Abstract
Discrete event simulation is a powerful tool for maximizing throughputs of existing systems and designing new ones. The paper describes how the technology has been applied in iron ore handling in Brazil, Australia, and Canada to:
- test the outcomes and benefits of changing operational practices in ore handling, rail and port systems to achieve increases in throughput with minimal capital investment.
- demonstrate the incremental and aggregate benefits of alternatives for system improvements
- assess design alternatives for new terminals to minimize investment.
Figure 1 Schematic of a Typical Coal or Iron-Ore Supply Chain
Bulk Material Supply Chains
Bulk material supply chains comprise all activities involved in taking material from its source to the end customer. Bulk materials include coal and iron-ore as well as other materials such as grains, cement, and dry chemicals. Typically, many steps are involved. The material may be processed one or more times, may be transported through various means (e.g. truck, rail, pipeline, ship), and may have several intermediate storage locations. (Select image for enlarged view of schematic representing a typical coal or iron-ore chain)
Large up-front investments are required in these supply chains and almost none of the investment can be recovered after the end of the project lifespan. Bulk materials are also typically commodity-priced and subject to variations in open-market pricing. This can cause swings in project viability.
Recent global trends have caused a surge in demand for coal and iron-ore, resulting in increased prices and expansions in many facilities around the world to increase supply.
As in any business, maintaining low costs along with high efficiency and quality is a prime goal. Traditionally, “rules of thumb” and basic calculations have been made in order to make key decisions such as where to build a facility and when to invest in new infrastructure. A relatively recent tool that can aid in these decisions is simulation. Advances in computing technology have allowed more complex and longer-duration simulations to become practical.
Discrete-Event Simulation
In a discrete-event simulation model, a representation of an operating system is created and modeled through time.
Objects represent equipment and processes in the system, such as trains, conveyors, and stockpiles. Objects have parameters that specify their qualities. A train, for instance, would have parameters such as cargo capacity, traveling speed, and length. Logical rules govern the interactions between the objects. For examples, trains must travel along rail signal blocks and are scheduled such that they do not collide with other trains in the system.
External events such as weather and production variation (due to ore hardness, for example) can be included in the model. Random events, such as equipment breakdowns and variation in activity time, can also be included. Simulating external and random events over a long period of simulated time allows a detailed and quantitative understanding of the expected operation of the system to be gained.
Case Studies
The following case studies are intended to show the variety of applications simulation can be used for and the types of decisions it can aid in.
Figure 2 – Coal Terminal Layout – Actual and Modeled
Case Study 1 – Port Capacity Analysis
The focus of this model was the marine terminal, so the mine and rail operations included only minimal detail, such as the average product rate, number of trains, average train capacity, and the average travel time to an from the terminal. The terminal itself was modeled in considerable detail. Production flows from the mines for each grade of coal to the mine stockpiles, where it is taken via train to the marine terminal. Train dumpers unload the trains and conveyors feed to either the stockyard (for stacking into piles) or to the berths (for direct loading onto ships).
In the model, the movements of trains and ships in the model are coordinated in the following steps:
- The main driver for the model’s logistics is mine production. Annual production for each mine and grade are entered as inputs to the model.
- Ships are chartered with stem dates (i.e. scheduled arrival dates) that match the availability of coal for each parcel. The size and number of parcels for each ship are selected appropriately from the historical distributions for each grade.
- The actual arrival time for each ship is selected randomly in a window around its scheduled time based on historical ship arrival data for a similar bulk terminal.
- Coal is moved by rail from the mines to the terminal on a just-in-time basis, to match ship arrival times over the next ten days. Only the amounts needed for the ships’ cargoes (plus any extra amount needed to make up full train loads) are railed to the terminal. If sufficient trains and stockyard space are available, the model accumulates the full amount for each of the ship’s parcels two days ahead of its arrival time at the terminal.
The operation was modeled at increasing levels of throughput to determine the port capacity. Capacity was determined by the utilization of the inloading and outloading components of the port. When utilization is reaches an impractical level, queuing of trains and/or ships will occur, resulting in high demurrage costs and, ultimately, a shortfall from the target throughput level.
Stepped infrastructure cases can be modeled to determine the optimal path forward to meet future throughput targets.
Figure 3 – Graphical Representation of Model of Complex Iron-Ore Export Port
Case Study 2 – Port Capacity Analysis 2
This port is shown to demonstrate that increased complexity can be modeled with relative ease. The port shown below includes four dumpers, seven pelletization and screening plants, and recirculation of product through several stockyard areas.
The study objective was to aid the owner in infrastructure requirements as exports increase from 85 Mt/y to well over 100 Mt/y.
Figure 4 – Graphical Representation
of Model of Rail Line
Case Study 3 – Rolling Stock Requirements versus Infrastructure
A 15 Mt/y iron-ore exporter faced challenges due to weekly variations in its production levels. Insufficient storage at its mine resulted in product being regularly sent to ground (resulting in loss of quality and additional handling costs) while at the same time its train fleet had excessive idle time during the low-production periods.
A model of the “mine to port” system was created, including the shared-use rail-line. The 260-mile single-track line with 26 sidings transported iron-ore from two mines as well as passenger and cargo trains.
Several scenarios were modeled at various throughput levels to determine trade-offs and interactions between investment in increased storage at the mine and increased rolling stock.
Capital and operating costs were incorporated in the analysis to determine the optimal solution.
Case Study 4 – Berth Capacity
In the iron ore boom of the late 1990’s iron ore ports around the world were stretched to their capacity in the same fashion as they are today. Simulation modelling showed that a two-berth export terminal in Western Australia had a capacity of 53 Mt/y. The terminal operator’s view was the port capacity could be defined as:
8760 hours per year x average cargo per ship (tonnes)/ ship service time (hours)
The result was a nominal port capacity of 58 Mt/y. The iron ore salesmen believed the operators and sold 58 million tonnes of iron ore; the resulting berth utilization exceeded 99%. The profit from the sale of the incremental 5 million tonnes of iron ore paid for the shipping demurrage bill.
A simulation model was adopted as a good measure of capacity and used as a tool to demonstrate the capacity gain by reducing ore processing delays (a means of justifying postponed maintenance) and ranking incremental expansion opportunities.
Figure 5 – Graphical Representation of a Beneficiation Plant
Case Study 5 – Process Bottleneck Analysis
Generic processes can also be analysed using simulation. An example is a beneficiation plant used by an iron-ore exporter.
The owner was planning to expand throughput from 80 Mt/y to over 100 Mt/y. The beneficiation process included several steps, with the bottleneck being the secondary crushing. At increasing throughputs, buffer storage and processing rate requirements throughout the plant were determined.
Figure 6 – Ore Delivery System
Case Study 6 – Ore Delivery System with Complex Interactions
An ore delivery system (ODS) was analysed for a 40 Mt/y raw iron-ore operation.
The ODS was an integral component of a larger system comprised of a) the upstream mining (truck and shovel) operation, b) the rail ore delivery system from the mining operations to the processing facility, c) the processing facility creating concentrate and pellets and d) the downstream transportation of the concentrate and pellets to tidewater.
The key components of the ODS were the rail car loading operation, the rail transportation, the ore unloading operation at the crusher, the crushed ore stockpile, and the milling operation. The simulation study was able to determine the critical bottleneck for the system, which was dependent upon the ore characteristics (soft or hard) and the freezing conditions in the winter.
Simulation modeling of the ODS system provided the insight into the interdependency of the component pieces and provided the owner with an understanding of the key bottlenecks to eliminate in order to add capacity to the system.
Figure 7 – Graphical Representation of Modeled Rail-Yard
Case Study 7 – Rail-yard Operations
Rail capacity is often limited by complex operations at the rail-yard. The rail yard shown below included the movements of trains as they entered the yard, split into two sub-units, and proceeded to one of three dumpers. Through the simulation, railyard bottlenecks and capacity could be found.
Figure 8 – Graphical Representation
Case Study 8 – Operational Synergy
After an acquisition, two iron-ore export facilities were owned by the same parent company. A simulation study was performed to determine the synergies that could be obtained by operating the two ports as a single unit.
