Int. J. Nav. Archit. Ocean Eng.

(2014) 6:1160~1181
http://dx.doi.org/10.2478/IJNAOE-2013-0237
ⓒ SNAK, 2014 pISSN: 2092-6782, eISSN: 2092-6790

Experimental investigations of higher-order springing

and whipping-WILS project

Sa Young Hong and Byoung Wan Kim

Korea Research Institute of Ships and Ocean Engineering, Daejeon, Korea

ABSTRACT: Springing and whipping are becoming increasingly important considerations in ship design as container
ships increase in size. In this study, the springing and whipping characteristics of a large container ship were investigated
through a series of systematic model tests in waves. A multi-segmented hull model with a backbone was adopted for
measurement of springing and whipping signals. A conversion method for extracting torsion springing and whipping is
described in this paper for the case of an open-section backbone. Higher-order springing, higher-mode torsion res-
ponses, and the effects of linear and nonlinear springing in irregular waves are highlighted in the discussion.

KEY WORDS: Springing; Whipping; Backbone model; Containership; Resonant whipping; Higher-order springing;
Torsion springing and whipping.

INTRODUCTION

As container ships increase in size, springing and whipping are becoming increasingly important considerations in ship
design because of their significant effects on wave loads and fatigue. The importance of springing and whipping effects to the
structural integrity of ultra-large container ships has motivated classification societies to validate the tools and procedures used
to evaluate the structural loads associated with springing and whipping and the resulting fatigue damage. The analysis tools
that the major classification societies use for analysis of wave loads and the dynamic responses of ships, including hydroelastic
effects, are based on potential theory and three-dimensional (3-D) panel methods, combined with either simplified beam
models or full finite element method (FEM) models (e.g., G-Hydroflex, HydroSTAR, NLOAD3D, and WISH-FLEX; Hong
et al., 2011). However, the Reynolds-averaged Navier–Stokes (RANS) equation approach has become an alternative for
analyzing highly nonlinear wave forces, by virtue of recent advances in computational fluid dynamics (CFD) technology
(Oberhagemann et al., 2010).
Some examples of model tests of wave loads and springing and whipping responses can be found in the literature. Storhaug
and Moan (2006) conducted a springing and whipping model test with a model made of four segments with flexible joints.
Remy et al. (2006) investigated the hydroelastic effects of wave loads on torsion moments using two-rod beam structures that
allowed torsional deformation, but the effect of springing was not considered because the model ship was a box structure. Iijima
et al. (2009) investigated torsion springing experimentally using a backbone with partial cut-outs to model the elastic
characteristics of torsion. Miyake et al. (2009) conducted a model test of the hydroelastic responses of a very large container

Corresponding author: Sa Young Hong, e-mail: sayhong@kriso.re.kr

This is an Open-Access article distributed under the terms of the Creative Commons
Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/3.0)
which permits unrestricted non-commercial use, distribution, and reproduction in any
medium, provided the original work is properly cited.
Int. J. Nav. Archit. Ocean Eng. (2014) 6:1160~1181 1161

ship in regular and irregular seas and observed higher-order resonant harmonic bending moment behaviors in irregular seas.
Oka et al. (2009) investigated springing and whipping phenomena experimentally in regular and irregular waves, using a type
of open-section backbone made of two C-shaped channels. They were successful in detecting vertical and horizontal whipping
signals but obtained no further information with respect to torsion springing and whipping.
Most previous experimental studies on springing and whipping have been limited to analysis of the bending mode only.
Experimental investigations of torsion springing and whipping have rarely been conducted because of the complexity of the
measurements and analysis of torsion moment associated with the backbone model. For this reason, reliable model test data
for use in validation are lacking, despite the considerable number of previous experimental studies on springing and whipping.
The need of classification societies and ship builders for experimental springing and whipping data for validation purposes has
grown because springing and whipping model test data are the most reliable references for validation of the state-of-the-art
numerical analysis tools.
In a validation study of the effects of wave loads on a large container ship, Hong et al. (2008) presented a set of
meaningful model test data for global wave loads for both bending and torsion modes. Their experimental study was
conducted as a joint industry project, referred to as WILS-I (Wave-Induced Loads on Ships, Phase I), in which most of the
major classification societies and Korean shipyards participated. Hong et al. (2011; 2012) investigated torsion springing and
whipping experimentally using six segmented container ship models with H-backbones, with a focus on torsion springing
and whipping.
In the present study, a series of model tests of springing and whipping of a 10,000-TEU-class container ship was con-
ducted, with a focus on torsion moment and vertical bending moment, using a U-shaped backbone. A segmented ship model
with various open-section backbones was used to measure and compare torsion springing and whipping. This model was
developed as part of the joint industry project “Wave-Induced Loads on Ships, Phases II & III (WILS-II & III).” A series of
model tests were carried out for selected cases of springing and whipping in regular and irregular waves. The focus in this
study was on the physics of torsion springing and whipping, as measured in model tests in which higher-mode torsion
responses were observed. The improvement in the torsion conversion matrix was examined, along with that of the measured
linear torsion response–amplitude operator (RAO), which demonstrates enhancement of the conversion method for an open-
section backbone. Some of the cases investigated by Hong et al. (2012) and Kim et al. (2014) were revisited to examine
aspects of the physics of springing and whipping that were overlooked in the previous studies. The discussion presented in this
paper highlights higher-order springing resonance, higher-mode torsion responses in springing and whipping, and the effects
of linear and nonlinear springing in irregular waves.

MODEL SET-UP

Ship model

A 10,000-TEU container ship model, which is the same model used in WILS JIP-II and III, was considered in this
study. The model is composed of six segments of a fiber-reinforced plastic (FRP) hull and a steel backbone to which the
segments are connected. The segmented model configuration is shown in Fig. 1, and the details of the model are sum-
marized in Table 1.
The model was made of FRP with a wooden base that was divided into six equal segments on the basis of LPP.
Each segment was fitted to a U-shape steel backbone that was designed to match the vertical bending and torsional
stiffness simultaneously. The mass distributions of the six segments were adjusted for full loading conditions according
to an assigned weight distribution. The center of gravity (KG) and the radii of gyration of each segmented part for pitch
and roll (kxx and kyy) were adjusted to a tolerance of within 5% using a swinging table. The mass distributions, which
are shown in Fig. 2 (Hong et al., 2012), were extrapolated from those of the WILS-I model ship. In the present study,
the notations Myi and Mxi are used for the vertical bending moment (VBM) and torsional moment (TM), respectively, at
section Si (see Fig. 1).
1162 Int. J. Nav. Archit. Ocean Eng. (2014) 6:1160~1181

(a) Ship model in waves.

(unit=mm, Si=strain sensing sections)

3996
678 330 S1 330 446 S2 446 446 S3 446 446 S4 446 330 S5 330 678

z
Backbone Fixing system
x WL

Segment (1) Segment (2) Segment (3) Segment (4) Segment (5) Segment (6)
112 892 892 892 892 892 892 149

5350
AP FP
5611

(b) Arrangement of segments and backbone.

Fig. 1 Model ship.

Table 1 Main particulars of model ship.

Particulars Real Model
Scale 1/1 1/60
LOA 336.641 m 5.611 m
LBP 321 m 5.350 m
Breadth 48.4 m 0.807 m
Height 27.2 m 0.453 m
Draft 15 m 0.250 m
Displacement 143,741.920 ton 665.472 kg
KM 23.296 m 0.388 m
GM 2m 0.033 m
KG 21.296 m 0.355 m
LCG from AP 152.495 m 2.542 m
kxx 19.073 m 0.318 m
kyy 77.228 m 1.287 m
kzz 77.228 m 1.287 m
Troll 29.5 s 3.808 s
Int. J. Nav. Archit. Ocean Eng. (2014) 6:1160~1181 1163

2.5E+07
1000
900 6250 TEU 6250 TEU

Rotational mass (ton-m )

2
10000 TEU 2.0E+07
10000 TEU
800
700
Mass/L (ton/m)

600 1.5E+07

500
400 1.0E+07

300
200 5.0E+06

100
0 0.0E+00
0 2 4 6 8 10 12 14 16 18 20 0 2 4 6 8 10 12 14 16 18 20
St. St.
(a) Weight distribution. (b) Rotational mass distribution.
Fig. 2 Weight and rotational mass distributions along the ship length.

The test set-up and sensor arrangement is shown in Fig. 3. The model was towed using a counterweight system (Hong et al.,
2008) that balances the ship resistance in calm water. Additional resistance in waves is counterbalanced by four springs located
at the stem and stern centers. This counterbalance system helps the model to maintain its course in oblique wave headings.
Three-axis strain gauges were installed to measure the strains at target sections of the backbone, as shown in Fig. 4. Six degrees
of ship motions, local accelerations, relative waves, and local pressures and local forces were also measured. However, only
springing and whipping and the responses associated with them are discussed in this paper.

Fig. 3 Test set-up and layout of sensors.

The target natural frequencies of the bending and torsion modes were 0.45 and 0.35 Hz, respectively. The backbone was
designed to match the bending and torsion natural frequencies simultaneously. An H-shaped backbone was used in WILS
JIP-II because of the advantages afforded by its symmetric section shape (Hong et al., 2012). Because the neutral point and
shear center are identical in an H-section backbone, the strain distribution is symmetric, and the force conversion is simple.
However, because the shear center of a real container ship is located far below the neutral center of the backbone (below the
bottom), a more realistic modeling approach was needed to represent real ship torsion in the model tests. A U-shaped back-
1164 Int. J. Nav. Archit. Ocean Eng. (2014) 6:1160~1181

bone was devised in WILS JIP-III to model torsion modes realistically. The two backbone models and their characteristics
are described and compared in this section. The shapes and section properties of the H and U backbones are presented in Fig.
4 and Table 2. The measured shear center location of the U-section was 38.12 mm at model scale, which corresponds to 2.3
m at full scale.

(unit = mm) (unit = mm)

B=148 B=150
tf =7.9
12.6 10.29
z
Flange

Flange
H=65
tf =9

H=84
y Neutral point
Web
34.175
Neutral point zc 95 deg
t w=6 = Shear center Web tf =13
z
e tw=6.5
y
Strain gauge (Rosette type)
Shear center
Fig. 4 Section shapes of H and U backbones.

Table 2 Section properties of H and U backbones.

Section properties H backbone (WILS JIP-II) U backbone (WILS JIP-III)
Neutral point (zc) 0 20.02 mm
Shear center (e) 0 19.70 mm
Vertical bending stiffness (Iy) 89.140 cm4 83.902 cm4
Lateral bending stiffness (Iz) 841.204 cm4 781.300 cm4
Pure torsion constant (J0) 5.018 cm4 6.831 cm4
Warping torsion constant (J) 10.143 cm4 10.817 cm4

Conversion method of VBM and TM

The advantage of adopting a simple backbone to measure the torsion moment and vertical bending moment is that it permits
the use of systematic conversion matrices. Section forces can be calculated from measured strains at specific sensing sections
using the force and strain relation given as equation (1):

{ f } = [C ]{ε } (1)

where {f} is a vector of section forces, {ε} is a vector of measured strains, and [C] is a conversion matrix. The conversion
matrix can be constructed using either theoretical or numerical methods. To consider the so-called warping effect in the case
of an open-section beam, a finite element analysis was conducted in this study using the commercial solver NE/NASTRAN
(Noran Engineering, 2001). Details of the FEM-based construction of the conversion matrix can be found in Hong et al.
(2012).
The warping effect is significant in the case of an open-section beam. The strains induced by torsion moments vary
depending on the loading modes, even when the applied torsion moment along the beam is the same. Thus, in general, the tor-
sion moment (TM) components of the conversion matrix are not unique. The set-up of the TM conversion matrix is therefore
complicated. A variety of conversion modes for use in TM measurement were investigated and compared in this study. Four
possible loading configurations were investigated: section-by-section decoupled second-mode loading, section-by-section
decoupled first-mode loading, fully coupled loading, and full-span first-mode loading. The TM conversion matrices A, B, C,
Int. J. Nav. Archit. Ocean Eng. (2014) 6:1160~1181 1165

and D illustrated in Fig. 5 were constructed for these loading modes.

For conversion modes A and B, it is assumed that the TM in a section only induces strain at the section and does not affect
the adjacent sections. If the strains in the adjacent sections are very small compared to that in the loaded section, conversions A
and B are applicable. However, in the case of an open-section beam, the TM in a section induces strains in the adjacent sections,
as well as in the section in which it is applied, due to warping, which causes coupled effects between the section forces. The
conversion matrix C was constructed without neglecting these coupled effects. This is the case of a fully coupled generalized
model that is theoretically complete. If the contribution of each mode is similar in the case of conversion mode C, the TM in
reverse directions can be considered canceled. This leads to the mode shown in Fig. 3(d). For conversion mode D, it was
assumed that a uniform torsion moment was loaded along the whole span. Conversion mode C is the most consistent and
general model, irrespective of the loading mode. The accuracy and applicability of conversion modes A, B, C, and D are
discussed in this paper for the model test cases considered.

S1
S1 S2

S2
S2 S3

S3
S4 S5

S4
(a) Decoupled 2nd mode (conversion matrix A)
S1 S2 S3 S4 S5
S5

S1 S2 S3 S4 S5

(b) Decoupled 1st mode (conversion matrix B)

S1 S2 S3 S4 S5

S1 S2 S3 S4 S5

S1 S2 S3 S4 S5
S1 S2 S3 S4 S5

(c) Coupled mode (conversion matrix C) (d) Full span 1st mode (conversion matrix D)

Fig. 5 Loading modes of TM conversion matrix.

TEST RESULTS AND DISCUSSION

Hammering tests

Hammering tests results for the H-section beam and U-section beam cases are compared in Figs. 6 and 7, respectively.
The hammering test results for the H-section beam were taken from Hong et al. (2012) for the purpose of comparison.
Measured first-mode natural frequencies obtained from fast Fourier transform (FFT) analysis are summarized in Table 3.
Measured bending and torsion moments due to impact by hammering are shown for five locations along the beam length. A
natural bending frequency of 0.43 Hz was measured for both the H-section and U-section beams. A relative error of less than
5% was obtained for the bending-mode natural frequency. For the bending mode, the highest level was observed at midship,
and a sharp peak occurs at the first-mode target frequency. For the torsion mode, slightly different natural frequencies were
obtained for the H-section and U-section backbone models: 0.30 Hz and 0.32 Hz, respectively. The corresponding relative
1166 Int. J. Nav. Archit. Ocean Eng. (2014) 6:1160~1181

errors with respect to the target frequency were approximately 14.3% and 8.6%, respectively. Considering the complicated
mechanism of torsion and the simplified design of the backbone section shape, a relative error of approximately 10% can be
considered acceptable. Details of the design of the backbone model used to match the bending and torsion natural frequencies
simultaneously can be found in Hong et al. (2012). It is worth noting that the spectral shape of the torsion hammering test
results exhibit relatively wider bandwidths and multiple peaks. The wider bandwidth is due to higher damping and the rela-
tively short duration of measurement, which resulted in rapid signal decay. The second peak was observed near the bending
natural frequency but was not coupled with the bending mode, for two reasons: first, because the H-section backbone is
theoretically free from coupling between bending and torsion, and second because the second peak is associated with coupling
due to bending, which should be dominant at midship but was not. For these two reasons and based on the interpretation of the
multiple peaks measured in the hammering tests for torsion mode, it is reasonable to identify those peaks as corresponding to
one of the higher torsion modes. The corresponding locations of the second and third peaks are closely correlated for the H-
section and U-section backbones. This observation also supports the significance of higher-order torsion modes in the back-
bone model. Thus, it is important to consider higher-order torsion mode characteristics when designing a backbone model.
Table 3 shows the mode analysis results for the H-section backbone obtained using a commercial FEM code. Higher-mode
torsion exists up to the fourth mode within a 1-Hz frequency band. It is noteworthy that the U-section torsion signal exhibits a
wider spectral distribution than the H-section, which may be due to different degrees of damping due to the different setups,
although further investigation is needed to confirm that that is the reason for the difference. The occurrence of dominant
signals at the midship location for both vertical bending and torsion moments provides information on the mode charac-
teristics. The hammering locations were at the center of station 8 of the backbone for bending and at the deck sidewall at
station 14 for torsion. No significant differences were observed with changes in ship speed in hammering tests conducted at
speeds of 0 to 20 knots. The results for a ship speed of 15 knots are shown in this paper.

Table 3 Natural frequencies of model ship in real scale.

(a) Target versus measured.
Measured
Mode Target
H backbone U backbone
ωbn (bending freq) 0.45 Hz 0.43 Hz 0.43 Hz
ωtn (torsion freq) 0.35 Hz 0.30 Hz 0.32 Hz

(b) Calculated by eigenmode analysis.

Mode no. Natural frequency Mode shape

1 0.35 Hz 1st torsion
2 0.45 Hz 1st vertical bending
3 0.58 Hz 2nd torsion
4 0.78 Hz 3rd torsion
5 0.95 Hz 4th torsion
6 1.14 Hz 2nd vertical bending
7 1.19 Hz 5th torsion
8 1.92 Hz 1st lateral bending
9 2.10 Hz 3rd vertical bending
10 3.27 Hz 4th vertical bending
Int. J. Nav. Archit. Ocean Eng. (2014) 6:1160~1181 1167

0.43 Hz 0.30 Hz
3.0E+12 2.0E+11

1.5E+11

Mx1 (PSD)
My1 (PSD)
2.0E+12
1.0E+11
1.0E+12
5.0E+10

0.0E+00 0.0E+00
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
3.0E+12 ω (Hz) 2.0E+11 ω (Hz)
1.5E+11
My2 (PSD)

Mx2 (PSD)
2.0E+12
1.0E+11
1.0E+12
5.0E+10
0.0E+00 0.0E+00
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
3.0E+12 ω (Hz) 2.0E+11 ω (Hz)
1.5E+11
My3 (PSD)

Mx3 (PSD)
2.0E+12
1.0E+11
1.0E+12
5.0E+10

0.0E+00 0.0E+00
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
3.0E+12 ω (Hz) 2.0E+11 ω (Hz)
1.5E+11
My4 (PSD)

Mx4 (PSD)
2.0E+12
1.0E+11
1.0E+12
5.0E+10
0.0E+00 0.0E+00
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
3.0E+12 ω (Hz) 2.0E+11 ω (Hz)
1.5E+11
My5 (PSD)

Mx5 (PSD)

2.0E+12
1.0E+11
1.0E+12
5.0E+10

0.0E+00 0.0E+00
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
ω (Hz) ω (Hz)

(a) VBM. (b) TM.

Fig. 6 PSD of H-section beam hammering tests (ship speed 15 knots) (Hong et al., 2012).

0.43 Hz 0.32 Hz
1.5E+12 4.5E+10
My1 (PSD)

Mx1 (PSD)

1.0E+12 3.0E+10

5.0E+11 1.5E+10

0.0E+00 0.0E+00
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
1.5E+12 ω (Hz) 4.5E+10 ω (Hz)
My2 (PSD)

Mx2 (PSD)

1.0E+12 3.0E+10

5.0E+11 1.5E+10

0.0E+00 0.0E+00
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
1.5E+12 ω (Hz) 4.5E+10 ω (Hz)
My3 (PSD)

Mx3 (PSD)

1.0E+12 3.0E+10

5.0E+11 1.5E+10

0.0E+00 0.0E+00
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
1.5E+12 ω (Hz) 4.5E+10 ω (Hz)
My4 (PSD)

Mx4 (PSD)

1.0E+12 3.0E+10

5.0E+11 1.5E+10

0.0E+00 0.0E+00
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
1.5E+12 ω (Hz) 4.5E+10 ω (Hz)
My5 (PSD)

Mx5 (PSD)

1.0E+12 3.0E+10

5.0E+11 1.5E+10

0.0E+00 0.0E+00
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
ω (Hz) ω (Hz)

(a) VBM. (b) TM.

Fig. 7 PSD of U-section beam hammering test (ship speed 15 knots).
1168 Int. J. Nav. Archit. Ocean Eng. (2014) 6:1160~1181

VBM and TM RAO tests in oblique waves

A series of RAO tests were conducted in oblique head waves to investigate the effects of the two different backbone models
and the torsion moment conversion method. The case of a heading angle of 150 degrees and a ship speed of 20 knots in full
scale was investigated. Fig. 8 shows a comparison of the torsion conversion methods for the case of the H-section backbone.
The measured torsions are denoted by blue and red lines, and the black line represents calculated values from Bureau Veritas
(Bigot et al., 2011). The measured values change noticeably depending on the conversion method, but the overall trends are
similar for the different conversion methods. Among the four conversion methods for torsion moment, conversion matrix C
yields the best fit with the calculated values, as expected, because of the theoretically complete consideration of full coupled
effects between sensor signals. The measured values exhibit smoother behavior than the calculated values because of the
damping effect, which was not fully considered in the calculation. Conversion matrix D yielded results similar to those
obtained for conversion matrix C, which implies that the dominant torsion loading mode in waves can be approximated by
conversion matrix D. Conversion matrices A and B were found to yield conservative estimates, while conversion matrix D
was found to yield slightly more optimistic values than matrix C. The terms Mx1, Mx2, Mx3, Mx4, and Mx5 correspond to
torsion moments at S1 (AP) section through S5 (FP), respectively, as shown in Fig. 1. The wave-induced torsion moments
exhibited similar trends with respect to the ship length, except for the ship front locations, where the sectional shape changes
significantly. This explains why conversion matrix D is effective in spite of its modeling being the simplest. Fig. 9 compares
the measured TM RAOs for the two different backbone shapes. Conversion matrix C was applied to both cases. The overall
trends are similar, but noticeable differences exist depending on the locations and frequencies, which indicates that more
accurate backbone modeling is needed for realistic modeling of torsion. Fig. 10 compares the VBM RAO for the different
backbone sections. In general, the measured and calculated values are in good agreement. The VBM RAO values are not
very different for the H and U backbones.

4.0E+04 4.0E+04
Cal (H backbone) Cal (H backbone)
M x1 (tonf-m/m)

M x2 (tonf-m/m)

3.0E+04 Exp A (H backbone) 3.0E+04 Exp A (H backbone)

Exp B (H backbone) Exp B (H backbone)
Exp C (H backbone) Exp C (H backbone)
2.0E+04 Exp D (H backbone) 2.0E+04 Exp D (H backbone)

1.0E+04 1.0E+04

0.0E+00 0.0E+00
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
ω (rad/s) ω (rad/s)

(a) Mx1. (b) Mx2.

4.0E+04 4.0E+04
Cal (H backbone) Cal (H backbone)
M x4 (tonf-m/m)
M x3 (tonf-m/m)

Exp A (H backbone) 3.0E+04 Exp A (H backbone)

3.0E+04
Exp B (H backbone) Exp B (H backbone)
Exp C (H backbone) Exp C (H backbone)
2.0E+04 Exp D (H backbone) 2.0E+04 Exp D (H backbone)

1.0E+04 1.0E+04

0.0E+00 0.0E+00
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
ω (rad/s) ω (rad/s)

(c) Mx3. (d) Mx4.

4.0E+04
Cal (H backbone)
M x5 (tonf-m/m)

3.0E+04 Exp A (H backbone)

Exp B (H backbone)
Exp C (H backbone)
2.0E+04 Exp D (H backbone)

1.0E+04

0.0E+00
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
ω (rad/s)

(e) Mx5.
Fig. 8 Comparison of TM conversion in RAO.
Int. J. Nav. Archit. Ocean Eng. (2014) 6:1160~1181 1169

4.0E+04 4.0E+04
Cal (H backbone) Cal (H backbone)
M x1 (tonf-m/m)

M x2 (tonf-m/m)
3.0E+04 Exp C (H backbone) 3.0E+04 Exp C (H backbone)
Exp C (U backbone) Exp C (U backbone)

2.0E+04 2.0E+04

1.0E+04 1.0E+04

0.0E+00 0.0E+00
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
ω (rad/s) ω (rad/s)

(a) Mx1. (b) Mx2.

4.0E+04 4.0E+04
Cal (H backbone) Cal (H backbone)
M x3 (tonf-m/m)

M x4 (tonf-m/m)
3.0E+04 Exp C (H backbone) 3.0E+04 Exp C (H backbone)
Exp C (U backbone) Exp C (U backbone)

2.0E+04 2.0E+04

1.0E+04 1.0E+04

0.0E+00 0.0E+00
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
ω (rad/s) ω (rad/s)

(c) Mx3. (d) Mx4.

4.0E+04
Cal (H backbone)
M x5 (tonf-m/m)

3.0E+04 Exp C (H backbone)

Exp C (U backbone)

2.0E+04

1.0E+04

0.0E+00
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
ω (rad/s)

(e) Mx5.
Fig. 9 TM RAO (ship speed = 20 kts, heading = 150 deg).

4.0E+05 4.0E+05
Cal (H backbone) Cal (H backbone)
M y1 (tonf-m/m)

M y2 (tonf-m/m)

Exp (H backbone) Exp (H backbone)

3.0E+05 3.0E+05
Exp (U backbone) Exp (U backbone)

2.0E+05 2.0E+05

1.0E+05 1.0E+05

0.0E+00 0.0E+00
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
ω (rad/s) ω (rad/s)

(a) My1. (b) My2.

4.0E+05 4.0E+05
Cal (H backbone) Cal (H backbone)
M y3 (tonf-m/m)

Exp (H backbone)
M y4 (tonf-m/m)

3.0E+05 Exp (H backbone)

3.0E+05
Exp (U backbone) Exp (U backbone)

2.0E+05 2.0E+05

1.0E+05 1.0E+05

0.0E+00 0.0E+00
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
ω (rad/s) ω (rad/s)

(c) My3. (d) My4.

4.0E+05
Cal (H backbone)
M y5 (tonf-m/m)

Exp (H backbone)
3.0E+05
Exp (U backbone)

2.0E+05

1.0E+05

0.0E+00
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
ω (rad/s)

(e) My5.
Fig. 10 VBM RAO (ship speed = 20 kts, heading = 150 deg).
1170 Int. J. Nav. Archit. Ocean Eng. (2014) 6:1160~1181

Springing and whipping test cases

Springing is a structural resonance that a ship exhibits when it encounters wave frequencies or multiples of wave fre-
quencies that coincide with its own natural frequencies. Second-order springing is a type of springing that occurs when a ship
encounters waves with periods that are twice the magnitude of its own natural bending or torsion periods. The natural
frequencies of ship structures are usually high enough to avoid wave period resonance, but second-order springing is common.
However, because of rapid increases in container ship sizes in recent years, linear springing has become an important consi-
deration. Because higher-order springing is theoretically possible when the incident wave period is long and the wave is high,
experimental investigation of higher-order springing is also of interest.
Whipping is a type of impulsive response that a ship exhibits when subjected to impact loads such as slamming and deck-
wetting. In irregular waves, whipping manifests itself as a resonant response but rapidly decays, whereas springing resonance
can continue for long periods of time. In regular waves, whipping is caused by periodic impact, whereas whipping in irregular
waves is triggered by rare occurrences of slamming and deck-wetting. According to Hong et al. (2012), two types of whipping
occur in regular waves: resonant whipping and non-resonant whipping. Hong et al. described typical patterns of resonant
whipping and non-resonant whipping for both vertical bending and torsion. In this study, torsion springing and whipping were
investigated, along with vertical bending mode, for various sections of backbone models under identical test conditions. The test
cases considered in the investigation of higher-order springing and resonant whipping are listed in Table 4.

Table 4 Selected test cases of high-order springing and resonant whipping.

Speed Wave Wave Heading
Test type ID Backbone
(kts) height (m) period (s) (deg)
VBM2H-1 H 19 1 8.199 180
VBM2H-3 H 19 3 8.199 180
2nd order VBM springing
VBM2H-5 H 19 5 8.199 180
VBM2U-5 U 19 5 8.199 180
VBM3H-3 H 18.5 3 10.930 180
3rd order VBM springing
VBM3H-5 H 18.5 5 10.930 180
VBM4H-3 H 20 3 13.659 150
4th order VBM springing VBM4H-6 H 20.3 6 14.341 180
VBM4H-9 H 15 9 12.800 180
5th order VBM springing VBM5H-9 H 15 9 15.400 180
TM2H-1 H 17.5 1 10.345 150
TM2H-3 H 17.5 3 10.345 150
2nd order TM springing
TM2H-5 H 17.5 5 10.345 150
TM2U-5 U 19 5 9.799 150
TM3H-3 H 18 3 14.060 150
3rd order TM springing
TM3H-5 H 18 5 14.060 150
4th order TM springing TM4H-9 H 10 9 16.200 150
5th order TM springing TM5H-9 H 5 9 18.700 150
VBMWHH H 18 12 14.341 180
VBM whipping
VBMWUH U 18 12 14.341 180
Resonant TM whipping TMWHO H 20.3 12 14.341 150
Int. J. Nav. Archit. Ocean Eng. (2014) 6:1160~1181 1171

Figs. 11 and 12 compare the time series and power spectral densities (PSD) of the VBM and TM second-order springing for the
different backbone models. Typical second-order springing responses were observed for both vertical bending and torsion. The
time signals for the H and U backbones are very similar in shape and magnitude for vertical bending springing. The frequency
distributions also exhibit almost identical patterns along the whole backbone. The time signals of the H and U backbones also
exhibit similar trends in the case of torsion springing. However, for both torsion springing signals, noticeable differences can be
seen in the locations and magnitudes of the peaks in the case of vertical bending because the two backbones have different natural
frequencies and shear centers. The section shapes of the backbones have important effects on the pattern of second-order torsion
springing because the torsion pattern is sensitive to the section shape. An interesting finding with respect to the torsion springing is
that third-, fourth-, and higher-mode frequency components were observed, especially at bow locations, in the case of second-order
springing, while vertical bending second-order springing did not exhibit such higher-mode behaviors. This can be explained by the
results of the hammering tests, in which multiple peaks were observed for the torsion mode. Nonlinear hydrodynamic effects can
also be expected in the bow region because of the drastic change of the section shape with respect to the water line.

4.0E+06 2.0E+13
Exp (H backbone) Exp (H backbone)
Exp (U backbone) Exp (U backbone)
2.0E+06 1.5E+13
My1 (kN-m)

My1 (PSD)
0.0E+00 1.0E+13

-2.0E+06 5.0E+12

-4.0E+06 0.0E+00
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
t (s) ω (Hz)

(a) Time series & PSD of My1.

4.0E+06 8.0E+13
Exp (H backbone) Exp (H backbone)
Exp (U backbone) Exp (U backbone)
2.0E+06 6.0E+13
My2 (kN-m)

My2 (PSD)

0.0E+00 4.0E+13

-2.0E+06 2.0E+13

-4.0E+06 0.0E+00
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
t (s) ω (Hz)

(b) Time series & PSD of My2.

4.0E+06 8.0E+13
Exp (H backbone) Exp (H backbone)
Exp (U backbone) Exp (U backbone)
2.0E+06 6.0E+13
My3 (kN-m)

My3 (PSD)

0.0E+00 4.0E+13

-2.0E+06 2.0E+13

-4.0E+06 0.0E+00
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
t (s) ω (Hz)

(c) Time series & PSD of My3.

4.0E+06 2.0E+13
Exp (H backbone) Exp (H backbone)
Exp (U backbone) Exp (U backbone)
2.0E+06 1.5E+13
My4 (kN-m)

My4 (PSD)

0.0E+00 1.0E+13

-2.0E+06 5.0E+12

-4.0E+06 0.0E+00
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
t (s) ω (Hz)

(d) Time series & PSD of My4.

4.0E+06 2.0E+12
Exp (H backbone) Exp (H backbone)
Exp (U backbone) Exp (U backbone)
2.0E+06 1.5E+12
My5 (kN-m)

My5 (PSD)

0.0E+00 1.0E+12

-2.0E+06 5.0E+11

-4.0E+06 0.0E+00
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
t (s) ω (Hz)

(e) Time series & PSD of My5.

Fig. 11 Comparison of VBM second-order springing in head sea (H = 5 m, Tw = 8.2 s, Vs = 19 kts).
1172 Int. J. Nav. Archit. Ocean Eng. (2014) 6:1160~1181

Higher-order springing behaviors in the vertical bending mode are shown up to the fifth order in Figs. 14 and 15 for various
wave, heading, and ship speed conditions. As summarized in Table 4, greater wave heights were applied for excitation of
higher-order springing. Second-order springing was observed even for a relatively small wave height of 1 m. This implies that
second-order springing in the vertical bending mode can be easily excited in moderate sea states. The magnitude of the second-
order springing is proportional to the square of the wave amplitude, as shown in Fig. 13. Third- or higher-order springing of
vertical bending were clearly observed for various combinations of wave and ship speeds, but the wave height becomes larger
as the occurrence order increases. This means that not only second-order springing but also a number of higher orders of
springing are possible in irregular sea conditions, especially in high sea states. Miyake et al. (2009) demonstrated the possibility
of higher-order springing in irregular waves experimentally.

1.0E+06 2.0E+12
ExpC (H backbone) ExpC (H backbone)
ExpC (U backbone) ExpC (U backbone)
5.0E+05 1.5E+12
Mx1 (kN-m)

Mx1 (PSD)
0.0E+00 1.0E+12

-5.0E+05 5.0E+11

-1.0E+06 0.0E+00
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
t (s) ω (Hz)

(a) Time series & PSD of Mx1.

1.0E+06 2.0E+12
ExpC (H backbone) ExpC (H backbone)
ExpC (U backbone) ExpC (U backbone)
5.0E+05 1.5E+12
Mx2 (kN-m)

Mx2 (PSD)

0.0E+00 1.0E+12

-5.0E+05 5.0E+11

-1.0E+06 0.0E+00
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
t (s) ω (Hz)

(b) Time series & PSD of Mx2.

1.0E+06 2.0E+12
ExpC (H backbone) ExpC (H backbone)
ExpC (U backbone) ExpC (U backbone)
5.0E+05 1.5E+12
Mx3 (kN-m)

Mx3 (PSD)

0.0E+00 1.0E+12

-5.0E+05 5.0E+11

-1.0E+06 0.0E+00
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
t (s) ω (Hz)

(c) Time series & PSD of Mx3.

1.0E+06 1.0E+12
ExpC (H backbone) ExpC (H backbone)
ExpC (U backbone) ExpC (U backbone)
5.0E+05 7.5E+11
Mx4 (kN-m)

Mx4 (PSD)

0.0E+00 5.0E+11

-5.0E+05 2.5E+11

-1.0E+06 0.0E+00
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
t (s) ω (Hz)

(d) Time series & PSD of Mx4.

1.0E+06 2.0E+11
ExpC (H backbone) ExpC (H backbone)
ExpC (U backbone) ExpC (U backbone)
5.0E+05 1.5E+11
Mx5 (kN-m)

Mx5 (PSD)

0.0E+00 1.0E+11

-5.0E+05 5.0E+10

-1.0E+06 0.0E+00
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
t (s) ω (Hz)

(e) Time series & PSD of Mx5.

Fig. 12 Comparison of TM second-order springing in oblique sea
(H = 5 m, heading = 150 deg, TM2H-5 vs. TM2U-5).
Int. J. Nav. Archit. Ocean Eng. (2014) 6:1160~1181 1173

8.0E+05 1.6E+12
Exp (H backbone) Exp (H backbone)

4.0E+05 1.2E+12
M y3 (kN-m)

M y3 (PSD )
0.0E+00 8.0E+11

-4.0E+05 4.0E+11

-8.0E+05 0.0E+00
0 5 10 15 20 25 30 35 40 45 50 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
t (s) ω (Hz)

(a) Time series & PSD of My3 (H = 1 m)

2.0E+06 1.0E+13
Exp (H backbone) Exp (H backbone)

1.0E+06 7.5E+12
M y3 (kN-m )

M y3 (PSD)
0.0E+00 5.0E+12

-1.0E+06 2.5E+12

-2.0E+06 0.0E+00
0 5 10 15 20 25 30 35 40 45 50 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
t (s) ω (Hz)

(b) Time series & PSD of My3 (H = 3 m).

4.0E+06 6.0E+13
Exp (H backbone) Exp (H backbone)

2.0E+06 4.5E+13
M y3 (kN -m )

M y3 (PSD)

0.0E+00 3.0E+13

-2.0E+06 1.5E+13

-4.0E+06 0.0E+00
0 5 10 15 20 25 30 35 40 45 50 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
t (s) ω (Hz)

(c) Time series & PSD of My3 (H = 5 m).

Fig. 13 Comparison of vertical bending second-order springing
in head sea for wave heights (Tw = 8.2 s, Vs = 19 kts).

2.0E+06 6.0E+13
Exp (H backbone) Exp (H backbone)

1.0E+06 4.5E+13
M y3 (kN-m )

M y3 (PSD)

0.0E+00 3.0E+13

-1.0E+06 1.5E+13

-2.0E+06 0.0E+00
0 5 10 15 20 25 30 35 40 45 50 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
t (s) ω (Hz)

(a) Time series & PSD of My3 (H = 3 m).

4.0E+06 1.6E+14
Exp (H backbone)
Exp (H backbone)
2.0E+06 1.2E+14
M y3 (kN-m)

M y3 (PSD)

0.0E+00 8.0E+13

-2.0E+06 4.0E+13

-4.0E+06 0.0E+00
0 5 10 15 20 25 30 35 40 45 50 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
t (s) ω (Hz)

(b) Time series & PSD of My3 (H = 5 m).

Fig. 14 Comparison of VBM third-order springing in head sea
for various wave heights (Tw = 10.93 s, Vs = 18.5 kts).
1174 Int. J. Nav. Archit. Ocean Eng. (2014) 6:1160~1181

4.0E+06 6.0E+13
Exp (H backbone) Exp (H backbone)

2.0E+06 4.5E+13
My3 (kN-m)

M y3 (PSD)
0.0E+00 3.0E+13

-2.0E+06 1.5E+13

-4.0E+06 0.0E+00
0 5 10 15 20 25 30 35 40 45 50 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
t (s) ω (Hz)

(a) Time series & PSD of My3 (H = 3 m, wave heading = 150 deg, Tw = 13.7 s, Vs = 20 kts).

2.0E+07 6.0E+14
Exp (H backbone) Exp (H backbone)

1.0E+07 4.5E+14
My3 (kN-m)

M y3 (PSD)
0.0E+00 3.0E+14

-1.0E+07 1.5E+14

-2.0E+07 0.0E+00
0 5 10 15 20 25 30 35 40 45 50 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
t (s) ω (Hz)

(b) Time series & PSD of My3 (H = 6 m, head sea, Tw = 14.3 s, Vs = 20.3 kts).

2.0E+07 1.6E+15
Exp (H backbone) Exp (H backbone)

1.0E+07 1.2E+15
My3 (kN-m)

M y3 (PSD)

0.0E+00 8.0E+14

-1.0E+07 4.0E+14

-2.0E+07 0.0E+00
0 5 10 15 20 25 30 35 40 45 50 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
t (s) ω (Hz)

(c) Time series & PSD of My3 (H = 9 m, head sea, Tw = 12.8 s, Vs = 15 kts).

2.0E+07 1.2E+15
Exp (H backbone) Exp (H backbone)

1.0E+07 9.0E+14
My3 (kN-m)

M y3 (PSD)

0.0E+00 6.0E+14

-1.0E+07 3.0E+14

-2.0E+07 0.0E+00
0 5 10 15 20 25 30 35 40 45 50 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
t (s) ω (Hz)

(d) Time series & PSD of My3 (H = 9 m, head sea, Tw = 15.4 s, Vs = 15 kts).
Fig. 15 Comparison of VBM fourth and fifth-order springing for various wave conditions.

In Fig. 16, second-order torsional springing behavior is compared for various wave heights at midship. As Fig. 17 shows,
springing up to the fourth order was observed for the torsion mode, but fifth-order springing was not clearly observed for wave
heights up to 9 m.
A comparison of the results of the hammering tests and springing tests for the two backbones shows that almost identical
vertical bending whipping test results were obtained, as shown in Fig. 18. Typical characteristics of a whipping signal are shown
for the foremost bow section. These characteristics include higher harmonics up to the ninth order. Fig. 19 shows an example
for a case of torsion resonant whipping, which is similar to the case of third-order springing, although the wave height must be
much higher to excite an impact load. This case was previously discussed by Hong et al. (2012). It is interesting to observe
noticeable peaks at multiple torsion natural modes, which corresponds to the hammering tests results along the whole span.
Both the first and second torsion modes are dominant near the bow and stern locations. At location S5, higher harmonics are
dominant in comparison with the first torsion mode response. This indicates that the impact response at the bow is governed by
nonlinearity of hydrodynamics as well as by structural properties. It seems that the first- and second-mode torsion natural
frequencies occur near the multiples of the excitation frequency in this case and that the magnitude of the second-mode torsion
excitation is larger than that of the first mode. These results also support the existence of higher-mode torsion in model tests.
Int. J. Nav. Archit. Ocean Eng. (2014) 6:1160~1181 1175

1.0E+05 3.2E+10
ExpC (H backbone) ExpC (H backbone)

5.0E+04 2.4E+10
Mx3 (kN-m)

Mx3 (PSD)
0.0E+00 1.6E+10

-5.0E+04 8.0E+09

-1.0E+05 0.0E+00
0 5 10 15 20 25 30 35 40 45 50 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
t (s) ω (Hz)

(a) Time series & PSD of Mx3 (H = 1 m).

4.0E+05 2.8E+11
ExpC (H backbone) ExpC (H backbone)

2.0E+05 2.1E+11
Mx3 (kN-m)

Mx3 (PSD)
0.0E+00 1.4E+11

-2.0E+05 7.0E+10

-4.0E+05 0.0E+00
0 5 10 15 20 25 30 35 40 45 50 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
t (s) ω (Hz)

(b) Time series & PSD of Mx3 (H = 3 m).

1.0E+06 2.0E+12
ExpC (H backbone) ExpC (H backbone)

5.0E+05 1.5E+12
Mx3 (kN-m)

Mx3 (PSD)
0.0E+00 1.0E+12

-5.0E+05 5.0E+11

-1.0E+06 0.0E+00
0 5 10 15 20 25 30 35 40 45 50 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
t (s) ω (Hz)

(c) Time series & PSD of Mx3 (H = 5 m).

Fig. 16 Comparison of torsion second-order springing in oblique sea (Tw = 10.3 s, Vs = 17.5 kts).

3.0E+05 1.0E+11
ExpC (H backbone) ExpC (H backbone)

1.5E+05 7.5E+10
Mx3 (kN-m)

Mx3 (PSD)

0.0E+00 5.0E+10

-1.5E+05 2.5E+10

-3.0E+05 0.0E+00
0 5 10 15 20 25 30 35 40 45 50 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
t (s) ω (Hz)

(a) Time series & PSD of Mx3 (H = 3 m, third-order springing).

4.0E+05 2.0E+11
ExpC (H backbone) ExpC (H backbone)

2.0E+05 1.5E+11
Mx3 (kN-m)

Mx3 (PSD)

0.0E+00 1.0E+11

-2.0E+05 5.0E+10

-4.0E+05 0.0E+00
0 5 10 15 20 25 30 35 40 45 50 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
t (s) ω (Hz)

(b) Time series & PSD of Mx3 (H = 5 m, third-order springing).

4.0E+05 2.4E+11
ExpC (H backbone) ExpC (H backbone)

2.0E+05 1.8E+11
Mx3 (kN-m)

Mx3 (PSD)

0.0E+00 1.2E+11

-2.0E+05 6.0E+10

-4.0E+05 0.0E+00
0 5 10 15 20 25 30 35 40 45 50 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
t (s) ω (Hz)

(c) Time series & PSD of Mx3 (H = 9 m, Tw = 16.2 s, Vs = 10 kts, ωe = 4ωtn ).

4.0E+05 2.0E+11
ExpC (H backbone) ExpC (H backbone)

2.0E+05 1.5E+11
Mx3 (kN-m)

Mx3 (PSD)

0.0E+00 1.0E+11

-2.0E+05 5.0E+10

-4.0E+05 0.0E+00
0 5 10 15 20 25 30 35 40 45 50 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
t (s) ω (Hz)

(d) Time series & PSD of Mx3 (H = 9 m, Tw = 18.7 s, Vs = 5 kts, ωe = 5ωtn ).

Fig. 17 Comparison of torsion third-, fourth-, and fifth-order springing for various wave conditions (heading = 150 deg).
1176 Int. J. Nav. Archit. Ocean Eng. (2014) 6:1160~1181

2.0E+07 2.0E+14
Exp (H backbone) Exp (H backbone)
Exp (U backbone) Exp (U backbone)
1.0E+07 1.5E+14
M y5 (kN-m)

M y5 (PSD)
0.0E+00 1.0E+14

-1.0E+07 5.0E+13

-2.0E+07 0.0E+00
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
t (s) ω (Hz)

(a) Time series and PSD of My5.

Fig. 18 Comparison of vertical bending whipping for two backbones (H = 12 m, Tw = 14.3 s, Vs = 18 kts).

2.0E+06 4.0E+12
ExpC (H backbone) ExpC (H backbone)

1.0E+06 3.0E+12
Mx1 (kN-m)

M x1 (PSD)
0.0E+00 2.0E+12

-1.0E+06 1.0E+12

-2.0E+06 0.0E+00
0 5 10 15 20 25 30 35 40 45 50 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
t (s) ω (Hz)

(a) Time series & PSD of Mx1.

2.0E+06 6.0E+12
ExpC (H backbone) ExpC (H backbone)

1.0E+06 4.5E+12
Mx2 (kN-m)

M x2 (PSD)

0.0E+00 3.0E+12

-1.0E+06 1.5E+12

-2.0E+06 0.0E+00
0 5 10 15 20 25 30 35 40 45 50 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
t (s) ω (Hz)

(b) Time series & PSD of Mx2.

2.0E+06 8.0E+12
ExpC (H backbone) ExpC (H backbone)

1.0E+06 6.0E+12
Mx3 (kN-m)

M x3 (PSD)

0.0E+00 4.0E+12

-1.0E+06 2.0E+12

-2.0E+06 0.0E+00
0 5 10 15 20 25 30 35 40 45 50 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
t (s) ω (Hz)

(c) Time series & PSD of Mx3.

2.0E+06 4.0E+12
ExpC (H backbone) ExpC (H backbone)

1.0E+06 3.0E+12
Mx4 (kN-m)

Mx4 (PSD)

0.0E+00 2.0E+12

-1.0E+06 1.0E+12

-2.0E+06 0.0E+00
0 5 10 15 20 25 30 35 40 45 50 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
t (s) ω (Hz)

(d) Time series & PSD of Mx4.

2.0E+06 2.0E+12
ExpC (H backbone) ExpC (H backbone)

1.0E+06 1.5E+12
Mx5 (kN-m)

M x5 (PSD)

0.0E+00 1.0E+12

-1.0E+06 5.0E+11

-2.0E+06 0.0E+00
0 5 10 15 20 25 30 35 40 45 50 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
t (s) ω (Hz)

(e) Time series & PSD of Mx5.

Fig. 19 Resonant torsion whipping (H = 12 m, Tw = 14.3 s, Vs = 20.3 kts, heading = 150 deg).
Int. J. Nav. Archit. Ocean Eng. (2014) 6:1160~1181 1177

Springing in irregular waves

Because springing and whipping signals are mixed under irregular sea states, it is not easy to separate the effects of springing
and whipping, especially in high sea states. As the vertical bending and torsion natural frequencies lie inside the encounter wave
frequency in the case of an ultra-large container ship, experimental investigation of the contribution of linear springing becomes
an important concern with respect to the validation purpose of model test data. For this purpose, the following model tests in
irregular head waves were designed. The basic idea was to compare the springing responses obtained from two wave spectra,
one of which is a complete wave spectrum and the other of which is the same wave but with the high-frequency portion truncated
so that linear springing is avoided. Those wave spectra are shown in Fig. 20. The target wave spectrum is an ITTC 2-parameter
spectrum with Hs = 5 m and Tp = 9 seconds. The green line represents the complete wave spectrum, and the red line represents
the truncated wave spectrum. To investigate the amplitude dependency, the same wave signals were applied with half of the stroke
gains. The wave heading was set to be head waves, and the ship speed was set to 20 knots. To make a 20-minute-long time record,
a total of eight runs were made for one set of data in irregular waves. To avoid repetition of time signals, the starting time of
each run was shifted, and the same conditions were applied to the corresponding data set for different wave spectrum conditions.
Fig. 21 summarizes the test results for the case of Hs = 5 m and Tp = 9 s and the corresponding truncated wave spectrum.
Two wave spectra were compared on the basis of the encounter wave frequency, and as the figure shows, there are no linear
springing excitations in the truncated spectrum. The vertical bending moment at midship is shown for eight runs. The black
lines denote the full-spectrum case, and the red lines denote the truncated-spectrum case. Almost identical results were obtained
for the linear wave frequency responses, but different types of vertical bending responses were obtained for the springing
frequency. There are three categories of springing responses: the same spring responses for two different wave spectra, larger
springing responses for the full wave spectrum, and larger springing responses for the truncated wave spectrum. These three
categories of springing responses can be interpreted in view of linear and nonlinear springing responses. The first category
corresponds to the case in which the same springing responses were obtained for both the full and truncated wave spectra, as
illustrated in Figs. 21 (c), (e), (h), and (i). For those cases, we can see that the linear bending responses are almost the same and
that the coincidence in peaks is especially prominent. This nonlinear-springing-dominant case is the result of second- and third-
order springing excitations induced by high peaks that occur at frequencies of one half and one third of the bending natural
frequency. The second category corresponds to the case in which larger springing responses were obtained for the full wave
spectrum, as illustrated in Figs. 21(b), (f), and (g). We can see higher linear response peaks in the full spectrum case, which
resulted in larger higher-order springing for the full wave spectrum. The third category corresponds to the case in which larger
springing responses were obtained for the truncated wave spectrum, as illustrated in Fig. 21(d). We can see noticeably higher
wave frequency responses in the truncated wave spectrum case, which contributed to the second- and third-order springing
excitations. These results suggest that nonlinear springing effects are significant in the case of Hs = 5 m and Tp = 9 s.
Considering the wave frequency ranges, second-order springing seems to contribute mainly to the springing response. The
levels of the wave frequency response at one half and one third of the natural frequency are strongly correlated to the springing
response for both the full and truncated spectra. Fig. 21(j) shows the average spectra for eight runs in which the full spectrum
response and the truncated spectrum response are almost same for both the wave frequency and springing frequency regions.

35
No cut
30
Cut-off
25
Wave (PSD)

20

15

10

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

ωe (Hz)
Fig. 20 ITTC two-parameter wave spectrum (Hs = 5 m, Tp = 9 s).
1178 Int. J. Nav. Archit. Ocean Eng. (2014) 6:1160~1181

35 1E+13 1E+13
No cut No cut No cut
30
Cut-off 8E+12 Cut-off 8E+12 Cut-off
25
ωpe=0.19 Hz
Wave (PSD)

6E+12 6E+12

My3 (PSD)

M y3 (PSD)
20
ωbn=0.43 Hz
15 4E+12 4E+12

10
2E+12 2E+12
5

0 0 0

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
ωe (Hz) ωe (Hz) ωe (Hz)

(a) PSD of wave. (b) PSD of My3 (run 1). (c) PSD of My3 (run 2).

1E+13 1E+13 1E+13

No cut No cut No cut
8E+12 Cut-off 8E+12 Cut-off 8E+12 Cut-off

6E+12 6E+12 6E+12

My3 (PSD)

My3 (PSD)

My3 (PSD)
4E+12 4E+12 4E+12

2E+12 2E+12 2E+12

0 0 0

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
ωe (Hz) ωe (Hz) ωe (Hz)

(d) PSD of My3 (run 3). (e) PSD of My3 (run 4). (f) PSD of My3 (run 5).

1E+13 1E+13 1E+13

No cut No cut No cut
8E+12 Cut-off 8E+12 Cut-off 8E+12 Cut-off

6E+12 6E+12 6E+12

My3 (PSD)

My3 (PSD)

My3 (PSD)

4E+12 4E+12 4E+12

2E+12 2E+12 2E+12

0 0 0

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
ωe (Hz) ωe (Hz) ωe (Hz)

(g) PSD of My3 (run 6). (h) PSD of My3 (run 7). (i) PSD of My3 (run 8).

1E+13
No cut
8E+12 Cut-off

6E+12
My3 (PSD)

4E+12

2E+12

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

ωe (Hz)

(j) PSD of My3 (average).

Fig. 21 VBM springing in irregular waves (Hs = 5 m, Tp = 9 s, heading = 180 deg, speed = 20 kts).
Int. J. Nav. Archit. Ocean Eng. (2014) 6:1160~1181 1179

10 3E+12 3E+12
No cut No cut No cut
8 Cut-off 2.5E+12 Cut-off 2.5E+12 Cut-off

2E+12 2E+12
Wave (PSD)

6 ωpe=0.19 Hz

My3 (PSD)
M y3 (PSD)
ωbn=0.43 Hz 1.5E+12 1.5E+12
4
1E+12 1E+12

2
5E+11 5E+11

0 0 0

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
ωe (Hz) ωe (Hz) ωe (Hz)

(a) PSD of wave. (b) PSD of My3 (run 1). (c) PSD of My3 (run 2).

3E+12 3E+12 3E+12

No cut No cut No cut
2.5E+12 Cut-off 2.5E+12 Cut-off 2.5E+12 Cut-off

2E+12 2E+12 2E+12

My3 (PSD)

My3 (PSD)

My3 (PSD)
1.5E+12 1.5E+12 1.5E+12

1E+12 1E+12 1E+12

5E+11 5E+11 5E+11

0 0 0

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
ωe (Hz) ωe (Hz) ωe (Hz)

(d) PSD of My3 (run 3). (e) PSD of My3 (run 4). (f) PSD of My3 (run 5).

3E+12 3E+12 3E+12

No cut No cut No cut
2.5E+12 Cut-off 2.5E+12 Cut-off 2.5E+12 Cut-off

2E+12 2E+12 2E+12

My3 (PSD)

My3 (PSD)

My3 (PSD)

1.5E+12 1.5E+12 1.5E+12

1E+12 1E+12 1E+12

5E+11 5E+11 5E+11

0 0 0

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
ωe (Hz) ωe (Hz) ωe (Hz)

(g) PSD of My3 (run 6). (h) PSD of My3 (run 7). (i) PSD of My3 (run 8).

3E+12
No cut
2.5E+12 Cut-off

2E+12
My3 (PSD)

1.5E+12

1E+12

5E+11

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

ωe (Hz)

(j) PSD of My3 (average).

Fig. 22 VBM springing in irregular waves (Hs = 2.5 m, Tp = 9 s, heading = 180 deg, speed = 20 kts).
1180 Int. J. Nav. Archit. Ocean Eng. (2014) 6:1160~1181

Fig. 22 shows the results for the case of Hs = 2.5 m and Tp = 9 s and the corresponding truncated wave spectrum. These
waves were generated with a half gain stroke of the Hs = 5 m spectra. The wave frequency response spectra are almost identical
in form to those for Hs = 5 m, but noticeable differences are observed for the springing response spectra. Second-order
springing seems to play a role, but the linear springing contribution dominates the springing response. Typical examples are
shown in Figs. 22(c) and (f)-(h). Almost identical wave frequency responses are observed for both the full and truncated spectra,
and second-order springing is anticipated from the frequency distributions of the vertical bending response, although no
noticeable springing response appears in the truncated spectrum. These results can be explained by linear springing resonance
from full spectrum excitation. Second-order springing excitation can be considered negligible in comparison with linear springing
excitation because the waves are too small to excite second-order springing. Fig. 22(j) shows a typical example illustrating the
importance of linear springing excitations for low sea states.

CONCLUSIONS
Important aspects of springing and whipping of large container ships were investigated in this study through a set of model
tests conducted as part of the WILS-JIP project, the purpose of which is to provide reliable validation model test data on global
wave loads, springing, and whipping. Nonlinear effects on springing responses and multiple higher modes of torsion whipping
were analyzed in this study through a set of higher-order springing tests, resonant and non-resonant whipping tests, and truncated
wave spectra tests.
Higher-mode torsion behaviors were experimentally verified in both hammering tests and resonant whipping tests. Higher-
order torsion springing was observed up to fourth-order nonlinearity. For this purpose, H-section and U-section backbone
modeling were compared in detail. Nearly equivalent model test results were obtained for vertical bending, but some different
response patterns were identified in torsion mode. Noticeably higher mode responses were observed in hammering tests and
higher-order springing tests, especially at bow locations, where hydrodynamic nonlinearity is significant. In the resonant torsion
whipping case, multiple-peak responses near the torsion natural mode frequencies was observed.
Higher-order vertical bending springing was observed up to the fifth order, while higher-order torsion springing was observed
up to the fourth order. These results suggest that not only second-order springing but also higher-order effects, up to the fifth
order, should be carefully considered in analysis of vertical bending moment and torsion moment. Through truncated spectra
model tests, it was experimentally demonstrated that second-order springing is of practical importance in moderately high sea
states, such as Hs = 5 m, but that linear springing resonance plays a dominant role in overall springing in relatively mild seas,
such as Hs = 2.5 m. The possibility of higher-order springing was also observed in the case of Hs = 5 m. It could be concluded
that multiple sources of springing excitations are identified in irregular sea states depending on the frequency distribution of
wave energy and its intensity.

AKNOWLEDGMENTS
This research was conducted as part of WILS JIP-II & III. The support of WILS participants Daewoo Shipbuilding & Marine
Engineering, Hanjin TMS, Samsung Heavy Industries, the American Bureau of Shipping, Bureau Veritas, Germanischer Lloyd,
the Korean Register of Shipping, Lloyd’s Register, Class NK, Seoul National University, and KRISO are deeply appreciated.
The authors would like to express their special thanks to Dr. Sime for fruitful suggestions and discussions during the research.

REFERENCES
Bigot, F., Derbanne, Q., Sireta, F.X. and Malenica, S., Tuitman, J.T., 2011. Global hydroelastic ship response comparison
of numerical model and WILS model test. Proceeding of 21st International Offshore and Polar Engineering Confe-
rence, Maui, USA, 19-24 June 2011, pp.477-485.
Hong, S.Y., Kyoung, J.H., Kim, Y.S., Song, K.H., Kim, S., Malenica, S., Lindgren, M., Rathje, H. and GE, C., 2008. Vali-
dation of wave loads on a large container ship in oblique waves. Proceeding The 6th Osaka Colloquium on Seakeeping
and stability of ships, Osaka, Japan, 26-29 March 2008, pp.109-117.
Int. J. Nav. Archit. Ocean Eng. (2014) 6:1160~1181 1181

Hong, S.Y., Kim, B.W. and Nam, B.W., 2011. Experimental study on torsion springing and whipping of large container
ship. Proceeding of 21st International Offshore and Polar Engineering Conference, Maui, USA, 19-24 June 2011,
pp.97-107.
Hong, S.Y., Kim, B.W. and Nam, B.W., 2012. Experimental study on torsion springing and whipping of large container
ship. International Journal of Offshore and Polar Engineering, 22(2), pp.97-107.
Iijima, K., Hermundstad, O.A., Zhu, S. and Moan, T., 2009. Symmetric and antisymmetric vibrations of a hydroelastically
scaled model. Proceedings of 5th International Conference on Hydroelasticity in Marine Technology, Southampton,
UK, 6-10 September 2009, pp.173-182.
Kim, B.W., Kim, K.H., Kim, Y.S. and Hong, S.Y., 2014. Torsion moment conversion methods in model test with U-shape
backbone. Proceefing of the 24th International Ocean and Polar Engineering Conference, Busan, Korea, 15-20 June
2014, pp.782-791.
Miyake, R., Matsumoto, T., Zhu, T., Usami, A. and Dobashi, H., 2009. Experimental studies on the hydroelastic response
using a flexible mega-container ship model. Proceeding of 5th International Conference on Hydroelasticity in Marine
Technology, Southampton, UK, 6-10 September 2009, pp.161 -171.
Noran Engineering, Inc., 2011. User's manual and reference manual of NE/NASTRAN. USA: Noran Engineering, Inc.
Oka, M., Oka, S. and Ogawa, Y., 2009. An experimental study on wave loads of a large container ship and its hydroelastic
vibration. Proc 5th Int Conf on Hydroelasticity in Marine Technology, Southampton, UK, 6-10 September 2009, pp.
183-192.
Oberhagemann, J., Moctar, O. and Rathje, H., 2010. Fluid-structure interaction using free-surface RANSE for springing
and whipping. Proceeding of 13th Numerical Towing Tank Symposium, Duisburg, Germany, 10-12 October 2010.
Remy, F., Molin, B. and Ledoux, A., 2006. Experimental and numerical study of the wave response of a flexible barge.
Proceeding of 4th International Conference on Hydroelasticity in Marine Technology, Wuxi, China, 10-14 September
2006, pp.255-264.
Storhaug, G. and Moan, T., 2006. Spring/whipping response of a large ocean-going vessel-investigated by an experimental
method. Proc 4th Int Conf on Hydroelasticity in Marine Technology, Wuxi, China, 10-14 September 2006, pp.89-101.