That's probably the topic for a new thread.
Creating A Physical Model of the Initiation of the collapse of the World Trade Center Twin Towers
- Thread starter Mick West
- Start date
Jeffrey Orling
Senior Member
I agree... The problem would be that the model would require fire/heat to initiate the "changes" in the structure causing top to collapse. So I suppose the columns could be maybe candles (not lit) and some sort of fire was started which leads to melting of the candle columns... buckle and the floors they supported collapse... Maybe
FatPhil
Senior Member.
Dude, I've only been here a couple of months and I've already seen people nuking toilets - for science! I have faith in the creativity hereabouts.
Solder's weak, in particular for its weight, and melts at a lowish temperature, perhaps that might be a useful construction component. Current from electricity, just a 9V battery (which can double as ballast) can probably generate enough heat to make solder fail without the need for any open flame, and also can be put on a switch to ensure everything's safe before going live.
Last edited by a moderator:
One challenge would be sudden onset. The WTC collapses were characterized by a gradual shifting of loads between columns. As a column became weaker from heat, or was pulled pout of line by sagging floors, the load it carries was shifted by the stiffness of the upper structure (including the hat truss) to the other columns.
At some point, that load transfer became unsustainable, as loads shifted then more columns failed as there (simplifying) wasn't enough left intact to carry the entire weight of the building.
So you'd need a stiff upper structure supported by a multitude of columns, so that if columns are removed then the upper structure does not move.
At a VERY simple level, this could be a stiff sheet of MDF, just laid on top of the columns, like a board on a bed of nails.
As long as the upper structure is resting on a reasonable number of columns, then it's not going to tip over. In theory, you can keep removing columns and at some point all the remaining ones will fail rapidly, leading to a bit or tipping, then straight down collapse.
This has failings as most models do, because of scale. But if you have something like 50 columns that all support 1 lb each, and you load it with 30 lb, then you gradually removed 20+ columns, then you should get sudden collapse initiation.
A more accurate version with sagging floors pulling in the outer walls, would be challenging.
econ41
Senior Member
I had already responded to Jeffrey's question in the original thread. The same "big picture" questions arise as have done on other discussions of this nature -- i.e. modelling Twin Towers collapses. Some of those questions are:
1) What is the purpose of the model - the two main reasons are probably (a) for the fun and challenge of doing it and (b) to somehow help "convert" those who may not understand. The two may overlap;
2) Which "style" of model do we intend - (c) A model which "looks like" the real event in order to demonstrate some mechanism OR (d) a more rigorous engineering model which may be quantitatively closer to reality but not as accurate a "look like" because of the need for scaling distortions.
3) Are we trying to mimic the whole of "initiation" or separate examples of sub mechanisms.
4) (Some others which will arise as discussion progresses...)
Whichever of those options we pursue experience in similar discussions says we will need to clearly define objectives, not let them drift unless for reasons of clear necessity explicitly recognised. AND at each stage we will need to be assured that we do understand the bits of mechanism and principles of physics we are trying to demonstrate.
There will be at least two issues which probably cause problems:
(i) We cannot know the exact sequence of structural member failures. My own hypotheses use arguments which refer to "The first column to fail" and "The second column" etc etc. The logic perfectly valid BUT not easy if even possible to model without knowing "which SPECIFIC column" was first, second, "n"th. And the source of much frustration over many years with technical persons whose thinking "locks up" when they don't know which specific column.
(ii) Do we take the NIST version as the "bible"? I normally don't for reasons of preferred debating tactics. Not relevant at this stage.
In my own efforts at teaching Twin Towers collapse physics to persons who wanted to learn I have used a set of diagrams which make step by step building blocks to explain sub sets of mechanism. The driving feature of Twin Towers "initiation" IMNSHO is the cascading failure of columns triggered by heat weakening and progressing via load re-distribution. (The NIST trigger of joist sag pull in of perimeter is a valid sub-set. The net result of inwards bowing is loss of load carrying capacity by the affected columns.) So some how any model of initiation would need to mimic cascading (sequenced) failure. Do we also need to mimic "heat triggered"? More generally how far does realism need to go?
This is one of my diagrams from the middle of the sequence and used to explain heat triggered cascading failure by load redistribution.
It would be quite easy to model BUT is a long way short of being a full model. And the full explantion needs about a dozen of those diagrams.
Posted as a contribution to opening debate.
1) What is the purpose of the model - the two main reasons are probably (a) for the fun and challenge of doing it and (b) to somehow help "convert" those who may not understand. The two may overlap;
2) Which "style" of model do we intend - (c) A model which "looks like" the real event in order to demonstrate some mechanism OR (d) a more rigorous engineering model which may be quantitatively closer to reality but not as accurate a "look like" because of the need for scaling distortions.
3) Are we trying to mimic the whole of "initiation" or separate examples of sub mechanisms.
4) (Some others which will arise as discussion progresses...)
Whichever of those options we pursue experience in similar discussions says we will need to clearly define objectives, not let them drift unless for reasons of clear necessity explicitly recognised. AND at each stage we will need to be assured that we do understand the bits of mechanism and principles of physics we are trying to demonstrate.
There will be at least two issues which probably cause problems:
(i) We cannot know the exact sequence of structural member failures. My own hypotheses use arguments which refer to "The first column to fail" and "The second column" etc etc. The logic perfectly valid BUT not easy if even possible to model without knowing "which SPECIFIC column" was first, second, "n"th. And the source of much frustration over many years with technical persons whose thinking "locks up" when they don't know which specific column.
(ii) Do we take the NIST version as the "bible"? I normally don't for reasons of preferred debating tactics. Not relevant at this stage.
In my own efforts at teaching Twin Towers collapse physics to persons who wanted to learn I have used a set of diagrams which make step by step building blocks to explain sub sets of mechanism. The driving feature of Twin Towers "initiation" IMNSHO is the cascading failure of columns triggered by heat weakening and progressing via load re-distribution. (The NIST trigger of joist sag pull in of perimeter is a valid sub-set. The net result of inwards bowing is loss of load carrying capacity by the affected columns.) So some how any model of initiation would need to mimic cascading (sequenced) failure. Do we also need to mimic "heat triggered"? More generally how far does realism need to go?
This is one of my diagrams from the middle of the sequence and used to explain heat triggered cascading failure by load redistribution.
It would be quite easy to model BUT is a long way short of being a full model. And the full explantion needs about a dozen of those diagrams.
Posted as a contribution to opening debate.
Last edited:
econ41
Senior Member
econ41
Senior Member
So we already have a couple of possible paths for research. Two which suggest a model relying on heat triggering. Jeffrey Orling suggests a heat triggered failure modeled by using candles as columns:
A couple of obvious problems - the "A" and "B" candles are likely to melt rather than simply weaken. I don't know the shape of the graph for the transition from "stiff" through "weakening" to "molten" for candle wax. I suspect the transition is too fast and "melting" has obvious problems of being misunderstood or misrepresented by truthers. PLUS how do we sufficiently "warm" the next candles "C" >> "D" >>> etc so the sequence progresses? I don't think candles will be sufficiently elastic to yield allowing a reasonable mimicking of load-redistribution. But "It is early days."
So let's merge a couple of ideas as a way forward FOR DISCUSSION - far too early to start building methinks.
Why not a Mick's layout physical model with a few candles as the columns. (6-7-8 I think enough to prove the point.) Some sort of spread flame torch to heat the "A" and "B" columns as per my diagram. (For those who are racing ahead with the physics - I made the layout 7 equal spaced columns to simplify any "ball park" math guestimations and heated the two because even with rough maths it gives enough load redistrib from failure of A and B to fail C as next in line. It may not work with candles.)
Then - in my sequenced explanations of "cascading failure driven by load re-distribution" the immediate preceding diagram in the sequence of my diagrams at least partially mimics Mick's "remove columns" suggestion.
"
Excuse the reference to "explosives"...
Neither the "candles" nor the "remove columns" models accurately mimic the "3D plus two layers of "T" of the real event but I doubt we will get much closer even with this simple approach. Plus "candles" is one level closer to mimicking the real event - it could be preferred if we can make it work. And we are several dimensions short of mimicking the full complexity of the actual 9/11 "initiation" stage. Isn't mentally visualising a darn sight easier than physical modelling...
And for those who are curious - the reference to "Pivot or Fulcrum" was the main point of explanation in my original uses of the diagrams. IF "A" and "B" are sufficiently weakened so that they shed load the EXTRA load on "C" is MORE than the total of the loads released by "A" and "B". How much more depends on the stiffness of "Top Beam"...
FatPhil agrees the type but opens the door for other heat weakening materials:
Neither Jeffrey nor FatPhil at this stage suggest a physical layout. Mick does suggest an initial idea for physical layout of a model - essentially the same as my own diagram - but suggests removing columns as the approach to cascading failure:
This is the diagram I used for explanation/teaching. With "heat triggering". It could be made to sort of work using candles.
A couple of obvious problems - the "A" and "B" candles are likely to melt rather than simply weaken. I don't know the shape of the graph for the transition from "stiff" through "weakening" to "molten" for candle wax. I suspect the transition is too fast and "melting" has obvious problems of being misunderstood or misrepresented by truthers. PLUS how do we sufficiently "warm" the next candles "C" >> "D" >>> etc so the sequence progresses? I don't think candles will be sufficiently elastic to yield allowing a reasonable mimicking of load-redistribution. But "It is early days."
So let's merge a couple of ideas as a way forward FOR DISCUSSION - far too early to start building methinks.
Why not a Mick's layout physical model with a few candles as the columns. (6-7-8 I think enough to prove the point.) Some sort of spread flame torch to heat the "A" and "B" columns as per my diagram. (For those who are racing ahead with the physics - I made the layout 7 equal spaced columns to simplify any "ball park" math guestimations and heated the two because even with rough maths it gives enough load redistrib from failure of A and B to fail C as next in line. It may not work with candles.)
Then - in my sequenced explanations of "cascading failure driven by load re-distribution" the immediate preceding diagram in the sequence of my diagrams at least partially mimics Mick's "remove columns" suggestion.
Excuse the reference to "explosives"...
Neither the "candles" nor the "remove columns" models accurately mimic the "3D plus two layers of "T" of the real event but I doubt we will get much closer even with this simple approach. Plus "candles" is one level closer to mimicking the real event - it could be preferred if we can make it work. And we are several dimensions short of mimicking the full complexity of the actual 9/11 "initiation" stage. Isn't mentally visualising a darn sight easier than physical modelling...
And for those who are curious - the reference to "Pivot or Fulcrum" was the main point of explanation in my original uses of the diagrams. IF "A" and "B" are sufficiently weakened so that they shed load the EXTRA load on "C" is MORE than the total of the loads released by "A" and "B". How much more depends on the stiffness of "Top Beam"...
Last edited:
Jeffrey Orling
Senior Member
This could be an interesting discussion.
Several years back in my mental gymnastics to understand how the top section would come down as it did presuming the main driver was heat being applied to a structure which was damaged but not fatally. So this as a starting point tells us that there had to have been some amount of load redistribution immediately after the plane impact.
If we work with 1wtc... we know the plane penetrated the center of the north face and some of it emerged through the south face.
We could see some structure damages through the huge hole the plane made.
I made the following drawing to "model" how the core columns failure might have progressed. The strength of the columns is represented graphically by their area.
Basically the model suggests that the core column destruction on the floors of the plane strike and fires progressed south and west 'hollowing" out the core.
It should be noted that although there were 47 core columns the major "lifting" was done by the 24 perimeter columns of the core... and the entire facade perimeter columns. The 23 central core columns MOSTLY framed vertical shafts of elevators... The spans were short and traditional steel grid framing with concrete floor was employed. Note... as the belt girder was "cantilevered off" the perimeter core columns... there were girders framed to rows 600 and 900 on beam stubs.
I am guessing that the release came with the antenna drop. That is hard to visualize but I will give it a try.
The 360 ton antenna need additional support than those 3 columns in the center of the core... they were actually the 3 "weakest" of all the core columns. I am guessing they had only 3 columns there because the large freight elevator was located in that location....
The engineers used the hat truss the support the antenna by laterally distributing the loads to other columns. The hat truss also provided a somewhat rigid 3 D space frame at the top of the towers.
I believe the upper mech floor likely had large loads over the entire foot print and these too were INSIDE the hat truss 3 D space frame. The hat truss was supported axially on rows 500, 700, 800, & 1000... in the east-west axis. and also on rows 1, 4, 5, & 8 in the north-south axis.
When the heat driven core column failure reached to the center of the core under the antenna the columns directly beneath... 704, 705 & 805..facilitated the antenna to drop through the roof and with it the hat truss was substantially destroyed. What might have happened:
a) The top block to break free and move laterally
b) The floors supported insider the hat truss collapsed down (in the core)
c) The floors outside the core in the upper block may have buckled and collapsed inward.
(a) we witness as the top block's descent
(b&c) we can imagine becoming the ROOSD driving mass
break... more to follow
Several years back in my mental gymnastics to understand how the top section would come down as it did presuming the main driver was heat being applied to a structure which was damaged but not fatally. So this as a starting point tells us that there had to have been some amount of load redistribution immediately after the plane impact.
If we work with 1wtc... we know the plane penetrated the center of the north face and some of it emerged through the south face.
We could see some structure damages through the huge hole the plane made.
I made the following drawing to "model" how the core columns failure might have progressed. The strength of the columns is represented graphically by their area.
Basically the model suggests that the core column destruction on the floors of the plane strike and fires progressed south and west 'hollowing" out the core.
It should be noted that although there were 47 core columns the major "lifting" was done by the 24 perimeter columns of the core... and the entire facade perimeter columns. The 23 central core columns MOSTLY framed vertical shafts of elevators... The spans were short and traditional steel grid framing with concrete floor was employed. Note... as the belt girder was "cantilevered off" the perimeter core columns... there were girders framed to rows 600 and 900 on beam stubs.
I am guessing that the release came with the antenna drop. That is hard to visualize but I will give it a try.
The 360 ton antenna need additional support than those 3 columns in the center of the core... they were actually the 3 "weakest" of all the core columns. I am guessing they had only 3 columns there because the large freight elevator was located in that location....
The engineers used the hat truss the support the antenna by laterally distributing the loads to other columns. The hat truss also provided a somewhat rigid 3 D space frame at the top of the towers.
I believe the upper mech floor likely had large loads over the entire foot print and these too were INSIDE the hat truss 3 D space frame. The hat truss was supported axially on rows 500, 700, 800, & 1000... in the east-west axis. and also on rows 1, 4, 5, & 8 in the north-south axis.
When the heat driven core column failure reached to the center of the core under the antenna the columns directly beneath... 704, 705 & 805..facilitated the antenna to drop through the roof and with it the hat truss was substantially destroyed. What might have happened:
a) The top block to break free and move laterally
b) The floors supported insider the hat truss collapsed down (in the core)
c) The floors outside the core in the upper block may have buckled and collapsed inward.
(a) we witness as the top block's descent
(b&c) we can imagine becoming the ROOSD driving mass
break... more to follow
Jeffrey Orling
Senior Member
How the heat might have destroyed the core's integrity.
I suggest we consider that the "destroyer" of the core's integrity were the lateral beams inside and bracing the core. These beams likely expanded from heat and as such would push the columns the were framed into. Columns in a grid are restrained in both axes... but when a column has been destroyed... such as from the plane strike... the heated beams can push the columns that have lost bracing because they have lost some of their restraint. So these columns might be pushed inward by the heated beams.... and they might laterally translate the column to column connections reducing the bearing area for the columns above. This might cause buckling.. especially if they are weakened by heat. Note these columns were no more than 1/2" thick webs... and there was no fire suppression. So the core's central columns might have rapidly buckled... leaving the strong perimeter columns without some of the bracing they require.
My guess is that the upper block was rapidly hollowed out leaving the unstable box of the perimeter with no core side floor plates to brace it.. and sufficient lateral forces to break it free from the lower section.
I suggest we consider that the "destroyer" of the core's integrity were the lateral beams inside and bracing the core. These beams likely expanded from heat and as such would push the columns the were framed into. Columns in a grid are restrained in both axes... but when a column has been destroyed... such as from the plane strike... the heated beams can push the columns that have lost bracing because they have lost some of their restraint. So these columns might be pushed inward by the heated beams.... and they might laterally translate the column to column connections reducing the bearing area for the columns above. This might cause buckling.. especially if they are weakened by heat. Note these columns were no more than 1/2" thick webs... and there was no fire suppression. So the core's central columns might have rapidly buckled... leaving the strong perimeter columns without some of the bracing they require.
My guess is that the upper block was rapidly hollowed out leaving the unstable box of the perimeter with no core side floor plates to brace it.. and sufficient lateral forces to break it free from the lower section.
econ41
Senior Member
Why speculate?
If "we" are going to model something we need to be clear about both "Why?" and "What?"
Why? >> what purpose are we trying to achieve - is it just for fun OR as a tool to help others understand. << THOSE parts of the challenge are unchanged from previous "runs" of the same or similar themes". AND
"What?" >> We need to be specific about what mechanism or parts of mechanism we are seeking to model.
Both Mick and I have identified failure of columns to carry vertical loads in axial compression is A key feature. I certainly assert that it is THE key feature of initiation. And despite slight differences of description it is a central factor in most hypotheses. The NIST focus on floor joist sagging >> inward bowing results in the same feature >> Loss of axial vertical dwonwards load carrying capacity.
We should be able to agree that heat triggered cascading failures of columns under axial loading is a key component. Readily demonstrable that it is the major causal factor of "initiation" (And that reality doesn't change in principle if there was some assitance fro CD)
IF you or others funbdamentaly do NOT accept the extant hypotheses explaing HOW the Twin Towers collapses intited >> then we ar a long way from beingable to model anything. We need consensus on what we plan to model.
Last edited:
Jeffrey Orling
Senior Member
can't see much going on in the core.... so it's "speculation" regardless of much sense it may make.
econ41
Senior Member
But this is a new thread - established to discuss your suggestion about modelling the "initiation" of WTC Twin Towers collapses. You suggested "candes" as an apparent means of modelling heated columns.. or so I thought. So in effect the thread is YOUR OP.
Any chance you can agree the scope of discussion OR explicitly re-state what YOU see as the topic.
IF you prefer to deny the key part played by heat triggered cascading failure of columns I will be interested to see reasoned argument.
Jeffrey Orling
Senior Member
Nooo!! I DO see heat as the drive POST plane strike... The heat was "diddling" with the structure... And I believe this was principally in the core... Essentially the head cause beam expansion which caused them to push on the columns... those columns which had lost restraint on one side were able to be pushed to THAT side leading to loss of bearing between the pushed column and the one below. Columns were "restrained" the locations over their 36' length... where the floors were framed to the sides of columns via the bracing.
For the OOS floors to collapse a large portion of the belt girder would have to fail... and remember it was attached to the 24 outside core columns... so all the OOS floor loads on the core side were supported by the beam stubs connected the belt girder to the perimeter columns.
It seems possible that the heat caused distortion in the core could have also pulled or separated the belt girder at various columns such that it too was losing its ability to transfer OOS floor loads to the core columns... or they were being transferred to few perimeter core columns.
My hunch is that distortion led to loss of capacity of various core elements...
But the release seems to have been triggered by the mass of the floors and the mech equipment with the hat truss plus the dropping antenna. This was not a core collapse up there as much as an entire footprint inside the facade.... Unlike the office floors those mech floors had very heavy loads outside as well as inside the core. So the ROOSD mass may have come from those footprint wide mech floors collapsing. Those mech floor slabs were 6-9"thick and supported loads like water tanks... Elevator motors massive transformers.
econ41
Senior Member
@Jeffrey Orling
I refer to my previous question:
Do you deny that heat triggered cascading failure of columns was the key part of the mechanism?
AND do you want to base modelling on your core failure and beam expansion proto hypothesis?
The thread is at least nominally YOUR OP so if you want to change direction - I will respect your call.
Rejection noted.
So you agree the general principle BUT deny the specific focus I placed on cascading failure of columns?
I refer to my previous question:
Do you deny that heat triggered cascading failure of columns was the key part of the mechanism?
AND do you want to base modelling on your core failure and beam expansion proto hypothesis?
The thread is at least nominally YOUR OP so if you want to change direction - I will respect your call.
Jeffrey Orling
Senior Member
Things fall in building as in the twin towers because the columns are no longer holding up what they were supposed to... all loads make their way to columns and to the foundations....
The question really is how did the columns fail?
Or was the initiation like the ROOSD the collapse was the floors and contents which buster by the columns?
I am not sure... And it may be both!
Loads are not magically appearing. But they possibly might appear where they weren't designed to. Example.... if an area of a floor is carried by a steel beam which gets very heated and does one of two things...
a) it buckles and the floor and contents collapse down to the floor below... or...
b) the beam expands and gets pushed off its "seat" as we were told happened in 7wtc...the heated beams pushed girder off seat at col 79 and the whole shebang (floor around the column) collapsed.
We are told that after that happened the column bucked absent lateral bracing. That seems a bit old unless it was many floor collapses.
So.... I like the runaway floor collapse idea ....caused by heat displacing/expanding and weakening beams to the point of collapse.... leading to a large dynamic load to the floor below and that failing because.... it was over loaded... wash, rinse repeat.
I rather doubt heat weakened columns to the point of buckling. I find it makes more sense that this would happen to beams with fires below them and engulfed in super hot air. But I don't know....
There might be some clues in what was found in the other structures which burned out of control and had local collapses and beam failures like 6wtc I believe.
Heat was the driver... what how did it make things drop? Count the ways! Shakespeare.
The question really is how did the columns fail?
Or was the initiation like the ROOSD the collapse was the floors and contents which buster by the columns?
I am not sure... And it may be both!
Loads are not magically appearing. But they possibly might appear where they weren't designed to. Example.... if an area of a floor is carried by a steel beam which gets very heated and does one of two things...
a) it buckles and the floor and contents collapse down to the floor below... or...
b) the beam expands and gets pushed off its "seat" as we were told happened in 7wtc...the heated beams pushed girder off seat at col 79 and the whole shebang (floor around the column) collapsed.
We are told that after that happened the column bucked absent lateral bracing. That seems a bit old unless it was many floor collapses.
So.... I like the runaway floor collapse idea ....caused by heat displacing/expanding and weakening beams to the point of collapse.... leading to a large dynamic load to the floor below and that failing because.... it was over loaded... wash, rinse repeat.
I rather doubt heat weakened columns to the point of buckling. I find it makes more sense that this would happen to beams with fires below them and engulfed in super hot air. But I don't know....
There might be some clues in what was found in the other structures which burned out of control and had local collapses and beam failures like 6wtc I believe.
Heat was the driver... what how did it make things drop? Count the ways! Shakespeare.
Jeffrey Orling
Senior Member
this:
"Conclusions This paper examined the effects of fires on the behavior of multi-story steel-framed buildings. It can be concluded that, even in the context of the structural fire engineering approach of modern design codes, predictions of behavior based on furnace tests or numerical modeling of isolated members are unreliable. The behavior of the members within a continuous, compartmented structure is very different from the behavior of isolated members. Structural continuity, restraint to thermal expansion provided by the adjacent members, the beam-to-column joints, and the tensile membrane action of the composite slab have demonstrated a significant positive influence on the bevahior of the entire structure in the event of a fire. Data from the Cardington fire tests and the subsequent experimental and analytical studies provide fundamental information that is very important for researchers studying the performance of steel-framed buildings in a fire for many years to come so as to develop new design approaches that take into account the interaction between structural members in a fire. The developments that have already taken place in the past few years have been very significant in understanding the reality of structural behavior in a fire, which will undoubtedly lead to the emergence of new rational fire engineering design methods that may allow us to construct safer, more effective, and more efficient fire-resistant buildings in the future."
Plane destroyed integrity ie beam restraint... which let the heated beams expand and led to "expanding" local collapse.... which led to dynamic loading from collapsing floor sections.... and there you have it.
"Conclusions This paper examined the effects of fires on the behavior of multi-story steel-framed buildings. It can be concluded that, even in the context of the structural fire engineering approach of modern design codes, predictions of behavior based on furnace tests or numerical modeling of isolated members are unreliable. The behavior of the members within a continuous, compartmented structure is very different from the behavior of isolated members. Structural continuity, restraint to thermal expansion provided by the adjacent members, the beam-to-column joints, and the tensile membrane action of the composite slab have demonstrated a significant positive influence on the bevahior of the entire structure in the event of a fire. Data from the Cardington fire tests and the subsequent experimental and analytical studies provide fundamental information that is very important for researchers studying the performance of steel-framed buildings in a fire for many years to come so as to develop new design approaches that take into account the interaction between structural members in a fire. The developments that have already taken place in the past few years have been very significant in understanding the reality of structural behavior in a fire, which will undoubtedly lead to the emergence of new rational fire engineering design methods that may allow us to construct safer, more effective, and more efficient fire-resistant buildings in the future."
Plane destroyed integrity ie beam restraint... which let the heated beams expand and led to "expanding" local collapse.... which led to dynamic loading from collapsing floor sections.... and there you have it.
econ41
Senior Member
Yes. That is the question we are discussing. And SPECIFICALLY how did they fail in the "initiation" stage.. your OP, your suggestion of topic.
No it wasn't. "ROOSD" was part of "progression stage". We are discussing "initiation". And the mechanisms of both stages (or all three stages actually) were very different.
There is no reason to be unsure and it CANNOT be both.
We are not discussing WTC7 and we are not discussing the two examples you raise UNLESS you put them in the context of your alternate explanation of Twin Towers "initiation stage".
So you are disagreeing with the point I identify as the key feature of the Twin Towers "initiation" stage. Columns failed in axial overload as a consequence of heat weakening. AND it became a "cascading (sequenced) failure driven by load re-distribution". I'm prepared to prove my point - or as many points as needed to establish the hypothesis. BUT it means that we - or at least you and I - have some preliminary issues to resolve before we can contribute to discussion of modelling. We must agree what mechanism we need to model. OR we build (AKA "discuss building") TWO very different models.
I suggest there is more than enough information for the Twin Towers collapses for the single purpose we have - "Decide how the collapse initiated"
- Yes. That was the starting point of discussion;
- and that is still the question. Mick appears to assume column failure. I've taken it one step further but same base premise. Column failure in axial load carrying.
- AFAIK Shakespeare, W is not a member
and we have two competing hypotheses on the table. Pick one OR suggest a third option. And stand by for serious debate. 
Jeffrey Orling
Senior Member
@econ41...You seem to be ignoring or dismissing the role of the lateral bracing - beams and girders... and the entire hat truss in the initiation phase. The hat truss had diagonal members which transferred axial loads within the core and even to the perimeter.
My suggestion is that the lateral elements' heating led to the buckling of the axial structures. I referenced this "process" which NIST identified as a cause for column failure in 7wtc... the heating of beams and girders led to column failure. In 7wtc there was no claim that heat weakened the column sufficiently for it to fail. NO I am not changing the topic to a discussion of 7wtc.
One take away from the Caddington study was that when the steel beams and girders endured heat and in the case where there were columns which were "unrestained"... the beams and girder would push the columns leading to axial mis alignment and failure. The heat also weakened slabs so that they lost integrity and fractured as well.
I am suggesting that the way the heat caused the columns to fail was not simply from heat lowering their load bearing capacity to below demand. Axial capacity likely was reduced by the heat. I don't think in the time frame it reduced capacity enough to cause the buckling.
My hypothesis is that most simply stated... the lateral members expanded from heat, and forced mis alignments of the axial connections. And that led to inadequate bearing area .. And that led to web and flange buckling.
Look at col 704 which ran from flr 88-90:
The gray shows column 704 displaced by .75 inches in both the X and Y axis. We don't know how much it may have been displaced... or if it had been displaced... We can look at that later. But the diagram illustrated how the bearing area to the column ABOVE beginning at floor 91... might have its bearing area reduced to less than 4% ... This would surely cause web and buckling. I don't know how at what bearing area bucking would take place. I don't know how much and in what direction lateral displacement might have occured.
We do know that column to column connections were not restrained AT the column to column connections by beams. Lateral framing/bracing was at approximately 4', 16' and 28' from the bottom of the column. I believe this was to facilitate construction and the bolting of the column splices. The bolts of the splices would be the main lateral restraint.
I believe NIST discovered many failed/destroyed bolted connections in their study... pull out and shearing. This was also true at the truss seats in some cases.
Lateral Restraint of Core Columns
In a grid framing plan a column is laterally restrained in both the X and Y axis by beams/girders (regardless of whether the restraint is AT the column to column connection). We know that unbraced columns are weaker than braced columns according to the work of Euler. In fact, we were told that col 79 on 7wtc buckled when it lost lateral bracing.
So I return to what likely was going on in the core of the twin towers. Massive fires were heating all the steal under the slabs on the level of the plane strikes.... because of the fuel supplied by the jets almost full tanks. Those flames concentrated where the fuel fuel tanks of the jet were... in the case of 1WTC the center of the tower in a north-south swath through the building. Yes it did disperse and likely involve the entire foot print.
The Caddington study shows what happens to heated concrete slabs supported on steel beams. This is likely what was going on in those floors where the fires raged in the twin towers.
But unlike Caddington... 1wtc also had a number of columns severed and severely damaged by impact of the plane. My slide in a previous post suggested which columns might have been destroyed by the initial impact of the plane. I believe NIST suggested these same columns.
The OOS floor support on the core side was a belt girder which transferred the 80" OC load locations of the double floor trusses to the core columns which were spaced from 8' for rows 4 & 5 to over 20' for rows 1 - 4 and 5 - 8. The belt girder was supported on short cantilever beam stubs.
I am suggesting that the heated beams inside the core at the level and above the level where the plane destroyed columns and started massive uncontrolled fires... that the "frame" was being pushed by expanding beams... that the slabs were breaking fracturing and suffering massive spaulding... and the MOVEMENT caused by explansion of beams and girders in the core area was destroying splices and causing columns to move laterally destroying bearing area at their connections and this lead to LOCAL buckling... and partial floor collapse. Bolted connections were shearing and failing.
This process also led the antenna to lose axial support in 1wtc and sink into the top of the tower destroying much of the floors in the center of the core from the roof downward... freeing massive loads from the mech floors which I believe is the origin of the ROOSD mass of the runaway collapse stage.
The local floor destruction caused by the heat in the center of the 1wtc... expanded/radiated outward toward the facade as the fire progressed. It "grew" / moved... which is what fire does as it consumes more of the available "fuel". The destructive impact on the beams of the fire also grew outward from inside the center of the core reaching to the facade over time... eventually destroying axial capacity of the columns to support what was left of the "upper block".
The following graphic illustrates my conception (described above) of the progression of destruction in the core at the top of 1wtc over time... at 3 levels of the core...floors 94-99... floors 100-105 and floors 106 - 108
My suggestion is that the lateral elements' heating led to the buckling of the axial structures. I referenced this "process" which NIST identified as a cause for column failure in 7wtc... the heating of beams and girders led to column failure. In 7wtc there was no claim that heat weakened the column sufficiently for it to fail. NO I am not changing the topic to a discussion of 7wtc.
One take away from the Caddington study was that when the steel beams and girders endured heat and in the case where there were columns which were "unrestained"... the beams and girder would push the columns leading to axial mis alignment and failure. The heat also weakened slabs so that they lost integrity and fractured as well.
I am suggesting that the way the heat caused the columns to fail was not simply from heat lowering their load bearing capacity to below demand. Axial capacity likely was reduced by the heat. I don't think in the time frame it reduced capacity enough to cause the buckling.
My hypothesis is that most simply stated... the lateral members expanded from heat, and forced mis alignments of the axial connections. And that led to inadequate bearing area .. And that led to web and flange buckling.
Look at col 704 which ran from flr 88-90:
The gray shows column 704 displaced by .75 inches in both the X and Y axis. We don't know how much it may have been displaced... or if it had been displaced... We can look at that later. But the diagram illustrated how the bearing area to the column ABOVE beginning at floor 91... might have its bearing area reduced to less than 4% ... This would surely cause web and buckling. I don't know how at what bearing area bucking would take place. I don't know how much and in what direction lateral displacement might have occured.
We do know that column to column connections were not restrained AT the column to column connections by beams. Lateral framing/bracing was at approximately 4', 16' and 28' from the bottom of the column. I believe this was to facilitate construction and the bolting of the column splices. The bolts of the splices would be the main lateral restraint.
I believe NIST discovered many failed/destroyed bolted connections in their study... pull out and shearing. This was also true at the truss seats in some cases.
Lateral Restraint of Core Columns
In a grid framing plan a column is laterally restrained in both the X and Y axis by beams/girders (regardless of whether the restraint is AT the column to column connection). We know that unbraced columns are weaker than braced columns according to the work of Euler. In fact, we were told that col 79 on 7wtc buckled when it lost lateral bracing.
So I return to what likely was going on in the core of the twin towers. Massive fires were heating all the steal under the slabs on the level of the plane strikes.... because of the fuel supplied by the jets almost full tanks. Those flames concentrated where the fuel fuel tanks of the jet were... in the case of 1WTC the center of the tower in a north-south swath through the building. Yes it did disperse and likely involve the entire foot print.
The Caddington study shows what happens to heated concrete slabs supported on steel beams. This is likely what was going on in those floors where the fires raged in the twin towers.
But unlike Caddington... 1wtc also had a number of columns severed and severely damaged by impact of the plane. My slide in a previous post suggested which columns might have been destroyed by the initial impact of the plane. I believe NIST suggested these same columns.
The OOS floor support on the core side was a belt girder which transferred the 80" OC load locations of the double floor trusses to the core columns which were spaced from 8' for rows 4 & 5 to over 20' for rows 1 - 4 and 5 - 8. The belt girder was supported on short cantilever beam stubs.
I am suggesting that the heated beams inside the core at the level and above the level where the plane destroyed columns and started massive uncontrolled fires... that the "frame" was being pushed by expanding beams... that the slabs were breaking fracturing and suffering massive spaulding... and the MOVEMENT caused by explansion of beams and girders in the core area was destroying splices and causing columns to move laterally destroying bearing area at their connections and this lead to LOCAL buckling... and partial floor collapse. Bolted connections were shearing and failing.
This process also led the antenna to lose axial support in 1wtc and sink into the top of the tower destroying much of the floors in the center of the core from the roof downward... freeing massive loads from the mech floors which I believe is the origin of the ROOSD mass of the runaway collapse stage.
The local floor destruction caused by the heat in the center of the 1wtc... expanded/radiated outward toward the facade as the fire progressed. It "grew" / moved... which is what fire does as it consumes more of the available "fuel". The destructive impact on the beams of the fire also grew outward from inside the center of the core reaching to the facade over time... eventually destroying axial capacity of the columns to support what was left of the "upper block".
The following graphic illustrates my conception (described above) of the progression of destruction in the core at the top of 1wtc over time... at 3 levels of the core...floors 94-99... floors 100-105 and floors 106 - 108
Last edited:
Jeffrey Orling
Senior Member
Roberts Rules of Order states that humor is never out of order
deirdre
Senior Member.
or the writer of that sonnet.
Jeffrey Orling
Senior Member
e tu Brutus, then fall Caesar... yea it was Elizabeth Barrett Browning... I sit corrected...
back to the initiation....
can we accept the premise that for columns to fail... to carry axial loads...
there would have to be the following causes:
a) loss of axial capacity
b) loss of bearing area
c) loss of cross sectional area (partial)
d) new additional applied loads
A full capacity column could fail if it had its bearing are critically reduced... or had insufficient bearing. Let's say a plane engine severed a column and the columns above, though intact were incapable of transferring load to the foundations... they would fail by dropping from loads.
Axial capacity could be reduced if the column were heated as heat lowers the compression and tension strength of steel.
Partial loss of cross sectional area would reduce capacity... such as too large holes in the web or flanges.
Loss of bracing reduces capacity per Euler.
New additional loads from load redistribution could render the column inadequate to carry the new total loads applied.
Can you find any other cause Watson?
back to the initiation....
can we accept the premise that for columns to fail... to carry axial loads...
there would have to be the following causes:
a) loss of axial capacity
b) loss of bearing area
c) loss of cross sectional area (partial)
d) new additional applied loads
A full capacity column could fail if it had its bearing are critically reduced... or had insufficient bearing. Let's say a plane engine severed a column and the columns above, though intact were incapable of transferring load to the foundations... they would fail by dropping from loads.
Axial capacity could be reduced if the column were heated as heat lowers the compression and tension strength of steel.
Partial loss of cross sectional area would reduce capacity... such as too large holes in the web or flanges.
Loss of bracing reduces capacity per Euler.
New additional loads from load redistribution could render the column inadequate to carry the new total loads applied.
Can you find any other cause Watson?
Yeah, but my rules discourage it, as it's not conducive to discussion.Roberts Rules of Order states that humor is never out of order
This thread is about modeling the initiation. If you really think that thermal expansion pushed the columns out of alignment in a significant then suggest how that might be modeled (hint: with great difficulty)
From my perspective, the purpose of a physical model is illustrative, not investigative. Largely because it's impossible for a physical scale model of such a large structure to be built by an individual in such a way that it encompasses all the elements of the collapse initiation. So instead we can seek to illustrate the answer to questions like: "why didn't the top tip over and fall off" or "why didn't the top gradually sink one floor" or maybe even "why did the antenna fall before the roofline?"
I think it's a significantly more challenging task than the collapse progression model.
Jeffrey Orling
Senior Member
Of course the thread is not about what happened but modelling a top down driven collapse.
To do that we need to agree that the load which drove this initiation was part of the tower... dead load, super imposed dead load and live load.... and all supported by the axial structure - columns. We do accept the engineering principals that loads were moved laterally by beams, girders, floor trusses and the steel hat truss structure... to the columns.
And we do agree that the dropping top's fall was visible at the moment the top no longer had sufficient axial support to keep it in place...
So.... we don't have to model how that initiation process came about but rather how to release the top mass to then destroy the tower below.
Maybe
The initiation may not be like an on - off switch. In fact it likely was not... it was a process... The process conceptually could be the gradual loss of axial capacity... using up "reserve capacity" and then the buckling of the columns and the drop. But it may not have involved ALL the axial load bearing columns failing. It's conceivable that with only a threshold number of failed columns the top translated laterally and lost support and dropped.
This may not matter for the model... all that needs to be done would be to have a model in which columns/support near the top fail and that leads to the top's descent.
I suggested candles which would soften and lose strength when heated.
The model could be materials which will not ignite from the heat that will be used to weaken the columns... Maybe the floors are thin metal plates?
A fire on the metal plates with candle columns seems a bit dangerous.
Could hot air from a heat gun be directed on to the floors? Maybe through an opening mimicking the plane hole?
econ41
Senior Member
Ooops... never gave it a thought.
Being a philistine engineer I wouldn't have known even if I had. 
econ41
Senior Member
Fully agreed. I suggest the implied "target" is the notional truther or other person who doesn't understand - cannot visualise - the initiation process. So "looking sufficiently like" the real event would be a broad criterion.
If I presume that we are in general agreement that sequenced cascading failure of columns is the key mechanism. (And it subsumes the NIST joist sag inward bowing TRIGGER) Then my thinking is along these lines:
A top plate and bottom plate separated by a row of columns as per Mick's preliminary model. Two vertical rods as guides.
THEN inset a row of wax candles. Experiment with numbers and sizes. Spread them evenly for initial experiments. Apply a load - how big will be subject to trials.
Apply radiant heat to one end of the row sufficient to weaken the candles by softening. The candles nearest the radiant heat will soften more and that end start dropping (Tilting the Top Block)
I think it would work. And easier to simulate the overall process that way rather than narrow the focus onto the the NIST "joist sag trigger".
Thoughts?
Last edited:
I was thinking that the candles deformability might be a problem. I just does not scale well.
How about spaghetti?
I picture a bottom 1" plate of wood with spaghetti-sized holes in a pattern (perhaps with core and walls). The holes go through the plate so broken pieces can be pushed out when inserting new ones. They should per drilled with a drill-press to get them perfectly perpendicular.
The plate is clamped on a flat surface, which acts and the base of the columns. Equal length columns are inserted. A top plate is added, consisting of a piece of wood with some foam tape on the bottom to even out minor variations in length. Weight is carefully added to the top plate until it is at 1.2x its static load capacity.
Then columns are removed. This can be done initially with some kind of analog to the airplane impact. Perhaps shooting the columns with a salt gun. Perhasp firing in some light object with a rubber band. Or you may have to just clip them.
Then gradual removal. There three options here:
How about spaghetti?
I picture a bottom 1" plate of wood with spaghetti-sized holes in a pattern (perhaps with core and walls). The holes go through the plate so broken pieces can be pushed out when inserting new ones. They should per drilled with a drill-press to get them perfectly perpendicular.
The plate is clamped on a flat surface, which acts and the base of the columns. Equal length columns are inserted. A top plate is added, consisting of a piece of wood with some foam tape on the bottom to even out minor variations in length. Weight is carefully added to the top plate until it is at 1.2x its static load capacity.
Then columns are removed. This can be done initially with some kind of analog to the airplane impact. Perhaps shooting the columns with a salt gun. Perhasp firing in some light object with a rubber band. Or you may have to just clip them.
Then gradual removal. There three options here:
- Fire. Which might be difficult to isolate. But a small torch could be used to burn away some columns.
- Water. This might be a better way of simulating the effects of fire. Water causes sphagetti to lose its stiffness gradually, which should cause gradual load transfer up to sudden failure.
- Manual removal - by clipping.
Oystein
Senior Member
Haaaa I experimented with spaghetti when I first encountered WTC conspiracy theories like 10, 11 years ago, particularly to get a feel for static vs dynamic loading.
How about steam?
Promising, for an hour's work. The top platform was rather unstable. Adding the weight is a dynamic process, and yet we want to not apply any more force than the static load, which is nearly impossible to do adding all at once. I think the best way would be a tub gradually filled with water I can do it once to find the breaking load, then again to 75% followed by column removal.
I think I also need to use shorter columns, about 2" instead of 3" also a larger "floor", maybe 2x what is used here.
The good news though is that I got very rapid failure of all the columns. It was only 1/10th of a second between the failure of the first column and the last. This at least demonstrates the idea of rapid failure progression when overloaded.
Last edited:
Jeffrey Orling
Senior Member
Mick... good start.... you might:
drill small holes in the underside of the wood plate for column stability
place a thin tin plate on top... and pour sand slowly to increase the load
you could to this for several levels... but the lower level would need maybe SOME thicker spaghetti... ... each level thick spaghetti...
drill small holes in the underside of the wood plate for column stability
place a thin tin plate on top... and pour sand slowly to increase the load
you could to this for several levels... but the lower level would need maybe SOME thicker spaghetti... ... each level thick spaghetti...
The reason I'm just using the batting (the white stuff) instead of drilling holes is that the spaghetti snaps off easily, and would get stuck in the holes. I could drill the holes all the way through and then clamp a plate on top.
Jeffrey Orling
Senior Member
rather than a single load.... why not the sand idea?
That's what I was getting at with:
Although sand might be better to avoid damaging the wood by getting it wet.
Jeffrey Orling
Senior Member
water is too messy... sand is easier to clean up and vacuum.
Jeffrey Orling
Senior Member
This got me thinking of a food model....
floors of rice cakesc or matzos ... noodle columns... sugar loads...
floors of rice cakesc or matzos ... noodle columns... sugar loads...
Jeffrey Orling
Senior Member
Pizza model
Thin pizza crust floors
bread stick columns
flour loading
hold the sticks in place with some uncooked pizza dough
Thin pizza crust floors
bread stick columns
flour loading
hold the sticks in place with some uncooked pizza dough
deirdre
Senior Member.
water levels itself out though, as far as having even load over the whole plate quicker.
@Mick West
they do have skinny candles. but if spaghetti strength is the right scale tehn they would be too big probably. and heat guns are kinda big to get to teh inner core columns.
But if this is for laymen or truthers i think youd need to add alot more spaghetti as the twins had alot of columns. not just 7 on the exterior!! so as i laymen im thinking it fell because you dont have enough columns.
Jeffrey Orling
Senior Member
This is a matter of trial and error... Yes water is self leveling. But the load in reality was not uniform... so you could pour whatever granular substance used into a till pan... more or less evenly over the plan.
Same goes for the columns and even number of floors.... setting up the columns to look more like the twin towers will be more convincing... but require more columns than Mick's models.
In the model the weight of the slab to the weight of the super imposed live load needs to be similar to the ratio in the real world. An overly massive super imposed live load will raise cries of foul play.
It would also help if the model had multiple levels like the floors of the tower...
the model should demonstrate:
weakened columns fail causing the floor above to drop
this needs a column material that weakens hopefully with heat.
a material placed on the floor which would burn when ignited
columns which burn or better melt
excessive load on a floor causes the columns to collapse/buckle and with it the floor collapsesthe sand or rice or whatever represents mass fall from above represents the slabs and superimposed live loads
the floors should be of a material which could break up from being overloaded - matzohs could work... cast rice crispy floors
econ41
Senior Member
I see deformabilty as the necessary characteristic for valid modeling of the initiation stage. We need to model "heat weakening". Sure we are unsure of the exact parameters for candles (at this stage of research) BUT they do soften when heated and will ultimately melt. We don't need "melt" but "soften when heated" is the very feature of steel which drove the initiation process. I don't understand the reference to "does not scale well."
I'm very doubtful about the ability to mimic heat weakening by water soaking of spaghetti. And cutting is a very different mechanism. We may be able to model heat weakening by analogous "water soaking weakening" but the physics of successive cutting of bits of spaghetti "columns" is not the physics of the actual initiation stage.
Plus an over-riding consideration - more important for either method - cutting or water soaking would probably be seen as faking by the "down the Rabbit Burrow" truthers who are the notional target for the model.
We have the choice of either a linear array of candle columns OR a more extensive 2D "floor plan" as per your first attempt at a spaghetti column model. And the logistics of heating from one side via hair driers or heating lamps is less cumbersome than trying to water soak by either of the two obvious methods (spraying the spaghetti or putting the whole model in a water tank????) Plus heating from one side more accurately models both versions of the real event and would induce "tilting" which was a feature of both tho' more extreme with WTC2.
The initiation stage had two distinct sub-phases. The immediate effect of aircraft initial impact which cut some columns. Setting up some initial load re-distribution but having no further effect. Then the slower processes leading to exponentially faster failure. The difficult part to model is the second part. The initial impact cutting relatively trivial. Whether we model it or not is a decision to be made. But the two are conceptually different.
Best to not prematurely eliminate heat weakening which is the actual process of the real event and therefore a preferred option. Agreed that spaghetti may satisfy the requirement for an analogous "weakening" process. Neither fire nor cutting are good analogues and could therefore be of limited value in persuading either truthers or laypersons.
Yes it is progress. I would prefer more explicit consensus on the goals and the scope of how much we model. For example - would a linear 1D model demonstrate the mechanism sufficiently for a lay person? For a "rabbit burrow truther"? OR is a more 2D floor plan necessary. It should not be much harder.
I think that suggestions by Jeffrey Orling (any others??) that we go for a multi storey model are impractical. At least as a "first attempt".
Last edited:
econ41
Senior Member
Candles are a good model for "heat weakening" which was the #1 key feature of the "initiation stage". And I'm confident that a hair drier or two would provide the heat. And we don't have to "get into the middle" The real events STARTED from one side and heating from one side would be a reasonably accurate mimicking of the actual event.
We need to constantly remember that the layman or "down the Rabbit Burrow Truther" is the target. Some of them would understand with a single row of 6-7-8---15 candle columns. It would be dramatic as one end sank as candles softened. Other doubters would need a more realistic full floor plan - hopefully with no more than a couple of dozen candles. 2-300 candles OR bits of spaghetti would be impractical and excessive. And you couldn't cut the inside bits of spaghetti nor get at them with water unless you put the whole model in a tank. And I'm sure that would be too far from reality for most of our target audience to accept.
So Deirdre - as our resident "layperson" - if we use candles and hairdriers - would a single row starting to collapse from one end satisfy you?
OR would you need a bigger - two dimensional floor plan with several rows of columns?
Last edited:
Jeffrey Orling
Senior Member
good summary
how extensive the model is depends on how much imagination/intelligence the doubter has. So simpler models can model heat weakened columns will collapse. But is that really questioned?
The doubters refer to a high rise so I believe that the model needs so show how the initiation was not a single floor process. I don't even think that one floor on top of the one below would collapse... It would take several... dynamic loading changes things... so this might need to be worked into the model.
I don't think a column 2D column grid is a problem... the floors can be drilled and the candle columns place in the shallow holes on both sides of the floors.
So you will want to place some super imposed live and dead load on each floor (same). The columns should buckle from say 4 or 5 of these floor loads
I also believe that the floors need to be something that will fracture and not remain as solid plate under excessive load. Why? we want to model both heat failed columns and floors that shatter as the crash down to the floor below.
The column arrangement wants to be a rectangle for the perimeter of the core (24 columns) and a large square of closer spaces thinner columns for the facade. Let's make the core 3 on the short side and 5 on the long side. and say 12 on each side of the facade.
A real problem with the columns is that the loads - floors... are attached to the sides of the columns. Maybe these would work
1/2" diameter x 5.5" long they are not tapered. Cut the tips off. columns sit one on top the other.
What to do about the spandrels? More toothpicks thru drilled into the sides of the facade candles?
Drill the candle to insert a short piece of a toothpick to be the bearing point for the facade... drill thru and insert a toothpick to support the core and the OOS floor plates.
Core would have a solid rectangle
OOS floors would be a square with a rectangular center
Floor Plates that can fracture is a hard one. Maybe cast rice crispy floors made to stick together with sugar water?? Trial and error. Possibly matzos... 6.5" x 6.5"???
Carefully build up the model from the bottom floor - leave a few of the facade columns out... heat will be blown through the opening. Make it min 4 floors tall.
Perhaps use 4 thin acetate sheets on each floor so that the perimeter can contain the heat????
lotta work.... not terrible expensive... simple tools
MAYBE
Similar threads
