Why the Towers Fell: The Engineering Record

The World Trade Center was a place of work, travel and daily life before it became the site of a catastrophe. Construction began in 1966. The complex was dedicated on April 4, 1973, after the first tenants had already moved in. Its two 110-story towers joined columns, floors and foundations into a connected structural system. 9/11 Memorial & Museum, construction and dedication record.

On September 11, 2001, hijackers flew American Airlines Flight 11 into the North Tower and United Airlines Flight 175 into the South Tower. Both buildings remained standing after the impacts. The South Tower, struck second, collapsed first, about 56 minutes after impact. The North Tower collapsed about 102 minutes after it was hit. NIST NCSTAR 1, Chapters 2–3.

The attacks killed 2,977 people, including 2,753 in New York. These are the 9/11 Memorial & Museum's victim totals for the attacks, not a count confined to occupants of the two towers. An engineering account explains part of what happened that morning; it cannot encompass the lives lost or the experiences of survivors and responders. 9/11 Memorial & Museum, After the Recovery.

This explainer begins with how the towers were built, then follows how impact and fire changed their structure. Why did buildings that survived the initial damage later lose stability? The answer depends on the interaction of damaged columns, lost fire protection, heated floors and the connections that remained effective. The two towers shared a design approach, but had different damage and fire histories. NIST NCSTAR 1, §§6.14.6–6.14.7.

Reading the reconstruction

The interactive separates three kinds of information:

• Observed: photographs and videos record exterior damage, visible fire activity, wall bowing and early upper-section movement. Interviews and emergency records document evacuation and response.
• Calculated: NIST's analyses reconstruct interior impact damage, fire development, steel temperatures and structural response. These depend on evidence, assumptions and selected model cases.
• Illustrated: camera moves, assembly reveals and transitions between source states help explain the relationships. They do not supply missing measurements or a continuous simulation of the morning.

NIST compared alternative damage cases and developed a probable collapse sequence for each tower. Observations constrained those analyses; they did not reveal every internal condition. This viewer also brings together results from separate studies. Playback progress is explanatory, not historical elapsed time. NIST NCSTAR 1, Chapter 6.

Who investigated what

FEMA's Building Performance Assessment Team produced the preliminary FEMA 403 study in 2002. It documented early findings and identified issues for further investigation. FEMA 403, report record at NIST.

The National Institute of Standards and Technology (NIST) announced its investigation in August 2002 and issued the final towers report in October 2005. Its work examined building performance, evacuation and emergency response, and led to recommendations for improving building safety. The later investigation of WTC 7 addressed a different building and collapse mechanism. NIST, investigation dates and scope; investigation overview and reports.

The 9/11 Commission examined the attacks, preparedness and response. Its account of the morning supplies historical context; NIST's investigation supplies the principal engineering basis for this explainer. The 9/11 Commission Report, Chapter 9.

The main animation depicts early rotation and descent. The North model reaches a representative 8° south / 80 ft downward pose; its downward placement is a conditional archive-image estimate, not a NIST measurement paired with that angle. The South model passes 7° east / 3° south and continues to 20° east, holding the selected 3° south component. Reference cuts, intermediate poses and playback timing remain illustrative. The written account continues with the subsequent collapse and the limits of the evidence.

The towers' height depended on a connected system: a dense steel perimeter, a central core and floors spanning between them. Follow those connections from the foundations upward. They explain both how the buildings carried everyday loads and how the surviving structure responded after impact.

Below ground: enclosure, foundations and anchors

Construction began with the ground. Much of the site lay on fill beside the Hudson River. A concrete slurry wall held back surrounding earth and water so crews could excavate down to bedrock. Crews dug the wall trench in sections, supporting its sides with bentonite slurry before installing reinforcement and concrete. Tiebacks then anchored the enclosure into the ground and rock beyond it as excavation deepened. 9/11 Memorial & Museum, the slurry wall.

Inside that enclosure, the tower foundations had a different job: spreading the columns' loads into bedrock. The animation first locates those supports, then follows the 47 core columns upward. Around them, a second system of more widely spaced exterior columns carries the tower's outer wall toward the foundations. The lower core and exterior supports remain together in view as the frame takes shape. NIST NCSTAR 1-1, pp. 9–10 and Fig. 2–3.

Horizontal members connect the exterior columns into bays. Diagonal braces then join those bays into triangles, stiffening the lower frame against sideways movement. These visible diagonals belong to the exterior basement frame. Keep this same core and braced exterior frame in view as it continues above ground into the lobby.

The animation moves source-based members into their reference positions so their connections can be followed. Placement paths and timing are illustrative; they do not establish the exact historical erection order or reconstruct every footing connection. The enclosure's tiebacks remain a separate system. As basement floors were completed, they supplied permanent lateral support to the slurry wall. George J. Tamaro, The World Trade Center “Bathtub,” a Case History.

In the model: Follow the enclosure, foundations and lower frame · Inspect the tieback beat.

Continue the frame through the lobby

The core and exterior supports introduced below ground continue through the lobby together. Blue core members seat directly above the basement columns. Steel-gray exterior columns arrive first, followed by amber tree panels at the source splice, elevation 363 ft (53 ft above concourse). Around it, branching steel assemblies, often called column trees or tridents, gather groups of three closely spaced upper columns into the wider-spaced supports below. The prefabricated trees extend to the splice above floor 9; the regular wall panels continue above them. That transition opens the tall entrance while continuing the perimeter's load path to the foundations. NIST NCSTAR 1-1, pp. 10–11 and §7.2.3–7.2.4.

In the model: Watch the continuous sequence from the bathtub through the lobby.

The tube-frame concept

Each tower concentrated its principal column systems at the perimeter and core, leaving open office space between them. Above the lobby transition, each face had 59 closely spaced perimeter columns, for 236 around the tower. Columns and deep horizontal spandrels formed the exterior framed tube, which resisted wind forces and carried part of the building's weight. NIST NCSTAR 1-1, pp. 10–12.

Typical exterior panels arrived as prefabricated units of three columns and three stories. Neighboring units met at staggered joints and were bolted together into the continuous wall. The panel animation holds on those joints before showing how the same system repeated upward.

Inside, 47 core columns carried gravity loads alongside the elevators, stairs and other services. The core and perimeter were connected through the floors and the hat truss at the crown. Their ability to share loads would matter after the aircraft impacts. NIST NCSTAR 1, §§6.14.6–6.14.7.

In the model: Watch neighboring wall panels meet.

The floors: light, long, and critical

The typical floor system also arrived in prefabricated sections. A roughly 20-foot-wide module combined trusses, transverse bridging and metal deck. It contained two interior pairs of trusses and a single truss at each edge; adjacent modules brought their single edge trusses together at a shared seat. Concrete was placed after the assemblies were installed (NIST NCSTAR 1-1, p. 12; NCSTAR 1-3A, p. 11). The animation follows the complete steel span first, adding transverse bridging and metal deck in their assembled positions. The blue steel-and-deck section lowers once onto the fixed core and perimeter supports and remains there as its gold neighbor arrives alongside it. An interior view then holds on their shared perimeter seat: blue and gold bearing angles rest on the fixed wall seat, with the bolts highlighted between the truss ends. Concrete has its own later step: it appears across the installed deck after the camera returns to the full span. A lengthwise section shows the slab above the deck and the truss knuckles embedded in it before the surrounding floor returns. The floor-98 module boundaries explain the system; they are not recovered fabrication marks.

A typical office floor used a concrete slab over metal deck, supported by open-web steel trusses. The trusses spanned roughly 60 feet in the longer direction and 35 feet in the shorter direction between the core and perimeter. Their lattice of steel members carried the floor with less weight than a comparable solid beam. NIST NCSTAR 1-1, p. 12.

The floors carried people, furniture and their own weight. They also acted as horizontal diaphragms, joining the structural systems and bracing the perimeter columns. A floor's response therefore mattered beyond its own span. Heating, sagging, changes in restraint and some connection failures could alter the forces on the walls. Sagging alone does not mean that every connection failed or all lateral support disappeared. NIST NCSTAR 1, §6.6.4; NIST investigation FAQ, question 17.

In the model: Inspect the floor modules and their connections · Follow the connected structure upward.

The crown: connecting core and perimeter

The structural ascent ends at the hat truss, between floor 107 and the roof. Diagonal members meet columns and horizontal framing to form a trussed framework across the top of each tower. Follow one complete triangle, then trace its connections from the core toward the perimeter. These connections give loads another route between the inner and outer column systems. NIST NCSTAR 1-1A, §2.3.8.

The crown sequence locates this framework within the upper building before revealing its members and connections. It then identifies the North Tower's support for its telecommunications mast, about 360 feet tall, while retaining the surrounding crown. The mast is a later addition to the tower's structural frame. The South Tower had no comparable mast; its roof included an outdoor observation deck. 9/11 Memorial & Museum, World Trade Center facts and figures.

With the structure connected from bedrock to crown, the story turns to how people moved through it: first elevators, then stairs.

In the model: Explore the hat truss and its connections.

Circulation inside the core

Express elevators served sky lobbies at floors 44 and 78, where passengers could transfer to local banks. The animation begins in the ground-level lobby, follows an express journey to floor 78, then shows the sky-lobby transfer and a local elevator route to floor 91. The lobby frame and full shaft height remain visible. The moving trace explains those connections; it does not represent a particular cab, waiting time, operating speed or historical journey (NIST NCSTAR 1-7, §2.2.3, pp. 32–34; the original elevator plans are cataloged in the model).

In the model: Follow an express-to-local journey.

Stairs and transfers

Three stair routes also passed through the core. Their positions changed with the building's layout, using flights, landings and transfer corridors. The construction section moves through floors 74–83 and into the modeled B-stair transfer at 76. These are reference routes; their drawn continuity does not establish that they remained passable after impact (NIST NCSTAR 1-7, §2.2.2, pp. 27–31; source stair plans cataloged in the model).

In the model: Look inside the stair routes and transfers · Revisit the routes after impact.

The completed towers: a place of daily life

The final construction view returns to both complete buildings. Foundations, columns, floors and crown formed the structure; elevators and stairs connected the people inside it to the city below. These towers were workplaces, visitor destinations and part of everyday life before September 11.

In the model: Return to the completed towers · Begin The Day with both towers.

Fireproofing: slowing the heating of steel

The steel floor framing relied on sprayed fire-resistive material (SFRM) to slow its heating during a fire. The material was a protective coating, not part of the steel itself. Its performance depended on its properties, thickness, coverage and remaining attachment. NIST NCSTAR 1-5, thermal protection.

Fireproofing had been upgraded on the North Tower's fire-affected floors. The South Tower's affected floors retained the original application. The impacts nevertheless dislodged substantial amounts of protection. NIST concluded that widespread insulation loss was critical to the subsequent structural response, and that the towers likely would not have collapsed under the combined impact and fires without it. The extent of the loss was estimated through reconstruction. NIST investigation FAQ, question 14; NCSTAR 1, §6.14.5.

The same connections, under different conditions

The towers survived the initial impacts as their remaining structure redistributed loads. What followed depended on the changed condition of that structure: which columns and floors were damaged, where insulation was lost, how fires developed, and which connections remained effective. The Day follows those relationships through each tower's distinct sequence.

The aircraft impacts severed columns, damaged floor systems, dislodged insulation and disrupted fire protection. Both towers remained standing immediately afterward as the surviving structure redistributed loads. NIST reconstructed interior damage through impact analyses; photographs principally constrained the exterior damage.

In the model: Locate both towers and the selected impact region · Inspect exterior and core damage.

WTC 1 (North Tower): American Airlines Flight 11, 8:46 a.m.

Flight 11 struck the north face at 8:46:30 a.m., across floors 93–99. NIST estimated a speed of 443 ± 30 mph, a downward approach angle of 10.6 ± 3°, and 25 ± 2° of left-wing-down roll. The approach was close to perpendicular to the north wall in plan; the aircraft was descending and banked. NIST NCSTAR 1-2, Table E-8, p. lxxiii.

NIST's photographic assessment identified north-wall columns 112–144 and 151 as severed, columns 145–148 as heavily damaged, and 149–150 as moderately damaged. These are aggregate column identities, not a claim that every story of each column disappeared. Interior impact calculations gave 3 severed and 4 heavily damaged core columns in the base case, and 6 severed and 3 heavily damaged in the more severe case. Those estimates must retain their case labels. NIST NCSTAR 1-2, Finding 16 and Tables 7-1/7-4, pp. 183 and 205.

All three stair routes became impassable through the impact region, cutting off escape from the floors above. The access geometry shows reference routes; it cannot reconstruct all debris or obstructions. NIST NCSTAR 1, pp. 23–24.

The surviving frame redistributed loads immediately after impact. The hat truss at the crown resisted downward movement of the damaged north wall and helped transfer load to other walls and the core; adjacent columns, spandrels and floors also participated. The North Tower's Impact animation visits that retained hat-truss geometry before returning to the damaged region. This response is distinct from later redistribution caused by heating and core shortening. NIST NCSTAR 1, §§6.14.2 and 6.14.6, pp. 144 and 150.

In the model: Inspect the North Tower core routes.

WTC 2 (South Tower): United Airlines Flight 175, 9:03 a.m.

NIST places impact at 9:02:59 a.m., across floors 77–85 of the south face. Its estimated speed was 542 ± 24 mph, with a 6 ± 2° downward approach and 38 ± 2° of left-wing-down roll. The impact path was oblique, with damage concentrated toward the southeast portion of the core. NIST NCSTAR 1-2, Table E-8 and Finding 17.

The photographic assessment identified south-wall columns 410–436 and 438–439 as severed, and 437 as heavily damaged. NIST's core impact estimates were 5 severed and 4 heavily damaged columns in the base case, and 10 severed and 1 heavily damaged in the more severe case. These are different from totals for exterior damage or for all damage classes. The calculated redistribution left the east wall carrying more load; the core's slight lean toward the east and south is part of the reconstructed structural response. NIST NCSTAR 1-2, Tables 7-6/7-9, pp. 228 and 251; NCSTAR 1, §6.14.7, p. 151.

Stairway A remained passable through the impact region and provided an escape route for some occupants above it. Stairways B and C were impassable. A visible stair in the model alone does not establish usable access. NIST NCSTAR 1, pp. 40–42.

In the model: Inspect the South Tower core routes.

Designed routes and usable access

The access animation places the reference stair and shaft geometry beside the selected tower's damage. It does not simulate debris, doors, individual journeys or a clear walking path. Read the tower-specific accounts above to distinguish the designed route from the evidence for passage.

Four consequences of impact

1. Column damage and redistribution. Remaining columns, floors, spandrels and the hat truss redistributed gravity loads. Impact damage alone did not initiate collapse.
2. Loss of thermal insulation. NIST estimated widespread dislodgment from columns and trusses. The tower summaries report affected trusses over about 60,000 square feet in WTC 1 and 80,000 square feet in WTC 2, with insulation lost on at least one floor of 43 and 39 core columns respectively. These are model-based estimates, not measured stripping boundaries.
3. Fuel dispersal and ignition. Roughly 10,000 gallons of jet fuel in Flight 11 and 9,100 gallons in Flight 175 ignited widespread fires. NIST estimated that less than 15 percent burned initially inside each building; the external fireballs consumed a comparable fraction for WTC 1 and roughly 10–25 percent for WTC 2. Well over half remained initially unburned. Office contents sustained the subsequent fires.
4. Loss of fire protection and access. The impacts severed sprinkler supply piping and obstructed escape routes in the affected regions.

NIST NCSTAR 1, pp. 22–24, 38–42 and 148–149.

What the damage geometry represents

The exterior entry faces combine FEMA panel outlines with separately identified NIST observations. Interior floor footprints use the final NIST impact-case figures: solid outlines indicate severe structural damage; dotted outlines indicate floor openings. NIST excluded the severe regions from its subsequent structural analyses. Retained amber steel in this viewer is a display convention and does not establish load capacity. The report diagrams do not locate every fracture or identify every historical truss. NIST NCSTAR 1-2, pp. 187 and 232, Figs. 7-12, 7-26, 7-42 and 7-56.

Steel did not have to melt for the towers to lose stability. NIST calculated upper-layer air temperatures reaching about 1,000 °C in parts of the fire floors. Heating reduced steel strength and stiffness and produced expansion, deformation and time-dependent creep. NIST NCSTAR 1, pp. 28–30.

In the model: Move through the fire-affected region.

The animation's amber bands identify a floor region for the explanation. Their glow is not a temperature field, fire-spread calculation or insulation-loss map. Use the section's Deep dive to compare the fire maps and their source times before following the floor response.

How heat affects steel

Temperature-dependent yield strength, elastic stiffness, and a member's load-carrying capacity are different quantities. NIST used tested material properties in its analyses; a single temperature does not specify a universal buckling threshold. Its main report illustrates about a 20 percent yield-strength loss near 300 °C. Its FAQ describes strength falling to roughly 10 percent of the room-temperature value near 1,000 °C. These material summaries must not be interpreted as percentages of whole-column or tower capacity. NIST NCSTAR 1, p. 29 and Fig. 2-9; NIST investigation FAQ, questions 16 and 21.

High-temperature creep is deformation that accumulates under sustained load. Whether a heated member becomes unstable depends on its temperature history, stress, geometry, restraint and material properties. The viewer does not calculate those quantities.

Gas temperature and steel temperature

NIST first reconstructed the fires and gas temperatures, then calculated heat transfer into the structural components. Gas and steel temperatures were nonuniform and changed with location and time. A gas-temperature maximum cannot be applied to every member on a floor. NIST NCSTAR 1-5, Chapter 6; NCSTAR 1-5G.

Slender floor-truss members generally heated more rapidly than massive sections under comparable exposure, especially where insulation was missing. Heating also depended on section size, insulation thickness, remaining protection and the moving fires. The core included both box and wide-flange columns; neither section type nor a single heating rate applies to every location.

Why fireproofing mattered

NIST concluded that the towers likely would not have collapsed under the combined impact and fires if thermal insulation had not been widely dislodged. The extent of insulation loss was estimated through impact and thermal modeling, with uncertainty. The pale sleeve in the animation explains the protection's role; it does not map the actual dislodgment. NIST NCSTAR 1, §6.14.5, p. 149.

One separate NIST calculation extended WTC 2 fires for roughly two hours beyond the actual collapse time, with all windows broken. Its high temperatures in insulated west-side trusses describe that hypothetical extended fire. They are not pre-collapse steel temperatures and do not drive this animation. NIST NCSTAR 1-5, §6.15, pp. 180–181.

Thermal expansion and floor sag

In NIST's analysis of a composite truss-and-slab section, expansion initially pushed the columns outward. With further heating and greater sag, the floor pulled inward. This was a calculated response under particular loading, restraint and temperature conditions. It does not establish a measured outward-motion sequence at an aircraft entry opening. NIST NCSTAR 1, §6.6.4, p. 98.

The Fire & heat arrows explain that change in support reaction. The subsequent Sagging and Bowing sections combine separately sourced floor and wall evidence with illustrative transitions. They are not one coupled NIST simulation.

Photographs documented inward bowing of the towers' exterior walls before collapse. NIST used those observations alongside floor and whole-building analyses to develop its probable collapse sequences. The observed displacement and the inferred forces causing it are distinct forms of evidence.

In the model: Follow the floor and its supports.

How sagging floors can pull inward

A heated composite floor can expand against its supports, then develop tensile action as it sags. Where connections remain effective, that action can pull exterior columns inward. NIST's floor analyses also found buckling of truss web diagonals and failure of some connections. The response depended on the slab, trusses, connections and their restraints; it cannot be reduced to an identical hanging cable on every floor. NIST NCSTAR 1, §6.6.4; NCSTAR 1-6C.

Some floor connections failed while others could still transmit inward forces. NIST also identified slab-to-spandrel shear studs and diagonal struts as possible paths for that pull. Horizontal connection failure and loss of vertical seat bearing are different conditions. NIST investigation FAQ, question 17; NCSTAR 1-6D, Appendix A.

In the model: Stay with the wall as it bows.

The photographic record: measuring the bowing

The photographs measure exterior displacement, not force or an interior sag profile:

• WTC 1: At 10:23 a.m., NIST identified a maximum inward displacement of 55 ± 6 inches near floor 97 on the south face. Smoke obscured much of the wall, and image resolution limited the measurements. The summary bowing range is floors 94–100. NCSTAR 1, Table 6-2; NCSTAR 1-3, Fig. 6-15, p. 62.
• WTC 2: The earlier 9:21 a.m. observation recorded 7–9 inches near floor 80 on the east face, over floors 78–83. The later 9:52:54–9:53:04 measurement map records a maximum of 20 inches, with approximately ±1 inch measurement uncertainty. Blank cells are unmeasured. The overview uses the early snapshot; The Day uses the later map. Neither supplies a continuous measured history. NCSTAR 1, Table 6-2; NCSTAR 1-3, Fig. 6-18, p. 66.

What the floor models establish

The detailed floor displays use contour plots from named subsystem cases, including preliminary Bᵢ and Dᵢ conditions. These are distinct from the final global B and D analyses. Published plots show displacement bands with magnification; they do not provide the original solver's complete node coordinates. Interpolation, registration to the reference frame and shared movement of rendered members are reconstruction choices. NCSTAR 1-6D, Figs. 5-4 and 5-9, pp. 316 and 323.

NIST's global models used imposed pull-in forces and disconnections informed by subsystem analyses and exterior observations. The global tower model did not independently solve every detailed floor's sag and connection behavior. NCSTAR 1-6D, §2.5, pp. 36–40.

Bowing, restraint and instability

A compressive load acting through a displaced column creates additional bending, often described as a P–Δ effect. Heating, redistributed loads, inward bowing and changes in lateral restraint interacted in NIST's probable sequence. A uniform 75–90 percent reduction in buckling capacity cannot be inferred from the photographs or applied to all perimeter columns. The viewer illustrates geometry; it does not calculate stability or residual load capacity. NIST NCSTAR 1, §§6.14.6–6.14.7, pp. 150–152.

NIST's probable collapse sequences connect impact damage, subsequent fires, structural weakening and the onset of instability. Exterior photographs and videos constrain parts of those sequences; interior conditions depend substantially on reconstruction and calculation.

In the model: Inspect the beginning of upper-section movement.

The animated account depicts early rotation and descent constrained by NIST’s exterior observations. North downward placement is a conditional image estimate; South downward placement is an explicit kinematic assumption. The model does not recover a measured three-dimensional trajectory or historical clock. Source photographs and archive excerpts remain available for comparison.

WTC 1: Core weakening, redistribution and south-wall buckling

NIST places initiation at 10:28:22 a.m., about 102 minutes after impact. In its probable sequence, damaged and heated core columns weakened and shortened. The hat truss, extending through the upper structure from floors 107 to the roof, restrained that shortening and transferred load to the perimeter. The calculated core load decreased about 20 percent during this thermal stage; north/south wall loads increased about 10 percent, and east/west wall loads about 25 percent. These are model results for that stage, not measured forces. NIST NCSTAR 1, §§6.14.6 and Table 6-1, pp. 150–151 and 86.

South-side long-span floors 95–99 weakened and sagged. NIST's sequence includes failure of about 20 percent of the south-side connections on floors 97 and 98, while surviving connections transmitted inward pull. The south wall bowed, buckled and transferred loads to neighboring columns and walls as instability spread. NIST NCSTAR 1, §6.14.6.

NIST describes the section above the impact zone, near floor 98, initially acting approximately as a rigid block and tilting at least 8° south. The model pairs a selected 8° orientation with about 80 ft of common downward placement, a representative estimate admissible under a conditional archive-image fit. The camera calibration uses an uncertain Floor 98 ejection band; neither the distance nor its pairing with 8° is a NIST measurement or a synchronized video pose. The upper display moves above held lower context, separated near floor 98; this reference cut is not a measured hinge or fracture plane. Multi-angle video analysis found that the antenna’s apparent earlier descent from the north could be explained by southward upper-section tilt; it does not establish the antenna dropping independently through the roof. NCSTAR 1, p. 151; NCSTAR 1-6, §6.1, p. 153.

The same southward tilt looks different from different viewpoints. In NIST’s north-view sequence, the antenna appears to lower without a discernible east–west tilt; the west-view sequence shows the tilt toward the south. These exterior images constrain the upper-section orientation, not an independent antenna-base displacement or the timing of interior failures. Final NCSTAR 1-6, §6.1, p. 153 and Figs. 6-10–6-11, pp. 165–166.



WTC 2: Asymmetric damage and east-wall buckling

Initiation occurred at 9:58:59 a.m., about 56 minutes after impact. NIST identified the distribution of impact damage, insulation loss, persistent east-side fires and the greater loads above the lower impact zone as interacting factors. Speed or story count alone does not determine the time to collapse. NIST NCSTAR 1, pp. 148–152.

The southeast core damage shifted loads toward the east wall. Subsequent heating weakened core columns, while east-side floors 79–83 sagged. NIST's probable sequence includes failure of about one-third of the remaining east-wall floor connections on floor 83; sagging floors whose seats remained intact could still pull the wall inward. These aggregate estimates do not locate every failed seat. NIST NCSTAR 1, §6.14.7, pp. 151–152.

The east wall buckled, and instability spread toward the north and south walls. NIST reports that the upper section, near floor 82, tilted 7–8° east and 3–4° south before significant downward movement. Eastward rotation then continued to approximately 20–25°, while the southward tilt did not increase further. The model passes 7° east / 3° south and reaches a selected 20° east, holding 3° south for illustration. Its downward placement is the minimum geometric translation that prevents the bounding reference-cut corners from rising; it is not a measured fall distance or solved support response. Reference cuts, intermediate poses and timing remain display choices. NIST NCSTAR 1, p. 152; NCSTAR 1-6, p. 167 and Fig. 6-25, p. 182.

Floor connections and the initiation mechanism

NIST did not support a floor-by-floor “pancake” failure as the mechanism that initiated collapse. Its sequence includes column instability, sagging floors, some failed connections and some effective connections. Observed inward bowing supports the reconstructed interaction, but does not prove that every floor connection held or directly measure the inward forces. NIST investigation FAQ, collapse mechanism and connected-floor explanations.

NIST’s detailed investigation followed the towers from aircraft impact to collapse initiation. It did not calculate the entire subsequent collapse, complete debris motion or a calibrated floor-by-floor descent. The viewer also depicts a bounded interval of early visible rotation and descent from the photographic record. Its estimated placement, reference cuts and authored timing do not locate actual fracture planes or extend NIST’s calculations into a full collapse simulation. NIST NCSTAR 1, §6.2, p. 82.

NIST’s report continues beyond the computational endpoint with observations and an explanation of subsequent movement. Section 6.14.4 describes the initial movement continuing, the failure sequence spreading downward and dust obscuring the view. That discussion is distinct from calculating the complete propagation. NCSTAR 1, §6.14.4, p. 146.

Why the structure below could not arrest collapse

NIST separately presented a simplified capacity check for the connections supporting an intact floor. It estimated capacity for about 11 additional floors under gradually applied loading, or about 6 under suddenly applied loading. The upper sections exceeded that demand: approximately 12 floors in WTC 1 and 29 in WTC 2. This supports NIST's explanation for why collapse was not arrested, but it is not a complete dynamic simulation or a measured sequence of individual connection failures. NIST investigation FAQ, question 18.

Separate energy analyses

Bažant and Zhou's simplified analysis compared gravitational energy with an optimistic estimate of plastic energy absorption by underlying columns. Their 8.4 ratio refers to the assumptions of their two-floor-drop energy calculation, not a measured one-story fall or a result from NIST's structural model. Later progressive-collapse papers use additional idealizations about accumulating mass, resistance and energy loss. These provide separate mechanical analyses; their assumptions are not encoded in this viewer. Bažant and Zhou, 2002, Eq. 3; Bažant and Verdure, 2007.

What the timing evidence establishes

NIST's FAQ distinguishes the arrival of the first exterior panels at the ground—approximately 11 seconds for WTC 1 and 9 seconds for WTC 2—from completion of the whole collapse. Core remnants remained standing for approximately 15–25 seconds after initiation. Dust obscured much of the interior movement, so the first debris arrival does not establish a complete tower descent time or a continuous acceleration history. NIST investigation FAQ, question 31.

The model's playback progress is educational. It supplies no prediction of collapse duration, speed, acceleration, contact forces or rubble distribution.

Beyond the structural account

These buildings held workplaces, visitors and communities. Their destruction killed people inside them and across the surrounding site. The engineering record is one part of that history. The 9/11 Memorial & Museum preserves the names of those killed, alongside accounts of survivors, responders and the recovery. Remember the people.

Primary sources

NIST NCSTAR 1 — Final Report on the Collapse of the World Trade Center Towers (September 2005). The overview report of NIST's federal building and fire safety investigation, with more than 40 companion volumes (NCSTAR 1-1 through 1-8 series) covering reconstruction of the impacts (1-2), steel analysis (1-3), fire reconstruction (1-5), structural response and probable collapse sequences (1-6), and more.

• Mirror PDF: https://humanfactors101.com/wp-content/uploads/2026/04/Final-report-on-the-collapse-of-the-World-Trade-Center-towers-NIST-2005.pdf
• NIST archive of the full report series: https://www.nist.gov/world-trade-center-investigation (formerly wtc.nist.gov)

NIST, "Questions and Answers about the NIST WTC Towers Investigation" (created September 2011, updated 2021/2022). NIST's official FAQ addressing 34 specific questions about the towers investigation, including the collapse times, the molten metal, thermite, the seismic record, and the rejection of the pancake theory. Cited throughout as "NIST FAQ, Q#."

• https://www.nist.gov/disaster-failure-studies/faqs-nist-wtc-towers-investigation

NIST NCSTAR 1A — Final Report on the Collapse of World Trade Center Building 7 (November 2008). The separate investigation of WTC 7, identifying fire-induced thermal expansion and the buckling of Column 79 as the initiating mechanism, and acknowledging the 2.25-second interval of free-fall acceleration.

• https://www.nist.gov/publications/final-report-collapse-world-trade-center-building-7-federal-building-and-fire-safety-0

9/11 Commission Report, Chapter 9 ("Heroism and Horror," 2004). Timeline of the impacts, the emergency response, and the collapses. Note: the Commission's mandate was the attacks and the response, not structural engineering.

• https://en.wikisource.org/wiki/9/11_Commission_Report/Chapter_9

*Bažant, Z. P., & Zhou, Y. (2002). "Why Did the World Trade Center Collapse? — Simple Analysis." Journal of Engineering Mechanics, 128(1), 2–6. The independent energy analysis, submitted within days of the attacks, showing the kinetic energy of a falling upper block exceeded the underlying story's energy-absorption capacity by roughly an order of magnitude. (Companion discussion: Bažant & Zhou, 2002b, J. Eng. Mech.* 128(3).)

• https://doi.org/10.1061/(ASCE)0733-9399(2002)128:1(2)

*Bažant, Z. P., & Verdure, M. (2007). "Mechanics of Progressive Collapse: Learning from World Trade Center and Building Demolitions." Journal of Engineering Mechanics, 133(3), 308–319.* Refines the 2002 energy analysis into a full model of collapse propagation; contains a concise review of the collapse causes and the key energy figures cited here (8.4× energy margin, ~10.8 s modeled collapse duration vs. 9.21 s free fall).

• https://www.civil.northwestern.edu/people/bazant/PDFs/Papers/466.pdf

FEMA 403 — World Trade Center Building Performance Study: Data Collection, Preliminary Observations, and Recommendations (May 2002). The preliminary FEMA/ASCE assessment that preceded the NIST investigation; source of the early "pancake" hypothesis later superseded by NIST's findings.

• https://www.fema.gov/pdf/library/fema403_execsumm.pdf (executive summary; full report via FEMA's library)

Supporting NIST-authored summary

*Banovic, S. W., Foecke, T., Luecke, W. E., McColskey, J. D., McCowan, C. N., Siewert, T. A., & Gayle, F. W. (2007). "The Role of Metallurgy in the NIST Investigation of the World Trade Center Towers Collapse." JOM (Journal of the Minerals, Metals & Materials Society), 59(11), 22–29.* A NIST-authored summary of the steel recovery and testing program (236 cataloged steel elements), the steel strength–temperature data, and the probable collapse sequences for both towers — including the column damage counts, bowing measurements, and fire-migration timings cited in Chapters 2–5.

• https://www.tms.org/pubs/journals/jom/0711/banovic-0711.html

A note on citation practice

Inline citations in Chapters 0–7 identify the source document and, where the document itself provides one, the section (e.g., NIST NCSTAR 1, §6.14.4) or figure (e.g., NIST NCSTAR 1, Fig. 6-36). Readers are encouraged to verify each claim against the full reports; the NCSTAR series is in the public domain.

This document describes the evidence hierarchy, modeling choices and verification used in the World Trade Center reconstruction. Source consistency is distinct from validation of a structural simulation.

The project’s standard is to distinguish observed evidence, calculated results and illustrative geometry, and to record unresolved assumptions alongside the sources.

1. Evidence Hierarchy

Read the model and account using these three evidence tiers. A citation does not make every coordinate in the associated mesh a measured quantity:

Tier 1: Directly Observed Physical Evidence

Physical artifacts, verified photographic records, and instrumental measurements that exist independently of post-disaster numerical modeling.

• Pre-collapse high-resolution photography & video: Photographic records from dozens of vantage points establishing entry/exit gash geometry (NIST NCSTAR 1-2), visible flame and smoke migration across floor windows (NCSTAR 1-5A), window-by-window exterior column inward bowing measurements (NCSTAR 1-3, Fig. 6-18), and synchronized multi-angle video of collapse initiation (NCSTAR 1-6 §6.1).
• Physical steel recovery and metallurgical analysis: NIST cataloged 236 recovered steel elements, including 90 perimeter-panel samples and 55 core-column samples. Only a subset had known original locations. Recovery records, photographs and material examinations constrain some damage, but recovered failure modes alone do not establish when each connection failed (NIST NCSTAR 1-3 and NCSTAR 1-3C).
• Original architectural and structural contract drawings: Official Port Authority of New York & New Jersey (PANYNJ) contract sets, Leslie E. Robertson Associates (LERA) tenant and alteration records, Pacific Construction (PAC1) Tower A plans, and Foundation drawings R-60 through R-77 detailing the 3,100-foot perimeter slurry wall and bedrock tieback arrays.
• Instrumental records: Lamont-Doherty Earth Observatory seismic recordings of aircraft impacts and collapse ground waves; FAA radar transponder tracking.

Tier 2: Calibrated Engineering Calculations and Forensic Simulations

Engineering analyses reconstruct conditions that could not be directly observed. Agreement with exterior observations constrains their assumptions; it does not turn calculated interior damage into a measured survey.

• Impact dynamics (LS-DYNA): NIST calculated aircraft fragmentation and structural damage in separate base, less severe and more severe impact cases. Estimates of insulation dislodgment used that impact work with additional assumptions; the structural damage footprints are not insulation-loss maps (NCSTAR 1-2; NCSTAR 1 §6.14.5).
• Fire propagation (NIST FDS): Computational fluid dynamics (CFD) modeling of combustible fuel loads, ventilation-limited fire spread, and upper-layer compartment gas temperatures exceeding 1,000 °C (NIST NCSTAR 1-5).
• Thermal-structural response (ANSYS): NIST separately calculated heat transfer and structural response. The viewer samples named subsystem displacement contours, including preliminary Bᵢ and Dᵢ cases; those are distinct from final global B and D. Global models used imposed pull-in forces and disconnections informed by subsystem work and observations. No universal sag maximum or force per connection applies across these studies (NCSTAR 1-6C; NCSTAR 1-6D §2.5 and Chapter 5).
• Progressive collapse limit states: Academic conservation of momentum and energy analyses (Bažant & Zhou, 2002; Bažant & Verdure, 2007) evaluating kinetic energy versus plastic energy dissipation capacity during progressive downward failure.

Tier 3: Geometric, Interpolative, and Illustrative Representations

Visual modeling decisions made for interactive 3D WebGL rendering, clearly identified so they are not mistaken for historical surveying.

• Representative member schedules: Trusses outside the primary impact zones repeat standardized architectural panel modules (NIST NCSTAR 1-2A Appendix G typologies) rather than unique tenant configurations.
• Bounded early motion: The North model reaches a representative 8° south / 80 ft downward pose. The angle follows NIST’s reported lower-bound orientation; placement is a conditional archive-image estimate with uncertain camera and ejection-band calibration, not a synchronized measurement. The South model passes 7° east / 3° south and continues to a selected 20° east, holding 3° south. South downward placement is a minimum non-rising reference-cut convention, not a measured fall distance. Reference cuts near floors 98 and 82, nominal pivots, intermediate paths and timing are illustrative. The model does not solve instability, support resistance or subsequent collapse propagation.
• Display cutaways and annotations: Colors, arrows and transparent layers clarify source categories and structural relationships. They do not encode computed stress, temperature or connection capacity.

2. Primary Source Datums and Geometry Verification

Floor Elevations and Story Heights

The model uses the LERA story-height schedule (`floor_heights.txt`), which references Port Authority drawings A-AB-201 and A-AB-301, for both towers. Typical tenant stories are 12 feet floor to floor; special floors have different spacing. The schedule establishes relative story elevations, with floor 1 (concourse) as the viewer’s vertical origin. It is not a surveyed street-grade or sea-level datum. Story spacing accounts for:

• 12-foot standard tenant office stories.
• Two-story mechanical equipment rooms (Floors 7/8, 41/42, 75/76, 108/109) with heavy perimeter double-spandrel girders.
• The lower concourse, plaza and storage levels, each with its own scheduled height.
• Distinct upper-floor, sloping-roof, roof-panel and penthouse references. NIST’s reported tower heights of 1,368 and 1,362 feet must not be relabeled as drawing top-of-steel elevations. Local roof and penthouse surfaces retain their own source datums and unresolved interfaces. NCSTAR 1-1, p. 7.

Substructure and Slurry Wall ("The Bathtub")

The subterranean enclosure is modeled directly from foundation contract sheets R-60 through R-77:

• Perimeter Slurry Wall: 3,100 linear feet of 3-foot-thick reinforced concrete wall descending 65 to 80 feet to solid schist bedrock.
• Tieback Array: 1,548 illustrative multi-tier rock tendons representing the temporary drill anchors installed at 10-foot nominal spacing through Hudson River silt into bedrock sockets prior to floor slab installation.
• Sublevel Datums: Service level B1 (EL 294′) through lower pump room B6 (EL 242′), including PATH railway track tunnels and platform elevations from drawing R-65.

3. Forensic Case Studies in Reconciling Contradictions

Case Study A: Antenna Motion and Upper-Section Tilt

An apparent early antenna descent in a view from the north is not direct proof that the core dropped independently. NIST compared other views and explained that appearance through southward tilt of the upper building section. Its probable sequence includes core weakening and shortening, but the footage does not establish an antenna plunging through the roof. The viewer must not turn that inference into a measured independent core-drop trajectory. NCSTAR 1-6, §6.1, p. 153; NCSTAR 1, pp. 150–151.

Case Study B: Connected Floors and Wall Bowing

NIST did not support successive floor detachment as the mechanism that initiated collapse. Its probable sequence includes sagging floors, some failed connections, some remaining paths for inward pull, and column instability. Observed inward displacement reached 55 ± 6 inches on WTC 1’s south face. WTC 2’s east-face record distinguishes an early 7–9-inch observation from a later 20-inch maximum. These observations measure displacement, not force or the condition of every seat. NIST investigation FAQ, collapse mechanism and connected-floor explanations; NCSTAR 1-3, Figs. 6-15 and 6-18.

Recovered connections show several failure modes. Their final condition may include impact, fire, collapse and recovery damage; it does not by itself prove a pre-collapse tensile failure at each location (NCSTAR 1-3C, Chapter 3).

4. Source Review and Geometry Verification

The September 12, 2026 damage audit re-extracted all 24 tower/case/floor records from the final NCSTAR 1-2 floor-damage figures. Eight source pages contain 81 distinct solid or dotted annotation paths. The review preserves the source PDF hash, raw subpaths, page registration and visual overlays; disconnected shapes remain separate. Nominal core registration and small orthogonalization adjustments introduce roughly two feet of graphic uncertainty. This is a working tolerance for the diagram fit, not a NIST measurement-error estimate.

Solid outlines represent severe structural damage; dotted outlines represent openings. NIST excluded both from its subsequent structural models. The viewer leaves severe regions amber for inspection, which does not assign them surviving load capacity. The source figures do not identify every damaged truss or fracture elevation. NCSTAR 1-2, pp. 187–189, 210–211, 232–234 and 255–256.

Verification compares core table records with the source PDF, checks floor openings against actual slab/deck intersections, checks retained pieces and source identities, and exercises damage-case changes and view restoration. Such checks establish consistency between sources, data and rendered geometry; they do not validate a finite-element response or prove complete historical accuracy. AI assistance is part of the authoring process and is not independent evidence.

5. Explicit Limitations and Boundaries

1. Not an Aerodynamic or Hydrodynamic Solver: The browser model renders historical and calculated forensic states; it does not compute real-time structural stress tensors, wind turbulence, or fluid dynamics.
2. Post-Impact Interior Obscurement: Entry-face geometry combines FEMA diagrams and localized NIST observations; obscured areas, approximate registration and uncertain fracture depths remain. Interior impact damage is calculated. Source-traced retained pieces do not establish complete three-dimensional fracture geometry or load continuity.
3. Tenant Alterations: The model reflects original design and structural alteration drawings on file with the Port Authority; minor non-structural architectural partition shifts made by individual commercial tenants between 1973 and 2001 are not comprehensively reconstructed.
4. Non-Sensational Educational Mandate: The explainer is designed strictly as an engineering record. All representations prioritize sober structural physics, educational utility, and respect for historical fact.
5. Combining separate response studies: North Tower illustration mode adds a sampled subsystem sag field to representative impact deformations without a verified common nodal reference. South Tower mode registers its sampled total floor displacement to the impact baseline, but that registration also remains approximate. Neither animation is a single coupled NIST simulation.