This article from today's NY Times describes the most elaborate fire investigation ever...

January 6, 2004
Reliving 9/11, With Fire as Teacher

GAITHERSBURG, Md. — The jet fuel, of course, caught fire first.

As traces of the fuel's stench hung ominously in the air, a fiery burst quickly became a sullen orange glow, then grew until flames were licking the ceiling 11 feet above. The flames fed on a cluster of office workstations lying smashed, littered with collapsed ceiling tiles and office papers.

Within minutes, a maelstrom of heavy black soot, gyrating sparks and brilliant flame was leaping through the tops of four narrow seven-foot-high windows, warming the nearby faces of men wearing bunker coats, oxygen tanks and fire helmets.

The World Trade Center was burning again — at least one small part of it — in this fuel-soaked, full-size recreation of a Marsh & McLennan office on the 96th floor of the north tower, where the first plane struck on Sept. 11.

This office is inside a towering laboratory at the National Institute of Standards and Technology. And it is only one fragment of the disaster being replayed for a federal investigation centered here. In other studies, steel recovered from the twin towers is being ripped apart at high speeds, rebuilt structural supports are baking in artificial infernos, and fireproofing insulation is being pounded with simulated plane impacts to see how well it sticks to the steel it was meant to protect.

Initial research has already turned up major findings involving the surprising ease with which heat leaks through damaged insulation into the steel, and the ways in which isolated parts of the trade center's structure may have been prone to fail in a major fire — even without structural damage from the planes.

Some of those findings could lead to safety recommendations for hundreds of ordinary high-rises around the country when investigators issue their final report, expected by next fall.

The sprawling forensic analysis is also expected to advance the science of fire and how it affects structure, an area that has remained a backwater in the world of engineering, sometimes relying on techniques a century old.

After the office had become a smoldering ruin of ashes and charred metal cabinetry, Alexander Maranghides, the fire protection engineer at the institute who was leading the test, said, "It's just scary." He was referring to the fire, but he could have been describing the emotional vector of the entire effort.

"It just takes you back," Mr. Maranghides said, the red outline of his respirator still visible on his face.

Sidestepping that emotion in favor of dispassionate analysis, dozens of researchers like Mr. Maranghides are folding their data into a series of advanced computer simulations in an effort to recreate the destruction that was hidden within the twin towers on Sept. 11. They are recreating, in a digital world, the shape and extent of the dark holes gouged into the heart of the towers by the planes, the sweep of the deadly fire from floor to floor, the spots where the heat weakened the steel structure enough to set off a total collapse of each tower.

Charles H. Thornton, a structural engineer who is co-chairman of the Thornton-Tomasetti Group and also a member of the investigation's advisory committee, said the work would go a long way toward putting the field on a par with research on weaponry and space travel, which also require precise calculation of the effects of heat and blast on structures.

"It's never been done before in the building industry," said Mr. Thornton, who collaborated in the design of the Petronas Towers in Kuala Lumpur, Malaysia, currently the world's tallest structures.

Especially puzzling is that the analysis of hazards posed by earthquakes and high winds has far outstripped similar work on fires. But just as earthquake engineering lurched forward after the major California quakes of 1971, 1989 and 1994, fire science is advancing after the events of Sept. 11, 2001.

"We are bringing the structural fire-response analysis into the same league as structural analysis for wind and earthquake loads," said Dr. S. Shyam Sunder, the leader of the investigation at the National Institute of Standards and Technology. "It's that integration of the two disciplines that is going to be the real leading-edge development as a result of the investigation."

The Strength of Steel
A Profusion of Metal, and of Complexity

Standing in an asphalt parking lot here, Dr. Frank W. Gayle, a metallurgist, marveled at a battered triplet of steel columns from the north tower of the World Trade Center, 36 feet long, that were bent into the gently curving shape of a rocking horse's legs.

Without the benefit of any obvious markings on the steel, Dr. Gayle and his colleagues in the institute's materials science and engineering lab had determined that the piece dangled directly above the hole that the first plane punched into the north face of the north tower. The fuselage had probably grazed the piece and bent it into the curved shape.

"It's almost unbelievable that the material could be in essentially the same shape after falling so far," Dr. Gayle said.

In contrast to the rather anodyne concepts for a World Trade Center memorial at ground zero, most of which include few if any artifacts from the buildings themselves, this suburban parking lot was littered with twisted — and chilling — pieces of the steel structure. Dr. Gayle said the investigation had obtained samples of all 14 grades, or strengths, of steel used in the twin towers for analysis. All together, investigators have collected 236 major pieces of trade center steel, Dr. Sunder said.

One lab here is taking tiny medallions of the steel, no larger than electronic watch batteries, and crushing them at high speed with a high-tech battering ram called a Kolsky bar. Because steel, like taffy, becomes stronger when torn or deformed quickly, the measurements are critical for assessing how components behaved when the planes struck them.

Ordinary buildings, Dr. Gayle said, generally use no more than two or three different grades of steel, but that is only one measure of the astonishing complexity of the towers' structure. Each relied on a cluster of relatively heavy steel columns in its core, connected by lightweight floor supports called bar-joist trusses to a tight palisade of columns around the facade — 59 per face, producing those tall, narrow windows.

Even though the exterior columns all looked identical, both the grade and thickness of their steel varied from place to place, said Dr. Fahim Sadek, a researcher at the institute's building and fire research lab, who is producing a detailed structural model of the towers on a computer using the original blueprints. So there were actually more than 130 different column types, he said, each having to be accounted for in his model.

From there, it gets only more intricate. One of Dr. Sadek's detailed models for a single floor — the 96th floor of the north tower, considered typical — contains 40,000 separate elements. A coarser representation of the entire tower contains 90,000 elements.

Among early conclusions of the overall study is that recent tests based on models like these yield wind forces on the buildings 15 to 66 percent higher than the ones apparently assumed by the designers of the twin towers. Although the ultimate consequences are still unknown, said Dr. Sunder, one thing is sure: if the higher wind loads had been used, he said, "your factor of safety now" — meaning, on Sept. 11 — "would have been higher."

Dr. Sadek is also using his structural model as Square 1 for simulating the horrific impact and damage of the planes as they plunged into the towers.

In some calculations, Dr. Sadek recreates building components with such refinement that plates of steel just a quarter-inch thick are divided into four or five layers, each followed separately as parts of the steel are torn to pieces. Each plane's cargo, fuel tanks, engines, landing gear and aluminum skin are rendered with full three-dimensional fidelity. On the computer, time creeps forward in millionth-of-a-second increments so that every detail of the collision can be captured.

Such care has been given to structural subtleties that extra steel bracing, added to two crucial columns in the core of the south tower to help support an enormous bank vault on the 97th floor, have been included in the model. And those niceties could be important: an earlier simulation of the impacts, created as part of an insurance lawsuit, incorrectly predicted that the south tower should have crumbled as soon as the plane struck it at nearly 600 miles an hour.

The new simulation has already shed light on one of the darker mysteries of the attacks: how the extremely light aluminum of the plane wings could have sliced through the heavier steel of the exterior columns like knives. Dr. Sadek finds that without the mass of fuel-laden tanks in the wings, they might not have been able to cut through and do such grievous damage inside.

In several of the computer runs, Dr. Sadek said, "we did not observe any fracture of the column in the case when the wing did not have any fuel."

The Burning Office
When a Workstation Is Doused in Jet Fuel

Those same tanks, of course, ultimately ruptured and spewed jet fuel over multiple floors. Dr. Sadek is trying to calculate the pattern of that spray and the spectacular fireballs, which consumed roughly a third of the fuel but did little damage themselves.

Before the re-creation of the Marsh & McLennan office is set on fire, a technician in a yellow bunker coat with "NIST Fire Research" on the back carefully pours jet fuel from a calibrated Pyrex container into several sprinkler cans, the kind used to water gardens.

The idea is to douse the collapsed workstations with the expected amount of jet fuel and see if investigators can accurately calculate the intensity of the fire as it flares and burns out. Three technicians sprinkle the smelly jet fuel about the devastated office, which is 35 feet deep from the recreated trade center windows to the back wall — exactly the distance from the north tower's east and west facades to its rectangular core.

The distances and the precise sizes of the windows are important because they determine the flow of oxygen into the fire. To understand just how intensely the fire burns, the researchers have mounted temperature-sensitive thermocouples on four spindly columns in the room and placed weight scales beneath the workstations to measure how the blaze eats away at them. Devices in a huge duct outside the room — the experiment takes place within a towering laboratory on the campus here — gauge the amount of oxygen consumed by the fire.

After one technical glitch, Mr. Maranghides counts down to the spark that starts the fire. "Ignition," he says, and the blaze consumes the room.

Afterward, as the charred room smolders and sparks, he says the test has come up with at least one highly significant preliminary finding. The peak burning rate of workstations that collapsed before they burned is about half that of intact workstations, probably because the material in the workstations was compressed and partly covered with the fireproof ceiling tiles.

That finding could be extremely important as investigators try to estimate the heat of fires in the areas that were struck by the planes before being set on fire by the burning fuel.

Another investigator, Dr. Richard G. Gann, a senior research scientist at the building and fire research lab, said earlier tests showed that intact triplets of workstations typically put out the equivalent of 10 to 12 megawatts of power at peak burning, which lasted for about 20 minutes. And the investigators' computational tool, called the Fire Dynamics Simulator, successfully predicted those values, he said.

"At least from the data I've seen so far," Dr. Gann said, referring to humped curves showing the heat output, "we're getting the magnitude of the release right and we're getting the shape right."

The Fires' Spread
Simulating Disaster in Search of Its Story

Three workstations, of course, do not make an entire trade center, or even one of the wide-open, acre-size floors in either of the towers. So Dr. Kevin McGrattan, a mathematician at the building and fire research lab, is using the Fire Dynamics Simulator to knit together data from the small-scale tests and calculate the sweep of the fires over the entire 102 minutes of the disaster — from the impact of the first plane until the second tower fell.

His simulations of multiple-floor fires in the north tower show red splotches — the hottest areas — starting out near the holes punched by the plane in the north face. As the fires consume the combustible materials there, they creep southward around the floors — generally remaining hottest around the outside, where broken windows can provide oxygen. Another set of fires rage through elevator shafts and air ducts in the core, spreading the blaze to higher floors.

Two other researchers at the lab, Dr. Howard Baum, a fellow at the institute, and Dr. Kuldeep Prasad, a research engineer, have the daunting job of computing how that heat seeps into the steel of the towers.

"No one has yet put all this stuff together, where you have a detailed, physics-based model of the fire, a detailed structural analysis, and have them intimately coupled together," Dr. Baum said.

But the researchers have come up with a way to do just that, and have already discovered that missing or flawed fireproofing insulation on the steel could be much more important in the disaster than anyone suspected before. Simulations by Dr. Prasad indicate that heat can flow through gaps in the insulation and reach wide stretches of the steel, weakening it structurally and making it more likely to fail.

One thing the investigation has not done yet is find a specific sequence of failures that could have brought the towers down. But early studies of how isolated parts of the buildings might fail have produced their own unexpected twist.

Suspicion has long centered on the floor trusses, which some experts believe to have been poorly fireproofed and especially susceptible to buckling in a fire. Those suspicions have been bolstered by videos and photographs that show the trusses sagging and possibly giving way in the minutes before each tower collapsed.

But preliminary calculations suggest that long before they buckle, the trusses are likely to expand in the heat and bow the exterior columns perilously outward. The highly simplified calculations have not been carried beyond that point, but presumably the columns themselves could then have buckled under the tremendous weight they carried.

It all goes to show what investigators are up against in the most complex analysis of a building failure ever carried out, said Dr. Sunder, the leader of the investigation. One possible outcome is that several different plausible explanations for the collapse could emerge, he said.

"It is a process that includes analysis and judgment as we go forward," he said. "At the end, we'll have, for each tower, maybe three or five possible different scenarios that we can try and rank. We'd go from the most probable to the least probable."

Although that may not be the clean answer that historians, surviving family members and other scientists would like, it may be all that the fires of the World Trade Center have left to give.

Copyright 2004 The New York Times Company