Nuclear Engineering and Design 313 (2017) 214224

Contents lists available at ScienceDirect

Nuclear Engineering and Design

journal homepage: www.elsevier.com/locate/nucengdes

Vibration response of a pipe subjected to two-phase flow: Analytical

formulations and experiments
L. Enrique Ortiz-Vidal a,, Njuki W. Mureithi b, Oscar M.H. Rodriguez a
a
Department of Mechanical Engineering, Sao Carlos School of Engineering, University of Sao Paulo (USP), Av., Trabalhador So-carlense, 400, 13566-970 So Carlos, SP, Brazil
b
Department of Mechanical Engineering, Polytechnique Montreal, Dpartement de Gniemcanique 2900, H3T 1J7 Montreal, QC, Canada

h i g h l i g h t s

Analytical formulations for two-phase flow-induced vibration (2-FIV) are presented.

Standard deviation of acceleration pipe response is a function of the square of shear velocity.

Peak frequency is correlated to hydrodynamic mass and consequently to void fraction.

Dynamic pipe response increases with increasing mixture velocity and void fraction.

Hydrodynamic mass in 2-FIV in horizontal pipe is proportional to mixture density.

a r t i c l e i n f o a b s t r a c t

Article history: This paper treats the two-phase flow-induced vibration in pipes. A broad range of two-phase flow con-
Received 18 July 2016 ditions, including bubbly, dispersed and slug flow, were tested in a clamped-clamped straight horizontal
Received in revised form 14 December 2016 pipe. The vibration response of both transversal directions for two span lengths was measured. From
Accepted 16 December 2016
experimental results, an in-depth discussion on the nature of the flow excitation and flow-parameters
influence is presented. The hydrodynamic mass parameter is also studied. Experimental results suggest
that it is proportional to mixture density. On the other hand, two analytical formulations were developed
Keywords:
and tested against experimental results. One formulation predicts the quadratic trend between standard
Pipe flow
Two-phase flow
deviation of acceleration and shear velocity found in experiments. The other formulation indicates that
Gas-liquid flow the peak-frequency of vibration response depends strongly on void fraction. It provides accurate predic-
Flow-induce vibration tions of peak-frequency, predicting 97.6% of the data within 10% error bands.
2016 Elsevier B.V. All rights reserved.

1. Introduction full understanding of the two-phase flow mechanisms, which still

Flow-induced vibration is a common phenomenon in industry.
The flow generates dynamic fluid forces which excites the struc-
ture inducing mechanical vibrations. The phenomenon has been
studied extensively over the last four decades, mainly for nuclear
industry applications (e.g. Fujita, 1990; Padoussis, 2006;
Pettigrew et al., 1998; Weaver et al., 2000). To date, the phe-
nomenon of structural vibration induced by single-phase flow
(FIV) is reasonably well understood. Researchers have aimed to
reach that same level of understanding for vibrations due to two-
phase flow. This is a challenge considering that it depends on the
Corresponding author.
E-mail addresses: leortiz@sc.usp.br, enrique.ovl@outlook.com (L.E. Ortiz-Vidal),
njuki.mureithi@polymtl.ca (N.W. Mureithi), oscarmhr@sc.usp.br (O.M.H. Rodri-
guez).
is an open topic.
On account of the emphasis of research on nuclear industry,
where both axial and cross flows are predominant, the available
information on flow-induced vibration subject to two-phase pipe
flow is scanty (Chen, 1991; Ortiz-Vidal and Rodriguez, 2011;
Pettigrew et al., 1998). Experimental studies noticed influence of
two-phase flow parameters such as, mixture velocity, void fraction
and flow-pattern on the structural response of the pipe system. For
example, vibration response increases with increasing mixture
velocity and homogeneous void fraction. It also depends on flow-
pattern (Geng et al., 2012; Ortiz-Vidal et al., 2013; Zhang and Xu,
2010). The characteristics of bubbly flows can change due to the
level of flow-induced vibration and as a function of liquid superfi-
cial velocity; however, these changes are not big enough to alter
the flow pattern (Hibiki and Ishii, 1998). Despite the mentioned
findings, a comprehensive and deeper discussion on the effect of
two-phase flow parameters on vibration response is in order. The

http://dx.doi.org/10.1016/j.nucengdes.2016.12.020
0029-5493/ 2016 Elsevier B.V. All rights reserved.
 L.E. Ortiz-Vidal et al. / Nuclear Engineering and Design 313 (2017) 214224 215

Nomenclature

A cross-sectional inside pipe area, [m2] fT, fs, ftp total, structural and two-phase damping components
C damping coefficient, [Ns/m] [%]
d, D internal and external diameters, [m] l viscosity [Pas]
EI rigidity, [Nm2] q density [kg/m3]
J, J (mixture) velocity, shear velocity, [m/s] Tk,w wall shear stress of k-phase
Ji joint acceptance [adm] swall, slam, sturb, wall, laminar and turbulent shear stress [N]
Lspan length between clamps, [m] xi, xpeak, xair, natural (for ith mode), resonance-peak and in-air
m linear mass density, [kg/m] frequencies [Hz]
P pressure, [Pa] /(x), w(x) mode shapes of the pipe and the pressure field
p0 pressure fluctuation, [Pa]
Q volumetric flow rate, [m3/s] Subscripts
y(x,t) transversal displacement of the pipe [m] acc acceleration
STD, PSD standard deviation and power spectral density G, L gas (air), liquid (water)
r radius [m] h hydrodynamic
V velocity [m/s] H homogeneous
v velocity fluctuation [m/s] k k-phase
x quality [adm] Lspan length between clamps [m]
a void fraction [adm] M mixture
b homogeneous void fraction [adm] p pipe

nature of flow excitation and the effect of flow patterns, specially

slug-flow, and flow-pattern transition on vibration response
should be analyzed.
Dynamic response of a pipe subjected to two-phase flow can be
represented by damping, hydrodynamic mass and excitation
mechanisms (Fujita, 1990; Pettigrew and Taylor, 1994). These
dynamics parameters are also related to two-phase flow parame-
ters. Significant research efforts on the study of damping have been
spent in recent years (Bguin et al., 2009; Charreton et al., 2014;
Gravelle et al., 2007). Results indicate that both structural and
two-phase damping components are relevant in structural vibra-
tion induced by two-phase flow (2-FIV) in pipes. Furthermore, Fig. 1. Scheme of a pipe subjected to internal flow.
the latter has a viscous nature (velocity dependent) and is highly
dependent on void fraction. Hydrodynamic mass in two-phase pipe
flow, on the other hand, has not received much attention. No report vibration-excitation mechanisms, and consequently the vibration
on any experimental study has been found in the literature. In the response y(x,t). The turbulence and two-phase flow excitation
case of excitation mechanisms, both turbulence and two-phase mechanisms are the most relevant for both liquid and gas-liquid
flow are the most important mechanisms in two-phase pipe flow pipe flow (Pettigrew et al., 1998). Turbulence is associated to the
(Pettigrew and Taylor, 1994; Riverin and Pettigrew, 2007). An ana- presence of eddies of many sizes who decay to smaller ones, dissi-
lytical deduction has been reported in the literature to describe a pating energy. The eddies in the region near the pipe wall generate
relationship between vibration response and flow rate based on random pressure fluctuations that forces it to vibrate (Blevins,
the mechanism of turbulence (Evans et al., 2004). No other formu- 2001). In the case of two-phase flow, in addition to eddies, the
lation on vibration response including excitation mechanism has instant reconfiguration of the phases induces perturbations on
been found in the open literature. the flow. Depending of their nature, these perturbations can be
In this paper, flow-induced vibration subject to two-phase pipe another source of turbulence or can promote the emergence of a
flow is investigated. Experiments in a broad range of two-phase two-phase flow excitation mechanism. Thus, the relationship
flow conditions, including bubbly, slug and dispersed flow, in between the structural response and flow perturbations is evident.
two span lengths of a clamped-clamped straight horizontal pipe
are carried out. We present an in-depth discussion on 2-FIV vibra-
2.1. Relation between standard deviation of acceleration and shear
tion response from these experiments; e.g. the effects of flow-
velocity
pattern and hydrodynamic mass are presented. In addition, two
analytical formulations are developed and tested. One relates the
The development of an analytical formulation of flow-induced
vibration response to shear velocity based on the mechanical
vibration subjected to two-phase flow (2-FIV) in pipes has to do
energy equation of the two-phase flow and random vibration con-
with the establishment of a relationship between flow and struc-
cepts. The other formulation indicates the relationship between
ture parameters. To date, exact mathematical deductions are not
peak-frequency and void fraction.
available, mainly because some topics, e.g. turbulence, two-phase
flow and their interaction, are still in development. Here we ana-
2. Analytical approach lytically present a relationship between shear velocity and the
standard deviation of acceleration of the pipe. This is done in
Fig. 1 represents a pipe subject to internal adiabatic two-phase two steps. First, the mechanical energy equation of the flow is used
flow, where the flow characteristics determine the governing to relate pressure to shear velocity. Next, pressure and standard
216 L.E. Ortiz-Vidal et al. / Nuclear Engineering and Design 313 (2017) 214224

deviation of acceleration are correlated for the pipe, based on a ity perturbation, namely the Reynolds tensor. Therefore, there is a
random vibration approach. In the following analysis, we start clear relationship between two-phase wall shear stress and pres-
from fundamental equations for two-phase flow and structural sure perturbations that can be represented by
dynamics in order to establish relationships between flow and
f sw ; p0 0 4
structural parameters. Physical interpretations from previous
theoretical and experimental results and well-known standard
STDp0 c1 sw 5
mathematical methods are used.
The two-phase pipe flow of Fig. 1 can be modeled by the one- where c1 is a constant.
dimensional two-fluid model. The momentum equation of The perturbation component of pressure field in Fig. 1, i.e. p0 (x,t),
k-phase is represented by (Ishii and Hibiki (2006) and Rosa (2012)) acts as a distributed transverse load on the pipe system. It can be
assumed as stationary, zero mean, ergodic random process. Then,
@ @
hak iqk hV k i C Vk hak iqk hV k i2 the dynamic equation of motion can be expressed by
@t @x
@Pk Sk @4 @ @2
hak i  ak;w T k;w hV ki ihCk i hMki i 1 EI yx; t C yx; t qA 2 yx; t p0 x; t 6
@x A @x 4 @t @t
where the subscripts k, i and w refer to k-phase, interfacial and pipe where E is the elastic modulus, I is the second moment of area of the
wall. Angle brackets represent flow properties averaged over the tube section and qA is the linear mass density of the system, includ-
crosssectional area. a, q, P and V represent respectively void frac- ing added mass. The product EI is the flexural rigidity. C represents
tion, density, pressure and actual velocity. CVk is the distribution the damping coefficient. In presence of a linear system, the
parameter and computes the effect of the void and momentum- transversal displacement y(x,t), in terms of normal modal expan-
flux profiles on the cross-sectional area. A and S denote the cross- sions, is defined as,
sectional inside pipe area and wall wetted perimeter, respectively.
X
N
T represents the shear stress. The last two terms on the rhs of yx; t /i xyi t 7
Eq. (1) are related to interfacial forces and are associated with i1
momentum transfer due to mass transfer at the interface and the
where the summation is truncated after all important modes. /i(x)
stress acting at the interface, respectively. In the case of single-
and yi(t) represent the modal shape and modal coordinate, for the
phase flow, these two terms disappear and ak becomes equal to one.
ith mode, respectively. The former parameter is related to natural
The mechanical energy equation can be used to identify the
frequencies of the system. Blevins (1979a,b) provides a vade mecum
dependence between flow parameters. This energy equation is
for natural frequencies and modal shapes for different structural
obtained by multiplying the momentum equations (Eq. (1)) by
components. Substituting Eq. (7) in Eq. (6), multiplying by /j(x)
the respective average velocity, and adding the equations of both
and integrating over the length, the system equation becomes
phases. For most practical applications, the static pressure of the
RL
two phases can be considered equal (Pk = P) (Ishii and Hibiki, 1 2fi /i xp0 x; tdx
2006). Thus, we have the mechanical energy equation for two- t
y y_ i t yi t 0
R 8
xi 2 i xi qAx2i 0L /2i xdx
phase flow,
" # where xi and fi represent the natural frequency and viscous damp-
X2
@ hV k i2 X 2
@ hV k i2
hak iqk C Vk hak iqk hV k i hak ihV k iP ing factor, for ith mode, respectively. Eq. (8) assumes uncoupled sin-
k1
@t 2 k1
@x 2 gle dominant modes with a small damping ratio. In a similar
X
2 manner to the structural displacement, it is reasonable to separate
hV k ihF kM i 2 the pressure, p0 (x,t), into functions of space and time. In the follow-
k1 ing, we consider that pressure load contain a broad range of fre-
where FkM represents the source term. Considering small perturba- quencies and the majority of the pipe response due to the load is
tions in velocity and pressure, vk and p, upon an uniform steady contained at frequencies near the system natural frequencies. These
flow, Vk, P, respectively, and retaining only the first-order terms assumptions are coherent with several experimental results (e.g.
for simplicity (this does not detract generality), the mechanical Chung, 1985; Durant and Robert, 1998; Ortiz-Vidal et al., 2013).
energy equation becomes, Then, for each mode, pressure load can be expressed as
" ! # p0i x; t wi xp0i t 9
X
2
@ X2
@ 3V 2
ak qk V k v 0k ak C Vk qk k P v 0 0
V kp
where wi(x) is the shape of the distribution of the pressure load
k1
@t k1
@x 2 k

along the pipe, of the turbulence pressure load component in the

X
2
hV k v 0k ihF kM i 3 frequency band near the natural frequency of mode i. pi(t) repre-
k1 sents the load acting over the range of frequencies in the neighbor-
hood of the ith mode natural frequency. The last definitions are in
where the angle brackets are dropped for simplicity. For the present agreement with a previous approach by Blevins (2001). Substituting
analysis, void fraction perturbation is neglected. This assumption is Eq. (9) in Eq. (8),
reasonable for a stable flow pattern from the hydrodynamic point of
1 2fi
view, i.e., any void fraction perturbation does not promote transi- i t
y y_ i t yi t J i p0i t 10
tion to another flow pattern (Rodriguez and Bannwart, 2008; x2i xi
Rodriguez and Castro, 2014; Wallis, 1969). The phases are also
where Ji is the joint acceptance,
assumed incompressible. Eq. (3) is similar to the mechanical energy
RL
equation obtained by Laufer (1953) for turbulent single-phase pipe /i x wi xdx
flow. Eq. (3) establishes the interdependence between pressure and Ji 0
R 11
qAx2i 0L /2i xdx
velocity perturbations. For a number of systems of practical inter-
est, particularly for systems without mass transfer and phase Blevins (1989) studies different analytical approximations for Ji.
change, the source term FkM consists of dissipative surface forces, In general, the spatial distribution of pressure field, wi(x), is
specifically wall shear stress. In turn, FkM is also a function of veloc- unknown and experimental data on the pressure field is needed
 L.E. Ortiz-Vidal et al. / Nuclear Engineering and Design 313 (2017) 214224 217

(Sankaran and Jancauskas, 1992). The joint acceptance equals unity 2.2. Relation between pipe peak frequency and void fraction
is commonly used for simplicity and corresponds to the case where
the pressure-field and the pipe-mode shapes are identical. Ji = 1 Pressure fluctuations, due to turbulence, excite the pipe system
produces a conservative estimate (Cunningham and White, 2004; in a wide range of frequencies corresponding to their infinite
Pettigrew and Taylor, 1994). Taking the Fourier transform of both degrees of freedom (modes) (Blevins, 1979a,b; de Silveira Neto,
sides of Eq. (10), 2002). Therefore, the structural response in frequency domain
shows a resonance phenomenon for each mode, i.e. the response
Y i x J i Hi xP0i x 12 is amplified in the neighborhood of the natural frequencies.
According to flow-induced vibration studies, the peak frequency
where Hi(x) is the transfer function of the system. It is expressed by value for each natural frequency is affected by different parameters
1 (Carlucci, 1980; Chung and Chen, 1984; Yi-min et al., 2010), such
H i x   13 as flow velocity and fluid density. In the case of two-phase flow,
x2
x2i
2ifi xx 1 fluid density depends on void fraction. Thus, a relationship
i
between peak frequency and void fraction (a) is reasonable. The
x is the frequency of the excitation. Eq. (12) can be expressed in following expressions can be used for this purpose,
term of the Power Spectral Density of the acceleration response s  1=2 q s2
EI mp cJ mh J
(PSDacc) by multiplying each term by its complex conjugate and xpeaki ki Lspan 2
  1  f2T  1 
using the PSD definition, mp L4span mp mh |{z} EIk i
2

|{z} |{z} III |{z}

II IV
I
PSDacci x J 2i x4 jHi xj2 PSDp0 i x 14
21
PSDp0 i represents the power spectral density of the perturbation
component of the pressure. The fact that turbulence acts in a broad mh cm qL A1  a 22
frequency spectrum suggests that the pipe is excited in a similar
where xpeaki is the peak frequency of the ith mode. The terms I on
range of frequencies. Moreover, experimental results show PSDp-i
the rhs of Eq. (21) is related to theoretical natural frequency, being
is nearly uniform over a wide-band of the frequencies (Durant
that the factor kiLspan is the roots of frequency equation and depends
and Robert, 1998). Considering a pressure-fluctuation spectrum
on boundary conditions of the pipe (Blevins, 1979a,b). Lspan and mp
with constant magnitude PSDp in the range x1 6 x 6 x2, standard
are the length and linear mass density of the pipe, respectively. The
deviation of acceleration (STDacc) for ith mode is obtained by inte-
terms II and III take into account the effect of hydrodynamic mass
grating Eq. (14) over this range,
(mh) (Chung and Chen, 1984) and damping (f), respectively. The
Z x2 Z x2 term IV accounts for the influence of velocity, where the parameter
STD2acci PSDacci x dx J 2i PSDp0 x4 jHi xj2 dx 15 cJ is 0.55 for the first mode and 1 for the others (disregarding the
x1 x1
Coriolis effect) (Yi-min et al., 2010). In Eq. (22), cm is the hydrody-
By definition, the PSD of the pressure fluctuations is related to namic mass coefficient and when it is equal to 1 the hydrodynamic
its standard deviation by means of (Yang, 1986), mass is proportional to mixture density.
The formulation presented in Eqs. (21) and (22) is relevant for
STD2p0 2x2  x1 PSDp0 16 two-phase pipe flow and complements the formulation of the Sec-
tion 2.1 (Eq. (20)). In single-phase flow, for constant geometry and
Combining Eqs. (16) and (15),
known roughness, wall shear stress is exclusively a function of flow
STDacci ci STP p0 17 velocity (White, 1998). Consequently, Eq. (20) is enough to estab-
lish a relationship between structural response and flow. By con-
 Z 1=2 trast, a unique relationship is insufficient in two-phase flow
x2
1 because of wall shear stress is function of other parameters, such
ci J i x4 jHi xj2 18
2x2  x1 x1 as void fraction and mixture velocity (Ortiz-Vidal et al., 2014).
Thus, Eqs. (21) and (22) are necessary. Both formulations are com-
The overall acceleration response is obtained by summing of
plementary and establish a relationship between structural
squares of the contributions of the individual modes (Au-Yang,
dynamic and two-phase void fraction and velocity parameters.
1975; Blevins, 1989). Then, STD of acceleration response at refer-
ence position x1 can be expressed by
3. Experimental work
!1=2
X
N
STDacc STDp0 f/x1 i  ci g2 19 3.1. Two-phase flow facilities
i1

Eq. (19) establishes that STD of acceleration is proportional to
STD of pressure fluctuations. Furthermore, from Eq. (5), in turn,
the latter is proportional to wall shear stress. Introducing shear
p
velocity (J  sw =qM ) as a more convenient parameter to express
fluctuations in turbulence (e.g. Metzger and Klewicki, 2001) and
combining Eqs. (19) and (5), STDacc can be represent by
STDacc c2 qM J 2
where c2 is a constant. Eq. (20) is the proposed analytical formula-
tion based on turbulence that establishes a quadratic relationship
between standard deviation of acceleration and shear velocity. qM
represents the mixture density.
The experimental work was conducted at the Fluid-Structure
Interaction Laboratory of the BWC/AECL/NSERC Industrial Research
Chair, Polytechnique Montreal. Fig. 2 shows the test loop used to
investigate flow-induced vibration subjected by gas-liquid two-
phase pipe flow. The main components and measurement sensors
are designated by letters and numbers and listed in Tables 1 and 2,
respectively. The experimental facility was especially designed and
20 constructed according to Ortiz-Vidal (2012). A picture of the exper-
imental rig used in this study is shown in Fig. 3.
Compressed air is supplied into the system and its pressure con-
trolled by the regulator 1, which maintained a steady pressure.
Two air flowmeters (2 and 3 in Fig. 2) and the valve A are used
to measure and regulate the flow rate, respectively. From an inde-
218 L.E. Ortiz-Vidal et al. / Nuclear Engineering and Design 313 (2017) 214224
Fig. 2. Schematic representation of the test loop.
Table 1
Main components of the test loop.
Letter Component
A Air regulation valve
B Water reservoir
C Water centrifugal pump
D Frequency shifter
E Water centrifugal pump
F Water regulation valve
G Mixer
H Heavy clamp
I Test section
J Heavy clamp
Table 2
Measurement sensors of the test loop.
Number Component Range
1 Pressure regulator 0100 psig
2 Air flowmeter 075 slpm
3 Air flowmeter 01000 slpm
4 Pressure gage 0100 psig
5 Water flowmeter 10.18424.18 lpm
6 Pressure transducer 050 psig
7 Pressure transducer 050 psig
pendent line, water at standard conditions is pumped from the
reservoir B by the centrifugal pumps C and E. The water flow rate
is measured and regulated by the flowmeter 5 and valve F, respec-
tively. Water and air are mixed in mixer G and the two-phase mix-
ture flowed through the pipe in several flow-patterns. Then, the
fluids are separated by gravity in the atmospheric tank B.
An entrance length of 1760 mm (86d), between the mixer G
and the pressure tap 6 (vide Fig. 2) is considered. This length
ensures a fully developed flow for all tested single-phase flow con-
ditions (White, 1998). The adopted entry length is also enough for
dispersed two-phase flow conditions, provided that it can be con-
sidered as a pseudofluid with homogeneous properties (Wallis,
1969). In the case of intermittent flows, elongated bubbly and slug,
a minimum length of 60d is needed (van Hout et al., 2003). From
this length value, the coalescence rate of liquids slugs, related to
slug frequency, is negligible. In this case, the adopted entrance
length ensures the development of intermittent flow patterns.
3.2. Test section
The test section (I, Fig. 2) consists of a clamped-clamped hori-
zontal PVC pipe with 20.4 and 26.7 mm of internal (d) and external
(D) diameters, respectively. The span is delimited by the distance
between heavy clamps (H and J, Fig. 2). In the present study, two
spans were tested, Lspan = 40d (=816 mm) and 75d (=1530 mm).
Impact test and circle-fitting method techniques were performed
to extract the natural frequency and damping of the system. Mea-
surements for the two transversal directions with both stagnant
water and air were done. Structural damping ratio ranged between
0.7% and 2.3%. Experimental natural frequencies are in agreement
with those predicted by continuous-system theory (Blevins, 1979a,
b), The percent error between measurements and theory is less
than 5%. For example, experimental and predicted first-mode nat-
ural frequencies of air-filled span of 75d are 16.29 and 16.49 Hz
(gravity direction), respectively. The following equation was used,
s
EI
xi ki Lspan 2
23
mP L4span
where EI = 39.47 Nm2. The factor kiLspan takes the values of 4.7300
and 10.9956 for the first and third modes, respectively. The linear
mass density of the pipe, mp, is 0.336 kg/m.
A triaxial accelerometer located at the midspan position is used
to monitor the vibration response of the test section. The x-axis is
parallel to the direction of the flow (axial direction). Both z and y
axes are perpendicular to pipe axis and z-axis lies in the gravity
direction. The sensitivity for x, y and z axes is 102.0 mV/g,
102.7 mV/g e 99.7 mV/g, respectively.
3.3. Experimental matrix
The experimental matrix is composed from 8 water single-
phase and 32 air-water two-phase flow conditions, for both test
sections of spans 40d and 75d. Flow velocities from 0.25 m/s to
7 m/s are used to water single-phase flow. In the case of air-
water flow, two parameters are used: the mixture velocity,
referred simply as velocity in this paper, J, and the homogeneous
Accuracy void fraction (or simply void fraction), b,
QL QG
1% FS J JL JG 24
A
1.5% FS
3% FS
QG JG
0.25% RD b 25
0.05% FS Q L Q G JL JG
0,05% FS
where QL, QG and A represent the liquid and gas flow rates and
pipes inner cross-sectional area, respectively. Every flow condition
is set for the midspan position of the test section, i.e. in-situ values.
For this, the air flow rate is corrected from the pressure values
downstream and upstream (6 and 7, Fig. 1), assuming a linear pres-
sure variation along the test section. The mixture velocity and void
fraction ranges from 0.5 m/s to 25 m/s and 10% to 95%, respectively.
Tested experimental conditions include bubbly, dispersed and slug
flows as showed in Fig. 4. More details on experimental work are
given in Ortiz-Vidal et al. (2014).
3.4. Data acquisition and processing
A program developed in LabVIEWTM, via PXI National Instru-
mentsTM acquisition boards, was used for acquiring and processing
the signals. The sampling rate of 5000 samples/s was chosen to
ensure the accuracy of the sensors and avoid the aliasing effect.
 L.E. Ortiz-Vidal et al. / Nuclear Engineering and Design 313 (2017) 214224 219

Fig. 3. Experimental rig used in this study.

Fig. 4. Air-water experimental conditions on Mandhane et al. flow map (Mandhane et al., 1974). The solid grey, white and black symbols represent the bubbly, dispersed and
slug flow patterns, according to experimental observations (Ortiz-Vidal et al., 2014), respectively. The void fraction classification is based on the nominal value.

For each flow condition, 100-s time acceleration data were

acquired. The data were low-pass filtered at 750 Hz and windowed
in samples of four seconds with 50% overlap. Power Spectral Den-
sity (PSD) was computed for each sample and then averaged.
Finally, standard deviation of acceleration (STDacc) was calculated
from the averaged Power Spectral Density (simply PSDacc) using
Parsevals theorem (Allen and Mills, 2004).
Shear velocity was directly calculated from experimental
pressure-drop data collected for water single-phase and air-
water two-phase pipe flow, by means of Eq. (26) (White, 1998)
(pressure drop data can be found in Ortiz-Vidal et al. (2014)).
r s
sw d dP
J  26
qM 4qM dx

4. Results and discussion

4.1. Experimental observations

Fig. 5. Intensity time-frequency map for water single-phase flow excitation. The
4.1.1. Nature of flow excitation map shows the Zaxis response for the system span 75d at 7 m/s. Dark regions
Fig. 5 shows a representative intensity time-frequency map of indicate peak frequencies and correspond to natural frequencies for first, third and
the structure response under single-phase flow excitation. It is fifth modes.
220 L.E. Ortiz-Vidal et al. / Nuclear Engineering and Design 313 (2017) 214224
constructed from PSD of acceleration calculated based on samples
of 4 s time acceleration signal, where the grayscale intensity is nor-
malized by minimum and maximum fixed arbitrary values. The
map indicates two important matters. On one hand, as expected
(Blevins, 1979a,b; Chung, 1985), turbulence excites the structure
in a wide range of frequencies, amplifying the response in the
neighborhood of the natural frequencies. Dark regions in Fig. 5 cor-
respond to the frequency values for first three modes. It is impor-
tant to note that the accelerometer was installed at midspan
position, then even modes were not captured. Turbulence, on the
other hand, generates a constant amplitude excitation for each
time interval, suggesting that the system is excited by the same
force level. This fact is illustrated in Fig. 5 and was not reported
before. Similar behavior was found in maps generated for two-
phase flow in both spans 75d and 40d systems. These results are
not presented here for the sake of brevity.
4.1.2. Influence of flow parameters on acceleration-response spectra
Fig. 6a illustrates the effect of flow velocity on dynamics
response of the pipe system subjected to single-phase flow excita-
tion. The higher the flow velocity, the bigger the structural vibra-
tion. This may be explained as follows. As velocity is increased,
velocity fluctuations increase due to turbulence. Then pressure
fluctuations are also more intense and consequently the structural
vibration. Moreover, the rate of increase of the structural response
is more significant at higher velocities. This fact can be observed in
Fig. 6a, small variation of structural response is observed between
minimum (red line, J = 0.5 m/s) and medium (blue line, 3 m/s)
velocities for a factor of 6 increase in flow velocity. By contrast,
when comparing the structural response of intermediate and max-
imum (black line, 7 m/s) velocities, where factor is lower (2.33), a
larger variation is observed. This fact was also reported in the liter-
ature (Jendrzejczyk and Chen, 1985) and can be related to turbu-
lence scales.
Flow velocity also has influence on peak frequency. The value of
peak frequency decreases with increasing velocity, as shown
Fig. 6a. In this sense, flow velocity produces the same behavior as
a compressible load acting on the pipe. We showed recently
(Ortiz-Vidal et al., 2013) that the frequency shift due to velocity
can be reduced or even eliminated through the normalization of
frequency, where the dimensionless frequency is function of flow
condition and critical buckling velocities. For general purposes,
the velocity effect on frequency shift is negligible compared to
the velocity variation.
Fig. 6. PSDacc for the first mode of 75d span system: (a) water single-phase flow excitation; (b) air-water two-phase flow excitation. C# corresponds to the flow conditions
shown in Fig. 4. The results are for Z-axis, which lies in the gravity direction.
The effect of other flow parameters is observed when analyzing
the pipe dynamic response due to two-phase flow. We found influ-
ence of mixture velocity, homogeneous void fraction, flow pattern
and two-phase damping, being that it is not possible to isolate the
effects as in the single-phase flow excitation case. Fig. 6b illustrates
the findings. The influence of flow-pattern can be seen in the shape
of the response spectrum. The effect of slug-flow pattern is evident,
see condition C24. Strong response for frequencies lower than
10 Hz, corresponding to characteristic frequencies of liquid slugs,
are observed. The response spectrum due to dispersed flow (C14)
is similar to single-phase flow excitation. Although bubbly flow
(C9) is an intermittent flowpattern, the velocities are small and
the liquid-slug momentum transfer is not large enough to be cap-
tured by the accelerometer. Thus the response for bubbly flow has
the same response as dispersed flow.
The mixture velocity has also influence on peak frequency. For
same homogeneous void fraction, the value of peak frequency
decreases with increasing mixture velocity. This behavior is similar
to single-phase flow excitation and noted in Fig. 6b when compar-
ing conditions C9 and C14; both with b = 25%. On the effect of
velocity on the structural response level, two opposite behaviors
are found. For low homogeneous void fractions, in contrast to
single-phase flow, it is observed a slight variation on the level of
structure response, although C14 has a significantly higher veloc-
ity, as seen in Fig. 6b. This effect is caused by higher values of
two-phase damping in dispersed flow, which reduces the level of
the response (Bguin et al., 2009). On the other hand, for high
homogeneous void fractions (b P 50), where slug flow occurs,
higher variations of the structure response level are found.
Homogeneous void fraction also acts on the pipe dynamics
response shifting the peak frequency to higher values when this
flow parameter increases. This effect is noted in Fig. 6b when com-
paring conditions C9 or C14 (b = 25%) and C24 (b = 50%). The shift
of peak frequency is related to hydrodynamic mass and will be dis-
cussed in the next section. From the discussion above, the pro-
posed peak-frequency analytical formulation appears to be
consistent.
4.2. Hydrodynamic mass
Hydrodynamic mass (mh) calculations have been performed
based on the Eq. (27) proposed by Carlucci (Carlucci, 1980). The
assumption of constant effective stiffness, implicitly in Eq. (27),
is considered acceptable according to the discussion above, where
a negligible influence of velocity was observed. The resonance fre-
 L.E. Ortiz-Vidal et al. / Nuclear Engineering and Design 313 (2017) 214224
quencies of water single-phase flow or air-water two-phase flow
excitation, xpeak, were directly extracted from the PSDacc for each
flow condition. PSDacc and resonance frequencies for air (xair) were
obtained as described in the Experimental work section.
" # is more evident for the third mode, where lower scatter of the data
2
xair
mh mp 1
xpeak
Fig. 7 illustrates the variation of hydrodynamic mass normal-
ized by the corresponding single-phase flow value (mh,l) as a func-
tion of twophase flow parameters for the span-75d system.
Hydrodynamic mass results as a function of void fraction, for the
first and third modes and both transversal directions, are plotted
in Fig. 7a. It is observed that hydrodynamic mass decreases with
increasing void fraction, according to literature (e.g. Pettigrew
and Taylor, 1994). Furthermore, the results are in reasonable
agreement with single-phase flow theory, where hydrodynamic
mass is proportional to mixture density (cm = 1, Eq. (22)). Experi-
mental points are localized around the line established by theory.
Two more important matters should be noted. The difference
between hydrodynamic mass in the Y and Z directions is not signif-
icant. On the other hand, the scatter of the data is larger in first-
mode results. This can be caused by the presence of two-phase
flow phenomena, for example, liquid slugs with low frequencies
affect the first-mode peak frequency response. Similar results are
found for the system span 40d.
Fig. 7b presents the hydrodynamic-mass ratio as a function of
mass flux. The set of air-water two-phase flow conditions of
Fig. 4 are plotted. The hydrodynamic mass is practically indepen-
dent of mass flux, except to b = 95%. Nevertheless, for constant
homogeneous void fraction, hydrodynamic mass slightly decreases
with increasing mass flux and then increases. Comparing Figs. 7b
and 4, it is noted that the increment occurs when the dispersed-
flow-pattern boundary is reached. At this point, the hydrodynamic
coupling between pipe and flow is stronger, increasing the hydro-
dynamic mass. The opposite happens with the homogeneous void
fraction conditions of 95%; hydrodynamic-mass ratio decreases
with increasing mass flux. This behavior can be attributed to inter-
mittent flow, for which the hydrodynamic coupling is weak. For
75% of homogeneous void fraction, the presence of both opposite
effects described above can explain the almost constant value of
hydrodynamic mass.
The hydrodynamic mass results illustrated in Fig. 7 were not
reported before. They are similar to results reported for cross-
flow, e.g. Olala et al. (2014) and Pettigrew et al. (1989). In this
Fig. 7. Hydrodynamic mass ratio as a function of (a) void fraction and (b) mass flux. Void fraction is calculated using Armand-Massina correlation, a = (0.833 + 0.167x)b,
where x is quality. Room temperature value of 20 C is used. The results are for the system span 75d, where open and solids symbols correspond to first and third modes,
respectively. In (b), the homogeneous void fraction (b) classification is based on the nominal value.
221
sense, the equivalent mass of fluid vibrating with the structure
has similar behavior in both two-phase horizontal pipe flow and
cross-flow excitations. On the other hand, the use of single-phase
fluid theory appears reasonable from a practical point of view. It
related to little influence of mass flux is observed.
27
4.3. Validation of the proposed analytical approach
4.3.1. Relationship between standard deviation of acceleration and
shear velocity
The validation of proposed analytical formulations is presented
in this section. First, the relationship between standard deviation
of acceleration (STDacc) and shear velocity (J) is evaluated. Accord-
ing to the proposed formulation, there is a quadratic relation
between both parameters. Thus, a quadratic fit, with its respective
coefficient of determination, R2, accompanies the results of STDacc
versus J. The influence of direction response and length span is also
assessed. For the former, results of both transversal directions (Y
and Z axis) of system span 75d are presented; for the latter, Z-
axis results for both systems, spans 40d and 75d, are compared.
Fig. 8 presents the results for low homogeneous void fractions,
including water single-phase flow (b = 0%). As expected, standard
deviation of acceleration increases with increasing shear velocity.
Reasonable good quadratic fits (R2 P 0.92) are observed, especially
in two-phase flow (R2 P 0.98), indicating that the quadratic trend
predicted by the proposed formulation is valid. Results suggest low
influence of span length on STDacc. This behavior can be explained
by low shear velocities and, consequently, low levels of turbulence.
The latter is not large enough to affect the pipe response level. It is
important to highlight the fact that the proposed formulation does
not establish the influence of span length.
Other relevant matters should be observed in Fig. 8. Standard
deviation of acceleration increases with increasing the homoge-
neous void fraction, as previously reported by Geng et al. (2012)
and Ortiz-Vidal et al. (2013). Also, STDacc for Y-axis is larger than
for Z-axis. This can be caused by the gravity field. The fluid mass
would be acting in Z-axes as a natural damper reducing the vibra-
tion response. Concerning the shape of the fit, experimental points
follow a concave-up quadratic trend. An approach based on flow
pattern can be useful to explain this fact. For low homogeneous
void fraction (10% and 25%), according to Ortiz-Vidal et al.
(2014), bubbles break gradually into small bubbles with increasing
the mixture velocity. The turbulence increases progressively and
222 L.E. Ortiz-Vidal et al. / Nuclear Engineering and Design 313 (2017) 214224

Fig. 8. Standard deviation of acceleration as a function of shear velocity for low homogeneous void fraction, including water single-phase flow (b = 0%). The systems span 75d
and 40d are represented by square and triangular dots, respectively. Open and solid dots corresponds to Y and Z axes, respectively. The dashed line represents the quadratic
trend according to formulation proposed. R2 is its coefficient of determination. Vertical solid lines indicate flow-pattern boundary transitions as shown in Fig. 4.

excites the pipe in the same way. The last explanation is phe-
nomenologically compatible with experimental results by
Giraudeau et al. (2013). Measurements of flow-induced forces in
an elbow show the same concave-up quadratic trend at bubbly-
to-dispersed flow transition. Thereby, the pipe system responds
to this specific flow-excitation with a concave-up trend.
Fig. 9 presents the results for high homogeneous void fractions.
Similarly, (i) STDacc increases with increasing shear velocity, as
expected by theory, (ii) STDacc for Y-axis is slightly higher than
for Z-axis and (iii) STDacc increases with increasing the homoge-
neous void fraction. Good coefficients of determination
(R2 P 0.96) demonstrate that the quadratic trend predicted by
the proposed formulation is consistent. By contrast with low
homogeneous void fraction, some influence of span length on
STDacc is observed, especially for b = 50% and 75%. This behavior
may be due to the presence of high shear velocities and an inter-
mittent flow pattern, which excites the pipe systems in different
manners.
Fig. 9 also illustrates the effect of flow pattern and flow-pattern
transition. On one hand, values of standard deviation of accelera-
tion are higher for slug and dispersed flow patterns than for bubbly
flow, which can be attributed to periodic changes in the systems
mass (liquid-slug momentum exchange) and intense turbulence,
respectively. This findings are in accordance with previous studies
(Geng et al., 2012; Ortiz-Vidal et al., 2013). On the other hand, a
concave-down trend is observed in the presence of flow-pattern
transition. It can be related to the flow-induced forces. According
to Giraudeau et al. (2013), the two-phase flow excitation increases
with increasing mixture velocity with a greater rate in slug flow
than in dispersed flow. The two-phase damping can be also respon-

Fig. 9. Standard deviation of acceleration as a function of shear velocity for high homogeneous void fraction. The systems span 75d and 40d are represented by square and
triangular dots, respectively. Open and solid dots corresponds to Y and Z axes, respectively. The dashed line represents the quadratic trend according to formulation proposed.
R2 is its coefficient of determination. Vertical solid lines indicate flow-pattern boundary transitions as shown in Fig. 4.
 L.E. Ortiz-Vidal et al. / Nuclear Engineering and Design 313 (2017) 214224
Fig. 10. Peak-frequency deviation as a function of mixture velocity for several two-
phase damping ratios. Eqs. (21) and (22) were used to calculated xpeak. xpeak-ref
corresponds to peak frequency with zero two-phase damping (ftp = 0%). The
following values were used: fs = 1.5% and cm = 1.
sible for a lower response in dispersed flow, according to Bguin
et al. (2009). This explains the trend for 50% and 75% homogeneous
void fractions. In the case of 90%, the concave-down trend can be
caused by the proximity of transition to annular flow pattern,
where the flow-induced forces start to decrease because of the
small variation of instantaneous homogeneous-void-fraction
(Giraudeau et al., 2013).
4.3.2. Relationship between peak frequency and void fraction
In the second part of this section, Eqs. (21) and (22), the pro-
posed formulation, which relates peak frequency (xpeak) to void
fraction (a), is tested. First, the matters of hydrodynamic mass
and damping should be discussed. As demonstrated in Section 4.2,
the use of single-phase fluid theory is reasonable to estimate
hydrodynamic mass, especially for the third mode. This means that
this parameter (Eq. (22)) can be calculated using cm = 1. In the case
of damping, on the other hand, the literature reports that the total
damping ratio (fT, Eq. (21)) for pipe flow can be considered as the
sum of both structural (fs) and two-phase (ftp) damping compo-
nents (Bguin et al., 2009; Gravelle et al., 2007). Furthermore, the
latter component ranging from 1 to 3.5% for a pipe diameter simi-
lar to the present study. Note that here only structural damping
component was obtained from experiments.
We analyzed the effect of two-phase damping on peak fre-
quency predicted by Eq. (21). Simulations with two-phase damp-
ing ratios of 0%, 1.5%, 2.5 and 3.5% were carried out for two-
phase flow conditions of Fig. 4. For each damping ratio, hydrody-
namic mass was calculated by means of Eq. (22) and the
Armand-Massina correlation (see Fig. 7). These results were used
to obtain the peak frequency (xpeak, Eq. (21)). Structural parame-
ters presented in the Experimental work section were used.
Fig. 10 presents peak-frequency deviations, taking as reference
the peak frequency with zero two-phase damping. As can be
observed, there is very small influence of two-phase damping on
the proposed peak-frequency formulation. It is noteworthy that
this fact does not indicate the irrelevance of two-phase damping
on flow-induced vibration. Rather, other parameters have more
impact on the proposed peak-frequency formulation, such as
hydrodynamic mass.
Fig. 11 presents the performance of the proposed peak-
frequency formulation against measured data for the third-mode
223
Fig. 11. Comparison of measured (xpeak-experimental) and predicted (xpeak-predicted)
peak frequency for the third-mode of both systems, spans 75d and 40d. Open and
solid dots correspond to Y and Z axes, respectively. The homogeneous void fraction
classification is based on the nominal value, according to Fig. 4. The dashed line
indicates 10% of percent error.
of both systems, spans 75d and 40d. Experimental peak frequencies
were extracted directly from the PSDacc for each flow condition. In
the case of the predictions, and based on discussion above, the fol-
lowing values were used: cm = 1, fs = 1.5% (measured mean value)
and ftp = 2.5% (arbitrary). As observed, the proposed formulation
predicts the experimental peak-frequencies with good accuracy,
97.6% of the data within 10% of point spreading. This fact demon-
strates that the relationship between peak frequency and void frac-
tion predicted by Eqs. (21) and (22) is consistent.
5. Conclusions
Flow-induced vibration subject to two-phase pipe flow is the
main goal of this paper. Analytical formulations that relate flow
to dynamic response parameters are presented and tested. Two
span lengths of a clamped-clamped straight horizontal pipe were
subjected to a broad range of two-phase flow conditions, including
bubbly, dispersed and slug flow patterns. The nature of the flow-
excitation and flow-parameters influence and hydrodynamic mass
were deduced from the vibration response. The following remarks
summarize the conclusions of the present study.
(1) Vibration response depends on flow and structural parame-
ters such as mixture velocity, homogeneous void fraction
and flow pattern. The effect of two-phase damping and
hydrodynamic mass is also observed. Constant stiffness
appears to be a valid assumption.
(2) Turbulent flow excites the structure in a wide range of fre-
quencies at a constant amplitude excitation level.
(3) Standard deviation of acceleration (STDacc) is a function of
the square of shear velocity, corroborating the proposed ana-
lytical formulation. Furthermore, STDacc increases with
increasing homogeneous void fraction.
(4) Shear velocity appears to be an appropriate velocity scale to
relate to the vibration response. It captures the flow phe-
nomena at the wall, where turbulence excitation mechanism
happens.
(5) Peak frequency is strongly correlated to hydrodynamic mass.
The proposed peak-frequency formulation has been consis-
tent and accurate to predict it.
224 L.E. Ortiz-Vidal et al. / Nuclear Engineering and Design 313 (2017) 214224
(6) The use of single-phase fluid theory is reasonable to esti-
mate hydrodynamic mass, especially for the third mode.
Therefore, a hydrodynamic mass coefficient cm = 1 can be
adopted for two-phase pipe flow.
(7) Hydrodynamic mass due to two-phase pipe flow excitation
has the same behavior as cross-flow excitation, for which
this dynamic parameter is practically independent of mass
flux.
There are still some aspects demanding further investigation,
such as the effect of the boundary and length span on the level
of pipe response. The latter is not established by the proposed
STDacc formulation. The influence of two-phase flow pattern should
be also quantified.
Acknowledgment
The authors are grateful to Benedict Besner, Thierry Lafrance
and Cedric Bguin for their support in the experimental facilities
and discussions, and to Irma Consuelo Aguilar Avila for helping
in data collect. L. Enrique Ortiz Vidal gratefully acknowledges the
support of the Petrobras/CNPq and of BWC/AECL/NSERC Chair of
Fluid-Structure Interaction, Polytechnique Montreal. Oscar M. H.
Rodriguez is grateful to CNPq for the research grant.
References
Allen, R.L., Mills, D., 2004. Signal Analysis: Time, Frequency, Scale, and Structure.
Wiley-IEEE Press.
Au-Yang, M.K., 1975. Response of reactor internals to fluctuating pressure forces.
Nucl. Eng. Des. 35, 361375.
Bguin, C., Anscutter, F., Ross, A., Pettigrew, M.J., Mureithi, N.W., 2009. Two-phase
damping and interface surface area in tubes with vertical internal flow. J. Fluids
Struct. 25, 178204.
Blevins, R.D., 1979a. Formulas for Natural Frequency and Mode Shape. Van
Nostrand Reinhold, New York, NY, United States.
Blevins, R.D., 1979b. Flow-induced vibration in nuclear reactors: a review. Prog.
Nucl. Energy 4, 2549.
Blevins, R.D., 1989. An approximate method for sonic fatigue analysis of plates and
shells. J. Sound Vib. 129, 5171.
Blevins, R.D., 2001. Flow-Induced Vibration. Van Nostrand Reinhold, New York.
Carlucci, L.N., 1980. Damping and hydrodynamic mass of a cylinder in simulated
two-phase flow. J. Mech. Des. 102, 597602.
Charreton, C., Bguin, C., Ross, A., Etienne, S., Pettigrew, M.J., 2014. Two-phase
damping in vertical pipe flows: effect of void fraction, flow rate and external
excitation. ASME 2014 Pressure Vessels and Piping Conference. Volume 4:
Fluid-Structure Interaction. ASME, p. V004T04A048.
Chen, S.S., 1991. Flow-induced vibrations in two-phase flow. J. Pressure Vessel
Technol. 113, 234241.
Chung, H., 1985. Flow-induced vibrations of a circular cylinder due to turbulent
internal flows. In: Offshore Mechanics and Arctic Engineering Symposium.
Dallas, TX, USA, pp. 398402.
Chung, H., Chen, S.S., 1984. Hydrodynamic mass (Book No. CONF-8406479).
Cunningham, P.R., White, R.G., 2004. A review of analytical methods for aircraft
structures subjected to high-intensity random acoustic loads. Proc. Inst. Mech.
Eng., Part G: J. Aerosp. Eng. 218, 231242.
Durant, C., Robert, G., 1998. Vibro-acoustic response of a pipe excited by a turbulent
internal flow. Flow Turbul. Combust. 61, 5569.
Evans, R.P., Blotter, J.D., Stephens, A.G., 2004. Flow rate measurements using flow-
induced pipe vibration. J. Fluids Eng. 126, 280.
Fujita, K., 1990. Flow-induced vibration and fluid-structure interaction in nuclear
power plant components. J. Wind Eng. Ind. Aerodyn. 33, 405418.
Geng, Y., Ren, F., Hua, C., 2012. An auxiliary measuring technology of wet gas flow
based on the vibration signals of the pipe. Flow Meas. Instrum. 27, 113119.
Giraudeau, M., Mureithi, N.W., Pettigrew, M.J., 2013. Two-phase flow-induced
forces on piping in vertical upward flow: excitation mechanisms and
correlation models. J. Pressure Vessel Technol. 135, 30907.
Gravelle, A., Ross, A., Pettigrew, M.J., Mureithi, N.W., 2007. Damping of tubes due to
internal two-phase flow. J. Fluids Struct. 23, 447462.
Hibiki, T., Ishii, M., 1998. Effect of flow-induced vibration on local flow parameters
of two-phase flow. Nucl. Eng. Des. 185, 113125.
Ishii, M., Hibiki, T., 2006. Thermo-Fluid Dynamics of Two-Phase Flow. Springer.
Jendrzejczyk, J.A.A., Chen, S.S.S., 1985. Experiments on tubes conveying fluid. Thin-
Walled Struct. 3, 109134.
Laufer, J., 1953. The Structure of Turbulence in Fully Developed Pipe Flow (No.
NACA-TR-1174). Washington, DC, United States.
Mandhane, J.M., Gregory, G.A., Aziz, K., 1974. A flow pattern map for gasliquid flow
in horizontal pipes. Int. J. Multiph. Flow 1, 537553.
Metzger, M.M., Klewicki, J.C., 2001. A comparative study of near-wall turbulence in
high and low Reynolds number boundary layers. Phys. Fluids 13, 692.
Olala, S., Mureithi, N.W., Sawadogo, T., Pettigrew, M.J., 2014. Streamwise fluidelastic
forces in tube arrays subjected to two-phase flows. ASME 2014 Pressure Vessels
and Piping Conference. Volume 4: Fluid-Structure Interaction. ASME, p.
V004T04A014.
Ortiz-Vidal, L.E., 2012. Design of an Experimental Facility for Two-Phase Pipe Flow
(Diseo de un banco experimental para la generacin de flujo bifsico agua-
aire). Pontifical Catholic University of Peru (PUCP).
Ortiz-Vidal, L.E., Mureithi, N., Rodriguez, O.M.H., 2014. Two-phase friction factor in
gas-liquid pipe flow. Therm. Eng. 13, 8188.
Ortiz-Vidal, L.E., Rodriguez, O.M.H., 2011. Flow-induced vibration due to gas-liquid
pipe flow: knowledge evolution. 21st Brazilian Congress of Mechanical
Engineering - COBEM2011. ABCM, Natal, RN.
Ortiz-Vidal, L.E., Rodriguez, O.M.H., Mureithi, N.W., 2013. An Exploratory
Experimental Technique to Predict Two-Phase Flow Pattern From Vibration
Response. ASME2013 Pressure Vessels and Piping Conference. Volume 4: Fluid-
Structure Interaction. ASME, Paris, p. V004T04A061.
Padoussis, M.P., 2006. Real-life experiences with flow-induced vibration. J. Fluids
Struct. 22, 741755.
Pettigrew, M.J., Taylor, C.E., 1994. Two-phase flow-induced vibration: an overview.
J. Pressure Vessel Technol., Trans. ASME 116, 233253.
Pettigrew, M.J., Taylor, C.E., Fisher, N.J., Yetisir, M., Smith, B.A.W., 1998. Flow-
induced vibration: recent findings and open questions. Nucl. Eng. Des. 185,
249276.
Pettigrew, M.J., Taylor, C.E., Kim, B.S., 1989. Vibration of tube bundles in two-phase
cross-flow: part 1hydrodynamic mass and damping. J. Pressure Vessel
Technol. 111, 466.
Riverin, J.-L., Pettigrew, M.J., 2007. Vibration excitation forces due to two-phase
flow in piping elements. J. Pressure Vessel Technol. 129, 713.
Rodriguez, O.M.H., Bannwart, A.C., 2008. Stability analysis of core-annular flow and
neutral stability wave number. AIChE J. 54, 2031.
Rodriguez, O.M.H., Castro, M.S., 2014. Interfacial-tension-force model for the wavy-
stratified liquidliquid flow pattern transition. Int. J. Multiph. Flow 58, 114
126.
Rosa, E.S., 2012. Escoamento Multifsico Isotrmico. ARTMED EDITORA S.A, Porto
Alegre.
Sankaran, R., Jancauskas, E.D., 1992. A technique for the measurement of the cross
wind joint acceptance function. In: University of Tasmania (Ed.), 11th
Australasian Fluid Mechanics Conference. Hobart, Australia, pp. 973976.
de Silveira Neto, A. , 2002. Fundamentos da Turbulncia nos Fluidos, in: Associaao
Brasileira de Cincias Mecnicas (ABCM) (Ed.), Coleao Cadernos de
Turbulncia. ABCM, Rio de Janeiro, pp. 152.
van Hout, R., Shemer, L., Barnea, D., 2003. Evolution of hydrodynamic and statistical
parameters of gasliquid slug flow along inclined pipes. Chem. Eng. Sci. 58,
115133.
Wallis, G.B., 1969. One-Dimensional Two-Phase Flow. McGraw-Hill.
Weaver, D.S., Ziada, S., Au-Yang, M.K., Chen, S.S., Padoussis, M.P., Pettigrew, M.J.,
2000. Flow-induced vibrations in power and process plant components
progress and prospects. J. Pressure Vessel Technol. 122, 339348.
White, F.M., 1998. Fluid Mechanics, McGraw-Hill series in mechanical engineering.
McGraw-Hill Higher Education.
Yang, C.Y., 1986. Random Vibration of Structures. Wiley-Interscience, New York.
Yi-min, H., Yong-shou, L., Bao-hui, L., Yan-jiang, L., Zhu-feng, Y., 2010. Natural
frequency analysis of fluid conveying pipeline with different boundary
conditions. Nucl. Eng. Des. 240, 461467.
Zhang, M., Xu, J., 2010. Effect of internal bubbly flow on pipe vibrations. Sci. China
Technol. Sci. 53, 423428.