EPA-600/2-76-115
May 1976
Environmental Protection Technology Series
A PASSIVE FLOW MEASUREMENT SYSTEM FOR
STORM AND COMBINED SEWERS
Municipal Environmental Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
-------
EPA-600/2-76-115
May 1976
A PASSIVE FLOW MEASUREMENT SYSTEM
FOR STORM AND COMBINED SEWERS
by
K. M. Foreman
Research Department
Grumman Aerospace Corporation
Bethpage, New York 11714
Contract No. 68-03-2121
Project Officer
Hugh Masters
Storm and Combined Sewer Section
Wastewater Research Division
Municipal Environmental Research Laboratory (Cincinnati)
Edison, New Jersey 08817
U.S. ENVIRONMENTAL PROTECTION AGENCY
OFFICE OF RESEARCH AND DEVELOPMENT
MUNICIPAL ENVIRONMENTAL RESEARCH LABORATORY
CINCINNATI, OHIO 45268
-------
DISCLAIMER
This report has been reviewed by the Municipal Environmental Research
Laboratory, U.S. Environmental Protection Agency, and approved for publica-
tion. Approval does not signify that the contents necessarily reflect the
views and policies of the U. S. Environmental Protection Agency, nor does
mention of trade names or commercial products constitute endorsement or
recommendation for use.
11
-------
FOREWORD
The Environmental Protection Agency was created because of increasing public
and government concern about the dangers of pollution to the health and
welfare of the American people. Noxious air, foul water, and spoiled land
are tragic testimony to the deterioration of our natural environment. The
complexity of the environment and the interplay between its components require
a concentrated and integrated attack on the problem.
Research and development is that necessary first step in problem solution
and it involves defining the problem, measuring its impact, and searching for
solutions. The Municipal Environmental Research Laboratory develops new
and improved technology and systems for the prevention, treatment, and
management of wastewater and solid and hazardous waste pollutant discharges
from municipal and community sources, for the preservation and treatment
of public drinking water supplies, and to minimize the adverse economic,
social, health, and aesthetic effects of pollution. This publication is one
of the products of that research; a most vital communications link between
the researcher and the user community.
The need exists to develop innovative, passive, nonintrusive, and low cost
solutions to the problems of continuous measurement and recording of flows
in storm and combined sewers. This experimental investigation is of one
such technique that monitors the pseudosound produced by flow past a channel
discontinuity. The results of laboratory and field tests demonstrate the
feasibility of this method of flow measurement using an accelerometer trans-
ducer attached to the outside surface of a flow channel.
Francis T. Mayo, Director
Municipal Environmental Research
Laboratory
111
-------
ABSTRACT
This investigation concerns a new, nonintrusive, low cost,
passive flow measurement method to meet the urgent needs for
good management of storm and combined sewer systems. Opera-
tion of the system is based on sensing the near field sound
emitted by the disturbed flow at a channel discontinuity.
These local pressure pulses are called pseudosound and radi-
ate as dipole sound sources orthogonal to the flow direction.
The output signal of passive transducers, such as accelerome-
ters, attached to the outer wall of the channel indicates
flow rate after processing by a Fourier Analyzer. Feasibility
has been demonstrated by laboratory tests using full scale
sewer pipe elements, and by a brief series of field tests mea-
suring sanitary sewage flow. Recommendations are made for
further field site testing using an instrumented sewer line.
This report was submitted in fulfillment of EPA Contract
68-03-2121 by Grumman Aerospace Corporation under the sponsor-
ship of the Environmental Protection Agency. Work was com-
pleted as of August 22, 1975.
IV
-------
CONTENTS
Page
Foreword iii
Abstract iv
Figures vi
Acknowledgments xiii
1 Conclusions 1
II Recommendations 5
III Technical Study 7
IV References 119
V Glossary
120
-------
FIGURES
No. Page
1 Ringing Frequency, fr, for Various Diameter
Pipes and Pipe Materials „.„....„ ................ 14
2 Limiting Transmission Loss of Sound Across
Cylindrical Steel Pipe Walls at Less than
the Ringing Frequency , f r . „ ..................... 15
3 Theoretical Sound Transmission Loss, R, for
Mass Controlled Panels .......................... 17
4 Grumman Research Water Supply Facility .......... 19
5 240 Gallon FaciH ty Water Storage Tank .......... 20
6 Air Pressurization and Water Supply Pipes
to Storage Tank ................................. 20
7 20 Foot Length of 3-in. Diameter Pipe of
Facility Water Supply System .................... 21
8 View of Facility Test Section Showing Diversion
Trough, Water Collection Drums, and Weighing
Scale. 8 -in. Diameter Aluminum Pipe on Test
in Concentric Configuration ..................... 21
9 View of Water Flowing from 8-in. Diameter
Test Pipe into Diversion Trough ........... „ ..... 23
10 Overall Side View of Facility Test Section
Showing Diversion Trough, Water Collection
Drums, and Weighing Scale. 4-in. Diameter
Transite Pipe on Test „ „ „ „ „ . . . . ................... 23
11 Sensor Locations and Channel Discontinuities
Tested in Research Laboratory ................... 24
12 Schematic of Testing Instrumentation ......... „.. 27
13 Side View of Box for Containing Sand Used to
Simulate Buried Pipe Installation ............... 29
vi
-------
No. page
14 Top View of Box Used to Simulate Buried Pipes.
6-in. Diameter Transite Pipe is Shown on Test
with Sand Removed. Flow Direction is from
Left to Right 29
15 End View of 8-in. Diameter Pipe Showing Initial
Condition of Heavy (1-in.) Bed Load of Stone
Chips 30
16 Heavy Bed Load After Being Disturbed by Water
Flow at Rates up to 3 Pounds Per Second 30
17 Equipment Used to Feed Sand Grit into Upstream
Water Supply 31
18 Closeup of Speed Controlled Motor and Supply
Hopper Used to Feed Sand Grit into Water Supply .. 31
19 Screen Tray Added to Water Collection System to
Separate Grit from Collected Water. 6-in. Di-
ameter Transite Pipe is Shown on Test. Tray is
Just Below Pipe Exit 34
20 Exterior View of Valve Vault (Foreground at
Grumman Waste Treatment Plant "A". This is
First Site of Field Test Preview Experiments 34
21 Interior View of Valve Vault Showing 8-in.
Plug Valve and 8 x 10-in. Reducer Section with
Accelerometer Attached to Outer Wall for
Acoustic Emission Monitoring of Flow 35
22 Exterior View of Aeration Tank at Grumman Waste
Treatment Plant "A". 10-in. Diameter Pipe
Emerges from Underground Installation at
Left Center of Photo. This is Site of Second
Series of Field Test Preview Experiments 35
23 Closeup of 10-in. Diameter Pipe Where Second
Field Test Preview Series Was Conducted.
Sensor is Shown Attached to Top Center of
Horizontal Pipe. Tape Recorder and Amplifier
Used is Shown in Foreground 37
Vll
-------
No._
24 Data Processing Equipment Used to Obtain Spectral
Distribution from Flow Acoustic Emission Re-
cordings. HP-5465A Fourier Analyzer Computer
is in Center of Picture 37
25 Schematic of Acoustic Emission Data Processing
System 38
26 Closeup of Data Processing Equipment Showing
(L to R) HP-5465A Fourier Analyzer Computer,
Penco Continuous Loop Recorder and Playback,
and Nagra Tape Recorder Used for Laboratory
and Field 38
27 Typical Fourier Processed Spectral Distributions
of Laboratory Sound Recordings (10 Hz Bandwidth) . 40
28 Acoustic Signal Variation with Flow Rate in a
3-in. (7.62 cm) Diameter Steel Pipe 46
29 Typical Spectral Distributions of Laboratory
Sound Recordings. Distributions Obtained by
Fourier Analyzer (10 Hz Bandwidth) 47
30 Hydraulic Radius of Test Channel For Different
Water Level Heights 49
31 Calibration of Research Facility Water Supply
(A = Flow Area, Q = Volumetric Flow Rate) 51
32 Variation of Normalized Acoustic Signal with
Flow (Cylindrical Pipe Exiting to a Trapezoidal
Open Channel) 53
33 Acoustic Signal Variation with Flow Rate
(Cylindrical Pipe Discharging to Trapezoidal
Open Channel) 54
34 Variation of Normalized Acoustic Signal with
Flow Rate - Different Sensor Locations (Cylin-
drical to Trapezoidal Channal Transition) 55
Vlll
-------
No.
35 Variation of Normalized Acoustic Signal with
Flow Rate - Effect of Sensor Location
(Eccentric Cylindrical Pipe Elements and
Extended Inlet) ................................. 57
36 Effect of Sensor on Variation of Normalized
Acoustic Signal with Flow Rate (Eccentric
(Cylindrical Pipe Elements and Extended Inlet) .. 58
37 Effect of Extended Inlet on Acoustic Signal
Variation with Flow Rate (Eccentric Cylin-
drical Pipe Elements) ........................... 60
38 Effect of Sensor Position and Sound Frequency
on Acoustic Emission of Water Flow (Eccentric
Cylindrical Pipe Elements and Flush Inlet) ...... 61
39 Effect of Exit Weir Height on Normalized
Acoustic Signal of Water Flow (Eccentric
Cylindrical Pipe Elements and Flush Inlet) ...... 62
40 Effect of Exit Weir Height on Normalized
Acoustic Signal of Water Flow Eccentric
Cylindrical Pipe Elements and Flush Inlet) ...... 63
41 Variation of Signal with Flow at Different
Sensor Locations; (Eccentric Cylindrical
Pipe Elements and Flush Inlet) (with 4.24-in.
Weir at Exit Unless Noted Otherwise) ............ 65
42 Variation of Normalized Acoustic Signal with
Flow Rate for Different Frequencies and Test
Conditions (Eccentric Cylindrical Pipe
Elements and Flush Inlet) ....................... 67
43 Variation of Normalized Acoustic Signal with
Flow Rate for Different Test Conditions (Flush
Inlet, Eccentric Cylindrical Pipe Step) ......... 68
44 Variation of Acous tic Signal with Flow for 3
Sensor Locations and Various Test Conditions .... 70
45 Closeup of Concentric Discontinuity Plane
Assembly of 3-in. Steel Water Supply Pipe
to 8-in. Diameter Aluminum Test Pipe ............ 70
x
-------
No.
46 Internal Relation of 3-in. Supply Pipe to
8-in. Test Pipe with Heavy Bed Load of Stone
Chips„ View Taken Before Flow Established 72
47 View of Upset Bed Load After Water Flow Rate
of 5 Pounds Per Second 72
48 Variation of Normalized Acoustic Signal wtih
Flow Rate at Various Sensor Locations and
Characteristic Frequency (Concentric Cylin-
drical Pipe Array with Extended Inlet) 74
49 Variation of Normalized Acoustic Signal with
Flow Rate for Different Frequencies (Concentric
Cylindrical Pipe Array with Extended Inlet) 75
50 Effect of Sensor Circumferential and Axial
Location on Useful Signal (Concentric Cylin-
drical Pipe Array with Flush Inlet) 77
51 Varation of Acoustic Signal with Flow at Two
Sensor Locations for Different Test Conditions
(Flush Inlet, Concentric Cylindrical Pipes) 78
52 Variation of Acoustic Signal with Flow at Three
Sensor Locations and for Two Frequencies - with
4.24" Weir at Pipe Exit Plane - (Concentric
Cylindrical Pipe Elements) 79
53 Variation of Acoustic Signal with Flow at Three
Sensor Locations and Three Frequency Bands - with
4.24-in. Weir at Pipe Exit Plane - (Concentric
Cylindrical Pipe Elements)„ See Fig, 52 for
Sensor Position Location on Pipe 80
54 Variation of Acoustic Signal with Flow at Two
Sensor Locations and Two Frequency Bands (Con-
centric Cylindrical Pipe Elements) - with 4024"
Weir at Pipe Exit and 8" Deep Sand Overburden,
Except Where Noted 82
55 Varation of Acoustic Signal with Flow at Five
Frequency Bands and One Sensor Location (Con-
centric Cylindrical Pipe Elements) - with 4024"
Weir at Pipe Exit and 8" Deep Sand Overburden,
Except Where Noted 83
x
-------
No.
56 Effect of Heavy Bed Load of Stones (1M Deep)
in 8" Diameter Pipe on Acoustic Signal Varia-
tion with Flow - Sand Covered Array 84
57 Effect of Two Bed Loads of Sternes in 8-in.
Diameter Pipe on Acoustic Signal Variation with
Flow - Sand Covered Array with 4.24-in. Weir 85
58 Sketches of Heavy Bed Load Movement at Different
Stages of Water Flow Test Conditions 86
59 Variation of Acoustic Signal with Flow at Four
Frequencies and Two Sensor Positions (8" Sand
Cover and Monolayer of Stone Chips Within 8"
Diameter Pipe. 4.24" Weir at Exit of Pipe) 88
60 Closeup of Eccentric Connection of 3-in. Water
Supply Pipe to 4-in. Diameter Transite Test Pipe . 91
61 Closeup of Special Assembly Collar Used for
Eccentric Connection of 3-in. Pipe to 6-in.
Diameter Transite Test Pipe. Flow is from Right
to Left. Two Sensor Mounting Studs are Shown
Cemented at Top Centerline Surface of Transite
Pipe 91
62 Variation of Acoustic Signal with Flow (Eccentric
Cylindrical Pipe Array) - Transite and Steel
Pipe Elements without Sand Overburden Except
Where Noted 93
63 Variation of Acoustic Signal with Flow (Eccentric
Cylindrical Pipe Array) - Transite and Steel
Pipe Elements Buried Under 8-in. (20.3 cm) of
Sand, Except Where Noted 95
64 Variation of Acoustic Signal with Flow (Eccentric
Cylindrical Pipe Array) - Transite and Steel
Pipe Elements Without Sand Overburden, Except
Where Noted 95
65 Effect of Grit Addition to Water Flow on the
Acoustic Signal Variation with Flow - Transite
and Steel Pipe Elements, Without Sand Over-
burden, in Eccentric Assembly 96
XI
-------
No.
66 Variation of Acoustic Signal with Flow at
Grumman Waste Treatment Plant "A" Valve Vault
for Different Sensor Positions - First Test
Series 98
67 Variation of Signal with Flow at Grumman Waste
Treatment Plant "A" Aeroation Pond for Different
Frequencies and Sensor Positions - Second Test
Series i00
68 Typical Variation of Acoustic Signal with Flow
for Two Conduit Configurations (Cylindrical to
Trapezoidal Cross-Sectional Discontinuity) l03
69 Effect of Extended Inlet on Acoustic Signal
Variation with Flow Rate (Concentric Cylin-
drical Pipe Elements) 106
70 Effect of Pipe Configuration on Normalized
Acoustic Signal Variation with Flow Rate
(Cylindrical Pipe Elements) with 4.24-in.
Weir, 8-in. Sand Overburden, and Flush Inlet 11U
71 Effect of Pipe Material on Normalized Acoustic
Signal Variation with Flow Rate (Eccentric
Cylindrical Pipe Elements and Flush Inlet) -
8-in. Sand Overburden 113
Xll
-------
ACKNOWLEDGMENTS
This work was performed for Grumman Ecosystems Corporation,
Bethpage, New York, who are the prime contractors to the U.S.
Environmental Protection Agency (EPA) under Contract 68-03-
2121.
We acknowledge the assistance of Mr. James Rogers of Grumman
Aerospace Corporation (GAC) for programming and otherwise
assisting in the computer processing of our recorded sound
data. Thanks are also directed to Mr. Chris Clamser and
Mr. Norman Peele of GAC for making possible the many facility
and testing changes, on a timely basis, during the course of
this program.
Finally we are indebted to Mr. Hugh Masters and Mr. Richard
Field of EPA who provided helpful guidance during the pro-
gram, and enlightenment regarding simulation of sewage solids.
Xlll
-------
SECTION I
CONCLUSIONS
The results of this program demonstrate the feasibility of
*
Grumman s acoustic emission flowmeter concept for channels
where a discontinuity exists because of a conduit cross-
sectional increase. Since these geometric cross section
changes are common in sewer systems, this passive, nonintru-
sive, flow measurement technique is directly applicable to
storm and combined sewers.
On a scale that is without precedent regarding physical size
and variety of laboratory test conditions, the acoustic emis-
sion monitored by an accelerometer has been found to produce
unambiguous signals that increase in magnitude as the fluid
flow rate increases. The source of the acoustic emission is
the pseudosound created by the interaction of fluid with a
solid surface in the near field of a conduit discontinuity.
Pseudosound does not propagate with the flow, but radiates as
a dipole, orthogonal to the flow direction. Data has been
obtained for open channel, full flow, and pressurized flow
conditions involving full scale metallic and Transite pipe
sections used to conduct fluid waste.
During these laboratory tests, steady, unsteady, uniform, and
nonuniform flows were established. In addition to tests of
freely suspended pipes flowing clean water, corroborative
data has been obtained from tests with simulated buried in-
stallations of the pipes and with water containing suspended
and settleable solids characteristic of sewage. Tests with
Patent pending.
-------
cylindrical as well as trapezoidal conduits cross sections
have confirmed general similarity of results for these geome-
tries.
A field test preview, in which the acoustic emission flowmeter
technique was used to monitor sanitary sewage from large indus-
trial buildings, demonstrated good correlations with test trends
obtained in the laboratory, and with theoretical predictions.
Data of value to future engineering design of prototype sys-
tems have been obtained. This experience indicates that low
cost accelerometers having a wide range of voltage sensitivi-
ties, between 10 and 100 mv/g, are well suited to the
acoustic monitoring function. These sensors are easily em-
placed or removed by screwing on to inexpensive, dedicated
mounting plates which may be cemented or magnetically at-
tached to the outer surface of the flow channel. Epoxy ce-
ment with high metallic content has proven very satisfactory
for this purpose. For discontinuities involving sudden en-
largements of conduit cross section, the sensor location
yielding the strongest signal for the greatest range of flows
is at the bottom-most circumferential position and two to
three (upstream) pipe diameters downstream of the discontinu-
ity plane. Transmission of flow acoustic emission through
most sewer conduit walls is of a sufficiently high percentage
so that adequate signal quality is available to the sensors.
Sound transmission circumferentially in solid conduit walls
is of such high magnitude that useful sensor signals can be
obtained at nearly any circumferential position, at a given
axial position. These results hold for metallic as well as
inhomogeneous material (e.g., cement and Transite) used in.
pipe constructions and for conventional wall thickness to
pipe diameter ratios of sewers.
-------
In the early phases of this project, no special measures were
taken to simulate real sewer installation features in our
laboratory experiments. This initial oversimplification of
the test setups was intended to expedite acquisition of test
data and to avoid special equipment costs if unneeded. As a
result, however, some of the acoustic emission data appeared
to conflict with the expected trends of pseudosound produc-
tion. To overcome this equipment inadequacy, we developed in-
expensive but effective techniques to simulate properly sewer
installations. The acoustic emission signals obtained under
these changed test conditions were completely unambiguous, and
in general agreement with dipole theory predictions.
The most reliable unambiguous signal for flow rate measure-
ment is the acoustic emission component at a characteristic
frequency, f , which is inversely proportional to the maxi-
mum geometric step at the discontinuity, 6, and directly
proportional to the bulk sound speed, c, of the fluid, or
f ^ c/5. Because of secondary flow interaction processes,
minor transient changes in c, instrumentation sensitivity,
and acoustic energy conversion process, the useful signal
actually exists in a frequency band of up to 200 Hz width,
centered about f . For most practical sewer applications the
characteristic frequency will be in the midaudible range of
the sound spectrum. Random background noise is of the order
Q /
of 10 to 10 (i.e., -30 to -40 dB) the intensity
of flow acoustic emission. Response of sewer pipes to this
noise follows a general mass distribution law where amplitude
varies inversely with frequency. In contrast, the spectra of
flow acoustic emission have high amplitude content at fre-
quency bands associated with flow-related sources„ In this
regard, a bed load that does not adhere to the conduit wall
-------
in the near field of the discontinuity will not alter the
basic character of the flow acoustic emission. However, some
conversion of the acoustic emission energy to sound propa-
gating downstream with the flow can take place if the bed load
consists of sharp edged material that induces much turbulence.
In the laboratory, at low flows, the acoustic signal at char-
acteristic frequency increases with flow at a rapid rate, ap-
proximating the ideal (~ w ) dipole theoretical prediction
of 12 dB per doubling of flow rate. However, while this
trend is moderated at higher flow rates, it increases without
. T
ambiguity and assumes a direct proportionality to flow (~ w)
at the highest rates tested in the laboratory. In the limited
field test preview, the trend of high flow rate data taken at
an underground pipe installation follows dipole theoretical
predictions reasonably well.
In addition to its flow measurement capability, acoustic emis-
sion monitoring can be an operational tool for determining
maintenance needs of the sewer. To the experienced human
monitor at a central location, qualitative evaluation of total
sound content from various and diverse remote sites can pro-
vide a mental image of local sewer conditions that almost
rivals in situ observations by a mobile crew, but with
greater convenience and lower cost0
The contractual limitations of this project did not permit an
exhaustive investigation to reveal the full potential of Grum-
man *s passive measurement technique for all pipe geometries.
However, there are sufficient reasons, by virtue of the data
obtained, to justify extreme optimism on the future prospects
for this new, low cost approach to flow measurement.
w = rate of change of weight flow = Aw/At
-------
SECTION II
RECOMMENDATIONS
Feasibility of the passive acoustic emission flowmeter has
been demonstrated in the laboratory with a breadboard, re-
search type, instrumentation system. The features of this
technique should be appealing to management of municipal and
regional sewer districts. Therefore, to expedite the imple-
mentation of this method we recommend the following action:
1. Extend field testing conditions to a well in-
strumented sewer site, using the laboratory
breadboard system. Testing should be conducted
for a long enough duration (i.e., months) to
experience at least a realistic variation of
runoff and flushing conditions.
2. Based on the laboratory and extended field test-
ing results, design and build a prototype flow
measurement system, and test it at the well in-
strumented sewer site previously used. Labora-
tory calibration of the system should precede
the field testing. After initial shakedown and
possible system design modifications at the
field site, continue evaluation over a twelve
to twenty four month period for system relia-
bility, maintainability, and operational per-
formance data. These data should be compared
to ultimate system objectives defined by EPA.
-------
3. Depending on the experience of step 2 above, either
design and construct an advanced prototype or build
several initial production systems for more wide-
spread field tests throughout the nation.
Although it may be premature to be unduly concerned about
operational aspects of this passive measurement technique, it
is clear that its utility would be enhanced if field site cal-
ibration could be eliminated. One expeditious and economical
way of doing this is to calibrate the transducer in a flow
facility where small scale models simulate the relevant local
features of the actual sewer installatio n<, To use these cali-
brations with confidence requires that scaling laws be devel-
oped so that model data can be correctly applied directly to
the field site geometry. Accordingly, we recommend that theo-
retical studies be conducted to determine appropriate scaling
laws for the acoustic emission process. These similarity laws
then should be verified using data from field tests (such as
recommended above) and laboratory work reported here as well as
from future supplementary laboratory tests.
In order to explore more fully the growth potential of this mea-
surement technique, we recommend further laboratory research
with several pipe discontinuity configurations other than sud-
den enlargements. Among these are contractions, turns, tees,
Y-sections, valve bodies, and sumps. Investigations should in-
clude independent internal flow diagnostic instrumentation as
well as passive acoustic sensing. At least one study should
concern identifying and quantifying flow infiltration into
sewer lines; simulation methods should be established for the
infiltration. Other studies should investigate reverse flow
identification and correlations of field measurement perform-
ance with laboratory calibrations under simulated field condi-
tions .
-------
SECTION III
TECHNICAL STUDY
A. INTRODUCTION
The project investigations reported here are the first phase
of a program to develop a passive flow measurement system for
storm and combined sewers. It was performed by Grumman Aero-
space Corporation under subcontract to Grumman Ecosystems
Corporation in conformance with U.S. Environmental Protection
Agency (EPA) Contract 68-03-2121, dated June 28, 1974.
The purpose of this project is to determine, by experiment
and analysis, the feasibility of Grumman1s innovative acous-
*
tic emission flowmeter concept. Tests have been conducted
in the controlled environment of Grumman Research's water
flow laboratory under a variety of flow and simulated sewer
pipe installations. A brief field test also was conducted
to preview the type of data that-could be expected from a
full field test program. Secondary objectives are to obtain
engineering data to guide future flowmeter designs and in-
stallations, and to explore the growth potential of the con-
cept to meet ultimate design goals for a sewer flowmeter.
The acoustic emission flowmeter concept uses passive sensors
to monitor the sound emitted by fluid flow, and in this way
determines flow rate. This sound signal results from the
pressure disturbances when flow passes a channel discontinu-
ity. The discontinuity can be any change in conduit cross
section or shape, or where flow direction is significantly
altered. In the vicinity of the discontinuity, the pressure
_
Patent pending.
-------
disturbances depend on the mean velocity of the flow and can
be monitored by appropriate sensors at the boundary of the
flow. However, because this near field sound does not pro-
pagate throughout the flow with the sound speed of the fluid
•I r\
it is termed pseudosound. ' This differentiates it from the
more conventional acoustic processes that become coupled to
the fluid's far field radiation. Pseudosound has radiation
1 3
characteristics like a dipole source, ' so it emits acoustic
energy orthogonal to the fluid flow direction. For usual
sewer pipe construction, a passive sensor, such as an accel-
erometer, mounted to the outer surface of the sewer wall can
listen to pseudosound. The sensor's output signal can be
transmitted to nearby or remote devices that use the rela-
tionship of pseudosound to local mean velocity to determine
flow rate. The features of this simple, passive, nonintru-
sive flow measurement system are ideally suited to the oper-
ational conditions and environment found in storm and com-
bined sewers.
Background
The EPA need to characterize and assess storm-generated water
pollution, and to develop, demonstrate, and evaluate water
treatment and control systems, establishes a requirement for
real-time data from fast response and remote sensors. Severe
flow conditions in combined sewers obviate use of convention-
al flowmeters. Problems of cost, accuracy, reliability, op-
erational simplicity, and maintenance have limited the appli-
cation of other advanced active measurement techniques de-
veloped in recent years.
-------
A reliable flowmeter that is nonintrusive and passive has at-
tractive prospects for sewer applications because it can be
used on new as well as old sewer lines, does not have to be
physically subjected to the variable quality of sewer flow,
and can have low power requirements.
Pursuing its own independent studies for advanced flow mea-
surement techniques, Grumman Research invented a way by which
the relatively little known theory of pseudosound could be
used to determine fluid flow rate. This method is inherently
passive because it is activated by the sound energy emitted
by the flow.
About two years prior to the inception of this contract,
Grumman had demonstrated that the variation of pseudosound
intensity with flow rate could be measured and that it fol-
lowed theoretical predictions. The test configuration in-
volved a 0.875-in. (2.22 cm) diameter aluminum pipe dis-
charging to an open rectangular aluminum channel 1.0-in.
(2.54 cm) wide by 1.5-in. (3.81 cm) high. Measurements
were made at water flow rates to 0.66 Ib/sec (0.3 Kg/sec)
and used a single transducer type (B&K model 4332) attached
to the underside of the open channel by double adhesive faced
tape.
However, because of the very limited conduit size and con-
figuration, water flow rate range, and transducer type and
installation in these earlier demonstration tests, reserva-
tions remained about the applicability of the concept for
more representative conditions of storm and combined sewers.
To demonstrate feasibility, tests were necessary for greatly
extended conditions of:
-------
a) Flow rate
1) partial flow
2) full flow
3) pressurized flow
b) Conduit configurations
1) size
2) construction materials
3) geometry
c) Transducer
1) type
2) sensitivity
3) method of attachment
4) location relative to discontinuity plane
In addition, examination of such operational factors as
buried pipe installations and the addition of solids to
water flow were required.
Te s_it_Pha s ing
The test program discussed in this report was designed to
examine these questions in a phased sequence, giving system-
atically increased knowledge and preparation for progres-
sively more complex situations. First, we scaled up the sup-
ply conduit flow area by a factor of 10, and roughly dupli-
cated the circular to open rectangular conduit transition
cross section geometry used in the original demonstration.
Next, we examined the characteristics of various candidate
transducers and selected the best for our purposes. Methods
of sensor attachment were considered and tested for acoustic
coupling, ease of installation, and operational reliability.
10
-------
Supply flow rates were established in the water flow research
laboratory that were ten times the maximum previously used in
concept demonstration; later, flow rates twenty times greater
than initial demonstrations were used.
Following demonstration of reproducibility of results of the
prior demonstration tests, with cylindrical-to-rectangular
elements, we proceeded to test conduit arrays embodying only
cylindrical metal elements. After validating the acoustic
emission concept in completely cylindrical conduit geometry,
we returned to more detailed studies of the original conduit
test sections. The next test series was a detailed examina-
tion of metallic cylindrical pipe configurations, and in-
cluded extending the flow rates to the maximum capable of
being accurately measured. We then proceeded to test non-
metallic (Transite) cylindrical pipe elements. After con-
firming the similarity of flow acoustic emission measurements
for Transite and metallic pipes, we examined simulated buried
installations of cylindrical pipe discontinuities. The simu-
lated overburden was found to be helpful in reducing airborne
noise and did not adversely affect the flow acoustic emission
signal. Several succeeding tests with simulated buried con-
duits involved metallic and Transite cylindrical pipe ele-
ments. These data revealed detailed flow characteristics for
different sensors and sensor locations relative to the pipe
discontinuity. Toward the end of this series, exploratory
tests were made of the effect of solids in the water. Screened
grit was added to the upstream water supply in one set of
runs, and coarse stone chips were pre-placed as a pipe bed
in other tests. Finally a preview of field test acoustic
11
-------
emission measurements was made at various locations of the
sanitary sewer line to Grumman's main sewage treatment plant.
The test apparatus used in this program is described in Sec-
tion IIIB. The results of tests for each phase of this pro-
ject are described in Section IIIC, and discussed in Sec-
tion HID.
Theoretical Approach
Fluid flow in pipes or conduits produces sound that may be
caused by over-all or locally induced unsteadiness of em-
bedded turbulence. The local pressure disturbances induced
by channel discontinuities and not propagated downstream is
pseudosound. The sound pressure of pseudosound, p , is of
2 P
the order of pu where p is the average fluid density
and u is the fluid velocity. The radiation field of pseu-
dosound is like a dipole source so the total sound power,
P , in a free field is
2
2
where
TJu
pc
p = pressure disturbance
I = characteristic dimension of the local flow
conduit
pc = characteristic acoustic impedance of fluid
fluid ;
water)
c = fluid sonic speed (~15 x 10 cm/sec for
2
However, since p ^ pu , the theoretical sound power is
2 4
- Pi
T c
12
-------
This equation is the theoretical basis for the fourth power
relation of flow rate and sound emission. Note that the
•
volumetric flow rate, Q, is proportional to velocity, U,
in a given conduit geometry, so P ^ Q .
The characteristic frequency, f , of the dipole sound pro-
t_^
duced at the conduit discontinuity is inversely proportional
to the characterizing dimension of the discontinuity, 6.
From the fundamental relation f= c/A, where A is sound
wavelength in the fluid, f ^ c/6 or 6 ^ A. (Empirically,
v_
we have determined that the proportionality factor for A/8
is 4, or 46 = A.)
When the sound is "shadowed" by the shape of the discontinuity,
from being radiated locally downstream, the net radiated power
within the conduit is (eP. ). There is a sound power loss
through the walls of the conduit by the transmission coeffici-
ent, T, so that at the outer surface of the pipe the radiated
sound power is
Pr - PT eT
For bends and branch points in a conduit, e tends to unity;
valves and metering sections have values of e less than one.
The ratio of transmitted to incident acoustic power equals T.
It is approximately proportional to the pipe diameter to wall
thickness ratio, D/t, for excitation wavelengths larger than
2
the circumference of the cylindrical conduit, TfD = A . The
ringing frequency, f , equals c/A^ (see Fig. 1).
The "limiting transmission loss" in dB, for steel cylindri-
cal pipe is given by Fig. 2. The minor correction for actual
frequency relative to f is shown.
13
-------
ALUMINUM
STEEL
CONCRETE
10 12
D, PIPE DIAMETER, in.
14
16
18
20
Fig. 1 Ringing frequency, fr, for various diameter
pipes and pipe materials
14
-------
PQ
CO
o
60
50
-2
dB _o
CORRECTION
FOR
FREQ -4
-5
.05
0.1
\
I I I I I I
0.2 0.3 0.4 0.5 0.7 | 1.0
0.6 0.8
f/fr = 6.23x10"4f D
fr = C/TTD = 5050 m/sec (FQR SJEEL)
CO
CO
CO
40
30
20
10
-2
5 6 7 8 9 10'
-1
5 6 7 8 9 10U
t/D
Fig. 2 Limiting transmission loss of sound across
cylindrical steel pipe walls at less than the
ringing frequency, fr
15
-------
For smaller wavelengths than A^ the loss corresponds to
that of a flat panel of equal material and thickness to the
conduit. Here, the controlling factor is the product of
the surface mass (equal to the mass density divided by the
thickness) and the frequency. Figure 3 shows the theoreti-
cal transmission loss, in dB, for mass-controlled flat
panels. The three loss curves apply to different criteria,
discussed in Ref. 4fused for the directionality of the inci-
dent sound. The transmission losses for steel, aluminum, and
concrete panels are shown for several frequencies between
1000 and 20,000 Hz. At the ringing frequency, f , the
4 r
loss is a minimum. Therefore, the maximum signal through
the pipe wall will be obtained at this frequency, f , if
the sound sources within the pipe contain f .
The transmission of sound from one medium to another depends
on the acoustic coupling of the media. Using the transmis-
sion relation involving acoustic impedances, r = pc/S, in
3
two media, where S = surface area at the interface:
4r,r
P = L /
*12
/ ^ s
The sound transmission from water to iron or concrete is
about 157, and 53%, respectively. The poor acoustic
coupling to air is indicated by values of P,0, of about
-3. -4r -5
10 7,, 10 7,, and 10 70 for water to air, concrete to
air and iron to air interfaces, respectively.
B. EXPERIMENTAL FACILITIES
Most of the investigations made under this contract were con-
ducted in the Grumman Research water flow facility. AS shown
16
-------
CO
•o
70
60
50
40
30
20
10 —
f =
Ro (NORMAL INCIDENCE) - IN WATER
R = lOlog — ,dB
T
3 4 5 7 10 15 20KHz
I I I I I II
5 7 10 1520KHz
I I I I I
10 1520KHz
I I I
CONCRETE PANEL
10a
10C
10'
10°
10s
FREQ x SURFACE MASS, Hz-Kg/rn
Fig. 3 Theoretical sound transmission loss,
for mass controlled panels
17
-------
in Fig. 4, this facility uses a 240 gallon (908 liter) cyl-
indrical tank (see Fig. 5) as the primary water supply to
feed a 3-in. (7.6 cm) diameter steel pipe. Water is fed
into the tank by means of a hose connection from the plant
water supply, and stored under pressure until needed for
test. Pressure is supplied by a regulated air supply intro-
duced above the water level in the vertically oriented tank
(see Fig. 6). A plastic sight tube provides indications of
the water level. Flow rate is controlled by a 3-in. (7.6
cm) globe valve inserted in the 6 ft (1.8 m) high verti-
cal riser between the tank outlet and the 3-in. (7.6 cm)
horizontal pipe. With initial tank pressures up to 30 psi
2
(2.1 kg/cm ) maximum water flow rates of over 15 Ib/sec
(6.82 kg/sec) are manageable with the flow measurement tech-
nique used; this technique will be described later. Constant
flow rates during tests of up to about 2.5 Ib/sec (1.10 kg/
sec) are possible by having the plant water supply feeding
directly through the tank to the 3-in. (7.6 cm) pipe.
The 20 ft (6.1 m) length of 3-in. (7.6 cm) pipe (see
Fig. 7) effectively eliminates the flow profile irregulari-
ties induced by the flexible line and three 90 degree bands
between the tank and the pipe. At the downstream end of the
3-in. (7.6 cm) pipe the conduit discontinuity is introduced
by means of a test pipe section. Duration of tests usually
are between 30 seconds and one minute, and for flows above
about 3.5 Ib/sec (1.4 kg/sec) the flow rate slowly de-
creases during the latter part of the run. The outflow drops
into a wooden trough (Figs. 8 and 9) where it can be diverted
by a flapper valve into a drain, or into a collection drum.
The 55 gallon (208 liter) drum is connected by hose (see
18
-------
OP
0
ft)
05
05
l-i
o
P3
rt
05
r-i
en
£
VI
Hi
n
H-
U SIGHT TUBE GRIT
240 GAL. TANK INJECTOR
/_
AIR PRESSURE
REGULATOR WATER
SUPPLY
VALVE
V
FLOW
CONTROL
VALVE
20
FT
V
3 IN. DIA. STEEL PIPE
WOODEN SUPPORTS
SENSOR
MOUNTING
STUD
(TYP.)
/////////////
FLOW DIVERSION
TROUGH
?%#^J^%#S-^%^%%S^^
TOLEDO 55 GAL. DRUMS
SCALE FOR COLLECTING AND
(2500 LB CAP) MEASURING TEST FLOW
PLANT
DRAIN
I-1-
rt
-------
Fig. 5 240 gallon facility water storage tank
Fig. 6 Air pressurization and water supply pipes to
storage tank
20
-------
• .»ER HA _ naii
Fig. 7 20 foot length of 3-in. diameter pipe of facility
water supply system
Fig. 8 View of facility test section showing diversion
trough, water collection drums, and weighing
scale. 8-in. diameter aluminum pipe on test in
concentric configuration
21
-------
Figs. 10 and 8) to another one to double the total collec-
tion capacity and both drums rest on the weighing platform
of a Toledo scale. This scale has a 2500 Ib (1136 kg)
weighing capacity and has been calibrated to l/8th Ib
(0.06 gm) accuracy. Prior to each test run, a tare weight
reading is taken. When flow conditions are stabilized, the
flapper valve is opened allowing water to be diverted into
the drums, and a stop watch is activated. At the conclu-
sion of the test period, the watch is stopped as the flapper
valve is closed and a final weight is read off the scale
dial. The average rate of flow during this time is deter-
mined from these data. By reducing the collection time to,
for example, 10 seconds, a fairly constant flow rate is ob-
tained for all but the highest rates. After each run, a new
tare reading is taken or the drums are drained, depending on
the anticipated magnitude of flow rate for the next run.
For this program, the test sections of conduit were:
• ~ 2-in. (5.1 cm) high by ^ 5-in. (12.7 cm)
wide open aluminum channel
• 8-in. (20.3 cm) diameter aluminum pipe
• 4-in. (10.2 cm) diameter transite sewer pipe
• 6-in. (15.3 cm) diameter transite sewer pipe
and they were attached either concentrically or eccentrical-
ly. The resulting discontinuity geometries and distances
are depicted by Fig. 11.
The transducer types used to monitor the pseudosound at the
discontinuity are accelerometers. We selected accelerometers
in preference to several other transducer types that appeared
22
-------
9- flew sf Water flowing from 8-'inch, diameijea:
pipe Into
Fig. 10 Overall side view of facility test section
showing diversion trough, water collection drums,
and weighing scale. 4-inch diameter transite
pipe on test
23
-------
1. CLOSED CYLINDRICAL TO OPEN TRAPEZOIDAL
FLOW-
SENSOR
POSITION
Jr
)N \D
A^
(T)Q
J U 1
~
I
1-
f 2.53"
IW^> i f
i U U V
k*-*4 1-3" A /i art"
~ H \ 4.SJ —
K>
1
\ ,
I
+< — w
3" DIAM
STEEL
PIPE
1 1
1.62"
t
SECTION A-A
h* = 4.15,4.625,4.8 IN.
2. 3-IN. DIAM STEEL PIPE TO 3-IN. DIAM STEEL PIPE
• COUPLING
FLOW-
I_
.28"
-h* = 0.25" J 3" DIAM |
3.5" DIAM
SENSOR POSITION
3. 3-IN. DIAM STEEL PIPE TO 8-IN. DIAM ALUMINUM PIPE
a) CLOSED ECCENTRIC CYLINDRICAL CONFIGURATION
1) WITH EXTENDED INLET
3" DIAM PIPE
FLOW.
NOTE: 1 IN. = 2.54 cm
A •
fi
0_
-*•'
2.5" —
1
-^
© © ©
•A fcl^ — aJ — 7" 1
~>
y
!„ A
i I
\
8" DIAM PIPE
2) WITH FLUSH INLET
FLOW
t
i* = 4.45 IN.
1 ©
(NUMBER IN CIRCLE
DENOTES SENSOR
POSITION REFERENCE)
-9"-
Fig. 11 Sensor locations and channel discontinuities
tested in research laboratory, sheet 1 of 2
24
-------
-H
I
•8" DIAM PIPE
3" DIAM
© —
'"%
o .
'LI
^^_ (T " ^^^^
^^— iJ — ^^p
^©-3
0 ]
J
^
A /
© © © Ih* = 2.2"( /
LJ U U U
3. b) CLOSED CONCENTRIC CYLINDRICAL CONFIGURATION
1) WITH EXTENDED INLET
3" DIAM PIPE
FLOW
-8" DIAM PIPE
r
/f,_7.5.._»,
0_
-'
i
\
h* -
2
k /^N /^
U 1
i * »
J
)'
,'e rt V
i_i ^
r
j
8°40' —
ALTERNATE
AT SECTION A-A
3" DIAM PIPE
FLOW
©
3.0"
2) WITH FLUSH INLET
• 6" DIAM PIPE
2.3"-
©
TJ^ U
— 6"-
+ 9--
4. 3-IN. DIAM STEEL PIPE TO 6-IN. DIAM TRANSITE PIPE
CLOSED ECCENTRIC CYLINDRICAL CONFIGURATION WITH FLUSH INLET
Fig. 11 Sensor locations and channel discontinuities
tested in research laboratory, sheet 2 of 2
25
-------
potentially useful because these devices are inexpensive
(~ $250), have the best frequency and sensitivity character-
istics, and are capable of being environmentally sealed with-
out difficulty. Condenser microphones require an opening to
equalize static pressure and cannot be effectively sealed.
Piezoelectric microphones also require pressure equalization
and usually are limited to 10,000 Hz frequency. Hydrophone
type microphones are very expensive, and piezoresistive
microphones have low output sensitivity. Magnetic variable
reluctance transducers are affected by ambient magnetic
fields and do not have high frequency sensitivity.
The accelerometer is screwed onto mounting studs that are
cemented to the outer surface of the test conduit. We have
found Devcon "F" epoxy cement with 8070 aluminum content to
provide good acoustic coupling between the mounting stud and
the various test conduit materials. However, this cement
takes about four hours to cure properly. The many sensor
locations needed to be examined for signal optimization can
be accommodated best by attaching many mounting studs at
judiciously selected locations prior to test. Several ac-
celerometer models were used during the program. They vary
in sensitivity, size, weight, and one model has an integral
preamplifier to match impedance with external recording
circuits. All acoustic data have been recorded on £-in.
by 600 ft (6.3 mm by 180 m) Scotch brand magnetic in-
»
strumentation tape using a Nagra III model tape recorder. The
typical instrumentation equipment used is shown schematically
in Fig. 12.
About midway in the program, a wooden container, 22.4-in
(57 cm) wide by 48-in. (122 cm) long was constructed to
26
-------
TEST CONDUIT (CROSS-SECTION)
MOUNTING STUD FOR SENSOR
ACCELEROMETER PREAMPLIFIER AMPLIFIER
(TYPICAL)
NAGRA MODEL III MAGNETIC
TAPE RECORDER
Fig. 12 Schematic of testing instrumentation
27
-------
hold sand or soil (see Fig. 13). This equipment simulates
buried installations of the pipeline especially in the vi-
cinity of the conduit discontinuity (see Fig. 14). A maxi-
mum buried depth of 8-in. (20.3 cm) can be accommodated
with a uniform layer of sand circumferentially around the
largest test pipes used. It is possible to test with either
completely uncovered, partially buried, or fully buried in-
stallations of the pipes. This simulation closely meets
ASTM C-361 specifications for two pipe bedding techniques in
field installations.
In one of the tests conducted to simulate a heavy bed load
of coarse grit relatively large stone chips were placed at
the bottom of the test pipe and retained by a metal dam at
the exit plane of the pipe. Figure 15 shows an extra heavy
bed loading before water flow. Figure 16 shows the partial
washout of the uniform bed after being subjected to flow
rates above 2-3 Ib/sec (0.9-1.4 kg/sec).
In another study, medium-to-coarse sand grit (sieve sizes be-
tween 8 and 20 and a specific gravity of 2.62) was in-
jected into the flow. This required an additional piece of
equipment at the upstream end of the 3-in. (7.6 cm) pipe.
This grit feed device uses a dc servo motor-generator
(Motomatic Model E6500 Series 125) and a motor speed con-
troller (see Figs. 17 and 18). The motor shaft is attached
to a 5/8-in. (1.6 cm) diameter feed screw that traverses
the bottom of a grit storage hopper. The flow rate of the
grit depends on the rotational speed of the motor which is
set by a 1000 division vernier rheostat on the controller
box. For a given rheostat setting, the motor speed is main-
tained constant by the servo circuit.
28
-------
Fig. 13 Side view of box for containing sand used to
simulate buried pipe installation
Fig. 14 Top view of box used to simulate buried pipes.
6-inch diameter transite pipe is shown on test
with sand removed. Flow direction is from
left to right
29
-------
Fig- 15 End view of 8-inch diameter pipe showing initial
condition of heavy (1-inch) bed load of stone chips
Fig. 16 Heavy bed load after being disturbed by water
flow at rates up to 3 pounds per second
30
-------
Fig. 17 Equipment used to feed sand grit into upstream
water supply
Fig. 18 Closeup of speed controlled motor and supply
hopper used to feed sand grit into water supply
31
-------
Dry loose sand is dropped into the water flow through a
£-in. (0.64 cm) wide slit near the upstream end of the
3-in. (7.6 cm) pipe. As seen by the photos, the upper por-
tion of this pipe was partially removed and replaced by a
curved plastic shield. In this way the movement of the sand
particles downstream of the slit can be tracked visually.
However, because of the opening in the pipe it is not possi-
ble to introduce sand and exceed a water flow rate of some-
what more than 6 Ib/sec (2.73 kg/sec). This flow rate cor-
responds to a completely filled pipe. To eliminate clogging
of the drains and prevent serious error in measuring water
flow with the collection barrels, the grit is separated from
the water at the outflow of the test pipe. This is done by
a screening tray, mounted on rails, just under the test pipe
exit (see Fig. 19); the separated sand is removed for drying
and reused. Baffle plates prevent the outflow from splashing
out of the tray. The concentration of grit in the test water
flow is determined from a grab sample taken about midway
through a test run. The sample is contained in a 250 ml
glass beaker that has been manually inserted into the out-
flow. The grit is filtered from the water sample and mea-
surements are made of the captured water volume and the grit
weight in order to determine mg/liter concentration value.
It is assumed that the grab sample is representative of the
entire flow composition while the acoustic recordings are
made. For the test made, the sand concentrations were be-
tween 86 and 1682 mg/liter. These values simulate very
closely the recommended sewage characteristics for grit
(20-360 mg/liter) and settleable solids (200-1150
32
-------
In order to provide a preview of the type of results that
could be obtained in a field test phase, a limited amount of
field testing was made during this project. Several flow
measurements were made using pipelines feeding Grumman's
main sanitary sewage treatment plant. Figure 20 shows where
some of these data were obtained. This location is an under-
ground concrete vault, at the SE corner of the treatment
plant, where valves are installed to control the diversion of
sewage to two aerating tanks on either side of the building
at the rear of the picture. Figure 21 shows the 8-in.
(20.3 cm) plug valve and the 8 by 10-in. (20.3 x 25.4 cm)
expansion section which adapts the valve to the 10-in.
(25.4 cm) diameter cast iron sewer pipe. The accelerometer
sensor is seen about two-thirds of the section length down-
stream from the valve flange. To expedite versatility in
field measurements, a magnetic mounting stud was used to
couple acoustically the sensor to the pipe. A more intimate
coupling would have resulted if we used the epoxy-based ad-
hesive for the mounting stud, but the time necessary to cure
the epoxy would have delayed the test program. With the
magnetic attachment, we are able to shift rapidly sensor
locations upstream and downstream of the valve body. Another
field location used to make acoustic measurements is shown
in Fig. 22. This is at the north aeration tank of the treat-
ment plant. The 10-in. diameter sewer pipe rises from its
buried installation to enter a comninuter as shown in
Fig. 23. The sewer pipe changes from a cylindrical steel
channel to an open concrete conduit before entering the corn-
minuter. A smaller, 4-in. diameter (10.2 cm) pipe can be
seen in Fig. 23 as also rising from underneath the surface
33
-------
Fig. 19 Screen tray added to water collection system to
separate grit from collected water. 6-inch
diameter transite pipe is shown on test. Tray
is just below pipe exit
Fig. 20 Exterior view of valve vault (foreground) at
Grumman waste treatment plant "A". This is first
site of field test preview experiments
34
-------
Fig. 21 Interior view of valve vault showing 8-inch plug
valve and 8 x 10-inch reducer section with
accelerometer attached to outer wall for acoustic
emission monitoring of flow
-••-••_ • • • • . --.T-. ^- r
Fig. 22 Exterior view of aeration tank at Grumman waste
treatment plant "A". 10-inch diameter pipe
emerges from underground installation at left
center of photo. This is site of second series
of field test preview experiments
35
-------
and conducting its flow over the concrete wall. This pipe
brings sewage from another Grumman plant and discharges it
into the comminuter channel vertically above the exit plane
of the 10-in. (25.4 cm) pipe. Several pipe locations
were acoustically examined at this site for flow of just the
10-in. (25.4 cm) pipe as well as combined flow with the
4-in. (10.2 cm) pipe.
Processing of the tape recorded data is made with Grumman's
Fourier analyzer computer facilities shown in Fig. 24. The
block schematic of electronic and display equipment is shown
in Fig. 25. A closeup of the Nagra data recorder, Pemco con-
tinuous loop transfer recorder and HP-5465A, HP-5460A, and
HP-2114B Fourier Analyzer system is shown from right to left
in Fig. 26.
Total sound data, in the 40 to 22,000 Hz frequency band,
is transferred from the Nagra recorder to the Pemco continu-
ous loop recorder. The amount of original data transferred
depends on the length of the loop on the Pemco recorder;
usually at least 10 seconds of continuous recording was
used in our program for each test point. Sixty random sam-
plings of the loop, representing about one-fifth to one-
tenth of the total original recorded information, are pro-
cessed by the HP-5465A computer. Each sampling is broken
down by this digital computer into 1028 frequency components
and associated amplitudes using Fourier transforms of the
time-varying sensor signals, and then the spectra are co-
added (this process is called ensemble averaging). This tech-
nique extracts the useful signal from the recordings and prac-
tically cancels extraneous random noise. The rms average data
36
-------
Fig. 23 Closeup of 10-inch diameter pipe where second
field test preview series was conducted. Sensor
is shown attached to top center of horizontal
pipe. Tape recorder and amplifier used is shown
in foreground
Fig. 24 Data processing equipment used to obtain spectral
distribution from flow acoustic emission recordings
HP-5465A Fourier analyzer computer is in center of
picture
37
-------
SPECTRAL DYNAMICS SD104A-5
SWEEP OSCILLATOR
HP5460A
DISPLAY
UNIT
HP5465A
FOURIER
ANALYZER
& A/D
CONVERTER
HP2114B
COMPUTER
HP2748A
TAPE
READER
HP6823A POWER SUPPLY
AND AMPLIFIER
THIS EQUIPMENT USED FOR FREQUENCY
' CALIBRATION ONLY.
HP5233L ELECTRONIC COUNTER
Fig. 25 Schematic of acoustic emission data processing
system
Fig. 26 Closeup of data processing equipment showing (L to
R) HP-5465A Fourier analyzer computer, Pemco
continuous loop recorder and playback, and Nagra
tape recorder used for laboratory and field,
38
-------
value of the signal is displayed digitally on the HP-5460A
unit and a spectral distribution of amplitude is plotted on
semilog paper (dB versus frequency) by the Moseley 2FRA
plotter. Typical spectral distributions for the April 28
test series (see Section IIIC-7) are shown in Fig. 27 (a)
and (b). The upper curve (a) for the no flow condition
shows the mass distribution law fall-off of amplitude with
3
frequency for random, broadband noise. By contrast, the
spectra at a relatively high flow (^ 13 Ib/sec = 5.9 kg/sec)
given by curve (b) shows specifically strong content in
various frequency bands that are produced by the flow across
the conduit discontinuity as well as from other nonrandom
sources (e.g., bubbles, turbulence, valve noises, etc.). In
addition, the rms signal (i.e., integrated over-all frequen-
cies between 40 and 22,000 Hz) is 32 db greater for
the flow case [curve (b)] than the ambient noise level
[curve (a)].
A few select studies during the course of this program in-
dicated that biasing the transferred acoustic data to a par-
ticular portion of the original recording can result in a
± 2 db change in the rms signal value, for given test flow
conditions. This bias can be minimized but at considerable
inconvenience by transferring and connecting small pieces
of the original recording taken over its entire duration for
a test point. The limited magnitude of this bias effect,
and the extent of reproducibility of results, considering
the number of components integrated into all laboratory sys-
tems, did not warrant the delay that would be incurred in
processing with greater care the thousands of data points.
Referenced to 1 volt.
39
-------
FULL SCALE
= 94 dB r-
-10dB
-20 dB
-30 dB
RMS DATA - 26 dB
NAGRA SETTING: 120db
AMPLIFIER: 0 db
-40 dB
POS.3
SENSOR TYPE:
1020
(98 mv/g)
a) AMBIENT- NO FLOW
FULL SCALE
= 126dB i—
-10 dB -
-20 dB -
-30 dB -
RMS DATA - 34 dB
NAGRA SETTING: 140 db
AMPLIFIER: 20 db
-40 dB
POS. 3
SENSOR TYPE:
1020
(98 mv/g)
b) FLOW RATE = 13 Ib/sec FREQUENCY/4, Hz
10
Fig. 27 Typical Fourier processed spectral distributions
of laboratory sound recordings (10 Hz bandwidth)
40
-------
Another study to improve processing accuracy involved doub-
ling the quantity of samples randomly selected by the Fourier
analyzer. The processing time is doubled, but the results
obtained for several exploratory cases are not sufficiently
better than with 60 samples to justify the extra processing
time. Apparently, our standard sampling number produces
statistically good results.
Co EXPERIMENTAL PHASE
The experimental investigations of this program were mainly
concerned with clean water flow in Grumman1s hydraulics re-
search laboratory. However, a brief field test preview was
conducted after the laboratory tests were concluded. De-
scriptions of these facilities have been presented in the
previous section.
The test program was conducted from August 14, 1974 to July
1, 1975 and used various size pipes as well as an open chan-
nel. Salient features of the conduits, transducers, flow
parameters, and experimental setup conditions of each test
series are summarized in chronological order in the table.
In reviewing these tests here, each group of closely related
experiments is described separately with minimal comment.
Full discussion of the results is covered in Section HID.
It is to be noted that approximately the initial half of the
test program has been run with no special attempt to simulate
real underground installations of sewers (see item number 8
of Table T-l). This measure simplified the test setups and
41
-------
TABLE T-1. SUMMARY OF TESTING FEATURES (SHEET 1 OF 2)
No.
1
2
3
4
5
6
7
8
9
10
11
12
13
Date of Test
Feature
Channel size
a. 3" diam steel (stl)
b. 3" diam steel to 2" x 5" trapezoidal
c 3" diam steel to 8" diam aluminum
d. 3" diam steel to 4" diam transite
e. 3" diam steel to 6" diam transite
Channel type
a. Concentric cylindrical, closed
b. Eccentric cylind., closed
c. Closed cylind. to open trapezoidal
M ci's t" "tv Hi I" \
Characteristic frequency (kHz)
Sensor type
a. B&K model (no.)
b. Vibrametncs model (no.)
Sensor sensitivity, (mv/g)
Sensor installations
a. No. of axial positions
b. No. of circumferential posit.
Pipe installation features
a. Freely suspended in air
b. Simulated underground
c. In sim. box but not buried'
Max. flow rate (Ib/sec)
Pipe exit condition
a. Open exit
b. Exit weir installed
Supply pipe inlet condition
a. Extended inlet (overlap dist., in.)
b. Flush inlet
Solids in flow?
a. None (except rust)
b. Heavy bed load of stone chips
c. Light bed load of stone chips
d. Upstream dispersed grit
Remarks
1974
8/14
•
«
0 25
-60.
4332
62
1
1
8.4
•
Prelim
data
8/22
•
«
3 8
3.91
4332
62.
1
NA
• (3)
8/23
•
•
3 8
3.91
4333
19.5
1
NA
• (3)
Exploratory
acoustical
data
9/9
•
•
4 25
3.50
4333
4332
19.5
62
2
7.6
• (3)
9/12
•
•
4 625
3.21
4332
4333
19.5
62
2
5.5
• (3)
10/28
•
•
4 2
3.54
4332
62
3
7.0
• (2.5)
10/29
•
•
4 2
3.54
4333
19.5
3
7.0
• (2.5)
10/29
•
•
2 2
6.76
3
8.0
• (2.5)
10/30
•
•
2 2
6.76
7,0
•(2.5)
11/8
«
«
A 1 C
H. I D
3.58
3
7.0
• 116
1/11
•
•
4 8
3.10
7.5
• (16)
11/11
•
•
415
3.58
3
7.25
• (16)
11/11
•
•
0 0
6 76
-i
i
7.5
• (2.5)
1/16
•
•
4333
(2)
19.5
21.5
8
*
1/17
•
•
4333
4339
19.5
10.
8
•
112
•
•
4333
(2)
19.5
21.5
1
10
•
•
123
•
•
•^.
^•^
15
•
•
/29
•
•
t 45
3.32
333
(2) *
9.5.
1.5'
11
•
•
to
-------
TABLE T-1. SUMMARY OF TESTING FEATURES (SHEET 2 OF 2)
No.
1
2
3
4
5
7
8
9
10
11
12
13
Date of Test
Feature
Channel size
a. 3" diam steel (stl)
b. 3" diam steel to 2" x 5" trapezoidal
d. 3" diam steel to 4" diam transite
e. 3" diam steel to 6" diam transite
Channel type
a. uo ce t ic cyii orica , c oseo.
c. Closed cylind. to open trapezoidal
Sensor type
. .
bensor sensitivity, imv/g)
Sensor installations
Pipe installation features
a. Freely suspended in air
b. Simulated underground
c. In sim. box but not buried
Max. flow rate (Ib/sec)
Pipe exit condition
a. Open exit
b. Exit weir installed
Supply pipe inlet condition
a. Extended mlet (overlap dist., in.)
Solids in flow?
b. Heavy bed load of stone chips
c. Light bed load of stone chips
d. Upstream dispersed grit
Remarks
2/26
•
0 69
4333
(2)
19.5
21.5
3
.
10
•
2/27
.
2
•
10
•
2/28
.
•
8
•
3/6
4333
(2)
1020
19.5,
21.5,
98
•
12.5
•
3/12
4 25
•
13.25
•
3/21
2 2
•
12
•
3/27
•
12
•
1975
3/31
4333
(2)
19.5
21.5
•
11.5
•
3/31
•
8.25
•
.
4/9
4333
(2)
19.5,
21.5
98
•
11.0
•
*
4/23
•
•
11.0
•
4/24
•
19.5,
21.5
•
14.2
•
4/24
•
^
•
11.6
•
4/28
•
1020
19.5
98
•
14.0
•
5/8
•
4333
(2)
19.5
21.5
•
2
•
5/8
•
•
15.4
•
6/26
•
1020
19.5
98
•
5
•
•
8-20
Sieve
Size
S.G.=
2.62
7/1
a) In buried vault,
3) Emerging from1
underground
125.0
•
Sanitary
Sewage •
Field
Test
3 review •
CO
-------
was designed originally to expedite test schedules as well
as to avoid constructing test equipment that might prove
superfluous. Some of the data obtained with this initial
simple test setup, particularly for cylindrical pipe ele-
ments, appear in conflict with dipole theory trend pre-
dictions. In the interest of completeness several plots
of these discrepant data are included in the material of
this section. The reader is cautioned not to interpret
these data as evidence of a basic defect in our flow mea-
surement concept. As will be explained later in Section
HID, Discussion, these discrepant test data are solely
the result of shortcomings in our initial laboratory simu-
lation technique. This situation was rectified in the
latter half of the experimental program with relatively
simple, yet effective, measures developed to simulate
underground sewer installations.
1. Steel Supply Pipe (Tests of August 14, 1974)
Measurements were made for a 5 ft (1.52 m) length of
3-in. (7.6 cm) diameter steel pipe coupled to the 20 ft
(6.1 m) long, 3-in. (7.6 cm) steel water supply pipe of
the test facility. Because of the threaded coupling geome-
try [3.25-in. (8.3 cm) long] and pipe wall thickness
[0.216-in. (0.55 cm)] a 0.25-in. (0.64 cm) annular step
existed for a distance of 1.28-in. (3.25 cm) between the
two pipe sections (see Fig. 11).
44
-------
The objective of this test was to record the inherent noise
of the facility supply line for a case where just a minor
perturbation discontinuity was introduced into the pipeline.
A B&K 4332 accelerometer was attached to a steel mounting
stud that had been cemented to the outer surface of the pipe
at a position abutting the downstream end of the coupling and
on the bottom centerline. Devcon "F" epoxy cement with 80%
aluminum content, was the adhesive. Initial supply pressure
2 2
for the water was 15 Ib/in. (1.05 kg/cm ) and 20 seconds
collection time was employed to obtain average flow rates for
several control value settings. Between flows of 0 to
8.4 Ib/sec (0 to 3.8 kg/sec), the overall signal between
frequencies 40 and 22,000 Hz varied as shown by Fig. 28.
These data are duplicated by data obtained during the October
28-30 test series (see Section IIIC-3) in which the sensor
was located 2.5 pipe diameters upstream of a significant
conduit discontinuity (see Fig. 28). Both sets of acoustic
data appear to follow a lower rate of change with flow than
predicted by dipole theory. Data for two frequencies ex-
hibiting repeated peak amplitudes in the spectral distribu-
tion also are plotted on Fig. 28. The ambient sound (zero
flow) recorded in the laboratory displays two major peak
amplitudes in its spectral distribution, at 80 and 2160 Hz,
and a general trend of decreasing signal with increasing
frequency (see Fig. 29); this is characteristic of a mass-
3
law distribution of broadband acoustic energy. Spectral
distributions for several flow rates exhibit less of the
45
-------
LEGEND:
DATA OF 8/14/74
0 - ALL FREQUENCIES
(40-22,000 Hz)
A - -2160 Hz
4. - -880 Hz
DATA OF 10/28-30/74
" - ALL FREQUENCIES
(40-22,000 Hz)
(2.5D UPSTRM OF DISCONT'Y)
O
>
LU
cc
CO
•a
tn
O
-------
FULL SCALE = 108 dB RMS DATA: -22 dB
NAGRASETTING: 110dB
AMPLIFIER: 20 dB
-10 -
dB -20 -
-40
-30 —
a) AMBIENT-NO FLOW
NAGRASETTING: 130 dB
AMPLIFIER: 20 dB
FULL SCALE = 130 dB RMS DATA: -20 dB
0
-10 -
dB -20 -
-30 -
b) FLOW RATE = 5 (b/sec
102 103
FREQUENCY/4, Hz
10
Fig. 29 Typical spectral distributions of laboratory
sound recordings. Distributions obtained by
Fourier analyzer (10 Hz bandwidth)
47
-------
mass-law fall off that characterizes ambient noise and more
of a profile distinguished by several minor peaks, with a
major band at about 880 Hz. Figure 29b for a flow rate of
5 Ib/sec (2.27 kg/sec) typifies these latter type spectra.
The epoxy-based adhesive for accelerometer mounting studs was
judged satisfactory on the basis of adhesion and sound trans-
mission, and used in all subsequent tests.
2. Open Channel Conduit (Tests of August 22-23, September
12, and November 8-11. 1974)
In this test series, the supply pipe discharged into an open
channel having a trapezoidal flow cross section with a (see
Fig. 11) 4.82-in. (12.2 cm) minimum width and 1.66-in.
(4.2 cm) height. The maximum cross-sectional width of this
6 ft (1.8 m) long aluminum channel is 5.78-in. (14.7 cm).
The hydraulic radius of this open channel for different water
levels varies according to Fig. 30. The cylindrical supply
pipe was supported at a minus 40 minute slope above and in-
dependent of the trapezoidal channel so as to permit various
waterfall heights equal to or greater than 4.15-in. (10.5
cm). The supply pipe exit also overlapped by 16-in.
(40.6 cm) the upstream end of the open channel. The chan-
nel was mounted with a downward slope of 25 minutes of
arc. Two types of transducers were installed at four dif-
ferent axial locations along the outside bottom centerline
of the aluminum channel (see Fig. 11). In outward physical
appearance this test setup was an approximately ten times
scale version of the original exploratory tests conducted by
Grumman Research in August 1972.
48
-------
1.7
1.6
1.4
1.2
1.0
I
o
I 0.8
tr
in
H
0.6
0.4
0.2
0.2
r
5.78".
.4775"
i_ 212'
t
.465"
TRAPEZOIDAL
TEST CHANNEL
I
0.4 0.6 0.8
HYDRAULIC RADIUS, in.
1.0
1.2
Fig. 30 Hydraulic radius of test channel for different
water level heights
49
-------
The open channel downstream of the discontinuity made it con-
venient to determine the relation of flow rate to water level
in the 3-in. (7.6 cm) diameter supply pipe, as well as the
outfall trajectory parameters. These results are shown in
Fig. 31. The data for water level and flow rate are experi-
mental in the upper half of this figure; the area and volume
flow ratios are from tables in Ref. 5, and full flow area is
2 2
7.07-in. (45.6 cm ). Using the measured flow rate and
the flow ratios from this figure the supply pipe will run
full for flow rates above about 7.0 Ib/sec (3.18 kg/sec)
and have a velocity at the pipe exit of at least 2.3 ft/sec
(0.70 m. /sec). Higher exit velocities corresponding to
higher flow rates can be obtained by pressurization of the
water supply. For example, the highest flow rate measured
during the entire laboratory test phase was 15.4 Ib/sec
(7.0 kg/sec) for which the exit velocity is 5 ft/sec
(1.52 m/sec). The measured variation with flow rate of the
normalized axial offset distance for the waterfall's center
of contact upon the receiver channel, Xc/h, is shown in the
bottom half of Fig. 31, Extrapolated to the maximum flow
rate experienced in our tests, the normalized axial offset
would be a maximum of 2.2.
The acoustic measurements taken concurrent with the flow mea-
surements employed two accelerometers with sensitivities of
62.7 mv/g and 19.5 mv/g and with corresponding working
frequency ranges of 26,000 and 35,000 Hz, respectively
2
(1 g equals 980.6 cm/sec ). The 3:1 ratio for the trans-
ducer sensitivities is equivalent to a 10 dB signal dif-
ference. Flow rates up to 7.6 Ib/sec (3.45 kg/sec) were
50
-------
1.0 I—
Y/D
"FULL'Q/QFULL
a) HEIGHT OF PARTIALLY FILLED PIPE FLOW VS WATER FLOW RATE
1.8 i—
1.6 —
3"D
Xc
h
w, Ib/sec
b) NORMALIZED DISTANCE TO WATERFALL CENTER OF CONTACT VS FLOW RATE
Fig. 31 Calibration of research facility water supply
(A = flow area, Q = volumetric flow rate)
51
-------
established at initial supply pressures between about 8 and
2
18 psi (0.56 and 1.27 kg/cm ).
Typical results of processed signal level to flow rate are
presented in Figs. 32 to 34 for two waterfall heights, two
accelerometer sensitivities, and four sensor locations. In
Fig. 32 the signals at two frequency bands are shown nor-
malized to the zero flow (ambient) condition. This has been
done to compensate for the difference in accelerometer sensi-
tivity that would produce a 10 dB level difference, if
shown on an absolute scale. In Fig. 33 the signal components
at 3600 Hz (i.e., f/4 - 900) are considered characteristic
for the waterfall height, h, of 4.15-in. (10.5 cm) accord-
ing to our empirical rule that h =A/4 where A=c/f. The
other frequencies (f/4= 800 and 710 Hz), shown in Fig. 32,
correspond to h = A/4.5 and A/5, respectively, and have been
presented because peak signal amplitudes are observed in the
computer processed spectral distributions at these frequencies.
Figure 34 shows typical results for a waterfall height, h,
of 4.8-in. (12.2 cm). This configuration has a character-
istic frequency of 3125 Hz. Processed data are presented
for other frequencies, 2040 and 2460 Hz, which exhibit
peaking amplitudes in the spectral distribution. These other
frequencies correspond to approximately A/6 and A/5, re-
spectively.
3. Eccentric Steel and Aluminum Pipes (Tests of September 9,
October 28-29, 1974, and January 29, March 6 and 12, 1975)
The conduit arrangement for these tests consisted of the 3-in.
(7.6 cm) diameter steel pipe discharging into a nominal
8-in. (20.3 cm) diameter, 4 ft (1.22 m) long, aluminum
52
-------
3" DIAM PIPE
^
-5"x 1.6"
TRAPEZOIDAL CHANNEL
1 4.15"
QU ® ® L
16" «
CD
T>
UJ
CQ
O
ffl
CD
1/3
O
O
O
<
Q
LU
N
CC
O
50
40
30
20
10
f/4 = 800, POS 4
f/4 = 710, POS5
f/4 = 800, POS 5
f/4 = 800, POS 1
I I
0 1
2
3
4
w, Ib/sec
5
6
7 £
Fig. 32 Variation of normalized acoustic signal with
flow (cylindrical pipe exiting to a trapezoidal
open channel)
53
-------
3" DIAMPIPE
s
J . ©
T)b-N D-
4—*
\
-5"x 1.6"
TRAPEZOIDAL CHANNEL
DATA OF f/4 = 900 Hz
150
140 —
DIPOLE THEORY
SENSOR
4332, POS. 4
Fig. 33 Acoustic signal variation with flow rate
(cylindrical pipe discharging to trapezoidal
open channel)
54
-------
.3" DIAM PIPE
5" x 1.6" TRAPEZOIDAL
CHANNEL
SENSOR POSITIONS
POS 1
POS 2 & 5
POS 1
POS 1
POS 2
0 1
2
3
4
w, Ib/sec
5
6
7
8
Fig. 34 Variation of normalized acoustic signal with flow
rate - different sensor locations (cylindrical
to trapezoidal channel transition)
55
-------
pipe. This setup was the first one involving completely
cylindrical elements and was designed to demonstrate the
production of acoustic emission in channels that were not
always completely open (see Fig. 11). The pipes were at-
tached in an eccentric manner so that the waterfall height
was 4.2-in. (10.67 cm) or 4.45-in. (11.3 cm). The height
depends on whether the smaller pipe has an extended inlet
within the larger one, or is assembled as a sudden enlarge-
ment so that the roof line for the joined pipes is tangent.
The upstream end of the 8-in. (20.3 cm) pipe was closed
by an aluminum plate having an eccentric hole large enough
to fit the 3-in. (7.6 cm) pipe (see Fig. 8). All joints
were sealed watertight with 1-in. (2.54 cm) wide adhesive
faced lead tape. The slope of both pipes was minus 0.3%.
Flow rates up to more than 13 Ib/sec (5.9 kg/sec) were
established at water supply pressures between 8 and 27 psi
2
(0.56 and 1.90 kg/cm ). Three different accelerometer
models were attached to mounting studs at four axial loca-
tions on the 8-in. (20.3 cm) diameter pipe and one loca-
tion on the 3-in. (7.6 cm) pipe. In addition, during
several tests two circumferential locations were monitored,
at 45 and 180 degrees counterclockwise from the bottom
centerline viewed looking upstream, at an axial position
6-in. (15.2 cm) from the upstream end of the larger pipe.
After a series of tests in which the smaller pipe extended
2.5-in. (6.4 cm) into the 8-in. (20.3 cm) pipe, tape re-
corded acoustic data were processed and typical results are
shown in Figs. 35 and 36. The sensor acoustic signals at
different positions are shown varying with flow rate in
Fig. 35; these trends are for the characteristic frequency
56
-------
3" DIAM PIPE
\
8" DIAM PIPE
fc
(D
"••-P
-»u
U-2.5"-^
© © /
U i
*-3"->J
<« 6.4" »
j
POS.
POS. 6
35 Variation of normalized acoustic signal with
flow rate - effect of sensor location
(eccentric cylindrical pipe elements and
extended inlet)
57
-------
NUMBERS IN CIRCLES DENOTE
SENSOR POSITIONS
3" DIAM PIPE
8" DIAM PIPE
SENSOR SENSITIVITY,
SYMBOL TYPE POSITION mv/g
z
UJ
CO
LU
O
m
y
l-
t/i
O
O
<
Q
LU
N
Lt
O
40 r-
\>od
o
•
+
A X
a
•
ys
4332
4333
4333
4332
4333
4333
4333
4332
3
6
4
4
3
1
5
5
62.7
19.5
19.5
62.7
19.5
19.5
19.5
62.7
w, Ib/sec
Fig. 36 Effect of sensor on variation of normalized
acoustic signal with flow rate (eccentric
cylindrical pipe elements and extended inlet)
58
-------
of 3540 Hz. Figure 36 shows similar data to Fig. 35 but for
an additional sensor and at another sensor location. Data
points allow results of different test setups to be distin-
guished from one another.
In another test, the smaller diameter pipe was mounted so
that its exit plane and the entrance plane of the larger pipe
were flush. The purpose of this geometric array is to exam-
ine the influence of the air space behind the waterfall, as
a resonance chamber driven by waterfall instabilities, to
modulate the flow acoustic emission signal strength. Test
data given in Fig. 37 show that for comparable sensor loca-
tions (relative to the discontinuity plane) the setup with
the maximum air volume behind the waterfall produces the
largest signal amplitude. These data are for the signal com-
ponent at the characteristic frequency and for flow rates up
to 11 Ib/sec (5.0 kg/sec).
Visual observation of nonuniform or high flow rates reveal
various types of mild hydraulic jump conditions in the down-
stream channel. To investigate the effect that such rapid
changes in water level might have on flow sound emission, we
made three weirs to be attached to the exit plane of the
8-in. (20.3 cm) diameter pipe. The weirs have heights of
2.24, 3.24, and 4.24-in. (5.7, 8.2, and 10.8 cm). Be-
cause of the slope and length of the test pipe section, the
shortest weir retains just a film of water directly below
the discontinuity, and the highest weir retains a 2-in.
(5.1 cm) deep water pool there, which almost halves the free
fall of water from the 3-in. (7.6 cm) pipe.
Typical trends of the processed acoustic data to a flow rate
of 11 Ib/sec (5.0 kg/sec) are shown in Figs. 38-40. In
59
-------
• DATA AT CHARACTERISTIC FREQUENCIES
• (SENSOR TYPE - 4333, SENSITIVITY - 19.5 mv/g)
3" DIAM PIPE
m
•o
z
o
co
O
CO
O
O
140 I—
120 —
100
80
60
EXTENDED
INLET
D
t4,5" f
i #3 #4 L
2.5"
3"
* »
fi
8" DIAM PIPE
FLUSH
INLET
I
4.45"
— POS3, f/4 = 900Hz
POS4, f/4 =900 Hz
"• — — ___ POS 2, f/4 = 830 Hz
POS 1, f/4 = 830 Hz
#2
*J 6"
L
6 8
w, Ib/sec
10
12
14
Fig. 37 Effect of extended inlet on acoustic signal
variation with flow rate (eccentric cylindrical
pipe elements)
60
-------
3" DIAM 8" DIAfl.
\ /
I t
4.45"
© © tf
ALL FREQUENCIES (40-22,000 Hz)
POS 1
LU 20
IT
O
10
fc/4 = 830Hz POS 1,f/4 = 2010Hz
fc = CHARACTERISTIC FREQUENCY = 3320 Hz
6
w, Ib/sec
10
12
Fig. 38 Effect of sensor position and sound frequency
on acoustic emission of water flow (eccentric
cylindrical pipe elements and flush inlet)
61
-------
3" DIAM-
' DIAM
. SENSOR POS. 2
• DATA AT f = 3320 Hz
NO WEIR
0
246
w, Ib/sec
8
10
12
Fig. 39 Effect of exit weir height on normalized
acoustic signal of water flow (eccentric
cylindrical pipe elements and flush inlet)
62
-------
3" DIAM
.8" DIAM
z
UJ
. SENSOR POS. 2
• DATA AT f = 6640 Hz
W/3.24"WEIR
W/OWEIR
w, Ib/sec
Fig. 40 Effect of exit weir height on normalized
acoustic signal of water flow (eccentric
cylindrical pipe elements and flush inlet)
63
-------
the first of these figures the data for two sensor positions
and no weir displays the signal for the entire acoustic spec-
trum and the minimum characteristic frequency (3320 Hz),
computed on the basis of maximum waterfall height
(4.45-in. = 11.3 cm). The rapid increase in signal above
ambient up to about 2 Ib/sec (0.91 kg/sec) flow agrees
with dipole theory predictions. At greater flows, the in-
crease in useful signal is not as large as theoretically
predicted.
Figure 39 shows that when a greater depth of water pool
exists (highest weir) the acoustic signal is reduced, at the
low flow rates, compared to shallower or nonexistent water
pools. At flow rates greater than about 12 Ib/sec (5.45
kg/sec), however, the acoustic emission signals for all
four pool conditions appear to converge into a single trend.
If the water backed up by the weir were to behave exactly
like the solid wall of the conduit in producing pseudosound,
then the data at a characteristic frequency based on the free
fall distance to the water pool should yield improved acous-
tic signal data. This is not indicated by the data compari-
son of Fig. 40 with Fig. 39, where the characteristic fre-
quency of Fig. 40 is computed to correspond to the shortened
drop to the 2-in. (5.1 cm) deep water pool. The trends of
these two figures are very similar and probably differ only
by the differences in background (ambient) noise level at the
two frequencies compared.
Figure 41 depicts the variation of normalized acoustic signal
with flow at four locations of accelerometer installation.
At position 1, closest to the discontinuity plane, the
64
-------
3" DIAM PIPE
8"DIAM PIPE
70 I—
WEIR
3"
• DATA AT CHARACTERISTIC
FREQUENCY = fc = 3320 Hz
DIPOLE THEORY
POS 1, W/O WEIR
V
POS 1
POS2
SENSOR SENSOR SENSITIVITY
POSITION TYPE mv/g
2,4
3
1
4333 19.5
1020 98.0
4333 21.5
6 8
w, Ib/sec
10
12
14
Fig. 41 Variation of signal with flow at different
sensor locations; (eccentric cylindrical
pipe elements and flush inlet) (with 4.24-
inch weir at exit unless noted otherwise)
65
-------
normalized acoustic signals for the weir equipped pipe are
similar to the open-ended conduit, up to a flow of 3 Ib/sec
(1.36 kg/sec). At higher flows the latter pipe configura-
tion gives 3 to 4 dB greater signal.
At positions 2 and 4, which are at the same axial position
but circumferentially separated by 45 degrees, the signal
trends for both locations are almost identical and follow
dipole theory to flow rates of 1.5 Ib/sec (0.68 kg/sec).
For greater flows up to about 10 Ib/sec (45.5 kg/sec),
the signal at position 4 is greater than at the bottom cen-
terline (position 2). At still greater flows there is neg-
ligible difference between the signal levels at each sensor
location.
Most practical sewers are in buried installations where the
overburden depth is at least equal to one pipe diameter. We
have simulated such installations in our laboratory by en-
closing the section of test pipe, in the vicinity of the dis-
continuity plane, within a special wooden box (see Section
IIIB) and Figs. 13 and 14). The four feet (1.2 m) long
box contains coarse sand that we have used to simulate typi-
cal backfill. A maximum overburden depth of 8-in. (20.3
cm) is possible as well as partial burials and freely sus-
pended pipe installations. The sand is tamped down before a
test series to eliminate air pockets. Typical data obtained
for these conditions are presented in Figs. 42-44. In the
first figure of this series, the normalized acoustic signal
trends are shown for flows up to 13.25 Ib/sec (6.02 kg/
sec). These curves are at a characteristic frequency, f
of 3320 Hz, and at twice f . They indicate a signal
66
-------
3" DIAM
V
8" DIAM
)
I- d
r
WEIR
Upos 1 U POS 3
• COVERED BY 8" OVER BURDEN OF COARSE SAND;
• WITH 4.24" WEIR UNLESS OTHERWISE NOTED
f/4 = 1660 Hz, POS 3, W/O WEIR
f/4 = 1660-1500 Hz, POS 3
fc/4 = 830 Hz, POS 3, W/O WEIR
fc/4 = 830 Hz, POS 3 f/4 = 1660 to 1500 Hz, POS 1
SENSOR SENSITIVITY
POSITION TYPE mv/g
Fig. 42 Variation of normalized acoustic signal with
flow rate for different frequencies and test
conditions (eccentric cylindrical pipe
elements and flush inlet)
67
-------
3" DIAM
8" DIAM
/
#2
#4
45°
• COVERED BY 8" SAND EXCEPT WHERE NOTED.
• DATA AT fc/4 = 830 Hz
• POS.4
100 |—
LINEARIZED SCALE, W/WEIR
W/WEIR, W/O SAND OVERBURDEN
Fig. 43 Variation of normalized acoustic signal with
flow rate for different test conditions (flush
inlet, eccentric cylindrical pipe step)
-------
level reduction of about 5 dB when the highest weir is
placed at the pipe exit. This is measured at a sensor loca-
tion slightly greater than one diameter downstream from the
discontinuity plane.
The effect, at one sensor location, of the 8-in. (20.3 cm)
deep overburden of sand on the normalized signal trend, at
the characteristic frequency, is given by Fig. 43. At flows
above 1.5 Ib/sec (0.68 kg/sec) and no weir at the pipe
exit, the sand covered installation has a greater signal for
a flow rate than the unburied conduit. This trend also
exists for the buried pipe with a weir. The acoustic signal
trend assumes a large change in slope, for dB versus w
axes, above w ~ 1.5 Ib/sec (0.68 kg/sec). This curve shape
change does not mean greatly reduced resolution, however. If
we convert the dB scale to a linear ratio scale, as depicted
on Fig. 43, it is possible to resolve 0.025 Ib/sec (0.011
kg/sec) in flow rate with working scale curves. In most prac-
tical system applications, a linear scale would be used for
readout devices such as gauges and strip charts. The data of
Fig. 44 presents absolute signal levels in dB, for three
accelerometer locations.
4. Concentric Steel and Aluminum Pipes (Tests of October 29^
30, November 11, 1974. and January 16-17, 22-23, March
21, 27, 31, and April 9, 1975)
The arrangements of flow conduits for these tests are similar
to those described in the preceding Series 3 except that the
3-in. (7.6 cm) diameter steel pipe is attached concentric
to the 8-in. (20.3 cm) aluminum pipe. Because of the ac-
tual physical dimensions of the pipes the waterfall height
is 2.2-in. (5.6 cm) which corresponds to a characteristic
69
-------
DATA AT
CHARACTERISTIC
FREQ = fc/4 = 830Hz
'_)#4
3-K-J I
•*-6"-«H
-—9"—H
#3
NOTE:
* = POS 4, W/SAND, W/O WEIR
120i—
POS 3, W/SAND, W/O WEIR
POS 3, W/O SAND, W/WEIR
POS 4, W/SAND, W/WEIR
POS 1, W/O SAND, W/WEIR
POS 1 W/SAND, W/WEIR
POS 1, W/SAND, W/O WEIR
POS 4, W/O SAND, W/WEIR
POS 3, W/SAND, W/WEIR
60
w, Ib/sec
Fig. 44 Variation of acoustic signal with flow for 3
sensor locations and various test conditions
•ig. 45 Closeup of concentric discontinuity plane assembly
of 3-inch steel water supply pipe to 8-inch
diameter aluminum test pipe
70
-------
frequency of 6700 Hz in water. The upstream end of the
8-in. (20.3 cm) pipe was closed by an aluminum plate and
sealed with adhesive faced lead tape (see Fig. 45). The
slope of the pipe assembly was minus 0.3% initially;
this was altered in later tests to minus 370 to improve
drainage. Flow rates to 15 Ib/sec (6.8 kg/sec) were
established at water supply pressures between 8 and 27 psi
2
(0.56 and 1.90 kg/cm ). Four different accelerometer
models were used at four axial locations relative to the dis-
continuity plane, and at five circumferential positions at
one axial position. As with the eccentric arrangement pre-
viously described, three weirs of differing heights were
used at the exit plane of the conduit to force retention of
water, of different depths, beneath the discontinuity plane.
Also, tests were conducted with the pipes buried under one
pipe diameter depth of coarse sand.
One additional test was made with this concentric array in
which the bottom of the 8-in. (20.3 cm) diameter was
filled with stone chips to a maximum depth of 1-in. (2.54
cm) along its entire 4 foot (1.22 m) length. This ex-
treme pipe bed load condition was prevented from being washed
away by the 4.24-in. (10.8 cm) weir plate at the exit
plane. Views of this bed load, with the weir removed, are
shown in Figs. 15 and 46 taken before water flows, and
Figs. 16 and 47 taken after a series of tests with flow rates
above 2-3 Ib/sec (0.9 to 1.4 kg/sec), and 5 Ib/sec
(2.3 kg/sec), respectively. The purpose of these stone
chips is threefold. First, it is used to indicate whether a
diffuse solid material, which is not cemented to the con-
duit construction, could change the amplitude of acoustic
71
-------
Fig. 46 Internal relation of 3-inch supply pipe to 8-inch
test pipe with heavy bed load of stone chips.
View taken before flow established
Fig. 47 View of upset bed load after water flow rate of
5 pounds per second
72
-------
emission at the original characteristic frequency. Then,
would the characteristic frequency be changed if the stones
change the free fall height of the waterfall? Finally would
the movement of the stones, under the influence of water
flow, alter the amplitude of acoustic emission of the flow,
or otherwise interfere with interpretation of the acoustic
data? It was realized at the time that a bed load of stones
(average specific gravity of 2.75) that reduced the flow
cross section by almost 7.5 percent was likely to be an un-
common event. The test, however, was considered as poten-
tially informative for more carefully designed future experi-
ments. However, in recognition of the fact that the extremely
heavy bed load conditions imposed by this test setup were un-
realistic, a follow-up test was made in which the stone chips
were uniformly distributed in as near a single layer of
stones as possible, along the bottom inner surface of the en-
tire length of 8-in. (20.3 cm) pipe.
Typical data for the concentric mounting with an extended in-
let are presented in Figs. 48 and 49. The first gives the
sensor signal component at a characteristic frequency of
6700 Hz for four sensor mounting locations. This signal has
been normalized to the ambient noise level (zero flow) to
compensate for the differences in sensitivity and installa-
tion for the two accelerometers. At a good sensor location,
3-in. (7.6 cm) downstream from the discontinuity plane, the
normalized signals for several frequencies are shown by
Fig. 49. It is evident from this figure that the trend at
the characteristic frequency is uniquely unambiguous and use-
ful for relating signal amplitude to flow rate. The other
frequencies shown correspond to other peak amplitudes in
73
-------
3" DIAM
(
©
«-»
8" DIAM
NUMBERS IN CIRCLES DENOTE
SENSOR POSITIONS
SENSOR TYPE 4333 AT ALL
LOCATIONS (SENSITIVITY =
19.5 mv/g)
DATA ATfc/4= 1675 Hz
POS 4
POS 1
POS 3
w,Ib/sec
Fig. 48 Variation of normalized acoustic signal with
flow rate at various sensor locations and
characteristic frequency (concentric cylindrical
pipe array with extended inlet)
74
-------
3" DIAM
8" DIAM
TRANSDUCER 4332
SENSITIVITY = 62.7 mv/g
GO
TD
fc/4= 1675 Hz
f/4 = 775 Hz
f/4 = 1350 Hz
f/4 = 510 Hz
Fig. 49 Variation of normalized acoustic signal with
flow rate for different frequencies (concentric
cylindrical pipe array with extended inlet)
75
-------
the spectral distribution resulting from Fourier processing
of the recorded data.
The normalized data shown in Fig. 50 are results of the in-
vestigation into signal improvement by varying the circum-
ferential location of the sensor attachment. The concentric
cylindrical pipe array has a flush inlet in these tests.
While all locations yield a generally similar result at low
flows, the trends indicate that a position of 45 degrees to
the bottom centerline appears to yield greater signals, at
the intermediate to high flow rates, than at other circum-
ferential positions.
The effect of different weir heights on the acoustic signal
monitored at two positions, along the bottom centerline of
the larger pipe, is shown by Fig. 51 for the characteristic
frequency. Data obtained at position 2 are somewhat better
than for position 1 at flow rates greater than about 8 lb/
sec (3.6 kg/sec).
Comparative data for component frequencies of the sound other
than the characteristic frequency, are given in Figs. 52 and
53 for the highest weir-equipped conduit. The overall signal
strength for the entire recorded frequency range of 40 to
22,000 Hz is also given in Fig. 53. Data at position 3 al-
ways is of the highest amplitude because the sensitivity of
the transducer used at this mounting location is about five
times greater than for the other accelerometers. If the
sources of acoustic signals were equally strong at each mount-
ing stud location the accelerometer with enhanced sensitivity
would produce a 14 dB greater signal than the less sensitive
transducers.
76
-------
3" DIAM
8" DIAM
SENSOR
POSITION
DATA ATfc/4= 1675 Hz
CD
T3
LLJ
CD
O
CO
CO
O
I-
cn
O
O
<
O
HI
N
tr
o
2
45° POS 1
0 1
2 3 4 5 6 7 £
w, pps
3 9
10
Fig. 50 Effect of sensor circumferential and axial
location on useful signal (concentric cylindrical
pipe array with flush inlet)
77
-------
3" DIAWI-
8" DIAM
,u
-6"-*-l
• DATA AT ^-=1680 Hz
2.24" WEIR 3.24" WEIR
NO WEIR
m
T3
en
o
in
O
U
L,... \
POS2
NO WEIR
POS 1
Fig. 51 Variation of acoustic signal with flow at two sensor
locations for different test conditions (flush inlet,
concentric cylindrical pipes)
78
-------
3" diam
STEEL
8" diam
ALUMINUM
#1 #2 #3
150 i—
130
110
90
< 70
2
2
CO
o
H
co
O
O
POS3
POS2
POS 1
DIPOLE THEORY
TREND
re: w=1 Ib/s
024
a) DATA AT f/4 ~ 200 HZ
6 8
W, Ib/sec
10
12
14
DIPOLE THEORY TREND
(re: w=1 Ib/s)
110 r—
90
70
50
POS 2
POS 3
POS 1
DIPOLE THEORY TREND
(re: w=2 Ib/s)
2 4
b) DATA AT fc/4 = 1680 Hz
10
12
14
w, Ib/sec
Fig. 52 Variation of acoustic signal with flow at three
sensor locations and for two frequencies - with
4.24" weir at pipe exit plane - (concentric
cylindrical pipe elements)
79
-------
130
110
90
70
50
POS3
POS 1
_L
DIPOLE THEORY TREND
(re: w=2 Ib/s)
_L
_L
_L
J
0 2 4 6 8 10 12 14
w, Ib/sec
a) DATA ATf/4= 1400 Hz
CO
•a
O
C/3
D
O
O
140 r
120 -
POS 3
DIPOLE THEORY TREND
(re: w=2 Ib/sec)
100 -
b) (ALL FREQUENCIES
BETWEEN 40 AND
22,000 Hz)
120
100
80
60
50
POS 3
POS 1
DIPOLE THEORY TREND
' (re: w=2 Ib/s)
_L
_L
6 8
w, Ib/sec
10
12
14
c) DATA AT f/4 = 2700 Hz
Fig. 53 Variation of acoustic signal with flow at three
sensor locations and three frequency bands - with
4.24 inch weir at pipe exit plane - (concentric
cylindrical pipe elements). See Fig. 52 for sensor
position location on pipe.
80
-------
Typical acoustic signal data where the section of joined
pipes near the discontinuity plane is buried into a minimum
of 8-in. (20.3 cm) of sand is given in Figs. 54 and 55.
These curves are for three accelerometer locations and pre-
sent the normalized acoustic signal component at four fre-
quencies as well as the absolute overall signal. The nor-
malized dipole theory trend is close to the signal trend at
sensor positions 2 and 3 for the characteristic frequency
(6720 + 80 Hz).
The effect of the very heavy bed of stones, placed at the
bottom of the 8-in. (20.3 cm) pipe, is indicated from the
data plots of Fig. 56. At the characteristic frequency, the
signals at both sensor positions are within 2 dB of each
other, whereas for the overall range of frequencies about a
7 dB difference exists. It appears from the data shown
in Fig. 56 that the weir's use results in a more gradual
and resolvable signal trend, especially at the low flow
rates up to about 8 Ib/sec (3.6 kg/sec).
The acoustic signal obtainable for a bare conduit is com-
pared to that with both heavy and light bed loads of stones,
in Fig. 57. These data are for the characteristic frequency
component as well as the integrated value over the entire
frequency spectrum.
An appreciation of the physical action of the waterfall in
displacing even a heavy bed of stones can be obtained from
the sketches of Fig. 58, and Figs. 15, 16, 46, and 47. The
drawings are interpretations of the experimenter's observa-
81
-------
COVERED BY SAND TO 8" DEPTH
POSITION ( 1
FOR ALL FREQUENCIES (WITHOUT WEIR)
(DATA NOT SMOOTHED)
FOR ALL FREQUENCIES
IN 40- 22000 Hz BAND
(DATA NOT SMOOTHED FOR CURVE)
280 Hz = f/4
420 Hz = f/4
DIPOLE THEORY (NORMALIZED TO 1 pps)
_L
J
CO
T3
_T
<
"Z.
v
o
h-
D
O
O
A) SENSOR POSITION 3
120
100
FOR ALL FREQUENCIES (WITHOUT WEIR)
(DATA NOT SMOOTHED FOR CURVE)
FOR ALL FREQUENCIES
IN 40- 22000 Hz BAND
(NOT SMOOTHED)
fc/4 = 1675 280 Hz = f/4
O
O
<
Q
HI
ISI CD
"
0^
Zoo
40
20
I
420 Hz = f/4 672 Hz = f/4
024
B) SENSOR POSITION 1
6 8
w,Ib/sec
10
12
14
Fig. 54 Variation of acoustic signal with flow at two
sensor locations and two frequency bands
(concentric cylindrical pipe elements) - with 4.24"
weir at pipe exit and 8" deep sand overburden,
except where noted
82
-------
FOR ALL FREQUENCIES IN
40- 22,000 Hz BAND
(DATA NOT SMOOTHED FOR CURVE)
CD
•o
oo
O
O
120
100
80
60
\
O
O
u
<
a
LU
NCQ
"
40
20
FOR ALL FREQUENCIES IN
40-22,000 Hz BAND
(WITHOUT WEIR)
(DATA NOT SMOOTHED
FOR CURVE)
SENSOR POSITION 2
(SEE FIG. 54 FOR SENSOR LOCATIONDATA)
280 Hz = f/4
DIPOLE THEORY (NORMALIZED TO 1 pps)
672 Hz = f/4
8
w, Ib/sec
10
12
14
16
Fig. 55 Variation o±. acoustic signal with flow at five
frequency bands and one sensor location
(concentric cylindrical pipe elements) - with 4.24"
weir at pipe exit and 8" deep sand over burden,
except where noted
-------
3" diam •
8" diam •
POS. 1 (W/O WEIR AND STONES)
120 |—
100
80
§ 60
POS. 2 (W/O WEIR AND STONES)
POS 2 W/WEIR
J L
a) DATA AT ALL FREQUENCIES (40-22,000 Hz)
POS 1
W/WEIR
AND STONES
POS 1 (W/O WEIR AND STONES)
POS 2 W/O WEIR & STONES
POS 2 W/WEIR & STONES
6
w,Ib/sec
10
12
b) DATA AT fc = 6720 Hz
Fig. 56 Effect of heavy bed load of stones (1" deep) in
8" diam. pipe on acoustic signal variation with
flow - sand covered array
84
-------
\L) v£,
-Q—O
•^•••••^.^
8" diam
,POS
120 I—
100
80
CD
T3
_r 120
<
z
o
to
o
CO
o 10°
o
80
W/O STONE LAYER
W/1" STONE LAYER
I
a) DATA AT ALL FREQUENCIES (40 - 22,000 Hz)
60
50
W/THIN
STONE LAYER
POS. 1 & 2 (W/O STONES)
POS. 1 & 2 (W/1" STONE LAYER)
I
j
14
024
b) DATA AT fc/4 = 1680 Hz
6 8
w,Ib/sec
10
12
Fig. 57 Effect of two bed loads of stones in 8 inch diameter
pipe on acoustic signal variation with flow - sand
covered array with 4.24 inch weir
85
-------
3" diam STEEL PIPE
8" diam ALUMINUM PIPE
A-
STONE.
CHIPS
a) PIPE CONFIGURATION BEFORE TEST
SECTION
A-A
b) PIPE CONFIGURATION AFTER 3.08 Ib/sec FLOW RATE
AVJ&
:?:>.•..
•o.X->.'~
•
-------
tion after various stages of the tests. The observed local
washout of the stone bed as the flow increases practically
assures that the full discontinuity geometry will be main-
tained, and the characteristic frequency will be unaltered.
Further tests where a bed load can be securely attached
(e.g., by cement) to withstand such washout, could provide
valuable extrapolation of the data already obtained. Fig-
ure 59 shows that absolute signal components at frequencies
much above as well as much below the characteristic fre-
quency (f ) are of larger amplitude than f for the sen-
c c
sor location closest to the discontinuity plane (i.e., posi-
tion 1). Note that since the zero flow (ambient) signal
level is greatest for the lowest frequencies because of the
mass distribution law, the signal-above-ambient at, for ex-
ample, 800 Hz, actually is less than at the characteristic
frequency for all flow rates. Position 3 is well downstream
of the base of the waterfall but the acoustic data obtained
there (see Fig. 59) are generally similar to that obtained
at position 1.
5. Four Inch (10 cm) Diameter Transite Pipe (Tests of
February 26-28, 1975)
A 4-in. (10 cm) ID Transite pipe, 44-in. (1.12 m) long,
was installed at the end of the 3-in. (7.6 cm) ID facility
water supply pipe (see Fig. 60). The main purpose of this
test setup is to assess the transmission of flow acoustic
emission through a nonhomogeneous, nonmetallic sewer pipe
wall. Another objective is to examine a low sound loss tech-
nique for attaching transducers to the outer surface of the
nonmetallic conduit. From physical properties data supplied
87
-------
3" diam
PIPE
130 i—
110 -
90 -
70
_l
<
C/J
O
I-
D
O
O
50
f/4 =
200 Hz
1400
1680 = fc/4
1850
2700
a) POSITION 3
110 i—
Fig. 59 Variation of acoustic signal with flow at four
frequencies and two sensor positions (8" sand
cover and monolayer of stone chips within 8" diam
pipe. 4.24" weir at exit of pipe)
88
-------
by the manufacturer (Johns Manville ), it has been determined
that the acoustic properties of Transite are comparable to
concrete used in sewer pipe construction.
The discontinuity plane between the steel and Transite pipes
is of the eccentric sudden enlargement type (flush entry)
in which the roof line of the joined pipe sections is
continuous. The eccentric, crescent-shaped gap between
the pipes is filled by an aluminum plate and sealed by a
Velmix (quick-setting) cement. The step discontinuity is
0.69-in. (1.75 cm) which corresponds to a characteristic
frequency of 21,500 Hz. While the processed recorded data
show acoustic spectral distributions generally similar to
spectra obtained with metallic pipes, the characteristic
frequency is found to be at or just above the upper thresh-
hold of frequency response of the magnetic tape recorder
used. Therefore, the reduced data could not be correlated
reliably with mass flow and no graphic data presentations
are to be made here. This temporary inconvenience was over-
come by purchase of a 6-in. (15 cm) ID Transite pipe sec-
tion as a replacement.
No difficulty was experienced in attaching the accelerome-
ter mounting studs by Devcon "F" (8070. aluminum) epoxy
cement. However, it was noticed that the mounting studs were
somewhat more easily sheared off the Transite pipe than the
metallic pipe because of the tearing of a thin veneer of
epoxy impregnated Transite from the parent material. If a
sensor installed in this way were to remain undisturbed in a
Private communication from R. C. Elliott, Johns Manville
Sales Corporation, Manville, New Jersey (January 1975).
89
-------
buried installation, we would not expect this kind of shear
failure.
6. Six Inch (15.2 cm) Diameter Transite Pipe (Tests of
April 23, 24, 28, May 5, and June 26, 1975)
The second series of tests involving a nonmetallic nonhomo-
geneous conduit material used a 6-in. (15.2 cm) diameter
Transite, 5 foot (1.52 m) long, sewer pipe section. The
3-in. (7.6 cm) diameter steel pipe of the facility water
supply was attached to the Transite pipe by a machined plas-
tic mounting collar and blanking plate (see Fig. 61) so that
the roof line was continuous, and a sudden 3-in. (7.6 cm)
step resulted at the discontinuity plane (see Fig. 11). This
step size corresponds to a characteristic frequency of about
5570 Hz. The joint was sealed at the external surface by
silicone-based caulking. Data were obtained with three ac-
celerometer transducers at .six mounting locations including
one on the bottom centerline of the 3-in. (7.6 cm) steel
pipe and 3-in. (7.6 cm) upstream of the discontinuity
(position 7). Other sensor locations, not monitored during
previous test series, were at the bottom centerline of the
Transite pipe, 7-in. (17.8 cm) upstream of the exit plane
(position 9) and 2.3-in. (5.9 cm) downstream of the dis-
continuity plane (position 1). The Transite pipe was set at
a 3 degree downward slope and the bottom lip of the pipe
was 19.25-in. (49.0 cm) above the floor of the diversion
trough. Test conditions included simulated buried and un-
buried pipe installations, flow rates up to 15 Ib/sec
(6.8 kg/sec), and initial pressurization levels of 12 to
29 psi (0.84 to 2.04 kg/cm2).
90
-------
Fig. 60 Closeup of eccentric connection of 3-inch water
supply pipe to 4-inch diameter transite te&t pipe
Fig. 61 Closeup of special assembly collar used for
eccentric connection of 3-inch pipe to 6-inch
diameter transite test pipe. Flow is from right
to left. Two sensor mounting studs are shown
cemented at top centerline surface of transite
pipe
91
-------
For a special set of runs using this pipe configuration,
laboratory equipment was developed to introduce solid parti-
cles (screened sand) into the water supply at a point 18
feet (5.5 m) upstream from the test section; this equipment
and technique has been described in detail in Section IIIB.
Because of the slot in the 3-in. (7.6 cm) pipe line for
feeding sand, flow rates above 6 Ib/sec (2.7 kg/sec) could
not be run without water spillage from the opening. Typical
results of this test series are presented by Figs. 62 to 65.
7. Field Test Preview (Test of July 1, 1975)
The field test preview was conducted at the Grumman sewage
treatment plant "A" adjacent to the corporation's building
12 in Bethpage, New York. The flow conduits monitored con-
sist of 10-in. (25.4 cm) diameter steel pipes. Because
of the preliminary nature of the tests and the need for
flexibility, we decided to attach the accelerometer sensor
to the outer surface of the pipe by a special magnetic
mounting stud marketed by the transducer supplier. Unfor-
tunately, this mounting technique had not been used in the
previous laboratory tests so no rigorous basis of compara-
tive sound transmission ability was possible. Testing took
place between approximately 10 am and 1 pm. Acoustic mea-
surements were recorded on instrumentation tape over a 30
second time period. The first test location was in an
underground control valve vault shown by Figs. 20 and 21.
Sanitary sewage from several Grumman buildings feed into a
distant wet-well where collection continues until a pump-
activation level is reached. The pump supplies an approxi-
mately constant average flow of 650 gpm (2460 1/m) through
92
-------
i STEEL »_K~
I
7"
POS. 7
— — — — ~ \ DOO Q J
•POS. 9
\"
POS. 9 (UNDER SAND)
POS. 7
(UNDER SAND)
_L
_L
a) DATA AT ALL FREQUENCIES (40 to 22,000 Hz)
110 ,-
-POS.9
\
POS. 7
•POS. 7
(UNDER SAND)
POS. 9
(UNDER SAND)
J I
b) DATA AT fc/4 = 1400' Hz
DIPOLE THEORY TREND
POS. 9
POS. 9
(UNDER SAND)
024
c) DATA ATf/4= 180 Hz
6 8
w, Ib/sec
10
12
POS. 7 (UNDER SAND)
14 15
Fig. 62 Variation of acoustic signal with flow (eccentric
cylindrical pipe array) - transite and steel pipe
elements without sand overburden except where noted
93
-------
3" diam STEEL Wj~
REGION BURIED UNDER
SAND
^
-6" diam TRANSITE
130 r-
110 -
DIPOLE THEORY TREND
\
POS. 3 W/O SAND
POS. 3
-- POS. 4 W/O SAND
POS. 9
<
2
U
in
D
O
O
a) DATA AT fc/4 = 1400 Hz
—-^-_2.-^IL POS.7
130 i-
110 -
J
POS. 4 W/O SAND
POS. 3
POS. 9
b) DATA AT ALL FREQUENCIES (40 to 22,000 Hz)
140 r - POS. 3 W/O SAND
120 -
100
— POS. 4 W/O SAND
POS. 3
DIPOLE THEORY TREND
POS. 7
W, Ib/sec
c) DATA ATf/4 =180 Hz
Fig. 63 Variation of acoustic signal with flow (eccentric
cylindrical pipe array) - transite and steel pipe
elements buried under 8 in. (20.3 cm) of sand,
except where noted
94
-------
140
120 —
100
< 60
a) DATA ATf/4 = 1400 Hz
-0 2
b) DATA ATf/4 = 180 Hz
10
W, Ib/sec
POS. 3
POS. 4
POS. 1
POS. 3
POS. 1
12
14
Fig. 64 Variation of acoustic signal with flow (eccentric
cylindrical pipe array) - transite and steel pipe
elements without sand overburden, except where noted
95
-------
rdiam STEEL
6" diam TRANSITE
DIPOLE THEORY TREND
> I \J\J. *J
a) DATA ATf/4 = 1400 Hz
DIPOLE THEORY TREND
POS.3
60
2 4
b) DATA ATf/4= 180 Hz
Fig. 65 Effect of grit addition to water flow on the acoustic
signal variation with flow - transite and steel
pipe elements, without sand overburden, in eccentric
assembly
96
-------
the 10-in. (25.4 cm) pipe until the level in the wet-well
drops to a cutoff point. The pipe discontinuity accessible
for our measurements was an 8-in. (20.3 cm) plug valve
connected by an 8 by 10 by 6-in. tee fitting to the 10-in0
(25.4 cm) upstream supply pipe, and by an 8 by 10-in. re-
ducer section to the downstream pipe. Several positions on
the reducer, exit pipe, and upstream tee section were examined
(see Fig. 66). Ambient noise measurements were recorded when
no flow was evident; this fact was established by correlation
of the observed level indicated by the tape recorder's moni-
toring meter and visual verification at the exit of the pipe,
about 500 feet (152 m) away, at the aeration tank. The
sources of ambient noises at the value include moderate road
traffic about 500 feet (152 m) to the south and a railroad
about 750 feet (228 m) to the east. One of the two am-
bient noise measurements coincided with passage of an eight-
car passenger train.
The second field site where acoustic emission measurements
were taken was at the point where the 10-in0 (25.4 cm)
pipe emerges from beneath the surface and discharges hori-
zontally into an open, tortuous concrete channel feeding a
comminuter (see Figs. 22 and 23). Close by, a 4-in.
(10.2 cm) diameter steel pipe also discharges into the com-
minuter feed channel except the direction of flow is verti-
cally downward. This smaller pipe conducts an additional
250 gpm (746 1/m), from a pump-equipped wet-well located
in the Grumman factory building just north of the treatment
plant.
97
-------
CONCRETE WALL
,10" diam
,8x10 reducer
8" plug valve
,8x10x6 tee
SENSOR POSITIONS IDENTIFIED
BY NUMBER IN CIRCLE
120 r—
100 —
DIPOLE THEORY TREND
VOLUMETRIC FLOW RATE, q, gpm x 10
,-2
Fig. 66 Variation of acoustic signal with flow at Grumman
waste treatment plant "A" valve vault for
different sensor positions - first test series
98
-------
Four locations on the horizontal leg of the 10-in. (25.4
cm) diameter steel pipe were monitored as shown by Fig. 67.
D. DISCUSSION
The results of this program can be summarized by the state-
ment that no fundamental questions remain concerning the
feasibility of Grumman1s acoustic emission flow measurement
technique applied to discontinuities at enlargements of
conduit cross sections.
The test data have removed each of our initial reservations
about the technique discussed in Section IIIA. The numerous
tests established to demonstrate feasibility were, perhaps,
influenced by the failures of so many more complex tech-
niques when applied to sewer flow, and the contrasting rela-
tive simplicity of our passive nonintrusive method. But
now, after extending the range of flow rate by a factor of
20, using four different pipe geometries or constructions,
employing accelerometer sensors with four different sensi-
tivities, and conducting a field test preview, the acoustic
emission technique remains viable and appropriate for more
exhaustive field testing leading to prototype systems.
To date, only laboratory equipment has been used to monitor
flow acoustic emission and to process the raw data. This
approach has had the advantages of minimizing equipment costs
to the program, and of presenting greater measurement capa-
bility than the basic needs of tests to allow flexibility in
conducting the exploratory research.
The assemblage of many high quality but multipurpose labora-
tory components to obtain acoustic recordings and to feed
99
-------
SENSOR POSITIONS
IDENTIFIED BY NUMBER
IN CIRCLE
SIDE VIEW
140 i—
f = 2200 Hz
f = 3400
80 -
60
DIPOLE
THEORY
TREND
I
3 2
I I
4 6
I I
8 10
I
12
I
14
VOLUMETRIC FLOW RATE, q, gpm x 10'
-2
Fig. 67 Variation of signal with flow at Grumman waste
treatment plant "A" aeration pond for different
frequencies and sensor positions - second test
series
100
-------
these recordings through the Fourier analysis computer, in-
troduces and compounds the inaccuracies and mismatches of
each component in the system. The result of these unwanted
signal distortions, or noise, is that the reproducibility of
the processed signal components in narrow 10 Hz frequency
bands is no better than + 2 dB over the frequency response
of the tape recorders. A 2 dB change is equivalent to a
ratio of sensor signal levels of 1.26; while this seems
small, at high flow rates this change can represent large
absolute quantities of flow. The prospects for improving
this situation are extremely good if less versatile, well-
integrated electronic circuits and special purpose data
processors are used. Now that most of the feasibility
questions have been answered, prototype hardware need only
consist of a simple accelerometer, for attachment to the
sewer pipe, and a communications link to a simple, cheap,
dedicated minicomputer.
While we have not determined the accuracy limits for the acous-
tic data because of the equipment involved, we have estimated
the accuracy of the test facility's flow rate measurement equip-
ment as from + 7% for low flow, short sampling time (e.g.,
10 seconds) conditions to + 4% for high flow, long sampling
time (e.g., 30 seconds or more) situations. These values com-
pare favorably to commercial flow meters for sewage that
often have accuracies worse than + 10%.
The research facility has demonstrated sufficient versatil-
ity for the program, and accommodated various changes of pipe
test configurations with relative ease. The only limitation
we have experienced is an inability to sustain steady flow
at high flow rates because of the limited plant water supply
101
-------
to the pressurization tank. Uniform as well as nonuniform
flow conditions were established during the program although
the latter profile was limited to flow rates less than
6 Ib/sec (2.7 kg/sec). The maximum flow rate of about
15 Ib/sec (6.8 kg/sec) is imposed by the present flow col-
lection and measurement method. This capability could be
doubled by facility modifications if justified by the scope
of future test programs.
The superior acoustic data resulting from use of a high weir
at the pipe exit and with simulated buried installations
demonstrate that special precautions against acoustic reflec-
tions should be taken in laboratory investigations involving
passive acoustic monitoring. Anechoic chambers typify such
special measures. However, our simple inexpensive measures
proved effective in accomplishing almost complete sound isola-
tion. In most field measurements for sewer pipes, this prob-
lem does not exist. At outfalls, except if there are reflect-
ing surfaces surrounding the exit, there should not be signi-
ficant reflected sound to distort the acoustic emission char-
acteristics of the flow.
The test results summarized in Fig. 68 indicate that the
characteristic frequency component of acoustic emission of
flow into an open trapedzoidal channel increases with flow.
No undue precautions have been taken against sound reflec-
tions, but the open channel apparently produces diffuse
radiation that does not reflect back efficiently to the chan-
nel. Although the dipole theory trend is approximated only
at low flows, the signal component is unambiguous and there-
fore useful for flow measurement. For flows above 2.2 lb/
sec (1.0 kg/sec) the 3 dB higher acoustic signature for
102
-------
4.8"
150 i—
DIPOLE THEORY TREND
70
50
fr = 3600 Hz
10
w, Ib/sec
Fig. 68 Typical variation of acoustic signal with flow
for two conduit configurations (cylindrical to
trapezoidal cross-sectional discontinuity)
103
-------
the higher waterfall is not considered trend-setting at this
time, and is more likely peculiar to the test equipment set-"
ups at the two different test periods. Since these results
confirmed and extended Grumman's 1972 exploratory data, fur-
ther investigations with this conduit configuration were sus-
pended in favor of completely closed cylindrical pipe ele-
ments .
The data for the 3-in. (7.6 cm) diameter steel water sup-
ply pipe (see Fig. 28) has 20 to 30 dB lower overall (all
recorded frequencies) sound emission at intermediate to high
flow rates than for the test conduits downstream of a discon-
tinuity. This difference cannot be the result merely of trans-
mission loss through the pipe walls because all pipes used
(i.e., steel, aluminum, and Transite) have theoretical wall
transmission losses within +3 dB. It is concluded, then, that
the greater output of the sensors on the various downstream
test pipes are the result of strong acoustic radiation by pseu-
dosound, dipole type, sources caused by the discontinuity. At
zero flow, peak signal component amplitudes on the 3-in. (7.6
cm) pipe are observed at 80 and 2160 Hz, excited by pre-
sently unidentified sources in the laboratory (see Fig. 29b).
Clearly, the latter frequency duplicates a major excitation
source of the laboratory environment and although the signal
amplitude increases with flow to about 5 Ib/sec (2.27 kg/
sec), it decreases for higher flows. Therefore, this signal
component is ambiguous for flow measurement purposes because
the same signal strength can represent more than one flow
rate value at this frequency. The other peak in th^ spectral
distribution, at the lower (880 Hz) frequency, increases in
amplitude with flow rate at a somewhat slower rate than would
104
-------
• A 1 2
be predicted by dipole theory (i.e., ^ w ) ' at low to in-
termediate flow rates, but becomes more compatible toward the
higher flow rates tested. On the other hand, this frequency
component exhibits poor adherence to the theoretical near-
~ r
field boundary layer variation (~ w ). The 880 Hz fre-
quency is close to that which would be calculated as charac-
teristic on the basis of the 17-in. (43.2 cm) drop from
the pipe to the wooden diversion trough, so it is tempting
to hypothesize the source of this peak signal component as
the flow interaction with the solid surface of the trough.
The monitored signal would be the sound reflected back
through the open end of the 3-in. (7.6 cm) pipe. In any
event, the slight 0.25-in. (0.64 cm) discontinuity caused
by the pipe coupling produces no dipole sound signal within
the 22,000 Hz frequency response limit of our recording
equipment and no useful signal for flow measurement.
Figures 37 and 69 summarize the effect of an extended inlet
on acoustic emission signal at characteristic frequency, for
eccentric and concentric cylindrical pipe element assemblies.
While Fig. 37 data show higher dB levels for the extended
inlet, the useful signal above ambient is almost equal for
both inlets because the zero flow (ambient) signal level is
lower for the flush inlet. On the other hand, this does not
hold true for concentric installations as shown by Fig. 69,
where all signals are normalized to the zero flow level. For
the extended inlet shown in Fig. 69, the sensor position atop
the downstream pipe (5), measures 5 to 7.5 dB greater
levels than at the bottom centerline, (3), but at the same
axial distance from the pipe discontinuity. We interpret
105
-------
FLUSH INLET
A) WITHOUT WEIR
B) WITH WEIR
CO
<
o
CD
(D
in
CJ
t/3
O
CJ
<
Q
LU
N
CC
O
z
DIPOLE THEORY TREND
EXTENDED
INLET
NO. 3
0
2
4
6
w, Ib/sec
8
10
12
Fig. 69 Effect of extended inlet on acoustic signal
variation with flow rate (concentric cylindrical
pipe elements)
106
-------
this as showing greater susceptibility of position 5 to re-
flected sound. As flow increases, the trend is for signals
at both positions to converge. The flush inlet with a
4.24-in. (10.8 cm) weir produces lower signals than the
open pipe exit at the same sensor position but its trend
follows more closely that of dipole theory, to about 6 lb/
sec (2.7 kg/sec) flow. At higher flow rates, the open and
weir-equipped pipes display signal components approaching
equality. Figures 43 and 44 effectively summarize the in-
fluence of laboratory sound reflections on signals obtained
with the eccentric cylindrical pipe array.
With an end-positioned weir and sand burial, a constant
7.5 dB lower signal is monitored at position 4 (see Figo 43)
6-in. (15.2 cm) from the discontinuity. At position 1
(see Fig. 44) the nearly constant difference of 5 dB for
weir-equipped pipes with and without sand overburden is
equivalent to the exposed installation having a sound source
that is 1.8 times greater than the buried pipes, for flows
greater than 1.5 Ib/sec (0.68 kg/sec). At position 4 this
signal difference because of sand overburden also is about
5 dB but only at flows between 2 and 9 Ib/sec (0.9 and
4.09 kg/sec). When the only difference between test condi-
tions is the presence or absence of the end weir, the signal
level with the open ended pipe is 9 db greater than with
the weir, for sensors at both positions 1 and 4. This dif-
ference is equivalent to an apparent ratio of sound source
strengths of about 2.8. Similarly, at position 3 reductions
in signal are caused by overburden or weir attachment (see
Fig. 44) but of somewhat smaller magnitude than the previous
two positions analyzed. Presumably, sensor position 3 has
107
-------
different exposure susceptibility to reflected sound because
of its downstream-most location. The curves for positions 3
and 4 show that the effectiveness of the sand overburden in
preventing sound reflection from affecting the sensor at each
location^ becomes unimportant above 9 Ib/sec (4.09 kg/sec)
flows. This can be explained on the basis that the magnitude
of externally reflected sound becomes 10 dB less than the
internally produced flow interaction signal at flows of about
9.0 Ib/sec (4.09 kg/sec) (e.g., two separate 100 dB
acoustic signal sources produce 106 dB of combined acoustic
signal, but if only one signal increases to 110 dB the
combined signal is only 112 dB). Thus, we conclude that the
exit plane weir and 8-in. (20.3 cm) deep sand cover screen
the accelerometer pickups from being overpowered by flow
sound and other external noises reflected back from the
laboratory walls. When compensation is made for these re-
flected sounds, under laboratory conditions, completely un-
ambiguous signals are obtained for the acoustic emission flow
measurement technique in eccentric cylindrical pipe installa-
tions.
The results of Fig. 50 signify that acoustic signals are
transmitted circumferentially with negligible loss at any
particular axial pipe station. For unburied installations,
the upper sector of the pipe surface probably is more vul-
nerable to externally produced noises. However, for typical
buried pipe installations, acoustic pickups can be attached
at any convenient circumferential position, which simplifies
field instrumentation of existing sewer pipes.
Figures 51 through 55 indicate that the effect of an end weir
plate and simulated buried installation is similar for the
108
-------
concentric arrangement of cylindrical pipe elements as it was
for the eccentric pipe assembly. The highest weir prevents
reflected sound, off the laboratory walls, from reentering
the pipe and augmenting the acoustic signal of flow at the
discontinuity. Because of the downward slope of the 8-in.
(20.3 cm) conduit, the shorter weirs have no practical
shadowing effect and result in data almost identical to the
open ended pipe. When the acoustic reflection is minimized
by pipeline burial and weirs, the data appear to follow
closely the dipole theory prediction curve and are unambigu-
ous with respect to flow rate.
Summary Fig. 70 shows how the pipe configuration changes the
normalized acoustic emission signal at a fixed axial position
downstream from the discontinuity plane. Both curves shown
are for signal components at the characteristic frequency cor-
responding to the disturbance distance of the configuration.
The influence of acoustic reflections is minimized here by
burial under sand and attachment of an end weir. The moni-
toring position selected for data presentation always is
downstream of the base of the waterfall but more so for the
concentric pipes than the eccentric array because of the dif-
ferent discontinuity distances. Both data trends are unam-
biguous and useful for flow measurement although the concen-
tric configuration data follows the dipole theory trend bet-
ter than the eccentric configuration over a larger range of
flows.
Figures 56 and 57 effectively show that stone bed loads do
not thwart the acoustic emission technique. Further analy-
sis of the spectral distributions resulting from the compu-
ter processed recorded sound reveals that a major part of
109
-------
CONCENTRIC CONFIG
60 I—
m
T3
UJ
m
5
o
CD
50
40
dd
DIPOLE THEORY TREND
LJ_
4.25"
T
\ d
VNO.
-«— 9"-^|
I£ - «-5n u.
a
3 SENSOR POSITION
ECCENTRIC CONFIG
1
6
w,Ib/sec
10
12
Fig. 70 Effect of pipe configuration on normalized acoustic
signal variation with flow rate (cylindrical pipe
elements) with 4.24 in. weir, 8 in. sand overburden,
and flush inlet
110
-------
the 7 dB difference in overall signal value between the tv?o
sensor positions, displayed in Fig. 56, is in the low fre-
quency range, below 1200 Hz, rather than in the high range
near the characteristic frequency of 6720 Hz. From Fig. 57
we see that when the tumbling stones of the heavy bed load
are continuously being moved by the force of the water flow
there results a greater sound level, integrated overall fre-
quencies, than for the clean pipe, and a greater variation
of the overall signal at different sensor positions. How-
ever, the spectral discrimination exercised to extract a
signal relatable to flow (i.e., by obtaining signal compo-
nents at characteristic frequencies) makes our technique ef-
fectively insensitive to the consequences of extreme salta-
tion of the bed load; the bulk of the additional sound pro-
duced by rolling stones is in an unused part of the spectral
distribution. While Fig. 57 indicates comparable trends for
the thick layer of stones and the bare conduit, at the char-
acteristic frequency, for a nearly monolayer bed load of
stones the acoustic component is much larger than either of
the other two test conditions at low flow rates. As the
flows increase to intermediate and high rates the thin layer
of stones is displaced just like the thicker, more mobile,
bed load so that the persistent slightly higher acoustic sig-
nal measured must represent additional sound sources near the
discontinuity. One explanation is that the turbulence pro-
duced by the sharp-edged stones scatters pseudosound and con-
verts part of it into intense sound fields which are in a
propagating mode.
One of the pleasant features of passive acoustic monitoring
of flow is that the overall sound quality is good enough to
111
-------
be listened to by humans. Further, the sound of processes as
rolling stones are fully interpretable as such by experienced
personnel listening to the sensor output. This type of in-
formation can be extremely helpful in remote operational
monitoring of the condition of sewers to determine where
maintenance is necessary.
If the thickness of the monolayer bed of stones were to be
considered as reducing the discontinuity distance and acting
like the solid wall of the conduit, the characteristic fre-
quency would change to 7400 Hz from the 6720 Hz for the
clean pipe. But the data of Fig. 59 at 7400 Hz coincides
with that at the frequency for the bare conduit at all but
the lowest flow ranges below 2.2 Ib/sec (1.0 kg/sec).
Where the very low flows do not move the stones, the signal
at 7400 Hz is within 2 dB of the level at the original
6720 Hz. The data trend of Fig. 59 does not exactly match
the 12 dB per doubling of flow rate predicted by dipole
theory. However, the acoustic signals obtained are unambigu-
ous for the range of flows tested, and are therefore useful
for flow measurement in a geometrical setup similar to the
laboratory arrangement.
Summary Fig. 71 shows that the acoustic emission flow mea-
surement method can be used with metallic as well as inhomo-
geneous nonmetallic conduit materials of construction. Al-
though the discontinuity distances of the Transite and alumi-
num pipes are different, we interpret the results of Fig. 71
at two sensor positions as indicating basically similar
trends that are useful for flow measurement in both instances.
The quality of sound transmission through the inhomogeneous
112
-------
3" DIAM STEEL
8" DIAM ALUMINUM
60,—
fc = 6400 Hz
0
2
4
6
w, Ib/sec
8
10
12
Fig. 71 Effect of pipe material on normalized acoustic
signal variation with flow rate (eccentric
cylindrical pipe elements and flush inlet) - 8 in
sand overburden
113
-------
Transite material is very good and because concrete has simi-
lar acoustic properties to Transite, the prognosis for as yet
untested concrete pipe sections is excellent.
In the data display of Fig. 62, the trends for the acoustic
signal components at the characteristic frequency (~ 5600 Hz)
and 720 Hz are shown for the Transite pipe exit monitoring
location (position 9) and the 3-in. (7.6 cm) steel pipe
monitoring location. Also shown is the overall signal for
frequencies between 40 and 22,000 Hz. Since a 4 foot
(1.2 m) long section of the pipe configuration, equally dis-
tributed on either side of the discontinuity plane, was sup-
ported within the wooden box made for simulation of buried
installation, tests were made with and without sand overbur-
den. However, of the two sensor locations plotted, only po-
sition 7 is physically affected by the overburden because
the other position 9 is well downstream of the confines of
the box. From Fig. 62 we see that the sound signal at posi-
tion 9 is greater than on the steel pine. This position on
the Transite pipe has greater exposure to the sound produced
by the interaction of falling water with the diversion trough.
If the 19.25-in. (49 cm) drop to the trough is considered
the exit discontinuity distance, the exit characteristic fre-
quency would be 720 Hz. The difference in signal amplitude
between positions 9 and 7, at intermediate flow rates, is
11 dB at the assumed exit characteristic frequency, whereas
it is 6 dB different for all recorded frequencies, and
18 dB different at the discontinuity characteristic fre<-
quency of 5600 Hz. A flow interaction mechanism to explain
the higher acoustic signal at position 9 seems unsupported
114
-------
by the data. In addition, when the sound produced by the
upstream pipe discontinuity is prevented from radiating nor-
mal to the pipeline axis into the laboratory (and reflecting
back to the sensors) by an 8-in. (20.3 cm) layer of
surrounding sand, the reduced signal amplitude is within
2 dB of being the same at both positions and all frequencies
presented in Fig. 62.
Position 9 is in the far field of the pseudosound source near
1 2
the discontinuity plane„ It is known ' that at these large
distances from their source, pseudosound energy is converted
into real sound which propagates throughout the flow. This
is especially true where turbulence is added to the flow by
sharp edges such as at the exit plane of the pipe or by a bed
load of newly crushed stones (not weathered smooth) as noted
earlier. From these considerations we then attribute the
higher signal near the Transite pipe exit to sound reflected
longitudinally back into the pipe by the nearby laboratory
walls and diversion trough surfaces.
The higher magnitude of near field component signal levels,
at positions closer to the pipe discontinuity plane (e.g.,
positions 3 and 4 compared to positions 7 and 9), is evident
from Fig. 63. About 14 dB of the higher signal at posi-
tion 3 is produced by the higher sensitivity of the acceler-
ometer mounted there. The remaining 3-8 dB enhanced sig-
nal, depending on frequency, at positions 3 and 4 compared
to position 9, undoubtedly is evidence of pseudosound pro-
duction in the near field of the discontinuity plane. Fur-
thermore, since position 4 is further downstream from the
discontinuity, the signal obtained is about 4 dB less than
115
-------
that at position 3, for the characteristic frequency. Other
data in Fig. 64 show that except in the low flow range, the
acoustic signal obtained at position 3 is superior to signals
at the other sensor locations examined, for the characteristic
frequency component. This signal also is better or equal to
that of the other positions at other frequencies. The sound
insulation obtained with simulated buried installations is
evident in both Figs. 62 and 63 at all sensor positions. The
signal variation with flow at position 3 increases faster
than the dipole trend at low flows, and varies slower than
the dipole trend at intermediate to high flows. However,
within the range of flows tested, an unambiguous signal is
produced at the characteristic frequency and is usable for
flow measurement.
Figure 65 shows the effect of adding solids to the flow.
These sand particles, passing through mesh sizes between
0.85 and 2.36 mm, simulate sewage grit and were metered
into the water supply well upstream so as to be well inte-
grated into the flow passing the abrupt cross-sectional
change in the pipe. However, although one can safely presume
that each particle impacts the pipe wall many times, the ag-
gregate effect of such collisions appears inconsequential
relative to the near field acoustic emission of the flow.
The acoustic signal at the characteristic frequency, as well
as at a much lower frequency, is almost exactly the same for
the flow containing grit as for the clean water flow. Also,
despite a wide variation in grit concentration (between 86
and 1682 mg/liter of water) which was introduced for vari-
ous test points, there is no discernible change in the result.
116
-------
In the field test documented by Fig. 66, overall (all fre-
quencies) sound levels of 80 to 84 dB were recorded for
ambient noise, with the higher level occurring during the
train movement„ Overall sound levels of 96-104 dB were
recorded with flow passing through the pipes. The lowest
sound level was recorded at the top of the 10-in. (25.4
cm) pipe, some 27-in. (68.6 cm) downstream of the valve
centerline. The highest overall sound level was measured
at the bottom of the tee section, 18-in. (45.7 cm) up-
stream of the valve centerline. All four locations in the
vicinity of the plug valve have peak amplitude components
at a frequency of 2800 + 80 Hz, as well as at several other
frequencies which are not so universally present. This is an
example of a situation where acoustic properties of the fluid
and dimensional details are not accessible for direct mea-
surement, and a characteristic frequency has to be determined
by inspection of the spectral distribution. Compared to the
ambient sound level at 2800 Hz (see Fig. 66) the sound
level at three of the four locations in the vault are between
17 and 20 dB louder when flow is present. At the fourth
location near the top centerline of the upstream tee section,
the sound level at 2800 Hz is 14 dB louder when the flow
is present than the ambient recorded near the top of the
10-in. (25.4 cm) downstream pipe [45-in. (1.14 m) awayj.
Comparison with a dipole theory trend drawn through the
650 gpm (2460 liter/min) data point shows reasonably good
agreement with a linearized extrapolation of the trend
through the ambient data point.
At the second field test site, the largest acoustic signal
change with increasing flow (see Fig. 67) is for the com-
117
-------
ponent at a frequency of 2000 to 2200 Hz. Unfortunately,
since the wet-well pumps produce constant volumetric flow
rate, the acoustic data are at only two rates and full trends
can only be estimated. It should be further noted that the
signal at 900 gpm (3406 liter/min) is produced by flow
discharged into the open channel from two pipe source sizes
and with different flow orientations. At the characteris-
tic frequency of 2200 Hz, selected by inspection of the
spectral distribution, the increase in sound level is
11.5 dB between flows of 650 to 900 gpm (2460 to 3406
liter/min). The exponential law that follows these data for
Q
sound power, P , appears to be PT ~ Q rather than the
/
PT ~ Q relation of dipole theory. At 2200 Hz, the ver-
tical discharge of the 4-in. (10.2 cm) pipe and the re-
verberations by the walls of the open channel appear to be
much greater sources of acoustic emission than the flow dis-
turbance caused by the 90 degree bend in the 10-in.
(25.4 cm) pipe. However, at another empirically selected
frequency, 500 Hz, the increase in sound component ampli-
tude agrees almost exactly with the 5 dB increment pre-
dicted by dipole theory for the ratio of flows for which data
are available.
Perhaps the best explanation for these few data points is
that dipole theory applies strictly to flow changes where
the discontinuity geometry is of a singular nature and the
conduit is extremely long in both directions. When two dis-
similar discontinuities and sound reflections contribute to
the overall sound being monitored, as is the case in the
second of our field tests, the complex sound field is not
properly described by simple dipole theory; further empiri-
cal investigation is needed for these situations.
118
-------
SECTION IV
REFERENCES
1. Ffowes-Williams, J., "Hydrodynamic Noise," in Annua1
Reviews of Fluid Mechanics, Vol. 1, Annual Reviews,
Inc., Palo Alto, California, 1969.
2. Blokhintsev, D. I., "Acoustics of a Non-Homogeneous
Moving Medium," NACA TM 1399, February 1956.
3. Olson, H., Acoustic Engineering, D. Van Nostrand Co.,
Princeton, 1957.
4. Beranek, L. L., ed., Interaction of Sound Waves with
Solid Structures, Chapter 11, "Noise and Vibration
Control," McGraw-Hill Book Co., New York, 1971.
5. Fair, G. M. and Geyer, J. C., Water Supply and Waste-
Water Disposaj., John Wiley and Sons, New York, 1954.
6. Haddle, G. and Skudrzyk, E., "The Physics of Flow Noise,"
Journal of the Acoustical Society of America, Vol. 46,
No. 1 (Part 2), pp. 130-157, 1969.
119
-------
SECTION V
GLOSSARY
Accelerometer - an electromechanical transducer that gener-
ates an electrical output when subjected to accelera-
tion. Piezoelectric discs clamped between a mass and
base develop a potential field when the acceleration
of the mass exerts a force on the discs. The ratio of
electrical output to mechanical input is called sensi-
tivity.
Acoustic Emission - the radiation of sound generated by the
interaction of fluid flow with a solid surface.
Acoustic Reflection - the change of direction of sound pres-
sure waves impinging on a less than perfect sound ab-
sorbing surface.
Conduit Discontinuity - any change in a flow channel because
of channel cross section or shape, or where flow direc-
tion is significantly changed.
Decibels (dB) - a measure of the ratio of two amounts of
sound power. The range of sound pressure or intensity
is so large that it is more convenient to use the loga-
rithm to the base ten to express this ratio (bel).
Decibel equals one tenth of a bel. When other quanti-
ties (e.g., voltage) are related to the square root of
power, the number, n, of dB are: n = 20 log,0(v/vo),
where v is the referenced quantity.
Dipole - the type of sound source created when a fluid in-
teracts with a solid surface to produce unsteady forces.
120
-------
Because of its oscillating nature, this source is
analogous to two point-sources equal in strength but
opposite in phase and separated by a very small dis-
tance. The radiated power is proportional to the
fourth power of flow speed. Because of the pressure
cancellation in the plane normal to the dipole axis,
the directionality of radiation is strongest along
the dipole axis which is normal to the flow direction.
Hertz (Hz) - an international unit of frequency equal to
the number of cycles per second.
Nonintrusive - not penetrating the fluid flow boundary.
Normalized Acoustic Signal - when transducers of different
sensitivities measure the same sound source, their dB
sound signals are different by the ratio of sensitivi-
ties. Similarly, when a constant sound source signal
is measured against different background noise levels,
the total signals are different by the relative dif-
ference in backgrounds. When using decibel units for
sound level, the irrelevant variables of measurement
such as background noise or sensor sensitivity can be
eliminated by subtracting their dB contribution from
the total signal. The resulting dB level then is the
normalized signal, and is a more valid measure of the
sound source alone.
Overburden - the soil or backfill covering a buried sewer
pipe or flow conduit.
Passive Flow Measurement - a method of determining the mass
or volumetric rate of flow by using energy normally
121
-------
radiated by the fluid flow as opposed to imposing ex-
ternal energy sources or flow energy dissipating de-
vices .
Pseudosound - the pressure pulses produced in locally dis-
turbed fluid flow that have characteristics of sound
in the Rear field but do not propagate into the far
field of the fluid. The radiation pattern of pseudo-
sound is like a dipole sound source.
Sound Power (PT) - the total amount of energy radiated by a
sound source throughout a spherical envelope in a
period of time (watts). In practice, the sound power
level, L , is used to relate the ratio of sound power
to a reference power. By international agreement, this
-12
reference power is 10 watts, and L = 10
log10(PT/10'12), dB.
Unambiguous Signal - a sensor output signal that can be in-
terpreted as relating to only one flow quantity. Over
a continuous range of signal output there are no in-
termediate minima or maxima with regard to the depen-
dent parameter.
122
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-600/2-76-115
3. RECIPIENT'S ACCESSI Of* NO.
4. TITLE AND SUBTITLE
A Passive Flow Measurement System for Storm and
Combined Sewers
5. REPORT DATE
May 1976 (Issuing Date)
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
8. PERFORMING ORGANIZATION REPORT NO.
K. M. Foreman*
9. PERFORMING ORG \NIZATION NAME AND ADDRESS
Grumman Ecosystems Corporation, Bethpage, N.Y. 11714
Through subcontract with:
^Research Dept., Grumman Aerospace Corporation
Bethpage, N.Y. 11714
10. PROORAM ELEMENT NO.
1BC611
11. CONTRACT/GRANT NO.
EPA Contract 68-03-2121
12. SPONSORING AGENCY NAME AND ADDRESS
Municipal Environmental Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
13. TYPE OF REPORT AND PERIOD COVERED
Final Report 6/'74 to 8/'75
14. SPONSORING AGENCY CODE
EPA-ORD
15. SUPPLEMENTARY NOTES
PO: Hugh Masters
16. ABSTRACT.
is investigation concerns a new, nonintrusive, low cost, passive flow measure-
ment method to meet the urgent needs for good management of storm and combined
sewer systems. Operation of the system is based on sensing the near field sound
emitted by the disturbed flow at a channel discontinuity. These local pressure
pulses are called pseudosound and radiate as dipole sound sources orthogonal to
the flow direction. The output signal of passive transducers, such as accelero-
meters, attached to the outer wall of the channel indicates flow rate after
processing by a Fourier Analyzer. Feasibility has been demonstrated by labora-
tory tests using full scale sewer pipe elements, and by a brief series of field
tests measuring sanitary sewage flow. Recommendations are made for further
field site testing using an instrumented sewer line.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
COSATI Field/Group
Flow, Acoustics, Sewers, Flowmeters,
Experimental Design, Acoustic Signatures
Sound Level Meters, Electronic Computer
Electroacoustic transducers
Acoustic Emission
Flowmeter
20C, 20A
13B
14B
17A
9A, 9B
13. DISTRIBUTION STATEMENT
Release Unlimited
19. SECURITY CLASS (This Report)
unclassified
21. NO. OF PAGES
137
20. SECURITY CLASS (This page)
unclassified
22. PRICE
EPA Form 2220-1 (9-73)
123
U. S. GOVERtflHNT PRHffWG BffKE: \376-6S7-k^/^22 Region No. 5-11
------- |