INTERNATIONAL SOCIETY FOR

SOIL MECHANICS AND

GEOTECHNICAL ENGINEERING

This paper was downloaded from the Online Library of

the International Society for Soil Mechanics and
Geotechnical Engineering (ISSMGE). The library is
available here:

https://www.issmge.org/publications/online-library

This is an open-access database that archives thousands

of papers published under the Auspices of the ISSMGE and
maintained by the Innovation and Development
Committee of ISSMGE.
A comparative study on response spectrum
analysis and dynamic analysis for seismic
design of pile-supported wharf
Jung-Won Yun
Department of Geotechnical & Geo-Space Engineering, University of Science & Technology, Goyang,
Gyeonggi, Republic of Korea
Jin-Tae Han
Korea Institute of Civil Engineering & Building Technology, Goyang, Gyeonggi, Republic of Korea
Inn-Joon Park
Hanseo University, Seosan, Chungnam, Republic of Korea
Jae-Soon Choi
Seokeong University, Seoul, Republic of Korea

ABSTRACT
Generally, for the response spectrum analysis for the seismic design of a pile-supported wharf, it is suggested that the
virtual fixed point be utilized. The virtual fixed point is known to be useful for static-behavior evaluation, but it is not certain
that the results of the response spectrum analysis utilizing the virtual fixed point are reliable. In this study, the result of the
response spectrum analysis of a pile-supported wharf considering the virtual fixed point and of the dynamic analysis results
not considering the virtual fixed point were compared to improve the response spectrum design method. A commercial
software, MIDAS, was used for the analysis of the response spectrum, and a 3D FEM software, PLAXIS 3D was used for
the dynamic analysis. The results of the response spectrum analysis of a pile-supported wharf considering the virtual fixed
point and of the dynamic analysis of an actual structure differed depending on the type and depth of the seismic waves.
This is because the response spectrum analysis considering the virtual fixed point does not reflect the fact that the end
condition is different from the actual structure, and does not consider the ground under the virtual fixed point.

1. INTRODUCTION wharf. The interaction between the pile and the ground was
simulated by a spring, and the spring coefficient of the
Recently, due to the frequent earthquakes occurring ground was calculated via pushover analysis.
throughout the world, the need for a reasonable seismic Shafieezedeh et al. (2012) evaluated the seismic
design method for the harbor structure is on the rise. In the vulnerability using 3D FEM for the worn-out pile-supported
case of the existing port structures, the ASD (allowable wharf designed from the 1960s to the 1970s. Similarly,
stress design) method or the USD (ultimate strength Doran et al. (2015) also performed seismic-performance
design) method have been uniformly applied. Lately, evaluation using 3D FEM for a worn-out pile-supported
however, the performance-based seismic design method, wharf where frequent earthquakes had occurred.
which allows flexible design, has been applied. The In most of the previous related studies, the seismic
performance-based design method presents a number of performance of the pile-supported wharf was evaluated via
performance requirements and uses a variety of design the pushover and time history analyses of the 3D modeling
methods as a method of performing the rational design. It finite-element analysis. However, PIANC (2001) and
is gradually being applied to the pile-supported wharf, one Seismic Design Standard of Port and Harbor (1999)
of the typical port structures. In PIANC (2001) by the currently propose a simple response spectrum analysis
International Navigation Association, four performance method as well as dynamic analysis for the pile-supported
levels are presented, while PARI (2009) in Japan and wharf. There has been limited research worldwide,
ASCE and COPRI (2014) in the U.S. present three. In however, on the response spectrum analysis method for
Korea, only two performance levels are still being applied the seismic design of a pile-supported wharf.
to the design. Using the response spectrum analysis, the maximum
In the case of the pile-supported wharf, which is known response can be obtained by applying a response
for its excellent seismic-performance level, structures are spectrum load to the structure. In addition, the method
designed according to the above performance level, and determines the acceleration response value by connecting
the 3D FEM (finite-element method) numerical analysis the maximum response using the mode combination
technique is also applied for seismic-performance method. The mode combination method that is mainly used
evaluation as one of the seismic performance evaluation in the response spectrum analysis is the CQC (complete
methods. quadratic combination) method. Unlike the time history
Donahue et al. (2005) compared the results of the analysis method, it can be a relatively simple way to
numerical simulation of the SMD (strong-motion data) understand the dynamic characteristics of the structure.
seismic wave with actual measurement data to investigate Generally, it is proposed that a structure be modeled by
the interaction between the ground and the pile-supported applying the virtual fixed point in the response spectrum
analysis of a pile-supported wharf. The virtual fixed point is as shown in Figure 5(a). In the case of the dynamic
known to be a useful method of static-behavior analysis model, the entire length of the pile was modeled.
evaluation(PIANC, 2001). It is not certain, however, that To simplify the analysis model and ground properties, the
the results of the response spectrum analysis utilizing the ground was divided into the soil layer and the bedrock
virtual fixed point are reliable. layer. The properties used in the analysis are shown in
Therefore, the purpose of this study is to find problems Table 1.
by comparing the response spectrum analysis considering To carry out seismic analysis, three seismic waves
the virtual fixed point and the 3D dynamic analysis not were applied to the analysis. These were the long-period
considering the virtual fixed point in the seismic design of seismic wave (the Hachinohe earthquake), the short-
the pile-supported wharf. period seismic wave (the Ofunato earthquake), and the
artificial earthquake that reflected the characteristics of the
2. DETERMINATION OF THE VIRTUAL FIXED POINT site in Korea. According to the performance level, it is
divided into the collapse prevention level and the limited
The procedure for the determination of the virtual fixed serviceability level in South Korea. In this study, the
point is as follows. First, a virtual surface, which is a line collapse prevention level was applied, and the average
connecting the point corresponding to 1/2 of the slope returning period cycle was determined to be 1,000 years.
inclination and the pile, is defined. Next, the virtual fixed Figure 2 shows the regional acceleration factor for the
point is located at the 1/ß point under the virtual surface, 1,000-year earthquake ground motion proposed in Port
and the pile length below the virtual fixed point is ignored and Harbor Design Standard (2014) in Korea. The
while the pile with fixed ends is designed. Here, the ß value maximum acceleration of the seismic waves in Pohang on
is a characteristic value of the pile calculated using Eq.1 the earthquake disaster map is 0.13 g, as shown in Figure
and Eq.2 2. Therefore, the maximum acceleration of the three
seismic waves was adjusted to 0.13 g, as shown in Figure
4 Kh D 3 and the response spectrum curves of those earthquakes
ß=√ (cm−1 )     were shown in Figure 4.
4EI
k h = 0.15N (kgf/cm3 )    

The 𝑁 value is the average N value from SPT up to 1/ß

from the virtual surface, Kh is the coefficient of the
horizontal subgrade reaction, D is the diameter or width of
the pile, and 𝐸𝐼 is the flexural rigidity of the pile. The
method of determining the virtual fixed point is shown in
Figure 1.

Pohang

Figure 1. Determining the virtual fixed point

3. ANALYSIS METHODS AND CONDITIONS

For the response spectrum analysis and the dynamic

analysis using 3D numerical analysis, the analysis model
(3x3 piles) from the wharf construction site located in Figure 2. 1,000 years returning period earthquake hazard
Pohang, Korea was selected. The distance between the map (Port and harbor design standard, 2014)
piles was 5 m wide and 6 m high. In the case of the
response spectrum analysis model, the structure was Table 1. Material properties for numerical analysis
modeled considering the virtual fixed point, and the virtual
fixed point became shorter as it went from pile 1 to pile 3,
Property 𝜸𝒕 𝜸𝒔𝒂𝒕 𝑬 𝒗 𝑉𝑆 𝒄 Φ
(𝒌𝑵/𝒎𝟑 ) (𝒌𝑵/𝒎𝟑 ) (𝒌𝑵/𝒎𝟑 ) (𝒎/𝒔) (𝒌𝑵/𝒎𝟐 )
Soil (sand) 18 20 3.59 x 105 0.34 270 0 35
Rock 25 25 7.59 x 106 0.23 1100 1000 40
Plate (concrete) 24.5 2.6 x 107 0.16 D = 1m
Pile (steel) 78.5 2.1 x 108 A = 0.03958m2 I = 2.051 x 10−3 m4
D = 0.914m t = 0.014m

(a) Hachinohe eq.(long period)

(a) Hachinohe eq.(long period)

(b) Ofunato eq.(short period)

(b) Ofunato eq.(short period)

(c) Artificial eq.

(c) Artificial eq.
Figure 4. Response spectrum curves
Figure 3. Input accelerations
 In the case of the response spectrum analysis
considering the virtual fixed point, it modeled only the pile
structure and not the ground, and ignored the pile length
below the virtual fixed point and modeled only the upper
pile, as shown in Figure 5. Therefore, in the case of
response spectrum analysis, since the ground is not
modeled, the ground response analysis should be
performed to consider the ground amplification
phenomenon during earthquake. Next, the response
spectrum analysis is performed using the acceleration
value derived from the ground response analysis. As
response spectrum analysis is an elastic analysis method,
the maximum displacement of the structure is only
measured. MIDAS GEN ver.1.4 was used as a numerical
analysis program. MIDAS GEN is a 3D FEM program that
is widely used for structure analysis. In this study, the (a) Response spectrum analysis model diagram
response spectrum analysis module provided by MIDAS
GEN is utilized.
For the comparison, 3D dynamic analysis of the entire
structure was performed using the PLAXIS 3D ver.1.0.0
program, considering the ground. PLAXIS 3D is a 3D FEM
program that is widely used in the world today. Jeong and
Kim (2013) used this program for pile and soil design, and
used to design of the virtual fixed point model of a bent pile
structure. Besseling and Lengkeek (2015) used the
program to model the interaction between the soil and the
pile. Figure 6(a) shows the pile structure of the entire
structure of the model not considering the virtual fixed
point. The PLAXIS 3D model consists of a 3D triangular
mesh, like that shown in Figure 6(b). The adjacent part of
the pile and ground can be designed with fine mesh
elements to improve the accuracy of the analysis. In this (b) Response spectrum analysis model (MIDAS GEN)
analysis, the soil model that satisfies the Mohr-Coulomb
characteristics of the elasto-plastic model and the contact Figure 5. Response spectrum analysis model with the
element (boundary strength modulus) provided by PLAXIS virtual fixed point
3D were used for the contact surfaces between the piles
and the ground, and the slip between the ground and the
pile was modeled. In PLAXIS 3D, the contact element was
determined using the theories by Goodman et al. (1968)
and Van Langen and Vermeer (1991). The analysis region
was extended to where there is almost no stress and
deformation. Therefore, the boundaries of the model were
specified to be 100m apart from the area where the
structure was constructed.
Also, as the pile was inserted in the bedrock, the
analysis of the entire structure including the ground was
performed, and the properties and standard cross-section
were the same as in the response spectrum analysis. The
groundwater level was designated to be 2 m below the
surface, and the residual displacement and maximum
displacement were evaluated.
(a) Dynamic analysis model diagram
 4.3 Response Spectrum Analysis vs. Dynamic Analysis

4.3.1 Pile Moment Results

Figure 9(a) shows the moment time history curve of the

Hachinohe seismic wave obtained by PLAXIS 3D. As the
three seismic features are similar, the Hachinohe seismic
wave was only shown in this paper. Comparing Figure 3(a)
with Figure 9(a), it can be seen that the moment increased
sharply between 4 and 6 seconds when the acceleration of
Hachinohe wave became maximum in Figure 3(a). After
the 6 seconds, the moment converged to a constant value.
Figure 10(a) shows the maximum moment value of the
response spectrum analysis and the dynamic analysis by
seismic waves. In the case of the response spectrum
(b) Dynamic analysis model (PLAXIS 3D) analysis, the maximum moment occurred in the pile 3, and
in the dynamic analysis, the maximum moment occurred
Figure 6. Dynamic analysis model without the virtual fixed mostly in the pile 1. In the response spectrum model, the
point moment has the minimum value in the Hachinohe wave
and has the maximum value in the Ofunato seismic wave.
4. ANALYSIS RESULTS In the dynamic analysis, the moment has also a minimum
value in the Hachinohe wave but has the maximum value
4.1 Sine Sweep Test in the artificial seismic waves. It is because that the
resonance phenomenon of the soil-pile system occurring in
In the case of dynamic analysis model, sine sweep test was artificial seismic waves.
conducted to estimate the natural frequency of the soil-pile The maximum moments in the response spectrum
system. For the analysis, PLAXIS 3D was used. The analysis and the dynamic analysis were compared in
natural frequencies of the sine waves were changed from Figure 10(a). It can be seen that the difference in moment
0 to 10 Hz, and the natural frequency of the pile was between the Hachinohe seismic wave and the Ofunato
calculated by comparing the input acceleration with the pile seismic wave is large, and that the artificial wave has the
acceleration. As a result, it was found that the resonance smallest gap. This shows that the moment difference is
of the soil-pile system occurred at around 3.3Hz (natural large according to the resonance phenomena of the soil-
period: 0.3s) as shown in Figure 8. pile system.
To compare the result of sweep test with the spectral From the above results, a conservative design can be
acceleration value of the three seismic wave, the spectral expected from the use of the response spectrum analysis
acceleration values at 0.3s were calculated from the method because the moment value calculated by response
response spectrum curves of the seismic waves shown in spectrum analysis was generally larger, but it should be
Figure 4. Hachinohe seismic wave (long-period seismic emphasized that the difference between the two analysis
wave) was 0.21g, the Ofunato seismic wave (short-period values varied depending on the resonance phenomena of
seismic wave) was 0.29g, and the artificial seismic wave the soil-pile system.
was 0.32g. The resonance phenomenon of artificial Next, the moments by depth for the response spectrum
seismic waves can be expected to occur the most. analysis and the dynamic analysis were compared. As the
results of 3 seismic waves were similar, the moment value
4.2 Ground Response Analysis of Hachinohe seismic wave was only shown in Figure 11.
Figure 11(a) shows the moment values by depth when the
Generally, when an earthquake occurs, the seismic wave Hachinohe wave was applied to the response spectrum
is transmitted to the soil layer through bedrock, and the analysis, and Figure 11(b) shows the moment by depth
response acceleration value generally increases in the soil when the Hachinohe wave was applied to the dynamic
layer. This phenomenon is called the ground amplification analysis.
phenomenon. In order to consider the ground amplification In the response spectrum analysis, the moment values
phenomenon of the pile supported wharf, the response changed linearly with the depth. It varied linearly according
acceleration value was calculated using the SHAKE to the depth and the maximum value was occurred near the
program, which is a one-dimensional equivalent linear virtual fixed point. The minimum moment was generated in
analysis program, and the calculated response pile 1, the longest pile, and the maximum moment was
acceleration value was applied to the response spectrum generated in pile 3, the shortest pile. The top deformation,
analysis. however, was constant due to the constraint on the top
Figure 7 shows the ground response analysis results deformation. Therefore, it seems that pile 3, which was
of three seismic waves. These seismic waves were relatively short, had the maximum moment.
adjusted to the peak acceleration of 0.13g in bedrock. As a In the case of the dynamic analysis, the moment values
result of analysis, the largest ground amplification occurred changed nonlinearly with the depth, and converged to 0 as
with 0.28g of artificial seismic wave, 0.27g of Ofunato the depth increased as shown in Figure 11. The maximum
seismic wave, 0.25g of Hachinohe seismic wave.
moment was generated in pile 1, and the minimum moment
was generated in pile 3. This is because pile 1 is close to
the coastal side, and as such, its moment value was the
largest.
However, the two models tend to have similar pile
moment tendency up to the virtual fixed point, but in the
case of the dynamic analysis, the moment of the pile is
reversed beyond the virtual fixed point. As the moment
value can be maximized under the virtual fixed point, it is
necessary to strictly design a pile-supported wharf
considering the ground condition under the virtual fixed
point.

4.3.2 Pile Displacement Results

Figure 9(b) shows the displacement time history curve of

the Hachinohe seismic wave. Comparing Figure 3(a) with
Figure 9(b), it can be seen that the displacement increased
sharply between 4 and 6 seconds when the acceleration of
Hachinohe seismic wave became maximum in Figure 3(a).
After the 6 seconds, the displacement converged to a
constant value.
Ferritto (1997) considered the residual displacement for
the displacement criteria, and the allowable horizontal
displacement of the limited serviceability level is up to 10
cm while the collapse prevention level is limited to 30 cm.
As the response spectrum analysis that was performed in
this study was an elastic analysis method, it was Figure 7. Ground response analysis
impossible to estimate the residual displacement, and as
such, the maximum displacement was considered. In the
case of dynamic analysis, both the maximum and residual
displacements were evaluated. Figure 10(b) shows that the
maximum displacement of the response spectrum analysis
is relatively smaller than that of the dynamic analysis. It is
because that the response spectrum analysis takes into
account the virtual fixed point and therefore the
displacement decreases. In the case of the dynamic
analysis, the value of the maximum and residual
displacements were similar.
Figure 12(a) shows the displacement values by the
depth when the Hachinohe wave was applied to the
response spectrum analysis. Figure 12(b) shows the
displacement values by the depth when the Hachinohe
wave was applied to the dynamic analysis. In the case of
the response spectrum analysis considering the virtual Figure 8. Sine sweep test
fixed point, the displacement value decreased linearly with
the depth, and the displacement values of the virtual fixed
points below each pile were all zero. In the case of the
dynamic analysis, a pile also existed below the virtual fixed
point, and similarly, the displacement value of the lower
region converged to zero. In the case of the response
spectrum analysis, the point at which the displacement
became zero was 10-14 m. For the dynamic analysis, on
the other hand, the point at which the displacement
became zero was about 15 m. However, the decrease in
displacement per depth seems to have a similar tendency
in general.
 (a) Moment time history curve (b) Displacement time history curve
Figure 9. Pile moment and displacement time history
curve (Pile1, Hachinohe eq.)

(a) Comparison of pile moment (b) Comparison of pile displacement

Figure 10. Comparison between the response spectrum
analysis and the dynamic analysis results

(a) Response spectrum (b) Dynamic analysis (a) Response spectrum (b) Dynamic analysis
analysis analysis
Figure 11. Comparison of pile moment depending on Figure 12. Comparison of pile displacement depending on
depth (Hachinohe eq.) depth (Hachinohe eq.)
5. CONCLUSIONS Goodman, R.E., Taylor, R.L., Brekke, T.L. 1968. A Model
for the Mechanics of Jointed Rocks, Journal of Soil
In this study, the differences between two analysis models Mechanics &Foundations Div.
were derived by comparing the response spectrum Jeong, S.S. and Kim, J.Y. 2013 Analysis of Optimized
analysis results considering the virtual fixed point and the Column-pile Length Ratio for Supplementing Virtual
dynamic analysis results not considering the virtual fixed Fixed Point Design of Bent Pile Structures, Journal of
point. The conclusions below were drawn. Korea Society of Civil Engineers, 33(5): 1915-1933. (in
Korean)
(1) A conservative design can be expected from the use of Ministry of Oceans and Fisheries. 1999. Seismic Design
response spectrum analysis because the maximum Standard of Port and Harbor. (in Korean)
moment value of the response spectrum analysis Ministry of Oceans and Fisheries. 2014. Port and Harbor
model was generally larger. It was seen, however, that Design Standard. (in Korean)
the difference between the values obtained from the Ferritto, J.M. 1997. Design Criteria for Earthquake Hazard
two analysis models varied depending on the Mitigation of Navy Piers and Wharfs (No. NFESC-TR-
resonance phenomena. 2069-SHR), Naval Facilities Engineering Service
(2) The pile moments by the depth from the response Center Port Hueneme Ca.
spectrum analysis and the dynamic analysis showed PARI. 2009. Technical Standards and Commentaries for
that it is necessary to strictly design a pile-supported Port and Harbour Facilities in Japan, Port and Airport
wharf considering the ground condition under the virtual Research Institute, Japan.
fixed point because the moment value can be PIANC. 2001. Seismic Design Guidelines for Port
maximized under the point. Structures, International Navigation Association.
(3) The maximum displacement of the pile from the Brinkggreve, R.B.J., Kumarswamy, S., Swolfs, W.M. 2015.
response spectrum analysis was relatively smaller than PLAXIS 3D User Manual, PLAXIS Inc., Netherand.
that of the dynamic analysis. The moment and Shafieezadeh, A., DesRoches, R., Rix, G.J., Werner, S.D.
displacement tendencies by the depth in the two 2012. Seismic Performance of Pile-supported Wharf
analysis methods, however, were similar. Structures Considering Soil-structure Interaction in
(4) In the case of the pile moment, the response spectrum Liquefied Soil, Earthquake Spectra, 28(2): 729-757.
analysis result were larger, and for the displacement, Van Langen, H., and Vermeer, P.A. 1991. Interface
the dynamic model results were larger. The results of Elements for Singular Plasticity Point, International
this study have limitations, however, because the Journal for Numerical and Analytical Methods in
analyses were performed as a single-stratum study for Geomechanics, 15(5): 301-315.
the simple estimation of the virtual fixed points.
Therefore, further research on multi-layered ground
analysis is necessary and the analysis results should
be verified through dynamic centrifuge model tests or
existing case studies.

ACKNOWLEDGMENTS

This research was supported by the project of the

development of performance-based seismic design
technologies for advancement in the design codes for port
structures from Korea Institute of Marine Science &
Technology.

REFERENSES

ASCE/COPRI. 2014. Seismic Design of Piers and

Wharves, ASCE standard 61-14, American Society of
Civil Engineers.
Besseling, F. and Lengkeek, H.J. 2015. Soil-Structure
Interaction Modelling for Piles in Performance-Based
Seismic Design, 2015 Fall International Conference on
Earthquake Geotechnical Engineering.
Doran, B., Shen, J., Akbas, B. 2015. Seismic Evaluation of
Existing Wharf Structures Subjected to Earthquake
Excitation: Case Study, Earthquake Spectra, 31(2):
1177-1194.
Donahue, M.J., Dickenson, S.E., Miller, T.H., Yim, S.C.
2005. Implications of the Observed Seismic
Performance of a Pile-supported Wharf for Numerical
Modeling, Earthquake Spectra, 21(3): 617-634.