• This is a political forum that is non-biased/non-partisan and treats every person's position on topics equally. This debate forum is not aligned to any political party. In today's politics, many ideas are split between and even within all the political parties. Often we find ourselves agreeing on one platform but some topics break our mold. We are here to discuss them in a civil political debate. If this is your first visit to our political forums, be sure to check out the RULES. Registering for debate politics is necessary before posting. Register today to participate - it's free!

Are massless supports valid for use in progressive collapse simulation?

Kat Dorman

DP Veteran
Joined
Mar 28, 2013
Messages
1,903
Reaction score
630
Gender
Undisclosed
Political Leaning
Undisclosed
I say yes.


First, some disclaimers.

(Important points bolded for A.D.D. readers.)

1) This topic is in the CT section because its relevance is to 9/11, particularly the modeling of progressive collapse in an attempt to better understand the physical principles behind it.

2) The topic is VERY narrow and specific, unlike the vague thread topic which spawned this discussion; it will be very easy to judge what is and isn't on-topic. I don't report posts as a philosophical principle, but that doesn't mean a moderator won't take action of their own initiative.

3) The subject is important to at least two people on this forum and is intended to settle a years-long dispute spanning more than one forum. It may be of no interest to anyone else, but all are obviously welcome to participate if desired.

4) If the one other person who SHOULD be interested in this declines to participate, my interest will immediately go to zero, because it will be a no-contest concession that I'm right and have been all along. Anyone who indulges in unsupported one-liner jabs for years ought be able to spend a few words defending their position.

5) I have little confidence that anyone here will be savvy enough to be swayed by technical arguments, based on historical interaction in this subforum. I have no illusions of winning a popularity contest based solely on ideological bias.

6) and this is important: this thread is NOT the place to debate the validity of the 1D model in question for academic or simulation purposes, it's ONLY to discuss whether it's valid to utilize massless connections/supports with THAT model. I know this model doesn't simulate the towers, that's not the issue.

----------------------------------------------

The premise is simple: I say massless connections are valid to use within the model (to be specified) because the difference in results between connections with and without mass isn't significant. I further claim that using massless connections greatly simplifies the problem and is orders of magnitude less expensive computationally. Taken together, these are the reason massless connections are preferred when operating within this model.

Anyone who believes that use of massless connections is absurd and worthy of mockery will finally have a dedicated platform to elucidate why they feel this is the case. If there is no contention, I won't bother arguing the point and this thread can die. One or more posts will be made later (when I have time) explaining the nature of the model itself, so readers (if any!) can become familiar with the context.
 
Last edited:
Well the thread does not make any sense without the graph from the simulation which contained the SIMULATED massless compression springs. What was the date the graph was first posted and what was the date that the massless springs were specified?

psik
 
Well the thread does not make any sense without the graph from the simulation which contained the SIMULATED massless compression springs. What was the date the graph was first posted and what was the date that the massless springs were specified?

psik
First, I disagree with everything you've said here - quite strongly. The issue can most definitely be discussed and even brought to a rigorous conclusion without EVER considering output graphs from ANY ONE particular simulation. This is down to general mechanical principles, not beholden to the appearance of a single graph. Furthermore, specification of the fact that connections are massless is REDUNDANT and UNNECESSARY with a discrete mass model because that's the very definition of a discrete model!

Second, I know you and your tactics: you will use the graphs to distract from the thread topic by seizing on the oscillations and 1kg masses. THESE ARE SEPARATE ISSUES and are off topic here. If you want to discuss those distinct issues, start another thread. This is about massless connections, and massless connections only.

Third, I'll oblige you by supplying the information, anyway. Shortly. I will NOT let you distract from the central issue before I've even had a chance to make my case.
 
Let's be clear about one thing: your post above is already a lame attempt at distraction, very first thing you do.

What does it matter WHEN I disclosed something that should be obvious to someone who wrote a program with NO connections at all? The only difference is my connections have the property of applying a constraint force to the discrete masses and no mass property. Yours have no properties at all. Shouldn't I be mocking you for something worse than something you mock? I don't care that you disclaimed them as "magic supports'; if what I'm doing is stupid, what you've done is far more stupid, no matter what label you put on it.

The issue is not whether I was amiss in not stating the obvious, the issue is - is the use of massless connections valid? THAT'S what you mock me for. The ridicule over not stating the obvious should stop once the obvious is stated. That's NOT what you're mocking me for - you're ridiculing the use of "massless compression springs." Period.
 
Last edited:
Why don't you try to supply a justification for your ridicule of my use of massless connections? You shouldn't need ANY graph to do that, and I'm damn straight not going to post any graphs until you provide at least a sentence or two WHY you believe massless connections to be absurd.
 
Description of the model.



1) The model in question is 1D. That is, all action - motion, force vectors - occurs on a line. The only directions are up and down.

Aside from FEA and (some) physics simulations, virtually all progressive collapse models are 1D. In this respect, my model is the same as Bazant, Seffen, Greening, Benson, and some CD proponents like Szamboti and Beck, countless anonymous hobby analysts, and even psikeyhackr! His Python program is 1D, and his physical model strives to be 1D even though it's assembled in the real 3D world.

Certainly ALL of the popular analytical models are 1D, and I prefer to do 1D even when I have a full 3D simulation environment. My purpose is to study the basic physics of collapse, particularly the properties of 1D models, NOT to model the tower collapses.


2) There are two categories of models with respect to mass distribution, continuous and discrete. The one in question is discrete.

Full stop. A discrete model is defined to be a mass distribution with gaps or holes in it. Bodies in collision with empty space between them, versus a continuum. Need the argument go any further than this? No.

If two masses are to exert a force on each other, that force is just a force. Forces have no mass. So, when two discrete bodies have to interact via a force that force is specified as part of the model - literally a formula. It can mimic the resistive force of a spring or a column or whatever, but it's just a force and doesn't have mass.

It IS a connection. It can act as a support in the model because it defines the force which acts on a body (story) to support it, and so on up the chain to the top. This is exactly how joints (connections) are defined in 3D physics engines like the one I use, and also mass-spring-damper components in a modified version of Fydik which I use.

GET THIS: these are legitimate, accurate professional/research physics simulators, and there isn't even a way to specify mass for connections because connections are FORCES! Masses are masses! Duh!

The argument should stop here. We'll see.
 
Description of the model, continued.


3) The sort of discrete model in question is what I call (because no one's given it an official name) a 'slab model'. It concentrates all "story" mass into a finite extent, either a point or a line segment. The advantage to using a line segment is that it can represent the height of the fully compacted story after crushing.

This category of model includes Greening and psikeyhackr's computer model. The calculations using a point for story mass are done by both of them. The stories didn't compress down to nothing! But that's okay, abstractions are fine so long as their ramifications are understood. It's a 1D model, for goodness sakes, the whole thing is an abstraction!

This is also part of the incredible irony. psikeyhackr (the objector/mocker here) uses points separated by empty space in his computer program, where I use points/lines separated by empty space, with the addition of a force function which behaves like any kind of support I desire. psikeyhackr has mocked me for years for having functionality superior to his "analysis program", which is the most rudimentary analysis possible.

His physical model is well approximated by the slab model because his washers have mass much greater than the paper loops supporting them. As well, the towers themselves had supports with mass insignificant compared to the load they supported. The idealization of slabs is a good approximation for a lot of things.

4) The slab model, without connections, is a momentum-only model.

Without connections, the upper section falls down and collides with each mass in succession, and the law of conservation of momentum for an inelastic collision is applied. In other words, psikeyhackr's Python program. Also, Greening's first pass before adding E1 (energy to fail one story's supports). VERY crude, but still of some value as a loose first approximation. Because the forces which act on the upper section are additive, each contributing factor to retarding collapse can be examined separately. This lets you see what momentum exchange does; a useful learning experience but only one part of the consitituent resistive forces.

The other primary component is structural resistance. Without this, there's no way to know what the actual descent characteristics would be in a more realistic structure which must support itself (with a margin of safety). Besides that, momentum-only models can never arrest - they have no supports!
 
Description of the model, continued.


5) Adding structural resistance (a force) to the slab model is adding a FORCE, not a mass.

I added structural resistance in the criticized experiment as a load-displacement formula that followed Bazant's textbook specification. This is what psikeyhackr has mocked me over for over two years. For having a distinct improvement over his crude algorithm... nothing more, nothing less.

Now my simulations are not points magically suspended in space until something hits them from above, they are masses which support each other by way of forces acting between them, just like story masses are supported. That means they can vibrate when whacked, because real structures do. They can fail when whacked too hard, like real structures do. They'll fall down if the force is insufficient to define a self-supporting structure.

But these forces don't have mass! The mass is concentrated into a slab, that's what a slab model is.
 
6) There are many ways to implement the slab model to obtain solutions.

a) a discrete algebraic approach like Greening, purely analytical and not a simulation
b) numeric solutions from F=ma step solvers, of which there are multiple varieties
c) finite element analysis, 2D or 3D

I've done (a) and two variants of (b) - physics engine and RK/Euler - and dabbled in (c). Truth be told, at this level of analysis which is the physics of collapse - not the particulars of how a beam or column deforms - FEA is the worst. It's complicated, time-consuming, fraught with error and computational artifact, requires precise detail to converge on a meaningful solution, takes forever to run (then blows up before the end)... the list goes on. That's why I only dabbled in it; it's far too much detail when the 1D methods are so easy and give essentially the same result (when the FEA actually works).

All three environments I've used give results in very good agreement for all cases. That's a strong validation of the methods, because all three are radically different implementations and approaches. The only thing they have in common is they're all software.

Now, it is possible to also incorporate mass elements with finite extent and modulus properties in one of the environments, and possible to distribute discrete masses along the line in another. These function as connections with mass, like real supports. I have done this on a limited basis, but there many reasons to avoid it:

- it adds complexity
- greatly increases the time to run the simulation
- like FEA, the possibility of error or unphysical artifact is increased
- accuracy increase in error-free runs is insignificant

Therefore, there is no reason on god's green earth to use connections with mass in these environments, even when they are available.
 
I really don't see the need to add any more, this is already like trying to blow out a kitchen match with an ICBM.

I've been criticized for following standard operating procedure for physics-based simulation. Can this be too surprising when the person criticizing doesn't know what SOP is for simulation and analysis? Can't set up and solve even one equation of motion? Has had no education or professional experience in physics? No education or experience in software-based analysis and simulation? Doesn't even know what potential energy is?

No. Not a surprise.

--------------------------------

Interested readers can search on "mass-spring-damper" systems to see how ubiquitous this sort of technique is in physics and engineering. In no case does the spring have mass - that's what the "mass" is for in "mass-spring-damper"!

:lamo
 
First, I disagree with everything you've said here - quite strongly. The issue can most definitely be discussed and even brought to a rigorous conclusion without EVER considering output graphs from ANY ONE particular simulation.

Second, I know you and your tactics:

It has everything to do with one particular simulation. You made the simulation. Let's see YOU explain it to anyone.

And then deal with the fact that you did not tell everyone it was the result of ridiculously long massless compression springs holding 1 kg that no one knew were there until two years later. In fact when I first looked at the graph I didn't know the mass was only a kilogram. I was thinking in tons which would be relevant to the subject under discussion.

Oh sure, it is everybody's fault but yours. It is all about you complicated BS tactics.

So if you are not going to supply the graph then forget it. YOU MADE IT, not me.

psik
 
Too late to edit this from above:

Now, it is possible to also incorporate mass elements with finite extent and modulus properties in one of the environments, and possible to distribute discrete masses along the line in another. These function as connections with mass, like real supports. I have done this on a limited basis, but there many reasons to avoid it:

- it adds complexity
- greatly increases the time to run the simulation
- like FEA, the possibility of error or unphysical artifact is increased
- accuracy increase in error-free runs is insignificant
- most (cheap) deformable elements only cover elastic range and do not function well in non-linear or far-from-equilibrium conditions

Important addition in blue. Most elements in simulation programs, whether FEA or not, whether composites of small mesh or a single complex element, SUCK at reproducing failure dynamics. They'll indicate when you're approaching failure, perhaps, but are generally for deformations which are essentially elastic. Plastic deformation, fracture, irreversible compression - these things are in the realm of sophistication, expense, and difficulty that simply isn't worth it for this level of investigation. Even then it would only be reliable and verifiable for the simplest of systems.

Why bother? When it's a piece of cake to program the load-displacement relation and rehardening effects of steel columns in three hinge buckling and run it in under a second?
 
So if you are not going to supply the graph then forget it.
I said:

Third, I'll oblige you by supplying the information, anyway. Shortly. I will NOT let you distract from the central issue before I've even had a chance to make my case.

and

... I'm damn straight not going to post any graphs until you provide at least a sentence or two WHY you believe massless connections to be absurd.

Pretty straightforward.
 
It has everything to do with one particular simulation.
So, are you saying you accept massless springs in compression in all other simulations? Some? Including all the other simulations I've done in other environments with massless connections? Are those okay or not? This is not a rhetorical question.

Let's see YOU explain it to anyone.
I just went a long ways towards doing that right above your post. Went completely over your head, did it?

And then deal with the fact that you did not tell everyone ...
I've dealt with that long ago. WHO CARES? You've known for a long time, when are YOU going to deal with it?

...it was the result of ridiculously long massless compression springs holding 1 kg that no one knew were there until two years later.
So what? I just showed above it's not only NOT a problem, it's SOP.

SO WHAT???? to all of this.
 
You're tapdancing and dodging as usual. You're not showing that use of massless connections* is ridiculous. You simply call it that, as if it was self-evident. It's wrong. I showed that above. Read it. Refute even one point if you can.

All you can do is whine month after month that I didn't mention every detail of a simulation at the time I presented some of the results. Even if that were some kind of crime (it isn't), that's over and done with.

If the technique is acceptable, then it doesn't matter one way or another. Is the technique acceptable?



*(they're not springs, they're formulas - get that? They have traits in common with springs, just like steel columns, that's all.)
 
And please examine the thread title... the purpose is to ascertain if what I did was valid, not to pass judgement on when the fact was revealed to you. If you think this is such a terrible thing that it wasn't mentioned along with the thousands of other words I wrote describing the experiment, fine.

I'm sorry. It was wrong, I should've disclosed it at the top of the first post. You feel better now?

Of course, I did write at the top of that post that it was 36 stories, and mentioned it again several times thereafter, and you still insisted on calling THREE stories even after being corrected. Had I disclosed the information, there's no evidence you'd have done anything but ignore it. If you'll recall, your big complaint about me before this particular simulation was TALK, TALK, TALK - too many words.

Now, FOCUS on the subject. The distractions were tiresome two years ago.
 
First of all, I'd like to debunk the "progressive collapse" business right now
because it really isn't progressive, its a function of the Simultaneous removal of ALL of the structure
out from under the falling bit. the observed 2.25 sec of free fall acceleration,
is all the prof that is needed here.
end of debate. WTC7 .. Controlled Demolition ....
 
Blatantly off-topic. Please stop.
 
You're tapdancing and dodging as usual. You're not showing that use of massless connections* is ridiculous.

I never said ALL massless connections were ridiculous. I made derogatory remarks about your using 3.7 meter massless compression springs in THAT PARTICULAR SIMULATION and NOT TELLING everyone that you were doing it.

Now you want to TAP DANCE by accusing me of making some blanket accusation and not providing the graph on the particular case in question.

And then accuse me of TACTICS. :lol:

If you are not going to provide the graph then this is pointless. I already tried to provide a link to it and it was missing. Surprise, surprise! But your other graphs were there. Curiouser and curiouser. Tactics, tactics.

psik
 
I never said ALL massless connections were ridiculous.
Actually, when you first started complaining about this, you most certainly did. Then I trotted out a half dozen examples of the very same thing being done in the research literature, and you completely ignored all of them but one: The one with NASA using massless springs to simulate a parachute line in tension. Somehow, you managed to wrap your mind around that one but completely ignored the others like they didn't exist. That's when you switched your complaints to springs in compression, like somehow that matters. No comments from you on the approximation of large steel columns by massless compression springs!

Which, in case you didn't notice, is exactly what I'm doing.

So now you say you don't think all massless connections are ridiculous. Just mine. Somehow, when I do it, it's not okay even though it's standard operating procedure and NO spring has mass in a mass-spring-damper system - EVER.

You're trying to complain about something that's done all the time successfully and with good reason because you don't know it's SOP!


I made derogatory remarks about your using 3.7 meter massless compression springs in THAT PARTICULAR SIMULATION...
Are you saying it is ONLY that simulation you have problem with?

...and NOT TELLING everyone that you were doing it.
How long are you going to beat that dead horse? There's no shame in following SOP, and nothing odd about not explaining that it's SOP.

Now you want to TAP DANCE by accusing me of making some blanket accusation and not providing the graph on the particular case in question.
I said I would - AFTER you provide at least one sentence of explanation WHY you think massless connections are ridiculous. So far you have not. Only reiterated your complaint that it is ridiculous and I didn't tell you.

Why is it ridiculous in this one simulation but not others?

Surprise, surprise! But your other graphs were there. Curiouser and curiouser. Tactics, tactics.
I told you already the image host (TinyPic) deletes them after a period of inactivity. You don't believe that? Look at their stated policy. I don't even have the ability to delete them once uploaded. So, give this BS a rest.
 
It seems the complaints you have distill down to one simple thing: you don't know how physical modeling and analysis is done.


My choice of using something akin to a mass-spring-damper system is an excellent choice for the problem at hand. Since the momentum-only model (like what you use) is a MASS-ONLY system, I improved the model by adding the functionality of springs and dampers. The spring part serves the role of the elastic phase of compression, about 0.2% of the overall length. Thus, if these spring were real, they'd be (0.002)*3.7m = 0.74 centimeters long.

You didn't know that, did you?


So, maybe you need to adjust your goalposts again. It's the virtual damper functionality I've programmed in which mimics the plastic phase of compression which represents the rest of the column length (up to rehardening).

Do you even know what a damper is?


Now, if you go on complaining about "3.7m DAMPERS in compression", none of your fans will even know what the **** you're talking about. That may pose a problem for the game of semantics you've been playing.

Regardless, ALL of this is standard operating procedure for modeling of physical systems, which is why the totality of your complaints are void. You don't know what's normal in this field, having never once properly modeled a physical system in your life, so you don't understand that what you think is ridiculous and stupid is NORMAL and it WORKS (just fine).

At this point, the ONLY angle you could possibly argue is that a mass-spring-damper system (or analog to it, like my models) is not the proper choice of model. You cannot argue that the springs - or the dampers - should have mass in such a system because by definition they don't.
 
There's an easy way to cut through the ****.

Answer this: if using massless connections is ridiculous, why are the results between massless and massful connections nearly identical?

There is a theoretical and analytical reason for why they should be nearly the same, a reason why it wouldn't matter if you concentrated most (or all) of the mass into the supports. Ultimately, this is the justification for why massless connections are perfectly valid. Would you like to go through the analysis in detail to finally understand why you're wrong?
 
Last edited:
Pick the one(s) you have a problem with:

1)
2rfr3op.png


2)
2lveh4i.png


3)
14tbjmc.png


4)
n1b0w4.png


5)
ddcrqr.png
 
11)
120pjpy.png


12)
2i07w9i.png


13)
2dqrp7t.png


14)
saw3o9.png


15)
9kmlv6.png




NOTE: Every last one of these pictures will be deleted by the image host TinyPic after a period of inactivity, supposedly no sooner than six months. I cannot delete them.
 
Back
Top Bottom