COMPUTER INTEGRATED MANUFACTURING:FLEXIBLE MANUFACTURING SYSTEMS

FLEXIBLE MANUFACTURING SYSTEMS

Flexible Manufacturing Systems (FMS) is a manufacturing system with a high degree of flexibility. It was developed due to the need to increase productivity, improve product quality, and reduce cost for product production under the constraints of various uncertainties or disturbances both internal and external to the manufacturing system.

Flexibility and Components of FMS

Flexibility of Manufacturing System

A number of papers have studied different aspects of FMS. Gupta and Goyal (1989) provide a comprehensive review of the literature on flexibility. Flexibility can be defined as a collection of properties of a manufacturing system that supports changes in production activities or capabilities (Carter 1986).

In a manufacturing system, various types of flexibility are needed to fulfill different requirements. The types most discussed are machine flexibility, routing flexibility, process flexibility, product flex- ibility, production flexibility, and expansion flexibility. Machine flexibility is the capability of a ma- chine to perform a variety of operations on a variety of part types and sizes. Machine flexibility can reduce the changeover frequency, setup time, and tool-changing time, hence reducing the lead time and making small-lot-size production more economic. Machine flexibility is the basis for routing and process flexibility.

Routing flexibility provides the chance for a part to be manufactured or assembled along alter- native routes. Routing flexibility is required to manage shop-floor uncertainties caused by such prob- lems as machine breakdown, tool error, and controller failure. It can also be used to tackle the problems caused by external events such as change of product mix or product due date and emergency product introduction. These changes alter machine workloads and cause bottlenecks. The use of alternative routing helps to solve these problems and finally increase productivity.

Process flexibility, also called mix flexibility, is the ability to absorb changes in the product mix by performing similar operations or producing similar produces or parts on multipurpose, adaptable CNC machining centers.

Product flexibility, also known as mix-change flexibility, is the ability to change over to a new set of products economically and quickly in response to markets or engineering changes or even to operate on a market-to-order basis. In the current global market, high product flexibility is a very important factor for a company to compete.

Expansion flexibility is the ability to change a manufacturing system with a view to accommo- dating a changed product envelope. It has become more important in the current agile manufacturing era. Improving expansion flexibility can significantly reduce system expansion or change cost, shorten system reconfiguration time, and hence shorten the delivery time for new products.

FMS Definition and Components

An FMS is an automated, mid-volume, mid-variety, central computer-controlled manufacturing sys- tem. It can be used to produce a variety of products with virtually no time lost for changeover from one product to the next. Sometimes FMS can be defined as ‘‘a set of machines in which parts are automatically transported under computer control from one machine to another for processing’’ (Jha 1991).

A more formal definition of FMS is that it consists of a group of programmable production machines integrated with automated material-handling equipment and under the direction of a central controller to produce a variety of parts at nonuniform production rates, batch sizes, and quantities (Jha 1991).

From this definition, it can be seen that an FMS is composed of automated machines, material- handling systems, and control systems. In general, the components of an FMS can be classified as follows:

1. Automated manufacturing devices include machining centers with automatic tool interchange ability, measuring machines, and machines for washing parts. They can perform multiple func- tions according to the NC instructions and thus fulfill the parts-fabrication task with great flexibility. In an FMS, the number of automated machining centers is normally greater than or at least equal to two.

2. Automated material-handling systems include load / unload stations, high-bay storage, buffers, robots, and material-transfer devices. The material-transfer devices can be automatic guided vehicles, transfer lines, robots, or a combination of these devices. Automated material-handling systems are used to prepare, store, and transfer materials (raw materials, unfinished parts, and finished parts) between different machining centers, load / unload stations, buffers, and high- bay storage.

3. Automated tool systems are composed of tool setup devices, central tool storage, tool- management systems, and tool-transfer systems. All are used to prepare tools for the machining centers as well as transfer tools between machining centers and the central tool storage.

4. Computer control systems are composed of computers and control software. The control soft- ware fulfills the functions of task planning, job scheduling, job monitoring, and machine con- trolling of the FMS.

Figure 12 shows the FMS layout at the State CIMS Engineering Research Center (CIMS-ERC) of China. (HMC stands for horizontal machining center and VMC stands for vertical machining center.)

Another example of FMS is shown in Figure 13, from Kingdream. This system produces oil well drill bits, mining bits, hammer drills, high-pressure drills, and so on.

General FMS Considerations

Although FMS was originally developed for metal-cutting applications, its principles are more widely applicable. It now covers a wide spectrum of manufacturing activities, such as machining, sheet metal working, welding, fabricating, and assembly.

The research areas involved in the design, implementation, and operation of an FMS are very broad. Much research has been conducted and extensive results obtained. In this section, we present the research topics, problems to be solved, and methods that can be used in solving the problems.

FMS Design

FMS is a capital investment-intensive and complex system. For the best economic benefits, an FMS should be carefully designed. The design decisions to be made regarding FMS implementation cover

system configuration and layout, manufacturing devices, material-handling systems, central tool stor- age, buffers, and high-bay storage.

Before these decisions can be made, the part types to be made, the processes needed to make them, and the possible numbers of processing parts (workload) should first be determined. Based on these basic requirements, the number of machines and their abilities, tools, buffers, and storage system can be roughly determined. A rough system layout and material-handling system can be designed. The designed FMS can be simulated using an FMS simulation tool to test its ability to fulfill the requirements.

The design of an FMS is a system approach. Besides the above-mentioned basic requirements for part manufacturing, many other factors should be considered in designing an FMS. An economic assessment should always be done for every FMS plan obtained. System reliability, productivity, and performance evaluation should also be done for every FMS plan. The design of FMS is an iterative process that requires many experts from different disciplines to work together. Many alternative plans are compared and modified before an optimized plan is decided upon.

Talavage and Hannam (1988) summarize the work of other researchers in FMS design method- ology and present a five-step approach to FMS design:

1. Development of goals

2. Establishment of criteria on which goal achievement can be judged

3. Development of alternative candidate solutions

4. Ranking of alternatives by applying the criteria to the alternate solutions

5. Iteration of the above four steps to obtain a deeper analysis of alternate solutions and to converge on an acceptable solution

Other considerations regarding FMS design can be found in Tetzlaff (1990).

FMS Planning, Scheduling, and Control

Planning, scheduling, and control are important and difficult problems in FMS operations. A good planning and scheduling system will improve FMS operation efficiency and yield economic benefits. Extensive research and development of FMS planning and scheduling has been done. The general optimization indexes are:

1. Maximizing the productivity at certain period of time

2. Minimizing the makespan for a group of parts

3. Minimizing the cost for parts manufacturing

4. Maximizing the utility of key manufacturing devices

5. Minimizing the work in progress

6. Minimizing the production time for certain parts

7. Satisfying the due dates of parts

Figure 14 presents a function model for FMS planning, scheduling, and resource management.

The resource management and real-time control functions of FMS are closely related to the dynamic scheduling system. The resource-management system should be activated by a dynamic scheduling system to allocate resources to production process to achieve real-time control for FMS.

The resources to be controlled involve tools, automatic guided vehicles, pallets and fixtures, NC files, and human resources.

Planning Planning seeks to find the best production plan for the parts entered into the FMS. Its aim is to make an optimized shift production plan according to the shop-order and part- due dates. The FMS planning system receives the shop-order plan in the weekly time scale from the MRPII system. According to the product due dates, it analyzes the shop order and generates a daily or shift production plan. Group technology is used for grouping parts into families of parts. The capacity requirement is calculated for every shift plan generated. Capacity balance and adjustment work should be carried out if the required capacity is higher than that provided by machines.

After feasibility analysis, capacity balancing, and optimization, a shift plan is generated. The shift plan gives detailed information for the following questions:

1. What kind of parts will be machined?

2. In what sequence will the parts enter the FMS?

3. What operations are needed to process the parts? What is the operation sequence?

4. What are the start time and complete time for processed parts?

5. What materials are needed? In what time?

6. What kinds of tool are needed?

Static Scheduling Static scheduling is the refinement of the shift production plan. It seeks to optimize machine utility and reduce system setup time. Three functions are performed by a static scheduling system: part grouping, workload allocating and balancing, and part static sequencing. Because all these functions are performed before production starts, static scheduling is also called off-line sequencing.

A number of factors affecting production sequence should be taken into account for static sched- uling, such as the part process property, FMS structure, and optimization index. The part process property determines what kind of production method should be used. Flow-shop, flexible-flow-line, and job-shop are three major strategies for producing parts. Different methods can be used to generate static scheduling for the different production strategies.

The second factor affecting static scheduling is FMS structure. The main structural properties are whether a central tool system, a fixture system, or bottleneck devices are present. The third factor is the optimization index chosen. The general optimization index is a combination of several optimi- zation indexes, that is, the FMS static scheduling is a multiobjective optimization process.

The following parameters have an important effect on implementing optimal static scheduling.

1. Time distribution, such as time distributions for part arrival, tool setup, part fixture, part transfer, machine failure, and delivery time

2. Shop conditions, such as device type, transfer system, storage method, shop layout, and device condition

3. Shop control conventions, such as priority rule, operation method, hybrid processing route, task decomposition, performance-evaluation method, and workload

4. Alternate processing route, such as alternate processing device, alternate processing routing, and alternate processing sequence

A great deal of literature about static scheduling algorithms and systems can be found in the academic journals on FMS, operations research, and manufacturing technology and IEEE magazines on systems, robotics, and automatic control.

Dynamic Scheduling Dynamic scheduling is used to control the operations of FMS according to the real-time status of the AFMS. It is a real-time (online) system that focuses on solving uncertainty problems such as device failures, bottlenecks on certain machines, workload unbalance, and resource-allocation conflict. These problems are not anticipated by off-line static scheduling. They can only be solved using real-time dynamic scheduling or rescheduling.

Three strategies can be used to complete the rescheduling functions. The first is periodical sched- uling. A certain time interval must be set as a production cycle. A periodical scheduling system calculates a period operation sequence before the next period starts. The sequence is the job list execution instructions followed by the FMS. The second strategy is continuous scheduling, which monitors the FMS and executes scheduling whenever an event (such as the arrival of a new part or a machine completing the production of a part) occurs and the system states has been changed. Since the calculation of work content is effective for rescheduling FMS operations for every event (so as to get optimal scheduling at every point), the third strategy, hybrid scheduling, is frequently used. The hybrid strategy combines periodical and continuous scheduling so that only when an unexpected event occurs is the continuous scheduling algorithm used. Otherwise, periodical scheduling is exe- cuted at certain intervals.

For a dynamic manufacturing environment with possible disturbances both internal and external to the FMS, dynamic scheduling seeks to optimize the sequencing for the queue before the device is manufactured. Because the dynamic scheduling of an FMS is an NP-hard problem, it is impossible to find the optimal solution in a short time, especially for continuous scheduling with a very high speed requirement. Normally a suboptimal solution is used in real-time FMS operations. A number of heuristic rules are frequently used for finding the suboptimal solutions in dynamic scheduling. The heuristic rules that are frequently used are:

1. RANDOM: assigns a random priority to every part entering the queue and selects a part with smallest priority to be processed

2. FIFO (LIFO): first-in-first-out (last-in-first-out)

3. SPT (LPT ): selects the part that has the smallest (largest) current operation processing time to be processed

4. FOPNR (MOPNR): selects the part that has the fewest (most) remaining operations to be processed

5. LWKR (MWKR): selects the part that has the smallest (largest) remaining processing time to be processed

6. DDATE: selects the part that has the earliest due date to be processed

7. SLACK: selects the part that has the smallest slack time (due date minus remaining processing time) to be processed

In most cases, several rules will be used in a dynamic scheduling system to reach the satisfied sequencing solution. Besides rule-based scheduling, simulation-based and knowledge-based sched- uling systems are also widely used.

FMS Modeling and Simulation

Modeling and simulation are important topics for both design and operation of FMS. FMS modeling is the basis for simulation, analysis, planning, and scheduling. Because FMS is a typical discrete event dynamic system (DEDS), a number of methods for DEDS modeling and analysis can be used to model an FMS, such as Petri nets, network of queues (Agrawal 1985), and activity cycle diagram (Carrie 1988). This section briefly introduces Petri nets, their application in FMS modeling, and the FMS simulation method.

Petri Nets and Their Application in FMS Modeling A Petri net (PN) may be identified as a particular kind of bipartite directed graphs populated by three types of objects. These objects are places, transitions, and directed arcs connecting places to transitions and transitions to places. Pictorially, places are depicted by circles, transitions by bars or boxes. A place is an input place to a transition if a directed arc exists connecting this place to the transition. A place is an output place of a transition if a directed arc exists connecting the transition to the place. Figure 15 represents a simple PN. Where places p1 and p2 are input places to transition t1, place p3 is the output place of t1.

The state of the modeled system is represented by the tokens (small dots within the places) in every place. For example, in Figure 15, a small dot in place p1 means components available. The change of the states represents the system evolution. State changing is brought by firing a transition. The result of firing a transition is that for every place connected with the transition, after the firing of the transition, a token will be removed from its input place and a token will be added to its output place. In the example of Figure 15, the firing of transition t1 will cause the tokens in places p1, p2 to disappear and a token to be added to place p3.

Due to the advantages of its formal theory background, natural link with DEDS, and mature simulation tool, PN is well suited to FMS modeling. Figure 16 shows a two-machine production line to demonstrate the modeling of FMS using PN.

The production line consists of two machines (M1 and M2), two robots (R1 and R2), and two conveyors. Each machine is serviced by a dedicated robot that performs the load / unload task. One conveyor is used to transport workpieces, a maximum two at one time. The other conveyor is used to transport empty pallets. Three pallets are available in the system. Each workpiece is machined on M1 and then on M2. The machining time is 10 time units on M1 and 16 time units on M2. The load and unload tasks takes 1 time unit.

As with modeling general FMS or other systems, the modeling of this system using PN takes several steps:

1. Major activities are identified. In this example, they are R1 loading, M1 processing, R1 un- loading, R2 loading, M2 processing, and R2 unloading. The resources are raw materials with pallets, conveyors, M1, M2, R1, R2.

2. The relationships between the four major activities form a sequential order.

3. A partial PN model is defined to describe the four major activities and their relations as shown in Figure 17(a), where four transitions are used to represent four short operations, i.e., R1 loading, R1 unloading, R2 loading, and R2 unloading. Two places are used to represent two long operations, i.e., M1 and M2 processing.

4. Through a stepwise process, gradually adding resources, constraints, and links to the partial PN model will finally form the refined model as shown in Figure 17(b).

5. The model is checked to see whether it satisfies the specification. The PN simulation tool can also be used in this phase to check the model. If some problems are found, the model will be modified.

FMS Simulation Simulation is a useful computer technology in FMS modeling, design, and operation. Simulation modeling allows real-world objects to be described in FMS, such as moving of workpieces from one place to another. There are three approaches to simulation modeling for FMS. The first is network or graphical models, where some objects (such as machines) may be represented by graphical symbols placed in the same physical relationship to each other as the corresponding machines are in the real world. The graphical aspects of this kind of models are relatively easy to specify, and once completed they also provide a communication vehicle for the system design that can be readily understood by a variety of people. SLAM (Pritsker 1984) and SIMAN (Pegden 1982) are two widely used network modeling tools for FMS.

The second approach to FMS modeling is data-driven simulation. The model consists of only (or mainly) numerical data, usually representing, for example, a simple count of machines in a system or a table of operation times for each process on the route of a given part type. The nature of these data is such that, if they were collected in the factory information system, it would only be necessary to access them and place them in proper format in order to run a simulation of the corresponding

real-world system. This concept is quite close to automated simulation. It has the ultimate ease of use. The first such program for FMS was developed at Purdue, the general computerized manufac- turing system (GCMS) simulator (Talavage and Lenz 1977).

The third approach for FMS modeling uses a base programming language, such as SIMULA and SIMSCRIPT, which provides more model-specific constructs that can be used to build a simulation model. This approach thus has a much stronger modeling capability. Unfortunately, it is not widely used. One reason may be that few people know it well enough to use it.

Another method for DEDS simulation, called activity cycle diagram (ACD), can also be used in FMS simulation. This is a diagram used in defining the logic of a simulation model. It is equivalent to a flowchart in a general-purpose computer program. The ACD shows the cycle for every entity in the model. Conventions for drawing ACDs are as follows:

1. Each type of entity has an activity cycle.

2. The cycle consists of activities and queues.

3. Activities and queues alternate in the cycle.

4. The cycle is closed.

5. Activities are depicted by rectangles and queues by circles or ellipses

Figure 18 presents an ACD for a machine shop. Jobs are arriving from the outside environment. Jobs are waiting in a queue for the machine. As soon as the machine is available, a job goes to the machine for processing. Once processing is over, the job again joins a queue waiting to be dispatched.

Because ACDs give a better understanding of the FMS to be simulated, they are widely used for FMS simulation.

Benefits and Limitations of FMS

FMS offers manufacturers more than just a flexible manufacturing system. It offers a concept for improving productivity in mid-variety, mid-volume production situations, an entire strategy for chang- ing company operations ranging from internal purchasing and ordering procedures to distribution and marketing. The benefits of FMS can be summarized as follows:

1. Improved manufacturing system flexibility

2. Improved product quality, increased equipment utility

3. Reduced equipment cost, work-in-progress, labor cost, and floor space

4. Shortened lead times and improved market response speed

5. Financial benefits from the above

Talavage and Hannam (1988) contains a chapter discussing the economic justification for FMS.

The difficulties with FMS should also be given attention. First, FMS is expensive, normally requiring large capital resources. Even if a company is able to afford the investment, FMS may not be financially beneficial if the company does not have much product variety and volume. Second, the design, implementation, and operation of FMS is a quite complex process. Money may be lost if any of the work in the process is not well done. Third, the rapidly changing market may compel the company to change its product. This change may have a negative impact on the production system, causing the large investment in FMS not to be returned before it is abandoned.