Flexible Manufacturing Systems

The design problem arises subsequent to the recognition of a need for automation. In one context, the need arises in an existing facility that is already engaged in the production activity and that wishes to convert part of the production system to an automated system. In this case the design must be done appropriately so that integration of the FMS with the factory wide production activity can be achieved feasibly. The second context arises when the conceived system is totally automated and thus integration issues relating to automation and conventional production are relatively absent.

The process of designing an FMS is complicated by the cost and difficulty of acquiring information. Such information usually has relatively short shelf life due to the rapid advance of technology.

The first step in the design process is for managers and engineers to define goals and develop a conceptual design for processes, equipment, control systems, and information systems. Initial versions of the conceptual design are often unrealistic, especially when substantial new technology is being proposed for the firm. Typical reasons fort this situation include lack of familiarity with modern manufacturing technology, incentive systems which discourage innovativeness, overly optimistic assumptions, and the limited level of detail which can be considered in a conceptual design.

Next, a specification for the functionality and performance required for each part of the system must be written. This document needs to be as specific as possible and is often several hundred pages long. These specifications along with a request for quotation are sent to vendors who have been screened for ability to provide the specified systems, sub-systems, equipment, or services. The diversity of manufacturing processes and the specialization of vendors generally result in a large number of vendors contributing to the design and implementation of the factory.

Detailed technical proposals and firm offering prices must be developed by vendors for a variety of items such as machines, tools, fixtures, gauges, sensors, robots, controllers, CNC programming, material handling equipment, computer hardware and software, communications wiring and equipment, building construction, and training. Responding to a request for quotation usually requires considerable engineering and managerial effort which results in considerable cost.

Vendors are not obliged to quote and may do so only if they want the business and believe that they have a chance to get it. A vendor may require the requesting company to fund a study. Alternate specifications and vendors can be studied to accumulate data on cost and performance, but considerable time and effort may be expended in the process. Such thoroughness may be undesirable considering the associated time and cost penalties.

After a vendor is selected, a functional design and a detailed design must be developed. A functional design is a description of functions being performed, inputs and outputs, and the mechanism for providing the functionality. A detailed design includes all information necessary to procure material and build the equipment and systems. The vendor and customer must jointly reconcile and refine the design.

The system hardware: technological specifications, types, and capacities of machines, material handling equipment, and storage system; selection of tools, fixtures, pallets;

Computer system and other control devices: capabilities of host computer, distribution of control between the host and cell controllers;

Layout: Relative location and arrangement of machines, storage areas, work areas, robots, and their connection via a material handling system. This involves finding out which processing modules should be adjacent to each other and how they should be connected with the transfer links of the material handling system to minimize the material handling cost. This cost is of particular relevance for FMS because of its impact on in-process storage space, work-in-progress inventories, the cost of control, and throughput time.

Selection of FMSible part families: This is especially important if the FMS is viewed as a supporting sub-system to factory wide production.

The built-in responsiveness of an FMS to variations in the external demand makes it possible to delegate certain design issues to the setup phase. Decisions on the number of fixtures and pallets, and the number of material handling vehicles may be included in this class.

The first level of planning that specifies production requirements as a function of time is the aggregate plan. The objective is to find monthly production rate and inventory levels that minimize total production and inventory related costs over a planning horizon. In a job shop environment the aggregate plan trades off the cost of several capacity sources such as production in regular time, production in overtime, use of second shift and subcontracting with the cost of inventory and shortages.

The applicability of conventional aggregate planning techniques for non-flexible systems in the FMS case is questionable since they assume a fixed routing of each component in the system. In an FMS it is up to the scheduler to decide upon the specific routing of each component. Furthermore, the FMS is capital intensive and cost of labor is less significant than the cost of capital invested in the system. It is therefore assumed that capacity limits are imposed by the number of operational hours available in a given period. The utilization of capacity is related to the mix of processes scheduled for a given period as each machine can perform a wide mix of operations and the efficiency of each machine depends on the type of operation performed on it. Therefore, to adapt the conventional aggregate planning techniques to FMS, capacity limits as a function of the product mix must be incorporated.

The main output of aggregate planning is a master production schedule which has a built-in flexibility on part mix, production rates, and lot sizes. Consequently, aggregate plans for an FMS yield a more responsive production plan than a conventional one. The flexibility of aggregate plans leads to a greater degree of freedom in the management and control of day-to-day activities in the FMS, while increasing the difficulty of operations planning, due to the greatly increased number of alternatives.

The system setup problem deals with short term planning problems such as part mix selection, operations assignment, tooling, fixture and pallet allocation and routing. The system setup phase constitutes the bridge between the aggregate planning requirements and the scheduling and control of the FMS. Hence the system setup decisions are blended with both aggregate planning and operational control issues and it is difficult to recognize the system setup phase in most practical cases. The arbitrary partitioning of the setup functions into aggregate planning and scheduling is a common mistake in today's FMS applications.

System setup refers to a segment of the master schedule for which certain operational decisions have to be made. During this phase, a conglomeration of interrelated, complex problems needs to be solved. The solution method may be simultaneous, sequential, or iterative, but the key issue is the definition of feasible and well defined goals for the next and subsequent phases. In this respect the setup couples aggregate planning with day-to-day operations of an FMS.

By solving the setup problem, a clear view of the tasks to be performed in the scheduling phase is obtained. The solution of an appropriately formulated setup model will yield the set of parts which will be produced during the planned period, an allocation of tools to machines which will achieve the production goals set by the master schedule, an allocation of fixtures to parts, and an assignment of unit operations to machines. The routing of each part through the system can be directly determined from the output if a fixed operation sequence is specified for each part. Scheduling decisions can be made relatively easily once the routings are available.

After solving the setup problem, the next task is to determine start and completion times for each activity. Though many scheduling issues in conventional job shops and general purpose FMSs are conceptually similar since they meet the same customer needs (i.e. jobs are made to order), significant differences exist in the system characteristics. Some of these differences which are likely to impact the scheduling of FMS are the large number of alternate routings, buffer limitations, low or non-existent setup times, effects of transportation times and imbalance of workloads at individual machines.

Two types of scheduling may be distinguished : a priori scheduling and on-line scheduling. The a priori scheduling problem is a resource constrained job shop scheduling problem which is known to be NP-hard. Hence, frequent solution of the a priori scheduling problem requires the use of heuristic methods. Normally, a priori scheduling is not utilized in real world FMS's. This is mainly due to a lack of understanding of the issues in the system setup phase. The common practice is to implement on-line scheduling algorithms. Such algorithms are based on simple job shop dispatching rules.

The degree of difficulty in solving the scheduling problem is dependent on the level of flexibility built into the system setup model. The more flexibility the setup model proposes, the more difficult the scheduling will be. In return, the scheduling phase has an impact on the setup phase, especially if a time based measure of performance is to be considered. Most performance measures are dependent on activity start and completion times. The performance of the system resulting from a given schedule may dictate a revision of the solution of the setup model. This two way interaction between setup and scheduling is an open area for research.

The control phase deals with the actual operation of the system. Stecke defines the FMS control issues "...to be those associated with monitoring the system, keeping track of the production..." in an on-line environment. These issues include determination and implementation of policies to handle machine tool and other breakdowns, periodic and preventive maintenance, and quality and quantity control of in-process and/or finished goods. Other problems in the control phase are related to on-line data collection and processing. These include tool life and process monitoring, database management, and defining human interfaces with the system.

Due to the inherent flexibility of FMS, the dynamic nature and numerous part-types of it, unlike the traditional transfer line or the conventional job shop, there is no one set program and priority control policy according to which the FMS can operate. It would be more appropriate to have an on-line, real-time control system where the control decisions are made according to the actual state of the FMS.