Académique Documents
Professionnel Documents
Culture Documents
DOI 10.1002/tal.1370
1
School of Civil Engineering, Tongji University,
200092, Shanghai, China Summary
2
The State Key Laboratory on Disaster In the present work, the incremental dynamic analysis is performed for the supertall building
Reduction in Civil Engineering, Tongji structure of ZhongNan Center, which is more than 700 m high and is currently under
University, Shanghai, China construction in Suzhou, China. The ChiChi earthquake record is used as the ground motion input.
Correspondence The peak ground accelerations are scaled to five levels including 0.2 g, 0.4 g, 0.8 g, 1.4 g, and
Jie Li, School of Civil Engineering, Tongji
2.0 g. The plastic damage model is adopted to represent the material failure. The finite element
University, 1239 Siping Road, Shanghai,
200092, China. model is developed by using fiber beam elements for beamcolumns and laminated shell elements
Email: lijie@tongji.edu.cn for shear walls and slabs. Under the attack of strong seismic ground motions, the super tall
Xiaodan Ren, School of Civil Engineering, structure experiences overall collapse, which is well captured by the developed finite element
Tongji University, 200092, Shanghai, China
model. More importantly, the collapse patterns of the super tall structure are rather different
Email: rxdtj@tongji.edu.cn
under the excitations of earthquake records with same wave pattern but different peak
Funding information
National Natural Science Foundation of China, accelerations.
Grant/Award Number: 51208374
KEY W ORDS
collapse pattern, damage model, finite element method, incremental dynamic analysis, structural
collapse, supertall building
Struct Design Tall Spec Build. 2017;e1370. wileyonlinelibrary.com/journal/tal Copyright 2017 John Wiley & Sons, Ltd. 1 of 11
https://doi.org/10.1002/tal.1370
2 of 11 HUANG ET AL.
2 | D E S C R I P T I O N OF T H E S U P E R T A L L
B U I LD I N G
P : ; (3)
3.2 | Plasticdamage concrete model
According to the principle of strain equivalence,[12,13] the effective
P : ; (4)
stress tensor can be represented as
D d P d P : (9)
1; x>0
Hx : (7)
0; x0
TABLE 1 Summary of finite elements Thus far, the framework of damageplasticity model could be used
for the numerical simulation of structures.
Number of Nodes 225632
Number of fiber beamcolumn elements 53822
Number of multilayer shell elements 27021 3.3 | Steel constitutive law
Number of SR4 elements 62277 A standard elastoplastic strainhardening model has been used to
simulate the nonlinear behavior of steel tubes and steel reinforcing
The evolutionary law of damage scalars[16] is expressed as
bars under cyclic loading. The fundamental stressstrain relation is
8
>
< d f
d eq E0 : p : (17)
; (10)
>
: d f d eq
The evolutionary law of plastic strain is
f d ; (13) and nodes should be reproduced properly. The system with those
>
>
>
: 1 x 1 strong nonlinearities could hardly be solved by implicit solver, which
2
d x 1 x
is developed based on the NewtonRaphson method due to illness
of system Jacobian. Thus, in the present work, Abaqus/Explicit is
f E0 adopted, which has been used and verified for a wide variety of simu-
x ; ; n ; (14)
E0 E0 f lations including crash, blast, material failure, and impact applications.
The explicit dynamic analysis procedure in Abaqus/Explicit is
where f denotes the uniaxial tensile strength and the uniaxial compres-
based upon the implementation of the central difference method.[21]
sive strength, as appropriate; is the strain corresponding to the peak
value of the stress; and d is the parameter governs the slope of the
descending part of stressstrain curve.
According to literature,[12] the plastic evolutions could be devel-
oped based on the concept of operator splitting. In the present work,
we turn to the empirical plastic evolutionary model.[20] By analogy of
decomposition of damage tensor, we introduce the decomposition of
plastic strain as follows:
p p p ; (15)
where p and p are plastic strains under tension and compression,
respectively. Considering the coupling between damage and plasticity,
the evolutions of plastic strains are expressed as follows:
n
_ p H d_ E0 : _ H d_ p d np E1
np 1 _ p _
p d 0 : ; (16)
where
p, p , n
p and np are material parameters related to plasticity. FIGURE 9 Uniaxial stressstrain relation of steel
6 of 11 HUANG ET AL.
Consider the nth time increment as follows: parallelized so that the collapse of largescale structures could be
simulated by taking advantage of largescale computers or servers.
1 n1
tn1=2 tn1 tn ; tn1=2 t tn ; tn On the other hand, the central difference method is conditionally
2
t n1=2
t n1=2
: (21) stable. The stability limit for the dynamic system without damping is
given in terms of the highest eigenvalue max as follows:
Denote the vectors of displacement, velocity, and acceleration by
2
u,v, and a. The central difference formula for the velocity could be tstable : (31)
max
expressed as follows:
In Abaqus/Explicit, a small amount of damping is usually
un1 un 1 introduced to suppress the high frequency oscillations. With damping
_ n1=2
u v
n1=2
n1 n n1=2 un1 un : (22)
t t t the stable time increment is given by
un1 un tn1=2 vn1=2 : (23) where is the fraction of critical damping in the highest natural mode.
The stable time increment defined by Equation 31 or 32 is usually
The velocity at the nth time step is defined as follows: rather small (105 ~ 107 s) for regular structural simulations compared
with the total time length of earthquakes (101 ~ 102 s). Thus, usually a
1 n1=2
v n
v vn1=2 great number of time increments is required for the full process
2
1 1 n1 n 1
simulation of structural collapse. That is the price for the explicit
u u n1=2 un un1 : (24)
2 t n1=2
t
method for which the Jacobian of system is neglected.
4 | D E S C R I P T I O N OF SI NG L E S E I S M I C
un1 2un un1
vn : (25) RECORD INCREMENTAL DYNAMIC ANALYSIS
2tn
The acceleration is The seismic collapse of structures is by its essence a dynamic stability
For the case of equal time steps, the reduced form of acceleration the dynamic response process of structures, which are subjected to
could be obtained as follows: different levels of ground motion while structural responses are shifted
from the elastic state to the nonlinear state until collapse. IDA not only
un1 2un un1 can evaluate the anticollapse capacity of structure, but also can grasp
u n an : (27)
tn 2 the seismic performance of structure.
This paper develops the IDA process specifically for the structure
The integration formula of acceleration is
of ZhongNan Center, a supertall building. The key steps of implemen-
vn1=2 vn1=2 tn an : (28) tation are as follows:
The system of structural dynamics could be expressed by the (1) Select a ground motion record and determine the intensity
following second order equations: measure.
Man Cvn fint un fnext ; (29) (2) Define the scale factor and obtain a suit of ground motion
records of different intensities.
where M and C are mass and damping matrices; and fint and fext are (3) Calculate the dynamic response of the structure subjected to the
internal and external forces. Substituting Equations 25 and 27 into ground motion records.
Equation 29 gives the explicit integration formula as follows:
(4) Assess seismic performance and anticollapse capacity of
" # structurebased on the simulated results.
M C
n u
n1
fnext fint un (30)
tn 2 2t " #
M C n n1 The ground motion of ChiChi, Taiwan 1999 is adopted for this
n 2u u :
tn 2 2t study. The earthquake was of a Richter magnitude of 7.62 with
moderate epicentral distance of 35.39 km. The recorded PGA was
According to Equation 30, the displacement at the (n + 1)th step 0.1031 g, peak ground velocity was 43.44 cm/sec, and permanent
n
could be explicitly updated by u without any matrix interactions and ground deformation was 33.27 cm. Figure 10 shows the ground
inversions. Thus, material damage and softening cannot enhance the motion record scaled to 1.0 g and the response spectra is shown in
difficulties to solve the structural system. It is also very easy to be Figure 11. PGA was adopted as the Monotonic Scalable Ground
HUANG ET AL. 7 of 11
Motion Intensity Measure, which was scaled to five levels including 5.2 | Patterns of collapse
0.2 g, 0.4 g, 0.8 g, 1.4 g, and 2.0 g.
The supertall structure collapses in rather different patterns under
different levels of seismic excitations. They are briefly summarized as
5 | RE S UL T S A N D D I S CU S S I O N S follows:
FIGURE 12 The first three mode shapes of model Case 3. PGA = 0.8 g
T1 T2 T3 T4 T5 T6 T7 T8 T9 T10
9.28 s 9.06 s 4.05 s 3.77 s 3.65 s 2.80s 2.14 s 2.08 s 2.02 s 1.39 s
8 of 11 HUANG ET AL.
In Case 3 (see Figures 1718), the initial damage of the structure of structure develops rather rapidly, and the shear wall appears to
also occurs in the top and the middle part where the section of the be seriously damaged in zone 3, zone 6, and zone 9. At 42.5 s the
core tube changes. Up till to 35 s the core tube in zone 1 and 9 cracks cut through the core tube at different sections at the heights
has been damaged more seriously compared to Case 1 and Case 2; of 221 m, 349 m, and 542 m, simultaneously. Then the whole
in the meanwhile, many coupling beams, outrigger trusses, and belt structure collapses rapidly due to impacts of dropping structural
trusses in the region have failed. In the time of 38 s, the damage components.
Case 4. PGA = 1.4 g damage distribution shows different features from other cases. At
35 s, the bottom of core tube starts to crack and form an evident
In Case 4 (see in Figures 1920), the damage of the structure also damage zone. After that, governing cracks and damage bands rapidly
initiates in the top and the middle part where the section of the core penetrate the bottom of the core tube due to the fatal earthquake
tube changes. Later on, different phenomena are shown. At 35 s the excitations. The core tube finally ruptures at the bottom, and the
damage of core tube is mainly concentrated in zones 69. Especially structure collapses at 38 s.
in zone 9, most of the structural components fail, and the structure It is indicated by the analytical results of IDA that the damage of
experiences significant tilt. At 37.5 s, the damage in core tube develops the supertall building structure initiates in the similar regions, which
in a rather rapid way to form the damage zones. Finally, at the time of are mainly within upper region and the middle region whereas the
40 s, the crack cuts through the core tube at the heights of 151 m and section of shear wall tube changes. The reason is that in the early stage
349 m, and the structure collapses. of earthquake, the supertall building structure is governed by the
elastic deformation, for which the stresses concentrate in the regions
Case 5. PGA = 2.0 g with changes of stiffness. Then with the increase of structural
deformation, the P effect gradually governs the development of
In Case 5 (see Figures 2122), the PGA reaches 2.0 g. The initial the structural damage. As the peak intensity of ground acceleration
damage of the structure also occurred in the top and the middle part increases, the structural deformation increases. In the meanwhile, the
where the section of shear wall tube changes. Later, the structural bending moment of the total structure in the bottom part also
increases faster than the top part due to the P effect. Thus, the (1) To numerically reproduce the seismic collapse of super tall
damage concentration zones of structure gradually goes downward. structures, a number of essential factors should be carefully
On the other hand, the supertall building structure usually has a series considered, e.g., material softening under severe deformation
of transfer region, which the internal forces are transferred between modeled by damage model, deformation of element modeled by
internal tube and outer columns by trusses. Thus, the inertial tube proper structural elements, contact between dropping elements,
would experience sudden change of overall bending moment within and illness of the system determined by explicit algorithm.
transfer region. This will significantly affect the development process (2) Incremental dynamic analysis gives a comprehensive assessment
of structural damage evolution. As a result, when subjected to earth- of not only the regular behavior but also the collapse of struc-
quakes with the same waveform but different PGAs, the supertall tures under seismic loads. As a dynamic equivalence of static
building structure experiences rather different patterns of damage pushover analysis, the supertall structure could be dynamically
evolution. Finally, the structure collapses in quite different schemas. pushed over to its collapse in rather different patterns.
same waveform, their local waveforms and intensities are quite [12] J. W. Ju, Journal of Engineering Mechanics. 1989, 115(11), 2507.
different corresponding to the critical moments including damage [13] J. Lemaitre, Proceeding of ICAM1, Japan 1971.
initiation, unstable damage propagation, and structural collapse. [14] J. Li, J. Y. Wu, In: Proceedings of the ISCC2004 (No. Key9), Changsha,
Thus, the final patterns of collapse for the super tall structures China 2004, pp. 209221.
are quite different. [15] J. Lemaitre, Journal of Engineering Materials Technology. 1985, 107, 83.
[16] J. Li, X. D. Ren, International Journal of Solids and Structures. 2009, 46,
(4) In the design of supertall building structures, at least multiple 2407.
collapse patterns should be considered, even for the ground [17] J. Y. Wu, J. Li, R. Faria, International Journal of Solids and Structures.
motion record with similar waveforms but different intensities. 2006, 43(34), 583.
[18] J. Mander, M. Priestley, R. Park, Journal of Structural Engineering 1988,
114(8), 1804.
ACKNOWLEDGEMEN TS [19] Z. Guo, X. Shi, Reinforced concrete theory and analyse (in Chinese),
China, Tsinghua University Press, Beijing 2003.
Financial supports from the National Science Foundation of China are
[20] X. D. Ren, S. J. Zeng, J. Li, Computational Mechanics 2014, 55(2), 267.
gratefully appreciated (GNs: 51261120374, 91315301, and
[21] T. Belytschko, W. K. Liu, B. Moran, K. I. Elkhodary, Nonlinear finite
51678439). elements for continua and structures, John Wiley & Sons 2013.
RE FE R ENC E S
[1] S. C. Huang, M. J. Skokan, Collapse of the Tungshing building during the AUTHOR BIOGRAPHIES
1999 ChiChi earthquake in Taiwan. in Proc. 7th U.S. National Conf. on
Earthquake Engineering, 2002, Boston. Tiancan Huang is a lecturer at Guangzhou University. He received his
[2] L. D. Carpenter, F. Naeim, M. Lew, N. F. Youssef, F. Rojas, G. R. PhD from Tongji University in the year 2017. His research interests
Saragoni, M. S. Adaros, Structural Design of Tall and Special Buildings. include nonlinear analysis of structures and high performance
2010, 19(8), 826.
computation.
[3] M. Nakashima, K. Inoue, M. Tada, Engineering Structure. 1998, 20(46),
271. Xiaodan Ren is an Associate Professor at Tongji University. He
[4] R. Villaverde, Journal Structure of Engineering ASCE. 2007, 133(1), 57. received his PhD from Tongji University in the year 2011. His research
[5] Building Seismic Safety Council, NEHRP guidelines for the seismic reha- interests include damage models of concrete and computational
bilitation of buildings. Publ. No. 273, prepared by the Applied Technology
methods for structures.
Council for the Building Seismic Safety Council, FEMA Washington, D.C
1997. Jie Li is a distinguished professor of the Department of Structural Engi-
[6] R. W. Clough, J. Penzien, Dynamics of structures, McGraw Hill, New neering and the State Key Laboratory of Reduction in Civil Engineering
York 1975.
at Tongji University. He received his PhD in structural engineering
[7] Building Seismic Safety Council, Recommended seismic design criteria
for new steel moment frame buildings. Publ. No. 350, FEMA, from Tongji University in China in 1988. His research interests include
Washington, D.C 2000a, 2000. stochastic damage mechanics, probability density evolution method,
[8] V. V. Bertero, in Strength and deformation capacities of buildings under and reliability assessment of concrete structures.
extreme environments. Structural Engineering and Structural Mechanics,
(Ed: K. S. Pister), PrenticeHall, Englewood Cliffs, NJ 1977.
[9] D. Vamvatsikos, C. A. Cornell, Earthquake Engineering Structure How to cite this article: Huang T, Ren X, Li J. Incremental
Dynamic. 2002, 31(3), 491. dynamic analysis of seismic collapse of supertall building struc-
[10] D. Vamvatsikos, C. A. Cornell, Earthquake Spectra. 2004, 20(2), 523. tures. Struct Design Tall Spec Build. 2017;e1370. https://doi.
[11] GB50011, 2001, Chinese Code for Seismic Design of Buildings, China org/10.1002/tal.1370
Architecture & Building Press, Beijing 2001.