You are on page 1of 7

Acoustic Resonance in Heat

R. D. Blevins
Exchanger Tube Bundles-
Presently, Rohr Industries,
San Diego, Calif. 92012
Part II: Prediction and
Mem. ASME
Suppression
M. M. Bressler
Mem. ASME of Resonance
GATechnologies Inc., In the first part of this series, experimental data were presented which suggest that
San Diego, Calif. 92138 the acoustic resonance in heat exchanger tube bundles is tied to periodic vortex shed-
ding from the tubes. In this paper, a semi-empirical model for predicting the onset
of resonance is developed. This model is compared with experimental data and other
models from the literature. Methods of suppressing the resonance are developed and
experimental data on their effectiveness are presented.

1 Introduction
This paper compares various methods from the literature Thus, while there is general agreement that the acoustic
for predicting acoustic resonance in heat exchanger tube resonance in heat exchanger tube bundles is associated with
bundles with experimental data and then develops an im- natural acoustic modes of the ducting, there is not general
proved method. Experimental results on the effectiveness of agreement on the nature of the forcing phenomenon.
various methods of suppressing resonance are presented. However, most authors do agree that for a given tube bank,
In 1954, Baird [1] attributed the acoustic resonance in heat the predominant forcing frequency is proportional to flow
exchanger tube banks to flow instability, or pulsation, arising velocity.
from the gas flow over the tube bank. His measurements clear- A number of methods have been used to suppress acoustic
ly show that the fundamental transverse acoustic mode was ex- resonance in heat exchanger tube bundles, including
cited by the pulsation. In 1956, Halliday [2] suggested that the 1) detuning the resonance with baffles that shift the acoustic
pulsation was associated with periodic vortex shedding whose natural frequency upward, 2) increasing acoustic damping
frequency can be predicted by the Strouhal relationship (equa- with Helmholtz resonators, 3) removing tubes, and
tion (3) of Part I of this series). In the same year, Grotz and 4) altering the tube surface. Baird [1], Halliday [2], Grotz
Arnold [3] confirmed that the acoustic modes excited were the and Arnold [3], Cohan and Deane [8], and many others [9]
natural acoustic modes of the system. They suggested that the have successfully employed baffles. Baylac [10] applied a
frequency of excitation is proportional to the flow velocity, Helmholz resonator to suppress a resonance in a nuclear plant
and they presented a method of prediction of onset of heat exchanger. The success of the resonator is evidently at-
resonance. In 1959, Putnam [4] proposed that the periodic tributable to the acoustic damping it provides to the resonant
vortex shedding coupled with the acoustic mode to force the mode. Walker and Reising [11], Barrington [12], and
resonance. Zdravkovich and Nuttall [13] found that by removing about 3
In 1965, Owen [5] published the contrasting view that percent of "judiciously placed" tubes from the bundle,
discrete vortex shedding would not exist deep in a tube bank resonance was suppressed in at least some cases. Walker and
but that nearly homogeneous turbulence in dynamic Reising attribute the success of tube removal to disturbing the
equilibrium would dominate. Unswayed, in 1968 Y. N. Chen buildup of in-phase oscillation through the tube bundle. They
[6] advocated vortex shedding as the mechanism that caused suggest that the tubes to be removed should lie along the
the resonance and he presented correlations for the dominant center of the bank.
acoustic frequency. He illustrated possible vortex-shedding Finally, the buildup of dirt and soot in a tube bank has been
patterns in closely spaced tube banks. In 1985, Fitzpatrick [7] associated with the reduction of sound levels. Barid [14], in
straddled the issue by saying that vortex shedding, buffeting, discussion of Putnam's paper [4], notes
and broad band turbulence all can contribute to acoustic
resonance. . . . . the situation where pulsation of a new boiler
gradually diminishes (or increases) with time as tubes
become increasingly fouled. A good cleaning can ofttimes
Contributed by the Pressure Vessels and Piping Division for publication in the
JOURNAL OF PRESSURE VESSEL TECHNOLOGY. Manuscript received by the PVP
cause a reversion to the original condition. It has been
Division, September 22, 1986; revised manuscript received April 16, 1987. observed that this can be caused by a deposition of 1/16

282/Vol. 109, AUGUST 1987 Transactions of the ASME


Downloaded From: https://pressurevesseltech.asmedigitalcollection.asme.orgCopyright
on 07/02/2019 © 1987
Terms byhttp://www.asme.org/about-asme/terms-of-use
of Use: ASME
in. or less on 2-in.-diameter tubes just on their 4.0
downstream sides.
FLow H @ L <b
e
Rogers and Penterson [15] agree that "fouling of tubes has 3.5 -
been used to eliminate sonic vibration in coal-fired and some -w-
oil-fired boilers. With coal firing, the process requires a few
days while with oil firing the process may require six months i- 3.0 - s
to a year." Blevins has observed this process in a large process
i
plant heat exchanger. Unfortunately, there are no quantitative i
data on the noise reduction achieved by soot accumulation i * *
v 1

and very little quantitative data on the noise reduction 2.5 Y" I m V • ®

achieved by baffling or by removing tubes or by other tech- °z 1


niques for suppressing acoustic resonance in heat exchanger < 1

RESON
tube bundles. 2.0 ****
» * a
2 Prediction of Resonance z 1• •V
Y V '
Grotz and Arnold [3], Putnam [4], Cohan and Deane [8], - OO
• I e ®

Barrington [12, 16], Y. N. Chen, [6, 17], Fitzpatrick [7], and **!
Rogers and Penterson [15] have all proposed methods for 1
•>
RESONANCE
1 1 1 l 1 1
predicting the occurrence of acoustic resonance in heat ex- 1.0
changer tube bundles. A common approach is used in most of 1.0 2.0 2.5 3.0 3.5 4.0
these methods. First, the acoustic natural frequencies of the LONGTUDINAL SPACING, L/D
heat exchanger ducting are predicted; second, the dominant Fig. 1 Map of in-line tube banks prone to acoustic resonance (equation
fluid frequency is predicted and compared with the acoustic (4)) and occurrence of resonance
o Grotz and Arnold [3], no first-mode resonance
natural frequencies to assess the potential for resonance. Last- a Grotz and Arnold [3], first-mode resonance
ly, if resonance is predicted, some consideration is given to a Part I data, no resonance
damping or other mechanisms for suppression of the B Part I data, resonance
resonance. Each of these steps will be reviewed in light of the o Fitzpatrick [29], no resonance
experimental data of Part I and other recent evidence. First, « Fitzpatrick [29], resonance
v Funakawa [23], no first-mode resonance
let us consider prediction of the acoustic natural frequencies. T Funakawa [23], first-mode resonance
Exact closed form solutions are available for the acoustic A Soper [30], no resonance
natural frequencies of simple volumes with uniform proper- * Bryne [28], resonance
ties. For example, the acoustic natural frequencies of a closed
rectangular volume are [18]
lie between those calculated with the decreased sound due to
//ilk' ij,k = 0,1,2,3 the presence of tubes and those calculated without tubes.
(1)
Numerical techniques, such as discussed in reference [19], can
where /, j , and k are modal indices associated with the number be used to accurately calculate acoustic modes.
of half-waves in the longitudinal, transverse, and axial direc- It has been found experimentally that the acoustic modes
tions as shown in Fig. 1 of Part I. The corresponding expres- primarily excited during heat exchanger resonance are those
sion for a closed cylindrical volume is acoustic modes with fluid motion perpendicular to both the
tube axis and the tubes, i.e., the modes where the fluid
(2) oscillates in the lift direction. For example, for our rectangular
volume, these modes correspond to i=0,j= 1, 2, 3, & = 0.*See
Fig. 1 of Part I.
Xoo = 0, X10 = 1.841 is the lowest diametrical mode. R is the The second step in the prediction of acoustic resonance in
radius of the cylinder and L is its length, c is the speed of the methods proposed by Barrington [12, 16], Cohan and
sound in the gas (see references [8, 18]). Deane [8], Y. N. Chen [6], Grotz and Arnold [3], Fitzpatrick
Equations (1) and (2) do not apply exactly to tube-and-shell [7], and Rogers and Penterson [15] is to compare the frequen-
heat exchangers because 1) tubes fill much of the shell cy of the disturbance with the acoustic natural frequencies. If
volume, 2) there is heating or cooling, leading to the flow is subsonic (Mach number less than 0.3), then it
temperature and speed of sound variation, and 3) the heat follows from dimensional analysis that deep within the tube
exchanger shells do not conform exactly to simple shapes; for bank, the dimensionless group fD/Uis a function of Reynolds
example, there are connections to other ducting. Each effect number alone for rigid tubes excluding effects of sound.
can be considered as follows. For acoustic modes which have Hence, the dominant frequency (if any) within the tube bank
wavelengths much longer than the diameter of tubes (the usual is given by the Strouhal relationship
case), Blevins [19] and Parker [20] have shown that for wave
propagation perpendicular to the tube axis, the speed of sound /„ = S£//Z>Hz (3)
is decreased as the tube density increases according to equa- for a given tube pattern, where/„ is frequency in hertz, t/ is
tion (1) of Part I. This decrease in sound speed is typically on flow velocity between tubes, and D is tube diameter. Curves of
the order of 10 percent. Owen [5], Y. N. Chen [6], and Fitzhugh [21] have been used to
Since the speed of sound is a function of temperature, the obtain the dimensionless Strouhal number S. As noted in Part
speed of sound will tend to vary from inlet to exhaust. This I of this series, the present tube array tests have shown that the
can be incorporated in the calculation by considering each sec- dominant acoustic frequency is predicted better by the curves
tion of the heat exchanger at a time and calculating the speed of Fitzhugh (which are reprinted in reference [22]) than by
of sound and frequency of each section. In long units, only a those of other authors.
portion of the heat exchanger can resonate at a time. Finally, Since resonance can shift the natural shedding frequency up
as shown in reference [19], if a heat exchanger shell has or down and the acoustic mode itself has a finite band width, a
substantial regions without tubes, as is true of many ex- band must be placed on the excitation frequency (equation (3))
changers with rectangular shells, the natural frequencies will when comparing it with the acoustic frequencies (equations

Downloaded From:Journal of Pressure Vessel Technology


https://pressurevesseltech.asmedigitalcollection.asme.org AUGUST 1987,
on 07/02/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use Vol. 109 / 283
Table 1 Prediction of acoustic resonance

Grotz and Arnold [3] criterion Y. N. Chen [17] criterion


Tube bundle Resonance Resonance Resonance
(P/D) observed? C/=l) ( a ) U = 2) (a) predicted? \Jr(«) predicted?
1 Triangle (2.0) Yes 9.75(24.0) 4.48(12.0) Yes 83,900(41,400) Yes
2 Square (1.5) No 48.0 24.0 Yes 10,800 Yes
3 Diamond (3.0) Yes 12.0(21.4) 6.0(10.7) Yes 49,200 Yes
4 Square (3.0) Yes 12.0 6.0 Yes 62,900 Yes
5 Triangle (1.5) No 15.0(48.0) 48.0(24.0) Yes 77,100(22,600) Yes
6 Triangle (3.0) Yes 5.73(12.0) 2.8(6.0) Yes 87,300(59,600) Yes
7 Square (2.0) Yes 24.0 12.0 Yes 47,700 Yes
8 Diamond (4.0) Yes 8.0 4.0 Yes 53,700 Yes
9 Triangle (1.2) No 22.4(120.0) 11.2(60.0) Yes(No) 48,600(5,000) Yes
(o
> Numbers in ( ) are criteria utilizing P in place of L.

gap velocity; S is the Strouhal number; L is the longitudinal


FLOWv ©
U-l ©-J-
(direction of mean flow) distance between tube centers; and T
t$> © f is the transverse distance between tube centers. The Grotz and
" ^ ,©, ©- 1 - Arnold criterion [3] is
H/D
T= , —r-r <62 or 80, for resonance, / = 1,2,3, . . . (6)
(L/D — l)j
3 -
RESONANCE where His the distance between the duct walls, perpendicular
X to the flow and the tube axes, and j= 1,2,3, . . . is the acoustic
mode number. Both criteria suggest that the more closely
spaced tube arrays {L/D < 1.5) are not prone to resonance. In
*y
NO RESONANCE^D^" "]
•A-
addition, the Grotz and Arnold criterion suggests that in these
arrays, the higher modes (/' = 2,3 . . . ) may resonate while the
first one will not, as observed by Grotz and Arnold and by
1 2 3 4 5 8 Funakawa [23].
LONGITUDINAL SPACING, L/D The Grotz and Arnold and Y. N. Chen criteria are com-
Fig. 2 Map of staggered tube banks prone to acoustic resonance
pared with the experimental data in Table 1. (Fitzpatrick [7]
(equation (4)) and occurrence of resonance does not place his criterion in simple form, so it cannot be
o Grotz and Arnold [3], no first-mode resonance directly compared with the data. See discussion by Paidoussis
• Grotz and Arnold [3], first-mode resonance [24].) As can be seen in Table 1, both the Y. N. Chen and the
• Part I data, no resonance
Grotz and Arnold criteria predict all nine of the tube bundles
B Part I data, resonance
o Fitzpatrick [29], no resonance
will resonate, while in fact the closely spaced bundles (bundles
• Fitzpatrick [29], resonance number 2, 5, and 9) are free of resonance. Thus, these criteria
v Funakawa [23], no first-mode resonance are not good discriminators for resonance in the present test
T Funakawa [23], first-mode resonance data. (The Grotz and Arnold criterion is based on data for in-
A Barrington [16], resonance
* Soper [30], no resonance
line tube arrays only. Its extension to staggered arrays is not
• Soper [30], resonance defined. If we assume that L in equation (5) is the longitudinal
©Blevins, private data, resonance pitch for staggered bundles as well, then the Grotz and Arnold
Weaver [31], resonance criterion also predicts that all bundles will resonance; but if we
interpret L as the tube pitch (see Fig. 2 of Part I), then the
(1), (2)) to determine if the potential for resonance exists. Bar- criterion successfully predicts that bundle 9 will not resonate.)
rington [12] and Rogers and Penterson [15] recommended a Following a suggestion by Arnold [25] and the Strouhal
±20 percent band. That is maps of Fitzhugh and Y. N. Chen, experimental data for tube
S(l - a) U/D </„ < S(l + (3) U/D, for resonance (4) patterns free of resonance are shown on a map of T/D versus
L/D in Figs. 1 and 2 for in-line and staggered bundles. Care
Rogers and Penterson recommend a — /3 = 0.2, where /„ is has been taken to include only bundles which are predicted to
the natural frequency of the acoustic mode of interest. The resonate by the criterion of equation (4). Open symbols are for
averages of the experimental value of Part I, Table 3, are a = bundles that are free of first mode acoustic resonance
0.19 and /? = 0.29, but the maximum values are a = 0.40 and (although some do resonate in the second mode). Closed sym-
& = 0.48. Probably these maximum values should be used for bols are for bundles that resonate in the first mode. More data
a conservative design. are provided for resonance than nonresonance because
For designs where a resonance is predicted (equation (4)), a authors often do not provide enough information to confirm
third step in the analysis is made. In many cases in very closely that equation (4) was satisfied when a loud resonance did not
spaced tube banks where an acoustic resonance is predicted by emerge.
equation (4), the resonance fails to materialize [15]. Grotz and The maps imply that in-line bundles with L/D < 1.4 and
Arnold [3], Y. N. Chen [17], and Fitzpatrick [7] have pro- staggered bundles with L/D < 3.0 and T/D < 1.6 do not
posed semiempirical models for determining tube banks that resonate in the first acoustic mode. The close tube spacing
will not resonate. The Y. N. Chen criterion is probably suppresses the formation of large eddies within the
Re/ D \2 / D \ bundle. For example, Bearman and Zdravkovich [26] found
* = — (1 — — ) ( — J > 600 or 2000, for resonance (5) for a single tube above a plane surface in a flow that vortex
shedding was suppressed for gaps between the tube and the
where Re = UD/v is the tube Reynolds number, [/being the surface less than 0.3 tube diameter.

284/Vol. 109, AUGUST 1987 Transactions of the ASME


Downloaded From: https://pressurevesseltech.asmedigitalcollection.asme.org on 07/02/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use
O O O OOO
o o o o o o
o o o o o o
o o o o o o
o o o o o o
o o o o o o
(a) CENTRAL BAFFLE (b) UPSTREAM BAFFLE

o o o o o oo o o o o o o o
o o o o o oo a. 0 p 43
o o o o o o o_ 0 p 6 p
p 0 13
o o o o o oo o o o o o o
o o o o o oo o 1) A
h p Pi ft t5
o o o o o oo <r
o o o u u o o

(c) DOWNSTREAM BAFFLE (d) TWO CENTRAL BAFFLES


AT 1/3 AND 2/3
TRANSVERSE LOCATIONS
Fig. 3 Baffle positioning in bundle. Tube size is exaggerated for clarity.

CO

"o
0 *-7"^-~" '^?C / / 10.0 20.0 30.0 40.0 50.0 60.0 70.0
o GRP VELOCITY, M/S
o Jg -_^-H ' /
CD ~-
Fig. 5 Sound levels in 180-tube square pattern bundle with solid
<-> o bundles
». o • No baffles
U £ o Solid central baffle (Fig. 3(a))
_J
Q o A Solid downstream baffle (Fig. 3(c))
Z •
+ Two solid central baffles at 1/3 and 2/3 (Fig. 3(d))
CD _
h- o
OC o
_J
r*l
- fundamental transverse acoustic frequency of approximately
D_ „
en R
370 Hz.
CM The baffle positioning is shown diagrammatically in Fig. 3.
O
Sound levels with baffles installed in the 180-tube square pat-
O tern array are shown in Figs. 4 and 5. As can be seen in Fig. 4,
a the perforated metal baffles had relatively little effect on the
o sound level. The largest reduction is for the center placement,
10.0 20.0 30.0 40.0 50.0 60.0 70.0 but even here the reduction is over a very limited velocity
GRP VELOCITY, M/S range. Also shown are the sound levels which result with an
Fig. 4 Sound levels in 180-tube square pattern bundle with perforated upstream turbulence screen of 1/4 in. (6.3 mm) diameter rods
baffles (also see reference [27]). The turbulence screen has a negligible
a No baffles effect on the sound levels.
o Turbulence screen
A 33 percent open baffle at center
It is useful to note that in some cases, perforated baffles do
+ 33 percent open baffle upstream produce acceptable results. For example, Byrne [28] found a
x 65 percent open baffle at center perforated baffle reduced sound to acceptable levels in ap in-
line tube bundle. Unfortunately, he did not also utilize a solid
baffle to provide a basis for comparison.
3 Suppression of Resonance Figure 5 shows results with the solid baffles. When the baf-
Four methods of suppressing an existing resonance are fle is upstream of the bundle, the sound reduction is relatively
1) introduction of baffles to increase acoustic frequencies, small. With the baffle through the center of the bundle, the
2) selective tube removal, 3) addition of a tuned Helmholtz resonance with the 370 Hz acoustic mode is completely sup-
resonator to increase acoustic damping, and 4) alteration of pressed, but the second acoustic mode is excited beyond about
the tube surface to suppress organized vortex shedding. 164 ft/sec (50 m/s). With two baffles at the 1/3 and 2/3
Modifications of these four categories were installed on two transverse points, both the first and the second acoustic modes
tube bundles: the 180-tube square array (tube bundle 7) and are suppressed. (Compare with Fig. 9 of Part I.) These results
the 196-tube triangular array (tube bundle 1). These bundles are in general agreement with the observations of Grotz and
were chosen because they have intense, multimode acoustic Arnold [3], Walker and Reising [11], and Baird [1]. There is
resonances and their response is typical of the more widely general agreement in the literature that by decreasing the ef-
spaced in-line and staggered tube bundles. fective transverse width of the duct to increase the fundamen-
The baffle plates employed were sheet metal, 22 in. (55.9 tal acoustic natural frequency (equation (1)), resonance is sup-
cm) long by 15 in. (38.1 cm) high. They were of two types: pressed. These results are also in agreement with those of
0.080 in. (2.2 mm) thick perforated steel and 0.125 in. (3.2 Baird [1] showing that baffles placed in the center of the bun-
mm) thick solid aluminum plates. They were held in the test dle are more effective than upstream or downstream baffles
section by means of loose-fitting tacks along the top and bot- [1].
tom edges. The baffles spanned the test section from top to The acoustic mode shapes of the test section were measured
bottom, and they were approximately the length of the tube with the single, central solid baffle (Fig. 3(a)) installed using a
bundles. No tubes were removed for baffle installation in the Hewlett-Packard 5423A two-channel analyzer. The
square bundle, but a tube lane was removed to permit baffle methodolgy is described by Blevins [19]. In order to visualize
installation in the staggered bundle. While no direct the three-dimensional shape of the pressure modes, three or-
measurements were made of the baffle natural frequencies, thogonal lines are used as shown in Fig. 6. The horizontal line
the baffles appeared to possess bending modes well below the represents pressures measured along the axis of the duct. The

Journal of Pressure Vessel Technology AUGUST 1987, Vol. 109/285


Downloaded From: https://pressurevesseltech.asmedigitalcollection.asme.org on 07/02/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use
PISTON P0SITI 0N—I

TEST SECTION
WALL

MODE 1 362.8 Hz MODE 1 382.4 Hz


WITHOUT BAFFLE WITH BAFFLE
10 CM
-J5 ^ &
IT
Fig. 6 Acoustic mode shapes with and without solid central baffle (Fig.
3(a)). x is the direction in line with flow. Also see Ref. [19]. t 20.4 CM

Fig. 8 Helmholtz resonator design

)0.0 30.0 40.0 50.0 70.0


GRP VELOCITY, M/S
Fig. 7 Sound levels in 180-tube square pattern with tubes removed
D All tubes in place
o Every other tube in center lane removed
A Every other tube in two center lanes removed
+ AII tubes in center lane removed
x All tubes in center two lanes removed

10.0 20.0 30.0 40.0 50.0 60.0 70.0


two breaks in this line show the extent of the bundle. The GRP VELOCITY, M/S
horizontal line represents pressures measured across the duct.
Fig. 9 Sound levels in 196-tube triangular pattern tube bundle with
Figure 6 compares mode shapes with and without the baffle. Helmholtz resonator and baffles
The greatest difference between the two conditions is in the D No baffles or resonators
fundamental mode. Without the baffle, the fundamental o 1 central solid baffle (Fig. 3(a))
transverse acoustic mode is bound to the tube bundle; it A 2 central solid baffles at 1/3 and 2/3 (Fig. 3(d))
+ 3 percent volume tuned Helmholtz resonator
decays exponentially upstream and downstream of the bundle.
With the baffle, this mode is suppressed within the bundle, but
it does exist in the ducting upstream and downstream of the dimensions of the resonators were chosen so that the resonator
bundle. natural frequency [18],
Figure 7 shows the effect of tube removal on sound pro- 1/2
duced by the 180-tube square pattern bundle. Tubes were c / A A \\ l/i
removed in lanes along the bundle centerline. First, every f= ( ) Hz
other tube was removed from the lane of tubes down the V VL
where c is the speed of2TTsound in )the air, A is the area of the
centerline of the test section (i.e., seven tubes removed parallel resonator neck, L is the effective length of the neck, and Fis
to the flow). Then every other tube was removed from the ad- the volume of the resonator body, coincided with the fun-
jacent tube column to form a w pattern of removed tubes (15 damental transverse acoustic natural frequency of the test sec-
tubes removed). Runs were also made with all tubes in the tion (approximately 370 Hz). The resonators were constructed
centerline column removed and with the two center columns from acrylic pipe. They were tuned by trial and error for max-
removed. As can be seen in Fig. 7, the sound is not influenced imum sound reduction in the first acoustic mode by adjusting
by tube removal in this bundle until two full lanes of tubes (30 the piston. The formula is only approximate because of the ef-
tubes) are removed. Similar results were also obtained with fect of the length of the neck and also because a 50 percent
tube removal in the 196-tube triangular array. Evidently, tube perforated sheet was placed across the opening to minimize
removal is not a reliable method for reducing sound levels in disruption to the mean flow.
bundles of this type. Two resonator sizes were tested. The dimensions of the
Helmholtz resonators were installed external to the tube small resonator are as follows: length of neck = 2 in. (5.1
bundle on the side of the test section with the opening at cm), diameter of neck = 2.2 in. (5.6 cm), diameter of
midheight, 10 in. (25 cm) back from the forward edge of the resonator volume = 3.95 in. (10 cm), length of volume when
bundle. The design of the resonators is sketched in Fig. 8. The tuned = 1.06 in. (2.7 cm). The large resonator has diameters

Downloaded From: 286/Vol. 109, AUGUST 1987


https://pressurevesseltech.asmedigitalcollection.asme.org
Transactions of the ASME
on 07/02/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use
Table 2 Acoustic properties of 196-tube triangular bundle 1 A method for the prediction of acoustic resonance in heat
with resonator installed exchanger tube bundles can be developed by extension of
previous methods based on a vortex-induced resonance. The
Resonator Acoustic Sound Sound potential for resonance is assessed by comparing the vortex-
piston position, '"' damping, <6) level, (c) level, ( c ) shedding frequencies of Fitzhugh with the acoustic natural fre-
in. (cm) percent 10 3 Pa SPL quencies using a reduced speed of sound to account for the
0.0 (0.00) 1.50 1.50 157.5 presence of tubes. In bundles where a resonance is predicted,
0.5 (1.27) 1.92 1.46 157.3 the tube pattern is compared with a map showing patterns
1.0 (2.54) 1.98 1.46 157.3 known to be free of resonance.
1.5 (3.81) 1.85 1.38 156.7
2.0 (5.08) 2.57 1.14 155.0 2 Baffles installed within the tube bundle, parallel to the
2.5 (6.35) 2.81 1.581 149.2 flow, are very effective in suppressing lower mode resonances.
3.0 (7.62) 3.0 0.317 144.0 The number of baffles required is tied to the wavelength of the
3.5 (8.89) 2.5 0.783 149.9 resonance mode. The baffles can be relatively thin and un-
4.0 (10.2) 2.11 0.783 151.9
4.5 (11.4) 1.86 1.14 155.1 stiffened and still be very effective. The higher the mode, the
5.0 (12.7) 1.44 1.31 156.3 more baffles are required. Baffles upstream or downstream of
5.5 (14.0) 1.54 1.37 156.7 the bundle are much less effective than baffles within the bun-
6.0 (15.2) 1.92 1.39 156.8 dle. Peforated baffles are much less effective than solid
6.5 (16.5) 1.51 1.42 157.0
7.0(17.8) 1.33 1.43 157.1 baffles.
3 Tuned Helmholtz resonators external to the shell can be
(0)
SeeFig. 8 effective in reducing resonance with a particular acoustic
(6)
Measured in absence of flow; first m o d e mode. In order to be effective, the resonator must be suffi-
(c)
A t 27 m / s (89 ft/sec), first-mode resonance m a x i m u m
ciently large and it must be tuned to the mode of interest. The
present results give a 13-db reduction for a carefully tuned
resonator with a volume 3.2 percent of the tube bundle
about twice those of the small resonator: length of neck = 1 volume. It is felt that with resonators between 10 and 20 per-
in. (2.5 cm), diameter of neck = 4 in. (10.2 cm), diameter of cent of the tube bundle volume, reductions of 20 to 30 db can
volume = 7.5 in. (19 cm), length of volume when tuned = 3 be achieved.
in. (7.6 cm). The sound levels with the resonators installed 4 Removal of a small number of tubes (3 to 6 percent) from
with the 196-tube triangular tube bundle (tube bundle 1) are the 180-tube square bundle and the 196-tube triangular pattern
shown in Fig. 9. The small resonator has negligible effect on array had negligible effect on the sound levels. Removal of 16
sound levels, whereas the large resonator cut first mode sound percent of the tubes from the 180-tube bundle did significantly
levels by 50 percent. This result suggests that the resonator reduce sound levels.
volume must be a significant fraction of the shell volume in
order to have an appreciable effect. The volume of the large
resonator [175 in 3 (2800 cm3)] is 3.2 percent of the volume of
the test section containing tubes (5400 in. 3 (88,000 cm 3 )), Acknowledgments
while the small resonator volume is a factor of 12.8 smaller.
This work was supported by the U.S. Department of
Neither resonator has a significant effect on the higher mode
Energy, Office of Basic Energy Sciences, Division of
resonances.
Engineering and Geosciences, Contract Number DE-
The effectiveness of the resonators is tied to their tuning to
AT03-82ER12023. The experimental measurements were
the acoustic mode. When tuned, there is a large oscillating
made by R. Dexter, R. H. Jarrett, and J. P. Smith.
mass flow through the resonator neck which expends energy
by passing through the screen and the abrupt expansions on
each side of the neck. That is, the resonator damps the
acoustic mode. Tuning is critical to achieving maximum ef-
fect. As shown in Table 2, the sound during first mode References
resonance rises and falls as the resonator volume is tuned.
Also shown in this table is the acoustic damping of that mode. 1 Baird, R. C , "Pulsation-Induced Vibration in Utility Steam Generation
The damping was measured in the absence of flow by exciting Units," Combustion, Vol. 25 (10), 1954, pp. 38-44.
2 Halliday, J., "Boiler Vibration Caused by Combustion Gas Flow,"
the volume with white noise. The damping is greatest when the American Society of Mechanical Engineers, Paper No. 56-A-216, 1956.
resonator is tuned in resonance. However, the resonator must 3 Grotz, B. J., and Arnold, F. R., "Flow-Induced Vibrations in Heat Ex-
be carefully tuned to achieve the maximum sound level reduc- changers," Department of Mechanical Engineering, Stanford University
tion. Shifts of only 0.5 in. (1.27 cm) in the piston position Technical Report No. 31, DTIC No. 104568, 1956.
4 Putnam, A. A., "Flow-Induced Noise in Heat Exchangers," ASME Jour-
could shift sound level by 6 db. nal of Engineering for Power, Vol. 81, 1959, pp. 417-422.
Figure 9 shows various results for the 196-tube triangular 5 Owen, P. R., "Buffeting Excitation of Boiler Tube Vibration," Journal
tube bundle. In this case, the slightly-out-of-tune large of Mechanical Engineering Science, Vol. 7, 1965, pp. 431-439.
6 Chen, Y. N., "Flow-Induced Vibration and Noise in Tube-Bank Heat Ex-
resonator provides about 6-db reduction in the sound level at changers due to von Karman Streets," ASME Journal of Engineering for In-
first mode resonance, which occurs at 89 ft/sec (27 m/s). The dustry, Vol. 90, 1968, pp. 134-146.
single central baffle provides about twice this resonator sound 7 Fitzpatrick, J. A., "The Prediction of Flow-Induced Noise in Heat Ex-
reduction. The authors believe that larger resonators, 10 to 20 changer Tube Arrays," Journal of Sound and Vibration, Vol. 99, 1985, pp.
425-435.
percent of the bundle volume, are capable of reducing sound
8 Cohan, L. J., and Deane, W. J., "Elimination of Destructive Vibrations
levels by 20 to 30 db. Two solid baffles achieve a 35-db reduc- in Large, Gas and Oil-Fired Units," ASME Journal of Engineering for Power,
tion in the first mode resonance and also suppress second Vol. 87, 1965, pp. 223-228.
mode resonance, whereas the single baffle and the tuned 9 Blevins, R. D., "Review of Sound Induced by Vortex Shedding from
resonator have no effect on the higher mode resonances. Cylinders," Journal of Sound and Vibration, Vol. 92, 1984, pp. 455-470.
10 Baylac, G., and Gregorie, J. P., "Acoustic Phenomena in a Steam
Generating Unit," Journal of Sound and Vibration, Vol. 42, 1975, pp. 31-48.
4 Conclusions 11 Walker, E. M., and Reising, G. F. S., "Flow-Induced Vibrations in Cross
Flow Heat Exchangers," Chemical Process Engineering, Vol. 49, 1968, pp.
The conclusions from this study of the prediction and sup- 95-103.
pression of acoustic resonance due to gas flow over heat ex- 12 Barrington, E. A., "Acoustic Vibration in Tubular Heat Exchangers,"
changer tube bundles are as follows: Chemical Engineering Progress, Vol. 69(7), 1973, pp. 62-68.

Journal
Downloaded From: of Pressure Vessel Technology
https://pressurevesseltech.asmedigitalcollection.asme.org
AUGUST 1987, Vol. 109/287
on 07/02/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use
13 Zdravkovich, M. M., and Nuttall, J. A., "On the Elimination of 22 Blevins, R. D., Flow-Induced Vibration, Van Nostrand Reinhold, N.Y.,
Aerodynamic Noise in a Staggered Tube Bank," Journal of Sound and Vibra- 1977; reprinted by Robert E. Kreiger Publishing, Malabar, Fla., 1986.
tion, Vol. 34, 1974, pp. 173-177. 23 Funakawa, M., and Umakoshi, R., "The Acoustic Resonance in a Tube
14 Baird, R. C , discussion of reference [4], ASME Journal of Engineering Bank," Journal of Japanese Society of Mechanical Engineers, Vol. 13, 1970,
for Power, Vol. 81, 1959, p. 420. pp. 348-355.
15 Rogers, J. D., and Penterson, C. A., "Predicting Sonic Vibration in Cross 24 Paidoussis, M. P., reply to discussion by J. A. Fitzpatrick, Journal of
Flow Heat Exchangers—Experience and Model Testing," American Society of Sound and Vibration, Vol. 85, 1982, pp. 435-441.
Mechanical Engineers, Paper 77-FE-7, 1977. 25 Arnold, F. R., discussion of reference [4], ASME Journal of Engineering
16 Barrington, E. A., "Cure Exchanger Acoustic Vibration," Hydrocarbon for Power, Vol. 81, 1959, p. 420
Processing, 1978, pp. 193-198. 26 Bearman, P. W., and Zdravkovich, M. M., "Flow Around a Circular
17 Chen, Y. N., and Young, W. C , "Part 3—Damping Capability of the Cylinder Near a Plane Boundary," Journal of Fluid Mechanics, Vol. 89, 1978,
Tube Bank Against Vortex Excited Sonic Vibration in the Fluid Column, ASME pp. 33-47.
Journal of Engineering for Industry, Vol. 96, 1974, pp. 1072-1075. 27 Belvins, R. D., "The Effect of Sound on Vortex Shedding from
18 Blevins, R. D., Formulas for Natural Frequency and Mode Shape, Van Cylinders," Journal of Fluid Mechanics, Vol. 161, 1985, pp. 217-237.
Nostrand Reinhold, New York, 1979; reprinted by Robert E. Krieger 28 Byrne, K. P., "The Use of Porous Baffles to Control Acoustic Vibrations
Publishing, Malabar, Fla., 1984. in Crossflow Tubular Heat Exchangers," ASME Journal of Heat Transfer, Vol.
19 Blevins, R. D., "Acoustic Modes of Heat Exchangers Tube Bundles," 105, 1983, pp. 751-758.
Journal of Sound and Vibration, Vol. 109, 1986, pp. 19-31. 29 Fitzpatrick, J. A., and Donaldson, I. S., "A Preliminary Study of Flow
20 Parker, R., "Acoustics Resonances in Passages Containing Banks of Heat and Acoustic Phenomena in Tube Banks," ASME Journal of Fluids Engineer-
Exchanger Tubes," Journal of Sound and Vibration, Vol. 57, 1978, pp. ing, Vol. 99, 1977, pp. 681-686.
245-260. 30 Soper, B. M. H., "The Effect of Tube Layout on the Fluidelastic In-
21 Fitzhugh, J. S., "Flow-Induced Vibration in Heat Exchangers," Oxford stability of Tube Bundles in Cross Flow," Flow-Induced Heat Exchanger Tube
University Report RS57, Jan. 1973 (AERER-P7238); also, Proceedings of Vibration, ASME Publication HTD-V.9, 1980.
UKAEA/NPL International Symposium on Vibration Problems in Industry, 31 Weaver, D. S., private communication of unpublished data, August 19,
Keswick, Apr. 1973, Paper 427. 1986.

ERRATUM1
Figure 5 is reproduced here. It was disfigured in the pub-
lished paper.

I n si da d i a m e t e r Di = 3 0 5 m m (12 in)
Critical Detign pretture P = 3 0 0 M P a (43.5k«i)
crackdepth C r a c k ,haps a / 2 c = 0.l

..14
inch
2 5 + 1-0
Wall 9'
T h l c k n e e t . , 8"
20 0.8

15 0.6

10 0.4
."ELI

5-0.2
S a f e t y factor
against bursting

1 2 3 4 5 (Faupel)

Fig. 5 Relationship between critical crack depth and safety against


bursting

'For "Development of Design Criteria for a High Pressure Vessel Construc-


tion Code," by G. J. Mraz, published in the May 1987 issue of JOURNAL OF
PRESSURE VESSEL TECHNOLOGY, p. 258.

288/Vol. 109, AUGUST 1987


Downloaded From: https://pressurevesseltech.asmedigitalcollection.asme.org Transactions
on 07/02/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use of the ASME

You might also like