The Effects of Bundle Geometry

M. J. Pettigrew1
Mem. ASME on Heat Exchanger Tube Vibration
C. E. Taylor
Atomic Energy of Canada Limited, Chalk River,
in Two-Phase Cross Flow
Chalk River Laboratories, Many shell-and-tube heat exchangers operate in two-phase flows. This paper presents the
Ontario KOJ 1JO, Canada results of a series of experiments done on tube bundles of different geometries subjected
to two-phase cross flow simulated by air-water mixtures. Normal (30 deg) and rotated (60
deg) triangular, and normal (90 deg) and rotated (45 deg) square tube bundle configura-
B. S. Kim tions of pitch-to-diameter ratio of 1.2 to 1.5 were tested over a range of mass fluxes from
Korea Power Engineering Company, Inc., 0 to 1000 kg/(m2s) and void fraction from 0 to 100 percent. The effects of tube bundle
Taejon, Korea geometry on vibration excitation mechanisms such as fluidelastic instability and random
turbulence, and on dynamic parameters such as damping and hydrodynamic mass are
discussed. 关DOI: 10.1115/1.1388236兴

1 Introduction single flexible tube was in the interior of the tube bundle. For
brevity, these tube bundles are henceforth, respectively, called
Many shell-and-tube heat exchangers operate in two-phase
‘‘flexible’’ and ‘‘rigid’’ tube bundles in this paper.
flow; for example, boilers, condenser, and nuclear steam genera-
The tube vibration response was measured with strain gages
tors. Most of the research so far has been done in support of
installed inside the tubes near the cantilevered end. Upstream,
nuclear steam generators with pitch-to-diameter ratios, P/D, from
interior, and downstream tubes were instrumented, as shown in
1.4 to 1.6. This work is outlined in a few papers such as Pettigrew
Fig. 1. Two pairs of diametrically opposite strain gages were in-
and Taylor 关1兴, Pettigrew et al. 关2,3兴, Axisa et al. 关4兴, Nakamura
stalled at 90 deg from each other to measure vibration in the flow
et al. 关5,6兴, and Feenstra et al. 关7兴. However, in other industries,
direction 共drag兲 and in the direction normal to the flow 共lift兲. The
such as the petrochemical industry, heat exchangers with smaller
strain gages were connected to bridge amplifiers and to appropri-
P/D, i.e., 1.2 to 1.4, are also used. This paper presents results of
ate data processing equipment.
experimental studies done on tube bundles of P/D of 1.22, 1.32,
and 1.47 subjected to air-water cross flow. Normal 共30 deg兲 and
rotated 共60 deg兲 triangular, and normal 共90 deg兲 and rotated 共45
deg兲 square tube bundle configurations were tested over a range of 3 Test Procedure and Results
mass fluxes and void fractions. The effects of tube bundle con- All two-phase fluid properties, such as density, mass flux, and
figurations and of P/D on vibration excitation mechanisms, such void fraction, are taken as homogeneous in this paper. For ex-
as fluidelastic instability and random turbulence, and dynamic pa- ample, the homogeneous void fraction, ␧ g , is calculated from the
rameters, such as damping and hydrodynamic mass, are discussed.
volumetric flow rate of air, V̇ g and liquid, V̇ l , as
The effect of two-phase flow regime and, in particular, intermit-
tent flow is considered. ␧ g ⫽V̇ g / 共 V̇ g ⫹V̇ l 兲 (1)
This is convenient as it is well defined. In reality, the flow in a
2 Experimental Considerations tube bundle is far from homogeneous. This is discussed in a com-
Cantilever tube bundles of realistic heat exchanger configura- panion paper 关8兴, in which a number of nonhomogeneous two-
tions were subjected to two-phase flow simulated by air-water phase flow correlations are explored.
mixtures in a test loop. Mass fluxes, ṁ, from 0 to 1000 kg/共m2s兲 It is convenient to use the pitch velocity, U p 共sometimes called
and void fractions, ␧ g , from 0 to 100 percent were simulated. The the reference gap velocity兲, to correlate vibration data in tube
tube bundles were made of some 25 to 30 stainless tubes of 12.7 bundles. It is defined as
mm diameter and 600 mm length resulting in a tube natural fre-
U p ⫽U ⬁ P/ 共 P⫺D 兲 (2)
quency of roughly 30 Hz, which is typical for heat exchangers.
Random turbulence, fluidelastic instability, and damping informa- where U ⬁ is the homogeneous free-stream velocity. Similarly, the
tion were deduced from the vibration response. Experimental de- pitch mass flux
tails and measurement methods are described in 关2兴.
For most tests, all tubes in the bundle were flexible and free to ṁ p ⫽ṁ ⬁ P/ 共 P⫺D 兲 ⫽ ␳ U p (3)
vibrate. However, tests were also done with one flexible tube sur- The tube bundles were subjected to increasing mass flux until
rounded by rigid tubes to avoid the effect of hydrodynamic cou- the vibration amplitude was sufficiently high to indicate the onset
pling between the tubes. This was useful to study damping and of fluidelastic instability. During a given test, the void fraction
random turbulence excitation. This was achieved by supporting was kept constant. Vibration excitation mechanisms and damping
the free-end of the tubes with a special insert fixed to the end were deduced from the tube bundle vibration response. Typical
flange. This increased the frequency of the tubes to 160 Hz. Thus, vibration response spectra are shown in Fig. 2. As expected, the
they were essentially rigid compared to the flexible tube. The tubes were vibrating at roughly their natural frequency. Typical
vibration response results are presented as RMS vibration re-
1
Currently BWC/AECL/NSERC Chair of Fluid-Structure Interaction, Ecole Poly- sponse at the tube free-end versus mass flux curves in Fig. 3. The
technique, Montréal, QC H3C 3A7, Canada, e-mail: [email protected] difference in vibration response between the flexible bundle and
Contributed by the Pressure Vessels and Piping Division and presented at the 7th
International Conference on Fluid-Induced Vibration, Lucerne, Switzerland, June
the rigid bundle is small at mass fluxes below instability. This
19–22, 2000. Manuscript received by the PVP Division, March 13, 2000; revised means that random turbulence excitation is not much affected by
manuscript received May 22, 2001. Editor: S. Y. Zamrik. the motion of surrounding tubes. On the other hand, fluidelastic

414 Õ Vol. 123, NOVEMBER 2001 Copyright © 2001 by ASME Transactions of the ASME

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jpvtas/28412/ on 01/31/2017 Terms of Use: http://www.asme.org/a

 Fig. 2 Typical vibration response spectra for normal-square
tube bundle, P Õ D Ä1.47

4 Hydrodynamic Mass
Fig. 1 Sketch of test section and tube bundle configurations Hydrodynamic mass is defined as the equivalent external mass
of fluid vibrating with the tube. As discussed in Carlucci and
Brown 关9兴, the hydrodynamic mass, m h , is related to the tube
instability occurs at lower mass fluxes for the flexible bundle. This natural frequency, f, in the two-phase mixture; thus
indicates the importance of hydrodynamic coupling with the mo-
tion of adjacent tubes at instability. m h ⫽m t 关共 f g / f 兲 2 ⫺1 兴 (4)

Fig. 3 Vibration response: comparison between flexible and rigid tube bundle: rotated-square bundle P Õ D Ä1.22: P-1, P-8, P-11;
normal-square bundle, P Õ D Ä1.47: O-1, O-8, O-11

Journal of Pressure Vessel Technology NOVEMBER 2001, Vol. 123 Õ 415

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jpvtas/28412/ on 01/31/2017 Terms of Use: http://www.asme.org/a

 Fig. 5 Total damping ratio

Fig. 4 Comparison between experiment and theory on hydro-

dynamic mass
flow was obtained from the rigid tube bundle tests to avoid prob-
lems related to hydrodynamic coupling as discussed in Pettigrew
where m t is the mass of the tube alone and f g is the natural et al. 关2兴. Damping values were deduced from the tube vibration
frequency in air. The tube frequency is measured at mass fluxes response using the random vibration method. It is assumed that
sufficiently below fluidelastic instability. two-phase turbulence excitation is essentially broadband random.
We found the tube frequency to be relatively independent of Thus, the vibration response spectrum is essentially the frequency
mass flux below instability. Significant shifts in tube frequency are response function from which damping values can be extracted.
often observed at instability. The results are presented in the form Damping in two-phase flow is very complicated. Two-phase
of the ratio of hydrodynamic mass in two-phase flow over pre- damping is highly dependent on void fraction and flow regime.
dicted hydrodynamic mass in liquid, m h /m hl , in Fig. 4. Furthermore, damping measurements must be obtained under
The lines in the figures represent the predicted hydrodynamic some flow since it is impossible to maintain a stagnant two-phase
mass using the formulation 关2兴 mixture. As a practical approach, we have taken the average of

m h⫽ 冉 冊冋
␳␲ D 2
4
共 D e /D 兲 2 ⫹1
共 D e /D 兲 2 ⫺1
册 (5)
several damping values at mass fluxes roughly half that required
for instability. We found that damping is relatively independent of
mass flux at flow rates significantly below instability.
where ␳ is the homogeneous density of the two-phase mixture and The damping results are outlined in Fig. 5, which shows that
D e is an equivalent diameter to model confinement due to the damping is very dependent on void fraction. Damping is maxi-
surrounding tubes. For a tube inside a triangular tube bundle, mum between 40 and 80 percent void fraction. Figure 5 shows the
D e /D⫽(0.96⫹0.50 P/D) P/D 关10兴. Similarly, for a square tube total damping ratio which includes structural damping, viscous
bundle, we found that D e /D⫽(1.7⫹0.56 P/D) P/D. The agree- damping, and a two-phase component of damping as explained by
ment between experimental results and the proposed formulation Pettigrew and Taylor 关11兴.
is good except at void fraction above 80 percent. As explained The two-phase component of damping may be expressed by
later, this is related to intermittent flow regimes. Figure 4 also
shows that the proposed formulation to model confinement over a
realistic range of P/D is reasonable. ␨ TP⫽A 冉 冊
␳l D2
m
共 f 共 ␧ g 兲兲再 关 1⫹ 共 D/D e 兲 3 兴
关 1⫺ 共 D/D e 兲 2 兴 2 冎 (6)

5 Damping where A is a constant, ␳ l is the liquid phase density, m is the tube

Tube damping measurements were taken in air. The damping mass per unit length including the hydrodynamic mass, m h , and
ratio, ␨, was found to be less than 0.20 percent, which is small f (␧ g ) is a function of void fraction. The term 兵 关 1
compared to damping in two-phase flow. Damping in two-phase ⫹(D/D e ) 3 兴 / 关 1⫹(D/D e ) 2 兴 2 其 in Eq. 共6兲 is a confinement factor.

416 Õ Vol. 123, NOVEMBER 2001 Transactions of the ASME

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jpvtas/28412/ on 01/31/2017 Terms of Use: http://www.asme.org/a

 Fig. 6 Effect of void fraction on two-phase damping in cross
flow: two-phase damping ratio normalized for P Õ D using Eq. „6…

The results for the two-phase component of damping were then

normalized to take into account the effect of confinement due to
surrounding tubes by using the foregoing confinement factor.
The normalized two-phase damping ratios are presented in Fig.
6 which shows that the foregoing factor is a reasonable formula-
tion of the confinement due to P/D. As expected, greater confine-
ment due to smaller P/D’s increases damping. On the other hand,
the effect of tube bundle configuration for a given P/D is not
large, as shown in Fig. 5. From a practical design point of view, it
is probably not worth using different values for the different
bundle configurations. Fig. 7 Fluidelastic instability in two-phase cross flow: effect of
PÕD
6 Fluidelastic Instability
Fluidelastic instability may be formulated in terms of a dimen-
sionless flow velocity, U p / f D, and a dimensionless mass- were several other indications of intermittent flow regime above
damping term, 2 ␲␨ m/ ␳ D 2 . For the simple case of a tube bundle ⬃80 percent void fraction such as low-frequency vibration re-
subjected to uniform cross flow over its entire length sponse in the drag direction 共see Fig. 2兲 and visual observation of
intermittent flow regime in the test section. Also, some evidence
U p / f D⫽K 共 2 ␲␨ m/ ␳ D 2 兲 n (7) of a different flow regime could be seen in the hydrodynamic
where the pitch velocity, U p ⫽ṁ p / ␳ , in two-phase flow, f is the mass and damping results.
tube natural frequency in the two-phase mixture at mass flux, ṁ p , Intermittent flow regimes should be avoided in properly de-
significantly below instability, m is the mass per unit length in- signed heat exchangers. Ignoring the intermittent flow regime re-
cluding hydrodynamic mass, m h , ␳ is the homogeneous density of sults, the instability factor, K, can be deduced from the data. Fig-
the two-phase mixture, ␨ is the total damping ratio, and n is an ure 9 illustrates the effect of P/D on fluidelastic instability. The
exponent which is often around 0.5 as discussed by Pettigrew and effect of P/D appears to be related to the dimensionless flow path,
Taylor 关12兴. Using this relationship and knowing the critical mass ( P⫺D)/D. This suggests a formulation for a reasonably conser-
flux for instability, we can now deduce the fluidelastic instability vative design guideline to take into account smaller P/D
factor, K, and the exponent, n. The results for triangular bundles K⫽4.76共 P⫺D 兲 /D⫹0.76 (8)
with different P/D’s are shown in Fig. 7. It shows that fluidelastic
instability occurs at significantly lower flow velocities for smaller for P/D ratios between 1.22 and 1.47, as shown in Fig. 9.
P/D’s. It also shows two regions of instability. The first is a It should be noted that all tube bundles tested are similarly
region where the mass-damping parameter exponent is roughly affected by P/D. However, the rotated triangular 共60 deg兲 tube
0.5, as expected. This region occurs at void fractions below ap- bundle configuration was found to be the most susceptible to flu-
proximately 80 percent. For this region, fluidelastic instability idelastic instability. Interestingly, in single-phase flow, not all tube
may be formulated with Eq. 共7兲, where the instability factor is bundles are similarly affected by P/D as shown in Fig. 10 for
deduced from the experimental results. rotated triangular tube bundles. This observation is still in search
There appears to be another instability region at void fractions of an explanation.
above ⬃80 percent, where the slope of the instability curve is
considerably lower. This behavior is explained in terms of flow
regime change. As shown in Fig. 8, the transition between the two 7 Random Turbulence Excitation
fluidelastic instability regions corresponds to the transition be- For a given void fraction, the vibration response of the tube
tween continuous flow 共spray, bubbly兲 and intermittent flow. Fig- bundle generally increases with mass flux until fluidelastic insta-
ure 8 is a Grant flow regime map for two-phase cross flow. There bility is reached 共see Fig. 3兲. The vibration response below insta-

Journal of Pressure Vessel Technology NOVEMBER 2001, Vol. 123 Õ 417

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jpvtas/28412/ on 01/31/2017 Terms of Use: http://www.asme.org/a

 Fig. 10 Effect of P Õ D on fluidelastic instability constant for
rotated triangular tube bundle in single-phase cross flow †12‡

of the random excitation forces per unit length can be developed

from random vibration theory. The fundamental mode of a canti-
levered tube subjected to uniformly distributed and spatially cor-
related forces over its entire length, y 2 ( l ), is expressed by

y 2 共 l 兲 ⫽0.613S F 共 f 兲 / 共 16␲ 3 f3 ␨ m 2 兲 (9)

Thus, it is easy to deduce S F ( f ) from the vibration response
with Eq. 共9兲. The assumption that the random turbulent excitation
is spatially correlated is not strictly correct. However, it is conve-
nient to compare the vibration response of the different bundle
configurations. It is a reasonable approach providing that the spa-
Fig. 8 Flow regime for different bundle configurations tial correlation of the random turbulence excitation forces can be
assumed similar for all the tube bundles.
A relationship between the vibration response and the mass flux
bility is attributed to random turbulence excitation. Our approach was established by doing a linear regression analysis on the ex-
is to deduce the random turbulence excitation forces from the perimental data. By choosing only the data points corresponding
vibration response. to mass fluxes below the threshold for fluidelastic instability, a
The relationship between the mean square amplitude of the tube mass flux exponent, b, was determined for the relation (y 2 ) 0.5
vibration response, (y 2 ), and the power spectral density, S F ( f ), ⬀ṁ bp . The scatter in the results was significant, but a tendency
towards a value of b⫽1 was found in the void fraction range of
25 to 90 percent. At void fractions below 25 percent, the plots
were extremely difficult to interpret, due to the presence of some
periodic wake shedding peaks, as discussed in Taylor et al. 关13兴.
In this paper, an exponent of b⫽1 has been used to collapse the
random turbulence data. This allowed for the calculation of a sim-
plified normalized power spectral density 共NPSD兲 of the excita-
tions forces formulated by

S F共 f 兲
NPSD⫽ (10)
共 ṁ p D 兲 2
The normalized power spectral densities were plotted against
void fraction in Fig. 11. It should be pointed out that this NPSD is
unfortunately not dimensionless and is only applicable at the re-
sponse frequency of the tubes that is at roughly 30 Hz. A com-
parison between P/D⫽1.22 and 1.47 indicates that random tur-
bulence forces are not greatly affected by P/D. This is not
entirely unexpected since the response is not much affected by the
motion of surrounding tubes as shown in Fig. 3. Figure 11 illus-
trates that the random turbulence levels are practically the same
for triangular and square bundles for the worst orientation. In real
heat exchangers, all flow orientations are possible for a given tube
Fig. 9 Effect of P Õ D on fluidelastic instability constant in two- bundle. Thus, the worst flow orientation must be assumed in de-
phase cross flow sign analyses. The rotated-triangular data is marginally higher

418 Õ Vol. 123, NOVEMBER 2001 Transactions of the ASME

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jpvtas/28412/ on 01/31/2017 Terms of Use: http://www.asme.org/a

 3 Fluidelastic instabilities have been observed for all tube
bundles in two-phase flow. There appears to be two distinct re-
gions of instability. The first follows the classical fluidelastic in-
stability model and corresponds to continuous 共i.e., bubbly, froth,
etc.兲 flows. The instability behavior in the second region is quite
different and is related to bundle-scale intermittent flow. For a
given P/D in continuous flows, the difference in fluidelastic in-
stability behavior between the different bundle configurations is
not large. However, the rotated triangular 共60 deg兲 tube bundle
configuration was found to be the most susceptible to fluidelastic
instability. The effect of P/D is significant. The critical velocity
for instability for bundles of P/D of 1.47 is roughly 50 percent
larger than for P/D of 1.22. The effect of P/D on fluidelastic
instability is probably related to the flow path geometry. The dif-
ference in behavior between single and two-phase flow for some
tube bundle configurations is puzzling. Tube-scale flow regime
effects is the likely explanation.
4 Random turbulence excitation is not much affected by pitch-
to-diameter ratio or by neighboring tube vibration. The vibration
response is roughly proportional to the mass flux. The effect of
tube bundle configuration was also found to be small.

Nomenclature
A ⫽ constant
b ⫽ exponent
D ⫽ tube diameter 关m兴
De ⫽ equivalent diameter of flow boundary 关m兴
f ⫽ frequency关Hz兴
K ⫽ fluidelastic instability factor
l ⫽ cylinder length 关m兴
m ⫽ total mass per unit length (m t ⫹m h ) 关kg/m兴
ṁ ⫽ mass flux 关kg/共m2 s兲兴
n ⫽ exponent of mass-damping parameter
P ⫽ tube pitch 关m兴
S F( f ) ⫽ one-sided power spectral density of excitation force
Fig. 11 Normalized power spectral density of random turbu- per unit length 关共N/m兲2s兴
lence excitation at 30 Hz U ⫽ fluid velocity 关m/s兴
V ⫽ volume rate of flow 关m3/s兴
X ⫽ Martinelli parameter
y(x) ⫽ response 关m兴
than the others and the rotated-square seems to be lower; but for
simplicity, a single envelope was drawn over all of the data. The y 2 (x) ⫽ mean-square amplitude response
curve is defined as ␧g ⫽ void fraction
␨ ⫽ damping ratio
NPSD⫽10共 0.03␧ g ⫺5 兲 , for 25⬍␧ g ⬍90 percent (11) ␳ ⫽ fluid density 关kg/m3兴
The slope of the curve changes drastically at void fractions Subscripts
above ⬃90 percent. This is attributed to intermittent flow regime
g ⫽ gas or vapor
as discussed earlier. It shows that such intermittent flow regimes
h ⫽ hydrodynamic
should be avoided in properly designed heat exchangers.
l ⫽ liquid
p ⫽ pitch
8 Discussion and Conclusions t ⫽ tube or cylinder
⬁ ⫽ free stream
Vibration experiments on realistic heat exchanger tube bundles,
of both triangular and square configurations and of P/D ratios
between 1.22 and 1.47, subjected to two-phase 共air-water兲 cross References
flow have shown the following trends: 关1兴 Pettigrew, M. J., and Taylor, C. E., 1994, ‘‘Two-Phase Flow-Induced Vibration:
An Overview,’’ ASME J. Pressure Vessel Technol., 116, pp. 233–253.
1 Hydrodynamic mass is roughly related to the homogeneous 关2兴 Pettigrew, M. J., Taylor, C. E., and Kim, B. S., 1989, ‘‘Vibration of Tube
two-phase mixture density and to the confinement due to sur- Bundles in Two-Phase Cross-Flow—Part 1; Hydrodynamic Mass and Damp-
ing’’ ASME J. Pressure Vessel Technol., 111, pp. 466 – 477.
rounding tubes expressed in terms of the pitch-to-diameter ratio, 关3兴 Pettigrew, M. J., Tromp, J. H., Taylor, C. E., and Kim, B. S., 1989, ‘‘Vibration
P/D. As expected, a lower P/D results in a higher hydrodynamic of Tube Bundles in Two-Phase Cross-Flow—Part 2: Fluidelastic Instability,’’
mass. ASME J. Pressure Vessel Technol., 111, pp. 478 – 487.
2 The damping characteristics of all tube bundles are generally 关4兴 Axisa, F., Villard, B., and Sundheimer, P., 1986, ‘‘Flow-Induced Vibration of
Steam Generator Tubes,’’ Electric Power Research Institute Report EPRI-
similar. The effect of void fraction is dominant. Damping is maxi- NP4559.
mum between 40 and 80 percent void fraction. The effect of void 关5兴 Nakamura, T., Fujita, K., Kawanish, K., Yamaguchi, N., and Tsuge, A., 1995,
fraction is attributed to flow regime, although the underlying en- ‘‘Study on the Vibrational Characteristics of a Tube Array Caused by Two-
ergy dissipation mechanism is not yet fully understood. Damping Phase Flow—Part I: Random Vibration,’’ J. Fluids Struct., 9, pp. 519–538.
关6兴 Nakamura, T., Fujita, K., Kawanish, K., Yamaguchi, N., and Tsuge, A., 1995,
increases with confinement due to decreasing P/D. This behavior ‘‘Study on the Vibrational Characteristics of a Tube Array Caused by Two-
is similar to that of viscous damping in liquid flow. The effect of Phase Flow—Part II: Fluidelastic Vibration,’’ J. Fluids Struct., 9, pp. 539–562.
tube bundle configuration is minor. 关7兴 Feenstra, P. A., Weaver, D. S., and Judd, R. L., 1996, ‘‘Damping and Fluidelas-

Journal of Pressure Vessel Technology NOVEMBER 2001, Vol. 123 Õ 419

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jpvtas/28412/ on 01/31/2017 Terms of Use: http://www.asme.org/a

 tic Instability of a Tube Array in Two-Phase R-11 Cross Flow,’’ Proc., Symp. and Piping Conf., San Antonio, TX, June, ASME Publication H00316, pp.
Flow-Induced Vibration-1996, Montreal. ASME PVP, Vol. 328, pp. 89–102. 17–26.
关8兴 Taylor, C. E., and Pettigrew, M. J., 2000, ‘‘Effect of Flow Regime and Void 关11兴 Pettigrew, M. J., and Taylor, C. E., 1997, ‘‘Damping of Heat Exchanger Tubes
Fraction on Tube Bundle Vibration,’’ Proc., 7th Int. Conf. Flow-Induced Vi- in Two-Phase Flow,’’ Proc., 4th Int. Symp. FSI, AE & FIV⫹N. Dallas, TX,
November 16–21, ASME Publication AD-53.2, pp. 407– 418.
bration, 2000, Lucerne, Switzerland, June 19–22, pp. 529–536.
关12兴 Pettigrew, M. J., and Taylor, C. E., 1991, ‘‘Fluidelastic Instability of Heat
关9兴 Carlucci, L. N., and Brown, J. D., 1983, ‘‘Experimental Studies of Damping
Exchanger Tube Bundles: Review and Design Recommendations,’’ ASME J.
and Hydrodynamic Mass of a Cylinder in Confined Two-Phase Flow,’’ ASME Pressure Vessel Technol., 113, pp. 242–256.
J. Vib., Acoust., Stress, Reliab. Des., 105, pp. 83– 89. 关13兴 Taylor, C. E., Currie, J. G., Pettigrew, M. J., and Kim, B. S., 1989, ‘‘Vibration
关10兴 Rogers, R. J., Taylor, C. E., and Pettigrew, M. J., 1984, ‘‘Fluid Effects on of Tube Bundles in Two-Phase Cross-Flow—Part 3: Turbulence-Induced Ex-
Multispan Heat Exchanger Tube Vibration,’’ Proc., ASME Pressure Vessels citation,’’ ASME J. Pressure Vessel Technol., 111, pp. 488 –500.

420 Õ Vol. 123, NOVEMBER 2001 Transactions of the ASME

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jpvtas/28412/ on 01/31/2017 Terms of Use: http://www.asme.org/a