Anda di halaman 1dari 21

Engineering Structures 148 (2017) 639–659

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Effects of podium interference on shear force distributions in tower walls


supporting tall buildings
Mehair Yacoubian a,⇑, Nelson Lam a, Elisa Lumantarna a, John L. Wilson b
a
Department of Infrastructure Engineering, The University of Melbourne, Melbourne, Victoria 3010, Australia
b
Centre for Sustainable Infrastructure, Swinburne University of Technology, Hawthorn, Victoria 3122, Australia

a r t i c l e i n f o a b s t r a c t

Article history: High rise constructions featuring a podium surrounding tower walls are often favoured for the versatile
Received 11 September 2016 functionality of the building. It is shown in this paper that the podium can impose significant differential
Revised 28 May 2017 restraint on coupled tower walls. Incompatible tower wall displacements under lateral loads were found
Accepted 29 June 2017
to be the main contributor to the generation of in-plane strutting forces in floors above and below the
Available online 12 July 2017
podium-tower interface level. Shear force localisations in the interior tower wall immediately above
the interface was found to be the direct consequence of these actions. Key parameters contributing to this
Keywords:
detrimental shear force localisation in a tower wall were analysed by way of parameter studies on rep-
Backstay effect
Podium-tower buildings
resentative models of the building and sub-assemblages. It is revealed that the in-plane rigid diaphragm
Wall shear force distribution assumption commonly adopted in practice can significantly suppress compatibility forces generated
High-rise building within the building floor leading to unconservative design of the tower walls. Elaborate nonlinear model
Wall shear failure has been examined to showcase the consequences of understating the shear demands on these walls.
Diaphragm in-plane flexibility Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction the podium. The amplitude of the induced shear force is dependent
on the in-plane flexibility of the floor structure connecting the pair
Podiums are augmented floor area at the lower level of a high of walls. This was first investigated in the early works of Bevan-
rise building which are common in metropolitan areas in regions Pitchard et al. [5] by means of linear analysis of the tower walls.
of low-to-moderate seismicity. The lateral load resisting system A quarter of a century later, Rad and Adebar [6] extended their
for such building configurations comprises moment resisting work into the inelastic domain and concluded that stiff sub-
frames and shear (or core) walls. As the tower walls of the building grade diaphragms and perimeter walls can lead to shear-critical
is offset from the centre of the podium, high torsional moments conditions occurring in tower walls below their base.
can be imposed on the podium [1,2]. High shear forces can also The backstay phenomenon as described is well known [7].
be induced on the structural walls thereby jeopardising their struc- However, shear anomalies generated by differential restraints on
tural integrity when subject to severe earthquake ground shaking. a tower wall (which has an offset from the centre of the podium)
Recommendations against this form of construction have not been is not well understood. The structural wall which is closer to the
mandated in many design codes of practices in spite of potential centre of the podium (referred herein as the interior wall) is subject
undesirable behaviour in a rare seismic event [3,4]. to higher moment restraints from the podium structure than the
At the podium-tower interface, horizontal forces are transferred exterior wall. As a result, high strutting forces are developed in
from the tower to the podium. Reactive forces are developed at the the connecting floor structure (beam and slab) to maintain com-
podium-tower interface to resist the overturning actions (Fig. 1). patibility. This strutting action can only be modelled accurately if
The reacting mechanism is synonymous to the back span of a can- the horizontal in-plane deformation of the floor diaphragm has
tilever. Intuitively, the described backstay mechanism can induce been incorporated into the modelling. Thus, the extent of such
high intensity shear force in the structural (tower) wall within actions can be misrepresented by analyses in which the (usual)
rigid floor diaphragm assumption has been made. Effects of dia-
phragm flexibility on the lateral response behaviour of the wall
⇑ Corresponding author. were examined by Pantazopoulou and Imran [8] Shin et al. [9]
E-mail addresses: myacoubian@student.unimelb.edu.au (M. Yacoubian), ntkl@ and Su et al. [10]. It was found that diaphragm flexibility in build-
unimelb.edu.au (N. Lam), elu@unimelb.edu.au (E. Lumantarna), jwilson@swin.edu. ings (featuring vertical or on-plan irregularity) can adversely affect
au (J.L. Wilson).

http://dx.doi.org/10.1016/j.engstruct.2017.06.075
0141-0296/Ó 2017 Elsevier Ltd. All rights reserved.
640 M. Yacoubian et al. / Engineering Structures 148 (2017) 639–659

Fig. 1. Backstay action in a podium-tower sub-assemblage.

(a) Tower walls offset from the centre (b) Centrally positioned tower walls.
Fig. 2. Example podium-tower sub-assemblages.

dual wall-frame interaction. This issue has been highlighted in the Bull [14] further examined incompatibility issues resulted from
PEER/ATC 72-1 document [7] inciting practitioners to use explicit abrupt stiffness variations up the height of the building and dual
floor models at the podium-tower interface, and particularly so frame-wall interaction. Their work highlighted the detrimental
in situations where there are high transfer forces. increase in the transfer (in-plane) forces when the structure under-
Rutenberg and Bayer et al. [11,12] studied the prevalence of goes inelastic response behaviour. Diaphragm-wall interaction
incompatibility (strutting) forces in slabs and beams connecting issues is further highlighted in the New Zealand earthquake load-
structural walls of different base dimensions. They concluded that ing standard, NZS 1170.5 [15] in which more detailed analytical
these in-plane forces were the result of incompatible inelastic models and procedures are mandated to encapsulate the effects
deformations within the structural wall. Gardiner et al. [13] and of diaphragms interference on the seismic behaviour of the
structure.
The implications of podium-tower interactions as described
have not been thoroughly covered in the research literature or in
Table 1 code provisions.
Geometric configuration of the examined sub-assemblages.

Dimension, indication Units in [mm]


Length of coupling beam, a 2000
Tower walls length, a & c 6000 Table 2
Tower wall thickness 300 Material properties used (structural walls).
Podium wall length, d (typical) 6000
Material properties
Coupling beam depth 1000
0
Clear podium span, e (typical) 6000 fC 40 MPa
Effective slab width (podium) [16] 3100 Poission’s ratio 0.2
Podium wall thickness (Typical) 600 Ec 31.6 GPa
M. Yacoubian et al. / Engineering Structures 148 (2017) 639–659 641

be developed in structural walls that are resulted from interfer-


ences from the podium structure. The cause of significant increase
in the shear intensity of the structural wall above the podium level
is to be illustrated. Results from the parametric studies have been
used to track the trends of the increase in the shear force intensi-
ties in order that conditions deserving special attention in design
can be identified (Section 2). Findings from parametric studies of
the sub-assemblages have been verified by static and dynamic
analyses of example 3D finite element (FE) models of existing
buildings assuming linear elastic behaviour (Section 3), and non-
linear inelastic behaviour (Section 4).

2. Analyses on 2D planar podium-tower sub assemblages

The first part of the study to be reported in this paper is based


on 2D planar sub-assemblages of a podium-tower building in
which tower walls are offset from the podium (Fig. 2a). Results
obtained from the analysis of the sub-assemblages are compared
with that of centrally configured sub-assemblages as control
(Fig. 2b). For both building models, the podium structure was sim-
plified to a series of walls that were connected by floor slabs. The
later methodology was chosen such that it can simulate real build-
ing conditions wherein the lateral stiffness of the tower is typically
20–60% of the lateral stiffness of the podium (the building models
used in the first part of the study pertain to lower-bound values).
Details of structural elements making up the subassemblies and
material properties assigned to the structural walls are sum-
marised in Tables 1 and 2 respectively.
The occurrence of significant strutting (axial) forces that are
Fig. 3. Strutting forces in the connecting floors and beams between the tower walls generated in the interconnecting floor beams and slabs in between
in offset and centred podium-tower sub-assemblages. the pair of shear walls are first illustrated (see Fig. 3). The magni-
tude of the strutting force is shown to be highest in situations
where the structural tower walls are offset significantly from the
In this paper, parametric studies based on quasi-static analyses centre of the building. Interestingly, a significant amount of strut-
were first conducted on two-dimensional (2D) planar podium- ting force is shown to occur above the podium level and up the
tower sub-assemblages to demonstrate shear anomalies that can height of the wall for as much as 10 m. In contrast, no strutting

(a) Deflection of hypothetical structure without CB (b) Deflection Ratio


Fig. 4. Kinematic behaviour of tower walls in podium-wall sub-assemblage.
642 M. Yacoubian et al. / Engineering Structures 148 (2017) 639–659

Fig. 5. Influence of strutting forces on wall shear forces.

(a) Shear force distribution in (b) Externally applied lateral


tower walls forces

Fig. 6. Shear force distribution in tower walls.

force is generated in the centrally positioned tower walls. The


observed anomalies is best illustrated by showing the deflection
of the tower walls in a hypothetical configuration in which the
floor beams connecting the twin walls have been intentionally
removed (Fig. 4a). The deflection of the interior wall normalised
with respect to that of the exterior wall is defined herein as the
deflection ratio Dr (Fig. 4b).
The ingression of the exterior wall onto the interior wall
(Fig. 4a) sheds light on the differential restraints exerted by the
podium structure on the tower walls. It is shown that Dr < 1 in
the lower (podium) portion of the model. This trend extends to
one-to-two stories above the level of the podium beyond which
the ratio tends to unity (suggesting compatible displacements are
achieved above this level). These strutting (in-plane) forces gener-
ated in the connecting beams can result in a localised increase in Fig. 7. Moment-shear (M/V) ratio of isolated and coupled shear walls [17].
M. Yacoubian et al. / Engineering Structures 148 (2017) 639–659 643

(a) Tower walls at offset from (b) Tower walls centrally


centre of podium positioned

Fig. 8. M/V ratio in tower walls in podium-tower sub-assemblage.

Fig. 9. Podium-tower sub-assemblages with varying podium height.

the shear intensity of the interior wall, whereas the shear intensity ence shear critical, or flexural shear critical, mode of failure in ulti-
is decreased in the exterior wall. This mechanism results in an mate conditions. In a coupled structural wall, bending moments up
asymmetric distribution of shear demands in the connected walls the height of the wall are redistributed into the connecting beams
as shown in Fig. 5. The resulting distribution of shear forces in in order that the M/V ratio is much less than that of an isolated
the twin walls above, and below, the podium-tower interface of cantilever wall [17] (Fig. 7). The M/V ratios of both the interior
the example structure is shown in Fig. 6a. and exterior tower walls that are offset from the buildings centre
The moment-shear ratio (M/V) has been used to characterise are shown in a similar manner (Fig. 8a). Compared with the M/V
the relative likelihood of a structural wall, or a column, to experi- ratio of the control model (Fig. 8b), the much lower ratio predicted
644 M. Yacoubian et al. / Engineering Structures 148 (2017) 639–659

was also found by employing the same technique. The relative


stiffness ratio is henceforth defined as the lateral stiffness of the
tower normalised with respect to the lateral stiffness of the
podium ðKT =KP Þ. The typical range of this ratio (for the building
stock surveyed in this paper) was found to be in the range 0.2–
0.6. Fig. 14 shows results obtained from the analyses of four differ-
ent sub-assemblage models (similar to that shown in Fig. 2a) with
varying wall thickness (and length) of the podium walls in order
that distinct stiffness ratios can be KT =KP = 0.2, 0.48, 0.64 and
0.85. It can be shown that as the value of the relative stiffness
parameter is increased, the restraint of the podium on the tower
Fig. 10. Correlation of wall shear force ratios with podium height ratios.
walls in the form of shear reversals at the interface level is reduced
(refer Fig. 14b). Interestingly the strutting (compatibility) forces in
for the interior wall is consistent with the shear anomalies illus- the coupling beams were accordingly reduced (Fig. 14c). Conse-
trated in the schematic diagram of Fig. 5. quently, the most adverse shear force concentration was mani-
Parametric studies have been undertaken on a group of building fested in the tower walls which were supported by a stiff
models featuring variable heights of the podium. The elevations of podium structure.
the six planar models of podium-tower building systems are In the seismic design and assessment of high-rise buildings
shown schematically in Fig. 9. Arbitrary lateral quasi-static loads rigid diaphragm constraints are often utilised at floor levels up
were applied to each of the considered models. Their respective the height of the building. This assumption is warranted given that
maximum shear force (Vint =Vext ) ratios are shown for comparison floor slabs often have limited in-plane flexibility. ASCE/SEI 7-10
in Fig. 10. It is shown that the most acute shear anomalies occur and ASCE 41 [19,20] recommend modelling slabs that are likely
when the height of the podium is about 1/4–1/3 of the height of to be subject to high transfer and inertia forces with semi-rigid,
the building (i.e. building 3P9T, 0.25). Further details of the strut- or flexible, characteristics. This can be achieved by distinctly mod-
ting forces and the resulting wall shear forces in two of the exam- elling the building floor based on its sectional and material proper-
ple models are also shown (Fig. 11a–c). Further parametric studies ties. Similarly, PEER/ATC 72-1 [7] recommends against the use of
on more podium-tower structural systems have been undertaken ‘‘rigid diaphragm” assumptions in the modelling of the podium-
to investigate the sensitivity of the maximum shear intensity to tower interface slabs (main backstay slabs). The effects of imposing
changes in the position of the tower (Fig. 12), wall length ratio such numerical constraints on the structural wall are examined by
(Fig. 13b & c) and clearance between the twin walls (Fig. 13d). It means of a parameter study in which planar sub-assemblage mod-
is shown that sub-assemblages with high eccentricity, low wall els similar to that shown in Fig. 2a were analysed. In a control
length ratio ðLint =Lext Þ and high clearance between the walls pertain model, rigid diaphragms were assigned to all stories above and
to development of high shear intensities.The effect the relative lat- below the podium level. Results by the computations were sub-
eral stiffness of the tower (and the podium) has upon shear force stantially different when compared with a model featuring explic-
distributions up the height of the tower wall is next examined. itly defined floor slabs up the height of the building (refer to
The effective stiffness of the tower ðKT Þ was computed by indepen- Fig. 15a & b). The comparison is not surprising given that rigid dia-
dently analysing the tower structure (which was fixed at the inter- phragm constraints ‘‘slave” all in-plane degrees of freedom on the
face level) by the use of the classical modelling technique floor to a master node. Consequently equal wall displacements are
recommended by Priestley et al. [18]. The stiffness of the podium numerically enforced at each floor level [9,21]. To investigate

(a) Strutting force profiles of (b) Shear force distribution (c) Shear force distribution (d) Externally applied loads on
connecting coupling beams for 1P11T [Tower walls] for 3P9T [Tower walls] the building models

Fig. 11. Internal forces in tower walls for the different podium-tower sub-assemblages.
M. Yacoubian et al. / Engineering Structures 148 (2017) 639–659 645

(a) Effect of eccentricity on the (b) Schematic representation of podium-tower


shear force ratio eccentricity calculation

Fig. 12. Correlation of wall shear force ratios with podium eccentricity.

(b) Effect of the relative wall length on


(a) Elevation the shear force ratio

(c) Wall deflection ratio (d) Strutting (in-plane) force


distributions
Fig. 13. Behaviour of tower walls at an offset from the centre of the podium.
646 M. Yacoubian et al. / Engineering Structures 148 (2017) 639–659

(a) Externally applied lateral loads (b) Tower wall shear distri-bution (c) Strutting force profile in the
on all building models. (interior and ex-terior walls) con-necting coupling beams
Fig. 14. Results of analyses of sub-assemblage building models for podium height ratio of 0.32 ðhp =hb ¼ 0:32Þ.

(a) Wall deflection ratio (b) Wall shear force distribution


Fig. 15. Results from sensitivity study for various diaphragm extents.
M. Yacoubian et al. / Engineering Structures 148 (2017) 639–659 647

Table 4
Geometric details for analysed sets of walls.

Set I walls
Wall thickness 300 [mm]
Interior wall length 6100 [mm]
Exterior wall length 7700 [mm]
Set II walls
Wall thickness 350 [mm]
Interior wall length 7700 [mm]
Exterior wall length 7700 [mm]

The case study building is a podium-tower structure wherein the


tower is offset from the centre of stiffness of the podium (refer
Fig. 16. 3D rendering of case study building structure. to Fig. 16). The medium rise ten-storey reinforced concrete struc-
ture has been designed and detailed for gravity and wind loads,
but without taking into account the potential occurrence of seismic
Table 3 actions. The building features vertical irregularity above the 4th
Loads in case study building. floor and with setbacks from the 5th floor up the height of the
Podium Tower building. The lateral load resisting system consists of dual moment
[kPa] [kPa] frames-wall systems spanning in both directions. Design floor
Imposed loads 4.00 3.00 loadings are summarised in Table 3.
Electromechanical loadsa 3.00 2.00 The FE model of the building has been assembled using program
Superimposed dead loads (including floor 1.00 1.00 ETABS [22] with shell elements to represent the floor slabs and the
cover) structural walls, and frame elements to represent the beams and
Notes: the columns. Loads were uniformly distributed on the floor slabs
a
The case study building functions as a hospital, the Electromechanical loads are based on values listed in Table 3. Seismic mass was assigned as
categorised as additional loads (dead) applied to the floor slabs by virtue of the per the estimated self-weights of the structural elements, the per-
piping, heavy duty HVAC ducts and other equipment.
manent floor loads and 30% of the imposed floor loads in accor-
dance with the Australian earthquake loading standard, AS1170.4
[4]. Two modelling approaches were employed for the analysis of
trade-offs between practicality and accuracies with the modelling, the structure when subject to lateral seismic actions. In the first
sensitivity analyses were conducted to determine the portion of model, rigid floor diaphragms (as commonly assumed in design
the buildings height that requires explicit modelling of the floor practices) were adopted. In the second model, explicit modelling
slabs. It is shown that sufficiently accurate results are obtained of the floor structure was adopted.
when the floor slabs in between these height limits are explicitly The extent of podium interference on wall behaviour (tower
modelled: (i) two stories above and (ii) 60% of the height of the walls) was predicted to be substantial as shown in the 2D planar
podium (designated by: 2T + 0.6 HP in Fig. 15a & b) sub-assemblages parametric studies (Section 2). Tower walls I
and II (as shown in Fig. 17) were positioned very differently on
3. Verification studies on 3D FE model of buildings the floor plan: with the Set I walls being close to the centre of stiff-
ness of the podium whereas the Set II walls were much further
A 3D FE model of an existing building was employed in static away from the centre. Table 4 contains a listing of the dimensions
and dynamic analyses to verify findings reported in Section 2. of the walls in both wall sets.

(a) Plan view of typical tower floor (b) Plan view of typical podium floor
showing the examined walls showing the examined walls
Fig. 17. Floor plans of case study building structure.
648 M. Yacoubian et al. / Engineering Structures 148 (2017) 639–659

(a) Set I (b) Set II


Fig. 18. Wall shear force distributions of case study building structure.

(a) Podium height effect (b) Wall length effects


Fig. 19. Wall shear force ratio of elastic case study building structure.

Fig. 20. Comparison between parametric study and results from case study
buildings [podium height ratio]. Fig. 21. Acceleration response spectra of artificial time histories (log-log scale).
M. Yacoubian et al. / Engineering Structures 148 (2017) 639–659 649

(a) Maximum shear response: -ve Shear (b) Maximum shear response: +ve Shear
Fig. 22. Shear force distribution of Set II walls, models with rigid diaphragms assigned.

(a) Maximum shear response: -ve Shear (b) Maximum shear response: +ve Shear
Fig. 23. Shear force distribution of Set II walls, models with explicit diaphragm definition.

Lateral loads that were applied to the structure were assigned in


Table 5
accordance with the equivalent static force procedure in AS 1170.4
Shear ratios for models with explicit diaphragms.
(2% exceedance in 50 years on soil class D) [13]. Shear force distri-
bution on both the interior wall (the wall which was closer to cen- ½Vint =Vext -ve Shear ½Vint =Vext +ve Shear
tre of stiffness of the podium) and the exterior walls have been Art I 1.36 1.27
analysed. Results of the estimated shear forces up the height of Art II 1.35 1.27
the walls in both models based on the rigid diaphragm assumption Art III 1.25 1.29
Art IV 1.36 1.27
and explicit modelling are shown in Fig. 18. Art V 1.34 1.33
The backstay mechanism is shown to have resulted in reversal Art VI 1.25 1.28
of the shear force below the level of the podium-tower interface. Art VII 1.27 1.34
Conservative predictions were obtained from analyses based on Mean 1.31 1.29
the rigid diaphragm assumption at, and below, the podium level. STD 0.05 0.03
However, the highest shear force intensity is shown to occur above COV (%) 3.84% 2.12%
650 M. Yacoubian et al. / Engineering Structures 148 (2017) 639–659

Fig. 24. Base shear of the suite of artificial records.

wall. In Set II the wall shear force was 4050 KN and 2990 KN for
the interior and exterior walls respectively. The shear force
ðVint =Vext Þ ratio was accordingly 1.28 and 1.35 for the two wall sets.
Results of correlation between the ðVint =Vext Þ ratio and the podium
height ratio, and the wall length ratio are consistent with observa-
tions from the parametric studies of the podium-tower sub-
assemblages (refer Fig. 18a & b).
Results derived from the 3D FE analyses of the building model
considered in this section (Fig. 16) have been used to provide fur-
ther support of the predictive relationships derived in the previous
Hpodium LExterior
sections as to how Hbuilding
and LInterior
parameters control shear
anomalies in the structural walls (Fig. 19a & b).
Fig. 25. Out-of-phase vibration of case study building. To expand the scope of the study, additional case study build-
ings featuring comparable configurations have been examined fur-
ther. Details of the numerical models are provided in the Appendix
(and not below) the level of the podium-tower interface when A (Figs. 40, 41 & 43). Results from linear lateral analysis were
explicit modelling of the in-plane action of the building floor was superposed on results obtained from the parametric study
adopted. The interior walls were subject to much higher shear (Fig. 20). It is shown that results from the latter provided an upper
intensities than the interior walls. In Set I, the shear force in the bound approximation to the detrimental shear demand localisa-
interior wall was 4850 KN compared to only 3800 in the exterior tion (shear ratio) above the podium interface.

Fig. 26. Nonlinear numerical model with and without secondary columns.
M. Yacoubian et al. / Engineering Structures 148 (2017) 639–659 651

3.1. Linear time history analysis demands in the tower walls were found from analyses in which
the explicit diaphragm definition has been adopted (similar to
To further investigate the uneven share of shear distribution earlier findings). Time histories of the base shear are shown in
between the twin tower walls, linear time history analyses were Fig. 24.
performed using ETABS on the case study building (Fig. 16). Seis- Considerable interferences of the higher modes contributed to
moArtif [23]was used to generate a suite of artificial time history further anomalies in the shear distribution trends up the height
records based on the AS 1170.4 target spectra for 2% exceedance of the wall. This can be seen by comparing shear reversal in Figs. 22
in a design life of 50 years and site class D (Fig. 21). For each record, and 23. The backstay effect for the models with rigid diaphragms
two sets of analysis were performed: one with rigid diaphragms would only be evident at a level below the interface. This phe-
assigned at each floor level and one with diaphragms that have nomenon relates to the out-of-phase vibration of the tower and
been explicitly modelled. podium portions of the building as depicted in Fig. 25.
Consistent with the earlier findings, both shear force magnitude
and maximum location varied between the different modelling
4. Inelastic shear behaviour of coupled tower walls
approaches (with rigid diaphragms imposed and explicit dia-
phragm modelling). The 5% damped shear response of set II walls
In view of limitations with analyses based on the assumption of
(refer to Fig. 17) are plotted in Figs. 22 and 23. Table 5 summarises
linear elastic behaviour, non-linear time history (NLTH) analysis of
maximum shear ratios obtained from the analyses. On average, the
case study building 3 (refer to Appendix A) and inelastic push over
interior wall experienced 30% higher shear demands than the
analyses have also been employed. Results from non-linear analy-
exterior wall when in-plane actions of the floors have been explic-
ses confirm the findings of the parametric studies in terms of the
itly modelled. The observed behaviour was in close agreement
increased likelihood of the interior shear walls experiencing higher
with results obtained from the equivalent static analyses. Con-
shear demands (Section 4.1) and shear critical behaviour in the
versely, no shear force anomalies were found in models with rigid
interior tower walls (Section 4.2).
diaphragms assigned for the collectors. Generally, higher shear

4.1. Shear demands in nonlinear time history analysis

NLTH Analysis was performed on a simplified model represent-


ing the lateral load resisting system of the 56 storey building (see
Fig. 41 in Appendix). A simplified model was numerically con-
structed using the SeismoStruct [24] platform. Connecting slabs
were modelled as inelastic fibre section beam elements with effec-
tive slab width as recommended by Grossman, Shin et al. and
Zekioglu et al. [9,16,21] (see Fig. 42 in Appendix). The U-Shaped
coupled core walls were modelled using wide column analogy
based on recommendations given by Bayer et al. [25]. Five
displacement-based inelastic beam-column elements were
assigned to individual levels in the building (see Fig. 26). At the
cross section of the core wall each element was divided into ‘‘fi-
bres” with distinct uniaxial material properties which were
assigned to both the concrete and the reinforcing steel. The Mander
concrete material model was assigned for both the confined and
unconfined concrete. A modification factor of 1.11 was applied to
account for the effects of confinement [26]. The Meneggotto-
Pinto model was adopted for modelling the steel reinforcement
[27]. Masses of the structural members were manually calculated
and lumped at the nodes (see Fig. 43).
Rayleigh (modal) damping was used in the nonlinear time his-
tory analyses. The value of the mass and stiffness proportional
damping parameters were determined by assigning a 2.5% damp-
ing at 6.2 s, and 1.16 s, which correspond to 1.1T1 ; and 0.2 T1
Fig. 27. Deformed shape at onset of flexural yielding of coupling beams. respectively (refer Fig. 44 in the Appendix for details). The natural

Fig. 28. Base shear time history.


652 M. Yacoubian et al. / Engineering Structures 148 (2017) 639–659

period of vibration of the first mode (T1 ) was found to be 5.79 s. The models were subjected to excitations by ground motions as
This approach is consistent with recommendations given in recorded from the 1989 Loma Prieta earthquake [28]. It is shown in
PEER/ATC guide for tall buildings [7]. Two sets of simulations were Fig. 27 that the model with secondary columns exhibited signifi-
considered: first with the secondary columns (for resisting gravity cant higher mode contributions compared with the model with
loads) were included in the numerical modelling, and second a only the tower cores. The base shear time histories for both models
more conservative model with only the tower core walls consid- are plotted in Fig. 28 which shows that comparable base shear was
ered in the analysis (see Fig. 26). The main purpose of conducting observed for both models.
the analyses was to demonstrate the extent of podium interfer- Shear demands on the structural walls were assessed for both
ences on the shear demands of the primary and secondary lateral models. Fig. 29 plots the shear force distribution within the vicinity
systems in podium-tower buildings. of the podium interface at the onset of flexural yielding of the cou-

(a) at first yield of coupling beams (b) at maximum base shear


Fig. 29. Wall shear force distribution.

Fig. 30. Strutting force time histories.

Table 6
Geometric and material details of wall specimens used in calibration.

Specimen name 0
f c ½MPa h
lw
qt ½% qv ½% qbf ½% qbt ½% 0
P
f c Ag

LW2 [32] 41.6 1.125 0.50 0.50 1.40 0.97 0.05


MW2[32] 41.2 1.625 0.50 0.50 1.40 0.97 0.05
M4 [33] 24.4 0.69a 0.3 0.3 0 0 0.09–0.10
a
M=V ratio reported.
M. Yacoubian et al. / Engineering Structures 148 (2017) 639–659 653

Fig. 31. Simulated and experimental response of specimens: (a) LW2 [29] (b) M4 [30]

pling beams and maximum base shear (Fig. 29a and b respec- for the concrete and the smeared reinforcement meshes (top and
tively). Shear localisation and asymmetric demands were observed bottom); lap-splicing was not modelled explicitly.
in the tower core which is consistent with earlier findings. The The sub-assemblage as shown in Fig. 32 was numerically con-
ðVint =Vext Þ ratios for the core walls in the model which had the sec- structed on VecTor2 and was assigned a calibrated material (as
ondary (gravity) columns (designated as: Only Cores) were found demonstrated earlier). Geometric and sectional properties are
to be higher when compared with the model with secondary col- given in Table 7. The 20 storey sub-assemblage consists of two pla-
umns included in the analysis (1.4 and 1.3 respectively). Results nar coupled shear walls attached to a podium by means of 8 m dia-
obtained from linear elastic analysis (Section 2) as the upper bound phragms at the first and second floors. The diaphragms were
estimate (of 1.39) suggests that contributions from the higher
modes on the presented shear distribution anomalies are minimal.
As stated in Section 2, strutting forces in the connecting slabs
and beams were found to be the primary contributor to the pre-
sented wall shear demand anomaly. Interestingly, pronounced
strutting forces were observed in the slabs which connect the
tower cores to the secondary (gravity) columns (Fig. 30). These
forces caused unfavourable increase in the shear demand of the
interior secondary columns just above the level of the podium
(see Fig. 29a and b).

4.2. Discrete modelling of shear failure progression in podium-tower


buildings

The NLTH analysis discussed in Section 4.1 provided further


verification to issues pertaining to shear anomalies in the tower
walls featuring setbacks as presented earlier. The direct conse-
quences of these anomalies are to put forth in a discrete non-
linear sub-assemblage model that is capable of modelling shear-
critical behaviour and axial-flexure-shear interaction in the planar
walls. VecTor2 [29] has been employed for this purpose. The theo-
retical framework of element formulation in VecTor2 is based on
the Modified Compression field theory (MCFT) [30] and Disturbed
Stress field model (DSFM) [31]. The nonlinear FE model of the sub-
assemblage was first calibrated against results reported from Fig. 32. Elevation of inelastic case study building structure (VecTor2).
experimental programs on limited-ductile walls representative of
the walls commonly found in low-to-moderate seismic regions,
such as Australia. Publications by Bing et al. [32] walls: LW2 and Table 7
MW2, and Griefenhagen et al. M4 specimen [33]were selected Reinforcement and modelling parameter details of the nonlinear model.
for this purpose (details given in Table 6). Results from the numer- Tower wall
ical simulations on VecTor2 were superimposed on the experimen-
Vertical web reinforcement ratio qv ½% 1.5
tal data points and backbone curves representing experimental Transverse web reinforcement ratio qt ½% 0.25
data. It is shown that the numerical simulations were capable of Axial load ratio 0
P=f c Ag 0.05
simulating displacements, stiffness and ultimate strength to a rea- Wall thickness twall 200 [mm]
sonable degree of accuracies (see Fig. 31). The simulated results Coupling beams
were able to generally capture the trend of failure (cracking pat- Longitudinal reinforcement ratio ql ½% 1.5
terns and reinforcement fracture) and local effects with an even Transverse reinforcement ratio qt ½% 0.86
Depth of section dCB 850 [mm]
high degree of accuracies. The backstay floor slabs were numeri-
cally modelled by assigning an effective width based on recom- Backstay diaphragms
mendations given by Grossman [16]. The slab elements were Effective slab width beff ½m 4.10
Longitudinal reinforcement ratio q½% 0.75
modelled as quadrilateral (4-nodes) shell elements which are read- Slab thickness tslab 250[mm]
ily available in VecTor2. Suitable material models were assigned
654 M. Yacoubian et al. / Engineering Structures 148 (2017) 639–659

modelled with an effective width according to recommendations found to be approximately 12% higher compared to the case of an
set by Grossman [16]. Modal pushover analysis was performed elastic podium. This is so because of the redistribution of forces
on the 60 m sub-assemblage based on the fundamental modal between the tower and the podium (prompted by the softened
behaviour of the building. The latter was warranted given that
the intent of the analysis was to demonstrate relative behaviour
between exterior and interior tower walls and to encapsulate fail-
ure mechanism under extreme conditions, and not to accurately
capture period shifts and variation in the dynamic properties of
the structure in the inelastic range. Furthermore, the effect of the
higher modes on the shear distribution anomalies presented in this
paper was shown to be minimal (see Section 4.1) partially validat-
ing the choice of the lateral load distribution profile. The adequacy
of this assumption has been examined towards the end of this sec-
tion. The stiffness of the elastic podium was lumped into equiva-
lent material properties. This latter assumption prompted better
computational efficiency which was warranted given that the
tower walls and the connecting diaphragms were essentially the
main focus of the study. A smeared reinforcement modelling
approach was adopted.
Two sets of simulations were conducted. In the first set of sim-
ulations, discrete modelling of the podium based on equivalent
stiffness and strength was constructed (case 1). In the second set
of simulations elastic material properties of the podium elements
Fig. 35. Normalised in-plane backstay forces in the main backstay diaphragm (at
were specified as for the first set (case 2). The latter was conducted the podium interface level).
for the purpose of examining the influence of yielding at the
podium level and the implications on the overall seismic perfor-
mance behaviour of the sub-assemblage. In both sets of simula-
tions, the response behaviour of the tower walls was governed
by a brittle shear failure mechanism which is characterised by a
prominent diagonal crack forming in the interior wall leading to
a sliding mode of failure (along diagonal crack) followed by axial
failure of the tower. With reference to Fig. 33, when inelastic beha-
viour is prompted in the podium, the ultimate shear capacity was

Fig. 33. Push over curve for the sub-assemblage (shear response included).

Fig. 34. Considered response phases on the push over curve. Fig. 36. Simulated damage in the four response phases.
M. Yacoubian et al. / Engineering Structures 148 (2017) 639–659 655

stiffness of the podium at the onset of yielding and the consequent at the ends for all of the coupling beams above the podium-
partial ‘‘relief” of the restraint on the interior wall). Roof displace- tower interface. Upon the onset of shear yielding, cracking in the
ment in the former case is about 30% higher when compared with wall was found to occur in the interior wall and specifically in
the latter implying that drift demands are less restraint with the level just above the interface. As the response ingress into
inelastic podiums. the inelastic range, most of the successive inelasticity was chan-
In analysing failure progression in the sub-assemblage (case 1), nelled into the widening of the prominent diagonal shear crack.
four different states were considered: flexural yielding of coupling In stage (III), shear mechanism was prompted along the diagonal
beams, onset of shear yielding in interior wall, formation of diago- and in the subsequent displacement steps. Stage IV culminates in
nal shear crack and sliding shear failure along the prominent diag- abrupt shear and axial failure in the interior wall (which was char-
onal crack. The stages are numbered as: I, II, III and IV respectively. acterised by a rapid degradation in strength as shown in the force-
As shown in Figs. 34 and 36, the first stage (I) is mainly charac- displacement curve). Fig. 35 shows the value of the backstay force
terised with flexural yielding of the coupling beams. However, in the main backstay diaphragm at the interface level. The value
0
the effects of the strutting actions in the coupling beams can shown has been normalised with respect to f c Aeff (where Aeff is
implicitly be inferred from the yielding pattern up the height of the gross sectional area of the slab). An abrupt increase in the in-
the tower. For levels in the immediate proximity of the podium- plane forces of the backstay diaphragm are shown in between
tower interface, the occurrence of yield (as inferred from cracking stages II and III. This is attributed to the difference in the yielding
patterns at the ends of the beam) range from mild to non-existent. pattern between the podium and the tower structures. In the case
The latter implies that the additional in-plane stresses (in the form of an elastically responding podium structure, the increase is pro-
of strutting forces) augment stiffness (flexural) of the beams and nounced given that large forces are redistributed from the yielding
thereby prolonging yield displacement. The lag extends up until (softened) tower to the elastic (stiffer) podium structure. The mag-
the occurrence of stage (II), where flexural hinging is pronounced nitude of the normalised forces are shown to be lower than 0.15
(with a maximum value at 0.12), which is deemed to be not high
enough to cause premature failure of the backstay diaphragm
(see Fig. 37).
Damage to the shear walls in Stage I to IV is depicted in Fig. 36.
Clearly, the interior wall is shown to be much more susceptible to
shear-critical failure than the exterior wall.
The response behaviour of the 20 storey sub-assemblage model
of the building typifies ‘‘force-controlled” behaviour which is often
associated with a shear-dominated response of the structural
members (walls). FEMA 356 [34] characterises a member as ”
force-controlled” when the response of the member exhibits elas-
tic behaviour prior to reaching the ultimate limit state which can
be followed by a sudden loss in the gravitational load carrying
capacity of the member at the collapse-limit state. This is consis-
tent with response behaviour observed for shear-critical non-
seismically detailed walls. The shear force in the walls at stages
I-IV have been normalised with respect to the lower-bound shear
Fig. 37. Shear demand to capacity ratio for interior and exterior walls. capacity of the walls as recommended in ACI 318 [35]. The shear

Fig. 38. Stress distribution of the transverse reinforcement of the wall after stage III.
656 M. Yacoubian et al. / Engineering Structures 148 (2017) 639–659

force demand-capacity ratio of the interior wall in stage III (at the boundary conditions imposed by the adjoining podium struc-
onset of shear yielding) was approximately equal to 1.0 whereas ture. The magnitude of the strutting (compatibility) forces gen-
the ratio was approximately equal to 0.61 for the exterior wall. erated in the coupling beams are proportional to the relative
Subsequent shear demands on the interior wall can result in signif- drifts in the tower and the podium structure. In ultimate condi-
icant ‘‘shear yielding” which is characterised by diagonal cracking tions, the first mode of vibration (with the largest tower drift) is
and yielding of the transverse reinforcement up until the collapse deemed most critical to the shear response behaviour of the
limit state is reached in stage IV. tower walls.
Furthermore, following the formation of the prominent diago-
nal crack (between stages III and IV), stresses in the transverse The simplified sub-assemblage model presented herein has pro-
reinforcement in the interior walls are shown to have reached vided better insight into the consequence of localised shear con-
their characteristic yield capacity (approximately 410 MPa). This centration in the tower walls. As discussed earlier, these localised
indicates the existence of substantial stress transfer from effects imposed on the offset tower wall model might not have
the cracked concrete to the transverse bars prior to axial been evident in the simplified analyses of the building structure
failure which was observed in the interior wall at stage IV (see (based on rigid diaphragm assumptions) as is commonly adopted
Fig. 38). in design practices.
Fig. 39 presents the curvature distribution up the height of the
interior wall in both elastic and inelastic podium models for two
5. Conclusion and recommendations
distinct stages: I and II. Figures for inelastic podium model for both
stages exhibit larger spread of inelastic behaviour compared with
This paper sheds light on the unfavourable interference of the
the elastic podium model as is reflected in the curvature distribu-
podium structure on twin tower walls. It was found that podium
tion profiles. Curvatures below the interface level are higher in the
structures can impose significant differential restraints on the
inelastic podium model compared to the elastic model implying a
walls. Diaphragm-wall interaction in the form of (strutting) com-
lower degree of inelastic distribution occurring in the former
patibility forces were mobilised in the stories immediately above,
model. Synonymously, comparing the area enclosed by the curva-
and below, the podium level. These forces are shown to
ture profiles the higher inelastic deformation capacity maintained
redistribute internal actions between the interior and the exterior
by the interior wall in the inelastic podium model (see Fig. 33) is
wall, and consequently offsetting the distribution of the wall shear
further highlighted. The lateral load profile (based on the first
force equilibrium distribution above the level of the podium. Sig-
mode shape) is valid provided that the following conditions are
nificant shear concentration can be resulted from the redistribu-
met:
tion. A parametric study was devised to quantify factors affecting
these shear anomalies. Rigid diaphragm assumptions typically
1. No significant stiffness variation (tangent stiffness) is shown in
enforced in the design and assessment of high rise buildings were
the force-displacement behaviour of the building (see Fig. 33).
assessed in the light of the podium-tower interferences as
Analogously, no significant variation in the fundamental mode
described. It was found that such (commonly used) assumptions
shape or period lengthening can be associated with the
may lead to unconservative representation of the shear demands
response behaviour of the building which is in strike contrasts
in the tower walls of a building.
to buildings exhibiting large ductility demands.
Results from analyses of the case study buildings provided fur-
2. The brittle (non-ductile) response behaviour of the interior wall
ther validation to the initial parametric study. The ramifications of
in a coupled shear wall is primarily caused by differential
shear distribution between interior and exterior walls on failure
progression in the inelastic realm were examined by means of non-
linear pushover analysis on a simplified sub-assemblage. It was
shown that shear concentrations within the direct proximity of
the interface resulted in brittle shear failure of limited ductile inte-
rior walls. Further, secondary (gravity) walls and columns con-
nected to primary walls or cores were also shown to exhibit high
localised shear demands above the level of the podium.
Several key recommendations for practitioners are summarised
as follows:

1. Explicit modelling of the floor slabs in between these two


height limits: (i) two stories above the podium and (ii) 60% of
the height of the podium.
2. Alternative design procedures to reduce podium restraints on
the tower walls by the use of properly detailed expansion or
settlement joints at the podium-tower interface.
3. Use of procedures stipulated in PEER/ATC 72-1 [7] for the
design and detailing of floor slabs in locations where high strut-
ting forces are expected.
4. Consideration of interactions between primary and secondary
gravity systems in the lateral analysis of the type of buildings
considered in this study.

Funding

Fig. 39. Curvature distribution of interior wall within the vicinity of interface in The support of the Commonwealth of Australia through the
response stages I and II. Cooperative Research Centre program is acknowledged.
M. Yacoubian et al. / Engineering Structures 148 (2017) 639–659 657

Appendix A

(a) Floor Plans (b) 3D render


Fig. 40. Case study building 2.

(a) Floor Plans (b) 3D render


Fig. 41. Case study building 3.
658 M. Yacoubian et al. / Engineering Structures 148 (2017) 639–659

Fig. 42. Slab-wall/column representation for NLTH analysis (Section 4.1).

(a) Floor Plans (b) 3D render


Fig. 43. Case study building 4.

References

[1] Elnashai A, Soliman M. Effect of building configuration on seismic response


parameters; 1995.
[2] Moehle JP. Seismic response of vertically irregular structures. J Struct Eng
1984;110:2002–14.
[3] Standards Australia, AS 3600–2009, Concrete structures. Sydney, NSW 2001,
Australia: SAI Global Limited; 2009.
[4] Standards Australia, AS 1170.4-2007, Structural design actions - part 4:
earthquake actions in Australia Sydney, NSW 2001, Australia: SAI Global
Limited under licence from Standards Australia Limited; 2007.
[5] Bevan-Pritchard G, Man E, Anderson D. Force distribution between core and
sub-grade structure of high-rise buildings subjected to lateral load induced
forces. In: Proceeding of 4th Canadian conference on Earthquake Engineering,
Vancouver; 1983. p. 210–9.
[6] Rad BR, Adebar P. Seismic design of high-rise concrete walls: reverse shear due
to diaphragms below flexural hinge. J Struct Eng 2009;135:916–24.
[7] PEER/ATC. Modeling and acceptance criteria for seismic design and analysis of
tall buildings. Redwood City, CA: Applied Technology Council in cooperation
with the Pacific Earthquake Engineering Research Center; 2010.
Fig. 44. Rayleigh damping model employed in the NLTH (Section 4.1).
M. Yacoubian et al. / Engineering Structures 148 (2017) 639–659 659

[8] Pantazopoulou S, Imran I. Slab-wall connections under lateral forces. ACI [22] Habibullah A. ETABS-three dimensional analysis of building systems, users
Struct J 1992;89. manual. Berkeley, California: Computers and Structures Inc.; 1997.
[9] Shin M, Kang THK, Grossman JS. Practical modelling of high-rise dual systems [23] Seismosoft. SeismoArtif. Pavia, Italy; 2012.
with reinforced concrete slab-column frames. Struct Des Tall Special Build [24] SeismoSoft. Seismostruct, A computer program for static and dynamic
2010;19:728–49. nonlinear analysis of framed structures; 2007.
[10] Su R, Cheng M. Earthquake-induced shear concentration in shear walls above [25] Beyer K, Dazio A, Priestley M. Inelastic wide-column models for U-shaped
transfer structures. Struct Des Tall Special Build 2009;18:657–71. reinforced concrete walls. J Earthquake Eng 2008;12:1–33.
[11] Rutenberg A. The seismic shear of ductile cantilever wall systems in [26] Mander JB, Priestley MJ, Park R. Theoretical stress-strain model for confined
multistorey structures. Earthquake Eng Struct Dynam 2004;33:881–96. concrete. J Struct Eng 1988;114:1804–26.
[12] Beyer K, Simonini S, Constantin R, Rutenberg A. Seismic shear distribution [27] Menegotto M, Pinto E. Method of analysis for cyclically loaded reinforced
among interconnected cantilever walls of different lengths. Earthquake Eng concrete plane frames including changes in geometry and non-elastic behavior
Struct Dynam 2014;43:1423–41. of elements under combined normal force and bending. In: IABSE symposium
[13] Gardiner D, Bull D, Carr A. Internal forces of concrete floor diaphragms in on the resistance and ultimate deformability of structures acted on by well-
multi-storey buildings. Citeseer: Department of Civil engineering, University defined repeated loads, Lisbon; 1973.
of Canterbury NZSEE Conference; 2008. [28] PEER strong motion data base Pacific Earthquake Engineering Research Center;
[14] Bull DK. Understanding the complexities of designing diaphragms in buildings 2014. Available at: <http://peer.berkeley.edu/peer_ground_motion_database>.
for earthquakes. Bull NZ Soc Earthq Eng 2004;37:70–88. [29] Wong P, Vecchio F, Trommels H. VecTor2—software for nonlinear analysis of
[15] Standards New Zealand. Structural Design Actions (NZS 1170.5). Wellington, two-dimensional reinforced concrete membrane structures; 2014.
New Zealand; 2004. [30] Vecchio FJ, Collins MP. The modified compression-field theory for reinforced
[16] Grossman JS. Verification of proposed design methodologies for effective concrete elements subjected to shear. ACI J Proc: ACI; 1986.
width of slabs in slab-column frames. ACI Struct J 1997;94:181–96. [31] Vecchio F. Disturbed stress field model for reinforced concrete: formulation. J
[17] Krolicki J, Maffei J, Calvi G. Shear strength of reinforced concrete walls Struct Eng 2000;126:1070–7.
subjected to cyclic loading. J Earthquake Eng 2011;15:30–7. [32] Li B, Pan Z, Xiang W. Experimental evaluation of seismic performance of squat
[18] Priestley JN, Calvi GM, Kowalsky MJ. Displacement-based seismic design of RC structural walls with limited ductility reinforcing details. J Earthquake Eng
structures. IUSS Press; 2007. 2015;19:313–31.
[19] ASCE/SEI 41–06. Seismic rehabilitation of existing buildings. Reston, VA: ASCE; [33] Greifenhagen C, Lestuzzi P. Static cyclic tests on lightly reinforced concrete
2007. shear walls. Eng Struct 2005;27:1703–12.
[20] ASCE/SEI 7-10. Minimum design loads for buildings and other [34] FEMA 356. Prestandard and commentary for the seismic rehabilitation of
structures. Reston, VA: ASCE; 2010. buildings. Washington, DC: Federal Emergency Management Agency; 2000. p.
[21] Zekioglu A, Willford M, Jin L, Melek M. Case study using the Los Angeles tall 2000.
buildings structural design council guidelines: 40-storey concrete core wall [35] ACI 318. Building code requirements for structural concrete (ACI 318–08) and
building. Struct Des Tall Special Build 2007;16:583–97. commentary. American Concrete Institute; 2008.