The 9/11 Forum

Time for a sanity check. Go back to the fundamentals, Greening's ET. Page 4, Stage 1 collapse time for WTC 1, 14 story upper block => 11.6 seconds.

If I make the joints of the lower block quite weak, but still well above non-existent, and the upper joints really strong to ensure crush down, I get a time of about 10.7 seconds - with a stretch of 0.2. With a stretch of 0.02, it's about 11.7 seconds. After a slight vertical offset to account for our different starting heights, it lays right on top of Dr. Greening's WTC1 curve.



I think that passes sanity check. Not bad for 1kg floors, eh?

Might be useful to compare velocity of the last two runs, where the only difference was stretch. All the runs so far, unless otherwise noted, are 0.2 but, to match Greening's momentum only Stage 1 time, I had to go for high compaction.



As noted towards the beginning of the thread, the biggest influence of stretch is make the fall longer or shorter. There is some difference in velocity, the higher stretch being a little slower, but the accelerations end up the same, and constant.

The simulator is not junk. It's a little difficult to reconcile a figure of 0.6g with 0.3g, however. Maybe WTC1 starts at 0.6g and goes to 0g, whereas this starts at g and goes to 0.3g, and it all comes out in the wash. ???

In another thread, there's discussion about how much momentum transfer counts, as opposed to resistance from structural strength. Dr. G made a comment about the falling rain drop problem, in reference to the similarity in mass accretion. He states:

Dr. G wrote:The simple momentum transfer solution yields the result that the raindrop should fall at 1/3g, which is not observed for WTC 1.

Time to investigate the effect of connection strength. ET says 2x overestimate of E1 adds only a half second to collapse time.

These runs use a single top slab with 10x the normal mass to crush down 100 floors with connection strength varying vertically at a ratio of 50:1 going up. The overall strength was set with multipliers of 0.25, 1, 2, 3 and 4x. At 4x, crush down arrested partway through. Not much more than a second's difference amongst those that go to completion.

Displacement:


Velocity:


Finally evidence of a different terminal slope!

Dissipated energy:


There isn't much difference in magnitude between the vastly different connection strengths, all of which are capable of supporting the structure. Compare the green and blue curves, which represent the difference in energy thrown to the aethers given by a 12 fold increase in connection strength. Most of the energy loss is from inelastic collision.

The arrest is interesting. A slight departure from the tight pattern becomes evident just before a huge step change as the lower block top slab is able to withstand one of the impulses from the Zone A/B mass. The gradient of connection strength is quite steep, though not quite as steep as the load gradient.

Typical cumulative totals of dissipated energy versus connection break count:



Nice blend of a parabola (velocity dependence in inelastic collision) and a straight line (connection count and strength).


These next comparisons are between 1, 10 and 100 mass units for the top slab crushing down 100 floors below.

Displacement:


velocity:


Even with the 100x top mass, which is equal to the combined mass of all slabs below it, the tendency towards a similar terminal acceleration is suggested. Momentum transfer from collisions results in terminal acceleration as opposed to terminal velocity.

In the extreme example (100x top mass) graphed in yellow above, you talk of "terminal acceleration".

If the lower block were taller (200 floors) rather than 100, do you think the yellow velocity graph will level off to some near constant "terminal acceleration" like the blue and red examples?


It could be that it struck ground before it could reveal the slope at which it would level off.


Educational thread.


It is nice the way you compare the zone C structure as if it is in a falling elevator. (Mathematically the same as being in a reduced gravitational field).

You introduce an interesting balance of forces that the analytical approach has ignored. How does the analytical approach treat zone C connections? It doesn't.


It is interesting to see "upper block" momentum being applied to the lower portion only through connections. It is so much more believable and workable than some oversimplified differential equations which have been treated like little gods.

There are many factors to consider, like connections for example, that cannot be treated through analytical equations with a very finite number of variables.

Oh what a tangled web is woven when hitting 'quote' instead of edit. Duplicate post is deleted.

Major_Tom wrote:If the lower block were taller (200 floors) rather than 100, do you think the yellow velocity graph will level off to some near constant "terminal acceleration" like the blue and red examples?

I think so. Let's find out.

One 100x mass slab crushing down five hundred 1x mass slabs:


Yes. I'll explain the waviness of the line in a bit. This result is not a surprise, see below.

It is nice the way you compare the zone C structure as if it is in a falling elevator. (Mathematically the same as being in a reduced gravitational field).

Just to clear up any possible confusion in what I said earlier, if WTC1 Zone C is riding the elevator at two-thirds g, it's like being in a one-third g world. Coincidentally the sims come in at g/3 which is like being in a two-thirds g world. Will the virtual structure crush up at one-third g? Yes, it will. Even at one-tenth g.

Comparison of displacement over time, g and g/3 fields:


Not entirely fair to compare with simultaneous crush down, because the initial average acceleration there is close to g, the impacting slabs start with an initial velocity of half the upper block and freefall down to the next slab, as does the upper block. Not the same as a crush up into ground at one third gravity, but still useful.

Major_Tom wrote:You introduce an interesting balance of forces that the analytical approach has ignored. How does the analytical approach treat zone C connections? It doesn't.

It has been addressed, though quite different* methodology and slightly different result, pages 5 - 7(PDF) of B&L here: http://www.civil.northwestern.edu/people/bazant/PDFs/Papers/D25%20WTC%20Discussions%20Replies.pdf

What's of interest is that (if I interpret correctly), it's not too far off from going the other way, with crush-up, despite the differences in model and methodology. Look at the the upper left graph of four in the top corner of page 6 (PDF, or p918 journal). Doesn't seem too far off from crossing the threshold into crush up but, since it doesn't and won't, it's understandably not pursued. I don't really have a choice until I twiddle around to find the right parameters, where it crushes down. But the crossover point is of interest, anyway.

Given the tilt and all, I'd expect the bottom of the upper block to be quite vulnerable and the result of collision or entanglement to be unlike that of axial strikes. A tall tipping structure, like a smokestack, usually breaks on the way down. Take an intact (except for free column ends) upper block and set it gently on a 7 degree grade; how will it fare? Tilt the upper block 7 degrees and set it on a level surface. Drop it 0.5m.

Now the whole point is the lower block breaks away in the analytical scenario and the upper block, though damaged, is not compromised so can apply the force necessary to immediately accelerate the impacted slabs to upper block velocity. The mass accrual has already begun, and the inertia of those two floor slabs only makes the battle easier in the collision with the next.

In these simulations, it's difficult to get the lowest connection to survive (visibly past) the first impact, so the dynamics immediately take a different course. A slightly slowed Zone C, a greatly slowed (and small) Zone B, and stationary Zone A. Zone B may impact first and break the next floor down, but hasn't retained much velocity by the time Zone C comes thundering down from its slightly impeded two story drop. It runs into a free moving Zone B of mass 3m and the lower connection breaks again, and so on. It typically takes quite a few jolts to bring Zone C and B to a speed close enough for the remainder of C to survive.

If there were to be simultaneous crush up, this analysis suggests the upper block will descend noticeably quicker than it would with sole crush down. This difference matters for measurements made from the roofline or antenna, where it may matter little with respect to crush front motion or collapse time. Failure to account for the additional degree of freedom, should it exist in the real scenario, would toss quite a wrench in any analysis that assumed fitting a resistive force to crush down only.

I hadn't noticed this before but it's another validation of the engine:

BL wrote:The acceleration (v_B dot) rapidly decreases because of mass accretion of zone B and becomes much smaller than g, converging to g/3 near the end of crush down (Bažant et al. 2007).

Nothing new, if I'd been paying attention. Googling on falling raindrops wasn't overly helpful, here's one link:

http://www.iop.org/EJ/abstract/0143-0807/22/2/302

Abstract wrote:A standard undergraduate mechanics problem involves a raindrop which grows in size as it falls through a mist of suspended water droplets. Ignoring air drag, the asymptotic drop acceleration is g/7, independent of the mist density and the drop radius. Here we show that air drag overwhelms mist drag, producing drop accelerations of order 10-3g. Analytical solutions are facilitated by a new empirical form of the air drag coefficient C = 12R-1/2, which agrees with experimental data on liquid drops in the Reynolds-number range 10<R<1000 relevant to precipitating spherical drops. Solutions including air drag are within reach of students of intermediate mechanics and nonlinear dynamics.

my bold

OK. In any case, ~g/3 is apparently the result the simulations should be giving under these conditions. There's something to be said for reproducing Greening's momentum-only result without doing a single calculation....

This thread (http://forums.randi.org/showthread.php?t=104558) has some interesting posts, a few very worthwhile, a lot of the usual tripe as well. Haha, looks like I've reproduced beachnut's work, I'm so proud.

Major_Tom wrote:It is interesting to see "upper block" momentum being applied to the lower portion only through connections. It is so much more believable and workable than some oversimplified differential equations which have been treated like little gods.

Little gods are the most demanding and temperamental. Would that I could work through these simplified equations, and maybe someday I shall, but this is too easy and fun right now.

There are many factors to consider, like connections for example, that cannot be treated through analytical equations with a very finite number of variables.

Good point, I find the problem of masses coupled by springs to be distasteful. I won't say unworkable because I haven't tried it. Coupled harmonic oscillators with different initial velocities and springs that break... no wonder people want to make the upper block rigid.

This has been a simple start - intentionally, to ensure each step gives sensible, if not acccurate, results. So far, the results are encouraging, it has been possible to play with a variety of parameters and learn some things about scaling up in size and total member count. More complicated arrangements, to introduce new factors, are in the works.


*far more sophisticated, too, like oranges are far more sophisticated than apples.

About the wavy velocity line. 500 stories is about the limit. Not for the physics engine, but for the structure. 1850m tall, 0.74m wide, amazing that it stands at all. Maybe it wouldn't, haven't tried the static structure over time, but at least it doesn't break anywhere but the crushing front. There's so much flexure, the lower block has quite a few waves traveling back and forth during collapse. Sort of a Tacoma Narrows bridge thing going on, it can be seen in the bumps of the velocity curve. Those aren't jolts, they're resonance effects. Repeated slamming at the peak strength of the connections.

In this sense there's no cheating with the engine. Want a rigid block? Better make it that way or it might crush up. Think all the energy is dissipated at the crushing front in a 500 slab structure? Nope, it isn't. It's not how real-world materials behave, it's not how virtual world materials behave, either. Choose E1 carefully. The structure has to stand under its own weight; too strong and it refuses to break.

The other side of the coin is pushing the engine into bad behavior, something I've managed to avoid lately, but it was not without a good deal of time spent in trial and error. Even things like choice of time slice (with the consequent binary floating point representation, not so much the size) have a profound influence on the stability and accuracy. Now, having found the sweet spot, and having a good idea of the warning signs when things go awry, I'm pretty confident of the 1D results so far. Moving to more elaborate models, and then 3D, will bring new considerations and perils. My hope is to be able to get good simulations of matter transitioning from lattice structure to granular, near-fluid state under the influence of internal defect and its own weight, spanning a variety of conditions and configurations.

The 1D behavior is most curious but fortunate. Without it, there would be little hope of recreating these 'ideal' 1D analytical scenarios using this system. Turns out spheres in a vertical line will stay on the line, even bouncing at high speed elastically. Real slabs go 3D immediately, with rotation and asymmetry, but spheres are stably 1D in this arrangement. Actually did some 1D gases, helped to work out the kinks in the system.

I've been spending time in femr2's simulation thread, and it's been enlightening. I'd link to it, but anyone reading this has probably already been there. There are a bunch of spreadsheets that have been sitting open for many days, with obscure, coded titles that are difficult for even me to decipher. I need to post some pictures before I forget what they're about; sorry if some of this is short on explanation. Confidence in the general accuracy of the simulations has only increased.

Comparing largely inelastic to largely elastic collisions
In the engine, this is mostly governed by the restitution property of the colliding bodies. For inelastic, this value is set to zero, which does not provide the textbook definition of fully inelastic because of the object skin mentioned earlier. The elastic trial uses a restitution of 0.9 for both bodies, and there's quite a difference between that and 1.0, the maximum.

Perhaps surprisingly, there's not a tremendous difference between the two extremes. Slabs that can rebound more get involved in more collisions, which dissipate energy quickly. For this reason, I include a graph of collision count per timestep (which is small at 0.002s). The net effect is little difference in other quantities for such a drastic change in material properties. The crush front does not race wildly ahead of the inelastic case.

Top displacement:


Kinetic energy:


Dissipated energy (tail included to show there IS a difference at the end):


Crush front location:


Collisions per timestep:


Note the elastic case reached a maximum collision frequency of around 12,500 Hz! Normally, that wouldn't inspire much confidence in the simulation but it actually behaved quite well by every other standard. Outwardly, it was barely distinguishable from inelastic, which has some pretty high transient counts as well. Plenty of collisions for recrushing, or a smaller number of more energy-dissipating collisions, doesn't seem to matter much in observables.


Stretch
The subject of stretch came up in the other thread, so I thought I would delve into it a little more. It's the reciprocal of the compaction ratio, which may be more intuitive but is sometimes less convenient. Here is a comparison of two extremes, 0.1 and 0.5. The latter has the 'special' property that, upon collapse completion, the pile is half the height of the original structure. Obviously, this has no relation to anything significant, it's just to illustrate how stretch affects the measurables in what is an accretion problem.

These are the roofline and crush front locations, red is s=0.1 and blue is s=0.5:


This graphic overlay depicts the region extents at start, mid and end:



The obvious conclusion is that increasing stretch results in higher/slower roofline displacement, and relatively greater magnitude change of crush front displacement in the downward direction. Later, I'd like to explore some curve fits and interpolations, but this lets me close those spreadsheets.


Delayed crush-down
Oh yes, one other oddity. Initial drop breaks one lower connection, then there's a sustained, exclusive crush-up followed later by re-initiation of concurrent crush down. Unfortunately, I don't remember much about the conditions, and I don't want to look them up in the archived index. Undoubtedly, it was contrived. Locations of Zone A/B/C limits over the first five seconds:



700th post. Maybe I should get a life?

OWE, if you get a life, we won't have such interesting things to read about and we'll be forced to get a life.

By the way, I do hope you decide to summarize your work in a paper one day.

Haha, thanks. At some point, it may become coherent enough for a graphic novel.

OWE, this is a nice model to work with.

There is one difference between it and my impression of real building collision behavior.....

In your runs if zone C ceases to exist as an intact structure and combines with zone B, crush down can still proceed as an almost uncoupled independent type of motion. I understand why.


In real crush-up, crush-down demos there is no reason to believe this is true. There is reason to suspect that if zone C ceases to exist as an independent, intact structure, the flow crushing down on zone A changes it's nature.

I am observing the way that the planners of this type of crushing demolition are prepping the buildings and I do not believe that they feel they can ignore the possibility of "using up" zone C too soon into the collapse. Zone C structure seems to be an important tool for them to do their jobs.

In your model (and the Dr G momentum transfer and BGLB) the crushing mass from zones B and C are simply added together, so it doesn't greatly matter if zone C is totally converted into zone B. I see that in your examples and that is why you can treat the situation roughly as 2 independent uncoupled problems.


Every one of these discrete (or continuous?) simplified models do not distinguish between the natures of zone B and C. To you they are just mass M, so what is the difference if portions of the mass exist in an intact state (zone C) or crushed state (zone B).


But in reality, do you think that actual demo planners wouldn't care if the upper block was "used up" too early in the demo?

Would they likewise assume that crushed mass is as effective as intact zone C mass midway through the process, so what's the difference? It's all M, no?

I do not think so.


Reality check. Let's take the Balzac demo and and imagine the exact same situation, but this time we place about 20 or 30 more stories below the building to be crushed also.

We see an approximate crush-up, crush-down ratio of 1:1 (I don't care if it is 1:2, the same argument applies).

According to your model or the Greening or BL approach, if zone C was totally "used up" with 20 more stories to crush, zone B would just keep on crushing down and there wouldn't be much overall difference in results.

I suspect in reality the total conversion of zone C into zone B with 20 stories to go would make a noticable difference in behavior at that stage and there is no guarantee whatsoever that zone A would be completely destroyed.


I do not beleive that zones B and C masses are of the same nature and would "pack the same punch" to the structure below. They may not be so easily interchangable as your models treat them.


Demos do fail and "peter out" from time to time and there is no way to explain by merging zones B and C together as if both have the same interchangable effect.

We observe demo planners taking great care to destroy systems that, according the BZ or momentum transfer analysis or even your model appear so easy to take down.


If 50% or more column failure on only one floor makes total destruction inevitable as BZ claims, then demo planners seem a bit stupid for doing so much extra work.


In your models, it is just where zone C disappears but the process keeps on going as if it does not matter that seem most suspicious to me.

Your ability to decouple the two processes seems unphysical, though it makes perfect sense in the context of the model.


(There is a problem in the way zone C mass is seamlessly converted to zone B mass, also rigid, which doesn't seem to make sense with a substance like shattering concrete. That may be a big difference between theory and real demos).

You follow my point?

Major_Tom wrote:You follow my point?

Yes, I do. I just wanted to tell you that because your post contains a lot of really interesting points. A real reply is coming, it's not easy and thanks for the stimulus. If we were face to face for an afternoon, maybe with a few pints, the information exchange would go much quicker.

It is true the simulation doesn't care too much about what's above the break, it looks pretty much the same from below. Full entrainment of all members (neglecting shedding here). It is a little more like real life than than a stepwise model without shedding because energy is dissipated into all the connected slabs when collisions occur, you know, the 'harmless' elastic waves. There is a bit of rattling around of the slabs, but hardly like a debris flow. It's so unlike the actual collapses that it's amazing it's even in the ballpark on times.

Is it strange to say that I have a hard time visualizing the real collapses, even while looking at them?

I don't know what would happen with Balzac+80, but I'll take a guess. In the actual, the pile-up of rubble (raising the effective incompressible ground level) together with the low total number of stories on top makes it slow down rapidly in the latter portions. If it were instead a structure below which did have a yield point, the rubble could go down and, when it goes down it gains energy. Whether it would be self-sustaining indefinitely, who knows, but I'd bet it would go farther on this basis alone.

Skipping ahead...

I think open floor plans are a bad idea, architecturally. An apartment complex is just that, a complex of compartments, a repetitive cellular structure that may have unremarkable vertical load capacity but dissipates energy every inch of the way when crushed. Any rectangular subsection will be as strong separately as together, more or less. Nothing is very strong, but it doesn't break into nice uniform lengths and fold up like a lawn chair, either.

The 1D has been milked, maybe not to death, but the shortcomings you point out cannot be waved away. 2D is on its way, slowly. It will allow for more complicated geometry of interactions.

At this stage, it may be worth mentioning that the 1D world really hasn't outlived its usefulness. There is more accumulated info on simple 1D simulations than what's been posted here. The applicability is not immediately obvious, so going through ALL of the details to show the limited usefulness is questionable. The most interesting was attempts at systems akin to 1D gases. Originally, the purpose was to get a handle on the behavior of the simulation engine with as few variables as possible. I'd already done very simple 3D systems with quite chaotic results and it seemed appropriate to limit the system even more to ensure this was not just a stupid cartoon generator.

Fortunately, I found that spheres in original 'perfect' alignment stay in alignment so long as no lateral forces are introduced. I find this remarkable, and interpret it positively in regards to the stability of the engine. While no set of real spheres would exhibit this behavior, in theory the divergence is easily explained by the sensitivity of the problem to minor variations in placement, surface irregularities, deformations, and so on. In experiment, the same set of balls dropped with the best precision possible would display radically different trajectories and final positions because of these factors. Only statistical generalizations about the system can be made and results from simulations can contribute to this picture if run in large numbers with small (semi) random variations. The point is, without sufficient descriptive knowledge to build into the simulation, any particular run is of limited value in a chaotic system.

Consider how that may apply to the NIST simulation. While indeed 'detailed', it can easily be argued that it is not detailed enough to ensure convergence on the proper specific solution if one is sought, and that it is too detailed to represent insensitive, gross characteristics.

The spheres in PhysX behave well in the sense that no lateral force component is spuriously introduced via approximation tricks, floating point error and what have you. What is a 3D simulation then acts as if it is 1D. With that, multi-body collisions act in a (more) tractable domain so it's easier to set up systems that have grossly predictable behavior and see how well the engine does. In the course of doing so, several non-physical glitches became apparent, mostly in the area of time increment, and strategies to avoid bad behavior could be formulated and verified.

In the 1D gases, there is an opportunity to explore what happens when there are many collisions occuring between a multitude of bodies, in terms of just a few simple scalar parameters and numeric results. A 1D gas, in this context, consists of free bodies moving on a vertical line under the influence of gravity. The entities cannot cross paths, so maintain the same ordering. In one set of experiments, the top sphere was made much more massive than the collection below, so functioned as a (piston-like) load applied to the gas. The 'temperature' of the 'gas', as in statistical physics, is based on the ensemble kinetic energy. Initial temperature is set by assigning random initial velocities in the one dimension, then the system evolves under the influence of gravity and the applied load at the top.

Naturally, the system will bounce the load. If the restitution property is large enough, it can bounce indefinitely. In any case, the system behaves as a harmonic oscillator with a characteristic period. Since many real-world building materials have a non-trivial coefficient of restitution or, more appropriately, do not exhibit fully inelastic response to collision, the 1D gases helped characterize what can be expected of the simulator when restitution is varied. Previously apparent was, in general, the more chaotic simulations resulted from introducing elastic behavior in the members. While this may or may not be more realistic in any given sense, to get useful results while avoiding the need to do thousands of runs with minor variations (discarding the obvious failures), restitution must be set to zero and sometimes additional damping introduced to keep things from blowing up. Unfortunate, but true.

Where the gas study helped was in observing that the engine was capable of simulating pressure, that being the result of internal kinetic energy of an isolated portion of the system transferring momentum to another portion via high frequency (essentially discrete) collisions. Pressure from rigid body mechanics.

In one collapse sim earlier, I let the collisions be mostly elastic and it didn't change things much. I'm not sure what femr2 did in the async sim, but it looks like it went a little nuts. If I let the restitution be unity, my sims go ape, too. Clearly this is not a fruitful avenue; intuitively, there's no reason to expect the failure to propagate upwards in a 'fracture wave' (at least in this manner) and certainly there isn't much in the way of evidence to support it.

However, that's not to say that damage cannot propagate well away from the average location of the crush front, due to preservation of KE post-collision in the internal degrees of freedom. But, when you consider issues such as mean free path and average time between collisions, it's not obvious to what extent that can occur in real life. Higher restitution means higher post-collision velocity, which in turn means more frequent collisions. Thus the KE still gets drained off rapidly for reasonable values of restitution, a non-linear effect in a confined system.

Later I may post some of the 1D results of non-collapse systems, but I think this covers the gist of it. Where there is still some ground to explore in 1D is in composing 'stories' from spheres of non-uniform properties, as opposed to having a single sphere to represent a concentrated, rigid slab (which paradoxically is also inelastic). I believe that a very crude approximation to story density, compressibility, and elasticity can be incorporated which would make the sim more realistic than slabs. A story unit can then have internal degrees of freedom and exert pressure on its surroundings. Work is done against a virtual resistive force applied over distance instead of discretely once per story, stemming from repeated collisions and more gradual entrainment of mass.

Not for lack of interest, but for lack of time...

1D has many uses left, and 2D too, but going back to 3D is probably the next thing.

Might as well accumulate some of the earlier efforts here. Unfortunately, I lost original code and results and had to start over, never quite got my steam back for development, but it was good to start from scratch and build slowly and carefully, some of the early stuff was far too ambitious.


Early work before the loss, there's some really ridiculous stuff I may still have around, but these are the ones that are easy to dig up, very simple:





And these are really simple structures done after starting from scratch, all '1D' models executed in a 3D space using slabs, as opposed to the spheres used so far in this thread (which maintain 1D motion):

I want to go through this toy model and the different cases to see how it can be applied to possible OOSRD movement.

- constant mass
- varying strength
- true 1D in 3D
- dynamics calculated using spheres
- rendered as slabs with height = diameter of spheres
- camera tracks with topmost member
- sequence ends at arrest

How can we manipulate this to use it as an OOSRD learning tool? The only thing we need to change is the connection strength. We want constant strength, not varying strength.

>>>>>>>>>>>>>>>>>>>>>>>>>>>>
On a comparison of the toy model with Bazant's claims OWE writes:

The assumption of the prevailing theory, which is a continuum model, is that the upper block remains relatively undamaged until the end. I've read the part of B&L that addresses this (many times) and, while I understand the majority, I'm not in a position to check the validity of the work. I'll assume it to be correct and conclude that this is a significant difference with a discrete floor model - the upper block crushes up because it would if dropped onto a moving platform accelerating downward at a significant fraction of g, whereas in the case of continuous media this is not the case

We now know that the claim in B&L of an undamaged upper block is based on the mistaken assumption that because columns cannot buckle upwards, the upper block will be preserved.

OOSRD does not depend on column buckling so we find the whole idea of a preserved upper block to be silly. We will set all connection strengths equal (except mechanical floors) and see if we can create OOSRD.

Much of the collapse arrests in the OWE examples depend on connections that increase in strength as the collapse moves down the building. We don't have this increased resistance for OOSRD, so you'd think that OOSRD dynamics collapses to earth easier than his examples.

>>>>>>>>>>>>>>>>>>>

OWE wrote:

If this style of simulation offers anything of value by way of comparison to physical progressive collapse, at least in these two examples, it is:

1) mixed crush direction should be distinguishable from exclusive crush down via roofline measurements of sufficient duration, by virtue of a faster descent

2) having mixed crush direction does little to alter the energetics of the overall problem, so ignoring it is OK (sometimes)

Is that true for WTC1? How can we distinguish OOS crush-up, crush-down or mix on the basis of our super-cool Sauret measurements?

The second point is very important for OOSRD. Would the order of crush-up before or after crush-down really matter to the ability to induce and sustain OOSRD conditions?

>>>>>>>>>>>>>>>>>>>>>>>>>>

OWE writes:

The simulator is not junk. It's a little difficult to reconcile a figure of 0.6g with 0.3g, however. Maybe WTC1 starts at 0.6g and goes to 0g, whereas this starts at g and goes to 0.3g, and it all comes out in the wash. ???

It is not junk. This does need to be reconciled (understood).

I have not seen any cases of terminal velocity in OWE's examples. Always accelerations and one phenomenon called "terminal acceleration". How can we explain terminal velocities witnessed moving down the southwest corner?

Achimspok posted a nice little video recently on the OOSRD reaching what seems to be terminal velocity after reaching only the 85th floor.

It's a new release so maybe you haven't yet seen it at

http://www.youtube.com/user/achimspok?&MMN_position=313:313#p/a/u/0/Q-jMyMGBzYc

OWE may argue that if we consider resistive forces that are proportional to v or v^2 we can be expected to reach a terminal velocity. True, but after only 10 floors? This means that the resistive term with v^2 dominates?

For OODRD we guess that a constant resistive force will dominate with smaller contributions from terms with v or v^2 (why not). How can the v^2 resistive term dominate so early? Are there other natural examples of such dynamic behavior?

What would a -c*v^2 resistive force represent physically? (I know you'll say concrete pulverization, but does that really explain terminal velocity reached at floor 85?)