Towards A Replicable Physical Model Illustrating Aspects of the Collapse of The WTC Towers on 9/11

Mick West

Administrator
Staff member
Here's a version with the "core" in the middle. Of course with my super-strong 4x4 core it does not collapse. However it provides perhaps the best illustration so far (in this thread) of the actual collapse.


After the collapse I nudge the core a couple of times to show how stable it is.
 

Jeffrey Orling

Senior Member
CG of the core does not move far enough when nudged... try the core with 2x4s ... they might topple... Mick these experiments are fabulous! 2x4s separated by short beams/floors for the core.
 
Last edited:

deirdre

Moderator
Staff member
your trying to pull the solid core (4x4) from both sides at once. which cancels each other out. the real core wasnt a solid block.
 

Mick West

Administrator
Staff member
CG of the core does not move far enough when nudged... try the core with 2x4s ... they might topple... Mick these experiments are fabulous! 2x4s separated by short beams/floors for the core.
The core needs to be stable for it to work here, using two lots of 2x4 by themselves makes it impossible to build. I did a more layered approach of a 4x4 at the bottom and then 2 2x4s for the next level up. This gave an interesting result with a partial collapse of the core. Not really enough floors, but still interesting.

 

Mick West

Administrator
Staff member
I'm not sure I can do much more with this though. An 8' column of 4x4s is remarkably stable. It would be good if the demonstration included the core collapsing, but I'd have to force it a bit more artificially. In the real world there would be significant damage within the core, which is something that would be hard to duplicate. A pair of linked 2x4s for the core is very unstable.

The hooked connectors are pretty beat up too. I've often thought that the ideal reusuble set of building blocks for such a thing would involve small magnets. A bit too much work though.

Any suggestions for configurations with my current set of pieces? I guess I could try a 10 foot or even 12 foot high single sided tower, that would ensure "core" collapse.
 

Mick West

Administrator
Staff member
Here's nine floors double sided, did not fully work out.

Interesting though that there was a partial collapse of the lower floors some time after the other side had settled. Reminiscent of the "late" collapse of the spire.
 

Mick West

Administrator
Staff member
The plastic connectors have a number of problems. Firstly they are falling apart, so have very inconsistent behavior after a few trials. They also tend to push the columns apart, rather than hold them together - as the bend tends to spread outwards. Overall some interesting results so far, but difficult to build larger models, and difficult to be consistent and reusable.

So I'm contemplating moving to magnets. specifically using rare earth magnets in the "floors", and either matching magnets in the "walls", or just steel screws. Here's my initial thoughts:


Ideally I'd like to do an 8 foot high double sided model. That's four two foot sections of three floors on each side, so 12 floors high, 24 floors total, 4 magnets per floor, so 96 total magnets.

Maybe something like the K&J Magnetics D61?
https://www.kjmagnetics.com/proddetail.asp?prod=D61
20160317-105608-3tfgn.jpg


How to attach? Hot glue? I should probably make sure the magnet is strong enough first.
 

Oystein

Senior Member
How to attach? Hot glue? I should probably make sure the magnet is strong enough first.
...and also not TOO strong.
Don't hot glue. Magnets don't like high temperatures.
Clip them to the floors - I am imagining a pocket made of a thin strip of iron stapled or screwed to the wood. You wedge the magnet between strip and wood. That of course decreases the magnetic force on the outer surface of that pocket.
 

Mick West

Administrator
Staff member
Quick test results:

The screws alone are not strong enough. Using brackets makes it much stronger.

This is looking promising, although my concern is that the floor will pivot instead of failing at all four connectors, and hence end up just dangling
 
Last edited:

Mick West

Administrator
Staff member
More progress - A better option for the metal contacts is "mending plates" - in this case 3" by 0.75" strips of steel with four countersunk screw holes. Securing them with two screws (#6, 3/4") in the outer holes gives a nice flush surface to put the magnets on. Downside being I need a lot (96 for the 8 foot high two sided model) and they are a tad spendy at Home Depot, so I'll have to bulk order them somewhere cheap. Pivoting does not seem to be much of an issue when the load is applied suddenly and/or evenly.

Video to follow, while I make a three floor model.
 

Mick West

Administrator
Staff member
So here's how the mending plates work out:

Works great in some regards. It's very good at supporting about the right amount of weight (From NIST, one floor should support 12 if gradually applied, or 6 if suddenly applied).

However, a few new problems arise. Tolerances are now more important, and I need to ensure all the floors are exactly the same length, otherwise one set of magnets will not make perfect contact. This is essentially a scale problem.

So I need to sort out that problem before proceeding (hopefully just jigging the chop saw to give me exact lengths).

The extra stability might allow for even thinner "outer walls", and possible a less excessively stable core.
 

econ41

Senior Member
Mick
This is an outstanding effort demonstrating "ROOSD" - the pancaking or runaway collapse of the Open Office Space or Outer Tube.

Just a couple of cautions about where this model fits relative to recent discussions of claims.

The first comments are serious technical suggestions as to caution.

As far as ROOSD - outer tube - collapse is concerned you started with an understanding of the actual WTC "ROOSD" mechanism and modelled that. We should note that this sequence reverses the logical sequence that Cube Radio and aka are insisting on viz - you cannot understand it (rather "it cannot be understood") without a model. You understood it before you modelled it AND could not have modelled it before you understood it.

Recall also my comments in previous thread(s) that there are two classes of models viz:
"Demonstrations" which explain the mechanism for those who lack the mental visualising skill to "see" it without a physical model; AND
"Scientific quantifiable models" which can provide measurable data for use in analytical calculations.

We are definitely in the arena of demonstrations - as are most of Coles models except the one "borrowed" from Bazant which does not apply to WTC collapses (Bazant was wrong to apply that model to WTC - and Cole was right to identify it as being erroneous.)

Now your first step of incorporating core collapse is not based on prior understanding of the actual 9/11 WTC mechanism for the core. And it skips straight to the final stage which is (Sort of) Euler buckling instability of the "spires. Sure it happened - lots of visual evidence BUT it skips over the actual "how it got there" causal processes. Which I have postulated as "beam strip down" and analogous in the core with ROOSD on the office space. In effect it is the same as your ROOSD model BUT starting at the last step with the floors removed - no dropping weight - and the perimeter columns ready to topple. The floors/weight dropping >> joint shearing are the key parts of mechanism. To be a valid model for core it needs to model those key steps.

So IMO your ROOSD demonstration is very good for it's purpose of demonstration because it models reality. For the core mechanism to be a valid demonstration it needs tki also model the same steps of reality - what removed the beams to leave the columns unstable?

I'll leave the technical details there.....
.....let's look at the process reasoning.

We are here partly because some people cannot visualise the mechanism without a physical model. (Or claim that they cannot visualise.) Despite differences in style both aka and Cube Radio have been asserting that they need the model.

Most of the rest of us insist that we have no problem visualising. So be it. Let's take it as true that some can and some cannot visualise.

The reality is that you and others of us CAN visualise. AND had to be able to visualise before you or we could build models. If you cannot visualise what you are building you cannot build it. So visualise comes before build. BUT once built it can assist the "non-visualisers" to "see" what happens - even tho by definition they could not build it for themselves.

Which actually proves some of the key points I have made many times elsewhere and here over recent months. These two will suffice for now:

1) The physics is elementary and can be easily visualised by many persons;
2) The building of a model neither proves nor disproves the reality of mechanism. The model can help those who cannot visualise to understand the mechanism. In that sense alone it could be regarded as "proving" the point for those persons. The rest of us don't need the model as "proof" - it is redundant.

I'll leave the topic there - congratulations on the progress to ROOSD stage and take care if extending into core.


AND - finally - a "tongue in cheek" comment for those with a sense of irony and humour.

Recall that the Coles models demonstrated arrest. Your model demonstrates"no arrest".

And we have been reminded many times that in the presence of many "arrests" one lone "no arrest" is extraordinary and needs extraordinary evidence to prove it.

So as it stands your model is unproven. We only have visual evidence of no arrest when all those other models did arrest.

So I've just debunked your assertion of no arrest - until you provide extra and extraordinary evidence.

We must see extraordinary evidence.... :rolleyes:
 

econ41

Senior Member
The screws alone are not strong enough. Using brackets makes it much stronger.

This is looking promising, although my concern is that the floor will pivot instead of failing at all four connectors, and hence end up just dangling
At risk of either boring or preaching to the choir:

You clearly understand what you are trying to model BEFORE you model it.

So the model is an aid to "demonstrating" what you already understand.
 

Mick West

Administrator
Staff member
So the model is an aid to "demonstrating" what you already understand.
That was the goal, to communicate this to other people. But it has also been interesting to observe things I'd not really though about - particularly how the lateral forces operate. There's the obvious "squeezing" of the outer walls as a large and increasing mass hurtles down the inside, but then there's also the way the floors sometimes pivot about one end, and then the greater outwards forces when the mass reaches the ground level.

And while I think I understand the basics, it's always good to verify. Like if a floor can support a dynamic load of six floors and you lightly drop seven on it, then how will the collapse progress? (at a significant fraction of g), will there be "jolt" in the falling mass (no, there's a series of impacts which spread the jolt).

So I think I'll come away with a better understanding of what happened (or the range of things that might have happened). And hopefully be able to explain it to people using this model.
 

econ41

Senior Member
That was the goal, to communicate this to other people. << Understood - and fully aligned with my own interests in "explaining". But it has also been interesting to observe things I'd not really though about - particularly how the lateral forces operate. << Yes - my own "engineer's gut feeling" says vertical vectors dominate and horizontal are real but second order. Just what you are saying but in my words. There's the obvious "squeezing" of the outer walls as a large and increasing mass hurtles down the inside, << Yes but then there's also the way the floors sometimes pivot about one end, << Again my "gut feeling" but take care. Your "floors" are orders more rigid and stronger than the actual floors. I doubt that - in the real event(s) - "pivot about one end" would win over "fail due to massive vertical impact load of debris" shattering the decks and folding up the joists. More simply put the floors would break up before they pivoted as intact entities. Gut feeling - don't ask me to prove it. and then the greater outwards forces when the mass reaches the ground level. << sure but the war is long over by then. All the points of contention "truther v debunker" are long past.

And while I think I understand the basics, it's always good to verify. << sure - and don't over concern about my cautionary points. It is a very good exercise of modelling. One on the best "on target" examples I've seen - if not THE best. "On target" in that it is an accurate analogy of the ROOSD process. Like if a floor can support a dynamic load of six floors and you lightly drop seven on it, << I think I explained or at least identified the distinctions "static" v "suddenly" v "dropped to impact" IF you "drop" 7 it will hit with more than 14 - probably a lot more. It would be 14 if zero "drop" - held in contact THEN released. There is the secondary issue that the margin will always be unpredictable - 6.1 load on a 6 will near certain not fail. Neither can you guarantee that 7.0 will fail 6....too many variable factors in the boundary definition. Remember that even without explicit "Factors of Safety" the 6 is intended to be safe. So actual failure could need 7 or 8 or 9 or even 12. (Side issue but designing something to fail if over 6 is fundamentally different to design to be safe up to and at 6...a lot of engineers get confused when they need to design a mechanical or structural "fuse" that WILL fail to prevent overload damage.....I'll also take the rain check on that bit for now. :rolleyes:)

then how will the collapse progress? << That is a separate issue - two separate stages "causing failure" and "what happens AFTER failure" - another "rain check". (at a significant fraction of g), << Actually at G for the scenario described until some other factor intrudes. will there be "jolt" in the falling mass (no, there's a series of impacts which spread the jolt).<< The whole issue of "Jolts' is confused because both truthers led by Szamboti and most debunkers have "fallen" for the false premise from Bazant & Zhou 2002 - the false bit of "falling to make impact == a jolt" - the real WTC event didn't "fall" as in "drop through a gap" - it "scrunched" AND it was BEFORE what we are discussing. So another more complicated topic which I have referred to many times in several forums and another rain check at this stage here.

So I think I'll come away with a better understanding of what happened (or the range of things that might have happened). << You sure will. I and others will also befit so thank you. And hopefully be able to explain it to people using this model. << I will use your model as my #1 choice when I have the need for such a demonstration and the other party is one who needs visual physical models.
 
Last edited:

deirdre

Moderator
Staff member
And while I think I understand the basics, it's always good to verify. Like if a floor can support a dynamic load of six floors and you lightly drop seven on it, then how will the collapse progress?
i personally would nix the drop altogether. when people say "the floors would arrest", it seems (based on all the examples given anyway and cole's models that did arrest), they understand the initial falling mass will wipe out some floors.

like in this vid- timestamp set.. only like 15 seconds to view

the ONLY reason the ball arrested is because the floors COULD hold the weight of the ball and glass that had fallen. Once the "umph" of the fall petered out, the floors arrest. i think the drop is adding to the confusion (i'm just guessing of course)

That, to me is the first and primary misunderstanding about the cole models. even if i discount the un-TT set up... IF he had dropped a weight his floors couldnt hold, AND his floors were heavy enough to ADD the appropriate weight to the next level before the drop "umph" gave out.. his model would have collapsed.
 

Mick West

Administrator
Staff member
Last edited:

Jeffrey Orling

Senior Member
Mick,,,,
I am well aware that a physical model cannot reproduce everything going on during the actual collapse. ROOSD, as Ozzie, explains is not terribly hard to conceptualize without a model... and indeed to create a model one needs to have done just that.

I was thinking if you can add more features of the real collapse to your model... again not duplicating but demonstrating real world.

Ozzie pointed out as I had that the core needs to show a strip down much like the facade side. What I think you can achieve modifying your components:

ejections of material through the facade
collapse of the core

Modifications:

Facade panels should have slits or openings for "debris" to exit during collapse (maybe 2 or three facade columns connected by paper or cardboard spandrels
Add contents to the floors... random objects including some plaster powder, ice cream sticks, tooth picks, sugar cubes, cereal, rice, legos etc.
clear plexiglass front and real panels... to "sandwich" but no touch the floors and columns
smaller square core columns connected by lateral beams with magnets on the OOS side (perhaps 1x1)... a grid of 4 might do it... bracing between with one magnet at the end of the bracing... maybe no floors even required on core floors (elevator shafts) core bracing on last two rows was parallel to the long axis and attached to the outside of the columns. see attached.

Drop floor section which is wide enough to impact inside the core beams AND the OOS floors. the bracing of the core should not see this first impact... but they may collapse from jostling of the columns. Or they may remain as "ladders"...

The plexi will confine the contents and the collapse to be in one direction - out the spaces in the facade openings (where the glass windows would have been).

This is a delicate set up with a messy result.... but I could model more collapse behavior. Maybe You have to get the sizes, and magnets just right

I would have the live and dead loads more equal on each floor. The plaster is messy and optional.. but I bet it would show the lateral puffs.

Your work is excellent!!!!!!!!!!!!!!

BTW I think we should all toss some money to Mick for the materials in these experiments.
 

Attachments

Last edited:

Robert S

New Member
This is my first post on this board so I hope I'm not stepping on any toes. My first thought was that a vertical offset would reduce rotation of the floors, but rotation doesn't seem to be an issue. My next thought is that you might want to look at the orientation of the magnets. Having the magnets in the same orientation should require less force than alternating orientation. (The fields should reinforce if on a backer plate and one North, one South pole are connecting to the column)

Recessing the magnets so they are flush into the board or the column would help protect them as rare earth magnets love to break off tiny little shards on impact, but if you don't have the tools replacing broken magnets might be easier.
 

Mick West

Administrator
Staff member
My next thought is that you might want to look at the orientation of the magnets. Having the magnets in the same orientation should require less force than alternating orientation. (The fields should reinforce if on a backer plate and one North, one South pole are connecting to the column)
I will experiment with this to see if it makes a difference.

Recessing the magnets so they are flush into the board or the column would help protect them as rare earth magnets love to break off tiny little shards on impact, but if you don't have the tools replacing broken magnets might be easier.
I have the tools (drill press and 1/2" Forstner bit), but I've not had any breakage problems yet. Also I think metal on metal (for the magnets) provides a more consistent friction than wood/metal - so I'd have to have the magnets slightly proud of the surface (also needed to avoid wood/screw-head interaction), and that all adds another layer of extreme precision required, and hence potential for errors. With my recent order I'll have 20 spare magnets, so breakage should be covered.
 
Last edited:

Mick West

Administrator
Staff member
In case other people are experimenting: while at Home Depot I picked up some cheap ceramic magnets, and some adhesive magnetic strips (intended for use a backing for fridge magnets).

Neither of these was at all useful - far to weak to hold any weight. You really need neodymium magnets.

Neodymium comes in a variety of strengths:
https://www.kjmagnetics.com/neomaginfo.asp
I'm using 1/2" x 1/10th inch N35 magnets, the weakest (and cheapest), but still about 20x as strong as the ceramic magnets.

Using 1/2" N35 is also probably better than using a smaller N52, as it has greater surface area, and hence friction.
 

Robert S

New Member
Without engineering a weight activated release mechanism the magnets are probably the best, reusable release. I might sketch up an mechanical adjustable release mechanism as figuring out how to fine tune the magnetic system is beyond me. Perhaps notching the mending plate to control field strength?
 

Mick West

Administrator
Staff member
Without engineering a weight activated release mechanism the magnets are probably the best, reusable release. I might sketch up an mechanical adjustable release mechanism as figuring out how to fine tune the magnetic system is beyond me. Perhaps notching the mending plate to control field strength?
Jeffery suggested earlier adding a layer or so of tape to adjust things. This would change both the magnetic force and the friction, so might be fiddly. However I don't think it's necessary, the current magnet/plate/floor setup give almost exactly the right support per floor (about nine).

And regarding "6 gradually applied, and 12 suddenly applied", there's really no way of "gradually" applying floors (other than the obvious of one-at-a-time) the scale factors here mean that the slightest touch on the model is often equal to many tons of force at full scale. Image a 1000 foot tall human attempting to gently assemble the real WTC towers.
 

Mick West

Administrator
Staff member
So I re-sawed the floors to all exactly the same length (now 11.75 inches). This did help, as there was considerable variation. There's still some issues with the magnets aligning, as the "walls" are not perfectly straight. There's a few ways around this, depending on your resources.
  1. Buy material for "Walls" that is more likely to be flat, as it's made that way - MDF for example
  2. Plane or joint the walls so they are flat
  3. Use only two floors per wall segment instead of three, so the wall can shift into position (with three, there's always one out of line if the wall is not flat.
I'm going with #2, as I have a jointer.
 

Mick West

Administrator
Staff member
Using the magnets opens up the possibility of having even thinner outer walls. Unlike the "bent laminated paper" connectors the magnets give noticeable (and realistic) resistance to turning forces (essentially a moment resisting connection). This gives the structure a lot more stability. I won't be able to experiment much until next week, but here's a fairly stable thin outer wall, just using every other floor (and just weak magnet-screw connection for the outer wall). So things are looking very promising for a tall stable structure with thin outer walls.

20160318-110524-66yn4.jpg
 
Last edited:

Mick West

Administrator
Staff member
One thing that's less than realistic is the simple balancing of one column on another. I'd used some strips of plastic as "splice plates" but it was not very satisfactory. A better solution is again to use magnets. Metal plates are put on either side of the join, an then they are connected via an other plate (the "splice plate") with magnets providing the connection again.

This works very well, allowing the structure additional simple wind and sway (earthquake) resistance, but still allowing the column joins to fail realistically with lateral displacement.

 

Mick West

Administrator
Staff member
Thinking about the core, which I currently have modeled as a solid single column, when really it was a latticework of girders around a number of relatively skinny core columns.

20160319-111122-i28yf.jpg


The core itself being more that one third the width of the building. Note the "tiny" core columns here:
20160319-111440-nvbmi.jpg


I'd like to incorporate something into the model that reflects this. My initial though is to do away with the 4x4 (only 3.5 inches wide) and replace it with a 6"x3.5" structure made of four corner columns that are as thin as I can manage, but held together with "girders" (similar to the floors, but just skinny 1x0.5 boards, or similar). All joined up with the magnet/plate system.

The idea being that the falling structure will damage the "core", both by taking out some girders, and by lateral forces on the column.

The core will still be unrealistically strong. I might make things more realistic by adding in some 12" sections of rebar

This all would likely add quite a bit to the complexity, time, and cost of the build, so I'll initially do the simple core (which will demonstrate everything except core collapse), and perhaps the more complex core later.

Maybe I should start a GoFundMe page to raise money for a larger more complex model?
 

econ41

Senior Member
Thinking about the core, which I currently have modeled as a solid single column, when really it was a latticework of girders around a number of relatively skinny core columns.

View attachment 18137

The core itself being more that one third the width of the building. Note the "tiny" core columns here:
View attachment 18138

I'd like to incorporate something into the model that reflects this.
Mick the goal or drive for model accuracy is admirable BUT it could well be a "Bridge Too Far".

Recall my earlier comments - this activity is clearly separating us into "Those who can visualise what really happened" and "those who cannot visualise reality". Not a criticism - it is reality.

"Those who can visualise" includes you and all of us who are commenting on how well the model matches reality. Bottom line being obvious - we understand reality THEREFORE can "plan and construct the model' -- you or comment on how accurate it is -- most of us.

Whilst "those who cannot..." and who may genuinely befit from the demonstration that the model provides. I have commented at length on the limitations of the latter process in the earlier thread.

So the two questions facing you are:
1) How accurately can I model it?
2) So that it helps understanding of the actual WTC collapse.

The OOS collapse is simple because it involved ONE clear span between TWO columns - relatively easy to model -- near enough -- and quite good replication of the real event.

BUT the Core was NOT 4 columns - it was a grid of multiple columns. AND although the collapse mechanism was analogous to the OOS collpase - viz overwhelming falling weight shearing off the floor connectors - for the core it happened in every one of the multiple cells - falling horizontal beams hitting other horizontal beams and shearing them off. (Excuse the "global assertion" of "every" - proof if needed - the logic is the same with "nearly every". :rolleyes: )

To model that with any legitimacy as a demonstration I suggest that the minimum is a 4x4 array of columns leaving 3x3 spans of beams to shear. Feasible in the arena of physical modelling. BUT less accurate than the first stage model of OOS pancaking. AND not as persuasive for those who need persuasion. And probably useless for those who may be in to nit picking denial

Which brings us back to "What is the goal". Sure there is satisfaction in doing it and being close enough to the approximation.

BUT I doubt much benefit for those who cannot or do not or will not visualise. Because the mental translation from model to reality is a much steeper curve.

My initial though is to do away with the 4x4 (only 3.5 inches wide) and replace it with a 6"x3.5" structure made of four corner columns that are as thin as I can manage, but held together with "girders" (similar to the floors, but just skinny 1x0.5 boards, or similar). All joined up with the magnet/plate system.

The idea being that the falling structure will damage the "core", both by taking out some girders, and by lateral forces on the column.

The core will still be unrealistically strong. I might make things more realistic by adding in some 12" sections of rebar
Noted - further comment reserved at this time.
This all would likely add quite a bit to the complexity, time, and cost of the build, so I'll initially do the simple core (which will demonstrate everything except core collapse), and perhaps the more complex core later.
It is a good stage to "pause, regroup, rethink, reconsider."
Maybe I should start a GoFundMe page to raise money for a larger more complex model?
I for one would be amenable to such a process - provided the purpose/goals are clearly stated.
 

Mick West

Administrator
Staff member
BUT the Core was NOT 4 columns - it was a grid of multiple columns. AND although the collapse mechanism was analogous to the OOS collpase - viz overwhelming falling weight shearing off the floor connectors - for the core it happened in every one of the multiple cells - falling horizontal beams hitting other horizontal beams and shearing them off. (Excuse the "global assertion" of "every" - proof if needed - the logic is the same with "nearly every". :rolleyes: )

To model that with any legitimacy as a demonstration I suggest that the minimum is a 4x4 array of columns leaving 3x3 spans of beams to shear. Feasible in the arena of physical modelling. BUT less accurate than the first stage model of OOS pancaking. AND not as persuasive for those who need persuasion. And probably useless for those who may be in to nit picking denial
It was not four columns, but it certainly was not one solid column either. A 2x2 grid (i.e. the columns at the corners) is just the very simplest change that will allow for some mode of failure of the core other than being pushed over.

Still, I am unlikely to jump right into modeling the core. I'll see what happens with the 8' model when my magnets arrive.

I'd also be very interested in hearing feedback from people who tend to side with AE911. Do they see problem with my model? Do they think it demonstrates something, or nothing? Since the primary function here was to demonstrate A) that a tall structure can collapse rapidly if the upper portion is dropped on it, and B) roughly how the WTC towers collapse progressed, then I'd like to see how people receive it. @Cube Radio ?
 

Jeffrey Orling

Senior Member
I think Ozzie makes some good points... I don't need to see physical models to visualize the collapse... of course no one can conceptualize the DETAILs and sequence of events... tens of thousands of the in a few seconds....

But you CAN demonstrate NOT DUPLICATE the mechanisms in a model for those who can conceptualize the main failure mode features.

I believe you have done the OOS.... but that can be improved. You can add live loads... reduce the floor plate thickness and mass... but make the total floor load D+L the same and the OOS will collapse as you show. Use 3/8" MPF for the for plates. In fact you can use MDF for the entire model if you can get a good precise table saw and radial arm saw. They also have ultra light which you can use for the columns.. I believe the materials come as thick as 1 1/8"

You can increase the foot print but make it like a slice of an ant colony if you will... but adding plexi sides to make the collapse look like a cross section... confining the action between the plexi. You can can use exterior panels with some sort of spandrels... magnet connectors... and openings for the windows which floor debris might /should come out of (maybe).. certainly powder/ dust might.

You can make the core 9 smallish square columns with long bracing connected by magnets to the 3 columns at the OOS sides and then short braces between the columns with only the center 3 bays having floors (the others are shafts). Lotta magnets... Could they work with a magnet and a screw without a mending plate??????

I can make some drawings for you can will chip in for the materials.
 
Top