Click for larger image
Bob Houlihan/U.S. Navy/Getty Images

At 9:38 on the morning of September 11, 2001, as part of a terrorist action involving four hijacked aircraft, a commercial airliner was intentionally crashed into the Pentagon. One hundred eighty-nine people were killed and a portion of the building was damaged by the associated impact, deflagration, and fire. That afternoon, the American Society of Civil Engineers (ASCE) established a building performance study (BPS) team of volunteers who possess expertise in structural, fire, and forensic engineering to examine the performance of the structure in the crash and subsequent fire for the benefit of the building professions and the public. This study follows a similar examination of the April 19, 1995, bombing of the Murrah Federal Office Building in Oklahoma City, and parallels a study of the September 11 World Trade Center terrorist attack. The Pentagon BPS team’s report—The Pentagon Building Performance Report—is scheduled for release on July 31 and is available from ASCE in book form. This article is an abridged version of the report and does not include discussion of every aspect of the BPS study.

The BPS team’s analysis of the Pentagon and the damage resulting from the attack was conducted between September 2001 and April 2002. Members of the team inspected the site as soon as was possible without interfering with the rescue and recovery operations. They reviewed the original plans, the renovation plans, and all available information on the material properties of the structure. They scrutinized aircraft data, eyewitness accounts, and fatality records; consulted with the urban search and rescue engineers, the chief renovation engineer, and the engineer in charge of the crash and site reconstruction; and examined the focused assessments of the disaster conducted by the United States Army Corps of Engineers and the Pentagon Renovation Program staff.

Study Team

The BPS team included specialists in structural, fire, and forensic engineering. The following six individuals constituted the core group and are the authors of this report:

Paul F. Mlakar, Ph.D., P.E., Lead
Technical Director
U.S. Army Corps of Engineers
Vicksburg, Mississippi
Specialty: blast-resistant design; investigator, Murrah Federal Office Building study

Donald O. Dusenberry, P.E.
Simpson Gumpertz & Heger, Inc.
Arlington, Massachusetts
Specialty: blast effects and structural design

James R. Harris, Ph.D., P.E.
J.R. Harris & Company
Denver, Colorado
Specialty: structural engineering

Gerald Haynes, P.E.
Fire Protection Engineer
Bureau of Alcohol, Tobacco, and Firearms
Washington, D.C.
Specialty: fire protection

Long T. Phan, Ph.D., P.E.
Research Structural Engineer
National Institute of Standards and Technology
Gaithersburg, Maryland
Specialty: concrete structural and fire engineering

Mete A. Sozen, Ph.D., S.E.
Kettelhut Distinguished Professor of Structural Engineering
Purdue University
Lafayette, Indiana
Specialty: behavior of reinforced-concrete structures

On the basis of this information the BPS team analyzed the essential aspects of the response of the structural system of the Pentagon to the crash. Impact analyses revealed that the columns could tolerate large dynamic lateral deflections when loaded by the mass of the decelerating airplane. Static analyses indicated that the floor system was capable of significant load redistribution without collapse when several adjacent supporting columns were removed or severely damaged by the extreme action. Thermal analyses showed that the ensuing fire could have weakened frame members in approximately half an hour.

From this information the BPS team concluded that the impact of the aircraft destroyed or significantly impaired approximately 50 structural columns. The ensuing fire weakened a number of other structural elements. However, only a very small segment of the affected structure collapsed, approximately half an hour after impact. The collapse, fatalities, and damage were mitigated by the Pentagon’s resilient structural system. Very few blast-resistant windows installed during the renovation of the structure that was begun in 1999 and scheduled for completion in 2010 broke during the impact and deflagration of aircraft fuel.

The BPS team recommends that the features of the Pentagon’s design that contributed to its resiliency in the crash—that is, continuity, redundancy, and energy-absorbing capacity—be incorporated in the future into the designs of buildings and other structures in which resistance to progressive collapse is deemed important. The team further advocates that additional research and development be conducted in the practical implementation of measures to mitigate progressive collapse and in the deformation capacity of spirally reinforced columns subjected to lateral loads applied over the height of the column.

he Pentagon is one of the largest office buildings in the world, encompassing roughly 6.6 million sq ft of floor space. Its name refers to the five sides of the building, but the Pentagon is also five stories high and is subdivided into five circumferential rings, labeled A through E, beginning from the inside (figure 1 - below). In the upper three stories, the rings are separated by light wells; the well between rings B and C extends to the ground over most of its length and is known as AE Drive, which serves as an interior driveway. Ten radial corridors provide connection from ring to ring and span AE Drive. Expansion joints divide the Pentagon into 30 distinct structures (figure 2 - below).

The Pentagon is in the midst of a major renovation program (see “The Pentagon Project,” Civil Engineering, June 2001), and the work is phased in five “wedges.” Each wedge is centered on a building vertex and consists of the portion of the building between the midpoint of adjacent sides. Structurally, the renovation is not major; the most significant changes are the addition of new elevators, stairs, escalators, and mechanical equipment rooms. Additionally, the exterior walls and windows are being upgraded to provide a measure of resistance to external blasts. The renovation of Wedge 1 was essentially complete at the time of the September 11 attack.

Click for larger image

Figure 1 Overall plan of the Pentagon at the upper stories

Click for larger image
Figure 2 Overall plan at first story, showing expansion joints; hatching shows the aircraft impact zone; H is the helipad.

The original structural system, including the roof, was entirely cast-in-place reinforced concrete using normal-weight aggregate. Most of the structure used a specified concrete strength of 2,500 psi and intermediate-grade reinforcing steel (yield of 40,000 psi). The floors are constructed as a slab, beam, and girder system supported on columns, most of which are square. Member sizes vary with framing arrangements and special loads. The column sizes vary in each story—generally from about 21 by 21 in. in the first story to 14 by 14 in. in the fifth story—but there are many exceptions. Nearly all the columns that support more than one level are spirally reinforced. The remaining columns have ties. The floor spans are relatively short by modern standards: 5.5 in. slabs span to 14 by 20 in. beams at 10 ft on center. The typical beam spans are 10 or 20 ft, with some at 15 ft. Girders measuring 14 by 26 in. span 20 ft parallel to the exterior walls and support a beam at midspan.

The roof at ring E is gabled, as are those over Ring A and the radial corridors. Slabs 4.5 in. thick span perpendicular to the exterior wall with spans varying from about 8 to 11 ft. The slabs are supported by 12 by 16 in. purlins that span to rafter frames, which are 20 ft on center. The rafters are generally 16 by 24 in. and align with the floor beams and columns below. In general, the purlins do not align with the floor girders and columns below.

The roof over rings B, C, and D consists of a nearly flat pan joist and slab system. The joist stems are 6 in. wide by 8 in. deep and the slab is 2.75 in. thick. The joists are 26 in. on center and span 20 ft. The roof over the corridors is 4.5 in thick and spans 10 ft. The joists and slab are supported by 14 by 20 in. girders that are in line with the floor girders.

The perimeter exterior walls of Ring E are faced in limestone and backed with unreinforced brick infilled in the concrete frame. Nearly all remaining exterior walls are 10 in. concrete. The first-story at AE Drive is brick infilled in the concrete frame, with no windows. The concrete walls have 5 by 7 ft openings for windows and include columns built in as pilasters, corresponding to column locations below, and girders reinforced within the wall.

Slabs, beams, and girders all make use of straight and trussed bars. Except for the top reinforcement in the short spans adjacent to longer spans, there are no continuous top bars. However, approximately half of the bottom bars are made continuous by laps of 30 to 40 bar diameters at the supports. Beams and girders typically have open-topped stirrups. The longer spans generally have approximately equal areas of steel at the critical sections.

ny building is a product of its times. The Pentagon was constructed between September 1941 and January 1943. At that time the national standard predominantly used for reinforced-concrete buildings was ACI 501-36 as developed by the American Concrete Institute (ACI). Although no reference to ACI 501-36 was found in the drawings, it is very likely that this code affected decisions about member sizing and proportioning for the Pentagon structure.

ACI 501-36 was based on working stress design. The allowable stress for the intermediate-grade billet steel used in the Pentagon was 20,000 psi. For the design concrete strength of 2,500 psi, the allowable unit shear stress for beams with properly designed web reinforcement was 150 psi. The unit bond stress for deformed bars was 125 psi for the same strength of concrete.

The axial load, P, permitted on reinforced-concrete columns with spiral reinforcement was defined by the expression

P/Ag = 0.225 fc + rg fs

Ag =gross cross-sectional area;
fc=compressive strength of concrete (6 x 12 in. cylinder);
rg = ratio of longitudinal reinforcement area to area of cross section;
fs = permissible unit stress (16,000 psi for intermediate-grade steel).

This expression was conservative. The estimated service load of the column was set at approximately one-third of its expected strength on the basis of the specified compressive strength of the concrete and the specified yield stress of the reinforcement.

At that time continuity in reinforced concrete was still difficult to handle analytically. Thus ACI 501-36 specified the following moment coefficients to be used in the design.

Negative moment at face of first interior support: (1/10)wl2
Negative moment at face of interior supports: (1/12)wl2
Positive moment at center of exterior spans: (1/10)wl2
Positive moment at center of interior spans: (1/12)wl2

It is especially interesting to note that the interior-span positive and negative moments were to be of the same magnitude.

The physical characteristics of the Pentagon structure suggest that its design may have been influenced strongly by the book Reinforced Concrete Construction, by George A. Hool and Harry E. Pulver (McGraw-Hill Book Company, 1937). This work describes a floor system quite similar to that of the Pentagon and recommends a reinforcing arrangement for the girders similar to the Pentagon’s. A critical attribute of the Pentagon structure was the continuity of at least half of the bottom reinforcement across the column line to lap for a distance of at least 30 bar diameters.

Given the importance of the columns to the strength of the floor system, their original design was examined somewhat more closely. The table at the bottom of this page shows that one of the most typical columns (type 14) apparently was designed to be economical by the original designers, because the margin of allowable capacity to demand was very close to unity in the lower stories. This computation ignores any bending moment from lateral loads, which at the time of the design were probably shown to be accommodated by the one-third increase in allowable stresses that was the fashion at the time. The live loads were reduced by 20 percent for columns supporting more than one floor, the common rule for storage loads. (It was anticipated that the building would be used for record storage following World War II.)

Examination of the column design data leads to the conclusion that the minimum size used for columns was 14 in. square and that tied reinforcement was used until higher loads demanded a change. The first change was to spiral reinforcement, a 25 percent increase in allowable load by the standard of the day. The next change was in the size of the column. Given the nature of formwork at the time, today’s imperative of keeping column sizes constant was obviously not an issue.

Given the attention paid by the planners of the Pentagon renovation to the capacities of the existing structure, the BPS team did not study the lateral load capacity in much detail. A factor of note is that the seismic demands by current standards are less at the Pentagon site than the demands used in the planning studies. Given the higher material strengths found since those studies, as well as some very simple check analyses, the Pentagon structure would probably be evaluated as having a seismic capacity in line with guidelines for existing buildings.

he hijacked aircraft, a Boeing 757-200 designed to accommodate approximately 200 passengers and 1,670 cu ft of cargo, approached the west wall of the Pentagon from the southwest at approximately 780 ft/s. As it approached the Pentagon site it was so low to the ground that it reportedly clipped an antenna on a vehicle on an adjacent road and several light posts. When it was approximately 320 ft from the west wall of the building (0.42 second before impact), it was flying nearly level, only a few feet above the ground. The aircraft flew over the grassy area next to the Pentagon until its right wing struck a piece of construction equipment that was approximately 100 to 110 ft from the face of the building (0.10 second before the impact). At that time the aircraft had rolled slightly to the left, its right wing elevated. After the plane had traveled approximately another 75 ft, the left engine struck the ground at nearly the same instant that the nose of the aircraft struck the west wall of the Pentagon. Impact of the fuselage was at column line 14, at or slightly below the second-floor slab, and the right wing crossed at a shallow angle from below the second-floor slab to above the second-floor slab.

A large fireball engulfed the exterior of the building in the impact area. Interior fires began immediately.

Click for larger image
Northern portion of impact area before collapse. Photo: Associated Press/image enhanced by BPS team

The impact upon the west facade removed first-floor columns from column lines 10 to 14. First-floor exterior columns on column lines 9, 15, 16, and 17 were severely damaged, perhaps to the point of losing all capacity. The second-floor exterior column on column line 14 and its adjacent spandrel beams were destroyed or seriously damaged. Additionally, there was facade damage on both sides of the impact area, including damage as high as the fourth floor. However, in the area of the impact of the fuselage and the tail, severe impact damage did not extend above the third-floor slab.

Immediately upon impact, the Ring E structure deflected downward over the region from an expansion joint on column line 11 south to the west exterior column on column line 18. The deformation was the most severe at the expansion joint, where the deflection was approximately 18 in.

The structure was able to maintain this deformed shape for approximately half an hour, at which point all five levels of Ring E collapsed from column line 11 to roughly column line 18.

The photograph below, taken before the building to the south of the expansion joint collapsed, provides useful information that the BPS team could not observe at the site. This photograph shows that the portion of the building that subsequently collapsed was displaced vertically by approximately 18 in. to 2 ft relative to the building north of the expansion joint. The facade was missing on the first floor as far north as column line 8 (the expansion joint is at column line 11), and on the second floor, the facade was missing between column lines 11 and 15. However, windows and their reinforcing frames were still in place between column lines 11 and 13 on the second floor. The photograph also shows that the only column missing on the second floor in the west exterior wall of the building was at column line 14. The spandrel beam for the third floor and all third-floor exterior columns appears to be intact.

The photograph shows that blast-resistant glass installed as part of the Pentagon renovation was not broken by the impact or the fireball, even where the windows were located as close as 10 ft to the impact point of the fuselage.

Click for larger image
Southern portion of impact area before collapse.
Photograph by Steve Riskus/image enhancement by BPS team.
Click for larger image
Portion of Ring E at moment of collapse Photo: Defense Link

A second photograph taken before the collapse (see top of facing page) reveals that first-floor exterior columns on column lines 15, 16, and 17 were severely distorted but still attached at least at their top ends to the second-floor framing. The vertical displacement noted previously also is evident in this photograph.

The photograph at the bottom of the facing page was taken at the moment of collapse, approximately half an hour after the aircraft struck the west wall of the Pentagon. The collapse extended to approximately column line 18 on the west face of the building.

y the time the full Pentagon BPS team visited the site, all debris from the aircraft and structural collapse had been removed and shoring was in place wherever there was severe structural damage. The design team charged with reconstructing the Pentagon was assessing the building and preparations were being made to demolish the area for reconstruction. Consequently, the BPS team never had direct access to the structural debris as it existed immediately after the aircraft impact and subsequent fire.

Team members attempted to inspect and photograph all columns with significant visible damage and most of the beams and floor bays with significant visible damage. To the extent possible, it was noted whether physical loads or the effects of fire caused the observed damage. The team also noted the performance of windows and exterior wall reinforcements that had been installed prior to the attack to enhance blast resistance in Wedge 1. However, the team inspections were not comprehensive, and they did not address fire-related material degradation.

The collapsed portion of Ring E was immediately south of an expansion joint on column line 11. The collapsed area extended south from the expansion joint to approximately column line 15 on the east side of Ring E and to approximately column line 18 on the west side of Ring E. No portions of Ring D or Ring C collapsed; nor did either of the two-story sections between the rings.

Since all debris was removed prior to the detailed inspection, the team was unable to determine specifically the level and extent of impact damage in this region of the building.

None of the facade in the collapse area was accessible for inspection. However, the team did observe that limestone of the first-floor facade was seriously damaged to the north to column line 8. Some first-floor limestone panels of the facade were missing for an additional 30 to 50 ft to the north.

The site data indicate that the aircraft fuselage impacted the building at column line 14 at an angle of approximately 42 degrees to the normal to the face of the building, at or slightly below the second-story slab. Eyewitness accounts and photographs taken by a security camera suggest that the aircraft was flying on nearly a level path essentially at grade level for several hundred feet immediately prior to impact. Gashes in the facade above the second-floor slab between column lines 18 and 20 to the south of the collapse area suggest that the aircraft had rolled slightly to the left as it entered the building. The right wing was below the second-floor slab at the fuselage but above the second-floor slab at the tip, and the left wing struck the building entirely below the second-floor slab, to the north of column line 14.

The width of the severe damage to the west facade of the Pentagon was approximately 120 ft (from column lines 8 to 20). The projected width, perpendicular to the path of the aircraft, was approximately 90 ft, which is substantially less than the 125 ft wingspan of the aircraft. An examination of the area encompassed by extending the line of travel of the aircraft to the face of the building shows that there are no discrete marks on the building corresponding to the positions of the outer third of the right wing. The size and position of the actual opening in the facade of the building (from column line 8 to column line 18) indicate that no portion of the outer two-thirds of the right wing and no portion of the outer one-third of the left wing actually entered the building.

It is possible that less of the right wing than the left wing entered the building because the right wing struck the facade crossing the level of the second-floor slab. The strength of the second-floor slab in its own plane would have severed the right wing approximately at the location of the right engine. The left wing did not encounter a slab, so it penetrated more easily.

In any event, the evidence suggests that the tips of both wings did not make direct contact with the facade of the building and that portions of the wings might have been separated from the fuselage before the aircraft struck the building. This is consistent with eyewitness statements that the right wing struck a large generator before the aircraft struck the building and that the left engine struck a ground-level, external vent structure. It is possible that these impacts, which occurred not more than 100 ft before the nose of the aircraft struck the building, may have damaged the wings and caused debris to strike the Pentagon facade and the heliport control building.

The wing fuel tanks are located primarily within the inner half of the wings. The center of gravity of these tanks is approximately one-third of the wing length from the fuselage. Considering this tank position and the physical evidence of the length of each wing that could not have entered the building, it appears likely that not more than half of the fuel in the right wing could have entered the building. While the full volume of the left wing tank was within the portion of the wing that might have entered the building, some of the fuel from all tanks rebounded upon impact and contributed to the fireball. Only a portion of the fuel from the left and right wing tanks and the center fuselage tank actually entered the building.

The height of the damage to the facade of the building was much less than the height of the aircraft’s tail. At approximately 45 ft, the tail height was nearly as tall as the first four floors of the building. Obvious visible damage extended only over the lowest two floors, to approximately 25 ft above grade.

The team members do not have direct information on the impact damage to the upper floors in the collapsed portion of the building. However, based on observations of the condition of the adjoining structure and the photographs of the building before the collapse, the following general observations may be made:

Impact damage on the first floor was extensive near the entry point of the aircraft. It is likely that the exterior first-floor columns from column line 10 to column line 14 were removed entirely by the impact and that the exterior columns on column lines 9, 15, 16, and 17 were severely damaged. Most probably, many or most of the first-floor interior columns in the collapse area were heavily damaged by impact.

Click for larger image
Figure 3 Damage regions in first story.

The removal of the second-floor exterior column on column line 14, probably by the fuselage tail, suggests that the second-floor slab in this area was also severely damaged even before the building collapsed. In the portion of the building that remained standing to the north of the expansion joint, the slab and second-floor columns at column lines A, B, and C were heavily damaged. This condition, which is consistent with the trajectory of the aircraft, suggests that the second-floor slab from the expansion joint on column line 11 south to the fuselage entry point on column line 14—including columns 11B, 11C, and 13A on the second floor—was heavily damaged, perhaps destroyed.

It is difficult to judge the condition of other columns on the second floor in the collapse area. However, more likely than not column 15A was relatively undamaged. It is unlikely that columns above the second floor sustained impact damage, even in the area that ultimately collapsed.

Impact damage to the structure above the second-floor slab did not extend more than approximately 50 ft into the building. This shows that the aircraft slid between the first-floor slab on grade and the second-floor slab for most of its distance of travel after striking the building

Along the path of the movement of aircraft debris through the building, the most severe damage was confined to a region that can be represented approximately by a triangle centered on the trajectory of the aircraft in plan, with a base width at the aircraft entry point of approximately 90 ft and a length along the aircraft path of approximately 230 ft. (figure 3). However, within this triangular damage area there were a few relatively lightly damaged columns interspersed with heavily damaged columns along the path of the aircraft debris through the building. Column 1K, located 200 ft from the impact point, was the last severed column along the path of the aircraft. Note that columns on grids E and K are much weaker than the other columns because they support only one floor and a roof.

There were two areas of severe impact damage in the first story. The first area along the path of the aircraft was within approximately 60 ft of the impact point and corresponds generally to the area that collapsed. In the collapse area and for approximately 20 ft beyond the collapse area along its northern and eastern edges, columns were removed or very severely damaged by impact. In addition, there was serious second-floor beam and slab damage for 60 ft to the north of the collapse area, especially along a strip bounded approximately by column lines B and C.

The second area of severe damage was bounded approximately by column lines E, 5, G, and 9. In this region, which was beyond a field of columns that remained standing, several columns were severed and there was significant second-floor beam and slab damage. In both areas, severe slab damage appeared to be caused by moving debris rather than by overpressure from a blast.

In an effort to characterize the influence of the aircraft on the structure and, by extension, to characterize the loads on the structure, the team analyzed the available data to extract information about the destruction of the aircraft.

Most likely, the wings of the aircraft were severed as the aircraft penetrated the facade of the building. Even if portions of the wings remained intact after passing through the plane of the facade, the structural damage pattern indicates that the wings were severed before the aircraft penetrated more than a few dozen feet into the building. Ultimately, the path of the fuselage debris passed between columns 9C and 11D, which were separated by approximately 28 ft at a depth of approximately 65 ft along the aircraft’s path. Columns 9C and 11D were severely distorted but still in place: hence the wings clearly did not survive beyond this point.

At a depth of approximately 160 ft into the building, columns 3G, 3H, 3J, and 5J were damaged but still standing, although in the direct path of the fuselage. With a maximum spacing of less than 14 ft between pairs of these columns in a projection perpendicular to the path of the fuselage, it is highly unlikely that any significant portion of the fuselage could have retained structural integrity at this point in its travel. More likely, the fuselage was destroyed much earlier in its movement through the building. Therefore, the aircraft frame most certainly was destroyed before it had traveled a distance that approximately equaled the length of the aircraft.

The debris that traveled the farthest traveled approximately twice the length of the aircraft after entering the building. To come to rest at a point 310 ft from the area of impact at a speed of 780 ft/s, that debris experienced an average deceleration of approximately 30g.

The influence of the structure on the deceleration of the aircraft (and, conversely, the influence of the aircraft on the structure) can be appreciated by comparisons with aircraft belly-landed in controlled circumstances. In 1984, the Federal Aviation Administration (faa) conducted a controlled impact demonstration to evaluate the burn potential of antimisting kerosene fuel. In that test, the faa landed a Boeing 727 aircraft (weight approximately 175,000 lb) without landing gear on a gravel runway at Edwards Air Force Base. The aircraft in that test was flying approximately 250 ft/s when it made first contact, but it slid approximately 1,200 ft before it stopped. Although the test aircraft was traveling at approximately one-third the speed of the aircraft that struck the Pentagon, its sliding distance was approximately 3.9 times that of the Pentagon attack aircraft. Clearly, the short stopping distance for the aircraft striking the Pentagon derived from the energy dissipated through the destruction of the aircraft and building components; the acceleration of building contents; the loss of lift when the wings were severed from the aircraft; and effective frictional and impact forces on the first-floor slab, the underside of the second-floor slab, and interior columns and walls.

A study of the locations of fatalities also yields insight into the breakup of the aircraft and, therefore its influence on the structure. The remains of most of the passengers on the aircraft were found near the end of the travel of the aircraft debris. The front landing gear (a relatively solid and heavy object) and the flight data recorder (which had been located near the rear of the aircraft) were also found nearly 300 ft into the structure. By contrast, the remains of a few individuals (the hijacking suspects), who most likely were near the front of the aircraft, were found relatively close to the aircraft’s point of impact with the building. These data suggest that the front of the aircraft disintegrated essentially upon impact but, in the process, opened up a hole allowing the trailing portions of the fuselage to pass into the building.

Several columns exhibited severe bends. However, the predominant evidence suggests that these columns generally did not receive impact from a single, rigid object. Instead, the deformed shapes of these columns are more consistent with loads that were distributed over the height of the columns.

The analyses of the available data reveal that the wings severed exterior columns but were not strong enough to cut through the second-floor slab upon impact. (The right wing did not enter the building at the point where it struck the second-floor slab in its plane). The damage pattern throughout the building and the locations of fatalities and aircraft components, together with the deformation of columns, suggest that the entire aircraft disintegrated rapidly as it moved through the forest of columns on the first floor. As the moving debris from the aircraft pushed the contents and demolished exterior wall of the building forward, the debris from the aircraft and building most likely resembled a rapidly moving avalanche through the first floor of the building.

ire damage generally was similar to that normally resulting from serious fires in office buildings. Clearly, some of the fuel on the aircraft at impact did not enter the building, either because it was in those portions of the wings that were severed by the impact with the facade or with objects just outside of the building, or because it was deflected away from the building upon impact with the facade; that fuel burned outside the building in the initial fireball. Generally, fire damage to columns, beams, and slabs was limited to cracking and spalling in the vicinity of the aircraft debris. There were two areas with more severe damage. One area, to the north of the path of the aircraft, was bounded approximately by column lines 4, 7, A, and D. The other area, to the south of the path of the aircraft, was in the vicinity of column lines K and L and crossing column lines 11, 12, and 13. In both areas, there was more serious spalling and cracking than occurred typically throughout the fire area. Fire damage on the second floor in the vicinity of the path of the aircraft was generally more severe than in the same areas directly below on the first floor.

Click for larger image
Figure 4 Assumed stress-strain curves for confined and unconfined concrete

First-floor columns 5M, 5N, 3M, and 3N—located in Ring C, toward the end of the damage path—sustained thermal damage in the form of longitudinal cracks and corner spalling. Some sections of the columns appeared blackened, probably as a result of direct exposure to flame caused by partial loss of interior finishes. It took a little more than one hour of exposure to ISO 834—at a corresponding ambient temperature of about 1,740F (950C)—for the longitudinal cracks and corner spalling to develop in laboratory test columns. This indicates that the temperature of the fire at this location might have reached a similar level.

Similar damage was also observed for columns 7J, 7K, and 7L, suggesting that the maximum temperature at these locations might also have reached 1,740F (950C).

The structural upgrades that had been installed to enhance blast resistance performed reasonably well, considering that they were not specifically designed for aircraft impact. The only window frames removed by the impact were those struck directly by the wings or fuselage. On the second floor, immediately adjacent to where the fuselage entered the building, upgraded windows remained in their frames even though the surrounding masonry facade was completely removed.

Blast-resistant glass was generally not broken immediately after the impact or after the ensuing fire had been extinguished. By contrast, most of the original windows in a vast area of Wedge 2 were broken after the fire was extinguished. It is probable that some of these windows were broken by the fire or by firefighting efforts rather than by the effects of the impact.

he structural elements of the Pentagon that bore the brunt of the airplane impact were the first-story columns. Moment-curvature relationships for these columns were calculated assuming a mean concrete cylinder strength of 4,000 psi and a yield stress in the longitudinal reinforcement of 45,000 psi. For the concrete stress-strain properties, two different assumptions were made, as shown in figure 4. Assumption 1 was used for the gross area of the column treated as a “tied” column and corresponds to unconfined concrete, with the compressive strength of the concrete in the column 85 percent of that in the test cylinder. Assumption 2 was used for the core of the column confined by the spiral reinforcement. For the confined core, the limiting strain was defined to be that corresponding to the fracture of the reinforcement at a unit strain of 0.2. For calculating the relationship between the resisting moment and unit curvature for each type of column, an estimated service load was used reflecting the tributary dead load of the structure.

The spirally reinforced concrete core had a considerably higher calculated limiting unit curvature capacity in each case than that calculated for the gross section of the column treated as a “tied column.” The spiral cores possessed two other important properties not evident in those plots that define only cross-sectional response:

The cores enclosed by spiral reinforcement had shear strength higher than the shear corresponding to that associated with the development of the flexural strength of the core under lateral loading. For the limiting static uniform load corresponding to the critical failure mechanism, the maximum unit shear stress did not exceed three-fourths of the unit shear strength of the core.

The longitudinal bars had sufficient anchorage to develop their strengths.

These two properties eliminated the possibility of brittle failure of the cores. Indeed, none of the columns was observed to have failed in shear, and there was no evidence of pull-out of reinforcing bars. The cores and their connections did not unravel under impact. Destroying the column core required tearing it off its supports. The longitudinal reinforcing bars at each end of the column were observed to have fractured after necking, indicating ductile failure.

The impact effects may be represented as a violent flow through the structure of a “fluid” consisting of aviation fuel and solid fragments. The first-story columns in the path of this rushing fluid mass must have lost their shells immediately upon impact. It is very likely that there was never a finite time in which the affected columns responded as tied columns. The column shells must have been scoured off on first contact with the fluid. Bending resistance to the pressure created by the velocity of the fluid must have occurred in the cores only.

Considering that the axial loads were relatively light in all cases, the curvature limit is more properly based on fracture of the reinforcement. Such a limit, whether it is controlled by limiting strains in the concrete or in the reinforcement, is difficult to determine without directly relevant experimental data because the strain distribution over the column section in the regions of plastic hinging becomes acutely nonlinear at that stage of behavior. The conversion of calculated curvature to rotation is also hampered by the difficulties in defining the deformed geometry in the region of the nonlinear response. In keeping with common practice for determining the limiting draft of reinforced-concrete elements, it was assumed that the calculated limiting curvature occurred over a length equal to the core depth. The calculated limiting rotations ranged from 0.2 to 0.5. Accordingly, the spirally reinforced cores of the first-story columns would be expected to tolerate large deflections and still maintain their integrity and absorb energy in bending provided the axial load is transferred to neighboring columns.

Limiting the total energy absorbed by a column to the work done at flexural hinges at the top, bottom, and midheight of the column, the total energy absorbed would be

Wint = 4MrQlimit

where Mr is the resisting moment and Qlimit is the limiting concentrated rotation.

Assuming rigid plastic response (recognizing that the response is neither initially rigid nor eventually plastic in the exact sense) and assuming that the impact imparted an initial velocity to the column without continuing to exert drag after this, the maximum initial velocity that the column could sustain without disintegration would then be estimated by

vlimit = (2Wint/ Masseff)1/2

where Masseff may be taken as one-half the total mass of the column.

Evaluation of the above expression for velocity resulted in limiting column initial velocities ranging from approximately 100 to 200 ft/s for the column cores analyzed.

Several numerical simulations of a fluid mass (in this case modeled as aviation fuel) impacting a reinforced-concrete column fixed top and bottom were made by S.A. Kilic, a visiting scholar at Purdue University, in support of the BPS study of the Pentagon. These simulations indicated that the maximum response velocity of the column was comparable to the velocity of the impacting fluid. The conclusion for the facade columns is self-evident. Their maximum response velocities could not have been less than 600 ft/s (vis--vis the impact velocity of approximately 780 ft/s). These columns engulfed by the fluid would have been destroyed immediately, however much energy might have been deflected by the facade walls and slabs. The question of interest is whether there was any system to the distribution of the severely damaged columns in the first story.

It is plausible to expect that the energy content of the impacting fluid mass attenuated—as it penetrated the building—as the square of the distance from the point of impact. Recognizing that the debris was not thrown more than a distance of 310 ft and accepting the impacting velocity of approximately 780 ft/s, it may be inferred that the velocity of the fluid would have reached a value of approximately 100 ft/s, a velocity that, at a distance approaching 200 ft from the point of impact, most column cores would be expected to resist without disintegration.

There is no question that the progress of the impacting fluid in the structure must have verged on the chaotic. The reasoning in the preceding paragraphs is not presented as a prediction of an orderly process but as a preliminary rationalization of the distribution of severe damage to the spirally reinforced column cores immediately after impact. The important conclusion is that the observed distribution of failed columns does not contradict simple estimates made on the basis of elementary mechanics. The same reasoning would suggest that had the columns in the affected region been tied columns, all would have been destroyed, leading to immediate collapse of a large portion of the building.

The segment of the building that was exposed to heavy impact and fire but survived is bounded by column lines 1 and 11 (expansion joints) and AA and O north-south (figure 5, following page). In the following text, this segment will be designated as segment P. The small segment shown west of the expansion joint at column line 11 (double columns) includes the segment of the structure that eventually collapsed. For the purpose of this study it is taken to be bounded by column lines 11, 17, AA, and D. This segment will be designated Q (figure 6, at bottom of page).

Click for larger image
Figure 5 Yield line analysis of area with missing columns
Click for larger image
Figure 6 Yield line analysis of collapse area, segment Q

The limiting flexural strength of the floor system was determined in order to establish a measure of the strength of the system. The investigation was made not to find out how much load the system would carry but to determine the quality of the construction by using the limiting flexural capacity as an index value.

Flexural capacities at critical sections were determined, ignoring the effect of compression reinforcement. It was assumed at the support sections that the tensile reinforcement in the slab acting as a flange was effective in resisting flexure. The width of the flange was defined to be equal to the clear depth of the beam below the slab. Considering the moment gradient along the span at sections where flexural yield was expected, the effective stress in the tensile reinforcement was assumed to be 5/4 times the yield stress at room temperature.

Unit-resisting loads corresponding to the development of flexural failure mechanisms with yield lines paralleling the column centerlines identified by numerals were determined for segments P and Q. The minimum unit yield load calculated was more than 1,300 psf. The calculations were repeated, assuming that entire rows of columns were missing, resulting in a span length of 40 ft center to center of the columns. For those conditions, the unit load calculated was not less than 300 psf. Similar results were obtained when assuming a row of interior columns along lettered column lines missing, which results in a longer span for the beams.

The calculated probable value of 1,300 psf does not refer to the actual capacity of the structure, as other modes of failure might govern before this load could be achieved. But it does attest to the impressive intrinsic strength of the floor system and explains, along with the observation that the bottom bars were lapped at column lines, why the structure could tolerate losses of columns.

Simplified yield line analyses were also performed for areas within segment P that were missing several columns. Figure 5 shows two such areas, P1 and P2. The two areas are shown divided because the light well wall above provides a stiff and strong support, and the second-story columns were able to act as hangers because the column vertical reinforcement was sufficiently well developed (having the lap enclosed within a spiral may have been a factor). The calculated capacities of the floor system within P1 and P2 are both over 350 psf, more than twice the dead load. The light well wall was also analyzed for the loads that the area P1 and P2 mechanisms would deliver to it, and this analysis shows that some support from the two significantly impaired columns at 9 and 11 F would have been required to prevent a failure. The spiral reinforcement in those two columns was a key factor in preventing a much more widespread collapse.

Figure 6 depicts the segment of the building between column lines 11 and 17 and AA and D. To obtain another perspective on the intrinsic strength of the floor system, the capacity of the floor to resist load at room temperature was estimated by assuming that all columns at the intersections of column lines 12 through 16 and AA, A, B, and C were lost. The assumed locations of the negative- and positive-moment yield lines are shown in the figure. The floor-system edge along line 11—the location of the expansion joint—was assumed to be unsupported. The support along line AA was considered to be provided by the facade wall that had not been destroyed by the impact.

The calculated capacity for the failure condition shown ideally in figure 6 was approximately 160 psf (with the dimension x set at 35 ft, corresponding to the minimum calculated yield load). For the assumed boundary, material, and support conditions, the floor system at level two would have been able to support itself over the assumed unsupported area. There is no definitive information on how many first-story columns in this segment were demolished immediately after impact, although it is plausible to assume that the columns at the intersections of line AA with lines 11 through 16 were not functional at the time the photograph at the top of page 44 was taken. The photographs taken immediately after the impact indicate that segment Q might have derived some support from segment P before segment Q collapsed. The partial support might have provided a collateral mechanism for resisting the overall gravity loads on segment Q considered as a building block, but the floor system would still have had to be able to carry itself. The simplified yield line analysis indicates that this would have been within the capacity of the floor system. The eventual collapse of this section is attributed to the effects of heat from the fire or high-strain creep, possibly exacerbated by the water pumped into the structure to quell the fire.

he Pentagon’s structural performance during and immediately following the September 11 crash has validated measures to reduce collapse from severely abnormal loads. These include the following features in the structural system:

• Continuity, as in the extension of bottom beam reinforcement through the girders and bottom girder reinforcement through the columns;
• Redundancy, as in the two-way beam and girder system;
• Energy-absorbing capacity, as in the spirally reinforced columns;
• Reserve strength, as provided by the original design for live load in excess of service.

These practices are examples of details that should be considered in the design and construction of structures required to resist progressive collapse.

he Pentagon crash supports the need for research and development in progressive collapse and extreme lateral column response. The following topics are of particular interest:

Consolidation of information on prevention of progressive collapse: Much has been written on means to prevent progressive collapse, but little detailed guidance has been incorporated into the building codes in general use in the United States. There should be a focused effort to accumulate research and practical experience in the area of structural robustness so that an authoritative guide can be prepared that will be useful to the design community.

Influence of extreme column deformations on load-carrying capacity: The columns in the Pentagon deformed laterally to several times their diameter. In this highly plastic, postfailure state they continued to function as structural elements. Research should be performed to determine the load-carrying capacity of columns and other structural elements once they have been deformed beyond their maximum load-carrying state and are in the range of declining strength.

Influence of extreme column deformations on loads within a statically indeterminate structure: Once the columns in the Pentagon deformed laterally beyond a certain amount, most certainly they began to pull down on the structure above, acting as a catenary. Under this circumstance, the columns placed additional demand on the adjacent structure, at least for the brief time that it experienced the lateral load that caused the horizontal displacement. Research should be conducted to understand the implications of this short-term load on the survivability of structures.

Energy-absorbing capacity of reinforced-concrete elements: The columns of the Pentagon absorbed energy as they deformed at a very high strain rate. Research should be conducted to understand the energy-absorbing capacity of concrete elements when they are subjected to impact and impulse loads that result in large deformation.

Ability of a structure to withstand extreme impact: The Pentagon crash provided a rare view of the collision of an aircraft with a building. As such, the data collected and such other data as may be available or can be generated on this subject should be studied to extract information useful to engineers charged with designing buildings so as to reduce risks caused by extreme impacts that can induce extensive damage.

Order your copy of the Pentagon Building Performance Report today!

CE Magazine Table of Contents  |  ASCE Publications Home Page  |  ASCE Home Page

Copyright 2003 ASCE. All rights reserved.