Vous êtes sur la page 1sur 22

Flow-Excited Acoustic

Resonance Excitation
Mechanism, Design Guidelines,
and Counter Measures
The excitation mechanism of acoustic resonances has long been recognized, but the
Samir Ziada1 industry continues to be plagued by its undesirable consequences, manifested in severe
e-mail: ziadas@mcmaster.ca vibration and noise problems in a wide range of industrial applications. This paper
focuses on the nature of the excitation mechanism of acoustic resonances in piping sys-
Philippe Lafon tems containing impinging shear flows, such as flow over shallow and deep cavities. Since
this feedback mechanism is caused by the coupling between acoustic resonators and
Laboratoire de Mecanique des Structures shear flow instabilities, attention is focused first on the nature of various types of acoustic
Industrielles Durables (LaMSID), resonance modes and then on the aeroacoustic sound sources, which result from the
UMR EDF-CNRS-CEA 8193, interaction of the inherently unstable shear flow with the sound field generated by the res-
EDF R&D, onant acoustic modes. Various flow-sound interaction patterns are discussed, in which
Clamart, France 92141 the resonant sound field can be predominantly parallel or normal to the mean flow direc-
tion and the acoustic wavelength can be an order of magnitude longer than the length
scale of the separated shear flow or as short as the cavity length scale. Since the state of
knowledge in this field has been recently reviewed by Tonon et al. (2011, Aeroacoustics
of Pipe Systems With Closed Branches, Int. J. Aeroacoust., 10(2), pp. 201276), this ar-
ticle focuses on the more practical aspects of the phenomenon, including various flow-
sound interaction patterns and the resulting aeroacoustic sources, which are relevant to
industrial applications. A general design guide proposal and practical means to alleviate
the excitation mechanism are also presented. These are demonstrated by two examples of
recent industrial case histories dealing with acoustic fatigue failure of the steam dryer in
a boiling water reactor (BWR) due to acoustic resonance in the main steam piping and
acoustic resonances in the roll posts of the Short Take-Off and Vertical Lift Joint Strike
Fighter (JSF). [DOI: 10.1115/1.4025788]

1 Introduction downstream, they are amplified by the flow instability into vortex-
like structures [2628]. Further downstream, the phase of the reso-
Flow-excited acoustic resonances are often encountered in
nant sound field becomes favorable and the vortical structures
many engineering applications involving internal or external
generate acoustic energy, which sustains the resonance of the
flows. The acoustic pressure associated with such resonances can
acoustic mode. The main events of this excitation mechanism are
cause severe noise and vibration problems and, in some cases,
schematically illustrated in Fig. 1 for the case of flow over the
endanger the structural integrity of the installation. Flow-excited
mouth of a deep cavity or a closed side-branch. This flow geome-
acoustic resonance has been reported for piping systems convey-
try is encountered in many industrial applications and is therefore
ing gas [1,2] or liquid [3,4] flows and for various fluid-handling
extensively investigated in the literature [2941].
components, including flow control devices [57], closed-side
Since the theoretical, numerical, and experimental techniques
branches [813], turbomachines [14,15], boilers and heat
used to characterize the aeroacoustic sources generated by the
exchangers [1618], cavities in moving vehicles [19,20], and
interaction of unsteady flows with sound fields have been
many other applications. In the majority of these examples, the
reviewed recently by Tonon et al. [29], only a brief summary of
resonance is excited by unstable separated flows, such as shear
the progression of these techniques is given here, and interested
layers, jets, or bluff body wakes. Rockwell and Naudascher
readers are referred to the recent review by Tonon et al. for more
[2123] classified this excitation mechanism of acoustic reso-
specifics of these techniques.
nance as fluid-resonant in contrast to the fluid-dynamic mecha-
In early work, the growth of small, wave-like disturbances in
nism, which generates self-sustained oscillations of impinging
the shear layer is usually modeled with the linear stability theory
shear flows in the absence of resonance effects. In the fluid-
[26]. The disturbances then give rise to an acoustic source, which
resonant mechanism, which is the subject matter of this paper, the
excites the oscillation, thus completing the feedback loop
upstream feedback event, which sustains the oscillation, is pro-
[21,31,42]. However, in the case of fluid resonances, the upstream
vided by the resonant sound field. At the upstream separation
feedback introduced by the particle velocity of the resonant acous-
edge, the sound field induces velocity perturbations in the unstable
tic mode can be as high as 1%, or higher, of the mean velocity,
separated flow. Blevins [24] and Hall et al. [25], for example,
which is sufficiently large to promote the disturbance growth and
have illustrated that sound can shift the frequency of vortex
the shear-layer oscillation into the nonlinear regime very rapidly.
shedding from the natural frequency and enhance the spanwise
It is therefore more complex to describe the fluid resonant
coherence of vortex formation. As the initial perturbations travel
mechanism analytically.
Nelson et al. [43] used a different approach to analyze the prob-
Permanent address: Mechanical Engineering, McMaster University, Hamilton,
ON L8S 4L7, Canada.
lem of flow-excited resonance in a Helmholtz resonator. They
Manuscript received September 21, 2012; final manuscript received October 15, assumed the shear-layer vorticity to be concentrated in a series of
2013; published online December 3, 2013. Editor: Harry Dankowicz. line vortices, which are convected along the branch opening at a

Applied Mechanics Reviews Copyright V

C 2014 by ASME JANUARY 2014, Vol. 66 / 010802-1

Downloaded From: http://appliedmechanicsreviews.asmedigitalcollection.asme.org/ on 07/16/2015 Terms of Use: http://asme.org/terms

generates a larger amount of energy during the latter part of the
cycle than it absorbs during the earlier part.
Bruggeman et al. [47,48] modeled the flow at the opening of
resonant side-branches with discrete vortices and identified the
mechanism by which the moving vortices excite the acoustic os-
cillation. In this model, the acoustic power generated by the shear
layer is proportional to the acoustic amplitude, i.e., the acoustic
pressure induced by the shear layer oscillation is independent of
the acoustic amplitude. The model also shows that the occurrence
of resonance is dependent, to a large extent, on the timing between
the acoustic oscillation and the propagation of the vortex along
the opening of the branch. This latter feature was also confirmed
by means of more elaborate numerical techniques, which employ
a two-dimensional potential flow model based on the vortex blob
method in conjunction with Howes acoustic analogy (Eq. (1)) to
compute the acoustic power generated by the interaction of the
Fig. 1 Schematic presentation of flow-excited acoustic reso-
vortex blobs and the sound field. For example, Stoneman et al.
nance mechanism [49] explored the aeroacoustic sources for the case of flow over
two tandem plates in a duct, in which vortex shedding from the
plates excited the acoustic mode perpendicular to the flow, and
constant phase speed. The physics of the flow-acoustic interaction Hourigan et al. [50] investigated the excitation of longitudinal
was then investigated in terms of momentum and energy acoustic modes in a duct housing two baffles in close proximity.
exchanges. This alternative approach was explored because of the Kriesels et al. [37] used a similar approach to simulate the flow in
fact that the disturbances in the shear layer are large and cannot T-junctions and calculate the acoustic power generated by the
be modeled appropriately by the linear or weakly nonlinear stabil- separated shear layer.
ity theories. To support this approach, Nelson et al. [44] carried The importance of the timing between the acoustic oscillation
out measurements of the flow field in the neck of a Helmholtz and the vortex convection in the shear layer was also demon-
resonator excited to peak amplitude. strated experimentally by Graf and Durgin [51], using laser Dopp-
Substantial progress in understanding and modeling the acous- ler velocimetry measurements during the resonance of a deep
tic resonance mechanism has been achieved since Howe [45,46] cavity. In an extended vortex model based on Bruggemans origi-
formulated his acoustic analogy, in which vorticity is identified as nal idea, Graf [52] included the imaginary component of the exci-
a source of sound. He showed that the instantaneous acoustic tation source. With this complex valued excitation source, the
power, p, generated by the convection of the vorticity field, x, phase shift between shear-layer excitation and acoustic oscillation
within a fluid volume, 8, can be expressed as can be taken into consideration. The above-mentioned studies,
however, were not intended to predict the amplitude of pulsation,
p q x  v  u d8 (1) especially at large amplitude, but rather to better understand the
excitation mechanism and predict the range of flow velocity over
which an acoustic mode may be excited.
where q is the fluid density, v is the fluid velocity vector, and u is Subsequently, more elaborate models and numerical techniques
the particle velocity vector of the sound field. Whether vorticity were used to predict the pulsation amplitude as a function of the
acts as an acoustic source or sink depends on the triple product in flow velocity. Radavich et al. [53] simulated the two-dimensional
Eq. (1). Resonances are self-generated if the integral of Eq. (1) flow at a T-junction by means of an implicit, noniterative method
over an acoustic cycle is positive, which imposes a phase condi- to solve the unsteady, turbulent, and compressible NavierStokes
tion on the timing of vortex convection within the resonant sound equations. Although this simulation predicted correctly the extent
field. Figure 2 illustrates this phenomenon for the case of a deep- of the lock-in velocity range and the flow patterns of the simula-
cavity resonance. During the earlier part of the oscillation cycle, tion exemplified the results of the flow visualization study
the particle velocity is downwards and the vortex absorbs acoustic reported by Ziada [34], the amplitude of pulsation was underpre-
energy, whereas during the latter part, the particle velocity is dicted by 50%. Dequand et al. [38] used Euler equations for two-
upwards and the vortex generates acoustic energy. The net acous- dimensional compressible flow to perform numerical simulations
tic energy over a complete cycle is positive in this case because at the junction of the coaxial branch case. This latter approach
the vortex is more developed, or stronger, and therefore it improved the prediction of the maximum pulsation amplitude to
within 40% of the experimental results for the two-dimensional
coaxial case. At other flow conditions, the prediction of the
pulsation amplitude was less accurate.
Martnez-Lera et al. [54] has recently introduced a different
approach for modeling the interaction of compact aeroacoustic
sources with long wavelength sound fields. They used a commer-
cial solver for the laminar, incompressible, and two-dimensional
NavierStokes equations to predict the aeroacoustic sources in
T-junctions with various flow patterns. Although the Reynolds
number of these simulations is low, less than 3000 based on the
side-branch width, and the effect of turbulence is excluded, the
results agreed qualitatively with those obtained from more elabo-
rate numerical approaches. Nakiboglu et al. [55,56] used the
methodology introduced by Martnez-Lera et al. [54] to investi-
gate the details of flow-sound interaction of single and multiple
shallow cavities in a pipeline. These studies provided very prom-
Fig. 2 Flow visualization and acoustic particle velocity at two ising results reflecting the effects of flow velocity, sound intensity,
different instants during the acoustic resonance cycle of a deep and approach velocity profile on the aeroacoustic sources. Since
cavity. Reprinted from Ref. [13] with permission from ASME. these results are based on laminar, two-dimensional flow

010802-2 / Vol. 66, JANUARY 2014 Transactions of the ASME

Downloaded From: http://appliedmechanicsreviews.asmedigitalcollection.asme.org/ on 07/16/2015 Terms of Use: http://asme.org/terms

simulations, they still need to be validated quantitatively against
data collected for high Reynolds number turbulent pipe flow.
Although the models and numerical methods discussed above
have shown promise in predicting moderate amplitude pulsations
for two-dimensional applications, they do not seem suitable for
problematic industrial situations, which involve nonlinear, large
amplitude pulsations excited by three-dimensional turbulent
flows. To overcome this difficulty, Graf and Ziada [39,41] and
Mohamed et al. [57] developed an experimental approach to char-
acterize aeroacoustic sources under complex flow situations simi-
lar to those encountered in industrial applications. In this
approach, the time-dependent integral effect of the flow-sound
interaction at the source is modeled by a complex pressure differ-
ence Dp, whose amplitude and phase are dependent on the
Strouhal number and local particle velocity amplitude. The acous-
tic power generated by the unsteady flow is equal to the induced
pressure difference multiplied by the local acoustic particle veloc-
ity. For steady-state oscillation, the power generated by the
unsteady shear layer must be balanced by the viscothermal dissi-
pation and radiation losses of acoustic energy. Predictions of self-
excited oscillations using the aeroacoustic sources determined by
this method agree very well with the experimentally determined Fig. 3 Various arrangements of closed side-branches and
lock-on range as well as the pulsation amplitude as a function of associated acoustic pressure (p) distributions of the first
flow velocity. acoustic mode. The arrows in the bottom figures indicate the
In this article, attention is focused first on acoustic modes of acoustic flux of the resonant acoustic modes. Reprinted from
Ref. [35] with permission from Elsevier Publishing.
confined volumes, including global and trapped modes. There-
after, the most industry-relevant flow-sound interaction patterns
and the general features of the acoustic resonance phenomenon schematically shown in Figs. 36. Several other configurations of
are discussed. This is then followed by design guidelines for piping systems that can generate flow-excited acoustic resonances
selected flow geometries, which often experience acoustic are discussed in the review article by Tonon et al. [29].
resonances. The guidelines include charts of the Strouhal number
at the onset of resonance as well as the characteristics of aeroa- 2.1.1 Closed Side-Branches. Figure 3 shows three arrange-
coustic sources for typical flow-sound interaction patterns. Reme- ments of closed side-branches together with the acoustic pressure
dial methods to alleviate flow-excited acoustic resonances are also distribution of the lowest resonance mode in each case as well as
discussed. Finally, two industrial examples involving flow-excited the patterns of the acoustic flux at the junction generated by these
acoustic resonance are briefly described and effective means to al- modes. Note that the main pipe diameter, D, and the distance
leviate the acoustic resonance mechanism in both cases are between the tandem branches, , are much smaller than the acous-
discussed. tic wavelength of the resonant mode, k. In the single-branch case,
Fig. 3(a), the pulsation amplitude in the branch is strongly influ-
enced by the acoustic radiation into the main pipe and, conse-
2 Acoustic and Shear Layer Modes quently, by the viscothermal losses in the main pipe as well as by
the radiation losses at the main pipe terminations. However, as the
2.1 Acoustic Modes. As in the case of structural resonance branch pipe diameter becomes substantially smaller than the main
modes and frequencies, acoustic modes of piping systems depend pipe diameter, d  D, the branch radiation into the main pipe
on the boundary conditions of the pipe terminations. If these ter- diminishes, as in the case of sound radiation from flanged open
minations are reactive, sound waves will reflect back into the pipe pipe, and the reflection coefficient of the branch becomes close to
and form standing waves, which are called the acoustic modes of unity [58]. In this latter case, strong resonances in the branch can
the system. These conditions result in accumulation of acoustic be excited but only if the branch is relatively short (k/4); other-
energy in the piping system, such that the acoustic modes can be wise, viscothermal losses in the branch can be large enough to
excited readily by flow turbulence in the pipe. Acoustic resonance dampen the resonance. Jungowski et al. [2] and Ziada and Shine
modes of confined volumes can be obtained by solving Helmholtz [35], for example, have shown that increasing the diameter ratio
equation subject to appropriate boundary conditions [58,59]. If d/D for single side-branches results in a sharp reduction in the
sound attenuation in the medium or absorption at the terminations pulsation amplitude at resonance.
is included, the eigenvalue solutions are complex, where the real In the case of two branches, Figs. 3(b) and 3(c), which are well
part represents the frequency of the undamped oscillation and the tuned (i.e., of equal length) and in close proximity (i.e., << k),
imaginary part indicates the quality factor of the resonance mode. the acoustic flux at the mouth of one branch is equal but opposite
In practical applications, sound radiation is always present, and to that at the mouth of the other branch. In the case of the coaxial
therefore, the eigenvalues of acoustic modes are always complex. branches, for example, Graf and Ziada [39] found that only 2% of
Acoustic modes extending over the whole piping system are the acoustic power in the branches is radiated into the main pipe.
defined as global modes, and their response, or quality factor, The two branches therefore strongly couple and form a subsystem
depends strongly on the boundary conditions at the terminations. with negligible radiation losses into the main pipe. Therefore,
On the other hand, when the resonance modes are confined to a these lower modes of the coaxial branches are trapped modes with
portion of the piping system and have vanishingly small radiation negligibly small imaginary part of the eigenvalue [6063]. The
losses into the other pipes, they become decoupled from the pip- pulsation amplitude in this case can be drastically higher than that
ing system and are not influenced by the boundary conditions of in the case of a single branch. Ziada and Buhlmann [11] and
the system terminations. Since their energy becomes trapped in a Peters [64] have shown that this is particularly the case for large
subsystem, these modes are referred to in the literature as trapped values of d/D.
modes [6063]. In the case of the tandem branches of Fig. 3(b), radiation losses
To illustrate these different types of modes, selected pipe con- into the main pipe increase with the distance between the
figurations that are most commonly encountered in industry are branches. The lower modes in this case are not completely trapped

Applied Mechanics Reviews JANUARY 2014, Vol. 66 / 010802-3

Downloaded From: http://appliedmechanicsreviews.asmedigitalcollection.asme.org/ on 07/16/2015 Terms of Use: http://asme.org/terms

Fig. 5 Ducted shallow cavity, or two orifice plates, together
with the pressure distributions of the lowest three longitudinal
modes of the duct

resonance of the longitudinal modes of the branch pipes with a

pressure node at or near the T-junction. These modes with a pres-
sure node at the T-junction, the lowest of which is illustrated sche-
matically in Fig. 4, can be classified as (nearly) trapped modes
because of their negligible radiation losses into the main pipe.
Typical examples of this pipe arrangement can be found in power
plants, chemical installations, as well as in compressor and pump
stations of pipelines.
Figure 5 shows another pipe arrangement housing a shallow
cavity or two orifice plates in close proximity. This is a generic
flow geometry exemplifying flow separation and impingement
typical of flow-control devices in piping system. The figure also
Fig. 4 T-junction with a transition zone to combine the flow shows the acoustic pressure distributions for the lowest three lon-
from two branches into a main pipe. The length of the transition gitudinal modes of the pipeline. It should be noted that this
piece is 2LT. The top figure shows the lowest trapped mode generic flow geometry has also been used to mode a single
along the branches. corrugation of corrugated pipes conveying natural gas [55,56].

but nearly trapped modes because the imaginary parts of their 2.1.3 Trapped (or Cross) Modes of Ducted Shallow Cavities.
eigenvalues, albeit small, do not vanish. A special case of this ge- Acoustic trapped modes are known to exist in wave guides where
ometry is when the distance between the branches approximates the perturbation energy is localized in regions accommodating
k/2, and therefore, a pressure node forms at the mouth of each some changes in the domain geometry or the fluid properties [60].
branch. In this case, the modes become completely trapped again For example, the existence of shallow cavities in globe valves
because of the vanishing radiation losses. Another special case is reduces the local cross-mode frequency below the cut-off fre-
when a single branch is well tuned with the upstream or down- quency of the main duct. This reduction in frequency prevents the
stream pipe, which again generates a subsystem with trapped local acoustic energy generated at the cavity from propagating
modes. These latter special cases have been investigated in some away from the cavity along the duct [2] and thereby a new local
detail in Refs. [29], [37], [47], [48], and [65]. An interesting fea- resonance mode is introduced into the system. Duan et al. [61]
ture of the tandem branches is the difference in the acoustic parti- determined numerically the characteristics of such trapped modes
cle velocity fields at the mouth of each branch. While the at zero flow velocity for a two-dimensional cavity-duct system,
downstream corner of the upstream branch constitutes a singular- and Hein and Koch [62] and Koch [63] extended Duan et al.s
ity in the acoustic field, which leads to a large local acoustic parti- work [61] and predicted numerically the existence of nearly
cle velocity, this singularity exists at the upstream corner of the trapped diametral modes for axisymmetric cavity-duct configura-
downstream branch. As will be presented later, these differences tions, also for zero flow condition.
in the local acoustic velocity distributions affect the aeroacoustic In practice, these modes can be excited by flow turbulence
sources resulting from the flow-sound interaction occurring at the because they have very small damping [57,15,66]. Aly and Ziada
mouth of various arrangements of side-branches [39,41]. [67,68] investigated the excitation mechanism of the trapped dia-
metral modes of the internal axisymmetric cavity shown in Fig. 6.
2.1.2 Shallow Cavities and T-Junctions. T-junctions are often As can be seen from the acoustic pressure contours in this figure,
used to combine the flow from two pipes (branches) into a single the trapped modes consist of transverse standing waves and their
(main) pipe, as schematically shown in Fig. 4. Efficient high- acoustic pressure decays exponentially with distance from the
pressure piping maintains the flow speed close to the maximum cavity. This decay rate is a measure of the quality factor of the res-
allowable flow velocity, and therefore, these pipes are normally onance mode. The size of the cavity, relative to the pipe size, also
designed to maintain the flow velocity approximately constant in influences the decay rate and thereby the quality factor of reso-
different diameter pipes. To achieve this requirement for the pip- nance. Thus, as the cavity becomes larger, the quality factor of the
ing shown in Fig. 4, the main pipe diameter must be larger than nearly trapped modes increases and the self-excited resonance
that of the branch pipes, and this is accommodated by means of an becomes stronger and more robust against various means of its
expansion (or transition) section at the T-junction, as shown in the alleviation, as will be discussed later.
lower portion of Fig. 4. Because of the flow turbulence in the The contour plots shown in Fig. 6 correspond to the case of no
T-junction, this pipe arrangement is prone to flow-excited acoustic mean flow in the main duct, or zero Mach number conditions. At

010802-4 / Vol. 66, JANUARY 2014 Transactions of the ASME

Downloaded From: http://appliedmechanicsreviews.asmedigitalcollection.asme.org/ on 07/16/2015 Terms of Use: http://asme.org/terms

Fig. 8 Flow visualization of the lowest three shear-layer modes
(m 5 13) for (a) a coaxial side-branch resonator and (b) two ori-
fice plates in a pipeline. Insets (a1 and a2) are from Ref. [34]
with permission; (a3) from Ref. [70] with permission; and
(b1b3) from Ref. [71] with permission from Elsevier

2.2 Shear Layer (or Hydrodynamic) Modes. The shear-

Fig. 6 Ducted axisymmetric shallow cavity showing (a) main layer oscillations due to its inherent hydrodynamic instability are
dimensions; (b) longitudinal section showing contours of the the source of unsteadiness that act as the amplifier in the feedback
acoustic pressure of the first trapped cross mode; (c and d) loop generating self-sustained oscillations (Fig. 1). The growth of
acoustic pressure distributions along the duct length and vorticity disturbances in the shear layer undergoes an initial linear
across the cavity for the first and second cross modes.
growth phase followed by a nonlinear growth stage leading to
Adopted from Ref. [67] with permission from Elsevier
Publishing. roll-up of vortical structures and nonlinear amplitude saturation
[2628]. In the fluid-dynamic mechanism, which sustains the
oscillations of impinging shear layers [2123], acoustic resonance
effects are not present and the upstream feedback is generated by
the distortion of the vortical structures upon their impingement on
a downstream body, whether it is an edge, a plate, or a cavity cor-
ner. In the fluid-resonant mechanism, the upstream feedback is
provided by the resonant acoustic field, which is excited by the
sound power generated by the interaction between the shear-layer
oscillations and the resonant acoustic particle velocity field. This
interaction occurs along the whole impingement length between
the separation and impingement edges. As mentioned earlier, gen-
eration of positive sound power to sustain the oscillation requires
a favorable phase condition to be maintained between the convec-
tion of shear-layer disturbances (or vortices) and the sound field
oscillation. This phase condition can be satisfied at several shear-
layer oscillation frequencies, each of which corresponds to a dif-
ferent number of wavelengths (or vortices) along the impingement
Fig. 7 Simulation results of an axisymmetric cavity-duct sys- distance. The shear layer oscillation patterns associated with these
tem with and without mean flow showing contour plots of the various frequencies, which satisfy the phase condition, are
radial particle velocity amplitude for the first diametral mode.
The contours scale is similar in both figures. L/h 5 1 and h/
referred to as the shear layer modes (m) or the hydrodynamic
D 5 2/12, Adopted from Ref. [69] with permission from Elsevier modes, in contrast to the acoustic resonance modes (n), which
Publishing. were discussed in Sec. 2.1. Note that higher oscillation frequen-
cies, or higher-order shear layer modes, are associated with
shorter wavelengths of the shear-layer oscillation, and therefore,
more vortical structures are formed along the impingement length.
relatively moderate Mach numbers (M 0.10.3), Aly and Ziada The lowest three shear layer modes are illustrated in Fig. 8
[69] observed substantial changes in the pattern of the trapped [34,70,71] for two cases: the coaxial side-branch resonator shown
mode shapes, as can be seen in Fig. 7, which compares the acous- in Fig. 3(c) and two baffles forming a shallow cavity in a pipeline,
tic particle velocity distributions of the first diametral mode with- as shown in Fig. 5(b). For the first shear-layer mode (m 1), a sin-
out and with mean flow. Based on Howes acoustic analogy, gle vortex is formed along the impingement distance between sep-
Eq. (1), it is clear that the differences between the acoustic modes aration and impingement edges, whereas for the second and third
caused by the mean flow will affect substantially the amount of shear-layer modes (m 2 and 3), two and three vortices are
acoustic energy generation as well as the location of the aeroa- formed, respectively. In the first case, the coaxial side-branches,
coustic sources. the acoustic modes are trapped modes with negligibly small

Applied Mechanics Reviews JANUARY 2014, Vol. 66 / 010802-5

Downloaded From: http://appliedmechanicsreviews.asmedigitalcollection.asme.org/ on 07/16/2015 Terms of Use: http://asme.org/terms

Fig. 9 Finite element simulation of acoustic particle steam
lines near a baffle in a pipeline

Fig. 11 Acoustic response of coaxial side-branches with sharp

edges [29]. n is the acoustic mode number and m is the hydro-
dynamic mode of the shear layer. Reprinted from Ref. [29] with

Fig. 10 [72], with the top figure showing the observed lowest
shear-layer mode, which excites the strongest acoustic reso-
nance, and the bottom figure illustrating a higher shear-layer
mode producing a weaker resonance. As indicated in Fig. 10, the
particle velocity of the resonant acoustic mode is predominantly
Fig. 10 Flow structures during acoustic resonance in a along the axis of the inlet branches (in this case horizontal) and
T-junction combining the flow from two pipes into a single the vortical structures formed at the T-junction entrances travel
downstream pipe (Fig. 4). Top figure shows the lowest observed parallel to the particle velocity until they are diverted across the
shear-layer mode A, which produces the strongest acoustic acoustic streamlines into the outlet pipe. This latter stage of the
resonance, whereas bottom figure shows a higher shear-layer vortex diversion across the acoustic streamlines causes the gen-
mode B, producing a weaker resonance than that produced by eration of the acoustic energy needed to sustain the acoustic res-
mode A. Adopted from Ref. [72] with permission from ASME. onance. As will be seen later, the favorable phase condition for
the excitation of acoustic resonances is related to the length LT
of the enlarged diameter transition section of the T-junction
radiation losses. In addition, since the maximum particle velocity (Fig. 4).
of the acoustic modes occurs at the branch opening, where the
shear layer vortices form and convect with the flow in a direction 3 Main Characteristics of Fluid-Resonant Mechanism
normal to the acoustic velocity oscillation, the acoustic power pro-
duction is at its maximum according to Eq. (1). These unique fea- Figures 11 and 12 display the main features of typical flow-
tures of negligible radiation losses and efficient sound power excited resonance responses for two cases: the acoustic modes of
production make the well-tuned coaxial side-branches very liable well-tuned coaxial side-branches [37] and the trapped cross modes
to strong flow-excited acoustic resonances. of a ducted shallow cavity [67]. In the first case, the acoustic pres-
In the second case of Fig. 8, the acoustic velocity of the pipeline sure at the closed end of the side-branch and the corresponding
longitudinal modes is generally parallel to both the flow direction frequency are plotted against the mean flow velocity in the main
and the convection path of the shear layer vortices, which may pipe, V. As mentioned earlier, the resonance modes (n) of the
suggest zero acoustic power generation according to Eq. (1), coaxial side-branches consist of acoustic standing waves with a
because u and v are parallel. This apparent dilemma was tackled pressure node at the junction and odd multiples of quarter wave-
by Hourigan et al. [50], who illustrated that, because the stream- lengths in each branch. The frequencies of these modes for the
lines of the acoustic velocity bend around the baffle edges, a radial coaxial branches are given by
component of the acoustic particle velocity is introduced. As
shown in Fig. 9, this radial component of the acoustic streamlines 2n  1c
fn (2)
is normal to the direction of vortex convection past the down- 4LB
stream baffle. A similar situation also arises in the case of longitu-
dinal acoustic modes in pipelines housing shallow cavities or where c is the speed of sound and LB is the side-branch length.
corrugations [5557]. The hydrodynamic modes of the shear layer (m) are indicated by
In the case of a T-junction combining two flows into a single the solid lines, whose slopes approximate the Strouhal numbers
downstream pipe, as shown in Fig. 4, more complex shear-layer that satisfy the phase condition required to sustain the acoustic
modes can occur. Two of these modes are presented in resonance. The Strouhal number is defined by

010802-6 / Vol. 66, JANUARY 2014 Transactions of the ASME

Downloaded From: http://appliedmechanicsreviews.asmedigitalcollection.asme.org/ on 07/16/2015 Terms of Use: http://asme.org/terms

Fig. 13 Resonance characteristics of a pipeline containing
Fig. 12 Pressure contours showing acoustic pressure level as double orifice plates (adopted from Ref. [71] with permission
function of frequency and flow velocity for ducted axisymmetric from Elsevier Publishing). Solid lines correspond to shear layer
cavity. L/d 5 1, h/D 5 1/6; n is the acoustic diametral mode num- modes (m), and dashed lines represent the pipeline acoustic
ber, and m is the hydrodynamic mode number of the shear modes (n).
layer. Reprinted from Ref. [67] with permission from Elsevier
pipelines housing two orifice plates (Figs. 5 and 8(b)) or T-
junctions combining two flows (Figs. 4 and 10).
Harris et al. [73], Huang and Weaver [71], and Hourigan et al.
fLc [50] investigated the resonance mechanism of pipelines with
St (3)
V double orifice plates forming a shallow cavity. As illustrated in
Fig. 13, the three shear-layer modes (m 13) excite the acoustic
where Lc is a characteristic length related to the shear-layer modes of the pipeline when there is a frequency coincidence with
impingement length. For example, Lc (d r) for the case of an acoustic mode consisting of even multiples of half wavelengths
closed side-branches; Lc (L r) for shallow cavities and orifice (n 2, 4, 6). The odd-numbered acoustic modes are not excited
plates; and Lc (LT r) for T-junctions, where r is the radius of because they have a pressure maximum, i.e., zero particle veloc-
curvature of the upstream corner. ity, at the orifice plates, and therefore, no acoustic power can be
As can be seen in Fig. 11, each acoustic mode can be excited generated according to Eq. (1). Note that the data points in Fig. 13
over multiple ranges of flow velocity by different hydrodynamic marked b1b3 for the first resonant mode (n 2) indicate the test
modes. Similar features of the acoustic resonance mechanism can conditions at which the flow visualization pictures in Fig. 8 are
be seen in Fig. 12, which shows contours of the acoustic pressure taken. These flow visualization pictures, together with Fig. 13,
measured on the floor of a ducted shallow cavity as a function of delineate nicely the main characteristics of the excitation mecha-
flow velocity [67]. Within the tested velocity range, the first and nism. The resonance range of each acoustic mode, also called the
second hydrodynamic modes excite the lowest four acoustic dia- lock-on range, normally displays an increase and then a decrease
metral modes, whose frequencies are determined numerically by in the pulsation amplitude before switching to a new lock-on
means of finite element simulation. The destructive nature of this range of another acoustic mode, as illustrated in Figs. 11 and 12.
resonance mechanism is apparent from the acoustic pressure As indicated earlier, the flow-sound interaction mechanism of
recorded on the cavity floor, which exceeds 170 dB at some test the T-junction with combining flows (Figs. 4 and 10) is more
conditions. complex and needs further study; however, the system behavior is
Both Figs. 11 and 12 indicate that the shear-layer excitation is remarkably similar to the cases discussed above. This is truly the
most effective when it is operating at the first hydrodynamic mode case when the length scale of the excitation mechanism is prop-
(m 1), which corresponds to the formation of one vortex by the erly identified. Figure 14 shows typical results of the pulsation
shear layer along the impingement length. Indeed, most of the amplitude as a function of the flow velocity for six different
severe pulsations observed in field experiences correspond to T-junction geometries to illustrate the effect of the length of the
m 1. When the flow velocity is increased gradually from a low transition piece (LT). When this length is used to normalize the
value, a particular resonance mode is generally excited by a flow velocity (Vr VB/f1 LT), the lock-on ranges for all T-junction
higher-order hydrodynamic mode (m > 1) before it is excited by geometries are aligned together over the same ranges of reduced
the first hydrodynamic mode (m 1). Figures 11 and 12 clearly velocity. Each resonance range is associated with, or excited by, a
depict this feature: each specific acoustic mode (n 1, 2, 3.) is shear-layer hydrodynamic mode. For example, the shear-layer os-
first excited by the second hydrodynamic mode (m 2) and then cillation patterns associated with the resonance ranges centered
by the first hydrodynamic mode (m 1). around Vr 1 and 2 are depicted in Fig. 10. Note that it is not
These general features of the acoustic resonance mechanism, clear whether another resonance range would have occurred at
which are illustrated in Fig. 11 for the case of closed side- higher reduced velocities (Vr > 3), had it been possible to increase
branches and in Fig. 12 for the trapped cross modes of a ducted the flow velocity beyond the maximum blower capacity during
shallow cavity, are also observed for other flow configurations, the tests. Another resonance range at a higher reduced velocity
such as the excitation of longitudinal acoustic resonances in would generate fewer, but larger, vortices in the T-junction, and

Applied Mechanics Reviews JANUARY 2014, Vol. 66 / 010802-7

Downloaded From: http://appliedmechanicsreviews.asmedigitalcollection.asme.org/ on 07/16/2015 Terms of Use: http://asme.org/terms

Fig. 15 Normalized acoustic pressure (P=1=2 qV 2 ) as a function
of Strouhal number (fnL/V) for a ducted axisymmetric cavity. L/
h 5 1, h/D 5 1/6. All modes are trapped diametral modes.
Reprinted from Ref. [67] with permission from Elsevier
Fig. 14 Normalized acoustic pressure P=1=2 qV 2B of the Publishing.
primary acoustic mode f1 as a function of reduced velocity (VB/
f1LT) based on half the transition zone length LT. Reprinted from
Ref. [72] with permission from ASME. A convenient way to present acoustic resonance data is to plot
the dimensionless amplitude against the reduced velocity, as al-
ready discussed in Fig. 14, or against the Strouhal number, as
shown in Fig. 15, which depicts the same resonance data of
the pulsation amplitude would likely be stronger than those illus- Fig. 12 for the ducted axisymmetric shallow cavity. Note that the
trated in Fig. 14. This open issue needs further research, because Strouhal number is simply the inverse of the reduced velocity, and
industrial pipelines can be relatively long with low-resonance fre- therefore, increasing the flow velocity is equivalent to moving
quencies, which can result in relatively high reduced velocities at from right to left along the Strouhal number axis.
industry-relevant flow velocities. Figure 15 illustrates that the first four modes are excited over
The alignment of the reduced velocity resonance ranges indi- two well-defined ranges of Strouhal numbers: one corresponds to
cates not only that the controlling geometrical parameter of the the first shear-layer mode (St 0.370.62) and the other associ-
excitation mechanism is properly identified, but also, and more ated with the second shear-layer mode (St 0.851.18). There is
importantly, that the resonance ranges of reduced velocity are also an optimal Strouhal number for each range at which the max-
clearly defined for all T-junction geometries. It is therefore possi- imum resonance intensity occurs; approximately 0.5 and 1.0 for
ble to use a suitable combination of VB and LT to avoid the reso- the first and second shear-layer modes, respectively. This optimal
nance ranges of Vr during steady-state operation of the utility. For Strouhal number and the associated resonance amplitude depend
example, the use of a short T-piece (LT/D 1) would avoid reso- on the local and global system parameters, such as local geometry,
nance altogether up to a reduced velocity of at least Vr 3, but acoustic attenuation, and boundary conditions at the system termi-
this geometry is expected to be associated with a higher pressure nations. In Sec. 4, a methodology to determine the pulsation am-
drop than that resulting from a longer transition piece. A length of plitude at any Strouhal number within the resonance lock-on
LT/D 1.5 would also avoid the resonance up to a reduced veloc- range will be discussed in some detail. However, it is important to
ity Vr  1.5. note here that the Strouhal numbers at the beginning of each reso-
In principle, each acoustic mode of the system (n) can be nance range, e.g., 0.62 and 1.18 in Fig. 15, are of practical impor-
excited by each hydrodynamic mode (m). However, as can be tance, because they define the critical flow velocities at which
observed from Fig. 11, the resonance intensity becomes stronger acoustic resonances may be initiated. Since most of the acoustic
as the higher-order acoustic modes are excited. This is due to the resonance problems experienced in industry were excited by the
fact that higher acoustic modes are excited at higher flow veloc- first shear layer mode, the Strouhal number at the onset of reso-
ities, and therefore, more flow power is available to drive the nance by the first shear layer mode is the most relevant Strouhal
acoustic resonance. In order to take this effect into consideration, number for industrial applications.
the mean flow velocity is used to normalize the pulsation ampli-
tude. Two approaches are used in the literature: the first uses the
dynamic head to obtain a normalized pulsation amplitude P* 4 Design Guidelines
given by A proper aeroacoustic design of a piping system avoids the
P operational conditions, which may lead to the occurrence of self-
P (4) excited acoustic resonances. This can be achieved by keeping the
1=2 qV 2 maximum flow velocity in the piping system lower than critical
flow velocity at which an acoustic resonance may be initiated. For
where P is the acoustic pressure. The second approach uses the this purpose, the first part of this section focuses on design charts
characteristic impedance (qc) multiplied by the flow velocity to of critical Strouhal numbers (So), which can be used to determine
obtain a reduced pressure amplitude Pr as the critical flow velocity (Vo), above which self-excited acoustic
resonances can be initiated. Whenever the flow velocity cannot be
P kept below the critical velocity due to design constraints or
Pr (5)
qcV because of extending the operational conditions of existing plants
to higher flow velocities, the maximum pulsation amplitudes for
The reduced amplitude Pr represents the ratio of the maximum the projected velocity range should be estimated. The second part
acoustic particle velocity of the resonant acoustic mode to the of this section provides a systematic approach to predict the pulsa-
mean flow velocity (Pr u/V). tion amplitude as a function of flow velocity. In order to achieve

010802-8 / Vol. 66, JANUARY 2014 Transactions of the ASME

Downloaded From: http://appliedmechanicsreviews.asmedigitalcollection.asme.org/ on 07/16/2015 Terms of Use: http://asme.org/terms

Fig. 16 Acoustic response of coaxial side-branches showing
the effect of the branch length LB (i.e., the effect of acoustic
attenuation). D, LB 5 61 cm, test pressure 3.5 bar; ,
LB 5 110 cm, test pressure 4 bar; N, LB 5 158.5 cm, test pressur-
e 5 4 bar. Adopted from Ref. [35] with permission from Elsevier

this, the aeroacoustic sources associated with several industry-

relevant flow situations are given and their usage to determine the
pulsation amplitude is demonstrated by examples. If the estimated
amplitude is not acceptable, remedial measures should be consid-
ered and effective ones implemented to mitigate the resonance in-
tensity or eliminate it altogether. These remedial measures are Fig. 17 Design chart of critical Strouhal number (So 5 f d/Vo) at
discussed in the final part of this section. the onset of resonance for closed side-branches in various
arrangements and diameter ratios. Reprinted from Ref. [35] with
permission from Elsevier Publishing.
4.1 Onset of Resonance. Ziada and Shine [35] performed a
series of experiments on single, tandem, and coaxial arrangements
of closed side-branches (Fig. 3) to investigate the effects of the
approach flow conditions and geometry of pipe arrangement on local geometric parameters. These parameters include, for exam-
the critical Strouhal number at which acoustic resonances are ini- ple, the diameter ratio (d/D) and the distance of the branch from
tiated. The critical Strouhal number (So) is defined as the nearest upstream elbow, which affects the shape of the veloc-
ity profile approaching the branch. The test results in Ref. [35] are
So fr d=Vo (6) used to develop the Strouhal number chart shown in Fig. 17. This
chart can be used to predict the critical flow velocity (Vo) in the
where d is the side-branch diameter and Vo is the critical velocity main pipe, at which acoustic resonances may be initiated. The
at which the resonance mode of frequency fr is initiated. data corresponding to X/D > 40 were obtained from branches with
The side-branch arrangements tested in Ref. [35] experience very large distance to upstream elbow.
substantially different radiation losses, ranging from full radiation Several remarks should be made here regarding the use of the
into the main pipe as in the single branch case to negligible radia- Strouhal number chart given in Fig. 17. First, when single or tan-
tion losses as in the case of coaxial branches. The effects of vis- dem branches are installed at the inner side of an upstream elbow,
cous and heat conduction losses were also investigated by altering the critical Strouhal number is lower than that for X/D > 40 [35].
the test static pressure and the length of the side-branches. The In such cases, the use of a value corresponding to a large X/D, i.e.,
test results showed that, although radiation, viscous, and heat con- no upstream elbow or X/D > 40, is recommended because it
duction losses do strongly influence the maximum pulsation am- results in conservative designs. Secondly, since the effect of
plitude at resonance and the extent of the lock-in range, these upstream elbows on the critical Strouhal number of coaxial
losses were found to have negligible effect on the critical Strouhal branches is found to be negligible [35], a value corresponding to
number at the onset of resonance. Typical results are reproduced X/D > 40 should be used for coaxial branches, regardless of their
in Fig. 16 for the coaxial case. The maximum pulsation amplitude, position with respect to upstream elbows. Thirdly, it should be
the optimal Strouhal number at which it occurs, and the extent of noted that Fig. 17 is developed from the test results of one elbow
the lock-in range are seen to be dependent on the branch length. radius. However, since this radius is relatively small (3D), the
However, the resonance in all three cases is initiated at the same present results would be conservative for many practical applica-
critical Strouhal number (So). The finding that So is independent tions that involve elbows with radius larger than 3D. If the radius
of system losses, which has been illustrated for the case of coaxial is smaller than 3D, the designer should consider an additional
closed side-branches, can be generalized for other flow cases margin of safety. This design chart can be used also for side-
experiencing acoustic resonance. This is because the system branches with square or rectangular cross sections. However, the
response at the onset of resonance is very small and is therefore diameter d used in the Strouhal number must be replaced by the
independent of the oscillation amplitude. In the following, design equivalent diameter proposed by Bruggeman [74],
charts for So are given for the flow-sound interaction cases
discussed in Sec. 3. de 4=pH (7)

4.1.1 Strouhal Number for Closed Side-Branches. Although where H is the width of the branch mouth in the flow direction.
the critical Strouhal number So is independent of system losses, When the corners of the branch mouth are rounded with a radius
additional experiments showed that it is strongly influenced by the of curvature r, the critical flow velocity scales with (d r) instead

Applied Mechanics Reviews JANUARY 2014, Vol. 66 / 010802-9

Downloaded From: http://appliedmechanicsreviews.asmedigitalcollection.asme.org/ on 07/16/2015 Terms of Use: http://asme.org/terms

Fig. 19 Design chart of critical Strouhal number (So 5 f L/Vo) at
the onset of resonance of the trapped cross (or diametral)
modes for an axisymmetric shallow cavity in a duct as illus-
Fig. 18 Design chart for critical Strouhal number (So 5 fnL/Vo) trated in Fig. 6. Data extracted from Ref. [67].
at the onset of resonance of the longitudinal resonance of pipe-
lines with axisymmetric shallow cavity as illustrated in Fig. 5 velocity in different diameter pipes. Additional information for
the case of d/D 1 can be found in Ref. [76].
of d [74]. The Strouhal number design chart can still be used to
estimate Vo by means of the formula, 4.1.4 Strouhal Number for the Excitation of Trapped Cross
Modes of Ducted Cavities. The critical Strouhal numbers for the
Vo f d r =So (8) excitation of trapped cross modes by the first and second shear-
layer modes of an axisymmetric cavity are plotted in Fig. 19 as
Finally, the value corresponding to d/D 0.57 can be used for functions of the cavity length-to-depth ratio. The critical Strouhal
larger diameter ratios. This suggestion is based on the fact that the number increases first in the transition region from deep to shal-
Strouhal numbers reported by Peters [64] for d/D 1 are similar low cavities, i.e., for L/h increasing from 0.5 to 1.0, but then it
to the results for d/D 0.57. starts to decrease gradually with the ratio L/h. In fact, the results
of two-dimensional cavities housed in rectangular ducts [66] agree
4.1.2 Strouhal Number for Shallow Cavity Coupling With very well with the data in Fig. 19. For example, the critical
Pipeline Resonance. When an axisymmetric (or a two-dimen- Strouhal numbers for m 1 and 2 at L/h 2.5 are 0.54 and 1.12
sional) cavity is housed in a pipeline or a duct, the longitudinal (or and at L/h 4 are 0.45 and 1.09, respectively. Thus, the chart of
axial) acoustic modes may be excited, as illustrated in Fig. 5. The Fig. 19 seems to be also applicable to the excitation of trapped
acoustic modes can be excited by any shear-layer hydrodynamic cross modes in rectangular ducts housing two-dimensional
mode, but the strongest excitation occurs at the first hydrodynamic cavities.
mode. Figure 18 presents the critical Strouhal number for the first
two shear-layer modes (m 1 and 2) for an axisymmetric cavity
initiating acoustic resonance of the axial modes of the pipe hous- 4.2 Estimation of Pulsation Amplitude at Resonance. The
ing the cavity [75]. These Strouhal numbers are determined from critical Strouhal number charts presented in Sec. 4.1 can be used
small amplitude oscillation at the initial phase of the lock-on to estimate the critical flow velocity Vo for the onset of any poten-
range. As in the previous case, the Strouhal number is based on tial acoustic resonance. If the maximum flow velocity of the in-
the length of the cavity, which must include the radius of curva- stallation Vmax is found to be higher than Vo, design modifications
ture or the length of the chamfer at the upstream corner, if applica- should be adopted to increase Vo above Vmax. This can be
ble. The geometry of the downstream corner has only minor achieved by enlarging the length scale (Lc) of the shear-layer os-
effects on the Strouhal number and the pulsation amplitude, and cillation to reduce the excitation frequency (e.g., by increasing the
therefore, it is not an important design parameter for the fluid- shallow cavity length or the diameter of the closed side-branch) or
resonant mechanism. Figure 18 shows considerable increase by increasing the frequency of the potential resonance mode (e.g.,
in the critical Strouhal number as the cavity length to depth ratio by shortening the closed side-branch length). If these modifica-
(L/h) is increased. However, it does not include the effect of the tions are not feasible or insufficient, the pulsation amplitude at
ratio between cavity depth and main pipe diameter (h/D). The resonance should be estimated as outlined in this section.
effect of this parameter is currently being investigated by the The general approach for estimating the pulsation amplitude at
authors. a certain Strouhal number, i.e., for a given frequency and flow ve-
locity, is to estimate the aerodynamic source term as a function of
4.1.3 Strouhal Number for T-Junctions With Combining pulsation amplitude and apply the energy balance concept
Flows. The acoustic response of a T-junction that combines two between the power generated by the source and that dissipated by
flows has already been addressed in conjunction with Fig. 14. The the system via viscothermal losses and radiation at the termina-
critical reduced velocity for various shear-layer modes can be tions. These two amounts of energy must be balanced at the am-
obtained from this figure. For example, mode A is initiated at a plitude of steady-state oscillation. Therefore, the aeroacoustic
reduced velocity Vr  1.5; mode B at Vr  0.75; and mode C at Vr source for the relevant fluid-sound interaction pattern must be
 0.45. As mentioned earlier, the excitation intensity of each known in advance as a function of the Strouhal number and the
shear-layer mode depends on the length of the T-piece (LT). It pulsation amplitude. This aeroacoustic source must also corre-
should be noted that the results of Fig. 14 correspond to only one spond to plant flow conditions, which often involve three-
diameter ratio of d/D 0.75. Although systems with other diame- dimensional, turbulent, and fully developed pipe flow. At present,
ter ratios may behave slightly different, especially regarding the reliable industrial data on aeroacoustic sources are available only
resonance intensity, the tested ratio is the most industry-relevant for the closed side-branch case shown in Fig. 3 [29,41] and the
one, because the T-junction maintains nearly constant flow axisymmetric cavity case illustrated in Fig. 5 [5557]. These

010802-10 / Vol. 66, JANUARY 2014 Transactions of the ASME

Downloaded From: http://appliedmechanicsreviews.asmedigitalcollection.asme.org/ on 07/16/2015 Terms of Use: http://asme.org/terms

Fig. 20 Model of the aeroacoustic source resulting from the
interaction of a shear layer at the mouth of a deep cavity with
the particle velocity u associated with an acoustic resonance in
the cavity. Reprinted from Ref. [39] with permission from ASME.

sources are described in this section, and their use to estimate the
pulsation amplitude is demonstrated by means of examples.
Concerning the other two cases, i.e., the T-junction geometry Fig. 21 Experimentally determined aeroacoustic source for a
with combining flows (Fig. 4) and the axisymmetric cavity caus- shear layer at the opening of coaxial side-branches as illus-
ing resonance of the trapped cross modes (Fig. 6), further research trated in Fig. 3(c). The aeroacoustic source term Q is presented
is needed in order to be able to estimate the pulsation amplitude at in the Q-complex plane with the acoustic particle velocity (u/V)
resonance. Therefore, for these two cases, it is recommended to and the Strouhal number (St) taken as parameters. Reprinted
avoid the resonance conditions altogether by designing the system from Ref. [39] with permission from ASME.
to meet the condition that Vmax < Vo, and if this condition cannot
be met, one of the remedial methods discussed later in this paper
should be implemented to alleviate the resonance intensity. acoustic oscillation, are both required in order to be able to predict
the resonant velocity range as well as the pulsation amplitude.
Figure 21 shows the measured source term in the complex
4.2.1 Aeroacoustic Source of the Shear Layer at the Mouth of Q-plane. The Strouhal number, S, and the amplitude of the acous-
a Resonant Side-Branch. When the side-branch is at resonance, tic particle velocity, u/V, are taken as parameters. It is noteworthy
the excitation mechanism becomes highly nonlinear. In order to that this source term was determined from experiments involving
estimate the amplitude of pulsation, the shear-layer instability and highly turbulent flow in cylindrical pipes, and therefore, it can be
its interaction with the resonant sound field must be described by used for industrial piping systems. Since the interaction mecha-
means of the nonlinear theory. This difficult approach becomes nism between the shear layer and the sound field is strongly non-
even more complex and cost-prohibitive for industrial applica- linear, the source term is a function of both the Strouhal number
tions involving high Reynolds number pipe flow, which is inher- and the pulsation amplitude (u/V). The Strouhal number has pri-
ently turbulent and three-dimensional. marily an effect on the argument (i.e., phase) of Q, while the am-
In order to develop a relatively simple but still reliable model plitude influences mainly the modulus (absolute value). The lines
that can be used in industrial applications to predict not only the with constant amplitude form a spiral around the origin of the
onset of resonance but also the pulsation amplitude as a function complex Q-plane. For acoustic resonances to be self-excited, the
of flow and geometric parameters of the associated piping system, real part of Q must be positive to generate positive acoustic
Graf and Ziada [41] proposed to model the integrated effect of the energy. On the other hand, if the real part of Q is negative, acous-
shear-layer excitation by a pressure difference across the shear tic energy is absorbed by the shear-layer oscillation, and acoustic
layer, as illustrated in Fig. 20. This pressure difference Dp is resonances cannot be self-sustained under these conditions.
induced by the generation and convection of vortex-like structures Several authors [11,37,77] reported strong self-excited reso-
in the shear layer and is assumed to be a function of the local flow nance in the single-vortex mode for Strouhal numbers around 0.42
properties, such as the Strouhal number and the acoustic particle and a weaker resonance at the double-vortex mode for Strouhal
velocity. The acoustic power generated by the unsteady flow at numbers near 0.85. Figure 21 shows that, for these Strouhal num-
the side-branch opening can therefore be obtained by multiplying bers, the real part of Q is near its maximum and its imaginary part
the shear-layer pressure difference, Dp, times the local acoustic is close to zero.
particle velocity, u. For steady-state oscillation, the power gener- For Strouhal numbers slightly less than the value corresponding
ated by the unsteady shear layer must be balanced by the dissipa- to the maximum excitation level (St  0.42), the imaginary part of
tion of acoustic energy in the pipes and by radiation losses. The Q is positive. Therefore, the shear-layer pressure difference acts
measured source term, Q, is developed from the ratio between the like an additional stiffness, which increases the frequency of self-
induced acoustic pressure difference across the shear layer, Dp, excited oscillation. This feature is also observed in the experi-
and the acoustic particle velocity at the mouth of the branch, u. ments. On the other hand, for Strouhal numbers slightly greater
This ratio is normalized by the dynamic head, [1/2] qV [2], and than 0.42, the imaginary component of Q is negative and the
the mean velocity, V, in the main pipe. Thus, the source term Q induced pressure difference acts like an additional mass, thus
has the form of dimensionless acoustic impedance, reducing the frequency. For Strouhal numbers around 0.6, the real
part of Q is strongly negative. This indicates that the unsteady
DP=0:5 qV 2 shear layer extracts energy from the acoustic oscillation. During
Q (9)
u=V the experiments of Graf and Ziada [41], shear-layer oscillations
near this Strouhal number can only be excited with very strong
Graf and Ziada [41] experimentally determined the amplitude forcing by the loudspeaker.
and phase of this complex source term for a wide range of As the oscillation amplitude increases, the modulus of Q is
Strouhal number, St, and sound pressure level; the latter is reduced approximately propositional to 1= u=V . The induced p
expressed by the dimensionless acoustic particle velocity (u/V). pressure difference is therefore roughly proportional to u=V .
The amplitude and phase of this source term, with respect to the This is in contrast to the model suggested by Bruggeman et al. [48],

Applied Mechanics Reviews JANUARY 2014, Vol. 66 / 010802-11

Downloaded From: http://appliedmechanicsreviews.asmedigitalcollection.asme.org/ on 07/16/2015 Terms of Use: http://asme.org/terms

Fig. 22 Oscillation amplitude as function of flow velocity for
different depth modes for coaxial branches. The resonance
frequency is 137 Hz in all four cases. The symbols indicate
measured data, and the lines are computed from the
experimentally determined source term presented in Fig.
21. 1 LB1 5 LB2 5 0.6 m; LB1 5 0.6 m and LB2 5 3LB1;
LB1 5 0.6 m and LB2 5 5LB1; 3 LB1 5 LB2 5 1.8 m. Reprinted from
Ref. [39] with permission from ASME.
Fig. 23 Model of the aeroacoustic source resulting from the
coupling of the shear layer at a cavity opening with the particle
which assumes the driving pressure difference to be independent velocity u associated with a longitudinal resonance mode of a
of the excitation amplitude. pipeline
Although the source term Q presented in Fig. 21 is highly non-
linear, it is relatively simple to implement it in software employ-
ing the linear acoustic theory to simulate acoustic pulsations in
piping systems. Once the acoustic particle velocity is determined
at the branch mouth, a first guess of the shear-layer source term
(Fig. 21) can be inserted in the system, and after a few steps of
iteration, the final source strength and the resulting pulsation am-
plitude can be obtained. As an example, the acoustic pressure pre-
dicted by means of the source term Q for the case of coaxial
branches is compared with the experimental data in Fig. 22. Four
cases are considered, including two symmetric arrangements
(LB1 LB2 0.6 m; LB1 LB2 1.8 m) and two asymmetric
branches (LB1 0.6 m and LB2 3LB1; LB1 0.6 m and
LB2 5LB1). In all four cases, the amplitude of pulsation, the onset
of resonance, and the lock-in range are well predicted. Additional
comparisons for other pipe geometries can be found in Graf and
Ziada [41].

4.2.2 Aeroacoustic Source of a Cavity Shear Layer Coupled

With a Pipeline Resonance. Recent problems of acoustic reso-
nance in offshore gas transport corrugated pipes have promoted Fig. 24 Experimentally determined aeroacoustic source for a
research efforts to investigate the excitation source generated by cavity shear layer coupling with longitudinal acoustic modes of
flow over a shallow cavity in a resonant pipeline [5557]. This a pipeline as illustrated in Fig. 5(a). The aeroacoustic source
work has also been extended to explore the effect of hydrody- term Q is presented in the Q-complex plane with the acoustic
namic coupling when multiple cavities exist, as in the case of cor- particle velocity (u/V) and the Strouhal number (St) taken as pa-
rugated pipes [78]. In this section, we focus on the experimentally rameters. Reprinted from Ref. [57] with permission from ASME.
determined aeroacoustic source for a shear layer spanning a single
cavity housed in a resonant pipeline, as illustrated in Fig. 5(a).
This source was determined under fully developed pipe flow to the positive ranges of real (Q) values. These ranges, and the
conditions, as encountered in industrial application. effect of u/V, are better demonstrated by plotting the real (Q) com-
As shown in Fig. 23, Mohamed et al. [57] employed an ponent, which represents the active part of the source, against the
approach similar to that introduced by Graf and Ziada [41] to Strouhal number, as shown in Fig. 25. As mentioned earlier, the
model the integrated effect of the cavity shear layer by a pressure optimum Strouhal number for the maximum source strength
difference across the cavity. Multimicrophones in both the depends on the pulsation amplitude; it decreases with u/V, and
upstream and downstream pipes are used to measure the ampli- therefore, the Strouhal number at the onset of resonance So is
tude and phase of the pressure difference as a function of Strouhal more useful for industrial applications. The chart given in Fig. 18
number and amplitude of the acoustic particle velocity. The meas- is obtained from similar tests of various cavity sizes ranging from
ured normalized source term defined by Eq. (9) is given in Fig. 24 L/d 1 to 3.5.
for the case of a cavity shear layer coupled with the longitudinal The measured source in this case was also validated as in the
acoustic mode of a pipeline (Fig. 5). For the sake of clarity, only previous case of side-branch resonance. The validation tests
two spirals are shown for u/V 1% and 5%. For the lower ampli- entailed comparing the experimentally measured amplitude of the
tude case, it can be seen that there are two Strouhal number ranges self-excited system with that predicted from using the normalized
over which acoustic resonances can be excited, which correspond source term shown in Fig. 25. The energy balance approach is

010802-12 / Vol. 66, JANUARY 2014 Transactions of the ASME

Downloaded From: http://appliedmechanicsreviews.asmedigitalcollection.asme.org/ on 07/16/2015 Terms of Use: http://asme.org/terms

Fig. 25 Real part of source term Q for a cavity shear layer cou-
pling with longitudinal acoustic modes of a pipeline as illus-
trated in Fig. 5(a). Real (Q) is plotted as function of Strouhal
number with the acoustic particle velocity (u/V) taken as a pa-
rameter. Reprinted from Ref. [57] with permission from ASME.
Fig. 26 Oscillation amplitude as function of flow velocity for
longitudinal acoustic resonance of a pipeline housing a cavity.
used here to predict the pulsation amplitude for each value of flow The symbols indicate measured data, and the lines are com-
velocity. The comparison shown in Fig. 26 illustrates that the ex- puted from the experimentally determined source term pre-
citation amplitude and resonance range are very well predicted sented in Fig. 25.
over both ranges of the self-excited resonance.
particle velocity at the leading edge of the branch is higher for the
downstream pipe, there is more absorption at the mouth of the
5 Remedial Means downstream pipe, which results in lower amplitude of pulsation.
Whenever conditions conducive to acoustic resonance excita- In addition, the maximum pulsation amplitude is produced when
tion cannot be avoided and the estimated pulsation amplitude is the location of the maximum particle velocity is shifted toward
found to be unacceptable, remedial means should be implemented the upstream branch opening. It is for this reason that sharp edges
during the design stage to mitigate the resonance intensity or elim- at the upstream corner of side-branches and cavities reduce the
inate it altogether. In addition to enlarging the branch inlet or intensity of the fluid-resonant mechanism.
shortening the branch length, which have already been mentioned, For the case of short side-branches with small diameter ratios d/D,
several other design changes are possible [12,29], such as detun- which are liable to resonance even in the single configuration,
ing closed side-branches by making them of different lengths detuning the branches may be less effective. As will be shown
[11,39,41], inserting antivortex elements in the branch inlet [79], later in the second example of the industrial cases [36], introduc-
or adding upstream spoilers in the main pipe [11,48,80]. Active ing a relatively small offset in the length of two branches with
control methods have also been shown to be effective, however, small d/D ratio generates two distinct tones, corresponding to the
to date, only under laboratory environment. In the following, these different lengths of the branches. Although the tone amplitude in
countermeasures are briefly discussed. this case was reduced by detuning to about 25% of that observed
for the tuned branches, this reduction may not be sufficient, espe-
cially when the higher-order acoustic modes are of concern for
5.1 Detuning the Resonance Frequencies. If resonance con- short side branches with small-diameter ratios.
ditions cannot be avoided for piping systems with multiple closed In some other pipe configurations, the pulsation amplitude was
side-branches, one of the most effective methods to mitigate the found to be more robust, such that even a change of 30% in length
resonance intensity is to detune the side-branches by making one of one of the side-branches was not sufficient to reduce the pulsa-
branch shorter (or longer) than the other. However, the effect of tions of the system by an order of magnitude [29,81]. The robust-
detuning the branches is strongly dependent on the specific geom- ness of these systems has been found to be due to the coupling of
etry of the piping system, including the diameter ratio d/D, the the branches with the upstream and/or downstream segment of the
distance between the side-branches , the branch arrangement main pipe.
(i.e., whether the pipe system includes double or multiple
branches and whether they are in the tandem or the coaxial config-
uration), and the geometry of the upstream and downstream pipe 5.2 Antivortex Inserts. Jungowski and Studzinski [79]
segments [29]. developed and patented several antivortex devices, which can be
For the double side-branches arranged in the coaxial [39,41] or inserted into the mouth of side-branches. They consist of a single
the tandem [11,12] arrangements, the pulsation amplitude splitter plate, two plates in a cross configuration, or three plates in
decreases by an order of magnitude when the lengths of the a triangular arrangement. These inserts represent an attractive so-
branches are detuned by about 10%, as demonstrated by the lution, because they are very effective in suppressing the reso-
results in Fig. 27. While the results of the coaxial branches shown nance, do not interfere with the flow in the main pipe, and do not
in Fig. 27(a) display symmetric reduction in the pulsation ampli- cause substantial increase in the pressure loss when the flow is
tude when shortening either of the branches, the results of the tan- diverted into the branches. The excellent performance of these
dem branches in Fig. 27(b) depict very interesting features in inserts in suppressing acoustic resonances appears to be due to
addition to the sharp reduction in the amplitude due to branch several effects. They hinder the formation of vortices inside the
detuning. The pulsation amplitude at the closed end of the down- branch mouth, they change the impingement length scale of the
stream branch is consistently lower than that of the upstream separated shear layer, and they introduce strong three-dimensional
branch. In addition, the maximum pulsation amplitude does not effects into the free shear-layer oscillation and its impingement
occur for the symmetric pipe configuration but rather for the lon- length.
ger upstream branch, which entails that the pressure node (or the
maximum particle velocity) is closer to the upstream branch open- 5.3 Spoilers and Upstream Edge Geometry. Spoilers can be
ing. These features clearly emphasize the fact that, since the used to reduce self-sustained oscillations due to flow separation in

Applied Mechanics Reviews JANUARY 2014, Vol. 66 / 010802-13

Downloaded From: http://appliedmechanicsreviews.asmedigitalcollection.asme.org/ on 07/16/2015 Terms of Use: http://asme.org/terms

Fig. 27 Pulsation amplitudes P1 and P2 at the closed ends of asymmetrical side-
branches of length L1 and L2. (a) Coaxial side-branches. (b) Two tandem side-
branches in close proximity with =L 0:15. The length of one side-branch was
increased/decreased in steps while the other branch was shortened/elongated
accordingly by an equal length to keep the resonance frequency constant. D 5 89
mm, d 5 51 mm, (L1 1 L2) 5 1.57 m. Adopted from Refs. [12] and [41] with

piping systems. Many types of spoilers have been proposed in the cross modes of ducted cavities [7,82,83]. Smith and Luloff [7]
literature to weaken flow-sound interaction in T-junctions and tested a large number of edge geometries to suppress the reso-
ducted cavities. The effects of some of these spoilers are discussed nance of trapped modes associated with cavities in gate valves.
in this section. Using a chamfer at the upstream edge produced inconsistent
The sharp-edged plate positioned at the upstream corner of the results, suppressing the resonance in some cases and not in others.
side-branch opening suggested by Bruggeman et al. [48] is very More recently, Bolduc et al. [82] investigated in some depth the
effective but increases the losses when the flow is turning into the effect of the upstream edge geometry on the excitation of the
side-branch. The effect of making the upstream edge sharp has trapped modes of axisymmetric-ducted cavities. The investigated
been addressed briefly in connection with Fig. 27(b). In principle, geometries included rounded cavity corners, chamfering the
the effectiveness of the sharp edge proposed in Ref. [48] depends upstream edge, and various types of spoilers. In the following,
on the length of the plate forming the edge compared to the side- only selected results are presented, and the reader is referred to
branch diameter. In the experiments carried out by Bruggeman Ref. [82] for more details on the tested geometries and their effect
et al. [48], a ratio of 0.2 appeared to be very effective. In the cases on the acoustic resonance mechanism.
when a minimum flow area is needed at the branch inlet, such as First, rounding off the cavity edges is seen in Fig. 28(a) to
the standpipes of safety relief valves, the side-branch inlet should increase the pulsation amplitude substantially, but the resonance
be enlarged to compensate for the inlet area blocked by the plate lock-on range is delayed, i.e., shifted to higher flow velocities.
forming the edge. This delay is due to the increase in the length scale (or the
The effect of an upstream orifice plate on the acoustic reso- impingement length), which is equivalent to the cavity length plus
nance of side-branches in tandem and coaxial configurations was the upstream radius. Thus, as in the case of acoustic resonances of
investigated by Ziada and Buhlmann [11]. When an orifice plate closed side-branches and pipelines housing cavities, rounding off
was positioned 5.5 main pipe diameters upstream of the branches, the upstream cavity corner strengthens the acoustic resonance.
the pulsation amplitude was reduced to about 30% of its original Similarly, chamfering the upstream corner delays the onset of
value without the orifice plate for both the tandem and the cross resonance, because the impinging length scale is increased by the
arrangements. In addition, the resonance of the second acoustic length of the chamfer. This delay is clearly depicted in Fig. 28(b),
mode was eliminated for the tandem branches. Ziada and which compares the pulsation amplitude for a cavity with and
Buhlmann [11] attribute this mitigation effect to increased turbu- without a chamfer. However, once the resonance is excited at
lence level and scale in the main pipe, which disturbs the forma- higher flow velocities, Fig. 28(b), the pulsation amplitude is com-
tion of coherent vortices at the branch mouth. The drawback of parable with that observed without the chamfer. These results
this countermeasure is the increased pressure loss due to the explain previous observations that adding a chamfer at the
restriction of the flow area. upstream corner is not always a reliable means in suppressing the
For the case of cavity shear layer coupling with the longitudinal resonance of gate valves [7]. For example, for the specific
acoustic modes of a pipeline (Fig. 5), Karadogan and Rockwell case shown in Fig. 28(b), if the maximum flow velocity is about
[80] tested several geometries of spoilers (also called vortex 73 m/s, the acoustic resonance would be suppressed; however, it
generators) positioned at the cavity upstream edge, including would reappear if the flow velocity is increased beyond this value.
d-shaped spoilers as well as slotted, wavy, and compliant boun- Tests with other cavity sizes and different chamfers showed that
daries. The results show that the d-shaped spoilers are the most chamfering the edge at 17 deg for a length of 20% of the cavity
effective device in suppressing the resonance, and the degree of length can effectively suppress the resonance for shallow and/or
suppression is dependent on the height, length, and spacing of the small cavities, but it is not effective for deeper or larger cavities.
spoilers. The pressure pulsations were totally suppressed when the This is because the frequencies of (nearly) trapped modes of small
height of the spoilers was increased to about two momentum and shallow cavities are close to the cut-off frequencies of the
thicknesses of the approaching boundary layer. associated duct, whereas deep and large cavities have substan-
The effect of the geometry of the cavity upstream corner has tially lower frequencies than the duct cut-off frequencies. Since
been investigated also for the case of trapped (or nearly trapped) the transmission losses are inversely proportional to this

010802-14 / Vol. 66, JANUARY 2014 Transactions of the ASME

Downloaded From: http://appliedmechanicsreviews.asmedigitalcollection.asme.org/ on 07/16/2015 Terms of Use: http://asme.org/terms

Fig. 30 Effect of the curved tooth and delta spoilers on the
pressure drop over the cavity. Cavity parameters as shown in
Fig. 6 are L/h 5 1, h/D 5 1/6, D 5 152 mm. Reprinted from Ref.
[82] with permission from ASME.

these spoilers are shown in Fig. 29. The first photo shows
especially designed spoilers with curved surface of the tooth to
deflect the separated flow at larger angle relative to the axial flow
direction, whereas the other photo shows delta spoilers, which are
intended to introduce three-dimensional vorticity field at the
upstream corner. These curved tooth and delta spoilers were found
to be the most effective in suppressing very robust acoustic
resonances. Typical results are presented in Fig. 28(c), which
shows that both spoiler geometries produced negligibly small am-
plitude at the resonance frequencies over the whole range of flow
velocity. However, this superior performance in suppressing
strong acoustic resonances is associated with increased static pres-
sure drop, as illustrated in Fig. 30, which compares the measured
pressure drop over the cavity with spoilers against that measured
for the same cavity without spoilers. This increase in pressure
Fig. 28 Effect of upstream edge geometry on the excitation of drop can be a major disadvantage for the use of such spoilers in
trapped cross modes of the axisymmetric ducted cavity. (a)
Effect of rounding off the upstream edge with a radius r 5 0.2 L;
some industrial applications, such as gate valves, which are
(b) effect of a 17 deg chamfer of length 0.38 L; (c) effect of designed to produce minimum pressure drop when they are fully
curved tooth and delta spoilers. Cavity parameters as shown in open.
Fig. 6 are L/h 5 1, h/D 5 1/6, D 5 152 mm.
5.4 Active Control Techniques. Active control of flow-
excited acoustic resonances has received considerable attention
over the past two decades, however, as mentioned earlier, only
under laboratory environment. Ffowcs Williams and Huang [84],
Huang and Weaver [71], and Welsh et al. [85] used loudspeakers
to counteract the resonant sound field of different types of resona-
tors and thereby suppress the resonant oscillations. Later on,
active suppression of shallow-cavity acoustic resonance has been
demonstrated, with a varying degree of success, by means of per-
turbing the shear layer at its separation location with the aid of
oscillating flaps [19], pulsed mass injection [86], piezoelectric
actuators [87], or synthetic jets [88,89]. It is also possible to use
active control means to suppress acoustic resonances of closed
side-branches [12,90].
There are three different methods proposed in the literature to
suppress flow-excited acoustic resonances by active means. The
Fig. 29 Photographs of the curved tooth and the delta spoilers first involves externally forcing the free shear layer at frequencies
attached to the upstream cavity corner. The arrows indicate the that are substantially different from that occurring during the reso-
flow direction. Reprinted from Ref. [82] with permission from nance [19,86]. In this approach, a continuous high level of power
ASME. is needed to force the shear-layer oscillation at a frequency differ-
ent from the frequency of its natural instability. The other two
active control methods are similar in that they employ a feedback
difference in frequencies, the resonance of shallow cavities is rela- control strategy, but differ in the type of the used actuator. In the
tively weak and is more likely to be alleviated by appropriately first feedback control method [12,71,84,85], loudspeakers are
designed chamfers [67,82]. used to counteract the acoustic resonance and thereby reduce the
Bolduc et al. [82] investigated also the effect of various spoilers acoustic particle velocity below the critical level required to
on the robust acoustic resonance of large cavities (e.g., cavity organize and synchronize the shear-layer oscillation. This
length and depth are 1/6 of the duct diameter). Two examples of approach can be classified as active damping control of the

Applied Mechanics Reviews JANUARY 2014, Vol. 66 / 010802-15

Downloaded From: http://appliedmechanicsreviews.asmedigitalcollection.asme.org/ on 07/16/2015 Terms of Use: http://asme.org/terms

Fig. 32 Steam dryer assembly of Quad Cities unit 2 (top left)
and details of the acoustic fatigue failure on the outer hood.
Reprinted from Ref. [95] with permission from Exelon Nuclear
and GEH.

Fig. 31 Schematic of the upper section of a boiling water reac-

tor (BWR) showing the steam separator, steam dryer, inlet noz-
zle to a main steam line (MSL), and the path of steam flow
through the dryer and into the MSL. Reprinted from Ref. [92]
with permission from the US NRC and ASME.

acoustic mode, because the actuator acts on the resonant acoustic

mode rather than on the shear layer. In the other feedback control
method [66,8890], the shear layer is directly excited at the sepa-
ration location by means of synthetic jets to counteract the
upstream feedback generated by the resonant acoustic mode. This
technique therefore can be described as feedback control of the
shear-layer oscillation. Since the actuators in both feedback con-
trol methods are activated by the system response, the energy con- Fig. 33 Power spectral density of fluctuating pressure on the
sumption by the actuators decreases sharply after a short time dryer showing the tonal excitation near 150 Hz. Reprinted from
period, which is needed to suppress the resonance. This is in Ref. [96] with permission from Exelon Nuclear.
contrast with the external forcing of the shear layer, which
necessitates continuous high power level for the actuator.
exhibited new cracks upon continued operation [94]. As shown in
6 Examples of Acoustic Resonance From Industry Fig. 32, the cracks appeared on the outer hood plates, which face
the inlet nozzles of the MSLs [95]. Similar cracks were also
6.1 Acoustic Fatigue of a BWR Steam Dryer. This section observed on the steam dryer of unit 1. The steam dryers of both
deals with the acoustic fatigue failure of the steam dryer of a boil- units were therefore replaced with new ones, which are substan-
ing water reactor (BWR) in Quad Cities (QC) nuclear power plant tially more robust. In addition, the new dryer of unit 2 was instru-
[9199]. First, the steam-flow path through the dryer and into the mented with pressure transducers to provide direct measurements
main steam lines (MSLs) is explained. Thereafter, the root cause of the pressure fluctuations at numerous locations on the dryer.
of the problem is identified followed by a discussion of the During this course of events, the safety relief valves (SRVs) on
countermeasure adopted to solve the problem. the MSLs of both units were experiencing high vibration levels,
A simplified representation of the internal components within and subsequent inspection for maintenance during an outage
the upper portion of a typical BWR is given in Fig. 31 [92]. Wet showed that some valves had been damaged.
steam rises from the boiling water and flows upwards through the Figure 33 shows a typical power spectral density of a pressure
water separator and the steam dryer. The steam then collects in transducer on the dryer [96]. Pressure spectra obtained at other
the reactor dome above the dryer before turning downwards to locations on the dryer showed similar characteristics. The pressure
enter the inlet nozzles of four main steam lines. fluctuations are seen to be dominated by several narrow band
The steam dryer in QC unit 2 experienced high-cycle fatigue peaks in the range of 135160 Hz. The frequency of these tones
cracks after the maximum power of the reactor was increased by did not change appreciably, but their amplitudes were strongly de-
approximately 17%. Repairing the dryer by using thicker plates pendent on the steam flow rate, with the higher-frequency tones
and better quality welds did not resolve the problem, as the dryer becoming more dominant as the steam-flow rate was increased.

010802-16 / Vol. 66, JANUARY 2014 Transactions of the ASME

Downloaded From: http://appliedmechanicsreviews.asmedigitalcollection.asme.org/ on 07/16/2015 Terms of Use: http://asme.org/terms

ceptable, based on cost and prohibitively high dosage of radiation
exposure to plant workers. An alternative solution, which was pro-
posed, developed, and tested by Continuum Dynamics Inc. [99],
entailed the addition of an acoustic side-branch (ASB) to the
standpipe, as illustrated in the right-hand-side inset of Fig. 35.
The ASB was filled with sound dissipation material to reduce the
intensity of acoustic resonances in the standpipe-ASB
In contrast with the effect of enlarging the inlet or shortening
the length of the standpipe, the addition of the ASB, in fact,
reduces the standpipe resonance frequency, and therefore, the res-
onance of this particular standpipe is excited at a lower steam-
flow rate, about 80% of the original maximum power. However,
the intensity of this resonance, which occurs at part load of the
plant, was sufficiently diminished because: (a) the dynamic head
of the steam flow, 1/2 qV [2], is lower at part load, and therefore,
the excitation level is also lower than that at the extended maxi-
mum power level, and (b) the absorption material added in the
ASB provides sufficient acoustic attenuation. It should be noted
also that, at the extended maximum power level, which is the nor-
Fig. 34 Locations of the safety relief valves on the main steam mal operation mode of the plant, the resonance frequency of the
lines. Reprinted from Ref. [13] with permission from ASME. standpipe with ASB is sufficiently lower than the shear-layer exci-
tation frequency such that the standpipe resonance was avoided.
Analysis of the steam piping system, which is shown schemati- The effect of the acoustic side-branch on the acoustic resonance
cally in Fig. 34, indicated that the frequencies of these tones corre- was first tested by means of scale-model tests at all flow rates.
spond to the one-quarter-wavelength resonance modes of the Thereafter, the acoustic-resonance problem was satisfactory
standpipes leading to the SRVs (see the inset in Fig. 34). Three resolved when the acoustic side-branches were implemented in
different types of SRVs were used and the standpipe of each valve the plant.
type had a different length, and therefore, several tones were gen-
erated corresponding to the resonance frequencies of different 6.2 Annular Duct With Coaxial Closed Side-Branches.
standpipes. The Strouhal number associated with the appearance Investigation of this pipe configuration was inspired by the roll-
of each tone was found to agree well with the design chart in post design of the vertical lift system of the Joint Strike Fighter
Fig. 17. (JSFV), which is schematically illustrated in Fig. 36. During the
Although the new steam dryer was made sufficiently robust to normal (forward) flight mode, the roll posts are closed at their out-
withstand the dynamic loading generated by the standpipe tonal lets and are exposed to low Mach number flow at their inlets,
excitations, the SRVs with the shortest standpipe were experienc- which are connected to the bypass annular duct of the engine.
ing intense acoustic resonance at the extended maximum power Under these flight conditions, the roll posts approximate closed
level, and therefore, these valves were expected to be damaged coaxial side-branches exposed to a grazing annular flow. Model
again upon continued operation at the extended power level. This tests were performed to investigate the aeroacoustic response of
necessitated a solution that either reduced the intensity of the such a system and the effect of some passive countermeasures on
standpipe resonance or eliminated it altogether. the resonance intensity [36].
As discussed earlier, the simplest solution to such a problem is A small-scale model reflecting the main features of the JSFV
to enlarge the inlet or shorten the length of the standpipe (see the roll post and engine geometry was tested with air at atmospheric
middle inset in Fig. 35). This would generally increase the stand- pressure and room temperature. Figure 37 shows the experimental
pipe resonance frequency above the shear-layer excitation fre- setup and the main geometrical parameters for the test section.
quency. However, these solutions were deemed impractical. For The effects of symmetry (LB1 LB2) or asymmetry (LB1 6 LB2),
example, the standpipe length could not be reduced sufficiently to as well as the effect of the annular flow thickness, were all investi-
eliminate potential acoustic loads. Further, major modifications of gated [36]. The splitter plate shown in Fig. 37 was not originally
the main steam piping inside the reactor containment were unac- used in the model tests but was added at a later stage as a

Fig. 35 Geometry of the standpipe of safety relief valves. Left: original design;
middle: alternative design but impractical in the present case; right: final solution
with acoustic side-branch [99].

Applied Mechanics Reviews JANUARY 2014, Vol. 66 / 010802-17

Downloaded From: http://appliedmechanicsreviews.asmedigitalcollection.asme.org/ on 07/16/2015 Terms of Use: http://asme.org/terms

Fig. 36 Rolls-Royce Vertical Lift SystemV R showing the engine

and roll posts configuration of the joint strike fighter.2

Fig. 38 Frequency spectrum and mode shape for the first

acoustic mode f1. Reprinted from Ref. [36] with permission from
Elsevier Publishing.

Fig. 37 Geometry of test section showing the inner cylinder,

splitter plates, and microphone locations m1 and m2. The split-
ter plates were not used in the original design but were added
later as a countermeasure. Reprinted from Ref. [36] with per-
mission from Elsevier Publishing.

countermeasure to reduce the resonance intensity. All the results

reported hereafter were obtained without the splitter plate, except
those given in Fig. 42.
Figure 38 shows a frequency spectrum of pressure pulsations
measured at the closed end of one branch. It is seen that the lowest
acoustic mode of the branches, f1 304 Hz, is excited. This acous-
tic mode was identified by means of numerical simulation supple-
mented with the results from model tests, including the measured
resonance frequency and an observed phase shift of 180 deg
between the pressure oscillations at the two closed ends of the
branches. As shown in Fig. 38, this mode consists of one-half
acoustic wavelength (k) between the closed ends of the side-
branches. The acoustic pressure node is therefore placed in the
main annular duct, which minimizes the acoustic pressure in Fig. 39 Overview of the system acoustic response. Top figure
the main duct and the resulting acoustic radiation from the side- shows acoustic resonance (lock-in) frequency, with the straight
branches into the main duct. lines presenting the Strouhal numbers St 5 0.55 and 0.27, and
Similar spectra taken over the tested range of flow velocity pro- bottom figure shows dimensionless acoustic pressure as func-
vided the overall aeroacoustic behavior of the system, which is tions of flow velocity. Reprinted from Ref. [36] with permission
illustrated in Fig. 39. The higher acoustic modes are seen to be from Elsevier Publishing.
excited consecutively as the flow velocity is increased. All modes
are excited within a Strouhal number range of 0.270.55, and the
dimensionless amplitude of the lowest mode is the strongest of all acoustic mode in resonance. This underlines the susceptibility of
modes. Figure 40 shows contour plots of sound pressure level as a this duct arrangement to flow-excited acoustic resonance.
function of flow velocity and frequency. The consecutive excita- The effect of offsetting the length of one branch, by an amount
tion of acoustic modes can be clearly seen. For almost any flow DL LB1 LB2, to detune the resonance frequency of one branch
velocity above 20 m/s in the annular duct, there is at least one from the other was also examined. For branch offsets as small as
DL/LB 7%, the appearance of two distinct frequencies f10 and f100
was evident. These two frequencies can be clearly seen in Fig. 41,
www.flightglobal.com/news/articles/pictures-lockheed-martin-rolls-out-vertical- which shows typical pressure spectra taken from each branch at
lift-f-35b-jsf-220382/ DL/LB 42% for a flow velocity in the annular duct of

010802-18 / Vol. 66, JANUARY 2014 Transactions of the ASME

Downloaded From: http://appliedmechanicsreviews.asmedigitalcollection.asme.org/ on 07/16/2015 Terms of Use: http://asme.org/terms

Fig. 42 Acoustic pressure of the duct with and without splitter
plates. Reprinted from Ref. [36] with permission from Elsevier
Fig. 40 Sound pressure level in dB measured at the branch dent from the frequency spectra, as well as from the time signals
closed end. Reprinted from Ref. [36] with permission from of the two microphones, that these two resonant tones were
Elsevier Publishing.
generated simultaneously throughout the tested range of offset ra-
tio (see for example Fig. 41(a)).
In order to alleviate the intensity of acoustic resonances, espe-
cially that of the first mode, splitter plates were implemented in
the annular duct, as shown in Fig. 37. The purpose of the splitter
plates was to attenuate the resonance of the first acoustic mode,
which is the only mode relevant to the JSFV application, by
decoupling the two branches to weaken their mutual enhancement
of pressure pulsation. In addition, the splitter plates would
increase acoustic radiation into the main duct by imposing a
greater portion of the acoustic standing wave in the annular duct.
Figure 42 shows the acoustic pressure versus reduced velocity for
the cases with and without the splitter plates. The acoustic pres-
sure of the first mode is seen to be reduced substantially when the
splitter plates are added, ensuring a low level of pressure pulsa-
tions up to a flow velocity of 60 m/s.

7 Concluding Remarks
The excitation mechanism of acoustic resonances in piping sys-
tems containing flow over cavities has been discussed, and a
design guide is proposed that can be used to estimate the flow ve-
locity at the onset of resonance and predict the pulsation ampli-
tude as a function of flow velocity. It is important to note that the
source term, Q, which describes the integrated effect of the shear-
layer excitation, is measured for a three-dimensional, high-Reyn-
olds-number turbulent flow in cylindrical pipes. Since these fea-
tures represent the flow characteristics encountered in industry,
this source term is suitable for use in industrial applications.
The Strouhal number charts presented in this paper can be used
to determine the critical flow velocity at the onset of acoustic reso-
nance. These charts take into account the local geometrical pa-
rameters of the piping system (such as the diameter ratio of the
side-branch, the cavity depth, the shape of the upstream corner,
the effect of upstream elbows, etc.). If an acoustic resonance
Fig. 41 Pressure spectra at the end of (a) the shortened and cannot be avoided by suitable design modifications, the pulsation
(b) the unchanged branches for V 5 30 m/s, DL/LB 5 42%. (c) amplitude at resonance can be estimated by means of the acoustic
Acoustic pressure () at f10 in the unchanged branch and (3) at source term Q. Although this source term is formulated in a sim-
f100 in the shortened branch for DL/LB 5 42%. Reprinted from Ref. ple manner to allow its implementation in linear acoustics
[36] with permission from Elsevier Publishing. software, it still reflects the complex nature of the shear-layer ex-
citation, such as its nonlinear dependence on the Strouhal number
V 30.5 m/s. These frequencies corresponded to the resonance and the pulsation amplitude.
frequencies of each of the separate branches. Figure 41(c) shows The paper also includes several remedial means that can be
the magnitudes of the two acoustic tones, f10 and f100 , as functions implemented during the design stage to mitigate the resonance in-
of the flow velocity in the main duct for a branch offset ratio of tensity or eliminate it altogether whenever an acoustic resonance
DL/LB 42%. It is evident that the two tones are excited over two is expected. These include changing the dimensions of the side-
different ranges of flow velocity, with the lower-frequency tone f10 branch, using antivortex inserts, detuning the side-branches by
being excited for lower flow velocities and the higher-frequency making them of different lengths, and adding spoilers at the
tone f100 being excited for higher flow velocities. It was also evi- upstream corner of the cavity. Additional remedial methods have

Applied Mechanics Reviews JANUARY 2014, Vol. 66 / 010802-19

Downloaded From: http://appliedmechanicsreviews.asmedigitalcollection.asme.org/ on 07/16/2015 Terms of Use: http://asme.org/terms

been discussed within the scope of the presented industrial k acoustic wavelength
examples of acoustic resonance. p instantaneous acoustic power
Despite considerable advances in our ability to predict the q density of the fluid
occurrence of acoustic resonance, researchers and design engi- x flow vorticity
neers still rely on scale-model tests in many situations involving
complex flow or piping systems. In this regard, two main issues
must be carefully evaluated when performing small-scale-model References
tests. First, sound attenuation in the model must be at least similar, [1] Coffman, J. T., and Bernstein, M. D., 1980, Failure of Safety Valves Due to
if not lower, than in the full-size plant. Secondly, the model geom- Flow Induced Vibration, ASME J. Pressure Vessel Technol., 102, pp.
etry must be an exact replica of the prototype, and the termination 112118.
impedance in the model must be equivalent to that of the proto- [2] Jungowski, W. M., Botros, K. K., and Studzinski, W., 1989, Cylindrical
Side-Branch as Tone Generator, J. Sound Vib., 131, pp. 265285.
type. Since it is not possible to meet all these requirements in [3] Chen, Y. N., and Florjancic, D., 1975, Vortex-Induced Resonance in a Pipe
small-scale-model testing, the scaling of the pulsation amplitude System Due to Branching, Proceedings on Vibration and Noise in Pump,
at resonance from the model size and test conditions to the full- Fan, and Compressor Installations, Institution of Mechanical Engineers,
size and plant conditions is often not conservative. Further University of Southampton, pp. 6978.
[4] Ziada, S., and Buehlmann, E.T., 1991, Flow-Induced Vibrations in Long Cor-
research is still needed in order to develop reliable scaling laws rugated Pipes, Proceedings of International Conference on Flow-Induced
between the pulsation amplitudes in the model and the full-size in- Vibrations, Inst. of Mech. Eng., England, pp. 417426.
stallation. Additional areas that need further research and are cur- [5] Ziada, S., Buehlmann, E. T., and Bolleter, U., 1989, Flow Impingement as an
rently receiving considerable attention are the effect of moisture Excitation Source in Control Valves, J. Fluids Struct., 3, pp. 529549.
[6] Lafon, P., Caillaud, S., Devos, J. P., and Lambert, C., 2003, Aeroacoustical
content on the acoustic damping and aeroacoustic source strength Coupling in a Ducted Shallow Cavity and Fluid/Structure Effects on a Steam
in wet steam and natural gas applications [100,101] and the devel- Line, J. Fluids Struct., 18, pp. 695713.
opment of computational fluid dynamics methods that can be used [7] Smith, B. A., and Luloff, B. V., 2000, The Effect of Seat Geometry on Gate
to reliably predict not only the onset of acoustic resonance, but Valve Noise, ASME J. Pressure Vessel Technol., 122, pp. 401407.
[8] Baldwin, R. M., and Simmons, H. R., 1986, Flow-Induced Vibration in
also the amplitude of pressure pulsation [29,55,56,102,103]. Safety Relief Valves, ASME J. Pressure Vessel Technol., 108, pp. 267272.
[9] Chen, Y. N., and St urchler, R., 1977, Flow-Induced Vibrations and Noise in
the Pipe System With Blind Branches Due to Coupling of Vortex Shedding,
Acknowledgment Internoise 77, Zurich, Switzerland.
The authors gratefully acknowledge the permission of the [10] Gillessen, R., and Roller, W., 1989, Verminderung und Beseitigung von
Schwingungen an Rohrleitungssystemen (Reduction and Elimination of Vibra-
USNRC, Continuum Dynamics Inc., and Exelon Generation LLC tion of Piping Systems), Minderung von Rohrleitungsschwingungen, VDI
to include the results of the steam dryer vibration problem. With- Berichte Vol. 748, VDI Verlag, D usseldorf, Germany.
out the permission of many authors, Elsevier, and ASME to [11] Ziada, S., and B uhlmann, E. T., 1992, Self-Excited Resonances of Two Side-
include results published by other authors, this work would have Branches in Close Proximity, J. Fluids Struct., 6, pp. 583601.
[12] Ziada, S., 1993, Flow-Excited Resonances of Piping Systems Containing
not been possible. This article has been written while the first Side-Branches: Excitation Mechanism, Counter-Measures and Design Guide-
author was hosted by CNRS at LaMSID, EDF R&D, Clamart, lines, Seminar on Acoustic Pulsations in Rotating Machinery, Toronto, On-
France. tario, Canada.
[13] Ziada, S., 2010, Flow-Excited Acoustic Resonance in Industry, ASME J.
Pressure Vessel Technol., 132, pp. 19.
[14] Parker, R., and Pryce, D. C., 1974, Wake Excited Resonances in an Annular
Cascade: An Experimental Investigation, J. Sound and Vib., 37, pp. 247261.
Nomenclature [15] Ziada, S., Oengoeren, A., and Vogel, A., 2002, Acoustic Resonance in the
2LT length of T-junction transition piece Inlet Scroll of a Turbo-Compressor, J. Fluids Struct., 16, pp. 361373.
[16] Fitzpatrick, J. A., 1985, The Prediction of Flow-Induced Noise in Heat
c speed of sound Exchanger Tube Arrays, J. Sound Vib., 99, pp. 425435.
d branch pipe diameter [17] Blevins, R. D., 1993, Experiments on Acoustic Resonance in Heat Exchanger
D diameter of main pipe or duct Tube Bundles, J. Sound Vib., 164, pp. 503533.
fn frequency of resonance mode n [18] Ziada, S., Oeng oren, A., and B uhlmann, E. T., 1989, On Acoustical Reso-
nance in Tube Arrays Part I: Experiments, J. Fluids Struct., 3, pp. 293314.
h cavity depth [19] McGrath, S., and Shaw, L., 1996, Active Control of Shallow Cavity Acoustic
distance between branches Resonance, 27th AIAA Fluid Dynamics Conference, New Orleans, LA.
L cavity length [20] Kook, H., Mongeau, L., Brown, D. V., and Zorea, S., 1997, Analysis of the
LB branch length Interior Pressure Oscillations Induced by Flow Over Vehicle Openings, Noise
Control Eng. J., 45, pp. 223234.
Lc characteristic length of impinging shear layer [21] Rockwell, D., and Naudascher, E., 1978, Review Self-Sustaining Oscilla-
m order of the acoustic mode tions of Flow Past Cavities, ASME J. Fluids Eng., 100, pp. 152165.
M Mach number [22] Rockwell, D., and Naudascher, E., 1979, Self-Sustained Oscillations of Im-
n hydrodynamic (or shear layer) mode number pinging Free Shear Layers, Annu. Rev. Fluid Mech., 11, pp. 6794.
[23] Rockwell, D., 1983, Oscillations of Impinging Shear Layers, AIAA J., 21,
p time-dependent acoustic pressure pp. 645664.
P acoustic pressure (rms amplitude) [24] Blevins, R. D., 1985, The Effect of Sound on Vortex Shedding From Cylin-
P* dimensionless pressure pulsation P=1=2qV 2 ders, J. Fluid Mech., 161, pp. 217237.
Pr normalized pressure pulsation (P/qcV) [25] Hall, J. W., Ziada, S., and Weaver, D. S., 2003, The Effect of Applied Sound
on Vortex Shedding From Two Tandem Cylinders, J. Fluids Struct., 18, pp.
Q complex aeroacoustic source strength 741758.
So critical Strouhal number at onset of resonance [26] Michalke, A., 1965, On Spatially Growing Disturbances in an Inviscid Shear
St Strouhal number Layer, J. Fluid Mech., 23, pp. 521544.
u flow velocity [27] Freymuth, P., 1966, On Transition in a Separated Laminar Boundary Layer,
J. Fluid Mech., 25, pp. 683704.
v acoustic particle velocity [28] Miksad, R. W., 1972, Experiments on the Nonlinear Stages of Free-Shear-
V mean flow velocity in main pipe Layer Transition, J. Fluid Mech., 56, pp. 695719.
VB mean flow velocity in the branch of a T-junction [29] Tonon, D., Hirschberg, A., Golliard, J., and Ziada, S., 2011, Aeroacoustics of
VM mean flow velocity in main pipe of a T-junction Pipe Systems With Closed Branches, Int. J. Aeroacoust., 10(2), pp. 201276.
[30] Elder, S. A., 1980, Forced Oscillations of a Separated Shear Layer With
Vo critical flow velocity at onset of resonance Application to Cavity Flow-Tone Effects, J. Acoust. Soc. Am., 67, pp.
[31] Elder, S. A., 1978, Self-Excited Depth-Mode Resonance for a Wall-Mounted
Greek Symbols Cavity in Turbulent Flow, J. Acoust. Soc. Am., 64, pp. 877890.
[32] Elder, S. A., Farabee, T. M., and DeMetz, F. C., 1982, Mechanisms of Flow-
Dp time-dependent acoustic pressure difference across the Excited Cavity Tones at Low Mach Number, J. Acoust. Soc. Am., 72, pp.
shear layer 532549.

010802-20 / Vol. 66, JANUARY 2014 Transactions of the ASME

Downloaded From: http://appliedmechanicsreviews.asmedigitalcollection.asme.org/ on 07/16/2015 Terms of Use: http://asme.org/terms

[33] Gorter, J., Hirschberg, A., Wijnands, A., and Bruggeman, J. C., 1989, Flow [64] Peters, M. C. A. M., 1993, Aeroacoustic Sources in Internal Flows, Ph.D.
Induced Pulsations in Gas Transport Systems, International Gas Research thesis, Technische Universitat Eindhoven, Eindhoven, The Netherlands.
Conference, Tokyo, Japan. [65] Peters, M. C. A. M., and van Bokhorst, E., 2000, Flow-Induced Pulsations in
[34] Ziada, S., 1994, A Flow Visualisation Study of Flow-Acoustic Coupling at Pipe Systems With Closed Branches: Impact of Flow Direction, 7th Interna-
the Mouth of a Resonant Side-Branch, J. Fluids Struct., 8, pp. 391416. tional Conference on Flow-Induced Vibration, Lucerne, Switzerland.
[35] Ziada, S., and Shine, S., 1999, Strouhal Numbers of Flow-Excited Acoustic [66] Ziada, S., Ng, H., and Blake, C. E., 2003, Flow Excited Resonance of a Con-
Resonance of Closed Side Branches, J. Fluids Struct., 13, pp. 127142. fined Shallow Cavity in Low Mach Number Flow, and Its Control, J. Fluids
[36] Arthurs, D., and Ziada, S., 2009, Flow-Excited Acoustic Resonances of Struct., 18, pp. 7992.
Coaxial Side-Branches in an Annular Duct, J. Fluids Struct., 25, pp. [67] Aly, K., and Ziada, S., 2010, Flow-Excited Resonance of Trapped Modes of
4259. Ducted Shallow Cavities, J. Fluids Struct., 26, pp. 92120.
[37] Kriesels, P. C., Peters, M. C. A. M., Hirschberg, A., Wijnands, A. P. J., Iafrati, [68] Aly, K., and Ziada, S., 2011, Azimuthal Behaviour of\Flow-Excited Diame-
A., Riccardi, G., Piva, R., and Bruggeman, J. C., 1995, High Amplitude tral Modes of Internal Shallow Cavities, J. Sound Vib., 330, pp. 36663683.
Vortex-Induced Pulsations in a Gas Transport System, J. Sound Vib., 184, [69] Aly, K., and Ziada, S., 2012, Effect of Mean Flow on the Trapped Modes of
pp. 343368. Internal Cavities, J. Fluids Struct., 33, pp. 7084.
[38] Dequand, S., Hulshoff, S. J., and Hirschberg, A., 2003, Self-Sustained Oscil- [70] Peters, M. C. A. M., 1993, Aeroacoustic Sources in Internal Flows, Ph.D.
lations in a Closed Side Branch System, J. Sound Vib., 265, pp. 359386. thesis, Technische Universiteit Eindhoven, Eindhoven, The Netherlands.
[39] Graf, H. R., and Ziada, S., 1992, Flow Induced Acoustic Resonance in Closed [71] Huang, X. Y., and Weaver, D. S., 1991, On the Active Control of Shear
Side Branches: An Experimental Determination of the Excitation Source, Layer Oscillations Across a Cavity in the Presence of Pipeline Acoustic Reso-
International Symposium on Flow-Induced Vibration and Noise, Anaheim, nance, J. Fluids Struct., 5, pp. 207219.
CA. [72] Ziada, S., McLaren, K.W., and Li, Y., 2009, Flow-Acoustic Coupling in T-
[40] Selamet, A., Kurniawan, D., Knotts, B. D., and Novak, J. M., 2002, Whistles junctions: Effect of T-junction Geometry, ASME J. Pressure Vessel Technol.,
With a Generic Sidebranch: Production and Suppression, J. Sound Vib., 250, 131, pp. 114.
pp. 277298. [73] Harris, R. E., Weaver, D. S., and Dokainish, M. A., 1987, On the Generation
[41] Graf, H. R., and Ziada, S., 2010, Excitation Source of a Side-Branch Shear of Acoustic Resonance in Pipelines, Proceedings International Conference on
Layer, J. Sound Vib., 329, pp. 28252842. Flow Induced Vibrations, R. King, ed., Bowness-on-Windermere, England,
[42] Howe, M. S., 1981, The Influence of Mean Shear on Unsteady Aperture Flow BHRA, Cranfield, UK, pp. 543550.
With Application to Acoustical Diffraction and Self-Sustained Cavity Oscil- [74] Bruggeman, J. C., 1987, Flow-Induced Pulsations in Pipe Systems, Ph.D.
lations, J. Fluid Mech., 109, pp. 125146. thesis, Technische Universitat Eindhoven, Eindhoven, The Netherlands.
[43] Nelson, P. A., Halliwell, N. A., and Doak, P. E., 1983, Fluid Dynamics of a [75] Mohamed, S., 2012, private communication.
Flow Excited Resonance, Part II: Flow Acoustic Interaction, J. Sound Vib., [76] Ziada, S., Scott, A., and Arthurs, D., 2007, Acoustic Excitation by Flow in T-
91, pp. 375402. Junctions, ASME J. Pressure Vessel Technol., 129, pp. 1420.
[44] Nelson, P. A., Halliwell, N. A., and Doak, P. E., 1981, Fluid Dynamics of a [77] Erickson, D. D., Durgin, W. W., Maguire C. F., and Moeller, M., 1986, Shear
Flow Excited Resonance, Part I: Experiment, J. Sound Vib., 78, pp. 1538. Layer Coupling With Side-Branch Resonators, Forum on Unsteady Flow,
[45] Howe, M. S., 1980, The Dissipation of Sound at an Edge, J. Sound Vib., 70, ASME Publication No. FED 39, pp. 4345.
pp. 407411. [78] Nakibo glu, G., and Hirschberg, A., 2012, Aeroacoustic Power Generated by
[46] Howe, M. S., 1998, Acoustics of Fluid-Structure Interactions, Cambridge Uni- Multiple Compact Axisymmetric Cavities: Effect of Hydrodynamic Interfer-
versity, Cambridge, UK. ence on the Sound Production, Phys. Fluids, 24, p. 067101.
[47] Bruggeman, J. C., Hirschberg, A., van Dongen, M. E. H., Wijnands, A. P. J., [79] Jungowski, W. E., and Studzinski, W., 1989, U.S. Patent No. 4,867,190.
and Gorter, J., 1989, Flow Induced Pulsations in Gas Transport Systems: [80] Karadogan, H., and Rockwell, D., 1983, Toward Attenuation of Self-
Analysis of the Influence of Closed Side Branches, ASME J. Fluids Eng., Sustained Oscillations of a Turbulent Jet Through a Cavity, ASME J. Fluids
111, pp. 484491. Eng., 105, pp. 335340.
[48] Bruggeman, J. C., Hirschberg, A., van Dongen, M. E. H., Wijnands, A. P. J., [81] Tonon, D., Willems, J. F. H., and Hirschberg, A., 2010, Flow-Induced Pulsa-
and Gorter, J., 1991, Self-Sustained Aero-Acoustic Pulsations in Gas Trans- tions in Pipe Systems With Closed Side Branches: Study of the Effectiveness
port Systems: Experimental Study of the Influence of Closed Side Branches, of Detuning as a Remedial Measure, 20th International Congress on Acous-
J. Sound Vib., 150, pp. 371393. tics, Sydney, Australia.
[49] Stoneman, S. A. T., Hourigan, K., Stokes, A. N., and Welsh, M. C., 1988, [82] Bolduc, M., Elsayed, M., and Ziada, S., 2013, Effect of Upstream Edge Ge-
Resonant Sound Caused by Flow Past Two Plates in Tandem in a Duct, J. ometry on the Trapped Mode Resonance of Ducted Cavities, ASME PVP
Fluid Mech., 192, pp. 455484. Conference, Paris, France, Paper No. PVP2013-97149.
[50] Hourigan, K., Welsh, M. C., Thompson, M. C., and Stokes, A. N., 1990, [83] Amandolese, X., Hemon, P., and Regardin, C., 2004, An Experimental Study
Aerodynamic Sources of Acoustic Resonance in a Duct With Baffles, J. Flu- of the Acoustic Oscillations by Flows Over Cavities, ASME J. Vibr. Acoust.,
ids Struct., 4, pp. 345370. 126, pp. 190195.
[51] Graf, H. R., and Durgin, W. W., 1993, Measurement of the Nonsteady Flow [84] Ffowcs Williams, J. E., and Huang, X. Y., 1989, Active Stabilization of Com-
Field in the Opening of a Resonating Cavity Excited by Grazing Flow, J. Flu- pressor Surge, J. Fluid Mech., 204, pp. 245262.
ids Struct., 7, pp. 387400. [85] Welsh, M. C., Hourigan, K., Alfredson, R. J., and Lin, P. D., 1991, Active
[52] Graf, H. R. 1989, Experimental and Computational Investigation of the Flow Control of Flow-Excited Acoustic Resonance: Higher Order Acoustic Modes,
Excited Acoustic Resonance in a Deep Cavity, Ph.D. thesis, Worcester Poly- International Conference on Flow-Induced Vibration, Brighton, UK.
technic Institute, Worcester, MA. [86] Sarno, R. L., and Franke, M. E., 1994, Suppression of Flow-Induced Pressure
[53] Radavich, P. M., Selamet, A., and Novak J. M., 2001, A Computational Oscillations in Cavities, J. Aircr., 31, pp. 9096.
Approach for Flow-Acoustic Coupling in Closed Side Branches, J. Acoust. [87] Cattafesta III, L. N., Garg, S., Choudhari, M., and Li, F., 1997, Active Con-
Soc. Am., 109, pp. 13431353. trol of Flow Induced Cavity Resonance, 28th AIAA Fluid Dynamics Confer-
[54] Martnez-Lera, P., Schram, C., F oller, S., Kaess, R., and Polifke, W., 2009, ence, Snowmass Village, CO.
Identification of the Aeroacoustic Response of a Low Mach Number Flow [88] Ziada, S., 2003, Control of Fluid-Structure-Sound Interaction Mechanisms by
Through a T-joint, J. Acoust. Soc. Am., 126, pp. 582586. Means of Synthetic Jets, JSME Int. J., 46, pp. 873880.
[55] Nakiboglu, G., Belfroid, S., Golliard, J., and Hirschberg, A., 2011, On the [89] Ziada, S., 1995, Feedback Control of Globally Unstable Flows: Impinging
Whistling Corrugated Pipes: Effect of Pipe Length and Flow Profile, J. Fluid Shear Flows, J. Fluids Struct., 9, pp. 907923.
Mech., 672, pp. 78108. [90] Ziada, S., 2001, Interaction of a Jet-Slot Oscillator With a Deep Cavity Reso-
[56] Nakiboglu, G., Manders, H. B. M., and Hirschberg, A., 2012, Aeroacoustic nator and Its Control, J. Fluids Struct., 15, pp. 831843.
Power Generated by a Compact Axisymmetric Cavity: Prediction of Self- [91] NRC, 2002, Failure of Steam Dryer Cover Plate After a Recent Power
Sustained Oscillation and Influence of the Depth, J. Fluid Mech., 703, pp. Uprate, NRC Information Notice 2002-26, US Nuclear Regulatory Commis-
163191. sion, Washington, D.C.
[57] Mohamed, S., Graf, H. R., and Ziada, S., 2011, Aeroacoustic Source of a [92] Hambric, S. A., Mulcahy, T. M., Shah, V. N., Scarbrough, T., and Wu, J.,
Shallow Cavity in a Pipeline, Proceedings of Pressure Vessels and Piping 2000, Acoustic Loading on BWR Steam Dryers Caused by Valve Singing,
Conference, Flow Induced Vibration Symposium, Paper No. PVP2011-57437, 9th NRC/ASME Symposium on Valves, Pumps and Inservice Testing, Wash-
Baltimore, MD. ington, D.C.
[58] Kinsler, L. E., Frey, A. R., Coppens, A. B., and Sanders, J. V., 2000, Funda- [93] DeBoo, G., Ramsden, K., Gesior, R., and Strub, B., 2007, Identification of
mentals of Acoustics, 4th ed., Wiley, New York. Quad Cities Main Steam Line Acoustic Sources and Vibration Reduction,
[59] Morse, P. M., and Ingard, K. U., 1968, Theoretical Acoustics, Princeton Uni- ASME Pressure Vessels and Piping Division Conference, San Antonio, TX.
versity, Princeton, NJ. [94] NRC, 2004, Additional Flow-Induced Vibration Failures After a Recent
[60] Evans, D. V., Levitin, M., and Vassiliev, D., 1994, Existence Theorems for Power Uprate, NRC Information Notice 2002-26, Supplement 2, Jan. 9, US
Trapped Modes, J. Fluid Mech., 261, pp. 2131. Nuclear Regulatory Commission, Washington, D.C.
[61] Duan, Y., Koch, W., Linton, C. M., and McIver, M., 2007, Complex [95] GENE, 2003, BWR Steam Dryer Integrity, Services Information Letter, SIL
Resonances and Trapped Modes in Ducted Domains, J. Fluid Mech., 571, pp. No. 644, Supplement 1, Sept., GE Nuclear Energy, San Jose, CA.
119147. [96] Ramsden, K., 2005, Acoustic Circuit Validation Quad Cities Unit 2 Instru-
[62] Hein, S., and Koch, W., 2008, Acoustic Resonances and Trapped Modes in mented Steam Path Final Model Revision 930 MWe/2884 MWt Power Level,
Pipes and Tunnels, J. Fluid Mech., 605, pp. 401428. Exelon Report No. AM-2005-004.
[63] Koch, W., 2009, Acoustic Resonances and Trapped Modes in Annular Plate [97] Kitajima, Y., Watanabe, M., Matsunaga, K., and Hagiwara, T., 2006,
Cascades, J. Fluid Mech., 628, pp. 155180. Acoustic Analysis for a Steam Dome and Pipings of a 1,100 MWe-Class

Applied Mechanics Reviews JANUARY 2014, Vol. 66 / 010802-21

Downloaded From: http://appliedmechanicsreviews.asmedigitalcollection.asme.org/ on 07/16/2015 Terms of Use: http://asme.org/terms

Boiling Water Reactor, Proceedings of ICAPP06, Reno, NV, Paper No. [101] Uchiyama, Y., and Morita, R., 2012, Experimental Evaluation of Acoustic
6358. Resonance at Single Stub Pipe in Each Dry and Wet Steam Flow, 10th Inter-
[98] Takahashi, S., Ohtsuka, M., Okuyama, K., Ito, T., and Yoshikawa, K., 2008, national Conference on Flow-Induced Vibration, Dublin, Ireland, pp.
Experimental Study of Acoustic and Flow-Induced Vibrations in BWR Main 577584.
Steam Lines and Steam Dryers, ASME Pressure Vessels and Piping Division [102] Mohany, A., and Ziada, S., 2009, Numerical Simulation of the Flow-Sound
Conference, Chicago, IL. Interaction Mechanisms of a Single and Two Tandem Cylinders in Cross-
[99] Bilanin, A. J., 2006, private communication. Flow, ASME J. Pressure Vessel Technol., 131, p. 031306.
[100] Omrani, P. S., Golliard, J., Belfroid, S., and Diez, G., 2012, On the Effect of [103] Paduano, C., and Meskell, C., 2012, Does Reynolds Number Affect Look-In
Droplets Modulating the Acoustic Source Powers Generated in T-Joints, 10th of Aeroacoustic Resonance of Tandem Cylinders?, 10th International Confer-
International Conference on Flow-Induced Vibration, Dublin, Ireland, pp. 531536. ence on Flow-Induced Vibration, Dublin, Ireland, pp. 227234.

010802-22 / Vol. 66, JANUARY 2014 Transactions of the ASME

Downloaded From: http://appliedmechanicsreviews.asmedigitalcollection.asme.org/ on 07/16/2015 Terms of Use: http://asme.org/terms