Вы находитесь на странице: 1из 22

International Journal of Structural Stability and Dynamics

Vol. 18, No. 3 (2018) 1850038 (22 pages)


#.c World Scienti¯c Publishing Company
DOI: 10.1142/S0219455418500384

An Approximate Analytical Formulation for the Rise-Time


E®ect on Dynamic Structural Response Under Column Loss

Meng-Hao Tsai
Department of Civil Engineering
National Pingtung University of Science and Technology
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com

No. 1 Hseuh-Fu Rd., Neipu, Pingtung County, 912 Taiwan


by Prof. Meng-Hao Tsai on 10/04/17. For personal use only.

mhtsai@mail.npust.edu.tw

Received 5 January 2017


Accepted 24 June 2017
Published 20 July 2017

Dynamic progressive collapse analysis of building structures is usually conducted under sudden
column loss conditions. The time length required for disabling the failed column is de¯ned as the
rise time. The rise-time e®ect on the maximum dynamic response of building frames under
column loss is investigated in this study. Based on the work-energy principle, an approximate
analytical formulation for the maximum dynamic response is derived considering the rise-time
e®ect. The force- and displacement-based dynamic increase factors (DIFs) of a single degree-of-
freedom model and a clamped steel beam are used to assess the accuracy and validity of the
proposed formulation. Analysis results indicate that the DIFs decrease with increased rise time.
Also, the rise-time e®ect decreases with increased ductility demand. Practical application of the
analytical formulation to regular building frames subjected to column loss is illustrated. Pri-
mary factors related to the extent of the rise-time e®ect on the column-loss response are iden-
ti¯ed for design consideration.

Keywords: Column loss; rise-time e®ect; work-energy principle; dynamic increase factor.

1. Introduction
Progressive collapse vulnerability of building structures has become an active re-
search topic since the 9/11 terror attack of the World Trade Center in 2001.1 In fact,
the earliest noticed event can be traced back to the partial collapse of the Ronant
Point Apartment Building in England, 1968. The incident highlighted the impor-
tance of structural integrity and robustness against local structural failure. As a
result, special attention has been paid to structural integrity in several design
codes.2–5 The impact of local member failure was also investigated for some other
structural systems.6,7 Di®erent measures to reduce the risk and hazard of progressive
collapse have been proposed. In general, tie force, alternative load path, integrity
provisions, and speci¯c local load resistance are recommended as feasible approaches

1850038-1
M.-H. Tsai

against the progressive collapse.8–11 Since the cause, reoccurrence, and intensity of
abnormal loadings are di±cult to predict precisely, provision of threat-independent
alternative load paths is recognized as an acceptable and popular solution among
those proposed measures. Alternative load paths of a damaged building can be
veri¯ed by assessing its load transfer ability under stipulated column loss scenarios.
Practical analysis procedures have been recommended by the US General Service
Administration (GSA)12 and the Department of Defense (DoD).13,14 In the guide-
lines, the progressive collapse analysis of building structures is performed under
sudden column loss conditions. This means that the bearing capacity of the failed
column disappears instantaneously. The dynamic e®ect is considered with the load
increase factor (LIF) and dynamic increase factor (DIF) in the linear static (LS) and
nonlinear static (NS) analyses, respectively. In the dynamic analysis, a time length,
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com

which is de¯ned as the rise time, not larger than one tenth of the gravitational
by Prof. Meng-Hao Tsai on 10/04/17. For personal use only.

vibration period of the column-removed structure is suggested for disabling the failed
column. However, for earthquake-resistant buildings, most structural members have
a minimum ductility capacity as regulated in seismic design codes. When a structural
column is subjected to devastating abnormal loadings, it may lose its load-carrying
ability within a ¯nite rise time. In other words, the external loading for progressive
collapse analysis should be imposed gradually.
Some studies have been carried out for investigating the rise-time e®ect on the
structure response under column loss. Liu et al.15 indicated that the time length for
completely removing the column can a®ect the structural response and the rate
of removal is insigni¯cant in the ¯nal results as the rise time is less than 10 ms.
Gudmundsson and Izzuddin16 investigated whether the assumption of sudden col-
umn loss is realistic and whether the scenarios can re°ect an actual extreme event.
They concluded that the sudden column loss assumption may result in an upper
bound of the deformation under blast events. Liu et al.17 conducted sudden load-
release tests of steel beams with web cleat connections. The test results revealed that
the load-release time was around dozens milliseconds. Their numerical analysis
showed that the dynamic column-loss response decreases with increased load-release
time. Similar numerical results were revealed by Tavakoli and Kiakojouri.18 Also, a
recent experimental study19 showed that the vibration period under column loss may
increase with the imposed loading and thus the ratio of load-release time to period
decreases. Rahai et al.20 compared the instantaneous and gradual column-removal
responses of an RC structure and revealed that the former induced higher stress and
deformation demand on the structure. Yu and Guo21 used a nonlinear single degree-
of-freedom (SDOF) model to estimate the structural dynamic response under pro-
gressive collapse. They pointed out that the maximum dynamic de°ection is not
sensitive to the rise time if it is less than one tenth of the vibration period under
column loss. The importance of the rise-time e®ect on the dynamic response under
column loss is recognized from these studies. Most of them have adopted dynamic
time-history analysis method to evaluate the rise-time e®ect.

1850038-2
Rise-Time E®ect Under Column Loss

The work-energy principle has been applied to estimating the collapse resistance of
building frames under sudden column loss.22–24 This energy-based concept was also
used to develop analytical expressions for the DIFs and a performance-based retro¯t
design approach for regular building frames.25,26 Menchel et al.27 proposed a meth-
odology to identify the static pushover procedure for sudden column removal by using
a similar energy criterion. Szyniszewski and Krauthammer28 conducted energy-based
assessment of progressive collapse performance for steel framed building and sug-
gested that deformation energy limit is a good stability indicator. Fang et al.29 used
the energy-based concept to evaluate the robustness of steel-composite car parks
under vehicle ¯re. These studies have con¯rmed that instead of nonlinear time history
analysis, the maximum response under sudden column loss can be estimated with the
energy-based method. Therefore, this study intends to apply the work-energy prin-
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com

ciple to investigating the rise-time e®ect on dynamic structural response under column
by Prof. Meng-Hao Tsai on 10/04/17. For personal use only.

loss. At ¯rst, an approximate analytical formulation considering the rise-time e®ect is


proposed with an idealized SDOF model. Accuracy of the analytical formulation is
veri¯ed with the force- and displacement-based DIFs of the SDOF model and a
clamped steel beam under support-loss conditions. Practical application of the ap-
proach to estimating the maximum dynamic response of regular building frames under
column loss is illustrated. Primary factors related to the extent of the rise-time e®ect
on the dynamic column-loss response are discussed for design consideration.

2. Determination of Linear Load Growth


A conventional approach to investigate the dynamic structural response under
sudden column loss starts with replacing the failed column with equivalent internal
forces acting on the joint. Then, nonlinear dynamic analysis is carried out by ap-
plying a set of step time-history functions with equal-but-opposite forces to the joint
to counteract the equivalent internal forces. Based on the concept of an idealized
SDOF model, the equation of motion may be written as30
Pu ::
u þ ku ¼ Pu ; ð2:1Þ
g
where Pu is the tributary gravitational loadings on the failed column and u is the
vertical de°ection at the column-removed joint. k designates the vertical sti®ness
provided by the remaining structural members. In another aspect, if the column is
designed with su±cient ductility, it may lose its bearing capacity in a ¯nite length of
time. The load supported by the failed column is redistributed gradually, instead of
suddenly, to the neighboring structural members. Therefore, the equation of motion
of the idealized SDOF model in the load growing phase may be approximately
described by
 n
Pu :: t
u þ ku ¼ Pu ¼ P ðtÞ for 0  t  tr ; ð2:2Þ
g tr

1850038-3
M.-H. Tsai

0.9 n = 0.1

0.8 n = 0.2

0.7
n = 0.5
0.6
P(t)/Pu

0.5 n = 1.0

0.4 n = 2.0

0.3
n = 4.0
0.2
n = 8.0
0.1

0
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com

0 0.2 0.4 0.6 0.8 1


by Prof. Meng-Hao Tsai on 10/04/17. For personal use only.

t/tr

Fig. 1. The variation of load growth with di®erent exponents.

where tr is the rise time for the load growth and n is a positive exponent describing
the rate of the column loss. Figure 1 shows the variations of the load growing phase
with di®erent exponents. Apparently, the exponential load growth can be approxi-
mated to the sudden loss case either with a su±ciently large or small exponent.
Linear load growth with n ¼ 1:0 can be regarded as an average representative and is
suitable for analytical investigation of the rise-time e®ect. Hence, linear load growth
is adopted in this study.

3. Analytical Formulation
Based on the linear load-growth assumption, the equation of motion of the SDOF
model during the load growing phase is expressed as
Pu :: t
u þ ku ¼ Pu ; t  tr : ð3:1Þ
g tr
The steady-state displacement time history in this phase can be obtained as31
 
Pu t sin !n t
uðtÞ ¼  ; t  tr ; ð3:2Þ
k tr ! n tr
pffiffiffiffiffiffiffiffiffiffiffiffiffi
where !n ¼ kg=Pu is the natural frequency of the SDOF model under the sustained
loading. If the SDOF model behaves elasto-plastically under instantaneously im-
posed loading, the stored strain energy, denoted by Es , can be described by the area
OCDF in Fig. 2 and written as
1
Es ¼ P  ½ð  1Þ 2 þ 2  1; ð3:3Þ
2 y y
where y is the yield de°ection and Py is the corresponding static yield force. Also, 
is the ductility demand de¯ned as the maximum de°ection u divided by y .  is the

1850038-4
Rise-Time E®ect Under Column Loss

Elasto-plastic
P response
Suddenly D
imposed C Strain energy:
loading area OCDEF

A Work done by
B
Pu instantaneous loading:
E area OABEF

Gradually Work done by


imposed gradual loading:
loading area OBEF

O ur ∆y F ∆
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com

Fig. 2. A schematic for the stored strain energy and external work.
by Prof. Meng-Hao Tsai on 10/04/17. For personal use only.

post-yield sti®ness ratio. Meanwhile, the external work done by the instantaneous
loading is equal to the area OAEF in the ¯gure. However, as the loading is gradually
increased, the external work will present nonlinear load-de°ection relation, as shown
by the area OBEF in Fig. 2.
If the SDOF model is not loaded into the inelastic range during the load growing
phase, which actually holds under most conditions, the work done in that phase can
be calculated as
Z
t
W ðtÞ ¼ Pu du: ð3:4Þ
tr
After some operations, it leads to
 
P u2 t 2 t sin !n t cos !n t  1
W ðtÞ ¼   for t  tr : ð3:5Þ
k 2t 2r !n t 2r ! 2n t 2r
The above equation can be used to estimate the work done at t ¼ tr . Let ur represent
the displacement at t ¼ tr . Hence, the external work under the column loss is then
expressed as the sum of the works done in the load growing and the constant loading
phases as
 
P2 sin !n tr
Wu ¼ Pu ðu  ur Þ þ W ðtr Þ ¼ Pu u  u 1  þ W ðtr Þ: ð3:6Þ
k !n tr
It can be re-written as
 
P u2 1 cos !n tr  1 P u2
Wu ¼ Pu  u  þ  Pu u  gð!n tr Þ: ð3:7Þ
k 2 ! 2n t 2r k
where gð!n tr Þ ¼ ð 12 þ cos!!2ntt2r 1 Þ is the rise-time e®ect function.
n r
From the work-energy principle, the external work Wu is equal to the stored strain
energy Es of the elasto-plastic SDOF model. Hence, with Eq. (3.3) ¼ Eq. (3.7), the
applied loading Pu is obtained and de¯ned as the pseudo-static loading, which is

1850038-5
M.-H. Tsai

approximated to the dynamic loading capacity of the SDOF model at the maximum
de°ection u . Now, de¯ning the normalized pseudo-static loading as R ¼ Pu /Py , a
quadratic equation of R and  can be obtained. In the form of R, it is written as
2gR 2  2R þ ½ð  1Þ 2 þ 2  1 ¼ 0 with  > 1: ð3:8Þ
Alternatively, in the form of , it is written as
 2 þ 2ð1    RÞ þ 2gR 2 þ   1 ¼ 0 with  > 1: ð3:9Þ
The smaller of the two R values from Eq. (3.8) is used to determine the dynamic
loading capacity Pu as a ductility limit  is prescribed, while the positive  from
Eq. (3.9) is the ductility demand under a given sustained loading Pu . In practice, the
ductility limit is determined from the allowable plastic rotations as recommended in
the GSA guidelines12 or ASCE standard.32 Also, the analytical formulation can be
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com
by Prof. Meng-Hao Tsai on 10/04/17. For personal use only.

easily applied to multiple degree-of-freedom structures if the dynamic column-loss


response is dominated by a single vibration mode, which is usually the case under
mid-column loss conditions. However, since elastic response is assumed for the load
growing phase in the analytical formulation, the value of ur in Eq. (3.6) should be
less than the yield de°ection. The validity of this assumption will be assessed in the
later numerical veri¯cation.

4. Dynamic Increase Factors with the Rise-Time E®ect


Since the dynamic structural response under column loss is usually dominated by a
single deformation mode for regular building frames, it has been shown that the
idealized SDOF model can approximately capture the relationship between the
sustained loading and the displacement of the column-removed joint.21,25,33,34
Therefore, the analytical formulation in the previous section is used to calculate the
DIFs with rise-time e®ect for progressive collapse analysis of building frames.

4.1. Force-based DIF


The forced-based DIF is calculated as the ratio of static to dynamic loading under the
same ductility or de°ection demand. If the given ductility demand  is not greater
than 1.0, the normalized static loading Rs can be expressed as
Rs ¼ ky =Py ¼  with   1:0 ð4:1Þ
and the elastic DIF with the rise-time e®ect31 is calculated as
jsinðtr =T Þj
DIF ¼ 1 þ ; ð4:2Þ
tr =T
where T is the natural period under column loss. On the contrary, as the ductility
demand is larger than 1.0, the normalized static loading can be determined as
Rs ¼ 1 þ ð  1Þ with  > 1:0: ð4:3Þ

1850038-6
Rise-Time E®ect Under Column Loss

Together with the normalized pseudo-static loading obtained from Eq. (3.8) and the
¯nite rise time tr , the inelastic forced-based DIF is calculated as Rs =R.

4.2. Displacement-based DIF


The displacement-based DIF on the other hand is equal to the ratio of dynamic to
static de°ection under the same external loading Pu . If the normalized external
loading Pu =Py is less than 0.5, the structural response will be elastic and the dis-
placement-based DIF is the same as Eq. (4.2). However, as Pu =Py varies between 0.5
and 1.0, the structural response will be elastic if the loading is imposed statically and
inelastic if imposed dynamically. Under statically imposed loading, the ductility
demand s is written as
Pu =k Pu
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com

s ¼ ¼ with 0:5 < Pu =Py  1:0: ð4:4Þ


by Prof. Meng-Hao Tsai on 10/04/17. For personal use only.

y Py

Under dynamically imposed loading, the ductility demand is obtained from


Eq. (3.9) with R ¼ Pu =Py and then the displacement-based DIF is calculated as
=s . Moreover, if Pu =Py > 1:0, the static structural response is inelastic and the
corresponding ductility demand is obtained from
Pu =Py  1
s ¼ þ1 with 1:0 < Pu =Py : ð4:5Þ


Given imposed loading


Pu and R = Pu/Py

Calculation of vibration
period T and tr / T

Calculation of elastic DIF


from Eq.(4.2)

µ≤1.0
µ = R*DIF Elastic DIF per Eq.(4.2)

µ>1.0

Calculation of dynamic Normalized static loading


ductility µ per Eq.(3.9) Rs per Eq.(4.3)

Calculation of static ductility Normalized pseudo-static


µs per Eq.(4.4) or Eq.(4.5) loading R per Eq.(3.8)

Inelastic displacement- Inelastic force-based


based DIF = µ / µs DIF = Rs/R

Fig. 3. A °owchart for implementation of the analytical procedure.

1850038-7
M.-H. Tsai

Similarly, the inelastic displacement-based DIF is calculated as =s with  obtained


from Eq. (3.9).
A °owchart for the implementation of the above analytical procedure is shown in
Fig. 3. Considering a loading Pu imposed on the column-loss building frame, the ¯rst
step is the determination of the vibration period T and ratio of tr =T . The imposed
loading is included in the calculation of the vibration period. Then, a corresponding
elastic DIF is calculated by using Eq. (4.2). This elastic DIF is used to estimate a
dynamic ductility demand, . If  is not greater than 1.0, then the dynamic response
is elastic and the DIF is determined from Eq. (4.2). However, if  is larger than 1.0,
then the normalized external loading Pu =Py is substituted into Eq. (3.9) for the real
dynamic ductility demand . The inelastic displacement-based DIF is calculated as
the dynamic ductility demand divided by its static counterpart obtained from either
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com

Eq. (4.4) or Eq. (4.5). Meanwhile, the dynamic ductility demand obtained from
by Prof. Meng-Hao Tsai on 10/04/17. For personal use only.

Eq. (3.9) is used to determine the normalized static and pseudo-static loading by
using Eqs. (4.3) and (3.8), respectively. The ratio of the normalized static to pseudo-
static loading is then equal to the inelastic force-based DIF.

5. Numerical Veri¯cation
5.1. SDOF model
The idealized SDOF model of an RC beam-column sub-assemblage specimen35 is
used to verify the accuracy of the proposed analytical formulation. The SDOF model
consists of a load-equivalent mass and an elasto-plastic nonlinear spring with yield
force Py ¼ 51:5 kN, yield de°ection y ¼ 0:7 cm, and a post-yield sti®ness ratio
 ¼ 0:05, as shown in Fig. 4. The imposed loading magnitude is varied in terms of a
multiplier of the yield force Py . Nonlinear time-history analyses were carried out for
the SDOF model subjected to a constant loading function with ¯nite rise time tr . The
loading magnitude is changed from 0.2Py to 3.0Py and three di®erent rise times,
tr ¼ 0:06; 0:10, and 0.15 s, are considered in the analyses. Because the imposed
loading is considered as the e®ective mass in the analyses, the natural period of the

Load P

P Pu

W tr Time t
Force
k p = αk
k Py
k
∆y Displacement

Fig. 4. The SDOF model for numerical veri¯cation.

1850038-8
Rise-Time E®ect Under Column Loss

model can increase with the loading magnitude. The smallest rise time has all the
tr =T ratios smaller than 1.0. However, the other two result in some tr =T ratios larger
than 1.0 in the analyses.
Figures 5(a) and 5(b) show the variations of the force- and displacement-based
DIFs, respectively. In the ¯gures, the symbols represent the DIFs obtained from the
nonlinear time-history analyses, while the curves represent those obtained from the
proposed analytical expressions. Also, the analytical DIFs obtained without the rise-
time e®ect25 are included in the ¯gures for comparison. In Fig. 5(a), it is seen that the
force-based DIFs decrease with increased tr . This indicates that the applied loading
may be overestimated in the static progressive collapse analysis without considering
the rise-time e®ect. Nevertheless, the di®erence induced by the rise-time e®ect is less
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com

2.5
by Prof. Meng-Hao Tsai on 10/04/17. For personal use only.

SDOF model tr = 0
tr = 0.06 sec
2 tr = 0.1 sec
tr = 0.15 sec
tr = 0.06 sec (Predict)
tr = 0.1 sec (Predict)
Force DIF

1.5
tr = 0.15 sec (Predict)

0.5

0
0 1 2 3 4 5 6
∆u/∆y

(a)

SDOF model tr = 0
5 tr = 0.06 sec
tr = 0.1 sec
Displacement DIF

4 tr = 0.15 sec
tr = 0.06 sec (Predict)

3 tr = 0.1 sec (Predict)


tr = 0.15 sec (Predict)

0
0 0.5 1 1.5 2 2.5 3
Pu/Py

(b)

Fig. 5. (a) Force-based DIFs of the SDOF model (b) Displacement-based DIFs of the SDOF model.

1850038-9
M.-H. Tsai

signi¯cant under larger ductility demands, which implies a reduced in°uence on the
plastic column-loss response. Meanwhile, the dynamic analysis results match well
with the predicted DIFs. Similar to the force-based DIFs, the displacement-based
DIFs decrease with increased tr , as shown in Fig. 5(b). This means a conservative
ductility demand may be obtained without considering the rise time. However, it is
seen that the rise time e®ect on the displacement-based DIFs can be practically
neglected as the normalized dynamic loading Pu =Py is larger than 1.5.
In the DoD and GSA progressive collapse analysis guidelines, it is recommended
that the tr =T ratio should be less than 0.1 for sudden column loss scenarios. As shown
in Fig. 6(a), the values of tr =T decrease with increased normalized loading because of
the increased vibration period T . In the ¯gure, most tr =T are larger than the
threshold of 0.1. Figures 6(a) and 5(b) reveal that the threshold of 0.1 can be
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com
by Prof. Meng-Hao Tsai on 10/04/17. For personal use only.

2.5

SDOF model tr = 0.06 sec

2 tr = 0.1 sec

tr = 0.15 sec

1.5
tr/T

0.5

0
0 0.5 1 1.5 2 2.5 3
Pu/Py

(a)

tr = 0.06 sec tr = 0.06 sec (Predict) SDOF model


3.5
tr = 0.1 sec tr = 0.1 sec (Predict)
3 tr = 0.15 sec tr = 0.15 sec (Predict)

2.5
u(tr) / ∆ y

1.5

0.5

0
0 0.5 1 1.5 2 2.5 3
Pu/Py

(b)

Fig. 6. (a) Variation of tr =T ratios of the SDOF model (b) Variation of the normalized displacement u(tr Þ/
y at the end of rise time with normalized loading for the SDOF model.

1850038-10
Rise-Time E®ect Under Column Loss

appropriately increased as signi¯cant plastic deformation is expected under column


loss. In addition, it is reminded that elastic behavior is assumed for the load growing
phase in the analytical derivations. Figure 6(b) shows the variation of the normal-
ized displacement, ur =y , at the time instant t ¼ tr . It is seen that ur =y is less than
1.0 as tr ¼ 0:06 s, while it may be larger than 1.0 as tr ¼ 0:10 s and 0.15 s. However,
the analytical approach still provides accurate prediction even when ur =y is larger
than 1.0. This reveals that the assumption of ur =y  1 is not critical for the pre-
diction of the DIFs.

5.2. Clamped steel beam model


In general, there are always adjacent structural members or boundaries to share the
supported loading of the two-span beam bridging over the failed column in a building
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com
by Prof. Meng-Hao Tsai on 10/04/17. For personal use only.

frame. The previous SDOF model cannot consider the action of load redistribution.
For a more realistic veri¯cation, the proposed analytical formulation is further ap-
plied to a clamped steel beam model. The steel beam has an ASTM W24  68 section
made of A572 Grade 50 material. A combined mid-span and uniformly distributed
loading, Q þ qL, is imposed on the clamped beam, as shown in Fig. 7. The mid-span
concentrated loading, Q, is equal to half the distributed loading, qL. The yield
reaction Py for a three-hinge mechanism of the clamped beam model25 is 1440 kN at
an equivalent yield displacement of 1.80 cm. The corresponding distributed and
concentrated loads are equal to 120 kN/m and 480 kN, respectively. Lumped °exural
hinges are assigned to the mid-span and beam ends. An elasto-plastic nonlinear
model with a post-yield sti®ness ratio of 0.05 is assumed for the hinge properties as
shown in Fig. 7. Nonlinear time-history analyses are conducted by applying an equal-
but-opposite loading to counteract the original supported loading Q þ qL=2 at the
mid-span. The counteractive loading is linearly increased within the rise time tr as

Q
q

W24x68 (A572 Grade 50)

L/2 L/2
P(t)
Moment (M )
0 tr Time t (sec)
My 5k
k p = 0.0
k
-(Q + qL/2)
θy Rotation (θ)
Hinge property Dynamic load P(t)

Fig. 7. The clamped steel beam model, °exural hinge property, and loading time history.

1850038-11
M.-H. Tsai

2.5

Clamped beam model tr = 0


tr = 0.06 sec
2 tr = 0.15 sec
tr = 0.30 sec
tr = 0.06 sec (Predict)
tr = 0.15 sec (Predict)
Force DIF

1.5
tr = 0.30 sec (Predict)

0.5

0
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com

0 1 2 3 4 5 6
∆u/∆y
by Prof. Meng-Hao Tsai on 10/04/17. For personal use only.

(a)

Clamped beam model tr = 0


5 tr = 0.06 sec
tr = 0.15 sec
tr = 0.30 sec
Displacement DIF

4
tr = 0.06 sec (Predict)
tr = 0.15 sec (Predict)
3
tr = 0.30 sec (Predict)

0
0 0.5 1 1.5 2 2.5 3
Pu/Py

(b)

Fig. 8. (a) Force-based DIFs of the clamped steel beam (b) Displacement-based DIFs of the clamped steel
beam.

shown in Fig. 7. Three di®erent ¯nite rise times, i.e. 0.06, 0.15, and 0.30 s, and
imposed loadings ranging from 0.2 Py to 3.0 Py are considered in the time-history
analyses.
Figures 8(a) and 8(b) show the comparison of the numerical and predicted DIFs
of the clamped beam model. Although not as perfect as the SDOF case, it is seen that
the proposed approach can give accurate predictions for both the force- and dis-
placement-based DIFs. In general, the analytical approach leads to slightly larger
DIFs than the numerical analysis results. Since the analytical formulation is derived
from an SDOF model subjected to a direct loading, discrepancy may increase under
the load-release condition of the clamped beam.36 Nevertheless, the rise-time e®ect

1850038-12
Rise-Time E®ect Under Column Loss

Clamped beam model


2.5
tr = 0.06 sec

2 tr = 0.15 sec

tr = 0.30 sec
tr/T

1.5

0.5

0
0 0.5 1 1.5 2 2.5 3
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com

Pu/Py
by Prof. Meng-Hao Tsai on 10/04/17. For personal use only.

(a)

6
tr = 0.06 sec tr = 0.06 sec (Predict) Clamped beam model

5 tr = 0.15 sec tr = 0.15 sec (Predict)


tr = 0.30 sec tr = 0.30 sec (Predict)

4
u(tr)/∆ y

0
0 0.5 1 1.5 2 2.5 3
Pu/Py

(b)

Fig. 9. (a) Variation of tr =T ratios of the clamped steel beam (b) Variation of the normalized displacement
uðtr Þ=y at the end of rise time with normalized loading for the clamped steel beam.

on the clamped beam is similar to that on the SDOF model. Insigni¯cant e®ects on
the force- and displacement-based DIFs can be observed as the ductility is larger
than 5.0 and the normalized loading is larger than 1.5, respectively. A larger imposed
loading is concurrent with an increased ductility demand and structural period
under column loss. Thus, the tr =T ratio decreases with increased loading as shown in
Fig. 9(a), and it leads to less rise-time e®ect on the plastic response. Most of the tr =T
ratios are larger than 1.0 for the case with tr ¼ 0:3 s. As tr =T is close to or larger 1.0,
the DIF becomes close to or only slightly larger than 1.0 as observed in Figs. 8(a)
and 8(b). This is due to the quasi-static response with a larger tr =T ratio. As tr =T
being larger than 1.0, the dynamic e®ect decreases and the column-loss response is

1850038-13
M.-H. Tsai

dominated by the quasi-static behavior. Hence, the DIFs converge asymptotically


toward 1.0 for the elastic and moderately plastic responses. Also, although the as-
sumption of elastic behavior during the load growing phase is violated with a larger
Pu =Py or tr as shown in Fig. 9(b), it does not signi¯cantly reduce the accuracy of the
analytical predictions. In other words, although both Figs. 6(b) and 9(b) indicate
that Eq. (3.2) may underestimate the displacement response at the end of the load
growing phase as the rise time increases, the underestimation of uðtr Þ brings only
minor in°uence on the prediction of the maximum dynamic response. Hence, the
analytical procedure should be applicable for practical range of tr =T .

6. Practical Application
From the aforementioned veri¯cation, the accuracy of the proposed analytical for-
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com
by Prof. Meng-Hao Tsai on 10/04/17. For personal use only.

mulation considering the rise-time e®ect is con¯rmed. In the current design guide-
lines issued by the GSA12 and DoD,14 the force-based DIF is used to amplify the
imposed loading combination in the NS progressive collapse analysis. The ratio of
allowable plastic rotation to yield rotation in the guidelines is equivalent to the
ductility demand, , in the analytical formulation. The deformation demands under
the ampli¯ed loadings are then compared with the deformation capacities of struc-
tural members. Some studies have indicated that pseudo-static analysis can be used
to estimate the maximum dynamic response under column loss.17,22,23,37,38 The
pseudo-static loading is calculated from the NS pushdown curve as
Z ud
PCC ðud Þ ¼ PNS ðuÞdu=ud ; ð6:1Þ
0

where PNS ðuÞ is the statically imposed loading obtained from the nonlinear pushdown
curve. PCC ðud Þ is the pseudo-static loading, which represents the dynamic loading
resistance at the de°ection demand ud . Therefore, to avoid time-consuming nonlinear
dynamic analysis, the rise-time e®ect on the dynamic column-loss response can be
accounted for if the proposed approach is applied to the pseudo-static analysis.
A steel moment-resisting frame model adopted from an earlier study26 is used to
demonstrate the application. Elevation views, material strength, and section details
of the model are shown in Fig. 10. An assumed penultimate column-loss scenario is
indicated in the ¯gures. Flexural plastic hinges with nonlinear properties displayed in
Fig. 11 are assigned to the ends of the structural members. In the ¯gure, y and c are
the yield rotation and the rotation at the ultimate strength, respectively. Two hinge
rotation capacities, c =y ¼ 4 and c =y ¼ 6, are considered to examine the adequacy
of the proposed approach under moderate and high ductility demands. The ultimate-
strength factor Mu =My is held constant as 1.12. It is assumed that the column-loss
response is governed by the °exural behavior of the structural members in a single
deformation mode. Three di®erent lengths of rise time, tr ¼ 0.06, 0.2 and 0.4 s, are
determined to cover a wide range of tr =T in the numerical illustration. Also, catenary
action is not considered in the structural model.

1850038-14
Rise-Time E®ect Under Column Loss
10 @4 m

Frame sections
Interior Exterior
Floor Beam
column column
fy (kN/m2) 248000 344700 344700
8F~10F W18x40 W24x94 W14x74
5F~7F W21x50 W24x117 W14x90
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com
by Prof. Meng-Hao Tsai on 10/04/17. For personal use only.

failed 1F~4F W24x68 W24x131 W14x132


8m 8m 8m

Fig. 10. Elevation views, material strength, and section details of the moment-resisting steel frame.

M
Mu M u/M y = 1.12
My

θc /θy = 4 and 6
Mr

θy θc θ

Fig. 11. Flexural hinge properties in the numerical analysis.

6.1. Bilinear approximation and pseudo-static response


As revealed in the calculation of the forced- and displacement-based DIFs, the elastic
and inelastic rise-time e®ects are estimated by two distinct expressions. This means
that the yield loading of the moment frame under column loss should be determined a
priori for the analytical prediction. Hence, uniformly distributed downward loading
is imposed to the structural bays and NS pushdown analysis is carried out for the
column-removed frame. From the NS pushdown curve, its static ultimate loading
capacity and the associated displacement are obtained and denoted as Pul and ul ,
respectively. Then, an equivalent bilinear curve between the origin and the ultimate
response point (ul , Pul Þ is constructed based on the equal energy principle, in which
the stored strain energy is expressed as
1 1
Es ¼  ðK   Pul Þ þ Pul ul : ð6:2Þ
2 y e ul 2

1850038-15
M.-H. Tsai

On the left-hand side of the above equation, Es is the stored strain energy up to the
ultimate response point (ul , Pul Þ of the nonlinear pushdown curve. The right-hand
side is the corresponding strain energy of the bilinear approximation. Ke is the elastic
sti®ness of the bilinear curve and can be determined from the ¯rst two steps of the
nonlinear pushdown curve. y represents the equivalent yield displacement and is
determined from
2Es  Pul ul
y ¼ : ð6:3Þ
ðKe ul  Pul Þ
The yield strength Py and post-yield sti®ness Kp are then calculated from Py ¼ Ke y
and Kp ¼ ðPul  Py Þ=ðul  y Þ. Ductility demand at the ultimate loading is equal
to c ¼ ul =y .
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com

Figures 12(a) and 12(b) shows the nonlinear pushdown curve and the equivalent
by Prof. Meng-Hao Tsai on 10/04/17. For personal use only.

bilinear approximation up to the ultimate static resistance of the building frame with

1.5
1.4
1.3
1.2
1.1
1
0.9
0.8
P/Py

0.7
0.6
0.5
0.4 Pushdown Py = 6996 kN
0.3 ∆ y = 6.57 cm
Bilinear
0.2 α = Kp/Ke = 0.0788
0.1 Pseudo-static θc/θy = 4
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
Ductility

(a)

1.5
1.4 Pushdown
1.3
Bilinear
1.2
Pseudo-static
1.1
1
0.9
0.8
P/Py

0.7
0.6
0.5
0.4 Py = 10965 kN
∆ y = 10.3 cm
0.3
α = Kp/Ke = 0.0764
0.2
θc/θy = 6
0.1
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
Ductility

(b)

Fig. 12. (a) Nonlinear pushdown curve and bilinear approximation of the steel frame with moderate hinge
rotation capacity (b) Nonlinear pushdown curve and bilinear approximation of the steel frame with high
hinge rotation capacity.

1850038-16
Rise-Time E®ect Under Column Loss

moderate and high hinge rotation capacities, respectively. The ordinate is expressed
as a multiplier of the yield strength Py and the abscissa as the ductility obtained from
dividing the displacement of the column-removed joint by the equivalent yield dis-
placement. The magnitudes of the yield strength Py , the equivalent yield displace-
ment y , and the post-yield sti®ness ratio Kp /Ke are shown in the ¯gures. It is seen
that the building frame model with higher hinge rotation capacity has a larger
equivalent yield displacement under column loss. Their pseudo-static response curves
under sudden column loss are also shown in the ¯gures. The pseudo-static response
curves show a maximum collapse resistance PCC ¼ 0.94Py at the ductility demand of
3.93 with moderate rotation capacity and PCC ¼ 1.03Py at the ductility demand of
5.66 with high rotation capacity.
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com

6.2. Dynamic ductility demands


by Prof. Meng-Hao Tsai on 10/04/17. For personal use only.

Several dynamic loadings, Pu , increasing from 0.5Py to the maximum collapse re-
sistance, PCC , are selected to predict their dynamic ductility demand under column
loss. With the three di®erent rise-time lengths, the ratio of tr =T varies from 0.12 to
2.0, as shown in Fig. 13. It is believed that this range can cover most practical
conditions. The vibration period, T , in the analytical prediction is determined from
the modal period corresponding to the column-loss vibration mode. The normalized
rise time, tr =T , is used to calculate the rise-time e®ect function in Eq. (3.7) and the
maximum dynamic ductility demand is obtained from Eq. (3.9) with the given
loading magnitude Pu . The analytical ductility demands under the dynamic loadings
are compared with that obtained from nonlinear dynamic analysis, as shown in
Figs. 14(a) and 14(b). It is seen that the analytical predictions and numerical results
are in good agreement for both the moderate and high rotational capacity cases. The
analytical predictions tend to be slightly larger than the numerical results under
higher ductility demands, which can lead to a safer collapse resistant design. The
root-mean-square error of each rise-time length is also calculated and shown in the
¯gures. In average, the relative error is around 8% with the moderate (c =y ¼ 4) and

3
Py = 6996 kN
tr = 0.06 sec
∆ y = 6.57 cm
2.5
α = Kp/Ke = 0.0788 tr = 0.20 sec
θc/θy = 4
tr = 0.40 sec
2
tr/T

1.5

0.5

0
0 0.2 0.4 0.6 0.8 1 1.2
Pu/Py

Fig. 13. Variation of tr =T ratios of the steel frame.

1850038-17
M.-H. Tsai

4.5 Py = 6996 kN
∆ y = 6.57 cm
4 α = Kp/Ke = 0.0788
θc/θy = 4
3.5
Predicted ductility
3

2.5

1.5 tr = 0.06 sec (err = 9.6%)

1 tr = 0.20 sec (err = 7.5%)

0.5 tr = 0.40 sec (err = 7.8%)

0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Numerical ductility

(a)
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com
by Prof. Meng-Hao Tsai on 10/04/17. For personal use only.

4.5 Py = 10965 kN
∆ y = 10.3 cm
4 α = Kp/Ke = 0.0764
3.5 θc/θy = 6
Predicted ductility

2.5

1.5 tr = 0.06 sec (err = 5.5%)

1 tr = 0.20 sec (err = 4.3%)

0.5 tr = 0.40 sec (err = 4.5%)

0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Numerical ductility

(b)

Fig. 14. (a) Comparison of the predicted and numerical dynamic ductility with moderate hinge rotation
capacity (b) Comparison of the predicted and numerical dynamic ductility with high hinge rotation
capacity.

5% with the high (c =y ¼ 6) hinge rotation capacity. The error appears not sensitive
to the rise-time length. Nevertheless, it is observed that the analytical estimations
are more accurate with the higher hinge rotation capacity. This is consistent with the
previous SDOF analysis results that the rise-time e®ect decreases with increased
plastic deformation. These evidences indicate that the analytical approach can
precisely capture the maximum dynamic response of the building frame under col-
umn loss. Therefore, instead of conventional nonlinear time-history analysis, the
pseudo-static analysis approach can be applied to accounting for the rise-time e®ect
on the dynamic response.
Finally, Fig. 15 summarizes the normalized load-de°ection response envelopes
obtained from the nonlinear dynamic analysis results. The ¯gure clearly reveals that
an increased tr =T ratio leads to a larger collapse resistance under the same ductility
demand. On the other hand, for a speci¯c normalized loading, a larger tr =T ratio

1850038-18
Rise-Time E®ect Under Column Loss

1.2
1.1
1
0.9
0.8
Pu/Py 0.7 tr = 0.06 sec
0.6 θc/θy = 6 tr = 0.20 sec
0.5 tr = 0.40 sec
0.4 tr = 0.06 sec
0.3 θc/θy = 4 tr = 0.20 sec
0.2 tr = 0.40 sec

0.1 tr = 0

0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
Ductility

Fig. 15. Normalized dynamic load-de°ection response envelopes.


Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com
by Prof. Meng-Hao Tsai on 10/04/17. For personal use only.

leads to a smaller ductility demand. As compared with the sudden column-loss


response (tr ¼ 0), the rise-time e®ect on the collapse resistance and ductility demand
decreases with the increased hinge rotation capacity. This implies that the rise-time
e®ect is more important for building frames with less ductility capacity under column
loss.

7. Conclusion
Progressive collapse analysis of building structures is usually performed under sud-
den column loss conditions. A time length less than one tenth of the column-loss
vibration period is suggested by some guidelines for disabling the failed column in the
numerical analysis. However, for seismically designed building frames, most struc-
tural members have a minimum ductility capacity such that they can gradually lose
the load-carrying ability under localized extreme loadings. In other words, the ex-
ternal loading for the progressive collapse analysis grows within a ¯nite rise time.
Therefore, the rime-time e®ect on the dynamic structural response under column loss
is investigated in this study.
An analytical formulation with consideration of the rise time has been derived and
its validity has been evaluated with the displacement- and force-based DIFs of an
idealized SDOF model and a clamped steel beam. The numerical analysis results are
consistent with the analytical predictions. Both the force- and displacement-based
DIFs decrease with increased normalized rise time, tr =T . Hence, the loading and
ductility demands may be overestimated in the static progressive collapse analysis if
the rise-time e®ect is neglected. Nevertheless, the rise-time e®ect is apparently re-
duced for signi¯cant plastic response under column loss. Generally speaking, it can be
neglected as the ductility demand is larger than 5.0 or the normalized loading
(Pu =Py Þ is larger than 1.5.
Furthermore, practical application of the proposed analytical formulation has
been demonstrated with a building frame subjected to column loss. Combined with
the pseudo-static response curve, the rise-time e®ect on the maximum dynamic

1850038-19
M.-H. Tsai

response of the building frame under column loss can be appropriately considered.
Compared with the nonlinear dynamic analysis results, the accuracy of the pseudo-
static response curve with the rise-time e®ect has been veri¯ed. In average, the root-
mean-square error is within 8% and 5% for the column-removed building frame with
moderate and high hinge rotation capacity, respectively. An increased tr =T ratio can
lead to a larger collapse resistance under a speci¯c ductility demand or a smaller
ductility demand under a speci¯c normalized loading. However, the rise-time e®ect
on the collapse resistance and ductility demand decreases with increased hinge ro-
tation capacity. Therefore, it is less critical if signi¯cant ductile behavior under
column loss can be assured for the building frames.

References
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com
by Prof. Meng-Hao Tsai on 10/04/17. For personal use only.

1. Z. P. Bazant and Y. Zhou, Why did the World Trade Center collapse?-simple analysis,
Int. J. Struct. Stab. Dyn. 1(4) (2001) 603–615.
2. B. R. Ellingwood, R. Smilowitz, D. O. Dusenberry, D. Duthinh and H. S. Lew, Best
practices for reducing the potential for progressive collapse in buildings, Report No.
NISTIR 7396 National Institute of Standards and Technology, Department of Com-
merce, USA (2007).
3. British Standards Institution (BSI), Actions on structures — Part 1-7: General actions–
Accidental actions, BS EN 1991-1-7, London, UK (2006).
4. American Society of Civil Engineers (ASCE), Minimum design loads for buildings and
other structures, Standard ASCE/SEI 7-10, Reston, Virginia (2010).
5. American Concrete Institute (ACI), Building code requirements for structural concrete,
Standard ACI 318-14, Farmington Hills, Michigan (2014).
6. B. Shekastehb and K. Abedi, Dynamic propagation of snap-through buckling in ten-
segrity structures, Int. J. Struct. Stab. Dyn. 14(1) (2014) 1350049.
7. L. Tian, R. Ma, H. Li and Y. Wang, Progressive collapse of power transmission tower-line
system under extremely strong earthquake excitations, Int. J. Struct. Stab. Dyn. 16(7)
(2016) 1550030.
8. J. Abruzzo, A. Matta and G. Panariello, Study of mitigation strategies for progressive
collapse of a reinforced concrete commercial building, J. Perform. Construct. Facilit.
ASCE 20(4) (2006) 384–390.
9. B. R. Ellingwood, Mitigating risk from abnormal loads and progressive collapse,
J. Perform. Construct. Facilit. ASCE 20(4) (2006) 315–323.
10. O. A. Mohamed, Progressive collapse of structures: Annotated bibliography and
comparison of codes and standards, J. Perform. Construct. Facilit. ASCE 20(4) (2006)
418–425.
11. R. S. Nair, Preventing disproportionate collapse, J. Perform. Construct. Facilit. ASCE
20(4) (2006) 309–314.
12. General service administration (GSA), Alternate Path Analysis & Design Guidelines for
Progressive Collapse Resistance, Washington D.C., USA (2013).
13. Department of Defense (DoD), Uni¯ed Facilities Criteria, Design of Buildings to Resist
Progressive Collapse, UFC 4-023-03, Washington D.C., USA (2009).
14. Department of defence (DoD), Uni¯ed Facilities Criteria, Design of Buildings to Resist
Progressive Collapse, UFC 4-023-03, Washington D.C., USA (2013).
15. R. Liu, B. Davison and A. Tyas, A study of progressive collapse in multi-story steel
frames, in Proceedings of Structures Congress, New York (2005).

1850038-20
Rise-Time E®ect Under Column Loss

16. G. V. Gudmundsson and B. A. Izzuddin, The ‘sudden column loss' idealization for dis-
proportionate collapse assessment, Struct. Eng. 88(6) (2010) 22–26.
17. C. Liu, K. H. Tan and T. C. Fung, Dynamic behaviour of web cleat connections subjected
to sudden column removal scenario, J. Construct. Steel Res. 86 (2013) 92–106.
18. H. R. Tavakoli and F. Kiakojouri, Numerical study of progressive collapse in framed
structures: A new approach for dynamic column removal, Int. J. Eng. Trans. A: Basics
26(7) (2013) 685–692.
19. C. Liu, T. C. Fung and K. H. Tan, Dynamic performance of °ush end-plate beam-column
connections and design applications in progressive collapse, J. Struct. Eng. 142(1) (2016)
04015074-1-14.
20. A. Rahai, M. S. Asghshahr, M. Banazedeh and H. Kazem, Progressive collapse assessment
of RC structures under instantaneous and gradual removal of columns, Adv. Struct. Eng.
16(10) (2013) 1671–1682.
21. J. Yu and Y. Guo, Nonlinear SDOF model for dynamic response of structures under
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com

progressive collapse, J. Eng. Mech. 142(3) (2016) 04015103-1-13.


by Prof. Meng-Hao Tsai on 10/04/17. For personal use only.

22. B. A. Izzuddin, A. G. Vlassis, A. Y. Elghazouli and D. A. Nethercot, Progressive collapse


of multi-storey buildings due to sudden column loss — Part I: Simpli¯ed assessment
framework, Eng. Struct. 30(5) (2008) 1308–1318.
23. M. H. Tsai and B. H. Lin, Investigation of progressive collapse resistance and inelastic
response for an earthquake-resistant RC building subjected to column failure, Eng.
Struct. 30(12) (2008) 3619–3628.
24. G. Xu and B. R. Ellingwood, An energy-based partial pushdown analysis procedure for
assessment of disproportionate collapse potential, J. Construct. Steel Res. 67 (2011) 547–
555.
25. M. H. Tsai, An analytical methodology for the dynamic ampli¯cation factor in progres-
sive collapse evaluation of building structures, Mech. Res. Commun. 37 (2010) 61–66.
26. M. H. Tsai, A performance-based design approach for retro¯tting regular building
frames with steel braces against sudden column loss, J. Construct. Steel Res. 77 (2012)
1–11.
27. K. Menchel, T. J. Massart and P. Bouillard, Identi¯cation of progressive collapse push-
over based on a kinetic energy criterion, Struct. Eng. Mech. 39(3) (2011) 427–447.
28. S. Szyniszewski and T. Krauthammer, Energy °ow in progressive collapse of steel framed
buildings, Eng. Struct. 42 (2012) 142–153.
29. C. Fang, B. A. Izzuddin, A. Y. Elghazouli and D. A. Nethercot, Simpli¯ed energy-based
robustness assessment for steel-composite car parks under vehicle ¯re, Eng. Struct. 49
(2013) 719–732.
30. C. H. Lee, S. Kim, K. H. Han and K. Lee, Simpli¯ed nonlinear progressive collapse
analysis of welded steel moment frames, J. Construct. Steel Res. 65 (2009) 1130–1137.
31. A. K. Chopra, Dynamics of Structures: Theory and Applications to Earthquake Engi-
neering (Prentice-Hall, New Jersey, 1995).
32. American Society of Civil Engineers (ASCE), Seismic Evaluation and Retro¯t of Existing
Buildings, Standard ASCE/SEI 41-13, Reston, Virginia (2013).
33. N. Buscemi and S. Marjanishvili, SDOF Model for Progressive Collapse Analysis, in Proc.
Structures Cong. New York (2005).
34. M. Sasani and S. Sagiroglu, Progressive collapse of reinforced concrete structures: A
multihazard perspective, ACI Struct. J. 105 (2008) 96–103.
35. M. H. Tsai and Y. T. Chang, Collapse-resistant performance of RC beam–column sub-
assemblages with varied section depth and stirrup spacing, Struct. Des. Tall Special
Build. 24(8) (2015) 555–570.

1850038-21
M.-H. Tsai

36. M. H. Tsai, Evaluation of di®erent loading simulation approaches for progressive collapse
analysis of regular building frames, Struct. Infrastruct. Eng. 8(8) (2012) 765–779.
37. M. H. Tsai and B. H. Lin, Dynamic ampli¯cation factor for progressive collapse resistance
analysis of a RC building, Struct. Des. Tall Special Build. 18(5) (2009) 539–557.
38. K. Qian, B. Li and J. X. Ma, Load-carrying mechanism to resist progressive collapse of
RC buildings, J. Struct. Eng. 141(2) (2015) 04014107-1-14.
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com
by Prof. Meng-Hao Tsai on 10/04/17. For personal use only.

1850038-22

Вам также может понравиться