Noise reduction of a multistage export / reinjection - Dresser-Rand

NOISE REDUCTION OF A MULTISTAGE EXPORT / REINJECTION CENTRIFUGAL
COMPRESSOR THROUGH THE USE OF DUCT RESONATOR ARRAYS (TP081)
Abstract
By
Zheji Liu and Mark J. Kuzdzal
Dresser-Rand Company
And
Einar Flood Jr.
Statoil Company
Most industrial centrifugal compressors absorb large amounts of power and produce significant
head rise. This is accomplished by imparting velocity into the working fluid using impellers and
by diffusing the flow downstream of the impeller. As a result, noise is generated by the
compressor as an undesirable by-product. The noise signature of a centrifugal compressor is
typically dominated by the discrete frequency noise occurring at the blade passing frequency
and its harmonics. Excessive noise is not only an environmental hazard, but also can indicate
poor health of the machine. It can take a considerable amount of resources to address the noise
issue of a compressor. Historically, this challenge was addressed by treating the sound
transmission path. This involves using acoustic lagging, sound insulation, sound enclosures and
ear protection.
Within the past few years, a new centrifugal compressor noise-reduction device was developed
to reduce compressor noise. The intent of this device is to attenuate the noise as close as
possible to the source. This approach includes mounting duct resonator arrays inside the
compressor flow path to minimize the internal acoustic energy. Over the last few years more
than sixty-one centrifugal compressors, both single stage and multistage, have been equipped
with this technology. This paper presents an application of duct resonator arrays applied to a

2528 PSIG (172 BARG) multistage centrifugal compressor on a platform in the North Sea. This
compressor is used in a gas export process.
Comparative field noise data taken from the compressor with duct resonator arrays and a
second compressor of the same design but without duct resonator arrays will be shown in this
study. The compressor with the duct resonator arrays was commissioned in September 2002
and has been running successfully. The field data presented in the paper will show that noise
reduction of 12 dBA can be achieved without the use of traditional noise insulation products.
Introduction
The Sleipner field, as depicted in Figures 1 and 2, consists of two major gas and condensate
processing platforms (Sleipner A and T) plus a wellhead platform (Sleipner B), and a pipeline
riser platform (Sleipner R) located in the Norwegian part of the North Sea. The Sleipner A
platform started production of gas and condensate in August 1993. Sleipner T started in
September 1996. Statoil operates the Sleipner field with partners ExxonMobil, Norsk Hydro and
Total.
The Sleipner T processing platform and the Sleipner B wellhead platform have a production
6 3
capacity of 554,200 SCFM ( 22. 6×
10 Sm
/day) of dry sales gas and 235 SCFM
3 3
( 9. 6×
10 Sm
/day) of condensate. An overview of the Sleipner Field is given in Figure 3.
As the reservoir pressure gradually decreases in the Sleipner field, the natural gas
recompression demand increases to maintain the same export capacity. In the summer of 2002
both of the export compressors and gears (trains A and B) on the Sleipner T platform were

evamped. The revamped configuration was a five-stage straight-through centrifugal
compressor. At the design point, the unit power was 17,300 horsepower (12.9 MW) and
compressor inlet and discharge pressures were 1154 psia and 2229 psia. The unit operating
speed was 8200 RPM at design point and the impeller tip diameter was 17.93 inches. A picture
of the unit in operation is shown in Figure 4 and a cross section of the revamped five-stage unit
is shown in Figure 5.
This revamp was part of the Sleipner West compression project. With the new higher flow and
higher head demands, it was estimated that the noise would increase 4-6 dB around the
compressors. As a result, the end user was looking for a way to increase flow demand without
increasing noise.
This challenge created an opportunity for the compressor operators to install noise attenuation
hardware inside the compressor. The limited space available on the platform means that the
operators work close to the equipment. As a result, noise control on the platform is an important
issue.
This paper focuses on the noise reduction of an export compressor on the Sleipner T platform.
A unique noise control technique was implemented to one of the compressors during the
summer 2002 revamp. It was different than the traditional approach of treating the noise
transmission paths by using sound insulation to cover the noisy structure; it used duct resonator
arrays to address the compressor noise, internal to the compressor and at the noise source.
The two compressors were revamped with identical designs except that compressor A had a
device called a duct resonator array installed in it and compressor B did not. This arrangement

was proposed by the end user so that the noise attenuation effectiveness of the duct resonator
arrays could be verified side by side once the two compressors were running.
Duct Resonator Array
The duct resonator arrays used in this application took the form of flat plates. They were
mounted in the last stage diffuser, just downstream of the impeller exit, as shown in Figure 6.
This noise control approach has several advantages over the insulation approach. It attenuates
the noise source directly instead of treating the symptom. Being installed inside the compressor
diffuser, this approach minimizes the internal acoustic energy by reducing pressure pulsations.
This internal energy relief not only provides the impeller with a smoother operating environment,
but also means less acoustic energy traveling to upstream or downstream pipes. Less in-pipe
pressure pulsation means less pipe vibration and less noise radiation.
In addition, the duct resonator does not create equipment accessibility issues because it is
internally mounted. Pipe or compressor body insulation not only causes inconvenience in
equipment maintenance and accessibility, but also has the potential to trap seawater under it
that can accelerate corrosion of the insulated structure.
Before this application, the success of using duct resonator arrays to reduce centrifugal
compressor noise had been experimentally validated through comprehensive in-house OEM
acoustics testing and numerous field applications on pipeline compressors. Both in-house
experimental data and field data gathered from a wide range of operation showed that the duct
resonator arrays reduced the noise level by more than 10 dB without adverse effects on
compressor efficiency, head, and range [1-3]. This proven technology has now been applied to

more than sixty-one multistage and single stage compressors. The results presented in this
study represents the first duct resonator array application to a multistage compressor.
The device has an array of acoustic chambers that are connected to the flow path by a series of
perforated openings. The array chambers behave like dead volume to the mean flow but are
transparent to sound waves. The array does not have gas passing through it but does have
acoustic waves oscillating through at all times. The insertion of a duct resonator array in a
diffuser wall changes the diffuser wall from being acoustically rigid to acoustically transmissive.
Consequently, the placement of the duct resonator array changes the diffuser wall impedance;
this determines the noise attenuation performance [4-5]. A well-tuned duct resonator array can
provide remarkably effective noise attenuation in a targeted frequency range with a limited
space requirement. The array was designed to attenuate a frequency range where high
amplitude noise occurs. Because centrifugal compressors tend to produce predominant noise at
the blade passing frequency (BPF) and higher harmonics, tuning the array to attenuate these
frequencies is the ultimate goal.
Compressor Noise Characteristics
In general, the noise signature of a centrifugal compressor is a superposition of discrete
frequency tonal noise peaks on a broadband noise floor. The tonal noise peaks occur at the
impeller blade passing frequency (BPF) and its harmonics. The tonal noise is a result of the
impeller rotation. The noise level increases if stationary vanes (such as diffuser vanes) are used
in the compressor. The broadband noise is mainly caused by the turbulence in the flow.
Because the tonal noise peaks dominate the overall noise level of a centrifugal compressor,
noise control strategy must focus on the reduction of the dominant noise peak level as a first
priority.

The blade passing frequency (BPF) is dependent on machine operating speed (RPM) and
impeller blade count ( N b ) and is given by:
BPF =
N
b ×
RPM
60
Knowing just the blade passing frequency is not enough to understand the compressor noise
characteristics. There are other features in the compressor noise spectrum and they vary from
machine to machine. These include the tonal prominence, peak bandwidth, and relative strength
among the noise peaks (first BPF and second harmonic). Each has essential information for the
design of the duct resonator array.
These noise characteristics have been mostly determined by experimental test data. Narrow
band FFT sound pressure level or narrow band sound intensity level measured from the
compressor provides all the noise characteristics in the spectrum, although the latter requires a
more sophisticated acoustic data acquisition system. The full octave and one-third octave sound
data are easier to obtain but do not provide all the acoustical details and frequency
characteristics.
Physical measurement, particularly with a new compressor design, is not always possible.
Under this circumstance, the computational approach becomes the only choice. Using an
unsteady computational fluid dynamics (CFD) model incorporating a transient sliding mesh
approach is required to simulate the impeller/diffuser aerodynamic interaction. Using this
analysis, the dominant tonal noise peaks of the compressor can be predicted. The current CFD
codes can not predict the broadband flow noise with good accuracy, this is determined
empirically.

The measured compressor noise data in Figures 9-10 show that the compressor overall noise is
dominated by the BPF. The noise peak at the second harmonic also is relatively strong. Both
peaks are sharp with an elevated level of more than 10 dB above the broadband noise. This
indicates that there is a large amount of sound energy surrounding the peak frequencies.
These two figures will be discussed in greater detail later in this paper.
The tonal noise dominance characteristics of a centrifugal compressor match very well with the
noise attenuation characteristics of duct resonator arrays. Understanding compressor noise
characteristics is a vital step to the success of devising a noise reduction solution.
Compressor Noise Generation Mechanisms
Discrete frequency noise at the first blade passing frequency and its harmonics is the dominant
noise source of a centrifugal compressor. Understanding how and where it is generated is
another important step to develop a noise reduction solution.
There are two types of discrete frequency noise. The first one is the rotational noise associated
with steady forces exerted by the rotating blades on the fluid being compressed. The rotation of
the rotor blades converts a steady, circumferentially varying pressure field around the surface of
the rotor blades in the rotating frame of reference into a periodic time-varying pressure field
experienced in the stationary frame of reference. In addition, the flow field surrounding the rotor
blades is rarely circumferentially uniform because a centrifugal compressor is geometrically
non-axisymmetrical and often has obstructions (i.e., vanes) in the flow path. Any flow
nonuniformity or distortion felt by the rotor blades cause a periodic fluctuating force on the rotor.
For a compressor with vaneless diffusers, this type of rotor-alone discrete noise is the prominent
noise source.

The second type of discrete frequency noise is the interaction noise, generated by the
aerodynamic interaction between the rotating impeller blades and the stationary vanes (e.g.,
diffuser vanes or low solidity diffuser vanes (LSD)). The discrete frequency noise due to this
interaction can be further classified into wake interaction and potential field interaction. As the
wake from an upstream object (e.g., an impeller blade) impinges on a downstream object (e.g. a
vane), the velocity deficit in the wake causes the downstream object to experience a momentary
change in angle of attack and in velocity because of the variation of the in-flow velocity in the
wake. Consequently, a fluctuating load is imposed on the downstream object and a dipole noise
source is generated.
The potential field interaction is caused by the local periodic interference between the potential
field of the upstream object such as an impeller blade and the potential field of the downstream
object such as a vane. These potential fields interact with each other as the impeller rotates.
That is, the pressure field of an impeller blade disturbs an LSD vane and the pressure field of an
LSD vane disturbs an impeller blade. These disturbances also cause periodic loading variations
that produce tonal noise at blade passing frequency and higher harmonics.
The discrete frequency noise that results from this interaction is commonly associated with
centrifugal compressors using LSD vanes. As a result, compressors with diffuser vanes are
typically noisier than vaneless diffuser designs.
Duct Resonator Array Location and Design Considerations
There are two major reasons why the aeroacoustic noise source is mainly generated in the
region between the impeller exit and diffuser entrance. First, the gas exiting the impeller is of the

highest velocity. High gas velocity directly relates to high amplitude pressure pulsations and
higher acoustic power. Second, impeller/diffuser aerodynamic interaction is strongest between
the impeller and the diffuser. This makes the diffuser an ideal location for the duct resonator
arrays. The closer a noise control device is installed to the noise source, the more effective it
becomes.
In this application, duct resonator arrays are mounted to the diffuser walls using a T-slot
geometry (see Figure 6). The flat plate arrays are horizontally split and the top and bottom
halves are rolled into the top and bottom halves of the diaphragm assembly. Anti-rotation
screws at the horizontal split are used to keep the arrays in place.
The array is a one-piece unitary design machined from a solid steel plate with no bonding or
welding. This eliminates the possibility of any mechanical failure associated with the bonding or
welding process.
Partitions between the array cavities are much bulkier than that of the honeycomb used as the
middle layer of the aircraft engine acoustic liners [6]. This significantly increases the rigidity of
the duct resonator array.
Stress and Modal Analysis
Under normal operation, the static pressure differential across the array is zero; the holes in the
array equalize the pressure. Nevertheless, a stress analysis was conducted assuming the
cavities do not equalize during a sudden depressurization. For the Sleipner T application, a
pressure differential of 441 psi (172 Bar –142 Bar = 30 Bar delta pressure) was applied to the
array.

The step change in pressure was modeled as a static pressure load. A representative section of
the array was analyzed with a constant pressure load of 441 psi across the array. The
maximum stress value is located at the edge of the small holes and has a value of 30 ksi. The
material used was a low alloy steel plate, grade 4140 annealed to achieve NACE limits of
235HB maximum hardness. Because the material yield strength is 55 ksi, generalized yielding
is not expected from this rare load event. Plate deformation is calculated to be 2.1 mils.
Based on measured dynamic pressure pulsations from full load, full pressure test stand data,
dynamic pressure pulsations of less than 10 psi are expected in the application [7]. The
corresponding stress associated with a 10 psi pulsation is 0.7 ksi.
Results of the above analysis are represented in Figure 7, which shows an alternating load of
0.7 ksi and a mean load of 0 ksi for the normal operating condition. Also shown on this plot is
the 30 ksi maximum stress with 0.7 ksi alternating load for a hypothetical emergency shutdown
condition. In either case, infinite life is expected.
The modal analysis was conducted to determine the natural frequencies of the array
membranes.
The frequency of the first mode is well above the expected alternating pressure excitation
frequency. Depending on the operating speed range, the blade passing frequency and
associated alternating pressure excitation is between 2.4 kHz and 2.9 kHz. Membrane
excitation is not expected.
Acoustic Measurements and Noise-Reduction Results

In 2002, two compressors installed on the Sleipner T platform were revamped. Compressor A
was modified from a four-stage compressor to a five-stage compressor with duct resonator
arrays retrofitted into the last stage diffuser. Compressor B was revamped from a four-stage to a
five-stage but without duct resonator arrays. After revamping, compressors A and B are of the
same design except that duct resonator arrays are used only in unit A. With these two
compressors running, a final acoustic test was conducted in January 2003 to determine the
effectiveness of the duct resonator arrays on noise attenuation.
The complex sound field around a compressor consists of contributions from various noise
sources or components such as the compressor, the driver, the gearbox, and other auxiliary
equipment in the vicinity. This generally makes the sound intensity level measurement more
accurate and more reliable than the sound pressure level measurement. The accuracy of the
sound intensity measurement results from the following two factors.
First, sound intensity is a vector quantity that contains the magnitude and the direction of sound,
while the sound pressure level is a scalar quantity that contains only the magnitude information.
The sound intensity measurement can distinguish the targeted noise source facing the intensity
probe and other noise sources behind the probe. On the other hand, the sound pressure level
measurement by a microphone picks up sound from all noise sources existing in the given
environment.
Second, the sound intensity measurement can be conducted very close to the targeted noise
source while the sound pressure level is typically measured by positioning the microphone in
the far field of the sound source, at a distance of 3 feet or 1 meter. Measuring the sound
intensity level close to the sound source improves the signal-to-noise ratio. The sound pressure
measured in the far field has lower signal-to-noise ratio due to background noise.

Because sound intensity data are generally more accurate than sound pressure data for the
given acoustic environment of a compressor installation site, the sound intensity measurement
was used in this study. Before the start of the measurement, surface segments need to be
predefined around the compressor. Sound intensity levels were obtained by scanning the
specified surface segments with sound intensity equipment.
Noise Signature Measurement of Compressor B
Both compressors are installed in a room on the weather deck of the platform and both are
driven by a gas turbine through a gearbox. Because compressor B was not equipped with duct
resonator arrays, its acoustic measurement provides the baseline noise data.
The compressor system was divided into several segments for sound intensity measurement.
The surface segments defined prior to the measurement are shown as white blocks in Figure 8.
Two 20-inch long (50-cm) quarter pipe segments represent the sound emitted from the suction
and discharge pipes. A portion of the compressor casing surface represents another segment of
measurement.
Each segment was scanned individually according to ISO 9614-2 [8]. Using the multiple
analysis capability of the B&K pulse system, the following quantities were obtained
simultaneously: (1) Sound intensity level in both narrow band (FFT) and 1/3 octave band, (2)
mean sound pressure averaged between the phase-match microphone pair of the intensity
probe, and (3) the p-I (pressure-intensity) index that gives an indication of the quality of the
measurement. The p-I index of 2 to 8 dB was observed for all intensity data measured in the

critical frequency range. This indicates the sound intensity measurement is of good quality since
the p-I index does not exceed 10 dB (ISO 9416-2 Engineering grade).
Figures 9-11 present the baseline noise data measured from compressor B for the operating
conditions listed in Table 1. The compressor noise signature shows that the blade passing
frequency (BPF) dominates the overall noise. The noise peak at the second harmonic is also
relatively strong. Both peaks are sharp and have a relatively narrow bandwidth. Extending the
noise attenuation range of the duct resonator arrays to cover both peaks will help reduce the
overall noise level.
Comparing the sound intensity levels of the suction and discharge pipes at the high speed
(8900 rpm), i.e., Figure 9 versus Figure 10, shows that the discharge pipe radiates higher noise
than the suction pipe. This reveals that the last stage of the compressor is the first priority in
noise reduction efforts. The duct resonator arrays installed in the last stage diffuser of
Compressor A were intended to reduce the dominant discharge pipe noise level. Figure 11
shows sound intensity levels of the discharge pipe of compressor B while running at a slower
speed. Comparing figures 9 and 11 indicates the higher noise level occurs at the higher speed.
As expected, the compressor casing (cylinder) did a good job containing the noise. The sound
intensity level from the casing is much lower than that of the suction and discharge pipes, as
shown by Figure 12. The compressor casing is typically a thicker structure and, therefore,
provides higher noise transmission loss than the thinner suction and discharge piping. The
noise level emitted from the casing is more than 10 dBA lower than the suction and discharge
pipes. As a result, the ambient noise is controlled by the noise emitted from the suction and
discharge pipes. Because the dominate noise source is the discharge pipe, the majority of the
remaining data presented will be focused on this pipe.

Acoustic Measurement of Compressor A
The same acoustics equipment, personnel and test procedures used for compressor B were
applied to compressor A. The same surface segments on the discharge pipe, the suction pipe,
and the casing were specified for measurement. The same operating conditions (as practically
could be achieved) run for compressor B, as listed in Table 1, were also run for compressor A.
Figure 13 is the sound intensity spectrum measured from compressor A at the high speed of
8900 rpm. Relative to the baseline data of Figure 9, the spectrum plot of compressor A has
different characteristics. The magnitude of the sound intensity level of compressor A is lower
and the shape of the spectrum is changed. The installation of the duct resonator arrays
significantly reduced noise peaks at the blade passing frequency and its second harmonic. This
indicates that the duct resonator arrays worked effectively.
At 8200 rpm, the sound intensity levels measured from compressor A, as shown in Figure 14, is
also significantly lower than that of compressor B, as shown in Figure 11. The duct resonator
array was effective for not just one condition or speed but for a range of operating conditions.
The duct resonator array does not necessarily give its peak attenuation at the aerodynamic
design condition since the aero design condition is not necessarily the high noise level
condition. The optimum acoustic design condition was selected based upon the high noise level
condition in this application.
To quantify the noise attenuation achieved by the duct resonator arrays, Table 2 compares the
blade passing frequency noise levels on the discharge pipe between compressor A and

compressor B and lists the noise attenuation under each operating condition. Table 3 compares
the overall intensity levels between compressors A and B and lists the overall intensity level
reduction.
Figure 15 overlays the sound intensity plots of compressors A and B at the high-speed
condition. At this high noise level, the reduction of 16 dB at the blade passing frequency and 14
dB at the second harmonic corresponds to an overall intensity level reduction of 11.7 dBA on
the discharge pipe. This demonstrates the great potential of duct resonator arrays on noise
attenuation.
At 8250 rpm, the duct resonator arrays reduced the blade passing frequency intensity level by
12 dB, as compared in Figure 16. The overall intensity level of the discharge pipe was reduced
by 8.9 dBA. Decibels are measured on a log scale. A 10 dB reduction corresponds to a 90
percent reduction of sound power on a linear scale.
Conclusions
Acoustic tests demonstrated that the noise from the Sleipner T compressors was dominated by
the blade passing frequency emitted from the suction and discharge pipes. Comparing the noise
levels measured from compressors A and B confirmed that the installation of the duct resonator
arrays in the last stage diffuser of Compressor A effectively reduced the major noise component
at the blade passing frequency and its second harmonic on the discharge pipe. Consequently,
the overall noise level was effectively reduced.
The highest noise attenuation occurred at the high-speed condition (8900 rpm). This
corresponds to the high-noise generation condition and, therefore, requires the maximum

amount of noise attenuation. A reduction of 16 dB at the blade passing frequency was obtained
at this condition. The overall intensity level of the discharge pipe was reduced by 11.7 dBA. This
demonstrated the great potential of noise attenuation by duct resonator arrays.
At the design condition (8250 rpm), the BPF noise level of the discharge pipe was reduced by
12 dB. The overall intensity level of the discharge pipe was reduced by 8.9 dBA.
The duct resonator array is made of a solid plate with a rugged structural design. Rigorous finite
element analysis and successful operating experience indicate that the duct resonator array is
fit for the intended service. In 2004, the compressor was again revamped to match the new
field condition. When the array, which had run successfully for two years, was removed from the
compressor, inspection of the array revealed the array was in “like new” condition. The new
revamped compressor was outfitted with duct resonator arrays that were tuned to the new
operating conditions.
Duct resonator arrays have been successfully applied to many single-stage pipeline
compressors. This paper documents the first application of a duct resonator array in a
multistage centrifugal compressor. The application of duct resonator arrays enabled the end
user to increase flow and head and still meet noise regulations.
Acknowledgements
The authors would like to thank Mr. Per Gustafsson for conducting both the initial and final noise
surveys. In addition, the authors would like to thank Mr. Bernard Ghim for his solid mechanics
analysis work and Mr. Kevin Majot for his drafting support. Finally, the authors also thank
Dresser-Rand Company and Statoil for allowing them to publish this paper.

References
1. Liu, Z. and Hill, D. L., "Centrifugal Compressor Noise Reduction by Using Helmholtz
Resonator Arrays." Proceedings of the 30th Turbomachinery Symposium, September 17-20,
2001, Houston, Texas.
2. Liu, Z., Jahnke, W., Marczak, M., and Kiteck, P., "Reducing Compressor Station Ambient
Noise Level By Controlling Compressor Internal Noise Source," Proceedings of the
International Pipeline Conference 2002, September 29 - October 3, 2002 Calgary, Alberta,
Canada.
3. Liu, Z., Hill, D. L., and Motriuk, R., "On Reducing Piping Vibration Levels – Attacking the
Source," Proceedings of ASME Turbo Expo: Land, Sea & Air, June 3-5, 2002, Amsterdam,
The Netherlands.
4. Rice, E. J., "Optimum Wall Impedance for Spinning Modes- A Correlation with Mode Cutoff
Ratio." Journal of Aircraft, vol. 16, No 5, May, 1979.
5. Rice, E. J.,"Acoustic Liner Optimum Impedance for Spinning Modes with Mode Cut-off Ratio
as the Design Criterion." 3 rd AIAA Aero-Acoustics Conference, Palo Alto, California, July,
1976.
6. Hubbard, H. H., Aeroacoustics of Flight Vehicles, Acoustical Society of America, 1995.

7. Borer, et. al. , An Assessment of the Forces Acting Upon a Centrifugal Impeller Using Full
Load, Full Pressure Hydrocarbon Testing, Proceedings of the 26th Turbomachinery
Symposium, PP 111-121.
8. ISO 9614-2, Acoustics - Determination of Sound Power Levels of Noise Sources Using
Sound Intensity - Part 2: Measurement by Scanning.

Table 1 - Operating Conditions during Final Test
Table 2 - Noise Attenuation at BPF
Table 3 - Overall Intensity Level Reduction
Figure 1 - Off-Shore Platform at Sleipner
Figure 2 – Close-up of one of the Sleipner Platforms
Figure 3 - Sleipner Fields
Figure 4 - Photograph of the Compressor in Operation
Figure 5 - Multistage Compressor for Sleipner West Compression Project
Figure 6 - Multistage Compressor with Duct Resonator Array
Figure 7 - Goodman Diagram for the Duct Resonator Array
Figure 8 - Definition of Suction/Discharge Pipe and Casing Segments
Figure 9 - Compressor B (without array) Running at 8903 RPM; Discharge Pipe Overall
Intensity Level = 102 dBA
Figure 10 - Compressor B (without array) Running at 8903 RPM; Suction Pipe Overall
Intensity Level = 97.3 dBA
Figure 11 - Compressor B (without array) Running at 8250 RPM; Discharge Pipe Overall
Intensity Level = 95.1 dBA
Figure 12 - Compressor B (without array) Running at 8250 RPM; Casing Overall Intensity
Level = 78.5 dBA
Figure 13 - Compressor A (with arrays) Running at 8858 RPM, Discharge Pipe Overall
Sound Intensity Level = 90.3 dBA
Figure 14- Compressor A (with arrays) Running at 8200 RPM; Discharge Pipe Overall
Intensity Level = 86.2 dBA
Figure 15 - Discharge Pipe -Comparison at High Speed
Figure 16 – Discharge Pipe Comparison at Design Speed

Table 1 - Operating Conditions during Final Test
Comp TS
[°C]
PS
[bara]
TD
[°C]
PD
[bara]
Speed
[rpm]
Flow
[sm 3 /h]
1× BPF
[Hz]
A 19.4 69 87 154 8200 610000 2597
B 19.3 69 86 154 8250 660000 2613
A 19.6 68 91 154 8858 756000 2805
B 19.4 68 90 155 8903 821000 2819
Table 2 - Noise Attenuation at BPF
Operating points BPF Reduction on Discharge Pipe (dB)
8900 rpm 16
8250 rpm 12
Table 3 - Overall Intensity Level Reduction
Operating pts Discharge Pipe Overall Intensity Level
Unit A (dBA) Unit B (dBA) Reduction (dBA)
8900 rpm 90.3 102 11.7
8250 rpm 86.2 95.1 8.9
Figure 4 - Off-Shore Platform at Sleipner

Figure 5 – Close-up of One of the Sleipner Platforms

543 km
12 km
18 km
Figure 6 - Sleipner Fields
12 km
8 km
650 km
240 km

Figure 4 - Photograph of the Compressor in Operation

Figure 5 - Multistage Compressor for Sleipner West Compression Project

Array
Location
Figure 6 - Multistage Compressor with Duct Resonator Arrays

Alternating Stress (ksi)
60
50
40
30
20
10
HCF
Yield
Normal Operation
Worst Case
Scenario
0
0 10 20 30 40 50 60 70 80 90 100
Mean Stress (ksi)
Figure 7 - Goodman Diagram for the Duct Resonator Array

V1
Z
Y
X
Suction pipe
Casing
Figure 8 - Definition of suction/discharge pipe and casing segments
Discharge pipe

[dB(A)/1.00p W/m²]
Calc. Intensity Spectrum(Mic B,Mic A) - Mark 1
Working : pipe_dis3 : Input : FFT Analyzer
100
90
80
70
60
50
40
BPF
2 nd
harmonic
0 1k 2k 3k 4k
[Hz]
5k 6k 7k 8k
Figure 9 - Compressor B (without array) Running at 8903 RPM; Discharge Pipe Overall Intensity Level =
102 dBA

[dB(A)/1.00p W/m²]
Calc. Intensity Spectrum(Mic B,Mic A) - Mark 1
Working : pipe_suc3 : Input : FFT Analyzer
100
90
80
70
60
50
40
BPF
2 nd
harmonic
0 1k 2k 3k 4k
[Hz]
5k 6k 7k 8k
Figure 10 - Compressor B (without array) Running at 8903 RPM; Suction Pipe Overall Intensity Level = 97.3
dBA

[dB(A)/1.000p W/m²]
Calc. Intensity Spectrum(Mic B,Mic A) - Mark 1
Working : pipe_dis21 : Input : FFT Analyzer
100
90
80
70
60
50
40
BPF
2 nd
harmonic
0 1k 2k 3k 4k
[Hz]
5k 6k 7k 8k
Figure 11 - Compressor B (without array) Running at 8250 RPM; Discharge Pipe Overall Intensity Level =
95.1 dBA

[dB(A)/1.00p W/m²]
Calc. Intensity Spectrum(Mic B,Mic A) - Mark 1
Working : case_suc22 : Input : FFT Analyzer
100
90
80
70
60
50
40
0 1k 2k 3k 4k
[Hz]
5k 6k 7k 8k
Figure 12 - Compressor B (without array) Running at 8250 RPM; Casing Overall Intensity Level = 78.5 dBA

[dB(A)/1.00p W/m²]
Calc. Intensity Spectrum(Mic B,Mic A) - Mark 1
Working : pipe_dis3 : Input : FFT Analyzer
100
90
80
70
60
50
40
BPF
2 nd
harmonic
0 1k 2k 3k 4k
[Hz]
5k 6k 7k 8k
Figure 13 - Compressor A (with arrays) Running at 8858 RPM, Discharge Pipe Overall Sound Intensity Level
= 90.3 dBA

[dB(A)/1.000p W/m²]
Calc. Intensity Spectrum(Mic B,Mic A) - Mark 1
Working : pipe_dis11 : Input : FFT Analyzer
100
90
80
70
60
50
40
BPF
2 nd
harmonic
0 1k 2k 3k 4k
[Hz]
5k 6k 7k 8k
Figure 14- Compressor A (with arrays) Running at 8200 RPM; Discharge Pipe Overall Intensity Level = 86.2
dBA

Sound Intensity Level [dB]
100
95
90
85
80
75
70
65
60
55
50
45
BPF
40
0 1000 2000 3000 4000 5000 6000 7000 8000
Frequency [Hz]
Compressor B
Compressor A (with Array)
Figure 15 - Discharge Pipe -Comparison at High Speed
2 nd
harmonic

Sound Intensity Level [dB]
100
95
90
85
80
75
70
65
60
55
50
45
BPF
Compressor B
Compressor A (with Array)
2 nd
harmonic
40
0 1000 2000 3000 4000
Frequency [Hz]
5000 6000 7000 8000
Figure 16 – Discharge Pipe Comparison at Design Speed