporters Committee for Freedom of the Press said the rule has a provision for releasing records and, given the public interest, they should be released.
“It’s not clear to me why that procedure is not being put in place to help the public understand the facts about the collapse,” said attorney Adam Marshall. “You’re talking about essentially what appears to be an infrastructure failure that cost lives. Access to public records that could shed light on what happened in this event is precisely what public records laws like Florida’s Sunshine Law are designed to do.”
March 15 failure
The 174-foot-long span, weighing 950 tons, fell in a heap March 15, five days after it was hoisted into place over the Tamiami Trail. The collapse killed one construction worker and crushed to death five people in cars stopped under the span at a red light.
The span had been fabricated by the side of the road over several months under what’s known as accelerated bridge construction, a newer but now widely used approach that minimizes the time a road must be closed. Lifting the span into place took only a few hours.
What is known about the collapse so far has been largely culled from preliminary plans, dribs of information from government agencies, and dozens of photos and videos showing the bridge installation, its abrupt fall and the mountain of broken concrete and steel it produced. A dramatic dash-cam video, slowed down and blown up, has provided outside analysts some of the best, if also frustratingly grainy, clues into how the bridge came down.
But speculation and outside analysis has been sorely hampered by a lack of final construction plans. All engineers and the public have had to go by to date are early plans submitted to FIU as part of the winning proposal by the design-build team of Munilla Construction Management and FIGG Bridge Group.
The experts say the FIU bridge failure is likely to be taught in engineering schools along with other infamous construction failures such as the 1981 collapse of an elevated walkway at the Hyatt Regency Hotel in Kansas City, Mo, that killed 114 people.
While any tentative assessments are bound to change, perhaps drastically, as new information comes to light, this is what we know so far:
▪ The 174-foot span installed March 10 was the main portion of the unfinished bridge and rested on a concrete pylon at either end, though unsupported in the middle. It fractured when two diagonal concrete beams or trusses joined in a V shape at the very north end of the structure appeared to separate from the bridge’s canopy and its walking deck, leading to a sudden collapse of the entire span.
▪ The structure was a variant of a truss bridge design, usually made of steel. But FIU’s consultants, MCN and FIGG, instead chose heavy concrete, which has several advantages. It’s more durable, easier to maintain and malleable, so that it can be shaped into smooth, swooping forms. That last was a plus for FIU, which wanted more than a simple pedestrian overpass. The bridge was part of an ambitious plan to weave together the university campus with the small town of Sweetwater, located just across Southwest Eighth Street and a narrow canal that runs along the road. New privately built student housing is rising in the town’s compact downtown by the canal.
The bridge was meant as a symbolic new portal to the university, not just a way to safely cross the dangerous roadway. It would have open views, benches, planters, glass-enclosed elevators, Wi-Fi and a wide 30-foot deck for pedestrians and cyclists.
“Bridges designed by FIGG are purposeful works of art, functional sculptures within the landscape, that are created through a careful analysis of the site, contextual and environmental sensitivity, and a regional approach that encompasses a community’s particular needs, as well as the realities of funding and maintenance,” according to the firm’s web site.
But experts say it was an unorthodox design: Truss bridges usually have parallel sets of bracing along each edge. The FIU bridge boasted just one asymmetric arrangement of concrete beams in a zig-zag across center of the span. That design may have accentuated the vulnerability of truss bridges, which are known to fail when a single structural support is damaged.
The FIU bridge also appeared to lack redundancy -- that is, extra structural elements that would have held up the span if one piece failed, experts said. The truss design was chosen in part because it lent itself to the quick, one-day installation.
“You can actually do it, but it’s a lot more difficult,” Hawkins said. “You can almost use the analogy of your backbone. We don’t have two backbones. We have one backbone.”
▪ At the time of the collapse, crews from Structural Technologies/VSL were working on adjusting the tension of steel support rods running inside those end diagonal trusses when the bridge fell. The well-regarded contractor manufactures and services the adjustable steel cables and rods used to provide tensile strength in modern concrete construction, engineers say, with experienced and technically proficient crews.
The crew had, apparently without incident, just finished adjusting tension rods at the south end of the bridge.
▪ The preliminary plans for the bridge called for the tension rods to be adjusted twice: They would be tightened before the bridge was lifted into position from below by special transporters. Because concrete is weak when it’s pulled on, that tightening would provide needed support to the bridge span ends, so that the structure would retain its integrity under the immense stress of the lift. Once the span was settled into place, the diagonal beams would become compressed by the weight of the bridge, providing the needed structural stability. The rods would then be loosened.
In a video taken just after the bridge was installed, an MCM employee tells an interviewer that the next step would be to de-tension, or loosen, the support elements at either end.
▪ Two days before the collapse, FIGG’s lead engineer called FDOT to report that cracking had appeared at the bridge’s north end. The engineer, W. Denney Pate, said the cracking did not pose a safety hazard but would have to be repaired. Cracking is common in concrete structures and is often superficial, but can sometimes be a sign of deeper trouble.
▪ Just before the May 15 collapse, the project contractors, engineers, FIU and FDOT officials held a two-hour morning meeting at which the cracking was discussed. No one raised any red flags over the cracking, FDOT said in a statement after the collapse.
▪ FDOT had requested a mid-stream change that may have triggered a significant structural design change in the bridge. The agency asked that the north pier of the bridge be moved 11 feet to the north to allow for the addition of a dedicated transit lane to the Trail sometime in the future.
Engineers who have compared preliminary plans on FIU’s website to photographs of the bridge just before it fell say the diagonal trusses underwent some changes. The southernmost diagonal was thickened, and a steel support rod was added to the northernmost diagonal, among other apparent alterations.
▪ The critical positioning of the lift trucks also changed between the preliminary plans and the actual move. In the plans, the trucks were placed directly under the truss joints at the very ends of the span. For the move, the trucks at the north end were moved toward the center, leaving that end of the span hanging unsupported, or cantilevered, during the move. Sensors attached to critical points in the span allowed engineers to monitor stresses to ensure they stayed within expected parameters.
The change in the truck positions was likely the reason that additional steel rod was inserted into the northernmost truss, to provide more tensile support to prevent the structure from sagging at that point.
The change in the transporter position may have been necessitated by the move of the northernmost support column 11 feet to the north, which placed it well off the edge of the road, in an area the lift truck could not traverse. The redesign of support trusses appears to have been made to re-position the joints, the strongest points of the span, over the new lift points.
Those conclusions give rise to several tantalizing questions and hypotheticals:
▪ One, postulated by Canadian vlogger AvE in a crowd-sourced YouTube video that got more than 600,000 views, is that the support rod in the northernmost diagonal truss was damaged while the bridge end was cantilevered during installation, stretching it to a breaking point. The anonymous YouTuber says crews were probably tightening the rods in the diagonal trusses when the rod snapped, leading the concrete truss to shatter and the bridge to collapse.
His assumption was based in part on a widespread supposition that crews were tightening the rod. It may have originated in a tweet by U.S. Sen. Marco Rubio that claimed support “cables” in the bridge had loosened and that engineers ordered them tightened.
One related hypothesis holds that engineers decided to tighten the rods possibly to address the cracking.
▪ Engineers on the widely respected online Eng-Tips Engineering Forums, however, have been skeptical of some aspects of that hypothesis. In long technical discussions on the forum, in which engineers use handles but provide their areas of expertise, several have said there would be no reason to tighten the rods since the diagonal trusses were already compressed by the substantial weight of the bridge.
More likely, they say, crews were doing the planned loosening of the rods, though no one can explain why they would have waited several days after the bridge span was in place to do so.
But some engineers lay out scenarios under which which either loosening or tightening could lead the diagonal trusses to fail. One explanation: Loosening a rod requires an initial, momentary tightening of a nut at the end. If the rod was already stressed to a breaking point, it could have snapped, fatally damaging the truss or a connecting point to the deck or canopy.
▪ Engineers examining plans and photos of the debris say the joint at the deck may not have had enough steel rebar or design strength to do its job of supporting the truss “web” that was to hold the structure together. Adjusting tension of support rods in that case may have been enough to subject the joint to more stress than it could handle.
Because of the lack of redundancy in the design, failure of that one element would have caused the support web to shatter and allow the entire span to drop.
▪ Some question whether FIGG engineers had enough time to properly consider the ramifications of design changes and the move of the lift trucks. Often, they say, a senior engineer like Pate at a busy mid-size firm like FIGG has moved on to new projects after the construction design is approved, leaving the redesign to an engineer working under him. In some cases, too, computer modeling undertaken to ensure stability of the original design may not be redone.
That points to a larger problem in bridge construction, Hawkins said, essentially triggered by lawsuits.
“The bridge designer only needs to show that the bridge is safe when completed and to be convinced from calculations they have made that the bridge they have designed can be constructed,” he said. That means designers don’t generally tell contractors how to construct the bridge, leaving the contractor to figure that out.
“There are numerous cases of stadiums, buildings, and bridges that have collapsed during construction as the contractor did not fully evaluate” what would occur during the actual construction, he said. The most famous example occurred in 1981 when a pedestrian bridge at the Kansas City Hyatt Regency collapsed and killed 114 people. The cause? An error in a design fix intended to correct a problem in the original plan.
▪ Others points of discussion among engineers and construction experts include whether the concrete was properly “cured” as the span was fabricated by the side of the road, whether the composition of the high-tech concrete was adequate, and whether the long slabs of the bridge deck and canopy were themselves properly tensioned before the span was lifted into place.
But the engineers say their analysis is little more than speculation without several key missing pieces of information:
▪ Photographs of and the precise location of cracking reported by the FIGG engineer to FDOT two days before the collapse, which might show whether Pate overlooked anything in concluding there was no danger.
▪ The construction plans showing precisely how the bridge was to built, along with any mathematical calculations. Those early plans by FIGG almost certainly changed, engineers say.
▪ Records from the Thursday meeting to know exactly what engineers said about the cracks, whether crews were tensioning or de-tensioning rods, and whether doing so was routine work or came in response to the cracking or some other issue.
▪ Whether design changes prompted by FDOT required the bridge be moved 11 feet to the north, or added 11 feet to its length. Most likely, after examining photos of the south pier on the FIU side, engineers on Eng-Tips.com concluded the entire bridge was moved north, with the span likely remaining the original length.
▪ Whether the novel design was tested structurally through use of a mockup or model.
Like the general public, engineers have also wondered volubly why the Tamiani Trail wasn’t closed to traffic while the tensioning adjustments were going on. Given a design that some experts now regard as risky, and the appearance of cracking, many say the design-build team and FDOT should have closed down the roadway in an abundance of caution.
“This procedure was considered so dangerous that the manual recommended red flags or flashing lights be used to warn personnel of the operation,” Texas engineer Jerry Rovner said in an email, referring to a document from the Federal Highway Administration. “This would seem to present a strong case that traffic should have been stopped while this post-tensioning was being done.”