2.2.1.2 Fire Development
It is estimated, based on information compiled from Government sources, that each aircraft contained
about 10,000 gallons of jet fuel upon impact into the buildings. A review of photographic and video records
show that the aircraft fully entered the buildings prior to any visual evidence of flames at the exteriors of the
buildings. This suggests that, as the aircraft crashed into and plowed across the buildings, they distributed
jet fuel throughout the impact area to form a flammable “cloud.” Ignition of this cloud resulted in a rapid
pressure rise, expelling a fuel rich mixture from the impact area into shafts and through other openings
caused by the crashes, resulting in dramatic fireballs.
Although only limited video footage is available that shows the crash of American Airlines Flight 11
into WTC 1 and the ensuing fireballs, extensive video records of the impact of United Airlines Flight 175
into WTC 2 are available. These videos show that three fireballs emanated from WTC 2 on the south, east,
and west faces. The fireballs grew slowly, reaching their full size after about 2 seconds. The diameters of the
fireballs were greater than 200 feet, exceeding the width of the building. Such fireballs were formed when
the expelled jet fuel dispersed and flames traveled through the resulting fuel/air mixture. Experimentally
based correlations for similar fireballs (Zalosh 1995) were used to estimate the amount of fuel consumed.
The precise size of the fireballs and their exact shapes are not well defined; therefore, there is some uncertainty
associated with estimates of the amount of fuel consumed by these effects. Calculations indicate that between
1,000 and 3,000 gallons of jet fuel were likely consumed in this manner. Barring additional information, it
is reasonable to assume that an approximately similar amount of jet fuel was consumed by fireballs as the
aircraft struck WTC 1.
Although dramatic, these fireballs did not explode or generate a shock wave. If an explosion or
detonation had occurred, the expansion of the burning gasses would have taken place in microseconds, not
the 2 seconds observed. Therefore, although there were some overpressures, it is unlikely that the fireballs,
being external to the buildings, would have resulted in significant structural damage. It is not known whether
the windows that were broken shortly after impact were broken by these external overpressures, overpressures
internal to the building, the heat of the fire, or flying debris.
The first arriving firefighters observed that the windows of WTC 1 were broken out at the Concourse
level. This breakage was most likely caused by overpressure in the elevator shafts. Damage to the walls of the
elevator shafts was also observed as low as the 23rd floor, presumably as a result of the overpressures developed
by the burning of the vapor cloud on the impact floors.
If one assumes that approximately 3,000 gallons of fuel were consumed in the initial fireballs, then the
remainder either escaped the impact floors in the manners described above or was consumed by the fire on
the impact floors. If half flowed away, then approximately 4,000 gallons remained on the impact floors to be consumed in the fires that followed. The jet fuel in the aerosol would have burned out as fast as the flame could
spread through it, igniting almost every combustible on the floors involved. Fuel that fell to the floor and
did not flow out of the building would have burned as a pool or spill fire at the point where it came to rest.
The time to consume the jet fuel can be reasonably computed. At the upper bound, if one assumes
that all 10,000 gallons of fuel were evenly spread across a single building floor, it would form a pool that
would be consumed by fire in less than 5 minutes (SFPE 1995) provided sufficient air for combustion was
available. In reality, the jet fuel would have been distributed over multiple floors, and some would have been
transported to other locations. Some would have been absorbed by carpeting or other furnishings, consumed
in the flash fire in the aerosol, expelled and consumed externally in the fireballs, or flowed away from the fire
floors. Accounting for these factors, it is believed that almost all of the jet fuel that remained on the impact
floors was consumed in the first few minutes of the fire.
As the jet fuel burned, the resulting heat ignited office contents throughout a major portion of several
of the impact floors, as well as combustible material within the aircraft itself.
A limited amount of physical evidence about the fires is available in the form of videos and still
photographs of the buildings and the smoke plume generated soon after the initial attack. Estimates of the
buoyant energy in the plume were obtained by plotting the rise of the smoke plume, which is governed by
buoyancy in the vertical direction and by the wind in the horizontal direction. Using the Computational
Fluid Dynamics (CFD) fire model, Fire Dynamics Simulator Ver. 1 (FDS1), fire scientists at the National
Institute of Standards and Technology (NIST) (Rehm, et al. 2002) were able to mathematically approximate
the size of fires required to produce such a smoke plume. As input to this model, an estimate of the openings
available to provide ventilation for the fires was obtained from an examination of photographs taken of the
damaged tower. Meteorological data on wind velocity and atmospheric temperatures were provided by the
National Oceanic and Atmospheric Administration (NOAA) based on reports from the Aircraft
Communications Addressing and Reporting System (ACARS). The information used weather monitoring
instruments onboard three aircraft that departed from LaGuardia and Newark airports between 7:15 a.m.
and 9:00 a.m. on September 11, 2001. The wind speed at heights equal to the upper stories of the towers
was in the range of 10–20 mph. The outside temperatures over the height of the building were 20–21 °C
(68–70 °F).
The modeling suggests a peak total rate of fire energy output on the order of 3–5 trillion Btu/hr,
around 1–1.5 gigawatts (GW), for each of the two towers. From one third to one half of this energy flowed
out of the structures. This vented energy was the force that drove the external smoke plume. The vented
energy and accompanying smoke from both towers combined into a single plume. The energy output from
each of the two buildings is similar to the power output of a commercial power generating station. The
modeling also suggests ceiling gas temperatures of 1,000 °C (1,800 °F), with an estimated confidence of plus
or minus 100 °C (200 °F) or about 900–1,100 °C (1,600–2,000 °F). A major portion of the uncertainty in
these estimates is due to the scarcity of data regarding the initial conditions within the building and how the
aircraft impact changed the geometry and fuel loading. Temperatures may have been as high as 900–1,100
°C (1,700–2,000 °F) in some areas and 400–800 °C (800–1,500 °F) in others.
The viability of a 3–5 trillion Btu/hr (1–1.15 GW) fire depends on the fuel and air supply. The surface
area of office contents needed to support such a fire ranges from about 30,000–50,000 square feet, depending
on the composition and final arrangement of the contents and the fuel loading present. Given the typical
occupied area of a floor as approximately 30,000 square feet, it can be seen that simultaneous fire involvement
of an area equal to 1–2 entire floors can produce such a fire. Fuel loads are typically described in terms of the
equivalent weight of wood. Fuel loads in office-type occupancies typically range from about 4–12 psf, with
the mean slightly less than 8 psf (Culver 1977). File rooms, libraries, and similar concentrations of paper materials have significantly higher concentrations of fuel. At the burning rate necessary to yield these fires, a
fuel load of about 5 psf would be required to provide sufficient fuel to maintain the fire at full force for an
hour, and twice that quantity to maintain it for 2 hours. The air needed to support combustion would be on
the order of 600,000–1,000,000 cubic feet per minute.
Air supply to support the fires was primarily provided by openings in the exterior walls that were
created by the aircraft impacts and fireballs, as well as by additional window breakage from the ensuing heat
of the fires. Table 2.1 lists the estimated exterior wall openings used in these calculations. Although the table
shows the openings on a floor-by-floor basis, several of the openings, particularly in the area of impact,
actually spanned several floors (see Figure 2-17).
Sometimes, interior shafts in burning high-rise buildings also deliver significant quantities of air to a
fire, through a phenomenon known as “stack effect,” which is created when differences between the ambient
exterior air temperatures and the air temperatures inside the building result in differential air pressures,
drawing air up through the shafts to the fire area. Because outside and inside temperatures appear to have
been virtually the same on September 11, this stack effect was not expected to be strong in this case.
Based on photographic evidence, the fire burned as a distributed collection of large but separate fires
with significant temperature variations from space to space, depending on the type and arrangement of
combustible material present and the available air for combustion in each particular space. Consequently, the
temperature and related incident heat flux to the structural elements varied with both time and location.
This information is not currently available, but could be modeled with advanced CFD fire models.
Damage caused by the aircraft impacts is believed to have disrupted the sprinkler and fire standpipe
systems, preventing effective operation of either the manual or automatic suppression systems. Even if these
systems had not been compromised by the impacts, they would likely have been ineffective. It is believed
that the initial flash fires of jet fuel would have opened so many sprinkler heads that the systems would have
quickly depressurized and been unable to effectively deliver water to the large area of fire involvement.
Further, the initial spread of fires was so extensive as to make occupant use of small hose streams ineffective.
