Nuclear Engineering and Design 74 (1982) 31-60 31

North-Holland Publishing Company

A REVIEW OF FLOW-INDUCED VIBRATIONS IN REACTORS AND REACTOR

COMPONENTS *

M.P. P A I D O U S S I S
Department of Mechanical Engineering, McGill University, Montreal, Quebec, Canada

Received 10 November 1982

This paper presents a review of the state of the art of two classes of vibration problems encountered in reactors and reactor
peripherals: namely, vibration of cylindrical structures induced by cross flow and by axial flow. A historical perspective is
given first, in which the milestone contributions that have advanced the state of the art are highlighted. Then recent
developments in the last decade, with emphasis on those in the last three years, are discussed, concluding with an assessment of
the state of understanding of the fluidelastic mechanisms involved, on the one hand, and of predictive tools available to the
designer, on the other.

I. Introduction have to be reported and thoroughly studied. The realiza-

tion that flow-induced vibration problems can cause
Consideration of flow-induced vibration has tradi- complete shutdowns of nuclear plants for prolonged
tionally not been an integral part of design of nuclear periods of time, ranging from days to years, and the
plant in the same sense as, for instance, reactor physics enormous power-replacement costs entailed, have con-
and thermohydraulics. Design deficiency, in terms of vinced designers that they may only ignore flow-in-
the latter, may mean that the plant will not work at all, duced vibration at their peril.
while inadequacy in terms of flow-induced vibration There are a host of possible vibration problems in
considerations may simply mean that smooth operation nuclear plant, in the reactor itself, as well as in the
will be short-lived. This distinction, coupled with inex- associated piping, heat exchangers and steam genera-
cusable disregard of one of Murphy's laws, caused tors; ancillary equipment, such as in-core instrument
flow-induced vibration to be relegated to the status of a tubes, "secondary" piping, etc., are often involved [1].
secondary design parameter; indeed, prior to the mid- Most of these diverse vibration problems may be classi-
1960's, it was often totally ignored: if a structure could fied under four broad categories [ 1], namely those caused
stand up to the steady fluid loads, then a healthy safety by (i) cross flow, (ii) internal, or (iii) external axial flow,
factor ought to take care of the unsteady ones underly- and (iv) annular or leakage f l o w - - although other clas-
ing vibration, it was optimistically reasoned. sifications and approaches for reviewing the state of the
Some spectacular failures, some of which are de- art are clearly also possible; vide ref. [2,3].
scribed and analyzed in another review [1], have since This paper will only examine two of these classes of
put flow-induced vibration in nuclear reactors and pe- vibration. Specifically, it will deal with vibration of
ripherals on the map, so to speak; this was aided by the structures, and especially cylindrical structures, induced
fact that all incidents and malfunctions in nuclear plant by either external axial or cross flow **. The paper will

* This paper has been presented as an invited lecture at the ** It is obvious, of course, that this classification involves
Second International Topical Meeting on Nuclear Reactor drastic idealization of the actual flow field in reactors and
Thermal-Hydraulics sponsored by the American Nuclear heat exchangers. It is, nevertheless, retained because, as
Society, American Society of Mechanical Engineers and will become obvious later, the state of knowledge is less
American Institute of Chemical Engineers, in Santa than adequate, even for such highly idealized flow condi-
Barbara, California, 11-14 January 1983. tions, let alone for more realistic flow fields.

0029-5493/82/0000-0000/$02.75 © 1982 N o r t h - H o l l a n d
32 M.P. Paidoussis / Flow-induced vibrations in reactors

thus provide an up-dated and expanded version of acterized by 40 _< Re _< 2 x l0 s and Re >_ 3.5 x 106, re-
another recent review of the state of the art [4], in which spectively, periodic shedding of vortices--with Strouhal
emphasis will be placed on more recent developments in numbers S---0.2 and S = 0.3, respectively, for circular
the field, and where flow-induced vibration of solitary c y l i n d e r s - - m a y cause large amplitude oscillations if
cylinders, in addition to cylinder clusters, will also be system damping is small. In the intermediate, critical
reviewed. Because of rapid development and topicality, range of Reynolds numbers, periodicity is not sharply
the lion's share of this review goes to cross-flow-induced defined; the values of S, if at all definable, vary widely
vibration of cylinder arrays. and can be as high as 0.50. All the foregoing, and
The obvious fact should be stressed that any struct- especially in the critical r6gime, are highly dependent on
ural component placed in flow will vibrate, to a greater flow confinement (blockage), surface roughness, inci-
or lesser extent. Vibration per se is not bad; indeed dent free-stream turbulence intensity and scale, aspect
recent studies, e.g. [5,6], attempt to take advantage of ratio, two dimensionality of flow, and vibration ampli-
vibration to enhance heat transfer! Only sometimes are tude. The same is true for the fluctuating components of
the effects of vibration bad; either catastrophic, involv- lift and drag induced by periodic vortex shedding; a
ing almost immediate failure of components, or more tabulation of most anterior studies on fluctuating loads
subtle, involving longer-term damage by fretting wear on circular cylinders has recently been compiled [19].
and fatigue which may limit the life span of compo- If vibration amplitude is not negligibly small, i.e. if it
nents. In the former case, it is sufficient for the designer is greater than 0.01D to 0.02D, and if the parameter
to know how to avoid conditions leading to such m S / o D 2 is not too large, then lock-in may arise, where
catastrophic failures. In the latter case, however, predic- the vortex-shedding frequency, fvs, is entrained to fol-
tion of vibration behaviour is not enough: the link low the cylinder natural frequency, fn, over a range of
between impact level and fretting wear for different flow velocities; the width of this lock-in range depends
materials and environments, or between vibration-in- on the vibration amplitude. Several studies have been
duced stress and fatigue, as the case may be, are essen- conducted, where the cylinder is either forced to move
tial blocks of knowledge that a designer must equally in a prescribed manner (usually transversely to the flow)
have at his disposal. or left free to vibrate in response to excitation by the
flow. Studies with amplitudes ranging from 0.01D to
several diameters have been conducted, e.g. [20-24]. In
2. Solitary cylinder in cross flow the lock-in region, there is an increase in the strength of
the vortices shed and in the span-wise correlation--
The forces exerted on a solitary bluff body, and hence also in the fluctuating loads exerted on the cylin-
especially on a circular cylinder, subjected to cross flow der.
have been studied quite extensively for a very long time, In addition to normal resonance, harmonic and sub-
starting with Strouhal's work [7] in 1878. Extensive harmonic resonances have also been observed, e.g.
reviews are available on this topic, e.g. [8-16], mostly [20,24]. Moreover, oscillation predominantly in the
treating the classical problem of vortex shedding and streamwise direction at 2fv s is also possible [25-27], in
the vibrations induced thereby. Interested readers are contrast to the usual case where the oscillation is pre-
also referred to refs. [17,18]. dominantly in the cross-stream direction.
Two phenomena are of interest insofar as flow-in- The literature on vortex shedding about solitary cyl-
duced vibration is concerned: vortex shedding and inders and on the forces and motions induced thereon is
turbulent buffeting. truly overwhelmingly extensive and cannot possibly be
properly reviewed in the limited space available
2.1. Vibration induced by vortex shedding here--considering that there literally exist hundreds of
important papers on the various aspects of the topic.
The vortex-shedding mechanism is well known and Extensive work has been done on various parameters
has been experienced by any fast swimmer on his arms. affecting vortex shedding and on the steady and
In the subcritical and supercritical flow r6gimes *, char- fluctuating loads induced on the cylinder, e.g. on the
effect of confinement (blockage) [28], end conditions
* There is some confusion in nomenclature here: some [29,30], aspect ratio [30,31], mass ratio [32], upstream
authors define the three r~gimes as subcritical, supercriti- turbulence [33-35], surface roughness [36-38], proxim-
cal and transcritical, in ascending Reynolds number se- ity to a wall [39,40], angle of attack of the flowing fluid
quence, rather than subcritical, critical and supercritical. [41], and phase composition of the fluid [42], to mention
 M.P. PaMoussis / Flow. induced vibrations in reactors 33

Table 1 found to be 0.126 **, 0.113~, 0.116y, 0.128, respec-

Predictions of resonant vortex-induced vibration amplitude of tively, according to these four models, where it is noted
circular cylindrical structures as a function of damping that 1.00 < y < 1.30 typically. These expressions, how-
ever, overpredict the amplitude of vortex-induced vibra-
Author Predicted resonant tion if it is small, because, as the vibration becomes of
amplitude
order < 0.1 D, the span-wise correlation of vortex shed-
ding becomes smaller, while in these expressions it is
Hartlen & Currie [43,44] --=AY 0.0505
assumed to extend over the total length of the cylinder.
D [3.36+8f]1/2S2
(For a stationary cylinder, the correlation length varies
Griffin, Skop & Ramberg A_Z= 1.29"/
from 2D to 12D, according to various studies [2], but
[45] D [1+0.43(2~rS 2 ~r* )] 3 "35 for vibrating cylinders, it is strongly dependent on the
Ay 0.07y amplitude of motion [48].) Finally, it should be stressed
Iwan & Blevins [46] D ( ~r + 1.9) S 2 that table 1 gives the resonant amplitude of vibration.
1/2
x ( 0.3 + 0.72 For a more general discussion the reader is also referred
k (rr + 1.9) S } to ref. [16,49].
A_zy= 0.32 All the above relate to circular cylinders. There also
Sarpkaya [47] exists a whole body of literature on vortex-shedding and
D [O.06+(21rS28r)2]l/2
its effects on non-circular bluff bodies of triangular,
rectangular, semi-profiled, elliptical and other cross sec-
tions [16,18]. These will not be discussed here, nor will
the interesting effects of sound on vortex shedding, e.g.
some of the principal ones, where the references quoted [50,51]. Consequently, this review will also not deal with
are illustrative, rather than exhaustive. These parame- galloping, which is an important consideration in the
ters constitute a complex multi-dimensional space, af- design of structural members of non-circular cross sec-
fecting all the foregoing discussion on vortex shedding tion subjected to cross flow [2].
and on vortex-excited vibration, both quantitatively and
qualitatively. 2.2. Vibration induced by turbulent buffeting
Having said all this, the designer must nevertheless
be provided with some tools for dealing with this vibra- In addition to vortex-induced forces, a cylinder in
tion problem. The usual design strategy is to avoid cross flow is also subjected to turbulent buffeting due to
resonance and lock-in, if possible, e.g. by altering fn and the random pressure perturbations of the fluid stream
keeping it + 40% away from fvs. Another possibility, if acting on the cylinder surface. Buffeting is operative at
space permits, is to suppress vortex shedding or reduce all flow velocities and flow rrgimes. The cylinder acts
its strength, e.g. by splitter plates, helical strakes or a somewhat like a band-absorption filter, extracting en-
perforated shroud *--although some or all of these may ergy from the turbulent field in narrow windows about
be impractical in industrial environments. If neither of the natural frequencies of the cylinder (mostly about the
the above is possible, then the only option is to increase lowest natural frequency). In this respect, this vibration,
the damping, so as to reduce the vibration amplitude. which is generally of small amplitude, is similar to that
In Blevin's review [2], the expressions for vibration excited by turbulence in axial flow situations and in
amplitude, Ay, according to three analytical models are cross flow through cylinder arrays, to be discussed later.
given in tabular form, reproduced here as table 1; how- Recent work by So and Savkar [35] shows that the
ever, since most of these models draw upon Hartlen and distinction between vortex shedding and buffeting be-
Currie's remarkable work [43,44], that model has also comes rather arbitrary when the intensity of incident
been added to the table. The various quantities are turbulence exceeds a certain level. Thus, for turbulent
defined in the Nomenclature. intensities of the order of 10% the spectrum of fluctuat-
All four models give similar answers and correctly ing lift on the cylinder becomes very much broader than
predict the self-limiting vortex-induced vibration at large in laboratory situations; indeed, there is no longer an
amplitudes. For 6r = ~r2, S = 0.2, for example, Ay//D is identifiable vortex-shedding "peak". Also, the fluctuat-

* The writer has belatedly become aware of an excellent ** In the Hartlen & Currie model, the numerical value of 3.36
review of means for suppressing vortex shedding by is approximate [44], but the possible error is small, so long
Zdravkovich [240]. as 8r is not too small.
34 M.P. Paidoussis / Flow- induced vibrations in reactors

ing lift coefficients diminish drastically, as compared to

those for turbulence intensities of the order of 1% or
less *. Most importantly, there is some experimental
evidence that, under conditions of high turbulence,
lock-in is entirely eliminated [52]; indeed, a very re- For Equilateral Arrays
cently proposed design procedure [53] for the response D/D._st._stcosO;O.30 °
of a solitary cylinder to turbulent cross flow takes
advantage of this factor--recognizing nevertheless that
it has not yet incontrovertibly been established. Fig. 1. Definition diagram for cylinder arrays in cross flow.

2.3. Shell-type vibrations

vis-a-vis the flow [59,60]; this is because the overall flow
If the cylinder is sufficiently thin, then in addition to field is affected by the presence of the additional cylin-
beam-like motions, it may also be subjected to shell-type der, and also because of wake interference effects.
vibrations [54-57]. Although it was originally thought Several studies have been made--especially in connec-
that these vibrations were sub-harmonically excited by tion with applications to structural vibration induced by
vortex shedding [54] it now seems more likely that they wind and marine c u r r e n t s - - i n which the two cylinders
are an aeroelastic flutter phenomenon [56]. are in tandem, side-by-side or in staggered arrangement.
A n excellent review has been given by Zdravkovich [60].
2.4. State of the art for vibration of a solitary cylinder in The present review, however, will mainly deal with
cross flow the case of cylinder arrays, involving many cylinders in
repeated geometric pattern, such as shown in fig. 1.
The degree of understanding of the unsteady fluid Traditionally, four distinct mechanisms have been
mechanics of flow about a solitary cylinder in cross flow thought to be operative, giving rise to flow-induced
is far more advanced than any of the other phenomena vibration: turbulent buffeting, vortex shedding, jet
discussed in this paper. Yet, new insights are still possi- switching and fluidelastic instability--vide [ 1-4, 61-72].
ble, e.g. even in conjunction with the very process of This section will deal with the first two, as well as with
rolling up of vortices behind the cylinder [58]. the related problem of acoustic resonances.
From the point of view of design data, a great deal
remains to be done and, in the context of this paper, 3.1. Buffeting and "vortex shedding" in cylinder arrays in
especially for cylinders in high turbulence level, "in- cross flow." developments to circa 1978
dustrial" cross flows. Savkar and So's work [35,52]
shows how different the forces acting on a cylinder in Early studies on vibration of tube banks in heat
such environments can be, as compared to those under exchangers proceeded on the assumption that vortex
normal laboratory conditions, and that even the lock-in shedding was the main, if not the only, mechanism
phenomenon may be suppressed. Nevertheless, some exciting vibration. Hence, the main concern was to
design guides, as well as compilations of data that can determine the pertinent Shrouhal numbers for various
be used as design guides, are available [1-3, 16-19, 53] geometrical layouts of the tubes [73-79], so as to enable
for judicious use by the seasoned designer. designers to avoid resonance under operating condi-
tions. Two kinds of resonance were recognized as hav-
ing to be avoided: resonance with the tube natural
3. Cylinder arrays in cross flow. 1. Vibration induced frequencies, and acoustical resonance with the fluid
by buffeting and "vortex shedding" medium contained by the outer walls of the system--the
former potentially damaging the tubes, and the latter
The dynamical behaviour of two cylinders in cross the walls of the system as well as the hearing of those
flow, positioned less than n diameters apart, may be working in the vicinity [1].
radically different from that of a solitary one, where n Buffeting excitation of cylinders in arrays was also
depends on the geometrical disposition of the pair recognized as giving rise to vibration, similarly to the
case of solitary cylinders. The difference here is that the
* These findings are important additional reasons why the incident turbulent field in a tube bank is generated
models of table 1 are likely to overestimate the amplitude of mostly by the upstream rows of tubes, rather than by
vibration in industrial environments. free-stream turbulence. It was generally believed that
 M.P. Paidoussis / Flow-induced vibrations in reactors 35

the vibration generated by buffeting excitation is of very mentioned in the first paragraph of this section. By
small amplitude, and hence considerably less worrisome sound physical reasoning, he then went on to predict
than that produced by the other phenomena. that in a cylinder array, characterized by the geometric
Thus, prior to the mid-1960's, buffeting and vortex parameters s t and s t, this predominant frequency fb
shedding were considered to be two distinct and unre- may be found from the corresponding Strouhal number
lated mechanisms giving rise to vibration in cylinder S b, as follows:
arrays. Moreover, the vibration in both cases was con-
sidered to be forced by the flow field, and the possibil-
S b = [3.05(1 - l / s t ) 2 + 0.28](sest)-'. (1)
ity of lock-in was not discussed. Owen's work did not gain wide acceptance, and Y.N.
In 1964, in a remarkable paper, Owen [80] disputed Chen and his followers [79] maintained that there exists
the very existence of "vortex shedding" within closely vortex shedding as such, even deep in a cylinder
spaced cylinder arrays, as well as the separate identity array--although later conceding [81,82] that it may not
of the two mechanisms of vortex shedding and buffet- be of the classical form, and proposing instead the
ing. Indeed, it is rather difficult to visualize shear layers so-called "jet switch", "wake swing" and "jet instabil-
separating from the cylinders and neatly rolling up into ity" models, which, although conceptually appealing,
vortices in the classical manner, in view of the narrow- were not experimentally substantiated.
ness of the flow passages--let alone the formation of In due course, by collecting data from various
conventional Khrmhn-Brnard vortex streets. According sources, maps were constructed of Strouhal numbers
to Owen, "Deep enough within a bank of tubes, the associated with "vortex shedding", Svs, for various val-
cumulative growth of random irregularities in the ues of s t and s~, notably by Fitz-Hugh [83,18] and by
labyrinthine-like, high-Reynolds-number flow must lead to Chen [84]. These maps are quite intricate, displaying
a state of complete incoherence on which it is difficult to pockets of high Svs in otherwise low-lying areas--re-
imagine any superposed regular pattern.., to be discerni- sembling somewhat the map of Europe at the time of
ble." He then went on to propose that the observed the Thirty Years' War. One of the reasons for this was
periodicity in cylinder arrays may simply be associated suspected to be [1] that the data in which there was
with the dominant frequency, fb, in the fairly broad- acoustic reinforcement are not separated from those in
band turbulent energy spectrum. Thus, the observed which there was not. (In a very recent paper, an attempt
resonances, according to Owen, would be generated by has been made to expurgate the data [85], and the maps
coincidence of fb with fn or fa, for the two resonances begin to look more orderly.)

I I I
1.4 1.4
- - Chen -- Chen
1.2 - - - Fitz - Hugh 1.2 - - - Fitz- Hugh
\ ° Owen o Owen
S s
1.0 1.0
o \ |
l
I
Q8 O.8 1

0.6 \\ o

%%%%% o
0.6
\~-~-~ Layout
o ',,~Angle = 45"
%% 0
0.4 Q4

0.2 50 0.2

0.0 0.0
3 4¸ 5 3
St St

Fig. 2. Strouhal numbers of dominant flow periodicity in an array in cross flow, according to Y.N. Chen, Fitz-Hugh and Owen. For
definition of layout angle 9, see fig. 1.
36 M.P. PaMoussis / Flow-induced vibrations in reactors

Fig. 2 shows the Strouhal numbers for different bers, specific types of construction (and hence mass
layout angles (vide fig. 1) according to Y.N. Chen's and ratios and damping parameters), and so on.
Fitz-Hugh's maps and Owen's equation (1). It is noted All the foregoing make it obvious that circa 1978,
that, although Chen and Fitz-Hugh draw their data there was not only a rather confused state of under-
mostly from the same sources, large discrepancies do standing of the unsteady fluid mechanics of interstitial
nevertheless exist between them. Moreover, these dif- flow in cylinder arrays, but also an incomplete and
ferences are of the same order as those between the often contradictory data base available for practical
vortex-shedding Svs of Chen's and Fitz-Hugh's and the design purposes.
buffeting S b of Owen's. It was concluded [1] that there Nevertheless, it was necessary to develop some design
is but one resonance mechanism at work, but one that guidelines, and a good beginning was made by Pet-
has been explained in terms of different hypotheses, by tigrew et al. [90-92]. They proposed sound analytical
the vortex-shedding advocates, on the one hand, and the models (a) for buffeting vibration, assuming the excita-
buffeting disciples, on the other. For this reason, in this tion to be a random process, and (b) for "vortex-shed-
paper the frequency of a dominant periodic component ding" response at resonance, assuming the excitation to
in the fluid medium which varies proportionately to the be harmonic. Using these models and their extensive
flow velocity will simply be called Strouhal frequency, measurements of vibration in cylinder arrays, they
fs, whatever its cause; the corresponding Strouhal num- worked backwards and extracted (i) the pertinent com-
ber will be denoted by S. ponents of the power-spectral-densities of turbulent
However, whatever is the true mechanism underlying buffeting excitation and (ii) the fluctuating lift coeffi-
periodicity in the flow within cylinder arrays, matters cients at resonance between fs and fn' This data, to-
are even more complicated than indicated so far: it was gether with the analytical models, could then be used as
found that flow periodicity does not always arise and conservative design tools. The same analytical model for
that, when it does, it does not necessarily occur buffeting response, in a more elaborate form, was also
throughout the array. The following observations have presented by Blevins [93]. These will be presented, in
been made: their up-dated forms, in sections 3.3 and 3.4.
(i) For gaseous flows, Gibert et al. [86] found that Before presenting more recent developments, the
resonances may occur only if m 3 l / 2 / p D 2 < 30, and question of acoustic resonances should be discussed.
they suggest that lock-in actually takes place. However, Periodicity in the flow through cylinder arrays is greatly
the phenomenon was observed at maximum intensity enhanced if there is coincidence with an acoustic mode
only in the first few rows [87]. of the flow-containing vessel or duct [1], resulting in
(ii) For gaseous flows at low Reynolds numbers intense sound amplification and large fluctuating pres-
(Re < 700), Grover and Weaver [88] found that discrete sure levels on the containing walls. The deafening sound
periodicity in the flow could be observed in the first 15 levels that may result from this type of resonance are a
rows; but at higher Re, increased turbulence levels very real problem, both for the personnel and for the
overwhelm or suppress any discrete fs. integrity of the system, although not necessarily causing
(iii) For gaseous flows, Zdravkovich and Namork large vibrations of the cylinders. This phenomenon is
[89] found that, for a closely packed double-row array, associated mainly with gas flows. In water, the speed of
the flow structure around the cylinders in the first row sound is higher and equipment dimensions generally
was intrinsically different from that in the second--with smaller, both implying high f,; at the same time flow
one Strouhal frequency (which they equate to 2fs) ap- velocities are smaller, so that fs tends to be lower; hence,
pearing in both rows, and another, equal to fs, only in resonance conditions are less likely to materialize in
the second row. water. The interested reader is referred to refs.
(iv) For liquid flows, Pettigrew et al. [90] found that [79,94-97,1]. The pertinent available design guidelines
resonances between fn and fs occur only in the first few will be presented in section 3.5.
rows, with 0.3 < S < 0.7, and suggest that such reso-
nances need be considered only if ( U / f n D ) - 1 < 1. 3.2. Buffeting and "'vortex shedding" in cylinder arrays in
(v) For two-phase flows, Pettigrew et al. [90] report cross flow: recent developments
that they have never observed flow-induced resonances
of this type. Zdravkovich and Namork pursued their studies of
These observations were not claimed by the the interstitial flow structure in closely packed normal
researchers involved to be general, having been made equilateral arrays ( p / D = 1.375) in air, taking surface
over limited ranges of geometries and Reynolds num- pressure readings on several cylinders and making hot-
 M.P. PaMoussis / Flow-induced vibrations in reactors 37

wire anemometer traverses between rows [98,99]. Large with amplitude 10-3D to 10-2D. At resonance (fs =fn),
turbulence intensities, as high as 50% were observed. It the fluctuating lift coefficient peak was found to be
was found that the flow structure in the first row is much broader and six times lower than for a solitary
radically different from that in the second, and that the cylinder. For cylinders free to vibrate, lock-in was found
third-row cylinders are affected not only by the narrow to occur, with a lock-in range (in terms of U / f D )
wakes behind the second-row cylinders but also by the several times wider than for a solitary cylinder.
turbulence generated by the first-row wakes. From the Similar measurements were made by Blevins et al.
third row onwards, the interstitial flow pattern remains [102] for quasi-in-line (s t = 1.86, s~e= 1.4) twelve-row
almost unchanged. Moreover, it was found that there arrays in air; in addition to fluctuating lift spectra, the
exists a strong correlation between velocity and turbu- span-wise correlation coefficients were also measured,
lence profiles in between the rows. In these experiments on both immobile and force-vibrated cylinders. The
fn and fs were far removed from each other, allowing the amplitude of the lift force was found to increase from
study of buffeting response of the cylinders away from the first, to the second and third rows, but then to
resonance. Second-row cylinders were found to be remain virtually unchanged (in agreement with Zravko-
excited most. vich's observations [98]). It was also found that when
Chen [100] conducted an extensive set of experi- the spectra, S L ( f ) , are plotted in dimensionless dia-
ments with both closely-packed and widely-spaced grams of S L ( f ) [ ½ p D U 2 ] - 2 ( U / D ) versus f D / U , a col-
staggered arrays ( s t / s e = 4.00/3.76, 2.00/3.76; lapse of the data for different U is achieved, on sensibly
4.66/1.40; 1.66/2.80; 2.33/1.40) in air, measuring cyl- the same curve. This data may be used to complement
inder response, as well as velocity fluctuations in the Blevin's analytical modeling of buffeting response [93].
fluid with hot-wire anemometers. In most cases several The spectra were found to display a very weak hump at
fs were found in the array, one of which was generally 0.12 < f D / U < 0 . 2 0 , most notable in the second row,
dominant; acoustic resonances were also clearly ob- which was identified "with the remnants of organized
served, as well as lock-in of fs to the dominant cavity fa. vortex shedding". As expected, similarly to a solitary
For large longitudinal spacings (s~e= 3.76), where it is cylinder, the span-wise correlation drastically increases
supposed that there is space enough for something akin with cylinder motion.
to "classical" vortex shedding to occur, the cylinder G o r m a n [103] studied the effect of upstream turbula-
response displays (i) an important peak corresponding tots on deep cylinder arrays, taken from actual heat
to resonance between fn and fs, and (ii) a sizable vibra- exchangers, in water flow (normal equilateral, p / D =
tion of the cylinder at both fn and fa, in the region of 1.36; square, s t = s ~ = 1.37, 1.47). These arrays "were
acoustical lock-in of fs. For arrays with smaller se, the known to have large vortex resonances in their inlet row
observed f~ are supposed to arise from Chen's "jet tubes", most pronounced in the first and second rows.
instability" ( s t / s t = 4.66/1.40), "wake swing" (st~st = Various types of turbulators were tried: screens, flat-strip
1.66/2.88) and "jet switch" (st/se = 2.33/1.40) mecha- grids and rigid-tube grids. The first were found to have
nisms. What is most important, however, is that in these little effect, while the latter two, with axes perpendicular
arrays no peaks were observed in the vibration am- to the cylinder axes, were found to be remarkably
plitude versus flow-velocity curves at the point of coin- effective in suppressing vibration; this was attributed to
cidence of fs with f,; although the increase in amplitude their ability to break up the correlation of the fluid
is not smooth, Chen concedes that the response may forces on the cylinders.
practically be computed as due to buffeting. Signifi- G o r m a n later extended his experiments to multi-span,
cantly, the computed fluctuating lift coefficients are in multi-pass arrangements [104,105] to model more accu-
all cases small, generally less than 0.2 and often less rately the situation in real-life heat exchangers.
than 0.1. The maximum lift coefficient varies from row A similar study to Gorman's [103] on the effect of
to row; for s,e = 3.76, for example, the maximum occurs upstream turbulators on arrays of rigidly held cylinders
in the fifth row. Finally, it should be said that the in air ( p / D = 1.2, 1.5, 1.7) was conducted by Savkar
significance of this work transcends the important re- and Litzinger [106]. It this case the fluctuating lift and
sults obtained, in that it repesents, in a certain sense, a drag coefficients in different rows and the turbulence
reconciliation of two divergent schools of thought. and other flow characteristics were carefully mea-
Fluctuating lift coefficients were determined by Gi- s u r e d - b o t h without and with turbulators. The follow-
bert et al. [101] on single- and double-row arrays in air ing observations, among others, were made: (i) there
( p / D = 1.24), by taking pressure measurements on cyl- appears to be a transition from a subcritical to a super-
inders which were either rigidly held or forced to vibrate critical r6gime with increasing Re, akin to that for a
38 M.P. Paidoussis / Flow-induced vibrations in reactors

solitary cylinder, with a sudden drop in the fluctuating correlated span-wise, the cylinders to be simply sup-
lift coefficient at a critical Re; (ii) the dominant Strouhal ported and the response to be prinicipally in the first
peak for both lift and drag spectra occurs at the same mode. Since the data with which it must be used corre-
frequency, implying that the mechanism underlying in- spond to these same simplifying assumptions, this pro-
terstitial flow periodicity in the array is not that of vides a neat and effective way of obtaining a conserva-
"classical" vortex shedding; (iii) square-mesh grid tive estimate of vibration amplitude; the mid-span r.m.s.
turbulators entirely eliminate the Strouhal peak, but the amplitude is given by the remarkably simple form
buffeting spectra remain otherwise similar.
Nakamura et al. [107] conducted systematic experi- y,rms(½L) = S,/2(fl)/[4~rSf~m2~1 ] 1/2, (2)
ments on in-line and staggered arrays in air-water
where S ( f l ) is the power-spectral-density of the exci-
mixtures, measuring the buffeting forces in drag and lift
tation force per unit length, and fl and ~'l are the
directions. Some very interesting observations were
cylinder first-mode frequency and damping ratio, re-
made; e.g., the drag coefficients, typically 2.5 times the
spectively; S ( f ) may be written in the form
corresponding lift ones, display power spectral densities
of distinctly different patterns in the slug- and bubbly- S ' / 2 ( f ) = ½pDU2CR, (3)
flow r~gimes.
Donaldson and McKnight [108] studied the develop- where CR(f) is the effective random excitation coeffi-
ment of turbulence and acoustic signals in a model cient. Based on the data collected by Pettigrew and
cross-flow heat exchanger. No correlation was found Gorman, C R should be read from either of the two lines
between turbulence and acoustic signals, until the cavity of fig. 3, corresponding to a frequency on the abscissa
fundamental, fa, is excited, whereupon they become equal to fl.
correlated and the sound-pressure level rises rapidly to Blevins's expression for the mid-span vibration am-
165 db. It is suggested that the signals are such that plitude is equally simple:
discrete vortex shedding is an unlikely forcing function
for the excitation of the acoustic resonance; rather, it is Yrms(½L ) = SiF/2(fn)J(an(½L)//[64qr3f3m2~n] ,/2 (4)
thought that, at a threshold flow velocity, the turbulent
whereSv (fn) is the power-spectral-density of the excita-
energy in the neighbourhood of f, becomes sufficient to
tion field per unit length, J is the joint acceptance which
initiate resonance and thence organize periodicity is equal to 1.0 for a perfectly correlated field, Q,(½L) is
(" vortex shedding") in the interstitial flow. the eigenfunction at mid-span; the other quantities are
Eisinger [109], among other things, discusses practi- the same as usual. Typical values of S v ( f ) were mea-
cal measures for avoiding acoustic resonances.
sured by Blevins et al. [102], and are given in dimension-
Finally, a series of studies has been published on the less form in fig. 4. To be able to easily utilize this figure,
speed of sound in passages containing arrays of rigid or
flexible cylinders, and on the pertinent acoustic natural
frequencies [ 110-113].
x 10 . 3

3.3. Buffeting response of cylinder arrays in cross flow." 50 [-UPl3er Li~mit for ~ '

state of knowledge and design guidelines .i. Upstream Cyli_n6ers

e4 20 "~
Turbulence-induced vibration, or buffeting, occurs at
all flow velocities and in every conceivable situation.
There is basically one analytical model for predicting a: Upper Limit for
vibration amplitude, based on random vibration theory; 5 Interior Cylinders "-
however, in developing this model and rendering it
practicable for simple use, two different forms have
emerged: one due to Pettigrew and Gorman [91,92,3] 2 I I ; I I
and another due to Blevins [93,102]. I0 20 30 ' 40 50 60 70
The Pettigrew and Gorman model, in its initial form, f(Hz)
recognizes (i) a finite correlation length for the exci- Fig. 3. The random excitation coefficient CR(/) for the calcu-
tation field, (ii) general boundary conditions , and (iii) lation of buffeting response of a cylinder in a cylinder array in
general response in all modes of the system. In its cross flow, according to Pettigrew and Gorman [eqs. (2) and
simplest form, however, it takes the field to be fully (3)1.
 M.P. PaMoussis / Flow-induced vibrations in reactors 39

10- 3 I I I I in fig. 4. This likely reflects differences in upstream

turbulence intensities in the two sets of experimental
data.
Pettigrew and G o r m a n also give equivalent design
o guides for two-phase flow [91,92].
It should finally be noted that in using eqs. ( 2 ) - ( 4 ) f ,
(or f l ) and fin (or ~'l) should be the values measured in
stationary fluid, rather than in vacuo.

10 - 7 3.4. Strouhal resonance in cylinder arrays in cross flow:

state of knowledge and design guidelines
10 - 8 I Although the work of the past three years' cannot be
oo o, o12 o13 oI, 05
fD/U said to have answered all outstanding questions by any
means, some solid progress has nevertheless been
Fig. 4. The lift coefficient C L for buffeting response of a
cylinder in a cylinder array in cross flow, according to Blevins, achieved.
Gibert and Villard [eqs. (4) and (5)]. It appears clear that Strouhal periodicity in the inter-
stitial flow within the first few rows of cylinder arrays
does commonly occur, unless the upstream turbulence
level is sufficiently high to suppress it. Deep within the
it is convenient to write
array, on the other hand, Strouhal periodicity may or
[ ( U / D ) SF( f ) ] '/2 = ½pDU2CL . (5) may not survive, depending on Reynolds number, cylin-
der layout geometry, fluid-to-cylinder mass ratio and
Hence, it is easy to see that the two expressions are very cylinder vibration amplitude. In all cases, however, if
similar, except for the nondimensionalizing factor U / D resonance with an acoustic mode should be established,
on both ordinate and abscissa of fig. 4. then periodicity is strong and omnipresent.
As this has not been done before, it was considered From the point of view of cylinder vibration, Strouhal
useful to compare the two expressions for two arbi- resonance--defined as amplified vibration, over and
trarily chosen cases, as shown in table 2. above the buffeting background, when fs coincides with
The agreement is remarkably close, considering that f , - - m a y occur
Pettigrew and G o r m a n determined C R from experi- (i) for tightly spaced arrays in gaseous flows, in the
ments in water, while Blevins et al. from experiments in first couple of rows--provided that upstream turbu-
air. This is not an extensive comparison, however, and lence is kept quite low;
no doubt cases may be found where the two are further (ii) for tightly spaced arrays in liquid flows, in the
apart. first few rows--unless turbulence levels are very high;
One important disagreement that should be noted is (iii) for widely spaced arrays, throughout the array;
that in fig. 3 the upstream cylinders are shown to be (iv) for any type of array if there should exist a triple
subjected to higher excitation, while the opposite is true coincidence of fs, fa and fn *
Clearly, these are tentative generalizations, since what
is "very high" or "quite low", "tight" or "loose" has
Table 2 not quantitatively been specified. Hence, until this is
Comparison of mid-span buffeting amplitudes of inner-row done, the designer is faced with having to design to
cylinders, assuming simply-supported ends and fully correlated avoid Strouhal resonances, "just in case".
excitation field, according to the Pettigrew & Gorman and If the Strouhal number is known fairly accurately,
Blevins, Gibert & Villard variants of the analytical model then one can design such that U / f , D be + 4 0 or + 50%
away from S - I . If S is not known with sufficient
Fluid D f. U (Yrms)PG accuracy, one may use Pettigrew's design guide [3,92]:
(Yrms)aGv noting that for typical cases 1/3 < S < 2/3, it is sug-
(mm) (Hz) (m/s)
gested that the design be such that U / f . D < 1, which is
Water 19 40 2.0 2.6
Air 25 60 50 0.5 * Under such circumstances the array becomes defunct
within hours [1].
40 M.P. Patdoussis / Flow-induced vibrations in reactors

sufficiently less than ( 2 / 3 ) - 1 ,,to take care of possible Furthermore, even in the simplest case of f~--fa,
'locking-in' of the wake." frequency coincidence is a necessary but not sufficient
As noted previously, however, the design criterion condition for acoustic resonance to occur. It is addition-
U / f , D < 1 may lead to impossibly over-conservative ally necessary for the strength of the excitation to be
designs [3,4]. Hence, a second step in the design proce- sufficiently high to overcome acoustic damping, which
dure is to calculate the vibration amplitude at the point is a function of system construction as well as of geome-
of the presumed resonance: it might be sufficiently try [85,115,116].
small that the cylinder array could live with it. Pettigrew Chen [79,82,94] has proposed empirical guidelines
[3] gives the very simple expression for the amplitude, for the occurrence of acoustic resonance for in-line
A y, in the lift direction arrays, which may be stated as follows:
(i) resonance may occur if U > U~, where Uc is calcu-
ay(X) dp,,(x) foLF(x,)dp,,(x,)dx '' lated such that f~->fa through Uc =faD~S;
8~rZf~, (ii) resonance will occur if U > Uc and additionally

where F ( x ) is the correlated periodic wake shedding =- ( R e / S ) [ 1 - 1/s,]2(1/st) > 600 or 2000,
force along the cylinder, which may be expressed as
where 600 applies to laboratory cases and 2 000 to real
F(x)= ½oDU2(x)Q, heat exchangers (where acoustic damping is likely to be
higher).
C e being the lift coefficient. It may be seen that, espe-
A simpler criterion has been proposed by Bryce et al.
cially for gases, where 0 is small, F ( x ) may also be quite
[97], stating that resonance will occur only if
small. Unfortunately values of CL are not sufficiently
well documented, although Pettigrew et al. [90-92,3] U/f~D > 2 ( s , - 0.5).
and Y.N. Chen [82,100] do give some values. It is noted
that C L is generally smaller than 0.10 and can be as low Ulrich, more recently, has proposed another criterion
as 0.01 for p / D < 1.5 [3,4,92], while it can take much [1161.
larger values for more widely spaced arrays, where s,, Unfortunately no cross-comparisons of the relative
s t - 0 ( 3 ) or 0(4) [82,100]. These lift coefficients (and effectiveness of these criteria have been published, and
the span-wise correlation integrated into them) are likely the designer is left unaided in this respect [117]. In cases
to be strongly dependent on vibration amplitude; how- where acoustic resonance appears to be unavoidable,
ever, there is insufficient information at present to be then the use of baffles to ~icoustically detune the system
able to take this factor into account. Pettigrew [3] seems the most effective design strategy [96,109];
recommends that the maximum vibration amplitude be Zdravkovich and Nuttall's suggestion [114] of (i) using
kept below 0.02D. unequal longitudinal pitches in successive rows and (ii)
Finally, the use of upstream turbulators is recom- removing some of the tubes in the array, to detune the
mended, wherever feasible, at least for tightly spaced excitation, should also be mentioned.
arrays, so that Strouhal resonances may be suppressed Finally, it should be reiterated that fa will vary with
altogether [103,105,106]. However, for more widely the geometric density of cylinders in the array through
spaced arrays, it is unlikely that the same effect may be variations in the sonic speed [97,110-113], and this
achieved by such turbulators. should be taken into account.

3.5. Acoustic resonances in cylinder arrays in cross flow:

state of knowledge and design guidelines 4. Cylinder arrays in cross flow. !I. Fluidelastic insta-
bilities
As stated previously, if fs should coincide with fa, it
is possible to excite acoustic resonance in the system. This section deals with self-excited vibrations of
However, there exists evidence suggesting that reso- cylinder arrays in cross flow, otherwise known as
nance may also occur if fa = 2fs, where the Strouhal fluidelastic instabilities.
frequency is captured by an acoustic one--vide, e.g.,
[114]. Moreover, it has also been found [97,85] that 4.1. Fluidelastic instabilities: developments to circa 1980
resonance may be initiated with 0.4 <fs/fa < 0.8, de-
pending on the number of rows in the array. Hence, the In 1962 Roberts uncovered the possibility of self-ex-
problem is by no means simple. cited oscillations of a staggered row of cylinders in cross
 M.P. Paidoussis / Flow-induced vibrations in reactors 41

flow, where only alternate cylinders were free to move sufficient energy may be extracted from the flow, inde-
in the in-flow direction [118,119]. The mechanism pro- pendently of jet switching, to overcome damping and
posed to explain the phenomenon was a Coanda-like sustain the vibration. For his particular single-row array
"jet switching", where the jet pairing between adjacent and an assumed intercylinder modal pattern, Connors
cylinders switches direction in synchronism with cylin- obtained the following relationship for the critical flow
der motions, in such a manner that energy is extracted velocity, U~:
by the cylinders from the flow, in the course of a cycle
of oscillation; if this energy is sufficiently high to offset UJf, D = 9.9( m $/pD 2)'/z. (6)
dissipation, then motion is amplified to a self-sustaining
limit cycle. Roberts developed a complex semi-empirical While Roberts' work was largely ignored, Connors'
analytical model for predicting critical flow velocity for equation became widely accepted. Moreover, unfor-
the onset of instability. Moreover, he suggested that the tunately, eq. (6) was subsequently misused to calculate
proposed mechanism may only occur if U/f,D > 12 Uc for systems for which it was never intended: notably,
approximately, which would allow sufficient time for jet for multi-row cylinder arrays, for which it has subse-
switching to actually take place. quently been shown to be non-conservative.
In 1970 Connors studied the problem of a single row An interesting comparison of Connors' and Roberts'
of cylinders, all of which were free to vibrate [120]. theoretical UJf, D and their own and others' experi-
Drawing on Roberts' work, he proposed a quasi-static mental data [119-121] was later undertaken [1] and is
semi-empirical analytical model to explain the observed reproduced here in fig. 5. It appears that, irrespective of
self-excited oscillations. In this model the displacement whether UJfnD is greater or less than 12, the two
of cylinders with respect to their neighbours results in theories are relatively close to each other; more im-
modified forces coming into being vis-h-vis those of the portantly, all the experimental results are self-consistent
original configuration; for a given pattern of motions, and agree fairly well with both theories, although those

I000 i i i

Experimental Data
• Roberts 1966
Uc u Connors 1970 ..
o Halle a Lawrence 1977 /"
fn D ,, Hartlen 1974 ../"
• Ishigai 1975- 79 . 4 ~ ¢-
I00

• Roberts Theory
I0 /

°c '
./" L Connors Theory . ,, - - 9 . 9 ( ,, )
, / o 8 ° oO tnU pD"

I ± L J

i0 -I i0 ° i01 i0 2 i0 a
m(~
pD 2
Fig. 5. The threshold of fluidelastic instability of a single row of cylinders in cross flow.
42 M.P. Pai'doussis / Flow-induced vibrations in reactors

obtain suitable values of K, applicable to cylinder arrays

of different geometries, different fluids, etc. [121,
Yrms 127-137, 8 6 - 9 2 ] - - w i t h o u t wishing to imply that this
was the only, or even the principal, concern of most of
these papers. However, it was found impossible to de-
Fluidelastie Instability fine a unique K for all systems. Soper attempted to
define K = K ( s t, s g ) ; although his own results were
self-consistent, nevertheless large discrepancies re-
Response to Flow mained when comparing with other researchers' data
N

PeroodocoIty
. . .

[1351.
Hence, the only sensible thing to do seemed to be to
define a minimum value of K, K m i n , such that all
occurrences of fluidelastic instability would occur at
higher Uc/fnD than those computed with eq. (7) and
K = K m i n. Connors, from work on square-geometry
arrays, proposed [131] K = 0.37 + 1.76 s t for 1.41 _< st _<
2.12, yielding Kmi n = 2.9. Soper's experiments [135] sug-
gest K m i n = 2.0 for rotated (parallel) triangular arrays,
and higher values for other geometries. However, it was
the value of K m i n = 3.3 proposed by Pettigrew et al.
Fig. 6. Idealized response of a cylinder in an array subjected to [90-92,3] which gained most widespread acceptance:
c r o s s flow. indeed, when tested against actual heat exchangers which
developed, or not, fluidelastic instabilities, it was shown
to stand up very well [90-92]. However, it should be
for small U¢/fnD agree much better with eq. (6). This noted that important disagreements have arisen as to
suggests that, basically, the same type of phenomenon exactly how the various terms in eq. (7) should be
occurs in all the experiments, but that it was explained interpreted (section 4.2).
in terms of different mechanisms by Roberts and Con- Some more fundamental work was undertaken by
nors. Zdravkovich and his colleagues [138-140,98,99] on
Connors' model was mathematically formalized by single- and multi-row arrays in airflow, measuring (i)
Blevins [124] and extended, conceptually at least, to pressure distributions about the cylinders, thereby deriv-
deal with cylinder arrays [125], as well as to include the ing force coefficients in the lift and drag directions (in
effect of changing damping with flow [126] **. How- both original and displaced geometries), (ii) flow veloc-
ever, the general form of eq. (6) survived, except that it ity distributions in the interstitial passages, (iii) vibra-
was recognized that the constant 9.9 had to be different tion amplitudes, and (iv) patterns of inter-cylinder mo-
for cylinder arrays; i.e., the criterion for instability tions through high-speed cinematography. The observed
should have the form patterns of motion were found to be well correlated
with those suggested by the forces measured in statically
2 l/2
uc/U,D = K ( m S / p D ) . (7) " d e f o r m e d " array configurations, corresponding to dif-
ferent parts of a cycle of oscillation. Moreover, it was
The general response of a cylinder in an array sub-
found that (a) systems with only one flexible cylinder
jected to cross flow is typically that shown in fig. 6,
display relatively small vibration amplitude, the ampli-
where response to flow periodicity ("vortex shedding")
tude increasing with the number of flexible cylinders in
may or may not develop, as pointed out in section 3.
the array, being a maximum when all cylinders are
Fluidelastic instability, on the other hand, apparently
flexible; (b) turbulators, with axes normal to the cylin-
always develops and it is the most serious and damaging
der axes greatly affect the fluidelastic behaviour of the
phenomenon to contend with [1]; yet its prediction
system and delay the onset of instability; (c) cylinders
remained rather uncertain.
in the second row are most susceptible to oscillation.
A great deal of experimental data was gathered to
However, such fundamental studies were quite rare.
But with very few exceptions, eq. (7) was considered to
** This inclusion is controversial, and prominent authorities be valid for any system, if only the magic K could be
in the field dispute its validity, on fundamental grounds. d e t e r m i n e d - - a view tenaciously held, even in the face
 M.P. PaMoussis / Flow-induced vibrations in reactors 43

of accumulating contrary evidence. It is important to remained constant, suggests that UJfnD cc ( m / p O 2 ) 0"3.
recognize here that all fluid-mechanical aspects of the Hence, by 1977/78 it became obvious that all was
instability are hidden within K. As shown by Heinecke not well with fluidelastic instability modelling. Yet, in
[141], for a one-degree-of-freedom system excited the absence of a better predictive tool, designers had
aerodynamically by a force of the type ½CpU 2 sin ~2t, little choice but to use some form of eq. (7) for predic-
energy considerations will inevitably lead to resonant ting U~.
conditions when an equation of the form of (7) is Two attempts were made to devise new analytical
satisfied: for the energy input per cycle of oscillation is models, in this period, by Balsa [147] and S.S. Chen
proportional to pU2/m and energy dissipation per cycle [148]. Faced with the problem of dealing with the sep-
is proportional to f f s ; hence, balance between these arated flow in the arrays, they both considered the flow
two energies is obtained when ( U / f n ) o z ( m S / p ) 1/2, to be irrotational and modelled it using potential flow
which leads directly to eq. (7). theory; thus, the presence of wakes was totally ignored.
In this light, it is not surprising that some purely Balsa's theory is quite involved, utilizing the" method
empirical design guides for avoiding excessive vibration of matched asymptotic expansions; in the end it results
in heat exchangers [142-144], developed independently in an expression for divergence instability, which, in the
of the aforementioned work, can be re-cast in the form notation used here, may be written as U J f , D =
of eq. (7), as pointed out in refs. [1,4]. However, the C ( m / o D 2 ) 1/2, where C is a function of the characteris-
system of cylinders in an array is generally a multi-de- tic gap between cylinders and an "interaction parame-
gree-of-freedom system; also, the exciting force is known ter". The absence of ~ is noteworthy: as the flow field is
to be displacement- and velocity-dependent. Hence, it is totally conservative, the system must lose stability by
by no means obvious that the simple form of eq. (7) is divergence, which is independent of velocity-dependent
necessarily valid. forces.
Indeed, towards the end of the 1970's solid experi- Chen [148] formulates the system of cylinder-to-cyl-
mental evidence began to be gathered against the Con- inder interactions and obtains expressions for coupled
nors-Blevins semi-analytical model and the simple form inertial, damping and stiffness (displacement-depen-
of eq. (7) for predicting Uc, as follows. dent) terms. When these are evaluated by potential flow
(i) Southworth and Zdravkovich [138] and Weaver theory, the results are shown to compare poorly with
and Grover [145] have shown that even if all cylinders experiments. This work, however, is important in view
but one are held rigidly immobile, then that one be- of later developments, as will be seen in section 4.4.
comes unstable at the prescribed flow, contrary to Con-
nors' theory which requires a certain pattern of inter- 4.2. On the utilization of eq. (7)
cylinder displacements.
(ii) Gorman [130] showed that in arrays of mixed In addition to questions about its validity, questions
brass and stainless steel tubes, the instability develops also arose on the interpretation of most of the terms of
for the more flexible brass tubes, while the others re- eq. (7).
main immobile, once again contrary to the It was generally accepted that fn should be the
Connors-Blevins theoretical model. frequency of oscillation in the quiescent fluid medium.
(iii) Whereas theory predicts that detuning the cylin- However, for multi-span cylinders there exist groups of
ders should result in a monotonic stabilizing effect, at natural frequencies with the same mode shape per span
least for a single row [124], Weaver and Lever [146] (vide section 5, [90,104]). If the flow velocity is non-uni-
found that this effect disappears if detuning is greater form over the total multi-span length of the cylinder, it
than a small percentage of the basic frequency. is not always the mode with the lowest natural frequency
(iv) Experiments by Weaver and Grover [145], in that will be excited with minimum U [90]. Hence, eq. (7)
which 8 was varied whilst the other parameters were should be applied to all modes of the multiply sup-
kept constant, suggest that UJf~D cx 8°.2~; it was pro- ported structure, taking account of non-uniform flow
posed that, for an expression of the form of eq. (7), the velocity distribution U(x) by calculating an effective U,
exponents on the two dimensionless numbers, m / p D 2 as follows [129,18]:
and 6, need not be the same.
(v) Evidence by Nicolet et al. [87] points to Uc/fnD
cc ~0.1 to 8 °3.
(vi) Other experimental evidence by Nicolet and
Gibert et al. [86,87], in which p was varied whilst where L is the total length.
44 M.P. PaMoussis/ Flow-induced vibrations in reactors"

For closely packed arrays in dense fluids there exist 4.3. Recent, mainly experimental studies on fluidelastic
bands of frequencies, containing many first-mode-shape instability
natural frequencies. Normally, the lowest one within
this band is utilized, and methods for calculating it are Blevins et al. [102] conducted an extensive experi-
reviewed in section 5. It has to be recognized that in mental programme. They confirmed that a single flexi-
such situations the hydrodynamic or "added" mass may ble cylinder among rigid ones in an array does become
be considerably higher than that of a solitary cylinder in unstable. Detuning effects were studied and it was
unconfined fluid--which is equal to pA. Here, there is found that progressive detuning of whole rows of cylin-
divergence of practice: some do utilize this higher added ders is most effective; however, detuning was found to
mass, whilst others use the one pertaining to a solitary be less effective in stabilizing the system than predicted
cylinder ( = oA), which can be several times smaller. by existing theory. Among other things, it is reported
U is usually defined as the reference gap velocity. that in these tests (i) Kmi n = 2.5, although for slightly
Some researchers, however, prefer to use to so-called spiralling arrays K = 1.7; (ii) there is no discernible
minimum-gap velocity, which is the velocity at the pattern of inter-cylinder motions.
minimum diagonal space between staggered arrays, and Gibert et al. [101] report on a massive experimental
which for certain layout geometries may be considerably programme with arrays of different geometries and
higher than the reference gap velocity. pitch-to-diameter ratios, in air and water flows. Among
The question of which 8 to use is perhaps the most other things, it was found that (i) for m S / p D 2 > 20 and
serious. Some researchers use 8 in vacuo (approximately 20 > m S / p D z > 7, the experimental data suggest that
measued in still air), some in the pertinent quiescent Uc/fnD oc ( m S / p O 2 ) 1/2 and Uc/fnD ec ( m S / p D 2 ) °.3,
fluid medium, while others [126] the predicted 8 that respectively; (ii) for m S / o D z < 7, there is interference
might exist under flow and vibration conditions--de- with Strouhal lock-in for very lightly damped cylinders;
spite the fact that then 8 = 0 at U = U~. The values of 8, (iii) the total set of their data suggest a relationship of
according to these three interpretations, may differ from the form UJfnD = 2.7(mS/pD2) 034.
one another by as much as one order of magnitude. Another extensive experimental study was conducted
Here, the difficulty of measuring 8 in liquid should be by Heilker and Vincent [149] for various types of arrays
noted: if a cylinder is excited in quiescent liquid, then, in water and air-water mixtures. It is found that
through the effects of hydrodynamic coupling, energy UJfnD = K ( m S / p D 2 ) 1/2 approximately, with K m i n =
"leaks away" to excite the surrounding cylinders, giving 5.0. In the presentation of the results, however, the
false estimates of 8. The matter is usually resolved by values of 8 measured in the flowing fluid are used; at
rigidly clamping all surrounding cylinders before mea- least for U close to Uc, these values of 8 would be lower
suring 8, although it is uncertain what exactly this value than 8 in quiescent fluid. Hence, had the latter been
of 8 represents. used, then a lower Kmi n would have been obtained.
The value of m is usually taken to be the effective Chen and Jendrzejczyk [150] studied the evolution of
mass, including the "added" or hydrodynamic measured 8 with increasing U and report that (i) ini-
mass--vide section 5. Here, however, there is the same tially, damping increases with U; (ii) it then displays a
divergence of practice referred to in the paragraph dip associated with Strouhal resonance, and subse-
discussing fn, as to which added mass is actually used. quently increases again; (iii) as U~ is approached, 8
The interpretation of D has remained unchallenged, decreases rapidly, suggesting that it will eventually be-
so far: it is the outside diameter of the cylinder! come zero, at which point structural modal damping
Finally, from the point of view of interpreting ex- would be equal to the negative flow-induced damping.
perimental results, the onset of instability, and hence The same authors have also presented an extensive
U~, is defined in different ways. Some define it in terms set of experimental data [151] for twelve different arrays
of a threshold amplitude, when it exceeds a certain of slightly varying spacings and different geometries,
fraction of the diameter; others from the intersection of mass ratios, damping and progressive row-by-row de-
the tangent to the fast-rising portion of fig. 6 with the tuning, in water flow. It was found that (i) detuning
buffeting response tangent; others [145] in terms of the stabilizes the system, which agrees with previous work
onset of coherent sinusoidal oscillation; others still, [138,146,124], albeit to a lesser degree than previously
from the clanging sound of cylinders impacting. suggested [146,124]; (ii) cross-grid-type upstream turbu-
lators stabilize the system, whilst grids parallel to the
cylinder axes destabilize it, where both stabilizing [139]
and destabilizing [129] effects had previously been re-
 M.P. Pai'doussis / Flow-induced vibrations in reactors 45

ported; (iii) Strouhal resonance may interact with and the cylinders first become unstable in one of the lowest
reinforce fluidelastic instability; (iv) the amplitude- frequencies in the band of coupled frequencies (section
velocity graphs (fig. 6) display multiple Uc's, which are 5), and then progressively higher-frequency modes be-
attributed to different coupled modes of the system; (v) come involved.
the most critical cylinders in this study lie in the first Heinecke and Mohr [157] studied the response of
row. Finally, it is suggested that the results obtained in-line arrays ( s t = s ¢ = 2 . 8 8 , 2.30, 1.44, 1.15) in air.
would indicate that UJfnD = 2.49( m S / pD2) °'52. Some of the cylinders were instrumented around their
Weaver and co-workers have studied a series of circumference, to yield unsteady pressure distributions
specialized questions relating to fluidelastic instabilities about the cylinders and static and dynamic force coeffi-
[152-156]. It was found [ 152] that the m i n i m u m number cients; the corresponding vibration amplitudes were also
of cylinder rows necessary to study fluidelastic instabili- measured. From these measurements it is concluded
ties is s i x - - a t least for parallel triangular arrays and that (a) for the s t = 2.88 and 2.30 arrays, the dominant
p / D = 1.375 in air. It was also found that the instability vibration-inducing mechanisms, at least in the flow-
first develops in the third and fourth rows, and that the velocity range studied, are Strouhal resonance and buf-
presence of downstream rows does not significantly feting, and (b) only for the smaller two spacings do
affect the instability threshold. fluidelastic instabilities take place. Moreover, two dis-
Weaver and E1-Kashlan [153] extended previous work tinct types of instability are suggested: (i) a "galloping"
in which 8 was varied independently [145] and also instability, for the s t = 1.44 array, which is akin to
varied m / p D 2 while keeping 8 constant, finding that wake-galloping of transmission lines, and which may
occur even if only one cylinder is flexible among rigid
U J f n D = C( m / p D 2 )0"29~ 0"21 ones; (ii) an aerodynamic-stiffness-controlled mecha-
nism, termed "fluidelastically coupled instability",
is more appropriate than eq. (7). The 0.21 exponent which occurs for the s t = 1.15 array, and which develops
agrees with that of previous studies [145], while only if all cylinders (or at least a sufficiently large
( m / p D 2 ) °29 agrees with the work of refs. [86,87], and group) are free to vibrate.
to some extent with that of ref. [101]. Rrmy [158] studied an in-line array (s t = s e= 1.44)
Yeung and Weaver [154] studied the effect of ap- in air and water, as well as in air-water mixtures.
proach-velocity direction on triangular arrays with p / D Contrary to others' findings [90,91,137,149], no fluide-
= 1.5, in water. The flow direction was varied through lastic instabilities in two-phase flow were found, be-
an angle 0 ° < a < 30 °, where a = 0 ° corresponds to a cause "the existence of two phases limits.., coupling
parallel triangular array and a = 30 ° to a normal trian- between the tubes"; instead, vibration amplitudes were
gular array. It was found that as a is increased from 0% found to increase with flow to high levels, but with no
U J f n D rises slowly, up to a = 20% beyond which there distinct instability threshold. Hence, this is perhaps a
is an upwards j u m p to the UJfnD value associated with question of definition of fluidelastic instability (vide
a = 30 °, pointing to the importance of the interstitial section 4.2): if its occurrence had been defined in terms
flow pattern in the array. of exceeding a certain vibration level, then R~my's
Weaver and Koroyannakis [155] studied the effect of conclusion might have been different.
different ("asymmetric") stiffnesses in the in-flow and Fujita et al. [159] and Connors [160] study the effect
cross-flow directions--partly in order to simulate the on fluidelastic instability of two unusual flow condi-
situation in the U-bend region of heat exchangers. These tions: a thin two-dimensional jet impinging on
experiments, with parallel triangular arrays ( p / D = multiply-supported cylindrical arrays [159] and a "cor-
1.375) in water flow, show that U J f , D may be up to ridor" flow on the outer edge of the array [160], in both
20% higher for asymmetric stiffnesses as compared to cases the cross flow eventually diffusing into the array.
the symmetric case, this effect being independent of the G o r m a n [104,105] and Halle et al. [161,162] examine
orientation in which the system is stiffer. Whirling, the more realistic (for heat exchangers) configuration of
diagonal and figure-8 motions of the cylinders were multi-span arrays under multi-pass flow conditions. It
observed. was found that fluidelastic instabilities occur at higher
The same authors also studied the geometrically Uc/fnD than in the equivalent single-span system; it is
identical parallel triangular array ( p / D = 1.375) both in concluded that eq. (7) with gmi n = 3.3 is a good con-
air and water flow [156]. It was found that the be- servative criterion.
haviour of arrays in water is much less "regular" than in Finally, a cross section of the very real problems
air, due to hydrodynamic effects--similarly to ref. [151]; faced by designers and experimenters working on large
46 M.P. Patdoussis / Flow-induced vibrations in reactors

facilities, that researchers should be aware of, are high- where CxxJ and C~yj are the aerodynamic force coeffi-
lighted in refs. [158,163-165]. Bai and R+my [158, cients on cylinder i in the x-direction due to vibration of
163,164] discuss a large number of experimental data c y l i n d e r j in the x- and y-directions, respectively; x 1 and
obtained in large scale facilities. Reflecting the formida- yj are the amplitudes of motion of the two directions.
ble difficulties in determining 8 therein, they propose a Similar expressions obtain for the other forces acting on
design criterion of the form the five cylinders. The C's are complex numbers and
, 2 1/2 functions of the reduced velocity of motion, U/fD.
Uc/fnD=K ( m / p D ) , Hence, they contain acceleration-, velocity- and dis-
where K ' = K8 I/2. Thus, in 1981, we come full circle placement-dependent components: i.e., added-mass,
back to Thorngren's criterion for avoiding excessive aerodynamic damping and aerodynamic stiffness com-
vibration in heat exchangers [143], except that now ponents. Solving the resultant matrix equation yields the
K ' i n = 0.70, as compared to Thorngren's K ' = complex eigenfrequencies of the system, from which the
2.76[p/D-1]1/2; taking p / D = 1.31 to 1.44, which stability may be determined.
covers the range of the B a i - R 6 m y experiments, values The force coefficients were measured experimentally
of 1.54 and 1.83, respectively, are obtained for K', and by forcing each cylinder in the five-cylinder kernel to
it is easily seen that the Thorngren criterion is non-con- vibrate in water, over a range of U/fD, for the specific
servative. Indeed, as previously shown [1,4] most of in-line arrangement of s t = s e = 1.33. Using these in the
these older design criteria [142-144] are non-conserva- analysis yields values of Uc/fnD in excellent agreement
tive. with experimental values, both in air and water [167,
168]. Further measurements of the C's for other s t and
4.4. Analytical developments on fluidelastic instabilities:
s , yields, once more, excellent agreement with experi-
1980-82
ment [169]. For the s t = s g = 1.33 in-line array, it was
The theoretical development of the subject has been found [168] that
both rapid and exciting.
Whiston and Thomas [166] re-examine Blevin's anal- Uc/fnD oc ( m / p D 2 )1/261/2 for low-density flow, and
ysis of a single row of cylinders in cross flow and, using
his data and geometry, extend it to arrays, considering Uc/fnD cc ( m / p D 2 )1/361/5 for high-density flow
only position-dependent forces. The single-row analysis
is extended to deal with patterns of motion of wave- which agrees with others' experimental observations
length 2~, other than the Connors-Blevins ), = 4 T (i.e. [145,153,86,87]--whilst for s t = s ~ = 2.0 [169] UJfnD oc
such that the pattern periodically repeats itself every ( m 8/pD 2 )3/4. Moreover, as s t, s~eare increased, Uc/fnD
four cylinder pitches); it is found that )~ = 4T is the least becomes larger, but beyond s t = s z= 2.0 it remains al-
stable pattern. The effects of frequency detuning are most constant. Different types of detuning were found
discussed, and it is found that Blevins' large detuning to variously stabilize and even to destabilize the system.
effect applies only to small 8; for large 8 it is very small. Chen [170,171] developed further his previous model
The model is successfully extended to staggered cylinder [148] in the light of Tanaka and Takahara's work
arrays, by considering a union-jack kernel of nine cylin- [t67,168]. Instead of considering very large arrays, now
ders and recognizing that the interstitial flow velocity in five-cylinder sub-systems were examined and, when it
the minimum gaps (taking account of the wakes) may came to numerical calculations, Tanaka and Takahara's
be larger than the maximum in the gap between two measured force coefficients were utilized, rather than
cylinders in the same row; the extension to in-line those previously obtained by potential flow theory. Some
arrays is more tenuous. In all cases a relationship of the remarkable insights were obtained thereby. Two distinct
form of eq. (7) is obtained, with Kmi n = 2.8 for arrays. mechanisms are distinguished:
The discussion on mechanical coupling and detuning (i) "Fluid-damping-controlled" instability, induced
effects is extensive, interesting and illuminating. by negative aerodynamic damping forces, which may
In 1980, Tanaka and Takahara published a remarka- occur event if one cylinder is flexible among rigid
ble paper [167]. Considering a kernel of five cylinders of o n e s - - c f . [157]; this mechanism is dominant for low
an in-line array, a full unsteady aerodynamic force m S / p D 2, i.e. for high density flows. For the pure form
formulation was utilized; e.g., the force on cylinder i in of this instability, Uo/fnD = K*(m6/pD2), where K* is
the x-direction is given by a function of U/fD, and f , , m and 6 are all ap-
5
propriately evaluated either in vacuum or in the quies-
cent fluid medium.
F~i = ½pV z ~_, { C:,~jxj + CxvjY,),
j=l
 M.P. PaMoussis / Flow-induced vibrations in reactors 47

(ii) "Fluidelastic-stiffness-controlled' instability, large m 8 / o D 2 the effective exponent tends to ½. Clearly

which is dependent on coupling and phase differences these agree with Chen's findings [170,171]. The theoreti-
in the motions of neighbouring cylinders; this mecha- cal results were found to be in reasonably good agree-
nism is dominant for high m S / p D 2, i.e. for gaseous ment with the experimental data of Weaver and his
flows. For the pure form of this instability, UJfnD oz co-workers [145,153]. The effect of various types of
(mS//pDX) I/2. In general, both mechanisms will be detuning and mehanical coupling were also studied, and
implicated and the instability criterion will have the it was found that both effects depend strongly on
functional form m / p D 2, 8 and the specific form of coupling and detun-
ing utilized: although usually producing a stabilizing
Uc/fnD = F( m S / o D2, m / p D 2, S t , S a, effect, they may also lead to destabilization of the
system.
turbulence characteristics). Lever and Weaver [176] developed a very different
In addition to oscillatory instabilities, divergence is also and promising, truly analytical model, based on a num-
predicted to occur. ber of reasonable assumptions and requiring virtually
Calculated Uc/fnD versus mS / pD 2 diagrams exhibit no input from measurements. The theory is based on
the jumps found by Tanaka and Takahara, and are in the observation that, both in air and water cross flows, a
good agreement with their and some others' experimen- single flexible cylinder among rigid ones does become
tal results. unstable at essentially the same U¢ as an all-flexible
The effects of some parameters are also clarified array [138,145,102,156]. A model is then developed con-
[171]. Thus, since the fluid-damping-controlled mecha- sidering the motions of just one cylinder, surrounded by
nism depends basically on the motion of one cylinder, it immobile neighbours. The forces leading to destabiliza-
is insensitive to detuning, while in the stiffness-con- tion are considered to arise from flow redistribution in
trolled mechanism (large m S / p D 2) it may be im- the "channel" formed by the dividing streamlines of
portant. On the effect of m / p D 2 and 8, Chen finds that, sub-channel flows, where the flow redistribution is
for the cases investigated, they are complementary, such caused by cylinder motion, with an appropriate phase
that effectively they may be dealt with as effects of lag; thus, both velocity- and displacement-dependent
m S / p D 2 over narrow ranges of that parameter. Thus, forces are part of the model.
for the specific in-line array analyzed (s t = sg = 1.33), it This model was applied to a parallel triangular array
is found that ( p / D = 1.375). The Ue/fnD versus m S / p D 2 diagram
exhibits several instability bounds, depending on the
- 2.0 18
for0.01 < m S / p D 2 <0.5: U J f n D = 3 . 4 ( m S / p D ) " ; various values the aforementioned phase lag may take.
Below m S / p D 2--- 3 there exist several instability lines,
for 0.5 < m S / p D 2 < 3.5: uJynD = 4.0(mS/pD2)°41; giving rise to jumps, similar to those found by others
[167-171]. Agreement with experiment is remarkably
for 3.5 < mS/pD2: UJfnD = 7.6(mS/pD2)°42; good, especially for high m S / p D 2.
where m, 8 and fn here are the values in vacuum.
Price and Paidoussis [173,174] developed a quasi- 4.5. Fluidelastic instability: state of knowledge and design
static model for a staggered double row of cylinders, guidelines
which nevertheless incorporates velocity-dependent
terms in the manner successfully utilized by Den Hartog Although it cannot yet be said that the fluid mecha-
for galloping oscillations [175]. The force coefficients nics underlying fluidelastic instability are well under-
associated with a kernel of three cylinders are measured stood, the new set of analytical and semi-analytical
statically--a much simpler proposition to Tanaka and studies [166-176] and extensive experimental investiga-
Takahara's measurements. The critical flow velocity for tions [101,102,134,149-158] have begun to shed fight
fluidelastic instability is found to have the form and impose order in what appeared to be, circa 1980, a
rather chaotic state of affairs.
U ~ / f n D = A ( I + [1 + B ( m S / p D 2 ) ] ' / 2 ) , There is a great deal yet to be clarified, however. For
example: is it true that there are two distinct mecha-
where A and B are functions of m / p D 2, s t and s,e. Thus, nisms at work? As contended by Chen [170,171] and by
for small m S / p D 2, the effective exponent in an equa- Heinecke and Mohr [157], on the one hand, based on
tion of the form (7) is quite small, i.e. system stability is their work with in-line arrays, the "fluid-damping-con-
not very sensitive to mS/pD2; on the other hand, for trolled" or "galloping" instability is operative at low
48 M . P . Paidoussis / Flow- induced vibrations in reactors

rn~/pD 2 a n d / o r small s t, s e, whilst the "fluidelastic- plex force coefficients for a given array, then excellent
stiffness-controlled" instability at high m 6 / p D 2 a n d / o r prediction of U~ in that array is possible. However, it
larger st, s z. Lever and Weaver [176], on the other hand, should be recognized that, going this route, it would be
suggest that there is but one mechanism at work, mod- necessary to undertake these extensive measurements
ified by a phase lag between cylinder motion and flow for each and every array geometry one is interested in,
redistribution (and hence modified forces on the cylin- over the pertinent range of U/fD. Clearly, the question
der), which effectively gives different stability bounds must be asked if, under the circumstances, it would not
for low and high m S / p D 2. The allied question that be simpler to measure UJfnD directly and by-pass the
must be clarified is whether a single flexible cylinder use of the model altogether. Price and Paidoussis'
among rigid ones can become fluid-elastically unstable [173,174] approach involves a less onerous measurement
for high m 6 / p D 2 in all cases; is the divergence of views task, but has basically the same drawback; moreover,
between Chen and Heinecke and Mohr, on the one their analysis may be inappropriate for very low U j f n D ,
hand, and Lever and Weaver, on the other, related to where the quasi-static assumption breaks down. In this
the different geometries investigated, both theoretically respect, Lever and Weaver's proposition [176] looks
and experimentally? promising, for it requires minimal empirical input and
Tanaka et al. [167-169] and Chen [170,171] have requires no force coefficient measurements.
shown that, if one is prepared to measure all the com- Matters should be clarified in the next few years,

I I

® F~nk~n a $ o p e r 1977
I000
El Hartlen i~?,~
Nicolet et al. 1976
s Gibert e i al. 1977
UC
PettJgrew ~t al. 1975
fn D Weaver s Srover 1978
ZdravkoviCh 1976
I00 Heilker~~_nt 198! ' _~. . . . . . . .
Tanaka et, al. ]-980-82 []
V Gibert et al. 1981 . an [] ~ i x n .
W e a v e r et al. 1 9 8 1 - 8 2 t i I:t ~ ~x J ~ " f ' ~
® Chen a dendrzejczyk Igl~ll ] ~ ~.'r'~ u Lpetti~rew
Zukauska,, s KatJnas 19~',0 x ~.~-4~ et al ~

10

XX
X
e /o
10-2 I0 -I i0 ° i01 i0 2 i0 3
m3
pD z
Fig. 7. Experimental values, from various sources, for the threshold of fluidelastic instability of cylinder arrays in cross flow, plotted in
the conventional manner; also shown is Pettigrew's et al. empirical guide.
 M.P. PaMoussis / Flow-induced vibrations in reactors 49

but, until they do, designers may have to continue for Strouhal resonance and fluidelastic instability, the
utilizing whatever semi-empirical guidelines they have lowest fn would be the one excited with the minimum
been using, provided that the parameters in their new possible U.
designs do not stray too far away from those of tested However, this need not always be the case. For
designs. In an attempt to help matters in this way, most example, in cases of high impingement velocity over one
of the available experimental data have been plotted as half the cylinder span, the mechanical receptance (ad-
shown in fig. 7, despite the fact that this data suffers mittance) of the second mode is higher for this loading,
from the various interpretational difficulties discussed and hence the system may develop vibration or instabil-
in section 4.2, which are doubtlessly partly responsible ity mostly, or exclusively, in that mode. Such considera-
for the large scatter. Other factors contributing to the tions are especially pertinent to multiply supported cyl-
scatter are (i) differing layout geometries and spacings, inders, a common arrangement in heat exchangers and
(ii) different degrees of tuning of the cylinders, (iii) in nuclear reactors. In such cases it is generally im-
mechanical coupling effects, (iv) upstream turbulence proper to consider each span in isolation; rather, the
and (v) Reynolds numbers involved, all of which, it is natural frequencies of the whole system should be con-
now well established, affect the stability of the system, sidered.
in addition to the parameters involved in fig. 7. For a cylinder supported at its extremities and at an
Despite all the above, it is clear from this figure that, intermediate point, for example, there will be two natu-
as suggested by the work of refs. [170-176], eq. (7) ral frequencies, both with a first-beam-mode shape
cannot possibly apply over the whole range of m / p D 2 within each of the two spans. The overall mode shapes
and m S / p D 2 : the line shown is simply too steep, for will be different, however; in one mode the two "humps"
one thing. Thus, for low m S / p D 2, the slope is almost will be on the same side, while in the other on opposite
horizontal; for 2 X 10 - l < m S / p D 2 < 2 X 101, roughly, sides. For a larger number of spans this number of
the slope is nearer 1/3 than 1/2, gradually rising with distinct natural frequencies multiplies factorially. There
increasing m S / p D 2. One could draw a curved lower- exists, of course, a similar number of modes with higher
bound envelope through the lowest data points in the mode shape within each span. The reader is referred to
figure and utilize it as a new design guideline. However, refs. [90,104,177,178].
this temptation was resisted in view of the above discus- In cases where the fluid is dense, e.g. for liquid flows,
sion. The true relationship between U J f n D and the and especially in the case of tightly spaced arrays, there
parameters involved is likely to be rather complex and exists non-negligible hydrodynamic coupling in cylinder
to have the form [170-176] motions; i.e., it may no longer be assumed that motions
of one cylinder do not affect those of adjacent ones,
U J / n D = F ( m S / p D 2 , m / p D 2, st, s£, Re, u.t.c.), even in still fluid. Motion of one cylinder accelerates the
fluid, which creates a pressure field around adjacent
where u.t.c, stands for upstream turbulence characteris-
cylinders, causing them to also vibrate. Hydrodynamic
tics, which may affect the onset of instability in the
coupling has extensively been studied by S.S. Chen and
most critical first few rows of the array. Clearly, the
his associates and by others [179-187].
above relationship implies that the simple form U J f n D
One important consequence of hydrodynamic cou-
= F ' ( m S / p D 2) can only be a rough guide; the fact that
pling is the following. A solitary cylinder immersed in
it is at all useful simply suggests that m S / p D 2 is a very
dense fluid would have one first-beam-mode natural
important parameter in determining fluidelastic instabil-
frequency, one second-mode natural frequency, and so
ity.
on. The same would be true for an array of cylinders in
vacuo. However, for a N-cylinder array immersed in
dense fluid, there would be a group of 2N first-beam-
5. Natural frequencies and hydrodynamic coupling in mode natural frequencies, with values clustered about
tightly-spaced arrays and multiply supported cylinders that of the solitary cylinder in the same fluid: thus,
there is a band of first-mode frequencies, in place of a
In most of the foregoing, the cylinders have been single discrete value; the tighter the array, the wider this
characterized by a single natural frequency, fn, generally band would generally be. Similarly, there would be a
presumed to be the lowest natural frequency of a cylin- band of second-mode natural frequencies, and so on.
der in the system: i.e., that of its first beam mode. This For cylinder arrays of relatively few cylinders, the
is so for two reasons: first, this mode is likely to be the individual frequencies (and the bands referred to above)
least damped and hence the most easily excited; second, may be found by the "classical" means of refs.
50 M.P. Paidoussis / Flow-induced vibrations in reactors

[ 179-184,186] or by finite-element techniques [185, 186, ordinarily be neglectd, were it not for the often ex-
188,189]. For large arrays, however, a more attractive tremely close spacing of the cylinders in the array, with
proposition is to determine the frequency bands by the intercylinder gap-to-radius ratios of the order of 10- ~"
so-called homogenization technique [ 190,191 ]. hence, although very small, these vibrations may cause
It should finally be mentioned that in two-phase intercylinder impact, which may result in wear and
flows hydrodynamic coupling effects are not well under- fretting damage.
stood and cannot yet be properly modelled [192,193]. There exist well developed analytical tools for pre-
In this paper the difficulties of dealing with real dicting the vibration amplitude of cylinders in axial
cylinder supports, with partial contact, flexible con- flow [197-205]; strictly, they really apply to solitary
straints and other nonlinear effects will not be consid- cylinders, rather than clusters. The various theories de-
ered (vide, e.g., [178]), nor will the important effects of veloped are essentially similar, but nevertheless differ in
mechanical coupling [166]. some important respects which should be pointed out.
We next turn our attention to vibration of solitary Some theories [199,200,202,203] consider the cylinder
cylinders and arrays induced by axial flow. to be a damped beam immersed in still fluid, completely
ignoring the effect of mean flow on the vibrational
characteristics; the mean flow, according to this ap-
6. Cylinders in axial flow proach, is viewed simply as a vehicle for the generation
and transmission of the pressure fluctuations which
Problems associated with axial-flow-induced vibra- excite the vibration.
tion are rather rare [1]. Nevertheless, because of the The main effects of the mean flow are the following:
absence of separated flow regions, it has been possible (i) flow-induced damping is generated, which in the
to develop analytical methods to a much higher level range of most practical situations is proportional to U
than for cross-flow-induced vibration. Furthermore, [204,206-208,201 ];
there exists broad agreement as to the dynamical be- (ii) the natural frequencies of the cylinders are di-
haviour of cylinders in axial flow. Hence, it is felt to be minished [204,206,207];
unnecessary to give here a historical review for this (iii) the eigenmodes are no longer normal.
topic, which in any case may be found elsewhere The theory of ref. [201] takes some of these effects into
[194-196]. account, and the theory of ref. [204] takes them fully
The behaviour of the system with increasing flow is into account. It should be added, nevertheless, that if
much like that of fig. 6, except that (i) the resonance the dimensionless flow velocity u << 0.5, these effects
due to flow periodicity has a different meaning in this are not very large. As in most engineering applications
case, (section 6.2), and (ii) the instability is usually of u _< 0.5, neglecting mean flow effects is not an unrea-
the divergence type, rather than an oscillatory instabil- sonable approximation.
ity. The nonresonant buffeting response is essentially Other assumptions made are the following: (a) there
similar in axial and cross flow; it will simply be referred is no correlation between the pressure fields on adjacent
to here as buffeting, although it is often called subcriti- cylinders in the array [199-205]; (b) the motion of a
cal vibration. cylinder has no effect on the pressure field, nor on the
motion of adjacent cylinders [199-205]; (c) the process
6.1. Buffeting of cylinders in axial flow is ergodic and the pressure field homogeneous [199-202,
204].
It is known that at any given flow velocity, U, there In most cases, the characteristics of the pressure field
exist in the flow stochastic pressure fluctuations due to are left as input parameters to the analytical model. As
turbulence and far-field disturbances. It is now widely shown by Reavis [199] the excitation field contains
accepted that these, acting on the surface of the cylin- much more energy than that generated by boundary-
ders, constitute the excitation force field for buffeting layer pressure fluctuations. Indeed, the pressure field
vibration--although other possible mechanisms have exciting the vibration has near-field and far-field com-
also been advanced in the past [195]. Similarly to non- ponents. The near field comprises the local pressure
resonant buffeting in cross flow, the cylinders act as fluctuations associated with the boundary layer and the
shaped band-pass filters, vibrating predominantly in non-propagating part of disturbances associated with
their first-mode. singularities (e.g. valves, bends, protuberances, supports,
The vibration amplitudes Yrms/D, typically less than etc.). Propagating disturbances in the form of acoustic
10 -2, rarely exceed 10-~. Such small vibrations would waves constitute the far field. The theory of ref. [203]
 M.P. Paidoussis / Flow-induced vibrations in reactors 51

takes far field characteristics into account. may also be important, altough rarely taken into account
It has been shown that, generally, both the non- [213,214].
propagating part of singularity-related disturbances and It should also be mentioned that buffeting by two-
the far field (i.e., the so-called "system characteristics") phase flow may be quite different than pedicted by
are important [209]. That is the reason why, in cases assuming it to be an equivalent homogenized mixture
where the pressure field characteristics have been mea- [192,193,215-218].
sured in situ, i.e., in the very same facility in which Because of the limitations of the analytical models
cylinder vibration is measured, agreement between mea- discussed above, it is often necessary for the designer to
sured and predicted vibration amplitudes is so good predict buffeting vibration amplitudes by means of em-
[200-203]; otherwise, prediction and measurement may pirical and semi-empirical relationships [199,197,
differ by as much as one order of magnitude. 219-222], especially if there is no means at hand for
Clearly, the proposition of having to measure the predicting the pressure field characteristics in the sys-
pressure field characteristics in situ in order to predict tem being designed. As discussed in refs. [195,4], these
the vibration is not practical, for then the vibration relationships, applied indiscriminately to data for both
amplitude itself might as well be measured in situ, with solitary cylinders and to clusters, are capable of predict-
at least equal ease! ion of vibration amplitude to no better than one order
Of course, if the pressure field could be characterized of magnitude. This degree of success is similar to that
and predicted, say as a result of comprehensive series of achieved by analytical methods, unless the characteris-
measurements of surface pressure fluctuations on cylin- tics of the system are very well known. The interested
ders in clusters, then the existing analytical models reader is referred to refs. [195,1-4,205] for details of the
would become much more useful. Progress is being various models and comparison between them [4].
made in this regard. Measurements of turbulence and
wall-pressure fluctuations for a single cylinder in nar-
6.2. Vibrations due to flow periodicity
row annular flow and in seven-cylinder arrays have now
been made [210,211], with and without upstream turbu-
lators. For the single cylinder, it was found that turbu- If the mean flow velocity in an array of cylinders is
harmonically perturbed, then parametric resonances may
lence intensity eventually dies out to a constant level, no
be excited provided that the perturbation frequency, f,
matter what the upstream turbulence level is [210]; the
lies in the neighbourhood of 2 f n / k , k = 1, 2, 3... The
vibration amplitude is higher for higher upstream turbu-
physical nature of these resonances is similar to that
lence intensities, provided that Dh/D is high; for D h / D
induced in columns subjected to a time-varying com-
= 0.85, however, the vibration response was found to be
pressive load.
insensitive to upstream turbulence. For cylinder arrays
These parametric resonances are likely to material-
( D h / D = 0.63) it was found that the effect of upstream
ize, especially the principal one, where f = 2fn, if 8 is
turbulence almost disappears for x / L > 0.125, where x
not too large, the amplitude of flow perturbations not
is the axial coordinate measured from the upstream
support, so far as the wall-pressure p.s.d., ~p(f), is too small and the upstream turbulence level not too
high. These resonances have been studied both experi-
concerned [211]. Although the pressure field was found
to be a function of geometry, as well as D h, it is mentally and theoretically by Paidoussis, Hara and their
nevertheless possible to characterize it in terms of associates [223-226]. Prediction of the instability zones,
• p(f)(U/Dh)(½PU2) -2 versus f l U at a given axial at least for a single cylinder in axial flow, is very good
location--similarly to eq. (5). The theory of ref. [205] [226].
takes into account a varying ~ p ( f ) with x, assuming it
to decay exponentially. 6.3. Fluidelastic instabilities in axial flow
At present, there is no theory taking into account
hydrodynamic coupling on buffeting response--other A solitary cylinder in axial flow of sufficiently high
than the added mass effect. It has recently been shown flow velocity will be subject to fluidelastic instabilities:
that the dynamics of the system in such circumstances divergence, followed by flutter at higher flow velocities
may be quite different from that of a single cylinder [206,207,204,217,227-230]. The mechanism involves the
[212], especially at high flow velocities: the response reduction of the effective flexural rigidity of the system
spectrum is much broader, corresponding to the band of through a centrifugal (otherwise viewed as a compres-
first-mode frequencies referred to in section 5, the sive) load on the instantaneously bent cyhnder, where
breadth increasing with U. Mechanical coupling effects this force is proportional to u 2.
52 M.P. Paidoussis / Flow-induced vibrations in reactors

Cylinder arrays similarly develop fluidelastic insta- ual cylinders, that is of interest.
bilities, but at lower u for dense fluid flows, through the Similar are the vibrations of AGR-type fuel
effect of hydrodynamic coupling [179,231-233,187]; the "stringers" [238,62,63], where the string is sheathed with
tighter the array, the lower is the critical flow veocity. a tubular component, although the mechanism of exci-
The transition from the buffeting r6gime to instabil- tation is much more complex, because of the narrow-
ity occurs through progressive broadening of the hydro- ness of the flow passage between this sheath and the
dynamically coupled band of vibration frequencies (sec- surrounding flow tube.
tion 5) as the flow velocity is increased; loss of stability
by divergence then occurs when the lower bound of this
band reaches zero frequency [233,212]. Agreement be- 7. Conclusion
tween theory and experiment is excellent [231-233,212].
However, even in cylinder arrays, these instabilities It may safely be said that some very solid progress
materialize usually only for u > 1.5, which lies beyond has been achieved in the past few years. Although the
the operating range of typical engineering systems. This period cannot honestly be described as one of orgiastic
is the reason for not dwelling in this review on this creativity, it has nevertheless been characterized by what
interesting phenomenon. may be referred to as inspired persistence in attacking
the very questions where the state of the art, as per-
6.4. Axial-flow-induced vibration: state of knowledge and ceived in 1979, had been weakest [239].
design guidelines For vibration induced by axial flow, the greatest
weakness had been, and still is, in the characterization
Understanding of the various mechanisms underly- of the pressure field exciting buffeting, so as to permit
ing vibration of cylinders in axial flow may be said to the well-developed analytical models to be successfully
be good, at least in single-phase fluid flows. applied; here some significant progress has been made,
The design procedure is straight forward. First, it is and it may confidently be expected that in a few years
ascertained that u < 1.5, so that fluidelastic instabilities buffeting response in axially disposed arrays will be
are not likely to occur. Second, if the system is known to pedictable with the same degree of success as in piping
contain sizable harmonic flow-velocity perturbations, systems [ 1].
the design should be such that their frequency f be There has also been significant progress in the de-
+ 25% away from 2fn; alternatively upstream turbula- velopment of analytical models for axial-flow-induced
tors may be utilized to suppress parametric resonances. fluidelastic instabilities. This phenomenon per se is not
Third, the buffeting vibration amplitude is computed to practically important. However, the development of suc-
ensure that no intercylinder impact occurs, or that this cessful analytical methods for its prediction decidedly
impact does not exceed fretting criteria--or, for that is, for it underpins the analytical models for buffeting,
matter, that the vibration amplitude does not exceed which are of considerable practical concern.
fatigue criteria. For cross-flow-induced vibration of a single cylinder
The buffeting vibration amplitude may be estimated the state of affairs is fairly good. From the perspective
either semi-empirically [197,199,219-222], or, if there of this review, the most exciting developments are re-
exists reliable information on the excitation field for the lated to the dynamical behaviour of a cylinder in
hydraulic system at hand, analytically [197-205]--re- high-turbulence-intensity "industrial" environments,
membering that the reliability of prediction may be as which has been shown to be considerably different from
low as within one order of magnitude, as discussed in that under well controlled, low-turbulence laboratory
greater detail in section 6.1. conditions.
The most spectacular progress has been made in the
6.5. Related studies in axial-flow-induced vibration case of cross-flow-induced vibration of cylinder arrays,
where, circa 1980, there existed a state of restrained
The dynamics of so-called fuel strings has also been chaos, arising from the realization that the theretofore
studied, vide e.g. [234-237]. Fuel strings consist of bun- accepted understanding of the phenomena (e.g., "vortex
dles of cylinders stacked vertically and held together by shedding" in arrays) and semi-empirical design guide-
a central rod, supported only at the top or bottom of lines (e.g., the Connors-Blevins expression for predic-
the assembly and lying within a close-fitting tubular ting fluidelastic instability) were inadequate--yet, with
channel containing the flow. In this case, it is the nothing in sight to take their place. Buffeting response
vibration of the string as a whole, rather than of individ- may now be predicted well, but further characterization
 M.P. PaMoussis / Flow-induced vibrations in reactors 53

of the excitation field for a greater variety of geometries groin minimum value of K for design purposes
and flow conditions is still needed. Resonant buffeting, m mass of cylinder per unit length (usually in-
or " v o r t e x shedding", response may also be predicted cludes the added mass)
with some confidence; however, the conditions under P pitch of equilateral triangular arrays (fig. 1)
which this phenomenon will materialize remain to be Re Reynolds number
clarified, as well as more extensive and reliable data St transverse spacing of cylinder array, T / D
banks to be assembled on the Strouhal number and lift (fig. 1)
coefficients, for a variety of system parameters. Simi- S£ longitudinal spacing of cylinder array, L / D
larly, a great deal remains to be done on acoustic (fig. 1)
resonances, both at the fundamental level and in cross- S Strouhal number
testing the available empirical design guidelines. U flow velocity; for arrays in cross flow U is
Exciting developments have occurred with respect to usually the "reference gap velocity", related to
fluidelastic-instability analytical modelling for cylinder the array-less (empty) upstream velocity U~ by
arrays in cross flow; however, the state of the art is still Uoos,/(s,- 1)
in a state of flux. critical flow velocity for fluidelastic instability
The weakest link in all the above remains [239] its and acoustic resonance
application to two-phase flows, despite the fact that U dimensionless flow velocity for axial flow, =
such flows are quite common. (oA/EI)I/2UL
Having completed this review, the writer is filled Yrms rms vibration amplitude due to buffeting
with a sense of awe and admiration for the designer: L length of cylinder
despite serious gaps in codified knowledge, contradic- Y mode shape factor, = 1 for rigid-body motion,
tory sets of design guides and researchers' claims (ex- 1.305 for cantilevered cylinder, and 1.155 for
acerbated by the latter's apparent reluctance to read one simply-supported cylinder
another's work), the designer has been able, through his 8 logarithmic decrement
legendary intestinal sixth sense, common sense and ex- ~r = 2 m S / p D 2, in table 1 (8*: based on mea-
perience, to put together so many systems that have surements in still fluid)
given satisfactory service for so long. damping factor (ratio of actual to critical damp-
ing)
p fluid density
Acknowledgment ,/,.(x) n th mode eigenfunction of cylinder

The author is grateful to the Natural Sciences and

Engineering Research Council of Canada and Le pro- References
gramme " F o r m a t i o n de chercheurs--Action concertre"
du Qurbec for their support. He also gratefully [ 1] M.P. Paidoussis, Flow-induced vibrations in nuclear reac-
acknowledges S.J. Price's useful comments on the whole tors and heat exchangers: practical experiences and state
of this manuscript, and M.J. Pettigrew's on section 3.3. of knowledge, in: Practical Experiences with Flow-In-
duced Vibrations, eds. Naudascher E. and Rockwell, D.
(Springer-Verlag, Berlin, 1980) pp. 1-81.
[2] R.D. Blevins, Flow-induced vibrations in nuclear reac-
Nomenclature tors: a review, Progress in Nuclear Energy 4 (1979)
24-49.
Ay transverse displacement amplitude due to [3] M.J. Pettigrew, Flow-induced vibration phenomena in
Strouhal resonance nuclear power station components, Power Industry Re-
D cylinder diameter search 1 (1981)97-133.
Dh hydraulic diameter [4] M.P. Paidoussis, Fluidelastic vibration of cylinder arrays
E1 flexural rigidity of cylinder in axial and cross flow: state of the art, Journal of Sound
frequency (Hz) and Vibration 76 (1981) 329-360. Also in slightly differ-
f ent form in: Flow-Induced Vibration Design Guidelines
L acoustic resonance frequency
(ed. P.Y. Chen), pp. 11-46; ASME Publication PVP-Vol.
f. natural frequency of cylinder
52; ASME, New York (1981).
L Strouhal frequency [5] Y. Moil and S. Tokuda, The effect of oscillation on
Ls vortex-shedding frequency instantaneous local heat transfer in forced convection
K fluidelastic instability constant in eq. (7) from a cylinder, Proc. 3rd Int. Heat Transfer Conference
3 (1966) 49-56.
54 M.P. Paidoussis / Flow-induced vibrations in reactors

[6] P.D. Richardson, Effects of sound and vibration on heat [26] O.M. Griffin and S.E. Ramberg, Vortex shedding from a
transfer, Applied Mechanics Reviews 20(1967) 201-217. cylinder vibrating in line with an incident uniform flow,
[7] V. Strouhal, Uber eine besondere Art der Tonerregung, Journal of Fluid Mechanics 75 (1976) 257-271.
Annalen der Physik 5 (1878) 216-251. [27] W.A. Martin, I.G. Currie and E. Naudascher, Streamwise
[8] M.V. Morkovin, Flow around a circular cylinder--a oscillations of cylinders, Journal of the Engineering
kaleidoscope of challenging fluid phenomena, ASME Mechanics Division, ASCE 107 (1981) 589-607.
Symposium on Fully Separated Flows (1964) pp. 102-118. [28] A. Richter and E. Naudascher, Fluctuating forces on a
[9] A.W. Marris, A review of vortex streets, periodic wakes rigid circular cylinder in confined flow, Journal of Fluid
and induced vibration phenomena, ASME Journal of Mechanics 78 (1976) 561-576.
Basic Engineering 86 (1964) 185-196. [29] L.R. Wootton, The oscillations of large circular stacks in
[10] J.H. Lienhard, Synopsis of lift, drag, and vortex frequency wind, Proc. Institution of Civil Engineers 43 (1969)
data for rigid circular cylinders, Bulletin 300, College of 573-598.
Engineering, Washington State University (1966). [30] T. Okamoto and M. Yagita, The experimental investiga-
[11] W.A. Mair and D.J. Maull, Bluff bodies and vortex tion on the flow past a circular cylinder of finite length
shedding--a report on Euromech 17, Journal of Fluid placed normal to the plane surface in a uniform stream,
Mechanics 45 (1971) 209-224. Bulletin of the Japan Society of Mechanical Engineers 16
[12] E. Berger and R. Wille, Periodic flow phenomena, in: (1973) 805-814.
Annual Review of Fluid Mechanics 4 pp. 313-340; An- [31] B.H.L. Gowda, Some measurements on the phenomena
nual Reviews, Palo Alto, Calif. (1972). of vortex shedding induced vibrations on circular cylin-
[13] W.J. McCroskey, Some current research in unsteady fluid ders, Report DLR-FB-75-01, Technische Universit~t.
dynamics, ASME Journal of Fluids Engineering 99 (1977) Berlin (1975).
8-39. [32] P.L. Poon, Effects of mass ratio on vortex-induced oscil-
[14] R. King, A review of vortex shedding research and its lations of a circular cylinder, M.A.Sc. Thesis, University
application, Ocean Engineering 4 (1977) 141-171. of Toronto, Toronto (1980).
[15] T. Sarpkaya, Vortex-induced oscillations: a selective re- [33] M.J. Turnstall, The cross-wind vibration of chimneys,
view, ASME Journal of Applied Mechanics 46 (1979) Paper 123, Int. Symposium on Vibration Problems in
241-258. Industry, Keswick, U.K., 1973.
[16] F. Boulot, et al., Vibrations de structures dans un 6coule- [34] P. Sonneville, Etude de la structure tridimensionnelle des
ment sous l'effet du d+tachement tourbillonnaire de sil- 6coulements autour d'un cylindre circulaire, Bulletin de
lage, Special Report, Soci~t~ Hydrotechnique de France, la Direction des Etudes et Recherches, Electricit6 de
Paris (1981). France, Serie A, No. 3 (1976).
[17] E. Naudascher (ed.), Flow-Induced Structural Vibrations [35] R.M.C. So and S.D. Savkar, Buffeting forces on rigid
(Springer-Verlag, Berlin, 1974). circular cylinders in cross flows, Journal of Fluid Mech-
[18] R.D. Blevins, Flow-Induced Vibration (Van Nostrand anics 105 (1981) 397-425.
Reinhold, New York, 1977). [36] J.P. Batham, Pressure distributions on circular cylinders
[19] C. Farell, Flow around fixed circular cylinders: Fluc- at critical Reynolds numbers, Journal of Fluid Mechanics
tuating loads, Journal of the Engineering Mechanics Di- 57 (1973) 209-228.
vision, ASCE 107 (1981) 565-588. [37] E. Szechenyi, Supercritical Reynolds number simulation
[20] R.E.D. Bishop and A.Y. Hassan, The lift and drag forces for two-dimensional flow over circular cylinders, Journal
on a circular cylinder in a flowing fluid, Proc. Royal of Fluid Mechanics 70 (1975) 523-542.
Society (London) 277(A) (1964) 51-75. [38] E. Achenbach and E. Heinecke, On vortex shedding from
[21] N. Ferguson and G.V. Parkinson, Surface and wave flow smooth and rough cylinders in the range of Reynolds
phenomena on the vortex excited oscillation of a circular numbers 6 x 103 to 5 × 106, Journal of Fluid Mechanics
cylinder, ASME Journal of Engineering for Industry 89 109 (1981) 239-251.
(1967) 831-838. [39] P.W. Bearman and M.M. Zdravkovich, Flow around a
[22] O.M. Griffin and S.E. Ramberg, On vortex strength and circular cylinder near a plane boundary, Journal of Fluid
drag in bluff-body wakes, Journal of Fluid Mechanics 69 Mechanics 89 (1978) 33-47.
(1975) 721-728. [40] D.T. Tsahalis and W.T. Jones, Vortex-induced vibrations
[23] E. Szechenyi and H. Loiseau, Portances instationnaires of a flexible cylinder near a plane boundary in steady
sur un cylindre vibrant dans un 6coulement supercritique, flow, Proc. 13th Offshore Technology Conference, Dal-
La Recherche Arrospatiale 1 (1975) 45-57. las, Texas, Vol. 1 (1981) pp. 367-381.
[24] P.K. Stansby, The locking-on of vortex shedding due to [41] R. King, Vortex-excited oscillations of yawed circular
the cross-stream vibration of circular cylinders in uni- cylinders, ASME Journal of Fluids Engineering 99 (1977)
form and shear flows, Journal of Fluid Mechanics 74 495-502.
(1976) 641-665. [42] F. Hara, Two-phase cross-flow-induced forces acting on
[25] R. King, M.J. Prosser and D.J. Johns, On vortex excita- a circular cylinder, in: Flow-Induced Vibration of Cir-
tion on model piles in water, Journal of Sound and cular Cylindrical Structures--1982 (ed. S.S. Chen, M.P.
Vibration 29 (1972) 169-188.
 M.P. PaMoussis / Flow-induced vibrations in reactors 55

Paidoussis and M.K. Au-Yang), ASME Publication PVP- [61] Proceedings, U K A E A / N P L International Symposium on
63, pp. 9-17 (ASME, New York, 1982). Vibration Problems in Industry, Keswick, U.K., April
[43] R.T. Hartlen and I.G. Currie, Lift-oscillator model of 1973.
vortex-induced vibration, Journal of the Engineering [62] Proceedings BNES International Conference on Vibra-
Mechanics Division, ASCE 96 (1970) 577-591. tion in Nuclear Plant, Keswick, U.K., May 1978.
[44] I.G. Cuttle, Private communication (14 July 1982). [63] Proceedings U K A E A / B N E S Third Keswick Interna-
[45] O.M. Griffin, R.A. Skop and S.E. Ramberg, The reso- tional Conference on Vibration in Nuclear Plant,
nant vortex-excited vibrations of structures and cable Keswick, U.K., May 1982.
systems, Paper OTC-2319, Offshore Technology Con- [64] D.D. Reiff, (ed.), Flow-Induced Vibration in Heat Ex-
ference, May 1975, Houston, Texas. changers (ASME, New York, 1970).
[46] W.D. Iwan and R.D. Blevins, A model for vortex-in- [65] S.S. Chen and M.D. Bernstein (eds.), Flow-lnduced
duced oscillation of structures, ASME Journal of Applied Vibrations (ASME, New York, 1979).
Mechanics 41 (1974) 581-586. [66] M.K. Au-Yang (ed.), Flow-Induced Vibration in Power
[47] T. Sarpkaya, Fluid forces on oscillating cylinders, Journal Plant Components, PVP-Vol. 41 (ASME, New York,
of the Waterway, Port, Coastal and Ocean Division, 1980).
ASCE 104 (1978) 275-290. [67] J.M. Chenoweth and J.R. Stenner (eds.), Flow-Induced
[48] G.H. Toebes, The unsteady flow and wake near an Heat Exchanger Tube Vibration--1980, HTD-Vol. 9
oscillating cylinder, ASME Journal of Basic Engineering (ASME, New York, 1980).
91 (1969) 493-502. [68] P.Y. Chen, (ed.), Flow-Induced Vibration Design Guide-
[49] H.J. Connors, Jr., Vortex-shedding excitation and vibra- lines, PVP-Vol. 52 (ASME, New York, 1981).
tion of circular cylinders, in: Flow-Induced Vibration [69] S.S. Chen, M.P. Paidoussis and M.K. Au-Yang, (eds.),
Design Guidelines (ed. P.Y. Chen), pp. 47-74. ASME Flow-Induced Vibration of Circular Cylindrical Struc-
Publication PVP-Vol. 52 (ASME, New York, 1981). t u r e s - 1982, PVP-Vol. 63 (ASME, New York, 1982).
[50] M.C. Welsh and D.C. Gibson, Interaction of induced [70] E. Naudascher and D. Rockwell (eds.), Practical Experi-
sound with flow past a square leading edged plate in a ences with Flow-Induced Vibrations (Springer-Verlag,
duct, Journal of Sound and Vibration 67 (1979) 501-511. Berlin, 1980).
[51] R. Parker and M.C. Welsh, The effect of sound on flow [71] Trans. 4th International Conference on Structural Me-
over bluff bodies, Paper 9.3, Third Keswick Int. Conf. on chanics in Reactor Technology, San Francisco, 1977.
Vibration in Nuclear Plant, Kewsick, U.K., May 1982. [72] Trans. 6th International Conference on Structural Me-
[52] S.D. Savkar, and R.M.C. So, On the buffeting response chanics in Reactor Technology, Paris 1981.
of a cylinder in turbulent cross flow, General Electric [73] A.A. Putnam, Flow-induced noise in heat exchangers,
Report GEAP-24149 (Sept. 1978). ASME Journal of Engineering for Power 81 (1959)
[53] T.M. Mulcahy, Design guide for single circular cylinder 417~422.
in turbulent crossflow, Argonne National Laboratory [74] R.C. Baird, Pulsation-induced vibrations in utility steam
Report ANL-CT-82-7 (March 1982). generating units, Combustion 25 (1954) 38-44.
[54] D.J. Johns and C.B. Sharma, On the mechanism of [75] B.J. Grotz and F.R. Arnold, Flow-induced vibrations in
wind-excited ovalling vibrations of thin circular cylindri- heat exchangers, Stanford University Tech. Report No.
cal shells, in ref. [17], pp. 650-662. 31 (Aug. 1956).
[55] M.P. Paidoussis and C. Helleur, On ovalling oscillations [76] J.L. Livesey and R.C.F. Dye, Vortex excited vibration of
of cylindrical shells in cross-flow, Journal of Sound and a heat exchanger tube row, Journal of Mechanical En-
Vibration 63 (1979) 527-542. gineering Science 4 (1962) 349-352.
[56] M.P. Paidoussis and D.T.M. Wong, Flutter of thin cylin- [77] A.A. Putnam, Flow-induced noise and vibration in heat
drical shells in cross flow, Journal of Fluid Mechanics exchangers, ASME Paper 64-WA/HT-21 (Nov-Dec
115 (1982) 411-426. 1964).
[57] M.P. Paidoussis, S.J. Price and H.C. Such, Ovalling oscil- [78] A. K6nig and R. Gregorig, Vibrational criteria for a pipe
lations of cantilevered and clamped-clamped cylindrical under conditions of transverse incident flow. Part 3:
shells in cross flow: an experimental study, Journal of Experiments on vibration of bundled tubes, Chemie Ing.
Sound and Vibration 83 (1982) 533-553. Technik, 40 (1968) 645-650; CEGB translation: C.E.
[58] A.E. Perry, M.S. Chong and T.T. Lim, The vortex-shed- Trans. 5063.
ding process behind two-dimensional bluff bodies, Jour- [79] Y.N. Chen, Flow-induced vibration and noise in tube
nal of Fluid Mechanics 116 (1982) 77-90. bank heat exchangers due to yon Karman streets, ASME
[59] P.W. Bearman and A.J. Wadcock, The interaction be- Journal of Engineering for Industry 90 (1968) 134-146.
tween a pair of circular cylinders normal to a stream, [80] P.R. Owen, Buffeting excitation of boiler tube vibration,
Journal of Fluid Mechanics 61 (1973) 499-511. Journal of Mechanical Engineering Science 7 (1965)
[60] M.M. Zdravkovich, Review of the Interference between 431-439.
two circular cylinders in various arrangements, ASME [81] Y.N. Chen, Excitation sources of the flow-induced vibra-
Journal of Fluids Engineering 99 (1977) 618-633. tions and noise in tube bank heat exchangers, in: Noise
56 M.P. PaMoussis / Flow-induced vibrations in reactors

and Fluid Engineering (ed. R. Hickling), pp. 239-246 cation and suppression of flow-induced vibrations in heat
(ASME, New York, 1977). exchangers, in ref. [70] (1980) pp. 109-117.
[82] Y.N. Chert, Criteria for the cross-flow-induced tube [100] Y.N. Chen, Turbulence as excitation source in staggered
vibrations in tube bank heat exchangers, Paper 2.6, in ref. tube bundle heat exchangers, in ref. [66] (1980) pp.
[62] (1978). 45-63.
[83] J.S. Fitz-Hugh, Flow-induced vibration in heat ex- [101] R.J. Gibert, M. Sagner and R. Doyen, Vibration of tube
changers, Paper 427, in ref. [61] (1973). arrays in transversal flow, Paper B6/7, in ref. [72] (1981).
[84] Y.N. Chen, The sensitive tube spacing region of tube [102] R.D. Blevins, R.J. Gibert and B. Villard, Experiment on
bank heat exchangers for fluid-elastic coupling in cross- vibration of heat exchanger tube arrays in cross flow,
flow, in: Fluid-Structure Interaction Phenomena in Pres- Paper B6/9, in ref. [72] (1981).
sure Vessel and Piping Systems (ed. M.K. Au-Yang and [ 103] D.J. Gorman, The effects of artificially induced up-stream
S.J. Brown), pp. 1-18 (ASME, New York, 1977). turbulence on the liquid cross-flow induced vibration of
[85] B.G. Murray, W.B. Bryce and G. Rae, Strouhal Numbers tube bundles, in ref. [66] (1980) pp. 33-43.
in Tube Arrays, Paper 2.4, in ref. [63] (1982). [104] D.J. Gorman, Experimental study of the flow induced
[86] R.J. Gibert, J. Chabrerie and M. Sagner, Vibrations of vibration of multi-span heat exchanger tube bundles in
tube arrays in transversal flow, Paper F4/h, in ref. (71) liquid cross-flow, Paper B6/6, in ref. [72] (1981).
(1977). [105] D.J. Gorman, Further experimental studies of the vibra-
[87] J.B. Nicolet, M. Sagner and G. R6gis, Vibrations de tion of multi-span heat exchanger tubes subjected to
faisceaux de tubes sous excitation a6rodynamique, Revue liquid cross-flow, Paper 1.6, in ref. [63] (1982).
G6n6rale de Thermique 128 (1976) 869-891. [106] S.D. Savkar and T.A. Litzinger, Buffeting forces induced
[88] L.K. Grover and D.S. Weaver, Cross-flow induced vibra- by cross-flow through arrays of cylinders, Paper 1.5, in
tions in a tube bank-vortex shedding, Journal of Sound ref. [63] (1982).
and Vibration 59 (1978) 263-276. [107] T. Nakamura, K. Fujita, K. Shiraki, H. Kanazawa and K.
[89] M.M. Zdravkovich and J.E. Namork, Flow structure Sakata, An experimental study of exciting force by two-
within both stationary and vibrating tube banks with phase flow, in ref. [69] (1982) pp. 19-29.
triangular pitch, Paper 2.5, in ref. [62] (1978). [108] I.S. Donaldson and W. McKnight, Turbulence and
[90] M.J. Pettigrew, Y. Sylvestre and A.O. Campagna, Vibra- acoustic signals in a cross-flow heat exchanger model, in
tion analysis of heat exchanger and steam generator ref. [65] (1979) pp. 123-128.
designs, Nucl. Engrg. Des. 48 (1978) 97-115. [109] F.L. Eisinger, Prevention and cure of flow-induced vibra-
[91] M.J. Pettigrew and D.J. Gorman, Vibration of heat ex- tion problems in tubular heat exchangers, ASME Journal
change components in liquid and two-phase cross-flow, of Pressure Vessel Technology 102 (1980) 138-145.
Paper 2.3, in ref. [62] (1978). [110] R. Parker, Acoustic resonances in passages containing
[92] M.J. Pettigrew and D.J. Gorman, Vibration of heat ex- banks of heat exchanger tubes, Journal of Sound and
changer tube bundles in liquid and two-phase cross-flow, Vibration 57 (1978) 245-260.
in ref. [68] (1981) 89-110. [111] T.E. Burton, Sound speed in heat exchanger tube bank,
[93] R.D. Blevins, Buffeting of heat exchanger tube arrays in Journal of Sound and Vibration 71 (1980) 157-160.
cross flow, Paper 3.4, in ref. [62] (1978). [112] J. Planchard, Computation of the acoustic eigen-frequen-
[94] Y.N. Chen and W.C. Young, The orbital movement and cies of cavities containing tube bundle, Computer Meth-
the damping of the fluidelastic vibration of tube banks ods in Applied Mechanics and Engineering 24 (1980)
due to vortex formation, Part 3, ASME Journal of En- 125-135.
gineering for Industry 96 (1974) 1072-1075. [113] J. Planchard, F. R6my and P. Sonneville, A simplified
[95] G. Baylac and J.P. Gr6goire, Etude des +coulements et method for determining acoustic and tube eigen frequen-
des ph6nom~nes acoustiques dans un faisceau de tubes, cies in heat exchanger, in ref. [69] (1982) pp. 197-207.
in: A6ro-hydro-61asticit& recherches actuelles et applica- [114] M.M. Zdravkovicb and J.A. Nuttall, On the elimination
tions industrielles, pp. 279-379; Eyrolles, Paris. of aerodynamic noise in a staggered tube bank, Journal
[96] J.D. Rogers and C.A. Penterson, Predicting sonic vibra- of Sound and Vibration 34 (1974) 173-177.
tion in cross flow heat exchangers--Experienceand model [115] R.D. Blevins, Acoustic resonance in heat exchanger tube
testing, ASME Paper 77-WA/DE-28 (1977), reprinted as bundles, Paper 2.5, in ref. [63] (1982).
Combustion Engineering Report TIS-5154. [116] E. Ulrich, Ober die Existenz kfirm~nerregter stehender
[97] W.B. Bryce, J.S. Wharmby and J. Fitzpatrick, Duct Wellen in Dampferzeugern, VGB Kraftwerkstechnik 61
acoustic resonances induced by flow over coiled and (1981) 412-417.
rectangular heat exchanger test banks of plain and finned [117] J.A. Fitzpatrick, Acoustic resonances in in-line tube
tubes, Paper 3.5, in ref. [62] (1978). banks, and response by M.P. Paidoussis. Letter to the
[98] M.M. Zdravkovich and J.E. Namork, Structures of inter- Editor, Journal of Sound and Vibration 85 (1982)
stitial flow between closely spaced tubes in staggered 435-441.
array, in ref. [65] (1979) pp. 41-46. [118] B.W. Roberts, Low frequency self-excited vibration in a
[99] M.M. Zdravkovich and J.E. Namork, Excitation amplifi- row of circular cylinders mounted in an air stream, Ph.D.
Thesis, University of Cambridge (1962).
 M.P. PaMoussis / Flow-induced vibrations in reactors 57

[119] B.W. Roberts, Low frequency, aeroelastic vibrations in a [137] D. Bai, Tube vibrations in large condensers, in ref. [70]
cascade of circular cylinders, I. Mech. E. Mechanical (1980) 94-98.
Engineering Science Monograph No. 4 (Sept. 1966). [138] P.J. Southworth and M.M. Zdravkovich, Cross-flow-in-
[120] H.J. Connors, Jr., Fluidelastic vibration of tube arrays duced vibrations of finite tube banks in in-line arrange-
excited by cross flow, in ref. [64] (1970) pp. 42-56. ment, Journal of Mechanical Engineering Science 17
[121] R.T. Hartlen, Wind tunnel determination of fluid elastic (1975) 190-198.
vibration thresholds for typical heat exchanger tube pat- [139] P.J. Southworth and M.M. Zdravkovich, Effect of grid-
terns, Ontario Hydro Research Division, Mechanical Re- turbulence on the fluid-elastic vibrations of in-line tube
search Dept. Report 74-309-K (1974). banks in cross flow, Journal of Sound and Vibration 39
[122] H. Halle and W.P. Lawrence, Crossflow-induced vibra- (1975) 461-469.
tion of a row of circular cylinders in water, ASME Paper [140] M.M. Zdravkovich, S. Singh, J.A. Nuttal and D.M. Cau-
77-JPGC-NE-4 (1977). son, Flow induced vibrations in staggered tube banks, pp.
[123] S. Ishigai and E. Nishikawa, Private communication (l 237-243, Proc. 6th I. Mech. E. Thermodynamics and
May 1979). Fluid Mechanics Convention, Durham, 1976.
[124] R.D. Blevins, Fluidelastic whirling of a tube row, ASME [141] E.P. Heinecke, Fluid-elastic vibrations in heat exchangers
Journal of Pressure Vessel Technology 96 (1974) 263-267. with tubes in cross-flow, Paper 2.4, in ref. [62] (1978).
[ 125] R.D. Blevins, Fluidelastic whirling of tube rows and tube [142] R.L. Coit, C.C. Peake and A. Lohmeier, Design and
arrays, ASME Journal of Fluids Engineering 99 (1977) manufacture of large surface condensers--problems and
457-460. solutions, Proc. American Power Conference 28 (1966)
[126] R.D. Blevins, Fluid damping and the whirling instability 469-483.
of tube arrays, in ref. [65] (1979) pp. 35-39. [143] J.T. Thorngren, Predict exchanger tube damage, Hydro-
[127] M.J. Pettigrew and D.J Gorman, Experimental studies in carbon Processing 49 (1970) 129-131.
flow-induced vibration to support steam generator design. [144] C.C. Peake, G.F. Gerstenkorn and T.R. Arnold, Some
Part 3, Paper 424, in ref. [6] (1973). reliability considerations for large surface condensers,
[128] R.J. Gibert, J. Chabrerie and R. Schlegel, Etude des Proceedings American Power Conference 37 (1975)
vibrations a~ro et hydrorlastiques de faisceaux de tubes 562-574.
en ~coulement transversal, Soci~t6 Bertin et Cie, Soci~t6 [145] D.S. Weaver and L.K. Grover, Cross-flow induced vibra-
Fives-Cail Babcock, and Commissariat h l'Energie Ato- tions in a tube bank--turbulent buffeting and fluid elas-
mique (France); Final Report, D.G.R.S.T. Etude 4870 tic instability, Journal of Sound and Vibration 59 (1978)
CR no. 1 (Octobre 1976). 277-294.
[129] R.E. Franklin and B.M.H. Soper, An investigation of [146] D.S. Weaver and J. Lever, Tube frequency effects on
fluidelastic instabilities in tube banks subjected to fluid cross flow induced vibrations in tube arrays, Paper IV.4,
cross flow, UKAEA Report AERE R-8708, HTFS RS192 5th Biennial Symposium on Turbulence, Dept. Chem.
(1977). Engineering, University of Missouri-Rolla, Oct. 1977.
[ 130] D.J. Gorman, Experimental study of peripheral problems [147] T.F. Balsa, Potential flow interactions in an array of
related to liquid flow induced vibration in heat exchangers cylinders in cross flow, Journal of Sound and Vibration
and steam generators, Paper F4/g, in ref. [71] (1977). 50 (1977) 285-303.
[131] H.J. Connors, Jr., Fluidelastic vibration of heat exchanger [148] S.S. Chen, Crossflow-induced vibrations of heat ex-
tube arrays, ASME Journal of Mechanical Design 100 changer tube banks, Nuclear Engineering and Design 47
(1978) 347-353. (1978) 67-86.
[132] A. Zhukauskas, V. Katinas, E. Perednis, and V. Sobolev, [149] W.J. Heilker and R.Q. Vincent, Vibration in nuclear heat
Flow and vibration of radial banks of tubes of heat exchangers due to liquid and two-phase flow, ASME
exchangers in cross-flow of fluid, Thermophysics, Journal of Engineering for Power 103 (1981) 358-366.
Academy of Science Lith. S.S.R. 6 (109) (1978) 53-61. [150] S.S. Chen and J.A. Jendrzejczyk, Flow velocity depen-
[133] A. Zhukauskas, V. Katinas, V. Suksteris and A. Mikishev, dence of damping in tube arrays subjected to liquid
Flow, hydrodynamic forces and hydraulic resistance of cross-flow, ASME Journal of Pressure Vessel Technology
staggered radial tube banks in cross flow of viscous fluid, 103 (1981) 130-135.
Thermophysics, Academy of Sciences Lith. S.S.R. Series [151] S.S. Chen and J.A. Jendrzejczyk, Experiments on fluid
B 5(108) (1978) 79-89. elastic instability in tube banks subjected to liquid cross
[134] A. Zhukauskas and V. Katinas, Flow-induced vibration flow, Journal of Sound and Vibration 78 (1981) 355-381.
in heat-exchanger tube banks, in ref. [70] (1980) 188-196. [152] D.S. Weaver and M. EI-Kashlan, On the number of tube
[135] B.M. Soper, The effect of tube layout on the fluidelastic rows required to study cross-flow induced vibrations in
instability of tube bundles in cross flow, in ref. [67] tube banks, Journal of Sound and Vibration 75 (1981)
(1980) 1-9. 265-273.
[136] M.J. Pettigrew and A.O. Campagna, Heat exchanger tube [153] D.S. Weaver and M. E1-Kashlan, The effect of damping
vibration: comparison between operating-experiences and and mass ratio on the stability of a tube bank, Journal of
vibration analyses, in ref. [70] (1980) 82-93. Sound and Vibration 76 (1981) 283-294.
58 M.P. Paidoussis / Flow-induced vibrations in reactors

[154] H.C. Yeung and D.S. Weaver, The effect of approach vibration of tube arrays with various pitch-to-diameter
velocity direction on the flow-induced vibrations of a ratios, in ref. [69] (1982) pp. 45-56.
triangular tube array, ASME Journal of Vibration, [170] S.S. Chen, Instability mechanisms and stability criteria of
Acoustics, Stress and Reliability in Design 105 (1983) a group of circular cylinders subjected to cross flow. Part
76-82. 1: Theory, ASME Journal of Vibration, Acoustics, Stress
[155] D.S. Weaver and D. Koroyannakis, Flow induced vibra- and Reliability in Design 105 (1983) 51-58.
tions of heat exchanger U-tubes, a simulation to study [171] S.S. Chen, Instability mechanisms and stability criteria of
the effects of asymmetric stiffness, ASME Journal of a group of circular cylinders subjected to cross flow. Part
Vibration, Acoustics, Stress and Reliability in Design 105 If: Numerical results and discussions, ASME Journal of
(1983) 67-75. Vibration, Acoustics, Stress and Reliability in Design 105
[156] D.S. Weaver and D. Koroyannakis, The cross-flow re- (1983), ASME Paper 81-DET-22.
sponse of a tube array in w a t e r - - a comparison with the [172] H.G. Gross, Undersuchung aeroelastischer Schwin-
same array in air, in ref. [69] (1982) pp. 71-85. gungsmechanismen und deren Ber~cksichtigung bei der
[157] E.P. Heinecke and K.H. Mohr, Investigations on fluid Auslegung von Rohrbfindel-W~metauschern, Ph.D. The-
borne forces in heat exchangers with tubes in cross flow, sis, Technical University of Hannover (1975).
Paper 1.2, in ref. [63] (1982). [173] S.J. Price and M.P. Paidoussis, Fluidelastic instability of
[158] F.N. Rrmy, Flow induced vibration of tube bundles in an infinite double row of circular cylinders subject to a
two phase cross flow, Paper 1.9, in ref. [63] (1982). uniform crossflow, ASME Journal of Vibration, Acous-
[159] K. Fujita, T. Ito, K. Shiraki, T. Veda, Y. Kondo, T. tics, Stress and Reliability in Design 105 (1983) 59-66.
Teranishi, M. Shiba and Y. Yamasaki, Flow-induced [174] S.J. Price and M.P. Paidoussis, A theoretical investigation
vibration of PWR fuel assembly by baffle leakage cross- of the parameters affecting the fluidelastic instability of a
flow, Paper B6/8, in ref. [72] (1981). double row of cylinders subject to a cross-flow, Paper 1.7,
[160] H.J. Connors, Fluid-elastic vibration of tube arrays in ref. [63] (1982).
excited by a non-uniform cross-flow, in ref. [66] (1980) [175] J.P. DenHartog, Transmission line vibration due to sleet,
pp, 93-107. Trans. American Institute of Electrical Engineers 51
[161] H. Halle, J.M. Chenoweth and M.W. Wambsganss, (1932) 1074-1076.
Flow-induced tube vibration tests of typical industrial [176] J.H. Lever and D.S. Weaver, A theoretical model for the
heat exchanger configurations, presented at the ASME fluidelastic instability in heat exchanger tube bundles, in
Design Engineering Conference, Sept. 1981, Hartford, ref. [69] (1982) pp. 87-108.
Conn. ASME Paper 81-DET-37. [177] D.J. Gorman, Free Vibration Analysis of Beams and
[162] M.W. Wambsganss, H. Halle and W.P. Lawrence, Tube Shafts (J. Wiley and Sons, New York, 1975).
vibration in industrial size test heat exchanger (30 ° trian- [178] H.G.D. Goyder, Measurements of the natural frequen-
gular layout-six crosspass configuration) Argonne Na- cies and damping of loosely supported tubes in heat
tional Laboratories Tech. Memo. ANL-CT-81-42 (Oct. exchangers, Paper 2.6, in ref. [63] (1982).
1981). [179] S.S. Chen, Vibration of nuclear fuel bundles, Nuclear
[163] D. Bai, Flow-induced vibrations of multi-span tube bun- Engineering and Design 35 (1975) 399-422.
dles of large consdensers: experimental studies on full [180] S.S. Chen, Dynamic responses of two parallel circular
scale models in steam cross-flow, Paper 2.3, in ref. [63] cylinders in a liquid, ASME Journal of Pressure Vessel
(1982). Technology 97 (1975) 78-83.
[164] F.N. R~my and D. Bai, Comparative analysis of cross- [1811 S.S. Chen, Vibrations of a row of circular cylinders in a
flow induced vibrations of tube bundles, BHRA Int. liquid, ASME Journal of Engineering for Industry 97
Conf. on Flow-lnduced Vibrations in Fluid Engineering, (1975) 1212-1217.
Reading, U.K., Sept. 1982. [182] H. Chung and S.S. Chen, Vibration of a group of circular
[165] A.E. Collinson and A.F. Taylor, A development pro- cylinders in confined fluid, ASME Journal of Applied
gramme for the minimisation of vibration in a U-tube Mechanics 44 (1977) 213-217.
steam generator bundle for an LMFBR, Paper 2.1, in ref. [183] S.S. Chen and J.A. Jendrzejczyk, Experiments on fluide-
[63] (1982). lastic vibration of cantilevered tube bundles, ASME Jour-
[166] G.S. Whiston and G.D. Thomas, Whirling instabilities in nal of Mechanical Design, 100 (1978) 540-548.
heat exchanger tube arrays, Journal of Sound and Vibra- [184] S.S. Chen, J.A. Jendrzejczyk and M.W. Wambsganss, An
tion 81 (1982) 1-31. experimental and theoretical investigation of coupled
[167] H. Tanaka and S. Takahara, Unsteady fluid dynamic vibration of tube banks, in: Fluid-Structure Interaction
force on tube bundle and its dynamic effect on vibration, Phenomena in Pressure Vessel and Piping Systems, Eds.
in ref. [66] (1980) pp. 77-92. M.K. Au-Yang and S.J. Brown, pp. 19-36 (ASME, New
[168] H. Tanaka and S. Takahara, Fluid elastic vibration of York, 1977).
tube array in cross flow, Journal of Sound and Vibration [185] S. Levy and J.P.D. Wilkinson, Calculation of added
77 (1981) 19-37. water mass effects in reactor system components, Trans.
[169] H. Tanaka, S. Takahara and K. Ohta, Flow-induced 3rd Int. Conf. on Structural Mechanics in Reactor Tech-
nology, Paper F2/5, London, U.K., 1975.
 M.P. Pa~doussis / Flow-induced vibrations in reactors 59

[186] M.P. Paldoussis, S. Suss and M. Pustejovksy, Free vibra- [203] E. Ohlmer, S. Russo and R. Schwemmle, Investigation of
tion of cylinders in liquid-filled channels, Journal of an analytical model for parallel flow induced rod vibra-
Sound and Vibration 55 (1977) 443-459. tions, Nuclear Engineering and Design 22 (1972) 272-289.
[187] M.P. Paidoussis and P. Besanc~.on, Dynamics of arrays of [204] M.P. Paidoussis, Dynamics of cylindrical structures sub-
cylinders with internal and axial external flow, Journal of jected to axial flow, Journal of Sound and Vibration 29
Sound and Vibration 76 (1981) 361-380. (1973) 365-385.
[188] C.I. Yang and T.J. Moran, Calculations of added mass [205] W.H. Lin and M.W. Wambsganss, Analytical modeling
and damping coefficients from hexagonal cylinders in a of the buffeting of a rod in axial flow, General Electric
confined viscous fluid, in ref. [65] (1979) pp. 97-103. Report GEAP-24383 (DOE/ET/34209-24), San Jose,
[189] C.I. Yang, and T.J. Moran, Finite-element solution of Calif. (Dec. 1981).
added mass and damping of oscillation of rods in viscous [206] M.P. Paidoussis, Dynamics of flexible slender cylinders
fluids, ASME Journal of Applied Mechanics 46 (1979) in axial flow. Part 1: Theory, Journal of Fluid Mechanics
519-523. 26 (1966) 717-736.
[ 190] J. Planchard, Eigenfrequencies of a tube bundle placed in [207] M.P. Paidoussis, Dynamics of flexible slender cylinders
a confined fluid, Computer Methods in Applied Mecha- in axial flow. Part 2: Experiments, Journal of Fluid
nics and Engineering 30 (1982) 75-93. Mechanics 26 (1966) 737-751.
[191] U. Schumann, Homogenized equations of motion for rod [208] H.J. Connors, S.J. Savorelli and N.F.A. Kramer, Hydro-
bundles in fluid with periodic structure, Ingenieur-Archiv dynamic damping of rod bundles in axial flow, in ref. [69]
50 (1981) 203-216. (1982) pp. 109-124.
[192] L.N. Carlucci, Damping and hydrodynamic mass of a [209] J. Kadlec and E. Ohlmer, On the reproducibility of the
cylinder in simulated two-phase flow, ASME Journal of parallel-flow induced vibration of fuel pins, Nuclear En-
Vibration, Acoustics, Stress and Reliability in Design 105 gineering and Design 17 (1971) 355-360.
(1983) 83-89. [210] T.M. Mulcahy, T.T. Yeh, A.J. Miskevics, Turbulence and
[193] L.N. Carlucci and J.D. Brown, Experimental studies of rod vibrations in an annular region with upstream dis-
damping and hydrodynamic mass of a cylinder in con- turbances, Journal of Sound and Vibration 69 (1980)
fined two-phase flow, ASME Journal of Mechanical De- 59-69.
sign 104 (1982) ASME Paper 81-DET-27. [211] W.H. Lin, M.W. Wambsganss and J.A. Jendrzejczyk,
[194] S.S. Chen, Parallel flow-induced vibrations and instabili- Wall pressure fluctuations within a seven-rod array, Gen-
ties of cylindrical structures, Shock and Vibration Digest eral Electric Report GEAP-24375 (DOE/ET/34209-20),
6 (1974) 1-11. San Jose, Calif. (Nov. 1981).
[195] M.P. Paidoussis, Vibration of cylindrical structures in- [212] M.P. Paidoussis and J.O. Gagnon, Vibration of clustered
duced by axial flow, ASME Journal of Engineering for cylinders in axial flow, Paper 1.3, in ref. [63] (1982).
Industry 96 (1974) 547-552. [213] A.D. Lane et al., The measurement and prediction of
[196] M.P. Paidoussis, The dynamical behaviour of cylindrical vibration in CANDU-PHW fuel and channel assemblies,
structures in axial flow, Annals of Nuclear Energy 1 Paper 1.1, in ref, [62] (1978).
(1974) pp. 83-106. [214] D.V. Jagannath and M.P. Paidoussis, Solid-fluid interac-
[197] M.W. Wambsganss and S.S. Chen, Tentative design guide tions in the vibration of nuclear fuel bundles, Paper 1.2,
for calculating the vibration response of flexible cylindri- in ref. [62] (1978).
cal elements in axial flow, Argonne National Laboratory [215] M.J. Pettigrew and M.P. Paldoussis, Dynamics and sta-
Report ANL-ETD-71-07 (1971). bility of flexible cylinders subjected to liquid and two-
[198] Y.S. Shin and M.W. Wambsganss, Flow-induced vibra- phase axial flow in confined annuli, Paper D2/6, Trans.
tion in LMFBR steam generators: a state-of-the-art re- 3rd Int. Conf. on Structural Mechanics in Reactor Tech-
view, Nuclear Engineering and Design 40 (1977) 235-284. nology, London (1975).
[199] J.R. Reavis, Vibration correlation for maximum fuel-ele- [216] Y.W. Shin, Two-phase flow-induced vibrations of rods in
ment displacement in parallel turbulent flow, Nuclear parallel flow: a state-of-the-art review, General Electric
Science and Engineering 38 (1969) 63-69. Report GEAP-24148, San Jose, Calif. (Oct. 1978).
[200] D.J. Gorman, Analytical and experimental investigation [217] M.P. Paidoussis and M. Ostoja-Starzewski, Dynamics of
of the vibration of cylindrical reactor fuel elements in a flexible cylinder in subsonic axial flow, AIAA Journal
two-phase parallel flow, Nuclear Science and Engineering 19 (1981) 1467-1475.
44 (1971) 277-290. [218] F. Hara and O. Kohgo, Added mass and damping of a
[201] S.S. Chen and M.W. Wambsganss, Parallel flow-induced vibrating rod in a two-phase air-water mixed fluid, in ref.
vibration of fuel rods, Nuclear Engineering and Design [69] (1982) 1-8.
18 (1972) 253-278. [219] D. Basile, J. Faure and E. Ohlmer, Experimental study
[202] L. Cedolin et al., Vibrations induced by the two-phase on the vibrations of various fuel rod models in parallel
(gas & liquid) coolant flow in the power channels of a flow, Nuclear Engineering and Design 7 (1968) 517-534.
pressure tube type nuclear reactor, Paper E4/5, Trans. [220] M.P. Paidoussis, An experimental study of vibration of
1st Int. Conf. on Structural Mechanics in Reactor Tech- flexible cylinders induced by nominally axial flow,
nology, Berlin, 1971. Nuclear Science and Engineering 35 (1969) 127-138.
60 M.P. Paidoussis / Flow-induced vibrations in reactors

[221] Y.N. Chen, Turbulence-induced vibration of tube bundle [231] M.P. Paidoussis and S. Suss, Stability of a cluster of
heat exchangers with cross and parallel flow. Part 1: flexible cylinders in bounded axial flow, ASME Journal
parallel flow, in ref. [64] (1970) pp. 57-66. of Applied Mechanics 44 (1977) 401-408.
[222] C.D. Henry, G.E. Lee and M.W. Wambsganss, Flow- [232] M.P. Paidoussis, The dynamics of clusters of flexible
induced vibrations in CGFR core components, in ref. cylinders in axial flow: theory and experiments, Journal
[65] (1979) pp. 83-90. of Sound and Vibration 65 (1979) 391-417.
[223] M.P. Paidoussis, Stability of flexible slender cylinders in [233] M.P. Paidoussis, L1.R. Curling and J.O. Gagnon, Experi-
pulsatile axial flow, Journal of Sound and Vibration 42 ments on fluidelastic instability of cylinder clusters in
(1975) 1-11. axial flow, ASME Journal of Fluids Engineering 104
[224] F. Hara and T. Yamashita, Parallel two-phase-flow-in- (1982) 342-349.
duced vibrations in fuel pin model, Journal of Nuclear [234] M.P. Paidoussis, Dynamics of fuel strings in axial flow,
Science and Technology 15 (1978) 346-354. Annals of Nuclear Energy 3 (1976) 19-30.
[225] M.P. Paidoussis, N.T. Issid and M. Tsui, Parametric [235] M.P. Paidoussis, Mathematical model for the dynamics
resonance oscillations of flexible slender cylinders in of an articulated string of fuel bundles in axial flow,
harmonically perturbed axial flow. Part 1: Theory, ASME Paper D2/5", Trans. 3rd Int. Conf. on Structural Mecha-
Journal of Applied Mechanics 47 (1980) 709-714. nics in Reactor Technology, London, 1975.
[226] M.P. Paidoussis, N.T. Issid and M. Tsui, Parametric [236] V.A. Mason et al., Dynamics of nuclear fuel assemblies
resonance oscillations of flexible slender cylinders in in vertical flow channels: computer modelling and associ-
harmonically perturbed axial flow. Part 2: Experiments, ated studies, Atomic Energy of Canada Ltd., Report
ASME Journal of Applied Mechanics 47 (1980) 715-719. AECL-5976 (Oct. 1978).
[227] M.P. Paiddoussis and M.J. Pettigrew, Dynamics of flexi- [237] K. Hennig, E. Platen and G. Grunwald, Treatment of
ble cylinders in axisymmetricaUy confined axial flow, flow-induced pendulum oscillations, in ref. [70] (1980)
ASME Journal of Applied Mechanics 46 (1979) 37-44. pp. 140-143.
[228] M.J. Hannoyer and M.P. Paidoussis, Instabilities of tubu- [238] M. Parkin, Flow-induced vibration problems in gas cooled
lar beams simultaneously subjected to internal and exter- reactors, in ref. [70] (1980) pp. 126-139.
nal axial flows, ASME Journal of Mechanical Design 100 [239] M.P. Paidoussis, Flow-induced vibrations in heat-ex-
(1978) 328-336. changer and reactor components--critical unresolved
[229] M.J. Hannoyer and M.P. Paidoussis, Dynamics of slender problems, in ref. [70] (1980) pp. 829-832.
tapered beams with internal and external axial flow, Part [240] M.M. Zdravkovich, Review and classification of various
1: Theory, ASME Journal of Applied Mechanics 46 aerodynamic and hydrodynamic means for suppressing
(1979) 45-51. vortex shedding, Journal of Wind Engineering and In-
[230] M.J. Hannoyer and M.P. Paidoussis, Dynamics of slender dustrial Aerodynamics 7 (1981) 145-189.
tapered cylinders with internal and external axial flow.
Part 2: Experiments, ASME Journal of Applied Mecha-
nics 46 (1979) 52-57.