Proceedings of 12th IOE Graduate Conference

Peer Reviewed
ISSN: 2350-8914 (Online), 2350-8906 (Print)
Year: 2022 Month: October Volume: 12

Effect of Soil Structure Interaction on Seismic Pounding in

Adjacent RC Buildings
Prabesh Dhakal a , Bharat Mandal b , Sanjay Kumar Sah c
a, b, c
Department of Civil Engineering, Pulchowk Campus, IOE, Tribhuvan University, Nepal
Corresponding Email: a [email protected], b [email protected], c
[email protected]

Abstract
This study presents the effects of soil structure interaction on the seismic pounding response of adjacent RC
buildings. Regular 2D frame of 3, 5 and 7 storey buildings considering two buildings at a time are taken for
analysis with both fixed and flexible base cases with no gap between the buildings. Non-linear modeling and
analysis of building and soil is accomplished in OpenSees, a finite element based software framework. 7
different ground motions are used for time history analysis. The results are obtained in the form of pounding
force and storey displacement values. The results indicate an increase in pounding force value due to
consideration of SSI. The force has increased by 34% for pounding between 3 and 5 storey buildings whereas
it has increased by 62% for pounding between 3 and 7 storey buildings. Due to combined effect of pounding
and SSI, there is increase in displacement of 7 storey building by upto 104% on the pounding side when
collided with 3 storey building whereas there is 12% reduction for 3 storey building on the pounding side for
the same case. Compared to shorter building, the taller building is more affected due to SSI which becomes
more vulnerable with the combined effect of both pounding and SSI. Finally, a relation to approximate the
pounding force that may be considered in the design of buildings is proposed.
Keywords
Pounding, Soil-structure interaction, Non-linear analysis, OpenSees

1. Introduction buildings which could lead to building damage [2].

Several researchers [1, 3, 4, 5] had the primary notable
Nepal lies in seismically active zone which is prone to contribution in the study of pounding phenomenon.
moderate to strong ground shaking causing damage to
infrastructures. Rapid population growth, unplanned 1.2 Soil Structure Interaction (SSI)
urbanization and increased land costs in the cities has
changed building constructions to be built very close to In conventional building design, buildings are
adjacent buildings with little to no gap between them. generally considered to be fixed at the base. But this
Due to this, pounding between adjacent buildings may assumption is valid only for structures founded on
occur during seismic events. rocky strata or soil with high stiffness. Soil structure
interaction refers to the response of structure due to
influence of soil and the response of soil due to
motion of structure [2]. SSI analysis of buildings
1.1 Pounding
takes the combined response of the superstructure, its
Pounding is the impact of adjacent buildings on each foundation system, and the soil beneath under seismic
other during seismic events due to the out of phase excitation through settlement, sliding and rocking
vibration when the separation gap between them is behavior of the foundation. SSI has been traditionally
insufficient than that required for the free vibration. thought to be of little benefit for typical seismic
Pounding is one of the major causes of damages in analyses as it reduces the demand on the buildings.
buildings during earthquake events. Pounding induced But it might not always be the case as the flexibility of
building damage happened during the 1985 Mexico whole system is increased due to consideration of
city and 1989 Loma Prieta earthquakes [1]. Pounding leading to increased time period, increased
significantly amplifies the seismic response of adjacent displacements and reduced base shear [6].

Pages: 497 – 504

Effect of Soil Structure Interaction on Seismic Pounding in Adjacent RC Buildings

1.3 Pounding Considering SSI Table 1: Building Description

SSI and influence on the pounding of the adjacent Storey Column Beam
buildings have been studied by several researchers 3 350 × 350 250 × 350
[2, 5, 7, 8]. Rahman et. al. [7] found shift in period 5 450 × 450 300 × 450
due to underlying soil altered the time at which first 7 550 × 550 350 × 550
impact occurred which had consequences on the
subsequent poundings. Naserkhaki et. al. [2]
investigated two MDOF models with lumped mass, Same storey height of 3m has been taken for all
viscous dampers and linear springs. SSI was taken buildings. The bay length is also constant for all cases
into account. A linear viscoelastic impact element equal to 4m.
was used the study. Results indicated that the The three building models are shown in figure 1. The
underlying soil negatively impacts the response of schematic model of superstructre is shown in figure 2.
buildings. Madani et. al. [8] studied the effects of
pounding and SSI on different combinations of
adjacent steel buildings resting on soil. Time history
analysis using 7 different earthquake records was
done on the model. Soil was modeled using BNWF
model. They concluded that soil flexibility had
increasing effect on the pounding forces and made it
happen even at farther clear distances.
As soil structure interaction is not generally considered
in design of buildings, the provided gap may not be
Figure 1: 3, 5 and 7 Storey Buidings (Fixed Base)
sufficient to prevent pounding. Thus, effets of SSI
should be studied properly and incorporated in design
to be safe from the damage during earthquake events.

1.4 Objectives
1. To determine the effects of SSI on pounding
between adjacent RC buildings.
2. To evaluate the pounding force required to be
considered in design of buildings.

1.5 Limitations
1. The study only considers regular 2D buildings
with equal storey heights.
2. Stiffness of infill wall is not considered.

2. Structural Model

2.1 Building Frame

For the purpose of our study, regular 2D framed
buildings of varying heights of 3, 5 and 7 storey with
3 bays are taken. M25 concrete with Fe500 steel rebar
has been considered for the RCC. The beam column
cross-section properties for the buildings are shown in
Figure 2: Schematic Diagram of Superstructure, Gap
table 1. Same storey height of 3m has been taken for
Modeling and Material Behavior
all buildings. The bay length is also constant for all
cases equal to 4m.

498
 Proceedings of 12th IOE Graduate Conference

2.2 Non-linear Modeling of Members from numerical model in good agreement with
experimental results.
The non-linear behavior of beams and columns of the
building frames are modeled with force based
distributed plasticity approach. The cross section
consists of a number of meshed fibers running from
one end to other end of the member. Each fiber is
assigned with uniaxial stress strain behavior of a
particular material.
The RC section is made up of three distinct materials;
steel rebars, confined concrete inside the stirrups and
unconfined concrete. The steel rebars are defined
using steel02 material available in OpenSees [9]
whereas concrete02 material is used to define both
confined and unconfined concrete. The properties of Figure 4: Cyclic Displacement Comparison Between
rebars is obtained from Giuffre Menegotto Pinto Experimental and OpenSees Model
model whereas, confined concrete properties are
determined from Mander’s confinement model [10].
2.3 Pounding Model

2.2.1 Validation of RCC Modeling Pounding phenomenon between adjacent buildings is

simulated by using force based contact elements. A
For validation of the used models for RCC, an viscoelastic gap element also known as Kelvin-Voigt
experiment conducted by Tanaka and Park was model is used in our study. This model was first
modeled in OpenSees and its cyclic displacement proposed by Anagnostopoulos [3] and is widely used
curve was compared with the experimental result. in modeling pounding phenomenon due to its
Tanaka and Park specimen #5 is shown in figure 3. simplicity and ability to model energy dissipation.
This model consists of a spring and damper material
in parallel combination and gets activated when the
predefined gap closes. equation 1 gives the equation
of impact force for the model during pounding.

F = kk (u1 − u2 − gap) + ck (u˙1 − u˙2 ); u1 − u2 > gap

(1)

Here, u1 and u2 are displacements of adjacent

buildings and gap is the separation distance between
the two. ck is the damping coefficient given by
Equation 2. kk is the stiffness of impact material and
m1 , m2 are the masses of the connecting nodes. The
damping ratio is derived from the restitution
coefficient e, given by equation 3. A rational value of
0.65 was taken for coefficient of restitution [5].
r
m1 m2
ck = 2ξ k (2)
m1 + m2

Figure 3: Experimental Setup and Cross-Section ln(e)

ξ = −p (3)
Details π + (ln(e))2
2

To calculate the stiffness value for the model, several

From figure 4 it is seen that the displacement obtained researchers have proposed different relations. The

499
Effect of Soil Structure Interaction on Seismic Pounding in Adjacent RC Buildings

relation used in this study was is shown in equation large capacity in compression and a small capacity in
4.[11] tension. PySimple2 material in OpenSees is used to
−1 model P-x springs which has pinching hysteretic
1 − ν12 1 − ν22

kk = + (4) behavior to model the potential gaping of embedded
EDyn,1 EDyn,2 shallow foundation under seismic loading. TzSimple2
material in OpenSees is used to model T-x springs to
Where, ν is the Poisson’s ratio and EDyn is the dynamic
account for frictional behavior of foundation due to
modulus of elasticity given by equation 5.
sliding.
EDyn = 5.82 (EStatic )0.63 (5) A non uniform distribution of vertical springs is
present with stiffer springs at the edges. This is to
Poisson’s ratio and static modulus of elasticity for account for the stiffened soil due to foundation
concrete were taken as 0.2 and 25 GPa respectively. rotation. The softer inner springs account for
The accuracy of this model was validated by remaining vertical stiffness. The distribution length
Jankowski and Mahmoud (2015) by comparing the and stiffness of springs are made such that provided
experimentally obtained displacement time history vertical and rotational stiffnesses should be equal to
with that obtained from numerical modeling. the vertical and rotational stiffnesses of foundation.
The equations 6, 7, 8 satisfy the above condition.
2.4 SSI Model
The interaction between the structure and underlying Kz
Kmid = (6)
soil is modeled using Beam on nonlinear Winkler BL
foundation (BNWF) model [12]. This model was first
proposed for pile foundations by Boulanger et. al.  
Kθ y K
[13] and Harden [14] suggested to use this model for Kend = Kmid + CR−V (7)
Iy
shallow foundations. Later, the model was calibrated
for shallow foundations by Raychowdhury. This
model accurately predicted experimentally measured K Kθ y − (Kz /A)Iy
footing response in terms of moment, shear, CR−V = (8)
Kθ y
settlement and rotation demands.
The BNWF model consists of series of vertical zero Here, Kmid and Kend are the stiffness of vertical
length elements (Q-z springs) below the footing and springs in middle and end zones. Kz and Kθ y are the
two horizontal zero length springs (T-x and P-x vertical and rotational stiffness of the foundation. Iy is
springs). Q-z springs simulate the settlement and moment of inertia of foundation and CKR−V is the
rocking behavior of the footing whereas P-x spring residual rotational stiffness ratio. The width of
simulate lateral passive behavior of soil and T-x exterior zone of foundation containing stiffer springs
spring simulate sliding behavior of soil. The model is is given by equation 9.
shown in figure 5.

1/3
 
1 K
Le = 0.5L − L 1 −CR−V (9)
8

Kz and Kθ y are calculated from Gazetas equations [15]

which are shown in equations 10 and 11.

"  0.75 #
GL B
Figure 5: BNWF Model Kz = 0.73 + 1.54 (10)
1−ν L

The force-displacement behavior of Q-z springs is

defined by QzSimple2 material in OpenSees library.
  
G 0.75 L
This material has asymmetric hysteretic behavior with Kθ y = Iy 3 (11)
1−ν B

500
 Proceedings of 12th IOE Graduate Conference

Horizontal stiffness according to Gazetas is given in searching [16]. From the search results, the ground
equation 12. motions which have similarity in shape of response
spectrum with that of target response spectrum and
scale factor ranging between 0.33 to 3 were selected
"  0.85 #
GL B 0.1 GL

B
 and are shown in table 2.
Kx = 2 + 2.25 + 1−
2−ν L 0.75 − ν L
Table 2: List of Earthquake Data
(12)
Year Location Predominant Period (s)
1970 Imperial Valley 0.50
The ultimate bearing capacity, ultimate passive
1989 Loma Prieta 1.06
resistance and ultimate sliding resistance are required
1995 Kobe 1.12
for nonlinear modeling of springs. The ultimate
1999 Kocaeli 0.42
bearing capacity is determined from Meyerhof’s
1999 Chi Chi 0.62
equations.
2010 Darfield 0.44
The other necessary parameters in defining the spring 2015 Gorkha 1.38
model in OpenSees can be found in research paper by
[8, 12, 14].
The response spectrum of the ground motions in table
2.5 Loading 2 are shown in figure 6.

Gravity loading and earthquake loading has been

considered in this study. For gravity loading, dead
load and live load are considered. Details about
earthquake loading are given in section 3.
For the dead loads, self weight of slab, beam, column
and 230mm brick wall are taken. 1 kN/m2 floor finish
is taken on the slabs. Live load of 2 kN/m2 is taken for
all cases out of which 30% from inner storey is taken
for gravity load.
Unit weight of RCC is taken as 25 kN/m3 and that of Figure 6: Response spectrum of selected earthquakes
brick masonary is taken as 20 kN/m3 .

These ground motions are matched to the target

3. Earthquake Records
spectrum to be able to use in time history analysis.
Selection of earthquake time history plays significant Spectral matching is performed in Seismomatch 2022
role in the non-linear analysis of structures. NBC software. The response spectrum of ground motions
recommends to use minimum 3 ground motions for after matching are shown in figure 7.
2D analysis and a minimum of 7 ground motions
allows to take average of the structural responses. For
this study, 7 ground motions are selected. Selection
for ground motion is done in Pacific Earthquake
Engineering Research Center (PEER) database.
The selected ground motion should correspond to the
seismic hazard level of the site. It, in our case, is
defined by the design response spectrum provided in
NBC 105:2020 modified for Kathmandu valley with
soil type D, zone factor 0.35, importance factor 1.0 Figure 7: Response spectrum of earthquakes after
and overstrength factor 1.5. The magnitude range (Mw matching
6.5-8), source to site distance (0 km to 150 km) and
shear wave velocity (50 m/s to 200 m/s) are taken for

501
Effect of Soil Structure Interaction on Seismic Pounding in Adjacent RC Buildings

4. Non-linear Time History Analysis and SSI cases. The results are presented in table 3.

Non linear time history analysis of the system was Table 3: Fundamental Time Period
done in OpenSees. Three pair of buildings; 3v5, 3v7 Case Fixed SSI Period Ratio
and 5v7 were taken for adjacency case. As the 3 Storey 0.59s 0.66s 1.12
direction of seismic excitation also affect the seismic 5 Storey 0.68s 0.79s 1.16
response, the mirrored configurations 5v3, 7v3 and 7 Storey 0.74s 0.92s 1.24
7v5 were also analyzed as shown in figure 8. 2
different base conditions were taken for all the
building pairs; one as a fixed base case and another as The fundamental time period increased for SSI case in
soil structure interaction case. All these combinations all the buildings. There is 12% increase in 3 storey
of building pairs were analyzed for 7 different ground building whereas there is 24% increase in 7 storey
motions as mentioned in table 2. The seismic gap in building. The foundation soil and the structure can be
all these cases were taken to be 1 mm simulating the considered as springs which are connected in series
case of buildings constructed without any gap due to which the overall stiffness of the system
between them. Hence, total of 84 different cases were reduces.
taken for time history analysis.
5.2 Effect of SSI on Pounding Force
Pounding force was calculated based on equation 1.
The force in the top floor of lower storey buildings
averaged from 7 ground motions are presented in
table 4.

Table 4: Pounding Force in kN

Case Fixed SSI
3v5 2766 3708
3v7 3484 5647

Consideration of SSI increases the displacement

demand of the buildings due to the introduced
flexibility so, even more space is required for free
vibration of buildings. Due to this, the pounding force
will increase on SSI case compared to fixed base case.
As seen from table 4, the pounding force in 3v5 case
Figure 8: Building Configurations Taken for Study is increased by 34% and that on 3v7 case is increased
by 62% when SSI is considered.

The time history analysis results obtained are Based on the findings of the study, the equation 1 is
presented taking the maximum pounding force, storey proposed to approximate the pounding force that may
displacement and interstorey drift averaged between be considered in the design of buildings.
the 7 ground motions. Due to the space limitations, Force(kN) = 550 N β 2 (13)
results are presented only for the effect of pounding
Where,
on 3 storey building.
N = Number of storey of taller building
β = Period elongation ratio due to SSI of taller building
5. Results and Discussion Table 5: Comparison of Pounding Force (Fixed Case)
5.1 Effect of SSI on Fundamental Time Period Case β Predicted Obtained
Fundamental time period for all 3 buildings were 3v5 1 2750 2766
obtained from eigen value analysis for both fixed base 3v7 1 3850 3484

502
 Proceedings of 12th IOE Graduate Conference

Table 6: Comparison of Pounding Force (SSI Case)

Case β Predicted Obtained
3v5 1.16 3700 3708
3v7 1.24 5920 5647

5.3 Effect of SSI on Displacement

When SSI is considered in the analysis of single
building, the displacement of buildings in all cases are
increased. In case of 3 storey building, there is only
13% increase in top storey displacement and 43%
increase in inter storey drift value. Whereas in 5 Figure 10: Displacement of 3 Storey Building on
storey building, there is 48% increase in top storey Non-Pounding Side (3v5 case)
displacement and drift is increased by 62%. In the
case of 7 storey building, there is increase in
displacement by 100% and drift by 140%. It is
observed that the highest drift always occured in the
first storey for SSI case. This occurs because the fixed
base boundary condition is released [8].

5.4 Effect of SSI on Displacement due to

Pounding
The results of combined effects of pounding and SSI
on the displacements of adjacent buildings are
presented in this section. Graphs of displacement for
only 3 storey building are shown here for brevity.
Figure 11: Displacement of 3 Storey Building on
Figures 9 and 10 show the storey displacement for 3 Pounding Side (3v7 case)
storey building when pounded with 5 storey building in
both fixed base case and SSI case. There is reduction in
displacement on both sides for fixed base case whereas
there is increase in displacement on non-pounding side
for SSI case. The combined effect due to pounding
and SSI is 5.6% increase on PS and 23% increase on
NPS.

Figure 12: Displacement of 3 Storey Building on

Non-Pounding Side (3v7 case)

Similarly, figures 11 and 12 show the storey

displacement for 3 storey building when pounded
with 7 storey building in both fixed base case and SSI
Figure 9: Displacement of 3 Storey Building on case. The results are similar to 3v5 case but the
Pounding Side (3v5 case) reduction or increment is greater for this case. The

503
Effect of Soil Structure Interaction on Seismic Pounding in Adjacent RC Buildings

combined effect is 12% reduction in displacement on [2] Sadegh Naserkhaki, Farah NA Aziz, and Hassan
PS and 37% increment on NPS. The displacements Pourmohammad. Earthquake induced pounding
between adjacent buildings considering soil-structure
are generally reduced for shorter buildings and interaction. Earthquake Engineering and Engineering
increased for taller buildings due to pounding but the Vibration, 11(3):343–358, 2012.
displacements in all cases are increased due to SSI. [3] Stavros A Anagnostopoulos. Pounding of buildings
in series during earthquakes. Earthquake engineering
As for 5 storey building pounding with 3 storey & structural dynamics, 16(3):443–456, 1988.
building, the combined effect is 58% increase in [4] Stavros A Anagnostopoulos and Konstantinos V
displacement on PS and 71% increase on NPS. Spiliopoulos. An investigation of earthquake induced
building. For pounding with 7 storey building, pounding between adjacent buildings. Earthquake
engineering & structural dynamics, pages 289–302,
combined effect is 36% increase in displacement on 1992.
PS and 60% increase in NPS.
[5] Robert Jankowski. Earthquake-induced pounding
For pounding between 7 and 3 storey buildings, the between equal height buildings with substantially
different dynamic properties. Engineering Structures,
combined effect is 102% increase in displacement on 30(10):2818–2829, 2008.
PS and 104% increase on NPS. Combined effect of [6] George Mylonakis and George Gazetas. Seismic
85% increase in displacement on PS and 108% on NPS soil-structure interaction: beneficial or detrimental?
is observed for pounding with 5 storey building. Journal of earthquake engineering, pages 277–301,
2000.
[7] Amar M Rahman, Athol J Carr, and Peter J
Moss. Structural pounding of adjacent multi-
6. Conclusion storey structures considering soil flexibility effects.
In Proceedings of the 12th World Conference on
In this research, a comprehensive study was done on Earthquake Engineering, Auckland, New Zealand,
the effects of soil structure interaction on the pounding volume 30, 2000.
response of RC buildings. Different cases of adjacency [8] B Madani, F Behnamfar, and H Riahi. Dynamic
with 3v5, 3v7, 5v7 with no gap between them and response of structures subjected to pounding and
structure–soil–structure interaction. Soil Dynamics
resting on rigid base as well as flexible base were and Earthquake Engineering, 78:46–60, 2015.
investigated. 7 ground motions were applied for each [9] Frank McKenna, G Fenves, F Filippou, S Mazzoni,
building pair for each base condition and maximum M Scott, A Elgamal, Z Yang, J Lu, P Arduino, and
averaged responses were taken. P McKenzie. Opensees. University of California,
Berkeley: nd, 2010.
Consideration of soil structure interaction increased [10] John B Mander, Michael JN Priestley, and R Park.
the fundamental time period, storey displacement and Theoretical stress-strain model for confined concrete.
interstorey drifts. SSI also increased the pounding Journal of structural engineering, 114(8):1804–1826,
1988.
force in all the floor levels. Considering only
[11] Panayiotis C Polycarpou, Loizos Papaloizou, and
pounding caused reduction in seismic responses for Petros Komodromos. An efficient methodology
shorter building whereas increased them for taller for simulating earthquake-induced 3d pounding of
building. buildings. Earthquake engineering & structural
dynamics, 43(7):985–1003, 2014.
Considering the increased displacement, the taller [12] Prishati Raychowdhury. Nonlinear Winkler-
buildings seem to be more vulnerable to damage due based shallow foundation model for performance
to the combined effect of pounding and soil-structure assessment of seismically loaded structures.
University of California, San Diego, 2008.
interaction when compared to shorter buildings.
[13] Ross W Boulanger, Christina J Curras, Bruce L
Finally, based on the findings, an empirical relation to Kutter, Daniel W Wilson, and Abbas Abghari.
approximate the pounding force that may be Seismic soil-pile-structure interaction experiments
and analyses. Journal of geotechnical and
considered in the design of buildings that are geoenvironmental engineering, pages 750–759, 1999.
constructed with no gap in between them is proposed. [14] Chad Harden. Numerical modeling of the nonlinear
cyclic response of shallow foundations. Pacific
Earthquake Engineering Research Center, 2005.
References [15] George Gazetas. Formulas and charts for impedances
of surface and embedded foundations. Journal of
[1] Kazuhiko Kasai and Bruce F Maison. Building geotechnical engineering, 117(9):1363–1381, 1991.
pounding damage during the 1989 loma prieta [16] Kebin Jung Thapa and Prem Nath Maskey. Seismic
earthquake. Engineering structures, 19(3):195–207, Separation Requirement to Reduce Pounding. 2021.
1997.

504