Sei sulla pagina 1di 8

CECIL R.

SPARKS
Assistant Director,
Department of A p p l i e d Physics,
Southwest Research Institute,
San Antonio, Texas

D. E. UNDGREN
Senior Research Engineer,
Tennessee Gas Pipeline C o m p a n y ,
Houston, Texas

Design and Performance of High-Pressure


Blowoff Silencers
Through the application offl/uid dynamic and acoustic theory, the noise generation of a
high pressure blowoff can be approximated.
The effects of silencer configurations can
likewise be predicted through the application of pertinent field data taken to define performance of the silencer components.
This paper describes recent test results and their
application to improved silencer design for natural gas pipeline
applications.

Introduction

attenuation technique utilized in the design of these silencers,


and presents field data as to their effectiveness.

UNE OF the most severe pipeline noise problems insofar as sound intensity is concerned is that associated with highpressure blowoff S3'stems. Within the wide network of domestic
pipeline installations, blowdown valve locations range from
areas of almost complete isolation to locations where residential
areas have expanded to within a few hundred feet of the blowdown valve. The problems associated with blowdown noise
have increased steadily as rural areas adjacent to the right-ofway have become heavily populated. In order to avoid noise
annoyance problems arising from these blowdowns, the industry
has taken many steps ranging from moving blowoff valves outside populated areas to notifying residents well in advance of
planned blowdowns. In the latter case, residents at distances
up to one-half mile are notified and residents often leave home
until the blowdown is completed. Others, particularly those
who are unprepared for the noise, readily voice their annoyance
and objection.
In order to minimize community annoyance resulting from
blowdown noise, major pipeline companies have turned to the
development of blowoff silencers to predictably control generated
noise levels. Much of the work in this area was based upon research performed for the American Gas Association by Southwest Research Institute, and silencers were built and tested for
a wide range of applications within the industry.
One of the primaiy areas of concern was the development of a
portable blowoff silencer whose design could be generalized to
extend its applicability to the wide range of planned blowoff
applications. As such, it was desirable to obtain maximum noise
attenuation, but within the size and weight limitations of normally available field equipment to move the silencer from location
to location and place it on the blowdown valve. This paper
presents the theory of blowoff noise suppression, describes the

Contributed by the Petroleum Division and presented at the


Winter Annual Meeting, New York, N. Y., November 29-December
3, 1970,

of T H E AMERICAN SOCIETY OF MECHANICAL

ENGINEERS.

Manuscript received at ASME Headquarters, July 30, 1970.


No. 70-WA/Pet-l.

Journal of Engineering for Industry

Paper

Blowoff Noise Generation


The noise generation mechanisms of a high-pressure blowoff
are characterized by turbulence-induced noise from high velocity
flow. The primary source of this noise is mixing of the high velocit}' gas stream with the atmosphere, which in turn produces
shear eddies or vortices along the shear boundary. These shear
vortices then radiate acoustic pressure perturbations or noise
throughout this mixing region.
To demonstrate effects, consider a simple blowdown S3^stem
consisting of an open pipe discharging directly into the atmosphere. In this case, high velocity flow within the riser piping
generates severe internal turbulence; however, this is not where
the major portion of the observed noise comes from. While this
internal turbulence, particularly that generated by passage
through the constricting valve, generates dipole vortex noise, the
major noise is generated outside the piping itself; i.e., in the mixing region of the high velocity jet as it shears with the atmosphere. Under these conditions, shown graphically in Fig. 1, the
high velocity flow of the jet shears with the bounding air, producing violent eddies which are then convected downstream with
the jet. As the eddy is convected downstream, its kinetic energy
is converted to potential (pressuve) energy with the typical fourlobe propagation pattern of quadrapole sources as shown in the
figure.
Within this mechanism, therefore, the actual source of noise
is the entire mixing region of the jet rather than the open end
of the pipe itself. As such, the source extends from the exhaust
of the pipe to a distance of up to 25 pipe dia downstream.
Typically, the high frequency portion of the noise is generated
in the extremely high shear area near the pipe exhaust, whereas
the low frequencies are produced in the relatively low shear
but large eddy sections in the downstream portion of the mixing
region.

Noise Intensity and Frequency Content


The definition of total acoustic energy generated by a jet ex-

Copyright 1971 by ASME

Downloaded From: http://energyresources.asmedigitalcollection.asme.org/ on 04/17/2015 Terms of Use: http://asme.org/terms

MAY 1 97 1 / 695

Table 1

haust has been defined largely by observation. These observations are sufficient, however, to permit a definition of the dependency of generated noise on the controlling fluid parameters.
The source of acoustic energy in a flow stream is, of course, the
kinetic energy of the stream which may be defined as (U2/2)pUA, where (V-/2) is the kinetic energy per unit mass and
pUA is the mass flow rate of the stream. The efficiency of
conversion of this kinetic energy flux to sound power has been
shown to be proportional to the fifth power of Mach. number
CM6) and (when the jet is discharging into the atmosphere)
to the ratio of (p/po), where p is the flowing gas density and
po is ambient air density. Thus the total acoustic power of a
high velocity jet discharging into the atmosphere follows the
equation

0
25
50
75
100
125
150
175
200
300
400

528
339
213
152
107
53.7
27.2

(1)

Poc5

With regard to frequency content, experimentation has also


shown that a typical spectral distribution for a jet is wide band
in nature, but with a spectral peak corresponding to a Strouhal

Converting this acoustic power expression to more conventional


acoustic terms, we get for the sound power level of the jet the
following expression:
p*U*A

= 10 log Id ~

dB re lO" 1 2 watts

(2)

Poc"

or for sound pressure level


SPL = 10 log Kt

Ft from Source

#2

42500
6780
2130
1070

p2Usd2

PWL

Comparison of predicted and Field test noise reduction

p*UaA

Poc6

dB re 0.0002 dyne/cm 2

fd

number (N,) of from 0.17 to 0.2 (N, = -r, w h e r e / = frequency


in Hz, d = pipe dia, and U = exhaust velocity). On either side
of this Strouhal peak, a decay of 3 dB per octave is observed.
From this analysis, several observations became apparent for
reducing turbulent mixing noise:

(3)

Where K2 varies with distance from the measurement point to


the blowoff, these may be taken from Table 1.

Table 1

1 Reduce jet velocity. A reduction of 50 percent in jet


velocity will result in a 24 dB reduction in quadrapole generated
noise.
2 Complete the mixing process in a confined volume and
prevent its direct radiation into the atmosphere.

Comparison of predicted and field test noise reduction

DESCRIPTION

DETAIL

PREDICTED

FIELD

TEST

10 F T . S I L E N C E R
WITH
ORANGE
PEEL
DIFFUSER

17

25

10 F T . S I L E N C E R
WITH
ELLIPTICAL
WELD
CAP
DIFFUSER

17

18

6 FT. S I L E N C E R
ORANGE
PEEL

WITH
DIFFUSER

i . 3'. i

14

\n

696 / MAY

10 F T . S I L E N C E R
HEMISPHERICAL
CAP
DIFFUSER

WITH
WELD

HEMISPHERICAL
CAP
DIFFUSER

WELD
ONLY

17

23

10

197 1

Downloaded From: http://energyresources.asmedigitalcollection.asme.org/ on 04/17/2015 Terms of Use: http://asme.org/terms

Transactions of the AS ME

HIGH SHEAR
REGION,
SEVERE HIGH FREQUENCY
MIXING

MIXING L E N G T H '
5 - 2 0 DIAMETERS

^Sj( V ^ S H E A R

QUADRAPOLE

/\-s

' P O T E N T I A L CONE '


(UNMIXED)

"

QUADRAPOLE
PROPAGATION
PATTERN
Fig. 1

LARGE S C A L E
LOW
FREQUENCY
TURBULENCE

BlowofF jet noise and turbulence structure

3 Treat the exhaust section from this with absorbing material


to attenuate mixing noise before it is discharged into the atmosphere.
4 Reduce jet exhaust size by using, for example, multiple
small-dia holes. This does two things: reduces the length of the
mixing region (thereby permitting completion of the mixing
process in a shorter length), and shifts the spectral peak to a
much higher portion of the frequency spectrum. High frequency noise is easier to absorb.
Details of treatments following these suggestions will appear
in a subsequent section of the text.

Blowoff Silencing
Since most blowoff noise predominates in the high frequency
portion of the spectrum, a logical approach to silencing it is
through the use of an absorbing section or lined duct as is sometimes used to suppress regulator noise. However, since the mixing region extends up to 25 pipe dia downstream, absorbing material should be applied along the entire mixing region,
and a rather cumbersome design would result. On the other
hand, if this jet is broken up into a series of smaller jets, or is
otherwise altered to produce full mixing in a shorter region, the
design and size requirements for an effective sound absorber are
considerably reduced. An effective silencer might then consist
of a jet diffuser at the inlet of the silencer in a relatively short
section, followed by an absorbing section immediately downstream to further attenuate the noise of the inlet jet and that
regenerated by the diffuser. In essence, this latter section of
the silencer is a sound stream absorber for the more stabilized,
lower velocity flow, and acoustical material is thereby more
effectively utilized than it would be without a diffuser (i.e., under
conditions of full inlet nozzle flow).

Since noise from a blowoff is proportional to the eight power


of velocity, it is important in designing an effective silencer to
keep the discharge velocity of the silencer as low as possible.
In many high-pressure applications it is possible to have sonic
flow not only in the valve constriction but also in the silencer discharge, and in such cases the benefit derived is only by virtue
of reduced flowing density. Therefore, to assure adequate
silencer performance it is necessary to calculate the flow characteristics of the silencer before calculations can be made of its
acoustical performance. Since these calculations are complex
and are covered in previous literature [1, 2], 1 they will not be
treated here. Design charts presented later have the calculations already made for natural gas. It is worthwhile to discuss
what happens in regard to internal flow in these silencers, however, if just to emphasize the importance of tailoring silencer designs to the job at hand.
Flow in the silencer shown in Fig. 2 is characterized by a series
of expansion and mixing processes as flow is passed through the
discontinuously diverging piping. For a given gas composition,
this internal flow may be defined on the basis of pipeline (source)
pressure and flow areas, as follows:
1 At extremely low pipeline pressures, flow velocity at the
silencer inlet (Fi) will be subsonic, and exhaust velocity (F 2 )
will be even lower. This condition is of trivial importance in
most industrial applications where noise is a problem.
2 As upstream pressure is increased above a critical level,
sonic flow will be experienced at the inlet constriction, and no
higher velocity will be achieved in most practical designs. This
first critical line pressure (pcl) may be defined as follows:
Pel = P a t m ( 2 / f c +

l)fc/(l-fc)j

lb/ft

(4)

,2

where palm is the atmospheric pressure, lb/ft 2 . For natural gas


with k 1.3, this pel is approximately 1.85 times atmospheric
pressure.
Although inlet velocity will not increase as line pressure is
raised, inlet mass flow will increase in direct proportion to pressure
(neglecting supercompressibility) because of increased flowing
density. If Ai is the minimum inlet flow area in square feet
(valve area X flow coefficient), this mass flow may be calculated
from the following:
m = 8.02

-JET SHIELD SLOTTED


l/V'W.T. PIPE

Fig. 2

HEAVY MESH WIRE

Example of silencer design showing important flow dimensions

Avpo
RT0

2/(A--ir

\k + lj\k

lb/sec

+ 1/

(5)

Under continuing line pressure increases, the exhaust, velocity


(F 2 ) from the silencer will increase because of the increased mass
1

Numbers in brackets designate References at end of paper.

Journal of Engineering for Industry


Downloaded From: http://energyresources.asmedigitalcollection.asme.org/ on 04/17/2015 Terms of Use: http://asme.org/terms

MAY

1 97 1 / 697

PIPELINE

PRESSURE, p . p s i c i

Fig. 3 Difference in open stack noise and that regenerated at silencer exhaust by turbulent
mixing of the discharge gas

a:
u m

a = 0.9

35

a=o.8
a =0.7

30

15 Q
oa o

25

8*

20

/ '

^j

~--T.L

= &i- 4 . 2 6

a
T
d2 '

3 a:

a: CQ
o n
J Q
1-'
SILENCER

LENGTH

TO

DIAMETER

RATIO f - f )
x
d2'

Fig. 4 Transmission loss due to diffuser and absorbing section for high-pressure cylindrical
silencer as shown in Fig. 1

flow, and pressure in the silencer body will remain virtually at


atmospheric.
3 When a second critical line pressure (pa) is reached, mass
flow will be increased to the point that sonic flow (Mach 1) will
be experienced at the silencer exhaust as well as at the silencer
inlet. This second critical pressure is a function of both gas
characteristics and silencer diameter ratio, and can be expressed
as follows:
k/(i-k)
Pel

Patm

)( k + 1

(6)

Upon definition of silencer flow, it is then possible to use


acoustic theory to predict the noise attenuation achieved. In
such a silencer, there are two major noise sources: the diffuser
and the silencer outlet. Diffuser noise is several dB below open
stack noise and even this is reduced by passage through the absorbing section. The silencer outlet is a source of regenerated
noise and can be reduced only by controlling exit velocities. For
an optimum silencer design, that noise transmitted through the
silencer should just equal that regenerated at the outlet.

If the power regenerated at the silencer exhaust is severe, it is


apparent that the silencer will afford b u t little benefit. Similarly, there is a practical limit as to the absorption loss which is
useful for any given outlet regeneration level. For example, if
diffuser noise amounts to 150 dB and regenerated noise is 140
dB, there is little value in providing much more than about 10
dB of absorption within the silencer.
In order to design a silencer such as shown in Fig. 2 for a
specified total noise reduction under a given set of operating
conditions, it is necessary to predict both the regeneration level
at the silencer outlet, and the attenuation of incident noise experienced by the action of the diffuser and absorbing sections.
Graphical techniques for designing silencers for predictable performance characteristics are given in the following paragraphs.
Lighthill's equation (equation (3)) provides a technique for
calculating the regenerated noise level a t the silencer, or more
conveniently, the regenerated drop SN, which we define as the
dB difference between open stack (no silencer) and silencer
regenerated level. This 8N can be obtained directly from the
design chart in Fig. 3 [1, 2].
Similarly, to find the reduction of open stack noise afforded

698 / MAY 197 1


Downloaded From: http://energyresources.asmedigitalcollection.asme.org/ on 04/17/2015 Terms of Use: http://asme.org/terms

Transactions of the AS ME

).5F

i(sec)

Z0T0J

(7)

\Patm

where
V = pressurized volume, ft 3
A = constriction area (valve throat area X flow coefficient), ft 2
SO = specific gravity
Zo = supercompressibility factor (at pressure)
To = initial temperature, deg R
po = initial line pressure, psia
Patm = atmospheric pressure, psia

m
UJ

Such computational techniques can also be reduced to graphical


form. An example for a 30-in. pipeline is shown in Fig. 5.

2:00
IN

2:30

453
3

HOURS

HOLES

/4"

C -

'-V' D I A .

AT

C.

Fig. 5 Approximate blowdown time for 30-in-OD pipeline with two 8in. blowoff valves

by the diffuser and absorbing section, defined as AN, we go


directly to Fig. 4 [1, 2]. Since these two values, <5iV and AN,
are defined as levels below open stack noise, they now define the
intensity of the two silencer noise components: transmitted and
regenerated noise. For example, if open stack noise is 130
dB at 50 ft, SN = 20 dB and AA^ = 20 dB, then with the silencer
in place we would have two noise sources: that transmitted
through the silencer ( = 110 db) and that regenerated at its outlet (also = 110 dB). Total combined noise level would be 110 +
3 = 113 dB, from standard dB adding techniques. Thus the
noise reduction afforded would be 20 3 = 17 dB.

I8"0.D.

ORANGE

PEEL
453
45MIN.

20"O.D.

ELLIPTICAL

Blowdown Time

- W " DIA. H O L E S
AT
3 / 4 " O . C . IN A D J A C I E N T
ROWS
STAGGERED

WELD

CAP

12 H O R I Z O N T A L
!/4DIA. H O L E S
AND
3/4" APART

The mass flow rate through a sonically choked blowoff valve


is given in equation (7). The only restriction on applicability of
the equation is t h a t critical ratio exists across the valve and
therefore sonic flow is established in the throat. Note that
downstream pressure does not affect mass flow under these
conditions. During blowdown, upstream pressure falls and
mass flow varies directly with resulting changes in flowingdensity.
The time required for blowdown of a pipeline section or pressurized reservoir can be defined from the equation:

I8"0.D.

HEMISPHERICAL

Fig. 7

WELD

R O W S OF
@ 3/4 " C - C
(453)

CAP

Types of diffusers tested

130

120

BLO'W- >FF
Wl r H O U T
S1L ; N C E R -

>

UJ 1 0 0

a:
D

(/)
m
~

'*

,
90
f

'^]

80

E L L I >TICAL
. WEL ICAP
D I F F U : ER

*-*v
ORANG E
PEE L
D I F F U >ER

>v

'.

\
V
70

31.5

63

125

250

FREQUENCY

500
IN

1000 2000 4000

CYCLES

PER

8000

16000

SECOND

ALL
PASS

Fig. 6 Noise levels comparing weldcap diffuser and a special fabricated orange peel
diffuser

Journal of Engineering for Industry


Downloaded From: http://energyresources.asmedigitalcollection.asme.org/ on 04/17/2015 Terms of Use: http://asme.org/terms

MAY 1 97 1 / 699

130

_
-0

iao

BLOW-CFF
WITHOUT
SILENCER
..

__
J

no

S'

100

90

D1FFUSER

D
in

ONLY

\
\\

z
D
O
10

70
125

31.5

250

500

FREQUENCY
Fig. 8

IN

1000

2000

CYCLES

4000

PER

8000

16000

SECOND

ALL
PASS

Noise levels s h o w i n g p e r f o r m a n c e of diffuser a l o n e

120

B L O W - )FF W THOUT
SIL : N C E R -

110

a:

100

^ ^ f c ^

(/)
en

90

^c;

^ ,

,y
ITH

6F T. S I L I NCER
IN
PLACi

\
\>

80

70

31.5

63

125

250

FREQUENCY
Fig. 9

500
IN

2000

4000

PER

SECOND

8000

16000

ALL
PASS

Noise levels s h o w i n g p e r f o r m a n c e of 6-ft silencer

Field Tests
A series of field tests were conducted in order to optimize
silencer design for the particular pipeline conditions under which
it was to be used. Prime requirements were lowest possible
weight and compactness as well as applicability to a wide range
of line pressures. Of particular interest in the tests, conducted
under typical field conditions, were predictability of noise reduction, optimizing transmission loss and regeneration loss,
shell wall thickness and diffuser design for maximum noise reduction without flow choking. Another goal of these tests was
a decision on the amount of noise attenuation actually needed.
This amount would vary at each valve location due to masking
noise, nearness of residents, etc. Results of these tests are
shown beginning with Fig. 6. A summary of measured and predicted noise reductions for these tests are given in Table 1 together with illustrations of basic configurations used.
Most previous silencer models fabricated according to the design techniques outlined herein utilized a perforated capped pipe
diffuser and field results show a 7 to 9 dB reduction from the
diffuser itself. During the present field tests three other dif700 / M A Y

1000

CYCLES

fusers were tried; viz., (a) an "orange peel" welded diffuser


which in appearance was between a hemisphere and a cone,
(b) an ordinary20-in. elliptical weld cap, and (c) an 18-in. hemispherical weld cap. Diffusers are illustrated in Fig. 7. With
this type of diffuser the need for the steel shield covering the
bottom section of the fiberglass absorbing material was not
needed, thereby effecting a reduction in silencer weight. Fig. 6
shows a direct comparison of the elliptical weld cap and orange
peel diffuser. The weld cap silencer gives a reduction of 18
dB compared to a theoretical drop of 17 dB. The orange peel
diffuser, however, showed a total reduction of 25 d B ; it is assumed that this is due to a more diffusing hole pattern. Later
tests using a hemispherical diffuser show its performance to be
very near that of the orange peel. In all cases the total number,
size, and spacing of the holes were held constant.
In one test the hemispherical diffuser was tested with the
silencer shell and absorbing section removed. A reduction of
10 dB was achieved as shown in Fig. 8. However, at 1000 cps,
17 dB reduction was accomplished.
Also during these tests, shell wall thicknesses of 0.375 in. and
0.560 in. were evaluated, with no measurable difference in noise

19 7 1

Downloaded From: http://energyresources.asmedigitalcollection.asme.org/ on 04/17/2015 Terms of Use: http://asme.org/terms

Transactions of the ASME

3 6 " O. D.

FT
-LIFTING

MAX . OPERATING
PRESSURE 750PSI
W/42 % M A X I M U M
OPENING 8"VALVE

LUG

- 3 6 " O . D . X.375"W.T.
GR. " B u PIPE
r*

-EXPANDED
METAL

18" HEMISPHERICAL HEAD


DIFFUSE R w / 4 5 3 _ |/4> D IA.
H O L E S IN A D J A C I E N T
iROWS AT 3U" O.C

-INSULATION
MATERIAL

1/2" THK. GUSSETS

8 " - 6 0 0 * R.F.W.N.
FLANGE

3"ADJUSTABLE
PIPE
LEG

Fig. 10

Details of portable high pressure silencer

130

120

B L O W - IFF
W THOUT
SIL :NCER-

"7
x

y no
too

^i

m
m
S 90

'

, ^

s
a
z

ONE
01 14
FABRIC. M E D

\\

^ -

80

SI .ENCER S
10
SPE . I F I C A T ONS

\ >. - " '

D
O
CO

\\

\>

70

31.5

63

125

250

FREQUENCY
Fig. 11

500
IN

1000 2000 4000

CYCLES

PER

8000

SECOND

16000

ALL
PASS

Noise levels showing performance of silencer fabricated to specifications

level. While these data do not quantitatively define noise


transmission loss of the shell, they do show that the 3/s-in.
thickness is sufficient. Available lab tests show that for steel
pipe, transmission loss (TL) can be approximated from:
TL = 17 log (tf) - . 6, dB
where

t = thickness, in.
/ = center frequency of band for which TL is defined.
Fig. 9 shows data for a silencer of similar design except that
total length was reduced from 10 ft to 6 ft. Under field conditions, measured noise reduction was 14 dB, compared to a predicted 14 dB. While agreement with theory is good, reduction
afforded by the 6-ft model is substantially below the 10-ft version

Journal of Engineering for Industry


Downloaded From: http://energyresources.asmedigitalcollection.asme.org/ on 04/17/2015 Terms of Use: http://asme.org/terms

MAY 1 9 7 1 / 701

A further series of tests were conducted on one of these to define its noise and flow characteristics. Noise test data are given
in Fig. 11, showing a total overall attenuation of 23 dB. Flow
parameters are shown superimposed on Fig. 10 and measured
velocity profile is shown in Fig. 12. I t seems apparent that regenerated level could be dropped further by changing diffuser
hole pattern or hole orientation to better equalize flow profile
across the cross section.

\
\
1
/

w^
0
Fig. 12

10

/
1

Conclusions
Major conclusions from the studies are as follows:

r^'

15
INCHES

20

25

30

Velocity profile taken at silencer exit

for the operating conditions encountered. I t should be noted,


however, t h a t if line pressure were 900 psi, the 6-ft model would
perform as well as the 10-ft model. This again emphasizes the
importance of designing for the specific application involved, or
for a portable silencer, designing for the extreme application.
Based on results of these tests, design of the silencer was chosen
as shown in Pig. 10, and 14 were fabricated for use throughout
the pipeline system for pressures up to 750 psi and valve sizes
up to 8 in. (42 percent opening).

1 The silencer designs shown perform effectively in reducing


pipeline blowoff noise, as substantial noise reductions can be
achieved for blowoff with up to at least 8-in. 42 percent openingvalves at 750 psi with a portable silencer model.
2 For most pipeline applications, a wall of thickness of 3 /s in.
is sufficient for transmission losses up to 20 dB.
3 The theory presented is generally adequate for arbitrary
diffuser design. In these studies, a hemispherical diffuser affords
additional attenuation.
4 Diffusers should have sufficient holes such that the 42
percent opening valve is the minimum flow area. Diffuser area
in these tests was equivalent to one 5.31-in. hole compared to a
5.21-in. equivalent dia for the valve.

References
1 Damewood, Glenn, Sparks, Cecil R., et al., "Blowoff Noise
Suppression and Regulator Valve Noise Generation," Noise Abatement at Gas Pipeline Installations, Vol. I l l , American Gas Association,
Catalog No. 39/PR, Nov. 1961.
2 Sparks, Cecil R., "Design of High-Pressure Blowoff Silencers,"
JASA, Vol. 34, No. 5, May 1962.

702 / MAY 197 1


Downloaded From: http://energyresources.asmedigitalcollection.asme.org/ on 04/17/2015 Terms of Use: http://asme.org/terms

Transactions of the AS ME

Potrebbero piacerti anche