Seismic Pounding in Buildings
Seismic Pounding in Buildings
ABSTRACT: Investigations of past and recent earthquake damage have illustrated that the building
structures are vulnerable to severe damage and/or collapse during moderate to strong ground motion. Among
the possible structural damages, seismic induced pounding has been commonly observed in several
earthquakes. A parametric study on buildings pounding response as well as proper seismic hazard mitigation
practice for adjacent buildings is carried out. Three categories of recorded earthquake excitation are used for
input. The effect of impact is studied using linear and nonlinear contact force model for different separation
distances and compared with nominal model without pounding consideration. Pounding produces acceleration
and shear at various story levels that are greater than those obtained from the no pounding case, while the
peak drift depends on the input excitation characteristics. Also, increasing gap width is likely to be effective
when the separation is sufficiently wide practically to eliminate contact.
                                                           66
                                         Electronic Journal of Structural Engineering, 6 (2006)
zones in connection with the corresponding design            satisfies compatibility and boundary conditions. By
ground acceleration values will lead in many cases           assembling the element dynamic equilibrium
to earthquake actions which are remarkably higher            equation for the time t+Δt over all the elements, the
than defined by the design codes used up to now.             incremental FEM dynamic equilibrium equation can
   The most simplest and effective way for                   be obtained as:
pounding mitigation and reducing damage due to                  [M]{ u }t+Δt+[C]{ u }t+Δt+[K] t+Δt{Δu }t+Δt={F}t+Δt -
pounding is to provide enough separation but it is                                                                         (1)
                                                                {F}
sometimes difficult to be implemented due to
detailing problem and high cost of land. An                  where [M], [C] and [K]t+Δt = system mass, damping
alternative to the seismic separation gap provision in       and tangent stiffness matrices at time t+Δt. The
the structure design is to minimize the effect of            tangent stiffness considers the material nonlinearity
pounding through decreasing lateral motion (Kasai            through bilinear elastic-plastic constitutive model, u ,
et al. 1996, Abdullah et al. 2001, Jankowski et al           u   and Δu = accelerations, velocities, and
2000, Ruangrassamee & Kawashima 2003,                        incremental displacements at time t+Δt, respectively;
Kawashima & Shoji 2000), which can be achieved               and {F}t+Δt-{F}t = unbalanced force vector. The
by joining adjacent structures at critical locations so      Newmark’s step-by-step integration method is used
that their motion could be in-phase with one another         for the integration of the equation of motion. These
or by increasing the pounding buildings damping              equations for the building structure system subjected
capacity by means of passive structural control of           to earthquake ground motion input are assembled
energy dissipation system.                                   and numerically solved for the incremental
   The focus of this study is the development of an          displacement using the Newton-Raphson iteration
analytical model and methodology for the                     method. In this study, an equivalent viscous
formulation of the adjacent building-pounding                damping is explicitly introduced in the system in the
problem based on the classical impact theory, an             form of damping matrix [C]. A spectral damping
investigation through parametric study to identify           scheme of Rayleigh’s damping is used to form
the most important parameters is carried out. The            damping matrix as a combination of mass and
main objective and scope are to evaluate the effects         stiffness matrices, which effectively captures the
of structural pounding on the global response of             building damping and is also computationally
building structures; to determine proper seismic             efficient.
hazard mitigation practice for already existing
buildings as well as new buildings and to develop            2.2 Input ground motion
and provide engineers with practical analytical tools
for predicting pounding response and damage. A               A suite of nine-ground motion records from seven
realistic pounding model is used for studying the            different earthquakes [Muthukmar & DesRochs
response of structural system under the condition of         2004] is selected for the purpose of understanding
structural pounding during moderate to strong                the input ground motion effect, as listed in Table 1.
earthquakes. An analytical technique based on the            The ground motion records are grouped into three
contact force-based approach is developed, where             levels depending on the peak ground acceleration as,
the contact element is activated when the structures         low (0.1g up to 0.3g), moderate (0.4g up to 0.6g)
come into contact. A spring with high stiffness is           and high (0.7g up to 0.9g). The records are chosen
used to avoid overlapping between adjacent                   such that the period ratio (T1 /Tg and T2 /Tg; adjacent
structures. Two adjacent multi-story buildings are           buildings period over the ground motion
considered as a representative structure for potential       characteristic period) has a wide range.
pounding problem. A simplified nonlinear analytical
model is developed to study the response of multi-
story building subject to earthquake excitation.             3 FINITE ELEMENT MODELLING
                                                             3.1 Building model
2 NONLINEAR DYNAMIC ANALYSIS                                 This study investigates pounding of adjacent
  PROCEDURES                                                 building structures from an analytical perspective. A
                                                             simplified nonlinear model of a multi-story building
2.1 Equilibrium equation solution technique                  is developed incorporating the effects of geometric
The governing nonlinear dynamic equation of                  and material nonlinearities. A three-dimensional
motion for the structure response can be derived by          (3D) finite element model has been defined and 3D
the principle of energy that the external work is            non-linear time-history analyses have been
absorbed by the work of internal, inertial and               performed. A new formulation is proposed to model
damping forces for any small admissible motion that          pounding between two adjacent building structures,
                                                             with natural periods TA and TB and damping ratios ζA
                                                           67
                                                    Electronic Journal of Structural Engineering, 6 (2006)
             0.30      Whittier narrows, 1987             6.0   E-Grand Ave (2A-GRN)               180        9.0      23.0     3.3   0.70
             0.29      Morgan Hill, 1994                  6.2   Gilroy Array #6 (3G06)             090       11.8      36.7     6.1   1.20
             0.48      Loma Prieta, 1989                  6.9   Coyote Lake Dam (4CYC)             285       21.8      39.7    15.2   0.65
  Moderate
             0.51      Loma Prieta, 1989                  6.9   Saratoga-Aloha Ave (5STG)          000       11.7      41.2    16.2   1.80
             0.59      N. Palm Springs, 1986              6.0   5070 N-Palm Spring (6NPS)          210        8.2      73.3    11.5   1.10
             0.60      Coalinga, 1983                     5.8   Pleasant Valley P.P. (7D-PVY)      045       17.4      34.8     8.1   0.65
  High
             0.84      Northridge, 1994                   6.7   Rinaldi (8RRS)                     228        7.1     166.1    28.8   1.05
             1.04      Cape Mendocino, 1992               7.1   Cape Mendocino (9CPM)              090        8.5      42.0    12.4   2.00
                                                                 uB
                                                                                                                 Building B
                                Impact spring                                           Sepation S
                                  element
                                      uA
                                                                                         Building A
                               Potential pounding                                                                              HB
                                    location                    HB
                    HA                                                         HA
and ζB under earthquake excitation, as linear and                              center of rigidity, and the model has the same
nonlinear contact force based impact between two                               stiffness and mass distribution.
multi-degree-of-freedom oscillators. Steel moment
resistant frame building of 8-story (building A,
                                                                               3.2 Impact model
period = 0.72) is assumed to collide with and
adjacent 13-story (building B, Period = 1.22), as                              Pounding is simulated using contact force-based
shown in Figure 1. In this model, the building floor                           model such as linear and nonlinear springs. In
is assumed to be infinitely rigid in its own plane.                            addition, a nonlinear contact model accounting for
The entire mass of the structure is uniformly                                  impact energy dissipation is also introduced to
distributed at the floor level. The model has                                  model impact. A bilinear truss contact model with a
coincident CR (Rigidity/stiffness Center) and CM                               gap is considered for representing impact between
(Mass Center) that is located at the geometric center                          closely spaced adjacent structures, as shown in
of the floor. For the purpose of evaluating the effect                         Figure 1. The model parameters such as the stiffness
of torsion, a torsional unbalanced model is defined                            properties and the yield deformation of the truss
where the mass center lies at a distance e from the                            element are determined using the Hertz contact law
                                                                             68
                                          Electronic Journal of Structural Engineering, 6 (2006)
Table 2. Non-pounding and relative pounding displacements for different input earthquakes
Input Earthquake         uA (m)           uB (m)            uRel (m)            uRel / max. (uA & uB)                 uRel /(uA + uB)
1MVH                      0.06             0.10               0.13                         1.30                            0.81
2A-GRN                    0.24             0.45               0.65                         1.45                            0.94
3G06                      0.09             0.04               0.11                         1.22                            0.85
4CYC                      0.11             0.19               0.27                         1.42                            0.90
5STG                      0.09             0.17               0.19                         1.18                            0.73
6NPS                      0.15             0.14               0.24                         1.71                            0.83
7D-PVY                    0.08             0.13               0.21                         1.62                             1.0
8RRS                      0.13             0.06               0.14                         1.08                            0.74
9CPM                      0.09             0.16               0.19                         1.19                            0.76
for the effective stiffness and by equating the               relative displacement response (Lopez Garcia 2004,
element hysteresis area to the energy dissipated              Penzien 1997), is given by:
during impact (Muthukmar & DesRochs 2004,
Muthukmar 2003).                                                 S = u Re l (t ) = u A2 + u B2 − ρ AB u A u B                        (4)
                                                                 where uA, uB and uRel = mean peak values of uA(t),
4 REQUIRED SEISMIC SEPARATION                                 uB(t) and uRel(t), respectively. The correlation
  DISTANCE TO AVOID POUNDING                                  coefficient, ρAB depends on the period on the period
                                                              ratio r = TB /TA, as well as ζA and ζB, (Lopez Garcia
Seismic pounding occurs when the separation                   2004, Penzien 1997) and is given by
distance between adjacent buildings is not large                                     8 ζ A ζ B (ζ A + r ζ B ) r 1.5
                                                                 ρ AB =                                                              (5)
enough to accommodate the relative motion during
earthquake events. Seismic codes and regulations
                                                                          (1 − r )
                                                                               2 2
                                                                                     + 4r ζ A ζ B (1 + r 2 ) + 4(ζ 2A + ζ 2B ) r 2
worldwide specify minimum separation distances to                where TA, ζA and TB, ζB are natural periods and
be provided between adjacent buildings, to preclude           damping ratios of systems A and B, respectively.
pounding, which is obviously equal to the relative            The DDC rule is much more accurate than the ABS
displacement demand of the two potentially                    and SRSS rules, although it gives somewhat un-
colliding structural systems. For instance, according         conservative results when TA and TB are well
to the 2000 edition of the International building code        separated (Lopez Garcia 2004, Penzien 1997).
and in many seismic design codes and regulations
worldwide, minimum separation distances (Lopez
                                                              5 NUMERICAL RESULTS AND DISCUSSION
Garcia 2004) are given by ABSolute sum (ABS) or
Square Root of Sum of Squares (SRSS) as follow:               5.1 Pounding and spacing size effects
                                                            69
                                                    Electronic Journal of Structural Engineering, 6 (2006)
degree of biased response of the pounding system.                                  stiff building is reduced. Conversely, the stiff 8-
Therefore, seismic poundings between adjacent                                      story building demand increases and the flexible
buildings may induce unwanted damages even                                         building demand decreases due to pounding for the
though each individual structure might have been                                   2A-GRN input earthquake that has dominant period
designed properly to withstand the strike of credible                              near the fundamental period of stiff building.
earthquake events.                                                                 Pounding slightly decreases both building responses
   The acceleration variation at the top level of                                  for 8RRS input earthquake. The amplification in
shorter building during impact between adjacent                                    building response is a function of each of adjacent
structures under different earthquakes is computed                                 buildings vibration period and their ratio as well as
to study the behavior of the building during impact.                               the dominant frequency of input excitation.
Pounding is a severe load condition that could result
                                                                                      Furthermore, pounding can amplify the global
in high magnitude and short duration floor
                                                                                   response of participating structural systems. The
acceleration pulses in the form of short duration
                                                                                   effects of impact are found to be severe for both of
spikes, which in turn cause greater damage to
                                                                                   adjacent buildings. Pounding produces acceleration
building contents. A sudden stopping of
                                                                                   response and shear force at various story levels that
displacement at the pounding level results in large
                                                                                   are greater than those from the no pounding case, as
and quick acceleration pulses in the opposite
                                                                                   shown in Figure 4, while the peak drift depends on
direction. The acceleration increases due to impact
                                                                                   the input excitation characteristics. Flexible 13-story
with adjacent structure and can be more than 10
                                                                                   building pounding increases shear above impact
times those from no-pounding case, as illustrated in
                                                                                   level and below the third floor slab as well as
Figure 2. The time history of inward displacements
                                                                                   acceleration at the vicinity of impact, while stiff 8-
and their extreme values for the pounding and no
                                                                                   story building pounding almost increases the peak
pounding cases shows that pounding reduces the
                                                                                   shear over the entire height. The increase of spacing
building response when vibrating near the
                                                                                   from 0.12 to 0.25m has the capability for reducing
characteristic period of the ground motion and
                                                                                   impact effects and could reduce the number of
increases the adjacent building response, as shown
                                                                                   pounding's occasion. Also, increasing gap width is
in Figure 3. The flexible 13-story building vibrates
                                                                                   likely to be effective when the separation is
near the dominant frequency of the 3G06 input
                                                                                   sufficiently wide practically to eliminate contact.
earthquake; pounding response is increased in the
flexible building while pounding response of the
                              10                             No-pounding                  10                       No-pounding
        Acceleration (g)
                                                                                   8RRS
                               0                                                           0
-10 -10
-20 -20
                              10                              Pounding                    10                            Pounding
        Acceleration (g)
0 0
-10 -10
                              -20                                                         -20
                                 0            5             10                15            0          5           10                15
                                                                 Time (sec)                                             Time (sec)
                                   (a) 8-story building (8th level)                        (b) 13-story building (8th level)
                           Figure 2. Acceleration time histories at pounding level (Pounding problem versus no-pounding case)
                                                                               70
                                                                                           Electronic Journal of Structural Engineering, 6 (2006)
        Displacement (m)
                             0.5                       max. (2.7, 0.24)                                                         0.5
                                                       max. (2.8, 0.19)                                                 8RRS
0 0
0 0
                                   (a) 8-story building (8th level)                          (b) 13-story building (8th level)
                             Figure 3. Displacement time histories at pounding level (Pounding problem versus no-pounding case)
40 40
                           No-pounding                                                                                                         No-pounding
                           Pounding ( spacing S = 0.12 m)                                                                                      Pounding ( spacing S = 0.12 m)
                           Pounding (spacing S = 0.25 m)                                                                                       Pounding (spacing S = 0.25 m)                           30
                                                                                 30
20 20 20 20
10 10 10 10
                                                               4CYC                                                                                                                8RRS
                                               0                                  0                                                                              0                                      0
-10                         -5                     0                -2    -1           0     1       2         3                -20            -10                   0                 -5                   0             5
                             Acceleration (g)                                  Acceleration (g)                                                      Acceleration (g)                             Acceleration (g)
                                                                                            40                                                                                                                  40
                             No-pounding                                                                                                             No-pounding
                             Pounding ( spacing S = 0.12 m)                                                                                          Pounding ( spacing S = 0.12 m)
                             Pounding (spacing S = 0.25 m)                                  30
                                                                                                                                                     Pounding (spacing S = 0.25 m)                              30
20 20 20 20
                                                                                                                                                                                   8RRS
                                 10                                                         10                                                          10                                                      10
                                                                 8RRS
                                  0                                                          0                                                           0                                                       0
 -0.2           -0.1                  0       0.1        0.2                    -0.5             0       0.5                    -6        -4     -2          0            2    4                  -2                 0        2   4
                                  Displacement (m)                               Displacement (m)                                                      Force (KN)                                      Force (KN)
                                                       Figure 4. Response envelops for different spacing size between adjacent buildings
                                                                                                                    71
                                                     Electronic Journal of Structural Engineering, 6 (2006)
                                                                                     2a-grn
                            0                                                                  0
-20 -20
-40 -40
0 0
-20 -20
-40 -40
                             0                 5              10                15               0                            5                 10                      15
                                                                   Time (sec)                                                                        Time (sec)
                                 (a) 8-story building (8th level)                                     (b) 13-story building (8th level)
                                 Figure 6. Acceleration time history response for linear and nonlinear impact modeling
                                                                                 72
                                        Electronic Journal of Structural Engineering, 6 (2006)
6 CONCLUSIONS 7 REFERENCES
In this study, a mathematical modeling of adjacent          Abdel Raheem S. E., "Evaluation and prevention of seismic
building pounding has been demonstrated and its                pounding between adjacent building structures", Third
                                                               Egyptian Conference on Earthquake Engineering,
implementation in a finite element nonlinear seismic           EGYQUAKE 3, Cairo, Egypt, 6-8 December 2004, pp 253-
analysis is presented. Numerical investigation,                266.
aiming at accurate description and evaluation of            Abdullah, M. M., Hanif, J. H., Richardson, A. and Sobanjo, J.,
colliding adjacent structures real behavior and its            "Use of a shared tuned mass damper (STMD) to reduce
effects on global response has been conducted. It              vibration and pounding in adjacent structures", Earthquake
studies the relative importance of dynamic                     Engineering and Structural Dynamics, Vol. 30, 2001, pp
                                                               1185-1201.
characteristics of adjacent building structures in          Astaneh-Asl, A., Bolt, B., McMullin, K., Donikian, R. R.,
causing relative responses. The effect of vibration            Modjtahedi, D. and Cho, S. W., "Seismic Performance of
properties of adjacent structures is significant to            Steel Bridges During the 1994 Northridge Earthquake",
those of high-rise adjacent structures if they have            Report No. UCB/CEE-Steel-94/01, University of
noticeably different vibration periods.                        California, Berkeley, 1994.
                                                            Hao, H. and Zhang, S., "Spatial ground motion effect on
   Pounding is a highly nonlinear phenomenon and               relative displacement of adjacent building structures",
a severe load condition that could result in                   Earthquake Engineering and Structural Dynamics, Vol. 28,
significant structural damage, high magnitude and              1999, pp 333-349.
short duration floor acceleration pulses in the form        Hayashikawa T., Hirooka, T., Ikeda, K. and Abdel Raheem S.
of short duration spikes, which in turn cause greater          E., "Dynamic behavior of viaducts with PC cables in
                                                               consideration of pounding of girders and energy
damage to building contents. A sudden stopping of              absorption", Proceedings of Hokkaido Chapter of the
displacement at the pounding level results in large            Japan Society of Civil Engineers, JSCE, No. 59, 2002, pp
and quick acceleration pulses in the opposite                  50-53.
direction. Furthermore, pounding can amplify the            Jankowski, R., Wilde, K. and Fujino, Y., "Reduction of
global response of participating structural systems.           pounding effects in elevated bridges during earthquakes",
The vertical location of pounding significantly                Earthquake Engineering and Structural Dynamics, Vol. 29,
                                                               2000, pp. 195-212.
influences the distribution of story peak responses         Kasai, K. and Maison, B. F., "Observation of structural
through the building height. The acceleration                  pounding damage from 1989 Loma Prieta earthquake",
response at pounding level indicate that pounding is           Proceeding of 6th Canadian Conference of Earthquake
especially harmful for equipment or secondary                  Engineering, Toronto, Canada, 1991, pp 735-742.
systems having short periods, where the existing            Kasai, K., Jagiasi, A. R. and Jeng, V., "Inelastic vibration
                                                               phase theory for seismic pounding mitigation", Journal of
industrial design spectra does not cover this effect.          Structural Engineering, Vol. 122, No. 10, 1996, pp. 1136-
More importantly, pounding can amplify the                     1146.
building displacement demands beyond those                  Kasai, K., Maison, B. F., Jeng, V., Patel, D. J. and Patel, P.C.,
typically assumed in design. Existing design                   "A study of earthquake pounding between adjacent
procedure should account for dynamic impact.                   structures", Proceeding of 6th Canadian Conference of
                                                               Earthquake Engineering, Toronto, Canada, 1991, pp 93-
Adjacent building period ratio should be carefully             100.
selected to reduce the pounding effects.                    Kawashima, K. and Shoji, G., "Effect of restrainers to mitigate
   The results depend on the excitation                        pounding between adjacent decks subjected to a strong
characteristics and the relationship between the               ground motion", Proceeding of the 12th World Conference
buildings fundamental period. In addition, unwanted            on Earthquake Engineering, Auckland, New Zealand,
period shift of an existing structure imposed by the           2000, Paper No. 1435.
                                                            Lopez Garcia, D., "Separation between adjacent nonlinear
construction of a new building in its neighborhood             structures for prevention of seismic pounding", Proceeding
may lead to unprepared and unexpected damages of               of the 13th World Conference on Earthquake Engineering,
the former during earthquakes. Therefore, seismic              Vancouver, B.C., Canada, 2004, Paper No. 478.
poundings between adjacent buildings may induce             Muthukmar, S. and DesRochs, R., "Evaluation of impact
unwanted damages even though each individual                   models for seismic pounding", Proceeding of the 13th
                                                               World Conference on Earthquake Engineering, Vancouver,
structure might have been designed properly to                 B.C., Canada, 2004, Paper No. 235.
withstand the strike of credible earthquake events.         Muthukmar, S., "A contact element approach with hysteresis
Pounding produces acceleration and shear at various            damping for the analysis and design of pounding in
story levels that are greater than those from the no           bridges", Ph. D Thesis Presented to The Academic Faculty,
pounding case, while the peak drift depends on the             Civil and Environmental Engineering, Georgia Institute of
input excitation characteristics. An increasing gap            Technology, November 2003.
                                                            Northridge Reconnaissance Team, "Northridge Earthquake of
width is likely to be effective when the separation is         January 17, 1994, Reconnaissance Report", EERI,
sufficiently wide practically to eliminate contact.            Oakland, California, 1996, pp 25-47.
                                                            Pantelides, C. P. and Ma, X., "Linear and nonlinear pounding
                                                               of structural systems", Computers & Structures, Vol. 66,
                                                               No. 1, 1998, pp 79-92.
                                                          73
                                      Electronic Journal of Structural Engineering, 6 (2006)
74