Optimization of Open Pit Haulage Cycle Using a

KPI Controlling Alert System and a Discrete-Event

Operations Simulator
Pedro Pablo Vasquez Coronado and Victor Octavio Tenorio
Mining and Geological Engineering Department/University of Arizona, Tucson, AZ, USA

ABSTRACT: he loading cycle in an Open Pit mine is a critical stage in the production process that
needs to be controlled in detail for performance optimization. A Comprehensive Alert System (CAS)
was designed in order to notify supervisors of cycle times that are below the required performance
standards. he system provides alert messages when one or several trucks are idle or the time of
completing production tasks are ofsetting from pre-established values. his alert is identiied by the
system and compared with context-speciic Key Performance Indicators (KPIs) in order to identify
the least productive times of the leet. he goal is to develop a methodology to help the production
supervisor deining the strategies and corrective actions required for the optimization of the haulage
cycle model. A discrete-event simulator has been built in order to analyse diferent scenarios for
route design and queue analysis. he results are displayed in a customized dashboard every time the
simulation has inished a complete shift or full-day sessions.
his paper focuses on the optimization of haulage using the Alert System. However, the
system is intended for implementation in subsequent stages of the production process, and
the resulting improvement could impact mine planning and management as well. A simpliied
model of a hypothetical Open Pit mine was used to run a discrete-event simulation in order to
generate operating cycles that represent actual operations (As-Is model). After the Alert System was
implemented, operational settings were adjusted over trucks, shovels, crushers and routes, and a
new simulation was performed following the recommended changes (To-Be model). Comparative
analysis of productivity and statistical results of two diferent production stages are presented.

INTRODUCTION
Productivity in Open Pit mines depend on the observation and adjustment of critical phases of
operation. Transportation is one of the most expensive components of the production process,
making it necessary to concentrate eforts in optimizing the load-haul-dump cycle. However, the
everyday work makes some elementary aspects go unnoticed. Current operation tasks leave almost
no time to verify the quality or precision of the data being collected. With the latest developments
in technology, it is possible to continuously monitor the various activities of the equipment, record-
ing vital signs and storing them in a database, aiming to data centralization and integration for a
more elaborate analysis. he status of equipment performance can be displayed either in real time
or at the end of the shift, which allows tracking the mine plan, and opening opportunities for
improving operations. he mining industry is always looking for new challenges in optimizing its
34
 Optimization of Open Pit Haulage Cycle Using a KPI Controlling Alert System 35

processes in order to have a sustainable operation, improve worker safety and equipment productiv-
ity, avoiding unwanted delay times that may afect the overall production plan. Key Performance
Indicators (KPIs) are deined for a comparison between actual performance and ideal operating
conditions.

RESEARCH OBJECTIVE
he purpose of this study is to implement a methodology for optimizing the haulage cycle by
identifying and adjusting the factors that afect the cycle time performance. he methodology
includes the implementation of an alert system emitting warning messages when one or several
trucks are idle or the time of completing production tasks is falling outside a predeined value. his
alert is compared with pre-established Key Performance Indicators (KPIs) in order to determine
the following corrective actions: Demonstrate the beneits of time analysis and how this afects the
productivity outcomes in the mining process; Demonstrate the utility of KPIs in the extraction of
ore and waste, and how they help to identify equipment performance; Develop software interfaces
to visualize the Alert System in order to increase awareness in the operations supervisor; and Make
recommendations for future applications that can help improve the productive process of an Open
Pit mine.

PROBLEM STATEMENT
Transportation of material in Open Pit mines demands the utilization of high capacity equipment,
which also represents an elevated cost per hour. A supervisory system that controls the proper dis-
tribution of trucks and shovels by providing the operational status of each unit shall contribute to
increase productivity and subsequently reducing operational costs. Nevertheless, there is an issue
with certain dead times such as queue trucks for loading, trucks waiting to be loaded, and other
delays. his creates ineiciencies in fuel consumption and energy utilization.
According to the existing literature, there are several research initiatives for improving the cycle
times in mining operations. However, regarding alert systems actuating in the haulage process for
identifying delays, a context-based system could be required for providing more information and
help optimizing the haulage cycle time by controlling delays in loading, haulage, and dumping.
After applying this methodology, eforts were focused in correcting out-of-standard performance
values.

PREVIOUS WORK
Techniques for increasing productivity in open pits are mainly based on the Fleet Management
System (FMS), seeking to increase the productivity of the haulage process and lower operation
costs. A computer in every vehicle is the provider of mobile data to the system. Ercelebi and
Bascetin (2009) refer to optimizing the number of trucks per shovel in hauling by inding the opti-
mum number of truck assignments to shovels using linear programming. Some automated systems
help to reallocate patch trucks. Sgurev et al. (1989) describe the fundamentals of a system based
on controlling, monitoring, and reporting the truck leet solving the issue where a large number of
trucks cause queues, and a small number of them causes downtime of the shovels. Ataeepour and
Baai (1999) propose to optimize the production process, taking for the irst time the decreasing
wait time of trucks to be loaded for an available shovel, thus improving production and equipment
utilization. A practical assumption is considering all equipment units are of the same characteristics.
36 Mine Discrete-System Simulation I

Simulations allow testing diferent scenarios for decision making. Kelton et al. (2007) deine
simulation as a broad collection of methods and applications that mimic the behavior of actual
systems. With the present-time computing capability, it is possible to simulate complex operating
conditions. Open Pit scenarios can be designed for productivity analysis, utilizing building blocks
that may represent trucks, shovels, crushers, waste dump location, and dynamic processes such as
loading, traveling, and dumping. Highly accurate results are obtained with graphical representa-
tions to support project goals. In addition, the functionality of the software allows analyzing cycle
times because the probability distribution based on ield data is shaped to scholastic variables. Data
of every cycle time is recorded during the simulation to study the impact of this cycle in production
and performance (Krause, 2006). Underground haulage systems have also been modeled and ana-
lyzed utilizing simulation software (Pop-Andonov, 2012). Proiciency and functionality of a mine
layout is tested using a Monte Carlo-type simulation procedure.
Models help to estimate the behavior of operations and identify problems. An “As-Is” model
represents in a simpliied way the operation of the mine, analyzing the results and identifying
potential bottlenecks (Tan et al., 2012). It is a prerequisite for understanding how processes are
currently executed in a system (Lodhi et al., 2010). he “To-Be” model is the set of results obtained
after incorporating the improvements found on the initial model. Any adjustment can be applied
directly without modiication of the original coniguration (Tan et al., 2012).
Cycle Time is deined as the time spent by any equipment to complete one cycle of operation.
For a truck, it includes the time to spot and load, travel to the dump site; maneuver, spotting, and
dump and drive back to the loading point, along with predictable and unpredictable delays, and
wait times (Lineberry, 1985). Storage of loading times, travel time, queues and unloading times in
a database are required in order to deine the sequence of operation. Ercelebi and Bascetin (2009)
calculated the number of trucks per shovel for decreasing the cost of material movement.
Key Performance Indicators are collections of measured data that are used for evaluating the
performance of an operation. hey are the tools utilized by management to evaluate the perfor-
mance of a particular activity. hese assessments usually compare the actual and estimated perfor-
mance in terms of efficiency, eicacy, and quality (Cox et al., 2003).

METHODOLOGY
An Alert System that detects performances that outlay established values was developed using an
approach that helps controlling the cycle time by identifying idle times over standard values during
the haulage activities; also, results are obtained against predeined KPIs (Figure 1). In order to pro-
duce data that constitutes the source database for loading cycles, a primordial model of an Open Pit
mine was populated with cycle times for up to 100 replications of a full working day (two shifts).
his set of simulations and results produced the As-Is model, containing the critical components
related to the production/transportation outcomes. he results of the production cycle are collected
to understand how the components interact and which of them are adjustable, as a starting point
of recognizing opportunities and identifying weaknesses for improvement (Schwegmann & Laske,
2003). he Alert System monitored and detected out-of-standard performances in a database. he
factors that impacted negatively in the performance were detected after a detailed post-operation
analysis. he To-Be model was generated with the optimization of haulage cycle times. he goal is
to demonstrate that the production of an Open Pit mine can be increased with better control of
time in the performance of haulage trucks.
 Optimization of Open Pit Haulage Cycle Using a KPI Controlling Alert System 37

Figure 1. Alert System process worklow

DEVELOPMENT OF THE COMPREHENSIVE ALERT SYSTEM

he system detects the variability of performance standards of the main operational parameters:
queue, number of trucks, breakdowns, shutdown, and slow/high speed. Values are monitored and
stored by the Comprehensive Alert System (CAS) to identify whether the idle time of trucks dur-
ing haulage or the time of completing production tasks are above or below a predeined value. he
system shows a warning message when the task time of one of several trucks is out of limits. Values
are compared with pre-established KPIs for determining corrective actions and achieve production
targets. A time analysis table includes a dashboard that displays the status of individual trucks or the
whole leet for a quick recognition of any delay during haulage.
he system layout (Figure 2) presents the sequence for the detection of substandard haulage
cycles, generating a cycle time database, performing a measurement of the current haulage situation
within the As-Is model. It also deines the parameters to be applied in the To-Be model, generat-
ing a cycle time database with the corrections, verifying whether there was a haulage reduction of
cycle time, increase of production, and/or measured change of productivity. he main objective is
to assist supervisors and managers in identifying delays in haulage and improve the cycle times. A
secondary objective is assisting mine safety by controlling truck speeds to avoid accidents due to the
inherent pressure for achieving production quotas.
A virtual layout, consisting on a simpliied representation of the mine, was designed using
mine modeling software based on data extracted from an actual exploration project (Figure 3). he
results of the simulated mining process were stored in a database, for further evaluation of timings
that could be out of the expected standards. here are several controllable and uncontrollable con-
ditions that can alter the cycle time, such as Low/High Truck Speed, Low Performance Operator/
driver, Haulage equipment selection, Haul road condition, and Rolling Resistance. Some factors
in the optimization of the cycle time such as weather, visibility during night shift, power loss per
altitude and extreme temperature, and labor strikes cannot be controlled. hese may potentially
afect the productivity of haulage in the mine.
he mining process was simulated again with the corrected parameters. New cycle times were
stored in another database along with the simulated haulage cycle and compared with KPIs to
ensure process improvement. he time database created during cycle deines the performance stan-
dards of every equipment unit. he layout includes roads that connect the mill, the crusher and
38 Mine Discrete-System Simulation I

Figure 2. Overview of the Comprehensive Alert System

Figure 3. Mine layout with infrastructure distribution

the dump site. Haulage proiles were designed for in-pit and ex-pit locations. Table 1 describes the
equipment requirements for the mine in the diferent phases.
he initial scenario establishes the starting point for measuring operation conditions; cycle
times were deined by assigning the shovels and trucks to estimate the typical production rates and
performance capabilities by component. A second scenario is set during the peak phase of the mine
life where the pit has been deepened and the road distance has been extended, adding complexity
 Optimization of Open Pit Haulage Cycle Using a KPI Controlling Alert System 39

Table 1. Mine equipment requirements

Equipment Start up Average Peak
Shovels 3 4 5
Haul Trucks 15 24 31
Blasthole Drill 3 4 5
Rubber Tire Dozer 4 5 6
Grader 2 3 3
Water Truck 3 3 3
Support Loader 2 2 2
Light Vehicles 3 4 5

Table 2. KPIs for truck leet performance

Speed Full Haul (KPH): Speed of the load truck,
Speed Empty (KPH): Speed of the empty truck,
Production/Truck (MTPH): Truck production by hour,
Fixed Time: Sum of load, dump and spot time,
Production (MTPH): Production of material per hour,
Ore (MTPD): Amount of ore transported,
Crusher (MTPD): Amount of waste transported,
Utilization: Proportion of working time for equipment, and
Queues: Number of trucks is queuing in a shovel or a discharge point.

to the layout. he number of units changes according to the requirement of new distances for trans-
portation routes of ore and waste. Machines have more breakdowns, spending more time in main-
tenance, adding complexity to the system with the decrease of performance due to incremental time
in spotting, loading, dumping, and transporting. his increase has been deined as 10% over delays.
Assumptions include not considering standby times, using always the same type and model
of trucks and shovels, only one road for hauling or another for waste, with variable length accord-
ing to the depth of the pit, keeping the stripping ratio unchanged along the year, no passing of
trucks allowed, and permanent availability of trucks and shovels. he simulation was performed
for a period of 16 hours representing two shifts per day. All equipment units have a productive and
unproductive time during their operating cycle. he total cycle for truck transportation is deter-
mined by the following equations:
Cycle Time = Productive Time + Unproductive Time

Productive Time = Wait at load + Spot time + Load time + Full haul + Wait at dump +
Dump time + Empty haul

Unproductive Time = Queue time shovel + Queue time dump + Delays

he total cycle time by truck for the As-Is model provides a general overview of the initial
operational conditions in the As-Is model. Opportunities for the application of knowledge were
identiied, based on previous experiences and the supervisor’s criteria. he comparative variation
of an individual truck versus the leet was also identiied. he total sum of spot, load, and dump
times represent the Fixed Time. For this experiment, the Fixed Time was 5 min in average. Table 2
presents the KPIs selected for analyzing leet performance.
40 Mine Discrete-System Simulation I

Figure 4. Dashboard system for the To-Be model

he To-Be model is obtained by simulating the sequence of stages and displaying the partial
values during the process. he information collected was used to create the KPIs used for analysis.
Based on the results, process structures were modiied, and treated as input for the execution of the
inal model. For the validation, the results were compared with the original model.
Figure 4 shows the results of running the To-Be model after applying the settings recom-
mended by the Alert System. Observed times relect an improvement after the implementation of
the system.

DATA ANALYSIS AND RESULTS

Outcomes of production, cycle time, equipment utilization and queues provided the opportunities
to improve the general performance of the model. he analysis was performed in the startup and
peak operations of the mine, representing the existing (current), and the optimized (future) statuses
respectively, this is after the improvements and validations were made efective. Both As-Is and
To-Be models were processed using 100 replications to obtain a margin of error below 5%. he
utilization of KPIs provided a better perspective of the operating status of the mine. For the Startup
Operations, the actual production was less that the mine plan, more than 3 trucks were waiting to
be loaded per Waste Shovel 1 and queues are generated in the crusher when discharging the material.
he following step was inding the solutions based on the experience and predeined situations.
An analysis was made to ind the root cause, and the solutions found were adapted to the reality
of the operations. he remedial actions included Improvement of the “Spot at Loading” Time, by
retraining drivers every time the high time was over the expected standards; Improvement of the
 Optimization of Open Pit Haulage Cycle Using a KPI Controlling Alert System 41

Figure 5. Cycle times in ore Shovel 1—Startup operations—As-Is vs. To-Be models

“Waiting at Dump” Time, by suggesting the building of a second R-O-M bin; the “Maximum
Number of Queues “ had to be established according to speciic operating conditions.
Figure 5 shows comparisons of the diferent cycle times between As-Is and To-Be models where
there is an improvement in reducing Unproductive Time, Productive Time, and Total Cycle Time.
After completing the above corrections, several alternative scenarios are proposed in order to
see the behavior of production, utilization, and queues varying the number of trucks from 12 to 20
and 02 ROM bins (feeders). In addition, the relationship of material mined with truck utilization
and material mined with number of queues at crusher was examined in a sensitivity analysis. While
the number of trucks was increased, production plan target is not reached, decreasing the truck
utilization and increasing use of shovels.
he generation of the To-Be model allows to reach the production target from 13 trucks, and
the utilization decreases less than in the As-Is model. With the previous sensitivity analyses, the best
scenario is to build a second ROM bin with 14 trucks to meet the production goals with better
equipment utilization. he peak of the mine operation is deined in period 22. he routes for the
crusher and waste zone were maintained, increasing its distance only by the deepening of the mine.
After running the To-Be model, an improvement was obtained in reducing unproductive time
in the haul cycle. In order to make a relationship between the variables involved in the model,
additional simulations were made, inding an inverse relationship between truck utilization and
the tonnage of material moved. While increasing the number of trucks, truck utilization decreases;
however, a direct relationship was found between the number of trucks, shovel utilization, and ton-
nage of material moved. In the To-Be model, the inverse relationship between the number of trucks
and their utilization is preserved; however, this utilization increases in all scenarios, which indicates
a better use of vehicles for transporting material, as shown in Figure 6. It can be found that the Peak
Operations case can work optimally with 31 trucks (Figure 7).

CONCLUSIONS, RECOMMENDATIONS, AND FUTURE WORK

After having performed the simulated haulage cycle in Startup and Peak Operations, performance
problems were identiied and analyzed, leaving ground for a general diagnosis approach, providing
42 Mine Discrete-System Simulation I

Figure 6. Cycle times in ore Shovel 1—Peak operations—As-Is vs. To-Be models

250,000 95%

Mining Plan 90%

200,000 85%

Ulizaon
80%
150,000
Tonnes

75%
70%
100,000
65%

50,000 60%
55%
0 50%
27 28 29 30 31 32 33 34 35
Waste (MTPD) 116,280 119,849 122,650 125,570
0 127,730 130,581
1 133,366 134,905 137,360
Ore (MTPD) 78,592 80,122 82,494 84,456 86,792 88,052 88,754 91,159 91,962
Ul Truck 87.4% 86.8% 86.2% 85.6% 84.8% 84.1% 83.1% 82.3% 81.4%
Ul Shovel 1 Ore 70.9% 72.3% 74.5% 76.1% 78.4% 79.4% 79.8% 82.1% 82.6%
Ul Shovel 2 Ore 70.6% 71.9% 73.9% 75.9% 77.9% 79.1% 79.9% 82.0% 82.9%
Ul Shovel 1 Waste 77.6% 79.8% 81.9% 83.3% 84.8% 86.3% 88.0% 88.7% 90.1%
Ul Shovel 2 Waste 68.7% 71.7% 73.8% 76.7% 78.4% 81.1% 83.5% 84.7% 87.0%
Ul Shovel 3 Waste 78.6% 80.3% 81.7% 82.9% 83.9% 85.2% 86.7% 87.6% 88.7%

Figure 7. Sensitivity analysis of Tonnes vs. Utilization—Peak operations—To-Be model

the recommendations for the best development of a mine haulage system by reducing unproductive
time and increasing equipment utilization. he conclusions are as follows:
• Simulation tools help to represent actual systems in process, contributing with the develop-
ment of a methodology to improve productivity, reduce unproductive time, increase truck
and shovel utilization, reduction of queues in shovels and crusher by applying simulation
techniques, without the need of incurring in any additional investment. he CAS method-
ology was developed to ind out whether it is feasible to take advantage of the simulation
tools, thus optimizing the production process. Startup parameters could be modiied at any
 Optimization of Open Pit Haulage Cycle Using a KPI Controlling Alert System 43

Figure 8. Expected outcomes in the To-Be model

time until itting the ultimate optimization. In this way costs, resources, and time could be
saved.
• he As-Is model allowed the representation of a mine operating at a particular time, and the
To-Be model showed how would the operations look after certain adjustments. Values gen-
erated with the As-Is model and subsequently adjusted in the To-Be model can be utilized
as benchmarks for other mines with their own individual diferences such as production,
geology, or haulage equipment. It was observed that in Startup and Peak Operations after
the simulation and analysis, the production increases.
• Figure 8 presents improvements implemented to the As-Is model for becoming the To-Be
model. It shows the KPIs that were analyzed and settings to be applied in the original model,
showing an improvement in the expected results.
he recommendations arising from this methodology are:
• he collection of data is essential part for representing a model; it must be performed accu-
rately so that the model is quite similar to the real system. Additionally, as much data as
possible must be gathered for increasing reliable data.
• he CAS data were generated from an As-Is model and adjusted for the To-Be model. Tests
were performed only during an operating day of the mine (two shifts). However, simula-
tions should be performed for longer periods such as a week or a month, as the variations of
conditions may occur with more frequency.
• he current methodology only considers the haulage cycle times. Fuel consumption and
operating cost could be included to observe its efect on the sensibility analysis.
Future work considers an enhanced display superimposed over an Open Pit layout. Realistic graph-
ics and an added perspective would be a powerful aid to help people to visualize and understand the
mining process. With the introduction of random functions into the simulation model, it would
be possible to represent more realistic scenarios and strengthen the CAS methodology by deining
actions to resolve unexpected outcomes. he system can be used in mining operations which an
44 Mine Discrete-System Simulation I

FMS system and an Enterprise Resource Planning (ERP). Using the methodology as a tool of long
term planning for actual mines may help improving the annual budget, thus impacting in the cost
and calculation of reserves.

REFERENCES
Ataeepour, N., & Baai, E.Y. (1999). ARENA simulation model for truck-shovel operation in dispatching
and non-dispatching modes. International Journal of Surface Mining, Reclamation and Environment, 13,
3, 125–129.
Cox, R.F., Issa, R.R., & Ahrens, D. (2003). Management’s perception of key performance indicators for con-
struction. Journal of construction engineering and management, 129(2), 142–151.
Ercelebi, S.G., & Bascetin, A. (2009). Optimization of shovel-truck system for surface mining. Journal of the
South African Institute of Mining and Metallurgy, 109(7), 433–439.
Krause, A.J. (2006). Shovel-truck cycle simulation methods in surface mining (Doctoral dissertation).
Kelton, W.D., Sadowski, R.P., & Sadowski, D.A. (2007). Simulation with Arena. Boston, Mass: WCB/
McGraw-Hill.
Lineberry, G.T. (December 01, 1985). Optimizing of-highway truck characteristics for minimum haulage
cost. International Journal of Mining Engineering, 3, 4, 295–310.
Lodhi, A., Kassem, G., Köppen, V., & Saake, G. (2010, December). Building AS-IS process models from
task descriptions. In Proceedings of the 8th International Conference on Frontiers of Information Technology
(p. 14). ACM.
Pop-Andonov, G., Mirakivski, D., & Despotov, Z. (2012). Simulation Modeling and Analyzing in
Underground Haulage Sistems with Arena Simulation Software. International Journal for Science, Technics
and Innovations for the Industry MTM (Machines, Tecnologies, Materials), 48–50.
Sgurev, V., Vassilev, V., Dokev, N., Genova, K., Drangajov, S., Korsemov, C., & Atanassov, A. (1989).
TRASY—An automated system for real-time control of the industrial truck haulage in open-pit mines.
European Journal of Operational Research, 43(1), 44–52.
Schwegmann, A., & Laske, M. (2003). As-Is modeling and process analysis. Process Management (pp. 107–
133). Springer Berlin Heidelberg.
Tan, Y., Miwa, K., Chinbat, U., Takakuwa, S., & 2012 Winter Simulation Conference, WSC 2012.
(December 01, 2012). Operations modeling and analysis of open pit copper mining using GPS tracking
data. Proceedings—Winter Simulation Conference.