Note: Descriptions are shown in the official language in which they were submitted.
CA 02783089 2012-07-11
DAMAGE DETECTION IN PIPES AND JOINT SYSTEMS
FIELD OF THE INVENTION
The present invention concerns vibration-based damage detection systems and
process that are
particularly useful for assessing damage to joints of pipes or other
structures.
BACKGROUND OF THE INVENTION
Vital resources such as oil, gas, water, and other fluid materials are
transported through pipelines that
span various terrains. Pipelines are critical transport elements, and their
health and reliability through
their designed service life are important issues for design and maintenance
engineers. Factors such as
changes in the structural support due to ground movement, aging, corrosion,
impacts from heavy
construction equipment, pressure cracks, thermal expansion and contraction,
and defective welds can
severely impact the integrity of pipelines and the joints connection them and
thus, dramatically affect
the service life of pipeline segments. These factors can cause economic and
environmental problems
for industry stakeholders, including the producers, pipeline operators,
regulatory agencies, the public,
and others who are adversely affected by pipeline leakage. Thus, the creation
of a safe and reliable
process for detecting damage in pipelines is important.
Piping systems in refineries, power plants, petrochemical and other
industrial, commercial,
institutional and municipal settings such as water and sewage treatment
facilities are subject to various
stressors, which cause deterioration in pipes and joints. Degradation due to
corrosion is common and
inspections must be conducted on a continual basis as part of preventative
maintenance programs.
Piping systems may also require inspection after catastrophic events such as
fires, floods and
earthquakes that render infrastructure prone to damage. Some existing piping
inspection process
include; (a) ultrasonic (b) radiography (c) liquid penetration (d) magnetic
particle (e) eddy current
testing (f) acoustic emission and (g) vibration-based.
Traditional process for structural damage detection possess drawbacks such as
the necessity of
expensive equipment, poor sensitivity, and high operator related costs, while
other process are not
compatible with common structural pipeline materials such as various plastics
or metals. Some of these
process include visual inspection, impedance analysis, ultrasonic analysis,
acoustic
emission/transmission analysis, microwave analysis, magnetic flux leakage
analysis, thermography,
interferometry, and leaky lamb-wave analysis. Most of the above mentioned
process require expansive
equipment, high operator costs and involve elaborate and time-consuming
operations, except for some
of the vibration-based process.
2
CA 02783089 2012-07-11
DESCRIPTION OF PRIOR ART
Definitions
As used herein, "detecting" means identifying the presence of a characteristic
or an event. For
example, "detecting structural damage" or "detecting damage" means identifying
the presence of
damage (e.g., cracks, disbonding on a joint, weak sections of pipe wall, lose
fasteners (e.g., bolts,
screws, or the like) on mechanically fastened joints, corrosion, or the like).
In another example,
"detecting a vibrational response" means identifying the presence of a
vibrational response and
converting the vibrational response to a signal that can be transmitted,
stored, processed, or otherwise
manipulated.
As used herein, "relative damage" or "damage index" refers to the following
mathematical expression:
iHealthy tenst
DImn = m n __ m
'Healthy X100
'mn (1)
where DI nin is the measure of relative damage at a joint when a test
measurement is taken,
/ Healthy and / Test are the values of the integral of the processed
response signal of the vibrated joint
at the time of the damage detection and/or assessment, at the condition when
the structure/joint is
deemed "healthy" and a subsequent time, respectively. Several process of
signal processing can be
used to process the vibrational response of a joint and applied to the
expression in equation (1) to
determine the relative damage or damage index of the joint.
As used herein, "signal" refers to any time-varying quantity. Signals are
often scalar-valued functions
of time (waveforms), but may be vector valued and may be functions of any
other relevant independent
variable. For example, a signal produced from a sensor could be an electrical
quantity or effect, such as
current, voltage, or electromagnetic waves that can be varied in such a way as
to convey information.
As used herein, "processing", "signal processing" and other verb tenses of
"process" refer to the
analysis, interpretation, and manipulation of one or more signals. Processing
of signals, such as
electrical signals, e.g., voltage, current, or electromagnetic waves, includes
storage and reconstruction,
separation of information from noise, compression, and feature extraction.
Signal processing process
include Fourier Transformation processing (FT), Fast Fourier Transformation
processing (FFT),
Wavelet Transformation processing (WT), or Hilbert-Huang Transformation
processing (HHT),
without limitation.
As used herein, "noise" or "signal noise" refers to data without meaning; that
is, data that is not being
used to transmit a signal, but is simply produced as an unwanted by-product of
other activities.
As used herein, a "processor" refers to an electronic device designed to
accept data, perform prescribed
mathematical and logical operations, and display the results of these
operations. Examples of
processors include digital and/or analogue computers, Central processing Units
(CPUs),
microprocessors, and the like.
3
CA 02783089 2012-07-11
As used herein, "vibrating", "vibrate", "vibrated" or "vibrational" each
refers to mechanical oscillations
about an equilibrium point. The oscillations may be periodic such as the
motion of a pendulum or
random such as the movement of a tire on a gravel road. For example, vibrating
a structure or a pipe is
to affect the structure or pipe such that at least a portion of the structure
or pipe undergoes mechanical
oscillations about an equilibrium point.
As used herein, "pipe" refers to a hollow tube used for the conveyance of a
fluid such as water, gas,
steam, petroleum, or the like. The cross section of a pipe can have any shape
such as circular, elliptical
or rectangular.
As used herein, "joint" refers to the place at which two things, or separate
parts of one thing, are joined,
mated or united, either rigidly or in such a way as to permit motion. For
example, two pipes may be
united at a joint, wherein a male terminus of one pipe is mated to a female
terminus of a second pipe,
or a male terminus of a pipe is mated with a female terminus located at a
terminus opposite the male
terminus in the same pipe. Another example would be the case where one end of
one pipe is welded to
one end of another pipe. Furthermore, two I-beams, two cables, or two pipes
may be united to form a
joint using any means of anchorage or adhesively bonding of such as flanges or
the like.
As used herein, "dynamic response" or "vibrational response" refers to the
mechanical oscillations
experienced at the joint of a structure, e.g., a pipe joint, or other
structural joint, when the structure is
vibrated.
As used herein, "healthy" refers to a state of a structure wherein the
structure is substantially free of
damage. For example, a healthy pipe is a pipe that can convey a fluid
throughout the pipe's length
without leaking. A healthy pipe joint is a pipe joint that is substantially
free of damage, wherein the
term "joint" is defined above. A healthy pipe joint does not leak the fluid
that it conveys. A healthy
pipe joint can undergo excitation forces (e.g., vibrations or explosions in a
closed field without
significant loss of structural integrity, i.e., the joint does not leak and/or
the joint can undergo future
excitation forces. Furthermore, healthy pipes are substantially free of
corrosion (e.g., a reduction of
less than about 15 % of the joint wall thickness, a reduction of less than
about 10 % of the joint wall
thickness, a reduction of less than about 5 % of the joint wall thickness, a
reduction of less than about 1
% of the joint wall thickness, or reduction of less than about 0.5 % of the
joint wall thickness), and
fluid leaks (e.g., less than about 0.5% of the fluid flow leaks, less than
about 0.1 % of the fluid flow
leaks, or less than about 0.05% of the fluid flow leaks).
As used herein, an "output device" is a device that creates an effect that is
detectable using one of the
human senses, i.e., sight, sound, smell, touch, or taste. For example, an
output device could include a
siren that produces an audio alarm, a computer monitor or television screen
that produces images
and/or displays information, or an output device can be a light bulb or LED
that emits a wavelength of
electromagnetic radiation in the visible light spectrum when activated.
As used herein, "pipeline" refers to a structure that comprises more than one
pipe, wherein each of the
pipes is mated to at least one other pipe to form one or more joints.
As used herein "biasing" in electronics is the process of establishing pre-
determined voltages or
currents at various points in an electric circuit so as to set an appropriate
operating point (i.e. Q-point)
As used herein "analog filter" is any filter, which operates on continuous
time signals
4
CA 02783089 2012-07-11
As used herein "differential amplification" is when an amplifier will amplify
the difference between
two voltages, but does not amplify the particular voltages. Using differential
amplification facilitates
signal processing in highly dampened structures. In these circumstances,
measurements would fall into
the very narrow band of the data acquisition (DAQ) measurement. As a result,
maximum resolution
would not be achieved, which could potentially affect the resolution of a
damage detection system
Previous Art
Bakis, C.E., et al., "Adiabatic Thermoelastic Measurements," Section VIIB of
Manual on Experimental
process for Mechanical Testing of Composites, R.L. Pendleton and M.E. Tuttle,
Eds., Soc. for
Experimental Mechanics, Bethel, PA, 1989
In the above work, the authors used Ometron SPATE 8000 apparatus, consists of
a scanning infrared
photon detector coupled to a correlator (lock-in amplifier) and computer to
measure the small
temperature changes and the adiabatic thermoelastic effect in elastic solids.
Biemans et al., "Crack Detection in Metallic Structures Using Broadband
Excitation of Acousto-
Ultrasonics", Journal of Intelligent Material Systems and Structures, August
2001, Vol.12, pp. 589 ¨
597.
The above work is concerned with a sensitivity study. In essence, the study
examines the sensitivity of
different statistical parameters (in three domains: time, frequency and
wavelet) to the growing crack in
an aluminum plate. Only two out of the seven statistical parameters extracted
from the time and
frequency domain analysis of the two sensor's data proved to be sensitive
(capable of predicting the
crack growth in a linear fashion). However, there is no quantitative report on
the degree of the
sensitivity of those parameters. Compared to this process, our process does
not require the selection of
a specifically sensitive parameter. Most importantly, our process is capable
of detecting the
progression of the damage with a reliable degree of sensitivity, while there
is no indication that the
statistical parameters-based process could do this.
Cheraghi et al., A Novel Approach for Detection of Damage in Adhesively Bonded
Joints in Plastic
Pipes Based on Vibration process Using Piezoelectric Sensors, Conference
Proceedings - 2005 IEEE
International Conference on Systems, Man and Cybernetics October 2005, Vol. 4
pp. 3472 - 3478.
In the above work, conducted by the authors, they compare the damage indices
obtained by applying
an Empirical Mode-Decomposition (EMD), a less refined version of the process
disclosed in this
patent document (EMD-based), with those obtained by processing the vibration
signals by the wavelet
and Fast-Fourier Transform signal processing techniques, thereby demonstrating
the advantage of
using EMD as a signal processing technique.
Cheraghi, N & Taheri, F. (2007). A Damage Index for Structural Health
Monitoring Based on
Empirical Mode Decomposition, Mechanics of Materials and Structures, vol. 4,
pp. 43-62.
CA 02783089 2012-07-11
This work is very similar to the aforementioned work, with one added example,
in which the integrity
of a pipe having damage in the form of a narrow segment of a pipe having a
thinner thickness is
considered.
Guyott, C.C.H, et al., Use of the Fokker Tester Joints with Varying Adhesive
Thickness, Proceedings
of the institution of Mechanical Engineering, Part B: Management and
Engineering, Vol. 201, pp. 41-
49.
In the above work, the authors detect damage in joints using a frequency-based
process, not by an
energy-based process. They used an instrument to monitor one of the natural
frequencies of the system
comprising a piezoelectric crystal coupled to the joint. They investigated the
resonant frequencies of
two different sizes of transducer coupled to both plain plates and adhesive
joints, theoretically and
experimentally. They demonstrate that the process is capable of identifying
the location of disbands,
but that it cannot distinguish between adhesive modulus (i.e. adhesive's
stiffness) and adhesive layer
thicknesses. This is a more or less a technique for adhesive properties
characterization rather a damage
detection technique.
Huang, N.E., et al. A new view of nonlinear water waves: the Hilbert spectrum,
Annual Review of
Fluid Mechanics, 1999, Vol. 31, pp. 417¨ 457. This publication provides a
general description of the
principles of the Hilbert spectrum approach.
Huang, N.E, et al., The empirical mode decomposition and Hilbert spectrum for
nonlinear and non-
stationary time series analysis, Proceedings of the Royal Society of London-
Series A, 1998, Vol. 454,
pp. 903-995.
Huang et al (authors of the above two, and numerous other articles) are the
folks that developed the
EMD signal processing process. This process is the foundation of the damage
index that has been
established by our process. So, these papers provide the fundamentals of
signal processing by the
EMD approach.
Heslehurst, R. B., Obsevations in the structural response of adhesive bondline
defects, Journal of
Adhesion & Adhesive, 1999, Vol. 19, pp.133-154.
This paper addresses a practical non-destructive process for identification of
adhesive related defects.
They examined several NDTs for damage detection of adhesively bonded joints
and in agreement with
us, state that most conventionally used NDTs (e.g., ultrasonic) are
ineffective in identifying the weak
bonds. They use a holographic interferometry test process. While this process
is an effective process
for establishing weak adhesive bonds, it is not suitable for real-time
monitoring; it is a laboratory
quality assurance type test process.
Kumar et al., Artificial Intelligence and Image processing Approaches in
Damage Assessment and
Material Evaluation, Conference Proceedings - 2005 International Conference on
Computational
Intelligence for Modelling, Control and Automation, November 2005, Vol.!, pp.
307 ¨ 313.
This is also a work produced by the author of this patent. In this work, a
wavelet signal processing
process was used for image processing purposes. The image could be that of
damage or any other type
6
CA 02783089 2012-07-11
of image. In essence, the process disclosed in this paper automates the
interpretation of images
obtained through ultrasonic or x-ray NDT process.
Rizzo et al., Defect Classification in Pipes by Neural Networks Using Multiple
Guided Ultrasonic
Wave Features Extracted After Wavelet processing", Journal of Pressure Vessel
Technology -
Transaction of the ASME, August 2005, Vol.127, pp. 294 ¨ 303.
The guided ultrasonic wave (GUW) process in this work is used for detecting
and quantifying a notch
with varying sizes. Briefly speaking, a transmitter is used to generate the
guided waves to the pipe, and
a receiver records the eco from the pipe ends and the notch. The signal from
the receiver is processed
through the discrete wavelet, HT, and FFT. A damage index is defined by
comparing the selected
features for the direct signal and the signal that is reflected from the
notch.
Similar to the process described in the present work, the damage index
proposed by the present authors
also yields distinctive results in detecting the progression of the damage.
However, it has been shown
that the sensitivity of the damage index decreases as the distance between the
receiver sensor and the
notch increases. This is postulated to be the wave attenuation. For example,
when the receiver is placed
at 100mm from the notch, the resultant damage index from one on the damage-
sensitive features varies
between 20-80% for the different notch sizes (ranging from lmm to 5mm notch
depth in a pipe with
60mm OD and 5.1mm wall thickness and length of 3000mm). The value of the
damage index
decreases to 4-20% for the same damage cases when the receiver is placed 900mm
away from the
notch. This reduction in the sensitivity to the damage detection is similar to
EMD-EDI in the cases
when the damage site is far from a sensor. The authors have however not shown
nor proven in the
preliminary stage of their work, the applicability and reliability of their
process for damage detection
under various conditions.
Wegman, R.F., et al., Non-destructive inspection, Chapter 11. Handbook of
adhesive bonded
structural repair. Park Ridge, NJ: Noyes Pub1.1992.
This chapter contains a description of non-destructive inspection (NDI)
process that are particularly
applicable to the quantitative evaluation and detection of defects in adhesive
bonded structures, not
damage control systems.
Wickerhauser, M. V., Adapted Wavelet Analysis from Theory to Software, A K
Peters, Ltd.,
Wellesley, MA (1994). This is a general purpose textbook on the wavelet signal
analysis process,
which provides basic principles of wavelet theory.
Yang et al (also in this paper: Yang JN, Lei Y, Lin S and Huang N. Hilbert-
Huang based approach for
structural damage detection. J. Eng. Mech. 2004; 130(1): 85-95, actually
introduced two damage
detection process: the first process was used in detection of a sudden change
created in the stiffness of
a structure from the measured data due to the existence of
damage, while the other process combined the Hilbert¨Huang Transform (HHT)
with the EMD process
to determine the changes in the structural natural frequencies. The first
process actually works well if
the damage involves an abrupt change in the structure's stiffness (such as say
a large crack in a bridge
girder); however, the process is not capable of detecting the damage caused by
a gradual change in the
structure stiffness, since it is based on the extraction of the damage spikes
due to sudden changes in the
structure. Therefore it is incapable of detecting local damages such as small
cracks in (welded joints),
7
CA 02783089 2012-07-11
delamination in composites, bolt loosening in a bolted joint, or defects in
bonded joints, which are all
detectable by our process.
The present invention is based upon previous work by the researchers whereby
process of assessing
damage on a joint that includes vibrating the test specimen, detecting the
vibration of the specimen
using one or more signal generating sensors, processing the signal(s), and
applying a damage index to
the processed signal(s), wherein the damage index incorporates a processed
control signal generated by
a sensor(s) at or near the test specimen when the specimen was healthy, i.e.,
in a substantially
undamaged state.
Previous work relating to the invention provides process of detecting damage
on any structure that can
be vibrated and using the dynamic response to detect damage and/or determine
the relative damage at a
joint on the structure. Past process are also useful in detecting damage
and/or determining the relative
damage on a structural joint such as a pipe joint.
For example, one process (Cheraghi et al, 2005a) of detecting damage in a
structural joint comprises
vibrating a structure comprising at least one joint, e.g., a pipeline,
detecting a vibrational response of
the joint, transmitting the vibrational response to a processor as a signal,
processing the signal, and
applying the processed signal to a damage index to yield the relative damage
of the joint.
'Healthy 'Damaged
DI = x 100
'Healthy (2)
where DI is the measure of relative damage at a joint when a test measurement
is taken,
'Healthy is the value of the integral of the processed response signal of the
vibrated healthy joint and
'Damaged is the value of the integral of the processed response signal of the
vibrated joint at the time of
the damage detection and/or relative damage determination. It is noted that at
least some damage is
present in the structural joint (e.g., pipeline joint) if DI is a nonzero
number.
Relating to the present invention, many different signal processing process
can be used to manipulate
the vibrational response signal to create the processed signal that is applied
to the damage index in
equation (2) to detect damage and/or determine relative damage at a structural
joint. For example, a
signal can undergo FT, FFT, WT, or HHT and then be applied to the damage index
expression in
equation (1) to detect damage and/or determine relative damage in a structural
joint.
For example (Cheraghi et al, 2005a), a structure comprising a joint is
vibrated, a piezoelectric sensor
detects the time domain) at the joint, which is transmitted to a processor.
The response signal is
processed using FFT. Under FFT, the integral for the joint response is
expressed as (3):
/ lx(c)ldco
(3)
wherein Ix is the value of the integral of the absolute value of the FFT
processed vibrational response
signal, X(w).
8
CA 02783089 2012-07-11
In addition, (Cheraghi et al, 2005b), described the response signal as being
processed using a Wavelet
process of signal processing, and the expression of equation (3) becomes:
= = d (t)2 dt
(4)
wherein the wavelet packet component energy Ufik, n is the energy stored in
the component
signal dl' (t) . The recomposing dl , n is calculated according to expression
(4b):
d +1,2n-1 d.r1,2n
d"'n ¨ E g 1-2k
(4b)
wherein h(k) and g(k) are discrete filters as described in Wickerhauser, M.
V., (1994). Adapted
Wavelet Analysis from Theory to Software, A K Peters, Ltd., Wellesley, MA,
hereby incorporated by
reference. The damage index is assessed according to equation (2).
In another example (Cheraghi et al, 2006) that employs an Empirical Mode
Decomposition process of
signal processing, the index used in equation (2) takes the following form:
0 (5)
wherein IMF is the first intrinsic mode function of the signal.
For another example, the vibrational response signal of the joint is generated
by a piezoelectric sensor,
transmitted to a processor, and processed using HHT as described in Huang,
N.E, Shen, Z., Long, S.R.,
Wu, M.C., Shih H.H., Zheng Q., Yen N.C, Tung C.C., and Liu H.H. "The empirical
mode
decomposition and Hilbert spectrum for nonlinear and non-stationary time
series analysis".
Proceedings of the Royal Society of London-Series A, 1998, 454: pp. 903-995,
which is hereby
incorporated in its entirety by reference.
The Ix value above includes the FFT-processed response signal when the tested
structure is
healthy or the response signal when damage to the structure is assessed. The
calculation of
discrete approximation of FFT can be represented by:
N-1
X(CO) =x(rAt)ert At
r=0 (6)
Furthermore, in equation (1), m is the sensor number and the degree of freedom
of the structure,
n is the mode shape number. Damage is present in the structural joint when
Din.n is a nonzero
number. Damage indices can be similarly developed for signals processed using
WT. In equation
(7), DI x is the measure of relative damage at the joint when a test
measurement is taken.
9
CA 02783089 2012-07-11
)(Healthy I )(Damaged
Dix= __________ x100
!Healthy
(7)
hHealthy is the value of the integral of the wavelet- processed response
signal of the vibrated
healthy joint, IxDamped is the value of the integral of the wavelet-processed
signal of the joint at
the time of the damage detection and/or assessment. Structural damage is
present in the joint if
DI x is a nonzero number.
The damage index for an HHT-processed signal is:
imHealthy _ irntest
Dime = ___ n n X100
'Healthy
mn (8)
wherein DI. is the measure of relative damage at the joint when a test
measurement is taken, I=Healthy
is the value of the energy of the HHT-processed response signal of the healthy
joint as expressed in
equation (5) above, IõõJest is the value of the energy of the HHT-processed
response signal of the joint
at the time of the damage detection and/or assessment.
The damage indices relating to the present invention, such as those described
in equations (1), (2), (7),
and (8), each have a term that represents the value of the integral of the
processed response signal of a
healthy joint. In the damage index, this term represents a control value that
is used to measure the
amount of relative damage in a structural joint at the time when the damage
detection and/or relative
damage is measured.
Moreover, during signal processing, a response signal can be transformed into
different domains (e.g.,
voltage in a time domain, acceleration in a time domain, or strain in a time
domain) to better interpret
the physical characteristics inherent with the original signal.
In one aspect, a previous approach related to the present invention provides a
process of detecting
damage and/or determining the relative damage in a joint of a structure. In
one example, the process of
detecting damage and/or determining the amount of relative damage in a joint
of a structure comprises
vibrating a structure (e.g., a pipeline) having at least one joint, detecting
or mapping a vibrational
response from the vibrated joint, transmitting the vibrational response to a
processor as a signal,
processing the signal, and applying the processed signal to a damage index,
which yields the amount of
relative structural damage present in the joint.
Another aspect, related to the present invention involves a process of
detecting damage and/or
determining the amount of relative damage in a joint that mates more than one
pipe (e.g., 2 or more
pipes, 3 or more pipes, or 4 or more pipes). For example, the process
comprises vibrating a pipeline,
detecting or mapping the vibrational response from the vibrated joint,
transmitting the vibrational
response to a processor as a signal, processing the signal, and applying the
processed signal to a
damage index, which yields the amount of relative structural damage in the
joint. A sensor that detects
or maps the vibrational response from a vibrated joint can be situated
anywhere such that the
vibrational response of the exited joint is detected or mapped (e.g., the
sensor is placed on the joint).
CA 02783089 2012-07-11
Processes related to the present invention are also useful for detecting
damage and/or determining the
amount of relative damage in joints on structures. Suitable structures include
beams (e.g., I-beams or
the like), pipes, cables or wires that can be vibrated (e.g., the cable or
wire is under at least some
tension), or the like. These structures can comprise members that are joined
using any suitable
coupling process. Such coupling process include, without limitation,
adhesively bonding structural
members, mechanically joining the members (e.g., with bolts, nails, screws,
rivets, collars, friction,
combinations thereof, or other fasteners), welding the members together,
screwing a male member into
a female member, combinations thereof, or the like.
In process related to the invention, the vibrational response of a vibrated
joint is detected and/or
mapped using any suitable sensor(s), the response is transmitted as a signal,
processed using any
suitable signal processing process, and applied to a damage index to yield the
relative damage to the
structural joint. Sensors useful in detecting and/or mapping the vibrational
response of a joint and
transmitting the response as a signal include, without limitation,
piezoelectric sensors, accelerometers,
dynamic displacement transducers, strain gauges, or the like. Structures can
be vibrated using any
suitable process. For example, a pipeline can be vibrated by suddenly closing
or opening a valve
upstream or downstream from the joint, while a fluid is flowing through the
pipeline. In other
examples, the structure or pipeline is vibrated by striking it with a hammer,
contacting it with a
piezoelectric actuator, contacting it with a tuning fork, contacting it with
an electromagnetic actuator,
exposing the joint to electromagnetic radiation, or the like.
A related response mechanism to the present invention provides a process of
detecting damage and/or
determining the amount of relative damage to a pipeline joint comprising
vibrating the pipeline (e.g.,
using a piezoelectric actuator, or by suddenly closing or suddenly opening a
valve upstream or
downstream of the joint that halts or permits the flow of fluid there
through), detecting the vibrational
response of the joint using a piezoelectric sensor, transmitting the response
as a signal to a processor,
processing the signal using any suitable signal processing process (e.g., HHT,
FT, FFT, WT, or the
like), and applying the processed signal to a damage index to yield the
relative damage to the structural
joint.
The damage indices are useful for detecting structural damage in a joint as
well as measuring the
relative amount of damage. For instance, if the relative damage determination
yields a nonzero
number, then some damage is present in the joint.
Related to the present invention, the response signals obtained from the
vibrated joint can undergo
multiple signal processing process to remove excessive noise from the signal.
Another aspect related to the present invention provides process of
quantifying the amount of damage
to a structural joint comprising creating a calibration curve and applying an
amount of relative damage
corresponding to an unknown amount of actual damage to the curve to yield an
actual amount of
damage in the joint.
Previous problems have been addressed in the present invention by providing
improvements within the
process as follows:
11
CA 02783089 2012-07-11
Excitation process ¨
Previous work suggested that there needed to be an improvement in the
excitation process of using a
hammer, tuning fork, or other process, which could resolve inconsistency
issues. The present invention
addresses this problem by employing an electric hammer to create excitation of
a test specimen.
Data Collection Sampling Rate ¨
Features such as integral rubber gaskets and bolted joints often present a
challenge during vibration
data collection because the system can become highly damped. Examination of
the sampling rates of
the vibration signals collected in the previous work has been deemed somewhat
inadequate, thus
altering the accuracy and integrity of the damage indices evaluated by
previous process. The use of an
electric hammer increased the excitation rate by five to ten times the
previous rate, thus improving
collection speed and accuracy.
Effective Signal Amplification -
Generally, a signal conditioning step is mandatory in any acquisition system.
This pre-processing stage
mostly includes the application of low and high pass analog filters and
amplification. Analog filters
efficiently help with the reduction of noise in the portion of the signal that
suffers from aliasing, as
well as the ambient noise. The differential amplification will come handy in
highly damped structures,
in such cases, the measurements would lay in the very narrow band of the data
acquisition (DAQ)
measurement. As a result, the maximum resolution could not be achieved, and
hence could potentially
affect the damage detection resolution.
To overcome this issue, in the present invention, a signal conditioner with an
embedded analog
amplifier is adopted to elevate and improve the quality of the collected
vibration signals, hence
exploiting the maximum achievable resolution of the data acquisition (DAQ)
system. As a result, one
can successfully collect the vibration data from a highly damped and bolted
joint assembly of a real-
scale pipe. As stated earlier, bolted joints used to mate pipes, include
integral rubber gaskets, which
highly dampen the vibration signals. The use of the appropriate amplification
scheme effectively
combats this challenge and resolves the issue, thereby improving signal
quality (with respect to the
signal noise ratio (SNR)).
Use of Wireless Data Acquisition system ¨
Razaei & Taheri (2009) postulated that the first oscillatory mode of a
structure's vibration would be the
most dominant element among the excited modes, while the incorporation of
higher modes would
jeopardize the damage detection's resolution, since the higher modes of
vibration are generally
adversely affected by noise. It was verified that the first frequency would
not always be the dominant
frequency component and its' effectiveness would depend upon the location of
the sensor and the
location at which the structure (component) is excited. Past approaches have
not addressed this
important problem. In these cases, the gathered vibration signals would have
to be conditional and
amplified with a suitable signal conditioner before digitizing in the DAQ,
otherwise the resolution and
accuracy of the damage detection would be significantly lowered.
The present invention employs a wireless DAQ, which increases the utility and
robustness of the
previous approach quite significantly because one can effectively monitor the
health of systems that are
12
CA 02783089 2012-07-11
in obscure and/or inaccessible locations (e.g. being surrounded by water,
where a high damping is
imposed inherently by a joint assembly and the effects of a gasket that sits
between the mating
flanges).
Effective Band-Pass Filtering ¨
In 2010, Rezaei & Taheri suggested that an appropriate action might be to
improve the band width
range of the noise filter to take into account the vibration modes that would
effectively contribute to
the vibration of the structure. Previous background studies by the researchers
had determined that only
the first empirical mode decomposition (IMF) would be sufficient to establish
a damage index for a
joint damage detection system. Also, in (2010), Rezaei and Taheri suggested
that a second IMF might
be sensitive to the presence of some damage types in structures, since it may
contain necessary higher
modes of vibration. Previous approaches to signal filtering were not able to
provide an upper limit that
could achieve a reliable frequency content of a signal.
In the present invention, an improved process for acquisition of a signal
within its band width range, as
well as a noise filtering system that can collectively facilitate
determination of damage in joints
enabled an upper limit to be established for the band-pass filter by
retrieving the analog signal at a
higher frequency. The present invention thus provides greater signal
resolution and hence greater data
reliability by incorporation of a second IMF in the joint damage detection
process.
Improving the quality of Calculated Energy Indices ¨
Discrepancies in calculating damage indices of a joint damage detection system
were due to inevitable
inconsistent vibration excitation created through use of a conventional hand-
held instrumented
hammer.
The present invention solves this problem by using an electronically
controlled impact hammer, by
which the structure/joint can be excited quite consistently, thereby improving
the reliability and
repeatability of the proposed damage detection indices.
REFERENCES
Reliability of piping systems and effect of inspection process, Structural
Safety, vol.3, no.2, (1986),
pp.85-99.
Cheraghi N., G. P. Zou and F. Taheri. Piezoelectric-Based Pipeline Damage
Assessment Using Fourier
and Wavelet Analyses, International Journal of Computer-Aided Civil and
Infrastructure Engineering,
20,2005: 369-382.
Cheraghi, N, Riley, MJ & Taheri, F. (2005), 'A Novel Approach for Detection of
Damage in
Adhesively Bonded Joints in Plastic Pipes Based on Vibration process Using
Piezoelectric Sensors',
IEEE International Conference on Systems, Man and Cybernetics, v 4, p.3472-
3478.
Cheraghi N., G. P. Zou and F. Taheri. Piezoelectric-Based Pipeline Damage
Assessment Using
Fourier and Wavelet Analyses, International Journal of Computer-Aided Civil
and Infrastructure
Engineering, 20, 2005: 369-382.
13
CA 02783089 2012-07-11
Cheraghi, N & Taheri, F. (2007), 'A Damage Index for Structural Health
Monitoring Based on
Empirical Mode Decomposition', Mechanics of Materials and Structures, vol 4,
pp. 43-62.
Engelberg, S. (2008), Digital Signal processing: An Experimental Approach,
Springer, London.
Cheraghi N. M.J. Riley and F. Taheri. Application of Hilbert-HuangTransform
for Evaluation of
Vibration Characteristics of Plastic Pipes Using Piezoelectric Sensors, J. of
Structural Engineering and
Mechanics, 25(6), April 2007: 653-675.
Rezaei, D., and Taheri, F., (2009), "Experimental Validation of a Novel
Structural Damage Detection
process Based on the Empirical Mode Decomposition", Journal of Smart Materials
and Structures, 18,
No.4, doi: 10.1088/0964-1726/18/4/045004.
Rezaei, D & Taheri, F.( 2010a), Damage Identification in Beams Using Empirical
Mode
Decomposition', Structural Health Monitoring, vol 10, no. 3, pp. 261-274.
Rezaei, D & Taheri, F. (2010b), Health Monitoring of Pipeline Girth Weld Using
Empirical Mode
Decomposition', Smart Materials and Structures, vol 19, no. 5.
Esmaeel, RA, Briand, J. and Taheri,F (2011). Computational Simulation and
Experimental verification
of A New Vibration-Based Structural Health Monitoring Approach Using
Piezoelectric Sensors,
Structural Health Monitoring, vol.11, no.2, p.237-250, March, 2012
Esmaeel, RA, Briand, J & Taheri, F. (2011), 'Computational Simulation and
Experimental Verification
of A New Vibration-Based Structural Health Monitoring Approach
UsingPiezoelectric Sensors',
Structural Health Monitoring, p. DOI:10.1177/1475921711414239
Razi, P, Esmaeel, RA & Taheri, F. (2011), 'Application of a Robust Vibration-
Based Non-Destructive
process for Detection of Fatigue Cracks in Structures', Smart Materials and
Structures, vol 20, no.11,
doi: 10.1088/0964-1726/20/11/115017.
Razi, P, Esmaeel, RA & Taheri, F. (2012a), Application of a Remote Health
Monitoring System for
Pipeline Bolted Joints, to be presented in ASME Pressure Vessles and Piping
Conference, July 15-19,
Toronto.
Razi, P, Esmaeel, RA & Taheri, F. (2012b), Improvement of a Vibration-Based
Damage Detection
Approach for Health Monitoring of Bolted Range Joints in Pipelines. Submitted
to Structural Health
Monitoring, May, 2012.
(Wikipedia, http://en.wikipedia.org/wiki/Biasing). Ambient noise generally
means background noise or
the total noise in the surrounding environment (http://medical-dictionary
.thefreedictionaly .com).
(https://ccrma.stanford.edu/ ios/fp/Analog_Filters.html).
(http://en.wikipedia.org/wiki/Differential_amplifier).
US Patent 8,176,786, Sohn et al. (1012). process, apparatus and systems for
damage detection.
14
CA 02783089 2012-07-11
US Patent 7,469,595, Kessler (2008). Piezoelectric damage detection device.
US Patent 6,748,791, Georgeson et al, (2004).Damage detection device and
process.
US Patent 6,532,825, Abe (2003). Fatigue damage determination sensor for
structural materials and
mounting process thereof.
US Patent 5,327,358, Stubbs (1994). Apparatus and process for damage
detection.
BRIEF DESCRIPTION OF THE DRAWINGS
The aspects and the attendant advantages of the embodiments described herein
will become more
readily apparent by reference to the following detailed description when taken
in conjunction with the
accompanying drawings wherein:
FIG. 1 is a schematic describing the decision-making steps of the present
process.
FIG. 2 illustrates a block diagram of a bolted joint showing dimensions and
impact locations.
FIG. 3 illustrates a block diagram of the damage detection process/process.
FIG. 4 is a series of graphs displaying electric hammer signals triggered by
different sampling rates.
FIG. 5 is a series of graphs displaying a typical signal from a PZT bonded
flange, (a) before, and (b)
after amplification.
FIG. 6 is a series of graphs that displays calculated energy indices at
different sampling rates.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
Past works regarding damage detection of pipe systems have addressed damage as
determined from
consideration of semi-real damage scenarios. The current work addresses damage
detection by
monitoring the occurrence and progression of damage in real life situations
(Esmaeel et al, 2011) and
(Razi et al, 2011). Signal processing is considered a mandatory step in any
data acquisition system.
This is a pre-processing stage, which includes mostly the application of low
and high-pass analog
filters and appropriate signal amplification. Analog filters efficiently
assist in the reduction of noise in
the portion of the signal that suffers from a biasing along with ambient
noise.
FIG.1 describes the decision-making steps of the present damage detection
process. According to FIG.
1, excitation of a joint would be monitored in a healthy state and/or in a
distressed state. Signals are
then processed through the Empirical Mode Decomposition approach and the EMD
energy is evaluated
for the test case in the current state and for the future state. Then, the EMD
energy of the first and/or
second IMF must be evaluated to provide the basis for the final step of
calculating the damage index.
This progression of testing enables the user to identify the presence of
damage and to ultimately
determine the location of the damage.
CA 02783089 2012-07-11
FIG. 2 is employed to illustrate various major refinements over prior art that
are part of the present
invention. The set-up in FIG. 2 consists of ASTM compliant schedule 40 Grade B
standard steel pipes
(ASTM A53/A53M ¨07) (1) mated by a bolted flange type joint (2) that were
chosen for illustration
and verification purposes. The selected ANSI forged steel flanges (2) were of
the following type:
nominal pipe size: 6 in., class: 150 lb, raised face, slip-on and material
type A105N. SAE Grade 5
UNC hex head bolts were used (diameter: 3/4" and length: 4"), with the
corresponding sized nuts and
washers. The pipes (1) main dimensions and properties are summarized in Table
1.
In this case, piezo-ceramic sensors (3) are used to monitor the system's
vibration; four piezo-ceramic
sensors (PZT1:PZT4) (3) are bonded onto one of the flanges with an additional
four sensors
(PZT5:PZT8) (3) bonded on the circumference of the pipe (1) at equal angles as
shown in FIG .2
For this specific joint configuration (FIG. 2), the maximum bolt torque was
determined to be 124.7 N-
m. The procedure for tightening the bolts followed the industry standard,
which was a criss-cross
pattern across the face of the flange, using a torque wrench, at increments of
30%, 60%, and then 100%
of the maximum torque.
Table 1: Pipe dimensions and properties
Length (m) 3.52
Outer Diameter (mm) 168.3
Thickness (mm) 6.35
Density (kg/m3) 7800
Young's Modulus 200
(GPa)
Poisson's ratio 0.3
According to FIG. 3, there is an electric hammer (4), which produces an
excitation in a test specimen
(5). Past approaches at excitation produced problematic inconsistencies in
data collection. Previous
approaches at overcoming these inconsistencies in order to derive a usable
signal, involved a judicious
procedure of discarding some of the collected signals, then applying an
averaging technique to the
remaining hammer signal data, which still produced less than accurate data as
well as being time
consuming and costly. Razi et al. (2012a) employed an electric hammer (4) to
collect digital data for a
joint damage detection process and found it greatly increased the accuracy and
efficiency of data
collection within the process, also without a need to discard any of the
collected signals.
Considering the employment of an electric hammer (4) to create excitation in a
test specimen (5), since
the process is an energy-based approach, the consistency of the excitation is
of paramount importance
and governs the accuracy and integrity of the technique. The use of a
conventional manually operated
instrumented hammer (Razi et al, 2012a) can distort the outcome of the damage
detection practice.
This is mainly due to the fact that the orientation, location, and amplitude
of the impacts driven by a
16
CA 02783089 2012-07-11
manual hammer could vary significantly from one excitation to another and from
one operator to
another.
Hence, the use of an electric hammer (4) could effectively address the
abovementioned factors, thereby
facilitating consistent and reproducible impacts. Furthermore, when using a
conventional hammer, one
must excite the system within several trials and take the average of the
acquired signals, so as to reduce
the inconsistency. The use of the electric hammer (4) will eliminate the
requirement for multiple
excitations (Razi et al, 2012a).
In the procedure disclosed during past research (Rezaei and Taheri, 2010), it
was noted that the
vibration signals gathered through piezoelectric sensors and that of the
hammer "must be normalized"
with respect to the hammer's signal to reduce the inherent inconsistency in
the signal(s) produced by
the conventional instrumented hammer. Accordingly, the sensors' signals would
have to be divided by
the hammers' signal in the frequency domain. The resultant signal would be
then transferred back into
the time domain for further analysis. Importantly, this onerous and tedious
normalization step can be
skipped when an electric hammer is used, since it produces consistent impacts
from the perspective of
both frequency and amplitude.
Data acquisition at a high sampling rate, using an electric hammer (4), serves
two main purposes; (a) to
account for the higher modes, which are more indicative of potential damage,
and (b) to facilitate the
registration of the electric hammer signal with appropriate accuracy and
resolution, since any
misrepresentation of the hammer signal would adversely affect the efficiency
of the normalization
process that must be directed onto the impact signal.
A Lazer Doppler Vibrometer was used initially by Rezaei and Taheri (2010a) to
test its applicability
for data correlation. According to FIG .3, in the present invention, there is
a Lazer Doppler Vibrometer
(LDV)(6) and wireless sensors (7) used to collect digital signal data
remotely, without the need to
attach a sensor to the structure (Rezaei & Taheri, 2012-b). Also in the
present invention, Razi et al.
(2012) adopted a wireless receiver (8) for data acquisition, which employs
sensors that are wired to the
carrier through an analog to digital converter module (ADC)(9) that actually
sits in the wireless carrier
to increase the effectiveness and robustness of the current damage detection
process. The wireless data
acquisition system (WDAQ)(9) implemented was a model WLS-9163, produced by the
National
Instruments Inc. (Texas, USA). This implementation increases the utility and
robustness of the current
technique quite significantly because, one can effectively monitor the health
of systems that are in
obscure and/or inaccessible locations (e.g., joints of risers on oil
platforms, etc.). This approach permits
remote collection of vibration data with high sampling rates, thus negating
the need to attach a sensor
to the structure.
The hammer' signal (10) and the vibration signals (11) registered with the
bonded piezoelectric sensors
(7) were digitized with the WDAQ (9) using a 50 kHz sampling rate.
Commercially available wireless
nodes presently used in the industry could not accommodate the high sampling
rate required by the
present invention process (i.e. they are mostly limited to 10 kHz). The
presently adopted WDAQ (9),
however, can provide a sampling rate of 100 kHz. The WLS-9163 also embodies a
12 MB RAM,
which ensures a secure transmission of data; this feature enables the device
to temporarily store 14
seconds of data (sampled at the maximum rate). Thus, the present invention
provides a very significant
increase in efficiency and performance over past Art.
17
CA 02783089 2012-07-11
According to FIG .4, various hammer signals (10) are produced by different
sampling rates, which
range from 10kHz to 501(H. Typical hammer signals registered by different
sampling rates, ranging
from 10 kHz to 50 kHz are depicted in FIG. 4.
As can be seen, the impulsive force occurs in about 0.1 millisecond, thus it
necessitates the sampling
rate to be set to at least Fs =V(0.1 mSec/5) =50 kHz) (i.e., five times
greater than the present harmonic
of the signal). Any lower sampling rates may jeopardize the appropriate
registration of the hammer
signal both in terms of shape and peak amplitude (as seen from the results
shown in (FIG. 4). As such,
it is postulated that the inefficient sampling rate applied in the previous
works (i.e., 10 kHz) together
with variable orientations of the impact in different trials, were the major
causes of inconsistency in the
obtained energy indices and the subsequent damage (Razi et al, 2012b).
According to FIG. 3, under many practical circumstances, the vibration signals
(11) are dampened by
various factors (e.g., being surrounded by water, or in this particular case,
as a result of the high
damping that is imposed inherently by the joint assembly, as well as the
gasket that sits in-between the
mating flanges). In such cases, the gathered vibration signals (11) would
"have to be" conditioned and
amplified with a suitable signal conditioner (12) before digitizing in the
WDAQ (9), otherwise the
resolution and accuracy of the damage detection would be significantly
lowered.
This step is therefore a necessary one and is performed to raise the voltage
level of PZTs' signals (11)
in order to benefit from the maximum resolution that can be achieved after the
signals are digitized in
the WDAQ (9). Moreover, this amplification step reduces the effect of the
noise that is an inherent part
of the measurement. This important step contributes to attaining more
acceptable signal to noise ratios
(SNRs), thereby producing more reliable signals. This will in turn improve the
accuracy and quality of
the energy indices that are commutated, based upon the signals.
According to FIG. 5, a qualitative comparison of a typical signal of a PZT on
the flange taken before
and after the amplification is performed on the signal. The resulting
increases in the SNRs of the
signals after the amplification are tabulated in Table 2. The SNRs for
different signals are calculated
using the following equation:
I
SNR =20 log signal (9)
cynoise
Table 2: Effect of amplification stage on the SNR
Signal to Noise Ratio (SNR)
Before After
Amplification Amplification
Sensors on the pipe 108 125
Sensors on the flange 55 121
18
CA 02783089 2012-07-11
As it can be seen from Table 2 and FIG. 5, the SNR value is noticeably
increased in some of the
signals. This is because those sensors experience the severe local damping
produced, in this case, by
the flange assembly. The digitized data are then sent remotely to a computer
station for analysis
purposes through a reliable and secure communication.
According to FIG.3, it should be noted that the normalized signals "must be"
band-passed with a
digital filter" (e.g., Butterworth filter) (13) to maintain the useful portion
of the data. The bandwidth of
the filter is selected such that it would yield digital signals whose shapes
are very close to that of the
analog ones, and with the same frequency content. The normalized signals are
then individually
processed with the EMD method.
Recently, in 2012, lab trials were conducted by researchers of the present
invention, to establish an
upper limit for the band pass filter (13) and to accurately retrieve the shape
of the analog signal. It was
also determined that when an appropriate upper limit for the band pass filter
(13) was established, the
shape of the analog signal could be accurately retrieved at a higher
frequency. It was also determined
that a sampling rate of at least five or ten times greater than the highest
frequency component of the
analog signal was sufficient for the recovery of the analog signal (Razi et
al, 2012b).
This rule of thumb may be applied to energy-based approaches or the like,
where the reconstruction of
the signal is of paramount importance to the integrity of the process. In
other words, application of this
rule provides an upper limit for the reliable frequency content of a signal.
Moreover, the selected band-
width would remove the portion of the signals that suffer from aliasing.
Selection of the appropriate
frequency band-width based on the explained "rule of thumb" (e.g. upper-band
limit) will reduce the
confusion and the subsequent trials and errors noted in previous works (Razi
et al. 2012b).
According to FIG. 6, recently, it has been postulated that discrepancies in
calculating damage indices
of a joint damage detection system have been due to the consistent vibration
excitation of a
conventional hand-held instrumented hammer (Razi et al, 2012b).
FIG. 6 sheds light on the tangible inconsistencies in the calculated energy
indices (14), up to 62 %, that
could be potentially raised in various trials of the electric hammer (4) due
to the implementation of
lower sampling rates.
The previous description of the disclosed embodiments is provided to enable
any person skilled in the
art to make or use the present disclosure. Various modifications to these
embodiments will be readily
apparent to those skilled in the art, and the generic principles defined
herein may be applied to other
embodiments without departing from the spirit or scope of the disclosure. Thus
the present disclosure
is not intended to be limited to the embodiments shown herein, but is to be
accorded the widest scope
consistent with the principles and novel features as defined by the following
claims.
EXAMPLE of the USAGE of the INVENTION
The invention can be employed to determine damage in joint systems in a
nuclear power plants.
19
CA 02783089 2012-07-11
SUMMARY OF THE INVENTION
In accordance with the description, there is provided a process of detecting
damage and/or assessing
relative damage on a structural joint that includes vibrating the structure
that comprises the joint,
mapping the vibrational response of the joint using one or more signal
generating sensors, transmitting
the vibrational response as a signal(s) to a processor, processing the
signal(s), and utilizing the
processed signal(s) to produce damage indices that yields the relative damage
at the joint.
Specifically, in accordance with the description there are provided
improvements to a process of
damage detection in joint systems, whereby an electric hammer is used to
improve the accuracy and
efficiency of collecting data to calculate damage indices in a damage
detection system (DAQ).
In accordance with the description, there are provided improvements to a
process of damage detection
in pipe and joint systems, whereby a Laser Doppler Vibrometer (LDV) is
employed to collect signal
data remotely, thus negating the need to attach a sensor to the test element
to improve the efficiency of
data collection. In addition, there is provided a wireless carrier set-up with
sensors wired to the carrier
through an analog to digital converter module (ADC), which enables collection
of data from high
sampling rates that greatly improves data collection speed and accuracy at a
base computer station,
thereby facilitating remote data collection. The wireless carrier is easily
mounted on remote locations
at test sites.
In accordance with the description, there are provided improvements to a
process of damage detection
in pipe and joint systems, whereby a special purpose signal conditioner with
an embedded analog
amplifier to assist in obtaining maximum resolution of vibration data obtained
through any sensor
(especially in the case when piezoelectric sensors are used), from various
site locations.
In accordance with the description, there are provided improvements to a
process of damage detection
in pipe and joint systems, which employs a process of improving acquisition of
a signal within its
bandwidth range, as well as a noise filtering system that can collectively
facilitate detection of damage
in structural joints. An upper limit is established for the band pass filter
by retrieving the analog signal
at a higher frequency. Incorporation of a second IMF in the joint damage
detection system (JDDS) has
proven to yield greater signal resolution and hence greater reliability than
previous work, by combining
a first and second IMF, which will enhance the predictive accuracy of the
(JDDS).
In accordance with the description, there are provided improvements to a
process of damage detection
in pipe and joint systems, whereby normalized signals are band-passed with a
digital filter to maintain
the useful portion of the data. In addition, an appropriate upper limit was
established such that the
shape of the analog signal could be accurately retrieved at a higher
frequency. A sampling rate of at
least five or ten times greater than the highest frequency component of the
analog signal was sufficient
for recovery of the analog signal.
In accordance with the description, there are provided improvements to a
process of damage detection
in pipe and joint systems, whereby damage detection is achieved and expressed
through indices by
processing of the captured vibration signals and processing the signal via a
special process to evaluate
the damage indices, specifically by using software that is either hosted on a
remote computer or by
directly installing the code into a remote processor.
CA 02783089 2012-07-11
Other aspects, advantages and features of the present disclosure will become
apparent after review of
the entire application, including the following sections: Brief Description of
the Drawings, Detailed
Description of the Drawings and the Claims.
21
