Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

Automation in Construction 20 (2011) 982–997

Contents lists available at ScienceDirect

Automation in Construction
j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / a u t c o n

Optimal haulage routing of off-road dump trucks in construction and mining sites
using Google Earth and a modified least-cost path algorithm
Yosoon Choi a,⁎, Antonio Nieto b
a
Department of Energy Resources Engineering, College of Environmental and Marine Science and Technology, Pukyong National University, Busan 608-737, South Korea
b
Department of Energy and Mineral Engineering, College of Earth and Mineral Sciences, Pennsylvania State University, PA 16802, USA

a r t i c l e i n f o a b s t r a c t

Article history: This study presents new software, called Google Earth-based Optimal HAulage RouTing System (GEOHARTS),
Accepted 30 March 2011 to improve the functionality of Google Earth for optimal haulage routing of off-road dump trucks in
Available online 30 April 2011 construction and mining sites. A modified least-cost path algorithm, which is applicable to working areas with
both paved and unpaved temporary roads and can consider the effects of terrain relief and curves along a
Keywords:
route on the route planning, was proposed and utilized for the software development. GEOHARTS can
Google Earth
Off-road dump trucks
determine optimal haulage routes between loaders and dumps that ensure the least travel time or fuel
Route planning consumption of off-road dump trucks and can visualize the results using an embedded 3D render window of
Haulage operation Google Earth. The application to the Pasir open-pit coal mine in Indonesia demonstrates that GEOHARTS could
GIS provide a rational solution to support the truck haulage operations.
© 2011 Elsevier B.V. All rights reserved.

1. Introduction In response to the increased use of Google Earth and the advent of
KML, several applications have recently been developed for construc-
Google Earth combines satellite imagery, aerial photography, the- tion and mining projects. Sun et al. [15] uses Google Earth to provide
matic maps and various data sets to make a three-dimensional (3D) the Assisted Driving System (ADS) for haul trucks operating in open-
interactive image of the world [1]. It permits users to view seamless, pit mines to reduce accidents related to low visibility conditions.
true-color satellite imagery at every location on the surface of Earth. The Bay Bridge Public Information Office [16] uses Google Earth to
The images dynamically change as the users zoom in or flying around provide the general public a tool for viewing the ongoing construc-
to different areas of the globe. With the advent of the Internet, Google tion of the San Francisco-Oakland Bay Bridge's new self-anchored
Earth grew to over 100 million Internet users within one year of its suspension span and to monitor the construction process. The Lower
release [2]. Currently, Google Earth is a popular visualization tool in a Manhattan Construction Command Center [17] provides a four-
variety of fields and plays an increasingly important role not only for dimensional (4D) view of the downtown construction projects with
private users in daily life, but also for scientists, practitioners, policy- detailed information to the general public using Google Earth. The
makers and stakeholders in research and application [1–12]. National September 11 Memorial and Museum [18] utilizes Google
Google Earth reads and displays the Keyhole Markup Language Earth to virtually explore the 9/11 Memorial and other planned
(KML) files. KML was originally created in 2001 by a company projects within the context of the surrounding landscape of lower
called Keyhole as the format for its Earth browser named Earth Viewer. Manhattan by creating a virtual look at the World Trade Center site.
Since that time, KML has evolved to its status as an international These previous studies show the potential of Google Earth as an
standard, authorized by the Open Geospatial Consortium (http://www. effective tool for the presentation of the construction and mining
opengeospatial.org), for presenting geographic information visually. processes, using interactive 3D maps with true-color satellite imagery
KML enables users to add custom data to Google Earth and to share and sharing real-time information with the general public through
information about any subject that is geo-referenced on the Earth's the internet. However, relatively little attention has been paid to the
surface. The structure of KML has been documented by Wilson et al. [13] potential of Google Earth for optimal haulage routing of off-road
and Wernecke [14]. dump trucks in construction and mining sites.
For optimal route planning of on-road vehicles, Google Earth
provides a function, called ‘Getting Directions’, most likely based on
Dijkstra's [19] algorithm. Using this function, the users can analyze
⁎ Corresponding author at: Department of Energy Resources Engineering, the optimal route from an origin to a destination with contextual
Pukyong National University, Busan 608-737, South Korea. Tel.: + 82 51 629 6562;
fax: + 82 51 629 6553.
information along the route such as location of landmarks, turning
E-mail addresses: energy@pknu.ac.kr, yspower7@gmail.com (Y. Choi), instruction and details about the general layout of the surrounding
anieto@psu.edu (A. Nieto). area. However, the function is only applicable to areas where paved

0926-5805/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.autcon.2011.03.015
Y. Choi, A. Nieto / Automation in Construction 20 (2011) 982–997 983

(maintained) roads exist: difficulties are encountered in considering relief and curves along a route on the route planning. Using the
unpaved areas which occupy the major portion of working areas in algorithm, GEOHARTS can determine optimal haulage routes between
construction and mining sites. As shown in Fig. 1, Google Earth cannot loaders and dumps which ensure the least travel time or fuel con-
determine the optimal haulage route of off-road dump trucks from a sumption of off-road dump trucks. This paper describes the concept
loading point to a dumping point at the World Trade Center site in and details of software development, as well as its application to the
New York because the ‘Getting Directions’ function can only consider Pasir open pit coal mine in Indonesia, which revealed that GEOHARTS
the paved roads represented by a vector-based network data structure can provide a rational solution to support the off-road truck haulage
in spatial databases and approximates positions of the loading and operations.
dumping points on the unpaved areas into positions of the nearest
points on the paved roads. Moreover, this function cannot account
for the effects of terrain relief or of curves along a route on the route 2. Review of Dijkstra's Algorithm
planning. Considerations of these effects are critical for optimal
haulage routing of off-road dump trucks since they can significantly Dijkstra's algorithm is one of the spatial optimization techniques
influence the travel time or fuel consumption of the trucks at frequently used for optimal route analyses in Geographic Information
construction and mining sites. For instance, the travel time of off- Systems (GIS). It was designed for tracing the least-cost path (route)
road dump trucks moving through a route with the shortest physical in a network based abstraction (i.e. nodes connected by weighted
distance can be longer than other routes if there are many steep links) of space. For a given destination node in the network, Dijkstra's
positive slopes and horizontal curves along the route where trucks algorithm calculates the least accumulated cost between the
reduce their speed significantly [20]. In addition, the Getting destination node and every other node, and then finds the least-cost
Directions’ function of Google Earth only allows a single origin and path from any origin nodes to the destination node. The logical
destination for the route planning. Consequently, it is unsuitable in procedure of Dijkstra's algorithm is as follows:
large construction and mining sites where multiple loaders and
dumps are installed, and many off-road dump trucks are operated (1) Let the node at which we are starting be called the source node.
simultaneously. Assign to the source node an initial value of zero and to all other
This study presents new software, named Google Earth-based nodes an initial value of infinity. Mark all nodes as unvisited.
Optimal HAulage RouTing System (GEOHARTS), to improve the Set source node as current.
function of Google Earth for optimal haulage routing of off-road (2) For the current node, consider its unvisited nodes directly
dump trucks in construction and mining sites. A modified least-cost connected by links having cost values such as time, fuel
path algorithm is proposed to consider working areas with both paved consumption, dollar cost, etc. and calculate their accumulated
and unpaved temporary roads and to account for the effects of terrain costs from the unvisited nodes to the source node. If the new

Fig. 1. The problem of the ‘Getting Directions’ function of Google Earth for optimal haulage routing of off-road dump trucks from a loading point to a dumping point in the World
Trade Center site, New York.
984 Y. Choi, A. Nieto / Automation in Construction 20 (2011) 982–997

accumulated cost is less than the previously recorded cost, network and (2) the connections between neighboring grid cell centers
overwrite the cost. act as the links of the network. There have been many definitions of the
(3) When all nodes directly connected to the current node are neighborhood grid cell types for the connections, for example, 4-cell
completely considered, mark the current node as visited. A Rook's pattern, 8 cell Queen's pattern, 16-cell Knight's pattern [25]. The
visited node will not be checked again so that its accumulated Rook's and Queen's patterns consider the neighborhood grid cells
distance is final and the least. connected directly to a cell in a 3 × 3 cell window. An analogy can be
(4) If all nodes have been visited, finish. Otherwise, set the established between these connections and the moves in a chess game:
unvisited node (with the least accumulated cost to the source (1) the two vertical and two horizontal movements are represented as
node) as the next “current node” and continue from step 2. the Rook's pattern and (2) the movements to all direct neighborhood
grid cells (i.e. two horizontal, two vertical and four diagonal) are similar
to the moves of the Queen. In such cases, the horizontal turning angle
3. The modified least-cost path algorithm for optimal haulage between an incoming path and outgoing paths at the source cell is
routing of off-road trucks restricted to a minimum 45°. The Knight's pattern incorporates eight
more indirect neighborhood connections (i.e. two squares straight and
The ‘Getting Directions’ function of Google Earth, most likely based one square left or right in the direction of movement) with a 5 × 5 cell
on Dijkstra's algorithm, uses a vector-based network data structure to window. As the horizontal turning angle between an incoming path and
represent an abstraction of paved roads and determines optimal outgoing paths is restricted to a minimum 22.5°, the Knight's pattern
travel routes of on-road vehicles through the paved roads. Therefore, generally makes the least-cost path smoother [27]. In this study, the
it is not applicable to working areas in construction and mining sites Queen's pattern was used to construct the virtual network because it has
where both paved and unpaved temporary roads exist or where only been widely used in various applications and implemented in many
unpaved areas exist. This study considered a raster-based grid data types of GIS software, including ArcGIS.
structure to represent the working areas without paved roads and Once the virtual network is established, a source cell (where
analyzed the least-cost path on the raster surface as an optimal Dijkstra's algorithm will be initiated) should be identified. A
haulage route of off-road dump trucks. destination to which off-road dump trucks will move for haulage
There have been many studies aiming to develop raster-based operations is generally selected as the source cell, having an
least-cost path algorithms by modifying Dijkstra's algorithm [20–27]. accumulated cost of zero. Eight neighborhood cells in the Queen's
The raster-based least-cost path algorithms and their recent applica- pattern are searched from the source cell and examined to calculate
tions have been reviewed by Choi et al. [20] and Saha et al. [21]. There costs of passing from the cells to the source cell. Unlike the vector-
are some commercial GIS software packages such as ArcGIS (http:// based network data structure which already has a cost value at each
www.esri.com) and IDRISI (http://www.clarklabs.org) which contain link, the raster-based grid data structure needs to calculate the cost at
the raster-based least-cost path algorithms. ArcGIS provides a each virtual link using a cost map and a cost function during the
function, named PATH DISTANCE, based on a raster-based directional optimization process. The cost map is a raster data set in which the
least-cost path algorithm [28]. This study modified the raster-based value associated with each cell represents the cost (e.g. time, fuel
directional least-cost path algorithm implemented in ArcGIS to consumption, dollar cost, or risk) per unit length of passing through
account for the effects of terrain relief and curves along a route (Fig. 2). the cell. It can be generated from a land-use map by correlating land-
In the raster-based grid data structure, area is divided into rows and use classifications such as paved road, unpaved area, dumping area,
columns which form a regular gird matrix structure by setting the cell etc. with in-situ measurement data of off-road dump trucks such as
dimensions. Each cell within the grid matrix contains location co- speed, travel time or fuel consumption at different land-use con-
ordinates as well as an attribute value. To analyze least-cost paths on a ditions. The cost function determines a weighted distance (i.e. the
raster surface, a virtual network should be constructed under two cost of passing) between neighboring grid cell centers instead of
conditions: (1) the centers of each grid cell serve as the nodes in the calculating an actual physical distance. It should be noted that the
cost function applies the weighted distance in cost units, not in
geographic units. The cost function can be defined as non-directional
or directional according to characteristics of specific applications. A
non-directional cost function determines the cost of passing between
neighboring grid cell centers without considering the travel di-
rection. In contrast, a directional cost function determines the cost of
passing between neighboring grid cell centers dependent on the
travel direction, and thus the cost becomes non-symmetric [21].
Because the cost of off-road dump trucks moving uphill and the
one moving downhill can be significantly different in construction and
mining sites due to hilly terrain, this study defined a directional cost
function expressed by either travel time (s) or fuel consumption (l) as
follows:

CostN→S = 0:5 × SDN−S × VFN→S × ðCLN × HTFN + CLS Þ ð1Þ

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
SDN−S = D2N−S + ðElvS −ElvN Þ2 ð2Þ


D0pffiffiffi direct connection
DN−S = ð3Þ
D0 × 2 diagonal connection

 
1 ElvS −ElvN
θN→S = tan : ð4Þ
Fig. 2. Procedures of the modified least-cost path algorithm. D
Y. Choi, A. Nieto / Automation in Construction 20 (2011) 982–997 985

where CostN → S is the cost of passing from cell N to cell S, SDN-S is the truck moving downhill has a risk for break system failures), water-
the 2.5-dimensional (2.5D) physical distance (m) between the centers bodies, etc. and the LC and UC values indicate the lower and upper
of cell N and cell S, VFN → S is the vertical factor taking into account thresholds of θN → S where dummy is assigned to VFN → S. HTFN can be
the increasing cost when a truck moves uphill from cell N to cell S or determined by matching the Relative Horizontal Moving Angle
the decreasing cost when a truck moves downhill, CLN is the cost per (RHMA) between an incoming path and an outgoing path at a cell N
unit length (s/m or l/100 km) of passing through cell N, HTFN is the with a horizontal factor graph (Fig. 3C). Different values can be
horizontal turning factor to account for the effect of curves along a assigned to HTFN according to RHMA at the center of cell N, and HTFN
route on the increasing cost where a truck makes a turn to the right or above one, generally at curves along a route, increases the cost of
left with a relative horizontal angle between an incoming path and passing. However, no cost variation related to horizontal turning of
an outgoing path at the center of cell N, CLS is the cost per unit length off-road dump trucks occurs at a straight path, therefore HTF at the
of passing through cell S, DN-S is the 2-dimensional (2D) physical middle of centers of cell N and cell S (see the straight path from cell N
distance (m) between the centers of cell N and cell S, ElvS and ElvN are to cell S in Fig. 3D) is omitted in the Eq. (1). The dummy value is also
the elevations (m, above sea level) of cell S and cell N respectively, D0 used in the horizontal factor graph to represent the highly increased
is the size (m) of square grid cells and θN → S is the slope-gradient cost of passing at severe curves along a route where RHMA is too large
(degrees, −90–90) of terrain from cell N to cell S (Fig. 3A). for heavy off-road dump trucks to take the route. LC and UC are the
Since the physical distance between the centers of cell N and cell S lower and upper thresholds of RHMA where the dummy is assigned to
is used in the cost function, small measurement errors on the dis- HTFN. The graphs of vertical and horizontal factors can be created by
tances between successive pairs of cells through a path can accu- in-situ measurements of off-road truck performances such as speed,
mulate in the total haulage cost from an origin to a destination [25]. travel time or fuel consumption at different slope-gradients of terrain
Therefore, the 2.5D physical distance is used, derived from the 2D and RHMAs at curves along a route.
physical distance with considerations of changing slope-gradient of The raster-based directional least-cost path algorithm (PATH
hilly terrain, to improve the accuracy of the least-cost path algorithm. DISTANCE) implemented in ArcGIS uses a horizontal factor to consider
The values of ElvS and ElvN can be obtained from Digital Elevation the effects of wind direction and wind speed (i.e. global horizontal
Models (DEMs). As the 2.5D physical distance is non-directional, both friction) in analyzing the least-cost paths. However, their effects on
SDN → S and SDS → N are identical to SDN-S. speed, travel time or fuel consumption of heavy off-road dump trucks
VFN → S accounts the cost necessary to overcome the slope of is generally less significant than the one of curves along a route (i.e.
terrain from cell N to cell S. It can be determined by matching θN → S local horizontal friction). Moreover, dynamic changes of the wind
with a vertical factor graph of off-road dump trucks (Fig. 3B). VFN → S of direction and wind speed make it difficult to quantitatively assess
one indicates no cost variation related to terrain relief, and VFN → S their effects in calculating the cost of passing at each cell in the raster-
above one, generally at positive slopes, increases the cost of passing based grid data structure. Therefore, this study considered the local
and VFN → S below one, mostly at negative slopes, decreases it. It horizontal friction (i.e. the effect of curves) using HTFN instead of the
should be noted that the situation in which an off-road dump truck global horizontal friction in optimal haulage routing.
moves through a route, having both sections of positive slopes and Once the costs of passing from eight neighborhood cells to the
those of negative slopes, it can be effectively managed in the source cell are calculated using the cost map and the cost function,
optimization process using VFN → S. In the vertical factor graph, the Dijkstra's algorithm is initiated and determines one neighborhood cell
dummy value is used to represent the highly increased cost of passing having the least accumulated cost of passing to the source cell. The
at inaccessible areas such as steep slopes (where an off-road dump determined cell assigned the least accumulated cost is confirmed and
truck cannot have effective traction force enough to move uphill or then is defined as a current cell. From the current cell, another search

Fig. 3. (A) Concept for calculating the cost of passage from cell N to cell S. (B) An example of the vertical factor graph. (C) An example of the horizontal factor graph. (D) Concept for
determining the relative horizontal moving angle (RHMA) between two neighboring cells.
986 Y. Choi, A. Nieto / Automation in Construction 20 (2011) 982–997

of its eight neighborhood cells is started to find unconfirmed cells, and Table 1
the costs of passing from the unconfirmed cells to the current cell are Four different calculations of the least accumulated cost of passing from cell A to cell S in
Fig. 4.
calculated using the cost map and the cost function. This optimiza-
tion process is repeated until every cell in the raster-based grid data No. Cost function CostA → B CostB → S Accumulated
structure has the least accumulated cost of passing to the source cell CostA → S

and is confirmed. When the optimization process is finished, each 1 CostN → S = 0.5 × DN − S × (CLN + CLS) 75.00 s 106.07 s 181.07 s
cell has two attributes: the least accumulated cost of passing to the 2 CostN → S = 0.5 × SDN − S × (CLN + CLS) 86.60 s 150.00 s 236.6 s
3 CostN → S = 0.5 × SDN − S × VFN → S 69.28 s 225.00 s 294.28 s
source cell and a pointer indicating the direction which an off-road
× (CLN + CLS)
dump truck should move to trace the least-cost path. These attributes 4 CostN → S = 0.5 × SDN − S × VFN → S 69.28 s 247.50 s 316.78 s
are saved as two output files. Finally, the least-cost paths between × (CLN × HTFN + CLS)
origins (loaders) and destinations (dumps or stockpiles) which off-
road dump trucks will take are generated as optimal haulage routes
using the back-link mechanism [22].
To demonstrate the improvements of the modified least-cost path
algorithm proposed in this study, sample calculations were performed 4. Software Implementation
using a synthetic data illustrated in Fig. 4. Four different calculations
were carried out under the following conditions: A new software, GEOHARTS, was written in Visual Basic.NET 2005.
To embed a 3D render window of Google Earth into GEOHARTS, we
(1) Calculating the least accumulated cost by considering 2D flat used an open-source ActiveX control provided by the Google-Earth-
physical distance Airlines (http://www.googleearthairlines.com). This ActiveX control
(2) Calculating the least accumulated cost by considering 2.5D activates the interaction between Google Earth COM API (http://earth.
physical distance google.com/comapi) and GEOHARTS. It should be noted that the
(3) Calculating the least accumulated cost by considering 2.5D Google Earth client, a free stand-alone desktop application, needs to
physical distance and the effect of terrain relief (i.e. PATH be installed to utilize GEOHARTS. GEOHARTS operates on IBM PCs
DISTANCE in ArcGIS). The effect of wind is not considered. under the Microsoft Windows platform (i.e. XP, Vista and 7).
(4) Calculating the least accumulated cost by considering 2.5D Table 2 describes the input and output data files used in GEOHARTS.
physical distance and the effects of both terrain relief and The images such as BaseMapIMG, SlopeMapIMG and LandUseIMG
curves along a route (i.e. the modified least-cost path algorithm are used to improve the visualizations of local construction and mining
proposed in this study). sites where the satellite images of Google Earth are not updated very
frequently, and their spatial resolutions are low. The spatial resolution of
Table 1 shows the results from the four different calculations. The DEM can vary depending on the topographical survey methods used in
accumulated cost of passing from cell A to cell S was increased at construction and mining sites. However, the use of high-resolution
Calculation #2 compared to Calculation #1 as Calculation #2 con- DEMs (1–20 m) is desirable to calculate the least accumulated cost
sidered the 2.5D physical distance. When considering the negative by considering the effect of terrain relief. The resolution of DEM should
slope from cell A to cell B in Fig. 4, the cost of passing from cell A to cell be identical to that of LandUseGrid. The input data files need to be
B was decreased at Calculations #3 and #4, compared to Calculation frequently updated according to managerial conditions of construction
#2. In contrast, the cost of passing from cell B to cell S was increased at and mining sites.
Calculations #3 and #4 due to the effect of positive slope. Although GEOHARTS provides extended functions for optimal haulage
the Calculations of both #3 and #4 could consider the effect of terrain routing of off-road dump trucks in construction and mining sites.
relief through the route, the effect of curve at the center of cell B could The extended functions can be organized into four broad categories:
be only examined at Calculation #4; thus the cost of passing from cell system login, project settings, information visualization and optimi-
B to cell S at Calculation #4 was higher than the one at Calculation #3. zation. When an authorized user logs in to GEOHARTS, the main user
The sample calculations revealed that the modified least-cost path interface with the embedded 3D render window of Google Earth
algorithm used in this study can simultaneously consider multiple appears on the screen (Fig. 5). Based on the user's selections of a site
factors (i.e. the cost per unit length of passing at each cell, 2.5D physical and an off-road dump truck, data files related to the coordinates
distance, terrain relief, curves along a route) in analyzing the least-cost system of the site are loaded into memory and information about the
path on the raster surface. truck model, driver and GPS availability of the truck are displayed on
the interface status bar at the bottom of the window. The user also can
change the selections to check the status of other sites and trucks.
GEOHARTS displays the study domain using the embedded 3D
browser of Google Earth. Therefore, high-resolution satellite imagery
draped over a DEM can be displayed in conjunction with KML that
displays local site-specific data sets, e.g., a base map, a 2-D slope map,
a land-use map, information about loaders and dumps.
To display the base, slope, and land-use maps of the site, the
Ground Overlay element in KML is utilized. KML defines the Ground
Overlay element as an image that is draped onto the terrain of Google
Earth or added at a specified altitude above the Earth's surface. Using
the Ground Overlay element, the boundaries of base, slope, and land-
use maps in terms of latitude and longitude, the rotation of the maps
around their center (degrees, a counterclockwise direction), and the
altitude of the maps (m a.s.l.) are specified in GEOHARTS. To specify
how the altitude is interpreted, the ClampToGround option in KML is
used, which ignores the altitude specification and drapes the image
over the terrain. The Screen Overlay element indicates a 2D image
Fig. 4. Synthetic data for sample calculations of the least accumulated cost. such as a logo, credit or map legend that is fixed to a specific location
Y. Choi, A. Nieto / Automation in Construction 20 (2011) 982–997 987

Table 2 element has an icon and a label that can indicate the location and
Data files used in GEOHARTS. identity of a loader, dump or stockpile. The icon is, by default, a yellow
Type File Remarks pushpin, however, it can be replaced by specifying a custom icon in
IconStyle within the KML, specifying a custom image file. The
Input -UserInfo a,† -User account information
-SiteInfoa,† -Site information with links referencing the Placemark element can also have the Description element that allows
corresponding PrjInfo1 and CoordInfo1 files for the use of Hyper Text Markup Language 5 (HTML 5) [13]. Once the
-PrjInfoa,‡ -Information of absolute references to all data user clicks an icon in the 3D browser, a description balloon appears
files used in the selected site
that includes the name and image of a loader, dump or stockpile
-CoordInfoa,‡ -Coordinates system information of the selected site
-FleetInfoa,‡ -Fleet information of off-road dump trucks in the defined in KML using the Description element.
selected site Because the locations of loaders and dumps can dynamically
-BaseMapb,† -Information about parameter settings for the change during haulage operations, GEOHARTS uses the Network Link
ground overlay of BaseMapIMG3 and the screen element in KML to effectively manage timely updated information
overlay of BaseMapLgdIMG3
related to the loaders and dumps. The Network Link element allows a
-BaseMapIMGc,‡ -Base map of the selected site
-BaseMapLgdIMGc,‡ -Legend of the base map hierarchy of KML files that are linked to each other in a tree-like
-SlopeMapb,† -Information about parameter settings for the structure. A source (parent) KML file can activate GEOHARTS to fetch
ground overlay of SlopeMapIMG3 and the screen the linked files, load them into memory and display them in the
overlay of SlopeMapLgdIMG3
embedded 3D viewer of Google Earth. The fetched (child) KML files
-SlopeMapIMGc,‡ -Terrain slope-gradient map (deg.) of the
selected site can also contain links to another KML files. When the fetched KML
-SlopeMapLgdIMGc,‡ -Legend of the slope-gradient map files are loaded into memory, the Network Link element can refresh
-HaulPlanb,† -Information about parameter settings for the them using time-based parameters such as the RefreshInterval in
ground overlay of LandUseIMG3 and the screen KML [13]. We defined the RefreshInterval as 10, thus the locations of
overlay of LandUseLgdIMG3 with links referencing
loaders and dumps are refreshed every 10 s in GEOHARTS.
the Loading2,5 and Dumping2,5 files
-LandUseIMGc,‡ -Land-use map of the selected site When the user presses down the button for parameter settings
-LandUseLgdIMGc,‡ -Legend of the land-use map in the optimization category in the main user interface, a new user
-Loaderb,e,‡ -Information about a loader (e.g. location, interface, designed to apply the modified least-cost path algorithm,
equipment model, operator, icon image, picture)
appears on the screen (Fig. 6) and the files of DEM, LandUseGrid,
represented by a point
-Dumpsb,‡ -Information about dumps (e.g. location, equipment
TimeStudyInfo and FuelStudyInfo in Table 2 are loaded into memory.
model, operator, icon image, picture) represented The user can select an objective function (either travel time or fuel
by multiple points consumption) from the radio buttons and a truck model from the
-DEMd,† -Digital elevation model stored in a raster grid dropdown box, and then can press down the Draw button to review
-LandUseGridd,† -Land-use map stored in a raster grid
three charts that show in-situ measurements of the selected truck's
-TimeStudyInfoa,† -Average speed (km/hour) information of off-
highway trucks at different truck models, road performance at the site: the upper chart shows the speed (km/hour)
conditions, terrain slope-gradients, curved angles or fuel consumption (l/100 km) of loaded and empty trucks at dif-
and truck models measured in the selected site ferent land-use conditions, and the middle and lower charts illustrate
-FuelStudyInfo1,† -Average fuel consumption per distance (l/100 km)
graphs of the vertical factor regarding terrain relief and the horizontal
information of off-highway trucks at different truck
models, road conditions, terrain slope-gradients,
factor related to curves along a route respectively.
curved angles and truck models measured in the The IDs of loaders, dumps and stockpiles at the site are displayed
selected site on two checklist boxes; therefore the user can choose all or subsets of
Output -LeastCostDistd,† -A raster data set which indicates the least them as origins and destinations in optimal haulage routing of off-
accumulated travel time or fuel consumption of
road dump trucks. For the development of GEOHARTS, we considered
truck movement
-LeastCostDird,† -A raster data set that provides each cell with a three options of the haulage type in assigning origins and destinations
pointer that identifies the optimal connection to into the optimization process. Once the user chooses the “single trip of
neighborhood cells to ensure the least-cost of a loaded truck” option at the user interface, the loaders selected in the
truck movement checklist box are used as origins and the selected dumps are used as
-OptimalRoutesb,† -Optimal routes information with links referencing
the corresponding Route2,5 files
destinations. In contrast, in the case of the “single trip of an empty
-Routeb,e,‡ -Optimal route (from an origin point to a truck” option, the dumps selected in the checklist box are used as the
destination point) represented by a line-string origins and the selected loaders are used as destinations. The third
a
Comma-Separated Values (CSV) file. option, called the “round trip with the fixed assignment” allows
b
Keyhole Markup Language (KML) file. GEOHARTS to analyze optimal haulage routes of loaded dump trucks
c
Portable Network Graphics (PNG) file. by defining the selected loaders as origins and the selected dumps as
d
ESRI ArcGIS ASCII grid file.
e
potential halfway points, and the least accumulated cost of passing of
Multiple files are allowed according to the number of corresponding objects in the
selected site.
empty dump trucks from the halfway points to the origins through

Source file. the same haulage routes. It should be noted that the first and second

Fetched file. options are useful in a variable truck dispatching system, where new
origins and destinations are dynamically assigned into loaded and
empty dump trucks during haulage operations. The third option
can be effectively used in a fixed truck dispatching system, where
on the screen and specifies the size, position and rotation of the image. loaded and empty trucks use same haulage route between origins and
It should be noted that Screen Overlays are displayed in the 3D browser destinations.
area but operate independently of the Earth model, i.e., they are un- When the user presses the OK button, the optimization is per-
affected by navigation around or movement of the globe. GEOHARTS formed according to the sequential procedures in Fig. 2, and two
uses Screen Overlays to display map legends. output files (i.e. LeastCostDist and LeastCostDir in Table 2) are gen-
The locations of loaders, dumps and stockpiles are critical for erated. Because these output files are saved as the ESRI ASCII grid
planning the haulage operations in construction and mining sites. format, they are exchangeable with ArcGIS. For tracking the least-cost
GEOHARTS uses the Placemark element in KML to display information path between an origin and a destination, GEOHARTS uses a queue
about loaders, dumps and stockpiles in the site. The Placemark data structure in which the first element added to the queue will be
988 Y. Choi, A. Nieto / Automation in Construction 20 (2011) 982–997

Fig. 5. Main user interface of GEOHARTS with the embedded 3D render window of Google Earth.

the first one to be removed. The longitude, latitude and the least 5. Application
accumulated cost of passing at the origin are firstly inputted into the
queue, and then the values at the next cell, identified by the LeastCostDir The Pasir open-pit coal mine is located in the east side of Kalimantan,
file, are inputted into the queue (see Fig. 7). This enqueue operation Indonesia (115°53′E longitude and 1°51′S latitude) [35]. The opera-
continues by tracing the least-cost path until the values at the des- tional tasks in the Pasir mine consist of drilling, blasting, loading,
tination are inputted into the queue. When the enqueue operation is haulage, dumping, and general services. Of these, haulage is the most
finished, the values in the queue are extracted from the front of the expensive operation, because a large amount of ore and waste must be
queue and converted into the LineString element (i.e. a collection of delivered from inside the pit to outside the pit through relatively long
two or more coordinate values) in the Route KML file in Table 2 to and steep-gradient haulage routes using off-road dump trucks. Con-
display the least-cost path from an origin to a destination. Because sequently, improving the efficiency of haulage operations to reduce
GEOHARTS uses the OptimalRoutes KML file that includes the Network the travel time and fuel consumption of off-road dump trucks has
Link elements referencing multiple Route KML files, optimal haulage been increasingly demanded because the improved efficiency can be
routes of off-road dump trucks between multiple origins and destina- linked directly to the curtailment of haulage cost.
tions can be displayed on the screen and updated in a timely manner. A region of the Roto North pit that covers 14.04 km2 (3900 m ×
There have been many studies focused on developing truck 3600 m) was selected as a study area (Fig. 8). The highest elevation is
scheduling software to determine the number of trucks required and 188 m a.s.l., and the lowest is currently 102 m b.s.l.. Fig. 9 shows the
the dispatching intervals [29–34]. Because this study mainly considered base map of study area draped onto the true color satellite imagery in
the routing problem, a discussion of how to optimize the time when GEOHARTS. The study area has large unpaved working areas around
each event in a route should take place is beyond the scope of this paper. coal seams and several waterbodies both inside and outside the pit. In
However, this study attempted to provide necessary information to addition, paved haul roads connect the unpaved areas into waste dumps
external software that mainly considers the scheduling problem. To and an ore stockpile. Fig. 10 illustrates the slope map of study area
support external truck scheduling software, GEOHARTS provides a generated from a DEM with 20 m grid spacing. Average gradient of
summary report that includes the 2.5D physical distance (m) and the paved haul roads is 5° and one of bench slopes is 40°. A haulage plan
least accumulated cost (travel time or fuel consumption) of passing including the land-use map and distributions of loaders, dumps and a
through each optimal haulage route. In particular, the travel time of stockpile is displayed on the embedded 3D viewer of Google Earth in
each haulage route is necessary to optimize the dispatching schedule of GEOHARTS (Fig. 11A). The green pushpins indicate the four dumps
off-road dump trucks for haulage operations [29]. and the red pushpin identifies the stockpile. Custom icons are used to
Y. Choi, A. Nieto / Automation in Construction 20 (2011) 982–997 989

Fig. 6. User interface designed for parameter settings in the optimization process.

represent the two loaders inside the pit. When the user clicks an icon of KOMATSU HD785 as the truck model, the two loaders as origins and
the loaders, dumps or stockpile, a balloon of Google Earth that includes the stockpile as the destination. Based on these parameter settings,
images and descriptions opens as shown in Fig. 11B. Two kinds of three charts representing characteristics of the selected truck model
off-road dump truck models were considered in this application: were displayed as illustrated in Fig. 12A. The speeds (km/h) of the
(a) KOMATSU HD785 and (b) CATERPILLAR 777D. loaded truck at three different land-use conditions were converted
into travel times per unit length (s/m), and then the travel times per

5.1. Optimal haulage routing of loaded off-road dump trucks

To determine optimal haulage routes of loaded trucks, the option


of “single trip of a loaded truck” was selected in GEOHARTS. In
addition, we chose the travel time as the objective function, the

Fig. 7. Enqueue operations for tracing the least-cost path between an origin and a
destination. Fig. 8. 3D view of the Roto North pit at the Pasir open pit coal mine, Indonesia.
990 Y. Choi, A. Nieto / Automation in Construction 20 (2011) 982–997

Fig. 9. Visualization of a base map in the study area.

unit length were used to generate the cost map of the loaded truck loaded trucks from the loaders to the stockpile. The optimal haulage
in the study area. Because off-road dump trucks cannot cross the route connecting Loader 1 and the stockpile was almost identical to
waterbodies and the stockpile in the study area, a dummy value of the one in Fig. 12B. However, the optimal haulage route of loaded
1010 (s/m) was assigned to cells of the waterbodies and the stockpile trucks departing from Loader 2 was significantly changed compared to
in the cost map. The vertical factor at different slope-gradient con- the one in Fig. 12B; it was determined along paved roads with
ditions was used to account for the effect of terrain relief and the fewer curves because the effect of curves along a route on the fuel
horizontal turning factor at different RHMA conditions was utilized to consumption (Fig. 13A) was higher than that on the travel time
consider the effect of curves along the route. (Fig. 12A) in the study area.
Fig. 12B shows the optimal haulage routes of loaded trucks that From this sensitivity analysis, it can be noted that the optimal
ensure the minimum travel times to reach the stockpile. Therefore, we haulage routes determined by GEOHARTS are sensitive to the
know that loaded trucks departing from the loaders 1 and 2 should selection of the objective function. Therefore, careful considerations
take the same paved roads connecting to the stockpile for haulage are required to appropriately select the objective function according
operations. The 2.5D physical distance and the estimated travel to the specific managerial conditions at the site. For example, travel
time at each optimal haulage route were presented at the summary time is a better selection for the objective function when efficiency
report. Because the optimal haulage routes in Fig. 12B were deter- and productivity are critical in the site; however, fuel consumption
mined by considering not only physical haulage distance but also can be an alternative for the objective function during the period of
various parameters such as cost of passing per unit length, vertical high oil prices. In future work, it would be interesting to use multi-
slope-gradient and horizontal turning angle, they do not always objective optimization techniques to simultaneously consider both
coincide with the shortest Euclidean distance between loaders and travel time and fuel consumption in optimal haulage routing.
the stockpile.
To examine the sensitivity of objective functions for optimal 5.2. Optimal haulage routing of empty off-highway trucks
haulage routing of off-road dump trucks, the fuel consumption was
selected as the objective function, and other parameters and input To determine optimal haulage routes of empty trucks in the study
data sets were not changed. Fig. 13A illustrates the characteristics of area, we chose the “single trip of an empty truck” option as a haulage
the truck model regarding the fuel consumption in the study area. The type, the travel time as the objective function, KOMATSU HD785 as
fuel consumptions per unit length (l/100 km) at three different land- the truck model, the dumps 2, 3, 4 and the stockpile as origins and the
use conditions were used to generate the cost map. Fig. 13B shows two loaders as destinations. The characteristics of the selected empty
the optimal haulage routes that ensure the least fuel consumptions of truck illustrated in Fig. 12A were also used to create the cost map and
Y. Choi, A. Nieto / Automation in Construction 20 (2011) 982–997 991

Fig. 10. Slope map in the study area displayed on the embedded 3D render window of Google Earth.

to calculate the cost functions during the optimization process. optimal haulage routes determined by GEOHARTS are sensitive to
Fig. 14A shows the optimal haulage routes of empty trucks characteristics of the selected truck model. Each truck model can show
determined by GEOHARTS. The optimal route of empty trucks different performance depending on specific site conditions, therefore
connecting Loader 1 with the stockpile (2.5D physical distance: the characteristics of each truck model should be carefully measured at
3644 m) is almost identical to the one of loaded trucks (2.5D physical the site to improve the accuracy of optimization process in GEOHARTS.
distance: 3632 m) in Fig. 12B. However, the estimated travel time of
empty trucks moving downhill from the stockpile to Loader 1 (i.e. 5.3. Optimal haulage routing in the fixed dispatching system
2 min 38 s) is significantly reduced compared to the one of loaded
trucks moving uphill (i.e. 7 min 13 s). It is because the speed of empty To analyze optimal haulage routes of off-road dump trucks in
trucks is faster than the one of loaded trucks (see Fig. 12A) and the fixed dispatching system, we selected the option of “round trip
GEOHARTS can distinguish between negative and positive slopes of with the fixed assignment” as a haulage type, the travel time as the
terrain against truck movement using the directional cost function objective function, KOMATSU HD785 as the truck model, the two
(Eq. (1)). loaders as origins and the four dumps as potential halfway points.
To analyze the sensitivity of different truck models in optimal Fig. 14D illustrates the optimization results from GEOHARTS. The
haulage routing of empty trucks, CATERPILLAR 777D was selected as optimal haulage route of loaded trucks departing from Loader 1 was
the truck model. There were no changes in other parameters and input connected to Dump 2. The empty trucks should return to Loader 1
data sets. In the study area, CATERPILLAR 777D shows a relatively through the same haulage route after dumping. The estimated travel
faster speed at unpaved working areas and slower speed at paved roads time for the one-cycle round trip was 9 min, 6 s. Loader 1 was con-
in an empty condition (Fig. 14B) than the ones of KOMATSU HD785 nected with Dump 3 through the optimal routes with an estimated
(Fig. 12A). Fig. 14C shows optimal haulage routes of CATERPILLAR 777D travel time of 7 min, 26 s.
in an empty condition. The optimal haulage route connecting Dump 4
with Loader 2 in Fig. 14C was determined for both paved and unpaved 6. Conclusions
temporary roads, demonstrating that the route for unpaved temporary
roads used a shorter 2.5D physical distance rather than that for This study developed new software, GEOHARTS, to improve the route
paved roads (Fig. 14A). Other optimal haulage routes of CATERPILLAR planning functionality of Google Earth for use in the construction and
777D are almost similar to the ones for KOMATSU HD785, however, mining industries. Using the modified least-cost path algorithm,
their estimated travel times vary significantly. Therefore, we know that GEOHARTS can analyze optimal haulage routes of off-road dump trucks
992 Y. Choi, A. Nieto / Automation in Construction 20 (2011) 982–997

Fig. 11. Visualization of a haulage plan in GEOHARTS. (A) A land-use map and distributions of loaders, dump and a stockpile. (B) A balloon of Google Earth that includes an image and
description of the stockpile.
Y. Choi, A. Nieto / Automation in Construction 20 (2011) 982–997 993

Fig. 12. (A) Parameter settings and characteristics of KOMATSU HD785 related to travel time measured in the study area. (B) Optimal haulage routes of loaded trucks that ensure the
least travel time with a summary report.
994 Y. Choi, A. Nieto / Automation in Construction 20 (2011) 982–997

Fig. 13. (A) Parameter settings and characteristics of KOMATSU HD785 related to fuel consumption measured in the study area. (B) Optimal haulage routes of loaded trucks that
ensure the fuel consumption with a summary report.
Y. Choi, A. Nieto / Automation in Construction 20 (2011) 982–997
Fig. 14. (A) Optimal haulage routes of empty trucks (KOMATSU HD785) that ensure the least travel time. (B) Parameter settings and characteristics of CATERPILLAR 777D related to travel time measured in the study area. (C) Optimal haulage
routes of empty trucks (CATERPILLAR 777D) that ensure the least travel time. (D) Optimal haulage routes of off-road dump trucks (KOMATSU HD785) that ensure the least travel time in the fixed truck dispatching system.

995
996 Y. Choi, A. Nieto / Automation in Construction 20 (2011) 982–997

that ensure the least travel time or fuel consumption. These optimal [2] S.R.J. Sheppard, P. Cizek, The ethics of Google Earth: crossing thresholds from
spatial data to landscape visualization, Journal of Environmental Management 90
routes can be shared and synchronized using Google Earth clients in real- (6) (2009) 2102–2117.
time thorough wireless networks, thus significantly improving the [3] Y. Yamagishi, H. Yanaka, K. Suzuki, S. Tsuboi, T. Isse, M. Obayashi, H. Tamura, H.
dissemination of information to all truck drivers involved in the haulage Nagao, Visualization of geoscience data on Google Earth: development of a data
converter system for seismic tomographic models, Computers & Geosciences 36
operation. (3) (2010) 373–382.
An application to the Pasir open-pit coal mine in Indonesia has [4] M.N.K. Boulos, Web GIS in practice III: creating a simple interactive map of
shown that GEOHARTS is efficient, and brings the following advantages: England's Strategic Health Authorities using Google Maps API, Google Earth KML,
and MSN Virtual Earth Map Control, International Journal of Health Geographics 4
(1) (2005) 22.
(1) Since GEOHARTS uses a raster-based grid data structure for
[5] T.C. Patterson, Google Earth as a (not just) geography education tool, Journal of
the least-cost path analyses, it can overcome current limitations Geography 106 (4) (2007) 145–152.
of the ‘Getting Directions’ function of Google Earth so that (a) [6] P.W. Webley, Virtual Globe visualization of ash–aviation encounters, with the special
case of the 1989 Redoubt–KLM incident, Computers & Geosciences 37 (1) (2011)
working areas with both paved roads and unpaved temporary
25–37.
roads, (b) the effects of terrain relief and curves along a route and [7] L. Postpischl, P. Danecek, A. Morelli, S. Pondrelli, Standardization of seismic
(c) multiple origins and destinations can be simultaneously tomographic models and earthquake focal mechanisms data sets based on web
considered for optimal haulage routing of off-road dump trucks. technologies, visualization with keyhole markup language, Computers & Geosciences
37 (1) (2011) 47–56.
(2) Users can readily update the parameter settings related to [8] T.E. Wright, M. Burton, D.M. Pyle, T. Caltabiano, Visualising volcanic gas
the objective function, the truck model, the loaders and dumps plumes with virtual globes, Computers & Geosciences 35 (9) (2009)
involved and the haulage type for the optimization process. 1837–1842.
[9] N.Q. Chien, S.K. Tan, Google Earth as a tool in 2-D hydrodynamic modeling,
Therefore, GEOHARTS is most applicable to operating construc- Computers & Geosciences 37 (1) (2011) 38–46.
tion and mining sites, in various areas and countries, where [10] B. Tomaszewski, Situation awareness and virtual globes: Applications
the alteration of managerial decisions and economic conditions for disaster management, Computers & Geosciences 37 (1) (2011) 86–92.
[11] L.M. Ballagh, B.H. Raup, R.E. Duerr, S.J.S. Khalsa, C. Helm, D. Fowler, A. Gupte,
can dynamically change the demands for the optimal haulage Representing scientific data sets in KML: Methods and challenges, Computers &
routing with time. Geosciences 37 (1) (2011) 57–64.
(3) GEOHARTS allows users to obtain both optimal haulage routes [12] D.G. De Paor, S.J. Whitmeyer, Geological and geophysical modeling on virtual globes
using KML, COLLADA, and Javascript, Computers & Geosciences 37 (1) (2011)
of trucks and approximately estimated travel times along
100–110.
the routes in a single framework. Therefore, it can be used to [13] T. Wilson, D. Burggraf, R. Lake, S. Patch, R. Martell, B. McClendon, M. Jones, M.
complement other truck dispatching software that mainly Ashbridge, J. Wernecke, C. Reed, OGC® KML, Open Geospatial Consortium, Document
OGC 07-147r2Available from: http://www.opengeospatial.org/standards/kml. 2008
considers scheduling problems.
[Accessed 7 Dec 2010].
[14] J. Wernecke, The KML Handbook: Geographic Visualization for the Web, Addison-
For the purpose of routing trucks within construction and mining Wesley, Boston, 2009.
sites, high-resolution satellite images with higher frequency of updating [15] E. Sun, A. Nieto, Z. Li, GPS and Google Earth based 3D assisted driving system for
are desirable. However, the satellite images of Google Earth are not trucks in surface mines, Mining Science and Technology (China) 20 (1) (2010)
138–142.
updated very frequently, and their spatial resolutions are low in some [16] Bay Bridge Public Information Office, Bay Bridge uses Google Earth to reveal iconic
areas. Therefore, it is required to use GEOHARTS with additional data construction processAvailable from: http://baybridgeinfo.org/bb_placemark.
such as local aerial photography, thematic maps, etc. to improve the 2010[Accessed 7 Sep 2010].
[17] Lower Manhattan Construction Command Center, Construction projects and
visualizations of the construction and mining sites. In addition, street impactsAvailable from: http://lowermanhattan.info/construction/ipm/4d/
GEOHARTS is not truly relevant as a fully automated haulage system at map/index.php. 2010[Accessed 7 Sep 2010].
construction sites since the haulage operations in the changing outdoor [18] National September 11 Memorial and Museum, Google Earth: a virtual look at
the World Trade Center siteAvailable from: http://www.national911memorial.
environments are so difficult to fully automate and still require the hand org/site/PageServer?pagename=New_Visit_GoogleEarth. 2010[Accessed 7 Sep
of an experienced operator. However, this does not detract from the value 2010].
of the technique described in this paper because GEOHARTS can support [19] E.W. Dijkstra, A note on two problems in connection with graphs, Numerische
Mathmatik 1 (1959) 269–271.
semi-automated operations, as an intermediate phase in the develop-
[20] Y. Choi, H.D. Park, C. Sunwoo, K.C. Clarke, Multi-criteria evaluation and least-cost
ment of fully automated systems, by providing truck drivers with optimal path analysis for optimal haulage routing of dump trucks in large scale open-pit
haulage routes to improve the human operator's performance. mines, International Journal of Geographical Information Science 23 (12) (2009)
A relatively straightforward extension of the current work would 1541–1567.
[21] A.K. Saha, M.K. Arora, R.P. Gupta, M.L. Virdi, E. Csaplovics, GIS-based route
involve (1) the introduction of multi-objective functions for the planning in landslide-prone areas, International Journal of Geographical Infor-
optimization and (2) the implementation of optimal truck scheduling mation Science 19 (10) (2005) 1149–1175.
functionality in GEOHARTS. Finally, it should be noted that Google has [22] J. Xu, R.G. Lathrop, Improving cost-path tracing in a raster data format, Computers
& Geosciences 20 (10) (1994) 1455–1465.
announced that the COM API is no longer to be supported (http:// [23] S.C. Feldman, R.E. Pelletier, E. Walser, J.R. Smoot, D. Ahi, A prototype for pipeline
googlegeodevelopers.blogspot.com/2010/08/sunset-for-google- routing using remotely sensed data and geographic information system analysis,
earth-com-api.html). Therefore, in future work, it would be desirable Remote Sensing of the Environment 53 (2) (1995) 123–131.
[24] W. Collischonn, J.V. Pilar, A direction dependent least-cost path algorithm for
to develop a web-based version of GEOHARTS using the JavaScript API roads and canals, International Journal of Geographical Information Science 14 (4)
(development and support for which supersedes COM API support) (2000) 397–406.
that eliminates the need to install and run the Google Earth client. [25] C. Yu, J. Lee, M.J. Munro-Stasiuk, Extensions to least-cost path algorithms for
roadway planning, International Journal of Geographical Information Science 17
(4) (2003) 361–376.
Acknowledgements [26] W.G. Rees, Least-cost paths in mountainous terrain, Computers & Geosciences 30
(3) (2004) 203–209.
[27] C. Zhan, S. Menon, P. Gao, A directional path distance model for raster distance
This work was supported by the Energy and Resources Engineering
mapping, in: A.U. Frank, I. Campari (Eds.), Spatial Information Theory: a
Program Grant funded by the Ministry of Knowledge Economy, South theoretical basis for GIS, Springer-Verlag, Berlin, 1993, pp. 434–443.
Korea. The authors thank the anonymous peer reviewers, who provided [28] ESRI, Path distanceAvailable from: http://webhelp.esri.com/arcgisdesktop/9.3/
useful comments that significantly improved the final paper. index.cfm?TopicName=Path_Distance. 2010[Accessed 7 Sep 2010].
[29] V.A. Temeng, A computerized model for truck dispatching in open pit mines, PhD
dissertation Michigan Technological University, Michigan, 1997.
References [30] G.H. Blackwell, Estimation of large open pit haulage truck requirements, CIM
Bulletin 92 (1028) (1999) 143–149.
[1] A. Chen, G. Leptoukh, S. Kempler, C. Lynnes, A. Savtchenko, D. Nadeau, J. Farley, [31] S. Alarie, M. Gamache, Overview of solution strategies used in truck dispatching
Visualization of A-Train vertical profiles using Google Earth, Computers & systems for open pit mines, International Journal of Surface Mining, Reclamation
Geosciences 35 (2) (2009) 419–427. and Environment 16 (1) (2002) 59–76.
Y. Choi, A. Nieto / Automation in Construction 20 (2011) 982–997 997

[32] C. Niemann-Delius, B. Fedurek, Computer-aided simulation of loading and transport [34] S. Yan, W. Lai, An optimal scheduling model for ready mixed concrete supply with
in medium and small scale surface mines, in: M. Hardygora, G. Paszkowska, M. Sikora overtime considerations, Automation in Construction 16 (6) (2007) 734–744.
(Eds.), Mine Planning and Equipment Selection, Taylor & Francis Group, London, [35] Y. Choi, H.D. Park, C. Sunwoo, Flood and gully erosion problems at the Pasir open
2004, pp. 579–584. pit coal mine, Indonesia: a case study of the hydrology using GIS, Bulletin of
[33] H. Zhang, Multi-objective simulation-optimization for earthmoving operations, Engineering Geology and the Environment 67 (2) (2008) 251–258.
Automation in Construction 18 (1) (2008) 79–86.

You might also like