Sei sulla pagina 1di 13

HEAT RECOVERY STEAM GENERATOR NOISE ATTENUATION

John Briggs, Foster Wheeler Limited


Dr. Mal Sacks, Tacet Engineering Ltd.
Dr. Steven Kraemer

1.0) Introduction
A Heat Recovery Steam Generator (HRSG) performs a primary function as part of a
combined cycle power plant and a secondary function as an in-line silencer for
combustion turbine noise emissions. Typical combustion turbine noise emissions for
current large machines approach sound power levels of 150 dB(A). Overall plant noise
restrictions are often imposed by the community or the state, and include a limit for
sound pressure at the plant boundary for reasons of community comfort. An additional
noise limitation adjacent to the equipment is also typically necessary to comply with
Occupational Health and Safety regulations. An error in the prediction of noise
attenuation by the HRSG can result in a plant that does not meet environmental limits and
safety standards. In addition, unnecessary margin in the noise specification, particularly
in the lower octave bands, can result in significant costs for noise attenuation equipment.
The task of generating the specifications to meet project requirements falls on the
architect/engineer or project developer. The problem is divided between two major
systems: combustion turbine casing/air inlet noise paths, and the HRSG casing and stack
noise transmission paths. Over the past ten years Foster Wheeler Limited (FWL), with
the assistance of Tacet Engineering Ltd., has developed and refined methods of
predicting HRSG noise emissions. The foundation of this expertise in noise prediction
includes numerous field surveys of HRSG noise emissions and four major field research
projects investigating the noise attenuation characteristics of HRSGs. These studies
have established and verified both transmission loss characteristics for HRSG casings and
insertion loss for HRSG tube bundles, and have facilitated the development of a library of
far field noise measurements of HRSGs. Of particular note is a field investigation that
utilized a unique method of cross spectrum measurement/data analysis that allowed
accurate measurement of in-duct sound pressure for an operating 250 MW CT/HRSG
train illustrated in Figure 1.
This paper will provide a review of these research projects followed by a discussion of
current FWL methodology for the prediction of HRSG noise attenuation. Appendix A
provides suggested guidelines for noise specification development.

Figure 1: Elevation of HRSG for 250 MW CC Plant

2.0)

Field Investigations of HRSG Noise

Four major studies of field noise performance were completed over the past ten years.
The following section will review the first three investigations briefly and will discuss the
most recent research study in greater detail.

2.1) HRSG Noise Attenuation Model Development


The initial investigation (1991) consisted of extensive measurements of both sound
pressure (near field and far field) and casing vibration for a number of HRSGs, followed
by the development of a HRSG noise attenuation model. The investigation correlated
noise predicted by casing vibration levels with the noise measured as sound pressure in
the near and far field. The results were used to derive empirical formulas that permitted
the calculation of sound pressure levels at a given far field receiver location. Correction
factors were determined for combustion turbine load, wall construction, directivity, and
distance.

All calculations of noise attenuation were based on the assumption that the in-duct noise
spectrum reported by the combustion turbine manufacturer was correct. This assumption
could not be verified during this investigation. Overall internal tube bundle insertion loss
was determined based on stack outlet noise emissions and the aforementioned
assumption; casing transmission losses were based on sound pressure measurements and
the assumed internal noise spectrum.
This predictive technique was adequate in the early 1990s as environmental noise
requirements were somewhat less onerous than today. Throughout the ensuing decade
our research and development projects in the field of noise prediction and control
reflected the increased importance of this aspect of the power plant environment..

2.2)

Transmission Loss Versus Casing Construction

Foster Wheeler Limited initiated a research project in January, 1994 to investigate


transmission loss for a total of 18 panel constructions commonly used in steam generator
design. This project required fabricating sample panel sections for insertion in the test rig
noted in Figure 2.
4mm ACOUTHERM BARIUM
DAMPED VINYL SHEET

NOISE SOURCE

1/16 RUBBER SHEET


PERIMETER
FRAME
SAND
TEST PANEL

3- 2 lb/ft INSULATION

PIT
(7x7x7)

PANEL SUPPORT ISOLATORS


MICROPHONE

Figure 2: Transmission Loss Test Rig

The test rig consisted of a 7 ft deep by 7 ft square concrete pit with the test panels
supported on panel isolators at the perimeter of the pit to reduce noise transmission from
the concrete to the panel. The panel edges were buried using both sand and barium
damped vinyl mat to minimize the noise flanking path. A noise source was placed above
the panel and test microphones placed in the pit. Data was accumulated for each panel
section for each octave band. This work provided verification of transmission loss
coefficients versus octave band for different casing constructions.

2.3) Noise Attenuation Through Tube Bundles and Casing Walls


The third investigation undertaken was to accurately determine the sound attenuation
through the banks of tube bundles and the wall transmission loss for an HRSG prior to
commissioning. This was accomplished by placing a noise source inside the inlet duct of
a new HRSG. The inlet opening was then closed off with a heavy panel (the gas turbine
unit had not yet been installed). The sound source was a continuous, steady broad-band
noise with a sound power level of approximately 118 dB(A). Measurements of sound
pressure level were taken at a number of different locations along the inside and outside
of the HRSG. Vibration measurements along the exterior wall were also taken for
comparison.
Figure 3 shows the results of sound pressure level attenuation inside the HRSG between a
reference location 1.4m in front of the loudspeaker (at the inlet duct) and averaged results
at planes normal to the HRSG axis with increasing distance from the source throughout
the tube banks, from the inlet transition to the exhaust stack base. The results have been
corrected for operational temperature and flow conditions, assuming that sound
attenuation is wavelength-dependent. All of the measurements were well above the
ambient sound levels inside the HRSG when the source was turned off. The results
show a small attenuation at low frequencies through the tube bundles but much greater
attenuation at high frequencies. It should be noted that, in the absence of flow, there was
no re-generated flow induced noise. Flow induced noise would reduce the sound
attenuation at high frequencies under operational conditions.
Figure 4 shows the averaged results of sound attenuation through the HRSG walls.
Because of high external ambient sound levels, measurements were limited to the inlet
transition area. The vibration measurement results on the exterior wall were used to
predict the radiated sound for comparison. The results show very close correspondence
between the sound and vibration measurements at low frequencies, but they diverge at
high frequencies because of external sound leakage through the fabric inlet expansion
joint. The results based on the vibration measurements were therefore used to estimate
the sound attenuation of HRSG wall construction.

Sound Pressure Level (dB)

80
70
60
50
40
30
20
10
0
0

10

Measurement Position
dB(A)
1000

31
2000

63
4000

125
8000

250

500

Figure 3: Sound Pressure Vs Position in Tube Banks

Sound Pressure Level (dB)

80
70
60
50
40
30
20
10
0
31

63

125

250

500

1000

2000

4000

Octave Band Centre Frequency (Hz)


Series1

Series2

Figure 4: Comparison of Internal External SP (measured; series 1)


(calculated from wall acceleration; series 2)

8000

11

2.4) Investigation of In-Duct On-Line Noise


Foster Wheeler Limited and Tacet Engineering recently completed an extensive
investigation of an operating HRSG to obtain near field sound pressure levels, far field
sound pressure levels, wall vibration levels and in-duct sound power levels.
Measurement of in-duct sound pressure levels provides critical information needed to
correlate the gas turbine exhaust sound power levels (i.e. the sound source) with HRSG
acoustic performance and the resulting environmental sound pressure levels.
The ability to perform accurate measurements of in-duct sound pressure levels is
complicated by the high temperatures present inside the HRSG, the turbulent nature of
the flow at the outlet of the combustion turbine, and the fact that the gases inside the
HRSG are under pressure. The methods employed during this study are the result of a
long term R&D program conducted by Tacet Engineering Ltd., Higgott-Kane Industrial
Noise Controls Ltd. and Ryerson Polytechnic University as described in references [1],
[2] and [3]. This collaboration has resulted in the development of a lightweight, portable
microphone probe system that allows accurate measurement of in-duct sound pressure
levels [1] (Figure 5).
In performing these measurements, there are two exhaust sound power level spectra that
must be considered: one that is associated with sound pressure fluctuations and one that is
associated with turbulent pressure fluctuations. These two noise source spectra react
differently with the exhaust system and produce different sound pressure levels in the
environment.
LINING

DUCTWALL
MICROPHONE

COUPLING
REDUCER

ANECHOIC
TERMINATION

Figure 5: In-Duct Microphone Probe System

In order to delineate these spectra, the research team developed a new method of
analyzing the data using a cross spectrum technique to separate the sound and turbulence
fields. This method is discussed in detail from a theoretical perspective in reference [3]
and from an experimental perspective in reference [2]. The cross spectrum technique
requires simultaneous microphone measurements at two locations on a duct cross section.
The cross spectrum analysis attenuates the turbulence component of the signal and
extracts the embedded sound pressure level spectrum. The result is a spectrum that is the
average of the two underlying sound spectra.
In applying these techniques to the subject HRSG, in-duct measurements were obtained
at cross sections upstream and downstream of the tube bundles and near the top of the
exhaust stack. External sound pressure levels in the plane of the exhaust outlet were also
obtained. With this data, the evolution of the dynamic pressure fields from the HRSG
inlet to outlet could be observed.
Analysis of the data produced by these measurements led to the following observations:
1. The turbulence levels are strongly dominant upstream of the tube bundles and in the
outlet stack.
2. The sound levels are strongly dominant downstream of the tube bundles.
3. The embedded sound pressure level in the exhaust stack, as extracted by the cross
spectrum analysis, correlates well with the sound power radiated by the exhaust
outlet.
4. The in-duct auto spectrum in the exhaust stack, which is dominated by turbulence,
would predict sound power levels radiated by the exhaust stack that are too high by
20 dB. This demonstrates that the in-duct turbulent pressure fluctuations do not
propagate to the external environment as sound..
5. The turbulent and sound pressure signals undergo both attenuation and regeneration
along the path from HRSG inlet to outlet. The details of these processes are not fully
understood and are currently under investigation.
Based on current evidence, we expect the walls of the HRSG to react in an approximately
similar manner to both turbulence and sound pressure excitation; i.e. for similar sound or
turbulence levels to produce similar wall vibration levels. Consequently both spectra
would produce similar external sound levels. Therefore the in-duct auto spectrum (which
measures the total sound plus turbulence fields) would be used to estimate HRSG duct
breakout and flanking noise, subject to future refinement
Figure 6 shows the sound power level upstream of the tube bundles based on the
measured in-duct auto spectra and as supplied by the turbine manufacturer. The figure
also shows the radiated sound power level at the stack outlet. Although the overall sound
power level measured in the inlet transition was very close to that supplied by the turbine
manufacturer, in the 125 Hz band the measured level was almost 9 dB lower. This
difference would have a major cost impact on noise control design where stringent noise
specifications are required if based on the manufacturers data for this turbine.

Sound Power Levels


160
155
150
145
140
135

Sound Power Level

130
125
120
115
110
105
100
95
90
85
80
75
70
31.5

63

125

250

500

1000

2000

4000

8000

SumL

SumA

Octave Band Centre Frequency


Inlet

Outlet Plane

Gas Turbine Manuf. Data

Figure 6: Sound Power Levels at Various Locations


Figure 7 shows the correlation between near field sound pressure level (measured two
inches from the HRSG inlet transition duct wall) and the sound pressure level predicted
from the measured adjacent wall vibration.
Evaluation of this data indicates that:
1. Good estimates of sound pressure level can be made from measured wall vibration
level.
2. These estimates are not contaminated by extraneous sound sources; i.e. we obtain the
sound pressure level due to the HRSG walls only.
3. This data can be used in conjunction with the in-duct data to estimate HRSG duct
wall transmission loss as well as breakout and flanking noise radiation from an online HRSG.

Measured (VD) vs. Expected Sound Levels (based on vib. measurements), Location 4A
Expected Lp is corrected for resonant frequency at 1k, 2k, and 4kHz bands
105
100
95

Sound Pressure Level (dB)

90
85
80
75
70
65
60
55
50
45
31.5

63

125

250

500

1000

2000

4000

8000

SumL

Octave Band Centre Frequency (Hz)

Expected Lp (based on vibration measurements)

Measured Lp (VD)

Figure 7: Sound Pressure Vs Vibration Measurement Sound Pressure

3.0) Noise Prediction Model


There are a number of fundamental approaches to modeling noise which can be adopted
for HRSGs, including statistical, mathematical, physical, and empirical. The relative
merits of these approaches are as follows:

Statistical models use measurements from many installations and can predict results
for similar installations but not new designs.
Mathematical models use theoretical methods (e.g. finite element or boundary
element methods) and are potentially very accurate, but are time-consuming and
costly to set up.
Physical scale models, built from structurally-modeled materials and measured at
very high (scaled) frequencies require expensive, specialized facilities and a separate
model for each design. These models have the drawback of high cost, but could be
useful in generic research.

SumA

Empirical models use measured data and correlations in a simplified analytical model.
This method can be used to predict results from existing and new designs fairly
quickly and inexpensively.

For the purposes of predicting HRSG noise attenuation characteristics, an empirical


model was set up in the form of a spreadsheet. The HRSG is modeled as an open-ended
duct, divided into sections including the inlet transition, main HRSG casing, outlet
transition, stack walls, and finally the stack outlet.
Input data are the gas turbine exhaust octave band sound power levels. The sound
pressure levels inside the HRSG are calculated in octave bands at each section along the
open-ended duct, using the equation:
Lp (inside) = Lw 10log S [dB]
Where:
Lp(inside)
Lw
S

=
=
=

average sound pressure level inside the duct section


sound power level entering each section
cross-sectional area

Where appropriate, values for internal gas-path sound attenuation are introduced for each
section, which reduce the sound power level accordingly. The sound power levels
radiated externally from the HRSG casing sections are calculated from:
Lw(radiated) = Lp(inside) TL + 10logPl
Where:
Lw(radiated) =
TL
=
Pl
=

[dB]

average casing breakout sound power level from each section


Transmission Loss (HRSG wall attenuation)
perimeter X length of each section

The final step is to calculate near-field or far field sound pressure levels, in octave bands
and in dB(A), using the radiated sound power levels obtained for each section.
A number of parameter values and correlations obtained from experimental or theoretical
data are used in the empirical model:
Attenuation of internal HRSG elements (tube bundles, silencers)
Attenuation of inlet, main casing, outlet and stack walls
Flanking vibration of the main casing
Stack outlet directivity
This allows the model to be adjusted for different design configurations, particularly with
regard to noise control measures needed to meet specified or guaranteed sound levels.
Typical noise control measures adopted include heavier casing walls, stack silencers,
inlet duct silencers, vent and safety valve silencers, and acoustic shrouds for the HRSG
casing.

4.0) Summary
The prediction of HRSG noise attenuation characteristics is critical to the success of
many combined cycle power plant installations. The preceding paper has provided an
overview of the engineering expertise Foster Wheeler Limited and Tacet Engineering
have developed through ongoing research in this important area.
The cornerstone of measuring noise attenuation of HRSGs is the determination of the
noise emitted by the combustion turbine. The methods described herein have permitted
the accurate assessment of HRSG noise attenuation.
Further work is planned to validate the noise prediction model for HRSGs with added
noise control elements such as stack silencers. Opportunities will be provided as more
low-noise HRSGs are completed and brought into service. More site measurements are
required to improve the statistical basis of uncertainty determinations in the modeling
process and to determine HRSG attenuation parameters with greater accuracy.
5.0) References
1. M.P. Sacks et al In-duct measurement of gas turbine noise emissions using a cross
spectrum method, Paper No 2000-GT-349, ASME Turbo Expo 2000, Munich, May
2000.
2. M.P. Sacks et al Microphone probe for in-duct measurement of gas turbine exhaust
noise, Paper No 2000-GT-656, ASME Turbo Expo 2000, Munich, May 2000.
3. R. Behboudi, J.G. Kawall and M.P. Sacks A novel technique for measuring sound
within turbulent duct flows, Proceedings of CANCAM 99, McMaster University,
Hamilton, Ontario, Canada, May 30 June 3, 1999.
4. ANSI Standard B133.8-1989 Gas Turbine Installation Sound Emissions.
5. ISO Standard 6190:1988 Acoustics Measurement of sound pressure levels of gas
turbine installations for evaluating environmental noise Survey method.
6. ISO Standard 10494:1993 Gas turbines and gas turbine sets Measurement of
emitted airborne noise Engineering/survey method.
7. Robert S. Johnson Recommended Octave Band Insertion Losses of Heat Recovery
Steam Generators used on Gas Turbine Exhaust Systems, Noise-Con 94, Florida,
May 1994.

Appendix A : HRSG Sound Level Guarantees


And Specification Development
In most countries today, new power plants must meet municipal, state or federal noise
regulations, which specify noise limits in dB(A) or octave band sound pressure levels.
The plant owner in turn, usually through the main contractor, engineer or consultant, will
prepare noise specifications based upon a model that divides the plant into a number of
discrete sources, including HRSGs. HRSG vendors are then requested to guarantee noise
emissions in accordance with the above specification.
There are several problems faced by HRSG manufacturers under these circumstances.
Firstly, the sound levels radiated from the HRSG are primarily related to the sound power
level of the exhaust of the gas turbine unit, which is outside the HRSG manufacturers
scope of supply. Noise specifications for HRSGs are often poorly prepared and have
unnecessarily low limits, especially in the low frequency octave bands.
Noise verification methods are not always clearly defined, which make contractual
guarantees difficult.
Table 1 below presents a suggested HRSG noise specification that can be used by project
developers to avoid misunderstandings and potential disputes caused by the above
problems. It can also be used as the basis for noise guarantees.
The prediction of noise from a future HRSG installation cannot be exact. A factor of
safety of 3 to 6 dB(A) should be included because of the various uncertainties in the noise
modeling and measurement procedures. Ideally, the safety factors inherent in the
specifications of gas turbine exhaust, HRSG and overall plant sound levels should be
combined in a statistically valid way, but that is rarely the case. Because of the need for
guarantees at each contractual level (e.g.: developer, architect/engineer, combustion
turbine vendor, HRSG vendor), the simple addition of all of these safety factors can lead
to a situation where excessive amounts of noise control are required, with a
correspondingly significant cost increase. This is especially the case when tight noise
specifications are imposed. There is no easy solution to this problem, unless all parties
concerned are prepared to disclose the safety factors inherent in their noise guarantees
and the overall plant noise uncertainty is computed by prior agreement.
1. Environmental Noise Guarantee
Sound emission from the heat recovery steam generator(s) will not exceed an average Aweighted sound pressure level of .. dB(A) when measured as specified below at a
distance of 122m (400) from the perimeter of the installation and 1.5m (5) above grade.
2. Occupational Noise Guarantee

Sound emission from the heat recovery steam generator(s) will not exceed an average Aweighted sound pressure level of . dB(A) when measured as specified below at a
distance of 1m (3) from the surface of the installation and 1.5m (5) above grade (or
personnel platform).
3. Measurement Specification
Guarantee sound pressure levels are to be measured according to ANSI Standard B133.81989 Gas Turbine Installation Sound Emissions.
4. Gas Turbine Sound Power Levels
The above sound pressure levels are only guaranteed when the gas turbine octave band
sound power levels (dB re 1 pW) are also guaranteed not to exceed the following values
at the turbine diffuser exhaust :
Band
Lw

31.5

63

125

250

500

1000

2000

4000

8000

Potrebbero piacerti anche