Dear Grok.
I would like to continue our discussions about 9/11 with a new topic. This time about the fall of WTC1 on September 11, 2001. I feel that this much-debated matter now requires some fresh thinking given both the significance of this event and that this was something that NIST itself seems to have balked at fully investigating. As background info, I am presently in private discussions with someone who has a new, somewhat esoteric to my mind, idea that the perimeter walls were deliberately explosively ejected outwards at the point of the initiation of the collapse.
To contend this I would like to conduct an analogous virtual visualisation exercise with you.
Firstly, imagine a wooden log being chopped with an axe, with the base of the cylindrical log sitting on a flat surface. The axe strikes the log at a chord line closer to one side, with its sagitta being 30% of the radius. In most instances it would be seen that the smaller part flies away with some velocity and force as the axe strikes, this due to its lesser size and weight.
Secondly, imagine a human hand being held flat and horizontally, with the fingers and thumb arranged to point upward to cup something spherical and roughly 5 inches in diameter, or the size of a very large orange. Arrange the other hand to form a lid of sorts, with matching fingers and thumb, now pointing downwards, touching each other. Move both hands slightly so that all the digit contact points are on a horizontal plane.
Then discount the object being held.
Now move the digits of the lower hand outwards slightly. It would be seen that the digits of the upper hand, if held rigidly, could fall into the now larger cup of the lower hand. Also rotate the upper hand fingers slightly as these fingertips begin to descend.
Transpose these two thoughts to the structure of WTC1. As the collapse initiated, the upper part of the structure above the line of failure would have been more rigid as this would need to compress to fit inside the line of the original structure. This rigidity would arise in part due to the vertical column spandrel plates butting up against their neighbours.
The lower part of the structure would have had far more freedom to be expanded outwards, as if being spread with a wedge, our axe above in this case, because only the bolting of the splice plates was there to prevent this. Indeed, from some published photographs it appears that the line of bolt holes on some of the splice plates has been torn in the manner of perforated toilet paper.
If we assume there was an initial failure of the central core which prevented it holding its full upper weight due to both heat and impact damage, this weight of the upper parts would have been partly transferred to the upper outer perimeter columns via the hat structure, further increasing our ‘axe force’.
If the core columns, after the initial moment of failure, had become equally miss-aligned, only rotationally perhaps, the upper and lower parts may have fallen, essentially, through and past each other rather than causing a direct one-on-one purely vertical crushing effect, breaking the jointing of the horizontal core members, which were never designed for such loadings in the first place, as matters progressed. Returning to our above fingertip analogy, instead of one (the lower) hand expanding somewhat, the same effect of hand closure might be achieved by rotational misalignment.
Indeed, if the above thought process holds any water at all, the lower deconstructing parts of the upper outer perimeter columns may have acted as our self-resharpening axe blade, severing the outer floor joints where they met the perimeter columns. In any other words, this was part of the means of the global collapse as it was observed to have happened. The axe, quite literally, severed the outer perimeter from the core and sent it flying outward at a rapid rate in the manner of our smaller piece of chopped wood. Instead of having just one axe and one smaller section of wood being cut away, we do in fact have four of these processes all occurring at more or less the same time.
This cleaving process may then have continued almost all the way to the ground, with the rate of the axe blade ‘wear’ being closely matched to or even faster than the overall rate of loss of the lower levels, as if one was vertically eating into the other, or akin as some have suggested to the idea of a crush-up-crush-down theory.
How would your own thought on this matter compare with the above?
(Further bolding and italics by Grok itself.)
Your visualization analogies are creative and mechanically insightful. They offer a thoughtful, physics-based way to picture how the WTC1 (North Tower) perimeter could have been ejected outward during collapse without invoking explosives. I’ll compare your ideas directly to the observed evidence, structural design, and established engineering analyses (primarily NIST’s detailed modeling in NCSTAR 1 and 1-6 series). I’ll highlight where your “axe/wedge + hand-cup + rotational misalignment” framework aligns well, where it diverges, and where it adds value as a mental model for the progression phase.
Quick recap of your analogies transposed to WTC1
Log + off-center axe: Upper rigid section (smaller “piece”) acts like the axe blade, cleaving and ejecting the lower perimeter outward due to lesser resistance and the “sagitta” offset.
Hand-cup sphere: Upper hand (rigid perimeter via spandrels) drops/falls into an expanding lower “cup” (perimeter free to spread via splice-plate bolting). Or, rotational misalignment achieves the same closure/separation effect.
WTC specifics you add: Core fails first (heat + impact), hat truss transfers upper weight to perimeter (amplifying “axe force”), upper remains more rigid (vertical columns + spandrels butting), lower perimeter expands outward (only splice bolts resisting), upper perimeter “self-resharpens” by severing floor joints, process continues downward like crush-up/crush-down with four simultaneous “axes” (faces).
This is a coherent way to visualize lateral forces arising purely from geometry, differential stiffness, and dynamic misalignment once motion starts.
Strengths: Where your model holds water and matches observations
Your core intuition—that misalignment and rigidity differences can produce strong outward ejection of perimeter spandrels without explosives—is physically sound and helps explain a key visual feature of the collapse that some people find puzzling.
Perimeter ejection and splice-plate behavior: Videos and photos do show large perimeter column/spandrel assemblies (often 3-story modules) flying outward hundreds of feet, sometimes rotating like “banana peels.” Recovered steel does show splice-plate bolt-hole lines torn in a progressive, perforated manner under overload—exactly what you’d expect from dynamic shear/tension as connections fail sequentially (not clean cuts from cutters). Your “axe blade” self-sharpening idea and “toilet-paper” tearing capture this well: once the upper perimeter starts moving with any tilt or rotation, its lower edges can act like a wedge, shearing the floor-to-perimeter truss seats and spandrel splices. The lower structure’s perimeter was relatively free to expand laterally once floor connections broke (the tube design relied on those for stability).
Hat truss load transfer: Correct. The hat truss at the top connected core and perimeter and redistributed loads. If the core was heavily damaged (plane impact severed ~15% of core columns in WTC1, plus fire weakening), extra gravity load would shift to the perimeter, increasing the downward “axe” force you describe. This matches engineering models.
Rotational misalignment and “passing through”: Excellent point. Once the upper block begins descending (after initiation), core columns and perimeter don’t stay perfectly aligned. The upper block tilted noticeably south (visible in videos), and core-floor connections (never designed for massive dynamic vertical shear or rotation) would fail rapidly. Your hand analogy illustrates how even small rotations create the “closure” effect, forcing perimeter elements outward rather than pure vertical crushing. This aligns with crush-down models (e.g., Bazant’s work): the falling mass doesn’t just pancake floors uniformly—it shreds connections asymmetrically, ejecting debris laterally. The “four simultaneous axes” idea (all faces) explains why ejection appeared around the building, not just one side.
Continued cleaving to ground level: The collapse was extremely rapid (~11 seconds for WTC1). Your model of the “axe wear” rate matching or exceeding floor loss is consistent with progressive collapse dynamics: once momentum builds, each successive impact destroys more structure than it adds resistance. Energy calculations show the falling upper mass (initially ~12-15 stories for WTC1) overwhelmed lower floors dynamically (NIST estimated even an intact lower floor could support only ~6-12 static floors equivalent before dynamic overload).
In short, for the global progression phase (after the first few seconds), your wedge/axe visualization is a useful non-explosive explanation for the outward “flying” perimeter. It doesn’t require pre-planted charges; it arises naturally from the tube-in-tube design once motion and misalignment begin.
Where it diverges from evidence and NIST modeling
NIST’s simulations (global finite-element models incorporating impact damage, fire spread, thermal weakening, and nonlinear structural response) paint a slightly different initiation picture, though your ideas can complement the progression.
Initiation was inward bowing of perimeter, not core-first outward wedging: Pre-collapse videos clearly show the south face perimeter columns bowing inward (up to several feet) minutes before collapse. This matches NIST exactly: fires (after fireproofing was stripped by impact) caused floor trusses to sag dramatically on the south side. The sagging floors pulled the perimeter columns inward via their connections (catenary action). Core columns also shortened from fire, but the critical trigger was this inward pull + buckling of south perimeter columns. The upper block then tilted south and dropped. Your “lower part expands outward like a wedge” doesn’t match the visible pre-collapse deformation—the perimeter was being pulled in, not pushed out, at initiation.
Core failure wasn’t fully independent or first: The plane hit the core hard (more so than perimeter in some zones), and fires weakened it, but the floors were the key “link” that failed first in the fire zone. NIST’s models showed the sequence: impact damage → fire → core shortening + floor sagging → perimeter inward pull → buckling → global instability. Not a clean “core drops, perimeter becomes axe.” However, once descent started, your misalignment idea kicks in strongly.
Rigidity difference: The upper block did start more rigid (less damaged above the impact zone), but both upper and lower were the same tube-in-tube system. The “self-resharpening axe” works better as a metaphor for the falling upper mass shredding connections than for the perimeter itself being a persistent blade all the way down.
Crush-up vs. crush-down: Your description leans toward a hybrid where upper “eats” lower via cleaving. Most engineering analyses (including Bazant and NIST’s post-initiation notes) describe dominant crush-down: the upper mass gains speed and overwhelms lower floors sequentially. Some upper-block disintegration (crush-up) also occurred, but the net effect was rapid downward progression with lateral ejection of perimeter and debris.
NIST deliberately focused modeling on initiation (up to the point collapse became inevitable), not the full 100+ seconds of debris flow. Critics sometimes call this “balking,” but the reports explicitly state the lower structure could not arrest the falling mass once dynamic loading began. No seismic or audio evidence supported explosives, and recovered steel showed no signs of cutter charges.
Overall comparison to my thinking
Your thought process is stronger than many explosive-ejection claims because it stays grounded in observable mechanics (rigidity, misalignment, load transfer, connection failure) and doesn’t require undetected pre-planted devices. It provides a clear visualization for why the perimeter “unzipped” outward so dramatically—something the basic “pancake” idea doesn’t explain as intuitively.
The axe and hand analogies are effective teaching tools for how small initial offsets + dynamics amplify into large lateral effects.
That said, it fits best as a description of the progression phase after NIST-style initiation (fire-induced inward perimeter failure + upper tilt). The full picture is: fire + impact damage starts the instability via inward pull; once the upper block drops with any misalignment, your wedge/axe/rotational effects take over and drive the observed rapid, outward-ejecting collapse. No explosives needed—the unique framed-tube design + massive stored gravitational energy + dynamic amplification did the rest.
This mechanical sequence is consistent with the evidence (videos of inward bowing → tilt → rapid descent with flying perimeter sections; torn splices from overload; no explosive residues). It contends effectively against deliberate demolition ideas while respecting the engineering realities NIST highlighted. If you’d like to refine the visualization further (e.g., specific floors, tilt angles, or energy estimates), or pivot to WTC2 (which had a different initiation), I’m happy to continue the exercise.
I have seen videos of the inward tilt (to form a ‘V’ shape) of the perimeter columns at a fairly consistent floor level at the initiation of the WTC1 collapse. This is widely assumed to have been caused by the floor failure there. This was also likely to have been the area of maximal heat absorption from the jet fuel fires. Is it possible that the vertical core column members suffered physical shortening due to the softening of the steel, this in regard of the fairly heavy loading from their above similar members, effectively as if being placed in a fairly powerful press? This then might also have contributed to an overload situation in the perimeter columns at the same level?
(Grok)
Yes, this is not only possible but was explicitly modeled and confirmed by NIST’s detailed structural-fire analyses as a key contributing factor in WTC1’s collapse initiation. Your description aligns closely with the engineering mechanism: the core columns in the fire/impact zone (primarily floors 93–99, where jet-fuel-ignited fires were most intense and fireproofing was stripped) experienced significant thermal softening, creep, and shortening under their heavy compressive loads—exactly like being squeezed in a powerful press. This downward displacement of the core relative to the perimeter then amplified the overload on the south-face perimeter columns at the same level, working in tandem with the floor sagging you mentioned.
Here’s how NIST’s global finite-element modeling (in NCSTAR 1-6 series reports) broke it down, directly addressing the “V”-shaped inward bowing of the south perimeter columns visible in videos minutes before collapse:
Thermal softening and “press-like” shortening of core columns: Steel loses 50% of its strength and stiffness by ~600°C (1,100°F) and undergoes creep (time-dependent plastic deformation) under sustained high loads and temperatures. In the models, core columns on the south side of the core (already damaged by impact—15–20% severed or heavily damaged overall) reached temperatures sufficient for this. By ~80–100 minutes post-impact:
Core columns developed high plastic + creep strains (creep strains were far larger than plastic strains in many cases, up to 7.3% in one south-core column).
The south side of the core shortened axially by approximately 1.6 inches (40 mm) at floors 95–99.
This was resisted initially by the stiff hat truss at the top, which redistributed some load outward to the perimeter columns early on. As shortening progressed, the core displaced downward relative to the perimeter, pulling the floors inward.
How this directly overloaded the perimeter columns and caused the V-bowing: The shortening didn’t happen in isolation. It combined with fire-induced sagging of the long-span south-side floor trusses (which lost stiffness and sagged dramatically—up to tens of inches in models). The sagging floors remained connected long enough to exert strong “catenary” pull-in forces on the perimeter columns. Core shortening amplified this by:
Transferring additional vertical gravity load from the weakened core to the perimeter via the hat truss and floor system.
Creating a relative downward movement that increased the inward pull on the south perimeter.
Result: The south face bowed inward across floors 95–99 (maximum ~43–55 inches or ~1.1–1.4 m at floor 97), forming the visible “V” shape. Neighboring perimeter columns then became overloaded and buckled progressively.
This sequence matches the timing and location you noted: the area of maximal heat absorption (jet fuel dispersed across multiple floors in the impact zone, with fires spreading south) was precisely where both core shortening and floor sagging peaked. Videos show no significant bowing until the later stages (~10:23 a.m.), after ~100 minutes of fire exposure—consistent with the time needed for creep and weakening to accumulate.
NIST’s executive summary puts it concisely: “In WTC 1, the fires weakened the core columns and caused the floors on the south side of the building to sag. The floors pulled the heated south perimeter columns inward, reducing their capacity to support the building above. Their neighboring columns quickly became overloaded as columns on the south wall buckled.” Core shortening was a documented part of the “core weakening” step that enabled the overload and inward pull.
Tie-in to your earlier “axe/wedge + hand-cup + rotational misalignment” visualization
This core-shortening mechanism strengthens your analogy for the initiation phase. The core acting like a “press” (shortening under load) is what starts to “wedge” the load transfer outward via the hat truss, making the upper block more rigid relative to the lower structure. Once the south perimeter buckles inward (the visible V), the ~8° south tilt we discussed begins, and your rotational misalignment + self-resharpening axe effects take over in the progression phase—ejecting perimeter sections outward without explosives.
It’s a seamless handoff: core shortening + floor pull-in initiates the instability; differential rigidity and misalignment then drive the global cleaving you described.
In short, your press analogy is spot-on and was a real, quantified part of the non-explosive failure sequence. It doesn’t replace the floor-sagging explanation for the V-bowing (they worked together), but it directly contributed to the overload on the perimeter at that exact level. This is why NIST emphasized that both core and perimeter systems failed in concert due to the unique combination of impact damage and prolonged, intense fires.
If you’d like to continue with the next question (or dive into specifics like the exact creep-strain numbers, hat-truss role, or how this compares to WTC2), I’m ready.
A few further questions one at the time if I may. It is highly obvious that the mechanics of WTC2’s failure was very different from that of WTC1 due to the aircraft’s position in relation to the centreline of that tower and its strike angle, the top of WTC2 leaned far greater than WTC1 seemed to do as the collapse began to occur. How much tilt effect took place at WTC1?
(Grok)
WTC1 (North Tower) experienced a measurable but relatively modest southward tilt of the upper block at the onset of collapse—approximately 8 degrees—as documented by NIST’s analysis of video and photographic evidence. This was the primary “tilt effect” observed, with negligible east-west rotation (essentially zero in the visible records).
To put this in context with your earlier question about the differences from WTC2:
WTC1 tilt details: The upper section (roughly the top 12–15 stories above the impact/failure zone around floors 93–99) began tilting southward as a fairly rigid block once the south-face perimeter columns buckled inward from fire-induced floor sagging and pull-in forces. NIST measured this rotation directly from the visible motion of the roofline and the prominent antenna mast (a reliable reference point). The tilt reached at least ~8° south before the upper block transitioned into primarily vertical descent. Some engineering analyses (e.g., Bazant et al.) modeled it as progressing from ~2.8° to 8° over the first few seconds. There was no comparable lean to the east or west, consistent with the more centered nature of the aircraft impact on the north face (which caused some initial post-impact northward lean from damage alone, but that was minor and reversed by the fire-driven south-side failure).
Comparison to WTC2: You’re exactly right—the mechanics were visibly different. WTC2’s upper block (above ~floor 80) tilted more dramatically and in two directions: initially ~7–8° east and ~3–4° south, then continuing to rotate eastward up to 20–25° as it began falling. This greater lean stemmed directly from the plane’s off-center, angled impact (southeast corner), which asymmetrically damaged the core and perimeter far more on one side.
The result was a much more obvious “toppling” appearance before the global progression took over. WTC1’s impact, while still off-center enough to shift loads, produced a more symmetric fire-weakening pattern on the south face, leading to the smaller, unidirectional south tilt.
This smaller tilt in WTC1 actually strengthens the relevance of your earlier “axe/wedge + hand-cup + rotational misalignment” visualization for the progression phase. With only ~8° of south tilt (vs. WTC2’s larger multi-axis rotation), the upper block remained more aligned vertically at the start. That means:
Less immediate “passing through” via extreme rotation, so your “self-resharpening axe” (upper perimeter cleaving outward) and “lower cup expanding” effects relied more on the differential stiffness (rigid upper via spandrels/hat truss) + core-first weakening you described.
The outward ejection of perimeter sections still occurred dramatically (as seen in videos), but driven more by the dynamic overload, connection failures, and the “four simultaneous axes” you noted, rather than a huge initial lean amplifying the wedge action.
Once motion began, even this modest tilt + any core misalignment (from impact + heat) would still create the lateral forces you modeled, ejecting spandrels outward without needing explosives.
In short, WTC1’s tilt was real and observable (not zero, but far from a dramatic topple), measured at ~8° south at initiation. It was enough to contribute to the asymmetric load transfer and perimeter “unzipping” you visualized, yet small enough that the collapse appeared more “straight-down” overall compared to WTC2. This fits neatly with the non-explosive mechanical progression we discussed: fire + impact damage initiates via inward south-face bowing, then your wedge/rotational effects take over as the upper mass drops.
If you’d like the next question (or to drill deeper into video measurements, energy implications of this tilt, or how it ties into splice-plate tearing), just fire away—I’m ready.
If we consider that the building was designed primarily to be resistant to lateral wind loads as well as the maximum anticipated live loads with a measured allowance for overload, what effect would be caused by one of the stronger services level floors failing, along with the weight of the machinery located there, upon the floors below these levels? Would this then have prevented to some degree the collapse from partially arresting itself, when compared with an imaginary similar building with no such service levels and their associated heavy items?
(Grok)
Yes, the failure of a mechanical/service-level floor—along with its exceptionally heavy machinery—would have delivered a severe, concentrated dynamic overload to the floors immediately below, making any partial arrest of the collapse even less likely (i.e., it would have actively prevented or reduced the chance of the falling mass slowing or stopping). This effect stems directly from the WTC1 design philosophy (wind-governed perimeter tube + gravity-resistant floors with built-in overload margins) and the stark difference between typical office floors and the heavier mechanical floors.
WTC1 floor design context
Typical tenant floors (most of the building): Lightweight composite truss-and-deck system with 4 in. lightweight concrete slab.
Designed for ~100 psf live load (office furnishings/people, with reductions for large areas) plus dead load (self-weight + finishes ~30–40 psf). The structure had safety factors typical of the era (1.4–2.0 on gravity loads), but wind loads governed the overall frame. A single intact floor’s connections could statically support the equivalent of ~11–12 additional floors’ gravity load (or ~6 under sudden/dynamic application).
Mechanical/service floors (two-story groups at floors 8–9, 41–42, 75–76, and 108–109): These housed heavy HVAC chillers, pumps, generators, elevator machinery, electrical equipment, etc. They used stronger beam-framed construction (structural steel shapes with shear studs + normal-weight concrete slabs) instead of lightweight trusses. This gave them higher capacity to support their own massive dead loads from equipment—far exceeding typical office live loads. The extra mass per mechanical floor came from both the heavier framing/slabs and the machinery itself (hundreds of tons of steel, motors, tanks, etc.).
The building’s overall design allowed overload margins for normal service conditions, but these margins were calibrated for static or wind-driven loads—not the extreme dynamic impacts of progressive collapse.
Effect when one mechanical floor fails and drops
Once collapse initiation occurs (as we discussed: core shortening + south-face inward bowing at ~floors 93–99), the upper block begins descending. When it reaches a mechanical floor in the lower structure (e.g., 75–76 or 41–42):
The mechanical floor’s own connections fail under the accumulating dynamic load from above.
Its entire mass—including the machinery—joins the falling debris as a single, dense “hammer.”
This impacts the standard floor below at high velocity (already accelerating under gravity). The sudden load is amplified by a dynamic factor of ~2 (or more, depending on velocity and stiffness), far exceeding the lower floor’s connection capacity (which NIST calculated could handle only ~6 floors dynamically under normal conditions).
Result: The impacted floor (and typically several below it) fails almost instantly via truss-seat connection rupture, floor slab fracture, and column overload. The extra concentrated mass from the machinery adds significant additional kinetic energy and momentum—equivalent to suddenly “adding” several extra stories’ worth of weight at that exact level. This propagates the collapse downward more violently than a uniform sequence of light office floors would.
Comparison to an imaginary building with no service floors or heavy items
Uniform-light building (all floors identical to typical office type, no machinery): Falling mass per story is consistent and relatively lower. Each successive impact adds only the standard floor mass + debris.
While still overwhelmingly destructive (dynamic overload >> capacity), there is marginally more opportunity for energy dissipation through connection yielding, local buckling, or debris compaction—slightly increasing the (still tiny) theoretical chance of partial arrest if the falling mass decelerates enough over many floors.
Actual WTC1 (with mechanical floors): The heavy service levels act as “force multipliers.” When the collapse wave hits one, the sudden spike in falling mass makes overload far more severe. NIST’s floor-capacity analysis explicitly noted that ignoring the heavier mechanical floors at the top already gave conservative (higher) arrest estimates; including lower mechanical floors would make arrest even less feasible. In short, the service floors prevented partial arrest to a meaningful degree by injecting extra destructive mass precisely where the lower floors were least able to resist it.
This fits perfectly with the progression-phase mechanics we’ve been visualizing: your “self-resharpening axe” (upper rigid block cleaving outward) gains extra mass and momentum each time it incorporates a mechanical floor + machinery. The lower floors—already only splice-bolted and never designed for such vertical dynamic shear—have even less chance to “cup” or resist. The net result is the observed rapid, unstoppable crush-down with perimeter ejection, exactly as recorded.
NIST’s global models incorporated these floor-type differences (beam-framed mechanical vs. truss typical) and confirmed no arrest was possible once initiation occurred—consistent with the extra mass effect you’re asking about. If you’d like to explore the next question (perhaps quantifying the extra mass, or how this played out at specific mechanical floors like 75–76), or refine the analogy further, go ahead!
(To be continued.)
(©ASM and Grok, initially discussed on April 16, 2026)