Such software does not exist. What you have to do is to first model the complete structure with FEM/beams and then decide the local failures step by step and redo the FEM analysis at every step. An iteration of kind.
If at a certain step you find that something free falls you have to include that in the analysis and apply the relevant loads (where they start again).
Many (10?) years ago the the Japanese did this with two bodies; one (A) a very solid bow of a ship (shell plates, strong horizontal webs, transverse frames) driven by a constant force, the other (B) the vertical plate side of another vessel and its internal structure. When A hit B, A penetrated a fair distance into B causing plenty of structural damages but A was finally arrested a bit into B. The whole thing maybe took 7 seconds full scale. In order to analyse this very non-linear collapse the Japanese split it into 700 steps, analysed the damage at every step, adjusted the FEM model + loads accordingly, and calculated the next step, etc. It took them 3 weeks with plenty of PCs running in parallell to do the analysis. And then a full scale test was done (real ship bow A hitting real ship side B) for comparison. It was quite good comparison. The vertical side of B lacked initially resistance to withstand the solid bow structure of A but after a while the assembly ran out of energy.
Heiwa, this is the most cogent point I've ever seen you make here. So, I think it's worth comparing the parameters of the ship collision scenarios you're describing, as compared to the WTC tower collapses.
I believe the tests you're referring to, where full-scale tests were corroborated with detailed computer modeling, are the same ones being described here: (Source:
http://books.nap.edu/openbook.php?record_id=5798&page=243)
Verification of analytical procedures using scale-model tests and actual collision data—where available—is a necessary part of the approach because of the inherent difficulty in modeling highly contorted collapse modes and the relatively crude criteria that are still employed to model plate- and weld-fracture during crushing. Full-scale collision tests have been conducted using two inland waterway tankers, each of approximately 1,000-metric ton displacement, in a collaborative effort with support from a number of Dutch and Japanese groups (Vredeveldt and Wevers, 1992, 1995; Wevers et al., 1994). These tests were accompanied by detailed numerical simulations (Lenselink and Thung, 1992). A series of four impacts were conducted wherein one tanker fitted with a nominally rigid bow struck the other tanker's side at 90 degrees. Two of the impacts were against side sections of the ship having a single hull, whereas in the other two collisions the tanker was struck in side sections having a double hull. Data were recorded on penetration depth, collision force, strains in critical locations, and all six rigid-body motions of each of the ships. In addition, observations on cracking, which largely occurred along weld lines, were reported. The accompanying numerical simulations were successful in replicating major features of the collision, with the exception of crack patterns. The experimental data will be available for calibration of analysis methods in the future. Among the conclusions drawn from the joint Dutch-Japanese research were these: fracture initiation is dominated by the welds and is poorly characterized; the hydrodynamics of both ships during collision must be modeled correctly if penetration and collision forces are to be predicted accurately; and a sizable fraction of the energy dissipated in a collision goes into wave generation.
I'm making a few assumptions on the basis of the tests being designed to simulate realistic ship-ship collision scenarios. If you any of my quantitative assumptions are invalid in a way that would affect the results of the comparison that follows, please provide a rationale and a more suitable value.
I'm assuming that the T-bone collision described in the passage above takes place with both ships steaming at full cruising speed. This would appear to be close to the worst-case. (The tests were probably conducted with the target ship stationary instead, but I'll allow the extra kinetic energy of having both ships in motion. A collision closer to head-on might increase the kinetic energy available, but would not appear to greatly increase the amount expended in the collision, as the ships would tend to slide past each other.)
Just to help visualize the kind of collision we're talking about, here's a 1,000 ton tanker for sale, with photos:
http://commercial.apolloduck.com/feature.phtml?id=77620
I'm using 15 knots (7.22 m/sec) as the ships' speed. This is slightly above the typical speed of large tankers on the high seas, so it's probably a significant overestimate for a small tanker navigating in inland waterways. However, if the testers were testing worst case scenarios, they might have had the ships cruising at full throttle. So, the magnitude of the relative velocity for the T-bone collision geometry is 7.2 * sqrt(2) = 10.2 m/sec
Mass of the ships is given; it's the same as their displacement, 1,000,000 Kg.
Assuming both are under full power and remain so for the duration of the collision, there's also additional energy input equal to no more than the power of the ships' engines times the duration of the event. I haven't found a reference yet for what a typical engine power of such a tanker might be, so I'm estimating it using an Admiralty Coefficient ( = displacement in tons ^ (2/3) * speed in knots ^ 3 / horsepower) of 450, which gives me 750 horsepower. Let's again overestimate for the worst case, say 1,000 hp instead.
For a tower, the mass is 350,000,000 Kg, and approximate height of the center of gravity (lower than the center height, to account for the heavier framing lower down) is 175m. In contrast with my assumptions for the ships, that's a conservative estimate.
SHIP COLLISION
Starting kinetic energy: .5 * m * v^2 = 52,000,000 Joules
Gravitational potential energy: negligible (and no mechanism for converting gpe into forward momentum contributing to the collision)
Additional energy from the ships' engines, for the collision duration: 7 seconds * 1000 hoursepower (746000 J/sec) * 2 ships = 10,450,000 Joules
TOTAL ENERGY AVAILABLE: 62,450,000 Joules
TOWER COLLAPSE
Starting kinetic energy: zero
Gravitational potential energy: m * g * h
cg = (approx) 600,000,000,000 Joules
TOTAL ENERGY AVAILABLE: 600,000,000,000 Joules
That's a little under
ten thousand times as much energy available to drive the collision in the towers as in your colliding ships.
If we consider only the potential energy of the upper blocks at the start of the collapses, we're still on the order of a thousand times the energy as in your colliding ships.
Do you think that if you put 1000 times more energy into your ramming ship -- say, by increasing its speed to about 475 knots -- it would still "come to rest" in the other ship?
Not only that, but for the colliding ships, only a portion of the kinetic energy has to be expended in destruction of the structures. The ships are likely still moving at the end of the collision, so some of the kinetic energy is still there. The towers, on the other hand, must come completely to rest for the collision to be over; the lower structures are pinned against the ground. Also, as the passage above notes, for the ships "a sizable fraction of the energy dissipated in a collision goes into wave generation." This represents a significant energy dissipation mechanism that the towers didn't have. The ship collision is taking place in a giant fluid shock absorber.
Does this help you to understand why your intuition about what should have happened to the towers, based on what happens to colliding ships, is unreliable at best?
Respectfully,
Myriad
There were a great number of skeptics who thought it would never stand.