THE AMERICAN SOCIETY OF MECHANICAL ENGINEERS 87-GT-61

345 E. 47 St., New York, N.Y. 10017

The Society shall not be responsible for statements or opinions advanced in papers or in dis-
cussion at meetings of the Society or of its Divisions or Sections, or printed in its publications.
Discussion is printed only if the paper is published in an ASME Journal. Papers are available
from ASME for fifteen months after the meeting.
Printed in USA.
Copyright 1987 by ASME

Low Frequency Noise Emission from a Natural Gas

Compressor Station
MANFRED SIEMINSKI
BEB Erdgas and Era! GmbH
Steimbke, Germany (FRG)
MANFRED SCHNEIDER
Werner Genest Consulting Engineers
Ludwigshafen, Germany (FRG)

ABSTRACT Figures 1 and 2 show the measured sound

pressure level at a distance of 300 m in
Low Frequency Noise at Gas Turbines front of the windows of nearby houses on the
basis of a 1/3 octave band analysis and on a
A natural gas compressor station that small band frequency analysis. The peak of
was equipped with Hispano Suiza Turbines THM the sound pressure level particularly at
1202 emitted high intensity noise between 31.5 Hz is quite remarkable.
20 Hz and 40 Hz, causing window vibrations
and standing waves within the living rooms
80
of a nearby residential area.
dB
Since additional sound attenuation by
70 :..!.
increasing the volume of the exhaust
silencers was impossible, further
investigations were carried out to explain ....2 't
the mechanism of this low frequency noise 60
emission. By changing the flame pattern %....

inside the combustion chambers of the

turbines it was possible to achieve a 50
remarkable reduction at 31.5 Hz amounting to
15 dB. The investigation procedure leading
to the final results will be the subject of
40 ----
this presentation.

HISTORICAL BACKGROUND
30
A natural gas compressor station
belonging to BEB in Wardenburg, West Germany **,...*
comprises two HISPANO SUIZA Gas Turbines of 20
the type THM 1202 and one unit of the type E-----2
THM 1203. Whenever the 1202 units were in
operation at full load (gas genera or t 10
revolution between 10600 and 11600 min ), 16 32 63 125 250 500 1000 2000
people from the nearby residential area
complained about an annoying low frequency Frequency/Hz
noise emission. Particularly annoying were
the standing waves inside the sleeping and Curve 1: Spectrui ii of soundpressure level
-

living rooms and the rattling of the glasses at 300 m (flat)

on the shelves in cupboards. Although the
Curve 2: Curve 1 A-weighted
normally used A-weighted sound pressure
level was within the acceptance limit
required by the local authorities on the Figure 1
grounds of the low frequency noise further
endeavours were made to reduce the low
frequency noise emission.

Presented at the Gas Turbine Conference and Exhibition, Anaheim, California May 31-June 4, 1987

Downloaded From: http://asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/conferences/asmep/83701/ on 06/28/2017 Terms of Use: http://www.asme.org/about-asme/term

 ED
The two combustion chambers of this
dB industrial gas turbine are located on top of
70 the gas producer in V-position.

After leaving the diffusor the

bo
compressed air has to be deflected several
times before entering the flame tubes as
50 shown in Figure 3.

40 In the original design the primary air

passes the swirler in horizontal direction,
30 nearly at a 90-angle related to the fueT
10 20 z 50 100 26 0 500 1
injectors.
Frequency/Hz
ACOUSTICAL INVESTIGATION INTO THE MECHANISM
Soundpressure level at a distance of 300 m CAUSING EXTRAORDINARY LOW FREQUENCY EMISSION
THM 1202, at the beginning of the investigation
Measurement Procedure
Figure 2
The measurements were concentrated on
TECHNICAL FEATURES OF THE GAS TURBINES, readings of the sound pressure level at the
TYPE THM 1202 exhaust stack aperture, the structure-borne
sound on the stack surface, the sound
Type of turbine : two shaft pressure at a distance of 10 cm from the
engine exhaust elbow just behind the power turbine
Year of startup : 1971 and on readings of the static pressure of
Output power : 4500 kW the exhaust gas in the entire exhaust
Max.. gas producer speed :11600 min system. The array of measurement points to
Max. power turbine speed : 8000 min
1 determine that the static pressure was in
Efficiency 22 % the exhaust elbow in the middle of the
Max. power turbine exhaust stack where the silencer was located
: 625C
inlet temperature and at the end of the stack.
Exhaust gas temperature : 480C
Max. air inlet mass flow at 15C: 31 kg/s

Igniter Torch
Fuel Injector (Burner Nozzle)

Burner Dome with Swirler (Original Design)

Flame Tube (Combustor) Exhaust gas

Power Turbine

Air
Inlet

.0

Compressor Rotor

HP-Turbine Exhaust Elbow

Coupling
HISPANO SUIZA THM 1202
Simplified Turbine Engine Airflow Diagram
Figure 3

Downloaded From: http://asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/conferences/asmep/83701/ on 06/28/2017 Terms of Use: http://www.asme.org/about-asme/term

 The Time-related Sound Pressure Level at
the Starting Point of the Investigation
From figure 4 it can be seen that the
time-related sound pressure level measured
at the exhaust elbow and at the stack
aperture have the same fluctuation period,
hence there seems to be an interaction
between the sound emission at the stack
aperture and at the exhaust elbow. The
fluctuation period lies between 10 and 12
Hz.
The frequency analysis of the emitted
sound pressure level at the stack aperture
is shown in figure 5. As can be seen, there
is a fairly even distribution of the sound
pressure levels for the frequencies below
the 1/3 octave band of 31.5 Hz. At the 1/3
octave band of 31.5 Hz, however, the sound
pressure level peaks sharply, which is
tantamount to a sound power level of 130.2
dB at 31.5 Hz. In comparison to this high
noise intensity the A-weighted sourq power
level yields only 93.8 dB(A) (re 10 W).

120 The same information was delivered by a

Fast Fourier Transformation (FFT) analysis
110 (figure 6). The curser position indicates
the maximum at 32.5 Hz.
a
100

90
THM 1202
80 01-17-86 10:25 Band: 100 Hz nGG. 11190 min -1

O
70 dB

Curve 1: Soundpressure level

at the exhaust elbow
Curve 2: Soundpressure level at
the exhaust stack aperture
4)
Frequency: 31,5 Hz a
Figure 4

mTitle , BEB-Wardenburg n GG-11490

"L H= 3 31.5 Hz 1/3 FAST
60
130,2 di 32.5 100.0Hz
120 0.0

<.;, 110
FFT - Analysis of the sound pressure level
?..! 100 at the stack aperture
90
Figure 6
80
70
s_
60
50
0_
40 Determination of the Stack Resonance
0
4.0 16 63 250 1k AL To clarify the question of whether stack
Frequency/Hz resonances are responsible for this high
sound pressure level at 31.5 Hz, the hot
THM 1202 original burner nozzle and swirler skin of the stack was excited with a rubber
hammer to measure the decay of the velocity
level, see figure 7. The result was a
Black columns: Energy equivalent sound pressure level (Leq)
resonance frequency of the stack skin of
over the period of measurement time
approximately 22.5 Hz and so it could he
Gray columns: Maximum sound pressure level (Lmax)
concluded that the sound emission at 31.5 Hz
over the period of measurement time
was not caused or enhanced by vibrations of
Figure 5 the stack skin.

Downloaded From: http://asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/conferences/asmep/83701/ on 06/28/2017 Terms of Use: http://www.asme.org/about-asme/term



to install a new exhaust elbow, of which a

250mm/s
lower back pressure of the exhaust system
could be expected. The new elbow
configuration allowed a higher exhaust speed
of up to approximately 70 m/s compared with
50 m/s before. There was a slight
improvement particularly in the summer, with
high outside temperatures due to a smaller
25 io Y 10 , 5 11 , 5 mass flow through the turbine. No
A improvement could be measured
tine.
at winter
After this disappointing experience,
IF V further tests were carried out by changing
the flame pattern in the combustion chamber.
=1 0 ,0 4 8
Natural Gas Fuel

resonance frequency
of the stack skin
f o 22,5 Hz

Figure 7

burner nozzle
Relationship between the Pressure
Fluctuation of the Exhaust Gas and the Sound
Generation at 31.5 Hz

A close relationship was found between

the static pressure fluctuation inside the
exhaust elbow with a period of approxi-
mately 9 s and the sound pressure
fluctuation with a period of approximately
10 s. This information can be drawn from
figure 8.

INITIAL CONCLUSION AND DECISION WITH REGARD

TO IMPROVEMENT
Previous swirler
Due to the fact that there was an
obvious relationship between the pressure
fluctuation inside the exhaust elbow and the Figure 9
sound generation at 31.5 Hz it was decided

OM= MIN MI NIMMIIMIIIII MN MIMI =

WWI 1=11111111111= MIEMIIIMMIMI11111 MI
M UM MI =I
111=111MMIMIIIII IIMMIM11111
I= imummismommumrimmium.
mom minummummuusrminammmonimm.
winws,..simumagolwaluomplumunimwstirawcammisommor
mum . rapmrsammosmribioamaimnrommuliagamomourrax w
"TANWAIIIILT latAW/MINIMEIVKL1111,2111=1111117/111Pur 1=11.11.1MILIVriall.=
'WA 11111=11111111.111.1411111.11MILINIIILICAME=113111111111MI
90 dB INNIMN11111=1111MMINIENIONMEIMINIMIIIMEGIIMMININIM11
IMINIMMIIIIMMINIIMMENMIIIIMMINEMIIIMINM11111111111
MINIMIIMMEIMINKIMMIIMIIIMININIMI111111=1111=11111
MM= IIMINIMINNIIM.CmINNIIMI1MIPPIMIMEIMIMIIMMILINIEL
fr.I1M1=1.111P. M=1.111100 1 11M111111/.11 1111.7AMi=^2.1110..11.11 1
" IMILM11/1.0:-Ar.WAIIHNSIVV0=1111111E -J.111" 14110/011111111110.111iLgralF
E IMMINIMIIIIWAILMANNIIIMMLINIMIUNINIIIEGMMIONGIIIMIIMIIMILII

riza ussorma
MEM
Lim om
'MEW
la
mm Egr
nil=114
WW1 - 9s
MI ENSIMEMIMIN,
MINI
Ws
ms.=..ummiri

ow
mummon-im N.
mi......m. wil imemagsm ow
m
mi sisrmem. moommasnal
Curve 1: Soundpressure level at the stack
aperture frequency:
1/3 oct. at 31.5 Hz
Curve 2: Static pressure in the exhaust
elbow

Figure 8

Downloaded From: http://asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/conferences/asmep/83701/ on 06/28/2017 Terms of Use: http://www.asme.org/about-asme/term

 CHANGING THE MIXTURE ZONE BETWEEN GAS AND
AIR AT THE BURNER NOZZLE
BEB-WARDENBURG THM 1202 - E2
During routine maintenance, the nozzle 86/01/24 19:29 BAND: 100 Hz BOG - 11070 min - 1
core of the burner had shown traces of
120
erosion presumably caused by cavitation.
dB
This fact might have had an effect on the
low frequency noise generation. As a logical
consequence, the insertion depth of the
burner nozzle into the swirler was varied.
When the nozzle had been moved up by only a
few mm, an even higher noise generation
occurred. On the other hand, moving the
nozzle down between 1.6 and 2 mm enabled an
inpressive improvement to be achieved, see
figures 10 and 11. The sound pressure level AVL

at 31.5 Hz had no peaks anymore; the

spectrum shows a very smooth configuration.

Tit1e:BEB-Wardenburg 1202-E2 NGG=11070

0 N= 1 31.5 Hz 1/3 FAST

120ch 1 Lec.Lt 83.6 Max= 95 2

110
40
100i 0.0 28.0 100.0Hz
s_
90 ,

FFT - Analysis of the soundpressure level at

80 the stack aperture burner nozzle 1,6 mm deeper
70 than specified by the manufacturer

60
0 Figure 11
s_
40

0 16 63 250 1k 4k AL
(r)

Frequency / Hz
THM 1202 with previous swirler, insertion depth
1.6 Me deeper than specified
Black columns: Energy equivalent sound pressure level (Leq)
over the period of measurement time
Gray columns: Maximum sound pressure level (Lmax)
over the period of measurement time
Figure 10
Explanation of the Test Result
From literature it is known that a
steady combustion occurring in fuel-fired
heating plants is sometimes disturbed by
regular oscillations in the combustion
process. They can arise as a consequence of
pulsations in fuel and air supplies caused
for example by fans, pumps, meters and
unstable control loops on flow and pressure
regulators. A second source are aerodynamic
instabilities within the supply system or at
the burner head such as a vortex shading
phenomenon. A third type of oscillation
occurs when there is an acoustic coupling
between the flame, the supply lines and the
combustion chamber. According to H.J.Merk
/1/ and investigations undertaken at the
Midlands Research Station /2/ it is known
that the combustion-driven oscillation in a
flame can be mathematically described in a
transfer function. The flame transfer
function describes the way in which the
flame responds to a disturbance in the
gas/air supply at the burner head. In the
investigations referred in References /2/,
/3/ and /4/ special types of burner and
swirl design were studied allowing flame
stabilisation to be successfully improved
and leading to a significantly lower noise
intensity.
As the manufacturer HISPANO SUIZA did
not allow an insertion depth of the burner
nozzle of 1.6 to 2 mm beyond the specified
level, another swirler was tested, see
figure 12 and 13. The use of this swirler
allowed the rectangular stream direction
between air and gas to be avoided. The
result was an even greater decline in noise
generation, see figure 14. The emitted sound
power yielded 112.8 dB at 31.5 Hz compared
with 130.2 dB at the beginning of the tests.
A reduction of approximately 17 dB could be
achieved. The reduction of the A-weighted
sound power level at the exhaust stack
aperture was approximately 6 dB(A).

Downloaded From: http://asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/conferences/asmep/83701/ on 06/28/2017 Terms of Use: http://www.asme.org/about-asme/term



Ti t le :biardenburg THM 1202-E1 nGG=11620

N= 1 31.5 Hz 1/3 FAST
,T3

120 Ch. 1 85.3 - 112,6 aB

1
110
, 100
L 90
80

Primary Air > 70

Inlet 60
,7) 50
L 40

16 63 250 1k 4k AL
0

new swirler Frequency / Hz

Figure 12
THM 1202 with a new swirler

Black columns: Energy equivalent sound pressure level (Leq)

over the period of measurement time
Gray columns: Maximum sound pressure level (Lmax)
over the period of measurement time

Figure 14

loss at this particularly low frequency

thicker splitters or longer silencers will
have to be installed. In any case the result
will be a greater volume of the exhaust
silencer. In such a situation it seems
useful to analyse the origin of the noise
source, which can for instance be either the
combustion chamber or the flame
configuration.
1 previous burner dome and swirler
2 new burner dome and swirler

REFERENCES:
Figure 13
/1/ Merk, H.J.:
An analysis of unstable combustion of
This low sound emission is exclusively premixed gases. 6th Symposium
attributable to the improved and more (international) on Combustion,
uniform combustion behaviour with a more pp.500-512 (1956).
stable flame configuration, resulting in an
equally favourable effect on exhaust gas /2/ Fricker, N. and
emission from the turbines. Later emission Roberts, C.A.:
measurements showed a considerably reduced An experimental and theoretical approach
CO content of the exhaust gas: to combustion driven oscillations.
At 40 % load for instance the CO content Schwingungen in Feuerr'6umen, Heft 433
was reduced from 170 mg/m 3 previously to (Volume 433) published by Vulkan-Verlag
75 mg/m 3 , whilst at full engine load the CO Dr. W. Classen, Essen (W. Germany)
content of the exhaust gas decreased from
96 mg/m 3 previously to a mere 15 mg/m 3 /3/ Leuckel, W. and
(referred to 15 % 0 and dry exhaust gas). Fricker, N.:
2
The characteristics of swirl stabilised
CONCLUSION natural gas flames - Part 1. Journal of the
Institute of Fuel, pp. 103-112 (1976).
From time to time, low frequency noise
emission from exhaust stacks of gas turbines /4/ Fricker, N. and
can cause environmental disturbances. The Leuckel, W.:
first step to overcome such a problem is to The characteristics of swirl stabilised
think about a redesign of the exhaust natural gas flames - Part 3. Journal of the
silencer. To achieve a greater insertion Institute of Fuel, pp. 152-158 (1976).

Downloaded From: http://asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/conferences/asmep/83701/ on 06/28/2017 Terms of Use: http://www.asme.org/about-asme/term