Are massless supports valid for use in progressive collapse simulation?

[psikeyhackr]

10)

Yeah, it is the usual Kat Dorman information overload info dump. The same strategy as the NIST.

Actually these two are somehat similar but not the same as the original graph.

#9 has the ~100 hz oscillation but there are two lines not 3.

#10 has 3 lines but there is damped oscillation rapidly reducing the ~100 hz cycles.

Why not the actual original diagram? But since they are your diagrams, you explain them.

psik

 

[Kat Dorman]

Before I reply, let me thank you for posting something which has both good observations and a good question.

 

[Kat Dorman]

Yeah, it is the usual Kat Dorman information overload info dump. The same strategy as the NIST.

There was a reason for posting a variety of graphs and images, which I'll get to after addressing your post.

Actually these two are somehat similar but not the same as the original graph.

Correct, and good observation. They are graphs from different simulations. #9 has a different number of stories (possibly other parametric differences as well, I don't recall). Same simulation software, though, and same model and basic setup.

#9 has the ~100 hz oscillation but there are two lines not 3.

I presume the oscillation is what tipped you off that it's the same environment.

#10 has 3 lines but there is damped oscillation rapidly reducing the ~100 hz cycles.

Again, correct, and good observation. This one is EXACTLY like the one you complained about but with damping turned ON. Only difference.

Why not the actual original diagram?

Because I told you I wasn't going to post the original diagram you criticized until you at least attempt to justify your criticism. You haven't done that yet.

 

[Kat Dorman]

But since they are your diagrams, you explain them.

I will. First, two questions:

1) by singling out #9 and #10, are you saying you have a problem with them?
2) by singling out #9 and #10, are you saying you DON'T have a problem with the others?

I understand if you need explanations of each BEFORE you judge any or all, I just want to know if you have already made any judgements.

 

[Kat Dorman]

Yeah, it is the usual Kat Dorman information overload info dump. The same strategy as the NIST.

The reason I put a bunch of them up is the same basic reasoning as a police line-up. Which of these things are a problem? Also, a little Sesame Street. Which of these things is not like the others?

Somehow, you managed to pick two as being interesting (or perhaps 'guilty', you haven't said which) out of the "information overload info dump." Why these two?

Did you choose #9 because it shows oscillations? I can understand that. That's why I included it, but you were able to discern that it was not the graph in question. Now that you know it's not the same simulation, but similar, do you have a problem with it?

But why did you pick #10? It doesn't show the same sort of oscillations as the suspect graph or #9... though there are damped oscillations as you note. Is it because it's the same color (rhetorical)? Now that you know it's exactly the same as the suspect simulation, only with damping, do you have a problem with it?

Perhaps the visual cues told you that 9 and 10 were from the same "batch" so are likely to utilize massless connections... am I guessing correctly?

 

[Kat Dorman]

But since they are your diagrams, you explain them.

All but #14 are slab models (amply explained starting here). #14 is a 2D mass-spring-damper model of a rectangular sheet subjected to a transient shock on one end.

Need anything more to make a determination as to okay/ridiculous?

 

[Kat Dorman]

While I wait for answers to important questions, here's a visualization which may assist readers in understanding how a mass-spring-damper model can be applied to a slab model:

As always, if you don't understand, ask questions.

 

[psikeyhackr]

Are you saying it is ONLY that simulation you have problem with?

Ho many times do I have to explain the obvious.

Your masses were on 1 kg. So it is reasonable that a massless support be simulating some real object less than 10% of that weight, which would be 100 grams. That is the weight of TOW OF MY WASHERS. But you say this thing must be 3.7 meters long but strong enough to support a kilogram.

You have created a simulation that cannot come close to actually existing but this entire issue is about a building over 400,000 tons.

You create pseudo-intellectual bullsh!t that you do not explain then expect people to spend time arguing with you about it. Then talk about my not reading your posts.

ROFL

I never said ALL MASSLESS CONNECTIONS were nonsense. Just the one you created in this simulation and then didn't tell people they existed. The problem is YOU!

psik

 

[Kat Dorman]

Your masses were on 1 kg. So it is reasonable that a massless support be simulating some real object less than 10% of that weight, which would be 100 grams.

Why is that "reasonable"? Where do you come up with this figure of 10%? Did you do an analysis of any sort, or did you pull it out of your ass?

There must be a reason it's reasonable - in your mind. What is it?

 

[Kat Dorman]

... and then didn't tell people they existed.

I've never told you until now that the 1kg masses were point masses (in that simulation). Point masses don't exist, either. Got a problem with that?

 

[SanderO]

11)

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.

This is brilliant!

 

[SanderO]

I can't post my own PDFs because they exceed 19 KB and I don't know how to reduce them!

 

[Kat Dorman]

I can't post my own PDFs because they exceed 19 KB and I don't know how to reduce them!

Here's one option. There are many others, too. Just find a place which won't dump them after a short time.

 

[psikeyhackr]

I've never told you until now that the 1kg masses were point masses (in that simulation). Point masses don't exist, either. Got a problem with that?

The masses in my Python program were point masses, but they were not one kilogram. But they would have behaved the same way if they were in proportion because the only force I worked with was gravity.

You just create extremely unrealistic BS that was way more complicated and then left out information. How was anyone supposed to know from just the resulting graph that it was a kilogram? I supplied the complete source code with the table so anyone could see what was there.

psik

 

[psikeyhackr]

Why is that "reasonable"? Where do you come up with this figure of 10%? Did you do an analysis of any sort, or did you pull it out of your ass?

There must be a reason it's reasonable - in your mind. What is it?

You are saying that doing a simulation with MASSLESS supports makes sense. So presumably the simulation is not TOO DIFFERENT from reality. If in the REAL WORLD supports cannot be made less than 11% of what can really exist then your simulation is too far out of touch with reality to be relevant.

Unless you think simulations are more important than reality.

Didn't you say the simulations did the same thing whether the supports had mass or not? How much mass did you give the supports in terms of percentage of what was being supported?

psik

 

[Kat Dorman]

The masses in my Python program were point masses, but they were not one kilogram.

You say that like 1kg masses are a bad thing. In fact, you spent months mocking the 1kg masses, though you've laid off recently. Your point masses were WAY more than 1kg so, if things which can't really exist are a problem in simulation, you've got a bigger problem than I do.

But they would have behaved the same way if they were in proportion because the only force I worked with was gravity.

BINGO!!

What you just said is very important. It shows you are aware of mass scaling invariance with respect to gravitational force. Why would you then question 1kg masses, as you did in the past (and seem to be verging on again)?

Because you don't understand that structural capacity - therefore its contribution to resistive force and energy lost to crushing supports - ALSO scales with the mass it has to support.

Do you now understand why the simplest of these 1D models can use 1kg masses and get the same result as 1000kg or 10^6 kg?

 

[Kat Dorman]

I will address the remainder of what you posted, but the prior point is pretty important, so I'd rather wait until you respond before proceeding.

 

[Menard_K]

Dare I speak up here, but what about the possibility,
and what I believe is a very real possibility ... that is in the "collapse" of either tower,
if at any level, the floor trusses at the outside wall connections were the first to fail
and the floor gave-way such to form a ramp pointing to the outside of the tower,
then tons of rubble from above would slide down said ramp and fall to the street
depriving the "pile driver" of mass, and stopping the whole show. What do you think?

 

[Kat Dorman]

I think (once again) it's completely off topic. I'm not discouraging you from posting your ideas, not at all. By all means, start a thread. Here, however, the subject is very narrow and well-defined as indicated in the original post (which I doubt you've read). It's very easy to judge what's on-topic: if it doesn't have anything to do with massless connections, it's off topic.

 

[psikeyhackr]

You say that like 1kg masses are a bad thing. In fact, you spent months mocking the 1kg masses, though you've laid off recently. Your point masses were WAY more than 1kg so, if things which can't really exist are a problem in simulation, you've got a bigger problem than I do.


BINGO!!

What you just said is very important.

Their velocity would not oscillate at 100 Hz. What kind of energy would it take for masses upwards of a ton do that?

BINGO, yeah right. I said the different sized masses would do the same thing under my conditions. Not that they would do the same thing under your conditions.

psik

 

[Kat Dorman]

Their velocity would not oscillate at 100 Hz.

Yours did not oscillate at all! How realistic is that? Have you EVER seen metal (or anything else for that matter) collide SOUNDLESSLY?

100 Hz is low frequency, in case you didn't notice. Most cheap stereos don't even reproduce bass at that frequency. What would steel columns do when smacked together? Thunk and ring! The ringing would be at a MUCH higher frequency than that, maybe into the thousands of hertz. Any transducer (accelerometer, etc) capable of this frequency response would show the same sort of oscillations in a real physical experiment using steel columns in collision. The data would have to be filtered to remove the ringing if the oscillation wasn't the velocity of interest.

What kind of energy would it take for masses upwards of a ton do that?

But, you see, it WASN'T a ton, it was 1kg. Attached above and below to other 1kg masses via a constraint force AKA massless connection. As such, vibrating at 100Hz (or whatever it actually was) takes very little energy at all. What's depicted is realistic action. Things vibrate in collision, and these masses follow the equation of motion for a harmonic oscillator.

This is EXACTLY what a set of 1kg masses subject to connection force per Bazant's load-displacement formula do when a pair collide. There is even other frequencies propagating up and down the structure in waves, reflecting a portion back at each mass point, just like real life. All according to long established physical laws.

Where is your silent collision?


Maybe part of the problem is you think these are large amplitude oscillations? They're not. You can't even see them on the displacement curves, which is one of the reasons you've never complained about them. One of the graphs above is the displacement graph from the very same simulation you're complaining about.

Which means, BTW - your objections in this have nothing to do with massless connections and everything to do with these vibrations which troubled you. That's in part why I posted a police line-up of graphs... to see what sort of things prejudiced you against the simulation which have nothing to do with massless connections.

The graph you do criticize is a VELOCITY graph. It has nothing directly to do with the amplitude of the oscillations. It oscillates some infinitesimal distance around the mean motion, but the velocity is comparable to the translational (collapse) velocity so the oscillation swamps the actual signal. These are very small displacement oscillations, and frankly quite low velocity as well. By the time a collapse was really moving, these "wild" oscillations would look like some noisy artifact and nothing more. The graph you singled out shows the very early portion of collapses.

BINGO, yeah right. I said the different sized masses would do the same thing under my conditions. Not that they would do the same thing under your conditions.

They do and I'll show you why.

 

[Kat Dorman]

Let me put these oscillations into an everyday framework which makes their nature clearer.

You've seen how a bass woofer in speaker cabinet vibrates, yes? This is like taking a speaker cabinet and throwing it out a 20th story window WHILE it's playing some music that has bass. The short throw of the woofer is nothing compared to fall, but the woofer's oscillation is much higher velocity than the initial part of the fall.

If you were to put an accelerometer on the speaker and then integrate the data to find velocity, you'd see exactly the same sort of thing as in my velocity graph. They're not oscillating wildly, for god's sakes, they're oscillating exactly at the frequency and magnitude of two 1kg mass coupled to each other via a force constraint which has reached full compaction; that is, rehardening of the squashed "columns" to an elastic response with MUCH higher stiffness than the intact column and shorter length (typically 20% original length or about 3/4ths meter).

Think of it as a bigass, heavy tuning fork. How would it behave? Deep, bass frequency sound in collision, with vibration VELOCITIES easily exceeding translational velocity if it were dropped.

I'm sorry I ever posted the UNDAMPED velocity graph; it was a poor choice. It shows perfectly natural and physical behavior for the system being modeled but it fails to convey the pertinent information. Do hear that? It's my fault I put a crappy representation up. I posted the damped one later; the translational velocities are nearly identical, but you can actually see it.

 

[Kat Dorman]

BINGO, yeah right. I said the different sized masses would do the same thing under my conditions. Not that they would do the same thing under your conditions.

Let me explain why they DO.

The static capacity can be expressed as a multiple of the load imposed (mg). The mass m is the mass that a support at a given level must hold, so static capacity is FOS*mg. The resistive force at a level, be it static or dynamic, is proportional to the mass above that level - just like the force of gravity is on that same mass.

No matter WHAT mass distribution there is, if a structure is to be self-supporting, the very MINIMUM static capacity must exceed the imposed load. You've made a big point of this, and you're right. But what that means is that capacity (and therefore energy lost in crushing) scales directly with mass just like gravitational force. The very same reasoning you apply to the momentum-only Python model also applies once failure energy is added in.

Both resistive force due to momentum exchange AND due to structural resistance scale directly with the mass (actually, also load=mg). If the FOS is uniform - which is of course unrealistic, but not a bad approximation - the resulting motion is unchanged in this model when m is replaced with k*m, k any number. Just run through the math yourself.

F = ma => a = F/m

F = (force from momentum exchange) + (force from structural resistance) = F' + F".

=>

a = (F' + F")/m = F'/m + F"/m

Force F' from momentum exchange you already know is invariant to mass scaling, as you acknowledged above. That's because it's of the form F'=d(mv)/dt, and mass cancels out when calculating acceleration. Force from structural resistance is of the form F"=kmg where k is a constant of proportionality.

Therefore, mass also cancels out of the structural resistance term: a = F'/m + F"/m = F'/m + kmg/m = F'/m + kg.


It gets stronger going down the building because the mass supported is greater. How many times have you said that?

(PS same with energy dissipated: W = FΔx = kmgΔx)

 

[SanderO]

Here's one option. There are many others, too. Just find a place which won't dump them after a short time.

Thanks...

Not interesting in publishing in that option but inserting a PDF into a post here would be handy.

 

[SanderO]

Dare I speak up here, but what about the possibility,
and what I believe is a very real possibility ... that is in the "collapse" of either tower,
if at any level, the floor trusses at the outside wall connections were the first to fail
and the floor gave-way such to form a ramp pointing to the outside of the tower,
then tons of rubble from above would slide down said ramp and fall to the street
depriving the "pile driver" of mass, and stopping the whole show. What do you think?

Makes no sense at all. You don't understand the geometry and the facade itself was only 20% open (where the glass was)