Abstract
Catastrophic disasters of large-scale framed structures occurred recently
are mainly caused by sudden, extreme external loads such as aircraft collision,
explosion, large seismic excitation, and big fire. Dynamic codes are generally
used to investigate such phenomena. However, strong nonlinearity in the
deformation of structures and rapidness of the external loads often generate
higher hurdle in the analyses. The authors have developed an adaptive finite
element code with the use of an ASI (Adaptively Shifted Integration)-Gauss
technique, which provides higher computational efficiency than the conventional
code in such analyses, and enable us to cope with dynamic behavior with
strong nonlinearities including phenomena such as member fracture and elemental
contact. Contact release and re-contact algorithms are also developed and
implemented in the code to realize complex behaviors of structural members
during impact and collapse sequence.
The code has been applied to various collapse analyses of buildings. One
of the various collapse simulations is a fire-induced collapse analysis
of a high-rise tower, which is carried out for an investigation seeking
for the true cause of the total collapse of New York World Trade Center
(WTC) towers, which collapsed in 2001.
A demolition planning tool based on a parameter called the key element
index, which indicates the contribution of a structural column to the vertical
capacity of the structure, is also developed by using the code. Two ways
of selecting specific columns to demolish the whole structure are demonstrated:
selecting the columns from the largest index value and from the smallest
index value. The demolition results are confirmed numerically by carrying
out some collapse analyses, and the tendencies of the demolition modes
to follow the key element index values are estimated. The numerical results
suggest that for a successful demolition, a group of columns with the largest
key element index values should be selected when explosives are ignited
in a simultaneous blast, whereas those with the smallest values should
be selected when explosives are ignited sequentially, with a final blast
set on a column with a large index value.
A seismic pounding analysis is also performed on a simulated model of the
Nuevo Leon buildings, in which two out of the three collapsed completely
in the 1985 Mexican earthquake, to understand the impact and collapse behavior
of structures built near each other. The difference of natural periods
between the buildings, which was caused by the damages from previous earthquakes,
may have triggered the collision between the north and the center buildings.
This collision eventually led to the collapse of the center building, followed
by the destruction of the north building.
Some other recent applications of the numerical code are also demonstrated in the presentation.