Tunnelling and Underground Space Technology 19 (2004) 175–191

A decision tool for the selection of imaging technologies to detect

underground infrastructure
Hyung Seok Jeong*, Dulcy M. Abraham
School of Civil Engineering, Purdue University, West Lafayette, IN 47907, USA

Abstract

Damage to underground infrastructure during construction results in undesirable consequences to contractors, project owners
and citizens. These consequences include construction delays, design changes, claims, property damages, service breakdowns,
disruption of neighboring business and even injuries and lost lives. Subsurface utility mapping is becoming an essential process
to reduce the probability of such damages before construction starts. However, a wide variety of geophysical imaging technologies
and different application conditions pose challenges in selecting appropriate imaging technologies for the successful identification
of underground utilities. This paper describes a decision tool that was developed to assist in selecting appropriate imaging
technologies based on site conditions. The proposed decision tool, named IMAGTECH, utilizes the deterministic parallel selection
technique (DPST) as a decision framework to evaluate the input values involved in the decision problem.
䊚 2003 Elsevier Ltd. All rights reserved.

Keywords: Subsurface utility engineering; Geophysical techniques; Underground infrastructure; Underground utilities; Decision tool; Determin-
istic parallel selection technique; IMAGTECH

1. Introduction tion of neighboring business and even injuries and lost

lives.
Population growth and industrial expansion since The costs of utility damages are very significant and
World War II have resulted in increased infrastructure on the rise. In 1993, there were more than 104 000 hits
spending, particularly in the United States (USA) (Lew or third party damage to gas pipelines with a total cost
et al., 2000). Utility demand in the USA is projected to exceeding $83 million (Doctor et al., 1995). A 1996
expand by three percent annually to 183 million feet survey in Kansas reported that the total cost of the
(56 million meters) of utilities in the year 2003, with a reported damages was $4 663 544, and that 1.2 million
valuation exceeding $7 billion (Sterling, 2000). The locates were requested from members of One-Call. In
urban underground has become a spider’s web of utility 1997, Memphis Light, Gas and Water paid damages of
lines, including phones, electricity, gas, cable TV, fiber $515 000 and collected damages of $793 000 for utility
optics, traffic signals, street lighting circuits, drainage damage (Stinson, 1998). Generally, the total cost of
and sanitary sewers and water mains. The deregulation damages is underreported because only the direct costs
of utility services has been adding to the problem of of the emergency response and of repairing the damage
utility congestion as multiple service providers seek to are included (Lorenc and Bernold, 1998). Heinrich
place their networks underground. (1996) revealed that the total costs associated with an
New construction in urban areas and a growing accident reported in the media to cost $15 000 were
number of rehabilitation and replacement projects under- actually closer to $313 000, which was almost 20 times
taken to maintain and improve the ageing infrastructure higher than the originally expected cost of the damage.
have often resulted in increased instances of damages to The American Institute of Constructors (AIC) reported
underground utilities, and undesirable consequences to that damage to utility lines is the third most important
contractors, project owners and citizens. These conse- crisis for contractors, the other two issues being on-the-
quences include construction delays, design changes, job accident requiring hospitalization and contractual
claims, property damages, service breakdowns, disrup- dispute with a client resulting in litigation (Reid, 1999).
Subsurface utility mapping is becoming an essential
*Corresponding author. process to reduce the adverse effect before construction

0886-7798/04/$ - see front matter 䊚 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tust.2003.09.001
176 H.S. Jeong, D.M. Abraham / Tunnelling and Underground Space Technology 19 (2004) 175–191

starts. Since records about utility positions are virtually from underground objects is recorded. Processing of the
nonexistent, or often incomplete and inaccurate with recorded data yields information about the distribution
errors as high as 15–30% (Stevens and Anspach, 1993), of the physical properties related to buried bodies.
the ability to physically determine on-site the location, Interpretation of the processed data can indicate the
nature and depth of underground utility services is horizontal position of an underground utility. Current
critical. One of the organized efforts to diminish the technologies applying this procedure are based on vari-
risk of utility hits and subsequent damage is the One- ous different geophysical theories such as electromag-
Call system, which is a state-regulated program that netic theory, elastic wave theory, electrical resistivity
requires utility owners to mark the location of known theory, energy transfer theory, magnetic theory, gravity
active facilities on the ground surface prior to construc- theory, etc.
tion (Lew, 2000). The involvement of the One-Call However, limited construction budgets and limited
system in the construction stage, however, limits its allowances for site instrumentation due to rights-of-way
benefits to mere avoidance of utility hits. Subsurface and restriction of noise pollution hinder the use of
utility engineering (SUE) which has emerged in the past certain imaging technologies for utility locating purpos-
two decades characterizes the quality of subsurface es. Selection of such imaging technologies may incur
utility information utilizing surface geophysical imaging additional costs that exceed the project budget (Anspach,
technologies, civil engineering, surveying and data man- 1995). Typically, the choice of imaging techniques is
agement skills during the design phase of a construction limited to those based on electromagnetic methods,
project. The employment of SUE in the design stage magnetic methods and acoustic emission methods. The
allows not only the prevention of utility damage but next section focuses on the key features of these
also minimizes the costs of utility relocates, design methods.
changes and utility related construction delays.
The success of both One-Call system and SUE system 2.1. Electromagnetic methods
is initialized by correct identification of underground
utility. However, a wide variety of geophysical imaging This category of utility imaging methods includes
technologies and different application conditions pose techniques utilizing pipe and cable locators, terrain
challenges in selecting appropriate imaging technologies. conductivity meters, e-line locators, metal detectors,
Fifty six percent of the damages in 1995 for gas electronic marker systems (EMS) and ground penetrat-
pipelines was caused when the One-Call system was ing radar (GPR). Key features of these technologies are
used and 25% of hits on located facilities were due to described next.
mis-locates (Sterling, 2000). Selection of appropriate
imaging methods requires specific considerations such 2.1.1. Pipe and cable locators
as knowledge of existing site environmental conditions, Pipe and cable locator systems involve transmission
utility size and composition, and cost (Anspach and coils, which releases different electromagnetic frequen-
Wilson, 1994). Furthermore, the advent of new materi- cies. The electromagnetic energy, which is created in
als, congested rights-of-way and new construction meth- this manner, generates the magnetic fields around the
ods such as horizontal directional drilling trigger a electrically conductive material buried underground,
challenge to the successful identification of horizontal which is ultimately captured by the receiver coil on the
location of underground utilities (ASCE, 2002). surface. The captured magnetic field is analyzed to
The objective of this paper is to discuss the framework produce a visual or audible indication of the horizontal
of a decision tool developed to assist in the initial location of the utility.
decision making process regarding the selection of an The fact that the applied frequency restricts the limit
appropriate imaging technology. Section 2 of the paper of horizontal distance of the electromagnetic waves and
presents a comprehensive review of different imaging the ability to identify a utility with a different level of
technologies and key criteria for the selection of the electrical conductivity signifies that a wide range of
appropriate imaging method. Section 5 describes a new available frequencies is essential to trace utilities. Typi-
methodology, namely the deterministic parallel selection cally, frequencies from 50 to 480 kHz can be successful
technique (DPST) utilized as the decision framework. (ASCE, 2002).
In Section 6, the application of the decision tool called Practically, frequencies less than 1 kHz are classified
IMAGTECH, a computer program based on this meth- as low frequencies and frequencies greater than 1 kHz
odology, is presented using two case studies. are classified as high frequencies. As the frequency of
the electromagnetic wave becomes higher, the distance
2. Subsurface utility imaging technologies the wave travels decreases mainly due to a reduction in
the impedance to ground, which causes the electromag-
In typical subsurface utility imaging applications, netic current to leak away. In addition, high frequencies
energy is input to the earth and the reflected energy tend to create an electromagnetic coupling effect largely
 H.S. Jeong, D.M. Abraham / Tunnelling and Underground Space Technology 19 (2004) 175–191 177

Fig. 1. False identification of utility due to electromagnetic coupling effect (Haddon, 2001).

due to their propensity to transfer electromagnetic fields system. An E-line locator system is a machine, which
to near-conductive materials. This can cause false iden- digs a hole to make an access point to the target line
tification of the utility as the peak signal is captured at and creates a mechanical fitting that allows an electro-
an inaccurate horizontal position, as shown in Fig. 1. line to be inserted through the pipe. Once the electro-
In general, pipe and cable locators work well for line is inserted to the pipe, the pipe and cable locator
metallic utilities, utilities that have tracing materials with conductive mode is applied. Connecting the trans-
installed above the utility and utilities that can accept a mitter to the electro-line in the access point, the receiver
metallic conductor or a transmitter (sonde) inserted into follows the signal reflected from the electro-line to
them. The major shortcoming of this method is its designate the pipe. This method has been only used for
inability to trace non-metallic utilities, which do not fall imaging plastic gas pipes due to a need to make a
in the aforementioned categories. Besides, this method mechanical fitting to the pipe (Continental, 1999). This
must be prudently used where utility congestion is method works very well where the tracing wire or
expected because of the electromagnetic coupling effect. metallic marking tape is deteriorated. Some of the
disadvantages of using this method include: requirement
2.1.2. Terrain conductivity meters for at least one point of exact location of the under-
Terrain conductivity equipment creates and measures ground utility, slow progress of operation and relatively
eddy currents due to differences in the average conduc- high cost of operation.
tivity from the ground surface to the effective penetra-
tion depth (typically 15 feet (5 m) or so) (ASCE, 2.1.4. Metal detectors
2002). Electromagnetic eddy currents emitted from the Metal detectors work by transmitting an alternating
transmitter reflect the eddy current back to the ground current (AC) magnetic field into the ground and then
with a slightly different property when it comes in analyzing a corresponding magnetic field to see if there
contact with an object which has conductivity that is have been any changes. Metallic objects reflect magnetic
different from the surrounding soil. For instance, a fields that are slightly different from the current reflected
buried metallic object has a lower conductivity than the from the surrounding soil, back to the surface. The
surrounding soil. Hence, the reflected current has a search-coil in the receiving unit senses the difference in
distinguishable value. The receiver attached to the end the magnetic field. Some form of output device (such
of the equipment analyzes the reflected currents to detect as a loudspeaker, gauge, etc.) will emit a noise alerting
underground utilities. the operator to the presence of the object. Since the
This method is useful for utility imaging in non- response decays exponentially with depth, metal detec-
utility congested areas, or in areas of high ambient tors are usually only applicable for shallow manhole
conductivity (Anspach, 1995). Isolated metallic utilities, lids, valve box covers, etc. (ASCE, 2002). In addition
underground storage tanks, wells, and vault covers are to metals, such detectors may react to magnetic non-
usually detectable via this method, and under some conductive minerals and beach salts due to their high
conditions, large non-metallic water pipes in dry soils conductivity. This can cause the metal detector to pro-
or large non-metallic empty and dry pipes in wet soils duce a false signal or ground noise. To eliminate this
may also be imaged (ASCE, 2002). Magnetic fields ground noise, the detector must be ground balanced to
produced along overhead power lines and above-ground compensate for the levels of these substances in the
metal objects, such as fences, vehicles, or buildings, ground being searched.
interfere with terrain conductivity readings.
2.1.5. Electronic marker systems (EMS)
2.1.3. E-line locators The EMS use electronic markers and electronic mark-
This method is principally dependent on pipe and er locators. The locator transmits a radio frequency
cable locator equipment but needs an E-line locator signal to the electronic marker that was buried along
178 H.S. Jeong, D.M. Abraham / Tunnelling and Underground Space Technology 19 (2004) 175–191

Table 1 computer to analyze the received data and depict the

Operating frequencies of electronic markers (Metrotech, 2001)
vertical profile of the imaged area.
Type of utility Frequency (kHz) In order to generate the GPR image profile, the
antenna is moved along the surface on the investigation
Power line 145.7
Telecommunication line 121.6
area. High frequency waves cannot penetrate as deep as
Water pipeline 169.8 low frequency waves but the former have the capability
Waste water pipe 101.4 to resolve smaller scale features. Hence, experience
Gas pipe 83.0 plays a key role in the selection of appropriate frequen-
cies. Commonly, the frequency of the radar ranges from
with the facility at the time of construction. The marker, 10 MHz to 1 GHz. In the GPR profile, other features
which consists of a passive resonant circuit, reflects the such as subsurface structures, rock formations and the
signal back to the locator, and the location is indicated water table can be imaged as well as subsurface utilities
with both a visual reading and an audible tone (3M, since they have different electrical properties compared
2002). The operating frequency of the electronic markers to the surrounding soils. Detected subsurface utilities
is adjusted according to the type of utility for the exact are typically imaged as a cone shape in the profile, as
detection of the utility as shown in Table 1. shown in Fig. 2.
These electronic markers are used for underground The performance capability of GPR is strongly
marking of special buried features, such as splices, dependent on the soil electrical conductivity at the site.
valves, etc., as well as non-metallic utilities. Markers Thus, highly conductive soils such as clays, saturated
can be detected even in the presence of metal conduc- sands, and tidal areas where salt is prevalent in soil, can
tors, fences, power lines, etc. because the electronic severely limit the maximum penetration depth of the
markers are specially designed to react to a certain level radar signal, reducing the penetration depth to less than
of frequencies. 3 feet (1 m).
One of the outstanding benefits of GPR is that
2.1.6. Ground penetrating radar (GPR) metallic and non-metallic utilities may be imaged since
In the ground penetrating radar (GPR) technique, a the radar signal reflection from the utilities depends on
transmitting antenna transmits electromagnetic waves contrasting dielectric properties of the soil and pipe. The
into the ground. Reflected waves at the boundary inter- major drawbacks include its highly limited use because
face represent the changes in the electrical properties of of restricted applicable soil conditions and the low
the subsurface material (Foillard et al., 1995). The penetration depth. Even in ideal soil conditions, the
reflected waves to the antenna are passed on to a GPR is not effective in imaging subsurface utilities at

Fig. 2. GPR profile using 500 MHz antenna (Mala, 2002).

 H.S. Jeong, D.M. Abraham / Tunnelling and Underground Space Technology 19 (2004) 175–191 179

Table 2
Application features of electromagnetic methods

Electromagnetic methods Application features

Pipe and cable locators
Conductive mode Only metallic objects can be detected Good for tracing utilities
(Low frequency) Direct hook-up to the utility line Crew size: 1–2 people
Frequency range: 220–640 Hz Effective depth-15 feet (4.6 m)
Conductive mode Only metallic objects can be detected High chance of coupling of electromagnetic
(High frequency) Frequency range: 8–480 kHz wave to adjacent metallic utilities
Direct hook-up to the utility line Crew size: 1–2 people
Good for tracing utilities Effective depth-15 feet (4.6m)
Not good for low conductive joints
Inductive mode Only metallic objects can be detected surface utility appurtenance is required
Frequency range: 8 kHz–480 kHz Crew size: 1–2 people
Indirect energy generation from the surface Effective depth-6 feet (1.8 m)
Passive mode No transmitter is required Crew size: 1–2 people
Frequency: 50 Hz Effective depth-4.5 feet (1.4 m)
Good for searching metallic or live electric lines
Sonde insertion Specially designed transmitter (sondeyprobe) Surface connection is necessary such as
Method Direct insertion of sonde to the underground Manhole
utility Crew size: 2 people
Good for non-metallic drain, sewer pipe or
ducts
Tracing wire Tracing wireymetallic marking tape must be Crew size 1–2 people
Metallic marking installed in the construction stage of the Effective depth-15 feet (4.6 m)
Tape method Utility
Pipe and cable locator with conductive mode
is generally utilized
Imaging technologies Application features
Terrain conductivity Typically metallic objects can be detected Useful in non-utility congested area
Can detect other features of different Crew size: 1 person
conductivity from the surrounding soil Effective depth -15 feet (4.6 m)
Good for searching utilities (isolated
metallic utilities, underground storage
tanks, wells and vault covers)
E-line locators Requires E-line locator as well as pipe Pipe and cable locator with conductive mode is generally
and cable locator utilized
Digging a holeycreating an access point to Currently used for plastic gas pipe
the utility is necessary Exact location of pipe is required
installing an electro-line through the access Crew size: 2 people
point
Metal detectors Metallic objects can be detected Good for searching utilities
Applicable for shallow manhole lids, valve Crew size: 1 person
box covers and so on Effective depth -2 feet (0.6m)
Electronic marker systems (EMS) Electronic markers must be installed in the Typical spacing of markers: 60 feet (6.1 m)
Construction stages Usually used for marking of special buried
Different frequency of electronic markers for features such as valves, splices, etc.
different type of utility Crew size: 1 person
Ground penetrating radar (GPR) Both metallic and non-metallic utilities may Geophysicist or highly trained engineer’s
be imaged analysis required
Frequency range: 1 MHz–1GHz Effective depth -6 feet (1.8 m)
Good for searchingytracing utilities Poor ability on highly conductive soils such
as clay, highly saturated soil, etc.
Crew size: 1–2 people

depths greater than 6 feet (2 m) (ASCE, 2002). Advanc- 2.2. Magnetic methods
es in processor speed and mathematical algorithms
promise to make this technique more user-friendly in Magnetic methods involve an instrument called a
the future. magnetometer, which is used to measure the intensity
The key application features of different electromag- of the earth’s magnetic field (Sharma, 1997). The
netic methods are summarized in Table 2. equipment detects deviation of the magnetic intensity
180 H.S. Jeong, D.M. Abraham / Tunnelling and Underground Space Technology 19 (2004) 175–191

Fig. 3. Principle of acoustic emission method.

caused by ferrous objects since ferrous metal radiates service or main line, applies sound waves (typically
its own magnetic field. Two techniques are basically from 132 to 210 Hz) into the pipeline. The sound waves
available in this category: total field measurements and travel along the length of the pipe and attenuate through
gradiometric measurements. Both of them utilize the the pipe wall into the surrounding soil. Those sound
same instrument. The proton precession magnetometer waves that reach the surface may be detected using
is a common type for total field measurement and flux- special sensors such as geophones or accelerometers, as
gate magnetometer is commonly used for gradiometric shown in Fig. 3. The location of the buried facility is
measurement. indirectly determined by monitoring the highest (peak)
In the proton precession magnetometer, the analysis vibration amplitude at the surface.
of the measured total magnetic field reveals the existence The detectable range is highly dependent on the
of ferrous magnetic objects. The field caused by mag- rigidity of the sound carrying material. As the rigidity
netic induction of ferrous magnetic objects is measured (inverse bulk of modulus) of the material increases,
from the total magnetic field in the analysis process. It detection capabilities in depth and horizontal distance
can be useful for a utility search over large areas in the from source sound also increases. This method can
absence of power lines, railroads, vehicles or other detect up to 8 feet (2.5 m) in depth in the case of gas
sources of magnetic interferences. The equipment should pipes and 6.5 feet (2.0 m) for cases involving water
be held in the vertical or near vertical position to reduce pipes. The horizontal range reaches up to 1000 feet (300
interference from solar magnetic storms and regional m) for plastic gas pipe and more than 500 feet (150 m)
magnetic changes. for water pipes (Radiodetection, 2001) in favorable
The different intensity of the magnetic field captured conditions such as a non-noisy environment, rigid soil
in a fluxgate magnetometer by two sensors separated in and shallow buried utilities.
a known distance is indicated by both a visual reading This method is typically useful to determine the
on the screen and an audible tone for an operator. The location of plastic water pipes and gas pipes. Prior
response is a change in the frequency of the signal knowledge about the surface appurtenance of the target
emitted by the piezoelectric speaker. Typically, signal utility limits this method only to tracing rather than
patterns for a vertically oriented target, such as iron pipe ‘blind’ location. In addition, the susceptibility to noise
markers, exhibit peaks over the top and for a horizontally interference, such as aircraft, automobiles, trains, etc.,
oriented target, such as cast iron pipe, the signal exhibits constitutes one of the major disadvantages in the use of
peaks at their ends ( joints). The closer the distance this method.
from the target to the equipment, the greater the tenden-
cy of getting a peak signal around the edge of the 3. Decision criteria
object. Magnetized non-metallic fiber optic cables and
cast iron pipes which produce the strongest signals at A comprehensive analysis of the characteristics of
their joints can be located by this method (Schonstedt, each imaging technology and available information that
2000) and it is typically effective for detecting valve site engineers can obtain from as-built drawings and site
boxes, steel drums, iron markers and manhole lids. visits led to identification of ten significant criteria that
are fundamental for the selection of appropriate imaging
2.3. Acoustic emission method technologies. These criteria include:
An acoustic emission method utilizes an acoustic 1. Type of utility: Certain imaging technologies are not
transducer that, when connected to an opening on a used for the detection of specific types of utilities.
 H.S. Jeong, D.M. Abraham / Tunnelling and Underground Space Technology 19 (2004) 175–191 181

For instance, the use of acoustic emission method is thumb, a utility whose diameter in inches (25 mm)
typically applicable only for imaging water and gas exceeds the depth in feet (305 mm) can be imaged
pipes. in dry soil conditions.
2. Material of utility: This criterion specifies the mate- A detailed discussion of the evaluation of imaging
rial property of the utility which can be characterized technologies can be found in Jeong (2001). Each crite-
into three different categories: i) ferrous metal, ii) rion includes several entries that specify the site condi-
non-ferrous metal or iii) non-metallic material. tions and features of the target utility as listed in Table
Imaging technologies can be classified into the 3. The degree of impact of each entry of criterion on
aforementioned categories based on their capability each imaging technology in terms of appropriateness of
of imaging different types of materials. the technology is evaluated by assigning one of three
3. Joint type of metallic utility: This criterion is used linguistic attributes. These attributes are:
to evaluate the electrical continuity of the metallic
utility. Joints can be categorized into three different a Applicable: a ‘superior’ condition for the use of the
sets based on electrical continuity; continuous, semi- imaging technology
continuous and discontinuous. Electromagnetic fre- b Inapplicable: an ‘infeasible’ condition for the use of
quencies cannot penetrate discontinuous joints and the imaging technology
only high frequencies can penetrate semi-continuous c Neutral: the entry is a theoretically possible condition
joints. In addition, the shield effect hinders the use for the use of the imaging technology but is rarely
of the sonde insertion method in the electrically used in practice due to better performance of the
continuous pipe. The shield effect is a phenomenon other technologies or the entry has no impact on the
where electromagnetic waves cannot escape the pipe operation of the technology.
due to the reflection of the waves against inside Technical judgments of each imaging technology on
wall of the pipe. each entry of criterion regarding its applicability result
4. Special materials for detection: The difficulty of in typically ‘applicable’ or ‘inapplicable’ condition. For
locating non-metallic utilities has prompted the instance, pipe and cable locators are applicable to
installation of special materials onyabove the utility metallic material but are not applicable to non-metallic
at the time of construction. Such materials include objects. There is a need to have a ‘neutral’ value for
tracing wireymetallic marking tape and electronic the evaluation of imaging technologies due to two
markers, which require specific imaging technology reasons. First, not all criteria are significant for all
for their detection. technologies. For instance, the condition of the ground
5. Access point to utility: This criterion identifies the surface does not have significant influence on the
existence of surface appurtenances connected to the operation of electronic marker systems or GPR. Second,
underground utility line in the vicinity. For instance, assume that Technology A can be used for a given site
the presence of a utility in the vicinity is one of the condition, but Technology B shows significantly better
essential conditions for several imaging technologies performance in the same environment, then the decision
such as pipe and cable locators (used in the inductive tool must place a higher ‘value’ on the more appropriate
mode, sonde insertion method, tracing wireymetallic technology. In such cases, the value ‘neutral’ is assigned
marking tape method), e-line locator and acoustic to Technology A. A matrix summarizing the applicability
emission method. of each criterion to each imaging technology is illus-
6. Ground surface condition: In general, reinforced trated in Table 4.
concrete pavement may cause disturbance to the
emitted electromagnetic waves when electromagnet- 4. Features of the decision tool
ic methods are used.
7. Internal state of utility: The internal condition of Four major features were identified to be crucial for
utility limits the use of certain methods. For instance, the process of the decision tool: (1) knowledge based
the acoustic emission method is applicable when the decision, (2) crisp and tangible input values, (3) paral-
pipe is full since the method is based on the pressure lelism among criteria, and (4) capability for proposing
to transport the sound wave. alternatives. The decision tool requires construction of
8. Soil type: High soil conductivity typically found in a knowledge base since the decision is made by technical
clays or highly saturated soil creates unfavorable judgments on imaging technologies utilizing site condi-
operating conditions for the use of the terrain con- tions as input values. The attributes of input values are
ductivity method and GPR. crisp and tangible since they correspond to distinctive
9. Depth of utility: This criterion addresses the field technical terms. In addition, the 10 criteria explained in
limits of penetration depth of each imaging the decision criteria section (Section 3) carry equal
technology. weights in the selection process, i.e. they are not
10. DiameteryDepth ratio: This criterion is designed to hierarchically ranked. The last feature required for the
evaluate the potential use of GPR. As a rule of decision tool stems from the need to have alternatives
182 H.S. Jeong, D.M. Abraham / Tunnelling and Underground Space Technology 19 (2004) 175–191

Table 3
Criteria in selecting appropriate imaging technologies

Criteria Entries of criterion

1) Type of utility Water (W), sewer (S), steam (ST), gas (G), oil and chemical (O),
Electric (E), telecommunication (T)
2) Material of utility Ferrous Steel, cast iron, ductile iron
metallic(F)
Metallic (M) Copper and metallic conduit
Non-metallic Fiberglass reinforced plastic (FRP), vitrified
(N) clay, Concrete, asbestos-cement, brick, cement,
Plastics, fiber optic cable, etc.
3) Joint type of metallic Electrically continuous (EC), semi-continuous (ESC)
pipe discontinuous (ED)
4) Special materials for Tracing wire or metallic marking tape installed (T), electronic
detection markers installed (E)
5) Access point to utility Presence of utility (P), exactly known location of utility (E),
probable location of utility (PR), none of the above (NK)
6) Ground surface Paved (P) reinforced concrete paved (R), and natural surface (N)
condition
7) Internal state of Full with flowing material (F), partially full with flowing material
pipeyconduit (P), conduits full of cables (C), full and empty conduit (FE), empty
pipe or conduit (E)
8) Soil type Highly conductive soil (H), clay dominated soil (C), silt
dominated soil (S), sand dominated soil (SA), granular and
compacted soil (GC)
9) Depth of utility -2 feet (0.6 m), -6 feet (1.8 m), -8 feet (2.4 m), -10 feet (3.0 m), -15 feet
(4.6 m), -50 feet (15.2 m), )50 feet (15.2m)
10) Diameter (inch)ydepth )1 (8.3), -1 (8.3)
(feet) of utility (cmym)

in practice. When the first attempted technology fails, a fuzzy logic as an applicable decision framework in this
second or the third alternative may be used to locate the case.
underground utility. An artificial neutral network (ANN) is a computing
Five frequently used decision frameworks were exam- system, made up of a number of highly interconnected
ined to determine whether they would be appropriate processing elements, which processes information by its
candidates for the development of the decision tool. The dynamic state response to external inputs. ANNs are
results of this analysis are illustrated in Table 5. good at solving problems that are too complex for
A decision tree, which is mainly governed by ‘if- conventional techniques such as problems that do not
then’ rules, requires subordination of the subsequent have algorithmic solutions or for which an algorithmic
decision node to the prior decision node. One of the solution is too complex to be found. The distinctive
biggest shortcomings of the decision tree is that many relationship between input values and imaging technol-
decision nodes, and many alternatives at each decision ogies in the required decision tool do not necessitate the
node, causes the decision tree to get cumbersome (Clem- use of ANN. Genetic algorithm (GA) is a branch of
ens, 1996) and difficult to create and use. The analytical evolutionary computing inspired by Darwin’s evolution
hierarchy process (AHP), originally designed by Saaty, theory. The primary purpose of GA is optimization
involves the overall hierarchy of the decision and per- (Tsoukalas and Uhrig, 1997). Since the problem solving
forming pair-wise comparisons between each possible process of GA is not typically based on retrieving and
pair of criteria as a matrix (Saaty, 1980). The AHP is analyzing data from a database but generating a new set
typically applicable to subjective decisions. The use of of solutions from the previous set of solutions, this
fuzzy logic as a decision tool provides a simple way to methodology was also ruled out as a candidate for the
draw definite conclusions from vague, ambiguous or decision framework.
imprecise information. The major process involves eval-
5. Decision framework
uation of fuzzified inputs against a set of production
rules. However, the need to have descriptive input values A technique named the deterministic parallel selection
for the decision tool eliminates the consideration of technique (DPST) was developed as a decision frame-
Table 4
Matrix of the applicability of each entry of criterion to imaging technologies

Imaging technologies Applicability Criteria

1 2 3 4 5 6 7 8 9 10
Type of Material Joint type of Special material Access point Surface Inner state Soil Depth of Diameterydepth
utility of utility metallic utility for detection to utility condition of pipe type utility ratio

H.S. Jeong, D.M. Abraham / Tunnelling and Underground Space Technology 19 (2004) 175–191
Pipe and cable locators
Conductive mode AP F,M ESC P -15 feet
(High frequency) NE All neutral EC, NK All neutral All neutral All neutral All neutral All neutral
IN N ED E,PR,N )15 feet
Conduction mode AP F,M EC P -15 feet
(Low frequency) NE All neutral NK All neutral All neutral All neutral All neutral All neutral
IN N ESC, ED E,PR,N )15 feet
Inductive mode AP F,M EC, ESC E P,N -6 feet
NE All neutral NK All neutral P, PR R All neutral All neutral All neutral
IN N ED N )6 feet
Passive method AP E F,M PR,N -15 feet
NE Neutral All neutral All neutral P,E All neutral All neutral All neutral All neutral
IN N )15 feet
Tracing wire or AP T P,E
metallic marking tape NE All neutral All neutral All neutral N, NK All neutral All neutral All neutral All neutral All neutral
tape method IN E PR,N
Sonde insertion method AP S,E,T N ESC, ED P P,FE,E -15 feet
NE All others F,M NK All neutral All neutral C,NK All neutral -50 feet All neutral
IN EC E,PR,N F )50 feet
E-line locator method AP G plastic pipes P,E -15 feet
NE All neutral All neutral All neutral All neutral All neutral All neutral
IN All others All others PR,N )15 feet
Terrain conductivity meter AP F,M PR,N P,N S,SA,G -15 feet
NE All neutral N All neutral All neutral All neutral H,C,NK All neutral
IN P,E R )15 feet
Electronic marker systems AP E P,E,PR
(EMS) NE All neutral All neutral All neutral N, NK All neutral All neutral All neutral All neutral All neutral
IN T N
Metal detectors AP F,M PR,N P,N -2 feet
NE All neutral All neutral All neutral All neutral All neutral All neutral
IN N P,E R )2 feet
Ground penetrating radar AP S,SA,G -6 feet )1
(GPR) NE All neutral All neutral All neutral All neutral All neutral All neutral All neutral H,C,NK
IN )6 feet -1
Magnetic methods AP F ESC, ED PR,N P,N -10 feet
NE All neutral NK All neutral P,E All neutral All neutral All neutral
IN M,N EC R )10 feet
Acoustic emission method AP W, G N P F -8 feet (g), -6.5 feet(w)
NE ST F,M All neutral All neutral All neutral NK All neutral All neutral
IN all others E,PR,N P,C,FE,E )8 feet, )6.5 feet
All initials follow the initials in Table 3. AP: Applicable, NE: Neutral, IN: Inapplicable, NK: Not known.

183
184 H.S. Jeong, D.M. Abraham / Tunnelling and Underground Space Technology 19 (2004) 175–191

Table 5
Analysis of five decision frameworks

Decision frameworks Knowledge Crisp and tangible Parallelism Capability for

base input values among criteria proposing
alternatives
Decision tree X
Analytical hierarchical X X
process
Fuzzy logic X X
Artificial neural X
network
Genetic algorithm X
X: not applicable for the feature.

work. This framework satisfies the four major features utility; and C4: access point to utility. Suppose each
required of decision tool for imaging buried utilities. criterion consists of only three or four entries, as shown
The DPST works as an inference engine. When input in Fig. 5.
values are provided, the DPST evaluates the input data First, the entries of criteria that are applicable to each
to determine the applicability to each imaging technol- imaging technology must be identified as shown in
ogy based on the established knowledge base. The next Table 5. According to Table 5, A3 (the acoustic emission
process involves the conversion of values of linguistic method) operates effectively in a condition when the
applicability to numeric values. A computational process type of a utility is water pipe or gas pipe, when the
using the obtained numeric values discards inappropriate material of utility is polyethylene (PE), and a physical
methods, and ranks the selected appropriate imaging access point to the utility is provided in the vicinity.
technologies. A schematic diagram of the DPST is Hence, these conditions are deemed ‘appropriate’ for
illustrated in Fig. 4. the acoustic emission method. However, the acoustic
A simplified example is employed to discuss the emission method is not applied to the sewer pipe or if
different steps in this technique. Suppose three different there is no utility present in the vicinity. ‘Neutral’
imaging technologies (A1, A2, A3) are available, and signifies that the entry in the criterion exerts no influence
there are four criteria (C1, C2, C3 and C4) affecting on the selection of the relevant imaging technology.
the selection of appropriate imaging technologies. The Once the identification of applicability (knowledge
three imaging technologies are A1: Pipe and cable base) is completed, the collected information about the
locator—conductive mode with high frequency; A2: site conditions is provided to the DPST as input values.
Sonde insertion method; and A3: Acoustic emission Suppose the area where utility locating is to be
method. The four criteria are C1: type of utility; C2: performed has the following characteristics. The utility
material of utility; C3: type of joint in the metallic is a water pipe made of cast iron which has a bonding

Fig. 4. Schematic diagram of the DPST.

 H.S. Jeong, D.M. Abraham / Tunnelling and Underground Space Technology 19 (2004) 175–191 185

Fig. 5. Simplified example for application of DPST.

Table 6
Identification of applicability of entries to each imaging technology

Criteria Applicability C1. Type C2. C3. Joint of C4. Access

Imaging technologies of utility Material of Metallic point to
utility pipe utility
A1. Pipe and cable Applicable ST, CI EC, ESC P
Locator–conductive Neutral W, S, G, NK NK
mode with high NK
frequency Inapplicable PE E, N
A2. Sonde insertion Applicable S PE ESC P
method Neutral W, G, NK ST, CI, NK
NK
Inapplicable EC E,N
A3. Acoustic emission Applicable W, G PE P
method ST, CI, EC, ESC,
Neutral NK NK NK

Inapplicable S E, N
W: water pipe, S: Sewer pipe, G: Gas pipe, ST: Steel, CI: Cast iron, PE: Polyethylene, NK: Not Known, EC: Electrically continuous, ESC:
Electrically semi-continuous, P: Probable Presence of utility in the vicinity, E: Exact Location of utility is known.

Fig. 6. Numeric conversion of linguistic applicability.

186 H.S. Jeong, D.M. Abraham / Tunnelling and Underground Space Technology 19 (2004) 175–191

Fig. 7. Step 1 and Step 2 matrices.

jumper installed (electrically continuous joints), and imaging technology. The closer this outcome is to 1, the
there is a fire hydrant in the vicinity (presence of greater the appropriateness of the imaging technology
utility). From Table 6 (knowledge base), the linguistic for the specified site conditions.
applicability of input values to each imaging technology In this study, reliability index (RI) indicates the level
can be established in a matrix as shown in Fig. 6. These at which input information supports the use of the
linguistic values are converted to numeric values based imaging technology. This index is a percentage value of
on the following rules: Applicable: 10; Neutral: 1; each imaging technology in the final matrix.
Inapplicable: 0.
The converted numeric matrix is also shown in Fig. RI (%)
6. The numeric matrix is used to calculate the appropri- sthe value of each technology in the final matrix
ateness and the reliability of imaging technologies =100
through the following steps.
Step 1: Multiplication of values in columns. The The most appropriate method and the other alterna-
numeric values in each column in the numeric matrix tives can be determined by ranking the reliability of
are multiplied with each other to create a step 1 matrix each imaging technology. In this example with four
of one row (Fig. 7). The resulting value of multiplication criteria and three technologies, the first alternative is A1
is always a value of 0 or 10x since the multiplication is (pipe and cable locator–conductive mode with high
a combination of 0, 1 and 10. frequency) with 100% RI which implies that all selected
Step 2: Step 2 matrix. The logarithm of the step 1 entries in criteria are appropriate conditions for the use
matrix with respect to a base 10 produces a step 2 of this method. The second alternative is A3 (Acoustic
matrix in which matrix elements take the exponent emission method) with 67% RI.
numbers in the step 1 matrix (Fig. 7). The number in
the resultant matrix reveals the number of applicable 6. IMAGTECH
conditions for each imaging technology. Infinity implies
that at least one inapplicable condition is selected. 6.1. Introduction to IMAGTECH
Step 3: Maximum number of applicable conditions
for the imaging technology. Table 5, which works as a A prototype computer decision tool, named IMAG-
knowledge base, classifies imaging technologies based TECH, a dialog based computer application in a Micro-
on the applicability of the entry of criterion. The total soft Windows environment, was developed based on the
number of criteria, which include ‘applicable’ conditions established criteria and the DPST procedure. The use of
for the imaging technology indicates the number of the DPST in the case of 13 candidate technologies and
criteria for its optimal operating circumstances. As 10 criteria necessitate a large matrix (10=13) for the
shown in Fig. 8, Table 6 can be used to create a new decision process. IMAGTECH can eliminate the possi-
matrix (step 3 matrix) in which each element corre- bility of human errors during the calculation of the large
sponds to the number of criteria for optimal operating matrix and hence facilitate the decision making process.
conditions. Fig. 9 displays the system architecture of IMAGTECH.
Step 4: Selection of appropriate imaging technologies. The first stage in this process seeks information about
The numbers in the step 2 matrix represent the number the available type of equipment by the utility locating
of applicable conditions selected for the imaging tech- team. This process enables the system to limit candidate-
nology. The outcome of dividing the step 2 matrix by imaging technologies to only selected equipment based
the step 3 matrix determines the appropriateness of the methods. Site conditions and utility features are also
 H.S. Jeong, D.M. Abraham / Tunnelling and Underground Space Technology 19 (2004) 175–191 187

Fig. 8. Step 4: Selection of appropriate imaging technologies.

specified. The ‘not known’ option can be selected for Case I: Locating a water main for Duck Creek Bridge
each criterion if specific data regarding that criterion is Project, Las Vegas, NV. The Clark County Public Works,
not available to the user. In such cases, the reliability Nevada, needed to build a bridge for providing residen-
index (RI) for each case is lowered since a neutral value tial access across Duck Creek. It was designed to be 52
is assigned to the criterion. feet (15.8 m) long and 31 feet (9.4 m) wide and
Some criteria are correlated with other criteria. For included two travel lanes. The reinforced concrete bridge
instance, the type of utility governs the range of mate- was cast in place. In the design stage of the project, the
rials used for the utility. In such cases, the selection of design team needed accurate information about the
the former criterion activates or inactivates each entry location of the water main supposedly buried under the
in the latter criterion. Based on the input values and the proposed bridge construction site. Information about
knowledge base, IMAGTECH suggests the most appro- utility location would be used to adjust and finalize the
priate imaging technology and two alternatives. The two-dimensional and three-dimensional position of the
reliability index is also presented for each proposed bridge foundation.
imaging technology. Sample input and final screens are Case II: Locating underground utilities along SR27
shown in Figs. 10 and 11. in Richmond, IN. The Indiana Department of Transpor-
tation (INDOT) hired the Tampa Bay Engineering
6.2. System validation (TBE) company to develop horizontal and vertical
location database for underground utilities along SR27
Two field studies were employed to validate the utility in Richmond, IN. In 2004, INDOT intends to add two
of IMAGTECH and to assess its performance. Table 7 travel lanes to I-70 from the interchange at I-70 and I-
lists the key features of these cases. The information for 27 to approximately 2 miles (3.2 km) to the east. The
the input data is obtained from the as-built drawings new construction site is located in urban areas and many
and visual site investigations. It was assumed that in utilities such as water, telecommunication, gas and
both cases, all the equipment is available. sewers are present in this area. Moreover, storm sewers

Fig. 9. System architecture of IMAGTECH.

188 H.S. Jeong, D.M. Abraham / Tunnelling and Underground Space Technology 19 (2004) 175–191

Fig. 10. Sample input screens for IMAGTECH.

are planned in this phase of construction. It is vital to In case I, IMAGTECH recommends the use of the
incorporate the exact location of underground utilities acoustic emission method (RIs100%) and GPR (RIs
in the final design adjustment for future road construc- 67%), as shown in Table 8. In the Duck Creek Bridge
tion projects. Project, the acoustic emission method was successfully

Fig. 11. Sample final screen for IMAGTECH.

 H.S. Jeong, D.M. Abraham / Tunnelling and Underground Space Technology 19 (2004) 175–191 189

Table 7
Cases for validating IMAGTECH system

Site Properties Case I Case II

Type of utility Water pipe Gas pipe
Material of utility Transite (non-metallic) Steel
Joint type of metallic utility Not known Electrically continuous
Special detection material installed Not known Not known
Access point to utility Fire hydrant Exactly known horizontal location
Ground surface condition Asphalt paved Asphalt paved
Internal state of utility Filled with flowing material Filled with flowing material
Soil type Not known Granular and compacted soil
Depth of utility (approximation) 6 feet (2.4 m) 1.2 feet (0.4 m)
Diameterydepth ratio 1.3 5.0

used. This method was deemed most appropriate due to and accurate by checking the results obtained using
the type and material of the utility, access point to utility IMAGTECH.
and internal state of utility. The site properties at the
Duck Creek Bridge Project constitute optimal environ- 7. Summary
ments for the acoustic method (as it is mainly targeted
for water in service and there was a fire hydrant on the Damage to underground infrastructures can, in
surface in the vicinity for a sound elastic wave genera- extreme cases, result in injury and death, as well as
tion). The lower RI value assigned to GPR in Case I severe property damage and loss of vital services and
can be attributed to unknown soil conditions. products. Obtaining accurate information on under-
In case II, pipe and cable locator using the inductive ground infrastructure is becoming more critical during
mode and GPR were proposed (100% RI), together with the planning and design phases of construction projects.
the magnetic method (75% RI). The method that was However, detecting the presence of underground infra-
utilized on the project was pipe and cable locator using structure is challenging due to the lack of complete as-
the inductive mode, because the pipe and cable locators built drawings, the availability of various imaging
were more readily available. technologies operating under different conditions, and
The two case studies showed that the IMAGTECH the high level of skill required in interpreting the
successfully predicts the most appropriate imaging tech- imagesydata collected by the technologies.
nology, which was chosen and used by the engineers on This paper discusses the analysis of an evaluation of
site. Choosing an appropriate imaging method is com- currently available imaging technologies and identifies
plex yet, nonetheless, crucial for the successful identi- 10 critical criteria for the selection of the most appro-
fication of buried utilities. Currently, decision-making priate imaging technology based on site conditions. A
on the selection of the imaging technology is highly decision tool named IMAGTECH was developed based
dependent on experience of the site engineer. Lack of on the deterministic parallel selection technique (DPST)
appropriate experience in the use of imaging technolo- framework. It can be used in practice for training or for
gies, which is quite common in this emerging industry, providing information about different imaging
could result in false identification of the underground technologies.
utilities and reduced productivity of the imaging and Imaging technologies are not magic wands. The com-
locating operations. The use of IMAGTECH can alle- plexity existing in the selection of the most appropriate
viate these problems by providing assistance in the imaging technology stems from the shortcomings of
decision making process. In addition, the site engineer each imaging technology. There is a strong need in
can ensure that his or her decision is comprehensive industry for a new and robust imaging technology which

Table 8
Comparison of utilized technologies with recommended methods by IMAGTECH

Imaging technology Case I Case II

Used technology Acoustic emission method Pipe and cable locator–
Induction mode
Alternative 1 Acoustic emission method Pipe and cable locator–
(100%) Induction mode (100%)
Alternative 2 GPR (67%) GPR (100%)
Alternative 3 – Magnetic method (75%)
190 H.S. Jeong, D.M. Abraham / Tunnelling and Underground Space Technology 19 (2004) 175–191

can overcome these drawbacks. Sterling (2000) studied Proceedings of the American Power Conference. Publ by Illinois
innovative technologies for locating utilities which can Inst of Technology, Chicago, IL, USA. vol. 56, pt 1, pp. 137–141.
ASCE, 2002. Standard Guidelines for the Collection and Depiction
be transferable to industry. Most of them were based on of Existing Subsurface Utility Data. ASCE Codes and Standards
the GPR technique but did not overcome the innate Activity Committee (CSAC), New York, NY, USA.
problems of GPR, such as low operating capability in Clemens, R.T., 1996. Making Hard Decisions: an Introduction to
highly conductive soils. Efforts at the Construction Decision Analysis. Duxbury Press, Elmont, California.
Automation Research Institute (CARL) in North Caro- Continental Industries Inc. 1999. E-line locator. Company Technical
lina State University have focused on integrating both Brochure, Tulsa, OK. USA.
GPR and electromagnetic equipment onto an excavator Doctor R.H., Dunker N.A., Santee N.M., 1995. Third-party damage
prevention systems. Final Report for Gas Research Institute. Con-
for safely locating two or three 3 m ahead of the
tract No. 5094-810-2870, Oct. 1995, by Nicor Technologies,
excavation. Such systems would be helpful to detect all Naperville, IL, USA.
types of utility while performing excavations (i.e. during Foillard R., George B., Schwarze C., 1995. New applications of
the construction phase of the project), but cannot be ground penetrating radar for construction and rehabilitation in the
used in the planning stage of the project. pipe sector. Proceedings of the International No-Dig ’95 Confer-
Research is underway at the Trenchless Technology ence, Dresden, Germany, pp. 147–159.
Center (TTC), Louisiana Tech University to deal with Haddon, M.K., May–June 2001. Better Training Needed for Locate
Technicians. Underground Focus, pp. 24–25.
uncertainties related to the location of underground
Heinrich J., 1996. Assessment of the cost of underground utility
utilities by utilizing approximate reasoning techniques damages. North Carolina State University Tech. Report (August
and a simple sensor fusion method (Lanka et al., 2001). 1996).
By combining data from site visits, as-built drawings Jeong H.S., 2001. A decision tool for subsurface utility engineering
and GPR, the framework developed at TTC provides a (SUE). independent research study, School of Civil Engineering,
new avenue for handling uncertainties associated with Purdue University, December 2001.
individual information sources available in underground Lanka M., Butler A., Sterling R.L., 2001. Use of approximate
reasoning techniques for locating underground utilities. Trenchless
utility location and for integrating these information Technology Research, Tunnelling and Underground Space Tech-
sources. nology, 16, (S1), pp. 13–32.
Lew J.J., 2000. Cost savings on highway projects utilizing subsurface
Acknowledgments utility engineering (SUE). Research report–FHWA–IF-00-014,
Purdue UniversityyFederal Highway Administration. Washington
This study was funded by a grant (SPR-2451) from DC, USA.
the Indiana Department of Transportation (INDOT) Lew J.J., Lew J.C., Harter J.E., 2000. Procedures for the incorporation
of subsurface utility engineering (SUE) in the design process.
Joint Transportation Research Program (JTRP). Its sup- Proceedings of the CIB W89 International Conference on Building
port is gratefully acknowledged. The authors thank Paul Education and Research, Georgia Tech and Clemson University,
Scott (Federal Highway Administration (FHWA)), Jim May 2000, Atlanta, GA, USA, pp. 165–172.
Anspach (SO-Deep Co.), Robert Clemens, Sandra Jones Lorenc S.J., Bernold L.E., 1998. Smart attachment for utility damage
(TBE Group) and the members of the Study Advisory prevention. Proceedings of the ASCE Specialty Conference on
Committee of this project–David Ward (INDOT), James Robotics for Challenging Environments. ASCE, Reston, VA, USA,
pp. 140–146.
Ude (INDOT), James Harrell (INDOT), Paul Berebitsky
Mala GeoScience, 2002. RAMAC GPR. Company technical brochure,
(Indiana Constructors), Jeffrey J. Lew (Purdue Univer- Sweden.
sity), Ed Ratulowski (FHWA) and Matt Thomas Metrotech Co., 2001. Marker locating system. Company Technical
(INDOT); who continue to provide insightful comments Brochure, Sunyvale, CA, USA.
to the research team. The contents of this paper reflect Radiodetection Corp., 2001. Acoustic pipe tracer. Radiodetection
the views of the authors, who are responsible for the Ltd., Technical report, Bristol, UK.
facts and the accuracy of the data presented herein. The Reid, J., July–August 1999. Are Contractors, Engineers and Archi-
contents do not necessarily reflect the official views of tects Ready for a Crisis?, Vol. 28. American Institute of Construc-
tors (AIC), pp. 1–7.
policies of INDOT or FHWA at the time of publication
Saaty, T.L., 1980. The Analytical Hierarchy Process. McGraw Hill,
and do not constitute a standard, specification, or New York, NY, USA.
regulation. Schonstedt Instrument Co., 2000. Instruction manual model MAC-
51Bx (magnetic and dual-frequency pipe and cable locator).
References Schonstedt Instrument Co., Technical Report, 4 Edmond Road,
Kearneysville, WV, USA.
Anspach J.H., 1995. Subsurface utility engineering: upgrading the Sharma, P.V., 1997. Environmental and Engineering Geophysics.
quality of utility information. Proceedings of the International Cambridge University Press, Cambridge, UK.
Conference on Advances in Underground Pipeline Engineering. Sterling R.L., 2000. Utility locating technologies: a summary of
ASCE, New York, NY, USA. pp. 813–824. responses to a statement of need distribution by the federal
Anspach J.H., Wilson S.E., 1994. Case study of an underground 138 laboratory consortium for technology transfer. Federal Laboratory
KV transmission line design utilizing subsurface utility engineering. Consortium Special Reports Series No.9 ISSN 1075–9492C.
 H.S. Jeong, D.M. Abraham / Tunnelling and Underground Space Technology 19 (2004) 175–191 191

Stevens R.E., Anspach J.H., 1993. New technology overcomes the sored by Underground Focus Magazine, Atlanta, GA, USA, Dec
problems of underground system interferences on power projects. 2–4.
Proceedings of the American Power Conference. Publ by Illinois Tsoukalas, L.H., Uhrig, R.E., 1997. Fuzzy and Neural Approaches in
Inst. of Technology, Chicago, IL, USA, vol. 55, pt 1, pp. 323–326. Engineering. John Wiley and Sons, Inc, New York, NY, USA.
Stinson W. 1998. Preventing damage to unlocatable infrastructure, M Co., 2002. 3M ScotchMark locating and marking. Company
session handout, session A5, damage prevention convention, Spon- Technical Brochure. St. Paul, MN, USA.