System Models And Simulation

Published Date: 02 Nov 2017

Simulation has now become one of the most widespread modelling approaches in operations research, management science, and engineering (Rubistein & Melamed, 1998). According to a survey undertaken by UK Ministry of Trade and Industry, the simulation modelling is applied by the 500 largest corporations in the United States to improve all aspects of management. Moreover, there is evidence shows that simulation has helped companies to reduce the capital costs by 5% to 10%. Hence, simulation is a powerful tool worth to study deeply.

In this chapter, some concepts related to simulation technology are described firstly, and then the reasons of applying simulation technology are explored. Following that, the advantages and disadvantages of using simulation technology are discussed. Lastly, the classification of simulation models is identified.

1.1 System, Models and Simulation

Before going to identify the definition of simulation, two concepts need to be explored firstly, they are system and models. According to Oakshott (1997), the term system can be defined as a collection of interacting components or processes. For example, a university can be considered as a system, with students, academic staff, and administrative staff as components. In addition, systems can be classified into four main categories: nature systems, designed systems, designed abstract systems and human activity systems (Checkland, 1999).

According to Rubinstein and Melamed (1998), a model can be defined as an abstraction of the real system. It is used to represent the real system to help analysts understand the operations of the system and evaluate the system performance. In particular, a model can be used to figure out potential influence of changes in the various aspects of the real system. A useful model should be similar to the real system, more precisely, should include all essential characteristics of the modelled system. However, the model must avoids being overly complex as one of the main purposes of using a model is to simplify the real system.

Want our help with your essays

If you need any help with your essays and dissertations.
Our premium essay writing services are here to help you!

Get Help with Your Essay

Having defined the concepts of system and models, the term simulation can be explained. Pegden, Shannon and Sadowski (1995) define simulation as follows:

Simulation is the process of designing a model of a real system and conducting experiments with this model for the purpose of understanding the behaviour of the system and/or evaluating various strategies for the operation of the system.

Based on the definition above, the term simulation means the presentation of a real system by using modelling methodology, it helps people to understand the operation of the system and evaluate the performance of the system under different configurations.

In more general sense, simulation can be defined an imitation of a system (Robinson, 2004). In other words, simulation refers to the process of copying actions of a system. For example, the simulation model in the weather forecast is used to imitate the movement of the weather system.

1.2 The need for simulation

In the real world, most operations systems are difficult to analyse and evaluate due to the variability, interconnectedness and complexity of the nature of the system (Robinson, 2004).

The variability refers to the variations exist in the system, these variations might be predictable, for example, the estimated number of passengers of a bus station in the normal working days. These variations might be unpredictable, for example, the arrival rate of customers of a shopping centre is changing all the time.

The interconnectedness means the correlated relationships between components of the system. As the definition above, system is a collection of independent or interacting components, and thus the work of one component might have influence on the other components within the same system. For instance, in a flow line production system, the breakdown of a single machine will lead to the breakdown of the entire system.

Lastly, many systems are complex and this complexity can be distinguished into two main categorises: combinational complexity and dynamic complexity (Robison, 2004). The number of components in the system and the number of the possible combinations of these components are the two main aspects related to the combinational complexity. As for dynamic complexity, it is always related to the interaction of the components in the system (Sterman, 2000).

Therefore, it is very difficult to evaluate the performance of a system directly due to the nature of the system described above. However, simulation provides an adequate approach to meet this challenge. Simulation models can be used to predict system performance, compare different design strategies and investigate the impact of changes in some components of the system.

1.3 Advantages of simulation

Besides simulation modelling, there are many other approaches can be employed to analyse the real system, such as experiment and other modelling techniques. However, simulation has a number of advantages compare with other methods.

Cost. Simulation modelling approach can be used to test new ideas and alternative strategies on the existing systems without interrupting the operations of the system, while the experiment method always requires shutting the system down in order to try the new strategies. In addition, according to a survey undertaken by Harrell, Ghosh & Bowden (2000), the overall cost of constructing a system by using simulation technology is less than cost without using simulation, though the use of simulation in the design phase costs more money.

Figure 1: Comparison of cumulative system costs with and without simulation (Harrell, Ghosh & Bowden, 2000).

Time. It is obvious that experiment method is a time consuming way to analyse the system. It always needs a long period before the results of experiments can be obtained. However, by using the simulation model, the system can be run much faster than the real time. As a result, the evaluation of the system performance can be obtained in often only a few of minutes or hours.

Applicable to complex systems. Simulation can help to study complex systems while some other approaches might not be applicable in these cases. For instance, some characteristics and behaviours of complex systems are extremely difficult to express by using mathematical models.

1.4 Disadvantages of simulation

Although simulation has many advantages, there are some problems with using this approach and these problems should not be ignores.

Expensive. The expense of developing a simulation model could be considerable. These costs may include of buying particular simulation software, employing programmers and collecting data.

Require data. A significant amount of data is always required in order to build a simulation. And the collection of these data might be very difficult and time consuming.

Compare with other exact mathematical approach, simulation is not a useful approach to find an optimal solution.

1.5 Types of simulation

The classification of system models can be illustrated by a tree diagram, as shown in the figure 2. It can be seen that simulation models can be classified based on three different dimensions: Static or Dynamic, Deterministic or Stochastic and Continuous or Discrete (Law, 2007). Details of different types of simulation models are illustrated as follow:

Figure 2: Model Taxonomy (In Tech-discrete event)

Static versus Dynamic Simulation models

A static model represents a system at a particular time point. In addition, the passage of time plays no role in this system. A typical static simulation model is the Monte Carlo Simulation which is defined as a scheme uses random numbers to solve certain stochastic and deterministic problems (Law, 2007).

A dynamic simulation model is a representation of a system which evolves over time. For instance, a simulation model of a computer-numerically-controlled (CNC) router for a 40-hour work week (Law and Kelton, 1991)

Deterministic versus Stochastic Simulation models

A deterministic simulation model does not contain any random variable. When a deterministic model is run with the same input value several times, the outputs of the model are the same.

A stochastic model contains random or unpredictable components. Most queuing system and inventory systems are simulated by stochastic models. The output of a stochastic model is uncertain and random, and thus these output data just provide estimates of the true characteristics of the model.

Continuous versus Discrete Simulation models

A continuous simulation model represents a continuous system in which the state variables change continuously with respect to time, for example, the cooling of hot water or the flow of water through a river.

A discrete simulation model represents a system which changes the state variables instantaneously at separate time points. These time points are the time events occur which will lead to the change of the state of the system. For instance, most queuing systems and manufacturing systems belong to this type.

The focus of this paper is in the discrete-event simulation models. More details about discrete-event simulation will be described in the next chapter.

1.6 When to simulate?

Simulation technology is one of the most popular tool applied in operational research, management science and engineering. However, not all problems can be addressed efficiently by using simulation technology. It is generally known that applying inappropriate methodologies for the problems will lead to inaccurate results and waste of resources and time. Therefore, the issue of determining under what circumstance simulation is the appropriate tool to use is really worth to analyse. Banks, Carson and Nelson (1999) state the following situations under which simulation technology is appropriate to be applied:

Simulate internal interactions of a complex system.

Simulate informational, organisational and environmental changes and evaluate the impact of these changes on the behaviour of the overall systems.

Improve system performance by analysing the knowledge and information obtained in designing a simulation model.

Observe interactions between variables and evaluate which variable is more important to the system.

Investigate feasibility of new strategies, policies or designs before implementation.

Verify analytical solutions.

In addition, Banks and Gibson (1997) claim 10 rules for determining when simulation technology is not an appropriate tool:

The problem can be addressed by using common sense analysis.

The problem can be solved analytically.

It is easier to perform experiments than build simulation models.

The expense of simulation is more than possible savings can be obtained.

There are sufficient and proper resources for the project.

Time is not enough.

Data collection is difficult.

The simulation model cannot be verified and validated.

The goals of the project cannot be met.

If the operations and behaviour of the system are too complex to be defined.

2. Discrete-event simulation

2.1 Introduction of discrete-event simulation

As the definition above, discrete-event simulation is a representation of the system which changes the state variables instantaneously at separate time points (Law, 2007). Discrete-event simulation is one of the most popular modelling techniques since the beginning of computer simulation in the 1950s (Robinson, 2005). Discrete-event simulation is a powerful technique as it helps people to understand the complex behaviours and interactions between individuals, groups and environments. Due to the advantages of discrete-event simulation, it is widely applied in a variety of fields in the real world.

2.2 Core concepts of discrete-event simulation

The core concepts of the discrete-event simulation are: entity, attributes, events, activities, list, and clock. Details of these concepts are explained respectively as follows (Banks, Carson & Nelson, 1999):

Entity. It refers to the components and objects in the system which need to represent explicitly in the simulation model (e.g., in a shopping centre’s queuing system, customers and servers are the typical entities).

Attributes. The priorities that specified to a given entity (e.g., in the healthcare system, the emergency patients usually have higher priorities than the routine patients).

Event. An instant of time at which the state of the system is changed (e.g., in a queuing system, an arrival of a new customer or a leave of a customer which will change the number of customers in the system).

Activity. It refers to the operation which will change the state of the entities (e.g., the service of a server, the production of a machine)

List. It refers to a collection of entities which are ordered by a given principle (e.g., in a queuing system, all customers are waiting in a line and ordered by first come, first served rule).

Clock. A variable used to represent the simulation time. The reason of creating this variable is the dynamic nature of the discrete-event system. More precisely, the states of the system, the number of entities, entity attributes, and the activities are all changing over time.

2.3 stages of a simulation project

There is not an agreed answer on what stages of a simulation project should have. This is due to the nature of the project itself, that is, different projects have different characteristics and factors. Therefore, particular stages might need to be performed in some particular projects. However, some key stages can be identified as they should be performed in any simulation project. There are many different diagrams and descriptions for the key phases of a simulation study (e.g., Shannon, 1975; Hoover and Perry, 1990; Law and Kelton, 1991), but they are quite similar. One of these diagrams is shown as the figure below (Banks et. al. 1996).

Figure 3: Steps in a simulation study

Details of each step in a simulation study are described as follows:

Problem formulation. Every simulation study should begin with problem formulation, in other words, identification of the problems. The problem should be described clearly and related issues should be pointed out.

Setting of objectives and overall project plan. The goals of the study should be determined and an analysis should be performed in order to determine whether simulation can achieve the goals or not. If simulation is decided to be employed, than an overall project plan should be constructed. It should include the following aspects: the number of people involved, time schedules from each steps, success criteria for each step, and the cost of the study.

Model conceptualization. A preliminary model needs to be constructed in this step. This model should contain essential features of the real system, such as the components, variables and interactions that constitute the real system.

Data collection. The collection of realistic input data should be done in this step. Since these input data is used to drive the simulation model and evaluate the performance of the model, it is important to collect accurate and realistic data from the real world. As a result, data collection is quite time consuming and it takes a large portion of the total time required to develop a simulation model (Banks, Carson & Nelson, 1999). Therefore, it is necessary to start this step as early as possible.

Model translation. Formulate the model in an appropriate simulation language in order to make it can be recognised by the computer.

Verified? Check whether the computer program operates the way as the analysts expect and confirm there is not any bug in the program.

Validated? Check whether the simulation model represents the real system accurately or not. Validation is usually performed by repeating the process of comparing the simulation model with the behaviour of the real system until the model is believed to be accurate.

Experimental design. Design an experiment which specified the length of each simulation run, the length of the warm-up period, and the number of replications.

Production runs and analysis. Run the simulation model, analyse the output of each run and evaluate the performance of the simulation model.

More runs? Based on the performances of the runs have been completed, decide whether additional runs are necessary or not.

Documentation and reporting. Two types of documentation need to be produced: program documentation and progress documentation. The program documentation is used to help other analysts to understand how the program operates, and also help to modify the program in future. The progress documentation is used to record the progress of developing the simulation model (e.g., starting time and finishing time of each step, related decisions made in each step). Besides two documentations above, a final report should be produced which includes the final formulation, alternative systems have been tried, the results of the experiment, and recommendation to the simulation model.

Implementation. If the simulation model suggests that it is beneficial to implement simulated system, then accomplish it in the real world. The success of this step is dependent on the performances of the last 11 steps.

The 12 steps described above are the general process of building a simulation model. Additionally, they can be classified into four phases (Banks, Carson & Nelson, 1999): the first phase is defined as discovery and orientation which includes steps1 and 2; the second phase is for model building and data collection which includes steps 3 to 7; the third phase is running the model which includes steps 8 to 10; the last phase is implementation which involves steps 11 and 12.

2.3 The discrete-event simulation approach

Although it is not necessary to know details about the inside execution of a simulation in order to develop a simulation model, basic understanding may help to improve the program. Modelling the progress of time is fundamental to every discrete-event simulation study. Since the special nature of the discrete-event system, the states only changed when the events occur. Thus, the discrete-event system is usually modelled as a series of events (Robinson, 2005). However, this is just the basic principle for the discrete-event system. In order to perform the discrete-event simulation more directly and efficiently, a number of methodologies have been developed, such as event-based, activity-based, process-based and three-phase approaches. Details of each mechanism are described respectively as follows:

Event-based executives:

In an event-based approach, the executive maintains an event calendar which contains references of each event routines. This event calendar helps executive recognise which event are due to occur newt. The event-based simulation is processed by two phases, as shown in figure 4:

Check the event calendar and find the due time of the next event. Then move the simulation clock to this time and schedule a current events list based on this new simulation clock.

Holding the simulation clock constant, execute event routines which are in the current event list.

These two phases are run iteratively until the simulation is finished.

Figure 4: An event-based executive (Pidd, 1998)

Activity-based executives:

In an activity-based approach, the executive detects the due time of the next activity, and then uses a repeated scan to find activities which will happen at that time (Pidd, 1998). The process (fig.5) of an activity-based simulation is as follows:

Check the event calendar to find the time of the next activity and move the simulation clock to this time.

Repeatedly scan through the activities to find all activities which are due and able to execute at that time. Then, executive these activities.

These two phases are repeated until the simulation is over.

Figure 5: An activity-based executive (Pidd, 1998)

Process-based approaches:

Process-based approach is the most popular method applied around the world, although the users of the simulation software cannot see it obviously (Pidd, 1998). Process-based simulation is different from event-based simulation and activity-based simulation, because it takes the whole process of an entity as the basic building block (Pidd, 1998).

In a process-based simulation, the executive might maintain a record for each entity which contains two fields: re-activation time and next re-activation point (Pidd, 1998). Then the executive uses two lists to maintain these records:

Future events list. This list contains records of entities whose progress is unconditional delayed. These entities are scheduled by their re-activation time. In addition, their re-activation time should be ahead of the current simulation time.

Current events list. This list contains records of two types of entities. The first type is the entities which have been unconditional delayed and their re-activation time is due at current simulation time. The second type is the entities that are due at the current simulation time, but subject to conditional delays.

These two lists are used in the three phases (fig.6) of operating a process-based simulation, details are as follows:

Phase one: future events scan. The executive scans the future events list and determines the due time of the next event. Then, the simulation clock is moved to this new time.

Phase two: move between lists. Those entities whose re-activation time is equal to current simulation time are moved from the future events list to the current events list.

Phase three: current events scan. The entities on the current events list will be moved on if the conditions permit, and these entities will either finish their process or be halted by a conditional or unconditional delay. If the delay is unconditional, entities will be moved to the future events list and the executive will record their re-activation points.

These three phases are run iteratively until the simulation is over.

Figure 6: A process-based executive (Pidd, 1998)

Three-phase approach

The three-phase approached was first presented by Tocher (1963). It is used by a number of commercial simulation software packages; however, other approaches are commonly applied by many other simulation software packages as well. In the three-phase approach, the activities are classified in two categories:

B (book-keeping or bound) activities: this type of activities has a starting or finishing time that can be predicted in advance (Pidd, 1998). Therefore, these activities can be scheduled directly, and the simulation model will execute these activities exactly when the simulation clock reaches the corresponding scheduled time points (e.g., a machine produces a component at a constant rate).

C (conditional) activities: this type of activities has no exact and constant starting or finishing time, in other words, the execution of these activities is not depended on the simulation clock but on the condition of each activity (e.g., the service can only be started when there is a customer in the system and also the server is idle).

These two types of activities are executed based on the three phases A, B and C, as shown in the figure 2. The simulation model executes the three phases respectively and iteratively until it is completed. Details of phase A, B and C are described as follows:

Phase A: This phase is also known as the ‘time scan’. The executive scans its event list and determines the time of the next event. Then, the simulation clock moves to that point and keeps time fixed until the next phase A.

Phase B: In this phase, all B-type activities due at the time determined in the phase A are executed.

Phase C: In this phase, the C-type activities whose conditions are met are executed firstly. Then, the remainder activities are re-scanned and those whose conditions are met are executed. Repeat this execution until no activity left.

The simulation operates these 3 phases iteratively until it is complete. This three-phase approach makes the simulation model more efficiently and it has good performance especially in the complex resource problems (Banks, Carson & Nelson, 1999).

Figure 2: A three-phase executive (Pidd, 1998)

2.4 Discrete-even simulation language and software

Simulation languages play a significant role on the development and execution of the simulations of complex systems (Banks, Carson & Nelson, 1999). According to Reed and Leavengood (2003), simulation languages can be classified into three categories: General purpose languages, Special purpose simulation languages, and Simulators.

Since the discrete-event simulation is a particular type of scientific computing, any general purpose languages which are suitable for scientific computing are usually appropriate for discrete-event simulations (Leemis and Park, 2004). Similarly, Shannon (1975) describes that general purpose languages are not just developed for simulations, but for a wide class of problems. Although general purpose languages were not initially designed for simulations, it is the first programming languages used by simulations. These languages include FORTRAN, C, C++, BASIC, JAVA, and PASCAL. However, in order to build a simulation model for a complex system, advanced programming skills in a specific computing language and massive time are required.

With the improvement of simulation technology, a number of special purpose simulation languages have been developed for the convenience of building simulation models, such as GPSS, SIMAN, SLAM, and SIMSCRIPT II.5. These languages enable users to build models with less programming, as they usually contain built-in modules (e.g., Random number generators, probability distributions, and modelling elements).

Simulators are a relatively new type of simulation software, which contains a graphical interface and many built-in modules. Some popular simulators are PROMODEL, SIMFACTORY, FACTOR, and WITNESS. With the help of these simulators, users can simulate systems with little or no programming required. Moreover, the time required is reduced dramatically by using a simulator. However, there are some limitations within these simulators. For instance, simulators are not as flexible as languages because users can model only system configurations allowed by the default system features (Reed and Leavengood, 2003).

However, as the number of simulation software packages keeps increasing, a critical issue arises, that is, the selection of appropriate simulation software. Arisha and Baradie (2002) describe that the selection process of the appropriate simulation software can be regarded as a milestone of the overall simulation process. Therefore, the selection of the simulation software is one of the most important factors to the success of simulation projects. However, selecting appropriate simulation software from massive packages is not an easy task (Banks, 1991).

Many researches have been performed on the selection of appropriate simulation software. For instance, Arisha and Baradie (2002) have developed a checklist to guide the selection of appropriate software; Banks (1991) provides a number of advices to help users determine which simulation modelling tool is the right one. Besides, Robinson (2004) summarises the process of software selection as follows:

Step 1: determine the modelling requirements

Step 2: survey and shortlist the simulation software

Step3: establish evaluation criteria

Step4: evaluate the software in terms of criteria

Step5: choose simulation software

The selection process described above is not linear (e.g., moving from step 1 through to step 5), as some of steps might need to repeat several times.

2.5 Applications of discrete-event simulation

Simulation modelling is a powerful tool which is widely used in the fields of operational research, management science and engineering. As one of the most important type of the simulation models, the discrete-event simulation is quite widely applied in a number of companies. According to a survey undertaken by Ingemansson, Bolmsjö and Harlin (2002), among the companies which have applied simulation technology, 79% of them answered that discrete-event simulation helps them for decision-making processes. Moreover, the areas of application of the discrete-event simulation are quite extensive. Based on the recent presentations of the Winter Simulation Conference (WSC, 2012), a list but not an exhaustive list of the main areas of applying discrete-event simulation is shown as below:

Manufacturing systems, which include manufacturing system design (e.g., material handling system design) and manufacturing system operation (e.g., production planning and scheduling, operational policies, and performance analysis).

Healthcare systems, which include design of healthcare system, simulation of patient flow, the care life cycle, healthcare capacity planning, healthcare operations management, simulation of emergency department et al.

Military, which includes military analysis, combat modelling and mission analysis, Simulation-Enhanced Military Testing, Military Logistics, Defence and Security Applications et al.

Construction systems, which includes construction operations, construction scheduling, energy simulations, and simulation in health and safety.

Transportation systems, which includes transport networks simulation.

Social science and organisation, which include economics and management, simulation in planning, simulation in social behaviour.

Environmental systems, which includes power grid simulations, life cycle assessment and traffic simulations.

From the list above, it can be seen that the application areas of the discrete-event simulation are quite wide. However, the applications in manufacturing system are more widely than in any other field (Law, 2007). Hence, this paper discusses more details about simulation of manufacturing system in the next chapter.

3 Simulation of Manufacturing System

3.1 Manufacturing simulations

Manufacturing sector is one of the earliest users of the simulation technology (Oakshoott, 1997). Babulak and Wang (2010) also describe that discrete-event simulation is traditionally applied widely in the industry. Moreover, Law (2007) points out that simulation technology is more widely used in manufacturing systems than in any other fields. This is due to the increasing competition in the manufacturing industry, particular issues exist in manufacturing system and benefits can be obtained by using simulation technology. The following subsections describe the definition of manufacturing systems, characteristics of a manufacturing system model, objective of simulation in manufacturing system, and simulators and languages for simulating manufacturing systems..

3.2 Definition of manufacturing systems

In order to define the concept of manufacturing systems, it is necessary to explain the terms ‘manufacturing’ and ‘system’ separately first. Manufacturing refers to the process of using labour and capital to convert raw materials into products, while system means the collection of a collection of interacting components or processes (Oakshott, 1997). Hence, the manufacturing system can be defined as a collection of components or processes that convert raw materials into products. In addition, manufacturing systems can be identified at four distinct levels (Rooda & Vervoort, 2005):

Factory level. At this level, the manufacturing system refers to the factory and plant. The elements of the system are areas and machines.

Area level. At this level, the manufacturing system refers to an area of the factory which includes groups of machines. The elements of the system are the individual machines

Cell level. At this level, the manufacturing system refers to a group of machines which are always scheduled as a single entity.

Machine level. At this level, the manufacturing system refers to individual machine or single equipment. The elements of the system are the components of the equipment.

In this paper, the manufacturing system is only taken into account at area level and cell level.

3.3 Manufacturing systems modelling

According to the summary of Babulak and Wang (2010), the discrete-event simulation is mainly applied to the following areas of the manufacturing sector:

Design the new manufacturing system and evaluate its performance.

Improve the performance of the existing manufacturing system.

Find and construct the optimal operational policies.

Support to production planning and scheduling.

In addition, most manufacturing systems can be modelled as dynamic and discrete-event simulation models (Reed & Leavengood, 2003). And the main application of simulation technology in manufacturing sectors can be classified into two categories: manufacturing system design and manufacturing system operations (Simth, 2003). For each categorise, there are a several subjects involved it, as shown below:

Manufacturing system design

General system design and facility design/layout

Material handling system design

Cellular manufacturing system design

Flexible manufacturing system design

Manufacturing system operation

Operations planning and scheduling

Real-time control

Operation policy

Performance analysis

In addition, in order to develop appropriate simulation models for manufacturing systems, a number of systems’ characteristics need to be taken into account (Banks, Carson & Nelson, 1999), as shown in the table below:

Models of manufacturing systems

Manufacturing system design

Components

Details

Physical layout

Labour

Shift schedules, Job duties and certification

Equipment

Rates and capacities, Breakdowns (time to failure, time to repair and resources needed for repair)

Maintenance

PM schedule, Time and resources required, Tooling and fixtures

Work centres

Processing, Assembly, Disassembly

Manufacturing system operation

Product

Product flow, Routing and resources needed, Bill of materials

Production schedules

Made-to-stock, Made- to-order (customer orders, line items and quantities)

Production control

Assignment of jobs to work areas, Task selection at work centres, Routing decisions

Suppliers

Ordering, Receipt and storage, Delivery to work centres

Storage

Supplies, Spare parts, Work-in-process, Final goods

Packing and shipping

Order consolidation, paperwork, loading trailers

Table 1: Characteristics of a manufacturing system model

3.4 Objectives of simulation in manufacturing systems

The most significant purpose of using simulation technology in manufacturing systems is providing managers and engineers a system wide view of the effect of local changes on the system (Law, 2007). With the help of simulation models, the impact of changes at a particular workstation can be predicted, and also the influence on the overall performance of the manufacturing system can be evaluated.

Besides the purpose above, there are a number of special objectives of using simulation modelling in manufacturing systems, such as improve productivity of the system, reduce the capital requirements (buildings, machines, tools, resources, etc.) or operating costs, reduce the failure rate of individual machines and overall systems, increase on-time deliveries and so on.

3.5 Simulators and language for manufacturing systems

Manufacturing systems always contains complex characteristics which are difficult to simulate (Banks, Carson & Nelson, 1999). In order to meet this purpose, some specialise simulators have been developed, such as SIMUFACTORY II.5, ProModel, AutoMod, Tayor II, AIM, and WITNESS. However, different software have different features, advantages, and disadvantages. In order to select the appropriate simulation software for simulation projects, a number of considerations need to be taken into account. Moreover, some guidences on the selection of simulation software are discussed in the chapter 2 of this paper.

4 Conclusion and future work

Simulation modelling is one of the most widespread tools in operational research, management science and engineering. It helps people to understand the behaviour of complex systems and evaluate the system performance. Particularly, simulation modelling can be applied to investigate potential impacts of changes in the various aspects of the system. As one of the most important simulation models, discrete-event simulation is applied in a wide range of areas, especially in the manufacturing sectors. With the help of discrete-event simulation, engineers and manager are able to improve the design of the manufacturing system as well as optimise its operations.

However, there are some issued need to be further explored, such as analysis of simulation data, verification and validation of simulation models, and evaluation of alternative system design.