Professional Documents
Culture Documents
Modeling Load-Haul-Dump Underground: Controlled/Automatic
Modeling Load-Haul-Dump Underground: Controlled/Automatic
Controlled/Automatic Load-Haul-Dump
Vehicles in Underground Mines
Nick Vagenas
School of Engineering
Laurentian University
Sudbury, Ontario, Canada
nvagenas@nickel.laurentian.ca
In the last
few years, a concept for 1. Introduction
automating Load-Haul-Dump (LHD) During the past few years, the concept of integrating
vehicles has stimulated considerable interest automation and teleoperation on Load-Haul-Dump
in the world mining industry. In this (LHD) machines (Figure 1) has attracted considerable
concept, the tramming and dumping interest in the international mining community. In
operations of an LHD should be automatic. this concept, the transport (tramming) and dumping
During loading, an operator from a control operations of an LHD should be automatic. During
room fills the bucket of a vehicle via remote- loading, an operator from a control room fills the
control aided by a television system. The bucket of a vehicle via remote-control aided by a
television system; see Figure 2. Using such a Remote-
application of such a Remote-controlled/ controlled/Automatic LHD (RAL) system, several
Automatic LHD (RAL) system in
advantages can be foreseen, e.g., better utilization of
underground hard rock mines exhibits some the vehicles, a better working environment for the
operational and traffic control characteristics operators, and better transport control [1, 2].
that have not been studied previously in At present, the general practice in most of the
mining (e.g., vehicle motion in bidirectional mines is one LHD machine to operate in a limited
lane segments). This paper presents a transport area with little or no traffic interference
discrete-event simulation model developed from other vehicles. This practice is applied mainly
for studying and evaluating the operation of for safety reasons since the vehicles operate in
tunnels with a width that prohibits two-way traffic.
a fleet of RALs. This model integrates Even though LHD machines have been used for
graphical modeling and presentation of several decades in mining, the utilization of a fleet of
RAL transport layouts with animation of
RALs exhibits a number of operational and traffic
the motion of vehicles during simulation.
control characteristics that have not been studied
The model is used to evaluate the capacity of
previously in mining. These characteristics may be
a future fleet of RALs which can operate in described as follows:
a typical transport layout at the LKAB
-
Vehicle motion in bidirectional lane-lane ents
Swedish iron ore mine in Kiruna.
(vehicles can move in either direction, but if two or
Keywords: Underground hard rock more vehicles are moving in opposite directions
mines, LHD vehicles, automation, inside a bidirectional lane segment, they will collide);
mining
331
332
operation of a fleet of RALs, a discrete-event simula- operators are necessary for loading and for
tion model has been developed. With this model, efficient operation of a fleet of RALs ?
called PROFITIS Version 2.0 [3], it is possible for a The above problems are concerned mainly with the
system planner to alter several operational character- effects of mining transport layouts on the perfor-
istics of an RAL system and study the behavior and mance of an RAL system. Other problems can also be
performance of the system. This model possesses the considered, as for instance, the problem related to the
following general features: effect of utilizing electric RALs powered by trailing
.
Computer-Aided Design (CAD) to create and cables, due to the limited cable length of the electric
display graphs of RAL transport layouts; LHD and the precautions taken for preventing the
vehicle from crossing over the cable. A detailed
· Graphical animation to animate the motion of
vehicles during a simulation run and display discussion can be found in [3].
the events executed; In general, an RAL system is not a fully automatic
. Text output simulation for fast execution of a system since teleoperation of the loading activity
simulation run, displaying only the final text requires human involvement. The number of opera-
tors, coffee and lunch breaks, as well as shift changes
output information.
may therefore affect the performance of a fleet of
At present, most of the work on LHD automation is RALs. In general, it is not always feasible to find an
only hardware oriented (e.g., selection of navigation optimal system performance that can satisfy all the
techniques and choice of data communication various operating conditions in a mine. The develop-
systems). It focuses on the development of prototype ment and utilization of a simulation model that
vehicles. None of this work is concerned with traffic
provides a system planner with an opportunity to
management and simulation modelling of a fleet of alter the operational characteristics of an RAL system
vehicles. In general, the currently applied practice is and experiment with alternative operational policies
the utilization of a single vehicle in a limited area can be a powerful and useful tool. It is also possible
containing a few loading and dumping points. to apply stochastic mathematical models using other
On the other hand, the manufacturing industry has
operations research techniques (e.g., queuing net-
experienced a dramatic growth in Automated work theory) instead of simulation.
Guided Vehicles (AGV) technology. Work in traffic
management and modelling of industrial AGV 3. Rules, Constraints, and Assumptions Used
systems has been described in the literature; see, for in the Construction of the Simulation Model
example, [5-9]. This research, however, is mostly 3.1 Introduction
applicable to AGV systems in which unidirectional
or two-way unidirectional guidepaths are usually PROFITIS Version 2.0 is a discrete-event simulation
considered. model programmed in Turbo Pascal. The variables of
In unidirectional lane segments, vehicles can move the model change at a discrete set of points in time.
in one direction only. Using a two-way unidirectional The state of the model can be altered by events
lane segment, vehicles moving in opposite directions scheduled to occur in future time. A simulation clock
never collide. In general, there are major differences is used to keep track of events and the current value
between underground mining and other industries of time as the simulation proceeds.
with regard to automation of vehicle-based systems; The occurrence of an event causes the execution of
for a discussion, see [3]. logics for changing the status of the model variables,
updates the simulation time, and collects statistics of
2. The Problem Statement interest.
The simulation time is far smaller than the actual
In this paper, problems concerning the utilization of
a fleet of RALs in an underground mining transpor- operating time (a six hour operating time (shift)
of an RAL system takes less than 100 seconds of
tation network will be discussed. The problems
simulation time on an IBM compatible 486-66 MHz
considered are the following:
Given a mining transport layout, how many
computer). The event scheduling algorithm used in
(a) the model for advancing the simulation clock is based
RALs are required to handle a given volume of on the variable time increment method in which
production? simulation time is advanced by the amount necessary
(b) Given a mining transport layout, how many RALs to cause the next most imminent event to take place.
can operate simultaneously without creating Three principal groups of input to the model can be
traffic saturation? identified:
333
by the user, namely the exponential, normal, lognor- (a) Pick up its load at a loading point;
mal, Poisson, Weibull, and uniform distributions. (b) Move along a segment;
The mining transport layout or, more specifically, (c) Cross a traffic zone;
the RAL guidepath layout, is modeled as a collection
(d) Dump its load at a dumping point.
of nodes and segments. Nodes are control locations
on the guidepath at which an RAL can stop. Nodes
During motion, the activities of the vehicles may be
described as follows.
include not only the repair station, loading, and
· Before entry to a segment, the vehicle checks the
dumping points, but also points where the paths
traffic conditions in the segment. If it is allowed to
merge, diverge, or cross (traffic nodes). Segments
represent the aisles on which vehicles can move. A enter, the RAL then begins to move in the segment.
segment is a directed section of the guidepath between · Move along a segment: When a vehicle enters a
two adjacent control locations. A segment is always segment its movement is regulated by the traffic
drawn as a straight line. Consequently, curves are conditions in this segment. z
represented as a sequence of straight segments. · End of a segment: When reaching the end of a :
334
335
336
337
Dumped (TTD) per six hour shift. Input data for the (b) The time for the loading operation of an RAL
transport layout of Figure 7 is based on records from should not exceed that of a conventional LHD of
the Kiruna mine. In this case study, an attempt is
made to investigate the following problems: comparable size by more than 30%;
(c) The time for the dumping operation of an RAL
· Given the transport layout of Figure 7, how many
should not exceed that of a conventional LHD of
RALs can operate simultaneously until traffic
saturation is reached? comparable size by more than 30%.
· Given the transport layout of Figure 7 and a fleet of Assumption (b) is taken into consideration since an
RALs, how many operators are necessary to
RAL is anew/prototype system and thus time is
supervise the loading operation of RALs for required by the operators to familiarize themselves
with and take full advantage of the RAL system’s
efficient operation of the RAL system in a given
time period? capabilities. Early experimentation with RAL systems
have indicated that the above mentioned assump-
The most interesting measures of performance in tions are appropriate, see [14]. The case study pre-
this case study are the following: sented in this section does not deal with the eco-
· Total production achieved (per six hour shift); nomic aspects of installing and operating an RAL
· Vehicle waiting times at loading points; system. The general input data related to the trans-
· Vehicle waiting and blocking times. port layout of Figure 7 are shown in Table 1.
For convenience, the following definitions are used.
· The Total Travel Time of a vehicle while traveling
Empty (TTTE) is the total time that a vehicle spends
traveling from dumping points to loading points.
· The Average Travel Time of all vehicles while
traveling Empty (ATTE) is
338
339
340
operators for the loading activity from one to three. operator for the teleoperation of the loading
The transport layout in Figure 7 includes three loading is substantial.
activity not
points. The maximum number of operators needed is ~
The increase in production is due mainly to the use
three. The output from ten repeated runs using
of additional vehicles in the layout of Figure 7.
different random number seeds and 95% confidence
limits, is shown in Table 2. The value of the t distribu- Capital investment and operating costs of the
vehicles therefore become very important when we
tion with (10-1) degrees of freedom, which produces a
wish to evaluate the economic feasibility of installing
95% confidence interval, is t(100-95)/2 = 2.26 [15].
an RAL system with several vehicles in simultaneous
The dispatch procedure manages to supervise a
maximum of five RALs in the layout of Figure 7. operation.
Traffic saturation is reached with six RALs. This is
~
The dispatch procedure can supervise a maximum
of five vehicles before traffic saturation is reached.
mainly due to the lack of available loading points
and the short cycle times of the vehicles (cycle time = ~
The traffic factor of vehicles in the system, even
loading time + transport time loaded + dumping with five RALs, is considered to be acceptable;
time + transport time empty). Since the dispatch AWT is less than 17% of the total operating time,
procedure cannot send the sixth vehicle to a loading with ATF ranging between 73% and 87%.
point, then the waiting time for the sixth vehicle It is interesting to mention that recently reported
increases rapidly. Furthermore, the short cycle times results and benefits from a demonstration of
of the other vehicles eliminate the chances of the
teleoperation of two automatic LHDs by one operator
sixth RAL finding a feasible route to a loading point. on the surface by INCO Limited, a mining company
The production varies from 2543 tons/shift for two in Canada, see [2], indicate agreement with the
RALs with one operator to 5945 tons/shift for five conclusions reported in this paper. Baiden and
vehicles and three operators. Thus, by adding three Henderson in [2] reported an expected increase in
vehicles to a fleet of two RALs and using three
productivity gains (in tons/person-shift) by a factor
operators instead of one for the loading activity, the of 2.4 in the case of teleoperation of five automatic
production is increased by a factor of 2.34. Five RALs LHDs from the surface by one operator instead of
with one operator, however, produce only 6% less
teleoperation of two automatic LHDs by one opera-
than five vehicles with three operators. tor. This simulation study indicates (see Table 2) that
Furthermore, by studying the results presented in an increase in the production by a factor of 2.2 is
Table 2, it can be seen that there is no significant
possible when we are operating five RALs with one
contribution to the production if the number of
operator (5608 tons/shift) instead of two RALs with
operators is increased from two to three for all one operator (2543 tons/shift).
simulation runs. This is true, since the maximum
average waiting time for all cases in Table 2 does not 5. Conclusions
exceed 0.28 minutes < 17 seconds (Table 2, the case of
This paper discusses simulation modelling of a fleet
five RALs with one operator). Waiting times at
of remote-controlled/automatic LHDs in under-
loading points of less than 17 seconds are very short
and will not affect the production significantly. ground mining. A discrete-event simulation model
Interesting results are also related to the Average (PROFITIS Version 2.0) has been developed for
this purpose. This model integrates graphical model-
Blocking Time (ABT) of the vehicles. The ABT varies
from 0.81 hours 49 minutes to 1.08 hours 65
= = ling and presentation of RAL transport layouts, as
well as animation of vehicles during simulation. The
minutes as shown in Table 2. Thus, on average, for
one hour from a total of six operating hours, the
application environment of the model corresponds to
that found in a real underground transport layout
vehicles are blocked less than 17% (1/6 hours) of
the total operating time. This is better illustrated if consisting of several loading points and dumping
the Average Traffic Factor (ATF) is used. The ATF points (ore passes). The simulation model is applied
to evaluate the capacity of a future RAL system in a
varies from 73% to 87% according to Table 2. These
values are quite satisfactory if we consider that these typical transport layout at the LKAB Swedish iron
ore mine in Kiruna.
vehicles are running in a network with bidirectional
road segments. In the manufacturing industry,
341
342