WTC Collapse Simulation using Unity/Besiege

[dylbie]

Hi all. I'm new here, first post, and this is not my normal topic of interest, so go easy on me! I'm also not an expert on any of this but thought I'd share a little project I've been working on whilst I've been bored waiting for the Kostack Studio's new simulations and during Covid isolation.

My model has been built using Besiege which runs on the Unity Engine. I've tried to replicate the building as best I can, but obviously there are a lot of limitations with the engine itself and my model which I am aware as follows:
- Heat/Fire. there is no simulated heat or fire and as a result no weakening of any blocks/structure.
- Deformation. All of the blocks are 'solid'. They do not bend, deform or snap. This means there is no compressive or tensile damage being simulated.
- Connections. There are no connections between blocks, meaning the structure is held up only by gravity so has no strength other than the weight of blocks and friction.
- Design. The design itself is basic due to the limitations of the software and the points above. I can only get to 96 floors before my CPU melts.

As such it is very very rudimentary, but despite all this the results look plausible.

Below are some pictures of the buildings. It's made with over 22,000 objects. I've colour coded different parts of the building. The white blocks are to represent the outer columns, the black are the floor/trusses, and the yellow represents the inner core columns.

And a collapse video:

 

[Mendel]

Well, that shows a lot of material "ejected" sideways!

For some background:

Article:

Besiege is a strategy sandbox video game developed and published by Spiderling Studios for Windows, OS X and Linux. The game was released on 18 February 2020, which followed a five-year long early access phase.

The game allows players to build outlandish medieval siege engines to pit against castles or armies. Players select from a collection of mechanical parts that can be connected together to build a machine. Each level has a goal, such as "destroy the windmill" or "kill 100 soldiers". Although the goals are relatively simple, the wide variety of possible approaches allows for experimentation.

Source: https://en.m.wikipedia.org/wiki/Besiege_(video_game)

Apparently the game also has an editor for building castles.

 

[investigating911]

Hi all. I'm new here, first post, and this is not my normal topic of interest, so go easy on me! I'm also not an expert on any of this but thought I'd share a little project I've been working on whilst I've been bored waiting for the Kostack Studio's new simulations and during Covid isolation.

My model has been built using Besiege which runs on the Unity Engine. I've tried to replicate the building as best I can, but obviously there are a lot of limitations with the engine itself and my model which I am aware as follows:
- Heat/Fire. there is no simulated heat or fire and as a result no weakening of any blocks/structure.
- Deformation. All of the blocks are 'solid'. They do not bend, deform or snap. This means there is no compressive or tensile damage being simulated.
- Connections. There are no connections between blocks, meaning the structure is held up only by gravity so has no strength other than the weight of blocks and friction.
- Design. The design itself is basic due to the limitations of the software and the points above. I can only get to 96 floors before my CPU melts.

As such it is very very rudimentary, but despite all this the results look plausible.

Below are some pictures of the buildings. It's made with over 22,000 objects. I've colour coded different parts of the building. The white blocks are to represent the outer columns, the black are the floor/trusses, and the yellow represents the inner core columns.

View attachment 47534

View attachment 47535

And a collapse video:

Interesting, but from what I have seen on this forum, it is generally held that physics sandbox games like Besiege still don't show what should happen but are more meant to show realistic looking (but with no guarantee of being accurate to the real world) graphics of building collapses.

see the following discussion (starting from the linked post downwards) I had with others here about this: https://www.metabunk.org/threads/ho...ngs-penetrate-the-wtc.3326/page-5#post-259628

Finite Element Analysis (FEA) simulations such as the ones that NIST carried out are said to show a more accurate description of what would occur in such a building collapse as they would better simulate real world physics.

 

[Mendel]

with no guarantee of being accurate to the real world

@dylbie wrote "obviously there are a lot of limitations with the engine itself and my model", so they seem to be aware of that.

(In my posts, I like to trim quotes down to the relevant bits. The easiest way to do that is to mark the text I'm replying to, and then use the "reply" popup.)

 

[investigating911]

@dylbie wrote "obviously there are a lot of limitations with the engine itself and my model", so they seem to be aware of that.

But OP also said that
"As such it is very very rudimentary, but despite all this the results look plausible."

I thought that concluding whether results of a building collapse simulation would show plausible results wasn't really possible unless you were working with Finite Element Analysis (FEA) models that simulate real world physics?

 

[galerkin]

I work with and have extensive knowledge of FEA and I can say that even if you see someone presenting a FEA simulation, you should be skeptical. A FEA analysis will be able to give you an accurate estimate for the onset of collapse/failure/damage of structures, but if you are looking a faithful representation of the collapse motion/pattern, FEA will give you garbage there.

P.S. - in very simple cases you can get an idea of a collapse pattern (like, soft-story pattern collapse of a building) but if you are looking to draw conclusions of the WTC pattern of collapse from the post-processing data of a FEA model, yeah, no chance.

 

[investigating911]

I work with and have extensive knowledge of FEA and I can say that even if you see someone presenting a FEA simulation, you should be skeptical. A FEA analysis will be able to give you an accurate estimate for the onset of collapse/failure/damage of structures, but if you are looking a faithful representation of the collapse motion/pattern, FEA will give you garbage there.

Yes I remember in that thread that @econ41 had talked about the NIST FEA simulation of WTC 7 and how and why it didn't "look like" the real event despite modeling the real world physics.

That is the fundamental issue. Methods of simulation can have two different goals viz (a) producing a graphic result that looks like the event OR (b) mimicking the actual physics using or producing quantified values of forces and other physical properties. The former "looks like" the real event, the latter produces engineering data and may not "look correct".

Engineering models often don't "look like" the real event because scaling factors needed to produce the correct numerical values result in distortions of the appearance. The NIST simulation of WTC7 is a classic example. It grossly distorts the movements of the frame way beyond the actual ductility limits of the material. Leading to the truther belief that the NIST model is wrong because it doesn't "look like" the real event. It does if you understand that it is grossly exaggerating the movements. BUT that "doesn't look like" feature is why Hulsey's project for AE911 cheated its alleged "simulation". It produced a "looks like" graphic to meet the false expectations of truther supporters. By direct manipulation of the input parameters. The Hulsey "simulation" is actually a bastardised neither one nor the other. The base method is engineering simulation BUT with various bits of graphic manipulation "cheating" to make it "look like" what the customers wanted.

P.S. - in very simple cases you can get an idea of a collapse pattern (like, soft-story pattern collapse of a building) but if you are looking to draw conclusions of the WTC pattern of collapse from the post-processing data of a FEA model, yeah, no chance.

But from the earlier discussion about plane impacts it appears that sandbox game rendering software like that found in Besiege is not suitable for correctly modeling plane-building impacts, so what makes it different that would allow such game rendering software to able to give a realistic idea of a building collapse pattern?

- Deformation. All of the blocks are 'solid'. They do not bend, deform or snap. This means there is no compressive or tensile damage being simulated.

Rigid body physics might be able to simulate the collapse progression in the beginning, but because there is no deformation of blocks in the building simulation, I would think there would still be some elements missing that are important for accurately simulating the entire duration of the collapse, such as build-up of the compressed floors (which would be less effective in forcing out exterior columns as the floors fell since a lot of the blocks would get compacted rather than bouncing off and away from the other blocks) and bending of the exterior columns and sagging of the floor trusses, which would better model peeling away of the exterior columns, especially in the region where the zone of collapsed floors reached the highest intact floor.

Also there is a delayed collapse of the lower portion of the tower in your simulation (0:10 to 0:16), after which the lowest portion of the exterior columns completely give way and the lower exterior portion of the tower finally falls. This doesn't seem to be accurate given how the lowest portion of the exterior columns at the ground lobby level remained intact in significant portions on various sides of the towers (north and east for the north tower, west and south for the south tower) after the towers had finished collapsing.

- Connections. There are no connections between blocks, meaning the structure is held up only by gravity so has no strength other than the weight of blocks and friction.

Not including this aspect might prevent an accurate simulation of the general manner and timing of collapse of the interior columns, as the trusses had connections to the exterior and interior columns and the free-standing interior columns that were left momentarily standing had a lot less support once the exterior columns and floor trusses were no longer attached to them, which occurred at different times for different heights of the temporarily free-standing interior columns.

One thing that would be interesting in examining the collapse progression is to create a simulation of the tower collapse with the basement levels underneath the towers included, such as in these simulations:

Spoiler

Description: Simulation of both WTC tower collapses onto the rest of modeled WTC complex in Detonate

Source: https://www.youtube.com/watch?v=8UWrswoyDNU

Spoiler

Description: Simulation of south tower collapse down onto the rest of modeled WTC complex in Detonate

Source: https://www.youtube.com/watch?v=9aGY8lI7YN4

So far there have been very few simulations with the added basement levels beneath the towers.

 

[dylbie]

I agree with pretty much everything above.

As I said Besiege is fairly simple software although it is useful.

I guess the point of my experiment is to show that a collapse is plausible, and that the collapse in my video does resemble what happened in real life to some extent. For example sideways ejection of material, the large debris field, the distribution of different parts of the structure in the debris (this is why I made them different colours), the fact that the core remains as the last part to fall. I do however note that there is a 'stall' in this collapse simulation which is due to the lack of connections and lack of deformation of the blocks.

This does do a pretty good job of ruling out any theory about controlled demolition though in my eyes. I have run ground level failure simulations and the collapse looks completely different and wildy different to what happened in real life. If anyone is interested in seeing 'demolition style' simulations with this model I can make the more videos.

There is one other thing, Besiege also has flyable, destructible planes. I could show simulations of the actual plane impacts but at the moment I'm fighting an internal battle with my brain over whether that would be disrespectful. What do you guys think?

 

[FatPhil]

There is one other thing, Besiege also has flyable, destructible planes. I could show simulations of the actual plane impacts but at the moment I'm fighting an internal battle with my brain over whether that would be disrespectful. What do you guys think?

Independent of respect, that line perhaps has already been crossed, I don't think it would necessarily even be that useful. The model you have kinda matches a brittle structure with no strength in tension, so not that great for building materials such as metals, but at least it shows that downward forces can end up with sideways movement. Given that even that simple concept has been objected to by some, it seems (I don't keep my ears that close to the ground), the first simulation at least shows that explosives are not necessary to fling things outwards. It demonstrates a few simple principles, but I don't think it has the legs for building a bigger grander theory than that.

 

[galerkin]

@investigating911

Everything econ41 said is correct but I also wanted to hint at another aspect of simulation (of extreme events). FEA, as a numerical method, will get you only so far. When phenomena like excessive distortions of elements (due to extreme deformations for example), cracking, breaking etc are starting to occur, the method does not give reliable results. The question the skeptics need to focus on and insist on is whether the likelihood of a plane crashing like it did could induce a collapse as the one we have all seen. If the debate is concerned with accurate depictions of "simulations", then the truthers will always be "correct" in the sense that no matter how detailed your FE model is, it will always diverge in the post-processing from the actual event. The rigid body analysis after the onset of collapse is also an interesting modeling approach.

But from the earlier discussion about plane impacts it appears that sandbox game rendering software like that found in Besiege is not suitable for correctly modeling plane-building impacts, so what makes it different that would allow such game rendering software to able to give a realistic idea of a building collapse pattern?

Well, the tower(s) didn't fall immediately after the planes hit them, so one could start the analysis of just the tower by itself with i) some missing structural elements(due to damage, worst case scenario zero contribution to the capacity assumed) and ii) fire in 2-3 floors. It might be the case that the impact of the plane caused damage to structural elements in stories far below the ones that caught fire. That, from a modelling perspective, "only" requires that you apply a point load at the point of impact (you have to take into account the plane momentum ofc and model it as a pulse). But the towers seemed quite stiff and I doubt the plane caused serious structural damage due to vibrations (ofc my estimation might be wrong here - no analysis!). In a few words, "modeling the impact of plane to a building" is much different, in some sense much simpler than it sounds.

 

[econ41]

Everything econ41 said is correct but I also wanted to hint at another aspect of simulation (of extreme events). FEA, as a numerical method, will get you only so far. When phenomena like excessive distortions of elements (due to extreme deformations for example), cracking, breaking etc are starting to occur, the method does not give reliable results.

If, in fact, it is even possible to apply FEA. I suggest that, in the context of most of these discussions, it is sufficient to rigorously explain the actual mechanism without the need to resort to quantification. My claim >> qualified or descriptive explanation of the collapse mechanisms will progress understanding further than it is possible to go with numeric quantification.

Add to the complexity the reality that full comprehension of Twin Towers collapses needs a clear understanding of the distinct stages of mechanism. IMO that means four distinct stages. And none of them needs FEA "help" - the key factors are definable without FEA and most of them are not accessible to FEA analysis.

The question the skeptics need to focus on and insist on is whether the likelihood of a plane crashing like it did could induce a collapse as the one we have all seen.

Maybe. But that is already heading a big advantage to the truthers. The extant hypotheses by scientific method say (a) the collapses we observed DID in fact happen; (b) the physics explains those collapses as caused by initial impact damage and accumulating fire damage from unfought fires; AND (c) No truther has ever falsified the "No CD help needed" hypotheses.

If the debate is concerned with accurate depictions of "simulations", then the truthers will always be "correct" in the sense that no matter how detailed your FE model is, it will always diverge in the post-processing from the actual event.

Only if we presume that modelling/simulation is the only path to "proving what actually happened". Which begs the question of this and one other contemporary thread >> "What is the objective of this discussion?"

Well, the tower(s) didn't fall immediately after the planes hit them, so one could start the analysis of just the tower by itself with i) some missing structural elements(due to damage, worst case scenario zero contribution to the capacity assumed) and ii) fire in 2-3 floors.

Why not simply explain what actually happened?

It might be the case that the impact of the plane caused damage to structural elements in stories far below the ones that caught fire. That, from a modelling perspective, ....

Again - what is the objective? Explain the Towers collapses OR build a model???

But the towers seemed quite stiff and I doubt the plane caused serious structural damage due to vibrations (ofc my estimation might be wrong here - no analysis!). In a few words, "modeling the impact of plane to a building" is much different, in some sense much simpler than it sounds.

The Towers stood after the impacts and the resonant vibrations had ceased. The collapse was caused by unfought fires AFTER that stage. Feel free to postulate a different scenario - but be aware that impact damage and vibration as causes of collapse have been comprehensively rebutted.

And modelling the aircraft impact(s) is a different scenario/situation from modelling the collapse mechanism.

 

[Jeffrey Orling]

Seems to me that the plane impact was the beginning of the load redistribution from the destroyed columns to the intact one. The fact that the collapse did not immediately go "runaway" seems to indicate that there was reserve capacity. However, as Econ mentions there were unfought fires which ensued and the heat from those fires did two things: cause steel elements to expand (and contract when cooled) and loose "strength".
It's pretty hard to "model" the "path of loss of capacity" without actual data from inside the towers. But we do see at some point the capacity fell below the "service" loads and the top sections in both towers did not have adequate support which meant they began to drop. But many columns were still continuous and would have buckled. The loss of capacity would also lead to some lateral movement as the upper mass descended downward. The gross movements are all recorded on video and are consistent with loss of capacity, column buckling and lateral translation. The tops would not have dropped if the columns supporting it had adequate capacity.

Does it matter which columns were destroyed in what sequence?

How much excess capacity is reasonable?

I think one take away is that without fire suppression a steel frame doesn't have a chance to survive. Should the steel have been more robust? Should the fire protection been more robust? Should the fire suppression been more robust? Was the design adequate for the envisioned environment?

 

[galerkin]

@Jeffrey Orling

Seems to me that the plane impact was the beginning of the load redistribution from the destroyed columns to the intact one. The fact that the collapse did not immediately go "runaway" seems to indicate that there was reserve capacity.

That is standard in all statically indeterminate structures (practically all buildings). When one member fails, the system stiffness changes and internal loading is redistributed. All structures do that.

However, as Econ mentions there were unfought fires which ensued and the heat from those fires did two things: cause steel elements to expand (and contract when cooled) and loose "strength".
It's pretty hard to "model" the "path of loss of capacity" without actual data from inside the towers.

Yes, some issues arising from the fire are i) beam-like elements suffering from buckling because of restrained expansion due to heat and ii) buckling/loss of ultimate strength (and thus "yielding") due to fire. Structural members exposed to heat for long periods lose strength and stiffness.

But all these are local phenomena, restricted to 2-3 floors. I was mostly interested in the modeling of the top block falling on the rest of the tower, and, secondarily, the initial impact of the plane.

@econ41

Only if we presume that modelling/simulation is the only path to "proving what actually happened". Which begs the question of this and one other contemporary thread >> "What is the objective of this discussion?"

The objective I entered this discussion with is mainly to add that someone going around with an FEM simulation depicting nice plots doesn't necessarily mean much, whether he is a skeptic or a truther, because the method itself has severe limitations after a certain regime has been reached. I can go ahead and provide here simulations with more "advanced" methods which aim to tackle phenomena of interpenetration, cracking, explosions etc, in all of which FEM fails, and you will see that even these methods do not really reproduce the experimental result accurately - but they do progress after a point where FEA simply fails. So my objective was to stress that FEM was good when you want to determine the onset of "collapse" or failure.

Why not simply explain what actually happened?

Well, we do not have a fully clear picture, that is, you can never tell by video observation that there was no damage to floors far below the impact point after the plane hit the tower(s). But you can tell that the towers didn't fall from the impact - that is clear and it pertains to your "what actually happened". The fire is also one other thing that "actually happened". So as you can see, I am talking about "what actually happened".

Again - what is the objective? Explain the Towers collapses OR build a model???

You seem quite partisan on the issue. I explained my objective earlier but, yeah, a general objective could be to build a reliable model(s) to explain the collapse or, to be more specific: to see under what conditions you can get "similar" patterns of collapse.

The Towers stood after the impacts and the resonant vibrations had ceased. The collapse was caused by unfought fires AFTER that stage. Feel free to postulate a different scenario - but be aware that impact damage and vibration as causes of collapse have been comprehensively rebutted.

I didn't claim to be in favor of any theory or anything. In my exchange with @investigating911, I was just exploring (thinking out loud one could say) what likely scenarios one could implement in a model. Whether they have already been debunked or not, I am not in a position to know but certainly not unwilling to accept. I have no dog in this fight to be honest. By the way...

The Towers stood after the impacts and the resonant vibrations had ceased

this sentence doesn't make much sense. The word "resonant" is i) redundant and ii) misused.

 

[investigating911]

Well, the tower(s) didn't fall immediately after the planes hit them, so one could start the analysis of just the tower by itself with i) some missing structural elements(due to damage, worst case scenario zero contribution to the capacity assumed) and ii) fire in 2-3 floors. It might be the case that the impact of the plane caused damage to structural elements in stories far below the ones that caught fire. That, from a modelling perspective, "only" requires that you apply a point load at the point of impact (you have to take into account the plane momentum ofc and model it as a pulse). But the towers seemed quite stiff and I doubt the plane caused serious structural damage due to vibrations (ofc my estimation might be wrong here - no analysis!). In a few words, "modeling the impact of plane to a building" is much different, in some sense much simpler than it sounds.

Spoiler: I recommend downloading this video in case it is removed or made private

Description: I have come across a simulation of the south tower collapse, which models plane damage and fire leading up to the collapse that someone made using the sandbox game Detonate

Source: https://www.youtube.com/watch?v=5RuP9kpKhUQ

The above video might be such an example of such a model, although it is far from a detailed structural analysis.

Also as others have stated and even the NIST report found, the plane impact damage to the towers wouldn't have been enough to cause the towers collapse, as the loads would simply be redistributed around the impact hole, which the hat truss at the top of the towers would help with. Rather, from what I have understood, the reason for the towers collapse was the removal of fireproofing on the impacted floors, severing of the standpipes that could otherwise deliver water from the sprinkler system to fight fires, and the ensuing fires that started from the jetfuel explosions that weakened the steel holding up to the top portion of the towers above impact enough that the top portion started moving down and the sheer momentum of the downward moving upper story of each tower resulted in a progressive collapse of each tower.

In a few words, "modeling the impact of plane to a building" is much different, in some sense much simpler than it sounds.

From the discussion earlier, https://www.metabunk.org/threads/ho...ngs-penetrate-the-wtc.3326/page-5#post-259628

I thought that the overall conclusion was that rigid body physics sandbox simulators are not suitable enough to model a plane impact into a building, such as could be modeled in sandbox games like Besiege or Teardown. This was said to be because of a variety of factors, such as the model of the plane not demonstrating movement of the wings, the jerry-rigged nature of how the plane and building break apart upon impact, the ability to turn off particle collisions, and a set function to create holes of 3 preconfigured sizes instead of taking into consideration the speed of the plane and the properties of the material being impacted (e.g. weak steel versus hard steel) when creating impact holes.

Therefore, I was wondering why if the physics of a plane-building impact collision could not be accurately simulated in a physics sandbox game like Besiege or Teardown, what would make it any different that a physics sandbox game would be able to model the entire duration of collapse of a hollow tube construction skyscraper (which the WTC towers were and which had with a dense array of exterior columns, light weight floor trusses, open floor space and a rectangular core column region) in accordance with real world physics?

 

[Mendel]

in accordance with real world physics

How do you know it's in accordance with real physics?

The video has been up for 2 years, no reason to assume it'll disappear; but I understand you enjoy a heightened feeling of drama?

 

[dylbie]

This is kind of my point though. In 20 years of massive advancements in processor power, the most we have is a video of a game recorded from a mobile phone of a laptop screen?

This is why I'm looking forward to the Kostack simulations, but I find it hard to believe he's the only person doing proper simulations in the 20 years since.

So since doing the simulations with Besiege it has thrown up a number of questions that remain unanswered as far as I can tell. And just to clarify I am definitely NOT a truther!

1) Did the oscillations caused by the impact have any contributing factor to the collapse? Was any damage to the core columns above or below ground significant? In my simulations I've noticed very significant oscillations not previously discussed which do seem consistent with those seen in Scott Myers video. As an example, the whole building sways and the portion above impact seems to be oscillating at a higher frequency to the rest. So dual oscillations if that makes sense.

2) Did the collapse itself cause any further oscillations which made the collapse worse? Again, in my simulations I'm noticing massive oscillations in the building especially after the initial 'hit' of the floors above the failure point. If this happened in real life, which I expect, then this could have accelerated the collapse.

3) Could the plane have hit in a place where it didn't cause the building to collapse? A partial or glancing blow for example.

4) Could the plane have hit in a place where it caused the building to fail and collapse immediately as a direct result of the impact? Lower impact where the columns bear more weight? Higher impact damaging the hat truss structure? A direct hit to the corner?

5) Which column failures caused the collapse? Which ones? How many? How many could be removed and in what sequence before collapse?

6) Is a partial collapse, as often suggested by truthers, possible?

7) Could better fire regulations in the building design, prevented collapse or prolonged the time to collapse?

All these above questions could be answered by accurate simulations.

 

[Landru]

This is kind of my point though. In 20 years of massive advancements in processor power, the most we have is a video of a game recorded from a mobile phone of a laptop screen?

This is why I'm looking forward to the Kostack simulations, but I find it hard to believe he's the only person doing proper simulations in the 20 years since.

So since doing the simulations with Besiege it has thrown up a number of questions that remain unanswered as far as I can tell. And just to clarify I am definitely NOT a truther!

1) Did the oscillations caused by the impact have any contributing factor to the collapse? Was any damage to the core columns above or below ground significant? In my simulations I've noticed very significant oscillations not previously discussed which do seem consistent with those seen in Scott Myers video. As an example, the whole building sways and the portion above impact seems to be oscillating at a higher frequency to the rest. So dual oscillations if that makes sense.

2) Did the collapse itself cause any further oscillations which made the collapse worse? Again, in my simulations I'm noticing massive oscillations in the building especially after the initial 'hit' of the floors above the failure point. If this happened in real life, which I expect, then this could have accelerated the collapse.

3) Could the plane have hit in a place where it didn't cause the building to collapse? A partial or glancing blow for example.

4) Could the plane have hit in a place where it caused the building to fail and collapse immediately as a direct result of the impact? Lower impact where the columns bear more weight? Higher impact damaging the hat truss structure? A direct hit to the corner?

5) Which column failures caused the collapse? Which ones? How many? How many could be removed and in what sequence before collapse?

6) Is a partial collapse, as often suggested by truthers, possible?

7) Could better fire regulations in the building design, prevented collapse or prolonged the time to collapse?

All these above questions could be answered by accurate simulations.

These are issues already discussed in other threads on Metabunk. Use tge search function to find them.

 

[dylbie]

Thanks @Landru. I'll have a good look this evening.

Have any of the above points been simulated. I have scoured the Internet for simulations and the only things I can find are the very early NIST ones, and the Kostack ones done from 10 years ago. Both are very limited in that they only seem to simulate a small portion of the building, and they are only trying replicating one aspect of the event.

 

[Rory]

There is one other thing, Besiege also has flyable, destructible planes. I could show simulations of the actual plane impacts but at the moment I'm fighting an internal battle with my brain over whether that would be disrespectful. What do you guys think?

I think there are plenty of videos showing simulated planes hitting simulated towers, so it's not like they're not already out there. I would say "go ahead": I quite enjoy watching them myself, so I'd be interested to see how yours looked.

 

[econ41]

All these above questions could be answered by accurate simulations.

I seriously doubt that those questions can be "...answered by accurate simulations."

And there is no doubt that all of them can be answered by reasoned analysis of the relevant physics. Without physical modelling. So what is the purpose, the goal, the objective of modelling?

I'll repeat the advice I've offered several times in similar threads:

Decide what your objective is. Are you wanting to build a physical model or do you want to understand the actual collapse?

And, even more, fundamental HOW can you model any collapse if you don't first understand what you are trying to model?

For the size and complexity of the modelling challenge, you must understand the mechanism BEFORE you can build a model.
And. as @Landru has said:

These are issues already discussed in other threads on Metabunk. Use tge search function to find them.

Several of those other threads discuss in some depth the need to determine what the goal is before undertaking to model.

 

[dylbie]

This sounds like a question from Neo from the Matrix from someone who is trying to be philosophical or something.

I want to create an accurate 3D model in decent software, where I can enter realistic variables which will produce reasonably accurate results, so I can BETTER understand the collapse. There are still questions which I haven't found the answers for. Maybe they are hidden in this forum within the thousands of opinions.

Maybe I need to be asking a different question then: why after 20 years of one of the most significant events in human history, and in a day and age of very sophisticated simulation software, have so few simulations been done?

Hopefully we won't have to wait to much longer for the Kostack simulations which will answer the questions I have. There doesn't seem to be anywhere else where this has/is being done.

 

[Jeffrey Orling]

A model/simulation would require that the modeler understand and input the energy in the right amounts, right places and proper sequence... and the resulting simulation should resemble real world.

If one doesn't understand how the structure collapsed... you can't possible make a model/simulation that has any credibility.

 

[dylbie]

I think you're getting mixed up between modeller, and simulation software/physics developer.

I see it the other way around. If the software is good enough to begin with and has a decent physics engine built in, then all you should be able to do is build an accurate scale model, set the strength variables of the steel, welds, and bolted connections. Provided the physics engine in the software is good, then there shouldn't be any more understanding needed. That's what the simulation software is used for.

There could be an argument that if you're having to write too much of the software to begin with then you end up manipulating it to give you the results that you want. That then wouldn't be a simulation, more of an animation of your expected results.

 

[dylbie]

To clarify my point... If you had an amazing bit of software, with a fantastic physics engine, then you shouldn't need to know ANYTHING about the collapse.

All you'd need to do is build an accurate 3D model which included some assumed failure points, and then the software would simulate the collapse.

 

[Landru]

I think you're getting mixed up between modeller, and simulation software/physics developer.

I see it the other way around. If the software is good enough to begin with and has a decent physics engine built in, then all you should be able to do is build an accurate scale model, set the strength variables of the steel, welds, and bolted connections. Provided the physics engine in the software is good, then there shouldn't be any more understanding needed. That's what the simulation software is used for.

There could be an argument that if you're having to write too much of the software to begin with then you end up manipulating it to give you the results that you want. That then wouldn't be a simulation, more of an animation of your expected results.

I doubt that something that cost less than $15 on steam is going to give you scientificly accurate results.

 

[econ41]

To clarify my point... If you had an amazing bit of software, with a fantastic physics engine, then you shouldn't need to know ANYTHING about the collapse.

Sorry, @dylbie, but that is fundamentally wrong and very wishful thinking. I'm prepared to explain why if you are interested.

All you'd need to do is build an accurate 3D model which included some assumed failure points, and then the software would simulate the collapse.

1) An accurate model of what? I assume you mean a model of the structure because assuming a model of the failure mechanism begs the question; SO
2) If YOU then "assume some failure points" YOU have decided the collapse mechanism which you have imposed on whatever "physics engine" your software incorporates; AND
3) "...then the software would simulate the collapse which you have implicitly pre-determined by your choice of "assumed failure points".

Hence my repeated advice to EXPLICITLY define your objective because it is "wandering".

I think the core of your objective is to build a physical model of a WTC Twin Tower which demonstrates how the power collapsed in the events of 9/11. But the discussion has also alluded to alternate impact points. Which by definition are NOT what happened o 9/11. There is nothing inherently wrong with exploring other possibilities BUT the implied goal of both scenarios - actual event or alternatives - is not pragmatically possible with physical modelling. (And is feasible by analysis.)

So what are you trying to demonstrate? And to whom do you want to demonstrate? And what level of credible accuracy is needed for whoever is your target "audience"?

 

[Jeffrey Orling]

What led to the visible movements of the 3 building collapses was largely obscured by the facades of each building. There were movements which were observed and those can be explained by "failures" inside the tower. But certainty may not be possible. Take the collapse of the East Penthouse. There are some clues that mass dropped many stories... but on what level was the initial failure? Once the mass dropped... it is likely it would continue as no floor slab was mostly likely capable to arrest the collapse. NIST decided that the failure was down around floors 12/13 and the drop was over 30 stories of mass and that may have been what caused structural failures to move south and west leading to a gutting of the tower and undermining or the support for the exterior moment frame.
The twin towers are more mysterious. We did some some inward pulling of the facade on 1WTC in the south side near the east corner. But that doesn't seem to be enough damage to the exterior columns to lead to the entire top section to collapse down.
Both twin towers stood after the plane destroyed parts of the exterior and we can conclude the facade design was capable of load redistribution. But the remain facade would be then carrying more load and more easily buckled. Unseen core destruction may have played role... where fires raged, warped and weakened the steel.
Perhaps a careful forensic analysis of the steel would have told something. It was not done. Because the event was so out of spec that it made little sense to strengthen future steel designs for events unlikely to ever happen. I think they now take fire suppression and egress a more seriously however.

 

[dylbie]

Sorry, @dylbie, but that is fundamentally wrong and very wishful thinking. I'm prepared to explain why if you are interested.


1) An accurate model of what? I assume you mean a model of the structure because assuming a model of the failure mechanism begs the question; SO
2) If YOU then "assume some failure points" YOU have decided the collapse mechanism which you have imposed on whatever "physics engine" your software incorporates; AND
3) "...then the software would simulate the collapse which you have implicitly pre-determined by your choice of "assumed failure points".

Hence my repeated advice to EXPLICITLY define your objective because it is "wandering".

I think the core of your objective is to build a physical model of a WTC Twin Tower which demonstrates how the power collapsed in the events of 9/11. But the discussion has also alluded to alternate impact points. Which by definition are NOT what happened o 9/11. There is nothing inherently wrong with exploring other possibilities BUT the implied goal of both scenarios - actual event or alternatives - is not pragmatically possible with physical modelling. (And is feasible by analysis.)

So what are you trying to demonstrate? And to whom do you want to demonstrate? And what level of credible accuracy is needed for whoever is your target "audience"?

Sure I'm very interested to know why you think a simulation with an accurate physics engine is fundamentally wrong and wishful thinking. You can PM me if you don't want to clog up the forum. I'm interested to know, do you think the Kostack simulations are inaccurate?

1) Yeah that's what I mean, an accurate model of the tower.
2) By assuming failure points, you can then run the simulation and see if it matches up with reality. Some failure points were known on the exterior columns/facade, some were unknown such as the interior columns, so you'd have to use some assumptions there, unless there is accurate data on which columns were broken by the impact (I don't think this data exists).
3) Yes the collapse would run with assumed failure points that is correct. But the aim is to get the simulation close to reality. If the building twisted, toppled, or fell over in a strange way, you'd know that something isn't quite right with what you've assumed that failed. The the simulation would need to be run again with different assumptions on internal failure points.

I'll EXPLICITLY state my objective again: to use simulations to better understand the sequences of collapse based on known data, and unknown assumed failures.

The point of alternative impact points would be a secondary objective once the simulation has proven to be reasonably accurate. So this can be ignored at this point.

To answer your last questions: I would aim to demonstrate, using accurate simulation software the events of the collapse. To everyone who is interested. It would be nice to have a high level of credible accuracy, which I would think should be possible with 2021 levels of processing power and modern software.

To further clarify my situation, I have reproduced a very rudimentary 3D model, with a consumer level PC, on software which has a reasonable level of gravity simulation. The results are plausible but have some obviously glaring shortfalls. If the software had some extra parameters and options for connecting objects, and also deformation of objects, I feel it could do a reasonably credible job of simulating what happened.

This is why blender with the constraint addons is probably our best bet at this time, and this is why I am looking forward to the new Kostack simulations, which are far more advanced than anything I can do.

 

[dylbie]

I doubt that something that cost less than $15 on steam is going to give you scientificly accurate results.

I've already said the software that I'm using isn't very accurate, and I state in the first post the obvious limitations of it.

My point now is why haven't more simulations been run with more advanced software?

 

[dylbie]

The twin towers are more mysterious. We did some some inward pulling of the facade on 1WTC in the south side near the east corner. But that doesn't seem to be enough damage to the exterior columns to lead to the entire top section to collapse down.
Both twin towers stood after the plane destroyed parts of the exterior and we can conclude the facade design was capable of load redistribution. But the remain facade would be then carrying more load and more easily buckled. Unseen core destruction may have played role... where fires raged, warped and weakened the steel.

And this is exactly where a simulation could help us better understand what was going on inside the core.

 

[Landru]

I've already said the software that I'm using isn't very accurate, and I state in the first post the obvious limitations of it.

My point now is why haven't more simulations been run with more advanced software?

What advanced software? Do you have an example of advanced software? Also, the topic of this thread is WTC Collapse Simulation Using Unity/Besiege. Other software discussions belong in another thread.

 

[dylbie]

What advanced software? Do you have an example of advanced software? Also, the topic of this thread is WTC Collapse Simulation Using Unity/Besiege. Other software discussions belong in another thread.

Extreme Loading for Structures
https://www.extremeloading.com/

Yeah I realise the topic has strayed somewhat from the original which I never intended. I guess maybe I should retitle the thread: This is what basic software can now do, why have there not been more advanced simulations?

 

[Landru]

Extreme Loading for Structures
https://www.extremeloading.com/

Yeah I realise the topic has strayed somewhat from the original which I never intended. I guess maybe I should retitle the thread: This is what basic software can now do, why have there not been more advanced simulations?

Probably because of the cost and no one sees the need to do it.

 

[Mendel]

with a fantastic physics engine

There are no fantastic physics engines. Game engines even less so, since they're tweaked for speed, not accuracy.

do you think the Kostack simulations are inaccurate?

Yes.

To clarify my point... If you had an amazing bit of software, with a fantastic physics engine, then you shouldn't need to know ANYTHING about the collapse.

All you'd need to do is build an accurate 3D model which included some assumed failure points, and then the software would simulate the collapse.

The problem is that without knowledge about the collapse, you can't judge whether the result is correct or not. This is a genefal problem people have when employing computers. Do you recall math problems in school where you had the right formula, but your result was still wrong?
To trust a model, you have to know which part of reality it corresponds to, and what its limitations are. That goes for all models, especially for computer models.

answer the questions I have. There doesn't seem to be anywhere else where this has/is being done.

1) Did the oscillations caused by the impact have any contributing factor to the collapse? Was any damage to the core columns above or below ground significant? In my simulations I've noticed very significant oscillations not previously discussed which do seem consistent with those seen in Scott Myers video. As an example, the whole building sways and the portion above impact seems to be oscillating at a higher frequency to the rest. So dual oscillations if that makes sense.

2) Did the collapse itself cause any further oscillations which made the collapse worse? Again, in my simulations I'm noticing massive oscillations in the building especially after the initial 'hit' of the floors above the failure point. If this happened in real life, which I expect, then this could have accelerated the collapse.

3) Could the plane have hit in a place where it didn't cause the building to collapse? A partial or glancing blow for example.

4) Could the plane have hit in a place where it caused the building to fail and collapse immediately as a direct result of the impact? Lower impact where the columns bear more weight? Higher impact damaging the hat truss structure? A direct hit to the corner?

5) Which column failures caused the collapse? Which ones? How many? How many could be removed and in what sequence before collapse?

6) Is a partial collapse, as often suggested by truthers, possible?

7) Could better fire regulations in the building design, prevented collapse or prolonged the time to collapse?

1) Did the oscillations happen in reality, or just in the model? Some physics engines are prone to oscillate by themselves. How is damping modeled? Are the dampers built into the WTC part of the model? Was there pre-collapse underground damage to loadbearing structures?

2) How can you find out if this happened in reality?

3) The plane did hit in a place that did not cause the building to collapse.

4) Are you planning to do this, or why do you want to know? Lower down, the supporting structures are stronger. An engineer could probably estimate the energy that the aircraft spent deforming the building, and then have a look at which height this would have sufficed to take out enough of the core to destabilize the building.

5) No way to tell whether the sequence that works for your model is what actually happened.

6) A partial collapse is only possible if the collapse sequence never enters the "runaway" phase. This means no floor can get overloaded, which means the fire damage must only set a small part of the building in motion, and the fires need to be confined/extinguished.

7) Yes. Most importantly, they would have allowed more people to escape.

I don't want to discuss these answers here. I'm trying to illustrate that I don't understand what knowledge you hope to be deriving from an unverifiable model. It seems to me that this model could contribute very little to answer these questions.

 

[econ41]

The problem is that without knowledge about the collapse, you can't judge whether the result is correct or not. This is a genefal problem people have when employing computers. Do you recall math problems in school where you had the right formula, but your result was still wrong?
To trust a model, you have to know which part of reality it corresponds to, and what its limitations are. That goes for all models, especially for computer models.

Exactly. In order to build a model or make a simulation of a complex system such as the Twin Towers collapse, you must know what yu are simulating. It can not be a learning experience for the person(s) doing the modelling because they need to already possess at least as much knowledge as the model/sim can show. And probably more. Hence my reason for asking "what is the objective" because the implicit objective is, odds on, "explain to some third party who needs a model or simulation to help their understanding". It then follows that the model must at the least be as accurate as that third party needs for their level of understanding. AND it must match the key aspects of the real event to at least that same level of credibility. And all of these "learning or teaching process" aspects have been discussed in detail on previous occasions.

I don't want to discuss these answers here. I'm trying to illustrate that I don't understand what knowledge you hope to be deriving from an unverifiable model. It seems to me that this model could contribute very little to answer these questions.

Hence my suggestions: (a) define the purpose of the model without all the vague ambiguity AND (b) If you don't already know what you are modelling you won't succeed for a system as complex as Twin Towers collapses. UNLESS you duplicate NIST's work to a higher standard. (Remember why Hulsey failed.)

BUT - given that we already understand the actual Twin Towers collapses in more detail than NIST - why do WE need the model or sim. It won't help "our" understanding. It MAY be helpful for a specific sector of layperson demography. Those who cannot understand complex m mechanisms without physical models. And discussions in previous threads suggest that demographic sector is a lot narrower than many people recognise.

 

[Jeffrey Orling]

A simulation comes AFTER understanding to visual show (presumably) the sequence of failures that one can't see real world because they facades obscure them. I recall Purdue U did one which put the viewer inside a stripped out 1WTC.

 

[Mendel]

. I recall Purdue U did one which put the viewer inside a stripped out 1WTC.

That's a good pointer, I looked this up, and it answers some questions (especially the engineering paper), and basically confirms what I previously wrote about the uncertainties. The team went to great lengths to try and simulate the impact of the fuel realistically.


Source: https://m.youtube.com/watch?v=cddIgb1nGJ8



Source: https://m.youtube.com/watch?v=_FVAzn1Yuz8

Article:

September 11 WTC Attack Simulations​

Using LS-Dyna​

Problem Statement

Simulate as faithfully as possible the effects of crashing an air frame loaded with fuel (simulating a Boeing 767-200ER) into a steel and concrete structure similar to the structure of the WTC-1, North Tower, of the World Trade Center.

Purpose of the Effort

Use the simulation results to understand what the extent of damage done by the impact has been. Effects of the subsequent fire are not under consideration in this phase of the project.
Use the simulation results also to construct animations and visualizations that vividly reenact of the impact, as it plausibly has been. This work will be Phase IV.

[...]
The simulation uses adaptive time stepping which averages to approximately 0.000001 sec time steps. We generate snapshots of the simulation approximately every 0.0025 sec.
[...]
From our modeling of the aircraft crash into the Pentagon building, we knew that a critical issue in defining the damage was the modeling of the fuel in the aircraft. Much of the mass of the aircraft is provided by the fuel; in this case about 27%. The energy imparted to the impacted structure is the initial cause of the damage. This time, we modeled the fluid-structure interaction using smooth particle hydrodynamics (SPH). To calibrate our approach, Dr. Pujol built a special test setup that made it possible to hurl 6-oz liquid containers at a steel target of varying speeds approaching 100 m/sec.

Source: https://www.cs.purdue.edu/homes/cmh/simulation/phase3/

The FEA physics simulation runs at 1 000 000 fps and generates keyframes at 400 fps for the slow-motion display. No computer game physics engine comes even close.

There's a paper on how they translated the simulation output into a visualization. It has a comparison of the model's after-impact facade to an actual photograph; you can see it's somewhat close, but not "fantastic".

Article:

Source: https://www.cs.purdue.edu/cgvlab/papers/popescu/popescuWTCVIS07.pdf

The engineering paper (worth reading in full) discusses the uncertainties inherent in the model.

Article:

No detailed observational data on the performance of tower core elements exist. Therefore, the Purdue research team used finite element simulations to assist in estimating the impact response and the post-impact state of the structural elements in the core. Objective evaluations of the simulation results indicated that identification of the number and distribution of columns damaged immediately by the impact was quite sensitive to the input parameters. As would be expected, the simulations did indicate consistently some damage to the core columns but credible changes in input parameters such as flight or column properties resulted in changes in calculated impact damage suggesting that an exact determination of the damage to the core was not defensible. On the other hand, it was found that a simple construct, not dependent on exact identification of the damage distribution, would explain the collapse.

The study concentrated on the core structure of the tower and its possible behavior under impact and thermal loads. It was not the purpose of the study to rule out other plausible mechanisms initiating collapse of the WTC-I tower, such as one due to loss of lateral bracing and buckling of perimeter columns induced by failure of open-web floor joists under thermal loads, or other failure mechanisms. Rather, the object of the paper is to show that a simple and credible hypothesis will suffice to explain the observed collapse.
[...]

However, it must be added that during the series of simulations performed, we found the estimates to be very sensitive to model parameters such as failure strain of materials, to the extent that in the heavily damaged stories 95 through 97, the number of damaged columns could be as few as half the numbers listed in Table 1. This observation was not surprising given the fact that simulation results reported by other researchers (see for example, NIST 2005 and Omika et al. 2005) with regards to damage to the core columns are scattered over a wide range.
[...]

In other words, even though the aircraft impact loads may have eliminated a significant number of core columns, the damage to core structure had, ultimately, little effect on the critical thermal load level to fail the core under axial gravity loads.

Although not illustrated in here, if the core columns were to lose their lateral supports and their free heights became more than 3.6 meter, even though there would be significant reduction in the total axial load capacity of the core columns, the estimated ultimate failure temperature would not have changed that significantly. For example, doubling the free-height to two-stories reduces the total axial capacity by about 30-40% but reduces the failure temperature to only about 600⁰C or about 650⁰C if no damage to core columns were considered. Again the damage to core columns from impact loads could be considered immaterial when viewed from ultimate thermal load capacity of the core structure. In thermal load analysis for response of structural elements at high temperatures under actual conditions, the effect of a 50⁰C difference in temperature is hardly distinguishable, given all the uncertainties.

It is evident from observation and our simulations that the debris of the aircraft went through the WTC structure at stories 94 through 97. Much of the fire insulation would have been scoured off leaving the steel elements unprotected during the immediately following fire event. Experimental data for steel in that condition (Buchanan 2000) indicate that the metal temperature in all unprotected structural elements would have reached 700⁰C in a typical office fire. That condition would suffice to initiate instability (e.g. Ali and O'Connor 2001, Wang and Davies 2003) even if all the girders were intact and the failure mechanism was limited to one story of the core structure.

Source: https://www.cs.purdue.edu/homes/cmh/distribution/PapersChron/WTC_I_Engineering_Perspective.pdf

Their conclusion is that they can't conclusively determine the damage that the aircraft did to the core, but it doesn't matter because the fires ultimately brought the building down.

So any simulation that depicts collapse initiation properly must be able to model fires (heat, spread, duration), its effect on structural steel (with more or less insulation), and a model of how the loads in the steel structure shift as the steel softens.

No game engine does that, and Kostack doesn't either.

 

[dylbie]

Great post @Mendel thanks for this.

So any simulation that depicts collapse initiation properly must be able to model fires (heat, spread, duration), its effect on structural steel (with more or less insulation), and a model of how the loads in the steel structure shift as the steel softens.

In that case, and the apparent lack of any recent simulation, I wonder if this software will ever exist. Or whether anyone has the time or inclination to do it.

 

[Jeffrey Orling]

simulation will only be as good as the data fed in... columns destroyed and damaged by the plane strike... damaged slabs and fire, size, location spread and of course fuel and time line of the fire damage. You need correct section sizes, correct live and dead loads...
Good luck.
NIST seems to have discounted any damage inside the core. Why?

 

[econ41]

OK @dylbie - sorry for the delay - I'm getting a clearer picture of what you are trying to achieve. But let's see if we can pick what seems to be your main thrust of interest. This touches on a couple of the problems I have identified:

To answer your last questions: I would aim to demonstrate, using accurate simulation software the events of the collapse. To everyone who is interested. It would be nice to have a high level of credible accuracy, which I would think should be possible with 2021 levels of processing power and modern software.

To further clarify my situation, I have reproduced a very rudimentary 3D model, with a consumer level PC, on software which has a reasonable level of gravity simulation. The results are plausible but have some obviously glaring shortfalls. If the software had some extra parameters and options for connecting objects, and also deformation of objects, I feel it could do a reasonably credible job of simulating what happened.

This is why blender with the constraint addons is probably our best bet at this time, and this is why I am looking forward to the new Kostack simulations, which are far more advanced than anything I can do.

So, restated simply, I think what you are trying to achieve is (a) to use software simplification; (b) to demonstrate WTC Twin Towers collapses; (c) to everyone who is interested AND (d) to have a high level of credible accuracy.

I have cautioned that all four of those aspect of your goals as currently defined are not defined with sufficient clarity. And the target area of interest is probably very limited and there are alternate approaches which may be more effective. So - yes there may be a place for yet another software simulation. Yes it can be fun and challenging to undertake the project but the end benefit will be less than what seems to be anticipated.

So the current status of my concerns and similar issues raised by other members is:
(a) For a range of reasons software simulation (i) is not the only way to explain collapses; (ii) though it may have advantages in some situations it is not a universal panacea. For most of "those who are interested" the explanation can be more effective by methods other than simulation. The demographic sector for which simulation has benefits is limited. Yes it can be a useful illustration tool - but it is only confirmatory for those who already understand the mechanisms better than the software can demonstrate.

I suggest that all four of those aspects of your implicit goal need further elaboration and discussion.