J. Marine Sci. Appl.

(2016) 15: 50-62

DOI: 10.1007/s11804-016-1339-9

Effect of Mass Ratio on

Hydrodynamic Response of a Flexible Cylinder
Haoyang Cen, Rupp Carriveau* and David S-K Ting
Turbulence and Energy Laboratory, Centre for Engineering Innovation, University of Windsor, Ontario N9B 3P4, Canada

Abstract: The effect of the mass ratio on the flow-induced
vibration (FIV) of a flexible circular cylinder is experimentally
investigated in a towing tank. A Tygon tube with outer and inner
diameters of 7.9 mm and 4.8 mm, respectively, was employed for
the study. The tube was connected to a carriage and towed from rest
to a steady speed up to 1.6 m/s before slowing down to rest again
over a distance of 1.6 m in still water. Reynolds number based on
the cylinders outer diameter was 80013,000, and the reduced
velocity (velocity normalized by the cylinders natural frequency
and outer diameter) spanned from 2 to 25. When connected, the
cylinder was elongated from 420 mm to 460 mm under an axial
pre-tension of 11 N. Based on the cylinders elongated length, the
aspect ratio (ratio of the cylinders length to outer diameter) was
calculated as 58. Three mass ratios (ratio of the cylinders structural
mass to displaced fluid mass, m*) of 0.7, 1.0, and 3.4 were
determined by filling the cylinders interior with air, water, and
alloy powder (nickel-chromium-boron matrix alloy), respectively.
An optical method was adopted for response measurements.
Multi-frequency vibrations were observed in both in-line (IL) and
cross-flow (CF) responses; at high Reynolds number, vibration
modes up to the 3rd one were identified in the CF response. The
mode transition was found to occur at a lower reduced velocity for
the highest tested mass ratio. The vibration amplitude and
frequency were quantified and expressed with respect to the
reduced velocity. A significant reduced vibration amplitude was
found in the IL response with increasing mass ratios, and only
initial and upper branches existed in the IL and CF response
amplitudes. The normalized response frequencies were revealed to
linearly increase with respect to the reduced velocity, and slopes for
linear relations were found to be identical for the three cases tested.
Keywords: flow-induced vibration, low mass ratio, flexible
cylinder, multi-frequency response, hydrodynamic response
Article ID: 1671-9433(2016)01-0050-13
1 Introduction1
The dynamics of cylindrical structures subjected to cross
flow (CF) have received significant attention. This to a great
extent is because of their engineering importance, particularly
in offshore engineering where they are extensively deployed
in CF as deep water exploration pipes, oil production risers,
catenaries, and so on. However, external forces exerted by
nearby fluid flow inevitably cause these cylindrical structures
to oscillate in both stream-wise (in-line, IL) and transverse
(CF) directions; this phenomenon is termed as flow-induced
vibration (FIV) (Blevins, 2001). By increasing the dynamic
load on offshore structures, FIV can be a significant source of
fatigue damage or premature failure. Therefore, a
comprehensive understanding of underlying dynamic
interaction mechanisms is important to enable a general
prediction of FIV occurrence and probable response
amplitude and frequency so that corresponding approaches for
mitigation can be established for offshore engineering
designs.
1.1 Elastically mounted rigid circular cylinders
Research in this field began with an analysis of elastically
mounted rigid cylinders with one or two degrees of freedom
(DOFs), and most scientific interest has been focused on
fundamental characteristics such as dynamic forces,
vortex-shedding mode, lock-in region, and effects of surface
roughness and mass ratio. The current research situation and
associated comprehensive reviews are provided by
Williamson and Govardhan (2004; 2008), Sarpkaya (1979;
2004), Gabbai and Benaroya (2005), and Bearman (1984;
2010), among others. These reviews suggest that in terms of
a cylinders dynamic response, the mass ratio is an
influential parameter. It is well recognized that a rigid
cylinders amplitude response is classified into three main
categories depending on the value of the mass ratio (see Fig.
1). For example, a cylinder with a high mass ratio vibrating
in a CF direction typically displays two branches of a
response amplitude: initial and lower. This was first
reported in a study by Feng (1968) using a mass ratio of 248,
and in addition, an abrupt jump and hysteretic transition
between these two branches were presented, where the jump
in a response amplitude was accompanied by a jump in the
phase of pressure fluctuation. This was later revealed to be
associated with a transition in the wake pattern from the 2S
mode (two single vortices shed per cycle, Fig. 2(a) to 2P
mode (two vortex pairs shed per cycle, Fig. 2(b), i.e., 2S
corresponded to the initial branch and 2P to the lower one
(Williamson and Roshko, 1988; Brika and Laneville, 1993;
Govardhan and Williamson, 2000).

Received date: 2015-09-23

Accepted date: 2015-10-30
*Corresponding author email: rupp@uwindsor.ca
Harbin Engineering University and Springer-Verlag Berlin Heidelberg 2016
Journal of Marine Science and Application (2016) 15: 50-62 51

evident with a decrease in the mass ratio. They concluded

that in the cases of mass ratios above five or six, the added
freedom had limited influence on the CF response; however,
once mass ratios fell below this value, the rigid cylinders
response was drastically modified: the upper branch was
replaced by a new branch (denoted as super upper; see Fig.
1) characterized by 1.5 diameter amplitude and 2T
vortex-shedding mode that comprised a triplet of vortices
every half cycle (Fig. 2(d)). This trend persisted until it
reached the critical mass ratio of around 0.54, below which
a large amplitude vibration persisted over the tested range
above initial lock-in.

Fig. 1 Cross-flow vibration amplitude (A*) of an elastically

mounted cylinder versus reduced velocity, Ur,
demonstrating various branches and corresponding
vortex-shedding mode: data from Feng (1968) at mass
ratio m*=248 (open circle symbols), Khalak &
Williamson (1996) at m*=2.4 (solid square symbols),
and Jauvtis & Williamson (2004) at m*=2.6 (open
triangle symbols)

With descreased mass ratio, an elastically mounted

cylinder would experience an increase in its response
amplitude (Sumer and Fredsoe, 1997). In addition to
response amplitude, a new branch, i.e., the upper branch,
was found to exist between the initial and lower branches
with a large response amplitude (Khalak and Williamson,
1996; 1997; 1999; Govardhan and Williamson, 2000). A
new vortex-shedding mode similar to the 2P mode but has
two vortex pairs shedding per cycle, in which latter vortices
of each pair are significantly weaker than the first (Fig. 2(c)),
was revealed to correspond with the upper branch.
Beyond this, low mass ratios also make the added mass
comparable to the structural mass of the cylinder, which
includes any enclosed matter but excludes the hydrodynamic
mass and results in a non-negligible variation in the total
mass that would alter the natural frequency. Consequently,
significantly broader lock-in regions within which the Fig. 2 Vortex-shedding modes of an elastically mounted
vortex-shedding frequency departs from the Strouhal cylinder: (a), (b), and (c) show modes corresponding
frequency and synchronizes with the natural frequency are to initial, lower, and upper branches, respectively
common when the vibrating cylinder has low mass ratios (Govardhan & Williamson, 2000); (d) shows the
(Sumer and Fredsoe, 1997). mode corresponding to the super-upper branch
Further investigation on rigid cylinders was extended to (Williamson & Jauvtis, 2004)
two DOFs by Jauvtis and Williamson (2003; 2004). They
1.2 Flexible circular cylinders
performed a comprehensive study on a rigid cylinder
Studies of FIV have been extended to flexible cylinders in
oscillating in IL and CF directions, where the cylinder was
recent decades because of their extensive use in offshore
designed to have identical natural frequencies and mass
engineering, but such studies are not as extensive as those
ratios ranging from 1.5 to 25.0 in both directions. The added
related to rigid cylinders. Structural flexibility introduces the
freedom in the IL direction was found to have little
capacity to vibrate at high modes, and when this capacity is
influence on the CF vibration at a high mass ratio, i.e., the
considered concurrently with the added mass and damping
findings from one DOF scenarios remained strongly relevant
distribution while in motion, added complexities (such as
to the case of two DOFs. However, this influence became
52 Haoyang Cen, et al. Effect of Mass Ratio on Hydrodynamic Response of a Flexible Cylinder
single/multi-mode vibration and multi-frequency),
competition among modes, and the traveling of waves along
the cylinders length occur (Vandiver, 1983; Kim et al., 1986;
Chung, 1987; 1989). For a vibrating flexible cylinder,
analysis usually employs a cable-like or beam-like structure
as an accepted structural system; structures are determined
to be cables if tension dominates and beams if bending
rigidity dominates. Results from a study by Lee and Allen
(2010) show that the vibration frequency of a cylinder rises
with the flow speed for a tension-dominated structure but
does not rise significantly for a bending-rigidity-dominated
structure. In addition, the lock-in bandwidth is broad for
bending-dominated cases, indicating a weak association
between changes in vibration frequency and lock-in
bandwidth.
In a pioneering work involving flexible cylinders by
Vandiver (1993), experimental results from previous
research on flexible cylinders were consolidated to reveal
non-dimensional parameters that govern the phenomena of
vortex-induced vibration in flexible cylinders. The study
emphasized the significance of the mass ratio and indicated
that cylinders with low mass ratios present a broader lock-in
range than those with high mass ratios, a fact also observed
in rigid cylinder studies. However, because of challenges
involved in varying mass ratios during a set of tests for a
constructed instrumented flexible cylinder, systematic
studies conducted to ascertain the dependence of the
cylinders dynamic response on the mass ratio are still rare.
Chaplin et al. (2005b) performed a set of laboratory tests on
a tensioned riser with an aspect ratio of 468.5 and mass ratio
of approximately 3, which were made in step flow (35% of
the cylinders length was submerged). CF vibration modes
up to the 8th one were observed, and multi-mode responses
and contributions from several modes were found in all
responses at reduced velocities except in lower ones.
Huera-Huarte and Bearman (2009) performed a similar test
on a flexible cylinder with an aspect ratio of 94 and mass
ratio of 1.8 and reported maximum attainable amplitudes up
to 0.7 diameters in the CF direction and 0.3 diameters in the
IL direction. This was considered to be different from a
flexibly mounted rigid cylinder, where a response amplitude
of over 1.5 diameters was demonstrated (Jauvtis and
Williamson, 2004). In addition, Huera-Huarte and Bearman
found that response frequencies in a lock-in region
synchronized with the natural frequency of a responding
mode and reported the use of a Strouhal number of
approximately 0.16 to predict response frequencies, as
previously proposed by Chaplin et al. (2005a).
Working in line with their studies, Huera-Huarte et al.
(2014) recently reported a new response dataset for a
flexible cylinder involving the investigation of the dynamic
responses of two slightly different flexible cylinders (one
with a mass ratio of 1.1 and aspect ratio of 158 and the other
with a mass ratio of 2.7 and aspect ratio of 187). They
reached a conclusion that very low mass ratio had the effect
of leading to a large increase in flexible cylinders response
amplitude for both cross-flow and in-line direction. This
phenominone was observed in studies with elastically
mounted rigid cylinders as well. In addition, it appeared that
the overall response in very low mass ratios was contributed
from a larger number of modes. Unfortunately, no direct
comparison was made in the dynamic responses between the
two cylinders because of the fact that they were not
completely identical except for their mass ratios. To the best
of the authors knowledge, this represents the only
experimental work in literature involving flexible cylinders
with different mass ratios. It is thus considered that
additional research efforts and systematic investigations
relating to the effect of the mass ratio on a cylinders FIV
are required.
In the present study, a series of experimental tests is
performed to further investigate the FIV of a highly flexible
circular cylinder at especially low mass ratios. The study
focuses on revealing whether varying mass ratios below 6
would cause significant variations on the overall response
mode, amplitude, and frequency. Measurements are obtained
for the cylinders response to vortex excitation at high
modes in both the CF and IL directions.
2 Description of experiment
2.1 Towing tank
The towing tank is made of glass (transparent to
enable optical access) and has dimensions of 2.5 0.8
0.8 m (see Fig. 3). During the tests it was filled with
water up to a depth of 0.7 m.
Fig. 3 Sketch of overall experiment setup.
The overall experimental configuration is illustrated in
Fig. 3. A pneumatic rodless actuator was installed and fixed
horizontally along the tanks length, in the center of the
tanks width span, to drive the rigid frame that serves as the
towing carriage. The rigid frame was mounted firmly under
the actuator. The rigid frame itself introduces a high natural
frequency that is well beyond that of the cylinder, and
ideally does not interfere with its response. The horizontal
connection of the test model to the rigid frame was achieved
through fixed constraints (a pair of set screws at both ends),
Journal of Marine Science and Application (2016) 15: 50-62 53

elevating the cylinder at a height of 400 mm from the tanks ranging from about 0.1 m/s to 1.6 m/s. Accordingly, the
bottom; there was a gap of 300 mm between the cylinder Reynolds number based on its diameter spanned from 800 to
and the free surface. According to a numerical study of fluid 13 000, while the reduced velocities ranged from 1 to 25, as
flow past a circular cylinder in the study by Farrant et al. calculated from
(2001), it is reasonable to neglect both the effects of the tank
U
bottom and the free surface at such distances, as they are Ur (3)
around 50 and 40 times the cylinders diameter. When
fn D
holding the test cylinder horizonatally, the pair of screws where U is the towing speed, fn is the natural frequency of
extended approximately 15 mm into the cylinder in its axial test cylinder in its first vibration mode, and D is the
direction, and therefore also acted as two plugs at the ends cylinders outer diameter. It should be noted that fn will alter
to help seal the desired materials inside the cylinder. This as a result of the varying mass ratio; therefore, cases with
towing apparatus carried the test cylinders through still different mass ratios have different values of Ur at the same
fluids over a total traveling distance of 1.6 m. During the towing speed. To maintain a uniform CF, a time period of at
traveling distance, the carriage accelerated rapidly from rest least 10 minutes was taken as the time interval between runs
to a constant testing velocity and back to rest again. The to minimize disturbance induced by the previous run. The
speed of the carriage was pressure-controlled, i.e., by cylinders material properties and the tested flow conditions
adjusting the applied pressure on the actuator, towing speeds are summarized in Table 1.
of up to 1.6 m/s were possible.
2.2 Characteristics of flexible cylinder
The flexible circular cylinder used in testing is a Tygon
tube with an inner diameter of 4.8 mm (3/16 in.) and an
outer diameter of 7.9 mm (5/16 in.). It has an initial length
of 420 mm and is horizontally connected to the rigid
supporting frame via a set of screws at both ends, as shown
in Fig. 4. It should be noted that the tip to tip distance
between the two screws is 460 mm (effective length), i.e.,
the flexible cylinder reaches a tensioned state (40 mm
elongation) when installed in place. From this point forward,
the cylinders length will always be referred to as the
effective length (460 mm). Inevitably a reduction in the
cylinders diameter occurs when it is being stretched, and Fig. 4 Supporting mechanism
measurements were conducted on the stretched cylinders
outer diameter using a digital caliper, which was found to be Table 1 Cylinder material properties and flow conditions
around 7.91 mm; therefore, the reduction rate is calculated
Aspect ratio L/D 58
to be 3%. Since this change is not remarkable, the cylinders
outer diameter is considered to be 7.9 mm throughout the Axial stiffness EA 124 N
whole study. Therefore, the test cylinders aspect ratio, Effective length L 460 mm
which is defined as a ratio of the length to outer diameter, is Flexural stiffness EI 0.001 Nm2
58 in the present study. The axial pre-tension applied on the Mass ms 37 g/m
stretched cylinder was measured to be around 11 N, using a
Outer diameter D 7.9 mm
force meter and inducing identical elongation (40 mm) on
the cylinder when it was disconnected from the supporting Pre-tension T* 1 200
frame. For the convenience of comparison with literature, Reduced velocity Ur 2 25
the value of pre-tension is converted to a dimensionless Reynolds number Re 800 13 000
form of 1 200, using Towing speed U 0.1 1.6 m/s
T T / Tcr
*
(1) As our focus of this study is the effect of the mass ratio on
the cylinders FIV, the cylinder was prepared to have
where T is the pre-tension, and Tcr is the critical
different mass ratios by filling it with different materials.
compression load for buckling. Tcr can be obtained from
Mass ratios of flexible structures in offshore applications
Tcr 2 EI / 4 L2 (2) can be as low as 3 (Huera-Huarte et al., 2014), and as noted
where E is Youngs modulus of the model material, I the there is a dearth of investigations for mass ratios near 1.
cross-sectional inertia, and L the models original length. Subsequently, the mass ratios in the present study were
The high value of dimensionless pre-tension is mainly a chosen to explore low values, namely 0.7, 1.0, and 3.4,
result of the low critical compression load. In the present which were were achieved by filling the cylinder with air,
study, the cylinder was towed through the water at speeds water, and alloy powder (nickel-chromium-boron matrix
54 Haoyang Cen, et al. Effect of Mass Ratio on Hydrodynamic Response of a Flexible Cylinder

alloy) respectively. Table 2 indicates the fill materials and on the model. Consequently, it was not possible for the
the fundamental natural frequency (natural frequency of the camera to remain directly above the cylinder during the test,
first structural mode) of each case, as well as the damping thereby causing a lowered accuracy. To minimize this effect,
ratio (damping coefficient/critical damping coefficient) in Camera 2 was positioned approximately 3D behind the
still water. The fundamental natural frequency and damping cylinders initial position in the IL direction. 8 mm wide
ratio for each case were determined via the free decay test, fluorescent tape rings were placed at three locations on the
where impulse excitation was imposed to the mid-span of cylinders span, which served as a visualizing target, as
the tensioned cylinder. This was achieved by slightly shown in Fig. 6. Identified as S-0.25, S-0.50, and S-0.75,
flipping the submerged cylinders mid-point using a slim their relative positions on the models span are 0.25, 0.50,
iron wire; the midpoints response was then captured and and 0.75 accordingly.
analyzed using fast Fourier transform (FFT). As an example,
the time history and spectra of the midpoint of case M-3.4
during the decay test is presented in Fig. 5. The damping
ratio was determined via logarithmic decrement from
ln( yn / yn m )
(4)
2 m
where yn is the displacement of the nth cycle, and yn+m of the
(n+m)th cycle. The results of tests show that the case with air
has a damping ratio of 4%, which is 1% higher than that of
Fig. 6 Measurement locations and their designations
the cases with water and alloy powder.

3 Results and discussion

The hydrodynamic response of a flexible cylinder is a
nonlinear phenomenon and is highly dependent on
parameters such as reduced velocity, Reynolds number,
mass ratio, and damping ratio. Many studies have attempted
to combine the mass ratio and damping ratio to investigate
the cylinders response. As the damping ratios of the three
tested cases in the present study are nearly the same (4%,
3% and 3%), it is thus assumed that the small variation in
Fig. 5 Time history of decay test (alloy powder)
the damping ratio has little influence on the cylinders
2.3 Measurement techniques response. Consequently, the variation in mass ratio is
In the present study, the response of the flexible cylinder deemed to be the major contributor to the difference in the
was measured optically, and two high-speed cameras were systems response.
deployed to track its vibration response at a rate of 240
frames per second. As illustrated in Fig. 3, to visualize the
CF response, Camera 1 was installed at one end of the tank,
10 m from the cylinders initial position and horizontally
aligned with the test cylinder. Camera 2 was mounted above
the cylinder on the supporting frame for the IL response
measurement (also illustrated in Fig. 4). During tests,
Camera 2 was carried through the fluids and thus stayed
above the model. To eliminate the impact of free surface on
video quality, Camera 2 was immersed slightly beneath the
waters surface. It was determined that while the test was
running, the mean drag force caused mean IL displacement Fig. 7 Time history examples of towing speed

Table 2 Test cases with corresponding parameters

Mass ratio Designation Fill material Material density / (kgm3) Fund. nat. freq./ Hz Damping ratio /%
0.7 M-0.7 Air 1.2 11.3 4%
1.0 M-1.0 Water 997.1 10.3 3%
3.4 M-3.4 Alloy powder 8 000 8.4 3%

A Cartesian reference system with origin at the mid-span the present study, with x denoting the IL direction, y the CF
of the model, as shown in Fig. 3, is under consideration in and z the span-wise. Therefore, u(z, t) and v(z, t) are used to
Journal of Marine Science and Application (2016) 15: 50-62
represent the instantaneous displacements in the IL and CF
directions respectively.
3.1 Towing motion and resulting response
As mentioned above, the towing carriage was actuated by
compressed air starting from rest and advancing to a
constant velocity; after a short period it decelerated to a
complete stop. This towing motion is demonstrated in Fig. 7,
which shows the time histories of several cases with towing
speeds as examples. It is noted that the positive and negative
acceleration stages take approximately 0.8 s and 1 s,
respectively, at all speeds, and thus the time period
remaining for constant speed motion is approximately 12 s
at the lowest towing speed of 0.1 m/s, and 0.6 s at the
highest speed of 1.6 m/s.
The left hand column of Fit. 8 also reflects the variance in
towing speed, and shows the time history and corresponding
spectra of the CF response displacement for the case M-3.4
at a towing speed of 0.53 m/s. The cylinder becomes excited
as the towing carriage is actuated, and experiences rapid
increase in response amplitude until around 0.8 s. Steady
vibration then appears to be reached approximately at a time
of 1 s. Finally, at t 2.5 s a drastic change in the
displacement is seen, reflecting the sharp slowdown of the
towing carriage. To study the cylinders hydrodynamic
response through the steady vibration stage, a time window,
within which the cylinder vibrates steadily, was thus applied
to each case to enable selection of data of interest. Only
displacement within the time window was processed for the
vibration amplitude and frequency analysis. The
mid-column of Fig. 8 shows the enlarged time histories of
the displacements in the chosen time intervals from 1 to 2.5
s. FFT was thereafter applied to the selected data to obtain
the spectra distribution, and this is demonstrated in the right
column of Fig. 8.
3.2 Multi-frequency vibration and higher harmonics
It is possible to see from the right column of Fig. 8, that there
are two distinct peaks at frequencies of 8.9 and 14.5 Hz,
indicating significant contributions from these two frequencies
to the cylinders CF response. Furthermore, the dominant
frequency, which has the highest peak in the spectra, is found to
be different at locations along its span. In the case shown in Fig.
8 for example, the mid-span of S-0.50 is dominated by a
frequency of 8.9 Hz, while that of S-0.25 and S-0.75 are
dominated by a frequency of 14.5 Hz. This could suggest that
the cylinder is undergoing multi-frequency vibration, and that
different span locations are vibrating at different frequencies. It
should be noted that this is not a special case in the cylinders
dynamic response; it also appears at other towing speeds and in
cases M-0.7 and M-1.0.
Another common phenomenon is that the outstanding
frequencies are found in the fold-increase in some spectra of
test case responses. For example, Fig. 9 shows the
displacement and spectra of case M-1.0 at U = 0.24 m/s,
which is similar to that shown in Fig. 8, and a dominant
vibration frequency of 6.1 Hz is seen to be outstanding over
55
the range. In addition, notable contributions from
frequencies that are 2, 3, 4, 5, and 6 times that of the
dominant one can also be identified in its dynamic response,
indicating the existence of higher harmonics. It is considered
that this could be a result of the synchronization between the
vortex-shedding frequency and the vibration frequency. For
underwater circular cylinders, the vortex-shedding
frequency follows the Strouhal frequency (fst) when lock-in
is not present. According to Figs. 14 and 15, which present
response amplitudes of the IL and CF vibrations, it is clear
that lock-in is absent for case M-1.0 at U = 0.24 m/s (Ur =
2.9). Hence, it is reasonable to assume the vortex-shedding
frequency, (fv), for the case under consideration follows the
Strouhal frequency, i.e., fv = fst. We can therefore evaluate
the vortex-shedding frequency, as
f st St U / D (5)
By adopting St = 0.2, a value of 6.07 Hz is identified for
fst, and thus fv at U = 0.24 m/s. This means that the higher
harmonics are associated with synchronization between the
vortex shedding and vibration frequencies. Unlike the higher
harmonics reported by Vandiver et al. (2009) and Song et al.
(2011), which found only odd multiple harmonics in the CF
response, in the present study both odd and even multiples
are discovered in the CF response. As suggested by Song et
al. (2011), higher harmonics require a greater attention to
design, as their high frequencies are likely to lead to more
severe fatigue damage.
3.3 General vibration characteristics
Fig. 10 demonstrates the response spectra at S-0.50 of
three cases with respect to reduced velocity. The left column
shows the spectra of IL response, while the right column
shows that of the CF response. From the IL spectra, it is
possible to see that for case M-0.7, which has air sealed
inside the cylinder, the vibration frequency increases nearly
linearly with increasing reduced velocity, which is similar to
that of a rigid cylinder. Govardhan and Williamson (2000)
reported a relation close to linear between the vibration
frequency and current velocity on a cylinder with m* 1.
The highest frequency observed in current study is around
60 Hz. It is also clear that at a reduced velocity of around 6,
the cylinder experiences a significant amplitude peak.
Similarly, the vibration frequency also rises for case M-1.0
as the reduced velocity increases and reaches an amplitude
peak at Ur 6. However, it is noticeable that at a reduced
velocity of around 7, there is a sudden change in the
vibration frequency of case M-1.0, jumping from 17 Hz at
Ur = 6 to 27 Hz at Ur = 8.1. Between these changes is a
transition zone where competition between the two vibration
frequencies is observed. Fig. 11 gives a better presentation
of such observations, where the competition and evolution at
consecutive reduced velocities is evident. Following this
trend, a second jump and competition are discovered at a
reduced velocity of around 11. This is also the case when the
mass ratio is increased to 3.4, and is even more obvious, as
shown in the bottom left of Fig. 10. Jumps and competition
56 Haoyang Cen, et al. Effect of Mass Ratio on Hydrodynamic Response of a Flexible Cylinder

can be observed at reduced velocity regions at around 7 and concluded that the mode transition is accompanied by
15, with a broader range at around Ur=15. This phenomenon continuous change in amplitude, but also by a jump in
was previously reported by Li et al. (2013) in a study on a frequency.
flexible cylinder with a mass ratio of 4.3; the study

t/s t/s f/Hz

t/s t/s f/Hz

Fig. 8 History of CF displacement and corresponding amplitude spectra for case M-3.4 at U = 0.53 m/s

t/s t/s f/Hz

t/s t/s f/Hz

t/s t/s f/Hz

Fig. 9 History of CF displacement and corresponding amplitude spectra for case M-1.0 at U = 0.24 m/s
Journal of Marine Science and Application (2016) 15: 50-62 57

Fig. 10 Spectral distribution of S-0.50: top to bottom: M-0.7, M-1.0, M-3.4; left column gives IL response; right column gives
CF response
58 Haoyang Cen, et al. Effect of Mass Ratio on Hydrodynamic Response of a Flexible Cylinder
Fig. 11 Response spectra of case M-1.0 at Ur = 6.5, 7.5, and 8.1 (from left to right)
The overall distribution of case M-3.4 is significantly
different from cases with lower mass ratios. There is no
amplitude peak found at Ur 6, but one is found instead at
Ur 11. It is also possible to see that there are two main
response frequencies involved in the response of M-3.4 at
the reduced velocity ranges of [5, 7] and [16, 25]; this does
not exist in cases M-0.7 and M-1.0. However, there is also a
difference in the increased rate of vibration frequency,
which, unlike cases M-0.7 and M-1.0 where there is an
increase in the reduced velocity, the vibration frequency of
case M-3.4 appears to stay steady, with a small increment
outside the competition regions. It is of note that with
increasing reduced velocity, the maximum vibration
frequencies for the three cases are all found to be around
60 Hz. It therefore appears that the cylinder is limited from
vibrating at a frequency beyond that value. A possible
reason for this is that the cylinder switches from a
tensioned-dominated cable at low reduced velocity (flow
speed) to a bending-dominated beam at high reduced
velocity. As explained by Lee & Allen (2010), the increased
flow speed tends to induce larger deflection on the structure.
The resistance to this tendency comes primarily from the
tension for a tension-dominated structure, or from the
flexural stiffness, EI, for a bending-dominated one. As the
tension is expected to rise due to an increased flow velocity,
the natural frequency, and thus the vibration frequency of a
tension-dominated structure, rises with flow speeds.
However, this is not the case for flexural stiffness, EI, as it
will not be affected by flow speed, which in turn means that
the flow speed has a limited influence on the response
frequency. Another significant difference is that the
amplitude peak is remarkable compared to other regions
when the mass ratio falls below 1, but with an increasing
mass ratio this peak becomes considerably less notable.
For the CF response shown in the right column of Fig. 10,
two amplitude peaks are found in case M-0.7 and 1.0, while
three peaks are found in M-3.4 across the tested reduced
velocity range. Of these, the first peaks in cases M-0.7 and
M-1.0 are found at Ur 5, while that at M-3.4 is found at
around 7. In fact, the change in spectra amplitude with
respect to reduced velocity indicates a transition in vibration
mode. If we take case M-3.4 as an example, two sets of high
amplitude vibration are experienced in the reduced velocity
range of [4, 11] and [16, 25], and between these a low
amplitude vibration exists. The first and second high
amplitude range indicates that the cylinder is vibrating at its
1st and 3rd structural modes, respectively, while the low
range indicates a 2nd mode at where point the mid-span
(S-0.50) is a node in the structural mode and thus has a low
amplitude. This is confirmed by an analysis of the obtained
video. As demonstrated in Fig. 12, from top to bottom are
consecutive frames taken from the recorded video at
reduced velocities of 7.3, 13.8, and 17.7, where it is clearly
evident that the cylinder vibrating at its 1st, 2nd, and 3rd mode,
respectively. It is also evident that for every vibration mode,
a peak in spectra amplitude can be expected, although for
locations at the node of the mode the peak is considerably
smaller than that at anti-nodes. From a general and
qualitative point of view, the vibration mode has strong
dependency on the Reynolds number; when the Reynolds
number is higher there is a higher vibration mode. At the
high Reynolds number under consideration, it is possible to
excite the 3rd vibration mode. With respect to reduced
velocity, the first mode transition (from the 1st to 2nd) takes
place at Ur 15, 13, 11 for cases M-0.7, M-1.0, and M-3.4,
respectively. A trend appears, whereby the higher mass
ratios lead to an earlier mode transition in relation to
reduced velocity. Similar to that of the IL response, the CF
response frequency also seems to be limited below 30 Hz
for the three cases tested.
3.4 Vibration amplitudes and frequencies
Fig. 13 presents the spatio-temporal root mean square
(RMS) of the dimensionless IL response displacement
computed as
1
S 1 N
~
x
S N u 2
ji z , t
(6)
i 1 j 1
where S is the number of samples in the selected time
window for analysis, and N is the number of measurement
points along the axis of the cylinder (which is 3 in the
present study). It is possible to observe that the
spatio-temporal RMS of the IL vibration displacement has a
strong dependence on reduced velocity, and that it generally
follows quadratic trends for the three cases, with the lower
mass ratio cases having considerably larger values. Similar
trends were previously reported by Chaplin et al. (2005b)
and Huera-Huarte & Bearman (2009), and it is considered
that this is a result of increased drag and tension, T, on the
towed cylinder. Under increasing towing velocities, the
cylinder experiences increasing drag and consequently
higher tension. It is known that the drag on the cylinder is
balanced by the sum of horizontal components of tension,
which is proportional to the product of ~ x T . This yields
Journal of Marine Science and Application (2016) 15: 50-62 59

(a) 1st mode at Ur = 7.3

(b) 2nd mode at Ur = 13.8

(c) 3rd mode at Ur = 17.7

Fig. 12 Consecutive frames showing the cylinder (M-3.4) vibrating at different modes

~
x T C1 Cd U 2 (7) tension, the relation below is true when assuming the
where, C1 is a constant coefficient, and Cd is the drag constant drag coefficient:
2
coefficient. Considering that the cylinders modal ~ U
x / D C2

(8)
frequencies are generally proportional to the square root of fn D
60
where C2 is a constant coefficient. Eq. (8) could help
establish a rough estimation for the mean deflection of a
flexible underwater cylinder during the design stage.
In the present work, to quantify the general vibration
amplitude of a flexible cylinder, the spatio-temporal
standard deviations of displacement are computed to
describe both the IL and CF responses, and are calculated as
follows,
S 1 N M-1.0 remain vibrating at a high amplitude with a slight
1
Ax
S N [u ji z, t ui z ]2
i 1 j 1
1
S 1 N
Ay
S N v 2
ji
i 1 j 1
where Ax represents the IL response amplitude, and Ay
represents the CF response amplitude.
Fig. 13 Spatio-temporal RMS of normalized in-line
displacements versus reduced velocity
Fig. 14 Normalized in-line reponse amplitude versus
reduced velocity
The dimensionless spatio-temporal standard deviation of
the IL and CF response expressed with respect to reduced
velocity is shown in Figs. 14 and 15. High scatter is found to
exist in the dataset. As the tested cylinder in the present
work has a low mass ratio, the added mass is expected to be
comparable with its structural mass, and thus have a
significant influence on the overall response by affecting the
natural frequencies of the in-motion cylinder. Therefore, it is
considered that the high scatter in the response amplitude
could be one of the consequences of the the potential
non-uniform distribution of the added mass and the resulting
hydrodynamics.
For the IL response amplitude, one of the evident effects
Haoyang Cen, et al. Effect of Mass Ratio on Hydrodynamic Response of a Flexible Cylinder
of mass ratio on the cylinders vibration is the generally
increased response amplitude. It is clear that case M-0.7 is
vibrating at an amplitude considerably larger than that of
case M-3.4; case M-0.7 achieves maximum amplitude
around 0.35D, while M-3.4 only reaches 0.25D. The
response amplitudes of all three cases start to rise as the
reduced velocity increases from 2, and the first peaks are
reached at Ur 6, after which point the cases of M-0.7 and
(9) increase, but that of M-3.4 seems to experience
de-synchronization characterized by a drop in amplitude at
Ur 8. Only the initial and upper branches exist in the
z, t
(10) response amplitude. The lower branch disappear for all three
cases, which could imply lock-in persists over the tested
range. This is similar to the findings from elastically
mounted rigid cylinders with mass ratios below 0.54 (Jauvtis
and Williamson, 2003; 2004). In the pioneering work on
flexible cylinders of Vandiver (1993), an explanation for the
widened lock-in bandwidth in low mass ratio cases was
addressed. In addition, Sarpkaya (1977) determined that the
added mass coefficient decreases sharply as reduced
velocity increases, which means that as the structural mass
is very low, the total mass, and thus the natural frequency of
the cylinder, will rise considerably; the consequence of
which is the resonant response frequency increases with
flow speed and leads to a persistence of the lock-in region.
Fig. 15 Normalized cross-flow reponse amplitude versus
reduced velocity
Similar to that of the IL response amplitude, only the
initial and upper branches are found to exist in the CF
response amplitude, with a large amplitude persisting from a
reduced velocity of 6 for all three cases. It is of note that the
de-synchronization for case M-3.4 at Ur 8 is also observed
in the CF response, leading to a drop in vibration amplitude,
which returns synchronized again after that region. Other
than this, varying the mass ratio seems to have little effect
on the CF vibration, and vibration amplitudes of more than
0.5D are observed for all three cases, which is lower than
0.7D reported in a study on a flexible cylinder with a mass
ratio of 1.8 by Huera-Huarte and Bearman (2009).
Dominant response frequencies representing the highest
peaks in the spectra for every case were obtained by
applying Fast Fourier Transform of the response
displacement, where the mean deflection in the IL response
due to mean drag force was removed from the response
displacement in advance. Figs 16 and 17 present the
Journal of Marine Science and Application (2016) 15: 50-62
vibration frequencies of IL and CF responses with respect to
reduced velocity. In the plot of the IL response (Fig. 16),
frequencies that have a comparative magnitude to that of the
dominant one are also added as solid symbols.
It can be immediately observed from Figs. 16 and 17 that
the response frequencies rise linearly with reduced velocity,
despite the variation in mass ratio. The slope for IL and CF
responses is 0.3 and 0.15 respectively, which also indicates
that the frequency ratio between the IL and CF response is
approximately 2.
Fig. 16 Normalized in-line response frequency versus
reduced velocity
Fig. 17 Normalized cross-flow response frequency versus
reduced velocity
4 Conclusions
An experimental investigation is reported on the effects of
mass ratio on the hydrodynamic response of a highly flexible
cylinder with 2 degrees of freedom undergoing cross FIV. A
Reynolds number range of 800~13 000, corresponding to a
reduced velocity range of 2~25, was tested, and low mass
ratio values of 0.7, 1.0, and 3.4 were tested to investigate
associated effects. Multi-frequency vibration was discovered
in the dynamic responses of the cylinder, which seemed to
transition from tension-dominated at a low reduced velocity to
bending dominated at high reduced velocity. Within the tested
range, only the initial branch and upper branch were
discovered in the response amplitudes of the IL and CF
vibrations. After the initial rise, high amplitude responses
were found to persist over the tested range in relation to the
low value of mass ratio. An observed effect of the mass ratio
was the overall decreased amplitude in the IL response, but
limited influence was found on the CF amplitude. The study
also examined the vibration frequency, and linear relations
61
and independence of mass ratios are shown for normalized
frequencies with respect to reduced velocities. A slope of 0.3
and 0.15 was revealed, which predicted the vibration
frequency for IL and CF, respectively, and the corresponding
frequency ratio between IL and CF response was shown to be
approximately 2.
References
Bearman PW, 1984. Vortex shedding from oscillating bluff bodies.
Annual Review of Fluid Mechanics, 16, 195-222.
Bearman PW, 2010. Circular cylinder wakes and vortex-induced
vibrations. Journal of Fluids and Structures, 27, 648-658.
DOI: 10.1016/j.jfluidstructs.2011.03.021
Blevins RD, 2001. Flow-induced vibration, Second ed. Krieger
Publishing, Inc., Malabar, Florida, USA.
Brika D, Laneville A, 1993. Vortex-induced vibrations of a long
flexible circular cylinder. Journal of Fluid Mechanics, 250,
481-508.
Chaplin JR, Bearman PW, Cheng Y, Fontaine E, Graham JMR,
Herfjord K, Huera-Huarte FJ, Isherwood M, Lambrakos K,
Larsen CM, Meneghini JR, Moe G, Pattenden RJ, Triantafyllou
MS, Willden RHJ, 2005a. Blind predictions of laboratory
measurements of vortex induced vibrations of a tension riser.
Journal of Fluids and Structures, 21, 25-40.
DOI: 10.1016/j.jfluidstructs.2005.05.016
Chaplin JR, Bearman PW, Huera-Huarte FJ, Pattenden R, 2005b.
Laboratory measurements of vortex-induced vibrations of a
vertical tension riser in a stepped current. Journal of Fluids and
Structures, 21, 3-24.
DOI: 10.1016/j.jfluidstructs.2005.04.010
Chung TY, 1987. Vortex-induced vibration of flexible cylinders in
sheared flows. Ph.D. thesis, Massachusetts Institute of
Technology, Cambridge, USA.
Chung TY, 1989. Vortex-induced vibration of flexible cylinders
having different mass ratios. Korea Research Institute of Ships
and Ocean Engineering, Report NO. UCE 440-1283ED.
Farrant T, Tan M, Price WG, 2001. A cell boundary element
method applied to laminar vortex shedding from circular
cylinders. Computers & Fluids, 30, 211-236.
DOI: 10.1016/S0045-7930(00)00009-8
Feng CC, 1968. The measurements of vortex-induced effects in flow
past stationary and oscillating circular and D-section cylinders.
Master thesis, University of British Columbia, Vancouver, BC,
Canada.
Gabbai RD, Benaroya H, 2005. An overview of modeling and
experiments of vortex-induced vibration of circular cylinders.
Journal of Sound and Vibration, 282, 575-616.
DOI: 10.1016/j.jsv.2004.04.017
Govardhan R, Williamson CHK, 2000. Modes of vortex formation
and frequency response for a freely-vibrating cylinder. Journal
of Fluid Mechanics, 420, 85-110.
DOI: 10.1017/S0022112000001233
Huera-Huarte FJ, Bangash ZA, Gonzalez LM, 2014. Towing tank
experiments on the vortex-induced vibrations of low mass ratio
long flexible cylinders. Journal of Fluids and Structures, 48,
81-92.
DOI: 10.1016/j.jfluidstructs.2014.02.006
Huera-Huarte FJ, Bearman PW, 2009. Wake structures and
vortex-induced vibration of a long flexible cylinderPart 1:
Dynamic response. Journal of Fluids and Structures, 25,
969-990.
62 Haoyang Cen, et al. Effect of Mass Ratio on Hydrodynamic Response of a Flexible Cylinder
DOI: 10.1016/j.jfluidstructs.2009.03.007
Jauvtis N, Williamson CHK, 2003. Vortex-induced vibration of a
cylinder with two degrees of freedom. Journal of Fluids and
Structures, 17, 1035-1042.
DOI: 10.1016/S0889-9746(03)00051-3
Jauvtis N, Williamson CHK, 2004. The effect of two degrees of
freedom on vortex-induced vibration at low mass and damping.
Journal of Fluid Mechanics, 509, 23-62.
DOI: 10.1017/S0022112004008778
Khalak A, Williamson CHK, 1996. Dynamics of a hydroelastic
cylinder with very low mass and damping. Journal of Fluids
and Structures, 10, 455-472.
Khalak A, Williamson CHK, 1997. Fluid forces and dynamics of a
hydroelastic structure with very low mass and damping.
Journal of Fluids and Structures, 11, 973-982.
Khalak A, Williamson CHK, 1999. Motions, forces and mode
transitions in vortex-induced vibrations at low mass-damping.
Journal of Fluids and Structures, 13, 813-851.
DOI: 10.1006/jfls.1999.0236
Kim YH, Vandiver JK, Holler R, 1986. Vortex-induced vibration
and drag coefficients of long cables subjected to sheared flows.
Journal of Energy Resources Technology, 108, 77-83.
Lee L, Allen D, 2010. Vibration frequency and lock-in bandwidth
of tensioned, flexible cylinders experiencing vortex shedding.
Journal of Fluids and Structures, 26, 602-610.
DOI: 10.1016/j.jfluidstructs.2010.02.002
Li XC, Wang YX, Wang GY, Jiang MR, Sun Y, 2013. Mode
transitions in vortex-induced vibrations of a flexible pipe near
plane boundary. Journal of Marine Science and Application, 12,
334-343.
DOI: 10.1007/s11804-013-1198-6
Sarpkaya T, 1977. Transverse oscillations of a circular cylinder in
uniform flow, Part I. Naval Postgraduate School Report No.
NPS-69SL77071.
Sarpkaya T, 1979. Vortex-induced oscillations. Journal of Applied
Mechanics - ASME, 46, 241-258.
Sarpkaya T, 2004. A critical review of the intrinsic nature of
vortex-induced vibrations. Journal of Fluids and Structures, 19,
389-447.
DOI: 10.1016/j.jfluidstructs.2004.02.005
Song JN, Lu L, Teng B, Park HI, Tang GQ, Wu H, 2011.
Laboratory tests of vortex-induced vibrations of a long flexible
riser pipe subjected to uniform flow. Ocean Engineering, 38,
1308-1322.
DOI: 10.1016/j.oceaneng.2011.05.020
Sumer BM, Fredsoe J, 1997. Hydrodynamics around cylindrical
structures. World Scientific Publishing Co. Pte. Ltd., Singapore.
Vandiver JK, 1983. Drag coefficients of long flexible cylinders.
Offshore Technology Conference, Texas, USA, No. 4490-MS.
Vandiver JK, 1993. Dimensionless parameters important to the
prediction of vortex-induced vibration of long, flexible
cylinders in ocean currents. Journal of Fluids and Structures, 7,
423-455.
Vandiver JK, Jaiswal V, Jhingran V, 2009. Insights on
vortex-induced, traveling waves on long risers. Journal of
Fluids and Structures, 25, 641-653.
DOI: 10.1016/j.jfluidstructs.2008.11.005
Williamson CHK, Govardhan R, 2004. Vortex-induced vibrations.
Annual Review of Fluid Mechanics, 36, 413-455.
DOI: 10.1146/annurev.fluid.36.050802.122128
Williamson CHK, Govardhan R, 2008. A brief review of recent
results in vortex-induced vibrations. Journal of Wind
Engineering and Industrial Aerodynamics, 96, 713-735.
DOI: 10.1016/j.jweia.2007.06.019
Williamson CHK, Jauvtis N, 2004. A high-amplitude 2T mode of
vortex-induced vibration for a light body in XY motion.
European Journal of Mechanics B/Fluids, 23, 107-114.
DOI: 10.1016/j.euromechflu.2003.09.008
Williamson CHK, Roshko A, 1988. Vortex formation in the wake
of an oscillating cylinder. Journal of Fluids and Structures, 2,
355-381.