Citation
Acoustic tomography of concrete using a James instruments velocity meter

Material Information

Title:
Acoustic tomography of concrete using a James instruments velocity meter
Creator:
Leiphart, Galina S
Place of Publication:
Denver, Colo.
Publisher:
University of Colorado Denver
Publication Date:
Language:
English
Physical Description:
xiii, 179 leaves : illustrations ; 29 cm

Thesis/Dissertation Information

Degree:
Master's ( Master of Science)
Degree Grantor:
University of Colorado Denver
Degree Divisions:
Department of Civil Engineering, CU Denver
Degree Disciplines:
Civil Engineering
Committee Chair:
Rens, Kevin L.
Committee Members:
Stalnaker, Judith J.
Mays, John R.

Subjects

Subjects / Keywords:
Tomography ( lcsh )
Concrete construction -- Testing ( lcsh )
Concrete construction -- Testing ( fast )
Tomography ( fast )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Bibliography:
Includes bibliographical references (leaves 175-179).
General Note:
Submitted in partial fulfillment of the requirements for the degree, Master of Science, Civil Engineering.
General Note:
Department of Civil Engineering
Statement of Responsibility:
by Galina S. Leiphart.

Record Information

Source Institution:
|University of Colorado Denver
Holding Location:
|Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
39719198 ( OCLC )
ocm39719198
Classification:
LD1190.E53 1997m .L45 ( lcc )

Full Text
ACOUSTIC TOMOGRAPHY OF CONCRETE
USING A JAMES INSTRUMENTS VELOCITY METER
by
Galina S. Leiphart
B.S., Carnegie Mellon University, 1994
A Thesis submitted to the
University of Colorado at Denver
in partial fulfillment
of the requirements for the degree of
Master of Science
Civil Engineering
1997
AL


1997 by Galina S. Leiphart
All rights reserved.


This thesis for the Master of Science
degree by
Galina S. Leiphart
has been approved by
Kevin L. Rens
/2-I0-77
Date


Leiphart, Galina S. (M.S., Civil Engineering)
Acoustic Tomography of Concrete Using a James Instruments Velocity Meter
Thesis directed by Assistant Professor Kevin L. Rens
ABSTRACT
Acoustic tomography is a nondestructive evaluation (NDE) technique that is
currently receiving heightened attention in the evaluation of concrete structures. It
is a process by which a slice through an object is taken to obtain an image of that
slice plane. It is a technique that uses acoustic (sound) waves shot from a
transmitting transducer through a specimen to a receiving transducer, and the time
travel across the specimen is recorded. The travel times and transducer
coordinates are used with a tomography algorithm to produce a velocity map over
the plane being imaged.
For this research, the James Instruments velocity meter (v-meter) was
utilized for the collection of travel times through the constructed test specimen. A
concrete wall with a height of 3-10 34, length of 8-0, and thickness step function
that goes from 12 to 18, was constructed in conjunction with this research. In
order to help ensure that the data collected was reliable, a gauge potential study
was performed prior to data collection. A gauge potential study is a short term
statistical study that examines the systems potential to be capable in the long term.
It examines the repeatability and reproducibility of the v-meter measurement
system as developed. The study indicated that the system has the potential to be
capable in the long term. Based on the results of the study a standard operating
procedure (SOP) was developed for the v-meter measurement system and is
included in this thesis.
IV


The U.S. Bureau of Mines 3DTOM software program was used to produce
the acoustic tomogram discussed in this research. Sixteen different tomograms
were produced of the twelve inch concrete wall. The tomograms were produced in
an attempt to image three known inclusions of a styrofoam ball, wood block, and
low density perlite cylinder. The sytrofoam ball and perlite cylinder were identified
while the wood block was not. A transducer spacing of three inches and 1.5 inches
was investigated, as was pixel grid density. The decreased transducer spacing and
increased pixel grid appears to increase the resolution of the tomogram. The
tomograms are presented at a 100 m/s contour interval.
This abstract accurately represents the content of the candidates thesis. I
recommend its publication.
Signed
Kevin L. Rens
v


ACKNOWLEDGEMENTS
Many people have offered me guidance and support over the past two years during
this pursuit of my masters degree. Without their encouragement, this journey
would have been much more difficult and not nearly as rewarding. I extend sincere
thanks and gratitude to :
Kevin L. Rens, for encouraging me to pursue the thesis option of a Masters Degree
and for agreeing to serve as my advisor. You opened my eyes to the world of
nondestructive testing and evaluation. Thank you for counseling me when I lost
sight of my goals and thought I would never finish.
Judith J. Stalnaker and John R. Mays, for accepting my invitation to be committee
members for my thesis. You are stimulating and passionate in your teaching. It is a
privilege and honor to have your input in this thesis.
Brian Abbott, without your help and hard work the two concrete walls used in this
research would not have been built. Thank you also for your countless hours
collecting data and teaching me how to use AutoCAD and SURFER. Funding for
the construction of the walls came in part from his grant from The Undergraduate
Research Opportunity Program (UROP). The initial straight shot assessment
included in Chapter 3 was a part of this grant, and is included here with Brians
permission.


James Knappmiller of the CPE Department, Ball Packaging, for your guidance and
expertise on gauge potential studies. Your interest in my research encouraged my
excitement about the results of the gauge potential studies.
Dave J. Transue, for your assistance in building the concrete walls, mix design for
the low density concrete, and for participating in the gauge potential studies; and
Atkinson-Noland & Associates in Boulder, Colorado, for allowing me to borrow your
James Instruments, Inc. Velocity Meter.
And finally, special thanks and love to:
Marjorie G. Ferrell and Frederic P. Ferrell, whose memories are with me everyday.
Kay A. Ferrell and Richard E. Gibboney for taking us in when I decided to go back
to school, for being my sounding board, for the numerous literature searches, for
valuable input regarding APA Style, and especially for a beautiful wedding in the
middle of all the chaos; and
Matthew, my husband and best friend. Thank you for supporting me in this
endeavor and encouraging me to excellence. Mizpah.


CONTENTS
CHAPTER
1.0 Introduction..........................................................1
1.1 Stress (Sound) Wave Theory...........................................5
1.2 Literature Review...................................................11
1.2.1 Tomography Algorithm................................................11
1.2.2 Frequency Concerns..................................................15
1.2.3 Concrete Samples....................................................16
1.2.4 Transducers and Couplants...........................................17
1.2.5 Data Collection.....................................................19
1.2.6 Software / Processors...............................................20
1.2.7 Resulting Images....................................................21
1.3 Gauge Potential Study...............................................22
2.0 Methodology...........................................................25
2.1 Description of Wall.................................................25
2.1.1 Inclusions..........................................................25
2.1.2 True Thickness of Wall..............................................30
2.1.3 Concrete Placement..................................................30
2.1.4 Concrete Strength...................................................31
2.1.5 Form-work...........................................................31
2.1.6 Screws..............................................................31
2.1.7 Support Against Blow-Out............................................32
2.1.8 Oil.................................................................32
2.1.9 Form Removal........................................................32
2.2 Equipment...........................................................34
2.3 Gauge Potential Study...............................................36
viii


2.3.1 Description of Spreadsheet..........................................38
2.3.2 Recommendations for Conducting a Gauge Potential Study..............43
2.4 3DTOM Software......................................................44
2.4.1 Data Files..........................................................45
2.4.2 Units...............................................................46
2.4.3 Coordinate System...................................................46
2.4.4 Constraints.........................................................47
2.4.5 Theory..............................................................48
2.4.6 Ray Tracing.........................................................50
2.4.7 Conclusions.........................................................52
2.4.8 Step-by-step Procedure..............................................52
2.5 Data Collection for 3DTOM...........................................53
2.5.1 Standard Operating Procedure for Data Collecting using V-Meter......54
2.5.2 Sample Data Collection Sheet........................................58
3.0 Results...............................................................60
3.1 Gauge Potential Study Results.......................................60
3.1.1 Study No. 1.........................................................60
3.1.2 Study No. 2.........................................................63
3.1.3 Comparison of Study No. 1 and Study No. 2...........................65
3.2 Tomogram Results....................................................68
3.2.1 Initial Assessment of Concrete Wall.................................68
3.2.2 3DTOM Input Information.............................................73
3.2.3 Tomogram of Styrofoam Ball..........................................74
3.2.4 Tomogram of Wood Block..............................................81
3.2.5 Tomogram of Perlite Cylinder........................................88
4.0 Conclusion............................................................99
4.1 Presentation of Tomograms...........................................99
4.2 Effects on Tomogram Resolution.....................................105
4.3 Summary and Further Research.......................................106
IX


APPENDIX
A. Gauge Potential Study Spreadsheets & Calculations...........109
B. Data for Styrofoam Tomograms................................113
C. Data for Wood Tomograms.....................................123
D. Data for Perlite Cylinder Tomograms.........................142
E. Photographs.................................................170
REFERENCES.....................................................175
x


FIGURES
1.1 P-wave..................................................................6
1.2 S-wave..................................................................7
1.3 Snells Law.............................................................9
2.1 Elevation view wall inclusions.......................................26
2.2 Plan view wall inclusions............................................27
2.3 Wall thickness prior to concrete placement.............................30
2.4 Form tie locations.....................................................33
2.5 Gauge potential study test locations...................................37
2.6 Sample spreadsheet.....................................................39
2.7 Example of 3DTOM data set..............................................46
2.8 3DTOM Cartesian coordinate system .....................................47
2.9 Ray coverage for tomogram..............................................55
2.10 Tomogram section locations.............................................56
2.11 Orientation of transducer on location marks............................57
2.12 Data collection sheet..................................................59
3.1 SURFER contour plot of straight shot travel times......................69
3.2 SURFER contour of 1st post processing of data..........................70
3.3 SURFER contour plot of 2nd post processing.............................71
3.4 Comparison of known inclusion location to apparent inclusion location.73
3.5 Styrofoam tomogram using 3 transducer spacing; 7x7 grid...............76
3.6 Styrofoam tomogram using 1.5 transducer spacing; 9x9 grid.............77
3.7 Styrofoam tomogram using 3 transducer spacing; 14x14 grid.............80
3.8 Styrofoam tomogram using 1.5 transducer spacing; 18x18 grid...........81
3.9 Wood tomogram at grid line 7V; 3 transducer spacing; 7x7 grid.........83
3.10 Wood tomogram at grid line 7V; 1.5 transducer spacing; 9x9 grid.......83
XI


3.11 Wood tomogram at grid line 8V; 3 transducer spacing; 7x7 grid......85
3.12 Wood tomogram at grid line 8V; 1.5 transducer spacing; 9x9 grid....85
3.13 Wood tomogram at grid line 8V; 3 transducer spacing; 14x14 grid....87
3.14 Wood tomogram at grid line 8V; 1.5 transducer spacing; 18x18 grid....87
3.15 Perlite tomogram at grid line 10V; 3 transducer spacing; 7x7 grid.....90
3.16 Perlite tomogram at grid line 10V; 1.5 transducer spacing; 9x9 grid...90
3.17 Perlite tomogram at grid line 11V; 3 transducer spacing; 7x7 grid.....92
3.18 Perlite tomogram at grid line 11V; 1.5 transducer spacing; 9x9 grid...92
3.19 Perlite tomogram at grid line 12V; 3 transducer spacing; 7x7 grid.....94
3.20 Perlite tomogram at grid line 12V; 1.5 transducer spacing; 9x9 grid...94
3.21 Perlite tomogram at grid line 12V; 3 transducer spacing; 14x14 grid...97
3.22 Perlite tomogram at grid line 12V; 1.5 transducer spacing; 18x18 grid.97
4.1 Copy of Figure 3.19 at a contour interval of 100 m/s..................101
4.2 Copy of Figure 3.19 at a contour interval of 200 m/s..................101
4.3 Copy of Figure 3.22....................................................103
4.4 Ray coverage per cell for Figure 4.3..................................104
xii


TABLES
3.1 Results of gauge potential Study No.1..............................61
3.2 Results of gauge potential Study No.2..............................63
3.3 Comparison of gauge potential Study No.1 and Study No. 2...........65
3.4 Comparison of low velocity areas for styrofoam tomogram............78
3.5 Comparison of low velocity areas for perlite tomogram..............95
xiii


1.0 Introduction
Due to our aging infrastructure, it is becoming more important to evaluate
current conditions and to predict the service life of existing structures. Now is the
time to turn efforts towards establishing universal criteria regarding what constitutes
failure of in place structures, rehabilitation and maintenance programs of those
deteriorating structures, and standardized testing methods for nondestructive
testing techniques. Attention should also be focused on preventive measures for
structures built in recent years. Nondestructive evaluation (NDE) is currently being
used as both a proactive and reactive technique in the condition assessment of
concrete structures.
Acoustic tomography is currently an NDE technique that is receiving
heightened attention in the evaluation of concrete structures. For example,
Jalinoos & Olson (1995), Kline, Wang, Mignogna, & Delsanto (1994) and Schuller,
& Atkinson (1995). The technique is non-invasive and the results are easily
interpreted. However, current equipment tends to be large and cumbersome which
does not lend itself to easy field use and data collection can be quite time
consuming. In addition, computation time of the tomography algorithm is quite
extensive.
1


In 1996 a questionnaire was sent to the fifty state transportation
departments (DOTS) as a follow up to a 1993 survey regarding their current NDE
needs and problems and also to see if tomographic imaging might be useful to
them (Transue, Rens & Schuller, 1997). By September 1997, 86% of the DOTS
responded to the 1996 questionnaire (Transue, Rens, & Schuller, September
1997). Of that 86%, 92% indicated that the ability to more thoroughly assess the
condition of concrete structures was desirable (Transue et al., 1997, September, p.
3). In comparing the two surveys Transue et al. (1997) further comments that five
DOTS that previously responded to the 1993 survey, indicating that they did not
perform in-house NDE, have since established an NDE program within their
organization. From these results it is obvious that there is in fact a need for further
research in the area of nondestructive evaluation and testing and acoustic
tomography. Further information regarding the 1997 questionnaires can be found
in Transue et al. (1997) and Transue et al. (1997, September). Further information
regarding the 1993 questionnaire can be found in Rens, Wipf and Klaiber (1997).
Other evidence that increased research in the NDE field is beneficial is the
increased number of conferences and articles that have been published in recent
years regarding inspection and rating of infrastructure. Previously, a presentation
or two on NDE might be included in a conference on structures or materials, now
there are entire three day conferences dedicated to the dissemination of
information concerning nondestructive testing research and progresses made in the
field. For example, in August 1996 there was a three day conference, sponsored
2


by the American Society of Civil Engineers, entitled Infrastructure condition
assessment: Art, science, and practice.
Tomography is the process of taking a slice through an object in order to
obtain an image of the interior plane of that object. Acoustic tomography is the
process of taking that slice of an object through the use of acoustic (sound) waves.
Sound waves are shot through the sample from source to receiver transducers.
The time required for the pulse to travel across the specimen is recorded. This
information used with the tomography algorithm produces a velocity map of the
plane over which the pulse(s) traveled. Anomalies within the plane are depicted as
changes in velocity.
Although, acoustic tomography has been in use and actively researched for
many years, further work needs to be focused on the ability to characterize the type
of flaw based on a specific velocity or change in velocity. In the laboratory, it is
easy to relate a low/high velocity area with the known inclusion. However, in the
field it is not known where a honeycomb or void may be located, although
knowledge of the reinforcement location can be obtained from as-built drawings.
Therefore, it is important to have knowledge about the tendencies of inclusions to
react to propagating acoustic waves in order for the technique to be effectively used
in the field. Similarly, further research is required in civil engineering applications of
the technique in the area of optimal transducer spacing and pixel location in order
to maximize tomogram resolution.
3


As well as focusing attention on the interpretation of the velocity tomograms,
it would also be beneficial to produce a standard operating procedure (SOP) for the
collection and measurement of travel time data. A SOP (or standard test method)
will lend credibility to the test procedure by regulating the process by which data is
collected and will help to ensure that the data is reliable. A gauge potential study is
one means by which reliable data acquisition can be obtained (Continuous Pursuit
of Excellence, 1997). The gauge potential study will measure the repeatability and
reproducibility of a measuring system by estimating the standard deviation (or
spread) of the data. Using these results to change the method used accordingly,
will help minimize the contribution of measurement system error in the total variance
seen in the tests results. However, assessing the quality of the data is not a one
step process. It is a process that must be continuously examined, and improved
upon. For example, variance in the data may occur as a result of the method used,
the operators procedure, the operators experience, the equipments functioning
tendencies, the equipments sensitivity, temperature, humidity, or external factors.
These influences do not have a fixed outcome, and therefore are likely to fluctuate
and effect the quality of the data. For this reason it is important to understand how
these factors will effect the data or standard procedure, in order to comprehend the
results obtained from such data.
The purpose of this research is six fold:
1. to construct a test specimen to further concrete NDE research at
the University of Colorado at Denver,
4


2. to conduct a gauge potential study to discuss the capability of the
James Instruments velocity meter to produce reliable data,
3. to produce a standard operating procedure for data collection
using the James Instruments velocity meter based on the gauge
potential study,
4. to use the U.S. Bureau of Mines 3DTOM software program to
produce tomograms of the known inclusions in the constructed
test specimen,
5. to assess the quality of the tomograms by comparison to known
cross-sectional data, and
6. to examine how transducer spacing of one and one half inch and
three inches effects tomogram resolution.
Included in this chapter is an introduction of stress wave theory as it applies
to acoustic tomography, a literature review detailing the current status of acoustic
tomography, and an introduction to gauge potential studies and how they benefit
this area of NDE research.
1.1 Stress (Sound) Wave Theory
The use of stress, or sound, waves is one of the oldest forms of
nondestructive testing. For example, it is a common technique to strike an object
and listen for a hollow or ringing sound, which is an indication of an internal void or
crack. However, this is a qualitative technique and depends primarily on the
5


experience of the user (Malhotra & Carino, 1991). Currently there is no standard
method for the use of stress waves to find flaws in concrete. Although American
Standard for Testing Materials (ASTM) subcommittee C09.64 is completing a
revision of standard C 597 on pulse velocity to bring it up to date with current
technology (Schuller & Woodham, 1996, p. 2). Members of the engineering
community are also discussing and producing reports on research that is necessary
to support standardization (Schuller & Woodham, 1996).
There are three types of waves in stress wave propagation: dilatational
waves, distortional waves, and rayleigh (surface) waves.
The dilatational wave is often referred to as the longitudinal, compression
or P wave. The motion of a P-wave is parallel to the direction of propagation of
the wave. For example, if the wave travels east, the direction of motion of the wave
is also to the east. This is similar to the motion of a snake slithering over the
ground. The P-wave occurs in all types of structures. (Malhotra & Carino, 1991).
The illustration in Figure 1.1 refers to a p-wave.
Direction of
motion.
------------->
Direction of propagation.
Figure! 1 P-wave.
6


The distortional wave is also called the shear or S wave. The motion of
an S-wave is perpendicular to the direction of propagation. If the wave travels east,
then the direction of motion is north-south. An S-wave only occurs in specimens
that have a shear stiffness. For example, S-waves would not occur in water
because water is a material that does not have shear stiffness. (Malhotra & Carino,
1991). The illustration in Figure 1.2 refers to an s-wave.
Direction of " I I I * I I I " ; I ; ; ; I ; I ;
motion.
------------>
Direction of propagation.
Figure 1.2 S-wave.
The rayleigh or R waves travel along the surface of a solid in a retrograde
elliptical motion. Often times, rayleigh waves are referred to as surface waves.
(Malhotra & Carino, 1991).
The speeds of the P-wave, S-wave and R-wave are as follows (Malhotra &
Carino, 1991):
Cp =
E{ l-fi)
p{\ p){\ 2ju)
(1.1)
7


Cs v 2p(l M)
(1.2)
Cr =
^.87 + 1.12//
^ 1 + H
Cs
(1.3)
Where E is Youngs modulus of Elasticity, p is the mass density, and p is Poissons
ratio. As can be seen from the above equations, the speed of the propagating
wave is directly dependent of the material properties of the specimen in which the
wave travels. For instance, if the density of a specimen increases in a particular
area, the P- and S-wave speed will both decrease at that location. It is this
occurrence of changing wave speed within a single heterogeneous material that
makes acoustic tomography possible. For concrete it has been found that the P-
wave velocity is in the range of 3000 to 4500 m/s (Malhotra & Carino, 1991, p.278).
In acoustic tomography, it is the P- and S-waves that are of interest. Both P-
and S-waves reflect and refract when at the interface of a different media.
Reflection occurs when a wave bounces off of a material interface, and refraction
is the process where the wave proceeds through a material interface but at a
different angle than its incident angle. (Malhotra & Carino, 1991).
The angle of incidence, 0, and the angle of refraction, p, are related by
Snells law (Malhotra & Carino, 1991). Where C2 and Ci are the speed of the wave
in the respective media.
8


(Ct)
sln/?=lcJs,n*
(1.4)
The illustration in Figure 1.3 shows the relation between the angle of incidence and
angle of refraction.
However, P-waves can also be reflected or refracted as S-waves. The
angles of these waves are also related by Snells law (Malhotra & Carino, 1991):
sin 6 sin/? sin 6. sin/?,
Cp. Cp2 C c.2 (1.5)
9


The portion of a P-wave that is reflected depends on the acoustic impedance of
each medium. The acoustic impedance, Z, of a medium is (Malhotra & Carino,
1991):
Z = p CP (1.6)
The reflection coefficient, Rn, determines the amplitude of the reflected ray
relative to the incident ray, and the refraction coefficient, Rd, determines the
amplitude of the refracted wave relative to the incident wave. The subscripts 1 and
2 refer to media 1 and 2 respectively. (Malhotra & Carino, 1991).
Rn
Z,-Z,
Z2 + Z1
(1.7)
2 Z2

Z2 + Zi
(1.8)
The wavelength, X, and frequency, f, relate to the wave speed by the
following equation (Malhotra & Carino, 1991):
C = f X (1.9)
Because concrete is not a homogeneous material (it has many material interfaces),
stress waves do not travel in straight lines. As a result, it is important to use low
10


frequency (long wavelength) waves to reduce attenuation (scattering) as it
propagates through the material. As a rule of thumb, the size of the flaw should be
approximately equal to or greater than the wavelength to be detected. Whereas,
the concrete aggregate size should be smaller than the wavelength to reduce
scattering. For example, if a flaw is approximately 40mm, then by rearranging
equation (1.9), the minimum frequency used would be equal to 4000/40 = 100 kHz
(Malhotra & Carino, 1991). The stifferthe material, the faster the velocity (Olson,
Wright, & Stokoe, 1990).
1.2 Literature Review
The literature review presented here is an edited and condensed version of
the literature review on acoustic tomography of concrete conducted by Ferrell
(1997).
1.2.1 Tomography Algorithm
The tomography algorithm is the actual process through which a tomogram
is produced. It involves collecting the necessary data, interpreting that data, and
performing required calculations. Sullivan, Kline, Mignogna and Delsanto (1996)
suggest that, Three major components comprise the tomographic algorithm: time
delay calculations, ray tracing, and the tomographic inversion (p. 2144).
11


Time delay calculations are the process of actually collecting the data that
describes the characteristics of the propagating waves. Data is typically collected
for a series of source receiver combinations, with the input frequency and initiation
time recorded. Arrival time of the pulse is recorded and the time required for the
wave to travel from source to receiver (time delay) is calculated. From this
information, the process of ray tracing can commence.
Ray tracing is the operation of finding the path that a wave travels from
source to receiver. As previously mentioned, sound waves tend to scatter and do
not travel in a straight line when traveling through a heterogeneous material such as
concrete. This makes the ray tracing process both computationally intense and
time consuming. Most ray tracing algorithms were originally developed for x-ray
applications, where scattering of the propagating wave is not an issue. Several ray
tracing techniques have been tried and are described here:
1. Sullivan et al. (1996) recommend a polynomial approximation to the time
delay curve instead of the traditional shooting method of ray tracing. It
was shown that the average velocity error was reduced by increasing the
degree of polynomial curve fitted to the data.
2. Wang and Kline (1994) recommend an algorithm based on Fermats
principle. The algorithm increases the quality of the image, especially at
material interfaces and at the samples edges.
12


3. Kline et al. (1994) suggest another ray tracing technique which utilizes
numerical techniques to evaluate the entire acoustic field instead of
individual rays on a piece by piece basis. Reasonable agreement
between the sample and acoustic tomogram were documented with the
main discrepancies again occurring at material interfaces and
boundaries
4. Moser (1991) utilizes yet another ray tracing technique, based both on
Fermats principle and Huygens principle. Fermats principle uses
bending to find the minimum travel time between two points. With this
technique only the absolute shortest paths are found. Longer travel
times caused by discontinuities are not recorded. A second run of this
algorithm must be used to find these longer travel times. Computation
time of this method was found to be dependent on the number of nodes
and number of connections per node in the pixel grid model.
Tomographic image reconstruction takes place by use of a grid
superimposed over the object to be imaged. If p(x,y) is the variance of the acoustic
parameter within a cross section, the acoustic parameter (velocity, impedance, etc.)
will vary depending on the x-y location on the superimposed grid (Eberhard, 1982).
The line integral of p is then (Eberhard, 1982, p. 7):
13


(1.10)
P*(r) = | p ds
i
Where r and 9 describe the position of the ray, /, through the cross section, the
distance from the origin of the grid to the ray is r, and the fixed angle is 0. It is the
set of line integrals, at a fixed 9, over various and many rays that form a projection
(Eberhard, 1982). It is then the inversion of these line integrals that form a
tomographic image These inversions obtain estimates of the velocity/attenuation
field through which the waves passed. Therefore, the ray coverage should be as
wide as possible to have a successful tomographic image (Worthington, 1984).
The main draw back with this type of technique is the large amount of data
that is required (Eberhard, 1982). As a result, the reconstruction computations
easily become intense and time consuming. The computations become even more
massive when consideration is taken of the tendency of waves to scatter when
propagating through concrete. Therefore a fair amount of attention has been
focused on ray tracing techniques, reconstruction methods, and improved scanning
methods.
There have also been improvements in the reconstruction algorithm in the
areas of diffraction tomography (Gelius, Johansen, Sponheim & Stamnes, 1991)
and limited angle tomography (Heiskanen, Rhim, & Monteiro, 1991). Diffraction
tomography (DTG) accounts for diffracting ray paths in the reconstruction and is
based on the generalized projection slice (GPS) theorem. Limited angle
14


tomography takes into account the possibility of limited access to a structure.
Heiskanen et al. (1991, p. 634) contend that this new limited angle algorithm
significantly increases the power of tomography in civil engineering applications.
It has also been suggested that tomography not be a first step in the
nondestructive evaluation of concrete structures. Jalinoos, Olson, and Sack (1995)
recommend using the impact echo (IE) method or ultrasonic pulse velocity (UPV)
method as a means of initial defect identification within a structure. These methods
are less computationally intense and require less data to identify a questionable
area. Once an area of concern is located via the IE/UPV method, a detailed scan
of the area can be done and the tomogram constructed. Further information on the
IE and UPV methods can be found in Ferrell (1997).
Lee and Chuang (1986) further suggest that reconstruction algorithms: a)
eliminate complicated mathematics, b) produce a small value for smooth phase
density distribution, c) generate greater values for clustered phase density
distributions, d) be based on phase density distributions to prevent high order
nonlinear operations, and e) be stable and consistent with the presence of noise.
1.2.2 Frequency Concerns
As mentioned previously, frequencies below 100 kHz are typically used in
acoustic tomography due to the tendency of higher frequencies to scatter.
However, frequencies of this magnitude tend to result in poor tomographic images
because low frequencies can not provide sufficient spatial resolution (Gaydecki,
15


Burdekin, Damaj, John & Payne, 1992, p. 126). Whitcomb, Jacobs, and Aref
(1993) also state that the maximum frequency that can propagate through a
material is dependent on aggregate size. This is comparable to the suggestion of
Malhotra and Carino which was previously discussed in section 1.1 that the
aggregate or flaw size be greater than or equal to the wavelength. Therefore
choosing a wavelength value based on aggregate size dictates what frequency
should be used by equation (1.9). Whitcomb et al. (1993, p. 243) also state that
The Standard Test Method for Pulse Velocity Measurement through Concrete
indicates that the maximum frequency used should be 150 kHz.
Because of these frequency limits, concrete testing has also be done with
the use of adaptive filters. Adaptive filters enhance the signal-to-noise ratio (SNR),
thus allowing the use of higher frequencies. Higher frequencies tend to scatter
more and therefore allow for more noise to interfere with the signal. As the signal
strength decreases, the random noise increases, and so too does the SNR
(Gaydecki, 1997).
1.2.3 Concrete Samples
Impact echo and ultrasonic pulse velocity techniques have been used in the
field (Sack, Olson, Kline, & Yates, 1992; Olson, 1990; Sansalone & Jaeger, 1996),
as has acoustic tomography. However, it is beneficial to discuss laboratory
samples that have been tested as a reference point for the research conducted
here.
16


Jalinoos, Olson, Aouad, and Balch (1995) and Jalinoos and Olson (1995)
have used wall samples with a length, width and thickness of 48 in (1.22 m), 48 in
(1.22 m) and 12 in (0.305 m) respectively. Simulated void inclusions six inches
(0.152 m) square with a thickness ranging from one inch (0.025m) to four inches
(0.102 m) were made of styrofoam. Large areas of simulated honeycombing were
also included in the sample by hand placing concrete without vibration into small
forms. However no information was given regarding the size of these areas.
Schuller and Atkinson (1995) conducted tests on both a small (flat plate) and
large sample. The smaller sample had a length, width and thickness of 17.75 in
(0.45 m), 17.75 in (0.45 m), and 3.5 in (0.09 m) respectively and either had a 1.97
in (50 mm) void or 1.97 inch (50 mm) steel bar cast into the center. In this case, the
inclusions were shown on the velocity tomogram in the correct place but appeared
slightly larger than their actual sizes. The larger sample with a length, width, and
thickness of 59.06 in (1.5 m), 14.96 in (0.38 m) and 47.24 in (1.2 m) respectively
sample included two clay balls of diameter 7.48 in (190 mm) and 3.94 in (100
mm). In order to receive a good image of both inclusions, it was necessary to have
the transducer spacing between 3.94 in (100 mm) and 1.97 in (50 mm).
1.2.4 Transducers and Couplants
The majority of transducers used to generate and detect acoustical waves
are made from piezoelectric materials. (Pla-Rucki, Eberhard & Eberhard, 1993).
Piezoelectric crystals have the property that when they are deformed by mechanical
17


pressure, electric charges on the surface of the crystal are produced. Barium
titanate and quartz are examples of piezoelectric material that is frequently used in
ultrasonics (Krautkramer & Krautkramer, 1990). In NDE, the source transducers are
stimulated by a pulse generator.
As noted in the previous section, Schuller and Atkinson (1995) found it
necessary to vary the transducer spacing in their test in order to receive results.
They found that a spacing between 3.94 in (100 mm) and 1.97 in (50 mm) was
optimal for a sample size with a length, width, and thickness of 59.06 in (1.5 m),
14.96 in (0.38 m) and 47.24 in (1.2 m) respectively. Similarly, Sullivan et al. (1996)
tested different source-receiver configurations.
On a sample with a length, width and thickness of 17.75 in (0.45 m), 17.75
in (0.45 m), and 3.5 in (0.09 m) respectively, Jalinoos and Olson (1995)
successfully imaged the wall with 42 source and 42 receiver locations. The receiver
was moved at 1.08 in (2.74 cm) incremental steps. The source transducer scanned
the opposite side of the wall for each receiver location. Material supporting an
optimal transducer spacing was not found during this literature search.
A couplant acts both as a way to adhere the transducers to a specimen as
well as to encourage the propagation of a clean wave from the transducer into the
specimen. For example, if the location where a test is being attempted happens to
have some honeycomb on the surface, scattering would occur. However, the use
of a grease couplant would fill in those voids and discourage the wave from
scattering.
18


In 1982, Eberhard described a typical couplant as a water bath." Water is
still used as couplant today, however a bath type situation is not practical for most
civil engineering situations. If acoustic tomography is to be readily used in the field,
a specimen such as a bridge abutment, can not be required to be completely
immersed in a water bath. A typical couplant currently in use is grease (Jalinoos &
Olson, 1995). Jalinoos and Olson (1995) also used a water couplant for their new
scanner source.
Gammell (1994) suggests Karo Brand syrup can be used as a shear wave
couplant and Groom and Clean hair tonic can be used as a longitudinal wave
couplant.
1.2.5 Data Collection
Acoustic tomography has proven to be an effective and useful technique for
determining the presence of inclusions in concrete structures. However, the fact
still remains that it is a time consuming technique. As a result, a new ultrasonic
scanning system was developed based on the ultrasonic pulse velocity (UPV)
testing method. The new scanner recorded all of the data required to produce an
image in approximately 1.5 hours. Without the scanner, the data collection alone
would have taken two weeks. (Jalinoos, Olson, Aouad et al., 1995). Typically, it is
the receiver that remains stationery while the source scanner proceeds along the
opposite side of the sample. After each run by the scanner, the receiver transducer
is incrementally moved down the sample.
19


Also, Transue et al. (1997) have developed an array system for collecting
data. The array system holds eight small transducers at regular intervals along its
arm. The array receives signals from one source transducer. A multiplexing switch
on the array is used to send one channel at a time to an analog to digital converter.
(Transue et al., p. 6) Once the data is collected it can be transferred to a computer
for processing. Further information can be found in Transue et al.
1.2.6 Software / Processors
Only one acoustical imaging software program was mentioned by name in
the articles that were found. MIGRATOM Geophysical Tomography Using
Wavefront Migration and Fuzzy Constraints is a program developed by the U.S.
Bureau of Mines (Schuller, Berra, Fatticioni, Atkinson, & Binda, 1994). As can be
seen from the title, MIGRATOM was originally developed for use in the geophysical
area. The program was published in 1994 (Jackson & Tweeton, 1994).
Like the process described in section 1.2.1, the program breaks up the slice
into a pixel grid. The input is in the form of transmitter/receiver locations and pulse
travel time. This information is gathered by independent equipment and stored.
The travel time of a particular ray path is then calculated by averaging the travel
time through each pixel point. This process is iterative until the difference between
the calculated and measured travel times is small. The program requires that
transmitters/receivers be located on at least two sides of the specimen (Schuller et
al., 1994).
20


Typically, a serial approach to processing of the data is used. Serial in the
method of one ray is traced at a time, to completion, before the next ray trace is
begun. However, Sullivan et al. (1996) introduced a parallel (rays traced
simultaneously) processing approach and found that the time required for such an
approach decreases from that of a serial approach for an identical data set.
Similarly, they found that the calculation time in parallel stays relatively constant
regardless data set size. In serial, calculation time would tend to increase
dependent on data set size.
The University of Colorado at Denver, Department of Civil Engineering
currently has a copy of 3DTOM: Three-Dimensional Geophysical Tomography by
Jackson and Tweeton (1996), the same authors that produced MIGRATOM.
3DTOM is a step-up from MIGRATOM in that the program can produce three-
dimensional images. However the program can also produce enhanced 2D images.
Section 2.4 describes the theory and input for 3DTOM.
1.2.7 Resulting Images
In the literature reviewed here, resulting acoustic images varied in quality.
Schuller and Atkinson (1995) found that transducer spacing had to be reduced
down to 3.94 in (100 mm) to 1.97 inch (50 mm) in order to obtain results. They also
stated that the algorithm used was developed for identification of zonal velocity
differences and is not well suited for identification of discrete velocity changes,
tending to smear anomalies over an area larger than the actual feature (Schuller et
21


al., 1994, p. 2220). And in masonry testing there is uncertainty as to which
combination of frequency, pixel size and path length that are optimal (Schuller et
al.). Nor was any information located regarding frequency, transducer spacing,
pixel size and path length which are optimal in concrete construction. It is also
difficult to characterize the anomalies once they are located in the velocity image
(Pla-Rucki & Eberhard, 1995).
1.3 Gauge Potential Study
A gauge potential study is a short term statistical study that examines the
potential of the measurement system to be repeatable and reproducible. This type
of study is a first step in examining the measurement system. It measures the
potential of the system to be capable in the long term. Much more data, collected
over a long period of time is necessary to gain an accurate population sample for
the method under consideration. A gauge potential study is used for preliminary
indications of the systems ability to be repeatable and reproducible. It allows for
the identification and elimination of large sources of variation. Longer term studies
are used to reduce the harder to find sources of variation. In using this study, it is
assumed that (Luftig, 1991):
1. measurement errors are independent,
2. measurement errors are normally distributed, and
3. measurement error is independent of true value.
22


Repeatability is the ability of the measurement system to show close
agreement of repeated data points within a single measurement. Examining the
measurements recorded by Operator A at Point 1, on several different
occasions, is an example of repeatability. Reproducibility is the ability of the
measurement system to show close agreement between measurements that are
obtained by different processes or operators. Comparing the measurements
recorded by Operator A at Point 1 and Operator B at Point 1 is an example of
reproducibility.
The objective of a gauge potential study is to identify, quantify, and
understand the sources of variability within a measurement system (Continuous
Pursuit of Excellence (CPE), 1997, p.2). It is important to note that a gauge study
examines the precision of a measurement system, not the accuracy. Precision
addresses the systems ability to produce or reproduce values in close agreement
with each other, while accuracy compares the system to a known standard (CPE,
1997).
Precision is important in acoustic tomography for several reasons. Sporadic
data values will most likely effect the resolution of a tomogram. Further more, if the
time and money is spent to collect necessary data to produce an acoustic
tomogram of a concrete section, it is simply wise to ensure that the results will be as
credible as possible. This can be obtained by ensuring that the measurement
system is performing in a precise manner. Similarly, transducer placement,
23


pressure, rotation and amount of couplant also effects the systems ability to be
precise (Rens & Greimann, 1997).
A typical gauge potential study consists of two or three operators that
randomly measure eight to ten locations/parts. Each location/part is measured at
least twice and usually three times in a random order. The data that is recorded is
input into a spreadsheet which calculates the repeatability, reproducibility, and P/T
ratio (precision to tolerance ratio) of the system. The P/T ratio is used as a
technique to judge the capability of the measurement system. Suggestions are
included on the spreadsheet based on the final numbers that are presented.
Further description of the spreadsheet and the calculations it performs can be
found in Chapter 2, section 2.3.
A representative of the Continuous Pursuit of Excellence (CPE) Department,
Ball Packaging Operations1, will be involved in the interpretation of the gauge
potential study. The representative will assist in the interpretation of the
spreadsheets' results and offer advice and recommendations based on those
results.
1 9300 West 108th Circle, Westminster, CO 80021-3682. (303)460-5344.
24


2.0 Methodology
2.1 Description of Wall
The concrete wall used in this research has dimensions of 3-10 1/z in
height, 8-0 in length, and a thickness step function that goes from T-0 to T-6.
Figures 2.1 and 2.2 illustrate the elevation and plan views of the wall. The wall has
a total volume of 1.4 cubic yards.
The wall is located on a concrete pad outside the materials lab, at the
Technology (TE) Building on the University of Colorado at Denver (UCD) campus.
2.1.1 Inclusions
Reinforcing steel, simulated voids, low density concrete, and lumber blocks
were all included in the wall. These inclusions were chosen based on the possibility
of these types of distresses or mishaps occurring during construction. A detailed
description of each type of inclusion is included here.
2.1.1.1 Reinforcing Steel
Number 4 reinforcing bars were placed vertically in the lower 24 inches of
both the twelve inch and eighteen inch thick wall sections. Tied perpendicular to
25


Figure 2.1 Elevation view wall inclusions
26


Figure 2.2 Plan view wall inclusions
27


the vertical steel was one #5 bar approximately six inches from the bottom of the
wall and one #4 bar placed eighteen inches from the bottom of the wall. Both bars
run through both the twelve in and eighteen inch sections. The #4 bar is spliced in
the 18 inch wall thickness. The horizontal steel is anchored to the vertical bars
facing the front face of the wall. The vertical steel was anchored into 14 drilled
holes in the bottom lumber support, 6from the back of the wall, of which the back
of the wall refers to the flat side. A hammer was used to be sure they were
anchored properly. The single mat of steel was kept plumb by attaching tie wire to
the mat in one location and securing the wire to the outside of the form through a
small hole in the form. Gauge 10 or 12 tie wire was used to secure all reinforcing.
A single mat of steel was chosen to simulate reinforcing steel that would be
used in a typical wall section. With the change in wall thickness, two different
depths of reinforcing can be tested. Similarly, gaps in the mat were left so that a
rebar locator could easily show positive and negative results.
2.1.1.2 Simulated Voids
The two simulated voids were made of FloraCraft styrofoam. A six inch
diameter styrofoam ball was located in each wall thickness. They were anchored by
use of 10 or 12 gauge tie wire threaded through the center of the ball and form-
work and then anchored to screws on the outside of the form. Styrofoam was
chosen to simulate air voids based on research by Jalinoos and Olson (1995). In
28


their research voids were six inches (0.152 m) square with a thickness ranging from
one inch (0.025m) to four inches (0.102 m).
2.1.1.3 Low Density Concrete
The low density concrete cylinders were made using the following mix
design: 0.295 lb. perlite, 2.5 lb. cement, and 1.25 lb. water. The cylinders are 3
inches in diameter with a length of 4 V* in. The low density concrete was chosen
based on research by Schuller and Atkinson (1995).
A smaller cylinder was also constructed to bench-mark the density of this
material.
2.1.1.4 Wood Blocks
A wood block was secured in both sections of the wall. Wood was chosen
based on including another type of material and the high probability of wood pieces
accidentally falling into concrete during placement in the field. The wood blocks
were constructed of piece 2x4studs. For the larger thickness, two six inch long 2 x
4 pieces were screwed together with a piece of tie wire anchored between them in
the long direction. For the 12 inch thickness only one six inch 2x4 block was used
with the tie wire also anchored in the long direction. During the placement of
concrete the blocks were centered and situated such that they were parallel to the
walls.
29


2.1.2 True Thickness of Wall
Once the wall forms were in place on the concrete slab outside of the TE
building, the following measurements were taken at the top of the forms:
k 3 Z 7| 5 7 :
12 'A" 12 3/16" 12 1/16'
t f 18 3/16
1713/16"
-17 15/16"
Figure 2.3 Wall thickness prior to concrete placement
2.1.3 Concrete Placement
Concrete was supplied by Rocky Mountain Ready Mix Concrete2. They
supplied 1.5 cubic yards of 4000 psi concrete (product 5675). The mix included per
cubic yard: a) 2550 lb. of 3A" aggregate, b) 2211 lb. sand, c) 677 lb. cement, d) 170
lb. fly ash, e) 7 oz. Air, and f) 51 oz. 344 water reducer.
A 1 % inch diameter Wacker vibrator was used to vibrate the lifts of
concrete. Concrete was placed in approximately 12 inch lifts.
Two cylinders were taken for concrete strength testing. The cylinders were
ASTM standard size of six inches in diameter and 12 inches in height. The
cylinders were left outside with the wall, covered by a plastic tarp.
2 RMRMC, 5700 Logan, Denver, CO 80216, (303) 296-8853
30


2.1.4 Concrete Strength
Cylinders were broken in accordance with ASTM C 39 for a seven day
compressive strength of 4600 psi and a twenty-eight day compressive strength of
7500 psi.
2.1.5 Form-work
The form-work for this wall was made out of Vi CDX plywood and 2x4 studs.
A 2 x 12 with a length of eight feet was used for the main bottom wall support.
Because the 2x12 has nominal dimensions of 1 14 x11, an additional seven inch
wide piece was attached to this support over a length of four feet to manufacture
the 18 inch thickness. An additional one inch piece was attached at the opposite
end of the support to manufacture the 12 inch thickness.
The back wall was formed by a 4 x 8 foot single sheet of Vi CDX plywood.
The two side walls had dimensions of 12 x 46 14 and 18 x 46 V.i. The front wall
was constructed from three sheets of plywood with dimensions 47 Vi x 46 14, 47
Vi' x 46 Vi, and 46 Vi x 6. The fastening of the plywood to the bottom form
resulted in the four foot height of the wall being reduced by 1 14 inches.
2.1.6 Screws
All lumber was fastened together with GRABBER V468 6 x 1s/8 screws.
31


2.1.7 Support Against Blow-Out
The forms were secured against blow-out by use of 12 inch and 18 inch
form ties. The form ties were secured on the exterior of the plywood by use of 2 x 4
studs and clamps that fit onto the end of the tie. Eight ties were used in the 12 inch
thickness and seven were used in the 18 inch thickness. The location of the ties
were based on a one foot grid except at the location of the step function, were two
form ties were located six inches on either side of the change in thickness. A 2 x 4
was also secured at the mid point of the 12 inch and 18 inch end of the wall. Figure
2.4 illustrates the form tie locations.
2.1.8 Oil
The inside of the wall forms were painted with Citigo SAE 10 W 30 to help
ease with form removal.
2.1.9 Form Removal
The bracing and wall forms were removed on Friday, August 22, 1997 at
2:30pm in the afternoon. The wall appeared plumb and well vibrated.
32



Figure 2.4 Form tie locations
33


2.2 Equipment
A Velocity Meter (v-meter), manufactured by James Instruments, Inc. in
Chicago, Illinois, will be used to collect data for this research.
The apparatus consists of one panel that produces a travel time in
microseconds. The accuracy of the time travel can be controlled to one decimal
place. The device includes one transmitting and one receiving transducer, two
inches in diameter, and approximately 1.625 inches in height. There is one cord to
supply power, and when charged the equipment can be run on battery power. The
equipment also includes a reference cylinder for calibration. The cylinder is
measured to have a time travel of 26 microseconds. Sounds waves can be induced
at 500 volts or one kilovolt.
A minimum of two people are required to use the v-meter. Each transducer
must be held in place securely by a person. It is also possible to require a third
person to record the travel time and take notes during the testing.
Malhotra and Carino (1991) state that the transducers have a typical
resonant frequency of 54,000 cycles per second for concrete testing (p.170). They
further mention that the transducers are made of lead zinconate titanate ceramic
piezoelectric elements that are inside stainless steel casing.
Malhotra and Carino also suggest that petroleum jelly is a good couplant
and that it should be applied as thin as possible. No description was offered for this
recommendation. Although, it was stated that the transducers should be in full
34


contact with the sample. An air pocket between the transducer and the sample will
cause an error in the travel time obtained.
From the given frequency (f) of the transducers and the compression wave
velocity (c) range established by Malhotra & Carino (1991) of 3000 to 4500 m/s
speed, the range of wave length (I) expected can be determined as follows:
A
nun
c 3000 m/s
7 54 kHz
2.19 in
A
c 4500 m/s
f ~ 54 kHz
3.28 in
As previously mentioned, wavelength should be larger than the concrete aggregate
size to minimize scattering. The 3A inch aggregate concrete mix used in the test
wall is consistent with this requirement. In order for a flaw to be detected, the
wavelength should be less than the size of the flaw. With a wavelength ranging
from 2.19 in to 3.28 in, it can be concluded that the following inclusions in the wall
will be detected:
1. six inch diameter Styrofoam ball,
2. three inch diameter perlite cylinder through its length of 4 3A inch, and
3. 2x4 wood block through it's six inch long dimension.
The #4 and #5 reinforcing bars are not expected to be located using the v-meter
because their 14 inch and 5/8 inch diameter is less than the minimum wavelength of
2.19 inches.
35


2.3 Gauge Potential Study
The gauge potential study will involve three sets of operators and one
measurement system. Each set of operators will include three people: one person
to hold each transducer, and one person to record the data and other observations.
The one measurement system will be travel time, in microseconds. Travel time will
be measured between source and receiver transducers located on either side of the
concrete wall as discussed in section 2.1. Ten different locations on the twelve inch
wall were chosen for the study based on the knowledge that there were no known
inclusions in the path of the wave (i.e., reinforcing steel, styrofoam balls, etc.)
However, there is of course the possibility that a honeycomb may be present, or
that an air pocket became trapped within the concrete without the knowledge of the
constructors. The locations were chosen at points of solid concrete so that the
specification limits could be based on expected travel times through concrete only.
The travel times at these locations should be of similar magnitude. Including
locations with inclusions within the wall would add unnecessary variation to the
study. Figure 2.5 illustrates the test point locations.
Each set of operators will use the v-meter to test each of the ten locations.
All ten of the locations will be numbered on the face of the wall for identification
purposes. Each operator will test the ten locations in random order. The random
sampling scheme will be generated through the use of a random sampling table.
Using this spreadsheet is important to ensure that there is no bias is the testing
procedure.
36


37


Upon completion of this first round of testing, the set of operators will
complete the same process two additional times. It is important that each of the
successive tests be done randomly and that the operator reading the values from
the v-meter not attempt to remember the value that was previously recorded at a
location, as this will effect the outcome of the study. The operator recording the
time measurements should also record observations of the testing procedure.
Observations might include differences between operator procedure, temperature,
amount of traffic, direction of sun, or anything out of the ordinary which may effect
the measurement system.
All of the data that is collected will be recorded in spreadsheets that are
reproduced here with the permission of the Continuous Pursuit of Excellence
Department, Ball Packaging Operations. The spreadsheets record the data,
calculate the repeatability and reproducibility of the measurement system, and
offers possible conclusions that can be drawn from the calculations. Figure 2.6
illustrates an example of a gauge potential study output spreadsheet. A description
of the calculations performed by the spreadsheet is offered in the following section.
2.3.1 Description of Spreadsheet
The spreadsheet, shown in Figure 2.6, which is used to perform the gauge
potential study allows for identifying test information to be recorded in the header.
38


Figure 2.6 Sample spreadsheet
39


Additional notes should be added in the lower shaded quadrant regarding
procedure used and descriptive information.
Each data point that is collected is recorded in the column and row that
corresponds to the appropriate operator, point number, and test round combination.
For example, each of the three sets of operators will input three columns of data
that consists of 10 rows each. It is important that although the data points are
collected in random order, that they be input into the spreadsheet next to their
corresponding point identifier. This is necessary so that apples are compared to
apples. As discussed in section 1.3, an example of repeatability is measurements
of Operator A at Point 1 on several different occasions. Comparing
measurements taken by Operator A at Point 1 and Point 2 does not offer an
accurate representation of repeatability, because there may be independent
external factors at each location that effect the two points differently.
A lower specification limit and upper specification limit is also input. Using
the compression wave velocity range established by Malhotra and Carino (1991) of
3000 to 4500 m/s yields a upper specification limit of 69.14 ns and a lower
specification limit of 103.72 ns for the twelve inch wall (see Appendix A for
calculations of these limits).
Once the data is input into the spreadsheet, the following calculations are
performed by the spreadsheet (Luftig, 1991). The order in which the calculations
are presented correspond to the numbered arrows in Figure 2.6.
40


(1) The range, Rj, for each set of measurements, for each appraiser. The
variable / refers to part number.
R, = max. measurement min. measurement (2.1)
(2) The average of the ranges, R*, for each appraiser. The variable n,
refers to the total number of parts, and the variable k refers to the
appraiser number.
n
_
R = ^ (2.2)
n
(3) The average of the average of the ranges, Rr. The variable s, refers to
the total number of appraisers.
s __
_ I*
Rr = (2.3)
k
(4) The variable d2m is a constant for converting average ranges into
estimates of population standard deviation (Luftig, 1991, p.D-3). The
value of 1.69 is based on three sets of data being used with ten data
points for each set.
(5) Repeatability an estimate of the standard deviation, (J-pt, within
operators. An estimate of the standard deviation is made because the
formula for calculating standard deviations becomes inaccurate as the
number of data points decreases. The standard deviation, or estimate
of the standard deviation, is a measure of how wide spread the data is.
41


For example, if a normally distributed set of data has a high standard
deviation, then there is much difference between the high and low
values of the data.
U2(rp0
(6) The average of each measure, Xmai*, for each appraiser.
XtraW
i=l
n
(2.5)
(7)
The average of the average measures,
Xk.
(8)
3
^ XtnW
triaJff1
#of trials
The range of the average of the averages, Rx .
Rx = Xk(max) Xk(min)
(2.6)
(2.7)
(9) The d2(.pd) variable of 1.91 for reproducibility is based on three sets of
data being collected.
(10) Reproducibility an estimate of the standard deviation, a^, between
operators.
R;
d2(n>d)
(2.8)
(11) The P/T ratio; the precision to tolerance ratio is the ratio between
natural tolerance of a measurement error and the blue print
42


tolerance (CPE, 1997, p.22). Tolerance in this case, is the difference
between the upper specification limit (USL) and lower specification limit
(LSL) previously computed in this section Precision is a value that
combines the effects of both repeatability and reproducibility.
f~Z j ^ 2
O X V CJrpd f* Orpt
USL LSL
(2.9)
(12) The P/T ratio can also be calculated individually for repeatability and
reproducibility using equation (2.9).
Guidelines for P/T ratio results are listed on the left hand side of the
spreadsheet. A result of 10% or less indicates that the measurement system has
the potential of being capable; that the system is very likely to pass a long term
gauge study. Suggestions are also presented on the spreadsheet regarding the
comparison of the reproducibility and repeatability outcomes.
It is important to examine the P/T ratio as well as the values for repeatability
and reproducibility. Reducing the repeatability and reproducibility will in turn
decrease the P/T ratio. However, it is important to note that the P/T ratio is also
directly dependent on the tolerance levels set by the user. In general, the smaller
the tolerance, the larger the P/T ratio and vice versa.
2.3.2 Recommendations for Conducting a Gauge Potential Study
The following are recommendations to consider while conducting a gauge
potential study (CPE, 1997):
43


1. Understand the operation of and the equipment and to be used;
2. Review the standard operating procedure. If one is not available, one
should be developed;
3. Ensure that all operators to take part in the study are familiar with the
testing procedure;
4. Calibrate the equipment before beginning the study. Do not attempt to
calibrate the equipment during the study as this will effect the results.
5. Do not let operators see any of the data until the study is complete;
6. Perform the study under normal operating conditions;
7. Report all of the measurements the same way. One operator should not
round when another has not. It is recommended that an independent
person record the data, while the two operators conduct the
measurement;
8. Take detailed notes regarding operating conditions and unusual
circumstances. Record these on the spreadsheet;
9. Allow equipment to reach room temperature before beginning study;
and
10. Have the output reviewed by a person with gauge study experience.
2.4 3DTOM Software
3DTOM: Three-Dimensional Geophysical Tomography is a software
program developed by Jackson and Tweeton (1996) at the United States Bureau of
44


Mines. The program takes an input file which includes source-receiver locations
and travel times, and by using a simultaneous iterative reconstruction technique
(SIRT) produces a velocity tomogram. The following sections outline the
information required to run this program.
2.4.1 Data Files
Data files should be constructed in a spreadsheet. The first two lines of the
file are header lines, as illustrated I Figure 2.7. Header lines can have a maximum
of 120 characters per line and should describe the data within the file. The
following lines contain the eight (8) columns of source, receiver, and travel time
data. The first column of data contains the ray identifier information. The rays can
be identified by integers only. The following six (6) columns contain the X, Y, and Z
coordinates of the source and receiver transducer, respectively. The last column
contains the travel time measurement. No units are input in the data. However, the
units must consistent and can be defined in 3DTOM. The program will use the
defined units in order to include the velocity units on the resulting tomogram. Figure
2.7 illustrates the data file format.
45


Figure 2.7 Example of 3DT0M data set
2.4.2 Units
Units must be kept self consistent. They are not included as input in the
data file, but can be specified within the program. Units of mm, cm, m, in, ft, fj.s, ms
and s can be defined for use in graphics displays.
2.4.3 Coordinate System
A Cartesian right-hand coordinate system is employed, with the positive z-
axis downward as shown in Figure 2.8.
46


z
Fig. 2.8 3DT0M Right-hand Cartesian coordinate system
2.4.4 Constraints (Jackson & Tweeton, 1996)
In order for tomography to be an effective tool in nondestructive testing it is
necessary for the image of an object to be unique. At times this can be limited by
insufficient ray coverage and as a result imperfect travel time data. As a result, the
program has an option to input known information about the test specimen in the
form of constraints. Two options are given: global and nodal constraints.
Global constraints specify upper and lower bounds (for the entire model) for
velocity reconstruction. Nodal constraints are typically used to specify known layers
within a sample. Nodal constraints can be specified for single nodes or groups of
nodes within the grid. Both of these constraint options contain a fair amount of
uncertainty. However nodal constraints are more limiting and therefore require
additional calculations based on the uncertainty of the knowledge being used as a
constraint.
47


Nodal constraints include two parts, an integer and a fraction. The integer
indicates the actual velocity constraint. For example a 0 integer indicates that the
velocity is unrestrained and can be freely adjusted by the program; a negative
integer indicates the velocity is to be maintained over the entire group of nodes; and
a positive integer indicates that the velocity is to be held fixed at the value of the
starting model. The fraction indicates the uncertainty of the velocity constraint, it is
also referred to as the fuzz factor. The fuzz factor has a value between 0 and 1,
with a value of one indicating a higher level of uncertainty. The calculation that
3DTOM uses to incorporate these constraints is (Jackson & Tweeton, 1996):
v = Vof + vi (1 f) (2.10)
Where v0 is the unconstrained velocity determined by SIRT, v-i is the constrained
velocity, and f is the fuzz factor. The purpose of these constraints is to find the best
agreement to the travel data.
2.4.5 Theory (Jackson & Tweeton, 1996)
3DTOM is based on the simultaneous iterative reconstruction technique
(SIRT). The model is constructed as a grid of nodes with intervening voxels. SIRT
includes three procedures that are cyclically repeated until certain criteria are met.
1. forward calculation of model travel time,
2. calculation of residuals, and
48


3. application of velocity corrections.
Forward calculation of travel times compares a calculated travel time for a
particular ray to the measured travel time of the ray. The following equation is used
(Jackson & Tweeton, 1996):
M
ti = 2>dij (2.11)
j=i
Where M is the number of voxels in the grid, tj is the measured travel time for ray i,
dij is the distance traveled by ray i through voxel j, and Pj is the average slowness
(inverse of velocity) of the ray in voxel j. The variable dy is only nonzero for those
voxels through which ray i passes. The difference between the right hand side and
the left is compared and used to calculate a correction factor for that particular ray.
The correction factor of all individual rays is calculated before being applied to the
rays, hence the simultaneous reconstruction.
The correction factor for ray i in voxel j is (Jackson & Tweeton, 1996):
At du
Apy = ------s------- (2-12)
NP£(dik)2
k=1
Where Apy is the slowness correction of ray i in voxel j, At is the travel time
difference (residual) for ray i, dy is the path length of ray i in voxel j, Np is the
49


number of rays in voxel j, M is the number of voxels in the grid, and dik is the path
length of ray i in each of the M voxels in the grid.
The net correction factor for voxel j is (Jackson & Tweeton, 1996):
N
APj = 2>PU (2.13)
i=i
Where Apj is the slowness correction for voxel j, N is the number of rays, and Ap^ is
the slowness correction of ray i in voxel j. For instance, if there are three rays that
pass through voxel number four, equation (2.13) would look like:
3
Af>4 = ^ APi4 = AP14 + AP 24 + AP 34. Correction factors for each node are then
i=i
calculated by averaging the correction values of the attached voxels.
2.4.6 Ray Tracing (Jackson & Tweeton, 1996)
3DTOM uses four different methods of ray tracing: a) straight line, b) ray-
bending, c) net-work theory and d) a hybrid approach. The straight line method
assumes the ray travels in a straight line and generally requires the least
computation time. It is generally recommended that at least one straight line
computation be attempted first, to get a rough (course) grid of velocities.
Ray bending is an iterative process that consists of a division phase and an
adjustment phase. The division phase connects a straight line between the source
50


and receiver and calculates the midpoint. At the midpoint the velocity gradient is
calculated and the midpoint is displaced accordingly. The adjustment phase
recalculates the gradient at the new midpoint and displaces the midpoint again. At
least one iteration takes place in the adjustment phase before the next division
phase. The same procedure is carried out with the segments just created, in
several iterations, until a circular arc is approximated. The travel time of the ray is
calculated for each iteration. When a stable minimum is reached calculations are
terminated. In general 5 iterations, which yields 32 segments is sufficient within 2%
of the measured travel time (Jackson & Tweeton, 1996).
Net-work theory uses a fixed grid of nodes and calculates a minimum travel
times along straight lines between nodes, effectively producing a shortest-path
tree. This is a time consuming method and may converge to a local minimum
instead of a global minimum. In other words it a local minimum may be a legitimate
ray path but not necessarily the first arrival.
The hybrid method incorporates both ray-bending and net-work theory. It is
the most intensive in calculation time but it is also the most accurate.
It is suggested that a fast and simple iteration be tried first followed by
more intensive ray tracing and calculation. Similarly, for a model with a uniform
starting velocity, at least one straight ray calculation should be performed before
ray-bending is considered.
51


2.4.7 Conclusions
3DTOM is unable to calculate reflected waves paths, however reflected
wave paths are never first arrivals so this is of little concern. For first arrival ray
paths, ray-bending is the most appropriate ray tracing procedure.
Jackson and Tweeton (1996) state that a typical P-wave velocity for air is
3000 m/s and for water is 1500 m/s. They further state that as density increases,
velocity increases and as temperature increases so too does velocity.
2.4.8 Step-by-step Procedure (Jackson 8i Tweeton, 1996)
1. Create a data file. It is suggested that a spreadsheet be used. Save the
spreadsheet as an ASCII text file with the suffix .3DD. (Choose Save
as type: Text (Tab delimited) and actually type file name as
name.3DD.) Store the file in an appropriate sub directory of the
\3DTOM main directory that corresponds to the data set. No
preprocessing is required for travel time data as it is for amplitude data.
2. Load data file. [F3]
3. Specify units, [options units specify]
4. Specify coordinate system, [options coordinate system define]
5. Examine data. [F7] Compare plots of data in 3DTOM with plots
constructed in spreadsheet.
52


6. Check default grid, [model view grid] To change voxel size use
regrid.
7. Set-up starting model and apply constraints. Starting model has a
uniform velocity that is the average from the data set. Use model editor
to apply constraints.
8. Set inversion parameters. Save with the suffix .3DO.
9. Run inversion routine.
10. Save calculations and model output. Model output is saved as
3DTOM.3DM. This file must be renamed and saved; it will be rewritten
for each run of the program.
11. View final model. Velocity map can be copied and pasted into a word
processing document.
12. Compare calculated model with other models. Models may be run with
different inversion parameters and compared.
13. Export model for high quality output presentation.
14. Check sampling density. It is useful to view plots of raypath sampling
density in order to evaluate how resolution may vary within the grid. (p.
28)
2.5 Data Collection for 3DTOM
The data required as input for 3DTOM consists of source and receiver
transducer coordinate locations and travel time between the two transducers.
53


Typically, the source transducer remains stationary while pulses are generated to
many different receiver locations. At the end of one such cycle, the source is
moved and again pulses are generated to the different receiver locations. Figure
2.9 illustrates the ray coverage generated by this method. It is suggested that the
transducers be located no closer than three inches from either the top or bottom of
the wall.
For this research, data will be collected at a variety of different sections of
the twelve inch concrete wall. Sections will be taken at the locations of the
styrofoam ball, wood block, and perlite cylinder. For each of these locations, a
tomogram will be produced with a three inch transducer spacing, and one with a
one and one half inch transducer spacing. Figure 2.10 illustrates the tomogram
section locations.
2.5.1 Standard Operating Procedure for Data Collecting using V-Meter
1. Feel the wall to check for roughness and raised spots.
2. Liberally apply Vaseline to test location on wall.
3. Apply Vaseline to bottom of transducers. Make sure that the lubricant is
evenly applied and that there is no dirt attached. (When you stabilize
transducer on wall, the should ooze out around the circumference of the
transducer.)
54


I
I
I
I
i
i
I
i
oo
Q/
00
z:
c
cm
Figure 2.9 Transducer-receiver ray coverage
55


<12' Thickness)
cu
s
J
CO
%
OJ
\
I
%
o
I
cu
so
>
>
>i
Sj
>
OS
>
CO
>
[V -
>
so
>
in
>
^ -
>
ro
X
in
Figure 2.10 Tomogram section locations
56
-.3/1


4. Center transducer on marks. Locations to be tested should have a cross
mark large enough to be seen outside of the transducer. It is
recommended that the transducer also be marked to indicate how the
transducer should be oriented on the cross mark. The mark on the
transducer indicating the top of the transducer should have a different
identifying characteristic (color) so that the transducer is always centered
in the same orientation. Figure 2.11 illustrates transducer alignment on
the location marks.
emu iiym diucd .
Figure 2.11 Orientation of transducer on location marks
5. Place transducer on wall by using thumb and forefinger of both hands in
the four comers of the transducer. Stabilize hand on wall by using ring
and pinky fingers.
6. Hold securely, but do not use excessive force. Apply enough pressure
so that Vaseline oozes out around transducer. Once Vaseline has
oozed out around the sides, reduce pressure that is applied and hold
Top mark.
57


securely. The Vaseline will have a suction effect between the transducer
and the wall.
7. Make sure transducers are on wall as flat as possible. Do not let them
rock. Find another spot if necessary.
8. Person with receiving transducer gives the okay to take measurement.
9. Person with transmitting transducer gives the okay to take
measurement.
10. During the gauge potential study, the third person records the
measurement from the v-meter panel. During other data collection, a
third person to read the measurements is not necessary but it does help
to facilitate the process. Wait until number remains stationery for a few
seconds before reading. It if does not settle, chose a best estimate. It
is recommended that some sort of data collection sheet be developed
for easier recording.
11. After measurement is taken, proceed to next test location and follow the
above procedure. Transducer should make a suction sound when
removed from the sample.
2.5.2 Sample Data Collection Sheet
The data collection sheet should include all of the information to produce
and identify a tomogram of the section in question. This includes: date, time,
58


weather, receiver and transmitter locations, time travel, and units of measurement.
For example:
Tomograr Date: Time: Weather: Section: n Data Collection Sheet I Section Line
::5 p

Wail Thickness: I I
Ray Transmitter Receiver Time
# X y z X y z
1 0 0 0 0 18 0
2 0 0 0 0 18 3
3 0 0 0 0 18 6
4 0 0 0 0 18 9
5 0 0 0 0 18 12
Figure 2.12 Sample Data Collection Sheet
59


3.0 Results
This chapter will present the results of the gauge potential study and
describe the tomograms produced by utilizing the United States Bureau of Mines
3DTOM software program (Jackson & Tweeton, 1996) and the James Instruments
velocity meter (v-meter).
3.1 Gauge Potential Study Results
Two gauge potential studies were conducted. The first study was conducted
utilizing a preliminary standard operating procedure that was developed based on
initial experience using the v-meter. The second study was conducted based on
the results of the first study and the corresponding improvements to the standard
operating procedure. The spreadsheets indicating the results of the studies can be
found in Appendix A.
3.1.1 Study No. 1
Table 3.1 illustrates the results of Study No. 1. These results indicate that
there is little potential for the v-meter measurement system to pass a long term
60


study. Although the individual repeatability and reproducibility P/T values are in the
range to be good, potential to pass combining these values for an overall P/T ratio
yields a not likely to pass value of 24%. These values indicate that there is a wide
variance in the time measurements within an operator and between operators.
DESCRIPTION VALUE P/T RANGE
Repeatability 16.09% 11-20% good, potential to pass
Reproducibility 17.81% 11-20% good, potential to pass
P/T ratio 24.00% 21-30% not likely to pass
Table 3.1 Results of gauge potential Study No.1
Study No. 1 was conducted on two days by three combinations of four
different people. Data sets were collected between noon and 3pm on Thursday,
October 2, 1997 and Saturday, October 4, 1997. The temperature was between 84
and 88 degrees, with no breeze or clouds. The concrete wall was fully exposed to
the intensity of the sun on both sides. The transmitting transducer was used from
the front of the wall, while the receiving transducer was used from the back of the
wall. Figure 2.2 illustrates this orientation of the wall.
During Study No. 1 operators 2, 3, and 4, handled the transducers. Each of
these operators had varying degrees of experience using the equipment, and used
the equipment according to the initial operating procedure developed. None of the
61


operators gave attention to transducer orientation, and various amounts of Vaseline
were applied to the transducer and to the wall. Operator 3 tended to use less
Vaseline than operators 2 and 4. Due to the heat, the Vaseline was melting and
was less viscous (thick) than its typical room temperature state.
Although, a method of holding and applying pressure to the transducer was
outlined in the procedure, variations of that method were still employed. Varying
degrees of pressure were applied to the transducer and a varying number of fingers
were used to stabilize the transducer on the wall.
Depending on the test point location on the wall to be tested, the operators
orientation to the transducer would also vary. Sometimes the operator would be at
eye level with the transducer, while other times the transducer would be centered
from the top down perspective. Obviously, an eye level perspective would aid in
centering the transducer more precisely on a specified location.
The first set of tests were conducted on a weekday, while the second two
sets of tests were gathered on a weekend. This is of possible significance because
of the location of the wall being in close proximity to a city street with a light-rail
station. Depending on the time of day and day of the week, varying degrees of
traffic can be expected. Vibrations of the light-rail train or traffic could influence the
readings of the v-meter.
Based on the results of Study No. 1, several changes were made to the
standard operating procedure prior to conducting Study No. 2. The changes were
as follows:
62


1. further description of the amount of Vaseline to be used;
2. increased mark size on wall for centering purposes; and
3. added temporary centering marks to transducer to have a consistent
orientation for all tests.
Prior to beginning Study No. 2, all of the operators re-read the standard operating
procedure and asked for further clarification on any questions.
3.1.2 Study No. 2
Table 3.2 illustrates the results of Study No. 2. These results indicate a
definite increase in performance from Study No. 1.
DESCRIPTION VALUE P/T RANGE
Repeatability 6.02% 10% excellent
Reproducibility 12.48% 11-20% good, potential to pass
P/T ratio 13.86% 11-20% good, potential to pass
Table 3.2 Results of gauge potential Study No.2
Data for this study was collected on Tuesday, November 4, 1997, during a
one hour period. The temperature was approximately 55 degrees with a light
breeze and direct sunlight. The back side of the wall was directly exposed to the
sun, while the front (or notched) side was shaded from the sun. As in Study No. 1,
63


the transmitting transducer was used from the front of the wall and the receiving
transducer was used from the back of the wall. Refer to Figure 2.2 for wall
orientation.
Two of the operators for this study were also involved in Study No. 1.
Operator 1 recorded the time travel for set number one, of Study No. 1 and operator
2 had recorded the data for set number two and held the transducer in place for set
number three. All of the operators read the revised standard operating procedure
prior to beginning testing for Study No. 2.
Sufficient Vaseline was used by all operators, in Study No. 2, such that the
Vaseline oozed out around the circumference of the transducer when applied to the
wall. When removed from the wall, a suction sound was heard. The transducer
was centered at test locations, using the large marks on the wall and the indicators
on the transducers. Appendix E contains photographs depicting transducer
orientation.
Again, variations of the stabilization method for the transducer on the wall
were used. Sometimes, the thumbs and first fingers were used in the corners of the
transducer, as described in the SOP, and sometimes only the two thumbs were
used in the center of the transducer. Similarly, the transducer was not consistently
centered on the marks while the operator was at eye level. This is particularly true
for the test locations at the bottom of the wall.
64


As mentioned previously, data for this study was collected on a single day,
which was a weekday in the early afternoon. Similar to Study No. 1 there was
heavy traffic and several light-rail trains passed by during the testing.
3.1.3 Comparison of Study No. 1 and Study No. 2
The improved results of Study No. 2 over study No. 1 is encouraging. Table
3.3 in the previous section presents the results of both studies. The 10% decrease
in P/T value, indicates that the measurement system does have the potential to
produce quality results.
Study No.1 Study No.2
DESCRIPTION VALUE VALUE
Repeatability 16.09% 6.02%
Reproducibility 17.81% 12.48%
P/T ratio 24.00% 13.86%
Table 3.3 Comparison of gauge potential Study No.1 and Study No. 2
The decrease in repeatability P/T from 16.09% to 6.02%, which indicates an
increase in repeatability in the system, indicates that a set of operators are able to
produce results with a small range of values. This can be credited to the change in
procedure regarding transducer orientation. The centering marks on the wall, as
65


well as marks on the transducer, appear to effect the travel time measured by the v-
meter. This makes sense in that without the marks, there was no guarantee that
each operator stabilizing the transducer would actually hit the same location time
after time. Not using the same location varies the distance between transducers
and therefore, affects the travel time measured. If a different transmitter and
receiver transducer location is used for each measurement, many different paths
will be followed and likewise many different times recorded.
Another possible factor in decreasing the repeatability value, is the
increased amount of couplant that is used. The couplant ensures there is good
contact between the transducer and the sample. This could also explain why the
time travels in Study No.1 are slightly larger than those in Study No.2 when more
couplant was used. It is possible that with the lesser amount of couplant being
used in Study No. 1 a small void was left between the transducer and the test
surface, which would increase the travel time. However, the change in travel times
could also be due to the change in ambient temperature or experience level of the
operators. Amount of couplant, ambient temperature, and experience of the
operator are variables that a long term gauge study could be used to examine.
The decrease in reproducibility value from 17.81% to 12.48%, which
indicates an increase in the reproducibility of the system, is likely due to the
operators making a greater effort of using an identical procedure to one another,
i.e. following the standard operating procedure developed. However, the
reproducibility value is still larger than repeatability value, see Table 3.3. As can be
66


seen on the sample spreadsheet in Figure 2.6, when reproducibility is large in
comparison to repeatability, the following are possible sources (CPE, 1997):
1. one gauge may be out of calibration;
2. operators may be using different methods; and
3. one or more gauges may need to be replaced.
In this study, a gauge refers to a transducer. Based on this information it is likely
that the sets of operators were not using the identical procedures; they were using
personal interpretations of the procedure. More experience using this specific
method would likely reduce this reproducibility value.
Another factor that may be affecting the reproducibility, is the technique the
operator uses to record the travel times. The standard operating procedure in
section 2.5.1 describes the technique as follows:
... Wait until number remains stationery for a few seconds before reading. If
it does not settle, choose a best estimate.
As there was not an manual describing the method of recording travel times, this
method is based solely on initial experience using the equipment. Obviously, this
type of wait-and-see technique allows for a great deal of variance between
operators recording the data. In fact, it was noted in Study No. 2 that all three
recording operators adopted a different wait time before recording the travel time
measurement. This type of error would be consistent with the high reproducibility
error and low repeatability error. The method used to record within a data set
(repeatability) would be consistent, but the method employed between sets
(reproducibility) would be different.
67


3.2 Tomogram Results
Tomograms of four different sections of the twelve inch wall were produced.
One set of data for each section was recorded with a transducer spacing of one
and one half inches. From this data an additional data set corresponding to a three
inch transducer spacing was extracted. Sections were taken through the following
locations:
1. the six inch diameter styrofoam ball,
2. the long direction of the 2x4 wood block,
3. the three inch diameter perlite cylinder, and
4. an area with no known conclusions.
Figure 2.10 illustrates the location of these sections.
3.2.1 Initial Assessment of Concrete Wall
Prior to collecting the necessary data for the production of the above
tomograms, an initial examination of the wall was completed. As Jalinoos, Olson,
and Sack (1995) recommend, the velocity meter was first used as a means of initial
defect identification within a structure. Straight shot travel times, from one side of
the wall to the other, were collected and examined for indications of low density
areas.
The straight shot travels times were collected based on a three inch grid
drawn directly onto both sides of the concrete wall. Figure 2.10 illustrates the this
68


grid. The transducers were held to the wall at corresponding grid intersections on
the front and back side of the wall. Travel times were collected at each of the
intersections on the grid for a total of 225 straight shot travel times.
The travel times and corresponding node locations were then input into a
software program entitled SURFER Version 5.0 (Golden Software, Inc., 1994). The
program takes the node locations and travel times and produces a surface map
where the X and Y axes correspond to the node coordinates and the travel times
are on the Z axis. Figure 3.1 illustrates the initial surface plot obtained from the
straight shot travel times.
Figure 3.1 SURFER contour plot of straight shot travel times
69


At first glance, Figure 3.1 appears to not reveal any credible conclusion.
Although, the points indicated by the arrows appear to correlate to the known
inclusions of the styrofoam, wood, and perlite. As a second step in the post
processing of the data, travel times with a value of 85 microseconds or less were
truncated to a value of 85. Figure 3.2 illustrates the surface plot of this revised
data.
400-
2.00-j
i_,..............................................
2.00 4.00 6.00 6.00 1 0.00 1 2.00 1 4.00
Figure 3.2 SURFER contour of 1st post-processing of data
Figure 3.2 indicates with even greater certainty that the three inclusions are
identified with a longer travel time than that of concrete. However, other inclusions
appear to also be present, as indicated by the contours along the top and sides of
the plot. This is possibly due to errors in data collection such as insufficient
70


couplant, or the fact that these data points are close to a boundary. Boundaries are
known to effect data in acoustic transmission due to the lower degree of
information density (Rhazi, Kharrat, Ballivy & Khayat, 1996).
Finally, as a third step in the evaluation of the data, travel times under 90
microseconds in value were also truncated to 85. This was performed to try and
eliminate the outlying contours around the perimeter of the plot. Figure 3.3
illustrates this final surface plot.
Too Too ?oo 8^00 io!oo iToo u!oo
Figure 3.3 SURFER contour plot of 2nd post-processing
Figure 3.3 clearly indicates the three known inclusions that were expected to
be identified based on the frequency and corresponding wavelength of the velocity
meter. Also as expected, the steel reinforcing bars were not identified due to their
diameters being less than the wavelength of the propagating wave.
71


The plot in Figure 3.3 was then used to identify the section locations for
detailed scans to be completed; the hash marks on the X and Y axes correspond to
the grid lines on the wall. Detailed scans will be taken at:
1. X = 4, Y = 12 STYROFOAM BALL
2. X = 7, Y=10 WOOD BLOCK
3. X=10, Y=12 PERLITE CYLINDER
The scans will extend 7.5 inches above and below the indicated XY intersections.
Comparing the above coordinates with the coordinates where the inclusions
were originally placed, prior to concrete placement, appears to indicate that all of
the inclusions have moved to a certain degree. Figure 3.4 illustrates the original
intended location of the inclusions in white and the apparent current location of the
inclusions in gray. Additional scans will be taken at the locations of the perlite and
wood inclusions due of their apparent movement. Figure 2.10 illustrates these scan
locations.
72


Figure 3.4 Comparison of original intended inclusions location (white) to
apparent current inclusion location (gray).
3.2.2 3DTOM Input Information
Tomograms were developed with global constraints of 4500m/s maximum
velocity and 3000m/s minimum velocity. No nodal constraints were implied.
Background an theory regarding constraint values can be found in section 2.4.4.
A pixel grid automatically developed by 3DTOM was used for initial runs.
For the three inch transducer spacing an initial pixel grid of 7x7, 49 pixel locations,
in the YZ plane was generated. For the one an one half inch transducer spacing a
grid of 9x9, 81 pixel locations, was generated.
Five straight ray iterations, followed by twenty-five curved ray iterations were
employed for each tomogram. The choice of number and type of iterations in this
73


case is based on Transue et al. (1997) and is user defined in the program. The
curved ray iterations are computed by the simultaneous iterative reconstruction
technique (SIRT) and the ray bending (curved) method. Information and theory
regarding this technique can be found in sections 2.4.5 through 2.4.6. Tomograms
are presented here with an interval of 100 m/s.
In examining these tomograms it is important to recognize that the
dimensions of the map are for the pixel grid, which extends outside the location of
the transducer locations. The area of the tomogram that refers to the cross section
of interest is actually within the boundaries of the map. Similarly, attention should
be paid to the notation on the axes. The tomogram appears to have square
dimensions when the true dimensions of the section being imaged is roughly
15x12. Dimensions of anomalies should be scaled from the tomogram, not just
examined visually.
3.2.3 Tomogram of Styrofoam Ball
Data was collected for the tomograms of the six inch diameter styrofoam ball
beginning at 6:30am on Tuesday, November 11, 1997. The temperature was 25
degrees and it was snowing. Data was collected at a transducer spacing of 1.5
inches. As mentioned in section 3.2, two individual runs will be processed from this
data; one with a spacing of three inches and one with a spacing of 1.5 inches.
The data was collected along the 4V vertical grid line, between horizontal
grid lines 14H and 9H. This yields a scan of fifteen inches in total length. There
74


was a form tie hole between horizontal lines 14H and 13H therefore no data was
collected at point 13.5H-4V. Similarly, data collected at point 13H-4V were shifted
down a half an inch to avoid this same form tie hole. Figure 2.10 illustrates the grid
layout and section locations. The thickness of the wall at this location is 12
Appendix B contains the data and 3DTOM input files for the styrofoam ball
tomograms.
3.2.3.1 Styrofoam Ball at Grid Line 4V
Figure 3.5 illustrates a tomogram of the styrofoam ball section based on a
three inch transducer spacing. The arrow indicates the low velocity area believed to
be the styrofoam ball. Using a three inch transducer spacing over the fifteen inch
scan area consists of six transmitter and six receiver locations. This yields a total of
thirty-six rays for the analysis.
75


Slice at X= O.DDD
Velocity n/vs
-2.50E-2
3.65E-2
9.81E-2
a
Esl
0.16
0.22
4.00E-3
3.906-3
3.806-3
3.706-3
3.606-3
3.SOE-3
Y m
Figure 3.5 Styrofoam tomogram using 3 transducer spacing; 7x7 pixel grid
Figure 3.6 illustrates a tomogram of the styrofoam ball based on a 1.5 inch
transducer spacing. The 1.5 inch transducer spacing consists of ten transmitter and
ten receiver locations for a total of 100 rays used for the analysis.
76


Slice at X=
D.DDD
Uelocity /vs
4.30E-3
4.20E-3
4.106-3
4.OOE-3
3.90E-3
3.80E-3
3.VOE-3
3.60E-3
3.50E-3
Y m
Figure 3.6 Styrofoam tomogram using 1.5 transducer spacing; 9x9 pixel grid
In both Figure 3.5 and 3.6 there is a dark, low velocity, area towards the
center of the velocity map as indicated by the arrow. This low velocity area appears
to be consistent in location and approximate size of the styrofoam ball placed in this
section. Table 3.4 compares the information describing this low velocity area
shown in both tomograms.
77


Fiqure 3.5 Fiqure 3.6
3 spacing; 7x7 grid 1.5 spacing; 9x9 grid
Velocity 3500 m/s 3400 m/s
Y dimension 4.88 in (,124m) 7.70 in (,1955m)
Z dimension 3.75 in (.0953m) 3.46 in (,0878m)
Center Point Y= 5.24 in (0.133m) Y= 5.11 in (0.13m)
Z = 6.06 in (0.154m) Z = 6.05 in (0.1536m)
Table 3.4 Comparison of low velocity areas for styrofoam tomogram
It is encouraging to see that both the velocity and center point location of the
low velocity area relatively close in value. The horizontal component of the center
point location in both tomograms is relatively close to the intended location. The
styrofoam was visually centered in the 12.25 in dimension of the wall during
concrete placement. This corresponds to Y=6.125 in (0.1556m). Considering the
method with which the ball was centered, it is likely that some movement or
placement error took place. It is viable that the horizontal component of the center
point in the tomogram is an accurate representation.
The vertical component of the center point appears to have moved also from
the intended position. The intended position of Z=7.5 in (0.1905m) was attempted
by anchoring the styrofoam with tie wire which was secured to the form work during
concrete placement. Although, the tie wire was secured as tightly as possible, there
78


is the possibility of slack, allowing the styrofoam to float. Compared to the assumed
Z=6.06 in (0.154 m), it appears that the styrofoam ball floated towards the top of the
wall approximately one inch. This corresponds to the prediction of the initial straight
shot assessment in Figure 3.4 that the styrofoam ball had indeed moved upward.
Although the above comparisons are interesting, these types of
observations are only conclusive when prior knowledge of a defect or inclusion is
available. Without dissecting the wall for verification some uncertainty remains.
Unfortunately, dismantling the wall defeats the purpose of this method.
The dimensions of the low velocity area are not as conclusive as the velocity
and center point values. The low velocity area in Figure 3.5 is more contained to a
specific area than that in Figure 3.6. Figure 3.6 appears to have more spreading of
the anomaly in the horizontal direction. The low velocity area, in general, has
smaller dimensions than would be expected for a section through the center of a six
inch diameter object.
3.2.3.2 Styrofoam Ball at Grid Line 4V Increased Pixel Grid
Another area of interest is the number of pixels used in the construction of
an acoustic tomogram. The following two tomograms that are described, were
constructed using identical data as the above, except that twice as many pixel
locations were used in the YZ plane of the tomogram. Figure 3.7 illustrates a
second processing of the data in Figure 3.5 with a pixel grid of 14x14, for a total of
79


196 pixels. Figure 3.8 illustrates a second processing of the data in Figure 3.6 with
a pixel grid of 18x18, for a total of 324 pixels.
Slice at X= 0.000
Velocity n/Ps
B
9.81E-2 -
-2.SOE-2 6.52E-2
30E-3
20E-3
ioe-3
OOE-3
90E-3
80E-3
TOE-3
GOE-3
50E-3
40E-3
306-3
Y m
Figure 3.7 Styrofoam tomogram
using 3 transducer spacing; 14x14 pixel grid
80


Slice at X= O.OOD
Velocity n/vs
4.30E-3
4.20E-3
4.lOE-3
4.OOE-3
3.90E3
3.80E-3
3. TOE-3
3-60E-3
3.50E-3
3.40E-3
3.30E-3
Y m
Figure 3.8 Styrofoam Tomogram using 1.5 transducer spacing; 18x18 pixel grid
The tomograms in Figure 3.7 and Figure 3.8 both show improvements in
resolution compared to their corresponding tomograms in the previous section.
Both have less spreading around the low velocity area and both have a more
consistent background velocity. The tomograms appear to be more specific with
increased pixel number when compared using the same contour level. The
computation time for the denser pixel grid, increased from that of the original pixel
grid.
3.2.4 Tomogram of Wood Block
Data for the tomograms of the wood block was collected at two wall
locations and on two different days. Data for the location implied by the initial
81


straight shot scan was taken beginning at 6:30am on Wednesday, November 12,
1997 along the 7V vertical grid line. Data for the location based on known original
placement of the wood was taken beginning at 6:00am on Thursday, November 13,
1997 along the 8V vertical grid line. The temperature on both days was 25
degrees. Data was collected at a transducer spacing of 1.5 inches between
horizontal grid lines 13H and 8H for a total scan length of an fifteen inches. The
thickness of the wall at this location is 12 1/8.
Appendix C contains the data and 3DTOM input files for the wood block
tomograms.
3.2.4.1 Wood Block at Grid Line 7V
Two tomograms along the 7V grid line were produced. Figure 3.9 illustrates
the tomogram constructed based on a three inch transducer spacing. The three
inch transducer spacing consists of six transmitter and six receiver locations for a
total of thirty-six rays used for the analysis. Figure 3.10 illustrates the tomogram
constructed based on a 1.5 inch transducer spacing. The 1.5 inch transducer
spacing consists of eleven transmitter and eleven receiver locations for a total of
121 rays used for the analysis.
82


Slice at X=
0.000
Uelocity n/Ds
Y m
Figure 3.9 Wood tomogram at grid line 7V; 3 transducer spacing; 7x7 grid
Slice at X 0.000
Uelocitu n/Vs
Y m
Figure 3.10 Wood tomogram at grid line 7V; 1.5 transducer spacing; 9x9 grid
I
t
83


Both tomograms have fairly consistent velocities between 3900 and 4000
m/s indicating that the wood block inclusion is not present in this section of the wall.
No isolated area shows up in the tomogram that would indicate the presence of the
wood block. As a result, no further analysis of this section was completed and it is
assumed that this is an area of the wall with no known inclusions. Based on these
tomograms, the initial straight ray assessment of the wall was inaccurate in this
case. This could be due to the procedure used for the collection of data or an error
in recording. Similarly, the amount of Vaseline used during the collection of the
straight shot data may have effected the results.
Based on these results, a second scan along grid line 8V was completed to
try and ascertain the location of the wood block.
3.2.4.2Wood Block at Vertical Grid Line 8V
As in the previous section, two tomograms were again produced, this time
along the 8V grid line. Figure 3.11 illustrates the tomogram constructed based on a
three inch transducer spacing, and thirty-six rays. Figure 3.12 illustrates the
tomogram constructed based on a 1.5 inch transducer spacing, and 121 rays.
84


Slice at X-
£
N
Figure 3.11
a
N
Figure 3.12
D.DOO
Uelocit^ n/Ps
4.30E-3
4.206-3
4.lOE-3
4.ODE-3
3. TOE-3
3.SOE-3
3. TOE-3
3.SOE-3
3.50E-3
Y m
Wood tomogram at grid line 8V; 3 transducer spacing; 7x7 grid
Slice at X= D.DDD
Velocity nyus
4.30E-3
4.206-3
4.lOE-3
4.00E-3
3.SOE-3
3.80E-3
3.70E-3
3.60E-3
3.SOE-3
Y m
Wood tomogram at grid line 8V; 1.5 transducer spacing; 9x9 grid
85


Both tomograms have a relatively consistent background velocity. Figure
3.11 has a background velocity of 3700 to 3800 m/s while Figure 3.12 has a
background velocity of 3800 to 3900 m/s. Each map has higher velocity areas on
the left and right sides, that range from 3900 to 4200 m/s. Both maps also have
small areas in the lower comers where the velocity is of the range 3500 to 3600
m/s.
Neither of these high or low velocity areas can be presumed to be the wood
block. First, there are a total of four isolated velocity areas, and only one block.
Second, the areas are located at the perimeter of the scan. If the wood block were
in fact located at any of these isolated areas, the block would be visible from the
exterior of the wall, which it is not. No specific area stands out as the location
containing the wood block.
Reprocessing the above data with a finer grid may offer further information
regarding the composition of the wall at this location. Figure 3.13 illustrates the
tomogram based on a three inch transducer spacing, with a 14x14 pixel grid.
Figure 3.14 illustrates the tomogram based on a 1.5 inch transducer spacing and a
18x18 grid.
86


Slice at X= 0.000
Uelocitij Hyps
S
4.306-3
4.20E-3
4.106-3
4.00E-3
3.906-3
3.806-3
3.TOE-3
3.606-3
3.506-3
Y m
Figure 3.13 Wood tomogram at grid line 8V; 3 transducer spacing; 14x14 grid
Slice at X= 0.000
Uclocity nyps
Y m
Figure 3.14 Wood tomogram at grid line 8V; 1.5 transducer spacing; 18x18 grid
87