Assistant chief firefighter Allen Hay was off duty the day terrorists flew hijacked airplanes into the World Trade Center buildings. By the time he reached the scene, both the North Tower and South Tower had collapsed, killing at least 2,550 people and spewing thousands of tons of debris onto the streets of New York.
Hay was given the task of attending another burning building surrounded by debris from the twin towers, Building 7 of the complex, which was still standing. “I don’t want any more buildings collapsing,” he recalls his supervisor telling him.
He arrived to find six floors ablaze, almost no water supply, and no radios to communicate with his firefighters. “We decided to give up on the building,” he says.
Although a part of Building 7 bulged dangerously, Hay and others expected fire to simply burn away whatever was flammable and leave the 47-story shell standing. But at 5:20 that afternoon, 7 World Trade Center became the third structure to crumble that day and the first to do so without being hit by a plane.
“We just expected it to burn out—we didn’t expect it to fall down,” says Hay, who is now chief safety officer of the New York City Fire Department. “It’s the only building I know in New York City to ever collapse [strictly] from fire.”
The events of 9/11 highlighted the danger that fire poses for the stability of structures and the need to design buildings that maintain integrity when they’re ablaze. “Fire was one of the primary factors that led to the collapse of the World Trade Center,” says Venkatesh Kodur, a civil and environmental engineer based in East Lansing, Mich., who consulted on the investigation into the failure of the towers and Building 7.
It’s the fire that ignites after damage from some other cause—such as an earthquake, a tornado, or even a vehicle impact—that often makes a structure unstable by burning through walls, crumbling concrete, and melting steel. According to the Quincy, Massachusetts-based National Fire Protection Association, 2005 saw 511,000 building fires in the United States that killed 3,105 civilians and resulted in $9.2 billion dollars of damage.
“Fire causes as much if not more loss of lives than any other hazard in this country,” says Doug Foutch, program director for structural systems and hazard mitigation of structures at the National Science Foundation in Arlington, Va. Foutch sees a need to increase funding of research into how fires affect structural integrity. “I consider it one of the top priorities in my area,” he says.
The devastation to structures caused by natural hazards, such as earthquakes and severe storms, spurred engineers in the early 20th century to design buildings that could withstand seismic shaking or fierce winds. But the structural effects of fire, often a secondary hazard triggered by other damage, have not been given the same engineering attention, Kodur says. “We’ve made significant progress in earthquakes in the last 20 years, but now is the time to tackle fire as well,” he says. “It’s an everyday event that can become a severe threat.”
The threat
Heat generated by combustion causes myriad changes to the properties of building materials, says structural engineer Amit Varma of Purdue University in West Lafayette, Ind. When heated, materials lose their stiffness and strength, which means that they fail at a lower load than they could normally bear, says Varma.
Concrete can disintegrate in extreme heat, as water trapped in its porous structure boils. Pressure from the water vapor cracks the concrete and breaks it into pieces, a phenomenon that engineers call spalling. And heat can weaken a structure’s steel framework by making metal components expand, soften, and sag. Deformation of one part of the framework can increase stress on neighboring parts, to the point that the entire structure can fail.
“Buildings are designed to be very reliable and safe,” Varma says. “We have a lot of confidence of the reliability of the structures and components from experiments.” But those experiments, he notes, are largely performed at room temperature, so they don’t reliably test how buildings behave in fires. “What has changed since 9/11, 2001” he says, “is that people are now looking toward structural engineers for more answers: Is a [burning] building going to be able to deform and carry the load?”
That kind of knowledge would enable architects to design buildings that could better withstand severe fires and might give firefighters better warning of when a building is about to collapse. “We’re behind the eight ball just walking in the door,” firefighter Hay says. “Often, we’re arriving on the scene as the building is collapsing, so it’s basically a time game. That’s a danger—we don’t know how long we’ve got.” That danger was tragically realized June 19, when nine firefighters died after the roof of a burning furniture factory in Charleston, S.C., caved in on them.
The testing
In the late 19th century, concerns over fire safety grew with the height of the buildings that were starting to clutter city horizons. Chimney-shaped skyscrapers could go up in flames quickly and trap occupants inside a structure that might then fall down around them. The American Society for Testing and Materials (ASTM), formed in 1898, developed the first standard tests for the structural integrity of building materials should they be in an intense fire.
These standards “continue to govern practice even today,” says Shyam Sunder, director of the Building and Fire Research Laboratory at the National Institute of Standards and Technology (NIST) in Gaithersburg, Md.
NIST doesn’t set the standards, but the government lab tests materials and makes recommendations to the ASTM and to local governments for their building codes. In a typical integrity test, a segment of a concrete column, a steel beam, or a wall is exposed to flames of controlled intensity rising to a critical temperature. If the segment retains its strength for at least 60 minutes, it’s given a 1-hour rating. Most load-bearing components of buildings have ratings of 2 to 3 hours. The main structural elements of the World Trade Center towers had 2-hour ratings, and Building 7 had 4-hour components.
Typically, fire-protection engineers ensure that the design and materials proposed for buildings measure up to the fire-resistance standards of local building codes. Those codes are often based on information from fire tests dating back to the early 1900s. When a new material is tested, however, the information is added to a database that engineers can access.
If a proposed design is unusual or incorporates novel materials, a company must submit samples of the relevant components for new testing. Given that testing a segment of a wall, for example, can cost $15,000 to $20,000, such a test is usually conducted only once.
Chicago-based fire-engineering consultant Nestor Iwankiw points out that many such tests don’t fully probe the failure of materials. That’s because a test of a beam or a pillar will usually be stopped once it reaches the 2- or 3-hour safety rating required under a given code.
For the most part, the standard fire tests represent good safety benchmarks, says Iwankiw, but “in unusual circumstances, like 9/11, all bets are off.” The extensive damage that the planes caused to the buildings’ frameworks and their fire-resistance coatings was “just never anticipated or designed for.” Engineers usually don’t even plan for multiple-story fires when designing buildings, he says.
The biggest limitation of current testing methods is that they don’t mimic conditions in real fires. Even the largest furnace at NIST can test samples no longer than about 5 meters, a fraction of the length of many structural components. And, testing single elements gives little guidance in understanding how a combination of components behaves, let alone how a whole building performs, in a fire.
“The rating tests don’t give you much additional information on materials’ properties,” Iwankiw says. Those tests don’t provide any information about how an assembly of connected components would fail. “We’re picking assemblies out of a catalog for a standard fire and we really don’t know how safe or unsafe [they] will be.”
When a beam starts to deform, for example, adjoining columns or other components may take up extra loads, so the failure of the beam won’t necessarily bring down the building. On the other hand, unanticipated stresses placed on adjoining components could tip them into failing. A building, therefore, might be more or less stable in a fire than the safety rating of its component parts would indicate.
In the late 1990s, researchers at a testing facility in Cardington, England, burned an eight-story building purposely constructed to test how an entire building would behave. They found that even without expensive fireproofing insulation on non-weight-bearing steel beams the building frame didn’t fail because the structure as a whole redistributed the extra stresses.
Further blurring the overall assessment of a building’s fire integrity is the behavior of the connections—bolts, welds, and mortar—that hold a structure together. The joints between different materials have been subject to “very little engineering” for fire safety, Iwankiw says. “We really don’t know how the system or the construction will perform in a real fire.”
In June, Kodur and his colleagues unveiled one of the biggest fire-testing furnaces in the United States. Located at Michigan State University in East Lansing, the furnace is specially designed for studying combinations of steel and concrete in conditions that occur in real fires. An important part of that research will focus on the effects of cooling on building materials and their contact points as a fire dies out—a factor known to be important but not accounted for in standard fire-integrity tests. Firefighters know well that even after a building has burned out, it can still suddenly collapse. Seven firefighters died in Gretzenbach, Switzerland, in 2004 when the roof of a smoldering parking garage fell in on them.
Reality check
To complement better physical testing of materials, researchers have begun to develop computer simulations of how structures would perform in fires. “You’d essentially like to be able to burn the building down ahead of time on the computer,” says William Grosshandler, chief of fire research at NIST and a lead investigator of the World Trade Center collapse.
To figure out why the twin towers and Building 7 fell, Grosshandler and his team had to develop programs that simulated, in three dimensions, damage from the planes’ impacts and the ensuing fire. In a fire, Grosshandler says, “even something as benign as concrete has extremely complex chemical reactions going on.” His team’s computer analyses required four separate programs that ran a total of 2 months, day and night, to model the buildings’ fates from the moment the first plane hit.
The NIST team used one program to simulate how the planes’ impacts damaged fireproof-coatings on beams and pillars in the buildings, and also to account for the fuel and combustible materials that came from the aircraft.
The second program modeled the progression of the fire after it ignited. The software predicts the movement of fire and smoke through a structure and calculates the resulting heat distribution.
The third program modeled the effect of heat on the strength and durability of the buildings’ components. The detailed calculations, for example, worked out the deformation of steel beams that received more heat on one side than the other.
The fourth and most complicated stage of modeling tracked how deformation of the buildings’ structural components threw new stresses onto adjoining components, ultimately causing failure. The program calculated, minute by minute, the changing position of each building’s structural elements until that building almost reached the point of collapse.
The modeling was quite an achievement, says Jean-Marc Franssen, research director of the National Fund for Scientific Research in Liège, Belgium. “Until about 10 years ago we modeled all buildings as 2-D skeletons,” he says.
The NIST simulation, like all models of building failures to date, couldn’t follow the 9/11 collapses through to the end. No computer is yet powerful enough to follow the chaotic sequence of events that ensues when components break apart and a building falls, but this is where research is headed.
Franssen and his colleagues have recently made a breakthrough in modeling the collapse of a structure. They developed an algorithm that figures out the velocity and acceleration of individual building components.
That allows the program to calculate what engineers call the dynamic forces involved in a fast-changing scenario, which makes the analysis one step closer to replicating a real collapse, Franssen says. Without the added algorithm, a program must stop at the first sign of failure, but that point may not reliably indicate a building’s imminent collapse because of the structure’s ability, or inability, to redistribute stresses. “We don’t know much about the failure mode,” he says.
To fully understand the collapse of buildings, researchers are also attempting to model the behavior of connections between components in a fire as well as the process of spalling and the cooling of materials after the fire is out.
A well-lit future
“Structural engineers spend millions of dollars thinking about wind and earthquakes, but not at all about fire,” says Susan Lamont, a structural engineer who studied the fire damage to the intentionally ignited eight-story building in Cardington. Fire experts and structural engineers are now joining forces to make fire-resistance design as important and routine as earthquake-damage protection and wind proofing of buildings. That trend has begun to take hold in Europe, she says, and “gradually, the U.S. is starting to do this.”
The future of design for structural integrity in fires should be based on risk analysis, Lamont says. Any innovative structure or novel skyscraper design, she adds, should be analyzed by the modeling software to predict how a fire might affect its stability. “It’s about designing a building for the forces and fire, rather than applying fireproofing and hoping that’s adequate,” she says.
