You are on page 1of 6

Fluid-Elastic Instability in a Tube

Array Subjected to Uniform and


Paul Feenstra
e-mail: feenstpa@mcmaster.ca
McMaster University,
Jet Flow
Department of Mechanical Engineering,
Hamilton, Ontario, Canada, L8S 4L7 A wind tunnel study was carried out to investigate the fluidelastic stability of a model heat
exchanger tube array subjected to a uniform cross-flow of air and a concentrated jet flow
David S. Weaver of air directed down a tube lane. The latter experiments were intended to simulate the
e-mail: weaverds@mcmaster.ca effects of a soot blower on the dynamic response of tubes which had apparently been the
McMaster University, cause of catastrophic tube failure in a heat exchanger. The experimental results showed
Department of Mechanical Engineering, that the model tube array experienced fluidelastic instability when subjected to a uniform
Hamilton, Ontario, Canada, L8S 4L7 cross-flow beyond a dimensionless pitch flow velocity which was substantially above the
maximum design flow velocity of the heat exchanger. These experiments established that
Zia Abdullah normal operating conditions could not have been responsible for the tube failures. Addi-
e-mail: zia.abdullah@weyerhaeuser.com tional experiments showed that a continuously translating nozzle dispensing a jet of air at
Weyerhaeuser Technical Center, the tubes caused some static deflection of the tubes but no serious vibrations were ob-
WTC 2H22, P.O. Box 9777, served. However, when the nozzle was fixed at one location, whereby the jet of air issued
Federal Way, WA, U.S.A. 98063-9777 directly down a tube lane, fluidelastic instability occurred in the first few tube rows. A
simplified analysis showed that the jet could cause fluidelastic instability. It can be in-
ferred that, for heat exchangers equipped with steam soot blowers, normal soot blower
operation should not cause fluidelastic instability but that a parked soot blower can cause
fatigue failure of the tubes adjacent to the impinging jet in a relatively short period of
time. 关DOI: 10.1115/1.1689356兴

Introduction way between tube rows. This heat exchanger had an in-line tube
pattern with a transverse pitch ratio of 1.8 and a streamwise pitch
It is well known that bundles of tubes, such as those used in
ratio of 2.2. It was speculated that the jet from the soot blower
boilers and heat exchangers, can be excited to vibrate excessively
caused fluidelastic instability with amplitudes sufficient to cause
when exposed to a fluid cross-flow. At relatively low cross-flow fatigue failures in a matter of a few hours. However, to the au-
velocities, small amplitude random vibrations occur due to turbu- thors’ knowledge, no study is available in the open literature
lence, and these are usually of little concern from the point of which indicates the effect of a localized jet on the fluidelastic
view of tube life. At sufficiently high flow velocities, these ran- instability of a tube array. The vast majority of the fluidelastic
dom vibrations give way to large amplitude periodic vibrations at instability data in the literature has been determined using labora-
the natural frequencies of the tubes. This phenomenon is known as tory models of single span tubes subjected to a uniform flow ve-
‘‘fluidelastic instability’’ and it is of great concern to plant opera- locity across their spans. Some work has been done on multiple
tors since it may cause rapid failure of the tubes through fatigue or span tube arrays and the effects of partial admission 共i.e., nonuni-
fretting wear at points of contact between adjacent tubes or be- form velocity distribution兲. See for example, Weaver and Goyder
tween tubes and their supports. The onset of the large amplitude 关8兴 and the references in that paper. However, none of these can be
vibrations is usually quite abrupt when the cross-flow velocity related directly to the effects of a localized jet issuing down a tube
through the tube bundle is increased above the so-called ‘‘critical lane where the flow is essentially over one side of the tubes.
velocity.’’ Since tube failures due to fatigue or fretting wear com- The purpose of this paper is to present the results of a study
monly occur at flow velocities exceeding the critical velocity, ac- designed to explain the observed tube failures. A wind tunnel
ceptance criteria for tubular heat exchangers require that the flow model was developed with a tube array geometry that scaled the
velocities never exceed the critical velocity for a particular tube heat exchanger array geometry in question exactly. Two series of
bundle. There are a number of excellent references which discuss tests were conducted. In the first test, the tube bundle was sub-
flow-induced vibrations in heat exchanger tube arrays and specifi- jected to a uniform cross-flow of air, while the tube response was
cally fluidelastic instability, such as by Au-Yang 关1兴, Blevins 关2兴, monitored with a dynamic signal analyzer. This initial test was
Chen 关3兴, Paı̈doussis 关4兴, Pettigrew and Taylor 关5兴, Price 关6兴, and designed to establish a datum case of tube vibration response un-
Weaver and Fitzpatrick 关7兴. der ‘‘uniform flow’’ conditions, which would be used as a basis
The present study was motivated by the catastrophic failure of for comparing with the subsequent tests where the tube bundle
tubes in a large heat exchanger. A second row tube parted com- was subjected to a concentrated ‘‘jet-flow’’ of air. By considering
pletely from its header while several neighboring tubes suffered the jet velocity profile and the partial admission factor, the flu-
fatigue cracks. Visual observations indicated tube scouring at the idelastic threshold of tubes subjected to the jet flow could be
location of a soot blower which had become parked for several predicted reasonably well. While the results have obvious value in
hours. Figure 1 shows a schematic illustration of a steam soot terms of explaining the case history in question, they are also of
blower and its normal path through the heat exchanger. From generic interest in that they bring to light an hitherto unknown and
photographs of the failed tubes, the scouring indicated that the jet potentially dangerous phenomenon.
from the soot blower was directed normal to the tube axes, mid-
Apparatus
Contributed by the Pressure Vessels and Piping Division for publication in the
JOURNAL OF PRESSURE VESSEL TECHNOLOGY. Manuscript received by the PVP
A model tube bundle and test section was designed and built for
Division September 29, 2003; revision received November 11, 2003. Associate Edi- McMaster University’s 2 ft wind tunnel. The tube layout geometry
tor: M. J. Pettigrew. and the mass-damping parameter were in the range of heat ex-

Journal of Pressure Vessel Technology Copyright © 2004 by ASME MAY 2004, Vol. 126 Õ 269

Downloaded From: https://pressurevesseltech.asmedigitalcollection.asme.org on 05/22/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use


Fig. 1 Path of steam soot blower lance

changers with gas cross-flows and the reduced velocity could be


varied over the range from about 10 to about 70.
The test section was designed for 78 tubes in total, 13 rows Fig. 3 Typical support arrangement for a flexible tube
wide by 6 rows deep in an in-line pattern. The layout of the tube
bundle is shown schematically in Fig. 2. This arrangement is an
exact geometric scaling of the heat exchanger array experiencing and the estimated error at 2 m/s could be as high as 10 percent.
the tube failures. The central 7 rows of tubes, shown in the dashed The gap flow velocity, V g , used in the analysis is the mean flow
box of Fig. 2, were cantilever mounted on threaded rod supports, velocity in the minimum gap between tubes in a transverse tube
which permitted the required scaling of the tube natural frequency row and is computed from the measured upstream flow velocity,
and reduced velocity well within the capacity of the wind tunnel. V u , using continuity, V g ⫽( P T /( P T ⫺D)) V u , where P T is the
Tubes labeled D1, D2, D3, and C2 were instrumented with strain transverse pitch 共45.7 mm兲 and D is the tube diameter 共25.4 mm兲.
gauges so that their flow-excited response in the transverse and The test section was equipped with a full cone air spray nozzle,
streamwise directions could be monitored with a dynamic ana- with a 15° diffusing angle, that was supplied with a 100 p.s.i. air
lyzer. These upstream tubes were monitored because they were line coupled with a control valve and a quick closing valve. The
the ones which experienced failure in the heat exchanger being nozzle was mounted in the test section on a slider device so that it
simulated. The flexible tubes were tuned to a first mode natural could be translated up to 100 mm in both the streamwise and
frequency of 12.5 Hz⫾2%. A schematic diagram of a typical transverse directions. The initial position of the nozzle tip was 200
flexible tube assembly is shown in Fig. 3. The various details for mm from the face of the first row tubes and was aimed perpen-
dimensions and other measured parameters of the model tube dicular to the tube axis near the tube free ends as illustrated in
bundle are summarized in Table 1. The remaining tubes in the test Figs. 2 and 3. This arrangement produced a jet whose core diam-
section were effectively rigid. eter was approximately equal to the transverse pitch of the tube
The wind tunnel has a uniform and flat flow velocity distribu- bundle at the first tube row. The vibration response of the tube
tion within 2% outside of the boundary layers and a turbulence array was also recorded visually using a VHS video camera lo-
intensity less than 1%. The flow velocity was measured using a cated outside the wind tunnel such that the tube tip vibrations
pitot-static probe coupled with a precision calibrated differential could be observed.
pressure transducer. The velocity measurement error is estimated
to be about 2% at 5 m/s and decreases to negligible values above Procedure
24 m/s. Velocities below 5 m/s were difficult to measure reliably Procedure for Uniform Flow Tests. The strain gauges were
carefully calibrated so that the free-end tube displacement could
be determined with the necessary sensitivity and recorded as a
percentage of the tube diameter. The strain gauge signals were

Table 1 Summary of experimental tube bundle details and


fluidelastic instability data

Parameter Quantity
tube diameter, D 0.0254 m 共1⬙兲
tube length1兲, L 0.619 m 共24.25⬙兲
tube natural frequency2兲 共1 st mode兲, f 12.6 Hz
effective tube lineal mass, m 0.161 kg/m
tube damping ratio 共tube D3兲, ␨ 1.1%
air density, ␳ 1.19 kg/m3
critical gap flow velocity, V g 7.8 m/s
mass damping parameter, m ␦ / ␳ D 2 14.4
critical reduced gap flow velocity, V red 24.3
Connor’s constant, K 6.4

Notes:
1兲
Portion of tubes exposed to the uniform flow of the wind tunnel.
2兲
To achieve the desired stiffness, tubes are mounted on cantilever support rods 共see
Fig. 2 Tube bundle layout and nozzle location Fig. 3兲.

270 Õ Vol. 126, MAY 2004 Transactions of the ASME

Downloaded From: https://pressurevesseltech.asmedigitalcollection.asme.org on 05/22/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use


Fig. 4 Amplitude response of a 2 nd row tube „tube D2… sub-
jected to uniform cross-flow of air
Fig. 5 Stability diagram for heat exchanger tube arrays „ASME
Boiler Code, 1998…

analyzed using an FFT spectrum analyzer. The r.m.s. amplitude


response of the tubes was determined using at least 30 sample
averages over a frequency range of 0 to 25 Hz and with a resolu-
which there was a sudden rise in the slope of the amplitude versus
tion of 0.125 Hz. The flow velocity was determined using a pitot-
flow velocity curve. At the FEI threshold, the frequency spectra of
static probe positioned just upstream of the test section and
the tube D2 revealed that the tubes oscillated with most of their
coupled with a calibrated pressure transducer as noted above.
energy at their natural frequency of about 12.6 Hz. This indicates
The experiments were conducted by setting the wind tunnel
strong fluid coupling, typical of fluidelastic instability rather than
velocity at the desired level and allowing sufficient time for steady
turbulent buffeting. There was a second, more significant change
state conditions to be reached. Measurements were then taken of
in slope at about 9.2 m/s which might also be interpreted as the
the flow velocity and the vibration response of the instrumented
critical flow velocity for these tubes. However, visual observation
tubes. The flow velocity was then incremented and the process
of the entire bundle indicated that other tubes deeper in the bundle
repeated until the vibrations reached a high enough level that tube
共noninstrumented兲 vibrated with greater amplitude than the moni-
clashing was imminent.
tored tubes when the flow was just above V g ⫽7.8 m/s. As stated
Procedure for Jet Flow Tests. The purpose of the jet-flow above, this is typical so it is appropriate to take the critical gap
tests was to determine if fluidelastic instability could be produced flow velocity, (V g ) cr ⫽7.8 m/s as the threshold for fluidelastic in-
locally by a concentrated jet of air, similar to that produced by a stability of the tube bundle.
steam soot blower that is used in many boilers for cleaning pur- Comparing the fluidelastic threshold flow velocities for various
poses. In this case, the experimental results were intended to be tube bundles is commonly performed by plotting the data on a
more qualitative rather than quantitative, so that most of the re- stability diagram as shown in Fig. 5. This figure is drawn from the
sults were captured on VHS video. ASME Boiler Code 关9兴 and it gives conservative guidelines to
The separation distance between the nozzle and the front face prevent fluidelastic instability in heat exchanger tube bundles. The
of the first row tubes was set at about 200 mm, at which distance abscissa is the mass damping parameter given by
the jet expanded to a diameter of about 46 mm. The core velocity
of the jet was measured at that distance 共200 mm兲 using a pitot- mass damping parameter⫽m 共 2 ␲␨ 兲 / 共 ␳ D 2 兲 , (1)
static probe and the velocity was correlated with a pressure gauge where m is the mass per unit length including internal and external
located in the air line just upstream of the nozzle. In this case, fluid added mass, ␨ is the damping ratio, ␳ is the fluid density, and
only two jet flows were tested: a sub-critical flow of about 53 m/s D is the outside tube diameter. Note that the symbol used in Fig.
and a post-critical flow of about 60 m/s. The uncertainty in these 5 for damping is that of the logarithmic decrement, ␦, where, for
core-flow velocities was about ⫾10%. The transient vibration re- light damping, ␦ ⫽2 ␲␨ . The ordinate in Fig. 5 is the critical re-
sponse of tube D2 was captured using a computer based data duced 共gap兲 velocity given by
acquisition system and plotted as a function of time to illustrate
the transient response of the tube when subjected to a jet flow. critical reduced gap velocity⫽V g / 共 f D 兲 , (2)
For the bulk of these tests, the tube vibration was captured on
VHS video tape. In this case, the nozzle was translated from side where V g is the gap flow velocity at the fluidelastic instability
to side continuously for about 10 s and then stopped in a position threshold and f is the vibration frequency. For the present model
where the jet issued down a tube lane. This was designed to simu- study, the mass damping parameter is 14.5 and the critical reduced
late the action of a steam soot blower that suddenly becomes stuck flow velocity is 24.4. This data point is plotted in Fig. 5 as the
in one position but continues to blow steam at the tubes. large open circle. In comparison with the other data around the
same mass-damping parameter, the present results are slightly
higher. This is understandable because the present array has a
Test Results for Fluidelastic Instability
pitch ratio larger than most of the data, the bulk of which corre-
Uniform Flow. The amplitude response of the tube bundle spond to laboratory studies performed on scale model arrays with
when subjected to uniform flow in the wind tunnel is shown in pitch ratios around 1.5 or less which are commonly used in the
Fig. 4 for tube D2 in the bundle. The vibration amplitude was nuclear industry. It has been observed that tube bundles with
monitored in the lift and drag directions separately. The point at smaller pitch ratios generally have lower stability thresholds than
which tube D2 became fluidelastic unstable was at a gap flow those with larger pitch ratios.
velocity of about V g ⫽7.8 m/s, which was the same for tube D3. It The straight lines in Fig. 5 are given by the so-called Connors’
is often observed that the first one or two tube rows are more equation


stable and have higher stability thresholds than tubes deeper in the
bundle. The vibration amplitudes generally increased for tubes Vg 共 2 ␲␨ 兲 m
⫽K , (3)
deeper in the bundle. The fluidelastic threshold is the point at fD ␳D2

Journal of Pressure Vessel Technology MAY 2004, Vol. 126 Õ 271

Downloaded From: https://pressurevesseltech.asmedigitalcollection.asme.org on 05/22/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use


Fig. 6 Illustration of tube vibration due to jet flow of air

where the constant of proportionality, K, is called the Connors’


constant. The ASME Boiler Code recommendation does not dis-
Fig. 8 Radial velocity profile of axisymmetric jet of air
tinguish between tube array geometric patterns or pitch ratios and
suggests a lower bound value of K⫽2.4 for all heat exchangers.
As can be seen, there is a lot of scatter in the data and this ASME
recommendation may be quite conservative for some arrays such for a few seconds and then the air jet was switched off, which
as the one considered in the present study. Indeed, the recommen- explains why the large amplitude vibrations are short-lived in this
dation is conservative by a factor of 6.4/2.4⬇2.7 for the present figure.
array. Since the heat exchanger being simulated operated In the present study, the jet flow velocity required was many
well below the ASME recommendation, it is clear that the tube times that of the uniform critical flow velocity but much less that
failures observed could not have been caused by normal operating the sonic or supersonic velocities expected from an actual soot
conditions. blower. The tests showed that a second row tube often vibrated
with significantly larger amplitudes than the 1 st or 3 rd row tubes.
Jet Flow. The tests with the jet flow revealed that localized This is consistent with the tube failures which motivated the
fluidelastic instability could also occur due to a concentrated jet of present study. It was also observed that some time is required to
air issuing down a tube lane. It was determined that the minimum produce fluidelastic instability after the tubes are exposed to a
jet flow velocity to initiate fluidelastic instability in the model tube high velocity jet-flow. This is the result of the nature of fluidelas-
array was about 55 m/s 共as measured in the core of the jet at a tic instability. The forces generating instability depend upon tube
distance of 200 mm from the nozzle tip兲. Figure 6 shows a sketch motions becoming coupled systematically with the flow and only
of the tube bundle response and relative vibration amplitudes a small amount of energy is transferred from the flow to the tube
when being subjected to an air jet with a core velocity of about 60 at each oscillation cycle. The initial tube motion is random as it is
m/s directed down the tube lane between row C and D. The tubes buffeted by the jet, which also explains why the transient time
in the 1 st, 2 nd, and 3 rd rows adjacent to the jet flow lane undergo between jet impingement and the onset of large amplitude vibra-
limit cycle oscillations, where the vibration amplitude exceeded tions is random. Thus, the transient passage of a soot blower jet is
50% of a tube diameter in some cases. Deeper into the bundle, very unlikely to produce fluidelastic instability while a parked
共i.e., 4 th to 6 th row兲 the tubes vibrated with much smaller ampli- soot blower certainly will. It follows that normal soot blower op-
tudes. In general, the tubes were most susceptible to vibration eration should not cause fatigue failures of the tubes. On the other
when the air jet was directed down the middle of a tube lane, hand, soot blowers should never be operated at a fixed location
rather than directly at a tube. There was some time delay before while impinging on a tube bank.
the high amplitude vibrations commenced, which varied depend-
ing primarily upon the jet velocity. Figure 7 shows a sample time Estimation of Fluidelastic Instability Threshold of a
trace of the transverse motion of tube D2 when subjected to the air Concentrated Jet Flow
jet 共at the same core velocity as in the previous figure兲. In this
time trace, the time delay between the starting of the jet and the In cases where it is likely that a steam soot blower might be-
onset of large tube vibrations was about 18 s. The time delay come parked, it is desirable to determine if the jet flow will cause
between the start of the air jet and the onset of the large vibrations the tubes to vibrate excessively. It is possible to estimate the
generally ranged from about 20 s down to about 6 s. It was found equivalent uniform velocity of the jet for comparison with the
that this time delay became shorter as the jet velocity was raised. critical velocity for a particular tube array. The steps are 共i兲 deter-
The time trace for the streamwise direction was very similar to the mine the radial jet velocity profile at the location where it im-
transverse direction shown in Fig. 7. In order to preserve the life pinges on the tubes; 共ii兲 determine a nominal jet diameter at the
of the model tube bundle, the tubes were only allowed to vibrate point of impingement with the first tube row; 共iii兲 calculate the
average jet flow velocity by continuity within the nominal jet area;
and 共iv兲 model the jet flow area as a partial admission flow and
calculate the equivalent uniform flow velocity. According to Ap-
pel et al. 关10兴, the nondimensional radial velocity profile of a jet,
û(r), can be closely approximated as a Gaussian curve where,
û(r)⫽exp(⫺r2/2␴ 2 ), where ␴ ⫽Cx, C⫽0.0806, and x
⫽203 mm 共distance from nozzle to the front row of tubes兲. This
radial velocity profile is illustrated in Fig. 8, where it is scaled to
the tube diameter. Since the Gaussian curve extends radially with
no bound, a nominal core diameter of the jet needs to be specified
in order to calculate the average jet velocity. In this case, we use
the criteria that the core jet will reside within the transverse pitch
of the tubes, so that 2R o ⫽46 mm. This is shown in Fig. 8 as the
dashed circle.
Fig. 7 Time trace of tube tip displacement while subjected to a The average jet velocity is calculated based upon the volume
stationary jet flow of air flux in the core area

272 Õ Vol. 126, MAY 2004 Transactions of the ASME

Downloaded From: https://pressurevesseltech.asmedigitalcollection.asme.org on 05/22/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use


concentrated jet flow of air. These experiments were designed to
simulate a heat exchanger tube bank under normal operating con-
ditions and with a soot blower parked while blowing steam down
a tube lane. The primary conclusions derived are as follows:
1. When subjected to the uniform flow of air in the wind tun-
nel, the model tube array experienced fluidelastic instability at a
critical threshold flow velocity of (V g ) cr ⫽7.8 m/s. The mass
damping parameter of this model tube array was 14.5 and the
critical reduced flow velocity was 24.4. The Connor’s constant
was found to be K⫽6.4. This compared well with previous data in
Fig. 9 Average jet velocity distribution over the tube span the literature but the stability constant was higher than recom-
mended by ASME, likely due to the relatively large pitch ratio and
the geometric pattern of this bundle. The results showed un-

û avg⫽
2
R 2o
冕 0
Ro
û 共 r 兲 rdr⫽0.635. (4)
equivocally that the tube failures experienced in the heat ex-
changer being simulated could not have been caused by normal
operating conditions.
The approximate mode shape for the lowest transverse mode of 2. When the tube bundle was subjected to the concentrated jet
the flexibly mounted tube is, ␺ (z)⫽z/L, where z is the axial of air issuing down a tube lane, fluidelastic instability occurred for
coordinate and L is the tube length. This arises from the fact that the 1 st, 2 nd, and 3 rd row tubes adjacent to the jet. This occurred
most of the bending occurs in the threaded rod support while the when the jet flow velocity was sufficiently high and when it was
acrylic tube remains relatively straight due to its greater rigidity. parked in one position for some time. However, if the air jet was
The equivalent nondimensional uniform velocity of the jet flow continuously translated from side to side, so that it did not issue
can be calculated by integrating the average jet flow velocity continuously down a single tube lane, then fluidelastic instability
times the tube mode shape as follows: did not occur. Thus, normal soot blower operation is not expected

冋冕 册
to cause fatigue failure of the tubes due to fluidelastic instability

1
冕 0.96L

0.88L
û 2avg␺ 共 z 兲 2 dz
0.5 but soot blowers should never be operated continuously at a fixed
location in a tube bank.
3. The fluidelastic threshold of a transversely impinging jet
V pn ⫽ L
⫽0.143 (5) flow was estimated by considering the concentrated jet as a partial
2
␺ 共 z 兲 dz
2 admission flow, and determining an equivalent uniform flow. In
0 this case, the equivalent uniform flow velocity at the onset of
fluidelastic instability was determined to be somewhere in be-
The expression in Eq. 共5兲 is essentially the partial admission fac- tween 7.6 and 8.6 m/s, which brackets the critical velocity mea-
tor, but in this case the 1/2 term is necessary because the jet flow sured in the uniform flow tests.
acts only on one side of the tube as opposed to both sides as
occurs under uniform flow conditions 共see Lever and Weaver
关11兴兲. The integrand in the numerator of Eq. 共5兲 is taken from 0.88 Nomenclature
L to 0.96 L 共i.e., from a to b shown in Fig. 9兲 which corresponds C ⫽ coefficient of air jet velocity profile 兵—其
to the nominal diameter of the jet acting near the free end of the D ⫽ tube diameter 兵m其
flexible tube. Finally, the equivalent uniform velocity of the jet f ⫽ frequency of vibration 兵Hz其
can be calculated by multiplying the measured core velocity times L ⫽ tube length 兵m其
the partial admission factor, V pn . This calculation is performed K ⫽ Connors’ constant 兵—其
below for two cases: the measured sub-critical jet flow of 53 m/s m ⫽ tube lineal mass including added fluid mass 兵kg/m其
and the measured post critical jet flow of 60 m/s, mt ⫽ tube lineal mass 兵kg/m其
PT ⫽ transverse pitch ratio 兵—其
sub-critical flow velocity: V eq ⫽53 m/s⫻V pn ⫽7.6 m/s r ⫽ radial coordinate of air jet 兵m其
Ro ⫽ nominal core radius of air jet 兵m其
post-critical flow velocity: V eq ⫽60 m/s⫻V pn ⫽8.6 m/s V avg ⫽ average flow velocity in the gap between tubes 兵m/s其
(6) V eq ⫽ equivalent uniform flow velocity for the onset of flu-
idelastic instability for a jet flow 兵m/s其
The results of Eq. 共6兲 agree fairly well with the critical velocity Vg ⫽ flow velocity in the gap between tubes 兵m/s其
measured from the uniform flow tests as shown in Fig. 4. The Vu ⫽ upstream flow velocity 兵m/s其
estimated sub-critical jet flow corresponds to an equivalent flow V pn ⫽ equivalent non-dimensional uniform velocity of jet
velocity of about 7.6 m/s, which resides in the stable vibration 兵—其
region of Fig. 4, while that of the post-critical jet corresponds to û(r) ⫽ non-dimensional radial jet velocity profile 兵—其
8.6 m/s which is above the critical velocity shown in Fig. 4. It û avg ⫽ non-dimensional average jet velocity in the core area
should be noted that there is an estimated error of ⫾10% in the 兵—其
measured centerline jet velocity. Moreover, some of the flow in x ⫽ axial distance from tip of air nozzle to tube 兵m其
the jet core overlaps the tube surface, and this flow would neces- z ⫽ axial coordinate of tube 兵m其
sarily pass through the tube gap thereby increasing the average ␦ ⫽ logarithmic decrement damping 兵—其
flow velocity, V pn , by a small amount. This effect has not been ␨ ⫽ damping ratio 兵—其
accounted for in this analysis, but it would serve to increase the ␳ ⫽ fluid density 兵kg/m3其
velocities calculated in Eqs. 共6兲 slightly. ␺ (z) ⫽ transverse mode shape of tube 兵—其
␴ ⫽ standard deviation of velocity profile of air jet 兵—其
Conclusions ␩ ⫽ dimensionless radius of air jet, r/x 兵—其
Flow-induced vibration experiments were conducted in a wind
tunnel to determine the critical flow velocity and Connors’ con- Subscripts
stant of a model heat exchanger tube array. Experiments were also cr ⫽ critical value at the fluidelastic threshold
performed on the model bundle to test the tube stability against a g ⫽ quantity measured in the gap between tubes

Journal of Pressure Vessel Technology MAY 2004, Vol. 126 Õ 273

Downloaded From: https://pressurevesseltech.asmedigitalcollection.asme.org on 05/22/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use


u ⫽ upstream quantity Exchanger Tube Bundles: Review and Design Recommendations,’’ ASME J.
Pressure Vessel Technol., 113, pp. 242–256.
avg ⫽ average velocity of air jet 关6兴 Price, S. J., 1995, ‘‘A Review of Theoretical Models for Fluidelastic Instability
eq ⫽ equivalent uniform velocity of air jet 兵m/s其 of Cylinder Arrays in Cross-Flow,’’ J. Fluids Struct., 9, pp. 463–518.
关7兴 Weaver, D. S., and Fitzpatrick, J. A., 1988, ‘‘A Review of Cross-Flow Induced
Vibrations in Heat Exchanger Tube Arrays,’’ J. Fluids Struct., 2, pp. 73–93.
References 关8兴 Weaver, D. S., and Goyder, H. G. D., 1990, ‘‘An Experimental Study of Flu-
关1兴 Au-Yang, M. K., 2001, Flow-Induced Vibration of Power and Process Plant idelastic Instability in a Three-Span Tube Array,’’ J. Fluids Struct., 4, pp.
Components, A Practical Workbook, ASME Press, New York. 429– 442.
关2兴 Blevins, R. D., 1990, Flow-Induced Vibration, Van Nostrand Reinhold, New 关9兴 ASME, 1998, Boiler Code, Section III, Appendix N-1300 Series, ASME, New
York. York.
关3兴 Chen, S. S., 1984, ‘‘Guidelines for the Instability Flow Velocity of Tube Arrays 关10兴 Appel, D. W. et al., 1959, Advanced Mechanics of Fluids, John Wiley and
in Cross-Flow,’’ J. Sound Vib., 93共3兲, pp. 439– 455. Sons.
关4兴 Paı̈doussis, M. P., 1982, ‘‘A Review of Flow Induced Vibrations in Reactors 关11兴 Lever, J. H., and Weaver, D. S., 1986, ‘‘On the Stability Behavior of Heat
and Reactor Components,’’ Nucl. Eng. Des., 74, pp. 31– 60. Exchanger Tube Bundles: Part 1—Modified Theoretical Model,’’ J. Sound
关5兴 Pettigrew, M. J., and Taylor, C. E., 1991, ‘‘Fluidelastic Instability of Heat Vib., 107, pp. 375– 410.

274 Õ Vol. 126, MAY 2004 Transactions of the ASME

Downloaded From: https://pressurevesseltech.asmedigitalcollection.asme.org on 05/22/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use

You might also like