Is Truck Queuing Productive?Study of truck & shovel operations productivity using simulation platform MineDES
Dmitry KostyukSpecialist Scientist, Group Resource and Business Optimisation25 November 2014
Contents
Dmitry Kostyuk, Specialist Scientist, Group Resource and Business Optimisation, 25 November 2014 Slide 2
Simulation Modeling in Mine Planning
Truck and shovel operations simulation platform MineDES
• Overview
• Advantages
• Benchmarking and validation
Case Study
Conclusions
Mine Planning Challenges
Dmitry Kostyuk, Specialist Scientist, Group Resource and Business Optimisation, 25 November 2014 Slide 3
• Mines will continue to increase in depth, scale and complexity.
• There is a strong demand for mine planners to:
–Produce achievable, optimized plans for these mines–Appropriately size equipment fleets–Design efficient, effective mine access systems–Accurately estimate mining system productivity
• Getting it wrong can negatively impact project NPV.
Conventional Mine Planning
Dmitry Kostyuk, Specialist Scientist, Group Resource and Business Optimisation, 25 November 2014 Slide 4
• Analytical methods – formula-based approach
• High-level of abstraction – BIG PICTURE point of view
• Normally doesn’t account for the impact on overall mining system productivity such factors as equipment interactions, parameters variability, randomness, uncertainty etc…
• Fails to describe systems with dynamic behavior featuring:
– non-linear behavior– non-intuitive influences between variables– time and causal dependencies– uncertainty, randomness and large number of parameters
Mine Planning using Simulation Modeling
Dmitry Kostyuk, Specialist Scientist, Group Resource and Business Optimisation, 25 November 2014 Slide 5
Simulation Modeling• Method of solving problems that
can’t be calculated analytically• Cheap and risk free experiments
(“what if ?” studies)• Efficient for analyzing systems with
dynamic behavior
Analytical Modeling vs. Simulation Modeling
Dmitry Kostyuk, Specialist Scientist, Group Resource and Business Optimisation, 25 November 2014 Slide 6
Arrivals: on average trucks / hour
Loading time exponentially distributed:
1/ – mean loading time
Shovel utilization:
Average waiting time: Average queue length:
Loading time arbitrary distributed:
1/ – mean loading time
Shovel utilization:
Average waiting time: ,where is coefficient of variation of loading time.
Single loaderPoisson stream (independent arrivals)
Case #1: Case #2:
Queuing theory
Analytical Modeling vs. Simulation Modeling
Dmitry Kostyuk, Specialist Scientist, Group Resource and Business Optimisation, 25 November 2014 Slide 7
Arrivals: on average trucks / hour
Multiple (K) loaders
Loading time exponentially distributed:
1/ – mean loading timeShovel utilization:
Average waiting time: ,
where P! ,
and !∑
!
Loading time arbitrary distributed:
1/ – mean loading time
Poisson stream (independent arrivals)
ANALYTICAL SOLUTION DOES NOT EXIST STARTING FROM HERE AND FOR ANY FURTHER COMPLICATION OF THE PROCESS!!!
Case #3: Case #4:
Queuing theory
Advantages of Simulation Modeling
Dmitry Kostyuk, Specialist Scientist, Group Resource and Business Optimisation, 25 November 2014 Slide 8
• Enabling system analysis, and to find solutions where other methods fail
• Once appropriate level of abstraction is selected, development of a simulation model is a more straightforward process than analytical modeling – less intellectual efforts, scalable, incremental and modular
• The structure of a simulation model naturally reflects the structure of the real system – it is visual, easy to verify and communicate to other people
• Any state of the model is measurable and any entity, which is not below abstraction level is tractable – sensitivity analysis, statistical analysis
• Ability to play and animate the system
• Simulation models are a lot more convincing than Excel spreadsheets or Power Point slides
Contents
Dmitry Kostyuk, Specialist Scientist, Group Resource and Business Optimisation, 25 November 2014 Slide 9
Simulation Modeling in Mine Planning
Truck and shovel operations simulation platform MineDES
• Overview
• Advantages
• Benchmarking and validation
Case Study
Conclusions
MineDES - What is it?
Dmitry Kostyuk, Specialist Scientist, Group Resource and Business Optimisation, 25 November 2014 Slide 10
• Truck & shovel operations simulation tool that can be used to estimate mining movement and processing capability in the following dimensions:
– productivity statistics– bottleneck processes and infrastructure– truck queuing statistics– the impact of different crew and maintenance schedules– the impact of unscheduled random equipment and road sector downtime – the influence of road maintenance vehicles and light vehicles on congestion – the capacity of particular pit ramps, and the whole road network, to support planned material
movements.
• Primary design concept was to focus application on addressing strategic mine planning questions, but secondly to be applicable in the short term planning environment.
Dmitry Kostyuk, Specialist Scientist, Group Resource and Business Optimisation, 25 November 2014 Slide 11
MineDES Features
Dmitry Kostyuk, Specialist Scientist, Group Resource and Business Optimisation, 25 November 2014 Slide 12
• Truck dynamics are calculated using full rim-pull curve data, taking into account truck payload as well as road gradient and quality
• A fast, purpose-built simulation engine, and the modelling of trucks as agents to ensure realistic and intelligent behavior
• Truck dispatch is predicated upon attempting to achieve a user-defined mining rate at each mining face
• Realistic and flexible modelling of traffic rules at complex intersections
• Modelling of payload and loading/dumping time variability
• Flexibility in assigning legal digging and dumping destinations to different truck sets within each truck fleet
• The optional application of a wide range of scheduled and unscheduled downtime for all mobile and static material processing infrastructure.
• Integrated 3D visualization engine
Advantages of truck & shovel simulator MineDES
Dmitry Kostyuk, Specialist Scientist, Group Resource and Business Optimisation, 25 November 2014 Slide 13
• In-house software development project with more than 4 years development history.
• Developed from scratch and doesn’t use any commercial simulation engines.
• Designed to address both long- and short-term mine planning problems
• Intuitive, flexible and fit for purpose. Development of a simulation scenario is straightforward process, which does not require programming skills and a lot of intellectual effort.
• Benchmarked against other industry standard software products and tested in real operations environment
MineDES Benchmarking and Validation
Dmitry Kostyuk, Specialist Scientist, Group Resource and Business Optimisation, 25 November 2014 Slide 14
600 m
0
5
10
15
20
25
30
35
40
45
50
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
MineDES
TALPAC
Truck dynamics model benchmarking against industry standard software
Truc
k sp
eed
(km
/h)
Distance (km)
Testing against real truck data.
Dmitry Kostyuk, Specialist Scientist, Group Resource and Business Optimisation, 25 November 2014 Slide 15
113.6
113.8
114
114.2
114.4
114.6
114.8
115
115.2
115.4
17 17.2 17.4 17.6 17.8 18 18.2
Y (k
m)
X (km)
GPS real truck data plot:
05
10152025303540
0 5 10 15 20
00.5
11.5
22.5
33.5
44.5
0 5 10 15 20
Truck ground speed profile. GPS vs. MineDES
Truck travel distance profile. GPS vs. MineDES
MineDES model - design view: Experiment results:
Truc
k sp
eed
(km
/h)
Dis
tanc
e (k
m)
Travel time (min)
Contents
Dmitry Kostyuk, Specialist Scientist, Group Resource and Business Optimisation, 25 November 2014 Slide 16
Simulation Modeling in Mine Planning
Truck and shovel operations simulation platform MineDES
• Overview
• Advantages
• Benchmarking and validation
Case Study
Conclusions
Case Study. Optimal fleet size.
Dmitry Kostyuk, Specialist Scientist, Group Resource and Business Optimisation, 25 November 2014 Slide 17
Truck Loading Time Distribution Graph [sec]:
Truck Payload Distribution Graph [tonnes]:
Case Study. Optimal fleet size.
Dmitry Kostyuk, Specialist Scientist, Group Resource and Business Optimisation, 25 November 2014 Slide 18
• A minimum 8 trucks are required to maximize productivity.
• Adding more trucks to the fleet simply increases overall queuing time with no additional aggregate material movement - in fact, adding extra trucks above 8 can lead to an insignificant decrease in productivity (< 1%) due to increased traffic congestion.
Case Study. Optimal fleet size
Dmitry Kostyuk, Specialist Scientist, Group Resource and Business Optimisation, 25 November 2014 Slide 19
The simulation experiment has shown that even in a simple case where non-deterministic behaviour is quite limited and the road network is simple (point-to-point), in an optimized configuration, we should expect to see trucks queuing at shovels to a not insubstantial extent. The simple financial model shows that this queuing is protective of productivity and operational value.
Conclusions
Dmitry Kostyuk, Specialist Scientist, Group Resource and Business Optimisation, 25 November 2014 Slide 20
• A new truck-shovel simulation tool called MineDES has been introduced
• A simulation case study was undertaken using MineDES to address the question of whether having queuing trucks could be a feature of an optimally productive mining operation.
• Our experiments have shown that in the case where non-deterministic factors are present in the system, typical of all real mining operations to a greater or lesser extent, then some degree of truck queuing will be observed in the most productive of configurations.