The Collapse of Buildings

Why the World Trade Center towers collapsed


Buildings, like all structures, are designed to support certain loads without deforming excessively. The loads are the weights of people and objects, the weight of rain and snow and the pressure of wind--called live loads--and the dead load of the building itself. With buildings of a few floors, strength generally accompanies sufficent rigidity, and the design is mainly that of a roof that will keep the weather out while spanning large open spaces. With tall buildings of many floors, the roof is a minor matter, and the support of the weight of the building itself is the main consideration. Like long bridges, tall buildings are subject to catastrophic collapse.

The causes of building collapse can be classified under general headings to facilitate analysis. These headings are:

Bad design does not mean only errors of computation, but a failure to take into account the loads the structure will be called upon to carry, erroneous theories, reliance on inaccurate data, ignorance of the effects of repeated or impulsive stresses, and improper choice of materials or misunderstanding of their properties. The engineer is responsible for these failures, which are created at the drawing board.

Faulty construction has been the most important cause of structural failure. The engineer is also at fault here, if inspection has been lax. This includes the use of salty sand to make concrete, the substitution of inferior steel for that specified, bad riveting or even improper tightening torque of nuts, excessive use of the drift pin to make holes line up, bad welds, and other practices well known to the construction worker.

Even an excellently designed and constructed structure will not stand on a bad foundation. Although the structure will carry its loads, the earth beneath it may not. The Leaning Tower of Pisa is a famous example of bad foundations, but there are many others. The old armory in St. Paul, Minnesota, sank 20 feet or more into soft clay, but did not collapse. The displacements due to bad foundations may alter the stress distribution significantly. This was such a problem with railway bridges in America that statically-determinate trusses were greatly preferred, since they were not subject to this danger.

Extraordinary loads are often natural, such as repeated heavy snowfalls, or the shaking of an earthquake, or the winds of a hurricane. A building that is intended to stand for some years should be able to meet these challenges. A flimsy flexible structure may avoid destruction in an earthquake, while a solid masonry building would be destroyed. Earthquakes may cause foundation problems when moist filled land liquefies.

Unexpected failure modes are the most complex of the reasons for collapse, and we have recently had a good example. Any new type of structure is subject to unexpected failure, until its properties are well understood. Suspension bridges seemed the answer to bridging large gaps. Everything was supported by a strong cable in tension, a reliable and understood member. However, sad experience showed that the bridge deck was capable of galloping and twisting without restraint from the supporting cables. Ellet's bridge at Wheeling collapsed in the 1840's, and the Tacoma Narrows bridge in the 1940's, from this cause.

The conservative, strong statically-determinate trusses were designed with pin-connected eyebars to be as strong and safe as possible. Sad experience brought the realization of stress concentration at the holes pierced in the eyebars. From earliest times, it has been recognized that tension members have no surprises. They fail by pulling apart when the tension in them becomes too high. If you know the tension, then proportioning a member is easy. A compression member, a column, is different. If it is short and squat, it bears its load until it crushes. But if you try to support a load with a 12-foot column that will just support the load with a 1-foot column, you are in for a surprise. The column bends outward, or buckles, and the load crashes to earth.

Suppose you have a beam supported at the ends, with a load in the center. You know the beam will bend, and if the load is too great, it may break apart at the bottom, or crush at the top, under the load. This you expect. However, the beam may fail by splitting into two beams longitudinally, or shearing, or by the top of the beam deflecting to one side or the other, also called buckling. In fact, a beam will usually fail by shearing or buckling before breaking.

A hollow tube makes a very efficient column or beam. If you think about it, it is the material on the surface that most resists buckling and bending. A column that is modified from a compact cross-section, like a cylinder, to an extended cross-section, like a pipe, can still support the same load per unit area, but with much greater resistance to buckling. As a beam, one side is in compression and the other in tension, while the pipe cannot buckle to one side or the other. When you do bend a pipe, notice that it crushes inward reducing the cross-section to a line, which bends easily. Tubes need to be supported against buckling. Such a tube has a very high ratio of strength to weight, and hence strength to cost.

Tall buildings have generally been made with a rigid steel skeleton, sheathed in the lightest materials to keep out the weather. Alternatively, reinforced concrete, where the compression-resisting and protecting concrete surrounds the tough, tension-resisting steel, integrated into a single body, has been used. Such structures have never failed (when properly built on good foundations), and stoutly resist demolition. When the lower supports of a steel skeleton are destroyed, the weight of the building seems to crush the lower parts and the upper parts descend slowly into the pile of debris. Monolithic reinforced-concrete buildings are diffcult to demolish in any fashion.

The World Trade Center towers used neither a steel skeleton nor reinforced concrete. They were designed as square tubes made of heavy, hollow welded sections, braced against buckling by the building floors. Massive foundations descended to bedrock, since the towers had to be safe against winds and other lateral forces tending to overturn them. All this was taken into consideration in the design and construction, which seems to have been first-rate. An attempt to damage the buildings by a bomb at the base had negligible effect. The strong base and foundation would repel any such assault with ease, as it indeed did. The impact of aircraft on the upper stories had only a local effect, and did not impair the integrity of the buildings, which remained solid. The fires caused weakening of the steel, and some of the floors suddenly received a load for which they were not designed.

What happened next was unexpected and catastrophic. The slumped floors pushed the steel modules outwards, separating them from the floor beams. The next floor then collapsed on the one below, pushing out the steel walls, and this continued, in the same way that a house of cards collapses. The debris of concrete facing and steel modules fell in shower while the main structure collapsed at almost the same rate. In 15 seconds or so, 110 stories were reduced to a pile 9 stories high, mainly of steel wall modules and whatever was around them. The south tower collapsed 47 minutes after impact, the north tower 1 hour 44 minutes after impact. The elapsed times show that the impacts were not the proximate cause of collapse; the strong building easily withstood them. When even one corner of a floor was weakened and fell, the collapse would soon propagate around the circumference, and the building would be lost.

It is clear that buildings built in this manner have a catastrophic mode of failure ("house of cards") that should rule out their future construction. It is triggered when there is a partial collapse at any level that breaks the continuity of the tube, which then rolls up quickly, from top to bottom. The collapse has a means of propagation that soon involves the whole structure, bypassing its major strengths and impossible to interrupt. There is no need for an airliner; a simple explosion would do the job. There were central tubes in the towers, for elevators and services, but they appeared to play no substantial role in the collapse, and were not evident in the pictures or wreckage.

The NOVA Production, Why The Towers Collapsed

I viewed this hour-long production on 30 April 2002. It was very interesting, but seemed to concentrate on the effects of fire and on building safety, rather than on the collapse mechanism. When the collision of an airplane with a tower was shown, the bang was simultaneous with the flash, showing that we were viewing an edited version. The connections of the floor trusses with the external wall were indeed not very substantial--two 5/8" bolts each, so the floors and the wall could easily be separated, as I assumed above.

The role of the fire in weakening the steel where the collisions occurred is undoubted, as is the ineffectiveness of the fire protection foam, which seems to have been blown off. However, this can only cause a local collapse, depositing an unusual load on the floors below. It is their response, of a structure unweakened by fire or impact, that is significant, and this topic was brushed aside in the program. In fact, an erroneous graphic of floors collapsing on one another successively, "pancaking," was shown, while the collapse of the towers was quite different, the upper floors ending up on the bottom of the pile and the lower floors on the top. One commentator actually mentioned the buckling of the wall (without mentioning buckling), but did not follow up.

More detail was presented on the core, which contained the stairways and elevators, plus building services such as firefighting water (which was only interrupted in the North Tower). This core never appears in the videos as an element of strength, though the floor trusses were supported on it. One might suspect that when the outer walls failed, the core was simply pulled apart and collapsed. The collapse of the North Tower shows the TV antenna initially falling, though the walls were already clearly in collapse. The conclusion that the central core failed in this case pulling down the outside seems very ill-founded. In the views of the South Tower, there is no evidence at all of the core. In both cases, the collapse was simultaneous around the building, not asymmetrical.

It should be recognized that the damage to the towers was different, as the program made clear, so we have two examples of this kind of failure, not just one. The program stays away from the embarrassing conclusion that this kind of structure has an inherent failure mode, as I have suggested. Perhaps we must have further examples to make this clear. The Empire State Building was also struck by an airplane (a bomber) that did considerable damage, but there was no hint whatever that the building was in danger. One suspects that if an airplane struck a building with a volume skeleton, there would be no total collapse, only local damage. Rubbish from the collapsed part would fall outward to the sides, not pry the building apart from the inside.


Return to Tech Index

Composed by J. B. Calvert
Created 23 September 2001
Last revised 30 April 2002