geosciences

Article
Effective Formula for Impact Damping Ratio for
Simulation of Earthquake-induced
Structural Pounding
Seyed Mohammad Khatami 1 , Hosein Naderpour 2 , Rui Carneiro Barros 3 ,
Anna Jakubczyk-Gałczyńska 4 and Robert Jankowski 4, *
1 Center of Semnan Municipality, University of Applied Science and Technology, 35149 Semnan, Iran
2 Faculty of Civil Engineering, Semnan University, 3513119111 Semnan, Iran
3 Faculty of Engineering, University of Porto (FEUP), 4200-465 Porto, Portugal
4 Faculty of Civil and Environmental Engineering, Gdansk University of Technology, 80-233 Gdansk, Poland
* Correspondence: jankowr@pg.edu.pl




Received: 10 May 2019; Accepted: 6 August 2019; Published: 8 August 2019


Abstract: Structural pounding during earthquakes may cause substantial damage to colliding
structures. The phenomenon is numerically studied using different models of collisions. The aim of
the present paper is to propose an effective formula for the impact damping ratio, as a parameter
of the impact force model used to study different problems of structural pounding under seismic
excitations. Its accuracy has been verified by four various approaches. Firstly, for the case of collisions
between two structural elements, the dissipated energy during impact has been compared to the loss
of kinetic energy. In the second stage of verifications, the peak impact forces during single collision
have been analyzed. Then, the accuracy of different equations have been verified by comparing
the impact force time histories for the situation when a concrete ball is dropped on a rigid concrete
surface. Finally, pounding between two structures during earthquakes has been studied. The results
of the analysis focused on comparison between dissipated and kinetic energy show relatively low
errors between calculated and assumed values of the coefficient of restitution when the proposed
equation is used. In addition, the results of the comparison between experimentally and numerically
determined peak impact forces during single collision confirm the effectiveness of the approach. The
same conclusion has been obtained for the whole impact time history for collision between a ball and
a rigid surface. Finally, the results of the comparative analysis, conducted for pounding between
two structures during an earthquake, confirm the simulation accuracy when the proposed approach
is used. The above conclusions indicate that the proposed formula for impact damping ratio, as a
parameter of impact force model for simulation of earthquake-induced structural pounding, is very
effective and accurate in numerical simulations in the case of different scenarios.

Keywords: earthquakes; structural pounding; impact damping; dissipated energy; coefficient

of restitution

1. Introduction
It is obviously accepted that insufficient separation distance between buildings, or bridge segments,
provides many damages due to collisions during earthquakes, which is called structural pounding [1,2].
This phenomenon occurs when critical distance cannot cover relative movements under seismic
excitations, and large lateral displacement exceeds separation distance between adjacent structures or
structural elements [3–8].
In order to investigate earthquake-induced structural pounding, many researchers experimentally
tested collisions between models of structures and also conducted numerical analyses using

Geosciences 2019, 9, 347; doi:10.3390/geosciences9080347 www.mdpi.com/journal/geosciences

Geosciences 2019, 9, 347 2 of 15

different impact force models. In order to calculate impact force during ground motions, several
equations were suggested to describe the impact damping ratio so as to determine the energy
dissipation. Anagnostopoulos [9,10], Maison and Kasai [11], Jankowski [12,13], Mahmoud et al. [14,15],
Komodromos et al. [16,17], Ye et al. [18], Barros et al. [19], Naderpour et al. [20–24] as well as Bamer
and Markert [25] analyzed a number of proposed models and also demonstrated the effects of
earthquake-induced structural pounding. Moreover, Filiatrault et al. [26] presented results of shake
table tests of pounding between adjacent single-bay steel framed model structures and compared them
with predictions resulting from two existing pounding analysis programs. Masroor and Mosqueda [27]
performed a series of earthquake simulator experiments to assess performance limit states of seismically
isolated buildings under strong ground motions. Jankowski [28] presented the analysis of pounding
between superstructure segments of a highway elevated bridge with three-span continuous deck under
3D non-uniform earthquake excitation and showed that the structural response depends much on the
gap size between adjacent segments.
Moreover, Papadrakakis and Mouzakis [29] conducted the experimental study and analytical
analysis focused on collisions between adjacent buildings during earthquakes. After static and shaking
table dynamic tests, the results of experiments were compared with analytical results and good
agreement was obtained. An interesting approach was also demonstrated by Papadrakakis et al. in [30].
The authors presented an effective method of testing the impacts of two, or more, adjacent buildings
during earthquakes based on the Lagrange multiplier method. A solution scheme was proposed and
the correlation between numerical and analytical results was found to be satisfactory. Shakya and
Wijeyewickrema [31] presented the analysis of seismic pounding of reinforced concrete buildings of
various heights, including the aspects of soil-structure interaction. Two configurations of buildings
were considered and pounding between structures was modelled by impact Kelvin–Voigt elements.
In most of the cases, the incorporation of soil, taken into account in the analysis, resulted in a lower
normalized story shear. Kajita et al. [32] performed collision tests and simulation analyses using the
three-dimensional finite element method. The authors proved that the maximum impact force can
be roughly estimated from the simulation analyses. Miari et al. [33] carried out the in-depth analysis
of the results of other researchers and also noted the recommendations concerning appropriate gap
size between structures so as to prevent pounding. Cole et al. [34] analyzed damages observed in the
February 2011 Christchurch earthquake. They found that susceptibility to building collisions during
an earthquake can be identified with adequate accuracy by comparing specified configurations with
characteristics previously noted by researchers. It is also worth mentioning the recent publication
by Crozet et al. [35] who carried out the shaking table experimental tests which can be utilized as
reference tests to be compared with the results of numerical analyses.
A number of the above-mentioned analyses considered the numerical models which can accurately
simulate impact force in the specific situation taking into account the amount of dissipated energy
during collision. However, there might be significant differences in the values of impact force and
dissipated energy in the case of different pounding scenarios during earthquakes. Therefore, the aim
of the present paper is to propose an effective formula for the impact damping ratio which can cover
all situations and required parameters while representing the best accuracy. The effectiveness of the
formula is confirmed by four different approaches.

2. Existing Formulae for Impact Force

In most of the cases, a special impact element, consisting of a spring and dashpot, is used to
investigate collisions between two structures during earthquakes in order to simulate impacts and
calculate energy absorption (see [9–24] for details). The general formula to calculate impact force
during contact, Fimp (t), is normally presented as [12,36]:
.
Fimp (t) = ks δn (t) + cimp δ(t) (1)
Geosciences 2019, 9, 347 3 of 15

.
where Ks , Cimp , δ(t) and δ(t) are: Impact stiffness, impact damping, relative displacement and relative
velocity, respectively. The power of n is usually taken as equal to 1 (linear models) or 1.5 (non-linear
models). It should be added that the impact stiffness, Ks , should be obtained through the iterative
procedure so as to equalize the peak pounding force determined from the simulations with the
peak pounding force obtained from the experiment (see [12] for details). On the other hand, the
impact damping, cimp , used for simulation of earthquake-induced structural pounding can be directly
calculated as [9,12]:
p
cimp = 2ζimp ks Meq (for linear models)
q p
cimp = 2ζimp ks δ(t)Meq (for non-linear models) (2)
mi m j
Meq = mi +m j

where mi , m j are masses of colliding structures, or structural elements, and ζimp is the impact damping
ratio which is related to the coefficient of restitution, CR, and can be defined by a general formula as
in [37]: .
δrebound
CR = . (3)
δimp
. .
where δimp is the relative prior-impact velocity and δrebound is the relative post-impact rebound velocity
(CR = 1 means a fully elastic collision and CR = 0 deals with a fully plastic impact). It should be
underlined that the CR value may depend on a number of factors, such as: Masses and shapes of
colliding structures, structural material properties and contact surface geometry [37]. The value can
be determined experimentally in a simplified way by dropping a sphere on a massive plane plate of
the same material from a specified height and observing the rebound height (see [37] for details). In
spite of its simplicity, the method is yet accurate to obtain appropriate CR values so as to conduct the
impact analyses.
Different formulae for impact damping ratio, ζimp , based on CR, were used by a number of
researchers. Anagnostopoulos [9] proposed the impact damping ratio for the linear viscoelastic model
to be described as (valid for 0 < CR ≤ 1):

ln(CR)
ζimp = − q (4)
π2 + (ln(CR))2

Mahmoud and Jankowski [15] suggested formula for the impact damping ratio in the modified
linear viscoelastic model in the following form (valid for 0 < CR ≤ 1):

1 − CR2
ζimp = (5)
CR(CR(π − 2) + 2)

Assuming that the damping term is activated only during the approach period of collision in order to
simulate the process of energy dissipation which takes place mainly during that period.
On the other hand, Jankowski [13] considered the non-linear viscoelastic model of impact, in
which the impact damping ratio is calculated as (valid for 0 < CR ≤ 1):
√
9 5 1 − CR2
ζimp = (6)
2 CR(CR(9π − 16) + 16)

while also assuming that the damping term is activated only during the approach period of collision.
The same assumption was taken into account by Barros et al. [19] who suggested another equation
Geosciences 2019, 9, x FOR PEER REVIEW 4 of 15

while also assuming that the damping term is activated only during the approach period of collision.
The same assumption was taken into account by Barros et al. [19] who suggested another equation
Geosciences
based the 9,nonlinear
on 2019, 347 viscoelastic model and justified the impact damping ratio using 4 the
of 15
coefficient of restitution expressed by (valid for 0  CR  1 ):
based on the nonlinear viscoelastic model 2
 and justified the impact damping ratio using the coefficient
of restitution expressed by (valid for 0 < CR ≤ 1): 
 (1  CR2 ) 
 imp    2
(7)
  CR 1  
 CR    (1 − CR
2 
 )  CR  
ζimp =    √   2  
 
 CR π CR + 1 − CR 
  (7)
2 π
Comparison between different impact force models using Equations (4–7) for one chosen CR
value Comparison between
equal to 0.5 is shown different
in Figureimpact
1. Theforce figure models
clearlyusing Equations
indicates (4)–(7) for
that different one chosen
equations showCR
value equal to 0.5 is shown in Figure 1.
visible different results for the same CR value. The figure clearly indicates that different equations show
visible different results for the same CR value.

Anagnostopoulos (1988)
Jankowski (2006)
Mahmoud and Jankowski (2011)
Barros et al (2013)
Impact force

Impact force
Anagnostopoulos (1988)
Jankowski (2006)
Mahmoud and Jankowski (2011)
Barros et al (2013)

Time (s) Displacement

Figure 1. Comparison between different impact force models using Equations (4)–(7).
Figure 1. Comparison between different impact force models using Equations (4–7).
3. Effective Formula for Impact Damping Ratio
3. Effective Formula for
All previously Impactimpact
proposed Damping Ratio
force models, applied to study structural pounding during
earthquakes, defineproposed
All previously the impact damping
impact forceratio basedapplied
models, on the coefficient of restitution.
to study structural However,
pounding the
during
results obtained for different models indicate that the use of them leads to visible
earthquakes, define the impact damping ratio based on the coefficient of restitution. However, the different results
for theobtained
results same CRfor value. Therefore,
different modelsthere is a need
indicate to justify
that the the equation
use of them leads tofor impact
visible damping
different ratio
results forin
order
the sameto determine
CR value. accurately
Therefore, all the is
there responses
a need to during
justifyearthquakes.
the equationFor forthe purpose
impact of defining
damping ratio the
in
order to determine accurately all the responses during earthquakes. For the purpose of defining theof
parameter in a more general form, an iterative process has been designed using CRVK (coefficient
restitution—velocity—stiffness)
parameter in a more general form, optimization
an iterativeprogram
process has(seebeen
[22] for details)using
designed so as CRVK
to reach a formula of
(coefficient for
solving the simulation of impactoptimization
restitution—velocity—stiffness) and calculateprogram
the impact(seedamping ratio. Insothis
[22] for details) as way, a new
to reach formula
a formula
has been derived based on the CR value. The iterative process has been divided
for solving the simulation of impact and calculate the impact damping ratio. In this way, a new into two different
parts. Inhas
formula thebeen
first one, the impact
derived based ondamping
the CRratio
value.hasThe
been definedprocess
iterative as: has been divided into two
different parts. In the first one, the impact damping ratio has been defined as:
ζimp = µ(1 − CR) (8)
 imp   (1  CR) (8)
where µ is the parameter of the model.
 isthe
whereThen, theeffectiveness
parameter ofofthe CRmodel.
has been investigated by the following procedure:
Firstly,
Then, thevalues of CR and
effectiveness of µCR
arehas
selected and the process
been investigated starts
by the to solve procedure:
following the problem. The dissipated
energy
Firstly, values of CR and  are selected and the process starts toimpact
is calculated as being equal to the area of hysteresis loop during solve (A)
the and its value
problem. Theis
comparedenergy
dissipated to the loss of kineticas
is calculated energy
being(E) determined
equal to the areaasof
[37]:
hysteresis loop during impact (A) and its
value is compared to the loss of kinetic energy
1 (E) determined . 2 as [37]:
E = Meq (1 − CR2 )δimp (9)
12
E  M eq (1  CR 2 )imp
2
(9)
2
Then, it is investigated whether the energy dissipation and the energy absorption are equal. If
they are, µ is accepted and the output is obtained; otherwise, another value for µ is selected and the
procedure is repeated.
 Then, it is investigated whether the energy dissipation and the energy absorption are equal. If
they are,  is accepted and the output is obtained; otherwise, another value for  is selected and the
procedure is repeated.
Based on the results of the first part of the iterative process, the following expression has been
obtained:
Geosciences 2019, 9, 347 5 of 15

1
   1.05CR0.653   (10)
Based on the results of the first partCRof the iterative process, the following expression has
been obtained:
In the second part of the iterative process, Equation (10) has been rectified so as to obtain the
1
µ =
same responses by using different coefficients → α
of = 1.05CR0.653
restitution. Another parameter has been added(10)to
CRα
investigate in details the impact damping ratio for different scenarios. In this way, the equation has
In the second part of the iterative process, Equation (10) has been rectified so as to obtain the
been improved into the following form:
same responses by using different coefficients of restitution. Another parameter has been added to
investigate in details the impact damping ratio for(1  CR) scenarios. In this way, the equation has
different
 imp   (11)
been improved into the following form: CR  

where  is a coefficient which, based on the results

(1 − CRof
) iterative process, has been determined as
ζimp = α
(11)
equal to: CR + β

where β is a coefficient which, based on the results3.351CR of iterative process, has been determined as equal to:
  (12)
CR0.204
3.351CR
β= 0.204
π (12)
So, the impact damping ratio can be defined CRby the following expression, as a proposed formula
for the impact damping ratio used to study earthquake-induced structural pounding (valid for
So, the impact damping ratio can be defined by the following expression, as a proposed formula for
0  CR  1 ):
the impact damping ratio used to study earthquake-induced structural pounding (valid for 0 < CR ≤ 1):
(1  CR)
 imp  ( ( 1 − CR)
0.204)
CR0.204 (13)
ζimp = CR  3.351CR CR0.204 (13)
CR(α+0.204) + 3.351CRπ
The relation defined in Equation (13) is presented graphically in Figure 2. The detailed scheme
Theabove
of the relation definediterative
described in Equation (13) is presented
procedure graphically
used to determine the in Figure 2.formula
proposed The detailed
for thescheme
impact
ofdamping
the above described iterative procedure
ratio is also shown in Figure 3. used to determine the proposed formula for the impact
damping ratio is also shown in Figure 3.

1.4

1.2

1
Impact damping ratio

0.8

0.6

0.4

0.2

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Coefficient of restitution (CR)

Figure2.2. Relation
Figure Relation between
between impact
impact damping
damping ratio
ratioand
andcoefficient
coefficientofofrestitution
restitution(CR)
(CR)according
accordingtoto
Equation (13).
Equation (13).
Geosciences 2019, 9, 347 6 of 15
Geosciences 2019, 9, x FOR PEER REVIEW 6 of 15

Figure 3. The
Figure 3. The detailed
detailed scheme
scheme of
of iterative
iterative procedure to determine
procedure to determine the
the impact
impact damping
damping ratio.
ratio.

4. Verification
Verification of
of Effectiveness
Effectiveness
In here, by focusing on Equation (13), in order to investigate the effectiveness of the model
proposed, four different approaches
approaches have been used. used. In all the cases, the CRVK program has been
applied andanddeveloped
developedtoto perform dynamic
perform dynamic analyses and solve
analyses impact
and solve problems
impact (see [22]
problems for [22]
(see details).
for
Firstly,
details).the dissipated
Firstly, energy and
the dissipated the absorbed
energy energy during
and the absorbed energyimpact
during have beenhave
impact compared with each
been compared
other for the
with each casefor
other of collisions
the case of between twobetween
collisions structuraltwoelements. In the
structural secondIn
elements. part
theofsecond
investigation,
part of
the experimentally
investigation, obtained peak impact
the experimentally obtainedforces
peakduring
impactsingle collision
forces during have beencollision
single compared withbeen
have the
numerical results. Then, the accuracy of different equations has been verified by comparing
compared with the numerical results. Then, the accuracy of different equations has been verified by the impact
force time histories
comparing for the
the impact situation
force when afor
time histories concrete ball is dropped
the situation on a rigidball
when a concrete concrete surface.
is dropped on Finally,
a rigid
the resultssurface.
concrete of the investigation
Finally, thefocused
resultson ofpounding between two
the investigation structures
focused during earthquake
on pounding between havetwo
been investigated.
structures during earthquake have been investigated.

4.1. Energy Dissipation

4.1. Energy Dissipation during
during Impact
Impact
Firstly,
Firstly, in
in order
order to
to determine
determine thethe impact
impact force
force and
and energy
energy dissipation,
dissipation, impact
impact between
between two
two
structural
structural elements
elements has
has been
been simulated
simulated andand the
the hysteresis
hysteresis loop
loop has
has been
been depicted.
depicted. The
The dissipated
dissipated
energy
energy during
during impact
impact has
has been
been calculated
calculated asas being
being equal
equal to
to the
the area
area of
of hysteresis
hysteresis loop
loop and
and its
its value
value
has
has been compared to the loss of kinetic energy determined by Equation (9). As an example, the
been compared to the loss of kinetic energy determined by Equation (9). As an example, the
representative results for
representative. results fortwo
twostructural
structuralelements
elementswith
withmass
mass ofof 170
170 kgkgandand
320320
kg kg impacting
impacting withwith
CR
CR = 0.4 and δ
= 0.4 and  imp imp = 1 m/s are presented below. The impact damping ratio has been calculated
= 1 m/s are presented below. The impact damping ratio has been calculated using using
Equation (13). The relation between impact force and displacement during impact, for the example
Equation (13). The relation between impact force and displacement during impact, for the example
considered, is shown in Figure 4.
considered, is shown in Figure 4.
 Geosciences 2019, 9, x FOR PEER REVIEW 7 of 15
Geosciences
Geosciences 2019,
2019, 9, x FOR PEER REVIEW
9, 347 7 of
7 of 15 15

7000
7000
6000
6000
5000

(kN)(kN)
5000
4000
force
4000
force
Impact

3000
Impact

3000
2000
2000
1000
1000
0
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
0
-1 -0.8 -0.6 -0.4 Displacement
-0.2 0 (cm)
0.2 0.4 0.6 0.8 1
Displacement (cm)
Figure 4. Relation between impact force and displacement during impact between two structural
Figure Relationbetween
Figure4.4. Relation
elements. between impact
impact force
force and and displacement
displacement duringduring
impactimpact
betweenbetween two
two structural
structural elements.
elements.
In this case, the enclosed area of the hysteresis loop has been determined as equal to 47.63 Nm,
InIn
while this
thethiscase,
loss ofthe
case, theenclosed
kinetic area
enclosed ofof
energy
area the
has
thehysteresis
been loop
calculated
hysteresis loop hasasbeen
hasequal
been determined asas
to 46.63 Nm.
determined equal
The
equal toto
47.63
calculatedNm,
47.63 CR
Nm,
while
valuethe
whilehas loss
thebeen of the
loss found kinetic energy
of thetokinetic
be 0.3764.has been calculated
Therefore,
energy has been as
thecalculated equal to
error of calculated 46.63
as equal CR Nm. The calculated
with relation
to 46.63 CR
Nm. Thetocalculated value
the assumed CR
has
CR been
value
value found
has isbeen tofound
as low beas0.3764.
5.9% Therefore,
to beconfirming
0.3764. theaccuracy
the error
Therefore, theoferror
calculated
of the CR with
of approach.
calculated CRrelation to the to
with relation assumed CR
the assumed
value is as calculated
The
CR value low
is asaslow5.9% confirming
values
as 5.9%of CR, the
for accuracy
confirming different of theof
cases
the accuracy approach.
of assumed (selected) CRs, are also presented in
the approach.
The
FigureThe calculated
5 for the formula
calculated values of CR,
proposed
values for different
as well
of CR, for cases
as forcases
different of assumed
otherofimpact
assumed (selected)
force models.CRs,
(selected) are also
The errors
CRs, are also presented
between bothin
presented
inCR
Figure
values
Figure 5 for
5 forare the formula
theadditionally proposed
summarized
formula proposed as well as for
as wellinasTable other impact
1. Figure
for other impact force
5 force
and Table models.
1 show
models. The errors
that the
The errors between
proposed
between both
both CR
method values
CR valuesresults are
are in additionally summarized
the lowest summarized
additionally errors obtained in
in forTable
Table 1.
all 1. Figure
values
Figure 5 and
of CR,
5 and Table
as Table 1
comparedshow that
to other
1 show the proposed
impact
that the force
proposed
method
models
method results
considered
results ininthethe
inlowest
the errors
analysis.
lowest obtained
errors obtained forfor
allall
values
values ofof
CR, asas
CR, compared
compared to to
other
otherimpact
impactforce
force
models considered in the analysis.
models considered in the analysis.

1
1 Anagnostopolos (1988)
0.9 Jankowski (2006) (1988)
Anagnostopolos
0.9 Mahmoud
Jankowskiand Jankowski (2011)
(2006)
0.8
Barros et al (2013)
Mahmoud and Jankowski (2011)
0.8 Proposed
0.7 Barros etmodel
al (2013)
Reference
Proposedline
model
0.7
0.6
Calculated CR

Reference line
0.6
Calculated CR

0.5
0.5
0.4
0.4
0.3
0.3
0.2
0.2
0.1
0.1
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0 Selected CR
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Selected CR
Figure 5. Relation between selected and calculated CR values for different impact force models.
Figure 5. Relation between selected and calculated CR values for different impact force models.
Figure 5. Relation between selected and calculated CR values for different impact force models.
Table 1. Errors between selected and calculated CR values for different impact force models.
Table 1. Errors between selected and calculated CR values for different impact force models.
Error (%)
Model
CR=0.1 CR=0.2 CR=0.3 CR=0.4 CR=0.5 CR=0.6 CR=0.7 CR=0.8 CR=0.9 CR=1
Error (%)
Model
CR=0.1 CR=0.2 CR=0.3 CR=0.4 CR=0.5 CR=0.6 CR=0.7 CR=0.8 CR=0.9 CR=1
Geosciences 2019, 9, 347 8 of 15

2019, 9,1.x Errors

Geosciences Table between
FOR PEER selected and calculated CR values for different impact force models.
REVIEW 8 of 15
Error (%)
Model
Anagnostopoulos [9] 53.4 44.5 32.5 23.4 14.5 7.2 4.5 2.1 1.3 0.77
CR = 0.1 CR = 0.2 CR = 0.3 CR = 0.4 CR = 0.5 CR = 0.6 CR = 0.7 CR = 0.8 CR = 0.9 CR = 1
Jankowski [13] 14.2 11.3 8.4 6.3 4.7 3.1 1.9 1.1 0.75 0.51
Anagnostopoulos [9] 53.4 44.5 32.5 23.4 14.5 7.2 4.5 2.1 1.3 0.77
Mahmoud and Jankowski
Jankowski [13] 14.2 40.4 11.3 33.3 8.4 24.5 6.3 16.8 4.7 9.5 3.1 4.1 1.9 2.7 1.1 1.8 0.75 0.94 0.51 0.63
[15]
Mahmoud and
Barros et al. [19]
Jankowski [15]
40.4 19.8 33.3 15.5 24.5 10.7 16.8 8.5 9.5 6.4 4.1 3.8 2.7 2.3 1.8 1.5 0.94 0.82 0.63 0.54
Proposed model 7.4 6.9 6.4 5.9 4.2 2.9 1.8 1.1 0.72 0.42
Barros et al. [19] 19.8 15.5 10.7 8.5 6.4 3.8 2.3 1.5 0.82 0.54
Proposed model 7.4 6.9 6.4 5.9 4.2 2.9 1.8 1.1 0.72 0.42
4.2. Peak Impact Force
4.2. Peak
In the Impact
secondForce
stage of verifications, the experimentally obtained and numerically determined
peak In impact forces during
the second stage ofsingle collisionthe
verifications, have been compared.
experimentally The results
obtained of the experimental
and numerically determined test
conducted by Kajita et al. [32] have been considered in the analysis. The
peak impact forces during single collision have been compared. The results of the experimental test experiment (see [32] for
details)
conducted wasbyfocused
Kajita etonal.impact between
[32] have two 300 kginsteel
been considered the frames
analysis. with
Theaexperiment
40 × 40 cm beam separated
(see [32] by
for details)
awas
10focused
cm gapon(see Figure 6). The collision test was implemented using
impact between two 300 kg steel frames with a 40 × 40 cm beam separated by a 10 cma horizontal hydraulic
high-speed loading
gap (see Figure 6). The machine
collision with
testawas
loading capacity using
implemented of 1000 kN and ahydraulic
a horizontal maximumhigh-speed
loading speedloadingof
3.0 m/s. The
machine withtest was conducted
a loading capacity of in 1000
a linekN onanda guide rail with
a maximum the length
loading speedofof3000 mmThe
3.0 m/s. allowing
test wasto
obtain the impact velocity of 0.68 m/s [32]. When the specimen was moving, the
conducted in a line on a guide rail with the length of 3000 mm allowing to obtain the impact velocity of sliding friction force
was slow,
0.68 m/s as compared
[32]. to the movement
When the specimen was moving, of thethe
steel bars.friction
sliding Therefore,
forcethewassliding
slow, as friction
compared forcetowas
the
reduced by using the compressed air and the energy loss caused by the
movement of the steel bars. Therefore, the sliding friction force was reduced by using the compressedsliding friction force was
suppressed duringloss
air and the energy collision.
caused by the sliding friction force was suppressed during collision.
It
It should be added that
should be added that Kajita
Kajita etet al.
al. [32] performed also
[32] performed the numerical
also the numerical analysis
analysis using
using thethe
three-dimensional finite element method. They underlined that there
three-dimensional finite element method. They underlined that there are three characteristic measures are three characteristic
measures
(impact force,(impact
law force, law of conservation
of conservation of momentum of momentum
and kineticand kinetic
energy energy
loss) loss) to be in
to be evaluated evaluated
the case
in
of the case of
collision collisioninvolving
problems problemstwo involving
bodies.two Thebodies.
authorsThe authorsthe
estimated estimated
maximum the impact
maximum forceimpact
using
force usingof
the results thetheresults of theanalysis.
numerical numerical Itsanalysis.
value was Itsfound
valueto was
be found to be approximately
approximately the same as the the value
same
as the value
obtained fromobtained from thetest.
the experimental experimental
They concluded,test. They concluded,
therefore, that thetherefore,
maximumthat theforce
impact maximum
can be
impact force can be roughly estimated from the simulation
roughly estimated from the simulation analyses using the approach applied. analyses using the approach applied.

Figure 6. Experimental setup for impact between two steel frames [32].
Figure 6. Experimental setup for impact between two steel frames [32].
The hysteresis loop obtained from the experiment by Kajita et al. [32] is shown in Figure 7. For
The hysteresis
the purposes loop obtained
of comparative fromthe
analysis, thehysteresis
experiment by has
loop Kajita et been
also al. [32] is shown in
numerically Figure 7. For
determined by
the
fitting the experimental results using the method of the least squares while applying Equation (13)by
purposes of comparative analysis, the hysteresis loop has also been numerically determined to
fitting thethe
calculate experimental resultsratio
impact damping using(see
thethe
method of in
results theFigure
least squares
7). It canwhile applying
be seen Equation
from Figure (13)the
7 that to
calculate the impact
experimentally damping
determined ratio
peak (see the
impact results
force in Figure
is equal to 30.757).kN,
It can be the
while seenvalue
fromobtained
Figure 7from
that the
the
experimentally determined peak impact force is equal to 30.75 kN, while the value obtained
numerical analysis is equal to 30.35 kN. The difference between both peak impact force values is as low from the
numerical analysis is
as 1.3% confirming theequal to 30.35
accuracy kN.approach
of the The difference
proposed.between both peak impact force values is as
low as 1.3% confirming the accuracy of the approach proposed.
Geosciences 2019, 9, x FOR PEER REVIEW 9 of 15

Geosciences2019,
Geosciences 2019,9,9,347
x FOR PEER REVIEW 99ofof1515
35
Experimental
Numerical
3035
Experimental
Numerical
2530

Impact force (kN)

2025

Impact force (kN)

1520

1015

510

05
0 1 2 3 4 5 6 7
Displacement (mm)
0
0 1 2 3 4 5 6 7
Figure 7. Experimentally and numerically Displacement (mm) relation between impact force and
determined
displacement during impact between two steel frames.
Figure Experimentally
Figure7. 7. and numerically
Experimentally determined
and numerically relation between
determined relation impact forceimpact
between and displacement
force and
during impact between
displacement two steel
during impact frames.
between two steel frames.
4.3. Impact Force Time History
4.3. Impact
thisForce
part,Time History of different equations is verified by comparing the impact force time
4.3. In
Impact Force the accuracy
Time History
histories forpart,
In this the situation whenofa ball
the accuracy is dropped
different ontoisa verified
equations rigid surface (see Figurethe
by comparing 8).impact
The results
forceoftime
the
In this part, the accuracy of different equations is verified by comparing the impact force time
experiment
histories for conducted bywhen
the situation Jankowski and
a ball is Mahmoud
dropped onto [36] have
a rigid been (see
surface usedFigure
in the8).
analysis. A number
The results of the
histories for the situation when a ball is dropped onto a rigid surface (see Figure 8). The results of the
of balls made
experiment of different
conducted materials and
by Jankowski andmasses,
Mahmoud m, were dropped
[36] have beenfrom
usedvarious height levels,
in the analysis. h, onto
A number of
experiment conducted by Jankowski and Mahmoud [36] have been used in the analysis. A number
a rigid
balls surface
made of the same
of different material
materials andand testedm,during
masses, the extensive
were dropped fromexperimental
various height investigation
levels, h, onto(see
a
of balls made of different materials and masses, m, were dropped from various height levels, h, onto
[36] for details). The results obtained for the concrete ball with mass of 1.763 kg impacting
rigid surface of the same material and tested during the extensive experimental investigation (see [36] the rigid
a rigid surface of the same material and tested during the extensive experimental investigation (see
concrete
for surface
details). with obtained
The results the velocity of concrete
for the 0.13 m/sball
have
withbeen
massused forkg
of 1.763 the purposes
impacting theofrigid
the concrete
present
[36] for details). The results obtained for the concrete ball with mass of 1.763 kg impacting the rigid
analysis.with the velocity of 0.13 m/s have been used for the purposes of the present analysis.
surface
concrete surface with the velocity of 0.13 m/s have been used for the purposes of the present
analysis.

Figure
Figure 8. Sketch of
8. Sketch of aa ball
ball falling
falling onto
onto aa rigid
rigid surface.
surface.

The
The numerical
numerical analysis has
has been
Figure
analysis 8. conducted
Sketch
been using
of a ball falling
conducted Equations
usingonto (5)–(7)
(5–7) and
a rigid surface.
Equations and (13). Parameters of
(13). Parameters of
different
different models used in the numerical analysis have been determined through iterative procedure
models used in the numerical analysis have been determined through iterative procedure
so The numerical analysis hasforce
been conductedfrom using Equations (5–7) and (13). Parameters of
so as
as to
to equalize
equalize the
the peak
peak impact
impact force determined
determined from the the simulations
simulations with
with the
the peak
peak impact
impact force
force
different models
obtained used in the (see
numerical analysis have been determined through iterative procedure
obtained from
from the
the experiment
experiment (see [36]
[36] for
for details). The results
details). The results of the
of the numerical
numerical simulations,
simulations, together
together
so asthe
with to experimental
equalize the peak impact
results, are force determined
shown in Figure 9. from the simulations with the peak impact force
with the experimental results, are shown in Figure 9.
obtained from the experiment (see [36] for details). The results of the numerical simulations, together
with the experimental results, are shown in Figure 9.
Geosciences 2019,
Geosciences 2019, 9,
9, 347
x FOR PEER REVIEW 10 of 15

1200

1000

800

Impact force (N)

600

400
Experimental
Jankowski (2006)
200 Mahmoud and Jankowski (2011)
Barros et al (2013)
Proposed formula
0
0 1 2 3 4 5 6 7 8 -4
Time (s) * 10

Figure
Figure 9. Impact force
9. Impact force time
time histories
histories for
for different
different equations.
equations.

The
The difference
difference between
between the
the experimental
experimental results
results and
and the
the results
results from
from the
the numerical
numerical analysis
analysis has
has
been assessed by calculating the normalized root mean square error (see
been assessed by calculating the normalized root mean square error (see [38]): [38]):
s
NV
NV
  HHii −HHii
P 22
ii=
11
EE= sNV ·100%
100% (14)
(14)

NV
P H 22
Hi i
ii=11

where H i and H i denote the values from the impact time history records obtained from the
where Hi and Hi denote the values from the impact time history records obtained from the experiment
experiment and from the
and from the numerical numerical
analysis, analysis,and
respectively respectively
NV is the and NV of
number is the number
values of values
in these history in these
records.
history records. The errors calculated using Equation (14) for different impact force models based
The errors calculated using Equation (14) for different impact force models based on Equations (5–7) on
Equations (5–7) and (13) are shown in Table 2. The results presented in the table confirm
and (13) are shown in Table 2. The results presented in the table confirm the simulation accuracy when the
simulation
the proposedaccuracy
formulawhen the proposed formula is used.
is used.

Table 2. Errors between experimental and numerical results for the impact force time history.

Model Error (%)

Jankowski [13] 13.9
Mahmoud and Jankowski [15] 13.8
Barros et al. [19] 14.9
Proposed formula 13.7

4.4. Pounding between Two Structures during Earthquake

In the last stage of verifications, the results of the numerical analysis have been compared with
the results of the shaking table experiment focused on pounding between two steel structures (tower
models—see Figure 10) with impacting concrete elements (see [36] for details). The columns of 1 m
high tested structures were arranged in a rectangular pattern with a spacing of 0.22 m along the shaking
direction (longitudinal one) and a spacing of 0.3 m along the orthogonal (transverse) direction (see
Figure 10). All supporting elements used in the left tower model had a rectangular cross section with
dimensions: 6 × 6 mm, whereas the model of the right tower was constructed of members with a
section of 8 × 8 mm. Elements made of different materials were fixed at the top of both towers, and,
with the help of additional masses in the form of plates and bolts (see [36]), the mass of each tower
model was kept constant during all experimental tests, apart from the material used. The top mass of
direction (see Figure 10). All supporting elements used in the left tower model had a rectangular
cross section with dimensions: 6 × 6 mm, whereas the model of the right tower was constructed of
members with a section of 8 × 8 mm. Elements made of different materials were fixed at the top of
both towers, and, with the help of additional masses in the form of plates and bolts (see [36]), the
Geosciences
mass of each2019, 9, 347 model was kept constant during all experimental tests, apart from the material
tower 11 of 15

used. The top mass of the left tower model, m1 , was equal to 9.485 kg, while the top mass of the right
tower,
the m2 , was
left tower 18.337
model, kg.equal
m1 , was The to
structures were the
9.485 kg, while fixed
topto theofshaking
mass the righttable
tower,platform with the
m2 , was 18.337 kg.
The structures were fixed to the shaking table platform with the in-between gap size of
in-between gap size of d = 40 mm and excited by the NS component of the El Centro earthquake d = 40 mm and
excited by the NS component of the El Centro earthquake (18.05.1940).
(18.05.1940).

Figure 10. Sketch of two steel structures (tower models) with impacting concrete elements during the
table experiment.
shaking table experiment.

The result of the experiment in the form of the pounding-involved displacement time history of
the left tower is shown
shown inin Figure
Figure 11
11 [36].
[36]. For the purposes of comparison, the numerical analysis has
been conducted
conducted using
using Equations
Equations(5)–(7)
(5–7) andand(13).
(13).Parameters
Parametersof ofdifferent
different models
models used in the analysis
have
have been determined through iterative procedure so as to equalize impact
been determined through iterative procedure so as to equalize the peak the peakforceimpact
determined
force
from the simulations
determined with the peak
from the simulations impact
with forceimpact
the peak obtained from
force the experiment
obtained (see [36] for (see
from the experiment details).
[36]
The results ofThe
for details). numerical analysis
results of are also
numerical shown
analysis areinalso
Figure
shown11. The normalized
in Figure 11. Theroot mean square
normalized root errors
mean
have
squarebeen calculated
errors using
have been Equationusing
calculated (14) for different(14)
Equation impact force models
for different based
impact on models
force Equations (5)–(7)
based on
and (13) and
Equations the and
(5–7) results
(13)are presented
and in Table
the results 3. It can beinseen
are presented from
Table 3. Itthe
cantable that the
be seen best
from theaccuracy of
table that
simulations is obtained by using the proposed formula.
the best accuracy of simulations is obtained by using the proposed formula.

Table 3. Errors between experimental and numerical results for pounding between two structures
during earthquake.

Model Error (%)

Jankowski [13] 11.5
Mahmoud and Jankowski [15] 13.9
Barros et al. [19] 17.6
Proposed formula 9.4
Geosciences 2019, 9, 347 12 of 15
Geosciences 2019, 9, x FOR PEER REVIEW 12 of 15

50 50
Experimental Jankowski (2006)
40 40

30 30

20 20
Displacement (mm)

Displacement (mm)
10 10

0 0

-10 -10

-20 -20

-30 -30

-40 -40

-50 -50
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
Time (s) Time (s)

(a) (b)
50 50
Mahmoud and Jankowski (2011) Barros et al (2013)
40 40

30 30

20 20
Displacement (mm)

10 Displacement (mm) 10

0 0

-10 -10

-20 -20

-30 -30

-40 -40

-50 -50
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
Time (s) Time (s)

(c) (d)
50
Proposed formula
40

30

20
Displacement (mm)

10

-10

-20

-30

-40

-50
0 5 10 15 20 25 30 35 40
Time (s)

(e)
Figure11.
Figure 11.Displacement
Displacementtimetime histories
histories of of
thethe
leftleft tower
tower forfor pounding
pounding between
between twotwo towers
towers under
under the
the El
El Centro
Centro earthquake
earthquake usingusing different
different approaches:
approaches: (a) experimental;
(a) experimental; (b) Jankowski
(b) Jankowski [13];
[13]; (c) (c) Mahmoud
Mahmoud and
and Jankowski
Jankowski [15];
[15]; (d) (d) Barros
Barros et al.and
et al. [19] [19](e)and (e) proposed
proposed formula.
formula.

5. Concluding Remarks
Table 3. Errors between experimental and numerical results for pounding between two structures
during
An earthquake.
effective formula for impact damping ratio, as a parameter of impact force model for simulation
of earthquake-induced structural pounding, has been proposedError
Model in this(%)
paper. It has been determined
based on the iterative process using Jankowski
a numerical program.
[13] The effectiveness
11.5 of the equation has
been verified by four different Mahmoud
approaches, i.e., by checking
and Jankowski [15] the energy
13.9 dissipation during impact, by
analyzing the peak impact forces during single
Barros et collision,
al. [19] by comparing17.6 the impact force time histories
for the case of collision between a ball and a rigid
Proposed formula surface and by studying pounding between two
9.4
structures during earthquake.
5. Concluding Remarks
Geosciences 2019, 9, 347 13 of 15

The results of the comparative analysis clearly indicate that using different, previously proposed,
impact force models leads to considerably different results for the same CR value. Therefore, the need
for an effective formula for the impact damping ratio in order to cover all the responses accurately has
been justified in the paper. The results of the analysis focused on the comparison between dissipated
and kinetic energy during impact show relatively low errors between calculated and assumed CR values
while using the proposed equation. In addition, the results of the comparison between experimentally
and numerically determined peak impact forces during single collision confirm the effectiveness of
the approach. The same conclusion has been obtained for the whole impact time history for collision
between a ball and a rigid surface. Finally, the results of the comparative analysis, conducted for
pounding between two structures during an earthquake, confirm the simulation accuracy when the
proposed approach is used.
The above conclusions indicate that the proposed formula for impact damping ratio, as one
of the most important parameters of impact force model, is accurate in numerical simulations in
the case of different scenarios. Therefore, it can be effectively used to study various problems of
earthquake-induced structural pounding.

Author Contributions: Methodology, S.M.K., H.N., R.C.B. and R.J.; software, S.M.K.; validation, S.M.K., H.N.,
R.C.B., A.J.-G. and R.J.; formal analysis S.M.K.; investigation, S.M.K., H.N., R.C.B. and R.J.; writing—original draft,
S.M.K.; writing—review and editing, A.J.-G. and R.J.
Funding: This research received no external funding.
Acknowledgments: The authors would like to thank Seyed Mohammad Nazem Razavi for his help in conducting
some numerical simulations.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Kasai, K.; Maison, B.F. Building pounding damage during the 1989 Loma Prieta earthquake. Eng. Struct.
1997, 19, 195–207. [CrossRef]
2. Hall, J.F.; Holmes, W.T.; Somers, P. Northridge Earthquake of Reconnaissance 17 January 1994—Report.
In EERI Report 95–03; Earthquake Engineering Research Institute: Oakland, CA, USA, 1995; Volume 1.
3. Jankowski, R.; Mahmoud, S. Linking of adjacent three-storey buildings for mitigation of structural pounding
during earthquakes. Bull. Earthq. Eng. 2016, 14, 3075–3097. [CrossRef]
4. Elwardany, H.; Seleemah, A.; Jankowski, R. Seismic pounding behavior of multi-story buildings in series
considering the effect of infill panels. Eng. Struct. 2017, 144, 139–150. [CrossRef]
5. Kun, C.; Yang, Z.; Chouw, N. Seismic response of skewed bridges including pounding effects. Earthq. Struct.
2018, 14, 467–476.
6. Li, C.; Bi, K.; Hao, H. Seismic performances of precast segmental column under bidirectional earthquake
motions: Shake table test and numerical evaluation. Eng. Struct. 2019, 187, 314–328. [CrossRef]
7. Raheem, S.E.A.; Fooly, M.Y.; Shafy, A.G.A.; Taha, A.M.; Abbas, Y.A.; Latif, M.M.A. Numerical simulation
of potential seismic pounding among adjacent buildings in series. Bull. Earthq. Eng. 2019, 17, 439–471.
[CrossRef]
8. Sasaki, T.; Sato, E.; Fukuyama, K.; Kajiwara, K. Enhancement of Base-Isolation Based on e-Defense Full-Scale
Shake Table Experiments: Dynamic Response of Base-Isolated Building under Impact due to Pounding.
In Proceedings of the 16th World Conference on Earthquake Engineering, Santiago, Chile, 9–13 January 2017;
p. 4082.
9. Anagnostopoulos, S.A. Pounding of buildings in series during earthquakes. Earthq. Eng. Struct. Dyn. 1988,
16, 443–456. [CrossRef]
10. Anagnostopoulos, S.A. Equivalent viscous damping for modeling inelastic impacts in earthquake pounding
problems. Earthq. Eng. Struct. Dyn. 2004, 33, 897–902. [CrossRef]
11. Maison, B.F.; Kasai, K. Dynamics of pounding when two buildings collide. Earthq. Eng. Struct. Dyn. 1992,
21, 771–786. [CrossRef]
12. Jankowski, R. Non-linear viscoelastic modelling of earthquake-induced structural pounding. Earthq. Eng.
Struct. Dyn. 2005, 34, 595–611. [CrossRef]
Geosciences 2019, 9, 347 14 of 15

13. Jankowski, R. Analytical expression between the impact damping ratio and the coefficient of restitution
in the non-linear viscoelastic model of structural pounding. Earthq. Eng. Struct. Dyn. 2006, 35, 517–524.
[CrossRef]
14. Mahmoud, S.; Chen, X.; Jankowski, R. Structural pounding models with Hertz spring and nonlinear damper.
J. Appl. Sci. 2008, 8, 1850–1858.
15. Mahmoud, S.; Jankowski, R. Modified linear viscoelastic model of earthquake-induced structural pounding.
Iranian J. Sci. Technol. 2011, 35, 51–62.
16. Komodromos, P.; Polycarpou, P.C.; Papaloizou, L.; Phocas, M.C. Response of seismically isolated buildings
considering poundings. Earthq. Eng. Struct. Dyn. 2007, 36, 1605–1622. [CrossRef]
17. Polycarpou, P.; Komodromos, P. On the Numerical Simulation of Impacts for the Investigation of
Earthquake-Induced Pounding of Buildings. In Proceedings of the Tenth International Conference on
Computational Structures Technology, Valenthia, Spain, 14–17 September 2010.
18. Ye, K.; Li, L.; Zhu, H. A note on the Hertz contact model with nonlinear damping for pounding simulation.
Earthq. Eng. Struct. Dyn. 2008, 38, 1135–1142. [CrossRef]
19. Barros, R.C.; Naderpour, H.; Khatami, S.M.; Mortezaei, A. Influence of seismic pounding on RC buildings
with and without base isolation system subject to near-fault ground motions. J. Rehabil. Civ. Eng. 2013, 1,
39–52.
20. Naderpour, H.; Khatami, S.M.; Barros, R.C.; Papadrakakis, M. Creation of A New Equation of Motion to
Calculate Dissipated Energy between Two Adjacent Buildings. In Proceedings of the 5th International
Conference on Computational Methods in Structural Dynamics and Earthquake Engineering Methods in
Structural Dynamics and Earthquake Engineering, Crete Island, Greece, 25–27 May 2015; pp. 2865–2875.
21. Naderpour, H.; Barros, R.C.; Khatami, S.M. A new model for calculating the impact force and the energy
dissipation based on CR-factor and impact velocity. Sci. Iran. 2015, 22, 59–68.
22. Naderpour, H.; Barros, R.C.; Khatami, S.M.; Jankowski, R. Numerical Study on Pounding between Two
Adjacent Buildings under Earthquake Excitation. Shock. Vib. 2016, 2016, 1–9. [CrossRef]
23. Naderpour, H.; Barros, R.C.; Khatami, S.M. Suggestion of an equation of motion to calculate damping ratio
during earthquake based on cyclic procedure. J. Theor. Appl. Mech. 2016, 54, 963–973. [CrossRef]
24. Naderpour, H.; Khatami, S.M.; Barros, R.C. Prediction of Critical Distance Between Two MDOF Systems
Subjected to Seismic Excitation in Terms of Artificial Neural Networks. Period. Polytech. Civ. Eng. 2017, 61,
516–529. [CrossRef]
25. Bamer, F.; Markert, B. A nonlinear visco-elastoplastic model for structural pounding. Earthq. Eng. Struct.
Dyn. 2018, 47, 2490–2495.
26. Filiatrault, A.; Wagner, P.; Cherry, S. Analytical prediction of experimental building pounding. Earthq. Eng.
Struct. Dyn. 1995, 24, 1131–1154. [CrossRef]
27. Masroor, A.; Mosqueda, G. Experimental simulation of base-isolated buildings pounding against moat wall
and effects on superstructure response. Earthq. Eng. Struct. Dyn. 2012, 41, 2093–2109. [CrossRef]
28. Jankowski, R. Pounding Between Superstructure Segments in Multi-Supported Elevated Bridge with
Three-Span Continuous Deck Under 3D Non-Uniform Earthquake Excitation. J. Earthq. Tsunami 2015, 9,
1550012. [CrossRef]
29. Papadrakakis, M.; Mouzakis, H.P. Earthquake simulator testing of pounding between adjacent buildings.
Earthq. Eng. Struct. Dyn. 1995, 24, 811–834. [CrossRef]
30. Papadrakakis, M.; Mouzakis, H.; Plevris, N.; Bitzarakis, S. A lagrange multiplier solution method for
pounding of buildings during earthquakes. Earthq. Eng. Struct. Dyn. 1991, 20, 981–998. [CrossRef]
31. Shakya, K.; Wijeyewickrema, A.C. Mid-column pounding of multi-story reinforced concrete buildings
considering soil effects. Adv. Struct. Eng. 2019, 12, 71–85. [CrossRef]
32. Kajita, Y.; Kitahara, T.; Nishimoto, Y.; Otsuka, H. Estimation of Maximum Impact Force on Natural Rubber
during Collision of Two Steel Bars. In Proceedings of the First European Conference on Earthquake
Engineering and Seismology (1st ECEES), Geneva, Switzerland, 3–8 September 2006; p. 488.
33. Miari, M.; Choong, K.K.; Jankowski, R. Seismic pounding between adjacent buildings: Identification of
parameters, soil interaction issues and mitigation measures. Soil Dyn. Earthq. Eng. 2019, 121, 135–150.
[CrossRef]
34. Cole, G.L.; Dhakal, R.P.; Turner, F.M. Building pounding damage observed in the 2011 Christchurch
earthquake. Earthq. Eng. Struct. Dyn. 2012, 41, 893–913. [CrossRef]
Geosciences 2019, 9, 347 15 of 15

35. Crozet, V.; Politopoulos, I.; Chaudat, T. Shake table tests of structures subject to pounding. Earthq. Eng.
Struct. Dyn. 2019, 48, 1156–1173. [CrossRef]
36. Jankowski, R.; Mahmoud, S. Earthquake-Induced Structural Pounding; Springer: Cham, Switzerland, 2015.
37. Goldsmith, W. Impact: The Theory and Physical Behavior of Colliding Solids; Edward Arnold: London, UK, 1960.
38. Bendat, J.S.; Piersol, A.G. Random Data: Analysis and Measurement Procedures; Wiley-Interscience: New York,
NY, USA, 1971.

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).