Non-Invasive Flow Measurement Via Distributed Acoustic Sensing Utilizing Frequency

Spectra Analysis of Wall Pressure Fluctuations

Steven M Snider

Thesis submitted to the faculty of the Virginia Polytechnic Institute and State University in

partial fulfillment of the requirements for the degree of

Master of Science

In

Materials Science and Engineering

Gary R. Pickrell- Chair

Daniel S. Homa
Anbo Wang

February 2, 2023
Blacksburg, VA

Keywords: flow measurement, flow temperature, vibration, distributed acoustic sensing, optical
fiber

i
 Non-Invasive Flow Measurement Via Distributed Acoustic Sensing Utilizing Frequency
Spectra Analysis of Wall Pressure Fluctuations

Steven M Snider

ABSTRACT

This research describes a method of using distributed acoustic sensing to noninvasively measure

volumetric flow rate via multiple unique sensor styles. This work modifies previously used

methods of flow detection via fiber optic acoustic sensors affixed onto the exterior body of a

flow apparatus. Flow rate measurement methods for two unique sensor styles are described.

Weak trends are additionally observed as a function of flow temperature that may represent

opportunity for future optimization.

A discussion of current noninvasive flow rate measurement methods is given as well as their

limitations. A background of distributed acoustic sensing is presented along with a summary of

its fundamentals as well as its functionality in noninvasive flow rate measurement. A

description of previous techniques that utilized distributed acoustic sensing in conjunction with

fiber optic acoustic sensing is shown.

The acoustic properties of the fluid-induced vibrations are measured as a function of flow rate

and flow temperature utilizing a special type of fiber optic sensor. Numerically smoothed

frequency domain acoustic peaks are evaluated by intensity, area, central frequency, and full

width at half maximum as flow conditions vary. All tested sensors were found to yield a strong

ii
dependence between peak intensity and flow rate. A dependence between central frequency and

flow temperature was observed in some cases. The sensor system developed was able to

measure fluid-induced vibration intensity and vibrational central frequency and offers potential

uses in a myriad of vibrational applications.

iii
 Non-Invasive Flow Measurement Via Distributed Acoustic Sensing Utilizing Frequency
Spectra Analysis of Wall Pressure Fluctuations

Steven M Snider

General Audience Abstract

This research provides a method of measuring fluid-induced vibrations caused by internal

pressure fluctuations stemming from a variety of flow conditions. In this case, a specially

fabricated optical fiber is applied to the external surface of the pipe. As water flows at a known

volumetric flow rate and temperature, the acoustic signal generated is detected by the optical

sensor signal demodulation system. The fiber used is a silicate material designed to transmit

optical signals over long distances with minimal loss. Modifications to the fiber can be made to

differentiate the measured optical signal loss by frequency band, as well as to designate the

spatial position on a fiber sensor to locate where loss is occurring. By measuring optical loss of

distinct fiber spatial positions at high sampling frequencies, an abundance of sensing

opportunities become available. In knowing optical signal travel time of select wavelengths to

corresponding strain characteristics amongst a section of fiber, optoelectronic devices with

strong computing power called interrogators can powerfully measure the intensity and rate of

fiber strain at a significantly high sampling frequency.

Fiber optic sensors have been used in many areas where monitoring of changes in positional

microstrain is desired. Such sensors are embedded in-ground for seismic monitoring, as well as

on the ocean floor for submarine structural characterization with long singular fibers. Flow rate

measurement is performed with fiber coils and various other geometries for active oil wells,

iv
fission reactors, and other areas. Improving the performance and applicational flexibility of

these sensors allows for greater opportunity for scientific advancement in an array of fields.

This research was completed to offer a new method of flow rate measurement while also

gauging if flow temperature was able to be measured via a single fiber optic sensor. Fiber strain

was observed to be strongly dependent on flow rate, whereas the rate at which strain occurred

suggests simultaneous flow and temperature measurement is possible in certain types of fiber

arrangements. The work produced in this research is a step towards singular-fiber flow rate and

temperature sensing.

v
 Acknowledgements

I would like to thank Dr. Gary Pickrell for his continued support in carrying out this work. I am

thankful that he selected me as a graduate research assistant years ago and I am grateful for his

expertise in the field of photonics. Having Dr. Pickrell’s knowledge to lean on when times

became challenging was of the utmost importance towards completion of this work.

I would also like to thank Dr. Dan Homa for his consistent guidance and support in this work.

Dr. Homa’s skill with distributed acoustic sensing technology proved to be beneficial many

times throughout the timeline of this work and its related projects. The experience and wit Dr.

Homa would frequently utilize in assisting with challenges towards this work were both

paramount towards its completion.

I would also like to thank Dr. Anbo Wang for his leadership role in the research through both of

his roles at Virginia Tech as well as at Sentek Instrument. The research completed in this work

would not have been possible without his prior achievements in the field or the experimental

foundation that equipment from Sentek Instrument permitted.

vi
 Table of Contents

Abstract............................................................................................................................................ii

General Audience Abstract..............................................................................................................iv

Acknowledgements.........................................................................................................................vi

Table of Contents...........................................................................................................................vii

Table of Figures..............................................................................................................................ix

List of Tables.................................................................................................................................xiv

Introduction......................................................................................................................................1

Background......................................................................................................................................6

1 Fundamentals of DAS .................................................................................................................... 6

2 Modal Analysis .............................................................................................................................10
3 Flow Characteristics .................................................................................................................... 14
Materials and Methods ...................................................................................................................18

4 Experimental Design ................................................................................................................... 18

4.1 Sensor Design and Application ........................................................................................... 19
4.2 Pipe System Design ............................................................................................................ 24
4.3 Interrogation ...................................................................................................................... 26
5 Modal Analysis of Pipe System ................................................................................................... 28
5.1 Numerical Modal Analysis .................................................................................................. 28
5.2 Experimental Modal Analysis ............................................................................................. 29
Results and Discussion .................................................................................................................. 31

6 Data Post-Processing .................................................................................................................. 31

vii
 7 Peak Intensity Relationships........................................................................................................ 32
7.1 Flow Rate ............................................................................................................................. 32
7.2 Flow Temperature ............................................................................................................... 43
8 Peak Area Relationships .............................................................................................................. 49
8.1 Flow Rate ............................................................................................................................. 49
8.2 Flow Temperature ............................................................................................................... 60
9 Peak Central Frequency Relationships ........................................................................................ 66
9.1 Flow Rate ............................................................................................................................. 66
9.2 Flow Temperature ............................................................................................................... 71
10 Peak Full Width at Half Maximum Relationships ........................................................................... 78
10.1 Flow Rate ............................................................................................................................. 78
10.2 Flow Temperature ............................................................................................................... 82
11 Discussion ................................................................................................................................... 88
Conclusions ...................................................................................................................................91

Future Work ...................................................................................................................................94

References .....................................................................................................................................95

viii
 Table of Figures

Figure 1-1: Principle of backscatter measurement in optical time domain reflectometry…….......................................... 9

Figure 3-1: Two-dimensional velocity profile of laminar, incompressible flow between two rigid surfaces………............14

Figure 3-2: Two-dimensional time-averaged velocity profile turbulent flow between two rigid surfaces……………..…... 16

Figure 4-1: Image of a coil sensor wrapped around the exterior of the pipe system...................................................... 20

Figure 4-2: Image of aerogel pad with optical fault detector identifying embedded fiber channels and gratings (left)
and without optical fault detector (right)...................................................................................................................... 21

Figure 4-3: Image of fiberglass pad with optical fault detector identifying embedded fiber channels and gratings (left)
and without optical fault detector (right)...................................................................................................................... 22

Figure 4-4: Image of aerogel pad affixed to the pipe system from a front view (left) and a side view (right)................ 23

Figure 4-5: Image of fiberglass pad affixed to the pipe system from a front view (left) and a side view (right)……....... 23

Figure 4-6: Image of the pipe system............................................................................................................................. 25

Figure 4-7: Image of tank, heater, and flow control valves from a front view (left) and a side view showing the water’s
tank exit (right).............................................................................................................................................................. 25

Figure 4-8: DASnova-02 Fiber Optic DAS System used through experimentation........................................................... 27

Figure 4-9: Sample real measurement data from the software control panel of the interrogator. [98]…...................... 27

Figure 5-1: Image of the bonded accelerometer.…......................................................................................................... 29

Figure 5-2: Sample real measurement data from the software control panel of the interrogator….............................. 30

Figure 7-1: Maximum FIV peak intensity vs flow rate at 40°C with quadratic fit line measured via the downstream coil
……………………………………………………………………………………………………………………………………………………............................. 32

Figure 7-2: Maximum FIV peak intensity vs flow rate at 60°C with quadratic fit line measured via the downstream
coil................................................................................................................................................................................. 33

Figure 7-3: Maximum FIV peak intensity vs flow rate at 60°C with quadratic fit line measured via the elbow coil....... 33

Figure 7-4: Maximum FIV peak intensity vs flow rate at 60°C with quadratic fit line measured via the elbow coil….... 34

ix
Figure 7-5: Maximum FIV peak intensity vs flow rate at 40°C with quadratic fit line measured via the fiberglass sensing
pad................................................................................................................................................................................ 36

Figure 7-6: Maximum FIV peak intensity vs flow rate at 60°C with quadratic fit line measured via the fiberglass sensing
pad…............................................................................................................................................................................. 37

Figure 7-7: Maximum FIV peak intensity vs flow rate at 40°C with quadratic fit line measured via the aerogel sensing
pad, location 1............................................................................................................................................................... 37

Figure 7-8: Maximum FIV peak intensity vs flow rate at 60°C with quadratic fit line measured via the aerogel sensing
pad, location 1…............................................................................................................................................................ 38

Figure 7-9: Maximum FIV peak intensity vs flow rate at 40°C with quadratic fit line measured via the aerogel sensing
pad, location 2…............................................................................................................................................................ 38

Figure 7-10: Maximum FIV peak intensity vs flow rate at 60°C with quadratic fit line measured via the aerogel sensing
pad, location 2…............................................................................................................................................................ 39

Figure 7-11: Maximum FIV peak intensity vs flow rate at 40°C with quadratic fit line measured via the aerogel sensing
pad, location 3............................................................................................................................................................... 39

Figure 7-12: Maximum FIV peak intensity vs flow rate at 60°C with quadratic fit line measured via the aerogel sensing
pad, location 3…............................................................................................................................................................ 40

Figure 7-13: Maximum FIV peak intensity vs flow temperature at 15 l/min measured via the downstream coil.......... 43

Figure 7-14: Maximum FIV peak intensity vs flow temperature at 50 l/min measured via the downstream coil.......... 44

Figure 7-15: Maximum FIV peak intensity vs flow temperature at 15 l/min measured via the elbow coil….................. 44

Figure 7-16: Maximum FIV peak intensity vs flow temperature at 50 l/min measured via the elbow coil……............... 45

Figure 7-17: Maximum FIV peak intensity vs flow temperature at 15 l/min measured via the fiberglass sensing pad.
………………………………………………………………………………………………………………………………………………………………………….……. 46

Figure 7-18: Maximum FIV peak intensity vs flow temperature at 50 l/min measured via the fiberglass sensing pad.
...................................................................................................................................................................................... 46

Figure 7-19: Maximum FIV peak intensity vs flow temperature at 15 l/min measured via the aerogel sensing pad,
location 1..…………………………………………………………………………………………………………………………………………………………...... 47

Figure 7-20: Maximum FIV peak intensity vs flow temperature at 50 l/min measured via the aerogel sensing pad,
location 1……................................................................................................................................................................. 48

x
Figure 7-21: Maximum FIV peak intensity vs flow temperature at 15 l/min measured via the aerogel sensing pad,
location 2….................................................................................................................................................................... 48

Figure 7-22: Maximum FIV peak intensity vs flow temperature at 50 l/min measured via the aerogel sensing pad,
location 2....................................................................................................................................................................... 49

Figure 8-1: Peak area vs flow rate at 40°C with quadratic fit line measured via the downstream coil.......................... 50

Figure 8-2: Peak area vs flow rate at 60°C with quadratic fit line measured via the downstream coil…........................ 50

Figure 8-3: Peak area vs flow rate at 40°C with quadratic fit line measured via the elbow coil..…………........................ 51

Figure 8-4: Peak area vs flow rate at 60°C with quadratic fit line measured via the elbow coil..................................... 51

Figure 8-5: Peak area vs flow rate at 40°C with quadratic fit line measured via the fiberglass sensing pad…............... 54

Figure 8-6: Peak area vs flow rate at 60°C with quadratic fit line measured via the fiberglass sensing pad.….............. 54

Figure 8-7: Peak area vs flow rate at 40°C with quadratic fit line measured via the aerogel sensing pad, location 1.
...................................................................................................................................................................................... 55

Figure 8-8: Peak area vs flow rate at 60°C with quadratic fit line measured via the aerogel sensing pad, location 1.
...................................................................................................................................................................................... 55

Figure 8-9: Peak area vs flow rate at 40°C with quadratic fit line measured via the aerogel sensing pad, location 2.
…………………………………………………………………………………………………………………………………………………………………………………56

Figure 8-10: Peak area vs flow rate at 60°C with quadratic fit line measured via the aerogel sensing pad, location 2.
...................................................................................................................................................................................... 56

Figure 8-11: Peak area vs flow rate at 40°C with quadratic fit line measured via the aerogel sensing pad, location 3.
...................................................................................................................................................................................... 57

Figure 8-12: Peak area vs flow rate at 60°C with quadratic fit line measured via the aerogel sensing pad, location 3.
...................................................................................................................................................................................... 57

Figure 8-13: Peak area vs flow temperature at 15 l/min measured via the downstream coil........................................ 60

Figure 8-14: Peak area vs flow temperature at 50 l/min measured via the downstream coil........................................ 61

Figure 8-15: Peak area vs flow temperature at 15 l/min measured via the elbow coil.................................................. 61

Figure 8-16: Peak area vs flow temperature at 50 l/min measured via the elbow coil.................................................. 62

xi
Figure 8-17: Peak area vs flow temperature at 15 l/min measured via the fiberglass sensing pad……………..…………..... 63

Figure 8-18: Peak area vs flow temperature at 50 l/min measured via the fiberglass sensing pad……………..…………..... 63

Figure 8-19: Peak area vs flow temperature at 15 l/min measured via the aerogel sensing pad, location 1................. 64

Figure 8-20: Peak area vs flow temperature at 50 l/min measured via the aerogel sensing pad, location 1……..…….... 65

Figure 8-21: Peak area vs flow temperature at 15 l/min measured via the aerogel sensing pad, location 2................. 65

Figure 8-22: Peak area vs flow temperature at 50 l/min measured via the aerogel sensing pad, location 2................. 66

Figure 9-1: Peak central frequency vs flow rate at 40°C measured via the downstream coil......................................... 67

Figure 9-2: Peak central frequency vs flow rate at 40°C measured via the elbow coil................................................... 67

Figure 9-3: Peak central frequency vs flow rate at 40°C measured via the fiberglass sensing pad................................ 68

Figure 9-4: Peak central frequency vs flow rate at 40°C measured via the aerogel sensing pad, location 1................. 69

Figure 9-5: Peak central frequency vs flow rate at 40°C measured via the aerogel sensing pad, location 2……….…..... 69

Figure 9-6: Peak central frequency vs flow rate at 40°C measured via the aerogel sensing pad, location 3................. 70

Figure 9-7: Measured unsmoothed frequency domain data for the aerogel pad in location 1 at a constant 40°C at 15
l/min (top), 30 l/min (middle), and 45 l/min (bottom)................................................................................................... 71

Figure 9-8: Peak central frequency vs flow temperature at 15 l/min measured via the downstream coil..................... 72

Figure 9-9: Peak central frequency vs flow temperature at 50 l/min measured via the downstream coil..................... 72

Figure 9-10: Peak central frequency vs flow temperature at 15 l/min measured via the elbow coil……………..………...... 73

Figure 9-11: Peak central frequency vs flow temperature at 50 l/min measured via the elbow coil………………..……...... 73

Figure 9-12: Measured unsmoothed frequency domain data for the downstream coil at a constant 50 l/min at 15°C
(top), 35°C (middle), and 55°C (bottom)........................................................................................................................ 74

Figure 9-13: Peak central frequency vs flow temperature at 15 l/min measured via the fiberglass sensing pad.......... 75

Figure 9-14: Peak central frequency vs flow temperature at 50 l/min measured via the fiberglass sensing pad.......... 75

Figure 9-15: Peak central frequency vs flow temperature at 15 l/min measured via the aerogel sensing pad, location 1
….................................................................................................................................................................................... 76

xii
Figure 9-16: Peak central frequency vs flow temperature at 50 l/min measured via the aerogel sensing pad, location 1
……………………………………………………………………………………………………………………………………………………………………………..... 77

Figure 9-17: Peak central frequency vs flow temperature at 15 l/min measured via the aerogel sensing pad, location 2
..…………………………………………………………………………………………………………………………………………………………………………..... 77

Figure 9-18: Peak central frequency vs flow temperature at 50 l/min measured via the aerogel sensing pad, location 2
….................................................................................................................................................................................... 78

Figure 10-1: Peak FWHM vs flow rate at 40°C measured via the downstream coil…………………………………………….…..... 79

Figure 10-2: Peak FWHM vs flow rate at 40°C measured via the elbow coil.................................................................. 79

Figure 10-3: Peak FWHM vs flow rate at 40°C measured via the fiberglass sensing pad.............................................. 80

Figure 10-4: Peak FWHM vs flow rate at 40°C measured via the aerogel sensing pad, location 1................................ 81

Figure 10-5: Peak FWHM vs flow rate at 40°C measured via the aerogel sensing pad, location 2…………………………..... 81

Figure 10-6: Peak FWHM vs flow rate at 40°C measured via the aerogel sensing pad, location 3................................ 82

Figure 10-7: Peak FWHM vs flow temperature at 15 l/min measured via the downstream coil................................... 83

Figure 10-8: Peak FWHM vs flow temperature at 50 l/min measured via the downstream coil……………………….…….... 83

Figure 10-9: Peak FWHM vs flow temperature at 15 l/min measured via the elbow coil………………………………….…….... 84

Figure 10-10: Peak FWHM vs flow temperature at 50 l/min measured via the elbow coil............................................ 84

Figure 10-11: Peak FWHM vs flow temperature at 15 l/min measured via the fiberglass sensing pad......................... 85

Figure 10-12: Peak FWHM vs flow temperature at 50 l/min measured via the fiberglass sensing pad…………………..... 85

Figure 10-13: Peak FWHM vs flow temperature at 15 l/min measured via the aerogel sensing pad, location 1.......... 86

Figure 10-14: Peak FWHM vs flow temperature at 50 l/min measured via the aerogel sensing pad, location 1……..... 86

Figure 10-15: Peak FWHM vs flow temperature at 15 l/min measured via the aerogel sensing pad, location 2……..... 87

Figure 10-16: Peak FWHM vs flow temperature at 50 l/min measured via the aerogel sensing pad, location 2……..... 87

xiii
 List of Tables

Table 5-1: Properties and dimensions of the tested pipe............................................................................................... 28

Table 7-1: Downstream coil peak intensity R² and standard deviation across tested temperature array..................... 35

Table 7-2: Elbow coil peak intensity R² and standard deviation across tested temperature array................................ 35

Table 7-3: Fiberglass sensing pad peak intensity R² and standard deviation across tested temperature array............ 41

Table 7-4: Aerogel sensing pad peak intensity R² and standard deviation across tested temperature array, location 1.
……………………………………………………………………………………………………………………………………………………………………............. 41

Table 7-5: Aerogel sensing pad peak intensity R² and standard deviation across tested temperature array, location 2.
...................................................................................................................................................................................... 41

Table 7-6: Aerogel sensing pad peak intensity R² and standard deviation across tested temperature array, location 3.
...................................................................................................................................................................................... 41

Table 8-1: Downstream coil peak area R² and standard deviation across tested temperature array………................... 52

Table 8-2: Elbow coil peak area R² and standard deviation across tested temperature array….................................... 52

Table 8-3: Fiberglass sensing pad peak area R² and standard deviation across tested temperature array.................. 58

Table 8-4: Aerogel sensing pad peak area R² and standard deviation across tested temperature array, location 1.
...................................................................................................................................................................................... 58

Table 8-5: Aerogel sensing pad peak area R² and standard deviation across tested temperature array, location 2.
...................................................................................................................................................................................... 59

Table 8-6: Aerogel sensing pad peak area R² and standard deviation across tested temperature array, location 3.
...................................................................................................................................................................................... 59

xiv
 Introduction

Fused silica optical fiber is commonly used in transmitting light long over distances between the

ends of fiber, the distance for which is limited by the optical attenuation. Fibers have been

developed with an ultra-low optical loss, which occurs at narrow wavelength windows around

1.3 and 1.55 µm [1]. These two high-transmission regions exist due to having minimal

absorption stemming from the OH absorption band centered at 1.37 µm, absorbing less light

from Rayleigh scattering than shorter wavelengths, and being located just before multiphonon

vibration absorption at longer wavelengths [2]. From this phenomenon, optical fiber has been

developed into many different types of sensors. One such sensor is a fiber optic interferometer, a

sensor that divides an originating optical signal over multiple unequal paths. When the resulting

signals are reunited and detected, the resulting interference pattern is measured. Some prominent

interferometer sensors styles include Fabry-Perot, Mach-Zehnder, Michelson, and Sagnac, each

style uniquely measuring the core concept of interference patterns between multiple signal beams

[3]. Fiber optic sensors are used to measure strain [4], temperature [4, 5], pressure [6], rotation

[7], refractive index [5], and other properties. The resultant signal can be based on intensity [8],

phase shift [9], polarization [10, 11], and other qualities.

In order to create uniquely sensed sections in the fiber, a fiber Bragg grating (FBG) is created by

inscribing reflectors of predetermined refractive index changes into the fiber at predetermined

length intervals. These intervals and refractive index changes create a unique band-stop filter that

reflects wavelengths of a desired frequency range from entering the next section of fiber, while

allowing light outside this frequency range to pass. In creating an FBG in the fiber, sections

between reflectors are independently sensed as instantaneous optical loss between reflectors is

1
determined. One such type of fiber optic sensor is distributed acoustic sensing (DAS), which

utilizes instantaneous scattering measurements within the fiber as a means to measure

instantaneous strain variations at sections throughout the fiber.

Developments in high sensitivity fiber optic sensing have led to the technology obtaining many

applicational upsides, such as capability for passive monitoring, high sensitivity, flexibility of

function, immunity to electromagnetic noise, and resistance to extreme environments. One

application that requires these qualities is a noninvasive sensor that is sensitive to flow and

temperature. Nuclear reactors safety systems implemented in sodium-cooled fast reactors

(SFRs), lead-cooled fast reactors (LFRs), and liquid metal-cooled reactors (LMRs) require fully

passive and noninvasive flowmeter that can withstand toxic, extreme temperature, and extremely

reactive environments [12]. Similar requirements are necessary in downhole well flow

monitoring, along with the ability to survive in high-pressure environments and service life of at

least 20 years [13]. For these reasons and more, it is important to develop a high sensitivity

noninvasive flowmeter that can withstand extreme environments with a long lifespan.

One method used by noninvasive flow sensors is to measure the relationship between flow rate

and fluid-induced vibrations (FIVs). After FIVs were first reported [14], multiple numerical

dynamic analyses of the phenomenon were performed [15-19]. Early FIV-detecting sensor

systems utilized a piezoelectric (PZT) accelerometer to measure the quadratically related

standard deviation of pressure fluctuations and accelerometer noise with flow rate [20, 21].

Methods to calculate flow rate and velocity profile have also been reported using one or a

number of ultrasonic transducers and Doppler effect measurements [22-25]. Coriolis flowmeters

have also been developed as a means of externally measuring flow via the Coriolis acceleration

2
stemming from some electrostatic, acoustic, piezoelectric, or some other originating force [26-

29]. Flow measurement and vibration detection have been reported with standard piezoelectric

devices [30, 31], as well as via the use of piezoelectric PVDF sensors or tubes [32-34]. While

these sensors offer a variety of noninvasive methods to measure flow, they are subject to

electromagnetic interference and lack the ability to serve for long lifecycles in extreme

environments.

Optical fiber sensing systems are relatively inexpensive in comparison to other sensing systems

that can passively function in extreme environments [35]. Such areas that benefit from optical

fiber-based sensors include nuclear core sensing systems [35], pipeline monitoring [36-38], and

oil and gas exploration [39]. Additionally due to the high sensitivity of these systems, use of

unique acoustic sensing systems has expanded to ocean observation applications [40], structural

integrity assessment of rigid bodies [41-46], and geological seismic surveying [47-50]. As a

means of using fiber optic sensors to measure flow via FIVs, a number of approaches have been

reported. An optical time-domain reflectometry (OTDR) system measuring distributed vibrations

has been reported [51], however DAS is typically preferred to OTDR systems due to higher

sensitivity and lower fiber length requirements. A preliminary fully fiber optic FBG

accelerometer was reported [52] measuring the previously known standard deviation of radial

FIVs and flow rate, as well as the relationship between flow rate and natural frequency,

determined in this system to shift lower from 49.47 Hz to 49.17 Hz as flow rate increased from

2.19 m3 /hr to 5.67 m3 /hr. However, the accelerometer’s accuracy, signal filtering, and general

optimization was self-reported as needing more study. One recent noninvasive optical fiber

flowmeter design [53] used an optical fiber DAS system to record the standard deviation of

3
pressure fluctuations from FIVs vs. flow rate, however this design does not report temperature

variance of the flowing liquid affecting standard deviation measurements of the pressure

fluctuations. Another recent optical fiber flowmeter design [54] uses two FBGs at the elbow of

the pipe – one on the internal radius and one on the external radius – to measure the difference in

strain between the FBGs located at the outer and inner radii as flow velocity increases. While

this method eliminates the effect of thermal strain on the wavelength measurements, systems

using differentials of two points contain higher error potential as the standard deviation of a flow

case increases.

The natural frequency is defined as the frequency at which a system oscillates at when the system

is disturbed. When a system is exposed to external forces of a frequency that excites the system’s

natural frequency, resonance will cause vibrations to occur at the system’s natural frequency

[55]. This set of vibrations is particularly relevant in sensing FIVs in a pipe, as the natural

frequency of the pipe is the central frequency of the sensed FIVs when a time-domain signal is

transformed into the frequency-domain. The natural frequency of a pipe has been modeled and

experimentally validated amidst varying pipe conditions [56]. By determining the natural

frequency of a given pipe system, one could use a noninvasive flowmeter to sense the FIVs at a

waveband approximately centered at the system’s natural frequency.

This research presents a noninvasive flow measurement technique utilizing an optical fiber DAS

system in which the intensity of sensed FIVs at the pipe’s natural frequency is measured as flow

rate and flow temperature change. The central frequency of the peak, full-width half maximum

(FWHM), and peak area are also calculated as these conditions vary. Flow characteristics and

4
FIV intensity analysis as a function of adjacency to the inflowing pipe elbow are also

investigated. The relationship between sensitivity and fabricated sensor design is also discussed.

5
 Background

1. Fundamentals of DAS
Measurements within the fiber can occur in several ways within a DAS system. Original DAS

systems used the Rayleigh backscatter from an optical fiber received by an optoelectronic device,

and the signal was then processed into strain intensity from acoustic disturbances. In this

instance, Raman [57] and Brillouin [58] scattering, as well as Rayleigh scattering that is not

directly backscattered, also occurs in lesser amounts during acoustic sensing. Of these types of

scattering, Rayleigh scattering is defined as linear because the scattered light remains at the same

frequency, whereas nonlinear Raman and Brillouin scattering are characterized by light scattered

at varied frequencies than the originating signal. Raman scattering is the result of high-frequency

optical phonons causing thermal molecular vibrations within the fiber, resulting in a wavelength

shift corresponding to the angle of scattered photons. Brillouin scattering is much like Raman

scattering, only the incident particle is an acoustic, low-frequency phonon causing a photon to

scatter at some smaller angle than Raman scattering, causing a smaller wavelength shift and

typically a weaker power of scattering [59]. In contrast to Raman and Brillouin scattering,

Rayleigh scattering is caused by inhomogeneities within the fiber that have distinct refractive

indices from the core [59]. While it is possible to use DAS by measuring Brillouin or Raman

scattering, typically Rayleigh scattering was the primary scattering choice as it easier to design

sensing systems around due to its predictability, as well as its ability to be measured in weaker

incident pulses.

6
Modern distributed acoustic sensing is characterized as a system that uses a sensing instrument to

measure optical qualities over a desired length of optical fiber. DAS systems utilize optical fibers

with unique grating spacing to independently measure optical qualities along these divided

sections, thus spatially dispersed information in the fiber distinguishes the system as distributed

[60]. In these sensing systems, the term acoustics refers to a broad scope of forces that

encompasses any propagation of mechanical disturbances, such as surface vibrations [61]. The

ability to independently measure the propagation of mechanical disturbances along a desired

fiber length originates from the creation of fiber-Bragg gratings (FBGs). An FBG operates by

reflecting narrow signal bands centered at the Bragg wavelength of each successive grating

reflector. The transmitted signal is defined as the portion of the input signal that passes through

all reflectors of the FBG, whereas the reflected signal is the portion of the input signal that is

reflected from the FBGs. Both signals are capable of use in DAS systems. The reflectors in an

FBG are created by varying an ultrashort length of the fiber core’s refractive index in relation to

the rest of the fiber. The refractive index of the core can be modified to reflect or transmit target

wavelength bands by manipulating the spacing, depth, and/or the shape of fiber inscriptions.

These reflectors are typically created by high-intensity ultraviolet radiation lasers [62, 63] with

periodicity equal to the grating’s desired periodicity [64]. The central wavelength of a signal

reflected is defined as the Bragg wavelength, λ𝐵 , and is given by:

𝜆𝐵 = 2𝑛𝛬, (1)

where n is the refractive index of the reflector and Λ is the grating period [64]. Oftentimes the

grating period between reflectors is uniform and reflects a portion of the input signal at a specific

7
wavelength. However, linearly varied grating periods create what is defined as a chirped grating,

often used to reflect multiple optical wavelengths. Each fiber section between these reflectors is

designated as a sensor length that independently records an optical response to a stimulus laser

pulse propagating throughout the fiber.

In order to reflect light of a desired frequency, inhomogeneities can be purposely fabricated

within the fiber at desired angles, typically 90º, to maximize backscattering to an optoelectronic

reception device. The time duration of a waveband portion of the input pulse can be measured

and processed into strain depending on the refractive index of the individual Bragg reflectors.

After initializing an optical fiber sensor with a known fiber length, an optoelectronic device can

receive backscattering reflected from each FBG and calculate individual strain from the phase

difference between each sensor section [65]. The relationship between strain and photon velocity

has been widely studied [66-68], and for plane wave incidence strain can be calculated as,

1
𝜀𝑥𝑥 = − 𝑐 𝑢̇ 𝑥 , (2)

where 𝜀𝑥𝑥 is strain, 𝑢̇ 𝑥 is particle velocity, and 𝑐 is apparent phase velocity along the cable axial

direction 𝑥⃑ (distance from the interrogation unit increases) [65]. This strain is calculated by an

interrogator that receives reflected Rayleigh backscatter from FBGs at the scale of the preceding

laser input. This entire process of measuring phase-altered backscattered light is known as

optical time domain reflectometry (OTDR). In OTDR, the backscattered light is received by a

photodiode in tandem with an integrator that digitizes the incoming optical array and averages a

set of incoming measurements to improve the signal-to-noise ratio (SNR). Averaging to

maximize the SNR in DAS is typically carried out via weighted spatial averaging, a method that

8
weighs received optical signal as inversely proportional to the mean square noise of its

corresponding measurement [69]. In addition, the pulse-to-pulse response between two Bragg

reflectors when considering the light speed is measuring the fiber elongation through a

nanostrain per second scale, as the light speed and frequency of pulses between two fiber

sections allows ample measurements for not only a high SNR, but a high precision as well via

instantaneous and constant integration. Nevertheless, this averaged signal is then sent to a

logarithmic amplifier so that averaged measurements at the fiber sensor’s length between

gratings are plotted on a chart recorder. The resulting chart gives location-dependent values

regarding where attenuation values are the highest in the fiber. In other words, the resulting plot

shows the attenuation per unit length of the optical fiber sensor, created from the slope of the plot

of the reflected power over the time domain [70]. Figure 1-1 below shows a block schematic of a

backscatter measurement that is typical for optical time domain reflectometry.

Figure 1-1: Principle of backscatter measurement in optical time domain reflectometry.

Using the above OTDR measurement method, DAS techniques achieve high robustness against

laser noise and high sensitivity [60]. Additional applicational upsides to using OTDR within

9
DAS are singular channel data acquisition, short calibration times, and quick detection of failure

points.

Several factors contribute to fiber optic strain via FIVs when measured by DAS. Interfacial bond

strength of the optical fiber sensor to the pipe directly correlates to accuracy and precision of

FIV-induced strain. The relationship between sensor adjacency to a 90º elbow and signal

intensity is measured and discussed within this paper. Due to the high sensitivity and scale of

strain for optical fiber-based sensing systems, several noise factors contribute to the observed

strain measurements. Inline metallic ball valves and similar metallic pipe fixtures may alter the

central frequency of FIVs. Inline throttling valves may contribute damping to the calculated

strain intensity caused by FIVs. All such noise was minimized during the experiments within this

work.

2. Modal Analysis
Each object has a unique set of mode shapes, qualities that can be described as inherent

dynamic characteristics that intrinsically determine a structure’s dynamic behavior. Mode

shapes for a system can be modeled with a series of eigenvectors representing stiffness,

inertia, support, etc. up to some n-order mode shape. Each mode shape represents the

frequency of oscillation where a specific cause of deflection in a desired direction is

maximized. In mechanics, these modes typically define n-order mode shapes for bending and

torsion in the X, Y, and Z directions. As the order of a mode increases, the number of

oscillation inflection points increases resulting in the object’s oscillation becoming more local.

Due to this behavior, the maximum deflection along a body at a higher order mode will be

lower than that of the same mode shape at a lower order mode. The lowest frequency at which

10
an object will oscillate at is oftentimes referred to as the fundamental frequency of an object,

or also the first harmonic and first modal natural frequency, whereas the nth-harmonic

corresponds to the nth-order mode of a mode shape. Single objects each have a unique natural

frequency, a dominant frequency where a system oscillates at amidst a non-driving

disturbance, whereas systems often tend to have a number of natural frequencies

corresponding to their complexity. The natural frequency is best described as the resonant

system of harmonics within the object, whereas the fundamental frequency denotes the natural

frequency with the highest deformation amplitude, oftentimes being the first harmonic. In the

instance within this paper, flow of varying turbulence travels throughout a pipe system, and

internal flow creates pressure fluctuations on the interior face of the pipe, the strength of

which corresponds to the flow velocity as turbulence increases with flow velocity. The

internal pressure fluctuations on the inside face of the pipe create external FIVs. These

resultant FIVs are a type of free vibration as the force from vortex shedding onto the interior

face of the pipe has been characterized as an impulsive force that elicits free vibration rather

than an overwhelming forced response [71, 72]. Thus, these vibrations occur at a waveband

central to the system’s natural frequency.

The natural frequency of an object can be modeled using the object’s size, shape, material,

weight, and numerous other factors. The modal analysis for a pipe system can be investigated

via a few approaches. The theoretical method considers a pipe system a simple beam and uses

either Euler-Bernoulli or Timoshenko beam theory to model the free vibration of a pipe

system undergoing transverse vibration conditions [73] . Using d’Alembert’s principle, free

11
vibration in a single-span beam with homogenous material and uniform cross section can be

written as,

4 𝑦(𝑥,𝑡) 2 𝑦
𝐸𝐼 + 𝜌𝐴 t2 = 0 , (3)
x4

where EI is the beam’s bending stiffness, 𝜌 is the beam’s material density, and A is the beam’s

cross-sectional area. From modeling a simply supported beam using Euler-Bernoulli beam

theory, the natural frequency 𝜔𝑛 of the beam can be written as,

𝐸𝐼
𝜔𝑛 = (𝛽𝑙)2 √𝜌𝐴𝑙4 , (4)

where,

𝑠𝑖𝑛𝛽𝑙 = 0 , (5)

due to the displacement and bending moment at ends equaling zero in the simply supported

condition [73], E is the beam’s modulus of elasticity, I is the beam’s moment of inertia, 𝜌 is

the beam’s material density, A is the beam’s cross-sectional area, and l is the beam’s length.

Using Euler-Bernoulli equations to model natural frequency, it is important to note that a

planar surface remains planar, and that beam deflections are small and that the simply

supported beam bends into a circular arc [74].

Finite element analysis (FEA) is useful for many engineering applications because it allows

highly complex systems with structural, vibrational, heat transfer, and other considerations to

be more easily modeled. Using FEA, the natural frequency and mode shapes of a simply

supported beam that were formulated in theory are also able to be supported with high-

12
accuracy computational analysis, and subsequently validated experimentally [73]. Analytical

models for free vibration can be used in cases where expensive software FEA models are not

necessary [75, 76]. In computer-assisted FEA for a supported pipe system, changing the

support materials and flow condition instantaneously yields a new set of vibrational modes

and natural frequency calculations [77, 78]. The numerical relationship between pipe length,

support span length, and support rigidity with the pipe’s natural frequency has also been

studied [79]. Vibration modal analysis has also been studied with varying pipe wall thickness

and diameter [80], as well as variable pipe cross-section [81]. Using finite element modal

analysis, the theoretical equation for the natural frequency in a pipe has been modified to

include pipe support condition amidst calculations spanning altering pipe material, length,

support condition, and whether fluid was present or not within the pipe. The natural frequency

for a uniform span of piping has been calculated and experimentally verified [56] and can be

expressed as,

𝜆 𝑔𝐸𝐼
𝑓𝑜 = 2𝜋 √ µ𝑙4 (6)

where 𝑓𝑜 is the pipe span natural frequency, 𝜆 is a dimensionless frequency factor derived

from the pipe system’s support condition [82], 𝑔 is the gravitation constant, E is the pipe’s

modulus of elasticity, I is the pipe’s moment of inertia, µ is the weight per unit length of the

pipe including fluid and insulation.

Modal analysis via an experimental approach is another method oftentimes used to verify FEA

solutions. Experimental modal analysis (EMA) is typically carried out by attaching

electromagnetic shakers or PZT accelerometers to send excitation signals through a system

13
body and measure vibrational force corresponding to the excitation signal frequency. Doppler

vibrometers can also be used for EMA by emitting a laser at a system’s surface and measuring

the shift in received frequency as vibration within the system occurs. Received signals

undergo a Fourier transform so that resonances within the system can be viewed via the

frequency domain. EMA for pipe systems amongst various flow, support, and materials

conditions has been reported [83-87].

3. Flow Characteristics
As fluid circulates through a pipe system, the behavior of the flow can be characterized as

laminar or turbulent. Laminar flow is best defined as smooth, platelike flow where momentum

diffusion is high. In laminar flow through a circular cross-sectional pipe, fluid tends to move in

concentric layers with minimal inter-boundary mixing as the fluid particles from a boundary lack

the momentum to move to a boundary with a faster or slower velocity profile. For laminar,

incompressible flow moving through a tube with positive wall frictional force, the velocity

profile is highest at the vertical and horizontal centerline as shear forces at the pipe wall slow

nearby flow boundaries [88]. A two-dimensional velocity profile is shown below in Figure 3-1.

Figure 3-1: Two-dimensional velocity profile of laminar, incompressible flow between two rigid surfaces.

14
The Reynolds number of a flow is commonly described as the ratio between inertial forces and

viscous forces, and can be expressed as,

𝜌𝑉𝐿
𝑅𝑒 = , (7)
µ

where ρ is fluid density, V is the characteristic velocity of the flow, L is the characteristic

length of the flow, and μ is the dynamic viscosity of the flow. Laminar flows are described as

flows with a low Reynolds number, characterized as at sufficiently laminar at or below Re ≈

2300 [89]. Beyond this Reynolds number there is what is called a transitional zone where the

fluid is neither sufficiently laminar nor turbulent. Turbulent dynamic fluid structures known as

puffs begin to appear above Re ≈ 2300, which when Re ≥ 3000 are characterized as slugs, the

difference between the two residing in the behavior of the leading-edge velocity of the

turbulent structures [90-92]. As the Reynolds number increases in the transitional zone,

turbulent structures endure for longer time periods before diffusing as flow becomes less

laminar, whereas flow can be described as turbulent at Re ≈ 4200-4300 [93, 94]. Turbulent

flow is characterized as chaotic, irregular, and highly diffusive flow with a high Reynolds

number, distinctive of flow with high inertial force. In turbulent flow, vortices of swirling

fluid opposite the current known as eddies occur which disrupt the sheetlike flow that is

characteristic of the laminar regime. Eddies are described as a primary attribute of the

distinctive rotationality in turbulent flow, as the force of eddies are stretched due to high flow

inertia. In addition to creating deviating pressure fluctuations on the pipe interior face, eddies

prevent internal flow from fully developing in the instantaneous time scale [88]. Figure 3-2

shows a time-averaged turbulent flow profile.

15
 Figure 3-2: Two-dimensional time-averaged velocity profile turbulent flow between two rigid surfaces.

Compared to the laminar regime, the shear stress at the wall is significantly greater, causing a

higher profile gradient. Additionally, the difference in average velocity as radial distance

increases from the centerline results in a significantly smaller velocity difference when

compared to the laminar case.

When flow changes directionality via pipe elbows, flow undergoes short-term turbulence that

diffuses as flow returns to a steady condition after a sufficient length of constant

directionality. As flow travels around a 90º bend, centrifugal force moves the faster and

previously more central flow boundaries off the centerline of flow and onto the internal outer

radius of the pipe elbow, or the extrados, while slower flow boundaries move towards the

internal inner radius, also known as the intrados [95, 96]. Pressure fluctuations occur as flow

returns from a relatively turbulent state in the elbow to steady in downstream flow. It is

reported that in higher velocities, pressure fluctuations depend on mean velocity in the pipe,

with the distribution of standard deviation of pipe fluctuations increasing in kurtosis as mean

velocity increases [97]. This is supported by earlier [20, 21] and later [53, 54] studies

16
observing that the standard deviation of internal pressure fluctuations increases quadratically

with mean velocity.

Proximity to pipe support is also a strong factor in FIV strength, as the 1st mode shape of a

simply supported beam contains the highest amplitude of deflection at the center of the beam.

The magnitude of FIVs is thus expected to be greater near the center of the pipe than near the

support of the pipe for equal flow conditions.

17
 Materials and Methods

Non-invasive detection DAS techniques via fiber optic sensors amidst an array of flow and

viscosity values have been developed. This section describes the method for fabrication of a fiber

optic sensor, application of the sensors onto the pipe system, as well as theoretical and

experimental modal analysis for the pipe system.

4. Experimental Design
FIV detection testing was conducted with a variety of sensors at unique positions on the pipe

system. Acoustic sensing was performed at volumetric flow rates of 0 liters per minute (l/min)

to 50-60 l/min (0.64-0.76 m/s), the maximum depending on the pump limitation at the time of

testing, at a step rate of 5 l/min with the exception of testing at 40ºC, all of which were

completed at a step rate of 1 l/min. Flow rate was measured down to 0 l/min, however the

smoothed peak properties become unmeasurable around a floor of 5-15 l/min (0.06-0.19 m/s)

dependent on the sensor sensitivity due to weak internal pressure fluctuations. This testing

was performed with four unique sensors, the last of which was placed at three unique

locations on the pipe system. This testing was performed at water temperatures from 15ºC-

60ºC at a step rate of 5ºC with the exception of one of the three locations of the fourth sensor.

All reported flow rates have an error of 0.2 l/min, and all reported temperatures have an error

of 1ºC. Tests run for about 15 seconds after flow and temperature are approximately steady at

a sampling frequency of 36.678 kHz.

18
 4.1 Sensor Design and Application
Two styles of sensing arrays were applied to the pipe system in a number of locations along

the system. Coil sensors, the first style of two, are jacketed single mode fiber sections

wrapped in tension around the pipe circumference in a gapless helix. When FIVs are present,

coil sensors measure strain from circumferential disturbances. Sensing pads, the second style

of two, are devices comprised of unjacketed single mode fiber glued onto a sheet of semi-rigid

insulation via epoxy. The sensing pads are affixed to the pipe system using external hose

clamps to increase the interfacial bonding strength of the pads to the pipe system. Similarly to

coil sensors, sensing pads measure strain via circumferential disturbances caused by FIVs.

Coil Sensors

Coil sensors are comprised of single mode fiber with a germanium-doped silica core and a

pure silica cladding. The fiber is coated with a layer of acrylate to protect the fiber physical

damage and to increase the fiber bending allowance. The fiber is jacketed with Hytrel, a

thermoplastic polyester elastomer which gives the fiber higher mechanical durability and

flexibility. With added mechanical stability from the coating and jacket, the fiber is able to be

tightly wound with ease at a bend radius of one inch. Continuous gratings are present

throughout the fiber at a distance of 2 meters to allow for DAS along each 2-meter fiber

section. Individual fiber channels are identifiable via optical fault detection and instantaneous

impulse additives such as fiber tapping in conjunction with DAS. Figure 4-1 shows a coil

sensor wrapped on the pipe system with outer pipe diameter 1.315 inches.

19
 Figure 4-1: Image of a coil sensor wrapped around the exterior of the pipe system.

Two coil sensors were used in this experiment, the first of which pictured in Figure 4-1 was

placed about 0.08 meters from the vertical centerline of the pipe elbow, and about 0.48 meters

from the nearest pipe support. The second coil sensor was placed about 2.16 meters from the

pipe elbow, and about 0.31 meters from the nearest pipe support.

Sensing Pads

Sensing pads were designed and fabricated as a means to increase sensor sensitivity to FIVs

when compared to traditional coil sensors. Sensing pads are fabricated by applying a layer of

epoxy on a sheet of partly or fully rigid material and embedding fiber within the epoxy before

the epoxy is fully hardened. In this experiment, two unique sensing pads were fabricated, both

using SMF-28 acrylate-coated fiber with Bragg gratings written every 2 meters within the fiber.

Both pads also used clear sealant as the epoxy. The first pad used a fiberglass pipe insulation

base fitted to wrap around the size of the pipes used. The second pad used a flat sheet of

microporous aerogel insulation as the base. During fabrication, a layer of epoxy was applied onto

20
the inner surface of the fiberglass pad and one side of the aerogel pad. Fiber was embedded on

both pads in a loop orientation centered at the middle of each pad with a bend radius of about an

inch for each pad, the approximate workable limitation for the fiber. Figure 4-2 and Figure 4-3

show both finished pads.

Figure 4-2: Image of aerogel pad with optical fault detector identifying embedded fiber channels and gratings (left) and without
optical fault detector (right).

In the left portion of Figure 4-2, fiber is illuminated with an optical fault detector that is typically

used to identify fiber breaks via high scattering. Exposed fiber with written FBGs also contain

points of high scattering at the FBG locations due to the mismatch of refractive indices. Portions

of fiber are also seen illuminated most prominently in areas of high fiber strain as a result of the

fabrication process. Three gratings are observed in the aerogel pad, yielding two sensor channels

and one length of fiber that leads to the end reflection, seen at the bottom of the left half of Figure

4-2.

21
Figure 4-3: Image of fiberglass pad with optical fault detector identifying embedded fiber channels and gratings (left) and without
optical fault detector (right).

Again, the left half of Figure 4-3 shows fiber illuminated by an optical fault detector. Much of

this fiber is illuminated due to the tight bend radius of the fiber orientation yielding high strain

when the sealant holding the fiber in place completely set. The fiberglass pad was placed on the

pipe about 1.65 meters from the elbow centerline and 0.85 meters from the nearest pipe support.

The fiberglass sensing pad was tested at one constant downstream position, while the aerogel

sensing pad was tested at three locations: the corresponding downstream position of the

fiberglass sensing pad (location 1), a location adjacent to the flow elbow (location 2), and an

additional downstream position adjacent to a pipe support (location 3). The first location is the

same location as the fiberglass sensing pad was tested at. The second location is 0.39 meters

from the elbow centerline and 0.17 meters from the nearest pipe support. The third location is

2.52 meters from the elbow centerline and 0.31 meters from the nearest pipe support. Hose

22
clamps were used with both sensing pads in order to ensure strong bonding to the pipe surface.

Adjusting the tightness of the hose clamp noticeably altered the intensity and central frequency

of the first mode smoothed frequency domain peak from FIVs. After discovery of this behavior,

hose clamp tightness was adjusted to yield the maximum sensed FIV frequency domain peak

intensity as a calibration technique. The fiberglass pad was applied onto the bare surface of the

pipe. The aerogel pad was placed onto a shell of about 9 layers of aerogel insulation created to

increase the bonding area of the pad to the pipe system. Conventionally the large amount of

insulation added damps the resultant signal, however early testing with the pad showed a

stronger signal being received as the diameter of the shell increased, likely due to the more

efficient interfacial bonding. Figure 4-4 and Figure 4-5 show both pads as they were applied onto

the pipe system.

Figure 4-4: Image of aerogel pad affixed to the pipe system from a front view (left) and a side view (right).

Figure 4-5: Image of fiberglass pad affixed to the pipe system from a front view (left) and a side view (right).

23
 4.2 Pipe System Design
The water used in the flow loop is housed in a tank and sourced from a local well. Water exits

the bottom of the tank to the system’s pump, and from there the water is sent to a diversion

where flow to the pipe system is controlled. All pipes in the pipe system are size 1 NPT. After

the diversion, there are two tubes with shut-off valves, the first of which sends water to the pipe

system and the second of which empties into the tank. As water enters the pipe system, there is

an initial 3-way ball valve that was originally intended to control the flow seen in the pipe

system, however initial tests observed that the valve when partially closed added a significant

amount of noise to every sensor that was tested. While the valve was open completely, no

significant noise was apparent thus the valve remained in that position for the duration of

experimentation. Water then flows through the bottom pipe, is diverted upwards via a pipe elbow

and then rightwards via another pipe elbow. The first pipe immediately after the second elbow is

a 10-foot stainless steel pipe which was used for all testing in this work. After this pipe there is

an inline magnetic-inductive flow meter used to verify flow and temperature of flowing water

throughout the experiment. Figure 4-6 shows a side view of the pipe system.

24
 Figure 4-6: Image of the pipe system.

The pipe system is supported by an apparatus of vertical and horizontal struts that allow for

rubber pipe clamps for pipe support. A suspended tank heater was used to heat the water within

the tank, and water from the well was used to cool the water. Figure 4-7 shows the tank with the

heater suspended as well as the flow control valves used.

Figure 4-7: Image of tank, heater, and flow control valves from a front view (left) and a side view showing the water’s tank exit
(right).

25
In addition to separating the flow control valves from the pipe system structure, additional noise

reduction methods were taken to minimize noise originating from the pump operation, seen to be

a strong, distributed low frequency (<200Hz) spread that created difficulty in data analysis.

Walls separating the tank room and pipe system room were re-insulated after creating a hole for

the water-carrying tubes. These operating conditions were held constant throughout the

experiment.

4.3 Interrogation
Fiber interrogation was completed using a DASnova-02 Fiber Optic DAS System interrogator

from Sentek Instrument. This system features zero crosstalk between sensors that is a

fundamental strength of fiber optic sensing systems, as well as little sensitivity deterioration

under 1 kilometer distance. The Sentek interrogator is reported to improve upon traditional DAS

systems in areas of fiber fabrication, interrogator design, and signal processing, all

contributing to a reported strain sensitivity of ≤ 0.2 nε [98]. Figure 4-8 and Figure 4-9 show

an image of the interrogator as well as a sample image of real measurement data, respectively.

26
 Figure 4-8: DASnova-02 Fiber Optic DAS System used through experimentation.

Figure 4-9: Sample real measurement data from the software control panel of the interrogator. [98]

Figure 4-9 shows a sample measurement in an instance of 360 sensors with gratings located

every 5 meters where a singular channel of fiber is acted on by a piezoelectric cylinder driven by

a function generator. In this instance, one channel shows a strain measurement over a short

duration with no channel crosstalk. In this experiment, the fiber channel length used throughout

27
all tests was 2 meters. The coil sensors had 1 and 3 active sensor channels on the elbow and

downstream coils, respectively, while all sensing pads had 1 active sensor channel during the

experiment.

5. Modal Analysis of Pipe System

The modal analysis of the pipe system was carried out numerically via previously reported

equations. Modal analysis was also evaluated experimentally using frequency spectrum analysis

via a piezoelectric accelerometer.

5.1 Numerical Modal Analysis

All experiments were completed on a continuous 10-foot SAE 304 stainless steel pipe with fixed-

fixed support. Table 5-1 below notes the material properties and dimensions for all variables in

Equation 6.

Table 5-1: Properties and dimensions of the tested pipe.

Frequency Factor 22.40
Gravitation constant (m/s) 9.8
Modulus of Elasticity (GPa) 215
Moment of Inertia (m4 ) 3.6355e-8
Weight per unit length (kg/m) 2.818

By using equation 6, the approximate natural frequency of the pipe used is 67.2 Hz. Across the

pipe, there are longer unsupported sections where flow was tested and these sections are

expected to have lower natural frequencies [56], whereas shorter unsupported pipe sections are

expected to have higher natural frequencies. Affixation condition and tightness were observed to

alter a computationally smoothed resonant frequency, with tight hose clamps correlating with

higher central smoothed frequency.

28
 5.2 Experimental Modal Analysis
Experimental modal analysis was performed on the pipe system in order to experimentally

confirm the numerical modal analysis. The analysis was carried out via a Virtins IEPE

accelerometer CA-YD-181 bonded to the pipe at the support-distant location. Free vibration was

measured via Virtins Multi-Instrument measurement software. Vibration was captured in the

frequency domain via the software’s built-in spectrum analyzer and smoothed with Hann

smoothing. Vibration was induced via instantaneous impacts with wrenches and hammers at

different locations. Figure 5-1 shows the accelerometer bonded to the pipe.

Figure 5-1: Image of the bonded accelerometer.

After performing the experiment, the first mode was experimentally determined to center around

67.35 Hz These results closely match that of the numerically obtained modal analysis, as well as

show additional system modes at 176.13 Hz and 279.08 Hz, respectively. Figure 5-2 shows the

frequency spectrum during the experimental modal analysis.

29
 Figure 5-2: Frequency spectrum of experimental modal analysis.

In Figure 5-2, it is important to note that the resolution note in the bottom left corner pertains to

of the software panel pertains to the visual display, not the calculated peak frequency.

30
 Results and Discussion

6. Data Post-Processing
After designating a sensor channel, individual tests yielded strain results in the time domain.

This data was then converted to the frequency domain by using the built-in Matlab fast

Fourier transform function. A smoothing filter was then applied to the data using a fifth order

Savitzky-Golay finite impulse response filter. The maximum of the resulting peak located

approximately 60-80 Hz was denoted as the peak intensity, with the corresponding frequency

being denoted as the peak central frequency. Peak minimums were automatically identified by

prominence level and manually verified in order to calculate the peak full width at half

maximum and peak area. Noise floors that coincided with peak frequencies were locally

approximated to better estimate peak area as a result of strain caused by FIVs. These four

resulting properties were plotted against flow rate and flow temperature. Flow temperature

analysis was not completed for the third aerogel pad location due to an insufficient number of

temperature steps. Several trends were observed across each sensor. Sensor position and

affixation quality remained constant throughout all testing for a given sensor. The ability to

measure FIVs with fiber optic sensors as a function of frequency peak intensity was gained.

Other trends have emerged that may suggest further study and optimization can lead to

increased sensing capabilities.

31
7. Peak Intensity Relationships
7.1 Flow Rate
The relationship between smoothed peak intensity and flow rate was calculated for all sensors.

Figure 7-1 shows the downstream coil’s measured peak intensity as flow rate varies at a constant

flow temperature of 40ºC at a step rate of 1 l/min. Figure 7-2 shows the same relationship at

60ºC and a step rate of 5 l/min. Figure 7-3 and Figure 7-4 show the same relationships for the

elbow coil.

Peak Intensity vs. Flow Rate, 40°C

0.001
Micro-Strain Intensity (µm/m)/Hz

0.0009 y = 0.0000006x2 - 0.0000178x + 0.0001530

R² = 0.9897
0.0008
0.0007
0.0006
0.0005
0.0004
0.0003
0.0002
0.0001
0
0 10 20 30 40 50 60
Flow Rate (l/min)

Figure 7-1: Maximum FIV peak intensity vs flow rate at 40ºC with quadratic fit line measured via the downstream coil.

32
 Peak Intensity vs. Flow Rate, 60°C
0.001
y = 0.0000005x2 - 0.0000135x + 0.0001210
Micro-Strain Intensity (µm/m)/Hz
0.0009 R² = 0.9967
0.0008
0.0007
0.0006
0.0005
0.0004
0.0003
0.0002
0.0001
0
0 10 20 30 40 50 60
Flow Rate (l/min)

Figure 7-2: Maximum FIV peak intensity vs flow rate at 60ºC with quadratic fit line measured via the downstream coil.

Peak Intensity vs. Flow Rate, 40°C

0.0014
Micro-Strain Intensity (µm/m)/Hz

0.0012 y = 0.0000007x2 - 0.0000212x + 0.0001961

R² = 0.9926
0.0010

0.0008

0.0006

0.0004

0.0002

0.0000
0 10 20 30 40 50 60
Flow Rate (l/min)

Figure 7-3: Maximum FIV peak intensity vs flow rate at 40ºC with quadratic fit line measured via the elbow coil.

33
 Peak Intensity vs. Flow Rate, 60°C
0.0012

Micro-Strain Intensity (µm/m)/Hz

y = 0.0000007x2 - 0.0000220x + 0.0002131
0.0010 R² = 0.9988

0.0008

0.0006

0.0004

0.0002

0.0000
0 10 20 30 40 50 60
Flow Rate (l/min)

Figure 7-4: Maximum FIV peak intensity vs flow rate at 60ºC with quadratic fit line measured via the elbow coil.

As reported in publications regarding the relationships of FIVs and flow velocity, a quadratic

relationship with a strong coefficient of determination (R²) is calculated in both coil sensors.

When tested at the same temperature, the peak intensity at the same flow rate of the elbow coil to

the downstream coil is considered to be about twenty percent stronger on the elbow coil with no

apparent velocity-dependent relationship. The trend between singular peak intensity and flow

rate is observed across the full array of tested temperatures with strong correlation and low

individual test variation. Table 7-1 and Table 7-2 show the calculated R² and standard deviation

for the downstream coil and elbow coil, respectively.

34
Table 7-1: Downstream coil peak intensity R² and standard deviation across tested temperature array
Flow Temperature (ºC) R² σ (l/min)

15 0.9988 0.52
20 0.9818 2.34
25 0.9896 1.84
30 0.9860 1.43
35 0.9871 1.29
40 0.9934 1.58
45 0.9959 1.32
50 0.9818 2.76
55 0.9940 1.10
60 0.9967 0.90
Average 0.9905 1.51

Table 7-2: Elbow coil peak intensity R² and standard deviation across tested temperature array
Flow Temperature (ºC) R² σ (l/min)

15 0.9935 0.96
20 0.9965 1.31
25 0.9857 3.13
30 0.9941 0.74
35 0.9942 0.88
40 0.9941 0.86
45 0.9910 1.93
50 0.9786 1.31
55 0.9989 1.02
60 0.9988 0.91
Average 0.9925 1.31

In both cases, the calculated average R² value is above 0.99 and average flow rate standard

deviation is below 2 l/min. Standard deviation was primarily attributed to high individual

variance at the lower end of peak measurability 10-20 l/min. Numerically this can be attributed

to the limitations of quadratic curve fitting, additionally it can also be experimentally explained

35
as a higher ratio of the total sensed signal at the central frequency originating from non-FIV

noise.

Smoothed peak intensity was also calculated at an array of temperatures for both sensing pad

designs. Figure 7-5 through Figure 7-12 show the relationship between peak intensity and flow

rate at a constant flow temperature of 40ºC at a step rate of 1 l/min, as well as the same

relationship at 60ºC and a step rate of 5 l/min for all four sensor sets.

Peak Intensity vs. Flow Rate, 40°C

0.007
Micro-Strain Intensity (µm/m)/Hz

0.006 y = 0.000003x2 - 0.000092x + 0.000764

R² = 0.9870
0.005

0.004

0.003

0.002

0.001

0.000
0 10 20 30 40 50 60
Flow Rate (l/min)

Figure 7-5: Maximum FIV peak intensity vs flow rate at 40ºC with quadratic fit line measured via the fiberglass sensing pad.

36
 Peak Intensity vs. Flow Rate, 60°C
0.008

Micro-Strain Intensity (µm/m)/Hz

0.007 y = 0.000004x2 - 0.000101x + 0.000794
R² = 0.9941
0.006
0.005
0.004
0.003
0.002
0.001
0.000
0 10 20 30 40 50 60
Flow Rate (l/min)

Figure 7-6: Maximum FIV peak intensity vs flow rate at 60ºC with quadratic fit line measured via the fiberglass sensing pad.

Peak Intensity vs. Flow Rate, 40°C

0.020
Micro-Strain Intensity (µm/m)/Hz

0.018 y = 0.000008x2 - 0.000153x + 0.000888

0.016 R² = 0.9953
0.014
0.012
0.010
0.008
0.006
0.004
0.002
0.000
0 10 20 30 40 50 60
Flow Rate (l/min)

Figure 7-7: Maximum FIV peak intensity vs flow rate at 40ºC with quadratic fit line measured via the aerogel sensing pad,

location 1.

37
 Peak Intensity vs. Flow Rate, 60°C
0.025

Micro-Strain Intensity (µm/m)/Hz

y = 0.000009x2 - 0.000183x + 0.001404
0.020 R² = 0.9962

0.015

0.010

0.005

0.000
0 10 20 30 40 50 60 70
Flow Rate (l/min)

Figure 7-8: Maximum FIV peak intensity vs flow rate at 60ºC with quadratic fit line measured via the aerogel sensing pad,

location 1.

Peak Intensity vs. Flow Rate, 40°C

0.014
Micro-Strain Intensity (µm/m)/Hz

0.012 y = 0.000007x2 - 0.000190x + 0.001388

R² = 0.9912
0.010

0.008

0.006

0.004

0.002

0.000
0 10 20 30 40 50 60
Flow Rate (l/min)

Figure 7-9: Maximum FIV peak intensity vs flow rate at 40ºC with quadratic fit line measured via the aerogel sensing pad,

location 2.

38
 Peak Intensity vs. Flow Rate, 60°C
0.014

Micro-Strain Intensity (µm/m)/Hz

0.012 y = 0.000007x2 - 0.000216x + 0.001852
R² = 0.9941
0.010

0.008

0.006

0.004

0.002

0.000
0 10 20 30 40 50 60 70
Flow Rate (l/min)

Figure 7-10: Maximum FIV peak intensity vs flow rate at 60ºC with quadratic fit line measured via the aerogel sensing pad,

location 2.

Peak Intensity vs. Flow Rate, 40°C

0.012
y = 0.000005x2 - 0.000122x + 0.000640
Micro-Strain Intensity (µm/m)/Hz

0.010 R² = 0.9843

0.008

0.006

0.004

0.002

0.000
0 10 20 30 40 50 60
-0.002
Flow Rate (l/min)

Figure 7-11: Maximum FIV peak intensity vs flow rate at 40ºC with quadratic fit line measured via the aerogel sensing pad,

location 3.

39
 Peak Intensity vs. Flow Rate, 60°C
0.009
y = 0.000005x2 - 0.000115x + 0.000703

Micro-Strain Intensity (µm/m)/Hz

0.008
R² = 0.9960
0.007
0.006
0.005
0.004
0.003
0.002
0.001
0.000
-0.001 0 10 20 30 40 50 60
Flow Rate (l/min)

Figure 7-12: Maximum FIV peak intensity vs flow rate at 60ºC with quadratic fit line measured via the aerogel sensing pad,

location 3.

A quadratic relationship between peak intensity and flow rate is universally observed amongst all

sensing pads with strong quadratic correlation. Similar to the coil sensors, the average peak

intensity difference between the aerogel sensing pad elbow-adjacent location 2 and downstream

location 3, is about twenty percent. In the downstream location 1 instance where the nearest pipe

support is about three times further than in location 2, peak intensity is on average about 143%

stronger. The trend between singular peak intensity and flow rate is also noticed across the full

array of tested temperatures as observed via the coil sensors. Table 7-3 through Table 7-6 show

the calculated R² and standard deviation for the fiberglass sensing pad and the three locations

tested via the aerogel sensing pad.

40
Table 7-3: Fiberglass sensing pad peak intensity R² and standard deviation across tested temperature array
Flow Temperature (ºC) R² σ (l/min)

15 0.9920 1.74
20 0.9969 0.74
25 0.9725 2.27
30 0.9887 2.11
35 0.9938 1.73
40 0.9859 1.59
45 0.9926 1.53
50 0.9984 0.53
55 0.9980 1.14
60 0.9941 1.72
Average 0.9913 1.51

Table 7-4: Aerogel sensing pad peak intensity R² and standard deviation across tested temperature array, location 1
Flow Temperature (ºC) R² σ (l/min)

15 0.9700 1.93
20 0.9973 1.68
25 0.9941 1.90
30 0.9957 1.18
35 0.9903 1.93
40 0.9968 1.50
45 0.9981 0.57
50 0.9943 0.99
55 0.9962 0.99
60 0.9962 2.24
Average 0.9929 1.49

41
Table 7-5: Aerogel sensing pad peak intensity R² and standard deviation across tested temperature array, location 2
Flow Temperature (ºC) R² σ (l/min)

15 0.9919 2.15
20 0.9941 2.58
25 0.9903 3.04
30 0.9960 0.89
35 0.9986 1.47
40 0.9944 2.00
45 0.9955 1.89
50 0.9967 1.97
55 0.9972 1.67
60 0.9941 2.06
Average 0.9949 1.97

Table 7-6: Aerogel sensing pad peak intensity R² and standard deviation across tested temperature array, location 3
Flow Temperature (ºC) R² σ (l/min)

15 0.9969 1.84
20 0.9952 2.23
30 0.9852 2.80
40 0.9924 2.71
50 0.9935 2.28
60 0.9960 2.16
Average 0.9932 2.34

In all cases, the measured average R² value is above 0.99 and average flow rate standard

deviation is below 2.5 l/min. Instances of standard deviation above 2 l/min contain high

individual variance at the floor of peak measurability 10-20 l/min. Standard deviation was

primarily attributed to high individual variance at the lower end of peak measurability 5-20

l/min.

42
 7.2 Flow Temperature
The relationship between maximum peak intensity and flow temperature at constant flow rates

was also observed. The y-axis scale in this section is smaller than that in 1.7.1 to enlarge

variation. Figure 7-13 and Figure 7-14 below show the graphs of calculated peak intensity and

flow temperature for 15 and 50 l/min for the downstream coil respectively. Figure 7-15 and

Figure 7-16 show the same data for the elbow coil.

Peak Intensity vs. Flow Temperature, 15 l/min

0.000040
Micro-Strain Intensity (µm/m)/Hz

0.000035
0.000030
0.000025
0.000020
0.000015
0.000010
0.000005
0.000000
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 7-13: Maximum FIV peak intensity vs flow temperature at 15 l/min measured via the downstream coil.

43
 Peak Intensity vs. Flow Temperature, 50 l/min
0.0010

Micro-Strain Intensity (µm/m)/Hz

0.0009
0.0008
0.0007
0.0006
0.0005
0.0004
0.0003
0.0002
0.0001
0.0000
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 7-14: Maximum FIV peak intensity vs flow temperature at 50 l/min measured via the downstream coil.

Peak Intensity vs. Flow Temperature, 15 l/min

0.00004
Micro-Strain Intensity (µm/m)/Hz

0.000035
0.00003
0.000025
0.00002
0.000015
0.00001
0.000005
0
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 7-15: Maximum FIV peak intensity vs flow temperature at 15 l/min measured via the elbow coil.

44
 Peak Intensity vs. Flow Temperature, 50 l/min
0.0012

Micro-Strain Intensity (µm/m)/Hz

0.0010

0.0008

0.0006

0.0004

0.0002

0.0000
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 7-16: Maximum FIV peak intensity vs flow temperature at 50 l/min measured via the elbow coil.

No apparent trend was determined in charting peak intensity and flow temperature at any

individual flow rate. Deviation in this relationship can be attributed to randomness in flow due to

lack of thermal effect on the thermoplastic elastomer fiber jacket to contribute to increased strain.

The relationship between maximum peak intensity and flow temperature at constant flow rates

was also observed in the fiberglass sensing pad. Figure 7-17 and Figure 7-18 below show the

graphs of calculated peak intensity and flow temperature for 15 and 50 l/min, respectively, in

the fiberglass sensing pad.

45
 Peak Intensity vs. Flow Temperature, 15 l/min
0.00020

Micro-Strain Intensity (µm/m)/Hz

0.00018
0.00016
0.00014
0.00012
0.00010
0.00008
0.00006
0.00004
0.00002
0.00000
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 7-17: Maximum FIV peak intensity vs flow temperature at 15 l/min measured via the fiberglass sensing pad.

Peak Intensity vs. Flow Temperature, 50 l/min

0.007
Micro-Strain Intensity (µm/m)/Hz

0.006

0.005

0.004

0.003

0.002

0.001

0
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 7-18: Maximum FIV peak intensity vs flow temperature at 50 l/min measured via the fiberglass sensing pad.

In this case, a positive trend is seen between peak intensity and flow temperature at individual

flow rate. This can be attributed to the direct contact that the silicone sealant has with the pipe

surface. Due to the sealant’s low coefficient of thermal expansion, the elevated external pipe

temperature is easily transferred to the face of the sensing pad, adding thermally induced

46
mechanical strain to the fiber. Linear fitting for tests plotting peak intensity and flow temperature

for the fiberglass sensing pad yields R² values of 0.75-0.95, with high deviation between flow

rate sets unlike that of the relationship between peak intensity and flow rate.

The relationship between peak intensity and flow temperature was calculated via the aerogel

sensing pad in three locations. Figure 7-19 and Figure 7-20 below show the graphs of

calculated peak intensity and flow temperature for 15 and 50 l/min for the support-distant

downstream aerogel pad, respectively. Figure 7-21 and Figure 7-22 show the same data for the

aerogel sensing pad in the support-adjacent case.

Peak Intensity vs. Flow Temperature, 15 l/min

0.0007
Micro-Strain Intensity (µm/m)/Hz

0.0006

0.0005

0.0004

0.0003

0.0002

0.0001

0.0000
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 7-19: Maximum FIV peak intensity vs flow temperature at 15 l/min measured via the aerogel sensing pad, location 1.

47
 Peak Intensity vs. Flow Temperature, 50 l/min
0.016

Micro-Strain Intensity (µm/m)/Hz

0.014
0.012
0.01
0.008
0.006
0.004
0.002
0
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 7-20: Maximum FIV peak intensity vs flow temperature at 50 l/min measured via the aerogel sensing pad, location 1.

Peak Intensity vs. Flow Temperature, 15 l/min

0.00025
Micro-Strain Intensity (µm/m)/Hz

0.00020

0.00015

0.00010

0.00005

0.00000
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 7-21: Maximum FIV peak intensity vs flow temperature at 15 l/min measured via the aerogel sensing pad, location 2.

48
 Peak Intensity vs. Flow Temperature, 50 l/min
0.01

Micro-Strain Intensity (µm/m)/Hz

0.009
0.008
0.007
0.006
0.005
0.004
0.003
0.002
0.001
0
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 7-22: Maximum FIV peak intensity vs flow temperature at 50 l/min measured via the aerogel sensing pad, location 2.

In these cases, the 15 l/min tests are inconsistent as the support-distant case shows a weak

positive trend as flow temperature increases. In the support-adjacent case, there is no apparent

trend. In the 50 l/min tests there is a positive trend until a local temperature threshold, 40ºC in

the support-distant case and 30ºC in the support-adjacent case, and subsequently peak intensity

in higher temperatures fails to maintain this trend. These cases resemble an upward exponential

decay trend. In these cases, as the aerogel sensing pad is affixed onto a shell of insulation about 5

inches thick, the face of the sensing pad is sufficiently insulated from the heat of the pipe exterior

surface. Since the sensing pad is fully insulated from increased temperature and subsequent

thermally induced mechanical stress, this relationship is fully intrinsic to flow.

8. Peak Area Relationships

8.1 Flow Rate
The relationship between smoothed peak area and flow rate was determined for all tests. Figure

8-1 shows the downstream coil’s calculated peak area as flow rate varies at a constant flow

49
temperature of 40ºC at a step rate of 1 l/min, and Figure 8-2 shows the same relationship at 60ºC

and a step rate of 5 l/min. Figure 8-3 and Figure 8-4 show the same relationships for the elbow

coil.

Peak Area vs. Flow Rate, 40°C

0.025
Micro-Strain Intensity (µm/m)²/Hz

y = 0.00001x2 - 0.00048x + 0.00409

R² = 0.9765
0.020

0.015

0.010

0.005

0.000
0 10 20 30 40 50 60
Flow Rate (l/min)

Figure 8-1: Peak area vs flow rate at 40ºC with quadratic fit line measured via the downstream coil.

Peak Area vs. Flow Rate, 60°C

0.020
Micro-Strain Intensity (µm/m)²/Hz

0.018 y = 0.00001x2 - 0.00033x + 0.00280

R² = 0.9938
0.016
0.014
0.012
0.010
0.008
0.006
0.004
0.002
0.000
0 10 20 30 40 50 60
Flow Rate (l/min)

Figure 8-2: Peak area vs flow rate at 60ºC with quadratic fit line measured via the downstream coil.

50
 Peak Area vs. Flow Rate, 40°C
0.030
y = 0.00002x2 - 0.00048x + 0.00412

Micro-Strain Intensity (µm/m)²/Hz

0.025 R² = 0.9810

0.020

0.015

0.010

0.005

0.000
0 10 20 30 40 50 60
Flow Rate (l/min)

Figure 8-3: Peak area vs flow rate at 40ºC with quadratic fit line measured via the elbow coil.

Peak Area vs. Flow Rate, 60°C

0.025
Micro-Strain Intensity (µm/m)²/Hz

y = 0.00002x2 - 0.00056x + 0.00564

0.020 R² = 0.9944

0.015

0.010

0.005

0.000
0 10 20 30 40 50 60
Flow Rate (l/min)

Figure 8-4: Peak area vs flow rate at 60ºC with quadratic fit line measured via the elbow coil.

Similar to the peak area relationship with flow rate, a quadratic relationship with a strong R² is

observed in both coil sensors when measuring the relationship with peak area and flow rate. The

trend between singular peak area and flow rate is seen across the full array of tested temperatures

51
with strong correlation and low individual test variation. Table 8-1 and Table 8-2 show the

calculated R² and standard deviation for the downstream coil and elbow coil, respectively.

Table 8-1: Downstream coil peak area R² and standard deviation across tested temperature array
Flow Temperature (ºC) R² σ (l/min)

15 0.9952 1.11
20 0.9754 2.78
25 0.9836 2.43
30 0.9859 1.41
35 0.9764 1.92
40 0.9846 2.02
45 0.9955 1.87
50 0.9698 3.06
55 0.9923 1.45
60 0.9938 1.59
Average 0.9853 1.96

Table 8-2: Elbow coil peak area R² and standard deviation across tested temperature array
Flow Temperature (ºC) R² σ (l/min)

15 0.9700 2.44
20 0.9983 1.28
25 0.9701 3.56
30 0.9785 1.68
35 0.9870 1.33
40 0.9976 1.15
45 0.9882 2.31
50 0.9689 1.31
55 0.9935 1.94
60 0.9944 1.59
Average 0.9847 1.86

In both cases, the calculated average R² value is above 0.98 and average flow rate standard

deviation is below 2 l/min. Standard deviation was again primarily attributed to high individual

52
variance at the lower flow rates, in these instances most commonly 10-20 l/min. Numerically this

can be attributed to the limitations of quadratic curve fitting, additionally it can also be

experimentally explained as a higher ratio of the total sensed signal at the central frequency

originating from non-FIV noise. Additional error in peak area cases can be attributed to peak

shape inconsistency in individual flow tests stemming from noise in the same frequency band as

the FIVs.

Smoothed peak area was also calculated at an array of temperatures for both sensing pad designs.

The fiberglass sensing pad was tested at one constant downstream position, while the aerogel

sensing pad was tested at three locations: the corresponding downstream position of the

fiberglass sensing pad (location 1), a location adjacent to the flow elbow (location 2), and an

additional downstream position adjacent to a pipe support (location 3). Figure 8-5 through Figure

8-12 show the relationship between peak area and flow rate at a constant flow temperature of

40ºC at a step rate of 1 l/min, as well as the same relationship at 60ºC and a step rate of 5 l/min

for all four sensor sets.

53
 Peak Area vs. Flow Rate, 40°C
0.14

Micro-Strain Intensity (µm/m)²/Hz

0.12 y = 0.00008x2 - 0.00226x + 0.01895
R² = 0.9899
0.10

0.08

0.06

0.04

0.02

0.00
0 10 20 30 40 50 60
Flow Rate (l/min)

Figure 8-5: Peak area vs flow rate at 40ºC with quadratic fit line measured via the fiberglass sensing pad.

Peak Area vs. Flow Rate, 60°C

0.16
Micro-Strain Intensity (µm/m)²/Hz

y = 0.00008x2 - 0.00191x + 0.01465

0.14 R² = 0.9951
0.12
0.10
0.08
0.06
0.04
0.02
0.00
0 10 20 30 40 50 60
Flow Rate (l/min)

Figure 8-6: Peak area vs flow rate at 60ºC with quadratic fit line measured via the fiberglass sensing pad.

54
 Peak Area vs. Flow Rate, 40°C
0.45

Micro-Strain Intensity (µm/m)²/Hz

y = 0.0002x2 - 0.0033x + 0.0192
0.40
R² = 0.9927
0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
0 10 20 30 40 50 60
Flow Rate (l/min)

Figure 8-7: Peak area vs flow rate at 40ºC with quadratic fit line measured via the aerogel sensing pad, location 1.

Peak Area vs. Flow Rate, 60°C

0.50
y = 0.0002x2 - 0.0051x + 0.0465
Micro-Strain Intensity (µm/m)²/Hz

0.45
R² = 0.9716
0.40
0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
0 10 20 30 40 50 60 70
Flow Rate (l/min)

Figure 8-8: Maximum FIV peak area vs flow rate at 60ºC with quadratic fit line measured via the aerogel sensing pad, location

1.

55
 Peak Area vs. Flow Rate, 40°C
0.35

Micro-Strain Intensity (µm/m)²/Hz

0.30 y = 0.0002x2 - 0.0045x + 0.0323
R² = 0.9818
0.25

0.20

0.15

0.10

0.05

0.00
0 10 20 30 40 50 60
Flow Rate (l/min)

Figure 8-9: Peak area vs flow rate at 40ºC with quadratic fit line measured via the aerogel sensing pad, location 2.

Peak Area vs. Flow Rate, 60°C

0.35
Micro-Strain Intensity (µm/m)²/Hz

y = 0.0002x2 - 0.0057x + 0.0499

0.30
R² = 0.9923
0.25

0.20

0.15

0.10

0.05

0.00
0 10 20 30 40 50 60 70
Flow Rate (l/min)

Figure 8-10: Peak area vs flow rate at 60ºC with quadratic fit line measured via the aerogel sensing pad, location 2.

56
 Peak Area vs. Flow Rate, 40°C
0.30
y = 0.0001x2 - 0.003x + 0.0153

Micro-Strain Intensity (µm/m)²/Hz

R² = 0.9821
0.25

0.20

0.15

0.10

0.05

0.00
0 10 20 30 40 50 60
Flow Rate (l/min)

Figure 8-11: Maximum FIV peak area vs flow rate at 40ºC with quadratic fit line measured via the aerogel sensing pad, location

3.

Peak Area vs. Flow Rate, 60°C

0.25
y = 0.0001x2 - 0.0030x + 0.0185
Micro-Strain Intensity (µm/m)²/Hz

R² = 0.9949
0.20

0.15

0.10

0.05

0.00
0 10 20 30 40 50 60
Flow Rate (l/min)

Figure 8-12: Maximum FIV peak area vs flow rate at 60ºC with quadratic fit line measured via the aerogel sensing pad, location

3.

A quadratic relationship between peak area and flow rate is again universally seen amongst all

sensing pads with strong quadratic correlation. The trend between singular peak area and flow

rate is also seen across the full array of tested temperatures as observed via the coil sensors.

57
Table 8-3 through Table 8-6 show the calculated R² and standard deviation for the fiberglass

sensing pad and the three locations tested via the aerogel sensing pad.

Table 8-3: Fiberglass sensing pad peak area R² and standard deviation across tested temperature array
Flow Temperature (ºC) R² σ (l/min)

15 0.9977 1.30
20 0.9926 2.85
25 0.9745 2.47
30 0.9830 2.47
35 0.9904 2.95
40 0.9923 1.34
45 0.9930 1.50
50 0.9978 1.74
55 0.9979 1.47
60 0.9952 1.67
Average 0.9914 1.98

Table 8-4: Aerogel sensing pad peak area R² and standard deviation across tested temperature array, location 1
Flow Temperature (ºC) R² σ (l/min)

15 0.9531 2.33
20 0.9954 1.90
25 0.9866 2.29
30 0.9912 2.69
35 0.9899 1.32
40 0.9953 1.58
45 0.9942 1.11
50 0.9949 1.07
55 0.9728 1.95
60 0.9716 2.88
Average 0.9845 1.91

58
Table 8-5: Aerogel sensing pad peak area R² and standard deviation across tested temperature array, location 2
Flow Temperature (ºC) R² σ (l/min)

15 0.9897 2.16
20 0.9839 2.91
25 0.9850 3.10
30 0.9930 1.03
35 0.9986 1.70
40 0.9912 1.86
45 0.9960 2.31
50 0.9870 2.23
55 0.9968 2.42
60 0.9923 2.85
Average 0.9914 2.26

Table 8-6: Aerogel sensing pad peak area R² and standard deviation across tested temperature array, location 3
Flow Temperature (ºC) R² σ (l/min)

15 0.9842 2.98
20 0.9937 2.26
30 0.9923 1.76
40 0.9886 2.69
50 0.9911 2.17
60 0.9949 2.43
Average 0.9908 2.38

In all cases, the calculated average R² value is above 0.98 and average flow rate standard

deviation is below 2.5 l/min. Instances of standard deviation above 2 l/min contain high

individual variance at the floor of peak measurability 10-20 l/min. Standard deviation was

primarily attributed to high individual variance at the lower end of peak measurability 5-20

l/min. Numerically this can be attributed to the limitations of quadratic curve fitting, additionally

it can also be experimentally explained as a higher ratio of the total sensed signal at the central

59
frequency originating from non-FIV noise. Additional error in peak area cases can be attributed

to peak shape inconsistency in individual flow tests stemming from noise in the same frequency

band as the FIVs.

8.2 Flow Temperature

The relationship between maximum peak area and flow temperature at constant flow rates was

also measured. The y-axis scale in this section is smaller than that in 1.8.1 to enlarge variation.

Figure 8-13 and Figure 8-14 below show the graphs of calculated peak area and flow

temperature for 15 and 50 l/min for the downstream coil respectively. Figure 8-15 and Figure

8-16 show the same data for the elbow coil.

Peak Area vs. Flow Temperature, 15 l/min

0.00045
Micro-Strain Intensity (µm/m)²/Hz

0.00040
0.00035
0.00030
0.00025
0.00020
0.00015
0.00010
0.00005
0.00000
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 8-13: Peak area vs flow temperature at 15 l/min measured via the downstream coil.

60
 Peak Area vs. Flow Temperature, 50 l/min
0.02

Micro-Strain Intensity (µm/m)²/Hz

0.018
0.016
0.014
0.012
0.01
0.008
0.006
0.004
0.002
0
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 8-14: Peak area vs flow temperature at 50 l/min measured via the downstream coil.

Peak Area vs. Flow Temperature, 15 l/min

0.0006
Micro-Strain Intensity (µm/m) ²/Hz

0.0005

0.0004

0.0003

0.0002

0.0001

0.0000
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 8-15: Peak area vs flow temperature at 15 l/min measured via the elbow coil.

61
 Peak Area vs. Flow Temperature, 50 l/min
0.025

Micro-Strain Intensity (µm/m) ²/Hz

0.02

0.015

0.01

0.005

0
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 8-16: Peak area vs flow temperature at 50 l/min measured via the elbow coil.

No apparent trend was noticed in charting peak area and flow temperature at any individual flow

rate. Deviation in this relationship can be attributed to randomness in flow due to lack of thermal

effect on the thermoplastic elastomer fiber jacket to contribute to increased strain.

The relationship between maximum peak area and flow temperature at constant flow rates was

also observed in the fiberglass sensing pad. Figure 8-17 and Figure 8-18 below show the

graphs of calculated peak area and flow temperature for 15 and 50 l/min, respectively, in the

fiberglass sensing pad.

62
 Peak Area vs. Flow Temperature, 15 l/min
0.0040

Micro-Strain Intensity (µm/m) ²/Hz

0.0035
0.0030
0.0025
0.0020
0.0015
0.0010
0.0005
0.0000
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 8-17: Peak area vs flow temperature at 15 l/min measured via the fiberglass sensing pad.

Peak Area vs. Flow Temperature, 50 l/min

0.14
Micro-Strain Intensity (µm/m) ²/Hz

0.12

0.1

0.08

0.06

0.04

0.02

0
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 8-18: Peak area vs flow temperature at 50 l/min measured via the fiberglass sensing pad.

Again in the fiberglass sensing pad case, a positive trend is seen between peak area and flow

temperature at individual flow rate. This can likely be attributed to the direct contact that the

silicone sealant has with the pipe surface. Due to the sealant’s low coefficient of thermal

expansion, the elevated external pipe temperature is easily transferred to the face of the sensing

63
pad, adding thermally induced mechanical strain to the fiber. Linear fitting for tests plotting peak

intensity and flow temperature for the fiberglass sensing pad yields R² values of 0.70-0.95, with

high deviation between flow rate sets unlike that of the relationship between peak area and flow

rate.

The relationship between peak area and flow temperature was measured on the aerogel

sensing pad in three locations. Figure 8-19 and Figure 8-20 below show the graphs of

calculated peak intensity and flow temperature for 15 and 50 l/min for the support-distant

downstream aerogel pad, respectively. Figure 8-21 and Figure 8-22 show the same data for the

aerogel sensing pad in the support-adjacent case.

Peak Area vs. Flow Temperature, 15 l/min

0.018
Micro-Strain Intensity (µm/m) ²/Hz

0.016
0.014
0.012
0.010
0.008
0.006
0.004
0.002
0.000
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 8-19: Peak area vs flow temperature at 15 l/min measured via the aerogel sensing pad, location 1.

64
 Peak Area vs. Flow Temperature, 50 l/min
0.35

Micro-Strain Intensity (µm/m) ²/Hz

0.3

0.25

0.2

0.15

0.1

0.05

0
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 8-20: Peak area vs flow temperature at 50 l/min measured via the aerogel sensing pad, location 1.

Peak Area vs. Flow Temperature, 15 l/min

0.006
Micro-Strain Intensity (µm/m) ²/Hz

0.005

0.004

0.003

0.002

0.001

0.000
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 8-21: Peak area vs flow temperature at 15 l/min measured via the aerogel sensing pad, location 2.

65
 Peak Area vs. Flow Temperature, 50 l/min
0.25

Micro-Strain Intensity (µm/m) ²/Hz

0.2

0.15

0.1

0.05

0
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 8-22: Peak area vs flow temperature at 50 l/min measured via the aerogel sensing pad, location 2.

In these cases, the 15 l/min tests are inconsistent as the support-distant case shows a weak

positive trend as flow temperature increases. In the support-adjacent case, there is no apparent

trend. In the 50 l/min tests, there is no apparent trend. In these cases as the aerogel sensing pad is

affixed onto a shell of insulation about 5 inches thick, the face of the sensing pad is sufficiently

insulated from the heat of the pipe exterior surface. Since the sensing pad is fully insulated from

increased temperature and subsequent thermally induced mechanical stress, any relationship in

with peak area and flow temperature calculated via the aerogel sensing pad is fully intrinsic to

flow.

9. Peak Central Frequency Relationships

9.1 Flow Rate
The relationship between peak central frequency and flow rate was measured for all sensors.

Figure 9-1 shows the downstream coil’s calculated peak central frequency as flow rate varies at a

66
constant flow temperature of 40ºC at a step rate of 1 l/min. Figure 9-2 shows the same

relationship for the elbow coil.

Central Frequency vs. Flow Rate, 40°C

72
71
70
Central Frequency (Hz)

69
68
67
66
65
64
63
62
61
0 10 20 30 40 50 60
Flow Rate (l/min)

Figure 9-1: Peak central frequency vs flow rate at 40ºC measured via the downstream coil.

Central Frequency vs. Flow Rate, 40°C

73
71
Central Frequency (Hz)

69
67
65
63
61
59
57
55
0 10 20 30 40 50 60
Flow Rate (l/min)

Figure 9-2: Peak central frequency vs flow rate at 40ºC measured via the elbow coil.

In both cases there is an upward exponential decay trend until a local flow rate threshold, where

after this threshold central frequency maintains no apparent trend. In both instances, 22 l/min

67
appears to be the approximate broad trend threshold. This can be explained by signal smoothing

in the flow rate range resulting in a lower signal-to-noise ratio (SNR) due to weak FIVs. The

relationship between peak central frequency and flow rate was also measured in the sensing

pads. Figure 9-3 shows this relationship for the fiberglass sensing pad.

Central Frequency vs. Flow Rate, 40°C

82
80
Central Frequency (Hz)

78
76
74
72
70
68
66
0 10 20 30 40 50 60
Flow Rate (l/min)

Figure 9-3: Peak central frequency vs flow rate at 40ºC measured via the fiberglass sensing pad.

No apparent trend was detected. Higher average calculated central frequency can be contributed

to the affixation method of the sensing pad creating varied interfacial bonding. Figure 9-4

through Figure 9-6 show the relationship between peak central frequency and flow rate for the

aerogel sensing pad in the support-distant, elbow-adjacent, and support-adjacent locations,

respectively.

68
 Central Frequency vs. Flow Rate, 40°C
76

74

Central Frequency (Hz) 72

70

68

66

64

62

60
0 10 20 30 40 50 60
Flow Rate (l/min)

Figure 9-4: Peak central frequency vs flow rate at 40ºC measured via the aerogel sensing pad, location 1.

Central Frequency vs. Flow Rate, 40°C

79
78
Central Frequency (Hz)

77
76
75
74
73
72
71
0 10 20 30 40 50 60
Flow Rate (l/min)

Figure 9-5: Peak central frequency vs flow rate at 40ºC measured via the aerogel sensing pad, location 2.

69
 Central Frequency vs. Flow Rate, 40°C
74
73

Central Frequency (Hz)

72
71
70
69
68
67
66
65
0 10 20 30 40 50 60
Flow Rate (l/min)

Figure 9-6: Peak central frequency vs flow rate at 40ºC measured via the aerogel sensing pad, location 3.

In the elbow-adjacent and support-adjacent cases, no apparent trend was seen. In the support-

distant case, measured central frequency steadily increases throughout the range of tested flow

rates. This can be attributed to gradually increasing response volume in a frequency band

adjacent to frequency band containing the FIV response. Figure 9-7 illustrates the raw frequency

domain as flow increases for the aerogel pad corresponding to Figure 9-4.

70
Figure 9-7: Measured unsmoothed frequency domain data for the aerogel pad in location 1 at a constant 40°C at 15 l/min (top),

30 l/min (middle), and 45 l/min (bottom).

As flow rate increases, the increasing signal intensity in the 74-83 Hz range gradually increases

the smoothed central frequency calculation, despite the primary signal remaining between 68-70

Hz.

9.2 Flow Temperature

The relationship between peak central frequency and flow temperature at constant flow rates

was also measured. Figure 9-8 and Figure 9-9 below show the graphs of calculated peak

71
central frequency and flow temperature for 15 and 50 l/min for the downstream coil

respectively. Figure 9-10 and Figure 9-11 show the same data for the elbow coil.

Central Frequency vs. Flow Temperature, 15 l/min

68

66
Central Frequency (Hz)

64

62

60

58

56
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 9-8: Peak central frequency vs flow temperature at 15 l/min measured via the downstream coil.

Central Frequency vs. Flow Temperature, 50 l/min

73

72
Central Frequency (Hz)

71

70

69

68

67

66
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 9-9: Peak central frequency vs flow temperature at 50 l/min measured via the downstream coil.

72
 Central Frequency vs. Flow Temperature, 15 l/min
69
68

Central Frequency (Hz)

67
66
65
64
63
62
61
60
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 9-10: Peak central frequency vs flow temperature at 15 l/min measured via the elbow coil.

Central Frequency vs. Flow Temperature, 50 l/min

72
71
Central Frequency (Hz)

70
69
68
67
66
65
64
63
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 9-11: Peak central frequency vs flow temperature at 50 l/min measured via the elbow coil.

In the 15 l/min case for both sensors, there is high variability in frequency calculations between

flow temperatures, and no apparent trend in central frequency as temperature increases. In the 50

l/min case for both sensors, another upward exponential decay trend emerges that is increasing

through the complete temperature domain in the downstream coil case, whereas in the elbow coil

73
a threshold is reached at 40ºC. In this case after 40ºC, central frequency remains within the 70-71

Hz range rather than increasing at the rate seen before 40ºC. Figure 9-12 shows the raw

frequency domain of the downstream coil at 50 l/min as temperature increases.

Figure 9-12: Measured unsmoothed frequency domain data for the downstream coil at a constant 50 l/min at 15°C (top), 35°C

(middle), and 55°C (bottom).

In the downstream case, the raw shape of the peak remains approximately the same and shifts

to higher frequencies as flow temperature increases. This behavior was also similarly seen in

74
the elbow coil under the same test conditions. The relationship between peak central

frequency and flow temperature at constant flow rates was also calculated in the fiberglass and

aerogel sensing pads. Figure 9-13 and Figure 9-14 below show the graphs of measured peak

area and flow temperature for 15 and 50 l/min, respectively, in the fiberglass sensing pad.

Central Frequency vs. Flow Temperature, 15 l/min

77
76
Central Frequency (Hz)

75
74
73
72
71
70
69
68
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 9-13: Peak central frequency vs flow temperature at 15 l/min measured via the fiberglass sensing pad.

Central Frequency vs. Flow Temperature, 50 l/min

78
77.5
Central Frequency (Hz)

77
76.5
76
75.5
75
74.5
74
73.5
73
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 9-14: Peak central frequency vs flow temperature at 50 l/min measured via the fiberglass sensing pad.

75
No apparent trend was seen in either the 15 l/min or the 50 l/min case. The relationship

between peak central frequency and flow temperature at constant flow rates was also

measured in the fiberglass and aerogel sensing pads. Figure 9-15 and Figure 9-16 below show

the graphs of calculated peak area and flow temperature for 15 and 50 l/min, respectively, in

the aerogel sensing pad at the support-distant location. Figure 9-17 and Figure 9-18 show the

same data for the aerogel sensing pad at the elbow-adjacent location.

Central Frequency vs. Flow Temperature, 15 l/min

72.5
72
71.5
Central Frequency (Hz)

71
70.5
70
69.5
69
68.5
68
67.5
67
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 9-15: Peak central frequency vs flow temperature at 15 l/min measured via the aerogel sensing pad, location 1.

76
 Central Frequency vs. Flow Temperature, 50 l/min
75

74

Central Frequency (Hz) 73

72

71

70

69

68
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 9-16: Peak central frequency vs flow temperature at 50 l/min measured via the aerogel sensing pad, location 1.

Central Frequency vs. Flow Temperature, 15 l/min

78

76
Central Frequency (Hz)

74

72

70

68

66

64
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 9-17: Peak central frequency vs flow temperature at 15 l/min measured via the aerogel sensing pad, location 2.

77
 Central Frequency vs. Flow Temperature, 50 l/min
77.5

77
Central Frequency (Hz)
76.5

76

75.5

75

74.5
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 9-18: Peak central frequency vs flow temperature at 50 l/min measured via the aerogel sensing pad, location 2.

No strong apparent trend was observed in any cases using the sensing pad designs. Weak upward

exponential decay relationships may be examined in Figure 9-16 and Figure 9-17 similar to those

seen in Figure 9-1 and Figure 9-2 in relation to flow rate and Figure 9-11 in relation to flow

temperature. As this relationship is seen in central frequency relationships with both flow and

temperature, there appears to be a relationship between the kinematic viscosity of a fluid and the

central frequency at which FIVs are centrally located.

10. Peak Full Width at Half Maximum Relationships

10.1 Flow Rate
The relationship between peak FWHM and flow rate was measured for all sensors. Figure 10-1

shows the downstream coil’s calculated peak FWHM as flow rate varies at a constant flow

temperature of 40ºC at a step rate of 1 l/min. Figure 10-2 shows the same relationship for the

elbow coil.

78
 FWHM vs. Flow Rate, 40°C
32

30

28
FWHM (Hz)

26

24

22

20
0 10 20 30 40 50 60
Flow Rate (l/min)

Figure 10-1: Peak FWHM vs flow rate at 40ºC measured via the downstream coil.

FWHM vs. Flow Rate, 40°C

34

32

30
FWHM (Hz)

28

26

24

22

20
0 10 20 30 40 50 60
Flow Rate (l/min)

Figure 10-2: Peak FWHM vs flow rate at 40ºC measured via the elbow coil.

No apparent trend in FWHM and flow rate is noticed in either case. Individual instances of

narrowing are observed but not sustained, suggesting inconsistent flow causing a lack of

consistency in sensed FIVs. These instances correlate highly with high standard deviation in

peak area calculations. The relationship between peak FWHM and flow rate was also calculated

79
in the sensing pads. Figure 10-3 shows this relationship for the fiberglass sensing pad at 40°C at

a step rate of 1 l/min.

FWHM vs. Flow Rate, 40°C

32

30

28
FWHM (Hz)

26

24

22

20

18
0 10 20 30 40 50 60
Flow Rate (l/min)

Figure 10-3: Peak FWHM vs flow rate at 40ºC measured via the fiberglass sensing pad.

Again, no apparent trend is noticed in FWHM as flow rate increases. FHWM appears to vary

between tests with large disparity without an apparent relationship. The relationship between

FWHM and flow rate was also measured for the aerogel sensing pad in three locations. Figure

10-4 through Figure 10-6 show the relationship between peak FWHM and flow rate for the

aerogel sensing pad in the support-distant, elbow-adjacent, and support-adjacent locations,

respectively.

80
 FWHM vs. Flow Rate, 40°C
34

32
FWHM (Hz) 30

28

26

24

22

20
0 10 20 30 40 50 60
Flow Rate (l/min)

Figure 10-4: Peak FWHM vs flow rate at 40ºC measured via the aerogel sensing pad, location 1.

FWHM vs. Flow Rate, 40°C

29

27

25
FWHM (Hz)

23

21

19

17
0 10 20 30 40 50 60
Flow Rate (l/min)

Figure 10-5: Peak FWHM vs flow rate at 40ºC measured via the aerogel sensing pad, location 2.

81
 FWHM vs. Flow Rate, 40°C
33
31
29
27
FWHM (Hz)

25
23
21
19
17
15
0 10 20 30 40 50 60
Flow Rate (l/min)

Figure 10-6: Peak FWHM vs flow rate at 40ºC measured via the aerogel sensing pad, location 3.

Again, no apparent trend is noticed in FWHM and flow rate. Individual cases of strong

narrowing are observed due to standard flow inconsistency.

10.2 Flow Temperature

The relationship between peak FWHM and flow temperature at constant flow rates was also

measured. Figure 10-7 and Figure 10-8 below show the graphs of calculated peak FWHM and

flow temperature for 15 and 50 l/min for the downstream coil respectively. Figure 10-9 and

Figure 10-10 show the same data for the elbow coil.

82
 FWHM vs. Flow Temperature, 15 l/min
30
29
28
FWHM (Hz) 27
26
25
24
23
22
21
20
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 10-7: Peak FWHM vs flow temperature at 15 l/min measured via the downstream coil.

FWHM vs. Flow Temperature, 50 l/min

30
29
28
27
FWHM (Hz)

26
25
24
23
22
21
20
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 10-8: Peak FWHM vs flow temperature at 50 l/min measured via the downstream coil.

83
 FWHM vs. Flow Temperature, 15 l/min
29
28
27
FWHM (Hz) 26
25
24
23
22
21
20
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 10-9: Peak FWHM vs flow temperature at 15 l/min measured via the elbow coil.

FWHM vs. Flow Temperature, 50 l/min

31
30
29
28
FWHM (Hz)

27
26
25
24
23
22
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 10-10: Peak FWHM vs flow temperature at 50 l/min measured via the elbow coil.

No apparent trend in FWHM and flow temperature is seen in either case. High variation is

observed in nearby temperatures tested. The relationship between peak FWHM and flow

temperature was also measured in the sensing pads. Figure 10-11 and Figure 10-12 below show

84
the graphs of calculated peak FWHM and flow temperature for 15 and 50 l/min for the fiberglass

sensing pad respectively.

FWHM vs. Flow Temperature, 15 l/min

33
31
29
FWHM (Hz)

27
25
23
21
19
17
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 10-11: Peak FWHM vs flow temperature at 15 l/min measured via the fiberglass sensing pad.

FWHM vs. Flow Temperature, 50 l/min

30
29
28
27
FWHM (Hz)

26
25
24
23
22
21
20
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 10-12: Peak FWHM vs flow temperature at 50 l/min measured via the fiberglass sensing pad.

No apparent trend in FWHM and flow temperature is noticed in either case. High variation is

observed in nearby temperatures tested. Figure 10-13 and Figure 10-14 show the relationship

85
between peak FWHM and flow temperature for the aerogel sensing pad in the support-distant

location at 15 l/min and 50 l/min respectively. Figure 10-15 and Figure 10-16 show the same

relationship in the elbow-adjacent case.

FWHM vs. Flow Temperature, 15 l/min

28
27
26
25
FWHM (Hz)

24
23
22
21
20
19
18
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 10-13: Peak FWHM vs flow temperature at 15 l/min measured via the aerogel sensing pad, location 1.

FWHM vs. Flow Temperature, 50 l/min

29
28
27
26
FWHM (Hz)

25
24
23
22
21
20
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 10-14: Peak FWHM vs flow temperature at 50 l/min measured via the aerogel sensing pad, location 1.

86
 FWHM vs. Flow Temperature, 15 l/min
28

27

26
FWHM (Hz)

25

24

23

22
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 10-15: Peak FWHM vs flow temperature at 15 l/min measured via the aerogel sensing pad, location 2.

FWHM vs. Flow Temperature, 50 l/min

28

27.5

27
FWHM (Hz)

26.5

26

25.5

25
0 10 20 30 40 50 60 70
Flow Temperature (°C)

Figure 10-16: Peak FWHM vs flow temperature at 50 l/min measured via the aerogel sensing pad, location 2.

No apparent trend in FWHM and flow temperature is noticed in either case. High variation is

observed in nearby temperatures tested.

87
11. Discussion
In all sensors that were tested, the correlation between smoothed peak intensity and flow rate

was calculated to be strong and quadratic as other similar FIV characteristics have been

reported. A significant majority of tests yielded an R² value over 0.99, and standard deviation

in flow rate calculations respective of the measured trendline was under 2 l/min in all but one

case. Much of the standard deviation summation of squared differences stems from peak

intensity calculations at 20 l/min or lower. At 25 l/min and above, squared difference

calculations are reduced significantly, and consequently standard deviation and overall sensor

precision improve significantly as well. The phenomenon of high variability in low flow

velocities can likely be attributed to a lower signal-to-noise ratio at flow rates with weaker

FIVs . In the relationship between peak intensity and flow temperature, only the fiberglass

sensing pad measured a consistent positive trend as temperature increased. This is likely due

to the direct bonding of the fiber face with the external pipe surface causing thermal strain that

trends linearly with flow temperature.

Peak area when calculated against flow rate closely followed the same trend as observed with

peak intensity. Coefficient of determination measures of peak area and flow rate were slightly

lower than that of peak intensity with all tests averaging between 0.98 < R² < 0.992. Standard

deviation measures were observably higher with all sensor-averaged calculations marking

between 1.85 < σ < 2.40 l/min. This trend degradation can be attributed to minor shape

inconsistencies such as small-scale peak width randomness. In the absence of large-scale peak

narrowing or broadening with a trend as flow rate increases, peak area is observed to be

primarily a function of peak intensity with minor shape inconsistencies to the calculation

88
which subtract from its correlation with flow rate. In turn, the peak area relationship with flow

temperature mirrored that of the peak intensity and flow temperature relationship for all tested

sensors.

The relationship between central frequency and flow rate follows an exponential decay

upward trend in instances with singular raw FIV-sourced peaks. In these instances, central

frequency shifted higher until a flow rate threshold where after the threshold central frequency

would vary on the small-scale. Figure 9-4 and Figure 9-7 show an instance where multiple

adjacent frequency domain peaks induced from FIVs were smoothed to create another

exponential decay upwards trend, but at a much wider flow rate range. With this observation,

possibilities of vibration characteristics manipulation arise for flow rate measurement via n-

number adjacent raw frequency domain peaks induced from FIVs. In measuring the

relationship between flow temperature and central frequency, both coil sensors showed

exponential decay upward trends with flow temperature with increasing strength as flow rate

increased. The same relationship in sensing pads showed either random variation or a weak

exponential decay upward trend. Similar to the relationship seen in Figure 9-4 and Figure 9-7

for variable flow rate, possibilities arise for widening the decay range in this instance to better

sense flow temperature by manipulating vibration characteristics above a local velocity

threshold. In the case of the sensing pads, central frequency itself was seen to vary by

affixation tightness of the hose clamps used, which explains the difference in central

frequency between the sensing pads and coil sensors.

As flow rate increased, FWHM measurements were without an apparent trend. Instances such

as those in Figure 10-5 were relatively stable with individual instances of narrowing. These

89
instances correlated with similar drops in peak area, but not in peak intensity. Instances where

this occurs may potentially be used to identify instances of irregular flow, or flow with

particulate, however there are also cases such as Figure 10-3 where FWHM is largely random

throughout the flow rate range. The aerogel sensing pad relationship with FWHM and flow

rate appears relatively constant, while the same relationship in that of the fiberglass sensing

pad and the coil sensors appears more chaotic. Further optimization of the aerogel sensing pad

and its affixation method may lead to more consistent FWHM measurements. No apparent

trend was noticed in the relationship between FWHM and flow temperature in any sensor.

90
 Conclusions

In this research, a DAS technique previously used to measure FIVs as a function of volumetric

flow rate was modified to examine the potential for simultaneous flow rate and temperature

measurement via acoustic sensors immune to electromagnetic interference with sensitivity at the

picometer scale. This research assisted in furthering the understanding of the relationship

between FIVs and a variety of frequency-domain characteristics yielded from two unique

acoustic sensor styles. A discussion of current methods used to sense FIVs and similar flow

characteristics as a function of flow is reviewed with limitations examined. The benefits of using

a fiber optic DAS sensor in order to measure FIVs as a function of flow were discussed. The

benefits of improved sensitivity and potential for temperature sensing were discussed.

Sensing pad prototypes were fabricated to increase acoustic sensitivity compared to traditional

fiber coils. Flow rate was measured via FIV-induced strain in the frequency domain using peak

intensity with a strong quadratic trend and low average standard deviation in all sensors.

Frequency domain peak intensity was observed to positively correlate with flow temperature in

instances where the fiber in a sensor was heated by high pipe surface temperature, though no

trend was noticed when the sensor was well insulated. The same trends are observed in

frequency domain peak area due to lack of peak narrowing as flow conditions change. The

quadratic trend between peak area and flow rate is calculated to be slightly weaker than the

corresponding relationship with peak intensity for all sensors. Standard deviation measurements

are likewise higher in peak area measurements when compared to peak intensity in all sensors.

Frequency domain peak central frequency appears to rise in low flow to a threshold and maintain

a constant frequency range as flow rate increases via an upward exponential decay relationship.

91
It was noticed that multiple FIV-induced raw frequency domain peaks can be numerically

smoothed to reveal a trend in central frequency and flow rate, in this case exponential decay

upward. A positive trend was seen in central frequency and flow temperature in coil sensors at

higher tested flow rates, though the trend subsides as flow rate decreases. No apparent trend was

noticed in FWHM as either flow rate or temperature increased. Sensors adjacent to the pipe

elbow were measured to yield higher strain at the same flow rate and temperature than

downstream counterpart sensors when similarly adjacent to a pipe support, though sensors

distant from supports regardless of distance to the pipe elbow were measured to yield higher

strain due to higher deflection according to beam theory.

The new contributions to the body of literature in this field of noninvasive flow measurement are

summarized below:

• Expanded on methods used to measure flow rate via FIVs using optical fiber sensors

utilizing DAS.

o Successfully confirmed the quadratic relationship between frequency domain

peak intensity generated from FIVs with noninvasive fiber optic based acoustic

sensors.

o Determined the relationship between peak intensity and area in the absence of

acoustic frequency domain peak narrowing in regard to FIV flow rate

measurement.

o Determined the relationship of numerically smoothed frequency domain central

frequency with flow rate in singular and multisource vibrational cases.

92
 o Obtained FWHM measurements revealing lack of trend in bandwidth as flow rate

changes

• Developed understanding of potential for flow temperature sensing using optical fiber

sensors utilizing DAS.

o Observed positive correlation in frequency domain peak intensity and flow

temperature in select sensing pad tests.

o Observed positive correlation in frequency domain peak area and flow

temperature in select sensing pad tests.

o Observed positive correlation in frequency domain peak central frequency and

flow temperature in select coil sensor tests.

o Observed a lack of trend in frequency domain FWHM with flow temperature in

all sensors.

• Produced prototype sensing pads to allow for acoustic strain measurement with

significantly increased sensitivity compared to traditional helically wrapped fiber.

o Observed lower measurable flow rate floor due to weak FIVs being more easily

received by sensing pad designs.

o Discovered tunable frequency reception stemming from affixation conditions.

93
 Future Work

This work was successful in demonstrating a new approach to noninvasive flow rate

measurement via DAS. An abundance of frequency domain strain relationships as FIV

characteristics varied were reported and present many unique applied sensing opportunities.

Future research made available by this study are below:

• Optimization of sensor pad design using alternate embedded fiber geometries and

superior coatings for smaller bend radii for sensitivity maximization.

• Exploration into flow temperature sensing within coil sensors at sufficiently high flow

rates.

• Further analysis of peak intensity and flow temperature relationship in instances of

sensing pad interfacial exposure to variable flow temperatures for flow temperature

measurement in instances of flow rate measurement that is not intensity-based.

• Investigation into adjacent raw FIV peak alteration as flow rate increases yielding

quantifiably significant upwards exponential decay as an alternative means of

noninvasive flow testing.

• Expansion of FIV sensing flexibility of application through optimized sensing pad

geometries.

94
 References

1. Senior JM. Optical Fiber Communications: Principles and Practice. 2nd ed. Hoboken
(NJ): Prentice Hall; 1992. p. 65-68, 330-331.
2. Simmons JH, Potter KS. Optical Materials. 1st ed. Cambridge (MA): Academic Press;
2000. p. 119-124.
3. Lee BH, Kim YH, Park KS, Eom JB, Kim JM, Rho BS, Choi HY. Interferometric Fiber
Optic Sensors. Sens 2012;12(3):2467-86. https://doi.org/10.3390/s120302467
4. Ferreira LA, Ribeiro ABL, Santos AL, Farahi F. Simultaneous measurement of
displacement and temperature using a low finesse cavity and a fiber Bragg grating. IEEE
Photon Technol Lett 1996;8(11):1519-21. https://doi.org/10.1109/68.541569
5. Kim DW, Shen F, Chen X, Wang A. Simultaneous measurement of refractive index and
temperature based on a reflection-mode long-period grating and an intrinsic Fabry-Perot
interferometer sensor. Opt Lett 2005;30(22):3000-2.
https://doi.org/10.1364/OL.30.003000
6. Wang X, Xu J, Zhu Y, Cooper KL, Wang A. All-fused-silica miniature optical fiber tip
pressure sensor. Opt Lett 2006;31(7):885-7. https://doi.org/10.1364/OL.31.000885
7. Cho JY, Lim JH, Lee KS. Optical fiber twist sensor with two orthogonally oriented
mechanically induced long-period grating sections. IEEE Photon Technol Lett
2005;17(2):453-5. https://doi.org/10.1109/LPT.2004.840073
8. Wang A, Xiao H, Wang J, Wang Z, Zhao W, May RG. Self-calibrated interferometric-
intensity-based optical fiber sensors. J Lightw Technol 2001;19(10):1495.
9. Ren Z, Li J, Zhu R, Cui K, He Q, Wang H. Phase-shifting optical fiber sensing with
rectangular-pulse binary phase modulation. Opt Lasers Eng 2018;100:170-5.
https://doi.org/10.1016/j.optlaseng.2017.08.010
10. Choi WS, Jo MS. Accurate Evaluation of Polarization Characteristics in the Integrated
Optic Chip for Interferometric Fiber Optic Gyroscope Based on Path-matched
Interferometry. J Opt Soc Korea 2009;13(4):439-44.
11. Jena J, Wassin S, Bezuidenhout L, Doucoure M, Gibbon T. Polarization-based optical
fiber acoustic sensor for geological applications. J Opt Soc Am B 2020;37(11):A147-
A153. https://doi.org/10.1364/JOSAB.396565
12. Lee MH, Jerng DW, Bang IC. Experimental Validation of Simulating Natural Circulation
of Liquid Metal Using Water. Nucl Eng Technol 2020;52(9):1963–73.
https://doi.org/10.1016/j.net.2020.03.005
13. Unalmis OH. The Use of Sound Speed in Downhole Flow Monitoring Applications. Proc
Mtgs Acoust 2015;23:045003. https://doi.org/10.1121/2.0000069
14. Blevins RD. Flow-Induced Vibration. New York: Van Nostrand Reinhold; 1977. p. 377.
15. Ting EC, Hosseinipour A. A Numerical Approach for Flow-Induced Vibration of Pipe
Structures. J Sound Vib 1983;88(3):289–98. https://doi.org/10.1016/0022-
460X(83)90689-2
16. Suradkar AE, Suryawanshi SR. Numerical analysis of fluid flow induce vibration of
pipes—A review. Int J Mod Trends Eng Res 2016 Apr;3(4):1058-62.

95
17. Dongyang, C, Abbas LK, Guoping W, Xiaoting R, Marzocca P. Numerical Study of
Flow-Induced Vibrations of Cylinders Under the Action of Nonlinear Energy Sinks
(NESs). Nonlinear Dyn 2018;94(2):925-57. https://doi.org/10.1007/s11071-018-4402-z
18. He, F, Dai H, Wang L. Vortex-Induced Vibrations of a Pipe Subjected to
Unsynchronized Support Motions. J Marine Sci Technol 2018;23(4):978-90.
https://doi.org/10.1007/s00773-017-0526-y
19. Zhu, H, Sun Z, Gao Y. Numerical Investigation of Vortex-Induced Vibration of a Triple-
Pipe Bundle. Ocean Eng 2017;142:204-16.
https://doi.org/10.1016/j.oceaneng.2017.07.019
20. Evans RP, Blotter JD, Stephens AG. Flow rate measurements using flow-induced pipe
vibration. J Fluids Eng 2004 Mar;126(2):280-5. https://doi.org/10.1115/1.1667882
21. Pittard MT, Evans RP, Maynes RD, Blotter JD. Experimental and numerical
investigation of turbulent flow induced pipe vibration in fully developed flow. Rev Sci
Instrum 2004 Jul;75(7):2393-401. https://doi.org/10.1063/1.1763256
22. Chande PK, Sharma PC. Ultrasonic flow velocity sensor based on picosecond timing
system. IEEE Trans Ind Electron 1986 May;33(2):162-5.
https://doi.org/10.1109/TIE.1986.350211
23. Wada S, Kikura H, Aritomi M, Mori M, Takeda Y. Development of pulse ultrasonic
Doppler method for flow rate measurement in power plant multilines flow rate
measurement on metal pipe. J Nucl Sci Technol 2004 Mar;41(3):339-46.
https://doi.org/10.1080/18811248.2004.9715493
24. Ricci S, Meacci V, Birkhofer B, Wiklund J. FPGA-Based System for In-Line
Measurement of Velocity Profiles of Fluids in Industrial Pipe Flow. IEEE Trans Ind
Electron 2017;64(5):3997-4005. https://doi.org/10.1109/TIE.2016.2645503
25. Campagna MM, Dinardo G, Fabbiano L, Vacca G. Fluid flow measurements by means of
vibration monitoring. Meas Sci Technol 2015;26(11). https://doi.org/10.1088/0957-
0233/26/11/115306
26. Wang T, Baker R. Coriolis Flowmeters: A Review of Developments Over the Past 20
Years, and an Assessment of the State of the Art and Likely Future Directions. Flow
Meas Instrum 2014;40:99-123. https://doi.org/10.1016/j.flowmeasinst.2014.08.015
27. Hu C, Zheng D, Fan S, Wiegerink RJ, Zhanshe G. Theoretical and Experimental
Research on the In-Plane Comb-Shaped Capacitor for Mems Coriolis Mass Flow
Sensor. Microsyst Technol 2016;22(4):747-55. https://doi.org/10.1007/s00542-015-2441-
7
28. Zeng Y, Groenesteijn J, Alveringh D, Wiegerink RJ, Lötters JC. Design, Fabrication, and
Characterization of a Micro Coriolis Mass Flow Sensor Driven by PZT Thin Film
Actuators. J Microelectromech Sys 2021 Dec;30(6)885-96.
https://doi.org/10.1109/JMEMS.2021.3107744
29. Schut T, Wiegerink RJ, Lötters JC. “μ -Coriolis Mass Flow Sensor with Resistive
Readout.” Micromach 2020;11(2):184 https://doi.org/10.3390/mi11020184
30. Lannes DP, Gama AL, Bento TFB. Measurement of flow rate using straight pipes and
pipe bends with integrated piezoelectric sensors. Flow Meas Instrum 2018 Apr;60:208-
16. https://doi.org/10.1016/j.flowmeasinst.2018.03.001

96
31. Keith W, Foley A, Cipolla K. Wavenumber-frequency analysis of turbulent wall pressure
fluctuations over a wide Reynolds number range of Turbulent Pipe Flows. OCEANS'11
MTS/IEEE KONA 2011:1-5. https://doi.org/10.23919/OCEANS.2011.6106964
32. Li Q, Xing J, Shang D, Wang Y. A flow velocity measurement method based on a PVDF
piezoelectric sensor. Sens 2019 Apr;19(7):1657. https://doi.org/10.3390/s19071657
33. Li, Q, Wang Y, Shang D, Tang R. A Flow Velocity Measurement Method Based on
Turbulent Pressure Fluctuation Characteristics. IEEE Sens J 2021 Oct;21(20):23162-73.
https://doi.org/10.1109/JSEN.2021.3107318
34. Medeiros KAR, Barbosa CRH, d’Almeida JRM, Ribeiro AS, de Paula IB. Flowmeter
based on a piezoelectric PVDF tube. Meas 2019 May;138:368-78.
https://doi.org/10.1016/j.measurement.2019.02.059
35. Turner RC, Fuierer PA, Newnham RE, Shrout TR. Materials for high temperature
acoustic and vibration sensors: A review. Appl Acoust 1994;41(4):299-324.
https://doi.org/10.1016/0003-682X(94)90091-4
36. Stajanca P, Chruscicki S, Homann T, Seifert S, Schmidt D, Habib A. Detection of leak-
induced pipeline vibrations using fiber-optic distributed acoustic sensing. Sens 2018
Aug;18(9):2841. https://doi.org/10.3390/s18092841
37. Zhao Y, Gu Y-F, Lv R-Q, Yang Y. A small probe-type flowmeter based on the
differential fiber Bragg grating measurement method. IEEE Trans Instrum Meas 2017
Mar;66(3):502-7. https://doi.org/10.1109/TIM.2016.2631779
38. Jiang T, Ren L, Jia Z-G, Li D-S, Li H-N. Pipeline internal corrosion monitoring based on
distributed strain measurement technique. Struct Control Health Monit 2017 Apr;24(11)
https://doi.org/10.1002/stc.2016
39. Wang Z, Zhang W, Huang W, Feng S, Li F, Optoelectronic hybrid fiber laser sensor for
simultaneous acoustic and magnetic measurement. Opt Express 2015;23(19):24383-9.
https://doi.org/10.1364/OE.23.024383
40. Liu P, Hao F, Li B, He S, Li D. A fiber-optic combined acoustic temperature and
pressure sensor for ocean observation. Proc Adv Sens Sys Appl VIII 2018;10821
https://doi.org/10.1117/12.2501322
41. Tsangouri E, Remy O, Boulpaep F, Verbruggen S, Livitsanos G, Aggelis D. Structural
health assessment of prefabricated concrete elements using acoustic emission: towards an
optimized damage sensing tool. Constr Build Mater 2019 May;206:261-9.
https://doi.org/10.1016/j.conbuildmat.2019.02.035
42. Yang W, Lang Z, Tian W. Condition monitoring and damage location of wind turbine
blades by frequency response transmissibility analysis,” IEEE Trans Ind Electron 2015
Oct;62(10):6558-64. https://doi.org/10.1109/TIE.2015.2418738
43. Worley R, Dewoolkar M, Xia T, Farrell R, Orfeo D, Burns D, Huston D. Acoustic
emission sensing for crack monitoring in prefabricated and prestressed reinforced
concrete bridge girders. J Bridge Eng 2019 Apr;24(4)
https://doi.org/10.1061/(ASCE)BE.1943-5592.0001377
44. Shetty N, Livitsanos G, Roy VN, Aggelis D, Hemelrijck VD, Wevers M, Verstrynge E.
Quantification of progressive structural integrity loss in masonry with acoustic emission-

97
 based damage classification. Constr Build Mater 2019:194;192-204.
https://doi.org/10.1016/j.conbuildmat.2018.10.215
45. Zou L, Ferrier GA, Afshar S, Yu Q, Chen L, Bao X. Distributed Brillouin scattering
sensor for discrimination of wall thinning defects in steel pipe under internal pressure.
Appl Opt 2004;43(7):1583-8. https://doi.org/10.1364/AO.43.001583
46. Zhang Q, Xiong Z. Crack Detection of Reinforced Concrete Structures Based on BOFDA
and FBG Sensors. Shock Vibrat 2018;2018:10. https://doi.org/10.1155/2018/6563537
47. Costley RD, Galan-Comas G. Distributed acoustic sensing for near-surface seismic
applications. J Acoust Soc Am 2018;144(3):1702 https://doi.org/10.1121/1.5067562
48. Niklès M, Ravet F. Depth and sensitivity. Nat Photon 2010 Jun;4(7):431-2.
https://doi.org/10.1038/nphoton.2010.149
49. Parker T, Shatalin S, Farhadiroushan M. Distributed acoustic sensing—A new tool for
seismic applications. First Break 2014 Feb;32(2010):61-9. https://doi.org/10.3997/1365-
2397.2013034
50. Jousset P, Reinsch T, Ryberg T, Blank H, Clarke A, Aghayev R, et al. Dynamic strain
determination using fibre-optic cables allows imaging of seismological and structural
features. Nat Commun 2018 Jun;9(1):2509. https://doi.org/10.1038/s41467-018-04860-y
51. Zhang ZZ, Bao X. Distributed optical fiber vibration sensor based on spectrum analysis
of polarization-OTDR system. Opt Express 2008 Jul;16(14):10240-7.
https://doi.org/10.1364/OE.16.010240
52. Jiang, K, Liang L, Hu C, Liu X. Fiber Bragg Grating Accelerometer-Based Nonintrusive
Flow Rate Measurements and Leak Detection. Appl Opt 2020;59(34):10680-7.
https://doi.org/10.1364/AO.408548
53. Li T, Fan C, Li H, He T, Qiao W, Shi Z, et al. Nonintrusive Distributed Flow Rate
Sensing System Based on Flow-Induced Vibrations Detection. IEEE Trans Instrum Meas
2021;70:1-8. https://doi.org/10.1109/TIM.2020.3036684
54. Li, C, Liu M-Y, Song H, Yang X-L, Wu Y-H. A Non-Invasive Measurement Method of
Pipeline Flow Rate Based on Dual FBG Sensors. IEEE Sens J 2022;22(6):5669-77.
https://doi.org/10.1109/JSEN.2022.3141733
55. Alva A, Ravish U, Gangadharan K. Experimental and finite elements analysis of a tuned
mass absorber for vibration isolation. J Eng Appl Sci 2011;6(1):77-83
56. Badardin NA, Mokhtar AA, Sallih N, Ishak MHI. Vibration Characteristics Study of
Different Pipe Length with Different End Conditions ARPN J Eng Appl Sci
2020;15(21):2364-74.
57. Culshaw B, Dakin J. Optical fiber sensors: Systems and applications. vol 2. Boston
(MA): Artech House; 1989.
58. Parker TR, Farhadiroushan M, Feced R, Handerek VA, Rogers AJ. Simultaneous
distributed measurement of strain and temperature from noise-initiated Brillouin
scattering in optical fibers. IEEE J Quantum Electron 1998 Apr;34(4):645-59.
https://doi.org/10.1109/3.663443
59. Senior JM. Optical Fiber Communications: Principles and Practice. 3rd ed. London
(UK): Pearson Education Limited; 2009. p. 86-168

98
60. Cherukupalli S, Anders GJ. Distributed Fiber Sensing and Dynamic Rating of Power
Cables. New York: Wiley-IEEE Press. 2020. p. 20-25.
61. Lewis MF. On Rayleigh Waves and Related Propagating Acoustic Waves. Rayleigh-
wave theory and application. Springer Series on Wave Phenomena 1985;2:37-58.
62. Nespereira M, Coelho JMP, Abreu M, Rebordao JM. Ultrashort Long-Period Fiber
Grating Sensors Inscribed on a Single Mode Fiber Using CO2 Laser Radiation. Hindawi
J Sens 2017;2017:1-9. https://doi.org/10.1155/2017/4196431
63. Schlangen S, Bremer K, Zheng Y, Roth B, Overmeyer L, Böhm S, et al. Long-Period
Gratings in Highly Germanium-Doped, Single-Mode Optical Fibers for Sensing
Applications. Sens 2018 Apr;18(5):1363. https://doi.org/10.3390/s18051363
64. Senior JM. Optical Fiber Communications: Principles and Practice. 3rd ed. London
(UK): Pearson Education Limited; 2009. p. 256-280
65. Yu C, Zhan Z, Lindsey NJ, Ajo‐Franklin JB, Robertson M. The potential of DAS in
Teleseismic Studies: Insights from the Goldstone Experiment. Geophys Res Lett
2019;46(3):1320-8. https://doi.org/10.1029/2018GL081195
66. Agnew DC. Strainmeters and Tiltmeters. Rev Geophys 1986;24(3):579-624.
https://doi.org/10.1029/RG024i003p00579
67. Langston C, Liang C. Gradiometry for polarized seismic waves. J Geophys: Solid Earth
2012;117(B6). https://doi.org/10.1029/2007JB005486
68. Daley T, Miller D, Dodds K, Cook P, Freifeld BM. Field testing of modular borehole
monitoring with simultaneous distributed acoustic sensing and geophone vertical seismic
profiles at Citronelle, Alabama. Geophys Prospect 2016;64(5):1318-34.
https://doi.org/10.1111/1365-2478.12324
69. Farhadiroushan M, Parker TR, Shatalin S, inventors; Method and apparatus for optical
sensing. International Patent WO2010136810A2. 2010 Dec 2.
70. Barnoski MK, Personick SD. Measurements in fiber optics. Proc IEEE 1978
Apr;66(4):429-40. https://doi.org/10.1109/PROC.1978.10935
71. Ishihara T, Li T. Numerical study on suppression of vortex-induced vibration of circular
cylinder by helical wires. J Wind Eng Ind Aerodyn 2020;197.
https://doi.org/10.1016/j.jweia.2019.104081
72. Navrose SM. Free vibrations of a cylinder: 3-D computations at Re=1000. J Fluids Struct
2013;41:109-118. https://doi.org/10.1016/j.jfluidstructs.2013.02.017
73. Sharma J. Theoretical and Experimental Modal Analysis of Beam. Lecture Notes in
Electrical Engineering. In: Ray K, Sharan S, Rawat S, Jain S, Srivastava S,
Bandyopadhyay A, editors. Engineering Vibration, Communication and Information
Processing. Lecture Notes in Electrical Engineering, vol 478. Singapore: Springer. 2019.
https://doi.org/10.1007/978-981-13-1642-5_16
74. Gere J, Timoshenko S. Mechanics of Materials. 4th ed. Boston (MA): PWS Publishing
Company: 1997. p. 549
75. Sollund H (University of Oslo, Department of Mathematics, Mechanics Division, Oslo,
Norway) Vedeld K (University of Oslo, Department of Mathematics, Mechanics
Division, Oslo, Norway). A semi-analytical model for free vibrations of free spanning
offshore pipelines. Research Report in Mechanics. Number 02. 2012.

99
76. Lin HY, Tsai YC. Free vibration analysis of a uniform multi-span beam carrying multiple
spring–mass systems. J Sound Vib 2007;302(3):442-56.
https://doi.org/10.1016/j.jsv.2006.06.080
77. Sutar S, Madabhushi R, Babu P. Finite Element Analysis of Piping Vibration with
Guided Supports. Intl J Mech Eng Automation 2016;3:96-106.
78. Kaneko S, Kobayashi R, Watanabe T, Nakamura T. Damping estimation of a loosely
supported single U-bend tube used in a steam generator. J Phys: Conf Ser 2012;382.
https://doi.org/10.1088/1742-6596/382/1/012048
79. Liu J, He X, Liu Q, Naibin J, Chen H. Vibration-modal analysis model for multi-span
pipeline with different support conditions. Comput Model New Technol 2014;18(5):14-8.
80. Sekacheva AA, Pastukhova LG, Alekhin VN, Noskov AS. Natural frequencies of a
vertical pipeline element. IOP Conf Ser: Mater Sci Eng 2018;481
https://doi.org/10.1088/1757-899X/481/1/012019
81. Jia Y, Madeira RE, Just-Agosto F. Finite Element Formulation and Vibration Frequency
Analysis of a Fluid Filled Pipe. ASME International Mechanical Engineering Congress
and Exposition: Recent Advances in Solids and Structures. 2005 Feb:127-32.
https://doi.org/10.1115/IMECE2005-81807
82. Tilson JD, Wachel JC. Vibrations In Reciprocating Machinery And Piping Systems,
Proc. 23rd Turbo Mach Symp; 1994. p. 243-272. https://doi.org/10.21423/R1808S
83. Nayak A, Das D. Experimental and Numerical Investigation of Flow Instability in a
Transient Pipe Flow. J Fluid Mech 2021;920(A39):1-23.
https://doi.org/10.1017/jfm.2021.460
84. Qu W, Zhang H, Li W, Sun W, Zhao L, Ning H. Influence of support stiffness on
dynamic characteristics of the hydraulic pipe subjected to basic vibration. Shock and Vib
2018;2018:1-8. https://doi.org/10.1155/2018/4035725
85. Wang D, Sun W, Gao Z, Li H. Vibration response analysis and hoop layout optimization
of spatial pipeline under random excitation. Aircr Eng Aerosp Technol 2022;94(8):1242-
51. https://doi.org/10.1108/AEAT-09-2021-0291
86. Sazesh S, Shams S. Vibration analysis of cantilever pipe conveying fluid under
distributed random excitation. J Fluids Struct 2019 May;87:84-101.
https://doi.org/10.1016/j.jfluidstructs.2019.03.018
87. Yano D, Ishikawa S, Tanaka K, Kijimoto S. Vibration analysis of viscoelastic damping
material attached to a cylindrical pipe by added mass and added damping. J Sound Vib.
2019;454:14-31. https://doi.org/10.1016/j.jsv.2019.04.023
88. Babu V. Fundamentals of Incompressible Fluid Flow. 1st ed. Springer Cham: 2021. p.
133-149. https://doi.org/10.1007/978-3-030-74656-8
89. Draad A, Kuiken G, Nieuwstadt F. Laminar–turbulent transition in pipe flow for
Newtonian and non-Newtonian fluids. J Fluid Mech 1998;377:267-312.
https://doi.org/10.1017/S0022112098003139
90. Wygnanski IJ, Champagne FH. On transition in a pipe. Part 1. The origin of puffs and
slugs and the flow in a turbulent slug. J Fluid Mech 1973;59:281-335.
https://doi.org/10.1017/S0022112073001576

100
91. Wygnanski I, Sokolov M, Friedman D. On transition in a pipe. Part 2. The equilibrium
puff. J Fluid Mech 1975;69:283-304. https://doi.org/10.1017/S0022112075001449
92. Rubin Y, Wygnanski I, Haritonidis JH. Further observations on transition in a pipe.
Laminar-Turbulent Transition, IUTAM Symp. Stuttgart/Germany 1979, (Eppler R, Faser
H, editors), Berlin: Springer; 1980. p. 17-26.
93. Nur A, Afrianita R, Ramli R. Effect of pipe diameter changes on the properties of fluid in
closed channels using Osborne Reynold Apparatus. IOP Conf Ser: Mater Sci Eng
2019;602. https://doi.org/10.1088/1757-899X/602/1/012058
94. Duggleby A, Ball KS, Schwaenen M. Structure and dynamics of low Reynolds number
turbulent pipe flow. Phil Trans R Soc A 2009;367:473-88.
https://doi.org/10.1098/rsta.2008.0241
95. Berger SA, Talbot L, Yao LS. Flow in curved pipes. Annu Rev Fluid Mech 1983;15:461-
512. https://doi.org/10.1146/annurev.fl.15.010183.002333
96. Ito H. Flow in curved pipes. JSME Intl J, 1987;30(262):543-52.
https://doi.org/10.1299/jsme1987.30.543
97. Shiraishi T, Watakabe H, Sago H, Konomura M, Ymaguchi A, Gujii T. Resistance and
fluctuating pressure of a large elbow in high Reynolds numbers. J Fluids Eng
2006;128:1063-73. https://doi.org/10.1115/1.2236126
98. Sentek Instrument [Internet]. [cited 2022 Dec 7]. Available from
https://docs.sentekinstrument.com/picoDAS%20Data%20Sheet.pdf

101