Modeling Methodology

FMSs are characterized by concurrency, resource sharing, routing flexibility, limited buffer sizes, and variable lot sizes. For interaction activities, and the coordination of individual units, a descriptive and dynamic modeling tool is required to model in detail the concurrency, and synchronization in the system with respect to time (Lin & Lee, 1997).

Owing to the complexity of these real-world manufacturing systems, it is very difficult to contain all the information necessary for scheduling in a unique model.

According to Lin, Fan, and Loiacono (2004), three views are required to be introduced to compose real-life dynamic scheduling model as the process view, the resource view, and the job view. Process view is constituted multi processes, each of which defines activities and their process constraints required to produce one type of product. Resource view involves the individual resource and resource pool. The resource pool is a classification of individual resources according to their functions or physical positions so that the individual resources in the same resource pool can be substituted for each other. This feature makes the model flexible in dealing with multi-resource sharing problem. Job view describes the properties of the jobs to be produced, which includes the product type, order arrival time, due date, order size, process plan, priority, and so on (Lin et al., 2004). Finally, control flow diagrams are employed to represent control issues such as decisions, dispatching and sequences.

The views and the relationship among them present various constraints and conditions of the scheduling problem as shown in Figure 3.3.

Processes

Process View

Composed

Process

Activities Relations

Properties Resource

Mapping Behaviour

Description

Job View

Process Jobs

Properties

Resource View

Mapping

Individual Resource Resource Pool Allocate

Figure 3.3 Fundamental structure of real-world scheduling problem (Lin et al., 2004)

A manufacturing system can be viewed as a collection of objects with rules that govern their dynamics and interactions to generate desired objects (products). The objects can be represented graphically as simplified images, icons, and stored in a data base as members of a class of similar objects, sharing common properties (Zhou

& Venkatesh, 1998). Therefore, Object Oriented Design (OOD) concepts can be employed to design of shop floor scheduling and control system in FMSs, and provide a way of defining manufacturing system using models organized around real-world concepts (Booth, 1998). OOD methodologies can offer the reusability, extendibility, and modifiability of software design.

By using OOD concepts, the properties and behavior of the real-world object are modeled by the data/attributes and methods/operations of the corresponding system object. It should be noted that real-world objects include not only physical objects, such as a machine, material handling device, robot, operator, and so on, but also logical objects, such as a scheduling process, routing process, product data, orders, bill of materials, and so on (Venkatesh & Zhou, 1998).

The fundamental building block of OOD is an “object” that contains both a data structure and a collection of related procedures. Procedures are also called operations or methods. Objects interact with each other by sending messages or by calls to their interfaces. Objects with identical data structure (attributes) and behavior (operations) can be grouped into a “class”. Each object is an instance of its class.

The most widely used OOD methodology is the object modeling technique methodology, known as OMT methodology, which is gaining increasing recognition as a powerful and robust system development methodology. The OMT methodology consists in developing three orthogonal models, as follows (Booth, 1998; Venkatesh

& Zhou, 1998):

1. Object model: It divides the application into object classes and shows the static, structural aspects of the system in which objects, their identity and

their attributes, relationships, and operations are described in detail. The relationships (interfaces) among the classes are described by the class structure, and the behavior of the objects is defined by operations associated with the object class.

2. Dynamic model: It represents the temporal, behavioral aspects of a system and shows the way the system behaves with internal and external events by capturing the time-dependent behavior of the system.

3. Functional model: It represents the transformational aspects of a system and shows how to process the data flow in the system during each event or action.

PNs and OOD concepts are complementary to achieve the goal of system development at incremental stage, and enable us to model the complex manufacturing systems in detail, so that the strategies to operate these systems effectively can be applied in a more realistic environment. In this thesis, OOD concepts are employed to design of shop floor scheduling and control system in FMSs, and high-level PNs are used to model dynamic behavior of the system. The properties and behavior of the objects are modeled by the data/attribute and methods /operations. OOD methodology used for development of the DSS is summarized in the following steps (Chen & Chen, 2003; Venkatesh & Zhou, 1998):

1. Apply the OOD concepts to find objects in an FMS and to model the static relations among them by developing an object modeling technique diagram (OMT). The OMT diagram is used to represent explicitly different kinds of static relations such as generalization, aggregation, and association among the objects in FMS.

2. Use PNs to model the dynamic behavior of the FMS based on the static relations of the OMT diagram. Part, operator, and transpallet flows in the OMT diagram are transformed into token color classes in CPN model.

And FMS objects WSs, Loading/Unloading station, and part routing control object are modeled as CPN classes.

3. According to the system configuration, (i.e. number of workstations, number of transportation vehicles between workstations, capacity of load/unload station and buffers of workstations, number of operators, process plans, and batch sizes), FMS objects are derived from the corresponding CPN classes.

4. Integrate all the instance of the CPN classes into an integrated system model by linking the input and output places of each CPN object.

CHAPTER FOUR

DESIGN AND IMPLEMENTATION OF THE PROPOSED DECISION SUPPORT SYSTEM BASED ON HIGH-LEVEL PETRI NETS AND

OBJECT-ORIENTED APPROACH

The conceptual framework of proposed decision support system for real-time scheduling of FMSs has been discussed in Chapter 3. Its modeling methodology using high level PNs and Object-oriented design approaches has also been presented.

This chapter is devoted to the illustration of the modeling methodology discussed in the third chapter. Firstly, object modeling diagram of the system is constructed and a heuristic rule base is proposed to solve the resource contention problem, then the dynamic behavior of the system is formulated by high-level PNs. Subsequently, the proposed rule-based system is compared with common dispatching rules with respect to part flow time and tardiness related performance measures, then the simulation experiments were carried out under different experimental conditions to analyze the impact of the different levels of independent variables on performance measures.